Regular black holes and the first law of black hole mechanics

1, ∗ 2, †
Sebastian Murk and Ioannis Soranidis
1
Okinawa Institute of Science and Technology, 1919-1 Tancha, Onna-son, Okinawa 904-0495, Japan
2
School of Mathematical and Physical Sciences, Macquarie University, Sydney, New South Wales 2109, Australia
Singularity-free regular black holes are a popular alternative to the singular mathematical black holes pre-
dicted by general relativity. Here, we derive a generic condition that spherically symmetric dynamical regular
black holes must satisfy to be compatible with the first law of black hole mechanics based on an expression for
the surface gravity at the outer horizon. We examine the dynamical generalizations of models typically consid-
ered in the literature and demonstrate that none of them satisfies the condition required for compatibility with
the first law, suggesting that modifications are required to maintain its physical meaning. We show that the need
for corrections is inherently linked to the introduction of a minimal length scale and can therefore be seen as
a direct consequence of the spacetime regularization. We explicitly identify the additional work terms in the
extended first law, comment on their thermodynamic interpretation, and show that the linear coefficient of the
arXiv:2304.05421v3 [gr-qc] 8 Aug 2023

Misner–Sharp mass suffices to determine the relevant thermodynamic properties.

I. INTRODUCTION It is important to note that the original 1973 paper consid-

ered stationary black holes [1]. While the mass loss of an
In their 1973 milestone paper [1], Bardeen, Carter, and evaporating black hole is usually ascribed to the emission of
Hawking introduced four laws of black hole mechanics and Hawking radiation [5, 6], the backreaction of spacetime ge-
elucidated close analogies with the four laws of thermody- ometry is not accounted for in its derivation, which assumes
namics. This important link connecting the two fields has that the underlying geometry is (at least asymptotically) sta-
since proven to be a powerful tool in advancing our under- tionary. This is a physically significant omission: if the back-
standing of black holes. In particular, the physical insights reaction from the Hawking flux is not ignored, variations be-
revealed in the rigorous mathematical derivation of the first tween nearby equilibrium states are no longer accurately de-
law in the integral and differential formalism of Ref. 1 have scribed by the first law.
provided strong motivation for further investigating their ther- To resolve this limitation and go beyond physically unreal-
modynamic properties [2–7]. istic stationary scenarios, the concept of a dynamical (and thus
While the existence of dark massive ultracompact objects quasilocal) horizon has been developed [25–27]. It allows us
has been established beyond any reasonable doubt, the ques- to describe the geometry of an evolving black hole spacetime
tion of whether these objects are black holes is still open [8– and has become an indispensable mathematical tool to accu-
13]. In the absence of a clear answer, singularity-free models rately model dynamical processes such as the formation and
of so-called regular black holes (RBHs) [14–16] have received possible evaporation of black holes, as well as to enable gener-
much attention in recent years, as they provide a way to avoid alizations of the laws of black hole mechanics and their ther-
the nontrivial causal structures inherent to the black holes pre- modynamic interpretation [25, 28], including that of surface
dicted by general relativity (GR). Unlike the singular mathe- gravity as a temperature parameter.
matical black holes of GR, which are bounded by globally de- In this article, we investigate the physical consequences of
fined and physically unobservable event horizons [17], RBH the first law of black hole mechanics for dynamical models
models are characterized by a separate inner (Cauchy) and of RBHs embedded in asymptotically flat spacetimes. We re-
outer (in the case of an evolving RBH spacetime quasilocal, strict our considerations to the case where the entropy scales
e.g. apparent or trapping) horizon. Both the inner and outer with the area of the outer horizon1 . Compared to the evolu-
horizon generally have a nonzero surface gravity. However, tion of the universe, the evaporation of black holes is consid-
inner horizons with nonzero surface gravity are typically un- ered to be a thermodynamically slow process [31]. Therefore,
stable under small perturbations, which gives rise to so-called if the first law is true, the behavior of dynamically evolving
mass inflation instabilities, i.e. exponential instabilities char- black holes should match that prescribed by the first law in
acterized by an exponential growth of the gravitational energy the quasistatic limit. Based on this assertion, we derive a
in a neighborhood of the inner horizon (as measured, for in- generic condition that RBHs must satisfy to be compatible
stance, by evaluating the Weyl scalar) [18–23]. Recently, a with the first law by considering the surface gravity at the
novel “inner-extremal” RBH model that cures the exponential outer horizon. We explicitly test the dynamical generaliza-
instability by making the inner horizon surface gravity vanish, tions of popular models such as those proposed by Bardeen
while maintaining the separation between the inner and outer [32] (Subsec. IV C), Dymnikova [33] (Subsec. IV D), Hay-
horizon, and a nonzero surface gravity at the outer horizon, ward [34] (Subsec. IV B), the model considered in Ref. 35
has been proposed [24]. (Subsec. IV E) which produces the strongest possible correc-
tions to the Schwarzschild geometry while still being com-

∗ sebastian.murk@oist.jp
† ioannis.soranidis@hdr.mq.edu.au 1 Alternatives have been considered, for instance, in Refs. 29 and 30.
 2

patible with its asymptotic behavior, and the aforementioned black hole metrics possess an inner horizon due to the fact that
inner-extremal RBH model [24] (Sec. V). Our analysis shows the outer horizon (which is located close to the classical gravi-
that none of these models is compatible with the conventional tational radius) cannot cross the center r = 0 without creating
form of the first law of black hole mechanics. a curvature singularity [37]. The inner horizon is generated
The remainder of this article is organized as follows: In by the additional hair, i.e. the minimal length scale l that is in-
Sec. II, we introduce mathematical concepts used in the con- troduced to regularize the spacetime [cf. Eqs. (32), (40), (45),
struction of RBHs (Subsec. II A) and review the first law of and (50)] and possibly other parameters such as charge [cf.
black hole mechanics and its relation to surface gravity (Sub- Eqs. (58) and (62)]. Constraints for the a priori undetermined
sec. II B). In Sec. III, we derive a generic condition that dy- function g(v, r) are discussed in what follows.
namical black holes must satisfy to be compatible with the The expansions of ingoing and outgoing radial null
first law of black hole mechanics. Based on this compatibility geodesic congruences are given by
condition, we test the dynamical generalizations of commonly
considered RBH models, and find that none of them satisfies 2 f (v, r)
θ− = − , θ+ = , (4)
the required relation (Sec. IV and Sec. V), suggesting that ei- r r
ther these models do not conform to the first law or modifica-
tions of the first law are required to maintain its essence. In respectively. The existence of a trapped spacetime region is
Sec. VI, we briefly outline the consequences of this result in contingent on the signature of their product θ− θ+ ≶ 0. We
the context of the so-called Page evaporation law. Lastly, we follow the widely used convention proposed in Ref. 38, ac-
discuss the implications of our findings more generally and cording to which the presence of a trapped region bounded
comment on possible directions for future research related to by the outer horizon r+ is signified by θ− θ+ > 0 (i.e. the
nonsingular black hole spacetimes (Sec. VII). Throughout this future-directed expansions of both ingoing and outgoing null
article, we use the metric signature (−, +, +, +) and work in geodesics are negative), and no trapped region is present for
dimensionless units such that c = G = ℏ = kB = 1. θ− θ+ < 0. Since θ− is always negative, this implies that
f < 0 inside of the trapped region r ∈ (r− , r+ ), and f > 0
outside of the trapped region r > r+ , and thus g > 0 and b
II. MATHEMATICAL PREREQUISITES odd (otherwise g would have to be negative inside, but positive
outside of the trapped region). The inner and outer horizon are
A. Trapped regions and regular black holes identified as the roots of the equation f = 0 [39, 40]. At the
“disappearance point” of the trapped region v = vd , they co-
alesce, i.e. r− (vd ) ≡ r+ (vd ). From Eqs. (3) and (4), it then
A general spherically symmetric metric in advanced null
follows that
coordinates (v, r) is described by the line element
2
a+b
ds2 = −e2h(v,r) f (v, r)dv 2 + 2eh(v,r) dvdr + r2 dΩ2 , (1) θ− θ+ v=vd
=− 2
g(vd , r) r − r+ (vd ) ≤0 ∀r , (5)
r
where dΩ2 = dθ2 + sin2 θ dϕ2 denotes the line element of the which implies that the sum a+b must be even, and thus a odd.
2-sphere. Since our argumentation in Subsec. II B and Sec. III We note that the formation of a trapped spacetime re-
is based on the analysis of surface gravity at the outer horizon gion in finite time according to the clock of a distant ob-
[i.e. at r = r+ (v)] and it is always possible to write the func- server inevitably requires a violation of the null energy condi-
tion h(v, r) as a series with respect to the coordinate distance tion (NEC) near the outer horizon [40–43], which posits that
r − r+ (v) from the outer horizon (see Ref. 36 for details), Tµν ℓµ ℓν ⩾ 0, i.e. the contraction of the energy-momentum
tensor with any future-directed null vector ℓµ is non-negative.
∞
X
i Similarly, violating the NEC is a prerequisite for the emission
h(v, r) = χi (v) r − r+ (v) , (2)
of Hawking radiation. While quantum effects are necessary, it
i=1
is worth noting that Hawking radiation is a purely kinematical
we can assume h(v, r) = 0 without loss of generality in what phenomenon [44], and neither the Einstein equations nor the
follows2 . Generic dynamical models of RBHs are then de- Bekenstein entropy relation [2–4] are required for its deriva-
scribed by the metric function tion.

a
b
f (v, r) ··= g(v, r) r − r− (v) r − r+ (v) , (3)
B. Surface gravity and the first law of black hole mechanics
where r− (v) and r+ (v) denote the inner and outer horizon,
respectively, and a, b ∈ N>0 = {1, 2, ...} are positive integers The first law of black hole mechanics derived in Ref. 1
labeling their degeneracy. In spherical symmetry, nonsingular has been proven to hold in any theory of gravity arising from
a diffeomorphism-invariant Lagrangian [45, 46]. Assuming
δJ = δQ = 0, it can be stated mathematically as
2 Note also that h = 0 for all RBH models typically considered in the liter- κ
ature, including all of those examined explicitly in Secs. IV and V.
δM = δA, (6)
8π
 3

where M , κ, and A denote the black hole’s gravitational en- and generates a conserved current
ergy, surface gravity, and horizon area, respectively. The
notion of gravitational energy within a sphere of radius r ∇µ J µ = 0 , J µ ··= Gµν Kν , (10)
is captured by the so-called Misner–Sharp (MS) mass [47]
M ··= C/2 [see Eqs. (16) and (17)]3 . In spherically sym- for any symmetric rank-2 tensor Gµν = Gνµ that is invariant
metric solutions of the Einstein equations, such as the static under the spherical symmetries of the spacetime. If Gµν is
Schwarzschild4 or the nonstatic Vaidya metric, C(v, r+ ) = the Einstein tensor, then the current’s Noether charge is the
2M (v) = r+ (v). Using A = 4πr+ 2
for the horizon area, this MS mass.
leads to the famous expression for the surface gravity at the
outer horizon:
III. DERIVATION OF THE COMPATIBILITY CONDITION
δM κ δA 1
= ⇒ κ= . (7)
δr+ 8π δr+ 2r+ For the metric specified in Eq. (1) (recall that, as established
It is important to note that the surface gravity κ is unambigu- in Subsec. II A, we can use h = 0 without loss of generality),
ously defined only in stationary spacetimes, where it is related K µ = (1, 0, 0, 0), and its only nonzero covariant component
to the black hole’s Hawking temperature TH via κ = 2πTH . at the outer horizon is Kr = 1. Hence the Kodama surface
Nonetheless, even in generic dynamical black hole spacetimes gravity [cf. Eq. (8)] at the outer horizon is given by
the first law and its associated expression for the surface grav- 1
ity are expected to approach Eqs. (6) and (7) in the quasistatic κK r=r+
= ∂r f (v, r) r=r+ (11)
2
limit due to the timescale of the evaporation process. We show
here that this is not the case for the RBH models typically con- (3) (r − r+ )−1+b bg(v, r)(r − r− )a
= lim , (12)
sidered in the literature. r→r+ 2
Generalizations of surface gravity to dynamical spacetimes
which implies that a nonzero Kodama surface gravity at the
[48, 49] are generally related to either the affine peeling sur-
outer horizon can be achieved only if the outer horizon is non-
face gravity [50] or the so-called Kodama surface gravity
degenerate, i.e. b = 1. We thus focus on this scenario in what
[28, 51, 52]. Since the peeling surface gravity is ill-defined
follows.
for a transient object that forms in finite time of a distant ob-
server [53–55], and there are strong arguments that Kodama Assuming that f (v, r) is decomposable as a rational func-
surface gravity is the critical quantity with respect to Hawking tion of the radial coordinate r, i.e.
radiation [28, 55], we focus on this generalization of surface Pn (r)
gravity in what follows. f (v, r) = , (13)
The main difficulty in the generalization of surface grav- P̃n (r)
ity to evolving black hole spacetimes is that, unlike their sta-
tionary counterparts, they are not guaranteed to admit a time- where Pn and P̃n are polynomials (whose coefficients depend
like Killing vector field that generates the null hypersurface on v in generic dynamical spacetimes) of the same degree n ⩾
(known as Killing horizon) needed to define surface gravity. 3 in r as motivated in Ref. 37, we can write
However, a dynamical notion of surface gravity can be de- m
λz r z
P
fined at the (quasilocal) outer horizon using the Kodama vec-
tor field [51, 52], which is well-defined even in nonstationary z=0
g(v, r) = n , (14)
spherically symmetric spacetimes, and thus in some sense su-
P
ci ri
persedes the concept of a Killing vector field. i=0
At the outer horizon, the Kodama surface gravity κK is de-
fined via where the coefficients λz ≡ λz (r− , r+ ) and ci ≡ ci (r− , r+ )
depend explicitly only on r− (v) and r+ (v) (and thus implic-
1 µ itly on v) as they are the only relevant length scales, and
κK Kν =··


K ∇µ Kν − ∇ν Kµ , (8)
2 n − m = a + 1 and λm /cn = 1 are required to recover
the Vaidya form of the metric in the asymptotic limit r → ∞.
where K µ denotes the contravariant Kodama vector [51, 52]. These considerations will prove useful in our analysis of the
The Kodama vector field is conserved, inner-extremal RBH model in Sec. V. We also note that —
complemented by the assumptions of regularity of the space-
∇µ K µ = 0, (9)
time at the origin r = 0 and a proper Schwarzschild/Vaidya
form of the metric in the asymptotic limit — the polynomial
decomposition of the metric function according to Eq. (13)
immediately leads to a class of metric families of the form
3 While the MS mass is technically C/2 by virtue of its definition in Eq. (16), (see Ref. 37 for a detailed derivation)
we often take the liberty to refer to C itself simply as the MS mass.
4 Note that in the Schwarzschild solution, M = const. corresponds to the
Arnowitt–Deser–Misner (ADM) mass, and the object’s Schwarzschild ra-
rg (v)r2
f (v, r) = 1 − (15)
dius rg corresponds to the outer horizon, rg ≡ r+ = 2M . rg (v)l(v)2 + c1 (v)r + c2 (v)r2 + r3
 4

for the case n = 3, where l(v) denotes the minimal length have not been neglected in this derivation, but rather, they sim-
scale, rg (v) = 2M (v), and the case c1 (v) = c2 (v) = 0 corre- ply do not enter the expression for the Kodama surface gravity
sponds to the dynamical Hayward metric of Eq. (34) consid- of Eq. (18). Second, the conventional form of the first law of
ered in Subsec. IV B. black hole mechanics [Eq. (6)] and its associated expression
The metric function f [cf. Eq. (3)] is usually defined in for the surface gravity κ = 1/(2r+ ) [cf. Eq. (7)] are attain-
terms of the MS mass via able only if w1 = 0, as can be seen from Eqs. (18) and (22).
Therefore,
C(v, r)
f (v, r) ··= ∂µ r∂ µ r = 1 − , (16)
r
w1 r=r+
=0 (24)
with the MS mass given by
∞
X
i is a necessary condition to be compatible with the first law,
C(v, r) = r+ (v) + wi (v) r − r+ (v) . (17) i.e. the linear coefficient in the MS mass expansion w1 [cf.
i=1 Eq. (17)] must vanish at the outer horizon. Physically, this
implies that the metric approximates the Vaidya solution near
By means of Eq. (16), the Kodama surface gravity of Eq. (11) the outer horizon. Third, the expressions derived in Eqs. (18)–
can then be expressed directly in terms of the MS mass as (24) apply generically to black holes described by a metric
1   (17) 1 − w1 function of the form of Eq. (16). For RBHs described by a
κK r=r+
= 2
C(v, r) − r∂r C(v, r) r=r+
= , metric function of the form of Eq. (3), performing a series
2r 2r+
(18) expansion of Eq. (17) about the outer horizon yields

where the rightmost expression is obtained by explicitly sub- w1 r=r+

= 1 − g(v, r+ )r+ (r+ − r− )a . (25)
stituting the MS mass expansion of Eq. (17). Evaluation at the
outer horizon yields If the inner horizon is nondegenerate (a = 1), as is typically
the case for the RBHs most commonly considered in the lit-
δC 1 erature (e.g. those of Refs. 32–35 examined in Sec. IV), the
δM r=r+
= = (1 − w1 ) δr+ . (19)
2 r=r+ 2 compatibility condition prescribed by Eq. (24) can be evalu-
ated straighforwardly from the expression derived in Eq. (25)
Substituting the expression for the surface gravity on the RHS by considering the series expansion of the MS mass about the
of Eq. (18) into the RHS of Eq. (6), we obtain outer horizon using Eq. (16) since the metric function f (and
thus by extension g) is known explicitly. The case of a de-
1 − w1
δM = δA. (20) generate inner horizon (a > 1) is treated in Sec. V on the
16πr+ basis ofPthe inner-extremal RBH model proposed in Ref. 24,
Subsequent substitution of Eq. (19) into Eq. (20) leads to where z λz rz = 1 [cf. Eq. (14)]. In this case, the equation
f = 0 has two positive real-valued solutions, and the smaller
r  1 − w w1
+ 1 one which corresponds to the inner horizon is degenerate. If
δ = δA + δr+ . (21)
2 16πr+ 2 the outer horizon r = r+ is a single root, then the inner hori-
zon root r = r− has to be at least cubic in order for the inner
Using the relation δA = 8πr+ δr+ = r2+ δV between area horizon surface gravity to vanish, which plays a crucial role
A = 4πr+2
and volume V = 34 πr+
3
, we obtain in preventing mass inflation instabilities and ensuring that the
backreaction of perturbations vanishes asymptotically.
r 
+ 1 − w1 w1 In the next two sections, we investigate whether the con-
δ = δA + 2 δV, (22)
2 16πr+ 8πr+ sistency condition Eq. (24) that is required to be compatible
with the first law of black hole mechanics is satisfied by the
which is an extension of the standard form of the first law of RBH models typically considered in the literature. Since, as
black hole mechanics that includes an additional work term. argued in Sec. I, dynamical models are necessary to accurately
The notion of internal/thermal energy is captured by the ex- model the evolution of evaporating black holes, our analysis
pression r+ /2, which corresponds to the MS mass C/2 eval- in Secs. II and III was designed to accommodate generic dy-
uated at the outer horizon. However, this internal energy is namical RBH models, and we consider the dynamical gener-
not necessarily the same as the ADM mass when matter fields alizations of popular models rather than their static counter-
are present [56], as is the case for dynamically evolving black parts in what follows. To be more precise, static RBH metrics
holes. Comparison with the traditional form of the first law of belongs to a different class of black hole solutions whose ef-
κ
mechanics δM = 8π δA − pδV [57] identifies the pressure p fective energy-momentum-tensor component scaling behav-
in terms of w1 , i.e. ior τµν ∼ f k close to the outer horizon is characterized by
w1 k = 1 as opposed to k = 0 (see chapter 2 in Ref. 36 for a
p=− 2 . (23) detailed exposition of the two classes of admissible solutions
8πr+
in spherical symmetry, or Table I in Ref. 58 for a succinct
A few brief comments are in order: first, it is important to overview). While the unique k = 1 solution describes black
note that higher-order terms i > 1 in the MS mass expansion holes at the instant of their formation, they are described by
 5

a k = 0 solution for the remainder of their entire evolution B. Hayward model

or until their disappearance (e.g. through complete evapora-
tion) [53]. The relation between the Kodama surface gravity A nontrivial minimal RBH model that reduces to a de Sit-
at the outer horizon and the linear coefficient w1 of the MS ter spacetime in the limit r → 0 and a Schwarzschild/Vaidya
mass that we derived in Eq. (18) is valid for k = 0 solu- spacetime in the limit r → ∞ was proposed by Hayward in
tions, which includes all dynamical models. However, it no Ref. 34. In this model, the metric is specified by the function
longer applies (at least not in general) to static metrics such as
the Reissner–Nordström solution briefly considered in Sub- rg r2
f (r) = 1 − , (30)
sec. IV F [cf. Eq. (54)]. r3 + rg l2
While we strive to be as generic as possible in our analysis,
recall that a nondegenerate outer horizon (b = 1) is necessary where l ⩾ 0 represents a minimal length scale (which can be
to allow for a nonzero Kodama surface gravity at the outer interpreted as an additional hair of the black hole) akin to a
horizon [cf. Eq. (12)]. However, in Sec. V we do allow for Planckian cutoff, l = 0 corresponds to the Schwarzschild so-
the possibility that the inner horizon is degenerate (a > 1), lution, and rg = 2M denotes the black hole’s Schwarzschild
as such models have been shown to exhibit interesting phys- radius. The horizons are identified through the equation
ical properties making it possible to evade the mass inflation f (r) = 0 [39, 40], which can be solved using Cardano’s for-
problem that typically plagues RBHs (as detailed in Ref. 24). mula [59] and admits three real solutions, namely
l2
+ O l3 < 0,


r0 = −l + (31)
IV. TESTING REGULAR BLACK HOLES PART I: 2rg
MODELS WITH A NONDEGENERATE INNER HORIZON l2
+ O l3 ,


r− = l + (32)
2rg
In this section, we proceed by examining RBH models with
a nondegenerate inner horizon (a = 1), including the pro- l2
+ O l4 .


r+ = rg − (33)
posals of Bardeen [32] (Subsec. IV C), Dymnikova [33] (Sub- rg
sec. IV D), and Hayward [34] (Subsec. IV B). We write ex-
Note that this is a necessary requirement in order for the met-
plicit dependencies on v only once when specifying the met-
ric function of Eq. (30) to describe a RBH, as one real and
ric function f (v, r) for the generalized dynamical model and
two complex solutions would imply that no inner horizon is
omit them thereafter for the sake of readability. However, we
present [40].
later reinstate the explicit dependencies in Subsec. IV F for
We can generalize this model to describe a dynamical RBH
clarity.
spacetime by allowing for an explicit dependence of rg and l
on v, i.e.
A. A simple nonsingular “dynamical” spacetime rg (v)r2
f (v, r) = 1 − . (34)
r3 + rg (v)l(v)2
Arguably the simplest nonsingular dynamical spacetime
one can construct
P based on Eq. (3) is given by the choice Using the roots of f (v, r) = 0, this can be rewritten as
a = b = 1 and z λz rz = 1 [cf. Eq. (14)], such that Eq. (3)
r − r0
can be written as f (v, r) = (r − r− )(r − r+ ). (35)


r3 + rg l2
f (v, r) = g(v, r) r − r− (v) r − r+ (v) , (26)
By comparison with Eq. (3), we identify
and, by virtue of Eq. (13),
1 r − r0
g(v, r) = , (27) g(v, r) = > 0, (36)
c0 + c1 r + c2 r2 r3 + rg l2
where c2 = 1 is necessary to recover the Vaidya form of the which is positive due to r0 < 0. The MS mass C(v, r) [cf.
metric in the asymptotic limit. Regularity at the center r = 0 Eq. (17)] for the generalized dynamical Hayward model is
requires that specified by Eq. (34) through the relation given in Eq. (16).
c0 = r− r+ , c1 = −r− − r+ , (28) By considering its expansion about the outer horizon in the
regime l ≪ 1, we obtain the linear coefficient
µν
as can be verified by evaluating the Ricci (gµν R ) or the
Kretschmann (Rµνρσ Rµνρσ ) scalar. However, substitution of (33) 3l2
+ O l4 ⩾ 0.


w1 r=r+
= 2
(37)
Eqs. (27) and (28) into Eq. (26) reveals that these coefficients rg
lead to a trivial metric function
(r − r− )(r − r+ ) This expression is zero only if l = 0, but then r− = 0 [cf.
f (v, r) = = 1. (29) Eq. (32)], indicating that there is no inner horizon. Conse-
r− r+ − (r− + r+ )r + r2 quently, the dynamical Hayward model specified by Eq. (34)
In other words, we have rediscovered flat Minkowski space- cannot satisfy the compatibility condition Eq. (24) while l ̸=
time. 0.
 6

C. Bardeen model From the expansion of the MS mass about the outer horizon,
we obtain its linear coefficient
The first nonsingular black hole spacetime was proposed by 2
3rg2 − r2g
(46) 2 2

Bardeen in 1968 [32]. It is defined by the metric function w1 r=r+
= 2
e l + O e−rg /l ⩾ 0, (47)
l
rg r2 which is strictly greater than zero provided that l ̸= 0, which
f (r) = 1 −
3/2 , (38)
r 2 + l2 would once again imply that the inner horizon is absent, i.e.
r− = 0 [cf. Eq. (45)].
where rg and l have the same physical meaning as in Sub-
sec. IV B. Once again, we consider its dynamical generaliza-
tion E. RBH model with the strongest Schwarzschild corrections

rg (v)r2
f (v, r) = 1 −
3/2 . (39) The RBH considered in Ref. 35 exhibits the strongest possi-
r2 + l(v)2 ble corrections to the Schwarzschild geometry while still be-
ing compatible with its asymptotics. It is described by the
Writing the inner and outer horizon in terms of rg and l, we metric function
find
rg r2
l3/2 f (r) = 1 − , (48)
r− = √ + O l5/2 , (r + l)3


(40)
rg
and is of particular interest as observational data of the S2 star
3l2
+ O l4 . orbiting Sgr A⋆ can be used to test its geometry and derive


r+ = rg − (41)
2rg upper bounds for the new length scale l. We once again gener-
alize this metric by allowing for an explicit time dependence,
Following the same steps as in Subsec. IV B, we obtain
i.e.
(41) 3l2 rg (v)r2
+ O l4 ⩾ 0,


w1 = (42) f (v, r) = 1 −
3 . (49)
r=r+ rg2
r + l(v)
which coincides with the linear MS mass coefficient of the Using the same procedure as in the previous subsections, we
dynamical Hayward model [cf. Eq. (37)] at leading order, al- find that the inner and outer horizon are given by
though higher-order contributions will differ once sufficiently
high orders O(lx ) in Eqs. (33) and (41) are taken into account. l3/2 3l2
+ O l5/2 ,


Similar to the dynamical Hayward model discussed in the pre- r− = √ + (50)
rg 2rg
vious subsection, this expression is only zero if l = 0 and thus
r− = 0 [cf. Eq. (40)]. 3l2
+ O l3 .


r+ = rg − 3l − (51)
rg

D. Dymnikova model For the linear coefficient of the MS mass at the outer horizon,
we find
Another well-known RBH model was proposed by Dym- (51) 3l 6l2
+ 2 + O l3 ⩾ 0.


nikova [33]. It is specified by the metric function w1 r=r+
= (52)
rg rg
3 3

rg 1 − e−r /r⋆ As in the previously considered models, this expression can-
f (r) = 1 − , (43)
r not be zero unless l = 0 and thus r− = 0 [cf. Eq. (50)].
where r⋆3 ··= l2 rg , rg denotes the Schwarzschild radius, and l
a minimal length parameter. A generalized dynamical version F. Charged Hayward–Frolov model
is given by
3 3
In Ref. 37, Frolov proposed a generalization of the Hay-
rg (v) 1 − e−r /r⋆ (v)
f (v, r) = 1 − , (44) ward model to include an electric charge q. In the generalized
r dynamical case, the metric function for this type of RBH is
and the inner and outer horizon can be expressed in terms of given by
the parameters l and rg as

rg (v)r − q(v)2 r2
  f (v, r) = 1 − 4
, (53)
2 2

r− = l 1 + O e−rg /l , (45) r + rg (v)r + q(v)2 l(v)2

where, as in Subsec. IV B, rg and l denote the Schwarzschild

 2 2


r+ = rg 1 + O e−rg /l . (46)
radius and minimal length scale, and the case q = 0 reduces
 7

to the metric function of the uncharged dynamical Hayward As can be seen by comparison with Eq. (60), this expression
model [Eq. (34)]. The static case with l = 0 reproduces the matches that of the evolving Reissner–Nordström black hole
Reissner–Nordström metric. For both q = l = 0, Eq. (53) cor- only if l = 0, in which case the horizons of the charged dy-
responds to the Vaidya (or, in the static case, Schwarzschild) namical Hayward–Frolov RBH specified by Eqs. (62)–(63)
metric. reduce to those of Eqs. (58)–(59). However, unlike the pre-
As alluded to at the end of Sec. III, the Reissner–Nordström viously considered examples, the inner horizon r− ̸= 0 even
metric belongs to the k = 1 class of black hole solutions, and if l = 0 due to the presence of a charge term that is indepen-
thus we must not use Eq. (18) a priori. The surface gravity of dent of l, cf. Eqs. (58) and (62).
a Reissner-Nordström black hole is given by As evident from Eqs. (64) and (65), the condition for the
compatibility of a dynamically evolving charged RBH with
r+ − r−
κRN = 2 , (54) the first law of black hole mechanics is no longer encoded by
2r+ the relation w1 = 0. However, in the special circumstance
where where l → 0, the new compatibility condition can be stated as

q(v)2
p
r− = m − m2 − q 2 , (55) w1 (v, 0) = . (66)
p r+ (v)2
r+ = m + m2 − q 2 . (56)

The dynamical generalization of the Reissner–Nordström

metric is described by the metric function V. TESTING REGULAR BLACK HOLES PART II:
INNER-EXTREMAL MODEL WITH A DEGENERATE
1

INNER HORIZON
f (v, r) = 2
r − r− (v) r − r+ (v) , (57)
r
with the evolving inner and outer horizon specified by The inner-extremal RBH model proposed in Ref. 24 solves
p the mass inflation instability problem at the expense of a de-
r− (v) = m(v) − m(v)2 − q(v)2 , (58) generate inner horizon with vanishing surface gravity. As ar-
p gued above, we consider its dynamical generalization. The
r+ (v) = m(v) + m(v)2 − q(v)2 . (59)
metric function is given by choosing a = 3, b = 1 in Eq. (3)
Since this is a k = 0 solution, we can make use of Eq. (11) to such that in the generalized dynamical case
determine the Kodama surface gravity at the outer horizon of
3

an evolving Reissner–Nordström black hole and find f (v, r) = g(v, r) r − r− (v) r − r+ (v) , (67)

1 r+ (v) − r− (v) and by virtue of Eq. (13) the choice z λz rz = 1 ⇒ m = 0

P
κKRN = ∂r f (v, r) = . (60)
2 2r+ (v)2 determines the degree n − 0 = 4 = a + 1 of the polynomial
in the denominator of Eq. (14), i.e.
On the other hand, from the charged Hayward–Frolov metric
[Eq. (53)], we obtain 1
g(v, r) = , (68)
1 − w1 (v, l) c0 + c1 r + c2 r2 + c3 r3 + c4 r4
κKHF = , (61)
2r+ (v, l) where we have once again omitted the dependencies of the
and the inner and outer horizon are given by coefficients ci ≡ ci (r− (v), r+ (v)) for the sake of readabil-
ity and will omit dependencies on v in what follows as well.
r− (v, l) = r− (v) + β− (v)l2 + O l3 ,


(62) While a = 3 for the model considered in Ref. 24, our deriva-
tion in what follows is valid for arbitrary RBH models with
r+ (v, l) = r+ (v) + β+ (v)l2 + O l4 .


(63)
m = 0 and b = 1 [as otherwise the Kodama surface grav-
Following the methodology of Sec. III, considering the MS ity would vanish at the outer horizon, cf. Eq. (12)], i.e. those
mass expansion allows us to identify the linear coefficient where Eq. (14) admits the form

q(v)2 1
+ β(v)l2 + O l4 ,


w1 (v, l) = (64) g(v, r) = , (69)
r+ (v)2 c0 + c1 r + · · · + ca+1 ra+1

where β(v) denotes a lengthy coefficient that simplifies to where ca+1 (r− , r+ ) = 1 is required to recover the Vaidya
β(v) → 3/rg (v)2 in the uncharged case q(v) → 0, in agree- form of the metric in the asymptotic limit r → ∞. Based
ment with the expression derived for the uncharged dynami- on dimensional grounds, the generic form of the coefficients
cal Hayward model [cf. Eq. (37)]. Substituting this result into ci (r− , r+ ) is prescribed by
Eq. (61) and using Eqs. (62)–(63), we find
∞
−j−i+(a+1)
X j
r+ (v) − r− (v) ci (r− , r+ ) = dij r− r+ ∀ i ̸= a + 1 , (70)
+ O l2 .


κKHF = 2
(65)
2r+ (v) j=1
 8

where the coefficients dij are dimensionless5 . If the inner We note that, in consonance with our analysis of nondegener-
horizon is absent (r− → 0), we should also recover the Vaidya ate RBH models in Sec. IV, this relation is trivially satisfied if
form f = 1 − r+ /r as the black hole center is then no longer there is no inner horizon (r− → 0). On the other hand, if an
regular. In this case, Eq. (3) is given by inner horizon is present (r− ̸= 0) satisfying Eq. (78) requires
ra+1 1 − rr+


a
f (v, r) r →0 = .
 
X a
− c0 (0, r+ ) + c1 (0, r+ )r + · · · + ra+1 dij = (−1)j (79)
a−j
(71) i=0

Therefore, to recover the Vaidya form of the metric, we must

This alternative form of the compatibility condition derived
have
in Sec. III allows us to test RBH models for which evalu-
1 ating the expression given in Eq. (25) is not straightforward
ci (0, r+ ) = 0 ∀ i ̸= a + 1 ⇒ g(v, r) r →0 = a+1 .
− r due to the fact that a > 1 and the coefficients in the polyno-
(72) mial decomposition of g [Eq. (14)] are a priori undetermined.
Using Eqs. (25) and (69), the compatibility condition Eq. (24) With the commonly used assumptions of a positive MS mass
can be rewritten as C > 0 [cf. Eq. (17)] and regularity (expressed mathematically
through the finiteness of relevant curvature scalars, such as
a
g(v, r+ )r+ (r+ − r− ) = 1, (73) the Ricci and Kretschmann scalar), it is possible to determine
a+1 the lowest-order coefficients of Eq. (69). However, this alone
a 1 (69) X
i does not suffice to determine whether or not relation Eq. (79)
⇔ (r+ − r− ) r+ = = ci (r− , r+ )r+ . (74)
g(v, r+ ) i=0 is satisfied in general, as this would require the ability to de-
termine every individual coefficient ci (r− , r+ ). Nonetheless,
Using the binomial theorem to expand the LHS, and Eq. (70) it is often possible to evaluate Eq. (79) in physically moti-
for the coefficients on the RHS, Eq. (74) can be rewritten as vated scenarios where the degeneracy of the inner horizon a
∞ in Eq. (3) is known explicitly.
a
  a P
a a−k k+1 j −j+(a+1) a+1
(−1)a−k r−
P P
r+ = dij r− r+ + r+ . We now examine the dynamical generalization [Eqs. (67)–
k=0 k i=0 j=1
(75) (68)] of the inner-extremal RBH model with a = 3 proposed
Separating the k = a term (whose contribution is r+a+1
) from in Ref. 24 and demonstrate that it cannot satisfy the condition
the summation on the LHS, this simplifies further to Eq. (79) required to be compatible with the first law. In this
model, c4 = 1 and c3 = −3r− [cf. Eq. (68)] are necessary
a−1 a X ∞
X a
k+1
X j −j+(a+1) to recover the Vaidya form of the metric asymptotically. The
(−r− )a−k r+ = dij r− r+ . (76) latter can be seen by expanding Eq. (67) about the point z ··=
k
k=0 i=0 j=1 1/r = 0 to represent the limit r → ∞, which results in
Note that the maximum possible exponent of r− on the LHS
 
is a (for k = 0). Consequently, if Eq. (76) is treated as a 1 c3 + c4 3r− + r+ 1 1
f (v, r) 1 = − 2 +O 2 .
polynomial with respect to r− , the only way to have a solution r =0 c4 c4 r r
is to truncate the summation over j on the RHS at j = a. (80)
Furthermore, since 0 ⩽ k ⩽ a − 1 and 1 ⩽ j ⩽ a, we can
redefine the summation with respect to k on the LHS and use Substitution of c4 = 1 in Eq. (80) and subsequent comparison
the transformation j 7→ a − k to rewrite Eq. (76) as with the Vaidya form f = 1 − r+ /r yields c3 = −3r− . From
a   a X a
the requirement of regularity, we obtain
X a −j+(a+1)
X j −j+(a+1)
(−r− )j r+ = dij r− r+ . 3 2
a−j c0 = r− r+ , c1 = −r− (r− + 3r+ ), (81)
j=1 i=0 j=1
(77) as can be verified, for instance, by evaluating the Ricci or the
As a result, the compatibility condition Eq. (24) is satisfied if Kretschmann scalar.
and only if The remaining coefficient c2 can be determined from the
" # requirement that f be nondivergent, which in turn requires
a  a
X j a j −j+(a+1)
X −j+(a+1) that its denominator
r− (−1) r+ − dij r+ = 0.
a−j
j=1 i=0
D(v, r) ··= 1/g(v, r) (82)
(78)
(68)
3 2
= r− r+ − r− (r− + 3r+ )r + c2 r2 − 3r− r3 + r4

5 Note that the coefficients of our simple nonsingular spacetime [Sub-

be nonzero. We rewrite Eq. (82) as
sec IV A, Eq. (28)] are consistent with the required coefficient form spec-
ified in Eq. (70) only if r− = 0 and r+ = 0, i.e. if the spacetime has
 3r− 2 h r 3 1 i2
D̃(v, r) = c̃2 r2 + r − 3
+ r− r+ 1 − + ,
no horizons, in agreement with the trivial form of the metric function in 2 2 r− r+
Eq. (29). (83)
 9

where Substituting these values into Eq. (91), we obtain

15 2 9 r3 2
−3r− r+ + 3r− 3 −1
− r−
c2 = c̃2 + r− + r− r+ + − , c̃2 ⩾ 0. (84) r+ > 0
4 4 4r+ 2 2
(96)
⇔ 3r+ − 3r− r+ + r− < 0.
Using Eq. (16) to express the MS mass C in terms of f and
expanding about the center r = 0, we find Considering Eq. (96) as a polynomial in r+ , its discriminant
2
3 2 2
is given by −3r− < 0, which implies that the polynomial
r− + 4c̃2 r+ + 3r− r+ − 3r− r+ has two distinct complex conjugate roots and is thus always
C(v, r) = 3 r2 r3 + O(r4 ).
4r− + positive. However, this is in direct contradiction with the in-
(85) equality in the second line of Eq. (96), indicating that it is
impossible for the inner-extremal RBH model to satisfy the
The positivity requirement for the MS mass then constrains
necessary relation Eq. (79) that is required to be compatible
the coefficient c̃2 via
with the first law of black hole mechanics.
3 2 2
r− + 4c̃2 r+ + 3r− r+ − 3r− r+ >0
3 2 2 (86)
ζ − r− − 3r− r+ + 3r− r+
⇒ c̃2 = , ζ > 0. VI. PAGE EVAPORATION LAW
4r+
Substituting Eq. (86) into Eq. (84) yields Building on Hawking’s result [6], Page demonstrated that
ζ the mass loss due to the emission of Hawking radiation is de-
2
c2 = + 3r− + 3r− r+ . (87) scribed by the formula
4r+
This is all we need to explicitly identify the dimensionless Z∞
dM X 1 ωΓjωℓmp
coefficients dij introduced in Eq. (70). We find =− dω, (97)
dt 2π
j,ℓ,m,p
e2πω/κ − 1
3 3 2 2 3 0
c0 = r− r+ = d01 r− r+ + d02 r− r+ + d03 r− r+
⇒ d01 = 0 , d02 = 0 , d03 = 1, (88) where j labels the emitted particle species, and Γjωℓmp de-
3 2 2 2 3
notes the absorption probability6 for an incoming wave mode
c1 = −r− − 3r− r+ = d11 r− r+ + d12 r− r+ + d13 r− labeled by the frequency ω, spheroidal harmonic ℓ, axial quan-
⇒ d11 = 0 , d12 = −3 , d13 = −1, (89) tum number m, and polarization p. Although this formula ac-
2 −1
c3 = −3r− = d31 r− + d32 r− 3 −2
r+ + d33 r− r+ counts for both angular momentum and charge, an isolated
black hole is expected to be well approximated by an un-
⇒ d31 = −3 , d32 = 0 , d33 = 0. (90) charged spherically symmetric metric due to the fact that in
To proceed with the identification of the coefficients d2j , we the Hawking process angular momentum is shed much faster
note that the dimensions of ζ are [ζ] = L3 (i.e. ζ is cubic in than mass [7], and charged black holes rapidly discharge in a
length), as can be seen from Eq. (87). Therefore, we can write Schwinger-like pair production process [62, 63].
the first term in Eq. (87) as Since the ingoing Vaidya metric with decreasing mass
(C ′ (v) < 0) is the appropriate choice to model the effects of
ζ 2 3 −1
= ζ1 r− r+ + ζ2 r− + ζ3 r− r+ > 0, (91) Hawking radiation (see Table 2 in Ref. 36), Eq. (97) is often
4r+ expressed in advanced null coordinates (v, r) using several
where the coefficients ζi are dimensionless. Hence physically motivated simplifying assumptions (most notably
2 3 −1 the restriction to ℓ = m = 0 modes and the approximate re-
c2 = d21 r− r+ + d22 r− + d23 r− r+ lation Γ ≃ ω 2 rg2 ; see Refs. 31, 64, 65 for a more detailed
2 3 −1 (92)
= (ζ1 + 3)r− r+ + (ζ2 + 3)r− + ζ3 r− r+ account) as
⇒ d21 = ζ1 + 3 , d22 = ζ2 + 3 , d23 = ζ3 . dM a dr+ α
≃− 2 ⇔ ≃− 2 , (98)
Next, we use Eq. (79) to determine the coefficients ζi . We find dv M dv r+
3   which implies that a black hole of initial mass M0 will evap-
X 3
di1 = (−1)1 = −3 orate in a time te ∼ M03 , and the explicit form of the coeffi-
2
i=0 cients and their expansion about w1 = 0 are given by
= d01 + d11 + d21 + d31 ⇒ ζ1 = −3, (93)
4 1
3   α = 8a = − 4π , (99)
X 3 π
di2 = (−1)2 = 3 e 1−w1 −1
1
i=0 4 1
α=− + O(w1 ), (100)
= d02 + d12 + d22 + d32 ⇒ ζ2 = 3, (94) π e4π − 1
3  
X 3
di3 = (−1)3 = −1
0
i=0 6 In practice, these are determined using analytical and numerical techniques
= d03 + d13 + d23 + d33 ⇒ ζ3 = −1. (95) in a formalism originally developed by Teukolsky and Press [60, 61].
 10

respectively. Consequently, the standard Page evaporation law models from alternative descriptions of trapped spacetime do-
is modified if the condition w1 = 0 derived in Sec. III is not mains. Since both are necessary ingredients in the regulariza-
satisfied. tion procedure to avoid singularities, one may conjecture that
there is a more fundamental physical or topological principle
at play that prevents nonsingular black hole spacetimes from
VII. CONCLUSIONS satisfying the first law.
In their most conservative form, the conclusions of our
Based on the assertion that the surface gravity of an evolv- analysis may be stated as follows: nonsingular black holes are
ing black hole horizon should approach the expression pre- incompatible with the widely accepted semiclassical descrip-
scribed by the first law of black hole mechanics [Eq. (7)] in tion of evaporating black holes that is based on the results of
the quasistatic limit, we have derived a compatibility condi- Refs. 1 and 5–7. Unless one is willing to give up either the
tion for generic spherically symmetric dynamical black holes idea of regularity and an interior that is physically well be-
[Sec. III, Eq. (24)]. In our analysis of the dynamical gen- haved all the way down to the center or the first law of black
eralizations of RBH models typically considered in the liter- hole mechanics (and its associated thermodynamic interpreta-
ature, we have evaluated the compatibility condition explic- tion of surface gravity as an effective temperature), our results
itly for the respective metric functions that describe them, and demonstrate that modifications of the first law are required
demonstrated that none of them satisfies the necessary condi- even at the level of semiclassical gravity.
tion required to be compatible with the conventional form of We note that our analysis is consistent with the interpreta-
the first law of black hole mechanics (Sec. IV and Sec. V). As tion of the deviation from the standard form of the first law
outlined in Sec. VI, this also implies that — if the decrease [Eq. (6)] as a thermodynamic pressure term as has been pro-
in mass δM < 0 due to the emission of Hawking radiation is posed, for instance, in Refs. 25 and 57, which can be seen
indeed proportional to the surface gravity as stipulated by the from Eqs. (22)–(23). In this sense, the linear coefficient of
first law — then the dynamical evolution of such RBHs can- the MS mass w1 encodes rather specific information about the
not be accurately described by the standard Page evaporation thermodynamic properties of black holes. In fact, as evident
law. One may argue that this is a somewhat counterintuitive from Eq. (22), knowledge of w1 suffices to fully specify the
result, considering that the derivation of Eq. (97) is based on generalized first law of black hole mechanics.
Hawking fluxes perceived in the asymptotic limit, and thus
one would naively expect that the minimal length scale intro-
duced for the purpose of regularization should not affect the ACKNOWLEDGMENTS
outcome.
Our analysis suggests that the incompatibility of dynamical We would like to thank Yasha Neiman, Fil Simovic, and
RBHs with the first law of black hole mechanics is directly Daniel Terno for useful discussions and helpful comments.
linked to the minimal length scale l (which can be interpreted SM is supported by the Quantum Gravity Unit of the Oki-
as an additional hair) introduced for the regularization and the nawa Institute of Science and Technology (OIST). IS is sup-
presence of an inner horizon, which are the main characteris- ported by an International Macquarie University Research Ex-
tics (together with their regular center) that distinguish RBH cellence Scholarship (IMQRES).

[1] J. M. Bardeen, B. Carter, and S. W. Hawking, Commun. Math. [20] A. Ori, Phys. Rev. Lett. 67, 789 (1991).
Phys. 31, 161 (1973). [21] A. J. S. Hamilton and P. P. Avelino, Phys. Rep. 495, 1 (2010).
[2] J. Bekenstein, Lett. Nuovo Cim. 4, 737 (1972). [22] R. Carballo-Rubio, F. Di Filippo, S. Liberati, C. Pacilio, and M.
[3] J. Bekenstein, Phys. Rev. D 7, 2333 (1973). Visser, J. High Energy Phys. 05, 132 (2021).
[4] J. Bekenstein, Phys. Rev. D 9, 3292 (1974). [23] P. K. Dahal, S. Murk, and D. R. Terno, AVS Quantum Sci. 4,
[5] S. W. Hawking, Nature 248, 30 (1974). 015606 (2022).
[6] S. W. Hawking, Commun. Math. Phys. 43, 199 (1975); [24] R. Carballo-Rubio, F. Di Filippo, S. Liberati, C. Pacilio, and M.
[7] D. N. Page, Phys. Rev. D 13, 198 (1976). Visser, J. High Energy Phys. 09, 118 (2022).
[8] M. Visser, PoS BHs,GRandStrings 2008:001 (2008). [25] S. A. Hayward, Class. Quantum Gravity 15, 3147 (1998).
[9] C. Bambi, Mod. Phys. Lett. A 26, 2453 (2011). [26] A. Ashtekar and B. Krishnan, Living Rev. Relativ. 7, 10 (2004).
[10] S. W. Hawking, arXiv:1401.5761. [27] J. M. M. Senovilla, Int. J. Mod. Phys. D 20, 2139 (2011).
[11] V. P. Frolov, arXiv:1411.6981. [28] F. Kurpicz, N. Pinamonti, and R. Verch, Lett. Math. Phys. 111,
[12] V. Cardoso and P. Pani, Living Rev. Relativ. 22, 4 (2019). 110 (2021).
[13] S. Murk, Int. J. Mod. Phys. D; arXiv:2210.03750. [29] D. Kastor, S. Ray, and J. Traschen, Class. Quantum Gravity 27,
[14] S. Ansoldi, arXiv:0802.0330. 235014 (2010).
[15] L. Sebastiani and S. Zerbini, arXiv:2206.03814. [30] C. Pacilio and S. Liberati, Phys. Rev. D 96, 104060 (2017).
[16] C. Lan, H. Yang, Y. Guo, and Y.-G. Miao, arXiv:2303.11696. [31] D. Harlow, Rev. Mod. Phys. 88, 015002 (2016).
[17] M. Visser, Phys. Rev. D 90, 127502 (2014). [32] J. M. Bardeen in Proceedings of the International Conference
[18] E. Poisson and W. Israel, Phys. Rev. Lett. 63, 1663 (1989). GR5 (Tbilisi University Press, Tbilisi, 1968).
[19] E. Poisson and W. Israel, Phys. Rev. D 41, 1796 (1990). [33] I. Dymnikova, Gen. Relativ. Gravit. 24, 235 (1992).
 11

[34] S. A. Hayward, Phys. Rev. Lett. 96, 031103 (2006). [50] C. Barceló, S. Liberati, S. Sonego, and M. Visser, Phys. Rev. D
[35] M. Cadoni, M. De Laurentis, I. De Martino, R. Della Monica, 83, 041501(R) (2011).
M. Oi, and A. P. Sanna, Phys. Rev. D 107, 044038 (2023). [51] H. Kodama, Prog. Theor. Phys. 63, 1217 (1980).
[36] R. B. Mann, S. Murk, and D. R. Terno, Int. J. Mod. Phys. D 31, [52] G. Abreu and M. Visser, Phys. Rev. D 82, 044027 (2010).
2230015 (2022). [53] S. Murk and D. R. Terno, Phys. Rev. D 103, 064082 (2021).
[37] V. P. Frolov, Phys. Rev. D 94, 104056 (2016). [54] S. Murk and D. R. Terno, The Sixteenth Marcel Grossmann
[38] S. A. Hayward, Phys. Rev. D 49, 6467 (1994). Meeting, pp. 1196-1211 (2023); arXiv:2110.12761.
[39] V. Faraoni, G. F. R. Ellis, J. T. Firouzjaee, A. Helou, and I. [55] R. B. Mann, S. Murk, and D. R. Terno, Phys. Rev. D 105,
Musco, Phys. Rev. D 95, 024008 (2017). 124032 (2022).
[40] P. Binétruy, A. Helou, and F. Lamy, Phys. Rev. D 98, 064058 [56] M.-S. Ma and R. Zhao, Class. Quantum Gravity 31, 245014
(2018). (2014).
[41] S. W. Hawking and G. F. R. Ellis, The Large Scale Structure [57] C. Lan and Y.-G. Miao, Eur. Phys. J. C 82, 1152 (2022).
of Space-Time (Cambridge University Press, Cambridge, Eng- [58] S. Murk, Phys. Rev. D 105, 044051 (2022).
land, 1973), Ch. 9.2. [59] B. L. van der Waerden, Moderne Algebra I, 2nd ed., p. 188
[42] V. P. Frolov and I. D. Novikov, Black Hole Physics: Basic Con- (Springer Berlin Heidelberg, 1937).
cepts and New Developments (Springer Dordrecht, 1998). [60] S. A. Teukolsky, Astrophys. J. 185, 635 (1973).
[43] V. Baccetti, R. B. Mann, S. Murk, and D. R. Terno, Phys. Rev. [61] S. A. Teukolsky and W. H. Press, Astrophys. J. 193, 443 (1974).
D 99, 124014 (2019). [62] J. Schwinger, Phys. Rev. 82, 664 (1951).
[44] M. Visser, Int. J. Mod. Phys. D 12, 649 (2003). [63] G. W. Gibbons, Commun. Math. Phys. 44, 245 (1975).
[45] R. M. Wald, Phys. Rev. D 48, R3427(R) (1993). [64] R. M. Wald (ed.), Quantum Field Theory in Curved Spacetime
[46] V. Iyer and R. M. Wald, Phys. Rev. D 50, 846 (1994). and Black Hole Thermodynamics (The University of Chicago
[47] C. W. Misner and D. H. Sharp, Phys. Rev. 136, B571 (1964). Press, 1994), Ch. 7.
[48] A. B. Nielsen and J. H. Yoon, Class. Quantum Gravity 25, [65] L. Parker and D. Toms, Quantum Field Theory in Curved
085010 (2008). Spacetime (Cambridge University Press, 2009), Ch. 4.
[49] B. Cropp, S. Liberati, and M. Visser, Class. Quantum Gravity
30, 125001 (2013).