Fermion Soliton Stars

Loris Del Grosso,1, 2 Gabriele Franciolini,1, 2 Paolo Pani,1, 2 and Alfredo Urbano1, 2
1
Dipartimento di Fisica, Sapienza Università di Roma, Piazzale Aldo Moro 5, 00185, Roma, Italy
2
INFN, Sezione di Roma, Piazzale Aldo Moro 2, 00185, Roma, Italy
(Dated: January 23, 2023)
A real scalar field coupled to a fermion via a Yukawa term can evade no-go theorems preventing
solitonic solutions. For the first time, we study this model within General Relativity without
approximations, finding static and spherically symmetric solutions that describe fermion soliton
stars. The Yukawa coupling provides an effective mass for the fermion, which is key to the existence
of self-gravitating relativistic solutions. We systematically study this novel family of solutions
and present their mass-radius diagram and maximum compactness, which is close to (but smaller
than) that of the corresponding Schwarzschild photon sphere. Finally, we discuss the ranges of
the parameters of the fundamental theory in which the latter might have interesting astrophysical
implications, including compact (sub)solar and supermassive fermion soliton stars for a standard gas
arXiv:2301.08709v1 [gr-qc] 20 Jan 2023

of degenerate neutrons and electrons, respectively.

CONTENTS of boson stars [3–5] (and of their Newtonian analog, Q-

I. Introduction
II. Setup
A. Thomas-Fermi approximation
B. Dimensionless equations of motion and
boundary conditions
III. Some preliminary theoretical considerations
A. On the crucial role of fermions for the
existence of solitonic stars
B. Scaling of the physical quantities in the
µR  1 regime
C. Energy conditions
IV. Numerical results
A. Numerical strategy
B. Fermion soliton stars
C. On the existence of a Newtonian regime
V. Parameter space and astrophysical implications 12
VI. Conclusions
Acknowledgments
A. Connection with scalar-tensor theories
References
I. INTRODUCTION
Solitonic solutions play a crucial role in many field
theories, in particular in General Relativity. In the con-
text of the latter, starting from Wheeler’s influential idea
of geons [1], considerable attention has been devoted
to find minimal models allowing for self-gravitating soli-
tonic solutions [2]. The prototypical example is that
balls [6]), which are self-gravitating solutions to Einstein-
1 Klein-Gordon theory with a complex and massive scalar
field (see [7–9] for some reviews). If the scalar field is real,
2 no-go theorems prevent the existence of solitonic solutions
2 for very generic classes of scalar potential [10, 11]. Indeed,
Einstein-Klein-Gordon theory contains time-dependent
3 solutions known as oscillatons which, however, decay in
time [12].
5
About forty years ago, Lee and Pang proposed a model
5 in which a real scalar field with a false-vacuum potential
is coupled to a massive fermion via a Yukawa term [13].
6 Working in a thin-wall limit in which the scalar field
8 is a step function, for certain parameters of the model
they obtained approximated solutions describing fermion
9 soliton stars.
9
10 The scope of this paper is twofold. On the one hand we
11 show that fermion soliton stars exist in this model also
beyond the thin-wall approximation, and we build exact
static solutions within General Relativity. On the other
hand, we elucidate some key properties of the model, in
14 particular the role of the effective fermion mass provided
by the Yukawa coupling. Then, we explore the model
14 systematically, presenting mass-radius diagrams and the
maximum compactness of fermion soliton stars for various
14 choices of the parameters, showing that in this model a
standard gas of degenerate neutrons (resp. electrons) can
16 support stable (sub)solar (resp. supermassive) fermion
soliton stars with compactness comparable to that of
ordinary neutron stars. This analysis paves the way for a
detailed study of the phenomenology of fermion soliton
stars as a motivated model of exotic compact objects [14].
Finally, in Appendix A, we explore the connection of the
model to a very peculiar scalar-tensor theory.
We use the signature (−, +, +, +) for the metric, adopt
natural units (~ = c = 1) and define the Planck mass
through G = m−2p .
 2

II. SETUP spacetime. The problem can be simplified significantly

within the Thomas-Fermi approximation2 , relying on the
We consider a theory in which Einstein gravity is mini- following assumptions:
mally coupled to a real scalar field φ and a fermion field i) the gravitational and scalar fields are slowly varying
ψ. The action can be written as [13] functions with respect to the fermion dynamics,
they do not interact directly with the (microscopic)
√ h R 1
Z
S = d4 x −g − ∂ µ φ∂µ φ − U (φ) fermionic field ψ, but with average macroscopic
16πG 2 quantities (mean-field approximation);
i
+ ψ̄(iγ µ Dµ − mf )ψ + f φψ̄ψ , (1) ii) the fermion gas is at equilibrium so that all the
macroscopic quantities are time independent.
where the scalar potential is
In practice, one can divide the entire three-space into
1  φ 2 small domains which are much larger than the de Broglie
U (φ) = µ2 φ2 1 − , (2)
2 φ0 wavelength of the typical fermion, but sufficiently small
that the gravitational and scalar fields are approximately
and features two degenerate minima at φ = 0 and φ = φ0 . constant inside each domain. Then, every domain is filled
The constant µ (resp. mf ) is the mass of the scalar (resp. with a degenerate (i.e. the temperature is much smaller
fermion). The Yukawa interaction is controlled by the than the chemical potential) Fermi gas, in such a way that
coupling f . It should be noted that Eq. (1) describes the the Fermi distribution is approximated by a step function,
action of a local field theory and, therefore, we expect that nk = θ(kF − k), where kF (xµ ) is the Fermi momentum
all physics derived from it will naturally respect causality observed in the appropriate local frame.
conditions (that, on the contrary, could be violated in The energy density of the fermion gas reads
the absence of such underlying formulation). Also, we
2
Z kF
point out that the matter Lagrangian in Eq. (1) describes W = d3 k k , (5)
a renormalizable field theory; this is in contrast to the (2π)3 0
widely used model describing solitonic boson stars [15–18]
where k = k 2 + m2eff . Notice that W = W (xµ ) through
p
in which the scalar potential is non-renormalizable and
field values should not exceed the limit of validity of the the spacetime dependence of kF . In an analogous way, we
corresponding effective field theory. The covariant deriva- obtain the fermion gas pressure P and the scalar density
tive Dµ in Eq. (1) takes into account the spin connection S = hψ̄ψi as
of the fermionic field. 2
Z kF
k2
From the quadratic terms in the fermion Lagrangian, P = d 3
k , (6)
(2π)3 0 3k
it is useful to define an effective mass,
2
Z kF
meff
meff = mf − f φ. (3) S= d3 k . (7)
(2π) 0
3 k
We will focus on scenarios in which the fermion becomes It it easy to show that these quantities satisfy the identity
effectively massless (i.e. meff = 0) when the scalar field W − 3P = meff S. (8)
sits on the second degenerate vacuum, φ = φ0 . This
condition implies fixing In the Thomas-Fermi approximation, the fermions enter
Einstein’s equations as a perfect fluid characterised by an
mf energy-momentum tensor of the form
f= . (4)
φ0 [f ]
Tµν = (W + P )uµ uν + P gµν , (9)
As we shall discuss, we are mostly interested in configu-
rations where the scalar field makes a transition between while they also enter the scalar field equation through the
the false1 vacuum (φ ≈ φ0 ) to the true vacuum (φ ≈ 0). scalar density S. Indeed, by varying the action in Eq. (1)
with respect to φ, we obtain a source term of the form
≈ f ψ̄ψ. Within the Thomas-Fermi approximation, this
A. Thomas-Fermi approximation becomes
f ψ̄ψ → f hψ̄ψi ≡ f S, (10)
The description of a fermionic field in Eq. (1) requires
which is consistent with the fact that, in the fluid descrip-
treating the quantization of spin-1/2 particles in curved
tion, the scalar field equation couples to fermions through
a term proportional to the trace (T [f ] )µµ = −W + 3P .

1 Although the minima at φ = 0 and φ = φ0 are degenerate,

we shall call them true and false vacuum, respectively, having 2 We point the interested reader to Appendix A of Ref. [13] for
in mind the generalization in which the potential U (φ) can be a complete derivation of the Thomas-Fermi approximation in
nondegenerate, i.e. U (φ0 ) 6= U (0), see Fig. 1 below. curved spacetime, while here we summarise the main properties.
 3

1. Equations of motion while the number of fermions is

4
Z
It is now possible to write down the equations of motion N= dρ ρ2 ev(ρ) kF (ρ). (15)
3π
for our theory in covariant form
To enforce a constant number of fermions, we introduce
Gµν = 8πG Tµν , the Lagrangian multiplier ωF and define the functional
∂U
+ f S = 0, (11)
 
φ − Ef0 [kF ] = Ef [kF ] − ωF N [kF ] − Nfixed , (16)
∂φ

where which is minimized by imposing the condition

 ∂L 1  δEf0 [kF ]
Tµν = −2
φ
− gµν φ + Tµν ,
L [f ]
(12) = 0. (17)
∂g µν 2 δkF (ρ)

in which Lφ is the Lagrangian density of the scalar field. One directly obtains F = e−u ωF , where F = kF is
In order to close the system, we need an equation de- the Fermi energy. Thus, ωF coincides with the Fermi
scribing the behaviour of kF . This is obtained by mini- energy in flat spacetime while it acquires a redshift factor
mizing the energy of the fermion gas at fixed number of otherwise. Finally, we find
fermions [13].
kF2 (ρ) = ωF2 e−2u(ρ) − (mf − f φ(ρ))2 . (18)
From now on, for simplicity, we will consider spherically
symmetric equilibrium configurations, whose background
metric can be expressed as
B. Dimensionless equations of motion and
boundary conditions
ds = −e
2 2u(ρ)
dt + e
2 2v(ρ)
dρ + ρ (dθ + sin θdϕ ), (13)
2 2 2 2 2

In order to simplify the numerical integrations, as well

in terms of two real metric functions u(ρ) and v(ρ). Fur-
as physical intuition, it is convenient writing the field
thermore, we will assume that the scalar field in its equilib-
equations in terms of dimensionless quantities. To this
rium configuration is also static and spherically symmetric,
end, we define
φ(t, ρ, θ, ϕ) = φ(ρ). Being the spacetime static and spher-
ically symmetric, kF = kF (ρ) can only be a function of kF φ
the radial coordinate. x= , y= , r = ρµ. (19)
mf φ0

Therefore, the potential and kinetic terms become

2. Fermi momentum equation
2 2 1 2
 
U = µ φ0 y (1 − y) ≡ µ2 φ20 Ũ (y),
2
2
In the Thomas-Fermi approximation the fermion gas
2 2 1 −2v(r)
 
energy can be written as [13] V = µ φ0 e (∂r y) ≡ µ2 φ20 Ṽ (y),
2
(20)
2
Z
Ef = 4π dρ ρ2 eu(ρ)+v(ρ) W, (14) while Eqs. (5)-(7) can be computed analytically as

2 kF (ρ)
m4 h p
Z q  p i
W = d3 k
k 2 + (mf − f φ(ρ))2 = eff2 s 1 + s2 (1 + 2s2 ) − log s + s2 + 1 ≡ m4f W̃ (x, y), (21a)
(2π)30 8π
2 d kk 2 2
Z kF (ρ) 3 2 4
  p 
meff  p
P = = s s − 1 1 + s 2 + log s + s2+1 ≡ m4f P̃ (x, y), (21b)
(2π)3 0 3 k 2 + (mf − f φ(ρ))2 8π 2 3
p

2
Z kF (ρ)
mf − f φ(ρ) m3 h p  p i
S= d3 k p = eff2 s 1 + s2 − log s + s2 + 1 ≡ m3f S̃(x, y), (21c)
(2π) 0
3
k 2 + (mf − f φ(ρ))2 2π

where W̃ , P̃ , S̃ are dimensionless quantities and we intro- expressions are the same as in the standard case of a
duced s ≡ x/(1 − y) for convenience. Remarkably, these minimally coupled degenerate gas with the substitution
 4

mf → meff . TABLE I: List of the model parameters, the fermion

As we shall discuss in Appendix A, this property will be soliton star parameters, and the dimensionless quantities
important when comparing this model to a scalar-tensor adopted to express the system of equations in compact
theory. Note that the massless limit, meff → 0, should form. Due to the condition in Eq. (4), in our case only
be taken carefully as not all the dependence on meff is three model parameters are independent.
expressed in the dimensional prefactor. By performing
the first integrals in Eqs. (21a)-(21c) in the meff → 0 limit, Model parameters
we obtain W = P/3, as expected for an ultrarelativistic
degenerate gas. µ Scalar field mass
It is convenient to further introduce the dimensionless φ0 VEV of the false vacuum
combination of parameters
mf Fermion mass
√
8πφ0 mf f Yukawa coupling
Λ= , η= 1/2
. (22)
mp 1/2
µ φ0 Solution parameters (boundary conditions)

Finally, the field equations (i.e. the Einstein-Klein-Gordon Pc Fermion central pressure
equations with the addition of the Fermi momentum  = 1 − φ/φ0 Central scalar field displacement
equation) take the compact form
Dimensionless parameters/variables
e−2v − 1 − 2e−2v r∂r v = −Λ2 r2 η 4 W̃ + Ũ + Ṽ ,
  √
Λ = 8πφ0 /mp Dimensionless VEV of the false vacuum
√
e−2v − 1 + 2e−2v r∂r u = Λ2 r2 η 4 P̃ − Ũ + Ṽ ,
 
η = mf / µφ0 Scale ratio
h  2  i ∂ Ũ x = kF /mf Fermi momentum
e−2v ∂r2 y + ∂r u − ∂r v + ∂r y = − η 4 S̃,
r ∂y y = φ/φ0 Scalar field
x2 = ω̃F2 e−2u(r) − (1 − y)2 , (23) r = ρµ Rescaled radius

where Ũ , Ṽ , P̃ , W̃ , and S̃ depend on x, y, and r, and

we also introduced ω̃F = ωF /mf . Static and spherically P̃ (r = 0) = P̃c , (24)
symmetric configurations in the model (1) are solutions to
the above system of ordinary differential equations. For where  > 0 will be fixed numerically through a shooting
clarity, we summarise the relevant parameters in Table I. procedure in order to obtain asymptotic flatness.
In practice, in a large region of the parameter space
one obtains   1. In this limit, using Eq. (21b) and
1. Absence of φ = const solutions Eq. (18), we find
ωF 3
Note that, because U = 0 = dU/dφ in both degen- ω̃F ≡ = (12π 2 P̃c )1/4 + 2 +O(3 ). (25)
mf 4(12π 2 P̃c )1/4
erate vacua, it is natural to first check what happens
when φ = φ0 = const or if φ = 0. The former case (i.e. In general, ω̃F is fixed in terms of the central values of
y(ρ) = 1) is an exact solution of the scalar equation and the pressure and scalar field.
reduces the Einstein’s equations to those of gravity cou- Finally, since a shift u(ρ) → u(ρ) + const in Eq. (23)
pled to a degenerate gas of massless (since meff (φ0 ) = 0) merely corresponds to a shift of the fermionic central
fermions. In this case, self-gravitating solutions do not pressure, we have imposed u(ρ = 0) = 0 without loss of
have a finite radius [19]. On the other hand, due to the generality.
Yukawa coupling, in the presence of a fermion gas φ = 0
is not a solution to the scalar field equation.
3. Definitions of mass, radius, and compactness
Thus, self-gravitating solutions to this model must have
a nonvanishing scalar-field profile. In particular, we will
search for solutions that (approximately) interpolate be- We define the mass of the object as
tween these two vacuum states. m(ρ → +∞)
M= , (26)
G
2. Boundary conditions at ρ = 0 where the function m(ρ) is related to the metric coefficient
v(ρ) by e2v(ρ) = 1 − 2m(ρ)/ρ, and can be interpreted as
the mass-energy enclosed within the radius ρ. In terms
Regularity at the center of the star (ρ = 0) imposes the
of the dimensionless variables introduced in Eq. (19), it
following boundary conditions
is convenient to define m̃(r) = µm(ρ). Thus, one obtains
v(r = 0) = 0, u(r = 0) = 0, µM
= m̃(r). (27)
y(r = 0) = 1 − , ∂r y(0) = 0, m2p
 5

Notice that, in the asymptotic limit r → ∞, Eq. (27)

becomes independent of the radius.
Typically, the radius of a star is defined as the value of 0
the radial coordinate at the point where pressure drops to
zero. As we shall discuss, in our case the fermion soliton
stars will be characterised by a lack of a sharp boundary. -1
Analogously to the case of boson stars [9], one can define
an effective radius R within which 99% of the total mass
is contained. (As later discussed, we shall also define the -2
location Rf where only the pressure of the fermion gas
vanishes.) Finally, we can define the compactness of the
-3
star as GM /R.
0.0 0.2 0.4 0.6 0.8 1.0 1.2

III. SOME PRELIMINARY THEORETICAL

CONSIDERATIONS
Before solving the full set of field equations numerically,
in this section we provide some theoretical considerations
that might be useful to get a physical intuition of the
model.
A. On the crucial role of fermions for the existence
of solitonic stars
FIG. 1: Inverted potential with degeneracy (blue line, our
case) and without degeneracy between vacua (orange).
E(0) = Û (φ0 − δφ). The energy E(t) of the particle at a
time t is obtained by subtracting the work done by the
friction:
E(t) − E(0) = L(t), (32)

1. Classical mechanics analogy where

t
φ̇2 (t0 )
Z
In order to understand why the presence of fermions in L(t) = −2 dt0 . (33)
this theory plays a crucial role for the existence of station- 0 t0
ary solutions, it is useful to study a classical mechanics Note that, owing to the initial conditions, this integral
analogy for the dynamics of the scalar field [6]. is regular at t = 0. On the other hand, the existence of
For the moment we consider flat spacetime. Further- a solution with asymptotically zero energy requires the
more, we start by ignoring the fermions (we will relax particle to arrive with zero velocity at φ = 0 for t → +∞.
this assumption later on). The set of Eqs. (23) drastically Therefore, we impose E(t → ∞) = 0. As the total energy
simplifies to a single field equation loss due to friction is L(t → ∞), the latter condition
2 ∂U means
∂ρ2 φ + ∂ρ φ − = 0. (28)
ρ ∂φ
To make the notation more evocative of a one-dimensional E(0) = −L(t → ∞) (34)
mechanical system, we rename
that is
ρ → t, φ(ρ) → φ(t), Û := −U, (29) ∞
φ̇2 (t0 )
Z
Û (φ0 − δφ) = 2 dt0 . (35)
in such a way that the equation of motion becomes 0 t0

∂ Û 2 This equation can be interpreted as an equation for δφ

φ00 (t) = − − φ0 (t), (30) in order to allow for the existence of a "bounce" solution3 .
∂φ t
One can demonstrate the existence of such a solution
which describes the one-dimensional motion of a parti- heuristically. Let us first consider a slightly modified ver-
cle with coordinate φ(t) in the presence of an inverted sion of the inverted potential without degeneracy (orange
potential, Û , and a velocity-dependent dissipative force, plot in Fig. 1). Obviously, if the motion starts exactly
−(2/t)φ0 (t). Within this analogy, the boundary (or initial) at φ0 with zero velocity, the particle would remain at
conditions (24) simply become

φ(t = 0) = φ0 − δφ, φ0 (t = 0) = 0, (31)

3 A bounce solution is the one reaching asymptotically the true
where φ0 is the position of the false vacuum and δφ = φ0 . vacuum with zero energy, after having "bounced" at the minimum
As we impose zero velocity at t = 0, the initial energy is of the inverted potential.
 6

rest. However, if we start on the left of the maximum stationary solitons generically exist also for a real scalar
the particle will roll down, bounce, and eventually climb field (at variance with the case of boson stars, that require
the leftmost hill shown in Fig. 1. Now, if the dynamics complex scalars) and for a wide choice of the parameters.
starts too far from φ0 (still on the left of the maximum),
with zero initial velocity it might not have enough energy
to reach the zero-energy point at φ = 0. Similarly, if B. Scaling of the physical quantities in the µR  1
the dynamics starts too close to φ0 , the particle might regime
reach φ = 0 with positive energy and overcome the hill
rolling up to φ → −∞. By continuity, there must exist a Assuming µR  1, it is possible to derive an analytical
unique point such that the total energy loss due to friction scaling for various physical quantities, as originally derived
compensates the initial gap of energy with respect to the in Ref. [20] and similar in spirit to Landau’s original
energy of φ = 0. computation for ordinary neutron stars (see, e.g., [19]).
However, by applying the same argument to our degen- It is instructive to consider the following toy model
erate case (blue curve in Fig. 1), it is easy to see that there in the absence of gravity. We consider a theory with
is no solution to Eq. (35) 4 . This is because the energy an additive quantum number N , brought by a spin-1/2
loss due to friction is nonzero, so the particle will never field ψ. We then add a real scalar field φ with the usual
reach φ = 0 and is doomed to roll back in the potential potential described in Eq. (2). Such a scalar field obeys
eventually oscillating around the minimum of Û . This Eq. (28) along with the initial condition (31). Since
shows that, in the degenerate case considered in this work, µR  1, its solution is well approximated by a stiff Fermi
a simple scalar model does not allow for bounce solutions function [13],[20]
in flat spacetime.
If we now reintroduce fermions in the theory, the scalar φ0
φ(ρ) ≈ . (37)
field equation reads (still in flat spacetime) 1 + eµ(ρ−R)
The definition of kF is (we take Eq. (18) with u = 0 since
∂ Û 2
φ00 (t) = − − φ0 (t) − f S. (36) we work in absence of gravity)
∂φ t
kF2 (ρ) = ωF2 − (mf − f φ(ρ))2 . (38)
Since S ≥ 0, the fermions act with a force pushing our
particle towards the origin, potentially giving the right Because of Eq. (37), the Fermi momentum is nearly fixed
kick to allow the particle reaching φ = 0 asymptotically. to the constant value ωF for ρ . R, and for ρ ≈ R it goes
As we shall see, this also requires S = 0 (i.e., no fermions) to zero stiffly. Therefore, the field ψ is approximately
around the origin, in order for the particle to reach a confined within the sphere of radius R. We assume that
stationary configuration at φ = 0. the quanta of ψ are noninteracting, massless (consistently
This simple analogy shows how the presence of the with the fact that we are interested in configurations in
fermions is fundamental as it allows the solution to exist. which the fermions are approximately massless in the
In the following section we will show how this is realised core of the star) and described by Fermi statistics at zero
in the full theory which includes gravitational effects. temperature. Thus, we obtain the standard relation for
Furthermore, we will show that, in certain regions of the the particle density
parameter space, relativistic effects are in fact crucial for
#particles 2
Z kF
the existence of the solution, since the latter requires a ωF3
n= = 4πk 2
dk = . (39)
minimum fermionic pressure to exist. unit.volume 8π 3 0 3π 2
Since kF ' ωF = const, the total number of particles is
2. Evading the no-go theorem for solitons
4
Z R
N =n 4πρ2 dρ = (RωF )3 . (40)
9π
The above conclusions, deduced from our simple heuris- 0

tic picture, holds also in the context of General Relativity. The fermion energy is
Indeed, without fermions in the system of Eqs. (23), and Z R  3 4/3 1
since our potential (2) is nonnegative, a general theorem Ef = 4πρ2 dρ W = (3π)1/3 N , (41)
proves that no axially symmetric and stationary solitons 0 4 R
(that is asymptotically flat, localized and everywhere reg-
ular solutions) can exist [10, 11]. where
However, the presence of fermions evades one of the energy 2
Z kF
ωF4
hypotheses of the theorem. As we will show, in this case W = = 4πk 2 dk · k = . (42)
unit.volume 8π 3 0 4π 2
The energy associated to the scalar field φ is instead
4 At least if we look for a solution in which the scalar field does
Z R 1 
the transition at a finite time. Es = 4πρ2 dρ (U + V ) ' µφ20 4πR2 , (43)
0 6
 7

TABLE II: Analytical scalings of the some physical quan- ii) a gas of massless fermions in the interior of the star.
tities at the maximum mass Mc in the µR  1 limit.
In practice, the first assumption is not restrictive. Indeed,
since µ−1 is the Compton wavelength of the scalar boson,
Mass µMc /m2p ∼ 1/Λ2 in the context of a classical field theory we should always
Radius µRc ∼ µMc /m2p ∼ 1/Λ2 impose µR  1. In other words, if µR ' 1 the quantum
effects of the scalar field become important on the scale of
ω̃F ω̃Fc ∼ (µ/mp )1/2 /(φ0 /mf ) ∼ Λ1/2 /η the star and one cannot trust the classical theory anymore.
Central pressure P̃c ∼ ω̃F4 ∼ Λ2 /η 4 The hypothesis µR  1 is an essential ingredient in order
to approximate the scalar field profile with Eq. (37), and
to assume, as a consequence, that the kF is a step function.
Besides, it guarantees that the energy density of the scalar
where we have used that fact that field is near to a delta function. Using the scaling reported
12 12 in Table II, condition i) implies Λ  1.
U' V ' δ(ρ − R) , (44) One may worry that the second assumption can be
µφ20 µφ20
violated, since the scalar field is not located exactly at
which can be shown using Eq. (37) and µR  1. φ0 in the origin ρ = 0, and therefore fermions are never
The total energy of our configuration is exactly massless. It is enough checking that the fermion
gas is very close to be a massless gas. Let us recall that
E = Ef + Es , (45) the effective mass of the fermion is defined as
 φ(ρ) 
while the radius can be found by imposing ∂E/∂R = 0, meff (ρ) = mf 1 − (50)
φ0
yielding
h 3  3 4/3 i1/3  1 1/3 and therefore meff (ρ = 0) = mf . We can say that the
R= (3π)1/3 N (46) fermion gas is effectively massless when W/P = 3. From
4π 4 µφ20 Eq. (5) and (6), at the lowest order in  one obtains
and the mass W  2m2f 2 
1 =3 1+ + O(3 ), (51)
 P kF2
M = E(R) = 12πR 2
µφ20 . (47)
6 which indicates we should require
From Eq. (46),(47), we get
2m2f 2
1 (52)
R ∼ N 4/9 M ∼ N 8/9 . (48) kF2
in the vicinity of the origin at ρ ' 0. At larger radii,
Thus, at least for large N , the mass of the soliton is lower
the scalar field gradually moves away from the central
than the energy of the sum of N free particles, ensuring
configurations and fermions start retaining a bare mass.
stability.
Inserting Eq. (38) in the previous condition and using
In the absence of gravity, M can be arbitrarily large.
Eq. (25), provides the condition we need to enforce to
However, due to relativistic effects we expect the existence
obey assumption ii), i.e.
of a maximum mass beyond which the object is unstable
against radial perturbations. We expect that gravity 2m2f 2
becomes important when 2GM/R ∼ 1. Therefore, the  1. (53)
critical mass Mc can be estimated by simply imposing (12π 2 Pc )1/2
R ∼ 2GMc in Eq. (47), yielding G2 Mc ∼ 1/µφ20 and thus We express  using the scalar field profile approximation
in Eq. (37). Indeed, with simple manipulations, one finds
µMc 1
∼ 2. (49)
2
mp Λ − log  = µR  1. (54)

Likewise, one can obtain the scaling of all other relevant Substituting (54) in (53), and neglecting, at this stage,
quantities, which we collect in Table II. the numerical factors one obtains
!
mf
log 1/4
 µR. (55)
1. Self-consistency criteria Pc
Using the scaling relations in Table II, we obtain
When deducing the scaling reported in Table II, we
made the following assumptions:
!
η 1
log  2. (56)
i) µR  1; Λ1/2 Λ
 8

Summing up, the following conditions on the parame- certain value. Using the scaling for ω̃Fc in Table II, this
ters can be translated into the condition

Λ  1, (57) Λ1/2
< C, (61)
! η
η 1
log  2 (58) where C is a constant that has to be determined numeri-
Λ1/2 Λ
cally.
are our self-consistency criteria to check if we are in At this point, it is natural to define ηc as the value of
a regime in which the scaling reported in Table II are η in which Eq. (61) is saturated. In this way, Eq. (61)
expected to be valid. While it can be shown that the becomes
second condition implies the first, we prefer writing both
η > ηc = CΛ1/2 . (62)
for the sake of clarity. Notice that, for fixed Λ  1,
one can violate (58) for increasing values of η, but only To summarize, when η & ηc (confining regime) the
logarithmically. size of the soliton near the maximum mass is mostly
determined by the properties of the scalar field, whereas
it strongly depends on the fermion mass when η . ηc
2. Confining and deconfining regimes (deconfining regime5 ).

An important consequence of the scalings collected in

Table II is that the critical mass and radius are inde- C. Energy conditions
pendent on η at fixed Λ. We shall call the region of the
parameters space where this happens the confining regime For an energy-momentum tensor of the form
of the solutions. Indeed, in this regime the size of the
soliton is dictated by the parameters of the scalar field, i.e. Tνµ = diag{−ρ, P1 , P2 , P3 }, (63)
µ and φ0 , regardless of the value of the fermion mass mf .
Physically, we expect that this would be the case when the energy conditions take the following form:
there exists a hierarchy between the scalar and fermion • Weak energy condition: ρ ≥ 0 and ρ + Pi ≥ 0
parameters. Since this hierarchy is measured by η, we
expect that the confining regime exists only when η is • Strong energy condition: ρ + i Pi ≥ 0 and ρ + Pi ≥ 0
P
larger than a critical value, ηc .
To better clarify this point, we consider again Eq. (18) • Dominant energy condition ρ ≥ |Pi |.
for the Fermi momentum, For a spherically symmetric configuration, P1 = Pr is
the radial pressure, while P2 = P3 = Pt is the tangential
 φ(ρ) 2 pressure. For our model,
kF2 (ρ) = ωF2 e−2u(ρ) − mf 1 − . (59)
φ0
ρ=U +V +W , (64)
In the mf → 0 limit this quantity becomes positive defi-
Pr = V − U + P , (65)
nite and so the fermionic pressure cannot vanish at any
finite radius. In other words, the radius of the star can Pt = −U − V + P . (66)
be arbitrarily large, provided that mf is sufficiently small.
Since, V, W, P are nonnegative quantities, we obtain ρ +
This is nothing but the well-known fact that a star made
Pr ≥ 0 and ρ + Pt ≥ 0. Thus, the weak and strong energy
of purely relativistic gas does not exist.
conditions are satisfied if
Hence, if we enter a regime where the fermion bare
mass mf is so small that, even after the scalar field has U + V + W ≥ 0, (67)
moved away from the false vacuum (where the effective
fermion mass is small by construction), the Fermi gas is 3P − 2U + W ≥ 0 , (68)
still relativistic, then the radius of the star grows fast and respectively. Since also U is a nonnegative quantity, the
a small variation in mf produces a big variation in the weak energy condition is always satisfied, while the strong
radius. We call this regime the deconfining regime of the energy condition can be violated. In particular, it is
solution. violated even in the absence of fermions (P = W = 0).
In terms of the dimensionless variables defined above,
the mf → 0 limit becomes

ω̃F → ∞. (60) 5 Note that, deep in the deconfining regime (when η → 0), the
Compton wavelength of the fermion, 1/mf , might become com-
Therefore, we expect that, for a given choice of (Λ, η), parable to or higher than the radius of the star. In this case we
the confining regime exists only if ω̃Fc is smaller then a expect the Thomas-Fermi approximation to break down.
 9

100 100

10-2 10-2

10-4 10-4

10-6 10-6

10-8 10-8
5 10 15 20 25 30 35 10 20 30 40

FIG. 2: Radial profiles of the adimensional pressure P̃ , scalar profile y and metric functions u and v for two example
configurations. Left: Λ = 0.141, η = 1.26, and P̃c = 0.00903. The mass and radius of the soliton fermion star are
µM/m2p = 6.14 and µR = 33.8, respectively. This solution falls within the confining regime. Right: Λ = 0.141,
η = 0.996, and P̃c = 0.0222. The mass and radius of the soliton fermion star are µM/m2p = 5.71 and µR = 39.3,
respectively. This solution falls within the deconfining regime.

The dominant energy condition, instead, gives two A. Numerical strategy

inequalities:
In this section, we summarise the numerical strategy
U + V + W ≥ |P + V − U |, (69) we adopt to find soliton fermion solutions. Given the
U + V + W ≥ |P − V − U |. (70) boundary condition (24), the set of equations (23) are
solved numerically by adopting the following strategy:
One can show that the dominant energy condition is
satisfied whenever 1. We fix a certain value of P̃c ;

2. for a given value of P̃c and of the central scalar field

W + 2(U + V ) ≥ P, (71)
This inequality is satisfied if
W − P ≥ 0, (72)
which can be shown to be true using the analytic expres-
sions of W and P .
To sum up, the weak and dominant energy conditions
are always satisfied, while the strong energy condition can
be violated (e.g. in the absence of fermions) as generically
is the case for a scalar field with a positive potential [11].
IV. NUMERICAL RESULTS
In this section, we present the fermion soliton solutions
in spherical symmetry obtained by integrating the field
equations (23). We will confirm the existence of a solution
beyond the thin-wall approximation used in Ref. [13]
(example solutions are shown in Fig. 2). Also, based
on the numerical solutions, we are able to confirm the
scalings derived in the previous sections in a certain region
of the parameter space and fix their prefactors.
(i.e., a value of ), we obtain ω̃F , and therefore x
through the last equation in (23);
3. for fixed P̃c and , we integrate the first three equa-
tions in (23) for the variables (u, v, y), starting from
r ≈ 0 to the point r = Rf where the fermion pres-
sure drops to negligible values, P̃ (Rf ) = 0;
4. we eliminate the fermionic quantities from the sys-
tem of equations (23) and start a new integration
with initial conditions given at r = Rf imposing
continuity of the physical quantities. That is, the
initial conditions on the metric and scalar field at
r = Rf are obtained from the last point of the
previous integration up to r = Rf ;
5. we use a shooting method to find the value of 
that allows an asymptotically-flat solution to exist,
which means imposing y(r → ∞) → 0;
6. as previously discussed, because the scalar field does
not have a compact support, we define the radius
of the star (R > Rf ) as that containing 99% of the
total mass, i.e. m̃(R) = 0.99 µM/m2p (Eq. (27)),
and the compactness is GM/R;
 10

10
0.25
8
0.20
6
0.15
4
0.10
2 0.05

0 0.00
0 10 20 30 40 0 2 4 6 8 10

101
0.25
8 100

10-1
0.20
6 10-2

10-3 0.15
4 10-4

10-5
0.10
10-1 100 101
2 0.05

0 0.00
0 5 10 15 20 25 30 35 0 2 4 6 8

FIG. 3: Mass-radius (left panels) and compactness-mass (right panels) diagrams for fermion soliton stars. The top
panels refer to various values of (Λ, η) in the confining regime (η > ηc , see Sec. III B 2). As a reference, in the top-left
panel we also draw the lines R = 2 GM , R = 9/4 GM , R = 3 GM , corresponding to the Schwarzschild radius, the
Buchdhal’s limit [21], and the photon-sphere radius. The bottom panels refer to various values of η for fixed Λ = 0.141.
The smallest value of η considered is near to but greater than the critical value. The inset shows the curves in
logarithmic scale, to highlight that in this case there exists a turning point in the M -R diagram at low masses that
proceeds towards the Newtonian limit of small M and large R.

7. Finally, we repeat the procedure for a range of values
of P̃c , finding a one-parameter family of solutions.
As we shall discuss, in certain regimes (including
the deconfining one) this family exists only if P̃c
is above a certain threshold, therefore lacking a
Newtonian limit.
As already noted, a vanishing scalar field (y = 0, ∂r y =
0) is a solution to the scalar equation in Eq. (23) only if
S = 0, that is in the absence of fermions. This ensures
that in any solution with y → 0 at infinity the fermion
pressure must vanish at some finite radius. Therefore,
the fermion soliton solution is described by a fermion
fluid confined at r ≤ Rf and endowed with a real scalar
field that is exponentially suppressed outside the star, as
expected from the discussion in Sec. III.
As described in the previous section, important param-
eters are the mass and radius of the critical solutions,
Mc and Rc . In practice, we compute these quantities by
identifying in the M -R diagram the point of maximum
mass.
B. Fermion soliton stars
First of all, we confirm that fermion soliton stars exist
also beyond the thin-wall approximation used in Ref. [13].
An example is shown in Fig. 2 which presents the radial
profiles for the metric, scalar field, and fermion pressure.
Inspecting both panels of Fig. 2 can help us understand
the qualitative difference between solutions in the con-
fining regime (left) and the deconfining one (right). In
the first case, as soon as the scalar field moves away from
its central value at ρ → 0, and the effective mass of the
fermion field grows, the pressure quickly drops to zero.
This reflects in the fact that the macroscopic size of the
star R is found to be very close to where the scalar field

102
101
100
10-1
10-2
100 101 102
FIG. 4: Left: Behaviour of the critical radius Rc with Λ and η. The scaling (62) is highlighted by the diagonal black
dashed line. We observe an agreement until Λ . 0.3 whereas, for larger Λ, ηc increasingly exceeds the predicted value.
The horizontal grid-line highlights when the µR > 1 regime ends. The shaded region above the two dashed lines is the
confining regime. Right: Behaviour of the critical radius Mc with Λ and η. We observe that the critical mass does
not exhibit a significant change of behaviour for η < ηc .
starts moving away from the false vacuum. This is the
reason why the macroscopic properties of the star are
mainly dictated by the scalar field potential. In the latter
case, the small bare mass of fermions makes them remain
ultra-relativistic even when the scalar field moves away
from the false vacuum, generating a layer where fermionic
pressure drops exponentially but remains finite. After the
energy of fermions has fallen within the non-relativistic
regime, fermionic pressure rapidly vanishes. The exis-
tence of such a layer makes the final mass and radius of
the star dependent on the fermion mass, see more details
below. Also, as the numerical shooting procedure requires
matching the asymptotic behavior of the scalar field out-
side the region where the energy density of the fermions
remains sizeable, deconfining solutions are characterized
by a larger tuning of the parameter controlling the central
displacement .
In Fig. 3 we present the mass-radius and compactness-
mass diagrams for various values of Λ and η, in the confin-
ing regime. In the top panels, we observe that Λ strongly
affects the mass-radius scale and the maximum mass,
while from the bottom panels we observe that η has a
weaker impact on the maximum mass, as expected from
the discussion in Sec. III.
The dependence of Mc and Rc on Λ and η is presented
in Fig. 4. As expected, we observe that, for a fixed Λ,
there is a critical value of η, below which the radius begins
to grow rapidly. For η > ηc and Λ . 0.5, we observe that
the predictions given in Sec. III are valid, confirming the
existence of a confining regime. Indeed, in that region
of the parameter space, both the mass and radius have
a little dependence on η. This dependence grows very
slowly for increasing value of η, in agreement with Eq. (58).
Moreover, the value of ηc scales, for Λ . 0.3, in agreement
with Eq. (62), while for larger values of Λ it exceeds the
11
102
101
100
10-1
10-2
10-3
100 101 102
analytical scaling. At variance with the critical radius,
the critical mass does not exhibit a change of behaviour
for η < ηc . As a consequence, the compactness decreases
quickly.
Finally, in Table III we report the scaling coefficients
computed numerically, which are valid in the confining
regime (η & ηc , Λ . 0.5).
C. On the existence of a Newtonian regime
From the bottom panels of Fig. 3, we observe that, even
though η has a weak impact on the maximum mass, it
can qualitatively change the M − R diagram, especially at
low masses. Overall, the mass-radius diagram reassembles
that of solitonic boson stars [15–18] with several turning
points in both the mass and the radius, giving rise to
multiple branches (see also [22]). The main branch is
the one with M 0 (R) > 0 before the maximum mass,
which is qualitatively similar to that of strange (quark)
stars [23, 24]. However, the low-mass behavior (and the
existence of a Newtonian regime) depends strongly on η.
For sufficiently large values of η (always in the confining
regime) there exists a low-compactness branch in which
M 0 (R) < 0 and where the fermionic pressure is small
compared to the energy density, giving rise to a Newtonian
regime. However, an interesting effect start occurring for
values of η near to, but greater than, the critical one (e.g.,
the blue curve for η = 1.26 in the bottom panels of Fig. 36 )
6 Notice that, in the bottom left panel, it is not possible to see the
complete tail of the M -R diagram. As underlined in the text,
in the center right panel of Fig. 5 we plot the complete M -R
diagram.
 12
−3
TABLE III: Various scaling of the critical parameters with 0.71 m2p

q
Rc ∼ ∼ 6.5 km , (75)
coefficients derived numerically in the Λ . 0.5 range. 8π q 3 5 × 105 GeV

where we included the prefactors obtained using the nu-

Critical mass µMc /m2p ≈ 0.19/Λ2
merical results. Given the cubic dependence on q, the
Critical radius µRc ≈ 0.71/Λ2 model can accommodate compact objects of vastly dif-
Compactness of the critical solution Cc ≈ 0.27
ferent mass scales, while the compactness at the maxi-
mum mass is independent of q, GMc /Rc ∼ 0.27, which
Critical value of the scale ratio ηc ≈ 2.7 Λ1/2 is slightly larger than that of a typical neutron star, but
still smaller than the compactness of the photon sphere.
As a consequence, one expects fermion soliton stars to
all the way down to the deconfining regime. In this case, display a phenomenology more akin to ordinary neutron
there is still a lower turning point in the M -R diagram, stars than to black holes [14]. The authors of Ref. [13]
but the compactness eventually starts growing (see right considered the value q = 30 GeV, yielding supermassive
bottom panel). In this case there is no Newtonian regime, objects with Mc ∼ 1012 M and Rc ∼ 1013 km ∼ 0.3 pc.
since the compactness is never arbitrarily small. Instead, the choice
This peculiar behavior is also related to another im-
portant feature of the model, namely the fact that, for η q = qastro ∼ 5 × 105 GeV (76)
sufficiently small, fermion soliton stars exist only above
a minimum threshold for the central fermionic pressure. leads to the existence of soliton solutions of mass and
We clarify this point in Fig. 5. In the left panels we radius comparable to ordinary neutron stars.
show the mass of the star as a function of the central Furthermore, the fact that the model is in the confining
fermionic pressure for Λ = 0.141 and three values of η. regime only above a critical value of η, Eq. (62), implies
For η = 0.966 and η = 1.26 (top and center panels), the (using Eq. (22) and our numerical results):
pressure has a lower bound, corresponding to the absence
 √8πq 3 1/2 3/2
of a Newtonian limit. For η = 2.92 (bottom panels) the

q
mf > 2.7 ∼ 0.6 GeV , (77)
behavior is qualitatively different and in this case the mp qastro
Newtonian regime is approached as Pc → 0.
To clarify where the minimum pressure and these mul- a range including the neutron mass. Therefore, the
tiple branches are in the mass-radius diagram, in the right fermion gas can be a standard degenerate gas of neutrons.
panels of Fig. 5, we show data points for M − R using It is also interesting to combine the above inequality (sat-
the same color scheme as in the corresponding left panels. urated when mf = mcf ) with Eq. (74), finding a relation
Interestingly, the minimum pressure does not correspond between the maximum mass of the soliton in the confining
to the minimum mass in Fig. 5, but it is an intermediate regime and the critical fermion mass,
point in the M − R diagram. In the center right panel we !2
show an extended version of the Λ = 0.141, η = 1.26 curve GeV
shown in Fig. 3. This highlights the peculiar behavior of Mc ∼ 0.46 M , (78)
mcf
the new branch, which has a further turning point at large
radii. Studying the stability of these different peculiar
independently of q. Interestingly, this model allows for
branches [22] is left for future work.
subsolar compact objects for fermions at (or slightly heav-
Finally, note that in both cases there are values of
ier than) the GeV scale, whereas it allows for supermassive
the central fermionic pressure corresponding to multiple
(Mc ∼ 106 M ) compact stars for a degenerate gas of elec-
solutions, each one identified by a different central value
trons (mcf ∼ 0.5 MeV).
of the scalar field.
Clearly, the same value of q can be obtained with dif-
ferent combinations of µ and φ0 . In general,
V. PARAMETER SPACE AND 3  2
500 TeV

ASTROPHYSICAL IMPLICATIONS q
µ = 500 TeV (79)
qastro φ0
Given the number of parameters of our model, it is
2  2
mcf 500 TeV

interesting to study the characteristic mass and radius of = 500 TeV , (80)
0.6 GeV φ0
fermion soliton stars in this theory. By defining
so µ ∼ GeV for q = qastro (or, equivalently, for mcf =
q ≡ (µφ20 )1/3 , (73)
0.6 GeV) and φ0 ∼ 3 × 105 TeV. Note that the latter
as long as we are in the confining regime, one finds value is still much smaller than the Planck scale, so the
−3 condition Λ  1 is satisfied. From our numerical results,
0.19 m4p Eqs. (74)–(75) are valid as long as Λ . 0.5 whereas, for

q
Mc ∼ ∼ 1.27 M , (74)
8π q 3 5 × 105 GeV larger values of Λ, Mc , Rc , and Cc decrease rapidly and
 13

10 10 10
9
8
9 7 9
6
8 5 8
4

7 0.02 0.03 0.04 0.05

7
6 6
5 5
4 4
-2 -1 0 1 2 3
10 10 10 10 10 10 20 40 60 80 100 120 140

10 10

8 8

6 6

4 4

2 2

0 0
10-2 10-1 100 101 102 20 40 60 80 100 120

10
101 101

8
100
0
10
6 10-1

10-1 4 10-2

0 10 20 30 40

-2
2
10
0
10-4 10-3 10-2 10-1 100 0 5 10 15 20 25 30 35

FIG. 5: Left panels: The mass of fermion soliton stars as a function of the central fermionic pressure. Right panels:
The corresponding mass-radius diagram using the same color scheme as in the left panels, in order to associate to each
point the corresponding central pressure. Top: Λ = 0.141 and η = 0.996. This solution is in the deconfining regime
and there is a lower bound on P̃c below which no solution exists. Center: Λ = 0.141 and η = 1.26. This solution is in
the confining regime but, also in this case, there exists a lower bound on P̃c . Bottom: Λ = 0.141 and η = 2.92. This
solution is in the confining regime but, given the larger value of η, there is no lower bound on P̃c and a Newtonian
regime exists. In all three cases, for a certain range of P̃c there are multiple solutions with the same central fermionic
pressure and different central value of the scalar field.

the condition µR  1 might not hold (see Fig. 4). This which, using Eq. (79), can be translated into a lower
gives an upper bound on φ0 bound on µ
3
0.5

q
φ0 . √ mp ∼ 1018 GeV, (81) µ & 8.4 × 10 −11
eV. (82)
8π qastro

Thus, also the scalar-field mass can vastly change depend-
ing on the value of q, reaching a lower limit that can
naturally be in the ultralight regime.
Finally, in the deconfining regime there is no minimum
fermion mass so solutions can exist also beyond the range
dictated by Eq. (77), but soliton fermion stars in such a
regime would be characterized by smaller values of the
compactness (see discussion in Sec IV).
VI. CONCLUSIONS
We have found that fermion soliton stars exist as static
solutions to Einstein-Klein-Gordon theory with a scalar
potential and a Yukawa coupling to a fermion field. This
confirms the results of Ref. [13] obtained in the thin-wall
approximation and provides a way to circumvent the no-
go theorems [10, 11] for solitons obtained with a single
real scalar field.
Focusing on spherical symmetry, we have explored the
full parameter space of the model and derived both an-
alytical and numerical scalings for some of the relevant
quantities such as the critical mass and radius of a fermion
soliton star. Interestingly, the model predicts the exis-
tence of solutions in the subsolar/solar (resp. supermas-
sive) range for a standard gas of degenerate neutrons
(resp. electrons).
We also unveiled the existence of a confining and decon-
fining regime – where the macroscopic properties of the
soliton are mostly governed by the scalar field parameters
or by the fermion mass, respectively – and the fact that
no Newtonian analog exists for these solutions for fermion
masses below a certain threshold.
Extensions of our work are manifold. First of all, for
simplicity, we have focused on a scalar-fermion coupling
tuned to provide an almost vanishing effective fermion
mass in the stellar core. This assumption imposes f =
mf /φ0 , a condition that can be relaxed, thus increasing
the dimensionality of the parameter space. We have
also considered a scalar potential with two degenerate
minima. A straightforward generalization is to break this
degeneracy and allow for a true false-vacuum potential
in which the scalar field transits from the false-vacuum
state inside the star to the true-vacuum state at infinity.
From the point of view of the fundamental theory, it
would be interesting to investigate an embedding within
the Standard Model and beyond, also including gauge
fields (e.g., see Ref. [25] for a recent attempt along this
direction).
Finally, although we focused on static and spherically
symmetric solutions, there is no fundamental obstacle in
considering spinning configurations and the dynamical
regime, both of which would be relevant to study the
phenomenology of fermion soliton stars, along the lines
of what has been widely studied for boson stars [9] and
for mixed fermion-boson stars [26]. In particular, due to
the existence of multiple branches [22] and the absence of
a Newtonian limit in certain cases, an interesting study
14
concerns the radial linear stability of these solutions.
We hope to address these points in future work.
ACKNOWLEDGMENTS
We thank Enrico Barausse, Mateja Bošković, and Mas-
simo Vaglio for useful conversations. G.F. and P.P. ac-
knowledge financial support provided under the Euro-
pean Union’s H2020 ERC, Starting Grant agreement
no. DarkGRA–757480 and under the MIUR PRIN pro-
gramme, and support from the Amaldi Research Cen-
ter funded by the MIUR program “Dipartimento di Ec-
cellenza" (CUP: B81I18001170001). The research of
A.U. was supported in part by the MIUR under con-
tract 2017 FMJFMW (“New Avenues in Strong Dynam-
ics,” PRIN 2017). This work was supported by the EU
Horizon 2020 Research and Innovation Programme un-
der the Marie Sklodowska-Curie Grant Agreement No.
101007855.
Appendix A: Connection with scalar-tensor theories
In this appendix we discuss whether the model for
fermion soliton stars presented in the main text can also
arise in the context of a scalar-tensor theory of gravity
(see, e.g., [27] for a review on modified theories of gravity).
In the so-called Jordan frame7 , where gravity is mini-
mally coupled to matter fields, scalar-tensor theories are
described by the action (see for example [28])
√
−ĝ h
Z i
Ŝ = d4 x F (φ̂)R̂ − Z(φ̂)ĝ µν ∂µ φ̂∂ν φ̂ − Û (φ̂) +
16πG
Ŝm (ψ̂m ; ĝµν ) . (A1)
The coupling functions F and Z single out a particular
theory within the class. For example, Brans-Dicke theory
corresponds to F = φ̂ and Z = ω0 /φ̂, where ω0 is a
constant.
We can write the theory in an equivalent form in the
so-called Einstein frame, where gravity is minimally cou-
pled to the scalar field. For this purpose, we perform a
conformal transformation of the metric, ĝµν = A2 (φ)gµν
with A(φ) = F −1/2 (φ̂), a field redefinition, φ = φ(φ̂), and
a conformal rescaling of the matter field, ψ̂m → ψm . The
scalar field φ is now minimally coupled to gµν whereas ψm
is minimally coupled to ĝµν [28]. The energy-momentum
tensor is Tµν = A2 (φ)T̂µν whereas the scalar potential
Û (φ̂)
becomes U (φ) = 16πGF 2 (φ̂)
7 In this appendix we used a hat to denote quantities in the Jordan
frame, whereas quantities without the hat refer to the Einstein
frame where gravity is minimally coupled to the scalar field.
 15

The scalar field equation in the Einstein frame reads In the Einstein frame, Eq. (A7) simply reads

d log A(φ) ∂U T µν = (W + P )uµ uν + g µν P , (A10)

φ = −T + . (A2)
dφ ∂φ
where W = A4 (φ)Ŵ and P = A4 (φ)P̂ . Therefore, also
Since in our theory (1) the scalar field is minimally in the Einstein frame we have a perfect fluid in the form
coupled to gravity, it is natural to interpret it in the of a zero-temperature Fermi gas. Let us now compute
context of the Einstein frame. Thus, we can compare the expressions of W and P explicitly. First of all, from
Eq. (A2) to the second equation in (11): Eq. (A8), following the same computation presented in
the main text, we get
∂U
φ = −f S + , (A3)
∂φ m4f h p  p i
Ŵ = x̂ 1 + x̂ 2 (1 + 2x̂2 ) − log x̂ + x̂ 2+1
which, using Eq. (8), can be written as 8π 2
m4f h  2 2 p  p i
f ∂U P̂ = x̂ x̂ − 1 1 + x̂ 2 + log x̂ + x̂ 2+1
φ = T+ . (A4) 8π 2 3
(mf − f φ) ∂φ (A11)

Therefore, if we identify where x̂ = k̂F /mf . Since A(φ) = meff /mf , we obtain
d log A(φ) −f 1 m4eff h p
= = (A5)
 i
,
p
dφ (mf − f φ) φ − φ0 W = x̂ 1 + x̂ 2 (1 + 2x̂2 ) − log x̂ + x̂ 2+1
8π 2
the scalar equation of our model is the same as in a m4 h  2 p  p i
P = eff2 x̂ x̂2 − 1 1 + x̂2 + log x̂ + x̂2 + 1 .
scalar-tensor theory with coupling A(φ) in the Einstein 8π 3
frame. Integrating this equation yields (henceforth as- (A12)
sumig A(0) = 1),
Note that W (x̂) and P (x̂) above implicitly define an equa-
φ meff tion of state that is exactly equivalent to that obtained
A(φ) = 1 − = . (A6) from W and P in Eqs. (21a)– (21b). This shows that
φ0 mf
our model can be interpreted as a scalar-tensor theory
Interestingly, the matter coupling vanishes when φ ≈ φ0 . in the Einstein frame with coupling to matter given by8
It is left to be checked if the gravitational sector of our A(φ) = meff /mf .
model is equivalent to that of a scalar-tensor theory with Furthermore, note that the dimensionless quantity x̂ =
A(φ) given by Eq. (A6). Let us consider a degenerate gas k̂F /mf = kF /meff = x is invariant under a change from
of noninteracting fermions with mass mf in the Jordan the Jordan to the Einstein frame. Therefore, W and P
frame, with energy-momentum are exactly those given in Eqs. (21a)–(21b).
Finally, Ŝ in the Jordan frame reads
T̂ µν = (Ŵ + P̂ )ûµ v̂ ν + ĝ µν P̂ (A7)
2 k̂F
Z
mf
where, assuming spherical symmetry, Ŝ = d3 k q =
(2π)3 0 k 2 + m2f
(A13)
2 k̂F (ρ̂)
Z q
Ŵ (ρ̂) = d3 k k 2 + m2f m3f h p  p i
(2π)3 x̂ 1 + x̂ 2 − log x̂ + x̂ 2+1
0
(A8) 2π 2
2 k̂F (ρ̂)
k2
Z
P̂ (ρ̂) = d3 k q . while in the Einstein frame9
(2π)3 0 3 k 2 + m2f
m3eff h p  p i
S = A3 Ŝ = x 1 + x 2 − log x + x 2+1 ,
In spherical symmetry, since the spacetime has the same 2π 2
form as in Eq. (13), it is straightforward to minimize the (A14)
energy of the fermion gas at fixed number of fermions
(the calculation is exactly the same as the one done to
obtain Eq. (18)):
8 Note that our model and the scalar-tensor theory are not exactly
equivalent to each other. Indeed, while in the scalar-tensor theory
k̂F2 = ω̂F2 e−2û − m2f . (A9) any matter field is universally coupled to A(φ)ĝµν , in our model
this is the case only for the fermion gas, while any other matter
It is important to notice that in the standard scalar- field is minimally coupled to the metric, in agreement with the
tensor theory in the Jordan frame there is no Yukawa fact that our model is based on standard Einstein’s gravity.
interaction, therefore the fermion particles do not acquire 9 The fact that S = A3 Ŝ can be derived from the condition
any effective mass. A4 (φ)T̂ = T ⇒ A4 (φ)mf Ŝ = meff S.

since x̂ = x. Thus, also in this case we obtain the same
expression as in Eq. (21c).
Having assessed that our model can be interpreted in
the context of a scalar-tensor theory, it is interesting to
[1] J. A. Wheeler, Phys. Rev. 97, 511 (1955).
[2] C. A. R. Herdeiro and E. Radu, Int. J. Mod. Phys. D 24,
1542014 (2015), arXiv:1504.08209 [gr-qc].
[3] D. J. Kaup, Phys. Rev. 172, 1331 (1968).
[4] R. Ruffini and S. Bonazzola, Phys. Rev. 187, 1767 (1969).
[5] M. Colpi, S. L. Shapiro, and I. Wasserman, Phys. Rev.
Lett. 57, 2485 (1986).
[6] S. R. Coleman, Nucl. Phys. B 262, 263 (1985), [Adden-
dum: Nucl.Phys.B 269, 744 (1986)].
[7] P. Jetzer, Phys. Rept. 220, 163 (1992).
[8] F. E. Schunck and E. W. Mielke, Class. Quant. Grav. 20,
R301 (2003), arXiv:0801.0307 [astro-ph].
[9] S. L. Liebling and C. Palenzuela, Living Rev. Rel. 15, 6
(2012), arXiv:1202.5809 [gr-qc].
[10] G. H. Derrick, J. Math. Phys. 5, 1252 (1964).
[11] C. A. R. Herdeiro and J. a. M. S. Oliveira, Class. Quant.
Grav. 36, 105015 (2019), arXiv:1902.07721 [gr-qc].
[12] E. Seidel and W. M. Suen, Phys. Rev. Lett. 66, 1659
(1991).
[13] T. D. Lee and Y. Pang, Phys. Rev. D 35, 3678 (1987).
[14] V. Cardoso and P. Pani, Living Rev. Rel. 22, 4 (2019),
arXiv:1904.05363 [gr-qc].
[15] R. Friedberg, T. D. Lee, and Y. Pang, Phys. Rev. D 35,
3658 (1987).
[16] C. Palenzuela, P. Pani, M. Bezares, V. Cardoso, L. Lehner,
and S. Liebling, Phys. Rev. D 96, 104058 (2017),
16
study the latter in the Jordan frame. In particular, since
1
A(φ) = q , (A15)
F (φ̂)
and A(φ) = 1 − φ/φ0 , the coupling function F (φ̂) is
singular in φ̂(φ0 ). In the language of the scalar-tensor
theory, we see that in the core of a fermion soliton star,
where φ ≈ φ0 and matter is almost decoupled in the
Einstein frame, the scalar field in the Jordan frame is
strongly coupled to gravity.
arXiv:1710.09432 [gr-qc].
[17] M. Bezares, M. Bošković, S. Liebling, C. Palenzuela,
P. Pani, and E. Barausse, Phys. Rev. D 105, 064067
(2022), arXiv:2201.06113 [gr-qc].
[18] M. Bošković and E. Barausse, JCAP 02, 032 (2022),
arXiv:2111.03870 [gr-qc].
[19] S. L. Shapiro and S. A. Teukolsky, Black holes, white
dwarfs, and neutron stars: The physics of compact objects
(1983).
[20] T. Lee and Y. Pang, Physics Reports 221, 251 (1992).
[21] H. A. Buchdahl, Phys. Rev. 116, 1027 (1959).
[22] D. Guerra, C. F. B. Macedo, and P. Pani, JCAP
09, 061 (2019), [Erratum: JCAP 06, E01 (2020)],
arXiv:1909.05515 [gr-qc].
[23] C. Alcock, E. Farhi, and A. Olinto, Astrophys. J. 310,
261 (1986).
[24] A. Urbano and H. Veermäe, JCAP 04, 011 (2019),
arXiv:1810.07137 [gr-qc].
[25] Y. Endo, H. Ishihara, and T. Ogawa, Phys. Rev. D 105,
104041 (2022), arXiv:2203.09709 [hep-th].
[26] S. Valdez-Alvarado, C. Palenzuela, D. Alic, and
L. A. Ureña López, Phys. Rev. D 87, 084040 (2013),
arXiv:1210.2299 [gr-qc].
[27] E. Berti et al., Class. Quant. Grav. 32, 243001 (2015),
arXiv:1501.07274 [gr-qc].
[28] T. P. Sotiriou and V. Faraoni, Rev. Mod. Phys. 82, 451
(2010), arXiv:0805.1726 [gr-qc].