betaFeedback

WorksPhysics and Astronomy

Comment on "Exact, nonsingular black holes from a phantom DBI field as primordial dark matter"

Abstract

We demonstrate that the metric and scalar configuration proposed in Ref.~[1] do not satisfy the complete set of field equations of the Dirac-Born-Infeld (DBI) field theory minimally coupled to general relativity on a radial domain containing the event horizon. The physical DBI branch for this gravitational setup terminates outside the event horizon. Thus, the proposed configuration does not describe a black hole solution of that theory.

Recently, in the framework of general relativity minimally coupled to a Dirac-Born-Infeld (DBI) scalar field, the authors of Ref. [1] obtained an exact regular solution and claimed that it was a black hole (BH). We show that this solution is indeed regular, but it does not describe a BH, since the field equations are satisfied in a radial domain that does not include an event horizon. Throughout this comment, we use the notation of Ref. [1], except that we replace κ2\kappa^2 with κ\kappa.

The action of general relativity minimally coupled to the DBI scalar field ϕ\phi is given by

S=∫d4x−g(R2κ+ϵLϕ),(1)S=\int d^4x\sqrt{-g}\left(\frac{R}{2\kappa}+\epsilon\mathcal{L}_{\phi}\right), \tag*{(1)}

where gg is the determinant of the metric tensor gg, RR is the Ricci scalar, κ=8πG/c4\kappa=8\pi G/c^4, and ϵ=−1\epsilon=-1 (ϵ=+1\epsilon=+1) for a normal (phantom) scalar field. Moreover, Lϕ\mathcal{L}_{\phi} is the Lagrangian density associated with the scalar field, namely,

Lϕ=V(ϕ)(1−1+2XΛ4),X≡−12gμν∇μϕ∇νϕ.(2)\mathcal{L}_{\phi}=V(\phi)\left(1-\sqrt{1+\frac{2X}{\Lambda^4}}\right),\qquad X\equiv-\frac{1}{2}g^{\mu\nu}\nabla_{\mu}\phi\nabla_{\nu}\phi. \tag*{(2)}

in which Λ\Lambda is the intrinsic energy scale, V(ϕ)V(\phi) is the DBI potential function. The equations of motion assume the form [1]

∇μ(V(ϕ)∇μϕQ)=Λ4V′(ϕ)(1−Q),(3)\nabla_{\mu}\left(\frac{V(\phi)\nabla^{\mu}\phi}{Q}\right)=\Lambda^4V'(\phi)(1-Q), \tag*{(3)}
Gνμ=κϵ[Lϕδνμ−V(ϕ)(∇μϕ)(∇νϕ)Λ4Q]=κTνμ,(4)G^{\mu}_{\nu}=\kappa\epsilon\left[\mathcal{L}_{\phi}\delta^{\mu}_{\nu}-\frac{V(\phi)(\nabla^{\mu}\phi)(\nabla_{\nu}\phi)}{\Lambda^4Q}\right]=\kappa T^{\mu}_{\nu}, \tag*{(4)}

with

Q=1+2XΛ4>0.(5)Q=\sqrt{1+\frac{2X}{\Lambda^4}}>0. \tag*{(5)}

In this work, the prime notation (′') implies a derivative with respect to the argument shown between parentheses. Consequently, ϕ′(r)=dϕ/dr\phi'(r)=d\phi/dr and ϕ˙(t)=dϕ/dt\dot{\phi}(t)=d\phi/dt.

Choosing the metric ansatz of the form

ds2=−c2f(r)dt2+dr2f(r)+ρ2(r)dΩ2,(6)ds^2=-c^2f(r)dt^2+\frac{dr^2}{f(r)}+\rho^2(r)d\Omega^2, \tag*{(6)}

† [email protected]

Eqs. (3) and Gtt=GθθG^t_t=G^\theta_\theta reduce to

1ρ2(r)ddr[ρ2(r)f(r)ϕ′(r)V(ϕ)Q]=Λ4V′(ϕ)(1−Q),(7)\frac{1}{\rho^2(r)}\frac{d}{dr}\left[\frac{\rho^2(r)f(r)\phi'(r)V(\phi)}{Q}\right]=\Lambda^4V'(\phi)(1-Q), \tag*{(7)}
ρ2(r)f′′(r)−2[(ρ′(r))2+ρ(r)ρ′′(r)]f(r)+2=0,(8)\rho^2(r)f''(r)-2\left[(\rho'(r))^2+\rho(r)\rho''(r)\right]f(r)+2=0, \tag*{(8)}

respectively. In Ref. [1], ρ(r)\rho(r) was chosen ρ(r)=r2+a2\rho(r)=\sqrt{r^2+a^2}, and it yields the function f(r)f(r) upon solving (8). Thus, we get

f(x)=1+3xg[x−(1+x2)arctan⁡(1x)],(9)f(x)=1+3x_g\left[x-(1+x^2)\arctan\left(\frac{1}{x}\right)\right], \tag*{(9)}

where aa is a regularization parameter, and we define

x≡raandxg≡GMac2.(10)x\equiv\frac{r}{a}\qquad\text{and}\qquad x_g\equiv\frac{GM}{ac^2}. \tag*{(10)}

Now, from (4), we obtain the following equations:

Gtt=ϵκV(ϕ)[1−Q(r)],(11)G^t_t=\epsilon\kappa V(\phi)[1-Q(r)], \tag*{(11)}
Gtt−Grr=ϵκV(ϕ)[1−Q(r)]1+Q(r)Q(r).(12)G^t_t-G^r_r=\epsilon\kappa V(\phi)[1-Q(r)]\frac{1+Q(r)}{Q(r)}. \tag*{(12)}

In terms of f(r)f(r) and ρ(r)\rho(r), we obtain

Gtt−Grr=2ρ′′(r)ρ(r)f(r),(13)G^t_t-G^r_r=2\frac{\rho''(r)}{\rho(r)}f(r), \tag*{(13)}

with ρ′′(r)=a2/ρ3>0\rho''(r)=a^2/\rho^3>0. We can cancel ff from both sides of (12), assuming that f≠0f\ne0 in order to show that

2ρ′′(r)ρ(r)=ϵκV(ϕ)(ϕ′(r))2Λ4Q(r).(14)2\frac{\rho''(r)}{\rho(r)}=\epsilon\kappa V(\phi)\frac{(\phi'(r))^2}{\Lambda^4Q(r)}. \tag*{(14)}

At the points where (ϕ′(r))2(\phi'(r))^2 remains positive and finite, all we can conclude from (14) is that ϵV(ϕ)/Q\epsilon V(\phi)/Q is finite, i.e.,

0<ϵV(ϕ)Q(r)<∞.(15)0<\epsilon\frac{V(\phi)}{Q(r)}<\infty. \tag*{(15)}

Since Q>0Q>0, (15) implies ϵV>0\epsilon V>0. Yet, we emphasize a point concerning the sign of the matter sector. In (1), ϵ\epsilon and VV enter only through their product. The low-energy kinetic term is ϵLϕ=−(ϵV(ϕ)/Λ4)X+O(X2)\epsilon\mathcal{L}_{\phi}=-(\epsilon V(\phi)/\Lambda^4)X+O(X^2). Therefore, the combinations (ϵ,V)=(−1,V<0)(\epsilon,V)=(-1,V<0) and $(\epsilon,V)=(+1,V>`

  1. describe the same phantom-sign kinetic term; the former is not a standard scalar field. Moreover, Eq. (15) constrains only the ratio ϵV(ϕ)/Q(r)\epsilon V(\phi)/Q(r). Consequently, if VV and QQ vanish simultaneously, the ratio may have a finite positive limit even though the required condition Q>0Q>0 fails at that point. The ratio of Eq. (12) to Eq. (11) yields

Q(r)=−GttGrr.(16)Q(r)=-\frac{G^{t}_{t}}{G^{r}_{r}}. \tag*{(16)}

As we shall see later, the ratio Gtt/GrrG^{t}_{t}/G^{r}_{r} does not have a constant sign, and since we want Q≥0Q\geq0, this implies that the solution may not be valid on the whole half real axis [0,∞)[0,\infty). From Eqs. (5), (11), and (16), we find that

ϵV(ϕ)=GttGrrκ(Gtt+Grr),(17)\epsilon V(\phi)=\frac{G^{t}_{t}G^{r}_{r}}{\kappa(G^{t}_{t}+G^{r}_{r})}, \tag*{(17)}
(ϕ′(r))2Λ4=1−Q2f=1f[1−(GttGrr)2].(18)\frac{(\phi'(r))^{2}}{\Lambda^{4}}=\frac{1-Q^{2}}{f}=\frac{1}{f}\left[1-\left(\frac{G^{t}_{t}}{G^{r}_{r}}\right)^{2}\right]. \tag*{(18)}

Here again, the sign of GttGrr/(Gtt+Grr)G^{t}_{t}G^{r}_{r}/(G^{t}_{t}+G^{r}_{r}) is not constant.

It was stated in Ref. [1] that for

M>Mrelic≡2ac23πG⟹xg>xBH≡23π,(19)M>M_{\rm relic}\equiv\frac{2ac^{2}}{3\pi G}\Longrightarrow x_{g}>x_{\rm BH}\equiv\frac{2}{3\pi}, \tag*{(19)}

the solution (9) is a non-extremal BH spacetime. We show that this statement is not correct. From Eq. (5), we see that if (ϕ′(r))2(\phi'(r))^{2} remains finite, then the functions ff and 1−Q1-Q must have the same sign if f(r)f(r) corresponds to a BH, provided that QQ remains real and positive. In this case, when ff changes sign at the horizon rhr_h, 1−Q1-Q also changes sign, and the limit of their ratio remains finite (this applies to ff and 1−Q21-Q^{2} as well). In fact, this was justified graphically in the upper plot of Figure 2 in Sec. IV of the Supplemental Material of Ref. [1], where (1−Q2)/f(1-Q^{2})/f is shown to maintain a positive sign. Yet, at the horizon, the combination of Eqs. (13) and (16) leads to

Figure 2. Top panel: Relevant functions 2a2(1+x2)2Gtt2a^{2}(1+x^{2})^{2}G^{t}_{t}, (ϕ′(r))2/(2Λ4)(\phi'(r))^{2}/(2\Lambda^{4}), f(r)f(r), Q(r)Q(r), and 8a2πκϵV8a^{2}\pi\kappa\epsilon V, as a function of x=r/ax=r/a, corresponding to the solution Eq. (9). Bottom panel: Same as the top panel, but for x∈[2,4]x\in[2,4]. We took xg=11/(3π)=5.5xBHx_{g}=11/(3\pi)=5.5x_{\mathrm{BH}} (M=5.5MrelicM=5.5M_{\mathrm{relic}}). Moreover, we have Gtt(3.23675)=Q(3.23675)=V(3.23675)=0G^{t}_{t}(3.23675)=Q(3.23675)=V(3.23675)=0 and f(2.24903)=0f(2.24903)=0.

Gtt(rh)[1+Q(rh)]=0.(20)G^{t}_{t}(r_h)[1+Q(r_h)]=0. \tag*{(20)}

Thus, we have two options, namely, either Gtt(rh)=0G^{t}_{t}(r_h)=0 or Q(rh)=−1Q(r_h)=-1. The function Gtt(rh)G^{t}_{t}(r_h) can be written as

Gtt(rh)=rhf′(rh)−1a2+rh2.(21)G^{t}_{t}(r_h)=\frac{r_h f'(r_h)-1}{a^{2}+r_h^{2}}. \tag*{(21)}

Solving f(rh)=0f(r_h)=0 for xgx_g and inserting the resulting expression into the equation above, one can show that it implies that for every horizon radius rh>0r_h>0, we have Gtt(rh)<0G^{t}_{t}(r_h)<0. Therefore, the product can vanish only through the second option, yielding Q(rh)=−1Q(r_h)=-1. However, Eq. (5) implies that Q(r)>0Q(r)>0 holds throughout the claimed radial domain. Consequently, Eq. (20) can be satisfied only if Gtt(rh)=0G^{t}_{t}(r_h)=0. This contradicts Gtt(rh)<0G^{t}_{t}(r_h)<0. Hence, the field equations cannot be satisfied at rhr_h, and the solution cannot describe a BH.

We illustrate the above conclusions in Figure 1 of the Supplemental Material. Using Eqs. (16)–(18) and Eq. (9), we display 2a2(1+x2)2Gtt2a^{2}(1+x^{2})^{2}G^{t}_{t}, (ϕ′(r))2/(2Λ4)(\phi'(r))^{2}/(2\Lambda^{4}), f(r)f(r), Q(r)Q(r), and 8a2πκϵV8a^{2}\pi\kappa\epsilon V for xg=4/(3π)=2xBHx_g=4/(3\pi)=2x_{\rm BH} (M=2MrelicM=2M_{\rm relic}) [cf. Eq. (19)].

Figure 1. Relevant functions 2a2(1+x2)2Gtt2a^{2}(1+x^{2})^{2}G^{t}_{t}, (ϕ′(r))2/(2Λ4)(\phi'(r))^{2}/(2\Lambda^{4}), f(r)f(r), Q(r)Q(r), and 8a2πκϵV8a^{2}\pi\kappa\epsilon V, as a function of x=r/ax=r/a, corresponding to the solution given by Eq. (9). We took xg=4/(3π)=2xBHx_{g}=4/(3\pi)=2x_{\mathrm{BH}} (M=2MrelicM=2M_{\mathrm{relic}}). We have Gtt(1.10278)=Q(1.10278)=V(1.10278)=0G^{t}_{t}(1.10278)=Q(1.10278)=V(1.10278)=0 and f(0.621184)=0f(0.621184)=0.

For sufficiently large values of xgx_g, as illustrated in Figure 2 of the Supplemental Material, ϕ′(r)\phi'(r) diverges at finite radii rsr_s where Grr(rs)=0G^{r}_{r}(r_s)=0. Near such a zero, we have (ϕ′(r))2∝(r−rs)−2(\phi'(r))^{2}\propto(r-r_s)^{-2}, so the scalar field itself diverges logarithmically at finite rr. This first occurs at xg≃0.989≃4.66xBHx_g\simeq0.989\simeq4.66x_{\rm BH}. However, these divergences occur inside the horizon, where the scalar gradient is timelike (X>0X>0). Therefore, they do not violate the DBI square-root reality bound.

In summary, the metric function (9) solves the reduced geometric equation, but the complete DBI field equations are not satisfied on a radial domain containing the horizon. The obstruction is the sign condition Q>0Q>0: the horizon requires Q=−1Q=-1, and the physical exterior branch ends at r∗>rhr_{*}>r_h with Q(r∗)=0Q(r_{*})=0. The result concerns the specific action (1). Moreover, since Q→1Q\to1 asymptotically and Q(rh)=−1Q(r_h)=-1, the function Q(r)Q(r) vanishes at r∗r_{*} if f(r∗)>0f(r_{*})>0, meaning that the valid exterior domain where Q>0Q>0 terminates before reaching the horizon f(rh)=0f(r_h)=0.

As a final comment, for the generalized ansatz ρ(r)=(rp+ap)1/p\rho(r)=(r^{p}+a^{p})^{1/p} considered in Ref. [1], asymptotically flat solutions of Eq. (8) can be constructed for any p≥1p\geq1, provided that f(r)f(r) is nonconstant. However, boundedness of ρ′′(r)\rho''(r) at r=0r=0 requires p≥2p\geq2, while smooth radial continuation through r=0r=0 requires even pp. These qualifications distinguish asymptotic flatness from regularity at the center. Further details on these issues are outside the scope of this Comment and will be investigated elsewhere [2].

M. A. A. de Paula acknowledges the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) from Brazil for partial financial support under the Grant No. CNPq/PJD 150589/2025-5.

Data availability

There are no publicly available research data or software supporting this manuscript. Requests for further information or data should be sent to the authors.

References

Supplemental Material

In this Supplemental Material, all equation numbers and citations refer to the Comment.

In Figure 1, using Eqs. (16)–(18) and Eq. (9) of the Comment, we display 2a2(1+x2)2Gtt2a^{2}(1+x^{2})^{2}G^{t}_{t}, (ϕ′(r))2/(2Λ4)(\phi'(r))^{2}/(2\Lambda^{4}), f(r)f(r), Q(r)Q(r), and 8a2πκϵV8a^{2}\pi\kappa\epsilon V for xg=4/(3π)=2xBHx_{g}=4/(3\pi)=2x_{\mathrm{BH}} (M=2MrelicM=2M_{\mathrm{relic}}) [cf. Eq. (19)]. We see that Q<0Q<0 (and ϵV<0\epsilon V<0) for 0≤x<1.102780\leq x<1.10278, which implies that the solution is not valid for 0≤x<1.102780\leq x<1.10278 and shows that Gtt/GrrG^{t}_{t}/G^{r}_{r} does not have a constant sign. We also observe that at the point where ff vanishes, x=0.621184x=0.621184, both (ϕ′(r))2(\phi'(r))^{2} and ϵV\epsilon V are finite, but Gtt(x=0.621184)≠0G^{t}_{t}(x=0.621184)\ne0, showing that Eq. (9) does not satisfy Eq. (11). This is because at x=0.621184x=0.621184, we have Q(0.621184)=−1Q(0.621184)=-1 instead of Q(0.621184)=+1Q(0.621184)=+1. The authors of [1] have checked in Figure 2 of their Supplemental Material the sign of 1−Q21-Q^{2} against that of ff, and should have checked the sign of 1−Q1-Q against that of ff instead.

We have obtained similar graphs for other values of xg>xBHx_{g}>x_{\mathrm{BH}} where QQ and (ϕ′(r))2(\phi'(r))^{2} remain finite for all x≥0x\geq0. For higher values of xgx_{g}, specifically beginning at xg≃0.989≃4.66xBHx_{g}\simeq0.989\simeq4.66x_{\mathrm{BH}}, the scalar derivative develops divergences at finite radii rsr_{s} where Grr(rs)=0G^{r}_{r}(r_{s})=0. For xg=11/(3π)=5.5xBHx_{g}=11/(3\pi)=5.5x_{\mathrm{BH}} (M=5.5MrelicM=5.5M_{\mathrm{relic}}), as shown in Figure 2, the quantities remain finite near the zero of f(x)f(x) and diverge toward the center. At x=2.24903x=2.24903, where f=0f=0, Gtt≠0G^{t}_{t}\ne0 and Q=−1Q=-1. The figure also shows the behavior in the outer-radius interval x∈[2,4]x\in[2,4].

References

  1. [1]Visible citation only: Ref. [1].
  2. [2]Azreg-Aïnou, M. (2026). Exact, regular static and dynamic solutions sourced by a DBI scalar field: No black holes. Phys. Lett. B, 883, 141154.

Paper details

vlnc:W_k6dkptc3
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 4

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents