Euler-Heisenberg AdS Black Hole

• Euler–Heisenberg AdS black holes are charged anti-de Sitter solutions incorporating leading one-loop QED corrections that modify the near-horizon geometry.
• Their metric function extends the Reissner–Nordström–AdS form with an extra term proportional to aQ^4/r^6, shifting horizons, the ISCO, and thermodynamic properties.
• Studies reveal that these corrections affect geodesic motion, chaos bounds, and phase transitions, linking nonlinear electrodynamics with observable black hole phenomena.

An Euler–Heisenberg AdS black hole is a charged anti-de Sitter solution of Einstein gravity coupled to Euler–Heisenberg nonlinear electrodynamics, obtained by retaining the leading one-loop QED correction to the Maxwell sector. In four dimensions, the purely electric, static, spherically symmetric solution is characterized by a metric function of the Reissner–Nordström–AdS form plus a short-distance correction proportional to $-\,aQ^4/(20r^6)$, so that the Euler–Heisenberg parameter modifies the near-horizon geometry while leaving the Maxwell and Schwarzschild limits intact as $a\to0$ and $Q\to0$ (Magos et al., 2020, Abbas et al., 2023).

1. Einstein–Euler–Heisenberg construction

A standard formulation begins from the four-dimensional action

$$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$

with

$$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$

Here $a=\tfrac{8\alpha^2}{45m_e^4}$ is the leading Euler–Heisenberg coupling, and for $a\to0$ the theory reduces to Maxwell electrodynamics. Varying the action gives

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$

with $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$ (Magos et al., 2020).

For a static, spherically symmetric electric configuration, the line element is taken as

$$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$

and in the $a\to0$0-framework one has $a\to0$1. The resulting physical electric field and electrostatic potential are

$a\to0$2

while in the notation used for charged-particle dynamics the gauge potential appears as

$a\to0$3

The literature uses several symbols for the nonlinear coupling, notably $a\to0$4, $a\to0$5, and $a\to0$6, but the leading correction always enters as a quartic electromagnetic self-interaction (Magos et al., 2020, Chen et al., 2022).

2. Metric, electromagnetic sector, and horizon structure

The defining metric function of the electrically charged Einstein–Euler–Heisenberg–AdS black hole is

$a\to0$7

with $a\to0$8. In this form the geometry interpolates continuously between Schwarzschild, Reissner–Nordström, and Reissner–Nordström–AdS limits. The Euler–Heisenberg term is the leading nonlinear correction and is subleading at large $a\to0$9 but important in the strong-field region (Abbas et al., 2023).

Horizons are determined by the roots of

$Q\to0$0

This equation can have up to three real positive roots, interpreted in the accretion analysis as inner, middle, and outer horizons, depending on $Q\to0$1. For the numerical example $Q\to0$2, $Q\to0$3, and $Q\to0$4 so that $Q\to0$5, the $Q\to0$6 Reissner–Nordström–AdS case has two horizons at approximately $Q\to0$7, whereas for $Q\to0$8 three horizons appear or degenerate into one depending on $Q\to0$9. As $$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$0 or $$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$1, the familiar Reissner–Nordström or Schwarzschild patterns are recovered (Abbas et al., 2023).

A recurrent conclusion in the accretion-disk analysis is that the one-loop QED correction term $$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$2 slightly modifies the Reissner–Nordström–AdS geometry and pushes characteristic radii outward by $$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$3. This outward shift is numerically modest, but it is systematic in the examples discussed for horizons and the innermost stable circular orbit (Abbas et al., 2023).

3. Geodesics, circular motion, accretion disks, and radiative signatures

For geodesic motion in the equatorial plane, the particle Lagrangian is

$$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$4

with conserved energy and angular momentum

$$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$5

The radial motion takes the form

$$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$6

For circular timelike orbits, the standard conditions $$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$7 and $$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$8 give

$$S=\frac{1}{4\pi}\int d^4x\,\sqrt{-g}\;\Big[\tfrac14\,(R-2\Lambda)-\mathcal L(F,G)\Big],$$9

In the Euler–Heisenberg–AdS background,

$$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$0

These expressions reduce to the Reissner–Nordström–AdS formulas as $$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$1, and further to Schwarzschild–AdS as $$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$2 (Abbas et al., 2023).

The innermost stable circular orbit is determined by

$$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$3

Because the analytic solution is not available in closed form for $$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$4, the ISCO is found numerically. For $$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$5, $$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$6, and $$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$7, the reported values are

$$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$8

so increasing $$\mathcal L(F,G)=-F+\tfrac a2\,F^2+\tfrac{7a}{8}\,G^2, \qquad F=\tfrac14F_{\mu\nu}F^{\mu\nu}, \qquad G=\tfrac14F_{\mu\nu}\,{}^*\!F^{\mu\nu}.$$9 slightly pushes the ISCO outward. The same study reports that decreasing $a=\tfrac{8\alpha^2}{45m_e^4}$0 moves the ISCO to larger $a=\tfrac{8\alpha^2}{45m_e^4}$1, while increasing $a=\tfrac{8\alpha^2}{45m_e^4}$2 pulls it inward (Abbas et al., 2023).

Thin-disk observables are computed in the Novikov–Thorne (Page–Thorne) framework. The time-averaged radial flux is

$a=\tfrac{8\alpha^2}{45m_e^4}$3

The reported flux profiles show that larger $a=\tfrac{8\alpha^2}{45m_e^4}$4 reduces the peak of $a=\tfrac{8\alpha^2}{45m_e^4}$5 and shifts it slightly outward. The specific energy develops a shallower minimum, the specific angular momentum grows a bit more slowly at small $a=\tfrac{8\alpha^2}{45m_e^4}$6, and the specific angular velocity is reduced by a positive $a=\tfrac{8\alpha^2}{45m_e^4}$7 but increased by a more negative $a=\tfrac{8\alpha^2}{45m_e^4}$8. The accretion efficiency $a=\tfrac{8\alpha^2}{45m_e^4}$9 is also lowered, with typical efficiencies dropping by $a\to0$0 when $a\to0$1 is turned on (Abbas et al., 2023).

The same analysis connects disk dynamics to gamma-ray bursts through

$a\to0$2

Using the observational estimate $a\to0$3 for an object of order $a\to0$4, it is argued that the few-percent Euler–Heisenberg shift in $a\to0$5 relative to Reissner–Nordström–AdS could in principle be used to constrain the nonlinear coupling (Abbas et al., 2023).

For charged probes, the effective potential becomes

$a\to0$6

which is the starting point for the Lyapunov-exponent analysis of orbit instability and chaos-bound violation regions around the same background (Chen et al., 2022).

4. Extended thermodynamics, criticality, and microstructure

In the extended phase-space treatment, the cosmological constant is interpreted as pressure,

$a\to0$7

with horizon entropy

$a\to0$8

Hawking temperature

$a\to0$9

and Euler–Heisenberg conjugate

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$0

The first law and Smarr relation are

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$1

In terms of the specific volume $$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$2, the equation of state is

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$3

and the critical point is determined by

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$4

which yields the cubic

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$5

For small $$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$6, the Maxwell-branch solution is

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$7

with

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$8

The mean-field critical exponents remain

$$\nabla_\mu P^{\mu\nu}=0, \qquad G_{\mu\nu}+\Lambda g_{\mu\nu}=8\pi T_{\mu\nu},$$9

(Magos et al., 2020).

The phase structure depends strongly on the sign and magnitude of the QED parameter. For $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$0, there is one critical point and a Van der Waals-like small/large black-hole transition. For $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$1, there are two critical points and a reentrant large/small/large black-hole transition. For $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$2, the first-order phase transition disappears. In the Gibbs free energy, the small/large transition is signaled by a swallowtail, while the reentrant regime contains a zeroth-order large-to-small jump followed by a first-order small-to-large transition (Ye et al., 2022).

The thermodynamic interpretation at fixed charge is consistent with the temperature minimum and heat-capacity sign change. The lower branch of small black holes has negative heat capacity, whereas the upper large-black-hole branch has positive heat capacity. The nonlinear correction modifies the thermodynamic quantities quantitatively but does not fundamentally and thoroughly change the phase structure for allowed values of the nonlinear parameter (Zhao et al., 19 Jan 2025).

From the Ruppeiner-geometry perspective, the normalized scalar curvature exhibits regions with $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$3, interpreted as attractive micro-interaction, and for small positive $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$4 an additional region with $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$5, interpreted as dominated repulsive interaction among black-hole microstructure. Near criticality the universal behavior

$$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$6

matches the mean-field universality class (Ye et al., 2022).

A distinct development is the inclusion of higher-order QED corrections. In that case the corrected metric acquires an additional term

$$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$7

the criticality condition has a single positive root for $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$8, and the corrected Euler–Heisenberg–AdS black hole has only one stable phase transition branch, again with Van der Waals exponents. The same correction removes the second branch and reentrant behavior present at $$P^{\mu\nu}=-\,\partial\mathcal L/\partial F_{\mu\nu}$$9 in the phase diagrams discussed there (Li et al., 2021).

5. Thermodynamic topology, chaos bounds, and consistency conditions

Thermodynamic topology classifies Euler–Heisenberg–AdS black holes as defects in thermodynamic space using generalized off-shell free energies and winding numbers. In the canonical ensemble, the topological class depends on the sign of $$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$0 for the uncorrected four-dimensional solution, whereas in the grand canonical ensemble it does not. With higher-order QED correction, the canonical distinction between positive and negative $$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$1 disappears (Gogoi et al., 2023).

System and ensemble	Parameter regime	Total $$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$2
Canonical EHAdS	$$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$3	$$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$4
Canonical EHAdS	$$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$5	$$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$6
Canonical higher-order QED-corrected EHAdS	any $$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$7	$$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$8
Grand-canonical EHAdS or higher-order corrected EHAdS	any $$ds^2=-f(r)\,dt^2+f(r)^{-1}dr^2+r^2(d\theta^2+\sin^2\theta\,d\phi^2),$$9	$a\to0$00

The probe-dynamical analysis of charged particles adds a separate instability diagnostic. The principal Lyapunov exponent $a\to0$01 for circular orbits satisfies $a\to0$02 at the horizon and is compared with the conjectured chaos bound

$a\to0$03

Numerical solutions exhibit regions where $a\to0$04. For example, with $a\to0$05, $a\to0$06, $a\to0$07, $a\to0$08, and $a\to0$09, the paper reports

$a\to0$10

Compared with the Maxwell case, the Euler–Heisenberg term shifts circular orbits outward for fixed $a\to0$11, enlarges the range of $a\to0$12 and $a\to0$13 for which $a\to0$14, and lowers the minimum $a\to0$15 needed at fixed $a\to0$16 to trigger a violation (Chen et al., 2022).

The same class of black holes has also been used to study the simultaneous realization of the Weak Gravity Conjecture and the Weak Cosmic Censorship Conjecture. For

$a\to0$17

the extremal ratio obeys

$a\to0$18

with $a\to0$19 as $a\to0$20. Numerical scanning shows that for $a\to0$21 there is an overlapping parameter window where the black hole still has horizons and the WGC inequality can be met simultaneously. In the same regime, the photon sphere persists outside the horizon and remains unstable, with topological charge $a\to0$22 in the photon-sphere topology approach (Alipour et al., 4 Apr 2025).

6. Hairy, dilatonic, matter-coupled, and higher-curvature generalizations

The Einstein–Euler–Heisenberg–AdS sector admits several exact or quasi-exact extensions. A magnetically charged family with scalar hair is obtained by coupling a self-interacting scalar field minimally to gravity. In that case,

$a\to0$23

and the scalar charge induces

$a\to0$24

The hairy black hole can develop up to three horizons when the Euler–Heisenberg parameter and magnetic charge are small, the event horizon increases with scalar charge $a\to0$25, the Hawking temperature is raised by the Euler–Heisenberg term for very small holes, and the heat capacity diverges at a local minimum of $a\to0$26, signaling a second-order phase transition from an unstable small hole to a stable large AdS hole. In this AdS hairy branch, the null energy condition is satisfied outside the horizon, but the weak energy condition is violated because the negative scalar potential dominates (Karakasis et al., 2022).

A string-inspired Euler–Heisenberg–dilaton construction yields an exact magnetically charged AdS generalization with

$a\to0$27

$a\to0$28

In this model the entropy is $a\to0$29, the AdS asymptotics are generated by a dilaton potential, and the effective stress tensor satisfies the weak, strong, and dominant energy conditions numerically for $a\to0$30 outside the horizon. The same analysis identifies the solution as an explicit example of bypassing modern versions of the no-hair theorem without violating energy conditions (Bakopoulos et al., 2024).

Matter-coupled variants incorporate a cloud of strings and perfect fluid dark matter. The corresponding lapse function is

$a\to0$31

In that background, the photon sphere, shadow radius, ISCO, and QPO frequencies acquire explicit $a\to0$32-, $a\to0$33-, and $a\to0$34-dependent corrections. The heat capacity shows divergences where $a\to0$35, and the Gibbs free energy remains Van der Waals-like but with critical coordinates shifted by the Euler–Heisenberg coupling (Ahmed et al., 12 Sep 2025).

A higher-curvature extension in four-dimensional Einstein–Gauss–Bonnet gravity gives the minus-branch solution

$a\to0$36

In the extended phase space, increasing the Gauss–Bonnet coupling raises the critical radius and lowers the critical temperature and pressure, while increasing the Euler–Heisenberg parameter has the opposite but milder effect. In geodesic dynamics, the Gauss–Bonnet correction pushes the ISCO outward, whereas the Euler–Heisenberg term pulls it inward in that model (Hamil et al., 21 Feb 2026).

Within AdS/CFT, the Euler–Heisenberg deformation has also been probed by wave optics. For the AdS Reissner–Nordström solution corrected by the term $a\to0$37, the holographic Einstein ring reconstructed from the boundary response has a radius that is essentially unchanged at leading order in $a\to0$38, because the correction falls off rapidly near the boundary and remains subleading in the wave dynamics. The same study finds that the ring radius decreases with source position $a\to0$39, wave frequency $a\to0$40, and chemical potential $a\to0$41, while it increases with electric charge $a\to0$42 and temperature $a\to0$43 (Baruah et al., 19 May 2025).