betaFeedback

WorksPhysics and Astronomy

Shadows and Images of Black Holes with Purely Nonlinear Scalarization

Abstract

We investigate the shadows and radiative images of static, spherically symmetric black holes with purely nonlinear scalarization in Einstein–scalar–Gauss–Bonnet gravity. The quartic exponential coupling supports scalarized backgrounds without a linear tachyonic instability of Schwarzschild black holes. We compare mass and coupling dependences and construct images on radially stable backgrounds using optically thin spherical emission and face-on thin disks. On the sampled radially stable branches, we find smaller shadow radii than those of Schwarzschild black holes at the same mass, with reductions reaching approximately 21.7%21.7\%. The sampled radially unstable branches show much smaller shadow-size deviations. The photon-orbit index, which measures radial instability per unit orbital angle, varies nonmonotonically even when the normalized shadow radius changes monotonically. The index also controls the asymptotic compression of higher-order images, providing information complementary to shadow size. For spherical emission, radial infall produces a darker center relative to the bright ring than static emission. For circular disks, we find that smooth metric variations can create an unstable interval separating two stable regions of timelike circular motion. When the disk occupies the stable region connected to infinity, its inner edge moves outward across this transition, suppressing emission inside the geometric shadow boundary while the normalized shadow radius changes only slightly. Our results show that purely nonlinear scalarization can alter images through both photon propagation and the orbital structure of emitting matter, with the visible features depending on the emission model.

Introduction

Event-horizon-scale images of M87* and Sagittarius A* obtained by the Event Horizon Telescope provide [1, 2, 3, 4, 5] a means of studying photon propagation and emitting matter in the strong gravitational fields of supermassive black holes. At the geometric level, analytical descriptions of black-hole shadows relate [6] their boundaries to photon capture and the critical null geodesics of the spacetime. In an observed image, however, the bright ring and central brightness depression reflect gravitational lensing and photon capture together with the distribution and motion of the emitting plasma. The geometric shadow boundary, the brightness maximum, and the edge of a dark central region therefore need not coincide [7]. Accordingly, using black-hole images to probe gravity requires understanding how spacetime geometry shapes these features.

A relevant theoretical setting is Einstein–scalar–Gauss–Bonnet gravity, which allows [8] black holes to acquire a nontrivial scalar field through a coupling between the scalar and spacetime curvature. For suitable coupling functions, scalar-free solutions remain solutions of the theory, while sufficiently strong curvature can trigger a tachyonic instability of the scalar perturbation. The resulting spontaneous scalarization gives rise [9, 10, 11] to black hole branches with nonzero scalar hair. However, the absence of a linear tachyonic instability does not exclude scalarization. In particular, for couplings whose leading dependence near the scalar-free state is quartic, the curvature-induced mass term vanishes in the linear scalar equation, but nonlinear interactions can support [12] nontrivial scalar configurations. Indeed, time evolutions in the decoupling limit show that sufficiently large initial scalar perturbations can settle into nonzero stationary profiles, while smaller perturbations decay. Beyond the decoupling limit, the corresponding fully coupled equilibrium solutions can have several scalarized branches at the same mass. Radial perturbation analyses further identify [13] stable regions of these branches as well as unstable and nonhyperbolic regions.

The optical consequences of scalarization have also been studied, with appreciable changes found [14, 15, 16] in black-hole shadows and circular-orbit properties. For example, studies of spontaneously scalarized black holes and their rotating extensions have shown [14] that, even without rotation, scalarized black holes can have shadow sizes that differ substantially from those of Schwarzschild black holes with the same mass. In the case of spin-induced scalarization, circular-orbit properties have been found [15] to differ appreciably from those of Kerr black holes even when the shadow size deviations remain small. Beyond these geometric properties, images of spontaneously scalarized black holes illuminated by thin disks and static spherical emission demonstrate [16] that the optical appearance depends on both the coupling and the emission prescription. These results therefore motivate extending the comparison to purely nonlinear scalarized branches and following their optical properties across mass, coupling, and radial stability.

More generally, black hole images can be described by three complementary properties: the shadow size, the structure of photon rings, and the brightness distribution. First, the shadow size measures the extent of the geometric capture region. Specifically, in the static, spherically symmetric setting, the critical impact parameter bcb_c sets [6] the capture boundary and the asymptotic geometric shadow radius. Second, photon-ring structure is characterized by two complementary quantities: the photon-ring width and the photon-orbit index. The photon-ring width specifies [7] the interval of impact parameters occupied by rays with large accumulated orbital angles, and hence the radial extent of the corresponding region in the image plane. Alongside this width, the photon-orbit index χ\chi measures [17, 18] the exponential rate at which nearby rays separate radially from the unstable circular photon orbit per unit orbital angle. The index also controls [18, 19] the asymptotic concentration of successive higher-order images toward the critical curve. Finally, the brightness distribution gives the observed intensity at each position in the image and depends on both photon propagation and the spatial distribution and motion of the emitting matter.

Predicting the brightness distribution therefore requires an emission model. For example, optically thin spherical emission provides [20] a simple setting for relating the critical curve to a central brightness depression, while thin-disk images illustrate [7] the importance of the emitting region and its multiple images. However, studies of spherical images have shown [21] that degeneracies between geometry and emission can persist, motivating the joint use of several image properties under controlled emission assumptions. In thin-disk models with circularly orbiting emitters, the spacetime also determines the allowed stable orbits and hence informs the choice of disk inner edge. In particular, more general black-hole geometries can contain [22] disconnected regions of stable timelike circular motion, so the innermost stable orbit and the inner boundary of the stable region accessible from large radius need not coincide. It is therefore important to assess how different emission models affect the brightness distribution and overall appearance of black-hole images.

Motivated by these considerations, we investigate the shadows and radiative images of static, spherically symmetric black holes with purely nonlinear scalarization in scalar–Gauss–Bonnet gravity. Using Schwarzschild black holes as a reference, we examine the mass and coupling dependence of the shadow radius, photon-orbit index, and photon-ring width along the scalarized branches. Building on this geometric comparison, we then study spherical emission and thin-disk images to determine how emitter motion and the disk inner edge affect the brightness distribution. Together, the geometric and emission analyses connect the image properties to photon dynamics and the orbital structure of the emitting matter.

The paper is organized as follows. In § 2, we introduce the theory, scalarized backgrounds, radial stability, and comparison strategy. In § 3, we then present the shadow size, photon-orbit instability, and photon-ring structure together with their mass and coupling dependence. In § 4, we connect these geometric results to spherical emission and thin-disk images, including the role of circular-orbit stability in selecting the disk inner edge. Finally, in § 5, we summarize the results and discuss future directions. Throughout the paper, we use geometrized units, G=c=1G=c=1, and the metric signature (−,+,+,+)(-,+,+,+).

Theoretical Model and Black-Hole Backgrounds

We consider four-dimensional Einstein–scalar–Gauss–Bonnet gravity with a massless real scalar field and no scalar potential. The action of the theory is

S=116π∫d4x−g[R−2∇μϕ∇μϕ+λ2f(ϕ)G],(1)S = \frac{1}{16\pi}\int d^4x \sqrt{-g}\left[R - 2\nabla_\mu\phi\nabla^\mu\phi+ \lambda^2 f(\phi)\mathcal{G}\right], \tag*{(1)}

where g=det⁡(gμν)g=\det(g_{\mu\nu}), RR is the Ricci scalar, ϕ\phi is a dimensionless real scalar field, and λ\lambda is a positive coupling constant with dimensions of length. The dimensionless function f(ϕ)f(\phi) couples the scalar to the curvature through the Gauss–Bonnet invariant,

G=R2−4RμνRμν+RμνρσRμνρσ(2)\mathcal{G} = R^2 - 4R_{\mu\nu}R^{\mu\nu} + R_{\mu\nu\rho\sigma}R^{\mu\nu\rho\sigma} \tag*{(2)}

where RμνR_{\mu\nu} and RμνρσR_{\mu\nu\rho\sigma} are the Ricci and Riemann tensors, respectively. In this work, we adopt [12] the quartic exponential coupling used to study nonlinear black-hole scalarization,

f(ϕ)=1−e−κϕ44κ,κ>0,(3)f(\phi) = \frac{1-e^{-\kappa\phi^4}}{4\kappa}, \qquad\kappa> 0, \tag*{(3)}

Here the dimensionless parameter κ\kappa controls the shape of the coupling function. There is no background matter or electromagnetic charge.

Varying Eq. (1) with respect to the inverse metric gives the modified Einstein equations

Gμν+λ2Hμν=Tμνϕ,(4)G_{\mu\nu}+\lambda^2 H_{\mu\nu}=T^\phi_{\mu\nu}, \tag*{(4)}

where Gμν=Rμν−12gμνRG_{\mu\nu}=R_{\mu\nu}-\frac{1}{2}g_{\mu\nu}R is the Einstein tensor. The tensor HμνH_{\mu\nu} represents the contribution from the scalar–Gauss–Bonnet coupling, while TμνϕT^\phi_{\mu\nu} is the scalar kinetic source in the normalization of Eq. (1):

Hμν=−R(∇μΨν+∇νΨμ)−4∇ρΨρGμν+4Rμρ∇ρΨν+4Rνρ∇ρΨμ−4gμνRρσ∇ρΨσ+4Rσμρν∇ρΨσ,Tμνϕ=2∇μϕ∇νϕ−gμν∇ρϕ∇ρϕ.(5)\begin{aligned} H_{\mu\nu}={}&-R\left(\nabla_\mu\Psi_\nu+\nabla_\nu\Psi_\mu\right)-4\nabla^\rho\Psi_\rho G_{\mu\nu}\\ &+4R_{\mu\rho}\nabla^\rho\Psi_\nu+4R_{\nu\rho}\nabla^\rho\Psi_\mu\\ &-4g_{\mu\nu}R^{\rho\sigma}\nabla_\rho\Psi_\sigma+4R^\sigma{}_{\mu\rho\nu}\nabla^\rho\Psi_\sigma,\\ T^\phi_{\mu\nu}={}&2\nabla_\mu\phi\nabla_\nu\phi-g_{\mu\nu}\nabla_\rho\phi\nabla^\rho\phi. \tag*{(5)} \end{aligned}

To express the coupling contribution compactly, we write f,ϕ=df/dϕf_{,\phi}=df/d\phi and introduce its space-time gradient,

Ψμ=f,ϕ(ϕ)∇μϕ=∇μf(ϕ).(6)\Psi_\mu=f_{,\phi}(\phi)\nabla_\mu\phi=\nabla_\mu f(\phi). \tag*{(6)}

Variation of Eq. (1) with respect to the scalar field gives

□ϕ+λ24f,ϕ(ϕ)G=0,□=∇μ∇μ.(7)\Box\phi+\frac{\lambda^2}{4}f_{,\phi}(\phi)\mathcal{G}=0,\qquad\Box=\nabla_\mu\nabla^\mu. \tag*{(7)}

For the quartic exponential coupling in Eq. (3), the first two scalar derivatives are

f,ϕ(ϕ)=ϕ3e−κϕ4,f_{,\phi}(\phi)=\phi^3e^{-\kappa\phi^4},
f,ϕϕ(ϕ)=(3ϕ2−4κϕ6)e−κϕ4,(8)f_{,\phi\phi}(\phi)=\left(3\phi^2-4\kappa\phi^6\right)e^{-\kappa\phi^4}, \tag*{(8)}

so that f(0)=f,ϕ(0)=f,ϕϕ(0)=0f(0)=f_{,\phi}(0)=f_{,\phi\phi}(0)=0. Scalar-free Schwarzschild black holes therefore remain exact solutions. The linear scalar perturbation δϕ\delta\phi about ϕ=0\phi=0 satisfies

□Sch δϕ=0,(9)\Box_{\mathrm{Sch}}\,\delta\phi=0, \tag*{(9)}

Here □Sch\Box_{\mathrm{Sch}} is the wave operator on the Schwarzschild background, and no curvature-induced tachyonic mass term appears. In conventional curvature-induced spontaneous scalarization, a coupling with f,ϕϕ(0)>0f_{,\phi\phi}(0)>0 instead gives [9, 10] the linear scalar perturbation an effective mass squared meff2=−λ2f,ϕϕ(0)GSch/4m_{\mathrm{eff}}^2=-\lambda^2 f_{,\phi\phi}(0)\mathcal{G}_{\mathrm{Sch}}/4. In a sufficiently strong curvature regime, this term can produce a tachyonic instability, allowing arbitrarily small perturbations with a component along the unstable mode to grow. Scalarized branches then bifurcate [9] from the family of Schwarzschild black holes at the onset of this instability.

However, the absence of this linear mechanism does not exclude scalarization. For the quartic coupling considered here, the scalar–curvature source becomes relevant at finite amplitude. For suitable black hole masses and coupling parameters, evolutions in the decoupling limit show [12] that small perturbations decay, whereas initial scalar perturbations exceeding a finite threshold can settle into a nonzero stationary profile, leaving a long-lived scalar configuration outside the horizon. This is the purely nonlinear scalarization mechanism relevant to the present model. The horizon value ϕH≡ϕ(rH)\phi_H \equiv\phi(r_H), where rHr_H is the horizon’s areal radius, helps distinguish how the resulting branches connect to scalar-free solutions. For conventional spontaneous scalarization, ϕH\phi_H tends continuously to zero where the scalarized branch meets Schwarzschild black holes. In the purely nonlinear model, the radially stable scalarized branch does not emerge [12, 13] from Schwarzschild black holes through such a linear onset at finite mass, but reaching it entails a finite change in the scalar amplitude.

To investigate the optical properties of these scalarized solutions, we restrict attention to static, spherically symmetric, and asymptotically flat backgrounds. Using the areal radius rr, for which a sphere has area 4πr24\pi r^2, we write

ds2=−A(r) dt2+dr2B(r)+r2(dθ2+sin⁡2θ dϕ2),ϕ=ϕ(r),(10)ds^2 = -A(r)\,dt^2 + \frac{dr^2}{B(r)} + r^2\left(d\theta^2 + \sin^2\theta\,d\phi^2\right), \qquad\phi= \phi(r), \tag*{(10)}

Here tt is the static time coordinate, and (θ,ϕ)(\theta,\phi) are the usual angular coordinates. The metric functions A=e2ΦA=e^{2\Phi} and B=e−2ΛB=e^{-2\Lambda} specify the time and radial components, with Φ\Phi and Λ\Lambda their logarithmic parametrizations. The functions AA and BB are determined independently and are positive in the static exterior. In general, A≠BA \ne B. The field equations are solved with the full scalar backreaction on the metric. In particular, Eq. (7) reduces to

ϕ′′+(2r+A′2A+B′2B)ϕ′+λ24Bf,ϕ(ϕ)G=0,(11)\phi^{\prime\prime}+\left(\frac{2}{r}+\frac{A^{\prime}}{2A}+\frac{B^{\prime}}{2B}\right)\phi^{\prime}+\frac{\lambda^2}{4B}f_{,\phi}(\phi)\mathcal{G}=0, \tag*{(11)}

where a prime denotes differentiation with respect to rr. This equation is integrated together with the independent components of the Einstein equations (4), evaluated for the ansatz (10).

For a regular nonextremal horizon at r=rHr=r_H, the leading exterior expansions have the form

A(r)=aH(r−rH)+O((r−rH)2),B(r)=bH(r−rH)+O((r−rH)2),ϕ(r)=ϕH+ϕH′(r−rH)+O((r−rH)2),(12)\begin{aligned} A(r) &= a_H(r-r_H)+O\left((r-r_H)^2\right),\\ B(r) &= b_H(r-r_H)+O\left((r-r_H)^2\right),\\ \phi(r) &= \phi_H+\phi_H^{\prime}(r-r_H)+O\left((r-r_H)^2\right), \tag*{(12)} \end{aligned}

Here aH=A′(rH)a_H=A^{\prime}(r_H) and bH=B′(rH)b_H=B^{\prime}(r_H) are positive horizon slopes, while ϕH\phi_H is finite. The horizon derivative ϕH′\phi_H^{\prime} is not an independent parameter and must be compatible with a regular solution of the coupled field equations. Substituting Eq. (12) into Eqs. (4) and (11), and eliminating the metric expansion coefficients, gives the following constraint

2λ2rH2f,ϕ(ϕH)(ϕH′)2+rH3ϕH′+3λ2f,ϕ(ϕH)=0.(13)2\lambda^2 r_H^2 f_{,\phi}(\phi_H)(\phi_H^{\prime})^2+r_H^3\phi_H^{\prime}+3\lambda^2 f_{,\phi}(\phi_H)=0. \tag*{(13)}

For f,ϕ(ϕH)≠0f_{,\phi}(\phi_H)\ne0, solving Eq. (13) yields [9] the two algebraic roots

ϕH,±′=rH4λ2f,ϕ(ϕH)(−1±ΔH),(14)\phi_{H,\pm}^{\prime}=\frac{r_H}{4\lambda^2 f_{,\phi}(\phi_H)}\left(-1\pm\sqrt{\Delta_H}\right), \tag*{(14)}

where the dimensionless discriminant of the horizon constraint is

ΔH=1−24λ4[f,ϕ(ϕH)]2rH4.(15)\Delta_H=1-\frac{24\lambda^4[f_{,\phi}(\phi_H)]^2}{r_H^4}. \tag*{(15)}

Here, we select the plus sign, for which ϕH,+′→0\phi_{H,+}^{\prime}\to0 as f,ϕ(ϕH)→0f_{,\phi}(\phi_H)\to0 at fixed rHr_H and λ\lambda, and the other root diverges in this limit. The selected root can equivalently be written as

ϕH′=−6λ2f,ϕ(ϕH)rH3(1+ΔH).(16)\phi_H^{\prime}=-\frac{6\lambda^2 f_{,\phi}(\phi_H)}{r_H^3(1+\sqrt{\Delta_H})}. \tag*{(16)}

We restrict the backgrounds to horizon satisfying

ΔH>0⟺rH4>24λ4[f,ϕ(ϕH)]2.(17)\Delta_H > 0 \Longleftrightarrow r_H^4 > 24\lambda^4 [f_{,\phi}(\phi_H)]^2. \tag*{(17)}

The limiting case ΔH=0\Delta_H = 0, at which the two local roots coalesce, is excluded because the quadratic constraint alone does not establish regularity of the higher-order horizon expansion there. The radial metric slope bH=B′(rH)b_H = B'(r_H) follows from the rrrr component of the Einstein equations (4) evaluated at the horizon

bH=1rH+2λ2f,ϕ(ϕH)ϕH′.(18)b_H = \frac{1}{r_H + 2\lambda^2 f_{,\phi}(\phi_H)\phi'_H}. \tag*{(18)}

The normalization of aHa_H is fixed by imposing A(∞)=1A(\infty) = 1. These local conditions supply horizon data, but do not by themselves establish a globally regular or dynamically admissible exterior.

At spatial infinity we impose

A(r)⟶1,B(r)⟶1,ϕ(r)⟶0.(19)A(r) \longrightarrow1,\qquad B(r) \longrightarrow1,\qquad\phi(r) \longrightarrow0. \tag*{(19)}

The leading expansions then define the Arnowitt–Deser–Misner (ADM) mass MM and scalar charge DD:

A(r)=1−2Mr+O(r−2),B(r)=1−2Mr+O(r−2),ϕ(r)=Dr+O(r−2).(20)\begin{aligned} A(r) &= 1 - \frac{2M}{r} + O(r^{-2}),\\ B(r) &= 1 - \frac{2M}{r} + O(r^{-2}),\\ \phi(r) &= \frac{D}{r} + O(r^{-2}). \tag*{(20)} \end{aligned}

Both MM and DD have dimensions of length in the adopted units, so M/λM/\lambda and D/λD/\lambda are their dimensionless values. The scalar boundary value is fixed rather than treated as an additional parameter.

Together, the horizon and asymptotic conditions define the boundary-value problem for the scalarized backgrounds. We adjust the horizon data to satisfy ϕ(∞)=0\phi(\infty) = 0 and follow the resulting solution branches, extracting MM and DD from Eq. (20).

Since the scalar-free solution remains available in the same theory, we use a Schwarzschild black hole with the same ADM mass as the reference geometry,

ASch(r)=BSch(r)=1−2Mr,ϕSch=0,rH,Sch=2M.(21)A_{\mathrm{Sch}}(r) = B_{\mathrm{Sch}}(r) = 1 - \frac{2M}{r},\qquad\phi_{\mathrm{Sch}} = 0,\qquad r_{H,\mathrm{Sch}} = 2M. \tag*{(21)}

The scalarized backgrounds need not satisfy rH=2Mr_H = 2M. Consequently, pointwise comparisons of static exterior fields are restricted to r>max⁡(rH,2M)r > \max(r_H, 2M). Matching the horizon radius would not, in general, match the ADM mass.

In addition to the scalar-free solution, different scalarized configurations can coexist [12] at the same mass. To identify these solution families and their stability properties, we present the scalar-charge–mass diagrams in Figure 1 at fixed coupling scale λ\lambda. The left and right panels show the representative couplings κ=100\kappa= 100 and 1000, respectively, with line styles distinguishing radial stability and loss of hyperbolicity. The axes are the dimensionless ADM mass M/λM/\lambda and scalar charge D/λD/\lambda, with DD defined by Eq. (20). Only positive-scalar solutions are needed because the symmetry ϕ↦−ϕ\phi\mapsto-\phi leaves the metric unchanged.

Figure 1. The dimensionless scalar charge D/λD/\lambda is shown as a function of the dimensionless ADM mass M/λM/\lambda for scalarized solutions at κ=100\kappa= 100 (left) and κ=1000\kappa= 1000 (right). Blue solid curves denote radially stable configurations, red dashed curves denote radially unstable configurations, and orange dotted curves denote radial loss of hyperbolicity. The black dash-dotted line represents Schwarzschild black holes.

Regularity of a static solution does not by itself establish its dynamical stability. We therefore classify the branches in Figure 1 using radial perturbations of the corresponding backgrounds. Eliminating the metric perturbations in the spherically symmetric sector gives [13] a scalar perturbation equation of the form

K(r)∂t2δϕ−∂r2δϕ+C1(r)∂rδϕ+U(r)δϕ=0,(22)\mathcal{K}(r)\partial_t^2\delta\phi-\partial_r^2\delta\phi+C_1(r)\partial_r\delta\phi+U(r)\delta\phi=0, \tag*{(22)}

Here δϕ\delta\phi is the linear scalar perturbation, K\mathcal{K} multiplies its second time derivative, C1C_1 multiplies its first radial derivative, and UU is the effective radial perturbation potential. All three coefficients are fixed by the static background. The highest-derivative terms describe a wave equation with real radial characteristic speeds when K>0\mathcal{K} > 0. A zero or sign change of K\mathcal{K} therefore signals a loss of radial hyperbolicity, meaning that this wave-propagation description breaks down. On a hyperbolic background, a mode δϕ=eσturad(r)\delta\phi= e^{\sigma t}u_{\mathrm{rad}}(r) with growth rate σ>0\sigma> 0 and radial profile uradu_{\mathrm{rad}} satisfying the black-hole boundary conditions represents a radial instability. We refer to configurations as radially stable when the available finite-domain analysis finds a positive K\mathcal{K} and no growing mode, and as radially unstable when a growing mode is detected on a hyperbolic background. This classification is restricted to the available linear radial analysis and does not establish stability against arbitrary perturbations.

For κ=1000\kappa= 1000 in the right panel, we find two scalar-charge branches at the same black-hole mass, with the displayed sequence reaching a maximum mass of M/λ≃0.06463M/\lambda\simeq0.06463. The lower-charge branch is radially unstable, whereas the upper-charge branch contains a radially stable segment. Following the upper-charge branch toward smaller masses leads to the orange nonhyperbolic region. The branch structure is more involved at κ=100\kappa= 100, shown in the left panel. A separate low-charge radially unstable branch extends to M/λ≃0.05688M/\lambda\simeq0.05688. The upper-charge branch contains a radially stable segment and turns near M/λ≃0.1214M/\lambda\simeq0.1214, where a short radially unstable continuation is also present. The low-mass end of the upper-charge branch is nonhyperbolic.

The coexistence of branches with different scalar charges and radial stability motivates three comparisons that separate mass, coupling, and branch effects on the optical properties. In all three comparisons, we keep λ\lambda fixed and characterize each scalarized background by κ\kappa, M/λM/\lambda, and its location on the solution family. Schwarzschild black holes with the same ADM mass provide the common reference through Eq. (21).

The first comparison follows the mass dependence along each branch at fixed κ\kappa. The analysis covers the radially stable segments over approximately

κ=100:0.05363≲M/λ≲0.12135,κ=1000:0.009219≲M/λ≲0.06457.(23)\begin{aligned} \kappa&= 100: && 0.05363 \lesssim M/\lambda\lesssim0.12135,\\ \kappa&= 1000: && 0.009219 \lesssim M/\lambda\lesssim0.06457. \tag*{(23)} \end{aligned}

We follow the radially unstable segments separately. Nonhyperbolic configurations are excluded from the optical comparisons.

The second comparison holds the ADM mass fixed while varying κ\kappa over the ten values 100, 200, …\ldots, 1000. With λ\lambda unchanged, this varies the coupling function at the same mass scale and tests how the scalarized geometry responds. The common interval of the ten radially stable segments is approximately 0.05363≲M/λ≲0.064570.05363 \lesssim M/\lambda\lesssim0.06457, with its lower and upper bounds set by the κ=100\kappa=100 and 1000 segments in Eq. (23), respectively.

The third comparison fixes both κ\kappa and M/λM/\lambda, and then contrasts radially stable and radially unstable configurations with Schwarzschild black holes. The masses M/λ=0.05365M/\lambda=0.05365 and 0.05550 allow this comparison at κ=100\kappa=100 and 1000. Comparing these coexisting backgrounds tests whether optical similarity to Schwarzschild black holes is accompanied by similar radial stability.

Shadow Size, Photon-Orbit Instability, and Photon-Ring Structure

We examine the geometric quantities that underlie the radiative images in § 4, presenting their theoretical definitions together with their behavior on the scalarized backgrounds. Starting from null geodesics, we determine the shadow size and compare its mass and coupling dependence. We then introduce the photon-orbit index and relate its variation to near-critical photon propagation, before examining the number of orbits and photon-ring width. The comparisons follow branches at fixed κ\kappa and use matched ADM masses across couplings, with λ\lambda held fixed.

Null geodesics and the optical potential

We treat light in the geometric-optics approximation, with photons minimally coupled to the metric (10) and with negligible backreaction. In the absence of a direct scalar–photon coupling, the scalar field affects photon propagation only through the background geometry. Spherical symmetry allows each geodesic to be placed in the equatorial plane, θ=π/2\theta=\pi/2. Let a dot denote differentiation with respect to an affine parameter η\eta. The time-translation and rotational symmetries give the conserved photon energy EE and angular momentum LL,

E=A(r)t˙,L=r2ϕ˙,b=LE.(24)E=A(r)\dot{t}, \qquad L=r^{2}\dot{\phi}, \qquad b=\frac{L}{E}. \tag*{(24)}

Their ratio defines the impact parameter bb, which has dimensions of length and is unchanged by a rescaling of the affine parameter. With A(∞)=1A(\infty)=1, bb specifies the transverse offset of an incoming ray at infinity. We denote the accumulated orbital angle along a ray by ψ\psi, with dψ=∣dϕ∣d\psi=\lvert d\phi\rvert in the chosen orbital plane and zero set at the starting point.

Substituting Eq. (24) into the null constraint −At˙2+r˙2/B+r2ϕ˙2=0-A\dot{t}^{2}+\dot{r}^{2}/B+r^{2}\dot{\phi}^{2}=0 gives the radial equation of motion,

r˙2=B(r)A(r)E2[1−b2A(r)r2].(25)\dot{r}^{2}=\frac{B(r)}{A(r)}E^{2}\left[1-b^{2}\frac{A(r)}{r^{2}}\right]. \tag*{(25)}

The factor A/r2A/r^{2} determines the allowed radial region and therefore motivates the optical potential,

Uopt(r)≡A(r)r2,A(r)B(r)r˙2+L2Uopt(r)=E2.(26)U_{\mathrm{opt}}(r)\equiv\frac{A(r)}{r^{2}}, \qquad\frac{A(r)}{B(r)}\dot{r}^{2}+L^{2}U_{\mathrm{opt}}(r)=E^{2}. \tag*{(26)}

Since A,B>0A,B>0 outside the horizon, a nonradial ray can access only regions where Uopt≤b−2U_{\mathrm{opt}}\le b^{-2}. Therefore, a simple root of Uopt(r)=b−2U_{\mathrm{opt}}(r)=b^{-2} is a radial turning point. Thus the potential determines where a ray can propagate, while the factor B/AB/A determines its radial motion along that allowed region.

A circular null orbit at r=r0>rHr=r_0>r_H requires both vanishing radial velocity and vanishing radial acceleration. Applying these conditions to the radial geodesic equation yields [17]

b2Uopt(r0)=1,Uopt(r0)=0′(27)b^2 U_{\mathrm{opt}}(r_0)=1,\qquad U'_{\mathrm{opt}(r_0)=0} \tag*{(27)}

where primes denote derivatives with respect to rr. In particular,

Uopt′(r)=rA′(r)−2A(r)r3⟹r0A′(r0)−2A(r0)=0.(28)U'_{\mathrm{opt}}(r)=\frac{rA'(r)-2A(r)}{r^3}\quad\Longrightarrow\quad r_0A'(r_0)-2A(r_0)=0. \tag*{(28)}

An exterior maximum, Uopt′′(r0)<0U''_{\mathrm{opt}}(r_0)<0, corresponds to an unstable circular photon orbit, whereas a minimum corresponds to a stable one. Rotating the orbital plane generates the associated photon sphere.

For the asymptotically flat backgrounds considered here, UoptU_{\mathrm{opt}} vanishes at the horizon and tends to zero at infinity. For a single nondegenerate exterior maximum, an incoming ray with b−2b^{-2} above the maximum crosses the potential barrier and is captured. If b−2b^{-2} lies below the maximum, the ray instead reaches an outer turning point and returns to infinity. If several maxima are present, the capture threshold is set by the largest barrier along the path from infinity to the horizon, rather than necessarily by the outermost circular orbit. We denote the radius of the unstable orbit controlling this threshold by rphr_{\mathrm{ph}}. Its relation to the apparent shadow size is developed in § 3.2.

Shadow size and its parameter dependence

The capture threshold identified in § 3.1 defines the geometric shadow boundary for a distant observer. The critical impact parameter bcb_c is fixed by the unstable orbit controlling capture. Therefore, the first condition in Eq. (27) gives

bc=1Uopt(rph)=rphA(rph),bc−2=max⁡r>rHUopt(r).(29)b_c=\frac{1}{\sqrt{U_{\mathrm{opt}}(r_{\mathrm{ph}})}}=\frac{r_{\mathrm{ph}}}{\sqrt{A(r_{\mathrm{ph}})}},\qquad b_c^{-2}=\max_{r>r_H}U_{\mathrm{opt}}(r). \tag*{(29)}

For a Schwarzschild black hole of the same mass, rph=3Mr_{\mathrm{ph}}=3M and A(rph)=1/3A(r_{\mathrm{ph}})=1/3, giving bcSch=33Mb_c^{\mathrm{Sch}}=3\sqrt{3}M. We therefore define the fractional shadow-size deviation as

δsh≡bc−bcSchbcSch=bc/M33−1.(30)\delta_{\mathrm{sh}}\equiv\frac{b_c-b_c^{\mathrm{Sch}}}{b_c^{\mathrm{Sch}}}=\frac{b_c/M}{3\sqrt{3}}-1. \tag*{(30)}

Positive and negative values indicate a larger and smaller geometric shadow, respectively, relative to a Schwarzschild black hole at the same ADM mass.

These definitions characterize the capture geometry. In an image produced by nearby emitting matter, both the brightness maximum and the extent of the central dark region depend [7] on the emission distribution, gravitational redshift, and radiative transfer, and need not coincide with the critical curve. In particular, rays traced toward the horizon can encounter emitting material before reaching it, so the region inside the geometric shadow need not be completely dark. We use bc/Mb_c/M and δsh\delta_{\mathrm{sh}} to compare the geometries, and examine their relation to the brightness profiles in § 4.

Mass dependence along scalarized branches at fixed κ\kappa

To determine how the shadow changes along each scalarized solution family, we compare the mass dependence at fixed κ=100\kappa=100 and 10001000 in Figure 2, keeping λ\lambda fixed. The upper and lower rows cover the respective mass ranges of the two couplings and distinguish radially stable and unstable backgrounds, with Schwarzschild black holes as the reference. The left column gives the normalized shadow radius bc/Mb_c/M, discussed here, while the right column gives the photon-orbit index χ\chi, whose definition and results follow in § 3.4.

Figure 2. The dimensionless shadow radius bc/Mb_c/M (left column) and photon-orbit index χ\chi (right column) are shown as functions of mass at fixed λ\lambda, for κ=100\kappa= 100 (top row) and κ=1000\kappa= 1000 (bottom row). Blue solid and red dashed curves denote radially stable and radially unstable backgrounds, respectively. Open circles identify the short unstable continuation near the high-mass turning point for κ=100\kappa= 100. Black dash-dotted lines give the values 333\sqrt{3} and 1 for Schwarzschild black holes. Each column uses a common vertical scale, while the rows cover their respective mass intervals. Curves connect the evaluated backgrounds in branch order, with distinct segments kept separate.

Along both radially stable branches, we find that bc/Mb_c/M increases with mass while remaining below 333\sqrt{3}. For κ=100\kappa= 100, the normalized radius increases from approximately 4.0705 at M/λ≃0.05363M/\lambda\simeq0.05363 to 5.1763 at M/λ≃0.12135M/\lambda\simeq0.12135, corresponding to a change from δsh≃−21.66%\delta_{\rm sh} \simeq-21.66\% to −0.381%-0.381\%. For κ=1000\kappa= 1000, the normalized radius increases from 4.1926 at M/λ≃0.009219M/\lambda\simeq0.009219 to 5.1867 at M/λ≃0.06457M/\lambda\simeq0.06457, with δsh\delta_{\rm sh} changing from approximately −19.31%-19.31\% to −0.181%-0.181\%. Thus, for either coupling, the shadow is smaller than that of a Schwarzschild black hole of the same mass, and the largest reduction occurs at the low-mass end of the stable segment. By comparison, the radially unstable segments in Figure 2 remain close to the reference values for Schwarzschild black holes in both bc/Mb_c/M and χ\chi. Their optical differences are much less apparent than the pronounced variations found along the radially stable branches. Within the sampled solution families, the stable branches therefore provide the clearest departures in shadow size and photon-orbit index.

To quantify the small departures that are difficult to distinguish on the scales of Figure 2, we display the same unstable backgrounds separately in Figure 3, with a row for each connected segment. The left panels give bc/Mb_c/M, while the right panels give χ\chi, discussed in § 3.5.

Figure 3. The radially unstable segments in Figure 2 are shown separately through bc/Mb_c/M (left) and χ\chi (right). From top to bottom, the rows correspond to the low-charge red dashed branch in the left panel of Figure 1 (κ=100\kappa= 100), the short red dashed continuation near the rightmost turning point of that panel (κ=100\kappa= 100), and the lower-charge red dashed branch in the right panel (κ=1000\kappa= 1000). Open circles mark the three calculated backgrounds on the short continuation. Black dash-dotted lines denote Schwarzschild black holes.

All three rows concern hyperbolic backgrounds with a growing radial mode, and the orange nonhyperbolic segments in Figure 1 are excluded.

The upper row of Figure 3 corresponds to the low-charge red dashed curve near the bottom of the left panel of Figure 1, for κ=100\kappa= 100. At masses where the low-charge branch coexists with the upper branch, the two backgrounds have different nonzero scalar charges, whereas Schwarzschild black holes lie at D=0D = 0. The computed low-charge branch extends to M/λ≃0.05688M/\lambda\simeq0.05688, and its shadow-size deviation satisfies ∣δsh∣≲0.04%\lvert\delta_{\rm sh}\rvert\lesssim0.04\%.

The middle row of Figure 3 corresponds to the short red dashed segment at the rightmost bend of the κ=100\kappa= 100 sequence in the left panel of Figure 1. The short segment continues the main solution sequence containing the blue radially stable branch. Near the maximum mass, the sequence develops a radially unstable portion and bends back toward smaller masses and scalar charges before entering the orange nonhyperbolic region. The high-mass unstable continuation is separate from the low-charge branch and spans approximately 0.12094≲M/λ≲0.121430.12094 \lesssim M/\lambda\lesssim0.12143, with δsh≃−(0.28–0.34)%\delta_{\rm sh} \simeq-(0.28\text{--}0.34)\%.

The lower row of Figure 3 corresponds to the lower-charge red dashed branch in the right panel of Figure 1, for κ=1000\kappa= 1000. The upper- and lower-charge branches form the two sides of a single sequence that turns near M/λ≃0.06463M/\lambda\simeq0.06463, with the red branch representing the radially unstable side. Where the red and blue branches coexist at the same mass, the two configurations have different scalar charges and radial stability. Over the displayed unstable branch, the maximum shadow-size reduction is approximately 0.161%0.161\%.

Across the displayed unstable segments, the shadow-size deviation from Schwarzschild black holes does not exceed approximately 0.34%0.34\%, well below the reductions of about 20%20\% reached at the low-mass ends of the stable branches. At the representative matched masses M/λ=0.05365M/\lambda= 0.05365 and 0.055500.05550, the radially stable solutions also have smaller shadows than their unstable counterparts for both couplings.

The full-branch sequences described above span different mass intervals for the two couplings. To isolate the coupling dependence, we next compare the shadow sizes at matched ADM masses.

Coupling dependence at matched ADM mass

To separate the effect of the coupling from differences in black-hole mass, we compare radially stable solutions at matched ADM masses in Figure 4, keeping λ\lambda fixed. The ten couplings κ=100,200,…,1000\kappa= 100, 200, \ldots, 1000 share the mass interval 0.05363≲M/λ≲0.064570.05363 \lesssim M/\lambda\lesssim0.06457. The figure uses sixteen matched masses spanning 0.05365≤M/λ≤0.064500.05365 \le M/\lambda\le0.06450, with denser coverage near the lower end, for a total of 160 backgrounds. The upper panels follow the mass dependence at each coupling, while the lower panels compare the couplings at M/λ=0.05365M/\lambda= 0.05365, 0.060000.06000, and 0.064500.06450. The left column shows the shadow-size deviation δsh\delta_{\rm sh} relative to Schwarzschild black holes, and the right column shows the photon-orbit index discussed in § 3.6.

Figure 4. The shadow-size deviation δsh\delta_{\mathrm{sh}}, displayed as a percentage (left column), and photon-orbit index χ\chi (right column) are compared for radially stable scalarized black holes at fixed λ\lambda. The upper panels show sixteen matched masses for each of the ten couplings κ=100,200,…,1000\kappa= 100, 200, \ldots, 1000, distinguished by the color bar. The lower panels show the coupling dependence at M/λ=0.05365M/\lambda= 0.05365, 0.060000.06000, and 0.064500.06450, using the same backgrounds. Black dash-dotted lines give the reference values for Schwarzschild black holes.

From the left panels, we find that the shadow radius increases monotonically with κ\kappa at every matched mass, while remaining below the value for Schwarzschild black holes. For example, at M/λ=0.06000M/\lambda= 0.06000, the size deviation is approximately −18.44%-18.44\%, −1.956%-1.956\%, and −0.460%-0.460\% for κ=100\kappa= 100, 500, and 1000, respectively. At each fixed coupling, the shadow also grows relative to that of Schwarzschild black holes as the mass increases across this common interval. The difference between the two endpoint couplings narrows with increasing mass: the κ=1000\kappa= 1000 shadow radius exceeds the κ=100\kappa= 100 radius by approximately 26.4%26.4\% at M/λ=0.05365M/\lambda= 0.05365 and 19.1%19.1\% at M/λ=0.06450M/\lambda= 0.06450.

By Eq. (29), the increase of bc/Mb_{\mathrm{c}}/M corresponds to a decrease of the dimensionless capture-barrier height M2Uopt(rph)M^{2}U_{\mathrm{opt}}(r_{\mathrm{ph}}). The local instability of rays near this barrier provides additional information, which we now examine through the photon-orbit index.

Photon-orbit instability and near-critical behavior

The instability of the circular photon orbit determines how rapidly nearby rays depart from it. Dividing Eq. (25) by t˙2=E2/A2\dot{t}^{2}=E^{2}/A^{2} gives

(drdt)2=A(r)B(r)[1−b2Uopt(r)].(31)\left(\frac{dr}{dt}\right)^{2}=A(r)B(r)\left[1-b^{2}U_{\mathrm{opt}}(r)\right]. \tag*{(31)}

Consider a nondegenerate unstable orbit with Uopt′′(rph)<0U_{\mathrm{opt}}^{\prime\prime}(r_{\mathrm{ph}})<0, and write r=rph+δrr=r_{\mathrm{ph}}+\delta r, where δr\delta r is a small radial displacement. At b=bcb=b_{\mathrm{c}}, the constant and linear terms in the bracket vanish. The quadratic expansion gives the growing and decaying radial modes

δr(t)∝e±Λpht,Λph2=−B(rph)rph22Uopt′′(rph).(32)\delta r(t)\propto e^{\pm\Lambda_{\mathrm{ph}}t},\qquad\Lambda_{\mathrm{ph}}^{2}=-\frac{B(r_{\mathrm{ph}})r_{\mathrm{ph}}^{2}}{2}U_{\mathrm{opt}}^{\prime\prime}(r_{\mathrm{ph}}). \tag*{(32)}

Here Λph>0\Lambda_{\mathrm{ph}}>0 denotes [17] the coordinate-time Lyapunov exponent, with dimensions of inverse length.

The orbital angular frequency, defined as a positive magnitude, is

Ωph≡∣dφdt∣r=rph=A(rph)rph=1bc.(33)\Omega_{\mathrm{ph}}\equiv\left|\frac{d\varphi}{dt}\right|_{r=r_{\mathrm{ph}}}=\frac{\sqrt{A(r_{\mathrm{ph}})}}{r_{\mathrm{ph}}}=\frac{1}{b_{\mathrm{c}}}. \tag*{(33)}

We define the dimensionless photon-orbit index by

χ≡ΛphΩph=bcΛph=[B(r)2A(r)(2A(r)−r2A′′(r))]r=rph,(34)\chi\equiv\frac{\Lambda_{\mathrm{ph}}}{\Omega_{\mathrm{ph}}}=b_{\mathrm{c}}\Lambda_{\mathrm{ph}}=\left[\frac{B(r)}{2A(r)}\left(2A(r)-r^{2}A^{\prime\prime}(r)\right)\right]_{r=r_{\mathrm{ph}}}, \tag*{(34)}

where the last equality uses rphA′(rph)=2A(rph)r_{\mathrm{ph}}A^{\prime}(r_{\mathrm{ph}})=2A(r_{\mathrm{ph}}). Since the accumulated orbital angle satisfies Δψ≃ΩphΔt\Delta\psi\simeq\Omega_{\mathrm{ph}}\Delta t near the circular orbit, the growing mode behaves as ∣δr∣∝eχΔψ|\delta r|\propto e^{\chi\Delta\psi}. Thus χ\chi measures radial instability per unit orbital angle, and larger χ\chi means faster separation over the same angular interval.

The same index governs the accumulation of orbital angle near the capture threshold. For scattering rays with b>bcb>b_{\mathrm{c}} and with the source and observer outside the photon sphere, the total accumulated angle, including both incoming and outgoing segments, has the leading behavior

Δψ=−1χln⁡ϵ+C+o(1),ϵ≡bbc−1⟶0+.(35)\Delta\psi=-\frac{1}{\chi}\ln\epsilon+C+o(1),\qquad\epsilon\equiv\frac{b}{b_{\mathrm{c}}}-1\longrightarrow0^{+}. \tag*{(35)}

Equation (35) gives [18, 23] the strong-deflection expansion for a single nondegenerate unstable photon orbit. Here ϵ\epsilon is the fractional distance from the critical impact parameter, and o(1)o(1) denotes terms that vanish as ϵ→0+\epsilon\to0^{+}. The finite constant CC depends on the geometry and the endpoints, whereas the logarithmic coefficient is fixed by χ\chi. The derivation and its relation to asymptotic image widths are given in Appendix § 8.§ 1. A smaller χ\chi therefore gives a stronger logarithmic increase in the accumulated orbital angle as the critical impact parameter is approached.

For a fixed emitting radial interval, consider successive images labeled by order nn such that each additional order adds an orbital angle of π\pi. If wnw_n denotes the width of the corresponding image interval in impact-parameter space, Eq. (35) implies

lim⁡n→∞wn+1wn=e−πχ,γhalf=πχ.(36)\lim_{n\to\infty}\frac{w_{n+1}}{w_n}=e^{-\pi\chi},\qquad\gamma_{\mathrm{half}}=\pi\chi. \tag*{(36)}

The half-orbit lensing exponent γhalf\gamma_{\rm half} is the logarithmic compression per additional half orbit, following the convention used in Ref. [18]. Larger χ\chi produces stronger compression between successive high-order bands.

The shadow size and the photon-orbit index probe different properties of the optical barrier. Equations (29) and (32) imply

(bcM)−2=M2Uopt(rph),χ2=−B(rph)rph2Uopt′′(rph)2Uopt(rph).(37)\left(\frac{b_{\rm c}}{M}\right)^{-2}=M^2U_{\rm opt}(r_{\rm ph}),\qquad\chi^2=-\frac{B(r_{\rm ph})r_{\rm ph}^2U_{\rm opt}^{\prime\prime}(r_{\rm ph})}{2U_{\rm opt}(r_{\rm ph})}. \tag*{(37)}

The dimensionless barrier height fixes bc/Mb_{\rm c}/M: a higher peak gives a smaller shadow. The index χ\chi additionally depends on the curvature of the optical potential at its maximum, the photon-sphere radius, and the radial metric function BB. Consequently, backgrounds with equal normalized shadow sizes can have [18] different near-critical photon dynamics.

For the metric of Schwarzschild black holes in Eq. (21), the optical potential and its first derivative are

UoptSch(r)=1r2−2Mr3,dUoptSchdr=−2(r−3M)r4.(38)U_{\rm opt}^{\rm Sch}(r)=\frac{1}{r^2}-\frac{2M}{r^3},\qquad\frac{dU_{\rm opt}^{\rm Sch}}{dr}=-\frac{2(r-3M)}{r^4}. \tag*{(38)}

The unique exterior extremum is at r=3Mr=3M, where UoptSch=1/(27M2)U_{\rm opt}^{\rm Sch}=1/(27M^2) and d2UoptSch/dr2=−2/(81M4)<0d^2U_{\rm opt}^{\rm Sch}/dr^2=-2/(81M^4)<0. Substitution into the expressions above gives the reference values

rphSchM=3,bcSchM=33,δshSch=0,\frac{r_{\rm ph}^{\rm Sch}}{M}=3,\qquad\frac{b_{\rm c}^{\rm Sch}}{M}=3\sqrt{3},\qquad\delta_{\rm sh}^{\rm Sch}=0,
MΩphSch=133,MΛphSch=133,χSch=1.(39)M\Omega_{\rm ph}^{\rm Sch}=\frac{1}{3\sqrt{3}},\qquad M\Lambda_{\rm ph}^{\rm Sch}=\frac{1}{3\sqrt{3}},\qquad\chi^{\rm Sch}=1. \tag*{(39)}

For the half-orbit sequences in Eq. (36), these reference values give γhalfSch=π\gamma_{\rm half}^{\rm Sch}=\pi and wn+1/wn→e−π≃0.0432w_{n+1}/w_n\to e^{-\pi}\simeq0.0432.

Mass dependence and radial-branch comparisons at fixed κ\kappa

To determine whether the monotonic growth of the shadow radius is accompanied by a monotonic change in photon-orbit instability, we now examine the right panels of Figure 2. The comparison follows χ\chi along the same branches at fixed κ=100\kappa=100 and 10001000 and fixed λ\lambda, with χ=1\chi=1 for Schwarzschild black holes as the reference.

Both radially stable branches show a nonmonotonic dependence on mass, with the extrema quoted below taken from the sampled sequences. For κ=100\kappa=100, χ\chi starts at approximately 1.402, reaches a maximum of 1.411 near M/λ≃0.05939M/\lambda\simeq0.05939, and then falls below unity near M/λ≃0.085M/\lambda\simeq0.085. The index then decreases to a minimum of approximately 0.741 near M/λ≃0.09417M/\lambda\simeq0.09417, before recovering to 0.981 at the high-mass end. For κ=1000\kappa=1000, χ\chi starts at approximately 1.027, rises to a maximum of 1.104 near M/λ≃0.02828M/\lambda\simeq0.02828, and falls below unity near M/λ≃0.042.Theminimumisapproximately0.9396nearM/\lambda\simeq0.042. The minimum is approximately 0.9396 nearM/\lambda\simeq 0.05060, followed by an increase to 0.9894 at the high-mass end.

The shadow radius and photon-orbit index therefore capture different aspects of the scalarized geometry. By Eq. (37), the monotonic growth of bc/Mb_{\rm c}/M corresponds to a decreasing dimensionless barrier height M2Uopt(rph)M^2U_{\rm opt}(r_{\rm ph}). At the minimum of χ\chi for κ=100\kappa=100, the shadow is about 3.55% smaller than the shadow of a Schwarzschild black hole, whereas χ\chi is about 25.9% below unity. At the corresponding minimum for κ=1000\kappa=1000, the departures are approximately 1.33% and 6.04%, respectively. These examples show that a modest size difference can accompany a more pronounced change in near-critical photon dynamics.

Where χ>1\chi>1, neighboring rays separate more rapidly per unit orbital angle than around Schwarzschild black holes. Where χ<1\chi<1, neighboring rays separate more slowly and the leading logarithmic increase in the accumulated orbital angle is stronger. The positive minima imply that the circular photon orbits remain unstable throughout both radially stable segments. For the half-orbit image sequence in Eq. (36), the decrease of χ\chi also implies weaker compression between successive high-order bands.

The right panels of Figure 3 provide the corresponding mass comparison for the three radially unstable segments identified in Section 3.2.1. In contrast to the larger excursions on the radially stable branches, all three unstable segments have photon-orbit indices close to unity. The low-charge κ=100\kappa= 100 branch has 0.9987≲χ≲10.9987 \lesssim\chi\lesssim1, the short high-mass continuation at the same coupling has χ≃0.984\chi\simeq0.984–0.9880.988, and the lower-charge κ=1000\kappa= 1000 branch has 0.9910≲χ≲10.9910 \lesssim\chi\lesssim1.

At the representative matched masses M/λ=0.05365M/\lambda= 0.05365 and 0.055500.05550, the radially stable solutions have larger photon-orbit indices than their unstable counterparts at κ=100\kappa= 100, but smaller indices at κ=1000\kappa= 1000. The ordering of shadow sizes therefore does not fix the ordering of χ\chi. The unstable solutions admit growing radial perturbations despite their optical indicators being close to those of Schwarzschild black holes: χ\chi describes null-geodesic separation on a given metric, whereas background stability concerns perturbations of the metric and scalar field. Neither a shadow resembling that of a Schwarzschild black hole nor χ≃1\chi\simeq1 establishes radial stability. Over the sampled unstable segments, the largest departure of χ\chi from unity is only about 1.6%. Together with the small shadow-size differences in Section 3.2.1, this confirms that the unstable branches provide much weaker departures from the Schwarzschild optical indicators than the pronounced variations on the stable branches.

The comparison above describes the behavior along each branch over its own mass interval. Because these intervals differ, the smaller excursion of χ\chi at κ=1000\kappa= 1000 does not by itself establish how changing κ\kappa affects a black hole of a specified mass. We address that question using matched ADM masses in § 3.6.

Coupling dependence at matched ADM mass

For the coupling dependence of near-critical photon dynamics, we return to the right panels of Figure 4. The comparison uses radially stable backgrounds at κ=100,200,…,1000\kappa= 100, 200, \ldots, 1000 and the same sixteen matched masses in 0.05365≤M/λ≤0.064500.05365 \leq M/\lambda\leq0.06450 as the shadow-size comparison in § 3.3, with λ\lambda fixed. The upper panel follows χ\chi with mass at each coupling, and the lower panel isolates its coupling dependence at M/λ=0.05365M/\lambda= 0.05365, 0.060000.06000, and 0.064500.06450.

At each matched mass, we find that χ\chi first decreases and then increases as κ\kappa grows, with its smallest sampled value occurring at an intermediate coupling. For M/λ=0.05365M/\lambda= 0.05365, 0.060000.06000, and 0.064500.06450, these minima occur at κ=700\kappa= 700, 500, and 400, with χ≃0.9286\chi\simeq0.9286, 0.91080.9108, and 0.89800.8980, respectively. Thus the coupling associated with the smallest index shifts toward lower values as the mass increases. The mass sequences in the upper-right panel also cross the reference value for Schwarzschild black holes for some couplings. For κ=100\kappa= 100 and 200, all sampled values satisfy χ>1\chi> 1, whereas for κ=500,600,…,1000\kappa= 500, 600, \ldots, 1000 they satisfy χ<1\chi< 1. For κ=300\kappa= 300 and 400, the crossings of χ=1\chi= 1 are bracketed by 0.059<M/λ<0.0600.059 < M/\lambda< 0.060 and 0.0540<M/λ<0.05430.0540 < M/\lambda< 0.0543, respectively.

The different coupling dependences of the shadow radius and photon-orbit index distinguish two geometric effects. By Eq. (37), the monotonic increase of bc/Mb_c/M corresponds to a decrease of the dimensionless capture-barrier height. The nonmonotonic index instead reflects the combined dependence on the barrier curvature and the local metric at the photon sphere. Near the sampled minima of χ\chi, neighboring rays separate more slowly per unit orbital angle, the logarithmic coefficient 1/χ1/\chi of the accumulated orbital angle is larger, and the asymptotic half-orbit width ratio e−πχe^{-\pi\chi} is larger. Higher-order image bands therefore undergo weaker compression toward the critical curve, under the fixed-source assumptions of Eq. (36).

Photon trajectories and photon-ring width

Consider a light ray traced backward from an observer at infinity in the asymptotically flat spacetime. We use the accumulated orbital angle ψ\psi introduced in § 3.1, setting ψ=0\psi= 0 at the observer and increasing it along the backward trajectory. The angular change along a radial segment follows from the conserved quantities in § 3.1:

∣dψdr∣=bA(r)/B(r)r21−b2A(r)/r2.(40)\left|\frac{\mathrm{d}\psi}{\mathrm{d}r}\right| = \frac{b\sqrt{A(r)/B(r)}}{r^{2}\sqrt{1-b^{2}A(r)/r^{2}}}. \tag*{(40)}

For a captured ray, the total accumulated angle covers the path from infinity to the horizon. For a scattered ray, it includes the ingoing and outgoing segments between infinity and the outer turning point. Therefore, the total accumulated angle Δψ\Delta\psi is

Δψ(b)={∫rH∞∣dψdr∣dr,0≤b<bc,2∫rt(b)∞∣dψdr∣dr,b>bc,(41)\Delta\psi(b) = \begin{cases} \int_{r_{\mathrm{H}}}^{\infty}\left|\frac{\mathrm{d}\psi}{\mathrm{d}r}\right|\mathrm{d}r, & 0 \le b < b_{\mathrm{c}},\\[6pt] 2\int_{r_{\mathrm{t}}(b)}^{\infty}\left|\frac{\mathrm{d}\psi}{\mathrm{d}r}\right|\mathrm{d}r, & b > b_{\mathrm{c}}, \end{cases} \tag*{(41)}

where rt(b)r_{\mathrm{t}}(b) is the outer turning point. Dividing the total angle by one full revolution defines [7] the fractional number of orbits,

N(b)=Δψ(b)2π.(42)N(b) = \frac{\Delta\psi(b)}{2\pi}. \tag*{(42)}

For an equatorial thin disk viewed along its normal, successive crossings of the disk plane occur at ψm=(m−12)π\psi_m = (m-\frac{1}{2})\pi, where m=1,2,…m = 1,2,\ldots counts the crossings from the observer. Following the geometric classification of Ref. [7], we call rays with N<3/4N < 3/4 direct, those with 3/4<N<5/43/4 < N < 5/4 lensed, and those with N>5/4N > 5/4 members of the photon-ring class. These classes allow for rays to have at most one, exactly two, and at least three crossings of the plane, respectively. We use n=m−1n = m - 1 for the discrete disk-image order, so n=0n = 0 denotes the direct image.

For the finite-radius calculations, the observer and the other endpoint of each scattered ray are placed at the common radius ro=200Mr_{\mathrm{o}} = 200M. The subscript o labels the resulting finite-radius quantities, Δψo\Delta\psi_{\mathrm{o}} and No=Δψo/(2π)N_{\mathrm{o}} = \Delta\psi_{\mathrm{o}}/(2\pi), with No→NN_{\mathrm{o}} \to N as ro→∞r_{\mathrm{o}} \to\infty.

To connect the near-critical orbit analysis with complete photon paths, we compare the number of orbits and selected trajectories in Figure 5 at M/λ=0.06000M/\lambda= 0.06000 and fixed λ\lambda. The columns show Schwarzschild black holes and radially stable scalarized backgrounds at κ=100\kappa= 100, 500, and 1000. The intermediate coupling is included because it gives the smallest sampled χ\chi at this mass in § 3.6. The upper row plots NoN_{\mathrm{o}} against b/Mb/M and identifies the direct, lensed, and photon-ring classes, while the lower row shows representative paths from those classes in Cartesian coordinates (x,y)(x,y) in the orbital plane.

Figure 5. The upper row shows the number of orbits as a function of impact parameter, and the lower row shows selected null trajectories at M/λ=0.06000M/\lambda= 0.06000 for Schwarzschild black holes and radially stable scalarized black holes with κ=100\kappa= 100, 500, and 1000. Rays are traced from ro=200Mr_o = 200M. Black, gold, and red curves denote the direct, lensed, and photon-ring classes, respectively. In the upper panels, horizontal dashed lines mark No=3/4N_o = 3/4 and 5/45/4, and vertical dotted lines mark bc/Mb_c/M. In the lower panels, the black disk is the horizon.

In the upper row, NoN_{\mathrm{o}} increases toward the capture threshold from b<bcb < b_{\mathrm{c}} and decreases as b>bcb > b_{\mathrm{c}} moves away from the threshold. The apparent spike marks b=bcb = b_{\mathrm{c}}, where the ray approaches the unstable circular photon orbit and accumulates an unbounded angle. The captured and scattered curves are separate branches, with no finite total angle at their common limiting value. The lower row shows the corresponding behavior in real space: rays with the largest numbers of orbits remain close to r=rphr = r_{\mathrm{ph}} before falling inward or escaping.

For κ=100\kappa= 100, we also find a shoulder, or a locally flatter portion of the scattered-side curve, near No=1N_{\mathrm{o}} = 1. This feature changes the shape of No(b/M)N_{\mathrm{o}}(b/M), so the curve for this scalarized background cannot be obtained from the curve for Schwarzschild black holes by a simple shift or uniform rescaling of the impact parameter.

Table 1 gives the four classification boundaries for the backgrounds at M/λ=0.06000M/\lambda= 0.06000 shown in Figure 5. The boundaries are the impact parameters at which No=3/4N_{\mathrm{o}} = 3/4 or 5/45/4, evaluated separately on either side of bcb_{\mathrm{c}}. Writing b3/4−b^{-}_{3/4} and b5/4−b^{-}_{5/4} for the captured-side thresholds and b5/4+b^{+}_{5/4} and b3/4+b^{+}_{3/4} for the scattered-side thresholds, the photon-ring class occupies (b5/4−,bc)∪(bc,b5/4+)(b^{-}_{5/4},b_c)\cup(b_c,b^{+}_{5/4}). The total impact-parameter width of this class is approximately 0.008683M0.008683M for κ=100\kappa= 100, compared with 0.038276M0.038276M for Schwarzschild black holes and 0.042631M0.042631M for κ=1000\kappa= 1000. Thus the κ=1000\kappa= 1000 critical radius is close to that of Schwarzschild black holes while its photon-ring class is somewhat wider. By comparison, the κ=100\kappa= 100 class is substantially narrower and displaced inward. To quantify this range along our mass and coupling sequences, we use [16] the photon-ring width in impact-parameter space,

Backgroundb3/4−/Mb^{-}_{3/4}/Mb5/4−/Mb^{-}_{5/4}/Mbc/Mb_c/Mb5/4+/Mb^{+}_{5/4}/Mb3/4+/Mb^{+}_{3/4}/M
Schwarzschild black holes5.0193785.1880215.1961525.2262986.094220
κ=100\kappa= 1004.1610534.2370844.2381204.2457675.878705
κ=500\kappa= 5004.8952645.0829025.0945185.1388556.063948
κ=1000\kappa= 10004.9823395.1626275.1722375.2052586.083820

Table 1. The table lists the geometric classification boundaries at M/λ=0.06000M/\lambda= 0.06000 for endpoints at ro=200Mr_o = 200M. Superscripts −- and ++ identify the captured and scattered sides, respectively.

Wph≡b5/4+−b5/4−,(43)W_{\rm ph} \equiv b^{+}_{5/4} - b^{-}_{5/4}, \tag*{(43)}

which measures the full interval with N>5/4N > 5/4 (No>5/4N_o > 5/4 for the finite-endpoint results). The critical value itself is excluded from the class, but removing this single divergent ray does not change its total width.

To determine how the impact-parameter range occupied by photon-ring rays changes across the solution families, we present Wph/MW_{\rm ph}/M in Figure 6 at fixed λ\lambda. Panels (a) and (b) follow the mass dependence along the radially stable and radially unstable branches at fixed κ=100\kappa= 100 and 1000, respectively. Panel (c) compares radially stable solutions at κ=100,200,…,1000\kappa= 100, 200, \ldots, 1000 for each of the three fixed masses M/λ=0.05365M/\lambda= 0.05365, 0.060000.06000, and 0.064500.06450. Schwarzschild black holes provide the reference width in all three panels.

Figure 6. The photon-ring width Wph/MW_{\mathrm{ph}}/M is shown as a function of mass and coupling at fixed λ\lambda. Panels (a) and (b) vary M/λM/\lambda at κ=100\kappa= 100 and 10001000, respectively. Blue solid and red dashed curves denote radially stable and radially unstable backgrounds. Panel (c) varies κ=100,200,…,1000\kappa= 100, 200, \ldots, 1000 for radially stable backgrounds at M/λ=0.05365M/\lambda= 0.05365, 0.060000.06000, and 0.064500.06450. Black dash-dotted lines give the corresponding width for Schwarzschild black holes.

Along the radially stable κ=100\kappa= 100 segment in panel (a), Wph/MW_{\mathrm{ph}}/M grows from approximately 0.005380.00538 at the low-mass end to a sampled maximum of 0.120940.12094 near M/λ=0.08494M/\lambda= 0.08494, before decreasing again. At κ=1000\kappa= 1000, the largest sampled photon-ring width, Wph/M≃0.05160W_{\mathrm{ph}}/M \simeq0.05160, occurs near M/λ=0.04530M/\lambda= 0.04530. The radially unstable segments remain closer to the reference width Wph/M≃0.03828W_{\mathrm{ph}}/M \simeq0.03828 of Schwarzschild black holes, reaching maxima of approximately 0.040910.04091 on the short high-mass continuation at κ=100\kappa= 100 and 0.039570.03957 at κ=1000\kappa= 1000.

At the three matched masses shown in panel (c), Wph/MW_{\mathrm{ph}}/M varies nonmonotonically with κ\kappa, in contrast to the monotonic shadow size in § 3.3. For example, at M/λ=0.06000M/\lambda= 0.06000, the largest sampled photon-ring width is Wph/M≃0.06715W_{\mathrm{ph}}/M \simeq0.06715 at κ=300\kappa= 300, whereas the smallest sampled χ\chi occurs at κ=500\kappa= 500, as shown in Figure 4. The largest photon-ring width therefore need not occur at the coupling with the smallest photon-orbit index. This difference arises because WphW_{\mathrm{ph}} depends on the accumulated angle along the full ray path, whereas χ\chi measures the local instability near the circular photon orbit.

Complementarity of shadow size, orbit instability, and ring width

The results above motivate describing the photon geometry by the pair (bc/M,χ)(b_{\mathrm{c}}/M,\chi), or equivalently (δsh,χ)(\delta_{\mathrm{sh}},\chi). The quantities bc/Mb_{\mathrm{c}}/M and δsh\delta_{\mathrm{sh}} encode the same size information, whereas χ\chi supplies an additional measure of near-critical photon dynamics. By Eq. (37), the size fixes the dimensionless height of the capture barrier, while the index also probes the radial metric at the photon sphere. This distinction is manifest in Figure 2: the normalized shadow radius increases monotonically along each radially stable segment, but χ\chi varies nonmonotonically.

A particularly clear illustration is provided by two radially stable backgrounds in the full-branch sample. At κ=100\kappa= 100 and M/λ≃0.060208M/\lambda\simeq0.060208, we obtain bc/M≃4.2435b_{\mathrm{c}}/M \simeq4.2435 and χ≃1.4110\chi\simeq1.4110. At κ=1000\kappa= 1000 and M/λ≃0.09973M/\lambda\simeq0.09973, the corresponding values are bc/M≃4.2398b_{\mathrm{c}}/M \simeq4.2398 and χ≃1.0292\chi\simeq1.0292. Their normalized shadow radii bc/Mb_{\mathrm{c}}/M differ by less than 0.1%0.1\%, yet their photon-orbit indices differ substantially. The comparison shows that nearly equal normalized shadow radii can coexist with distinctly different photon-orbit instability. The difference in χ\chi has a direct consequence for the high-order image hierarchy. For a fixed emitting radial interval and successive image orders separated by a half orbit, Eq. (36) predicts width ratios approaching 0.011880.01188 and 0.039430.03943 for these two backgrounds, respectively. The width ratio for the κ=1000\kappa= 1000 background is about $3.32 times that for the κ=100\kappa= 100 background, corresponding to weaker compression toward the critical curve. Similarly, the leading logarithmic coefficient 1/χ1/\chi in Eq. (35) is larger for the κ=1000\kappa= 1000 background. Near the critical curve, reducing the fractional impact-parameter offset by a fixed factor consequently produces a larger increase in the accumulated orbital angle, even though the two critical radii in units of mass are almost identical.

The same complementarity is present in the controlled branch comparison at fixed mass. For κ=1000\kappa= 1000 and M/λ=0.05365M/\lambda= 0.05365, the radially stable solution has a shadow radius only 0.921%0.921\% smaller than that of the unstable solution, but its index is 5.36%5.36\% smaller. Their asymptotic half-orbit width ratios are approximately 0.051420.05142 and 0.043470.04347, respectively, so the stable background has a width ratio about 18.3%18.3\% larger.

Together, shadow size and χ\chi specify the position of the critical curve and the asymptotic manner in which image bands approach it. The photon-ring width Wph/MW_{\rm ph}/M adds a finite-angle measure: it specifies the impact-parameter interval occupied by rays in the photon-ring class defined in § 3.7. These complementary quantities provide the geometric basis for the radiative images in § 4, where the emissivity, emitter motion, and emitting region determine how those light paths contribute to the observed intensity.

Spherical Emission and Thin-Disk Images

We focus on the radially stable scalarized backgrounds to examine how the geometric departures found in § 3 appear under different emission models. The radially unstable branches serve as geometric controls in Figure 2 and Figure 3, where their shadow radii and photon-orbit indices remain close to those of Schwarzschild black holes. We describe the images using Cartesian impact-parameter coordinates (bx,by)(b_x,b_y), with b=bx2+by2b = \sqrt{b_x^2+b_y^2} measuring distance from the image center.

Spherical emission models

We model the emitting matter as an optically thin, spherically symmetric distribution with negligible backreaction on the black-hole spacetime. Here, “optically thin” means that photons traverse the emitting region with negligible absorption or scattering. Emission is isotropic in the local rest frame, and we consider both static emitters and matter in radial free fall from rest at infinity.

For a photon with four-momentum kμk^\mu, the frequency shift between emission and observation is characterized by the dimensionless factor

Z≡νoνe=−kμuoμ−kμueμ,uoμ=Ao−1/2(∂t)μ,(44)\mathcal{Z} \equiv\frac{\nu_o}{\nu_e} = \frac{-k_\mu u_o^\mu}{-k_\mu u_e^\mu}, \qquad u_o^\mu= A_o^{-1/2}(\partial_t)^\mu, \tag*{(44)}

where νo\nu_o and νe\nu_e denote the photon frequencies measured in the local rest frames of the observer (o)(o) and emitter (e)(e), while uoμu_o^\mu and ueμu_e^\mu are their respective four-velocities. The observer is static at radius ror_o, where the metric function takes the value Ao=A(ro)A_o = A(r_o). We consider bolometric intensities, meaning intensities integrated over all photon frequencies. Neglecting absorption and scattering, the observed intensity is therefore

Io=∫γZ4je dℓe,(45)I_o = \int_\gamma\mathcal{Z}^4 j_e\,d\ell_e, \tag*{(45)}

where IoI_o is the radiative energy received per unit time, projected area, and solid angle, integrated over all observed frequencies. The comoving emissivity jej_e is the energy emitted per unit proper time, proper volume, and solid angle, integrated over all frequencies in the emitter’s rest frame. The integration path γ\gamma is the photon’s null geodesic reaching the observer, and dℓed\ell_e is the spatial path-length element measured in the emitter’s local rest frame.

For the spherical volume, the isotropic comoving emissivity is

je(r)=j0(rrref)−p1[rin,rout](r),(46)j_{\mathrm e}(r)=j_0\left(\frac{r}{r_{\mathrm{ref}}}\right)^{-p}\mathbf{1}_{[r_{\mathrm{in}},r_{\mathrm{out}}]}(r), \tag*{(46)}

where rinr_{\mathrm{in}} and routr_{\mathrm{out}} are the inner and outer radii of the emitting spherical shell. The independent reference radius rrefr_{\mathrm{ref}} sets the normalization, with j0j_0 the local emissivity there and pp the radial power-law index. The spherical-emission results below use rref=6Mr_{\mathrm{ref}}=6M and the same j0j_0 across the compared backgrounds and emitter motions. Thus the normalization radius is distinct from the inner boundary of the emitting region. The indicator function 1[rin,rout](r)\mathbf{1}_{[r_{\mathrm{in}},r_{\mathrm{out}}]}(r) equals 1 for rin≤r≤routr_{\mathrm{in}}\le r\le r_{\mathrm{out}} and 0 otherwise. To express the transfer factors compactly, we define P(r,b)=1−b2A(r)/r2\mathcal{P}(r,b)=1-b^2A(r)/r^2, the dimensionless radial factor in Eq. (25). Allowed ray segments have P≥0\mathcal{P}\ge0, and a radial turning point satisfies P=0\mathcal{P}=0. For static emitters, the frequency shift and local path length are

Zstat=A(r)Ao,dℓstat=∣dr∣B(r)P(r,b).(47)\mathcal{Z}_{\mathrm{stat}}=\sqrt{\frac{A(r)}{A_o}},\qquad d\ell_{\mathrm{stat}}=\frac{\lvert dr\rvert}{\sqrt{B(r)\mathcal{P}(r,b)}}. \tag*{(47)}

Scattered rays receive emission along both radial segments, but captured backward rays receive emission along the single segment outside the horizon. Thus the region b<bcb<b_c can have nonzero intensity even though rays entering it from the observer do not return to the outer boundary.

For radial free fall from rest at infinity, the emitter has conserved energy per unit rest mass −ue,t=1-u_{e,t}=1 and four-velocity

uet=1A(r),uer=−B(r)(1A(r)−1).(48)u_e^t=\frac{1}{A(r)},\qquad u_e^r=-\sqrt{B(r)\left(\frac{1}{A(r)}-1\right)}. \tag*{(48)}

Let s=+1s=+1 denote a future-directed outgoing photon segment and s=−1s=-1 an ingoing segment. The dimensionless factor Ds(r,b)=1+s[1−A(r)]P(r,b)D_s(r,b)=1+s\sqrt{[1-A(r)]\mathcal{P}(r,b)} accounts for the direction dependence introduced by the infalling emitter’s motion. Using the subscript ff for free fall, the frequency shift and path-length transformation are

Zs=A(r)AoDs(r,b),dℓffdℓstat=Ds(r,b)A(r).(49)\mathcal{Z}_s=\frac{A(r)}{\sqrt{A_oD_s(r,b)}},\qquad \frac{d\ell_{\mathrm{ff}}}{d\ell_{\mathrm{stat}}}=\frac{D_s(r,b)}{\sqrt{A(r)}}. \tag*{(49)}

Substituting these factors and Eq. (47) into Eq. (45) gives

Ioff(b)=1Ao2∑s=±1∫γsA(r)7/2je(r)Ds(r,b)3B(r)P(r,b)∣dr∣,(50)I_o^{\mathrm{ff}}(b)=\frac{1}{A_o^2}\sum_{s=\pm1}\int_{\gamma_s}\frac{A(r)^{7/2}j_{\mathrm e}(r)}{D_s(r,b)^3\sqrt{B(r)\mathcal{P}(r,b)}}\lvert dr\rvert, \tag*{(50)}

where γs\gamma_s denotes the part of a ray with direction ss inside the emitting region. Both signs contribute for scattered rays, whereas captured backward rays include only s=+1s=+1, corresponding to photons traveling outward toward the observer. The additional redshift of radiation escaping against the ingoing motion suppresses the central brightness.

In both spherical emission models, nonzero emission near the photon sphere leads [7] to a logarithmic divergence of the ideal intensity at the critical curve, as near-critical rays accumulate emission over an arbitrarily long optical path. This divergence is integrable, so averaging over a finite image area gives a finite intensity. We therefore compare the peak intensities of static and freely infalling spherical emission after area averaging over annuli of fixed radial width Δb/M=0.01\Delta b/M=0.01 in the image plane. We denote the maximum of this area-averaged intensity by IpkI_{\mathrm{pk}}.

To examine how the geometric shadow features appear under different emitter motions, we compare static and freely infalling spherical emission in Figure 7 at fixed λ\lambda and M/λ=0.05365M/\lambda=0.05365. The columns show radially stable backgrounds at κ=100\kappa=100 and 1000, followed by Schwarzschild black holes, with the same p=2p = 2 emissivity extending from just outside rHr_H to 30M30M and the same normalization j0j_0 at rref=6Mr_{\mathrm{ref}} = 6M. The first two rows display the static and infalling images, respectively, and the lower row compares their radial intensity profiles.

Figure 7. The images compare spherical emission at M/λ=0.05365M/\lambda= 0.05365, with p=2p = 2 and emission extending from just outside rHr_H to 30M30M. From left to right, the columns show radially stable scalarized backgrounds at κ=100\kappa= 100 and 10001000, followed by Schwarzschild black holes, at fixed λ\lambda. The upper and middle rows show static emitters and radial free fall, respectively, with a common linear intensity scale across backgrounds within each row. White dashed circles mark bcb_c. The lower row compares the static and infalling radial profiles, where dotted vertical lines mark bc/Mb_c/M. Intensities use the same comoving normalization j0j_0 at rref=6Mr_{\mathrm{ref}} = 6M in the two spherical emission models.

Both emission models retain the inward displacement of the bright ring near the critical curve for κ=100\kappa= 100, whose shadow radius is 21.65% smaller than that of a Schwarzschild black hole of the same mass. For κ=1000\kappa= 1000, the shadow radius is only 0.968% smaller, and the images remain close to the reference images for Schwarzschild black holes. Changing the emitter motion on a fixed background leaves the geometric critical curve unchanged but modifies the intensity contrast across the image.

Comparing the central intensities with the area-averaged peaks, we obtain central-to-peak ratios of 0.3154 and 0.0477 for static emission and infall, respectively, at κ=100\kappa= 100. The corresponding ratios are 0.2703 and 0.0550 for Schwarzschild black holes, and 0.2701 and 0.0547 for κ=1000\kappa= 1000. Radial infall therefore produces a darker center relative to the bright ring on each background, consistent with the additional redshift of radiation escaping against the infalling matter.

The pronounced image differences at κ=100\kappa= 100 motivate following how the scalarized images approach those of Schwarzschild black holes as mass increases. For this purpose, we present the mass sequence along the radially stable κ=100\kappa= 100 branch in Figure 8, keeping λ\lambda and the emission prescriptions fixed. The first four columns show the evenly spaced masses M/λ=0.06M/\lambda= 0.06, 0.080.08, 0.100.10, and 0.120.12, while the rightmost column shows Schwarzschild black holes. The upper and lower rows compare static emission and radial free fall, respectively, using the same p=2p = 2 emissivity from just outside rHr_H to 30M30M as in Figure 7.

Figure 8. The spherical-emission images follow the mass dependence along the radially stable branch at fixed κ=100\kappa= 100 and fixed λ\lambda. The first four columns show scalarized backgrounds at the evenly spaced masses M/λ=0.06M/\lambda= 0.06, 0.080.08, 0.100.10, and 0.120.12, and the last column shows Schwarzschild black holes. The upper and lower rows show static emitters and radial free fall from rest at infinity, respectively. We keep p=2p = 2, emission from just outside rHr_H to 30M30M, and the comoving normalization j0j_0 at rref=6Mr_{\mathrm{ref}} = 6M fixed. Each row shares a linear color scale for Io/(j0M)I_o/(j_0M) and white dashed circles mark bcb_c.

Across the four scalarized backgrounds, the critical radius bc/Mb_c/M increases from 4.23814.2381 to 5.16955.1695, and the bright ring moves outward toward the reference location in both models. The shadow radius is smaller than that of a Schwarzschild black hole with the same ADM mass by approximately 18.44%18.44\%, 8.91%8.91\%, 2.37%2.37\%, and 0.51%0.51\%, respectively.

To quantify the brightness changes accompanying the outward motion of the ring, we present the radial intensity profiles and peak intensities in Figure 9 at fixed κ=100\kappa= 100 and λ\lambda. Panels (a) and (b) compare static emission and radial infall at the same four masses, M/λ=0.06M/\lambda= 0.06, 0.080.08, 0.100.10, and 0.120.12, as the images in Figure 8. Panel (c) follows the peak intensity over the radially stable mass sequence, with the peaks defined by area averaging over annuli of width 0.01M0.01M. The reference curves and horizontal lines show Schwarzschild black holes under the same emission and averaging prescriptions.

Figure 9. The radial intensity profiles and peak intensities are compared for spherical emission along the radially stable branch at fixed κ=100\kappa= 100 and fixed λ\lambda. Panels (a) and (b) compare static emission and radial infall, respectively, at M/λ=0.06M/\lambda= 0.06, 0.080.08, 0.100.10, and 0.120.12, using the same backgrounds and emission prescriptions as Figure 8. Black dashed curves show Schwarzschild black holes, and the insets enlarge the peaks for M/λ=0.10M/\lambda= 0.10, 0.120.12, and Schwarzschild black holes with the same colors and line styles. Panel (c) shows how the peak intensity IpkI_{\rm pk} varies with mass, with the peak defined as the maximum of the area-averaged intensity over annuli of width 0.01M0.01M. Horizontal dashed lines give the corresponding mass-independent peaks for Schwarzschild black holes.

The profiles in Figure 9(a,Figure 9b) show that the distinction in central contrast between the two emission models persists across the four masses. Among the four displayed masses, the peak intensity in units of j0Mj_0M decreases as mass increases for freely infalling emitters, whereas it first increases and then decreases for static emitters. The insets resolve the smaller differences between the peaks at M/λ=0.10M/\lambda= 0.10, 0.120.12, and the reference peaks for Schwarzschild black holes.

Over the full mass sequence in Figure 9(c), we find an additional low-mass minimum for static emitters: Ipk/(j0M)I_{\mathrm{pk}}/(j_0M) falls to approximately 21.8621.86 near M/λ≃0.0667M/\lambda\simeq0.0667, rises to a maximum of approximately 25.3125.31 near M/λ≃0.0926M/\lambda\simeq0.0926, and then decreases. This nonmonotonic trend persists for averaging widths of 0.005M0.005M and 0.02M0.02M. For radial infall, the peak decreases from approximately 24.8824.88 to 20.8120.81. The horizontal reference lines give Ipk/(j0M)≃23.99I_{\mathrm{pk}}/(j_0M) \simeq23.99 for static emission and 20.77 for radial infall around Schwarzschild black holes. The static-emission peak crosses its reference, whereas the infall peak approaches its reference from above. Thus, the monotonic approach of the shadow size to the value for Schwarzschild black holes does not require a monotonic approach of the brightness peak, which also depends on the emitter motion and radiative transfer.

Thin-disk motion, orbital structure, and inner edges

We next consider a transparent, geometrically thin equatorial disk viewed face-on, along its normal. Here geometric thinness means that the vertical thickness is neglected and emission is assigned to intersections of the ray with the equatorial plane. We prescribe the comoving bolometric surface intensity

Iem(r)=I0(rrin)−p1[rin,30M](r),Io(b)=∑mZm4Iem(rm(b)).(51)I_{\rm em}(r) = I_0 \left(\frac{r}{r_{\rm in}}\right)^{-p} \mathbf{1}_{[r_{\rm in},30M]}(r), \qquad I_o(b) = \sum_m \mathcal{Z}_m^4 I_{\rm em}(r_m(b)). \tag*{(51)}

Here I0I_0 is the comoving bolometric surface intensity at the disk inner edge rinr_{\rm in}, pp determines the radial decline of Iem(r)I_{\rm em}(r), and the transfer function rm(b)r_m(b) is the radius at which the backward ray first reaches the accumulated angle ψm=(m−12)π\psi_m = (m-\frac{1}{2})\pi for crossing number mm. The factor ZmZ_m is the frequency shift at that crossing, evaluated for the chosen emitter motion. Only intersections inside the emitting disk contribute to the sum. This is a specified emitting-surface model, and no additional incidence-angle factor from a finite-thickness volume model is imposed. We consider both a static disk and a disk whose emitting matter follows circular geodesics. The static emitter uses ZstatZ_{\rm stat} in Eq. (47). For a circular geodesic emitter, let Ωcirc=dϕ/dt\Omega_{\rm circ}=d\phi/dt denote the matter’s orbital angular velocity. Then

Ωcirc2=A′2r,uet=1A−rA′/2,Zcirc=A−rA′/2A0.(52)\Omega_{\rm circ}^{2}=\frac{A'}{2r},\qquad u_{\rm e}^{t}=\frac{1}{\sqrt{A-rA'/2}},\qquad Z_{\rm circ}=\sqrt{\frac{A-rA'/2}{A_0}}. \tag*{(52)}

The expression for ZcircZ_{\rm circ} uses the vanishing photon angular momentum about the disk normal, Lz=0L_z=0, for an observer on the disk axis.

An important concept in circular geodesic motion is the innermost stable circular orbit (ISCO). It marks the inner limit of stable circular motion along the orbit sequence approached from large radius. For timelike circular orbits, A′>0A'>0 and 2A−rA′>02A-rA'>0 are required in the static exterior. Radial stability can be expressed [24] in terms of the stability function

S(r)≡3AA′+rAA′′−2r(A′)2,(53)S(r)\equiv3AA'+rAA''-2r(A')^2, \tag*{(53)}

where timelike circular orbits exist, S>0S>0 indicates radial stability, S<0S<0 indicates radial instability, and S=0S=0 gives the marginally stable radii. The function has dimensions of inverse length, and its relation to the timelike effective potential is given in Appendix § 8.§ 2. When multiple marginally stable circular orbits (MSCOs) separate disconnected stable regions, this prescription selects the outer ISCO, defined as the inner boundary of the stable region continuously connected to spatial infinity, as also adopted [25] in imaging studies of black holes with synchronized scalar hair. We denote this radius by rISCOr_{\rm ISCO} throughout this work, and it need not coincide [22] with the globally innermost stable circular orbit. The disconnected inner stable regions are assumed not to contribute to the emission in our model.

For the static disk model, we set rin=6Mr_{\rm in}=6M for every background. For the rotating disk model, we consider two choices: the same inner edge rin=6Mr_{\rm in}=6M, and an inner edge rin=rISCOr_{\rm in}=r_{\rm ISCO} determined separately for each spacetime. The first choice allows static and rotating emitters to be compared over the same radial interval, while the second incorporates the orbital structure of the background. Both choices use the same comoving intensity I0I_0 at the disk’s own inner edge.

The ISCO prescription makes the disk inner edge sensitive to changes in circular-orbit stability. To examine the origin and mass dependence of the unusually small ISCO at κ=100\kappa=100, we present the stability function SS and the marginally stable orbit radii in Figure 10 along the radially stable branch at fixed λ\lambda. Panel (a) compares S(r)S(r) for four neighboring masses, with an inset resolving the outer local minimum, while panel (b) follows all MSCO radii and identifies the selected disk inner edge.

Figure 10. The figure shows circular-orbit stability and marginal radii as mass varies along the radially stable branch at fixed κ=100\kappa= 100 and λ\lambda. Panel (a) shows the stability function SS defined in Eq. (53), in units of λ−1\lambda^{-1}, over 2≤r/M≤62 \le r/M \le6 for the four adjacent masses listed in the legend. The inset enlarges 4.2≤r/M≤5.64.2 \le r/M \le5.6, where the outer local minimum crosses zero and produces an unstable band bounded by two additional MSCOs. Dots mark the MSCOs. Panel (b) follows all marginal radii along the radially stable background sequence: the green dashed and orange dotted curves denote the inner and middle roots, while the blue curve marks the selected disk inner edge. Open diamonds mark degenerate MSCOs, where two marginal radii merge and satisfy S=∂rS=0S = \partial_r S = 0.

The contrast with larger couplings is substantial: at M/λ=0.06000M/\lambda=0.06000, rISCO/M≃2.7864r_{\rm ISCO}/M\simeq2.7864 for κ=100\kappa=100, compared with approximately 5.7406, 5.9395, and 5.9819 for κ=200\kappa=200, 500, and 1000, respectively. The small ISCO reflects the radial shape of the time-componenent metric function A=−gttA=-g_{tt}, including its derivatives. For example, at r=4Mr=4M the κ=100\kappa=100 value of AA differs from that of Schwarzschild black holes by only about 2.58%, while ∣M2A′′∣|M^2A''| is smaller by about 55.7%. The resulting S>0S>0 permits a stable circular orbit there, whereas S<0S<0 at the same r/Mr/M for Schwarzschild black holes.

In panel (a), the outer local minimum of SS crosses zero as mass increases, creating two additional MSCOs and an intervening interval with S<0S<0. At M/λ≃0.060208M/\lambda\simeq0.060208, stable circular motion extends outward from approximately 2.79M2.79M. At the next background, M/λ≃0.061025M/\lambda\simeq0.061025, the three marginal radii are approximately 2.78M2.78M, 4.68M4.68M, and 5.12M5.12M, and an unstable band separates the inner stable region from the stable region connected to large radius. The selected ISCO consequently moves outward to the outermost marginal radius, because the disk model uses the stable region connected to infinity. Panel (b) follows the resulting change from one to three MSCOs and the subsequent return to one. The inner and middle roots merge and disappear at M/λ≃0.0761276M/\lambda\simeq0.0761276 and r/M≃2.95419r/M \simeq2.95419.

To distinguish the effects of spacetime geometry, emitter motion, and the disk inner edge, we compare the three disk prescriptions in Figure 11 at fixed λ\lambda and M/λ=0.05365M/\lambda= 0.05365. The columns show the radially stable κ=100\kappa= 100 and 10001000 backgrounds alongside Schwarzschild black holes. The first two rows compare static and circular disks with the same inner edge rin=6Mr_{\rm in} = 6M, while the third row places the circular disk’s inner edge at the ISCO of each background. The fourth row gives the corresponding radial intensity profiles, with the same inner-edge intensity I0I_0 and emissivity index p=2p = 2 used throughout the comparison.

Figure 11. The images show face-on thin disks at fixed λ\lambda and M/λ=0.05365M/\lambda= 0.05365, with p=2p = 2 and outer disk radius 30M30M. From left to right, the columns show radially stable κ=100\kappa= 100 and 10001000 backgrounds and Schwarzschild black holes. The first three rows show static and circular disks with rin=6Mr_{\rm in} = 6M, followed by circular disks with rin=rISCOr_{\rm in} = r_{\rm ISCO}. All images share the inner-edge intensity I0I_0 and linear color scale 0≤Io/I0≤0.50 \le I_o/I_0 \le0.5. White dashed circles mark bcb_c. The fourth row shows intensity profiles as gray dotted, blue solid, and red dashed curves, respectively. Vertical dotted lines mark bc/Mb_c/M.

On a given background, replacing static emitters with circularly orbiting emitters at fixed rin=6Mr_{\rm in} = 6M lowers the observed intensity without changing the photon paths or the emitting radial interval. The dimming follows from the smaller frequency-shift factor Zcirc\mathcal{Z}_{\rm circ} in Eq. (52), and is visible in both the first two image rows and the gray dotted and blue solid profiles. Across the three backgrounds, the broad outer emission remains similar, whereas the secondary inner ring lies at smaller b/Mb/M and is broader for κ=100\kappa= 100. The κ=1000\kappa= 1000 images and profiles are close to those for Schwarzschild black holes.

Comparing the second and third rows, we find a much stronger response to the ISCO prescription for κ=100\kappa= 100, where the disk inner edge moves from 6M6M to rISCO≃2.8235Mr_{\rm ISCO} \simeq2.8235M. The inward extension allows captured backward rays to intersect the emitting disk, producing nonzero intensity at b<bcb < b_c, as shown by the red dashed profile in the leftmost column. The image is nevertheless dimmer overall than the circular disk with rin=6Mr_{\rm in} = 6M. Because I0I_0 is fixed at each disk’s own inner edge, reducing rinr_{\rm in} in Eq. (51) lowers the prescribed emission intensity throughout the outer region shared by the two disks. For κ=1000\kappa= 1000, the ISCO remains close to 6M6M, at rISCO/M≃5.9668r_{\rm ISCO}/M \simeq5.9668, so the two circular-disk images differ only slightly and their profiles nearly overlap. For Schwarzschild black holes, rISCO=6Mr_{\rm ISCO} = 6M, and the two circular-disk prescriptions coincide.

To examine how the mass dependence of the ISCO affects the appearance of a circular disk, we present images along the radially stable κ=100\kappa= 100 branch at fixed λ\lambda in Figure 12, together with a reference image for Schwarzschild black holes. The masses M/λ≃0.060208M/\lambda\simeq0.060208 and 0.0610250.061025 lie on opposite sides of the formation of the outer MSCO pair, while M/λ≃0.121345M/\lambda\simeq0.121345 shows the disk at higher mass. Panels (a)–(d) display the images, and panels (e)–(h) give the corresponding radial intensity profiles, with a common inner-edge intensity I0I_0 for all four disks.

Figure 12. The upper row shows images of circular disks with inner edges at the ISCO. Panels (a), (b), and (c) show the κ=100\kappa= 100 backgrounds at fixed λ\lambda and M/λ≃0.060208M/\lambda\simeq0.060208, 0.0610250.061025, and 0.1213450.121345, respectively. Panels (a) and (b) lie on opposite sides of the transition from one to three MSCOs in Figure 10. Panel (d) shows the reference image for Schwarzschild black holes with rISCO=6Mr_{\mathrm{ISCO}} = 6M, which is independent of mass in the displayed b/Mb/M coordinates under the same emission model. All images use the same inner-edge intensity I0I_0, linear color scale 0≤Io/I0≤0.50 \le I_o/I_0 \le0.5, radial emissivity index p=2p = 2, and outer disk radius 30M30M. White dashed circles mark b=bcb = b_c. The lower panels (e)–(h) show the corresponding radial intensity profiles Io(b)/I0I_o(b)/I_0 as blue solid curves on a common vertical scale. Vertical dotted lines mark bc/Mb_c/M. Panels (e) and (f) show that emission at b<bcb < b_c disappears when the disk inner edge set by the ISCO moves outward between the backgrounds in panels (a) and (b), while bc/Mb_c/M changes only slightly.

Comparing panels (a,b) and (e,f), we find that emission at b<bcb < b_c disappears when the ISCO moves outward across the creation of the outer MSCO pair. To explain the loss of interior emission, we compare rISCOr_{\rm ISCO} with the limiting first-crossing radius r1(bc−)=lim⁡b→bc−r1(b)r_1(b_c^-) = \lim_{b \to b_c^-} r_1(b), approached from the captured side. A captured backward ray reaches successive disk crossings at decreasing radii, so the first crossing determines whether that ray can encounter the emitting disk. At M/λ≃0.060208M/\lambda\simeq0.060208, rISCO/M≃2.785<r1(bc−)/M≃3.291r_{\rm ISCO}/M \simeq2.785 < r_1(b_c^-)/M \simeq3.291, allowing captured rays to intersect the emitting disk and produce the nonzero interior intensity in panel (e). At M/λ≃0.061025M/\lambda\simeq0.061025, rISCO/M≃5.118>r1(bc−)/M≃3.312r_{\rm ISCO}/M \simeq5.118 > r_1(b_c^-)/M \simeq3.312, preventing captured rays from intersecting the emitting disk and leaving no resolved interior emission in panel (f). The disappearance of emission at b<bcb < b_c therefore follows from the outward shift of the disk inner edge, while bc/Mb_c/M changes only from about 4.24354.2435 to 4.26454.2645. With I0I_0 fixed at the disk inner edge, the larger rISCO/Mr_{\rm ISCO}/M in panel (b) also increases the prescribed Iem/I0I_{\rm em}/I_0 at a common r/Mr/M within the emitting regions of both disks in panels (a) and (b).

At M/λ≃0.121345M/\lambda\simeq0.121345, we find that the image and radial profile in panels (c) and (g) approach the corresponding results for Schwarzschild black holes in panels (d) and (h). The values rISCO/M≃5.978r_{\rm ISCO}/M \simeq5.978 and bc/M≃5.1763b_c/M \simeq5.1763 are close to 66 and 333\sqrt{3} for Schwarzschild black holes. Both the higher-mass scalarized disk and the disk around Schwarzschild black holes have no resolved emission at b<bcb < b_c in the adopted model.

Conclusions and Outlook

We have investigated the shadows, photon trajectories, and radiative images of static, spherically symmetric black holes with purely nonlinear scalarization in scalar–Gauss–Bonnet gravity. For the quartic exponential coupling, we compared the mass dependence along the κ=100\kappa= 100 and 1000 branches and the coupling dependence at matched ADM masses for κ=100,200,…,1000\kappa= 100, 200, \ldots, 1000, keeping λ\lambda fixed. Using Schwarzschild black holes as a common reference, we obtained geometric and imaging results that distinguish the effects of scalarization, emitter motion, and the disk inner edge.

We find smaller shadows for radially stable scalarized black holes than for Schwarzschild black holes of the same mass throughout the sampled branch intervals. The largest reductions occur at the low-mass ends, where we obtain approximately 21.7% for κ=100\kappa= 100 and 19.3% for κ=1000\kappa= 1000. For both couplings, bc/Mb_{\mathrm{c}}/M increases monotonically with mass toward 333\sqrt{3} while remaining below that value. At matched ADM mass, we also find that the shadow radius increases with κ\kappa across the sampled coupling grid. These results show that the close agreement between images of the κ=1000\kappa= 1000 backgrounds and Schwarzschild black holes in the common mass interval is not representative of the entire stable branch. At lower masses, we obtain appreciable shadow-size deviations even for κ=1000\kappa= 1000.

We find that a monotonic change in shadow size can coexist with nonmonotonic changes in the photon-orbit index χ\chi and photon-ring width Wph/MW_{\mathrm{ph}}/M as mass or coupling varies. At the minimum of χ\chi along the stable κ=100\kappa= 100 branch, we obtain a shadow radius only about 3.55% below the value for Schwarzschild black holes, but an index about 25.9% below unity. The difference between these relative deviations shows that a modest shadow-size difference can accompany a substantially larger change in near-critical photon dynamics. The smaller index implies slower separation of neighboring rays per unit orbital angle and weaker asymptotic compression of successive high-order images toward the critical curve. We also find that the largest photon-ring width need not occur at the smallest χ\chi. The width depends on the accumulated angle along complete photon paths, including propagation away from the photon sphere, whereas χ\chi measures local photon-orbit instability. Our results thus establish the complementary information provided by shadow size, photon-orbit instability, and photon-ring width.

For the radially unstable segments examined here, we obtain much smaller geometric deviations from Schwarzschild black holes. The shadow-size deviation does not exceed approximately 0.34%, and the largest departure of χ\chi from unity is about 1.6%. We find that optical indicators close to those of Schwarzschild black holes coexist with radial instability of the background, so optical similarity alone does not establish dynamical stability. The clearest geometric departures in our sample occur on the radially stable branches.

For spherical emission, we find that radial infall produces a darker center relative to the bright ring than static emission on the same background. In both models, the ring lies near the critical curve and moves outward in units of MM as mass increases along the stable κ=100\kappa= 100 sequence. We obtain different mass dependences of the normalized peak intensity Ipk/(j0M)I_{\mathrm{pk}/}(j_0 M) for the two emitter motions. The static-emission peak varies nonmonotonically and crosses the reference value for Schwarzschild black holes, whereas the infall peak decreases toward its reference value from above. We therefore find that the approach of the shadow radius toward the value for Schwarzschild black holes does not determine how the brightness peak approaches its own reference. The peak comparison uses intensities averaged over annuli of width 0.01M0.01M, since the ideal pointwise intensity diverges logarithmically at the critical curve in the adopted optically thin models.

For face-on thin disks with a common inner edge at 6M6M, we find that circular motion lowers the observed intensity relative to static emission on the same background, without changing the photon paths or emitting radial interval. We obtain a stronger change when the disk inner edge is placed at the ISCO of the low-mass κ=100\kappa= 100 backgrounds, where rISCO/Mr_{\mathrm{ISCO}}/M is close to 2.8. Captured backward rays can then intersect the emitting disk, producing emission inside the geometric shadow boundary at b<bcb < b_{\mathrm{c}}.

Along the stable κ=100\kappa= 100 branch, we further find that the creation of an outer pair of marginally stable circular orbits separates the inner and outer stable regions and changes the disk edge selected by our ISCO prescription. Because the disk occupies the stable region connected to infinity, the selected edge moves outward when the intervening unstable interval appears. For the two representative backgrounds on opposite sides of this transition, we obtain rISCO/M≃2.79r_{\mathrm{ISCO}}/M \simeq2.79 and 5.12. We find that emission at b<bcb < b_{\mathrm{c}} disappears across these backgrounds even though bc/Mb_{\mathrm{c}}/M changes only slightly. The loss of interior emission is therefore caused by the change in the emitting region associated with circular-orbit stability. The accompanying brightness changes also depend on our choice to fix the comoving intensity at each disk’s own inner edge, which changes the prescribed emissivity in the shared outer region when that edge moves.

We conclude that purely nonlinear scalarization can produce distinct changes in shadow size, near-critical photon dynamics, and the orbital structure of emitting matter, despite the absence of a linear tachyonic onset of scalarization in this model. Our imaging results show that the geometric critical curve, bright ring, and extent of the dark central region respond differently to geometry and emission. Interpreting these image features together therefore requires both the metric and the emission prescription. These conclusions apply to the sampled static spherical backgrounds, prescribed optically thin emission with negligible backreaction, and face-on disks.

A natural extension is to determine how the region with multiple marginally stable circular orbits evolves with κ\kappa and how matter populates the disconnected stable regions. An accretion model that includes transport and emission in the plunging region would test whether the loss of interior emission found with a sharp ISCO boundary remains visible when the emitting region is determined dynamically. Varying the disk inclination and including frequency-dependent emission and absorption would further test which features persist beyond the present bolometric, face-on models. Extensions to rotating scalarized backgrounds and stability analyses beyond radial perturbations would broaden the class of physically relevant solutions. Finally, combining more complete emission models with mass and distance information and an observational response would be necessary to assess whether the geometric and radiative differences identified here can be distinguished in black-hole images.

Acknowledgments

The author Hao Yang thanks Li Li and Yiqian Chen for helpful discussions. The author also acknowledges assistance from ChatGPT and Codex in developing the numerical codes, checking analytical derivations, and improving the language and presentation of the manuscript. The author takes full responsibility for all scientific results and the entire content of this paper.

Data and Code Availability

The numerical data and code supporting the results of this study are available from the authors upon reasonable request. The available materials include the black-hole background solutions, optical and intensity-profile data, analysis and plotting scripts, and the associated parameter files and documentation.

Supplementary Analytical Relations

We derive the photon-orbit and timelike-orbit relations used in § 3.4 and § 4.2, retaining the metric and coordinate conventions of § 2.

Photon-orbit instability and asymptotic ring scaling

Consider the nondegenerate maximum of Uopt=A/r2U_{\mathrm{opt}}=A/r^2 at r=rphr=r_{\mathrm{ph}} that controls capture. At this radius, the circular-orbit condition gives

Uopt′′(rph)=rph2A′′(rph)−2A(rph)rph4<0.(54)U_{\mathrm{opt}}^{\prime\prime}(r_{\mathrm{ph}})=\frac{r_{\mathrm{ph}}^2 A^{\prime\prime}(r_{\mathrm{ph}})-2A(r_{\mathrm{ph}})}{r_{\mathrm{ph}}^4}<0. \tag*{(54)}

Expanding Eq. (31) at b=bcb=b_{\mathrm{c}} and retaining the leading radial displacement gives

(dδrdt)2=−A(rph)B(rph)bc22Uopt′′(rph)(δr)2+O((δr)3).(55)\left(\frac{d\delta r}{dt}\right)^2=-\frac{A(r_{\mathrm{ph}})B(r_{\mathrm{ph}})b_{\mathrm{c}}^2}{2}U_{\mathrm{opt}}^{\prime\prime}(r_{\mathrm{ph}})(\delta r)^2+O((\delta r)^3). \tag*{(55)}

The growing solution therefore has the Lyapunov exponent in Eq. (32), in agreement with the geodesic stability analysis of Ref. [17]. Using bc2=rph2/A(rph)b_{\mathrm{c}}^2=r_{\mathrm{ph}}^2/A(r_{\mathrm{ph}}) and Ωph=1/bc\Omega_{\mathrm{ph}}=1/b_{\mathrm{c}} then gives Eq. (34). The factor B(rph)B(r_{\mathrm{ph}}) enters the instability rate even though the photon-orbit condition itself depends only on AA.

To obtain the logarithmic growth of the accumulated angle, take a scattered ray with ϵ=b/bc−1>0\epsilon=b/b_{\mathrm{c}}-1>0 and introduce

xph≡r−rphrph,αph≡−bc2rph22Uopt′′(rph)=χ2B(rph)>0.(56)x_{\mathrm{ph}}\equiv\frac{r-r_{\mathrm{ph}}}{r_{\mathrm{ph}}},\qquad\alpha_{\mathrm{ph}}\equiv-\frac{b_{\mathrm{c}}^2r_{\mathrm{ph}}^2}{2}U_{\mathrm{opt}}^{\prime\prime}(r_{\mathrm{ph}})=\frac{\chi^2}{B(r_{\mathrm{ph}})}>0. \tag*{(56)}

Here xphx_{\mathrm{ph}} is the fractional radial displacement, and αph\alpha_{\mathrm{ph}} is the dimensionless curvature coefficient of the capture barrier. The radial factor in the geodesic equation becomes

1−b2Uopt(r)=−2ϵ+αphxph2+O(ϵ2,ϵxph2,xph3).(57)1-b^2U_{\mathrm{opt}}(r)=-2\epsilon+\alpha_{\mathrm{ph}}x_{\mathrm{ph}}^2+O(\epsilon^2,\epsilon x_{\mathrm{ph}}^2,x_{\mathrm{ph}}^3). \tag*{(57)}

Thus the outer turning point satisfies (rt−rph)/rph≃2ϵ/αph(r_{\mathrm{t}}-r_{\mathrm{ph}})/r_{\mathrm{ph}}\simeq\sqrt{2\epsilon/\alpha_{\mathrm{ph}}}. Within the same approximation, Eq. (40) gives

∣dψdxph∣≃1B(rph)αphxph2−2ϵ.(58)\left|\frac{d\psi}{dx_{\mathrm{ph}}}\right|\simeq\frac{1}{\sqrt{B(r_{\mathrm{ph}})}\sqrt{\alpha_{\mathrm{ph}}x_{\mathrm{ph}}^2-2\epsilon}}. \tag*{(58)}

Choose a dimensionless matching radius ℓph\ell_{\mathrm{ph}} satisfying 2ϵ/αph≪ℓph≪1\sqrt{2\epsilon/\alpha_{\mathrm{ph}}}\ll\ell_{\mathrm{ph}}\ll1. It separates the near-orbit region from the rest of the path and has no physical significance. Including both the incoming and outgoing segments, the quadratic approximation to the near-orbit angular contribution is

Δψnear(2)=2χarcosh⁡(αphℓph2ϵ)=−1χln⁡ϵ+2χln⁡(2αphℓph)+O(ϵαphℓph2).(59)\begin{aligned} \Delta\psi_{\mathrm{near}}^{(2)}=\frac{2}{\chi}\operatorname{arcosh}\left(\frac{\sqrt{\alpha_{\mathrm{ph}}}\ell_{\mathrm{ph}}}{\sqrt{2\epsilon}}\right) \\ &=-\frac{1}{\chi}\ln\epsilon+\frac{2}{\chi}\ln\left(\sqrt{2\alpha_{\mathrm{ph}}}\ell_{\mathrm{ph}}\right)+O\left(\frac{\epsilon}{\alpha_{\mathrm{ph}}\ell_{\mathrm{ph}}^2}\right). \tag*{(59)} \end{aligned}

The superscript (2) denotes the quadratic expansion of the radial factor. Matching this contribution to the remaining path recovers [23] Eq. (35). For a ray whose two endpoints are at infinity, the finite constant can be written as

C=lim⁡ℓph→0+[2∫rph(1+ℓph)∞bcA/Br21−bc2Uopt dr+2χln⁡(2αphℓph)].(60)C=\lim_{\ell_{\mathrm{ph}}\to0^+}\left[2\int_{r_{\mathrm{ph}}(1+\ell_{\mathrm{ph}})}^\infty\frac{b_{\mathrm{c}}\sqrt{A/B}}{r^2\sqrt{1-b_{\mathrm{c}}^2U_{\mathrm{opt}}}}\,dr+\frac{2}{\chi}\ln\left(\sqrt{2\alpha_{\mathrm{ph}}}\ell_{\mathrm{ph}}\right)\right]. \tag*{(60)}

The integral contains the opposite logarithmic dependence on ℓph\ell_{\mathrm{ph}}, so the limit is finite and independent of the arbitrary matching radius. Finite source and observer radii outside the photon sphere change the regular contribution, but leave the coefficient 1/χ1/\chi unchanged when the endpoints remain away from the orbit. For endpoints at infinity, the deflection angle is Δψ−π\Delta\psi-\pi, so subtracting π\pi changes only the finite constant.

The same logarithm determines the asymptotic widths discussed in § 3.4. Consider a sequence of high-order scattered images of a fixed emitting annulus outside the photon sphere, with successive orders adding an angle π\pi as in the face-on disk geometry. Let bn(1)b_n^{(1)} and bn(2)b_n^{(2)} denote the two impact-parameter boundaries of its nnth image. The finite path contributions at each source boundary are independent of nn in the critical limit. Inverting the logarithmic relation therefore gives

bn(i)−bc=Kie−πχn[1+o(1)],i=1,2,(61)b_n^{(i)}-b_c=K_i e^{-\pi\chi n}[1+o(1)],\qquad i=1,2, \tag*{(61)}

where KiK_i are constants with dimensions of length, fixed by the source boundaries and the observer geometry, and o(1)o(1) now denotes terms vanishing as n→∞n\to\infty. For distinct boundaries, the image width is consequently

wn≡∣bn(2)−bn(1)∣=∣K2−K1∣e−πχn[1+o(1)].(62)w_n\equiv\left|b_n^{(2)}-b_n^{(1)}\right|=\left|K_2-K_1\right|e^{-\pi\chi n}[1+o(1)]. \tag*{(62)}

This yields [18] the half-orbit compression factor e−πχe^{-\pi\chi} in Eq. (36). If successive orders instead differ by a full orbit, the corresponding factor is e−2πχe^{-2\pi\chi}. These are asymptotic relations for geometric image intervals and do not specify their brightness or flux.

Timelike circular orbits and disk inner edges

For a minimally coupled timelike test particle, let Et=A dt/dτE_t=A\,dt/d\tau and Lt=r2dϕ/dτL_t=r^2d\phi/d\tau denote the conserved energy and angular momentum per unit rest mass, where τ\tau is proper time. The normalization of the four-velocity gives

AB(drdτ)2+Vt(r;Lt)=Et2,Vt(r;Lt)=A(1+Lt2r2).(63)\frac{A}{B}\left(\frac{dr}{d\tau}\right)^2+V_t(r;L_t)=E_t^2,\qquad V_t(r;L_t)=A\left(1+\frac{L_t^2}{r^2}\right). \tag*{(63)}

Here VtV_t is the timelike effective potential at fixed angular momentum. A circular orbit requires Vt=Et2V_t=E_t^2 and ∂rVt=0\partial_rV_t=0. Solving these conditions gives

Et,c2=2A22A−rA′,Lt,c2=r3A′2A−rA′.(64)E_{t,c}^2=\frac{2A^2}{2A-rA'},\qquad L_{t,c}^2=\frac{r^3A'}{2A-rA'}. \tag*{(64)}

The subscript cc denotes values on the circular-orbit family. In the static exterior, the rotating timelike orbits used for the disk require A′>0A'>0 and 2A−rA′>02A-rA'>0. Their angular velocity and time component of the four-velocity satisfy

Ωc2=(ALt,cr2Et,c)2=A′2r,uet=Et,cA=1A−rA′/2,(65)\Omega_c^2=\left(\frac{AL_{t,c}}{r^2E_{t,c}}\right)^2=\frac{A'}{2r},\qquad u_e^t=\frac{E_{t,c}}{A}=\frac{1}{\sqrt{A-rA'/2}}, \tag*{(65)}

recovering the kinematic factors in Eq. (52).

Radial stability follows from the curvature of VtV_t while holding LtL_t fixed. The angular momentum is set equal to Lt,cL_{t,c} only after differentiating the potential. Using the stability function SS defined in Eq. (53), we obtain

∂2Vt∂r2∣Lt=Lt,c=2Sr(2A−rA′),dLt,c2dr=2r2S(2A−rA′)2.(66)\left.\frac{\partial^2V_t}{\partial r^2}\right|_{L_t=L_{t,c}}=\frac{2S}{r(2A-rA')},\qquad\frac{dL_{t,c}^2}{dr}=\frac{2r^2S}{(2A-rA')^2}. \tag*{(66)}

The second expression differentiates the circular-orbit family and shows [24] why the angular-momentum and effective-potential criteria identify the same marginal radii. To express stability as a dynamical response, let δrt\delta r_t be a small radial displacement of the particle and define Ωr\Omega_r as its radial oscillation frequency measured in tt. Linearizing the radial equation and using dt/dτ=Et,c/Adt/d\tau= E_{t,c}/A gives

d2δrtdt2+Ωr2δrt=0,Ωr2=AB2Et,c2∂2Vt∂r2∣Lt=Lt,c=BS2Ar.(67)\frac{d^2\delta r_t}{dt^2}+\Omega_r^2\delta r_t=0,\qquad\Omega_r^2=\left.\frac{AB}{2E_{t,c}^2}\frac{\partial^2 V_t}{\partial r^2}\right|_{L_t=L_{t,c}}=\frac{BS}{2Ar}. \tag*{(67)}

All coefficients are evaluated at the unperturbed circular radius. Positive Ωr2\Omega_r^2 describes radial oscillations, while negative Ωr2\Omega_r^2 describes exponential departure from the orbit. Since A,B>0A,B>0, the sign is fixed by SS. Thus AA and its derivatives determine the marginal radii, whereas BB also affects the oscillation or growth rate. For Schwarzschild black holes, these expressions reduce to

SSch=2M(r−6M)r4,Ωr,Sch2=Mr3(1−6Mr),(68)S_{\mathrm{Sch}}=\frac{2M(r-6M)}{r^4},\qquad\Omega_{r,\mathrm{Sch}}^2=\frac{M}{r^3}\left(1-\frac{6M}{r}\right), \tag*{(68)}

recovering stable circular orbits for r>6Mr>6M and marginal stability at 6M6M.

When several physical roots of S=0S=0 exist, their locations alone do not specify the disk inner edge. One must also identify which intervals admit timelike circular orbits and have S>0S>0. As discussed [22] for disconnected stable-orbit regions, the innermost stable orbit need not bound the stable region connected to infinity. Our prescription in § 4.2 uses the inner boundary of that outer region and leaves any disconnected inner stable region without emission.

The appearance of an additional unstable interval can be understood directly from a local expansion. Let (r∗,M∗)(r_*,M_*) identify a degenerate MSCO along a smooth family at fixed κ\kappa and λ\lambda, satisfying S=∂rS=0S=\partial_r S=0. Near a generic double root,

S(r,M/λ)=aMSM−M∗λ+cMS(r−r∗)2+⋯ ,aMScMS≠0,(69)S(r,M/\lambda)=a_{\mathrm{MS}}\frac{M-M_*}{\lambda}+c_{\mathrm{MS}}(r-r_*)^2+\cdots,\qquad a_{\mathrm{MS}}c_{\mathrm{MS}}\ne0, \tag*{(69)}

where aMS=∂S/∂(M/λ)a_{\mathrm{MS}}=\partial S/\partial(M/\lambda) and cMS=12∂r2Sc_{\mathrm{MS}}=\frac{1}{2}\partial_r^2S are evaluated at the degenerate orbit. The derivatives are taken at fixed areal radius and fixed mass, respectively, with the coupling parameters held constant. The two nearby marginal radii, denoted by rMS,±r_{\mathrm{MS},\pm}, therefore satisfy to leading order

rMS,±−r∗≃±−aMScMSM−M∗λ,(70)r_{\mathrm{MS},\pm}-r_*\simeq\pm\sqrt{-\frac{a_{\mathrm{MS}}}{c_{\mathrm{MS}}}\frac{M-M_*}{\lambda}}, \tag*{(70)}

on the side of the transition where the radicand is positive. For cMS>0c_{\mathrm{MS}}>0, the two roots enclose an interval with S<0S<0. If this interval forms within the previously connected outer stable region, the adopted disk inner edge switches to the larger new root. The selected ISCO can consequently change discontinuously even when the metric varies smoothly with mass. The later merger of the inner and middle roots need not produce another jump, because the outer stable region can retain the same boundary. These relations explain the changes of orbital structure in Figure 10 without requiring a discontinuity or singularity of the metric. The critical masses reported in § 4.2 are obtained from the simultaneous conditions S=∂rS=0S=\partial_rS=0, rather than from the separation of the representative plotted masses.

References

  1. [1]Event Horizon Telescope Collaboration. First M87 Event Horizon Telescope results. I. the shadow of the supermassive black hole. The Astrophysical Journal Letters, 875:L1, 2019. doi: 10.3847/2041-8213/ab0ec7.DOI
  2. [2]Event Horizon Telescope Collaboration. First Sagittarius A* Event Horizon Telescope results. I. the shadow of the supermassive black hole in the center of the Milky Way. The Astrophysical Journal Letters, 930:L12, 2022. doi: 10.3847/2041-8213/ac6674.DOI
  3. [3]Event Horizon Telescope Collaboration. The persistent shadow of the supermassive black hole of M 87. I. observations, calibration, imaging, and analysis. Astronomy & Astrophysics, 681:A79, 2024. doi: 10.1051/0004-6361/202347932.
  4. [4]Event Horizon Telescope Collaboration. First Sagittarius A* Event Horizon Telescope results. VII. polarization of the ring. The Astrophysical Journal Letters, 964:L25, 2024. doi: 10.3847/2041-8213/ad2df0.
  5. [5]Event Horizon Telescope Collaboration. Horizon-scale variability of M87* from 2017–2021 EHT observations. Astronomy & Astrophysics, 704:A91, 2025. doi: 10.1051/0004-6361/202555855.DOI
  6. [6]Volker Perlick and Oleg Yu. Tsupko. Calculating black hole shadows: Review of analytical studies. Physics Reports, 947:1–39, 2022. doi: 10.1016/j.physrep.2021.10.004.DOI
  7. [7]Samuel E. Gralla, Daniel E. Holz, and Robert M. Wald. Black hole shadows, photon rings, and lensing rings. Physical Review D, 100:024018, 2019. doi: 10.1103/PhysRevD.100.024018.DOI
  8. [8]G. Antoniou, A. Bakopoulos, and P. Kanti. Evasion of no-hair theorems and novel black-hole solutions in Gauss-Bonnet theories. Physical Review Letters, 120:131102, 2018. doi: 10.1103/PhysRevLett.120.131102.DOI
  9. [9]Daniela D. Doneva and Stoytcho S. Yazadjiev. New Gauss-Bonnet Black Holes with Curvature-Induced Scalarization in Extended Scalar-Tensor Theories. Phys. Rev. Lett., 120(13):131103, 2018. doi: 10.1103/PhysRevLett.120.131103.DOI
  10. [10]Hector O. Silva, Jeremy Sakstein, Leonardo Gualtieri, Thomas P. Sotiriou, and Emanuele Berti. Spontaneous scalarization of black holes and compact stars from a Gauss-Bonnet coupling. Physical Review Letters, 120:131104, 2018. doi: 10.1103/PhysRevLett.120.131104.DOI
  11. [11]Daniela D. Doneva, Fethi M. Ramazanoğlu, Hector O. Silva, Thomas P. Sotiriou, and Stoytcho S. Yazadjiev. Spontaneous scalarization. Reviews of Modern Physics, 96:015004, 2024. doi: 10.1103/RevModPhys.96.015004.DOI
  12. [12]Daniela D. Doneva and Stoytcho S. Yazadjiev. Beyond the spontaneous scalarization: New fully nonlinear mechanism for the formation of scalarized black holes and its dynamical development. Physical Review D, 105(4):L041502, 2022. doi: 10.1103/PhysRevD.105.L041502.DOI
  13. [13]Jose Luis Blázquez-Salcedo, Daniela D. Doneva, Jutta Kunz, and Stoytcho S. Yazadjiev. Radial perturbations of scalar-Gauss-Bonnet black holes beyond spontaneous scalarization. Physical Review D, 105:124005, 2022. doi: 10.1103/PhysRevD.105.124005.
  14. [14]Pedro V. P. Cunha, Carlos A. R. Herdeiro, and Eugen Radu. Spontaneously scalarized Kerr black holes in extended scalar-tensor–Gauss-Bonnet gravity. Physical Review Letters, 123:011101, 2019. doi: 10.1103/PhysRevLett.123.011101.DOI
  15. [15]Pedro G. S. Fernandes, Clare Burrage, Astrid Eichhorn, and Thomas P. Sotiriou. Shadows and properties of spin-induced scalarized black holes with and without a Ricci coupling. Physical Review D, 109:104033, 2024. doi: 10.1103/PhysRevD.109.104033.DOI
  16. [16]Xi-Jing Wang, Yuan Meng, Xiao-Mei Kuang, and Kai Liao. Distinguishing black holes with and without spontaneous scalarization in Einstein-scalar-Gauss-Bonnet theories via optical features. The European Physical Journal C, 84:1243, 2024. doi: 10.1140/epjc/s10052-024-13612-w.DOI
  17. [17]Vitor Cardoso, Alex S. Miranda, Emanuele Berti, Helvi Witek, and Vilson T. Zanchin. Geodesic stability, Lyapunov exponents and quasinormal modes. Physical Review D, 79:064016, 2009. doi: 10.1103/PhysRevD.79.064016.DOI
  18. [18]Prashant Kocherlakota, Luciano Rezzolla, Rittick Roy, and Maciek Wielgus. Prospects for future experimental tests of gravity with black hole imaging: Spherical symmetry. Physical Review D, 109:064064, 2024. doi: 10.1103/PhysRevD.109.064064.DOI
  19. [19]Michael D. Johnson, Alexandru Lupsasca, Andrew Strominger, George N. Wong, Shahar Hadar, Daniel Kapec, Ramesh Narayan, Andrew Chael, Charles F. Gammie, Peter Galison, Daniel C. M. Palumbo, Sheperd S. Doeleman, Lindy Blackburn, Maciek Wielgus, Dominic W. Pesce, Joseph R. Farah, and James M. Moran. Universal interferometric signatures of a black hole’s photon ring. Science Advances, 6:eaaz1310, 2020. doi: 10.1126/sciadv.aaz1310.DOI
  20. [20]Ramesh Narayan, Michael D. Johnson, and Charles F. Gammie. The shadow of a spherically accreting black hole. The Astrophysical Journal Letters, 2019. doi: 10.3847/2041-8213/ab518c.DOI
  21. [21]Prashant Kocherlakota and Luciano Rezzolla. Distinguishing gravitational and emission physics in black hole imaging: spherical symmetry. Monthly Notices of the Royal Astronomical Society, 513(1):1229–1243, 2022. doi: 10.1093/mnras/stac891.DOI
  22. [22]Jorge F. M. Delgado, Carlos A. R. Herdeiro, and Eugen Radu. Equatorial timelike circular orbits around generic ultracompact objects. Physical Review D, 105:064026, 2022. doi: 10.1103/PhysRevD.105.064026.DOI
  23. [23]Valerio Bozza. Gravitational lensing in the strong field limit. Physical Review D, 66:103001, 2002. doi: 10.1103/PhysRevD.66.103001.DOI
  24. [24]Yong Song. The evolutions of the innermost stable circular orbits in dynamical spacetimes. European Physical Journal C, 81:875, 2021. doi: 10.1140/epjc/s10052-021-09623-6.DOI
  25. [25]Valentin O. Deliyski, Galin N. Gyulchev, Daniela D. Doneva, Petya G. Nedkova, and Stoytcho S. Yazadjiev. Polarized equatorial emission around Kerr black holes with synchronized scalar hair. I. direct images. European Physical Journal C, 86:908, 2026. doi: 10.1140/epjc/s10052-026-16103-2.DOI

Paper details

vlnc:W_fprp55r6
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

ChatGPT, Codex

View statement
Looking for related papers…
Contents