Massive Boson Quasi-Bound States

A massive bosonic field can confine itself around a rotating black hole. Its mass makes low-frequency solutions evanescent at infinity, while the horizon remains a one-way inner boundary. The resulting quasi-bound frequency is complex: its real part is approximately hydrogenic when the gravitational coupling is small. A matched approximation gives growth in the superradiant band; this leading sign rule is not an exact flux theorem for arbitrary complex frequencies. The derivation below concerns a minimally coupled neutral scalar first; vector fields require their own constrained system and are identified explicitly rather than folded into the scalar formula.

Helpful background. Black-Hole Geometry and Mechanics supplies Kerr horizon data; PDEs and Green Functions for Gravitation supplies matched asymptotics and complex eigenvalues; Formulations of General Relativity supplies covariant massive-field equations; Relativistic Matter and Compact Objects supplies bosonic field models; Black-Hole Perturbation Theory supplies separation and endpoint conditions.

Quasi-bound boundary conditions

Denote the boson mass by $\mu_b$, the gravitational length by $g=GM$, and the dimensionless spin by $\chi=\chi_{\mathrm{BH}}$. Define

$$\alpha=GM\mu_b, \qquad \chi_{\mathrm{BH}}=\frac{J}{GM^2}.$$

For a scalar mode $\Phi=e^{-i\omega t+im\phi}S_{\ell m}(\theta)R_{\ell m}(r)$, a quasi-bound state obeys a future-ingoing condition at the horizon and an exponentially decaying condition at spatial infinity:

$$R\sim e^{-i(\omega-m\Omega_H)r_*}\quad(r\to r_+), \qquad R\sim r^{\nu_\infty-1}e^{-kr}\quad(r\to\infty),$$

where

$$k=\sqrt{\mu_b^2-\omega^2},\qquad \operatorname{Re}k\gt 0,\qquad \nu_\infty=\frac{g(2\omega^2-\mu_b^2)}{k}.$$

We use the positive-frequency weak-binding branch $0\lt\omega_R\lt\mu_b$ with small $\lvert\omega_I\rvert$. Because energy can cross the horizon, this is not a normal mode of a self-adjoint closed system; $\omega=\omega_R+i\omega_I$ is generally complex. With the stated Fourier convention, $\omega_I\gt 0$ means growth.

The two endpoint conditions are both essential. A local minimum of a radial effective potential does not by itself establish a quasi-bound eigenvalue. One must solve the angular and radial spectral conditions together and impose the physical branches at both endpoints. The scalar, vector, and tensor massive-field spectra have different separation structures, so a scalar potential sketch cannot determine a vector growth rate.

Hydrogenic real frequency

When $\alpha\ll1$, the cloud is much larger than the black hole and the far equation reduces at leading order to a Coulomb problem. Write $\omega_R=\mu_b+E$ with $\lvert E\rvert\ll\mu_b$. Expanding $\omega_R^2-\mu_b^2\simeq2\mu_b E$ and keeping the Newtonian potential gives the Schrödinger equation

$$\left[-\frac{\nabla^2}{2\mu_b}-\frac{GM\mu_b}{r}\right]\psi =E\psi.$$

Its principal quantum number is $N=n_r+\ell+1$, where $n_r=0,1,\ldots$, and therefore

$$E_N=-\frac{\mu_b\alpha^2}{2N^2}, \qquad \boxed{\omega_R=\mu_b \left(1-\frac{\alpha^2}{2N^2}+O(\alpha^4)\right)}.$$

The characteristic cloud radius is

$$r_{\mathrm{cloud}}\sim N^2a_g, \qquad a_g=\frac{1}{\mu_b\alpha}=\frac{GM}{\alpha^2}.$$

Thus the same small parameter that controls the hydrogenic expansion also separates the cloud from the horizon. Rotation first lifts degeneracies at higher order and controls the much smaller imaginary part. The hydrogenic approximation and its range of validity are developed in Brito, Cardoso, and Pani 2020, §§5.4–5.5, pp. 99–116.

A scalar near–far connection

We calculate $\ell=m=1$, $n_r=0$. Hold the subextremal spin fixed and impose $\alpha\ll\sqrt{1-\chi^2}$; do not take an extremal limit or approach synchronization within the neglected corrections. These are sufficient conditions for the following leading Detweiler matching approximation, not a percentage-error estimate for the full eigenvalue.

The exact separated radial equation for the raw scalar field is

$$\begin{aligned} &\frac{d}{dr}\left(\Delta\frac{dR}{dr}\right) +\left[\frac{K^2}{\Delta}-\mu_b^2r^2-a^2\omega^2 +2am\omega-A_{\ell m}\right]R=0,\\ &\Delta=(r-r_+)(r-r_-),\qquad K=(r^2+a^2)\omega-am . \end{aligned}$$

Here $a=\chi g$ and the massive scalar angular equation has deformation $a^2(\omega^2-\mu_b^2)$, not the massless spheroidal parameter alone. Its leading eigenvalue is $\ell(\ell+1)$. Set

$$d=r_+-r_-,\qquad z=\frac{r-r_+}{d},\qquad P=\frac{am-2gr_+\omega}{d} =\frac{2gr_+}{d}(m\Omega_H-\omega).$$

Dropping the small mass and frequency terms except in the horizon exponent gives the near equation and its future-ingoing solution

$$\begin{aligned} &\frac{d}{dz}\left[z(z+1)\frac{dR_N}{dz}\right] +\left[\frac{P^2}{z(z+1)}-\ell(\ell+1)\right]R_N=0,\\ &R_N=\left(\frac{z}{1+z}\right)^{iP} {}_2F_1(-\ell,\ell+1;1+2iP;-z). \end{aligned}$$

Indeed $R_N\sim z^{iP}$ at the horizon, with exactly the required ingoing exponent. At $\ell=1$ its hypergeometric factor is $1+2z/(1+2iP)$.

For the far field write $v=rR$. The leading Coulomb equation is

$$v''+\left[-k^2+\frac{2k\nu}{r} -\frac{\ell(\ell+1)}{r^2}\right]v=0,\qquad \nu=\frac{g\mu_b^2}{k}.$$

This $\nu$ approximates, but is not identical to, $\nu_\infty$. The decaying solution is the confluent hypergeometric function

$$R_F=(2kr)^\ell e^{-kr} U(\ell+1-\nu,2\ell+2,2kr).$$

The overlap must also lie outside the large-$z$ scale set by $P$: $g\max(1,\lvert P\rvert,\ell)\ll r-r_+\ll g/\alpha$. At fixed spin the stated condition on $\alpha$ leaves such a region. These near and far reductions follow Detweiler 1980, Eqs. (6)–(24), pp. 2324–2325.

The coefficient that determines growth

Integer powers are resonant in the connection formula. Temporarily put $\ell'=\ell+\epsilon$ and expand each solution as $A r^{\ell'}+B r^{-\ell'-1}$ in the overlap. The coefficients are

$$\begin{aligned} A_N&=d^{-\ell'} \frac{\Gamma(1+2iP)\Gamma(2\ell'+1)} {\Gamma(\ell'+1)\Gamma(\ell'+1+2iP)},\\ B_N&=d^{\ell'+1} \frac{\Gamma(1+2iP)\Gamma(-2\ell'-1)} {\Gamma(-\ell')\Gamma(-\ell'+2iP)},\\ A_F&=(2k)^{\ell'} \frac{\Gamma(-2\ell'-1)}{\Gamma(-\ell'-\nu)},\\ B_F&=(2k)^{-\ell'-1} \frac{\Gamma(2\ell'+1)}{\Gamma(\ell'+1-\nu)}. \end{aligned}$$

The arbitrary overall normalization cancels from $B/A$. First expand in the small spectral shift $\nu=\ell'+1+\delta\nu$, then take $\epsilon\to0$ at $\ell=1$. In particular,

$$\lim_{\epsilon\to0} \frac{\Gamma(-3-2\epsilon)}{\Gamma(-1-\epsilon)} =\frac1{12},\qquad \frac1{\Gamma(-\delta\nu)}=-\delta\nu+O(\delta\nu^2).$$

Consequently the leading matching equation is

$$\frac{B_N}{A_N} =-\frac{iP(1+4P^2)d^3}{12} ,\qquad \frac{B_F}{A_F} \simeq-\frac{2\delta\nu}{(2k)^3},$$

and hence

$$\delta\nu\simeq\frac{iP(1+4P^2)(2kd)^3}{24}.$$

The regulated Gamma ratio is the factor-of-two correction discussed in Bao, Xu, and Zhang 2022, Eqs. (12)–(15), p. 3, and Appendix A, p. 6, PDF. It is separate from the factor of two between amplitude and energy growth.

Using $N=2$, $k\simeq\alpha^2/(2g)$ and $d\omega/d\nu\simeq\mu_b\alpha^2/N^3$ gives

$$\begin{aligned} g\omega_I&\simeq \frac{\alpha^9}{48}\,Q(1-\chi^2+Q^2),\\ Q&=\chi-2r_+\omega_R\qquad(m=1). \end{aligned}$$

The bracket is positive for a subextremal hole. Thus the leading matching rate has the sign of $m\Omega_H-\omega_R$. For fixed positive $\chi$ its leading small-$\alpha$ coefficient is $\chi\alpha^9/48$. This statement belongs to the matching approximation; it is neither a uniform near-extremal result nor an independently proved full-eigenvalue asymptotic error bound. Near synchronization neglected terms must be reconsidered. The energy growth rate is $\Gamma_E=2\omega_I$, with coefficient $\chi\alpha^9/(24g)$ in the same limit.

Changing the amplitude of $R$ cannot change either eigenfrequency coefficient. It multiplies horizon flux and cloud energy by the same squared amplitude, which cancels in their ratio. In the Schwarzschild limit the signed $Q$ is negative for this positive-frequency family. Taking $\mu_b\to0$ instead sends the cloud radius to infinity: it does not leave a finite mass-confined cavity.

Complex frequency and horizon energy

The real-frequency flux rule needs a qualification for growing modes. In regular ingoing coordinates let the real field at the horizon be $\operatorname{Re}[H S(\theta)e^{im\tilde\phi-i\omega v}]$. For $m=1$ and $\int |S e^{im\tilde\phi}|^2d\Omega=1$, integration over the azimuth removes the twice-frequency terms. The Killing-energy flux into the hole is

$$\mathcal F_H^{\mathrm{into}} =\frac{r_+^2+a^2}{2}|H|^2e^{2\omega_I v} \left[\omega_R(\omega_R-m\Omega_H)+\omega_I^2\right].$$

This follows directly from $\operatorname{Re}[\bar\omega(\omega-m\Omega_H)]$ in $T_{ab}(\partial_v)^a(\partial_v+\Omega_H\partial_{\tilde\phi})^b$. The flux toward increasing radius is its negative. In particular negative energy influx requires $\omega_R(m\Omega_H-\omega_R)\gt \omega_I^2$, not merely a real-frequency inequality applied unchanged to a complex mode.

For a horizon-penetrating stationary family of spacelike slices, integrate the conserved Killing current with the induced slice measure and outward boundary normals over the intervening exterior slab. A decaying bound field has no flux at infinity; its energy change is minus the integrated flux into the hole. If the cloud has positive Killing energy dominated by the weak-binding far region, a mode therefore satisfies $2\omega_I E_{\mathrm{cloud}}=-\mathcal F_H^{\mathrm{into}}$ at matched slice time. This is not the general positivity of canonical energy in the ergoregion. The local horizon contraction alone neither proves that positivity nor constructs a global eigenmode. In the weak-growth expansion $\omega_I^2$ is beyond the retained order, explaining the leading sign rule without promoting it to an exact equivalence.

Vector fields and the near-threshold regime

A massive vector obeys the Proca equations and carries longitudinal as well as transverse degrees of freedom. Its Kerr reconstruction and separated equations are distinct from the scalar system; neither its angular labels nor its rates follow by substituting $s=1$ into a scalar result. The principal-tensor construction is given in Frolov, Krtouš, Kubizňák, and Santos 2018, §III, pp. 2–3, PDF.

Dolan recovers all three Kerr Proca polarization branches and compares their different quasi-bound growth rates in Dolan 2018, §V, pp. 18–19, PDF. Recovery of those branches does not itself prove completeness for arbitrary initial data. The field constraints, Schwarzschild low-multipole exceptions and singular separation-parameter limits are explained in Massive Vector and Higher-Spin Test Fields.

When $\alpha$ is not small, the overlap $GM\ll r-r_+\ll\mu_b^{-1}$ collapses. Near the synchronization surface $\omega_R=m\Omega_H$, the leading imaginary part is also small, so neglected higher-order terms can compete with it. Reliable work there requires solving the full spectral boundary-value problem and checking angular truncation, radial endpoints, numerical precision, and the sign of the horizon flux. The numerical and analytic regimes are compared in Brito, Cardoso, and Pani 2020, §§5.5–5.6, pp. 108–128.

The linear quasi-bound calculation establishes an early-time exponential rate on a fixed Kerr background. It does not determine the cloud’s saturation, the spin-down trajectory, gravitational-wave emission, self-interaction, or backreaction. Environmental changes to trapping are treated on Plasma, Matter, and Environmental Trapping.

Exercises

Take a neutral scalar with $\alpha=0.10$, $n_r=0$, and $\ell=m=1$ around a Kerr black hole with $\chi_{\mathrm{BH}}=0.9$, for which $GM\Omega_H=0.3133945$. Estimate $\omega_R/\mu_b$ and $r_{\mathrm{cloud}}/(GM)$, then decide the leading sign of $\omega_I$.

Solution

Here $N=n_r+\ell+1=2$. The hydrogenic formula gives

$$\frac{\omega_R}{\mu_b} =1-\frac{0.10^2}{2(2)^2} =0.99875.$$

Because $GM\mu_b=0.10$, $GM\omega_R=0.099875$. This is below $mGM\Omega_H=0.3133945$, so the signed matching factor is positive and $\omega_I\gt 0$ at leading order. The radius estimate is

$$\frac{r_{\mathrm{cloud}}}{GM} \sim\frac{N^2}{\alpha^2} =\frac{4}{0.01}=400.$$

The large radius supports the weak-binding picture. This exercise uses only the leading real frequency and sign; the coefficient derived above still does not supply a controlled percentage error at these finite parameters. A precise e-folding time requires testing the full spectral calculation.

References

• Bao, S., Xu, Q., and Zhang, H. (2022). “Improved analytic solution of black hole superradiance.” Physical Review D 106, 064016. DOI. Open PDF.
• Brito, R., Cardoso, V., and Pani, P. (2020). Superradiance: New Frontiers in Black Hole Physics. Lecture Notes in Physics 971. Springer. DOI.
• Detweiler, S. (1980). “Klein–Gordon equation and rotating black holes.” Physical Review D 22, 2323–2326. DOI.
• Dolan, Sam R. “Instability of the Proca Field on Kerr Spacetime.” Physical Review D 98 (2018): 104006. DOI. Open PDF.
• Frolov, Valeri P., Pavel Krtouš, David Kubizňák, and Jorge E. Santos. “Massive Vector Fields in Rotating Black-Hole Spacetimes: Separability and Quasinormal Modes.” Physical Review Letters 120 (2018): 231103. DOI. Open PDF.