Ultralight boson constraints from gravitational wave observations of spinning binary black holes

TL;DR

This paper directly constrains ultralight bosons through black-hole superradiance by comparing the measured spins of two binary BH mergers, GW231123 and GW190517, against the maximum allowed spins set by a boson mass via the $ om{superradiance}$ instability. Using the SuperRad framework to compute $\,\chi_{ m max}(M,m_b,T_{ m age})$ over a conservative age range $T_{ m age}=10^5$–$10^7$ years, and employing posterior samples from multiple waveform models, the authors derive 90% CL exclusions: scalars in $[0.55,11]\times10^{-13}$ eV and vectors in $[0.11,18]\times10^{-13}$ eV (for $T_{ m age}=10^5$ years). The constraints strengthen for older ages and with inclusion of the secondary BH spin in GW231123, and they extend to non-gravitational interactions by placing bounds on axion self-interactions ($f^{-1}$), dark-photon kinetic mixing ($\varepsilon$), and Higgs-Abelian couplings. These results complement electromagnetic spin constraints and prior GW analyses, offering a direct, population-independent probe of ultralight bosons in the gravitational sector with implications for axion and dark-photon models.

Abstract

In the presence of an ultralight scalar or vector boson, a spinning black hole will be spun down through the superradiant instability. We use spin measurements from gravitational wave observations of binary black holes, in particular the heavy binary black hole merger event GW231123, along with the lower-mass GW190517 event, to constrain the existence of ultralight bosons. We disfavor scalars with masses in the range of $[0.55, 11]\times 10^{-13}$ eV and vectors in the range of $[0.11, 18]\times 10^{-13}$ eV, making only a conservative assumption that the black hole lifetimes are greater than $10^5$ years. The lower ends of these ranges, where the exclusion confidence is the highest, were not previously excluded by spin measurements from electromagnetic or gravitational wave observations. We map these constraints to axion and dark photon models with interactions.

Figures (7)

• Figure 1: Exclusion regions for scalar and vector boson masses as a function of the BH age. The blue and orange shaded regions are excluded by GW231123 and GW190517, respectively, at confidence levels above 90% (i.e., $P<0.1P'$). The regions enclosed by dotted curves correspond to confidence levels above 99% (i.e., $P<0.01P'$), while the dashed curves mark the conservative exclusion regions common to all waveform models at a confidence level above 90%.
• Figure 2: Excluded parameter space for coupled bosons from GW231123 and GW190517. The shaded regions (down to vanishing coupling strength) show the parameter space excluded at the 90% confidence level, assuming a spin-down timescale of $T_{\rm age}=10^5$ years; i.e., we identified the regions, where the couplings are sufficiently weak so as to not affect BH spin-down via superradiance. Left: Constraints on the inverse energy scale $f^{-1}$ for an axion with an attractive quartic self-interaction. Right: Constraints on the dark photon parameter space with a kinetic mixing parameter $\varepsilon$ (light blue and orange regions) and in the Higgs-Abelian model (dark blue and orange regions). For the latter, the constraints on the combination of dimensionless coupling parameters $g\lambda^{-1/4}$ have been scaled up by a factor of $10^{12}$ to be accommodated on the plot. Due to theoretical uncertainties, we do not show bounds from superradiance involving higher azimuthal numbers ($m>1$), except for the kinetically mixed dark photon, in which case we use much less constraining assumptions for $m>1$. With improved theoretical modeling of the interaction effects, it should be possible to extend the constraints to cover the entire boson mass region excluded under the assumption of purely gravitational interactions (as indicated by the dotted vertical lines).
• Figure 3: Maximum dimensionless spin allowed as a function of boson mass and final BH mass when assuming a BH age of $T_{\rm age}=10^5$ years in the presence of a scalar (left) and vector (right).
• Figure 4: The growth rate $\omega_I$ of the higher order modes for two selected BH spins $\chi$ as a function of $\alpha$. Solid curves show the relativistically correct estimates, while the dashed curves show the nonrelativistic analytic results, expected to be accurate only in the $\alpha\ll 1$ regime.
• Figure 5: The relative errors, $\mathcal{U}_R$, defined in the text, of the growth rates $\omega_I$ as a function of $\alpha$ for two selected spins $\chi$.