Abstract

Pair instability supernovae are thought to restrict the formation of black holes in the mass range ~50 - 135 solar masses. However, black holes with masses within this "high mass gap" are expected to form as the remnants of binary black hole mergers. These remnants can merge again dynamically in densely populated environments such as globular clusters. The hypothesis that the binary black hole merger GW190521 formed dynamically is supported by its high mass. Orbital eccentricity can also be a signature of dynamical formation, since a binary that merges quickly after becoming bound may not circularize before merger. In this work, we measure the orbital eccentricity of GW190521. We find that the data prefer a signal with eccentricity $e \geq 0.1$ at 10 Hz to a non-precessing, quasi-circular signal, with a log Bayes factor $\ln{\cal B}=5.0$. When compared to precessing, quasi-circular analyses, the data prefer a non-precessing, $e \geq 0.1$ signal, with log Bayes factors $\ln{\cal B}\approx2$. Using injection studies, we find that a non-spinning, moderately eccentric ($e = 0.13$) GW190521-like binary can be mistaken for a quasi-circular, precessing binary. Conversely, a quasi-circular binary with spin-induced precession may be mistaken for an eccentric binary. We therefore cannot confidently determine whether GW190521 was precessing or eccentric. Nevertheless, since both of these properties support the dynamical formation hypothesis, our findings support the hypothesis that GW190521 formed dynamically.