Signs of eccentricity in two gravitational-wave signals may indicate a sub-population of dynamically assembled binary black holes

Abstract: The orbital eccentricity of a merging binary black hole leaves an imprint on the associated gravitational-wave signal that can reveal whether the binary formed in isolation or in a dynamical environment, such as the core of a dense star cluster. We present measurements of the eccentricity of 26 binary black hole mergers in the second LIGO--Virgo gravitational-wave transient catalog, updating the total number of binary black holes analysed for orbital eccentricity to 36. Using the \texttt{SEOBNRE} waveform, we find the data for GW190620A is poorly explained by the zero-eccentricity hypothesis (frequentist $p$-value $\lesssim 0.1\%$). Using a log-uniform prior on eccentricity, the eccentricity at $\unit[10]{Hz}$ for GW190620A is constrained to $e_{10}\geq0.05$ ($0.1$) at $74\%$ ($65\%$) credibility. With this log-uniform prior, we obtain a $90\%$ credible lower eccentricity limit of $0.001$, while assuming a uniform prior leads the data to prefer $e_{10} \geq 0.11$ at $90\%$ credibility. This is the second measurement of a binary black hole system with statistical support for non-zero eccentricity; the intermediate-mass black hole merger GW190521 was the first. Interpretation of these two events is currently complicated by waveform systematics; we are unable to simultaneously model the effects of relativistic precession and eccentricity. However, if these two events are, in fact, eccentric mergers, then there are potentially many more dynamically assembled mergers in the LIGO--Virgo catalog without measurable eccentricity; $\gtrsim 27\%$ of the observed LIGO--Virgo binaries may have been assembled dynamically in dense stellar environments ($95\%$ credibility).