The detection of gravitational waves (GWs) by Advanced LIGO opens the door on a new era of physical and astronomical discovery. A key astronomical question that can be answered by GW observations is the following: what process produces the majority of stellar mass black hole (BH) binaries in the universe? The two traditionally considered processes have been evolution of isolated field binaries and dynamical assembly of binary systems in globular clusters. However, the unexpected properties of GW150914 should motivate a careful consideration of alternative scenarios.
I will discuss two novel scenarios for binary BH assembly recently proposed by myself and collaborators. First, wide binary BHs embedded in the accretion disks of active galactic nuclei (AGN) can be driven to merge through torques from circumbinary mini-disks. Pre-existing BH binaries can be captured into AGN disks dissipatively, or they may form in situ in Toomre-unstable regions of an AGN disk. While BH mergers in these gas-rich environments raise the tantalizing possibility of electromagnetic counterparts, strong AGN are sufficiently rare that the modest localization capabilities of GW detectors alone may be able to confirm or falsify the importance of this scenario with a sufficiently large BH binary sample. Next, I will present work from an ongoing series of papers investigating the dynamical assembly of triple star systems due to binary-binary scattering in globular clusters. The inner binary of a triple system may be driven to merger by the Kozai-Lidov effect, and in certain circumstances can reach the LIGO band with nontrivial eccentricity. The rate of such mergers depends on the outcomes of the chaotic, non-hierarchical four-body problem, and we have developed a statistical mechanics formalism to describe the distributions of these outcomes.