GeminiFocus January 2015 | Page 7

lar populations: quenched galaxies are plotted as red triangles, and star-forming galaxies appear as inverted blue triangles. These two populations are distinctly different, with quenched galaxies tending to be found near the cluster’s center and moving at slower velocities. Star-forming galaxies are found throughout the cluster and tend to be moving at higher velocities. This had been seen before in lower-redshift clusters (e.g., Biviano et al., 2002; Poggianti et al., 2004). It can be explained naturally by the quenched galaxies having been accreted into the cluster earlier (presumably when they were still star-forming galaxies). Most likely they experienced dynamical friction and lost kinetic energy, gradually slipping into orbits well within the clusters’ gravitational potential wells. Star-forming galaxies are likely to be more recently accreted and hence have higher velocities, as they are on their first passage through the clusters. Our team also looked at galaxies that had spectral signatures indicating recent quenching. These “post-starburst” galaxies are shown as the encircled green stars in the left panel of Figure 1. These high-velocity objects had quite a striking distribution in this Figure, as they tend to lie roughly in a “ring” structure at intermediate radius. In particular, they very clearly avoid both the cluster core region, which is dominated by quenched galaxies, and the large radius region dominated by star-forming galaxies. This signature has never been seen before, so modeling it seemed an obvious way to try to understand more about the quenching process. We were most interested in getting some quantitative measurements of both how long it takes for galaxies to quench (once the process begins) and where in the cluster it starts to happen. In particular, the latter measurement is unprecedented. We employed some high-resolution dark-mat- January 2015 ter-only simulations of clusters to see if we could input these numbers and reproduce the observed distribution of the post-starbursts in the right panel of Figure 1. In the simulations, we followed galaxies as they first fell into the cluster. We then simulated quenching at different locations and with different timescales. These models easily ruled out many locations and specific timescales; the details of this can be found in Muzzin et al. (2014). What worked best in reproducing the distribution of the poststarbursts was if we quenched galaxies after they first passed roughly half the virial radius (R ~ 0.5R200) on a timescale of ~ 0.1 - 0.5 Gyr. The resulting distribution of post-starburst galaxies is shown as the right panel of Figure 1, and is statistically a good match to the observation. This was the first time that simultaneous constraints had been set on both the duration of the quenching process and where it happens. We were concerned that our findings depended heavily on a simulation, so we also used the stellar populations of the galaxies to test the model. Figure 2 shows the average spectra of both the typical star-forming galaxy in the cluster (top blue) and the average spectra of the post-starburst galaxies (bottom green). These spectra were fit to a range of models, with the best-fit star-formation history shown in the bottom left panel. Basically, this test showed that star-forming galaxies continue forming stars for several Gyr within the cluster before they quench — and that process needs to be rapid in order to create the stellar populations seen in the post-starburst galaxies. Not only is this consistent with what we d W&