Radio AGN in Merging Clusters

Research Question

Is there a difference in the morphology of radio AGN between merging and non-merging clusters at low-z (z<0.2)?

Short Overview

We investigate the effect of the merging cluster environment on the properties of radio AGN by comparing the radio morphology of cluster members in a sample of four merging and eight relaxed galaxy clusters at low redshift (z < 0.2). Using 144MHz data from the LOFAR Two-meter Sky Survey and Zooniverse, we classify the radio morphology of the radio-detected cluster members using the following morphology classes: compact, compact extended, extended, jetted, and disturbed. We find that the merging cluster environment has a statistically significant, higher population proportion of disturbed (bent andheadtail) sources, indicating that the merging environment can affect the morphology of cluster radio AGN. We also investigate the number of AGN that are detected in the radio data only and the number that are detected in both the radio and optical data in mergers and nonmergers. We find that the merging cluster environment has a higher population proportion of AGN that are identified only as radio AGN compared to AGN that are identified as both radio and optical AGN. Overall, we find that the merging environment affects certain radio AGN (disturbed and only radio-identified AGN), but not all.

Publication

• This work resulted in a co-first author publication between Mary Rickel and myself: Rickel, M. & Moravec, E., et al., The Merging Galaxy Cluster Environment Affects the Morphology of Radio AGN, 2025, ApJ, 983, 138

Summary of Project

Background and Motivation

It is now well understood that dense environments like galaxy clusters influence the characteristics and evolution of the galaxies they contain. Specifically, the cluster environment also affects the properties and evolution of galaxies hosting active galactic nuclei (AGN). Galaxy clusters experience mergers with other clusters as part of the hierarchical structure formation in the Universe. While the impact of these mergers on the intracluster medium is evident, their effect on the cluster galaxies is less clear. Recent studies suggest that cluster–cluster mergers influence the AGN population, but the precise impact on the properties of radio AGN remains uncertain. This study investigates how mergers affect the morphology of radio AGN.

Sample and Classification

Starting from a sample of 33 galaxy clusters identified as merging (8) or nonmerging (25) by Bilton et al. (2020), we assembled a sample of 12 galaxy clusters (8 merging and 4 nonmerging) that have LoTSS 144 MHz data. We then identified all cluster members that had radio emission and defined the following radio morphology classes:

• Compact: roughly circular and on a scale of less than or roughly equal to 12″

• Compact Extended: compact core with a singular source of extended emission that is on the order of or greater than 12″

• Extended: sources that have emission that is clearly on a larger scale than that of the compact sources but not one of the other morphologies

• Jetted: contains Fanaroff and Riley Type Is (FR I) and Type IIs (FR II), and sources with clear two-sided non-bent jets, but FR classification is unclear (FRM).

• Disturbed: Contains a) bent tails (BTs) which are sources that have jets and those jets are bent to some degree, and b) head tails (HTs) which are extreme bent-tail source where the jets are indistinguishable.

Results

We express our results as binomial population proportions (k/n) - aka fraction of successes. Given the small number statistics and asymmetric errors on the proportions we measure, it is necessary to be careful in quoting and defining uncertainties. We take a Bayesian approach by modeling the proportion (P) as a random variable with a posterior probability distribution which is proportional to the likelihood and prior for where the likelihood is modeled by a binomial distribution and the prior is modeled by a beta distribution with a Jeffreys prior choice of parameters. The output is a posterior distribution where the central value of the distribution is the proportion quoted and the errors on this proportion are the 68% credible interval or 1-sigma which means that the the lower and upper uncertainties are derived by extracting the 16th and 86th percentile values.

The goal is to answer the question, is there a statistically significant difference between M and NM for a given radio morphology? However, given the typically asymmetric nature of these errors and the fact that the posterior distribution about the proportion is far from being a Gaussian, it is not possible simply to combine the quoted ‘errors’ (credible interval) in order to compare two different proportions with one another, or to multiply by a constant to get a 3σ confidence range. Thus if we observe two proportions P1 and P2, then the Bayesian way of asking the question ‘is P2 significantly greater than P1’ which can be rephrased as ‘is there a high posterior probability that P2 − P1 > 0?’ And this becomes a simply subtraction operation. 

To do this, we take an Markov chain Monte Carlo (MCMC) approach and sample the posterior distributions of the radio morphology class (for example compact sources) in M and NM 100000 times and then we can simply calculate the difference in the sample pair values between mergers and non-mergers, and obtain a new posterior distribution of the difference. As a result, we obtain the PROBABILITY of P2 > P1. Then we map this probability to a statistical significance level of a Gaussian taking the standard approach that 68%, 95%, and 99.7% correspond to 1σ, 2σ, and 3σ, respectively.