Uh oh, some Black Holes are experiencing Middle Child Syndrome!

Title: Variability-selected intermediate-mass black hole candidates in dwarf galaxies from ZTF and WISE

Authors: Charlotte Ward, Suvi Gezari, Peter Nugent, Eric C. Bellm, Richard Dekany, Andrew Drake, Dmitry A. Duev, Matthew J. Graham, Mansi M. Kasliwal, Erik C. Kool, Frank J. Masci, Reed L. Riddle

First Author’s Institution: Department of Astronomy, University of Maryland, College Park, MD 20742, USA

Status: Published in ApJ [open access]

The Elusive Middle Child

Intermediate Mass Black Holes (IMBH) are the middle children of the black hole family. They typically have masses between the smaller stellar-mass black holes and larger supermassive black holes but remain far more elusive. This significant mass gap has been puzzling astronomers for decades. Why aren’t we able to detect IMBH? What is causing these black holes to act up?

One of the best ways to uncover black holes is to look for signatures of active black holes (also called AGN). Active black holes feed on the gas that surrounds them and thus shine brighter than the rest of the galaxy. The strength of this activity can be determined in several ways, and it is expected to scale with the size of the black hole. One method for detecting AGN activity in galaxies is to look for their emission line signatures. However, this method is not particularly useful for locating black holes that have masses lower than the typical supermassive black holes, as they are likely to show energies very similar to stars. The Baldwin, Phillips & Terlevich (BPT) diagram, which is used to classify galaxies as having AGN or star-forming activity based on the strength of the emission lines, often mistakes galaxies having lower mass AGN as star-forming galaxies. The other approach involves looking for signatures across different wavelengths. AGNs are the only objects in the universe with emissions in nearly all electromagnetic bands, but it can be tedious and expensive to take observations in many wavelengths.

Figure 1: An example of a light curve – the flux from a source over time. A variable source will vary in brightness with time. (Image credit: Wikimedia Commons)

Luckily for us, AGNs have another property that sets them apart. The brightness of the AGN measured by telescopes varies with time, typically in the order of days to months, while no other object varies as much. This variability can be used to identify black holes by constructing light curves, as shown in Figure 1. (For more details on time-domain astronomy, check out this Astrobite on transients.)

A better method to detect IMBH?

In this paper, the authors try to find AGN candidates using optical and mid-infrared (MIR) variability in dwarf galaxies, as IMBH are more likely to be found in these smaller galaxies. They use the Zwicky Transient Facility to look for optical variability in a sample of dwarf galaxies. They also have a control sample of galaxies, confirmed to have an AGN, which are optically variable. To detect variability, they do the following:

  1. Get the light curves of the target dwarf galaxies. 
  2. Compare them with the AGN light curves using statistical tests. They use a combination of Pearson’s correlation coefficient and the Chi-squared test in different optical bands to eliminate the non-variable AGNs in dwarf galaxies. This can be done by comparing them to how similar or different they look from the variability detected in the control AGN sample, as shown in Figure 2.
Figure 2: Figures showing the results of the statistical tests. The r value in Pearson’s test should be close to 1, and the χ2 should be large for a good correlation between the confirmed AGN light curves (green) and the dwarf galaxy sample (blue). The dotted line indicates the cut-off value used for selection. Image credit: Panel 3 in Figure 1 of the paper.

3. Inspect each light curve visually to see if any incorrectly identified variability could be caused by other sources, such as supernovae. Light curves from supernovae are spikier as the maximum light would die down in a shorter time than AGNs (see Figure 3). These were then discarded from the sample.

Figure 3: An example of AGN variability in the light curves (left) compared to supernova (SN) variability (right) observed at two wavelengths, r band, and g band. Notice how SNs have a sharp rise in flux and quickly die down. Such light curves would have passed the initial statistical tests, so they were discarded following a visual inspection. Image Credit: Figures 2 and 3 of the paper.

The MIR variable targets were chosen from the WISE survey. However, WISE is not a transient telescope, so doesn’t obtain light curves for the targets it observes. Instead, the authors adopted a technique called forward modeling, where light curves were modeled based on the photometry collected by WISE. WISE would revisit every field after 6 months. So by combining the fluxes obtained from each visit and predicting how the curves might look in the time between each visit, light curves were constructed. A similar combination of statistical tests and visual inspection used for optical variability was applied to the infrared light curves to check for variability.

Found them!

Once the object was identified as variable (and hence an AGN candidate), the black hole masses were calculated using the black hole-bulge mass relationship. They were found in the range of 105.33 to 107.6 M, some of which cover the highest expected ranges of IMBH. In the optical, they found 44 dwarf galaxies with signatures of variable AGN out of nearly 26000 galaxies. Of these, 81% were classified as star-forming in the BPT diagram. They also found 148 AGN candidates out of ~ 80000 galaxies in the mid-infrared, where 69% of them were classified as star-forming in the BPT diagram (see Figure 4). This shows that variability can be a valuable technique to uncover misclassified AGNs, especially in the lower masses. It is essential to use multiple wavelength variability searches for AGNs, as 90% of WISE-selected AGN were missed by optical variability.

Figure 4: BPT diagrams showing the classification of the optically variable targets (left) and infrared variable targets (right). The orange points indicate AGN classification using emission line signatures. The blue points indicate targets previously classified as star-forming (SF) using emission lines. This figure highlights that relying on typical AGN identification methods could result in missing AGN. Image credit: Figure 5 of the paper.

Through this paper, the authors have demonstrated that looking for variability can effectively detect low-mass AGN activities. They claim that their selection method can be used to identify variability from dwarf galaxies with redshifts up to z~0.15 and stellar masses down to 107.2 M. This could be useful in detecting lower-mass black holes, and hopefully, we’ll be able to give the poor black hole middle child the attention it deserves!

Astrobite edited by Storm Colloms, Lili Alderson, and Roan Haggar

Featured image credit: NASA/JPL-Caltech 

About Archana Aravindan

I am a third-year Ph.D. student at the University of California, Riverside, where I study black hole activity in small galaxies. When I am not looking through some incredible telescopes, you can usually find me reading, thinking about policy, or learning a cool language!


  1. Fascinating technique for IMBH detection! Really enjoyed your article, and I loved the title!

  2. From the chi-square method, most have the variability, however, from the Pearson r method, most don’t. Does it mean these two method cannot be used together?

    • It is a little hard to establish which of the two methods is better quantitatively (I believe it depends significantly on several parameters, like the sample size), but typically they are used together to indicate correlation. The authors of this paper do mention that they only select galaxies that pass both correlation tests.


Leave a Reply