In the summer of 2019, I began working with Dr. Chris Mihos on studying the star formation and ionized gas in the outskirts of Messier 101 (M101). This work was completed by the end of spring 2021. I will refer any interested reader to the paper for more details.

Extragalactic HII Detections

Understanding star formation and the variety of locales in which it takes place is key to understanding galaxy formation and evolution. Star formation is most easily studied in the inner luminous regions of galaxies, but star formation in low density environments is less well understood, whether that be low luminosity dwarf galaxies or outlying H II regions. The latter of these two are especially interesting, given that ultraviolet (UV) light revealed by the Galaxy Evolution Explorer (GALEX) has shown that over 30% of spiral galaxies have star formation beyond their optical disks (Thilker et al. 2007). This is likely associated with previous or ongoing galaxy interactions (Thilker et al. 2007; Werk et al. 2010).

In optical light, outlying isolated H II regions have been detected via narrowband imaging targeting specific emission lines, primarily Hα, where they appear as emission-line point sources. These outlying regions have been found in environments ranging from galaxy clusters and compact groups to the halos of galaxies, and they indicate recent star formation in extremely low density regions. These could be isolated H II regions or embedded in a larger star-forming dwarf galaxy.

In an effort to explore these areas of low-density star formation, my collaborators have used Case Western’s Burrell Schmidt telescope to perform the first deep, wide-field, multiline, narrowband observations of M101 and its group environment. They imaged M101 in three different emission lines that are characteristic of star formation: Hα (λ6562 Å), Hβ (λ4861 Å), and [OIII]λ,λ4959,5007. This allows me to reject contaminants easily without resorting to expensive follow-up spectroscopy. Additionally, instead of using a broadband filter for continuum subtraction, my collaborators used narrowband off-band filters, going redward for the Hα and [OIII] filters and blueward for the Hβ filter. I also have access to broadband B and V images taken from the Burrell Schmidt (Mihos et al. 2013).

Originally, we had used a Python implementation of SExtractor called “sep,” (Barbary 2016) but we were unconvinced with the results of that process. In January 2020, we switched our detection routine to another Python module PhotUtils’ “segmentation” routine (Bradley et al. 2019). This program “detects” sources by assigning as objects that have a minimum number of connected pixels that are each greater than the background threshold value. In our case, we marked pixels that were 3σ above the background level on our Hα on-band image as a detection. This is conceptually similar to SExtractor’s κσ clipping. After masking bright sources (stars, M101, its companions), the detection routine finds 32,439 sources which are further deblended into 35,308 sources. I will refer any reader to the documentation for further information.

The segmentation map that results from the detection routine. Each colored point is a different detected source. The black voids are masked objects. M101 is the circle at the center of the image.

The segmentation module gives basic information about each detected source after we run it on each image, having defined each region in the Hα on-band image.

At this point, we have a source catalog of any astronomical source that has excess emission in Hα. This could include objects like H II regions or dwarf galaxies in the nearby universe, background emission-line objects such as quasars or high redshift galaxies, or objects such as Galactic M stars with molecular absorption lines in our filters. We tested how these would appear through our narrowband filters by synthesizing stellar spectra from the Stellar Spectral Flux Library (Pickles 1998) and galactic spectra from the SDSS DR5 spectral templates.

The equivalent width (EW) of stellar sources (black points) and galactic sources (blue triangles) as a function of B-V color. The horizontal dashed lines indicate the equivalent width selection cut for our emission line sample in each filter pair.

Given the plots above, we made a selection cut on equivalent width for each source. To remain in the catalog, a source had to have EW(Hα) > 8 Å, EW(Hβ) > 2 Å, and EW([OIII]) > 5 Å. However, given the relative weaker strength of the Hβ line compared to Hα, we made a distinction between sources with all three lines in emission and those with only Hα and [OIII] filters in emission. These two groups are called the three- and two-line samples, respectively. Making these cuts reduced our source catalog from 35,308 sources to 147 sources.

At this point, we need to remove M stars. For those who are unfamiliar, M stars are cool (~3000 K) stars that are predominantly red (B-V ≳ 1.4) and make up about 76% of all stars in the solar neighborhood, regardless of the initial mass function (IMF) chosen. These M stars show up as bright sources in all of our narrowband filters, since we are creating difference images by subtracting each off-band image from each on-band image. Due to the placement of our filters, our filters often lie on top of absorption troughs in the spectra of M stars created by molecules such as titanium oxide and magnesium hydride.

The spectrum of an M5V star from Pickles (1998). The black solid line is the spectrum. The colored lines are our narrowband filter transmission curves.

It is easily seen that is we were to, for instance, subtract Hα off from Hα on, there would be net “emission” despite there being no noticeable Hα emission line. We chose to remove these M stars by cross matching our sources with the Gaia Early Data Release 3 catalog (Gaia Collaboration et al. 2021). This did an excellent job of identifying M stars, essentially removing every source with a B-V color greater than 1.0.

A final cut we made was to remove high redshift objects that could have shifted into our filters. We utilized the Balmer decrement as one cut. For unobscured ionized gas, the Balmer decrement should be Hα/Hβ = 2.86 (Osterbrock 1989), with higher ratios indicating higher extinction levels. If an object has an anomalous Balmer decrement, then it is likely a high redshift object and we remove it. However, there is no similar cut we can make for the two-line sample, since Hβ could be in absorption. We cross matched with SDSS, removing those sources for which SDSS has a spectroscopic or photometric redshift.

At this point, we have made all the cuts required and examined each source by eye to confirm its nature. We are left with 19 objects in the three-line sample and 8 objects in the two-line sample. Briefly, we give a sense of how deep we have attained in fluxes and the range of equivalent widths in each group. In the three-line sample, the fainted source has V = 22.9. The corresponding Hα net flux is 8 × 10-16 erg/s/cm2, which at the 6.9 Mpc distance to M101 and using the SFR-L(Hα) calibration of Kennicutt & Evans (2012), would correspond to a star formation rate (SFR) of 2.4 × 10-5 M yr-1. The three-line sample has a range of Hα equivalent widths from 35-425 Å with a median of 110 Å. In the two-line sample, the faintest source has V = 24.4. The coresponding Hα net flux is 9.3 × 10-17 erg/s/cm2, which corresponds to an SFR of 2.8 × 10-6 M yr-1. The two-line sample’s Hα equivalent widths range across 8-300 Å, with a median of 60 Å.

We assigned identifiers to each source based on their proximity to known objects in the survey area and their sample. In the three-line sample, 18 (95%) sources appear to be associated with the outer disk of M101 and 1 (5%) appears to be associated with M101 satellite galaxy, NGC 5474. In the two-line sample, 7 (88%) sources appear to be associated with the outer disk of M101, and 1 (12%) is associated with the background galaxy NGC 5486.

A zoomed-in view of the detected regions surrounding M101. Left: the continuum-subtracted Hα image. Right: the broadband B image, stretched to show the faint structure of the outer disk. The orange and green circles with radii of 20 arcsec are centered on the objects in the three- and two-line samples, respectively. The red circle represents the size of the mask used in our source detection algorithm. Both images measure 24 × 24 arcmin. North is up and east is to the left.
A zoomed-in view of the detected regions near NGC 5474 (top row) and NGC 5486 (bottom row). Left: the continuum-subtracted Hα images. Right: the broadband B image, stretched to show the faint structure of the outer disks. Colored circles are the same as in the above figure. The images of NGC 5474 measure 9 × 9 arcmin and the images of NGC 5486 measure 3.6 × 3.6 arcmin. North is up and east is to the left.

Since we are looking for faint star-forming objects in the M101 Group, we began by comparing the observed properties of our samples to the observed properties of known satellites in the Local Group and M101 Group (McConnachie 2012; de Vaucouleurs et al. 1991; Taylor et al. 2005; Merritt et al. 2014; Bennet et al. 2019; Bellazzini et al. 2020). Most of the three-line samples that match observed dwarf galaxy properties are not individual dwarf galaxies around M101, but are rather H II regions on the outskirts of the spiral arms. The source associated with NGC 5474 is the faintest V-band source and is very small. Its size indicates that it is likely an H II region, comparable to the predicted size of an H II region powered by an O9 star (Strömgren 1939; Osterbrock 1989). The two-line sources have a wider spread in brightness. The source associated with NGC 5486, if at the further distance of NGC 5486 (28.2 Mpc), has a size and brightness near the local dwarf galaxies WLM and the Sagittarius Dwarf Spheroidal Galaxy (McConnachie 2012).

Effective radius versus apparent magnitude for our detected sources and comparison samples shifted to the M101 distance. The dotted line indicates 1 arcsec (33 pc at M101’s distance). Pentagons denote sources whose properties were derived using SDSS data, while squares are sources with properties derived from our Burrell Schmidt imaging. Data for local dwarfs are shown as gray circles, while data for M101 satellites are shown as gray squares.

Another way of understanding these objects is to compare their star-forming properties to those of known galaxies. We do this by comparing our samples to the 11HUGS and SINGG surveys (Kennicutt et al. 2008; Meurer et al. 2006), both targeting local star-forming galaxies. On average, our sources are much fainter (and have much lower SFRs) than the 11HUGS/SINGG galaxies. Our sources also lie on the high end of Hα equivalent widths. This is to be expected given that our survey is an emission line survey; we will find strong emitters with high equivalent widths. The sources associated with M101 are consistent with H II regions in the spiral arms of M101 (Cedrés & Cepa 2002), albeit on the low end. The source with the lowest Hα equivalent width is the source associated with NGC 5486; its SFR and equivalent width are similar to the dwarf galaxy NGC 4163. It is possible that we have discovered a satellite galaxy of NGC 5486. The source with the highest equivalent with is associated with NGC 5474. Its equivalent width puts it into the range of outlying H II regions as defined by Werk et al. (2010).

The main panel shows Hα flux versus equivalent width in the three-line sample (squares) and two-line sample (circles), color-coded by associated galaxy. The left y-axis shows measured flux; the right y-axis shows equivalent star formation if the objects were at the M101 distance. Smaller marker sizes are used for unresolved objects. The upper panel shows the normalized distribution of Hα equivalent widths of the sources.

Given that we did not find any truly outlying H II regions, what does this mean for the star-forming luminosity function of galaxies in the group? We can put a limit on the faint end slope: a steep faint end slope would predict on the order of 120 objects detected, while a shallow slope would predict on the order of 1 object detected. This likely means that the faint end slope of the star-forming luminosity function for the M101 Group is α > -1.0.