CHemical Abundances Of Spirals


PIs: Evan Skillman (UMN) & Richard Pogge (OSU) 


The history of a galaxy can be traced by its abundances of heavy elements, as they are produced and accumulated as successive generations of stars return their newly synthesized elements to the interstellar medium (ISM).  In spiral galaxies, ISM abundance studies are dominated by the disk, where the majority of their star formation occurs, and are typically characterized by negative radial gradients of oxygen and nitrogen abundances (e.g., Pagel & Edmunds 1981; Garnett & Shields 1987; Zaritsky 1994). The abundance gradients across the disk of spiral galaxies provide essential observational constraints for chemical evolution models of galaxies, and support the inside-out growth theory of galaxy disk formation.

Emission lines originative from HII regions provide an excellent probe of the gas-phase abundances and, thus, the radial metallicity gradients in disk galaxies. Further, HII regions, which are ionized by recently-formed massive stars that carry the same chemical signature from the gas in which they were formed, allow us to measure the cumulative chemical evolution of the present-day ISM.

Many studies have used multi-object spectroscopy to attempt to directly measure the nebular physical conditions and abundances and map out their trends across the disks of spiral galaxies. However, because direct measurements of gas-phase abundances via one of the "direct" methods (i.e., auroral or recombination lines) have long been prohibitively expensive in terms of telescope time, the majority of these studies are limited to first order trends using a dozen or fewer abundance detections per galaxy. This challenge motivated the CHemical Abundances Of Spirals (CHAOS; Berg+15) project: a large database of high quality HII region spectra over a large range in abundances and physical conditions in nearby spiral galaxies. These spectra provide direct abundances, estimates of temperature fluctuations and their corresponding corrections to lower absolute abundances, and allow calibrations based on observed abundances over expanded parameter space rather than photoionization models. 

While the absolute abundance scale of HII regions is still a topic of debate(see, for example, the discussion of the Abundance Discrepancy Factor in Bresolin+16), the CHAOS survey is building a large sample of direct abundances, observed and analyzed uniformly, allowing us to characterize the possible systematics of the direct method. To date, CHAOS has increased, by more than an order-of-magnitude, the number of HII regions with high-quality spectrophotometry to facilitate the first detailed measurements of the chemical abundances in a sample of nearby disk galaxies. So far, results for individual galaxies have been reported for:

  1.  NGC 628 (M74) in Berg+15

  2. NGC 5194 (M51a) in Croxall+15

  3. NGC 5457 (M101) in Croxall+16

  4. NGC 3184 in Berg+19

The combined sample of the first four CHAOS galaxies totals 190 HII regions with measured auroral line based temperatures.




Erik Aver

Gonzaga University

Danielle Berg

The Ohio State University


Kevin Croxall

Expeed Software


Ness Mayker

The Ohio State University

John Moustakas 

Siena College

Richard Pogge

The Ohio State University


Noah Rogers

University of Minnesota


Evan Skillman

University of Minnesota


JD Smith

University of Toledo



3 cups


3 cups


1½ cups






























Status: To date, CHAOS has observed the 14 nearby spiral galaxies shown above. Results are complete for four galaxies, and data reduction and analyses are underway for the remaining galaxies.

Screen Shot 2019-09-23 at 4.20.05 PM.png

More coming soon ...




The CHAOS project was undertaken in order to build a large database of high quality H II region spectra from nearby spiral galaxies using the Multi-Object Double Spectrographs (MODS; Pogge et al. 2010) on the Large Binocular Telescope (LBT). The MODS have been designed to optimally obtain high quality spectra across fields of view (6′ × 6′) comparable to the extents of nearby spiral galaxies. Furthermore, the LBT and MODS combination provides the nessesary balance between sensitivity, resolution, and wavelength coverage to measure all emission lines relevant to O, N, and α-element abundance determinations and more. In particular, the optical design of MODS has been optimized in both the red and blue channels for high throughput in to produce high-quality spectrophotometry from the atmospheric UV cut-off (∼340 nm) to the silicon detector cutoff at ∼1μm.

The dual 8.4-m mirror on the Large Binocular Telescope

on Mount Graham, AZ.

The primary goal of MODS optical observations is to obtain high signal-to-noise spectra, with detections of intrinsically faint auroral lines (e.g., [O III] λ4363, [N II] λ5755, [S II] λ6312) at a significance of 3σ or higher. The multi-object mode of MODS, which uses custom-designed, laser-milled slit masks, allows spectra of many HII regions to be obtained simultaneously. Broad-band and Hα continuum-subtracted images are used to identify target HII regions, as well as alignment stars, and determine accurate astrometry for the masks. HII regions are selected to achieve large radial coverage of the optical disk of a galaxy, prioritizing knots of highest Hα surface brightness. On average, three multi-slit masks are observed for each galaxy, with each mask containing ∼20 slits, which span the radial and azimuthal extent of the optical disk. All HII region slits are 1.0" wide, but lengths vary depending on the size of the targeted H II region and proximity of other slits on the mask, where slits are made as long as possible to observe sky within the same slit when possible. Blue and red spectra are obtained simultaneously using the G400L (400 lines/mm, R∼1850) and G670L (250 lines/mm, R∼2300) gratings, respectively. As a result, broad spectral coverage extends from 3200 to 10000 Å, with a FWHM resolution of ∼2 Å.

Each mask field is observed for six 1200s exposures, for a total integration time of two hours. The masks are designed with slits at a fixed position angle which approximated the parallactic angle at the midpoint of the longest possible observation window in a night. This, in addition to observing galaxies at airmasses < 1.5, serves to minimize the wavelength-dependent light loss due to differential atmospheric refraction (Filippenko+82). The mask coverage of the first four CHAOS galaxies are shown below.

Screen Shot 2019-09-23 at 3.03.25 PM.png
45 Te Regions

NGC 628

30 Te Regions

NGC 3184

Read More
Screen Shot 2019-09-23 at 3.03.59 PM.png
29 Te Regions

NGC 5194

Screen Shot 2019-09-23 at 3.04.20 PM.png
74 Te Regions

NGC 5457

Demonstration of a one dimensional spectrum taken with MODS1/LBT of an region +16.4+119.8 in NGC 3184 with auroral line detections at a strength of 3σ or greater. The observed spectrum is plotted as a black line, with the model in blue, and the 1σ uncertainty in gray. In the expanded windows, we mark and label the five temperature-sensitive auroral emission line studied by CHAOS: [SII] λλ4068,4076, [OIII] λ4363, [NII] λ5755, [SIII] λ6312, & [OII] λλ7320,7330. This spectrum lacks an [OIII] λ4363 detection as the majority of the emission in that region is actually due to a contaminating [FeII] line at λ4360 (see blue box). Note that major telluric absorption features are not corrected for (see bottom panel).

Screen Shot 2019-09-23 at 2.24.23 PM.png

Recent Results

Screen Shot 2019-09-23 at 4.40.38 PM.png


CHAOS has proven highly successful at measuring significant detections of both [NII] λ5755 and [SIII] λ6312, demonstrating the utility of these lines in metal-rich HII regions. Given the robust Te[NII]−Te[SIII] relationship demonstrated for the 90 HII regions with simultaneous detections, we find that Te[SIII] is the best temperature ameasure to use in metal-rich HII regions that tend to be lower-ionization. In general, we promote an ionization-based decision of temperature measurements.


The large number of direct abundance measurements in individual CHAOS galaxies allows a robust measure of their oxygen abundance gradients. For the first four CHAOS galaxies, we find each galaxy has a unique slope and dispersion, regardless of whether plotted versus isophotal radius, R25, or effective radius, Re.

Screen Shot 2019-09-23 at 4.31.30 PM.png
Screen Shot 2019-09-23 at 4.32.27 PM.png


Top: N/O versus galactocentric radius for the first four CHAOS galaxies. Each galaxy has a separate fit to its N/O for the primary and secondary production of N. When radius is plotted relative to Re, a natural break in the N/O trends appears near Rg∼2.25·Re, separating the primary and primary+secondary nitrogen components. A variance-weighted average primary plateau is fit for Rg > 2.25·Re for each galaxy (and assumed for NGC 3184 based on the extrapolated fit for Rg < 2.25·Re) and listed in the legend. Bottom: N/O trends are normalized by the primary average N/O value of each galaxy, removing the effects of their SFH that set their plateaus apart. We find a universal N/O gradient for a composite sample of 97 HII regions of N/O = −0.35 dex/Re for Rg/Re < 2.0, where N is dominated by secondary production. The normalized N/O gradient has a much smaller dispersion (σtot. = 0.06 dex) than the O/H gradients and so may be useful for constraining stellar yields.


The N/O–O/H trends for the first four CHAOS galaxies are color-coded by the O+/O ratio, or low-ionization fraction. Interestingly, the overall trend of increasing N/O seems to be ordered by ionization or age. For comparison, the average relationship for metal-poor dwarf galaxies is plotted to extend the primary nitrogen plateau, and the empirical trend for galactic stars (Nichols+17) acts as a lower envelope to the CHAOS observations. At this time, the source of the scatter in the N/O–O/H relationship remains an open question, but with several promising  possibilities for future study.

Screen Shot 2019-09-23 at 4.36.09 PM.png


CHAOS I. Direct Chemical Abundances for HII Regions in NGC 628

    Berg, D.A.; Skillman, E.D.; Croxall, K.V.; Pogge, R.W.; Moustakas, J.; Johnson-Groh, M., 2015, ApJ, 806, 16

CHAOS II. Gas-Phase Abundances in NGC 5194

    Croxall, K.V.; Pogge, R.W.; Berg, D.A.; Skillman, E.D.; Moustakas, J.; 2015, ApJ, 808, 42

CHAOS III. Gas-Phase Abundances in NGC 5457

    Croxall, K.V.; Pogge, R.W.; Berg, D.A.; Skillman, E.D.; Moustakas, J.; 2016, ApJ, 830, 4

CHAOS IV. Gas-Phase Abundance Trends from the First Four CHAOS Galaxies

    Berg, D.A.; Pogge, R.W.; Skillman, E.D.; Croxall, K.V.; Moustakas, J.; Rogers, N.; Sun, J.; Leroy, A., 2019, ApJ, submitted