The Ultraviolet Halos of Nearby Galaxies Edmund Hodges-Kluck, Joel Bregman, Julian Cafmeyer

University of Michigan—Ann Arbor, MI (hodgeskl@umich.edu)

4. Method and Results (Case Study: NGC 5775)

1. Motivation: The Baryon Cycle

Galaxy disks have only 10-30% of the baryons expected from their dark-matter halo masses and the cosmic baryon fraction [1,2]. The remainder were expelled or prevented from accreting early on, and about 20-30% of the “missing” baryons are in bound galaxy halos [3]. These baryons slowly accrete onto the galaxy and are the long-term fuel for star formation. Meanwhile, stellar feedback expels baryons from the disk in a galactic wind or fountain [4,5]. The rate at which these processes occur determines the stellar population, but observational constraints are poor.

2. A New Way to Detect Halo Gas

The metal content of halo gas distinguishes between “fresh” accreted material and feedback exhaust, but it is hard to measure. Halo gas is most easily detected in emission around edge-on galaxies (X-ray or 21-cm emission for hot and cold gas respectively), but the most reliable indicators of metallicity are absorption lines in the spectra of background quasars. The number of sight-lines is limited, especially close to the disk.

The dust modifies the input spectrum in a way that depends on composition:

We would like to know…

• The accretion rate of fresh gas • Feedback mass loading factors • The longevity of halo gas in

either form • How these quantities change

with galaxy type and cosmic time, and how they depend on galaxy environment.

… but we need to be able to distinguish between accreted gas and feedback ejecta

Thus, from the nebula and galaxy spectrum we can determine the type of dust. With additional 21-cm data we can get the dust-to-gas ratio.

Wind (starbursts)

Galactic fountain Cold

streams

Warm/hot accretion

Tidal stripping

Halo Gas Sources

3. Sample

Our sample includes highly inclined (i > 65°), late-type galaxies (Sa-Sd) within 100 Mpc with Swift/UVOT and GALEX data. We detect diffuse UV around ~100 galaxies where we can rule out psf wings and instrumental artifacts. About 40 have high quality data.

Halos are visible to 5-15 kpc from the disk, and are bluer than the (edge-on) host galaxies. Specific halo UV luminosity is strongly correlated with the host galaxy luminosity (below we show values for nearby galaxies where we can estimate the SFR). The data are consistent with reflection nebulae but not stellar halos. We measure halo flux in the five GALEX+UVOT bands (FUV 1516Å, UVW2 1918Å, UVM2 2246Å, NUV 2267Å, UVW1 2600Å)

We measure SEDs at various heights around each galaxy with enough data, and fit the data with synthetic SEDs made from galaxy spectra templates, dust models (here we show MW, LMC, and SMC models from [9]), and filter response curves (two overlap at the 2175Å “UV bump”). We determine what combination of galaxy model/dust, if any, fits best. A “red leak” in the uvw1 (and uvw2) filter means we must also account for the stellar halo with optical data. The best fits are for a galaxy model matching the observed galaxy type and an extinction law more like SMC dust than MW dust (i.e., carbon-rich). Here we show NGC 5775, whose SED and best-fit dust type changes with height:

Halo Flux Halo SED

4. Results, cont.

)1( scateFF galaxynebula

H

gas

dust

ex

scatexscat N

M

M

Observable Observable/Galaxy Template Dust Model + NH

Extinction law depends on ratio of carbonaceous, silicate dust grains [8]

Observable

5. Summary and Future Work

6. If you remember only one thing…

Acknowledgments

References

With HI data, we can use the dust type and halo flux to constrain the dust/gas ratio, which tells us how much dust was expelled or how much grain growth has occurred outside of galaxies. We use Monte Carlo radiative transfer models constrained by the HI and UV/optical data to validate our measurements. At right is the MCRT model for NGC 5775 using SMC-type dust. The model matches the optical and HI disks and the halo total flux and profile.

Dust-scattered reflection nebulae are ubiquitous around nearby, late-type galaxies. Their SEDs tell us how much dust there is and its chemical nature. Although the derived metallicities and masses are more model dependent than in QSO studies, we have ~1500 candidates within 100 Mpc, where GALEX’s spatial resolution is sufficient. Some galaxies also have very deep images where we can map the SED (see Julian Cafmeyer’s poster).

Dust in galaxy halos produces UV reflection nebulae that provide new insight into the composition and origin of halo gas.

This work was supported by the NASA ADAP grant #061951. The authors thank Alice Breeveld, Eric Bell, and Adolf Witt for helpful discussions. We made extensive use of the HEASARC., a service of NASA/GSFC

[1] Dai, X. et al. 2010, ApJ, 719, 119 [2] Bregman, J.N. 2007, ARA&A, 45, 221 [3] Anderson, M. et al. 2013, ApJ, 762, 106 [4] Veilleux, S. et al. 2007, ARA&A, 43, 769 [5] Bregman, J.N. 1980, ApJ, 236, 577 [6] Ménard, B. et al. 2010, MNRAS,405, 1025

[7] Neininger,N.&Dumke,M.1999,PNAS,96,5360 [8] Coker, C. et al. 2013, ApJ, 778, 79 [9] Weingartner,J.&Draine,B.2001,ApJ,548,296 This work: Hodges-Kluck & Bregman 2014, ApJ, 789, 131 (and in prep.)

The discovery that dust is common in galaxy halos [6] suggests a way to distinguish between accreted gas and feedback. Near the disk, dust scatters light leaking out of the galaxy, producing reflection nebulae around edge-on galaxies like off-axis searchlight beams (these have already been seen in starburst winds [7,8]).

Because the halo surface brightness is a small fraction of the background, we must carefully remove instrumental scattered light artifacts from each UVOT exposure. We typically clean 96-99% of the artifact light.

Raw Corrected

Data

Models & Fits

Fits with height

Most of our halo D/G values are consistent with the average measurement in halos from extinction [6], and close to the MW value. Among galaxy disks, there is a dispersion in dust-to-gas ratio. Carbon-rich grains may contain a significant fraction of IGM carbon/oxygen.

We can map the SED inside/outside of the superwind in NGC 3079