Transcript Rich

A Study of the
Abundances and Lyα
Emission of Ultraviolet
Selected Samples of Star
Forming Galaxies in the
Local Universe
Ryan Mallery (Ph.D. thesis), R. Michael Rich,
Jean-Michel Deharveng, and the Galex team
UCLA
GALEX
Telescope
– Imaging and slitless grism
spectroscopy
– FUV (1300-1800 Å) 4.5” psf FWHM
– NUV (1800-2800 Å) 6” psf FWHM
– 50 cm diameter
– 1.2 degree diameter FOV
– 8 Å spectral resolution
Survey
Exp time
Area
mlim
Deep Imaging
Survey
30,000s
80 deg2
25 AB
Medium
1500s
Imaging Survey
1000 deg2
23 AB
All Sky Imaging 100s
Survey
26,000 deg2
20.5 AB
Medium
Spectroscopic
Survey
5 deg2
22 AB
150,000s
Lyα Emitters Motivation
• Issue: Lyα emission from galaxies is likely affected by a
combination of dust extinction, gas kinematics, and spatial geometry
but the order of importance of each of these is unknown, and may
vary from galaxy to galaxy.
• Opportunity: Galex grism mode in Far-UV (FUV) used to discover
the first known sample of low redshift LAE galaxies. Instead of
studying LAEs at z~4, study them at z~0.3
• Goal: Analyze photometry and spectroscopy of a local z~0.3 sample
of ~76 LAEs and 107 non-LAEs to :
– 1) determine why Lyα escapes from some galaxies and not others.
– 2) determine the relative effect of gas kinematics, ISM geometry, and
extinction by dust on the Lyα emission.
• Importance: Lyα is presently one of the most widely utilized spectral
features for the discovery and identification of galaxies at the highest
redshifts.
– Due to the low redshifts, and hence relatively bright UV+optical fluxes,
the sample represents a chance to understand the astrophysics of these
sources in much more detail than can be done with high redshift LAEs
GALEX LAE Sample
• The First Systematic low redshift
Survey for LAEs (first discovered by
Deharveng et al. 2008)
• Northern Hemisphere Deep GALEX
Spectroscopic Fields
–
–
–
–
Cosmological Evolution Survey (COSMOS)
All Wavelength Extended Groth Strip
International Survey (AEGIS)
NOAO Deep Wide Field Survey (NDWFS)
Spitzer First Look Survey (FLS)
• 76 LAEs
• LAEs are matched to sources in
publicly available datasets (SDSS,
ACS, IRAC, MIPS) within 4”
–
–
–
–
75/76 with SDSS photometry
25/76 with IRAC photometry
20/76 with ACS photometry
32/76 with MIPS 24μm photometry
COSMOS
• UV spectroscopy exposure
time: 35 hours
• 15 LAEs
• 14/15 with SDSS
• 15/15 with ACS imaging
• 14/15 with MIPS 24 detections
AEGIS
• UV spectroscopy exposure
time: 80 hours
• 37 LAEs
• 37/37 with SDSS
• 6/37 with ACS imaging
• 13/37 with MIPS 24 detections
NDWFS
•
•
•
•
•
UV spectroscopy exposure time: 43 hours
17 LAEs
17/17 with SDSS
0/17 with ACS imaging
0/17 with MIPS 24 detections
FLS
•
•
•
•
•
UV spectroscopy exposure time: 22 hours
7 LAEs
7/7 with SDSS
0/7 with ACS imaging
5/7 with MIPS 24 detections
Non-LAEs
• Sources with GALEX
spectroscopy, within the same
redshift range as the LAEs,
that lack detections of Lyα.
107 non-LAEs
•COSMOS: 39 sources
•AEGIS: 39 sources
•FLS: 29 sources
Non-LAEs: black hashed histogram
LAEs: black histogram
Non-LAEs
• Sources with GALEX
spectroscopy, within the same
redshift range as the LAEs,
that lack detections of Lyα.
107 non-LAEs
•COSMOS: 39 sources
•AEGIS: 39 sources
•FLS: 29 sources
•non-LAEs serve as a control sample.
•Do LAEs have other physical properties that differ
from non-LAEs?
Non-LAEs: black hashed histogram
•Are these properties related to Lyα LAEs: black histogram
emission/escape?
UV, Optical, IR distributions
•FUV: LAEs are on average slightly more luminous
•Optical: LAEs are bimodal. The more luminous peak being
due in part to AGN contamination.
•IR: LAEs and non-LAEs have statistically similar distributions.
•Black hashed
histogram: non-LAEs
•Black Histogram: all
LAEs
•Blue Histogram: nonAGN LAEs
•Red Histogram: AGN
LAEs
Color Magnitude Diagram: LAEs not
special
• LAEs and non-LAEs span
2 orders of magnitude in
optical luminosity
• Have similar NUV-r
distributions
• At Mr>-21 the LAEs tend
to have bluer UV colors,
related to star formation
history.
LEGEND
•Blue: non-AGN LAEs
•Red: AGN LAEs
Squares: non-LAEs
(symbols sized
according to EWLyα)
Dust and Extinction
• LIR is derived from MIPS
24um and Chary & Elbaz
(2001) dust models.
• LAEs tend to have FUV
luminosities comparable to
the highest FUV luminosities
of non-LAEs.
• LAEs and non-LAEs span a
similar range of IRX.
• 1/3 of LAEs have bluer UV
colors than non-LAEs with
similar IRX.
• Dust extinction is not likely
to be the reason why Lyα
is not detected in the nonLAEs.
Dust, Extinction, UV color and Starbursts
• Create a grid of Bruzual &
Charlot (2003) Stellar
population synthesis galaxy
evolution models, varying
star formation history and
extinction. Half with and
half without a bursts of star
formation.
• The blue UV colors of LAEs
can be explained by
starbursts which form at
least 10% of the galaxy’s
stellar mass.
Morphology of LAEs
• Since there appears to be little evidence to
suggest that dust extinction is inhibiting Lyα
emission from the non-LAE sample, perhaps
the sources have different
morphologies/geometries?
• 20/76 LAEs & 64/107 non-LAEs with ACS
F814W imaging with 0”.03 pix -1 resolution.
LAE/non-LAE Morphologies
• The size of the LAEs correlates with optical luminosity.
• IFUV is likely underestimated for the large sources as the UV flux
is likely emitted from HII regions within the disks.
• There are no compact low luminosity non-LAEs, which is likely a
selection effect of the GALEX spectroscopy.
• Overall there is no major difference between the F814W
geometries of the LAEs and non-LAEs.
Outflows and Star formation
log (Outflow Velocity / km s-1)
• Heckman (2002) found that for a sample of nearby
sources, outflows were present in all galaxies with a star
formation rate surface density, ΣSFR > 0.1 M yr-1 kpc-2
• Rupke et al. (2004) found a correlation between SFR
and outflow velocity.
ΣSFR of LAEs and non-LAEs
• All compact LAEs (r< 7 kpc) are offset from non-LAEs in ΣSFR UV color.
• As ΣSFR is an indicator of outflows, this implies that the
presence of outflows is the discriminating factor between
LAEs and non-LAEs.
• Higher resolution UV imaging is required to investigate ΣSFR for
the larger disk sources where the UV sizes may be poorly
estimated by the F814W imaging.
Lyα Luminosity and the UV
continuum
• Lyα to first order is correlated with the
FUV luminosity, and to second order
with the FUV luminosity density (IFUV).
What influences the emergent Lyα flux?
• For the GALEX LAEs, no correlation is found between the
equivalent widths and UV luminosity, IR luminosity, or UV
extinction (as indicated by IRX).
• This in contrast to the results of Shapley et al. (2003), Kornei et
al. (2009) for z~3 LAEs, and Atek et al. (2009) for z~0 LAEs
Lyα Escape Fraction
• fesc = Lyα/Hα/8.7
Scarlata et al. (2010)
Hα/Hβ probes extinction, E(B-V)
Atek et al. (2009)
fesc, decreases with increasing extinction
Large scatter: due to kinematics and/or different dust
properties/distributions?
The Lyα escape fraction
• For 21 LAEs, flux calibrated optical spectroscopy was
obtained with measured Hα and Hα fluxes.
• 13/21 were obtained with long slit aperture matched
spectroscopy at Lick and KPNO.
• 8/21 were observed at Keck with 1”.5 long slit
spectroscopy.
The mean escape fraction for these
sources decreases with increasing
 ratios, indicating that dust
extinction does have an effect on the
amount of emergent Lyα flux
LEGEND
Dotted lines: CASE B recombintion ratios.
Dashed line: Calzetti extinction law.
Dot-dashed line: Cardelli exinction law
Escape Fraction and Equivalent width (EW)
• The EW increases for
increasing escape
fractions.
– The notable exceptions
are the sources without
aperture matched
spectroscopic
observations.
IRX vs Hα/Hβ
• Hα/Hβ is uncorrelated with both
IRX and FUV-NUV.
• The global extinctions (e.g. IRX)
do not correctly represent the line
of sight extinctions into the HII
regions as probed by the Balmer
emission lines.
Escape, Extinction and ΣSFR
The Facts:
• Sources with large EW >50 Å have both low E(B-V)≲0.07 and
high ΣSFR≳0.1
• Sources with low E(B-V) ≲0.07 and low ΣSFR ≲0.1 have low
EW.
• One source, that has the
lowest EW of the sample,
has both a low E(B-V) and
high ΣSFR but is highly
inclined (b/a = 0.36)
Escape, Extinction and ΣSFR
Conclusions:
• Both low E(B-V)≲0.07 and high ΣSFR≳0.1may be required for
large escape fractions and EW.
• Both high E(B-V) ≳ 0.07 and low ΣSFR ≲0.1can greatly reduce
the observed escape fraction and EW.
• For the one source with
both a low E(B-V) and high
ΣSFR, the highly inclined
(b/a = 0.36) morphology of
the source may lead to
increased scattering
opacity along the line of
sight which gives the
source such a low EW.
Results
• The GALEX LAEs are a heterogeneous assortment of
star forming galaxies. The distributions of IR-luminosities
and optical luminosities span 2 orders of magnitude. The
optical distribution is bimodal in part due to AGN
contamination as 11/22 LAEs with 0.0Mr< -22.5 are AGN.
• Dust extinction is likely not the dominant effect inhibiting
Lyα from escaping the non-LAE sources. It may play a role
at high LIR.
• LAEs have higher ΣSFR than non-LAEs: this implies that
the LAEs have outflows, while the non-LAEs do not.
Results (cont)
• The Lyα EWs of these sources are not correlated
with the global photometric properties for the
sample, including LFUV, and IRX (extinction).
– For IRX, this is likely because the IRX values
probe a large area than the lines of sight into the
HII regions that give rise to Lyα and the Balmer
lines.
• Both geometry, extinction, and ΣSFR play a role in
the Lya escape fraction: all 3 factors can inhibit
the observed escape along the line of sight.
Project 2: Future Work
• Obtain aperture matched optical
spectroscopy of entire GALEX LAE sample.
• Obtain high resolution UV spectra of sources
with Cosmic Origins Spectrograph on HST
– Measure kinematics
• Obtain high resolution UV images of the
LAEs. Lyα imaging?
– Compare ΣSFR vs kinematics
• Compare/contrast the GALEX LAEs to high
redshift samples (i.e. SFH, M*, size, EW)
BACKUP SLIDES
Measuring Oxygen Abundances
• O3N2 {Pettini & Pagel et al 2005}
– Calibration to Te abundance data, O/H = f(O3N2)
– O3N2 = [OIII]/ / [NII]
– Not corrected for ionization
• M91 {McGaugh et al (1991)}
– Photoionization Model, O/H = f(R23)
– R23= ([OIII]5007+[OIII]4959 + [OII]3727)/ 
– O/H has 2 values for each R23 value
• T04 {Tremonti et al 2004}
– Bayesian approach using photoionization models of
Charlot & Longhetti (2001)
– O/H < 8.2 : N/H ∝ (O/H)
– O/H > 8.2 : N/H ∝ (O/H)2
Measuring the Nitrogen Abundance
•
Log N+/O+ = Log [NII6584]/[OII3727] + 0.307 - 0.02 Log TNII - 0.726/TNII
–
–
–
from Pagel et al. (1992)
Empirical calibration from Te abundances
Applicability to high metallicities is unknown.
•TNII from Cloudy
photoionization models
•TNII~ 500K uncertainty
•TNII = .6065+.1600 log R23 + .1878 (log R23 )2 + .2803 (log R23)3
– from Thurston, Edmonds & Henry (1996)