Simultaneous Modeling of the Stellar and Dust Emission in Distant Galaxies: Implications for Star Formation Rate Measurements

0.62 MB·

by  Dyas Utomo

Draftversion January23,2014
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
SIMULTANEOUS MODELING OF THE STELLAR AND DUST EMISSION IN DISTANT GALAXIES:
IMPLICATIONS FOR STAR FORMATION RATE MEASUREMENTS
Dyas Utomo1, Mariska Kriek1, Ivo Labb´e2, Charlie Conroy3, and Mattia Fumagalli2
Draft version January 23, 2014
ABSTRACT
Wehaveusednear-ultraviolet(NUV)tomid-infrared(MIR)compositespectralenergydistributions
4
(SEDs) to simultaneously model the attenuated stellar and dust emission of 0.5 . z . 2.0 galaxies.
1
These composite SEDs werepreviously constructedfrom the photometric catalogsof the NEWFIRM
0
Medium-Band Survey, by stacking the observed photometry of galaxies that have similar rest-frame
2
NUV-to-NIR SEDs. In this work, we include a stacked MIPS 24µm measurement for each SED type

n to extend the SEDs to rest-frame MIR wavelengths. Consistent with previous studies, the observed
a MIR emission for most SED types is higher than expected from only the attenuated stellar emission.
J
We fit the NUV-to-MIR composite SEDs by the Flexible Stellar Population Synthesis (SPS) models,

1 whichinclude both stellaranddust emission. We comparethe best-fit starformationrates(SFRs) to
2 the SFRsbasedonsimple UV+IRestimators. Interestingly,the UVandIRluminositiesoverestimate
SFRs – compared to the model SFRs – by more than ∼1dex for quiescent galaxies, while for the
]
A highest star-forming galaxies in our sample the two SFRs are broadly consistent. The difference
in specific SFRs also shows a gradually increasing trend with declining specific SFR, implying that
G
quiescent galaxies have even lower specific SFRs than previously found. Contributions from evolved
. stellar populations to both the UV and the MIR SEDs most likely explain the discrepancy. Based
h
on this work,we conclude that SFRs should be determined from modeling the attenuated stellar and
p
dust emission simultaneously, instead of employing simple UV+IR-based SFR estimators.
-
o Subject headings: galaxies: stellar contents — ISM: dust — galaxies: high-redshift — galaxies: star
r formation
t
s
a
[ 1. INTRODUCTION using a template spectrum. However, as this method is

1 SEDsofgalaxiescontainalotofinformationregarding comparable to fitting two photometric data points by a
singlegalaxytemplatewithonlytheamountofdustasa
v theirphysicalproperties,suchastheSFR,starformation
freeparameter,it willlikelyresultinlargeuncertainties.
2 history (SFH), metallicity, age of the stellar population,
To derive accurate stellar population properties from
7 stellarmass,andtheamountofdust(e.g.,Conroy2013).
UV-to-IRSEDs,itis importanttoexplorethe fullpossi-
4 These quantities can be extracted by fitting SEDs of
ble range in stellar populations. Fortunately, SPS mod-
5 galaxieswithSPSmodels(e.g.,Bruzual & Charlot2003;
elsthatincorporateboththeattenuatedstellaranddust
. Maraston 2005; da Cunha et al. 2008; Conroy et al.
1 emissions have recently become available (Conroy et al.
2009). SPS modeling is currently one of the most popu-
0 2009; Noll et al. 2009; da Cunha et al. 2008). These
lar and powerful methods to study the SFHs and stel-
4
models assume that the energy of the attenuated stel-
1 lar mass build-up of galaxies over cosmic time (e.g.,
lar light is reradiated in the IR.
: Labb´e et al. 2010a; Wuyts et al. 2011; Brammer et al.
v 2011; Muzzin et al. 2013). In this Letter, we extend the UV-to-NIR composite
i SEDs by Kriek et al. (2011) to MIR wavelengths, and
X SPSmodels aregenerallycombinedwith a dustmodel
simultaneously fit the stellar and dust emission with the
toaccountfordustattenuationofthestellarlight. How-
r updated Flexible SPS (FSPS) models by Conroy et al.
a ever, modeling just the attenuated stellar emission re-
(2009). Weadoptthefollowingcosmologicalparameters:
sultsinsignificantuncertaintiesonthederiveddustcon-
(Ω ,Ω ,h)= (0.27, 0.73, 0.7).
tent and consequently, other properties. Extending the m Λ
stellar SEDs to IR wavelengthsimproves the constraints 2. DATA
on the dust parameters, which in turn yields more ac-
In this work we make use of the composite NUV-
curate stellar population properties. The inclusion of IR
data has primarily resulted in empirical SFR indicators. to-NIR SEDs by Kriek et al. (2011), which were con-
To obtain the sum of the unobscured and obscured SFR structedusing the photometric catalogsfromthe NMBS
(Whitaker et al. 2011). The NMBS is a survey in the
of a galaxy, the uncorrected UV SFR is combined with
the IR luminosity. The IR luminosity is often only mea- COSMOS(Scoville et al.2007)andAEGIS(Davis et al.
sured at 24µm and the total IR luminosity is estimated 2007) fields, which uses five medium-bandwidth NIR fil-
ters in the wavelength range 1 − 1.8µm designed for
NEWFIRM (Autry et al. 2003) on the Mayall 4-m tele-
1Astronomy Department, University of California, Berkeley, cope (van Dokkum et al. 2009). The NIR medium-band
CA94720,USA;[email protected]. photometry has been combined with the publicly avail-
2Leiden Observatory, Leiden University, P.O.Box 9513, NL-
able data at NUV-to-NIR wavelength as described in
2300RALeiden,Netherlands.
3DepartmentofAstronomy&Astrophysics,UniversityofCal- Whitaker et al. (2011).
ifornia,SantaCruz,CA95060, USA. The original composite SEDs were constructed as fol-
2
lows. First, galaxies at 0.5.z .2.0 with S/NK−band >
25wereclassifiedintospectraltypesbasedonsimilarities
in their NUV-to-NIR rest-frame SEDs. The number of
galaxies in each type varies between 22 and 455. Next,
the SEDs of individual galaxies in each type were de- Template Archival image Model Cleaned image
redshifted and scaled to the same reference frame. Fi-
nally, the flux was averaged in wavelength bins. See
Kriek et al. (2011) for more details on the procedure. 123 13 75 5 82 15 101 15
Thistechnique resultedin32compositeSEDs, whichin-
clude ∼3500 galaxies.
Here, we extend these SEDs to rest-frame MIR wave-
lengthsbyaddingMIPS24µmdata. Galaxieswithactive Type 1 Type 7 Type 22 Type 31
galacticnuclei(AGNs)tendtohaveawarmdustcompo-
nent at MIR wavelengths (Fritz et al. 2006; Feltre et al.
Fig. 1.— Top: Source deblending technique. First, the sources
2012). Since we are interested in studying dust emis- are identified using SExtractor from the K-band template image.
sionfromreprocessedstellarlight,wewanttoavoidcon- Next, source models are constructed from the MIPS and K-band
tamination by AGNs. Therefore, any galaxies that have PSFs. Finally, all sources except the primary target are removed
detectedLX ≥1042erg/sintheChandraCOSMOSSur- rbeysusultbintrgascttainckgetdhiemirasgceasleodrdmeroeddelbeydSflEuDxesst.yBpeo.ttTomhe:pAhosetolemcteitornyoisf
vey(Elvis et al.2009)havebeenremoved. Furthermore, measuredinsidetheinnercircle,whilethebackgroundiscalculated
we reject galaxies which are identified to host obscured betweenthetwooutercircles. Thevaluesinthetopleftandright
AGNs based on their IRAC colors,following the criteria cornersarethenumberofgalaxiesandS/N,respectively.
byDonley et al.(2012). Asaresult,oursamplehasbeen
reduced by ∼3%. 3. SED FITTING
The simplest method to extend the composite SEDs In order to derive physical properties, we fit the com-
is taking the average of the scaled 24µm fluxes from posite SEDs with the FSPS models (Conroy et al. 2009;
the NMBS catalogs. However, many sources are unde- Conroy & Gunn 2010). We use the BaSeL spectral li-
tected, and thus, we stack the images per SED type to brary (Lejeune et al. 1997, 1998; Westera et al. 2002),
obtain deeper photometry. For this purpose, we use the Padova isochrones (Bertelli et al. 1994; Girardi et al.
archival mosaic image from the MIPS S-COSMOS Sur- 2000; Marigo et al. 2008), and dust emission models
vey (Sanders et al. 2007). In order to remove contami- of Draine & Li (2007). Motivated by the work by
nation by surrounding sources, the MIPS image of each Kriek & Conroy (2013), which was based on the same
individualgalaxyhasbeencleanedbeforestacking,using composite SEDs, we assume a dust attenuation curve
the following steps. First, a model is constructed for all with RV = 4.05 and a UV dust bump which is 20% of
surrounding sources,using the higher resolution K-band thestrengthoftheMilkyWaybump. Thethreeparame-
image. The K-band image is convolvedby a convolution tersoftheDraine & Li(2007)dustemissionmodel;U
min
curve derived from the point spread functions (PSFs) of (specifies the minimum radiation field strength in units
theK-bandandMIPSdata. Next,wesubtractthemod- of the Milky Way value), γ (specifies the relative contri-
eled fluxes of all surrounding sources, to get clean im- butionofdustheatedatU andatU ≤U ≤U ),
ages with a radius of ∼ 40′′ (Labb´e et al. 2010b). This and q (the fractionof grainmimnass in Pomlyincyclic Arommaaxtic
techniqueisillustratedinFigure1. Theremainingback- Hydrocarbon (PAH) form), are set to their default val-
groundhasbeenremovedbysubtractingtheaverageflux ues, i.e. 0.01, 1.0, and 3.5%, respectively.
withina7−13′′annulus(yellowcirclesinFigure1)from We also assume a delayed-τ SFH of the form SFR
these cleanedimages. Then, the cleanedimages for each ∝ t exp(−t/τ) and a Kroupa (2001) initial mass func-
type are stacked into one image, weighted by the scal- tion (IMF). The star formation timescale (τ), age, and
ing factors that are used in the NUV-to-NIR SEDs. We dust extinction (AV) are left as free parameters, with a
performafinalbackgroundsubtractionfromthestacked minimum log (τ/yr) and log (age/yr)of 7.5. The metal-
image. licity is assumed to be Solar.
The total flux is measured for each stacked image in- The fitting is done by minimizing
side a 3.5′′ aperture (red circles in Figure 1) and is cor-
rected for missing flux outside the 3.5′′ aperture using χ2 =X(Fi−aTi)2, (1)
theaperturecorrectionfactorfromtheMIPSinstrument i δFi2
handbook. The flux errors are derived using bootstrap
resampling of the individual galaxies within the bins. where
The stacked MIPS fluxes (spanning ∼ 8−15µm rest- a= PFiTi/δFi2 (2)
fsrtarumcet)NarUeVc-otmo-bMinIeRdcwoimthptohseiteNUSEVD-tso.-NCIoRmdpaotsaitteoficlotner- PTi2/δFi2
curves are constructed for the added 24µm data points, isthescalingfactorbetweentheobservedfluxFi andthe
by adding the normalized and de-redshifted 24µm filter templatefluxTiofthemodellibraries. Thetemplateflux
curvesfor eachindividualgalaxy. Errorsonthe effective is calculated by convolving the flux with the composite
wavelengthsare derived by bootstrapresampling the in- filter curves. The flux errors δFi are set to be 5% of the
dividual galaxies within the bins. fluxes Fi, to avoid that very small flux errors dominate
the fit, and to ensure that all data points have equal
weight in the χ2-calculations.
We calibrate the confidence intervals using Monte
3
Type 32 Hα PAH 0.8 1 0 1 2
stellar fit A
V
4 0.6
6
Type 31 Fstellar 0.4 2 8 11 1167
/
3 og Fobs 0.2 4 15101143191822412202252289
l 12 9
0.0 5 30
7 23
2 27
−0.2 26 31
Type 22
3 32
]
nits −0.4 −12.0 −11.5 −11.0 −10.5 −10.0 −9.5 −9.0
u 1

y
ar 0.8 0 1 2
tr stellar+dust fit AV
bi Type 11
r 0.6
a
og [λfλ 0 Fstellar+dust 0.4
l −1 Type 6 log F/obs 00..02 41212 96 8515101111431911168722412202252262728930
7 23 32
3 31
−0.2
−2
−0.4
−12.0 −11.5 −11.0 −10.5 −10.0 −9.5 −9.0
log SSFR [yr−1]
Type 1
−3 stellar fit Fig.3.—Ratioofobservedtoexpected 24µmfluxvs. SSFRfor
stellar fitting (top) and stellar+dust fitting (bottom). TheSSFRs
stellar+dust fit onthex-axisarebasedonthestellar+dustfitting. Colorindicates
thedustattenuationintheV-band(AV)basedonthestellar(top)
data and stellar+dust (bottom) best-fit models. The numbers are the
SEDtypes whichareorderedbydecreasingvalueofD(4000).
0.1 1.0 10.0
rest-frame wavelength [µm]
intervals.
In order to get the absolute SFR, we multiply the in-
Fig.2.— A selection of NUV-to-MIR composite SEDs, ordered stantaneous specific SFR, derived from the SED-fitting,
by decreasing D(4000) strength which is a measure of the age of by the average stellar mass of the galaxy type. The av-
thestellarpopulation. Redcirclesrepresentthestackeddata. The
erage mass is derived by assuming the same M/L for all
best fits to the stellar and stellar+dust emission are represented
by the blue and black curves, respectively. The error bars are individual galaxies within one type.
smaller than the data points. Consistent with previous studies, TheresultsareshowninFigure2foraselectionofSED
the expected MIPS fluxes from the ‘stellar fitting’ differ from the types which range from star-forming, to post-starburst,
observed fluxes. Data points affected by the Hα emission line, as
to quiescent galaxy types. The SEDs are fitted in two
clearlyseeninstar-forminggalaxytypes,areexcludedinthefitting
process. ways; by excluding and including the MIPS fluxes. We
refer to the former as ‘stellar fitting’ and the latter as
‘stellar+dust fitting’. For few star-forming and young
Carlo simulations, where the fluxes are perturbed ac- galaxy types, the NUV region does not have an excel-
cording to a Gaussian distribution, and determine the lent fit. As shown in Kriek & Conroy (2013) better fits
best-fit parameters. We run 200 simulations and deter- canbe obtained by allowingboth the dust slope and the
mine the χ2-level that encloses 68% of the simulation’s UV bump strength to vary. However, as this would not
best-fits (e.g., Papovich et al. 2001; Kriek et al. 2009). significantly change the results of this paper, and would
Error bars in all figures correspond to these confidence make the fitting impractical, we have decided to fix the
4
2 −9
24 32
19 20 31
/yr]⊙ 1 12 976 5815101111143711826212225328320921732623 1yr]− −10 15 1143191116872241220222352262728930
[M+IR 4 21 [V+IR 12 967 85 1011
V 3 U
RU FR 431
SF 0 SS −11 2
g g
o o
l l log τ=8, A =0,1,2
V
log τ=9, A =0,1,2
V
−1 −12
−1 0 1 2 −12 −11 −10 −9
log SFRSED [M /yr] log SSFRSED [yr−1]
⊙
2.0 2.0
0 50 100 150
)D 1.5 12 )ED 1.5 12 Hα+[NII]+[SII] EW [Å]
E S
S R
SFR 34 9 SSF 43 9
/ R 1.0 1 76 815 / R 1.0 1 67 815
+I 2 5 +I 2 5
UV 0.5 1143 1924 UV 0.5 114319 24
FR 1011171186 FR 1011 111687
S 21 S 21
og ( 0.0 2222528292272603 g (S 0.0 2202223522627289
l 330132 o 303132
l
−0.5 −0.5
−1 0 1 2 −12 −11 −10 −9
log SFRSED [M /yr] log SSFRSED [yr−1]
⊙
Fig. 4.—Top: Comparisonbetween(S)SFR basedonLUV andLIR vs(S)SFR basedonthestellar+dustSEDfitting. Solidblacklines
areone-to-onerelationships. Bottom: RatioofthetwoSFRs(SSFR)plottedagainstSFR(SSFR)basedonSEDfitting. Allplotsarecolor
codedwithHαequivalentwidth,andnumberedwithSEDtypessimilarasinFigure3. Curvesinthetoprightfigurearetheevolutionary
models of the SSFR for a delayed exponential model with τ = 8 and 9 for the red and blue curves, and AV = 0,1 and 2 for the solid,
dashed,anddotted curves,respectively. OnlygalaxieswithlogSSFRSED&−10lieclosetoone-to-onerelationship.
dust attenuation law. reprocessed stellar light at IR wavelengths. As MIPS
We compare the observed 24µm fluxes with the ex- is most sensitive at 24µm, the full IR luminosity is of-
pected fluxes, based on the best-fit stellar and stel- ten derived by extrapolating this one data point using
lar+dustmodels. Theratiosofthe observedtoexpected asingle averagegalaxytemplate (e.g.,Franx et al. 2008;
fluxes are shown in Figure 3. The top panel of Figure 3 Wuyts et al.2011). Here,weassesstheseSFRsusingour
illustrates that the observed MIPS fluxes become larger best-fit models to the dust and stellar emission.
than the expected stellar fit model fluxes with decreas- We use the monochromatic conversion template by
ing star formation activity. The difference between the Wuyts et al. (2008) to infer L from 24µm flux. This
IR
observedandmodeledfluxesismuchsmaller(<0.2dex) template is a luminosity independent template, derived
forthestellar+dustfit(bottompanelofFigure3). There by taking the log average of the exponents of the in-
is no correlation with AV. This result demonstrates the terstellar radiation field strength in the Dale & Helou
importance of including dust emission while modeling (2002) templates. Wuyts et al. (2011) found that this
galaxy SEDs, as just modeling the stellar emission may template is well matched with the SFR of star-forming
leadtosystematicbiasesinthederivedstellarpopulation galaxies based on UV+PACS data in the range of 0 <
properties. z <3. However,the accuracy of this template for quies-
cent and transition galaxies has not been assessed.
4. STAR FORMATION RATES Next, the total SFR based on both UV+IR can be
calculated by (Bell et al. 2005; Kennicutt 1998):
With the introduction of MIPS, it has become prac-
tice to measure SFRs using both the unobscured light
from young stars in the UV and the dust obscured and SFRUV+IR [M⊙/yr]=9.8×10−11(LIR+2.2LUV), (3)
5
−9 −9 )
D
E
1log SSFR [yr]−IR −−1101 0431212 967A1V851510111143129LL111687IR224122<022211352200627211891130LL31⊙32 1log SSFR [yr]−UV −−1101 0431212Hα 7967E5W85 1[5Å110]151114310911168722412202223522627289303132 (SSFR / SSFRUV+IRS 021−l3o13g2423 S13S02F92R27816 0[222y75221r171−2−161115]9623118398211024214159
−12 IR≥ ⊙ −12 log
−12 −11 −10 −9 −12 −11 −10 −9 −0.5 0.0 0.5 1.0 1.5
log SSFR [yr 1] log SSFR [yr 1] log (SSFR / SSFR )
SED − SED − IR UV
Fig.5.—SSFR basedonIR(left panel)andUV(central panel) areplottedversusSSFR basedonSEDfitting. BothIR andUV-based
SSFRs overpredict SED-based SSFR for galaxies withlow SSFR. Right panel: The ratio of UV+IR-basedSSFR to SED-based SSFR vs.
theratiooftheIRtoUVSSFR.ColorindicatesdustextinctionAV (left panel),Hαequivalent-width(central panel), andlogSSFRfrom
SEDsfitting(rightpanel). ThedataaredividedintotwocategoriesbyDaddietal.(2007a)basedonitsLIR (round andsquare markers).
The SFRUV+IR overestimates SFRSED by more than ∼1dex for quiescent galaxies, while for the highest star-forming galaxies the two
SFRsarebroadlyconsistent.
assuming a Kroupa (2001) IMF and luminosity in L⊙. dissecttheSSFRUV+IRinSSFRUV andSSFRIR. Wealso
Here, LUV is defined as 1.5 νLν at 2800 ˚A, which is plottheratiobetweenthelattertwoandcompareitwith
a rough estimate of the total integrated 1216−3000 ˚A the ratio between SSFRUV+IR and SSFRSED (Figure 5).
UV luminosity, and the factor of 2.2 accounts for the No dust correction was applied to derive SSFRUV.
unobscured light of young stars that is emitted outside For the majority of types, the SSFR excess is dom-
the 1216 − 3000˚A band (Bell et al. 2005). Note that inated by the MIR flux, while for a few, it is domi-
nated by the UV flux. This SSFR excess can be caused
L is derived using the flux interpolation at 2800 ˚A.
UV by the contribution of old (e.g., Fumagalli et al. 2013)
Thus,thismethodbasicallyfitstwodatapointsofthefull
and/or intermediate-age stars (e.g., Salim et al. 2009;
SED with one star-forming galaxy template, with only
Kelson & Holden 2010) to the MIRand UV light, which
the amount of obscuration as a free parameter. While
explains the strong correlation with SSFR. This finding
thismethodhasbeencalibratedusingactivestar-forming
is not surprising, as we only use a single star-forming
galaxies,it would not be surprising if it breaks down for
galaxy template (by Wuyts et al. (2011)) when estimat-
galaxies that have different stellar populations.
ing SFRs from UV+IR. Therefore, only young and star-
The comparison between (S)SFRSED and forming galaxies with SSFR & 10−10 yr−1 lie close to
(S)SFR are shown in Figure 4. Generally,
UV+IR one-to-one relation (see also Arnouts et al. (2013)). In
for galaxies with older stellar populations (i.e., lower
thiscase,L isarobustestimatoroftheSFR.How-
SSFR), SSFR is higher than SSFR , while UV+IR
UV+IR SED ever, there might be an upper limit where the agree-
younger and higher SSFR galaxies lie closer to one-to-
ment between the two methods breaks down again, as
one relation. In orderto assesswhether this discrepancy
Wuyts et al. (2011) found that SFR overpredicts
may be due to the fact that we use 2800˚A instead of UV+IR
SFR at high redshift (z &2.5) and at the high-SFR-
1600˚A, we calculate the expected SFRs using 1600 ˚A end (S&ED100M⊙/yr).
best-fit model fluxes and 24µm observed flux, but still Compton-thick AGNs with LX & 1043 erg/s could
find similar results. Wuyts et al. (2011) found that
also explain the discrepancy between SSFR and
SFR overestimates SFR , in particular for SED
UV+IR SED SSFR , due to their MIR excess (Daddi et al.
high SFRs, when short star formation timescales were UV+IR
2007b). However, we removed AGNs identified by their
allowed. Our results do not change significantly when
strong X-ray flux or by an IRAC upturn (Donley et al.
we restrictthe starformationtimescaleto logτ >8.5or
2012). Nonetheless, we cannot rule out contributions
when we assume an exponentially declining SFH. Thus,
from low luminosity AGNs, and X-ray stacks of the
we argue that SSFR overestimates SSFR for
UV+IR SED same composite SED sample indeed indicate low levels
galaxies with logSSFR . −10, and the discrepancy
of black hole accretion (Jones et al. 2013). Daddi et al.
becomes larger with decreasing SSFR.
(2007a) also reported that MIR excess galaxies have
5. DISCUSSION LIR & 1011L⊙. We check this possibility, but we do
not find such trend in our data (Fig. 5).
We find that SFR overestimates SFR by Nordon et al.(2010)proposedthatthediscrepancybe-
UV+IR SED
more than ∼ 1dex for quiescent galaxies, while for the tween SFR and SFR is due to excess in PAH
SED UV+IR
most active star-forming galaxies in our sample the two emission,ratherthanobscuredAGNs. Weconductatest
SFRs are broadly consistent (Figure 4). Our results are for this hypothesis by varying γ, U , and q. We find
min
consistentwithrecentfindingsbyFumagalli et al.(2013) thatthe SFR is not sensitive to these variations,but
SED
andSalim et al.(2009). Inordertoinvestigatethecause SFR can be affected due to the used monochro-
UV+IR
of the difference between SSFR and SSFR , we matic conversion template. Therefore, higher observed-
SED UV+IR
6
frame24µmfluxesduetovariationsinPAHemissioncan termine SFRs. We find that (S)SFR overpredicts
UV+IR
lead to discrepancies between SFR and SFR . (S)SFR for galaxies with log SSFR . −10, and the
SED UV+IR SED
Lastly, we mention that circumstellar dust around discrepancybecomes increasinglylargerfor lowerSSFR.
AGB star is notyetincluded in the FSPS models,but is The discrepancy is due to both UV and MIR luminosi-
subject of ongoing work. In this context, it is interest- ties, though the MIR is the dominant contributor for
ing to note that the post-starburstgalaxytypes (6 & 7), most SED types. Contributions from obscured and un-
for which we expect the highest contribution from AGB obscuredoldand/orintermediate-agestellarpopulations
stars, have an MIR excess, though it is not larger than to the MIR and UV luminosities are the likely explana-
for the other quiescent galaxy types. tion for the overestimated SFR .
UV+IR
Basedonourresults,weconcludethatSFRsshouldbe
6. SUMMARY determined from modeling stellar and dust emission si-
multaneously,insteadofjustmeasuringtheUVandMIR
In this letter, we use NUV-to-MIR composite SEDs
luminosities. An important implication of our work is
to simultaneously model the stellar and dust emission
that quiescent galaxieshave even lower SFRs than what
in distant galaxies. NUV-to-NIR composite SEDs had
was previously found, based on UV and IR luminosi-
previouslybeenconstructedfromtheNMBSphotometry
of ∼ 3500 galaxies at 0.5. z . 2, by matching galaxies ties. However, young star-forming galaxies with SSFR
& 10−10 yr−1 lie close to one-to-one relation, and thus
with similar SED shapes. In this work, we extend the
L is a robust SFR estimator.
SEDswitha stackedMIPS24µmdatapoint,resultingin UV+IR
multi-wavelengthSEDs spanning from ∼0.2 to 15µm in The composite SEDs currently only extend to MIR
wavelengths, and thus the SFRs derived from the mod-
rest-frame wavelength.
eled dust and stellar emission may still suffer from sys-
Stellarpopulationpropertiesarederivedbyfitting the
tematics. In future studies we will extend the SEDs to
composite SEDs with the FSPS models, which include
FIR wavelengths, to measure the full bolometric lumi-
bothstellaranddustemission. Consistentwithprevious
nosity and more accurately measure the total SFR.
studies, we find that the predicted MIPS flux, based on
fitting just the stellar emission, is inconsistent with the
observed MIPS flux for most galaxy SED types.
We use the best-fit SFRs fromthe full stellaranddust We thank the NMBS and COSMOS collaborations
fitting to assess SFRs determined from the UV and IR for making their catalogs publicly available, and Marijn
luminosities, currently the most popular method to de- Franx and Edward Taylor for useful discussions.
REFERENCES
Arnouts,S.,LeFloc’h,E.,Chevallard,J.,etal.2013,A&A,558, Kelson,D.D.,&Holden,B.P.2010,ApJ,713,L28
A67 Kennicutt,Jr.,R.C.1998,ARA&A,36,189
Autry,R.G.,Probst,R.G.,Starr,B.M.,etal.2003,inSociety Kriek,M.,&Conroy,C.2013,ApJ,775,L16
ofPhoto-OpticalInstrumentation Engineers(SPIE)Conference Kriek,M.,vanDokkum,P.G.,Labb´e,I.,etal.2009,ApJ,700,
Series,Vol.4841, InstrumentDesignandPerformancefor 221
Optical/InfraredGround-basedTelescopes, ed.M.Iye& Kriek,M.,vanDokkum,P.G.,Whitaker,K.E.,etal.2011, ApJ,
A.F.M.Moorwood,525–539 743,168
Bell,E.F.,Papovich, C.,Wolf,C.,etal.2005,ApJ,625,23 Kroupa,P.2001, MNRAS,322,231
Bertelli,G.,Bressan,A.,Chiosi,C.,Fagotto, F.,&Nasi,E.1994, Labb´e,I.,Gonz´alez, V.,Bouwens,R.J.,etal.2010a,ApJ,716,
A&AS,106,275 L103
Brammer,G.B.,Whitaker, K.E.,vanDokkum,P.G.,etal. —.2010b, ApJ,708,L26
2011, ApJ,739,24 Lejeune,T.,Cuisinier,F.,&Buser,R.1997,A&AS,125,229
Bruzual,G.,&Charlot,S.2003, MNRAS,344,1000 —.1998, A&AS,130,65
Conroy,C.2013,ARA&A,51,393 Maraston,C.2005, MNRAS,362,799
Conroy,C.,&Gunn,J.E.2010,ApJ,712,833 Marigo,P.,Girardi,L.,Bressan,A.,etal.2008,A&A,482,883
Conroy,C.,Gunn,J.E.,&White,M.2009,ApJ,699,486 Muzzin,A.,Marchesini,D.,Stefanon, M.,etal.2013, ApJ,777,
daCunha,E.,Charlot,S.,&Elbaz,D.2008,MNRAS,388,1595 18
Daddi,E.,Dickinson,M.,Morrison,G.,etal.2007a, ApJ,670, Noll,S.,Burgarella,D.,Giovannoli,E.,etal.2009, A&A,507,
156 1793
Daddi,E.,Alexander,D.M.,Dickinson,M.,etal.2007b, ApJ, Nordon,R.,Lutz,D.,Shao,L.,etal.2010,A&A,518,L24
670,173 Papovich,C.,Dickinson,M.,&Ferguson,H.C.2001,ApJ,559,
Dale,D.A.,&Helou,G.2002,ApJ,576,159 620
Davis,M.,Guhathakurta, P.,Konidaris,N.P.,etal.2007, ApJ, Salim,S.,Dickinson,M.,MichaelRich,R.,etal.2009,ApJ,700,
660,L1 161
Donley,J.L.,Koekemoer,A.M.,Brusa,M.,etal.2012,ApJ, Sanders,D.B.,Salvato,M.,Aussel,H.,etal.2007,ApJS,172,86
748,142 Scoville,N.,Aussel,H.,Brusa,M.,etal.2007,ApJS,172,1
Draine,B.T.,&Li,A.2007,ApJ,657,810 vanDokkum,P.G.,Labb´e,I.,Marchesini,D.,etal.2009,PASP,
Elvis,M.,Civano,F.,Vignali,C.,etal.2009,ApJS,184,158 121,2
Feltre,A.,Hatziminaoglou,E.,Fritz,J.,&Franceschini,A.2012, Westera,P.,Lejeune,T.,Buser,R.,Cuisinier,F.,&Bruzual,G.
MNRAS,426,120 2002,A&A,381,524
Franx,M.,vanDokkum,P.G.,Schreiber,N.M.F.,etal.2008, Whitaker,K.E.,Labb´e,I.,vanDokkum,P.G.,etal.2011,ApJ,
ApJ,688,770 735,86
Fritz,J.,Franceschini,A.,&Hatziminaoglou,E.2006,MNRAS, Wuyts,S.,Labb´e,I.,Schreiber,N.M.F.,etal.2008,ApJ,682,
366,767 985
Fumagalli,M.,Labbe,I.,Patel,S.G.,etal.2013,ArXive-prints, Wuyts,S.,Fo¨rsterSchreiber,N.M.,Lutz,D.,etal.2011, ApJ,
arXiv:1308.4132 738,106
Girardi,L.,Bressan,A.,Bertelli,G.,&Chiosi,C.2000, A&AS,
141,371
Jones,T.M.,Kriek,M.,vanDokkum,P.G.,etal.2013,ArXiv
e-prints,arXiv:1312.3950