View
6
Download
0
Category
Preview:
Citation preview
Introduction Data and Models Methods Results Summary
The VANDELS survey: Dust attenuation instar-forming galaxies at z = 3− 4
Cullen, F., et al.arXiv:1712.01292, accepted by MNRAS
March 7, 2018
Introduction Data and Models Methods Results Summary
Introduction
• Extinction: individual sight-lines, absorption + scattering outof the line-of-sight, related to the size distribution andcomposition of the dust grains
• Attenuation: extended sources, extinction + scattering backinto the line-of-sight, a strong dependence on the dustgeometry
• Main complication: the intrinsic shape of a galaxy SED is lesswell constrained
Some key questions on attenuation curve:
• changes with galaxy properties?
• redshift evolution?
• the presence/absence of the 2175 A UV bump?
Introduction Data and Models Methods Results Summary
Method 1
• Use an independent indicator to identify galaxies that aremore, or less, affected by dust, compare their observed SEDs
• E.g., Balmer decrement Hα/Hβ (Calzetti et al. 1994, 2000;Reddy et al. 2015, Battisti et al. 2016, 2017a, 2017b)
• Advantage: not having to assume the shape of a dust-freeSED
• Disadvantage: sensitive to the total-to-selective attenuationlaw, rather than the attenuation law directly
Introduction Data and Models Methods Results Summary
Method 2
• Assume/construct the underlying intrinsic SED of all sources,compare the observed SEDs with the intrinsic SED
• E.g., Scoville et al. 2015
• Advantage: directly probe the wavelength dependence of theattenuation curve
• Disadvantage: suffers from an increasing systematic errorsrelated to the choice of the intrinsic SED
• The problems will be mitigated somewhat at high redshiftswhere star-formation histories are predicted to become selfsimilar across the galaxy population
Introduction Data and Models Methods Results Summary
Sample selection
VANDELS Survey:
• ESO spectroscopic survey
• N ≈ 2000 galaxies at 1.0 < z < 7.0 inCDFS and UDS
• Instrument: VLT/VIMOS
• R ≈ 600, λ = 4800− 10000 A
• To be completed in January 2018
Selecting from 1502 VANDELS spectra:
• IAB ≤ 27.5 and HAB ≤ 27 (originalVANDELS selection criteria)
• the most secure redshift of3.0 ≤ z ≤ 4.0 (242)
• log(M∗/M�) ≤ 10.6 (236)
Final sample:
• 1.6× 108 ≤M∗/M� < 4.0× 1010,〈M∗〉 = 4.5× 109 M�
• 〈z〉 = 3.49
8.5 9.0 9.5 10.0 10.5 11.0log10(M? / M�)
−1.0
−0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
log 1
0(S
FR
/M�
yr−
1)
Speagle et al. 2014
Schreiber et al. 2015
Introduction Data and Models Methods Results Summary
Photometry
Four samples: CDFS-HST, CDFS-GROUND, UDS-HST, andUDS-GROUND
• CDFS-HST and UDS-HST: ∼ 50%, within CANDELS regions(CDFS and UDS), deep optical to NIR HST imaging,Spitzer/IRAC observations
• λ = 0.5− 8.0 µm, rest-frame UV to NIR SED out toλrest ∼ 2.0 µm
• CDFS-GROUND and UDS-GROUND: ∼ 50%, relies onpredominantly ground-based imaging
• λ only extends out the K-band (∼ 2.2 µm), rest-frame UV tooptical SED out to λrest ∼ 0.6 µm
Introduction Data and Models Methods Results Summary
Observed SED
Two publicly-available SED-fitting codes:
• EAZY (Brammer et al. 2008)
• linear combinations of a limitedstandard template set
• fixing z to spectroscopic redshiftmeasured from the VANDELS spectra
• LePhare (Ilbert et al. 2008)
• determine M∗
• Setting: constant SFH, BC03, Calzettiet al. (2000) attenuation curve
• Final observed SEDs: averaged thebest-fitting LePhare and EAZY SEDs
10−20
10−19
10−18
log 1
0(fλ
/er
g/s/
cm2/A
)
CDFS-HSTz=3.704 χ2
r = 0.35
10−20
10−19
10−18
log 1
0(fλ
/er
g/s/
cm2/A
)
CDFS-GROUNDz=3.403 χ2
r = 0.90
−0.2 0.0 0.2 0.4 0.6log10(λobs/µm)
10−20
10−19
10−18
log 1
0(fλ
/er
g/s/
cm2/A
)
UDS-HSTz=3.969 χ2
r = 2.33
Introduction Data and Models Methods Results Summary
A homogeneous SED shape
Assumption: all galaxies in the sample have the same intrinsic UVto optical SED shape
• First Billion Years (FiBY) simulation
• Sample: 628 galaxies with8.2 ≤ log(M/M�) ≤ 10.6 at z = 4
• Based on the similarity of SFHs, andthe narrow range in metallicity, it isreasonable to assume that theunderlying SED shape will be similaracross all masses.
200 400 600 800 1000 1200Lookback Time / Myr
−2.0
−1.5
−1.0
−0.5
0.0
0.5
1.0
log 1
0(S
FR
/〈S
FR〉)
8.2 < log10(M?/M�) < 9.0
9.0 < log10(M?/M�) < 10.0
10.0 < log10(M?/M�) < 10.6
9 10log(M?/M�)
0.1
0.2
0.3
0.4
0.5
0.6
Z/Z�
Introduction Data and Models Methods Results Summary
Intrinsic SEDs:
Construct intrinsic SEDs:
• Assign each star particle an instantaneousstarburst SPS model based on age andmetallicity
• Sum up SEDs of all individual starparticles for each galaxy
• Pair each VANDELS spectrum with FiBYgalaxy based on mass
• Stack matched SEDs of FiBY galaxies
0.2 0.3 0.4 0.5 0.6λ / µm
100
101
102
log 1
0(F
λ/F
5500
)
FiBYz4-BPASSv2-100
FiBYz4-BPASSv2-100bin
FiBYz4-BPASSv2-300
FiBYz4-BPASSv2-300bin
SB99-v000-z002
SB99-v000-z008
Introduction Data and Models Methods Results Summary
Intrinsic SEDs:
Construct intrinsic SEDs:
• BPASSv2 + CLOUDY
• BPASSv2-100bin: binary star evolution +an IMF cutoff of 100 M�
• BPASSv2-100: single-star evolution + anIMF cutoff of 100 M�
• BPASSv2-300bin: binary star evolution +an IMF cutoff of 300 M�
• BPASSv2-300: single-star evolution + anIMF cutoff of 300 M�
• S99-v00-z***: constant SFH +Starburst99 + 100 Myr +Z∗ = 0.002(0.008)
0.2 0.3 0.4 0.5 0.6λ / µm
100
101
102
log 1
0(F
λ/F
5500
)
FiBYz4-BPASSv2-100
FiBYz4-BPASSv2-100bin
FiBYz4-BPASSv2-300
FiBYz4-BPASSv2-300bin
SB99-v000-z002
SB99-v000-z008
Introduction Data and Models Methods Results Summary
Fitting method
Basic formula:
fλ,obs = φfλ,inte−τλ
Aλ = 1.086τλ
1.086× ln(fλ,int/fλ,obs) = Aλ −AV
Adopting a second-order polynomial as a function of x = 1/λ forAλ:
Aλ = a0x+ a1x2
1.086× ln(fλ,int/fλ,obs) = a0x+ a1x2 −AV
RV =AV
AB −AVkλ =
AλRVAV
Introduction Data and Models Methods Results Summary
Wavelength windows for fitting
Wavelength pixels used to fit:
• an estimated nebular contribution of < 10%
• free from stellar absorption/emission features
0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60Wavelength / µm
log(
Fλ)
fesc = 0.0
fesc = 0.5
fesc = 1.0
Introduction Data and Models Methods Results Summary
An example fit
2 3 4 5 6 7 8
λ−1 / µm−1
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
2.00
1.08
6ln
(Fi/
Fo)
0.2 0.3 0.4 0.5 0.6
λ / µm
1
2
3
4
5
6
7
Aλ/A
V
Calzetti
SMC
Reddy+ 2015
Charlot+Fall 2000
Introduction Data and Models Methods Results Summary
Simple Simulation
• generated 1000 M∗ within8.2 ≤ log(M∗/M�) ≤ 10.6
• assigned AV using the AV −M∗relation for a Calzetti law fromMcLure et al. (2017a)
• FiBY-BPASSv2-100bin template +nebular component
• perturbed flux assuming a 20% error
Using the defined wavelength regions, it ispossible to recover the underlying Calzettilaw and the input AV −M∗ relation.
0.2 0.3 0.4 0.5 0.6
λ / µm
1.0
1.5
2.0
2.5
3.0
3.5
Aλ/A
V
8.5 9.0 9.5 10.0 10.5log(M∗/M�)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
AV
Introduction Data and Models Methods Results Summary
Average shape of the attenuation curve
1
2
3
4
Aλ/A
V
FiBY-BPASSv2-100: N=178/236
Calzetti
SMC
Reddy+ 2015
Charlot+Fall 2000
FiBY-BPASSv2-100bin: N=224/236
1
2
3
4
Aλ/A
V
FiBY-BPASSv2-300: N=192/236 FiBY-BPASSv2-300bin: N=229/236
0.2 0.3 0.4 0.5 0.6
Wavelength / µm
1
2
3
4
Aλ/A
V
SB99-v00-z002: N=233/236
0.2 0.3 0.4 0.5 0.6
Wavelength / µm
SB99-v00-z008: N=232/236
• Failed fit: when observedSED was bluer than theintrinsic SED
• ffailed < 5% for half ofthe templates
• The majority ofattenuation curves areshallow and Calzetti-likein shape.
Introduction Data and Models Methods Results Summary
AV −M∗ relation
• McLure et al. (2017a): 8407galaxies at 2 < z < 3
• S99-v00-z008, FiBY-BPASSv2binary star templates: excellentagreement with both the McLureet al. (2017a) data and theempirical relation for a Calzettidust law at log(M∗/M�) > 9.5.
• S99-v00-z002, FiBY-BPASSv2single-star templates: incompatiblewith the McLure et al. (2017a)data
8.5 9.0 9.5 10.0 10.5 11.0log10(M?/M�)
0.0
0.5
1.0
1.5
2.0
AV
FiBY-BPASSv2-300bin
Cal
zett
i
SMC
McLure et al. 2017
Cullen et al. 2017
Introduction Data and Models Methods Results Summary
AV −M∗ relation
8.5 9.0 9.5 10.0 10.5 11.0
log10(M?/M�)
0
1
2
3
4
5
A16
00
This work
This work - Calzetti
This work - SMC
This work - Calzetti/SMC
AM+2016
Pannella+2015
Heinis+2014
Ibar+2013
Kashino+2013
Garn+Best 2010
8.5 9.0 9.5 10.0 10.5 11.0log10(M?/M�)
0.0
0.5
1.0
1.5
2.0
AV
FiBY-BPASSv2-300bin
Cal
zett
i
SMC
McLure et al. 2017
Cullen et al. 2017
• M∗ is a good proxy for attenuation.
• The AV −M∗ relation for normal star-forming galaxies with log(M∗/M�) > 9.5does not evolve between z = 0 to z ∼ 5.
Introduction Data and Models Methods Results Summary
Parameterization of the attenuation curve
Taking FiBY-BPASSv2-300bin as fiducial intrinsic SED template:
AλAV
=0.587
λ− 0.020
λ2, 0.12 < λ < 0.63 µm
kλ =2.454
λ− 0.084
λ2
RV = 4.18± 0.29
The results are fully consistentwith the Calzetti attenuationlaw.
0.2 0.3 0.4 0.5 0.6Wavelength / µm
4
6
8
10
12
14
16
18
kλ
Reddy
SMC
Calzetti
VANDELS
3.0 3.5 4.0 4.5 5.0RV
0
10
20
30
40
50
60
N
Cal
zett
iet
al.
2000
Introduction Data and Models Methods Results Summary
Attenuation curve at low masses
• constructed stacked intrinsicSED from FiBY withlog(M∗/M) < 9.0
• a steepening of the attenuationcurve at the lowest M∗
• remain consistent with theCalzetti law within 1σ, and ruleout SMC-like curve at > 1σ
0.15 0.20 0.25 0.30
Wavelength / µm
1.5
2.0
2.5
3.0
3.5
Aλ
/AV
Calzetti
SMC
FiBY-BPASSv2-300bin
S99-v00-z008
Introduction Data and Models Methods Results Summary
Evidence for a 2175 A UV bump?
Aλ =AV4.05
[k(λ)Calzetti +D(λ)]
D(λ) =Eb(λ∆λ)2
(λ2 − λ20) + (λ∆λ)2
where λ0 = 2175 A ∆λ = 350 A,AV = 0.8.
• 122 galaxies at 3.0 < z < 3.5
• robustly rule out the presence ofa strong UV bump withEb > 1.0
• a weak 2175 A UV bumpfeature with Eb ∼ 1
0.17 0.18 0.19 0.20 0.21 0.22
Wavelength / µm
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Fλ
/10−
19er
g/s/
cm2 /A
Eb = 0.0
Eb = 1.0
Eb = 2.0
Eb = 4.0
Introduction Data and Models Methods Results Summary
Summary
• The intrinsic shape of the UVoptical SED of SFGs at z = 3− 4 isapproximately constant across 8.2 < log(M∗/M�) < 10.6.
• The average attenuation curve shapes are consistent with a greyCalzetti-like attenuation law within ±1σ.
• No evidence is found for a steep SMC-like attenuation law.
• A subset of the intrinsic template sets yield results which areconsistent with the observed data, and with the AV −M∗ relationpredicted for a Calzetti-like attenuation law.
• The AV −M∗ relation does not evolve over 0 < z < 5.
• RV = 4.18± 0.29 (the original Calzetti value of RV = 4.05± 0.80).
• Tentative evidence for steeper attenuation curve shapes atlog(M∗/M�) < 9.0 is found.
• A weak 2175 A UV bump in the average attenuation curve at z ∼ 3is suggested by spectra of 122 galaxies at 3.0 < z < 3.5.
Recommended