measuring the evolution of the star formation rate ... · summary • measured extended rest-frame...

Measuring the evolution of the star formation rate efficiency of neutral atomic hydrogen gas from z ~1– 4

Marc RafelskiGalactic Scale Star Formation

August 2012

Collaborators: Harry Teplitz

Arthur WolfeHsiao-Wen ChenXavier ProchaskaUV UDF team

/ 30Heidelberg 2012: Marc Rafelski

• Definition of Damped Lyα System (DLA): N(HI)≥ 2×1020 cm-2

• Distinguishing characteristics of DLAs : (1) Gas is Neutral (2) Metallicity is low: [M/H]=-1.3 (more on this later) (3) Molecular fraction is low: fH2~10-5

• DLAs dominate the neutral-gas content of the Universe out to z~4.5• DLAs cover 1/3 of the sky at z=[2.5,3.5]

2Wolfe et al. 2005

Damped Lyman Alpha Systems (DLAs)

Observed Wavelength (Å)

Rel

ativ

e Fl

ux

Rest Wavelength (Å)

/ 30Heidelberg 2012: Marc Rafelski 3

Kennicutt-Schmidt (KS) Relation

The Star Formation Rate (SFR) surface density goes as the total

gas surface density to a power law

Normal Disks

Star Forming

Kennicutt, 1998

β = 1.4± 0.15

Can rewrite it with column density N:

Nc=1.25×1020 cm−2

K= 2.5×10−4M⊙ yr−1 kpc−2

ΣSFR = K × (N/Nc)β


Bigiel et al. 2010a

Tightly Correlated HI and FUV emission in M83

Blue: FUV map (GALEX)

Red: HI contours (THINGS)

Arc Minutes

Arc

Min

utes


Col

umn

Den

sity

(N

) [c

m-2]

32 30 28 26 24Surface Brightness [AB mag/arcsec2]

1020

1021

1022

1023

Obs

erve

d Co

lum

n De

nsity

(N) [

cm2 ] z = 2.5

z = 3.0z = 3.5

Surface Brightness [mag arcsec-2]

5

Can we see DLAs in emission at z~3?• Gas Density ↔ SFR (KS)

• SFR ↔ FUV Lν

(Madau Kennicutt Calibration)

Only high resolution image sensitive enough is the Hubble Ultra Deep Field (UDF)

• Most DLAs:N ~ 2x1020 →3x1021 cm-2 Navg ~ 1x1021 cm-2

At z=3 1500 Å → 6000 Å- This puts it in the visible!

• Lν/area↔ Surface Brightness


Cumulative comoving SFR density for DLAs

Wolfe & Chen 2006

(A proxy for NHI via the KS relation)

ΣSFR(N) = K(N/Nc)β

38 36 34 32 30 28 26Surface Brightness (µV) [mag arcsec 2]

5

4

3

2

1

log

*˙ (

µV)

[M �• y

r1 Mpc

3 ]

z = 3K=Kkenn

K=0.1 × Kkenn

K=0.01 × Kkenn


5

4

3

2

1

log

*˙ (

µV)

[M �• y

r1 Mpc

3 ]

z = 3K=Kkenn

K=0.1 × Kkenn

K=0.01 × Kkenn

10% K

1% K

Cum

ulat

ive

com

ovin

g SF

R d

ensi

ty



Wolfe & Chen 2006 result:• SFR efficiency of DLAs is a factor of ≥10 below KS relation

Another possibility:• Lyman Break galaxy cores may be embedded in DLAs, and may themselves exhibit in situ star formation

Caveat:• Wolfe & Chen 2006 search excluded objects with high surface-brightness cores (µV < 26.6 mag/arcsec2) (i.e. LBGS)


LBGs embedded in DLA Neutral Gas Reservoirs

DLA

LBG

Dust


In situ star formation in DLAs associated with LBGs

DLA

LBG

Dust

In-situ star formation


Solution: Ultra Deep u’-band image of UDF with Keck

Keck Telescopes

1 σ depth = 30.7 mag/arcsec2

Detection limit =27.6 mag/arcsec2

FWHM = 1.3 arcsec

10

Rafelski et al. 2009

Use the u-band image to select 407 z~3 LBGS via their flux

decrement from the Lyman break


48 compact, symmetric, and isolated z~3 LBGs in V-band



Stack 48 isolated, compact, symmetric z~3 LBGs in the V-band (rest-frame FUV)

12Rafelski et al. 2011


0.0 0.2 0.4 0.6 0.8 1.0 1.2Radius [arcsec]

34

32

30

28

26

24

22

Surfa

ce B

right

ness

(µV)

[mag

arc

sec

2 ]

4

3

2

1

0

Log

SFR [M

�• y

r1 kpc

2 ]

0 2 4 6 8Radius [kpc]

Median LBGsPSF

1 image sky limit48 image stack sky limit

13

Radial surface brightness profile of stacked image

Inner core OutskirtsSu

rfac

e Br

ight

ness

[m

ag a

rcse

c-2]

Radius [arcsec]Rafelski et al. 2011


Previously:

Column density of gas varies with radius, we need a differential version of the comoving SFR density(ρ̇∗)

Goal: compare comoving SFR density in outskirts of LBGs to DLAs to obtain a SFR efficiency

Differential:

∆ρ̇∗∆N

= �ΣSFR(N)�H0

cf(N,X) ⇒ ∆ρ̇∗

∆I


15 16 17 18 19log *˙ /< I 0

obs > [M �• yr 1 Mpc 3 / ergs cm 2 sec 1 Hz 1]

34

32

30

28

26

24Su

rface

Brig

htne

ss (µ

V) [m

ag a

rcse

c2 ]

K=0.01 × Kkenn

K=0.02 × Kkenn

K=0.05 × Kkenn

K=0.1 × Kkenn

K=Kkenn

Model differential comoving SFR density for DLAs

Differential comoving SFR density per intensity interval

Surf

ace

Brig

htne

ss [

mag

arc

sec-2

]



Determine efficiency for each point

Comparison of model to data to determine efficiency

Differential comoving SFR density per intensity interval15 16 17 18 19

log *˙ / I [M �• yr 1 Mpc 3 / ergs cm 2 sec 1 Hz 1]

34

32

30

28

26

24Su

rface

Brig

htne

ss (µ

V) [m

ag a

rcse

c2 ]

Overlapping Average LBGAverage LBGLBG Uncertainty (1 )LBG R/H rangeK=0.01 × Kkenn

K=0.02 × Kkenn

K=0.05 × Kkenn

K=0.1 × Kkenn

K=Kkenn

Surf

ace

Brig

htne

ss [

mag

arc

sec-2

]



10 100gas [M �• pc 2]

0.0001

0.0010

0.0100

0.1000SF

R [M

�• y

r1 kpc

2 ]

KS relationKS relation, K=0.1 × KkennLBG R/H rangeLBG Uncertainty (1 )LBG outskirtsDLAs

17

The KS relation for atomic dominated gas at z~3Σ

SFR[M

⊙yr

−1kp

c−2]

Σgas [M⊙pc−2]



3.0

2.5

2.0

1.5

1.0

Log

Cov

erin

g Fr

actio

n

22.5 22.0 21.5 21.0Log N [cm 2] with K=0.1×KKenn


LBG outskirtsCorrected LBG outskirtsDLAs with K=KkennDLAs with K=0.1×KkennDLAs with K=0.02×Kkenn

atomic-dominated gas

The covering fraction of the outskirts of LBGs is consistent with the DLA covering fraction

Log

Cov

erin

g Fr

actio

n

The emission unlikely to be from molecular-dominated gas


molecular-dominated gas

3.0

2.5

2.0

1.5

1.0

Log

Cov

erin

g Fr

actio

n

21.5 21.0 20.5 20.0Log N [cm 2] with K=KBigiel


LBGs with coresCorrected LBGs with coresextrapolated upper limit




gas [M �• pc 2]

SFR [M

�• y

r1 kpc

2 ]

1 10 102 103 10 4

10 3

10 2

0.1

1

10KS relationKS relation, K=0.1 × KkennLBG R/H rangeLBG Uncertainty (1 )LBG outskirtsDLAsG&K 2010 total gasG&K 2010 atomic gasG&K 2010 molecular gas

All Metals

19

Comparisons to predictions from simulations (Gnedin & Kravtsov 2010)

LBG Metallicity

gas [M �• pc 2]

SFR [M

�• y

r1 kpc

2 ]

1 10 102 103 10 4

10 3

10 2

0.1

1

10KS relationKS relation, K=0.1 × KkennLBG R/H rangeLBG Uncertainty (1 )LBG outskirtsDLAsG&K 2010 total gas Z<0.1Z �•

G&K 2010 atomic gas Z<0.1Z �•

G&K 2010 molecular gas Z<0.1Z �•

Metals with Z<0.1Z �•

DLA Metallicity

Σgas [M⊙pc−2]Σgas [M⊙pc

−2]

ΣSFR[M

⊙yr−

1kpc−

2]



What is responsible for the reduced SFR efficiency?

Metallicity of gas?

Background radiation field?

Role of molecular vs. atomic hydrogen gas?

Other possibilities?

To better answer this question, would like to measure SFR efficiency for a range of redshifts


0 1 2 3 4 5Redshift (z)

3

2

1

0

[M/H

] 0Gyr 7.7Gyr 10.3Gyr 11.4Gyr 12Gyr 12.3Gyr

LiteratureESI

HIRES<Z>

<Z> best fit

21Rafelski et al. 2012 in press

Metal Abundances versus redshift

<Z>= -(0.22±0.03) z - (0.65±0.09)Typical uncertainty

/ 30Heidelberg 2012: Marc Rafelski 22Haardt & Madau 2012

QuasarsGalaxies

combined

Evolution of Background Radiation Field


Comparison of z~3 outskirts with z=0 outskirts

ΣSFR[M

⊙yr

−1kp

c−2]

Σgas [M⊙pc−2]



The Ultraviolet Hubble Ultra Deep Field

Measure SFR efficiency at z~1 and z~2Improve z~3 measurement with larger sample of LBGs

90 HST orbits:30 F336W30 F275W30 F225W

2000 2500 3000 3500Wavelength (Å)

0.00

0.05

0.10

0.15

0.20

0.25

Thro

ughp

ut

F225W F275W F336W

Use existing i’ band UDF data for measurement at z~4


Epoch1

Epoch3Epoch2

IR UDF 09

Epoch 1:March 2 - March 116 Orbits / 12 exposures per filter

Epoch 2:May 28 - June 410 Orbits / 20 exposures per filter

Epoch 3:August 4 - September 1914 Orbits / 28 exposures per filter+ 2 failed orbits from above

Total:30 Orbits / 60 exposures per filter90 Orbits in total by mid September

25

NUV Coverage of UDF with WFC3

29th mag 10 sigma point source limit

25

/ 3026


F225W F275W F336W B I

F336Wdropouts

F275Wdropouts

F225Wdropouts

27

UV dropout galaxies at z~1-3


Radial surface brightness profile of stacked LBGs at z~4

0.0 0.2 0.4 0.6 0.8 1.0 1.2Radius [arcsec]

34

32

30

28

26

24

22

Surfa

ce B

right

ness

(µV)

[mag

arc

sec

2 ]

3

2

1

0

Log

SFR [M

�• y

r1 kpc

2 ]

0 2 4 6 8Radius [kpc]

Median LBGsPSF

1 image sky limit73 image stack sky limit

Inner core Outskirts


10 100gas [M �• pc 2]

0.0001

0.0010

0.0100

0.1000SF

R [M

�• y

r1 kpc

2 ]

KS relationKS relation, K=0.1 × KkennLBG R/H rangeLBG Uncertainty (1 )LBG outskirtsDLAs

29

How do things change at z~4?Σ

SFR[M

⊙yr

−1kp

c−2]

Σgas [M⊙pc−2]

10 100gas [M �• pc 2]

0.0001

0.0010

0.0100

0.1000SF

R [M

�• y

r1 kpc

2 ]

KS relationKS relation, K=0.1 × KkennLBG outskirts

LBG outskirts z~4


Summary

• Measured extended rest-frame FUV emission in outskirts of z~3 LBGs

• Star formation rate efficiency of atomic-dominated gas at z~3 is a factor of ~10 lower than predicted by Kennicutt-Schmidt relation for local galaxies at z=0

• Covering fraction of DLA gas consistent with LBG outskirts, while molecular gas insufficient to cover the LBG outskirts.

• Consistent with predictions from Gnedin and Kravtsov 2010 suggesting the metallicity could be the driver for the lower SFR efficiency

• Measured the metallicity evolution of neutral hydrogen gas out to z~5

• Obtaining NUV data with HST to measure the SFR effiency at z~1 & 2

• Preliminary measurement of the SFR efficiency at z~4

measuring the evolution of the star formation rate ... · summary • measured extended rest-frame...

Documents