particle collider the Bullet Cluster

LECTURE 16
COSMIC STAR FORMATION HISTORY
MEASURING SFR - RECAP
OBSCURED VS. UNOBSCURED ACTIVITY
sSFR EVOLUTION
BURST-LIKE VS. NORMAL/SECULAR STAR FORMATION
THE COSMIC SFH - LINK TO GAS CONTENT

Centaurus A The Cosmic Microwave

Background
RECAP L15
1. The Schechter function is the most commonly used distribution
function to describe mass- or luminosity-distributions of galaxies.
(Alternatives for populations with extended bright-end tails are
the double-exponential and double power law function.)

2. A Schechter-like distribution fct. for halo masses arises naturally

from a Gaussian initial field of density fluctuations.

3. MFs of nearby galaxies are dominated by quiescent galaxies at

the high-mass end.

4. LFs generally evolve to higher luminosity with increasing redshift

due to stronger star-formation activity in the young Universe.

5. Quiescent/early-type galaxies gain in importance as a function

of cosmic time.

6. Approx. Milky Way-sized haloes are generally the most efficient at

converting their gas to stars.
HOW DID WE GET TO THIS?

6 Gyr 3.5 Gyr

50% of the present day stellar mass
was built in the last 6 Gyr (z<1).

But also: ~45% of the stellar mass was

built in the "peak epoch of galaxy
assembly" 1<z<3.

➽ As much stellar mass was formed in

the 2nd quarter of cosmic history as in
the 2nd half (i.e. 3rd and 4th quarter):

Muzzin et al. (2013)

the star-formation activity in the early
universe must have been higher than
at late times.
LEARNING OUTCOMES
1. Know how/by how much the overall activity level of the Universe
evolved and how this is measured.

2. Understand:
a. the relative contribution of dust-obscured & unobscured star
formation to the cosmic star formation history (CSFH),
b. the relative importance of starburst activity and "normal" activity,
c. which galaxies (mass, type) contribute most to the cosmic star
formation rate density at different times.

3. Be aware of the link between gas accretion, gas reservoirs and the
evolution of galaxy specific star formation rates (plus specific black hole
accretion rates).

4. Understand what the CSFH must imply for the presently poorly
constrained evolution of the gas content of the Universe.
QUESTIONS YOU WILL BE ABLE
TO ANSWER AFTER THIS LECTURE
• What astrophysical quantity/ies do we measure to determine
the star formation history of the universe?

• How has the overall star-formation activity level of galaxies

evolved over time?

• What is the main driving factor for the higher star-formation

rates in distant galaxies?

• Galaxy collisions in the local universe often produce

spectacular starbursts. How would you go about determining
whether merger-induced starbursts in the younger, denser
universe are an important factor for high-redshift star formation
or not?

• Where do high-redshift galaxies get the gas to fuel their high

star-formation activity?
THE COSMIC SFH (SFH=STAR FORMATION HISTORY)
DIFFERENT ASPECTS OF THE TOPIC
steady, 15× decrease of
activity over last 10 Gyr

may require revision if

significant number of
dust-obscured, high-
redshift galaxies are
discovered

Madau & Dickinson (2014)

max. activity level at z~2,

aka “peak epoch of SFH"
THE COSMIC SFH (SFH=STAR FORMATION HISTORY)
DIFFERENT ASPECTS OF THE TOPIC
steady, 15× decrease of
activity over last 10 Gyr

may require revision if

significant number of
dust-obscured, high-
redshift galaxies are
discovered
~20 years
ago!

Madau et al. (1996)/

Lilly et al. (1996)

Madau & Dickinson (2014)

max. activity level at z~2,

aka “peak epoch of SFH"
NO SFH RECORD...
... WITHOUT EXTRAGALACTIC SURVEYS
H
H

Image credit: HerMES team

S

The wedding cake (multi-

tiered) survey approach:
- deep, small-area (pencil
beam) fields targeting faint,
but numerous sources
- shallow(er), wide-area
surveys targeting rare (bright)
objects
NO SFH RECORD...
... WITHOUT EXTRAGALACTIC SURVEYS

0.2 deg2

Madau & Dickinson (2014)

 NO SFH RECORD...
... WITHOUT EXTRAGALACTIC MULTI-WAVELENGTH SURVEYS
GOODS-North field (optical imaging)

Image credit: spacetelescope.org

 NO SFH RECORD...
... WITHOUT EXTRAGALACTIC MULTI-WAVELENGTH SURVEYS
GOODS-North field (optical imaging) GOODS-North field (IR imaging)

Image credit: spacetelescope.org Image credit: HerMES team

SFR TRACERS - RECAP
UV, IR, NEBULAR LINES & RADIO

SFRUV [M⦿/yr] = KUV×LUV [erg s-1 Hz-1] (KUV = 1.4×10-28)

SFRIR [M⦿/yr] = KIR×LIR [L⦿] (KIR = 1.7×10-10)

SFRH𝛼 [M⦿/yr] = KH𝛼×LH𝛼 [erg/s] (KH𝛼 = 7.9×10-42)

SFRradio [M⦿/yr] = Kradio×Lradio [W/Hz] (Kradio = 5.9×10-22)

Good measure of total (i.e. obscured & unobscured) SFR:

SFRtot = SFRUV + SFRIR

(All conversion factors are for a Salpeter IMF; to convert to a Chabrier IMF: K → K/1.7.)
SFR TRACERS - RECAP
UV, IR, NEBULAR LINES & RADIO

SFRUV [M⦿/yr] = KUV×LUV [erg s-1 Hz-1] (KUV = 1.7×10-28)

SFRIR [M⦿/yr] = KIR×LIR [L⦿] (KIR = 1.7×10-10)

SFRH𝛼 [M⦿/yr] = KH𝛼×LH𝛼 [erg/s] (KH𝛼 = 7.9×10-42)

SFRradio [M⦿/yr] = Kradio×Lradio [W/Hz] (Kradio = 5.9×10-22)

Example:

high-z sub-mm galaxy (SMG): LIR ~ 1013 L⦿ ⬌ SFR ~ 2000

local starburst M82: LIR ~ 2×1010 L⦿ ⬌ SFR ~ 4
SFR TRACERS - RECAP
NEW!
UV, IR, NEBULAR LINES & RADIO

SFRUV [M⦿/yr] = KUV×LUV [erg s-1 Hz-1] (KUV = 1.7×10-28)

SFRIR [M⦿/yr] = KIR×LIR [L⦿] (KIR = 1.7×10-10)

SFRH𝛼 [M⦿/yr] = KH𝛼×LH𝛼 [erg/s] (KH𝛼 = 7.9×10-42)

SFRradio [M⦿/yr] = Kradio×Lradio [W/Hz] (Kradio = 5.9×10-22)

Radio synchrotron continuum emission IR-radio correlation

for local galaxies
is a good (dust-free) tracer of star
formation activity. (Reason: massive
stars evolve rapidly and end as SNe,
thereby generating shock waves that

Yun et al. (2001)

accelerate charged particles which in
turn emit synchrotron radiation.)
Observationally this is reflected in the
existence of the "IR-radio correlation".
 ✓ ◆↵ ✓ ◆
⇤ L L dL
(2) (L)dL = exp
THE COSMIC SFH L⇤ L⇤ L⇤

LUMINOSITY DENSITY EVOLUTION

Z 1
Practical
(3) steps required to estimate
n [Mpc ]= 3the
(L) dLSFH
= ⇤of(↵
the Universe:
+ 1)
0
#1: fit a Schechter (or other parametric form) to measured lum. fct.
#2: integrate luminosity density based on Schechter function fit
Z 1
⇤
(4) L [L Mpc 3
]= L (L) dL = L⇤ (↵ + 2)
0

#3: convert (comoving) luminosity density to SFR density

✓ ◆1SFR ✓ i◆ - dust-obscured SFR
UV luminosity density - unobscured
L
⇤ 1 hIR luminosity density
(5) (L) dlog (L) = exp log (1 + L/L⇤ ) dlog (L)
L⇤ 2 2
Bouwens & Illingworth (2006)

8
⇤ L
< L⇤
f
dlog (L) if L  L⇤
(6) (L) dlog (L) =
: ⇤ L
L⇤
b
dlog (L) if L > L⇤
Gruppioni et al. (2013)

Z 1
3
⇢? [M Mpc ]= M (M) dM
UV LUM. FUNCTION EVOLUTION
REDSHIFT 0 < z < 10

z<2: positive luminosity evolution (growing L*), little density evolution

2<z<5: positive luminosity evolution (growing L*) & negative density evolution
(decreasing Φ*)
z>5: negative luminosity & density evolution (decreasing L* & Φ*)
drop-out
samples!
Cucciati et al. (2012) Bouwens et al. (2014)
IR LUM. FUNCTION EVOLUTION
REDSHIFT 0 < z < 4

z<1: little density evolution (approx. const. Φ*), positive luminosity evolution
(L* growing faster than for UV LF)
1.5<z<4: moderate, positive luminosity evolution (growing L*) & negative
density evolution (decreasing Φ*)

Gruppioni et al. (2013)

THE COSMIC SFH
CONTRIBUTION OF OBSCURED & UNOBSCURED ACTIVITY

Quick facts:
The dust-obscured SFRD has
been measured over "only" ~11
Gyr (z<3-4).
Whenever measurements of
both exist (z<3), dust-obscured
star formation dominates over
~8×
unobscured activity. During the
peak epoch of SFH (z~2), dust-

Madau & Dickinson (2014)

obscured SF was more
important than at low redshift.
Any discovery of even a few
~4×
strong, dusty starburst galaxies
at high redshift would imply
significant revision of the high-z
SFRD estimates based on drop-
out samples.
THE COSMIC SFH
CONTRIBUTION OF DIFFERENT MASS SCALES

Galaxies with the mass of the Milky

Way (~5×1010M⦿) contributed most
to the SFRD of the Universe over the

Karim et al. (2011)

last 10 billion years.
(Little surprisingly, they are associated with
~1012M⦿ dark matter halos, i.e. the mass scale
at which haloes are most efficient at
converting their baryons to stars.)

Massive galaxies like the Milky Way universal baryon ratio !bary/!m ~ 1/6
have typical LIR/LUV (i.e. SFRIR/SFRUV)
values of ~10. This explains why the
the dust-obscured contribution to
the SFRD always dominates the UV

Behroozi et al. (2010)

luminosity density at z<3.
 THE COSMIC SFH
UNDERSTANDING THE PHYSICAL CAUSES - THE RISE OF (U)LIRGs

Evolution of star
formation rate
density (SFRD):

Madau & Dickinson (2014)

(Note agreement between
extinction-corrected UV
measurements and IR-
based data.)

SFRD, split by IR
luminosity "classes":
LIRG = Luminous IR Galaxy
(SFR=10-100 M⦿/yr)

Magnelli et al. (2011)

ULIRG = Ultra-luminous IR
Galaxy (SFR=100-1000 M⦿/
yr)
 THE COSMIC SFH
UNDERSTANDING THE PHYSICAL CAUSES - THE RISE OF (U)LIRGs

Evolution of star
formation rate
density (SFRD):

Madau & Dickinson (2014)

In the nearby Universe (U)LIRGs
(Note agreement between
extinction-corrected UV are often found to be merging/
measurements and IR- interacting galaxies:
based data.)

SFRD, split by IR
luminosity "classes":
LIRG = Luminous IR Galaxy
(SFR=10-100 M⦿/yr)

Magnelli et al. (2011)

ULIRG = Ultra-luminous IR
Galaxy (SFR=100-1000 M⦿/
yr)

Image credit: STSci

THE COSMIC SFH
RISE OF (U)LIRGs = RISE OF (MERGER-INDUCED) STARBURSTS?

Natural to expect in an expanding

Universe where the merger rate used
to be higher, but...

Image credit: Nature

THE COSMIC SFH
RISE OF (U)LIRGs = RISE OF (MERGER-INDUCED) STARBURSTS?
Low-redshift star-forming galaxies define a "galaxy main sequence" (the SFR-M★
relation). Some rare starburst galaxies with exceptionally high SFR lie offset to
higher sSFR (specific star formation rate sSFR=SFR/M★), above the main sequence.

What would the distribution of galaxies in SFR-M★ space look like at high redshift if
merger-driven starburst activity were the key driver for the enhanced level of
activity in the younger Universe?

z=0 data ☛

☚ cartoon
s
xie
ala
gg

Brinchman et al. (2004)

in
fo rm
ar-
of st
n ce
ue
eq
in s
ma
THE COSMIC SFH
RISE OF (U)LIRGs = RISE OF (MERGER-INDUCED) STARBURSTS?
Low-redshift star-forming galaxies define a "galaxy main sequence" (the SFR-M★
relation). Some rare starburst galaxies with exceptionally high SFR lie offset to
higher sSFR (specific star formation rate sSFR=SFR/M★), above the main sequence.

What would the distribution of galaxies in SFR-M★ space look like at high redshift if
merger-driven starburst activity were the key driver for the enhanced level of
activity in the younger Universe?
THE GALAXY MAIN SEQUENCE
L14: REDSHIFT EVOLUTION (z<6)

z<2.5 (up to 11 Gyr look-back time): 4<z<6 (12.1 to 12.8 Gyr look-back time):

Wuyts et al. (2011)

Steinhardt et al. (2014)

➨
THE COSMIC SFH
THE RISE OF HIGH-ACTIVITY/-LUMINOSITY DISK GALAXIES!

The scattered distribution in the SFR-

M★ plane one would expect if short-
lived, stochastically-triggered
starbursts were all-important for high-
z star formation is not observed.

Instead, we observe the presence of

a galaxy main sequence out to at (U)LIRGs
least z~6, implying a large
homogeneity in the evolutionary
history of star-forming galaxies (see L13;
on-going SFR and integral of previous SFR - i.e. stellar
mass - are correlated over ~11 Gyr look-back time).

Luminosities on a par with those of

local, merger-induced starburst
ULIRGs are the norm for massive,
"normal" disk galaxies at z>1.
MAIN SEQ. GALAXIES: sSFR vs. z
EVOLUTION OF GALAXY SFRs vs. THE COSMIC SFR DENSITY

SFRD evolution The SFRD decreases at z>2, while the

sSFRs of individual galaxies are
constant or even continue to rise. How
does this go together...?

specific star formation rate

Madau & Dickinson (2014)
(sSFR=SFR/M★) evolution at fixed M★

(Whether or not the sSFR continues to rise beyond z~2 or

remain approx. constant is currently a subject of intense
debate. If galaxy sSFRs are proportional to specific dark
matter halo growth rates, their sSFR should continue to rise
at z>2.)
DIGRESSION
THE NIGHT SKY IN A
HIGH-Z DISK GALAXY
THE COSMIC SFH
THE MINOR CONTRIBUTION OF STARBURSTS TO THE SFRD

z~2

10#15%&

10-15%
85#90%&

85-90%
THE COSMIC SFH
THE MINOR CONTRIBUTION OF STARBURSTS TO THE SFRD

z~2 z~0

10#15%&

10-15%

5"10%&
5-10%
85#90%&

90"95%&
85-90%
90-95%

Based on: Sargent et al. (2012)

STAR FORMATION EFFICIENCY
THE CLUE TO THE CAUSE OF THE SFRD-INCREASE AT HIGH REDSHIFT
Recap:
#1: The cosmic SFRD used to be much higher in

S
M
the young Universe
#2: Starbursting (SB) outliers to the SFR-M★
relation are not responsible for the majority of
the star formation activity at high redshift; most
stars are formed in main-sequence (MS)

SB
galaxies

Sargent et al. (2014)

#3: Starburst galaxies have very high star-
formation efficiencies/short gas depletion times
(see Schmidt-Kennicutt relation in figure on left & L14) — too short;
if high-z disks had starburst-like star-formation
efficiencies their gas reservoirs would be
exhausted too rapidly for a SFR-M★ relation to
exist over billions of years.
THE FUEL FOR THE COSMIC SFH
STREAM-FEEDING FOR GAS-GUZZLING HIGH-Z DISK GALAXIES

In-flows of pristine gas must replenish

galaxy gas reservoirs. (This is a hypothesis motivated
by the arguments on the previous page; there is some, albeit
tentative, observational evidence for these gas streams which
are also seen in galaxy formation simulations.)

In simulations galaxies get about 90% of

their material through ~3 streams, the rest
trickles down as a diffuse "haze".
 .
EVOLUTION OF sSFR vs. sMBH
AGN ACTIVITY ON AVERAGE REFLECTS GALAXY GAS RESERVOIRS

The cosmic black hole accretion history has a similar shape as the evolution of the
SFRD. This is widely believed to reflect how much gas is available in general in
galaxies; both the star formation process in the many body of the galaxy and the
black hole can draw fuel from this gas reservoir.
Note:
#1: While the shape of the SFRD and BH accretion rate density (BHARD) evolution are similar, the
normalisation is not; the BHARD needs up-scaling by about a factor 1000 to match the SFRD
(approximately equal to the ratio between stellar and BH masses observed in nearby galaxies.)
#2: BH accretion rates are determined from the X-ray luminosities of AGN.

Black hole accretion rate density evolution

SFRD evolution (rescaled to match SFRD!)
Madau & Dickinson (2014)

Madau & Dickinson (2014)

star formation
history
 .
EVOLUTION OF sSFR vs. sMBH
AGN ACTIVITY ON AVERAGE REFLECTS GALAXY GAS RESERVOIRS

Black hole accretion events are of a

On average, the specific BH much more stochastic nature than the
mass accretion rate follows
the sSFR quite closely over
.
star formation process; the link
between sSFR and sMBH only becomes
the last 10 Gyr: apparent when average rates in large
samples are considered:
sSFR-evolution of main
sequence galaxies

Mullaney et al. (2012)

Mullaney et al. (2012)
RECAP L16
1. At the peak epoch of galaxy assembly (z~2) the SFRD of the
Universe was 15x higher than today. The contribution of dust-
obscured SFR was about twice as important as at z=0.

2. Careful analysis of data from a broad range of wavelengths

produces measurements of the CSFH that agree well.

3. Starburst galaxies (i.e. high-sSFR outliers to the SFR-M★ relation)

contribute only a small (~10%) fraction to the SFRD at z<2.

4. Most of the stellar mass is built in disk galaxies on the main

sequence with ~1 Gyr gas depletion time scales. High-z disks are
much more luminous than the Milky Way (SFR~100 vs. <10 M⦿/yr).
These high SFRs are (presumably) sustained by large accretion
flows.

5. The cosmic black hole accretion history resembles the cosmic

SFH. Both processes draw on galactic gas reservoirs, hence the
link.
NEXT MONDAY (L17)

More on the gas in galaxies:

What gas phases are there?

Where are they?
How do we find/distinguish between them?
What does gas do to the appearance of galaxies?
READING FOR THIS LECTURE
• 'Galaxy Formation and Evolution', Mo, van den Bosch & White:
chapter 2.4.1 & 15.1/15.2 (Galaxy mass/luminosity functions)
chapter 2.6 (Galaxies at high redshift)
chapter 2.6.8 (&15.4.2) (The cosmic star formation history)

• 'Extragalactic Astronomy and Cosmology', P. Schneider:

chapter 9.5 (The cosmic star formation history*)
chapter 9.5.1 (Summary of SRF-indicators)

*Note: in many books you will still find the view expressed that the star formation activity at high-z was 'burstier'; in this literature
burstiness is purely defined through the absolute level of star formation. As we saw in this lecture this is not the most meaningful
definition of a starburst, given that the entire population of main-sequence galaxies has evolved in synch over the last ~12 Gyr.