Quasars, Black Holes

& Host Galaxy

Evolution

Fred Hamann

University of Florida

(Quasar Metal Abundances)

Why quasars? Why high redshifts? Why metal abundances?

MBH sph SMBH growth linked to galaxy (spheroid) formation

Massive spheroids today have (mostly) old stellar pops.

Quasars mark the locations when and where the spheroids formed

Their metallicities trace the amount of star formation:

• When did the star formation occur during SMBHgalaxy evolution?

• How much star formation occurred before the visible quasar epoch?

Outline:

Metallicity diagnostics & results

Implications for SMBHgalaxy evolution

Significance of Fe/

Trends with z, L, L/Ledd, Mass

Future Prospects

Broad Line Region (BLR) metallicities:

Quasars have (broad) metal emission lines.

Prior star formation!

Even for quasars at z > 6!

Composite of z > 4 quasars

(Hamann & Ferland 1999, Constantin et al. 2002)

Shields 1976 Baldwin & Netzer 1978 Davidson & Netzer 1979 Uomoto 1984

~ Solar metallicities +/- 1 dex

Problem: As C/H increases, Tgas

decreases, and CIV/Ly constant.

Shields (1976):

Assume Nitrogen has secondary enrichment: N/O O/H

(as in galactic HII regions and stellar data)

Use N III] 1750, N IV] 1486, etc., to avoid saturation issues

(but weak and hard to measure)

Hamann & Ferland (1992, 1993, 1999), Ferland et al. (1996):

Include stronger UV lines: NV/CIV and NV/HeII

fainter quasars & larger samples

Saturation/thermalization issued unavoidable

Extensive photoionization simulations, with N/O O/H

Hamann et al. (2002):

Locally Optimally-emitting Cloud (LOC) model of BLR

(Baldwin et al. 1995)

the BLR is stratified, a wide range in nH, H coexist

not dependent on a particular choice

Calculate line strengths & ratios for each nH, H, Z

Ioni

zing

flu

x

H density

Add line emission over each LOC grid line ratios versus Zgas :

Hamann et al. (2002)

The metallicity dependence of these ratios is due mainly to N/O O/H

Nagao et al. (2006):

Include more lines, with sensitivities to nH, H, temperature (Zgas)

less dependent on N lines and N/O O/H

Find “best” solution for each quasar by tuning the weighted sums over LOC distrib. to match each quasar spectrum.

Nagao et al. (2006)

> 5000 SDSS quasars

Dietrich et al. (2003)

Obtained spectra the old fashioned way.

Zgas ~ 4-5 Zo

This quasar at z ~ 4.2 has many well-

measured diagnostics

We estimated:

Zgas ~ 2 Zo

Warner et al. 2002High redshift examples:

Pentericci et al. 2002

Zgas Zo

at redshift 6.28

based on NV/CIV,

lower limit on NV/HeII

How much metal-rich gas? From what stellar population?

LOC models suggest quasar MBLR ~ 1000 Mo (Baldwin et al. 2003)

But the amount of accretion over a quasar lifetime is MBH

If the BLR is continuously replenished by accretion, then the reservoir of metal-rich gas has mass: Mgas MBH ~ 109 Mo

Stellar mass needed to enrich this gas: Mstars few 109 Mo

at least ~bulge-size stellar pops.

In our models, the NV ratios often suggest 1.5 to 2x higher Zgas than NIII].

Measurement error?

(In “well-measured” - high EW - cases all the N

lines agree.)

(Dietrich et al. 2003)

Need independent checks

Narrow Line Region (NLR) metallicities:

Groves et al. (2006):

~23,000 low-redshift Seyfert 2s from SDSS

Visible emission-line ratios, e.g., [NII] 6584

Adopt: nH ~ 1000 cm3,secondary N enrichment

All but 40 have Zgas Zo

Typical values: Zgas ~ 2 - 4 Zo

Also: Storchi Bergmann & Pastoriza 1989 Storchi Bergmann et al. 1998, Nagao et al. 2002, Groves et al. 2004

Much larger scales:102 to 104 pc

Narrow Line Region (NLR) metallicities:

Nagao et al. (2006):

High-z quasar 2s and radio galaxies

UV emission-line ratios (same lines at BLR)

Adopt: nH ~ 102 or 105 cm3,secondary N enrichment

Zgas = 0.2 to 5 Zo depending on nH

Associated Absorption Line (AAL) metallicities:

Also posters: Nestor, Simon, Misawa, Ganguly

Observed Wavelength

AALs

AALs appear in ~25% of quasars

Probably form at a wide range of radii: ~10 to >104 pc

A simpler analysis:

Measure ionic column densities

Apply ionization correction

No assumptions about secondary N

Foltz et al. 1986

Associated Absorption Line (AAL) metallicities:

Early results: Zgas Zo and N/C solar are typical (for bona fidenear-quasar absorbers)

Petitjean et al. 1994, Wampler et al. 1993, 1996, Savaglio et al. 1997, Hamann 1997, Tripp et al. 1995, 1997, Savage et al.

1998, …

Best/most recent: D’Odorico et al. 2004

6 AAL quasars at redshifts 2.1 to 2.6

VLT/UVES spectra, resolution ~7 km/s

5 out of 6 have Zgas = 1 to 3 Zo

In progress: Leah Simon et al. 200x, poster

n AAL quasars at redshifts 2 - 4 at Keck, VLT, Magellan…

Other Indicators of star formation in quasar hosts: mm, sub-mm, CO, …

~30% of high-redshift, optically luminous quasars are ULIRGs based on mm and sub-mm (Cox et al. 2006, Beelen et al. 2006)

Inferred SFRs ~ 1000 Mo/yr

Dust masses 108 to 109 Mo

Enriched gas masses ~ 1010 to 1011 Mo

Stellar pop. masses ~ few 1010 to 1012 Mo

…formed prior to the quasar epoch.

For example:

SF coincident with quasar

SF that preceded the quasar

Understanding Zgas Zo near quasars: Galaxy Evolution

Massive spheroids today are old and metal rich:

Zstars ~ 1 to 3 Zo

The gas that produced this population must have had

Zgas > Zstars

toward the end of the evolution.

Central re/8 in field ellipticals

(Trager et al. 2000)

Age (Gyr) log Zstars

Quasar Z’s are consistent with normal galactic chemical evolution…

Friaca & Terlevich 1998 (and many others)

if most of this star formation occurred before the quasar epoch,

with 70% conversion of gas into stars.

Kauffmann & Haehnelt 2000, Granato et al. 2004

In physically motivated models,

e.g., to explain MBH sph

a major merger triggers a starburst

and funnels gas toward the SMBH

AGN (& SN) feedback halts the star formation…

The visible/luminous quasar appears after the starburst,

with central Zgas ~ 2-3 Zo

Di Matteo et al. 2004, Hopkins et al. 2005, Springel et al. 2006

obscuredvisible

Li et al. 2006

In this GADGET-2 simulation,

8 galaxies merge to make an enormous starburst, then a quasar at z = 6.54

Li et al. 2006

The total SFR reaches 104 Mo/yr,

creating a stellar mass of 1012 Mo

…before the quasar becomes

bright/observable at z = 6.54

(final MBH 2 109 Mo)

…leaving these metallicity

distributions in gas and stars at

the quasar epoch z = 6.54.

Near solar on large scales,

super-solar in dense pockets.

with Zgas ~ 2-3 Zo expected in

the nucleus

Li et al. 2006

Di Matteo et al. 2004

solar

Non-AGN data:

Quasars metallicities are like massive SF galaxies:

Zgas ~ 2-3 Zo

e.g., in this SDSS sample of 53,000 at z ~ 0.1

(HII region emission-line diagnostics)

Tremonti et al. 2004

Trends in the quasar data

…can further constrain evolution models:

Dietrich et al. (2003)

1) No significant trends with redshift, e.g., in these BLR studies

Nagao et al. (2006)

Trends in the quasar data:

2) More luminous quasars are more metal rich (based on BLR data).

Nagao et al. (2006)

Hamann & Ferland (1999)

Trends in the quasar data:

3) The fundamental trends are with Mass or L/Ledd

Shemmer et al. (2004) find a stronger relationship to L/Ledd than to L or MBH,

(based on 92 AGN with H SMBH masses)

higher Z at earlier evolutionary stages?

Warner et al. 2006 measured MBH (via

CIV) in 578 AGN

Create sub-samples to isolate trends

with L and MBH …

(Each sub-sample has ~150 quasars)

L 1047 ergs/s MBH 109 Mo

Composite spectra for fixed L and MBH (Warner et al. 2006) the underlying trend is mass-Z, possibly also driving the Baldwin Effect

These line ratios (metallicity) scale

with MBH

not Luminosity

(L = constant) (MBH = constant)

Mass Metallicity

is the main relation.

A physical explanation for the Baldwin Effect, driven by MBH :

Metallicity increases with increasing MBH

Korista et al. (1998), Warner et al. (2006)

SED becomes softer with increasing MBH

UV spectral index

All sources with both CIV and H in Warner et al. sample

CIV and H yield similar MBH

on average,

e.g., in composites.

with no systematic bias

(Warner et al. 2003)

Aside: MBH from CIV versus H

Groves et al. (2006): Z in the NLR increases with galaxy mass

(in their Seyfert 2 sample)

2x increase in O/H

Galaxy mass

NLR massmetallicity trend:

We might expect massmetallicity in quasars based on the well-known massmetallicity trend in galaxies:

solar

Lower mass galaxies expel their gas before it can be enriched to high

metallicities.

Tremonti et al. 2004Bender et al. 1993

Summary:

Quasar environs are metal rich, Zgas 1-5, out to the highest redshifts.

Enriched by at least bulge-size stellar pops. (1010 Mo), but maybe by the entire spheroid involved in MBH sph

High quasar metallicities require major star-forming episodes before the visible quasar epoch:

major merger ULIRG/starburst transition object? quasar

Quasars in more massive hosts are more metal rich, …with an added dependence

on L/Ledd (age)?

AALs and NLR lines at high redshifts

Compare quasar Z’s to host galaxy properties (mass, age, Zstars, etc.)

Transition objects (strong FIR, sub-mm) might be younger…

Sort out trends with Mass or L/Ledd

Fe/ and other ratios…

What’s next?

Hamann & Ferland 1999

Fe/ as a “clock”

Hamann & Ferland 1999

Understanding Zgas Zo near quasars:

1) Massive/dense environments evolve quickly and are metal rich at all redshifts

Quasars can uniquely probe galactic nucleiL

og M

etal

lici

ty

Pettini 2001

Quasar metal abundances as probes of host galaxy evolution:

How “mature” are the surrounding stellar pops (at different redshifts)?

When did the first major star formation begin, relative to SMBH growth & quasar activity?

Does metallicity (star formation) correlate with L, MBH & L/Ledd ?

NLS1s, Baldwin Effect, broad line ratios… AGN physics

Dependence on LAGN , MBH & L/Ledd :

Dietrich et al. (2002-04)

Warner, Hamann, & Dietrich (2002-04)

578 type I AGN measured at 950 < < 2050 Ǻ

including 26 NLS1s

Specifically targeted low L sources at high redshift

MBH = 1.4 106 Mo ( ) ( )FWHM(CIV) L(1450A)

1000 km/s 1044 ergs/s

2 0.7

All sources with both CIV and H measured(narrow H components removed)

There can be large differences between CIV and H FWHMs in a given source,

But in composites, CIV is ~ 2 broader,

consistent with reverberation and ~2x smaller RBLR

+

Peterson & Wandel (2000) Kaspi et al.

(2000)Vestergaard (2002,04)

-26

A

B

C

A

B

C

1000 1200 1400 1600 1800 2000

Rest Wavelength

1000 1200 1400 1600 1800 2000

Rest Wavelength

Composite Spectra

Sorted by SMBH mass Sorted by Luminosity

Baldwin Effect plus changing NV line ratios

0

Fit the lines to deblend & measure line ratios

NV and possibly NIII] line ratios increase with MBH

Log

Z/Z

o

L

og Z

/Zo

?

Metallicity, based on N/O O/H (Hamann & Ferland 1999, Hamann et al. 2001),

is above solar and increases with MBH

NV and possibly NIII] line ratios increase with MBH

Log

Z/Z

o

L

og Z

/Zo

?

AGN metallicity, from average of several Nitrogen line ratios...

...is above solar, and increases with both MBH and L.

L

og Z

/Zo

How “mature” are the surrounding stellar pops (at different redshifts)?

When did the first major star formation begin, relative to SMBH growth & quasar activity?

High metallicities (even at the highest redshifts, Dietrich et al. 2003) substantial conversion of gas stars

(>70% in simple closed box with “normal” galactic IMF)

Major star formation before bright/visible AGN phase, accompanying SMBH growth (Dietrich & Hamann poster, and 2004).

Stellar pop. masses > 104 to 105 Mo (>3x MBLR) (Baldwin et al. 2003)

probably > MBH (>109 Mo)

Does metallicity (star formation) correlate with L, MBH ?

Yes. How can we understand this?

More massive galaxies produce:

more massive SMBHs

more luminous AGN

higher metallicities (in their cores)

The fundamental relationship “should” be mass-metallicity.

How can we test this?

L

og Z

/Zo

MBH and L correlate with each other (in this analysis), so:

Create new composites to examine:

a range in MBH at constant L, a range in L at constant MBH...

Also spans range in L Also spans range in MBH

Friaca & Terlevich 1998

What about L/Ledd, NLS1s, …?

Shemmer & Netzer (2002) noted higher NV/CIV in NLS1s,

suggesting higher metallicities, for a given L.

Let’s look for trends with L/Ledd

L/Ledd

= 1.6 ( ) ( )FWHM(CIV) L

1000 km/s 1044 ergs/s

2 0.3 L

Ledd

Distribution of derived L/Ledd values

Composite spectra sorted by L/Ledd.

Note:

constant peak heights

constant line ratios

L

og Z

/Zo

Log

Z/Z

o

AGN metallicity from average of several Nitrogen line ratios...

▲ = NLS1s

...shows no trend with L/Ledd.

NLS1s may be slightly metal-rich for their L & MBH

but not compared to high L quasars.

One last test:

Examine composites spanning a range in L/Ledd

at L = constant, MBH = constant.

L 3 x 1047 ergs/sLedd MBH 3 x 108 Mo

MBH

L

constant peak heights constant NV line ratios

changing peak heights and NV line ratios

Z

Conclusions:

Luminous, high MBH quasars are metal-rich (see also AALs),

even at the highest redshifts,

substantial star formation before bright/visible AGN phase

(during SMBH growth).

Nitrogen line ratios (metallicities) correlate strongly with MBH

not with L or L/Ledd (AGN physics),

probably tied to galactic mass-metallicity relation.

Enriching stellar populations probably have masses > MBH

very rare major starbursts beginning at z > 8.

NLS1s may be slightly more metal-rich for given L, MBH

Based on CIV

Metallicity, based on N/O O/H (Hamann & Ferland 1999, Hamann et al. 2001),

is above solar and increases with MBH

NV and possibly NIII] line ratios increase with MBH

Log

Z/Z

o

L

og Z

/Zo

?

Intrinsic Quasar NALs

Hamann et al. (1997)

Time variable, partial covering, broad & smooth troughs...

Discrete blobs.

Cf (v) 1.

More accurate Ni &

abundances.

( Z Z0 )

Not discrete blobs.

Cf (v) < 1.

Complex (x,y) at each velocity.

Analyze point by point in v.

Use more lines, assume relative abundances of

similar ions, more constraints.

Derive (limits on) Ni(x) at each v.

“Broad” AALs:

Narrow AALs: