Evidence for Feedback in the IGM at High Redshift

1
Evidence for Feedback in the IGM at High Redshift

• Barlow (CIT), Becker (CIT), Boksenberg(IoA),
• Sargent (CIT), Simcoe (MIT), Rauch (OCIW)

(based on QSO absorption line data from Keck
HIRES, ESI and LRIS)
2
How does the undisturbed IGM look ?
A cosmic web of baryons formed mainly by
gravitational instability
Cen Ostriker et al
Main observational manifestation the Lyman alpha
forest
Keck HIRES
3
Interactions between Galaxies and the IGM
Galaxies
z4

• accrete gas (infall velocities 100 km/s)
• merge (approaching c.o.m. with velocities
  200km/s)
• interact tidally, lose gas by ram pressure
  stripping
• move about, stir and heat the IGM ( T 107 K)
• may have strong winds (outflows w. many 100
  km/s)
• chemically enrich the IGM
• produce ionizing radiation

z3
z1.8
Boxsize 2 Mpc comov vc200km/s Steinmetz
(sim.)
4
gas phases in a hypothetical large scale filament
100 kpc
5
Observable Effects

• Metal enrichment how much, when, how ?
• Ionization stellar/AGN ?
• Signatures of in/outflows
• Bulk motion and turbulence
• Accretion vs winds

6
By z3 the IGM is widely enriched with metals
(C,O)
See talk by Joop Schaye
Latest results (Simcoe et al 2004)

• Lognormal distribution with

• describes metallicity of about 50 of the mass
  and 5 of the volume of the universe
• probes down to overdensities 1.6, i.e., to the
  edge of large scale filaments

(Simcoe, Sargent Rauch 2004)
7
The Ultimate Closed Box model (Simcoe et al
2004)

• Universal chemical evolution
• treat galaxies as sources of metals and the IGM
  as the mass reservoir
• requires that on average
• more than 14 of a galaxys metals must be
  lost to the IGM
• to explain the observed IGM metallicity

8
The effect of the galactic radiation field

• least explored aspect of feedback highly
  important for reionization but observational
  evidence difficult to obtain.
• Idea different spectral shapes of the ionizing
  radiation produce different ratios among common
  metal ions
• strong CIV metal absorption systems (interior of
  LSS filaments, outer halos) are ionized by a
  stellar radiation field (T40,000 K)
• (Boksenberg,
  Sargent Rauch 1998,2003)
• matching observed relative C and O metallicities
  in the IGM to those of metal-poor stars (C/O
  -0.5) requires soft (stellar) radiation field
• 
  (Simcoe et al 2004)

9
Signatures of Bubbles and Winds in the ISM

• spherical, expanding shells
• compressed, shocked gas
• hot interior (106 K)
• transitory, cooling zone (OVI 105 K)
• cool dense layer (MgII few x 104 K)
• collisional photoionization

30 Dor (LMC) Wang 1999
10
Probe ISM gas with multiple lines of sight to
lensed QSOs
11
A possible galactic HI shell
MgII absorption system at z0.56
Curious two-component structure coherent over
1kpc LoS intersecting two bubble walls ?
12

• Do high z winds manage to get out of galaxies ?

• Two approaches
• look directly into galaxies and their immediate
  neighbourhood
• - learn about individual winds, connection
  of winds and stelpops.
• 2. look at random places in spaces and do a
  blind search for winds
• - learn about global statistics of winds

13

• Can we observe winds outside of galaxies ?

Lyman break galaxies have outflows with several
100 km/s, similar to present day superwinds
(Pettini et al 2000)
A lack of neutral hydrogen within 0.5 comoving
Mpc from those objects may correspond to
wind-blown cavities
(Adelberger et al 2003)
14
2. Search the IGM directly for

• shock heated (collisionally ionized) gas
• large, rapidly expanding shell structures
• metal enriched gas

use OVI ion as a tracer of galactic winds
OVI survey at z 2.5 with Keck HIRES (Simcoe et
al 2002)
15
Evidence for symmetric in/outflow (Simcoe et
al 2002)
OVI
HI Ly alpha
OVI
HI Ly alpha
¼ of strong OVI absorbers show conspicuous
double component structure in HI and other ions.
Shocked shell ? Bi-polar outflow ?
16
Temperatures of OVI, CIV and SiIV
If line widths predominantly thermal, the median
temperature of the OVI phase is whereas
Probably shocked gas or thermal conduction in a
hot bubble
Simcoe et al 2002
17
Properties of OVI systems

• High metallicity

as opposed to average metallicity in the IGM,
Sizes L 60 kpc, densities
(Simcoe et al 2002)
18

• Number density and cross section from rate of
  incidence per unit redshift

If all bright Ly break galaxies had such an OVI
halo around them (with comov. density
Adelberger Steidel 2000)
At z 2.5 Lybreak galaxies could account for all
of the observed OVI absorption in the Simcoe et
al survey if they are embedded in hot bubbles out
to radii 40 kpc
19
Summary Highly ionized (OVI) Gas (Simcoe et al
2002)

• OVI kinematically distinct from and hotter than
  other gas phases (CIV)
• 
  shocked gas ?
• peculiar double component structure relatively
  common in strongest systems.
  shells or cones ?
• sizes a few tens of kpc, overdensities around
  10 30 (as opposed to 100 for strong CIV/SiIV
  systems).
• 
  external to galaxies
• Metallicity O/H -1.5 higher than general
  IGM
• 
  outflow, as opposed to infall
• cross-section consistent with R 40 kpc hot
  bubbles around Lyman break galaxies
  

20
Kinematic effects of feedbackBulk motion and
turbulence in the IGM
21
Kinematics of the IGM
Probe bulk motion and turbulence with multiple
lines of sight
Lensed QSO
grav. lens
observer
IGM
Velocity and column density differences as a
function of spatial scale, density
22
Spatial coherence and kinematics in the IGM
sep 0.22 kpc
sep 260 kpc
Becker et al 2004
23

• Expect
• Large scale motion represent Hubble expansion
• Small scale motion are hydrodynamic disturbances
  (e.g., winds)

24
Large Scale Velocity Shear in the IGM
Differences between the velocities of the same
absorber in two lines of sight separated by S

• On kpc scales, velocity shear consistent with
  zero.
• On large scales (250 kpc) , a significant
  velocity shear ( 30km/s RMS) is visible.
• Its distribution can be reproduced assuming the
  clouds are randomly orientated, freely expanding
  slabs.

25

• Is the large scale motion consistent with the
  Hubble flow ?

26
Adopting a coherence length 500 physical kpc
(e.g., DOdorico etal 1998),
expansion velocity is about 70 of Hubble
flow.
Not clear whether one should expect to find
clouds to follow Hubble flow exactly (column
density limited sample, crude modelling,
observational errors)
27
The Lyman alpha forest on kpc scalesas seen in
two Lines of sight towards RXJ09110551 (z2.80)
2.2 kpc
0 kpc
28
degree of disturbances among two lines of sight
tells us about filling factor of winds
29
upper limit on the volume filling factor of
winds
Mechanical luminosity
gas density
e.g., winds starting at z4 cannot fill more than
18 of the volume.
(Rauch et al 2002)
30
General low density IGM at z3

• Large scale motions consistent with full Hubble
  expansion
• Most of the intergalactic medium (by volume) is
  highly homogeneous on kpc scales.
• The volume filling factor for strong winds
  arising later than z4 is less than 18
  (possibly much less).
• Low density Lyman alpha forest probably well
  described by numerical simulations with finite
  resolution and without any feedback (but see
  metal absorption systems)

31

• Going to higher density regions

32
Spectra of UM673 A (red) and B (black)(z(QSO)
2.72, sep. 2.24)
r 480 pc
metals !
metals !
metals !
metals !
zz(QSO) r 0 pc
33
Traces of galactic winds in higher density, metal
enriched CIV gas ?
a few 200 velocity width km/s characteristic of the filamentary matrix in
which galaxies are embedded
Origin of velocity differences, spatial scales ?
34
Measure differences between lines of sight A and
B as a function of transverse separation between
the LoS
Fractional difference in column density
Column density weighted projected velocity

• Results
• minimum size of CIV clouds (a few 100 pc)
• increasing velocity shear (70km/s _at_ 10 kpc)

transverse separation (kpc)
35
What Does It Mean ?
or are measures of the turbulence
of the gas on a spatial scale r, and of the rate
of energy input, . E.g., for Kolmogorov case,
A crude estimate of the energy transfer rate from
our data
i.e., the turbulent energy in CIV gas is much
less than for an actively starforming region
(e.g., factors 100 -1000 less than for Orion).
36

• There is a finite amount of turbulent energy in
  the gas.
• Defines a dissipation time scale (time
  it takes to transform the mean kinetic energy
  in the gas, at a rate into heat),
• 
  
• 
  years.
• The finite size of the CIV clouds defines
  relaxation time scale
• Without further energy input, pressure and
  density differences are wiped out by pressure
  waves during a sound crossing time
• 
  
  years.
• Structure on larger scales has not been wiped out
• there is (at least
  intermittent) energy input into the gas.

37
Origin of the turbulence ?

• Gas may have been stirred by mergers/tidal
  interactions or winds, or it may just be circling
  the drain

Timescales are similar to those of recurrent star
formation events that have been postulated for
various environments

• z1 field galaxies
  
  (Glazebrook et al 1999)
• the Galaxy
  
  (Rocha-Pinto et al 2000)
• fluctuations in SFR in nearby spirals
  
  (Tomita et al 1996 Hirashita Kamaya 2000)
• galactic nuclei
  
  (Krugel Tutukov
  1993)
• Lyman break galaxies
  
  (Papovich, Dickinson Ferguson
  2001)

38
CIV absorption from the filamentary LSS structure
SPH modelling of pre-enriched gas undergoing
gravitational collapse reproduces all know
properties of CIV systems (except clustering
box too small)
Rauch, Haehnelt Steinmetz 1997
Distribution of CIV line widths (thermal
turbulent)
39

• Velocity-density-scale diagram

structure function of the universe
40
Summary evidence for feedback in the IGM ?

• Cosmic web widely metal enriched down to mean
  density
• CIV metal absorbers ionized by local stellar
  radiation field
• General low density IGM (the universe by volume)
  kinematically undisturbed by feedback
• kinematic disturbances in the somewhat denser CIV
  gas low level (intermittent) energy input
  filamentary gas possibly stirred by galaxy
  motions, winds, circling the drain
• Double component structure, temperatures,
  expansion velocities, and the high metallicity
  seen in some MgII (low ionization, dense gas) and
  OVI (high ioniz., tenuous hot gas) point to ISM
  and IGM winds
• velocities and ionization structure around high z
  starburst gals. consistent with superwinds
• Inevitable that some of the wind phenomena
  described here are not due to winds but to
  gravitationally induced heating,
  motions,stripping
• To date origin and and timing of most of the
  metal enrichment unclear probably early (z5)
  and by dwarf galaxies

41

• When does the wide spread metal enrichment happen
  ?

• E.g.,
• Early vs. late (ongoing) enrichment
• Massive vs. dwarf galaxies
• Gravitational vs. winds
• ambient universe much denser at high z, ram
  pressure from infalling gas favors winds from
  dwarfs (e.g., Fujita et al 2004)
• mass-metallicity relation may indicate mass loss
  to IGM dominated by dwarf galaxies (Tremonti et
  al 2004)
• quiescence of Lyman alpha forest, ubiquity of
  metals appears to favour early, widespread (
  dwarf?) enrichment, ongoing locally (OVI winds,
  CIV turbulence)

42
Gravitational effects vs. winds ?
43
Case I possible old SN remnant at z 3.62

• Radius 13
• thickness (LoS) 0.015
• Mass range 0.4
• Expansion velocity v 195 km/s
• Number density 0.2
• metallicity
• age 10,000 years

44
(No Transcript)
45
Interactions between Galaxies and the IGM
Galaxies

• accrete gas (infall velocities 100 km/s)
• merge (approaching c.o.m. with velocities
  200km/s)
• interact tidally, lose gas by ram pressure
  stripping
• move about, stirring and heating the IGM ( T up
  to 106 K)
• may have strong winds (outflows w. many 100
  km/s)
• produce ionizing radiation

46
(No Transcript)
47
Structure function of the universe
48
(No Transcript)
49
Evidence for individual winds ?
50
General low density IGM at z3

• Large scale motions consistent with full Hubble
  expansion
• Most of the intergalactic medium (by volume) is
  highly homogeneous on kpc scales.
• The fraction of the Lyman alpha forest disturbed
  by more than 5 in optical depth is conservative)
• The volume filling factor for strong winds
  arising later than z10 is less than 20
  (possibly much less).
• Low density Lyman alpha forest probably well
  described by numerical simulations with finite
  resolution and without any feedback (but see
  metal absorption systems)

51
The Silence of the Lines

• Translate column density into baryon density
  fluctuations, making use of tight correlation

Obtain RMS scatter of the baryon overdensity
For a beam separation of 110 pc proper, and a
sample of unsaturated Lyalpha forest lines with
12baryon density are less than about 3
. Similarly, RMS velocity differences
!!!
52

• What about large scale motions ?
• Consistent with Hubble flow ?
• Peculiar velocities ?
• Signs of feedback (winds) ?

53

• Define disturbed fraction of the Lyalpha forest
  fraction of the spectrum where the
  optical depths differ by more than a certain
  amount

Measure
Where is the width of
the spectroscopic footprint of a wind bubble
intersecting a line of sight (two thermally
broadened absorption lines from a crossing
shell). and are the radius and
expansion velocity of the shell, and is the
space density of the sources of the wind
events. Adopt a model for and (e.g.,
superbubble model of Mac Low McCray 1988 and
solve for .
54
upper limit on the number density of galaxies
producing winds
upper limit on the volume filling factor of
winds
55

• There is a finite amount of turbulent energy in
  the gas.
• Defines a dissipation time scale (time
  it takes to transform the mean kinetic energy
  in the gas, at a rate into heat),
• 
  
• 
  years.
• The finite size of the CIV clouds defines another
  time scale
• Without further energy input, pressure and
  density differences are wiped out by pressure
  waves during a sound crossing time
• 
  
  years.
• Structure on larger scales has not been wiped out
• there is (at least
  intermittent) energy input into the gas.

56

• Can we observe winds outside of galaxies ?

Lyman break galaxies have outflows with several
100 km/s, similar to present day superwinds
(Pettini et al 2000)
A lack of neutral hydrogen within 0.5 comoving
Mpc from those objects may correspond to
wind-blown cavities
(Adelberger et al 2003)
57
Low vs. high mass gals. as the origin of metals
in the IGM

• Evidence in favor of massive galaxies
• superwinds at low z blow out of galaxies (e.g.,
  Heckman 2001), and strong winds seen in Lyman
  break galaxies (Pettini et al 2000)
• Lack of neutral hydrogen around high z starburst
  galaxies (Adelberger et al 2003)
• superwinds grafted onto massive galaxies in
  large scale cosmological hydro-simulations manage
  to get the metals out
• Evidence in favor of dwarf galaxies
• high resolution simulations of winds infall and
  high density of the IGM at high z favor dwarf
  galaxies outflows (e.g., Fujita et al 2004)
• z0.1 mass-metallicity relation (Tremonti et al
  2004) mass loss dominated by low mass galaxies.
• Lack of neutral hydrogen around starburst
  galaxies (Adelberger et al 2003) may have
  explanations other than winds (hot, highly
  ionized gas from accretion photoionization by
  cluster radiation field).

58

• winds linked to individual galaxies

Spectra of massive high z starburst galaxies have
outflow features similar to present day
superwinds

• But
• ambient universe much denser at high z, ram
  pressure from infalling gas favors winds from
  dwarfs (Fujita et al 2004)
• mass-metallicity relation indicates mass loss to
  IGM dominated by dwarf galaxies (Tremonti et al
  2004)

(Pettini et al 2000)
A lack of neutral hydrogen within 0.5 comoving
Mpc from those objects may correspond to
wind-blown cavities
(Adelberger et al 2003)
But May have alternative explanations
merger-heated halo, cluster radiation field
59

• Anecdotal evidence for the existence of ISM
  shells ok, but objects are much smaller and
  weaker than required for wind bubbles that leave
  the galaxy.

60
CIV Gas

• Too quiescent to be directly related to
  starformation
• residual turbulence and finite cloud sizes
  suggest ongoing (at z3) low level, energy input
  on timescales 10-100 Mio years.
• Time scales are similar to those involved in
  recurrent starformation events. Winds or galaxy
  encounters (accretion, mergers, stripping) may
  play a role.

structure function of the universe