Transcript UMich w/s - Royal Observatory, Edinburgh

How cold is Cold Dark Matter
dSph galaxies as a probe
Gerry Gilmore
IoA Cambridge
Dynamics, abundances with Mark Wilkinson, Rosie Wyse,
Jan Kleyna, Andreas Koch, Wyn Evans, Eva Grebel
Discovery work with Vasily Belokurov, Dan Zucker,
Sergey Koposov, et al
ApJ 663 948 2007 (july10), arXiv 0706.2687
Also Wilman, Walker & Mateo, Kamaya, Grillmair, Simons & Geha
The smallest galaxies are the places one must see the
nature of dark matter, & galaxy formation astrophysics
mssm has 120+ free parameters… lots to learn
Dwarf galaxy mass function
depends on DM type
Inner DM mass density
depends on the type(s) of DM
Figs: Ostriker & Steinhardt 2003
Satellite population problem:
Fewer expected in LG? reaching the cooling limit?
NB: predictions running out just where the data are today.
Ishiyama etal 0708.1987. dashed line from Moore etal
Real satellite luminosity function
Koposov et al 07 arXiv:0706.2687
Open symbols:
Volume-corrected
satellite LF from DR5
Filled symbols:
‘all Local Group dSph’
Grey curve: powerlaw ‘fit’ to data
Slope 1.1
Coloured curves:
Semi-analytic theory
(Benson et al 02, red
Somerville 02, blue)
--severe surfacebrightness discrepancies
Sgr & Field of Streams (and dots)
outer halo is lumpy: but is a tiny mass fraction
Streams  accretion history
Several new satellites  CDM test
Two wraps?
Halo DM shape
Disk accretion?
Warp?
Belokurov et al (2006b)
SDSS data, 19< r< 22, g-r < 0.4 colour-coded by mag
(distance), blue (~10kpc), green, red (~30kpc)
Sgr discovered 1994 Ibata, Gilmore, Irwin Nat 370
Dotted line is virial theorem
for stars, no DM
Walcher et al 2005
There is a discontinuity
in (stellar) phase-space
density between small
galaxies and star clusters.
Main difference is in
surface brightness
dSph
Why?
Dark Matter?
Size differences
Phase space density (~ ρ/σ3) ~ 1/(σ2 rh)
New systems extend overlap between galaxies and
star clusters in luminosity
Belokurov et al. 2006
Analyses of kinematic follow-up underway
 ~103 L
New photometric and kinematic studies of UCDs, nuclear
clusters, etc  ALL the small things are purely stellar
Seth etal astro-ph 0609302
systems, M/L~1-4
Virgo & Fornax UCDs
have stellar M/L –
Hilker etal,
A&A 463 119 2007
N5128 GC study by
Rejkuba et al 2007
MWG GCs extend
down to M~-2
MWG nuclear
cluster has size
~5pc, mass
10^6Msun
Schodel etal
A+A 469 125
faint
fluffies
Slightly different perspective…
(updated data)
M31; MWG; Other
Nuclear clusters,
UCDs, M/L ~ 3
Pure stars
boundary
Dark Matter
haloes
Tidal
tails
star clusters
Gilmore, Wilkinson, Wyse et al 2007 ApJ 663 948
dSph galaxies
Conclusion one:




Galaxy scaling relations work well, and indicate a
systematic star-cluster vs small galaxy distinction in
phase-space density
There is a well-established half-light size bi-modality
 all systems with size ~
< 40pc are purely stellar
−16< Mv < 0 (!!) M/L ~ 3; e.g. UCDs, Hilker et al
07; Rejkuba et al 07
 all systems with size greater than ~120pc have a
dark-matter halo
There are no known (virial equilibrium) galaxies
with half-light radius r < 120pc
So now look at the dSph galaxies’ masses
Note different scales: information at small and large r poor.
Mateo, walker etal
Derived mass density profiles:
Jeans’ equation with
assumed isotropic
velocity dispersion:
All consistent with
cores (similar results
from other analyses)
Not conclusive yet.
CDM predicts slope of
-1.3 at 1% of virial radius
and asymptotes to -1
(Diemand et al. 04)
Need different technique at large radii, e.g. full velocity
distribution function modelling.. And understand tides.
Conclusion two:











High-quality kinematic data exist
Jeans’ analyses  prefers cored mass profiles
Mass-anisotropy degeneracy allows cusps
Substructure, dynamical friction  prefers cores
Equilibrium assumption is valid inside optical radius
More sophisticated DF analyses underway
Cores always preferred, but not always required
Central densities always similar and low
Consistent results from available DF analyses
Extending analysis to lower luminosity systems difficult
due to small number of stars
Integrate mass profile to enclosed mass:
2007: extension of dynamic range [UMa, Boo, AndIX],
new kinematic studies:
Mateo plot improves.
Mass enclosed within stellar extent ~ 4 x 107M
Now a factor of 300+ in
luminosity, 1000+ in M/L
Scl – Walker etal
If NFW assumed, virial masses
are 100x larger, Draco is the
most massive Satellite (8.109M)
Globular star clusters, no DM
(old data)
Summary:

A minimum physical scale for galaxies, ~100pc, max size
for star clusters ~30pc

Galaxy mass size scale somewhat larger (?)
Galaxy nuclei are just massive star clusters?
Cored mass profiles, with similar mean mass densities
~0.1M/pc3, ~5GeV/cc





Phase space densities fairly constant, maximum for galaxies (cf
Walcher et al 2005)

An apparent characteristic
(minimum) mass dark halo
7
in all dSph, mass ~4 x 10 M ???

This is just a consistency check, not new info
dSph debris not yet found: cannot be (much of) the MWG halo, thick
disk, or thin disk
How did everything get pre-enriched?
context: substructure `issue’, old disks, one thick
disk, too few dead bodies, old red gals…
Consistency?





A minimum half-light size for galaxies, ~100pc
 mass scale similar, or a little larger
Probably cored mass profiles, with similar mean mass
densities ~0.1M/pc3, ~5GeV/cc
An apparent characteristic (minimum) mass dark halo at
dSph,
7
mass ~4 x 10 M
characteristic mass profile
convolved with
characteristic normalisation
must imply
characteristic mass
 internal consistency only
Minimum size natural with cored potential?
Implications for Dark Matter:
 Characteristic Density ~10GeV/c²/cm³
If DM is very massive particles, they must be
extremely dilute (Higgs ~100GeV)
Characteristic Scale above 100pc, several 107M
 Cooling? power-spectrum scale break?
 This would (perhaps!) naturally solve the
substructure and cusp problems
 Number counts low relative to CDM
 lots of similar challenges on galaxy scales
Need to consider seriously non-C DM candidates
Properties of Dark-Matter
dominated dSph galaxies:
Does the Mateo plot extend to the lowest luminosities?
Data still limited, lowest surface brightness gals may
have lowest sigma.
Simon & Geha: these are central values
Central velocity dispersion `masses’ are really dispersions, and are
only just resolved by the RV errors eg Simon/Geha here, our outer
Draco `cold’, etc. Independent confirmation is desirable
No parameters are *VERY* accurate. CVnI (top) has σ=13.9 (Ibata 2006) or 8.1 (Simon
+Geha, here), from the same instrument. LeoIV has σ=3.3+-1.7 derived here – 2bins?
Implications from Astrophysics:
Can one plausibly build a dSph as
observed without disturbing the DM?
 Star formation histories and IMF are easily
determined  survival history, energy input…
Chemical element distributions define gas flows,
accretion/wind rates,
 debris from destruction makes part of the field stellar
halo: well-studied, must also be understood
 Feedback processes are not free parameters
What are we really measuring with simple,
non-Distribution Function, analyses?




Dispersion profile close to flat, so sigma ~ cst, and
range of sigma is small (data <2)
derivative term is (log) luminosity profile : light, NOT
mass, and this is similar in scale for all the dSph
(factor of few)
So the derived mass really is a measure of the radial
extent of the data, and only a weak function of
anything else.
Increase in M in Mateo plot is a measure of increase
in data range
Hernandez, Gilmore & Valls-Gabaud 2000
Carina dSph
Leo I
UMi dSph
Atypical
SFH
Intermediate-age population dominates in typical
dSph satellite galaxies – with very low average
SFR over long periods (~5M/105yr), until recently
Comparing globular cluster structures, abundances, orbits, ages and likely survival
Implies ~5 [<Sgr-like] mergers in total, forming ~20% of the outer halo
(Mackey & Gilmore MNRAS 355 504 2004)
This is consistent with SDSS-observed halo lumpiness, and older (eg Unavane, GG, RW
1996 MN) age-metallicity limits
Globular cluster view of halo accretion
Constant mass scale of dSph?
Based on central velocity dispersions only low M/L
line corresponds to dark halo mass of 107M
Mateo 98
ARAA
dSph filled
symbols
2007: extension of dynamic range [UMa, Boo, AndIX],
new kinematic studies:
Mateo plot improves.
Mass enclosed within stellar extent ~ 4 x 107M
Globular star clusters, no DM
(old data)
It isn’t only gas-poor galaxies: all small galaxies are similar
Mass – to – light ratios for local dSph The star is LeoA, a gas-rich dwarf with
recent star formation, the arrow shows how it will fade with age. The square is Phoenix.
This is from Brown, Geller etal arXiv:0705.1093
Dynamics: three regimes



Body of galaxy, out to break/r_lim recent vast
increase in good data (Camb group,
Ibata/Chapman/Martin at keck, Simon/geha at
keck, Walker/mateo at Magellan/MMT, good
agreement, real progress, now pushing limits of
known systems
Outer limits: tidal tails, etc: data very limited,
agreement only fair, rather open analyses, fair
outcome: no strong effects in distant objects, Sgr a
model for the nearby.
Cores: just starting now.
Breaking the degeneracy – first steps
Survival of cold subsystem in UMi dSph
implies shallow mass density profile (Kleyna et al 03)
Dynamical friction limits on Fornax dSph
Globular Clusters also favour cores to extend
timescales Goerdt etal 2006
Main Focus: Dwarf Spheroidals
 Low luminosity, low surface-brightness satellite
galaxies, ‘classical’ L ~ 106L, V ~ 24 mag/sq
Extremely gas-poor
Apparently dark-matter dominated
 ~ 10km/s, 10 <
~ M/L <~ 100
 Metal-poor, mean stellar metallicity <
~ –1.5 dex
All contain old stars; extended star-formation
histories typical, intermediate-age stars dominate
Most common galaxy nearby
Crucial tests for CDM and other models
Walker etal arxiv:0708.0010
Omega Cen: Reijns etal A&A 445 503
Mass does not follow light
Leo II: Koch etal
Other lumps exists too, and are not understood at all.
astro-ph/0701790
CDM predicts many more satellite galaxies
than observed, at all masses (Moore et al 1999)
new large datasets of stellar line of sight
kinematics, now covering spatial extent, &
photometry for dSph satellite galaxies
 new discoveries; SDSS mostly – original key
project (also Willman et al 05; Grillmair 06; Grillmair &

Dionatos 06; Sakamoto & Hasegawa 06; Jerjen 07..)
 confirm and extend scaling relations
 Dark matter properties
G. Gilmore, M. Wilkinson, R.F.G.Wyse,
J. Kleyna, A. Koch, N. Evans & E. Grebel
2007, ApJ v663 p948; astro-ph/0703308
M31 and MWG GC size-lum, from Federici etal, 0706.2337,
Stars From Mackey etal (M31), triangles: nuclei of Virgo dEs
asterisk Virgo UCDs,
Core properties: adding anisotropy
Koch et al 07
AJ 134 566 ‘07
Fixed β
Radially varying β
Leo II
Core and/or mild tangential anisotropy slightly favoured


Mass – anisotropy degeneracy prevents robust
cusp/core distinction, but core provides better fit
(see also Wu 2007 astro-ph/0702233)
Break degeneracy by complementary information:
 Ursa Minor has a cold subsystem, requiring
shallow gradients for survival (Kleyna et al 2003
ApJL 588 L21)
 Fornax globular clusters should have spiralled in
through dynamical friction unless core (e.g.
Goerdt et al 2006)
Simplicity argues that cores favoured for all?
 New data and df-models underway to test
(GG etal, VLT high-resolution core/cusp
project)

NFW fits require very high mass, and a very wide range of mass
Draco = 8.10^9Msun and M/L=100,000 MWG vs M31 offset
no simple mass-luminosity link
astro-ph/0701780
Strigari etal (in prep) central mass fits – no simple rank
Thick and thin disk
element ratio data:
The thick/thin distinction
is evident.
The thick disk occupies
an empty part of the
halo-dSph-Sgr plot,
suggesting its parent
was different again…
This fig from A+A 465 271
Ramirez etal 2007
Sgr and the thick disk are
2 good `accretions’,
But both seem unique…
Thick disk
Galaxy halo (green), dSph (blue), LMC (cyan), Sgr (red) and dIrr (yellow) element ratios
The systematic difference is apparent (from Geisler, Wallerstein, etal 0708.0570)
NB Sgr is *very* distinctive: it must be the first such event.
CDM predicts many more satellite galaxies
than observed, at all masses
`Solutions’: warm DM; self-interacting
DM; star formation suppression by reionization; self-regulated star formation;
very high M/L plus some other variant;
predictions `wrong’, count different things;
predictions host-dependent; ….
 Conclude:
 very many proposed solutions suggests
there is still much to learn, both in models
and data

Satellite population depends on environment? Fewer expected in LG?
NB: predictions running out just where the data are today.
What should be believe from the simulations?
Ishiyama etal 0708.1987. dashed line from Moore etal
Where are the bodies? Little debris in
inner MWG?
Field halo
Thick disk
Thin disk
dSph members
NB: the pre-enrichment
problem has become more
extreme: no very metal poor stars
are found in dSph or GC
SMC
LMC
dIrr
Sgr
including thick disk (red) and thin disk (blue) stars: Chemically the local halo is much
more similar to the thick disk (progenitor?) than anything else, but has very different
orbital angular momentum.
Sgr and its clusters are shown from Sbordone etal A+A 465 815 2007
Very many attempts to model
feedback on CDM structure….

Some of our examples:

Read et al 2006 MN 367 387, MN 366 429, 2005 MN 356
107…; Fellhauer etal in prep
Conclusion:

DM halos certainly
respond to tides and
mass-loss, but secularly
If various histories leave
similar mass profiles,
history cannot be dominant
Stellar kinematic data across faces of dSph now quite
extensive e.g. Gilmore et al 2007 MOND problem…
dSph
Seitzer 1983;
van de Ven
et al 06
Globular
cluster
M/LV ~ 2.5
dSph: only one part of the challenge

Among the first systems to collapse, form stars
Star formation history and chemical enrichment are
sensitive probes of stellar ‘feedback’, galactic winds,
ram pressure stripping, re-ionization effects..
 BUT all seem pre-enriched


Most extreme (apparently) dark-matter dominated
systems: trends contain constraints on its nature
(Dekel & Silk 1986; Kormendy & Freeman 04; Zaritsky et al 06)

What are mass profiles within dSph? CDM
predicts a cusp in central regions


Accessible through current observations
Luminosity and mass functions critical tests
The MWG cdm challenge is not rare:
large disk galaxies
with no bulge are common, and are a
very serious challenge for CDM.
They should not exist.
In fact large old disks are a REALLY big
challenge…
cf Kormendy & Kennicutt ARAA 2004
and arXiv:0708.2104
The small pseudo-bulge here is a disk bar.
Kormendy & Freeman 04
Disk galaxy scaling
relations are wellestablished
Do the dSph
(green) fit?
Yes, KF
No, they tend to lie
OFF
the disk galaxy
extrapolations..
Limits instead
Bulk of stellar halo is
OLD, as is bulge:
Did not form from
typical satellites
disrupted later than a
redshift of 2
Unavane, Wyse
& Gilmore 1996
Scatter plot of [Fe/H] vs B-V for local high-velocity
halo stars (Carney): few stars bluer (younger)
than old turnoffs (5Gyr, 10Gyr, 15Gyr Yale)
Mass measurements: the early
context
•
•
•
The ``standard’’ value for local DM at the Sun is 0.3GeV/c²/cm³, all in a
`halo’ component
(cf pdg.lbl.gov: Eidelman etal 2004)
the original work, and origin of this value, is the first analysis to include a
full 3-D gravitational potential, parametric modelling, and a direct
determination of both the relevant density scale length and kinematic
(pressure) gradients from data, allowing full DF modelling for the first time:
Kuijken & Gilmore 1989 (MN 239 571, 605, 651), 1991 (ApJ 367 L9);
Gilmore, Wyse & Kuijken (ARAA 27 555 1989):
cf Bienayme etal 2006 A&A 446 933 for a recent study
• Dark halos are `predicted’ down to sub-earth masses; but…
• Neither the local disk, nor star clusters, nor spiral arms, nor GMC,
nor the solar system, have associated DM: Given the absence of a
very local enhancement, what is the smallest scale on which DM is
concentrated? How can sub-halos in dSph galaxies with star
formation over 10Gyr avoid collecting any baryons?
From kinematics to dynamics: Jeans
equations, simple and robust method

Relates spatial distribution of stars and their velocity dispersion
tensor to underlying mass profile
Either (i) determine mass profile from projected dispersion
profile, with assumed isotropy, and smooth functional fit to
the light profile
 Or (ii) assume a parameterised mass model M(r) and
velocity dispersion anisotropy β(r) and fit dispersion profile
to find best forms of these (for fixed light profile)
Full distribution function modelling, as opposed to velocity
moments, also underway: needs very large data sets. Where
available, DF and Jeans’ models agree.
King models are not appropriate for dSph  too few stars



Main sequence luminosity functions of UMi dSph
and of globular clusters are indistinguishable.
 normal stellar M/L
HST star counts
Wyse et al 2002
M92 
M15 
0.3M
Massive-star
IMF constrained
by elemental
abundances –
also normal
From Walcher etal 2006 apj 649 692
Challenges for small-scale DM









On large [>Mpc] scales LCDM is an astonishingly good
description of data, but n~-1 (and maybe w=-1….) so not
much physics made clear: lots still to learn…
On galaxy scales there is an opportunity to learn some
physics: everything should happen late. But.
0: big old galaxies, big old disks, SFR peaks z>1,
1: the MWG has a thick disk, just one of them, and it is
old. This seems common.
2: massive old pure-thin-disk galaxies exist. None should
3: Sgr proves late minor merging happens, but is clearly a
rare event
4 the substructure problem – where are the bodies?
5 the feedback problem: what is it
6 the early enrichment problem: what did it? When?