LoCuSS: The midinfrared ButcherOemler effect Chris Haines University of Birmingham LoCuSS: Graham Sm

1
LoCuSS The mid-infrared Butcher-Oemler effect
Chris Haines (University of
Birmingham)LoCuSS Graham Smith (PI), Eiichi
Egami, Richard Ellis, Sean Moran, Alastair
Sanderson and many others...ACCESS Paola
Merluzzi (PI), Gianni Busarello, Russell Smith

2
The Butcher-Oemler effectButcher Oemler
(1984) measured fraction of cluster galaxies with
MVltM1.5 gt0.2 mag bluer in rest-frame B-V than
red-sequenceRapid evolution of cluster galaxies
- fb0.03 at z0 to fb0.2 at z0.4Large
cluster-to-cluster scatter due to dynamical
status of cluster ?
3
The Butcher-Oemler effectSimple view
Star-forming spirals (20) in z0.4 clusters are
transformed into passive S0s by z0 by
ram-pressure stripping, harassment,
starvation.... Why should z0.4 cluster galaxy
population be so different to that at z0 if
clusters themselves havent evolved too much
(e.g. ICM) ?
z0 cluster
z0.4
4
The Butcher-Oemler effectCosmological context
Continuous accretion of galaxies from field50
of z0 cluster galaxies accreted since z0.4,
20 from groups
5
The Butcher-Oemler effectCosmological context
Continuous accretion of galaxies from field50
of z0 cluster galaxies accreted since z0.4,
20 from groupsEvolution must be much more
rapid and widespread
z0.4
z0 cluster
6
The Butcher-Oemler effectMany BO-type studies
since, with contrasting levels of evolution and
cluster-to-cluster scatter seen. Due to selection
biases ?

• Low-mass starbursts increasingly enter MV
  selected samples at high redshifts, creating
  evolutionary trends
• MK-based studies find little evolution of
  cluster population since z1 (both in fb and
  morphological composition)
• Optically selected samples biased to find those
  clusters with high fb
• Little evolution / scatter in massive (sgt600 km
  s-1) X-ray selected clusters
• B-V colour or EW(OII)-selection miss dusty
  star-forming galaxies found in MIR/radio cluster
  surveys

(Holden et al. 2007)
7
Motivation The need for a large MIR cluster
survey

• Sensitivity of ISO/Spitzer has opened a new
  window for studying star-formation in galaxy
  clusters
• 24µm emission provides a clean measurement of
  obscured SF
• ISO/Spitzer studies find population of cluster
  LIRGs (LIRgt1011L?) that imply SFRs 10-30x higher
  than derived from optical (e.g. OII)
• Census of cluster star-formation was strongly
  incomplete
• LIRGs could be key phase in evolution from
  spirals to S0s
• Large cluster-to-cluster variation in levels of
  obscured star-formation
• Related to dynamical status of clusters (Miller
  et al. 03) with cluster-merger induced starbursts
  ? (Kenji Bekkis talk)
• A1689, A2219, A2390 all merging clusters but no
  LIRGs....

8
LoCuSS Local Cluster Substructure Survey

• Multi-wavelength survey of 100 X-ray luminous
  clusters at 0.15ltzlt0.3, selected from ROSAT with
  LX gt 3x1044 ergs / s (no morphological bias)
• Measure cluster-to-cluster scatter in observable
  parameters and scaling-relations (TX, SZ
  Y-parameter, lensing masses, LK, SFRs)
• For first 31 clusters obtaining unique
  multi-wavelength dataset
• Subaru Suprime-CAM (V,i) weak lensing mass maps
  (Okabe et al. 2009)
• Wide-field NIR imaging (J,K) from UKIRT/WFCAM or
  KPNO/NEWFIRM
• Panoramic Spitzer 24µm imaging (25x25)
  reaching 400µJy
• GALEX NUV/FUV (Sean Morans talk) Herschel
  100/160µm FIR data
• Recently commenced spectroscopic follow-up of
  cluster members (NIR selected, KltK2) and 24µm
  sources with MMT/Hectospec
• 200-400 cluster members extending out to 5 Mpc

9
LoCuSS The mid-infrared Butcher-Oemler
effectLoCuSS provides a unique dataset to
examine the Butcher-Oemler effect, selecting
star-forming galaxies from their 24µm emission

• 22 LoCuSS clusters (0.15ltzlt0.3) with wide-field
  (25x25) NIR and 24µm data
• J-K increases monotonically with z, allowing
  efficient selection of cluster gals
• Estimate LIR(z) from F24µm using range of IR SED
  models (Le Floch et al. 2005)
• Galaxy star-forming if LIR gt 5x1010L?
• Limit corresponds to completeness limit
  (F24400µm) at z0.28, SFR8 M?yr-1
• Mass-selected sample with MKltK1.5
• Consider galaxies within 1.5 r500 (r200) as
  determined from Chandra data
• Statistically subtract field galaxies using
  counts from UKIDSSSWIRE (2deg2)

10
LoCuSS The mid-infrared Butcher-Oemler
effectTo measure evolution of fSF, analyse other
8 clusters with suitable NIR and 24µm data Coma,
A1367, 5 clusters from the Shapley supercluster
at z0.05 from the ACCESS project, and Cl002416
(Moran et al. 2007) Use only spectroscopic
members (gt95 coverage for KltK1.5)
11
LoCuSS The mid-infrared Butcher-Oemler
effectSteady increase in fSF from 3 at z0 to
10 at z0.3 (as in BO84)
High-z clusters from Saintonge et al. suggest
flattening in fSF at zgt0.4
Scatter in fSF at fixed z half that from
optical-based BO studies NIR / MIR
selection ?
Strong evolution in fSF within LoCuSS sample
12
LoCuSS The mid-infrared Butcher-Oemler
effectUse of fixed LIR limit assumes no redshift
evolution in SFR (LIR)In field global SFRs at
fixed stellar mass has declined by 10x since z1
(Zheng et al. 2007), while LIR ? (1z)3.20.7
(Le Floch et al. 2005)
(Zheng et al. 2007)
(Le Floch et al. 2005)
13
LoCuSS The mid-infrared Butcher-Oemler
effectUse of fixed LIR limit assumes no redshift
evolution in SFR (LIR)In field global SFRs at
fixed stellar mass has declined by 10x since z1
(Zheng et al. 2007), while LIR ? (1z)3.20.7
(Le Floch et al. 2005)

• If cluster LIRGs are recently accreted from the
  field, their MIR and Ha LFs should appear the
  same as the field, and evolve in the same way,
  as is observed by Bai et al. (2008) and Finn et
  al. (2008)
• Is rapid evolution in fSF due to increase in
  ?IR with redshift or just reflects the expected
  LIR ?(1z)3.2 evolution of infalling field
  galaxies
• To quantify excess evolution of fSF over global
  evolution, remeasure
• fSFs allowing LIR threshold to evolve to match
  global decline in LIR
• LIR gt 5x1010 ( 1z) / (1zmax) 3.2
  L?
• If the mid-IR Butcher-Oemler effect is simply
  due to the cosmic decline in SFRs, the fSF
  redshift trend should now completely disappear

14
LoCuSS The mid-infrared Butcher-Oemler
effect

• Some residual evolution (lt1s) , which could be
  due to increasing fraction of recently accreted
  galaxies (should double by z0.4)

15
LoCuSS The mid-infrared Butcher-Oemler
effectInfalling galaxies may dominate fSF at
rgtr500, so remeasure fSF for rltr500 Evolution
now completely removed fSF ? (1z)0.32Almost
all clusters have fSF0.05 with just 4 active
clusters

• Can fit entire cluster sample with single value
  of fSF 0.056 with scatter just due to
  observational uncertainties (reduced ?20.72)
• Butcher-Oemler effect due to the global decline
  in SF in field galaxies that are then accreted
  into the cluster

16
LoCuSS Correlation with cluster X-ray
luminosityIf star-formation quenched by
ram-pressure stripping may expect to see an
anti-correlation between fSF and LX

• None seen, but we are only including most
  massive clusters
• May be saturation effect, ram-pressure stripping
  may strip gas in all infalling galaxies even for
  our lowest LX clusters
• Anti-correlations seen for much poorer systems
  with s lt 550 km/s (Poggianti et al. 2006)

17
LoCuSS Correlation with dynamical status of
clusterIs excess star-formation seen in merging
clusters as observed by Miller et al. (2005) in
the radio ?

• Large offset between BCG and X-ray centroid one
  indicator of ongoing cluster merger
• No evidence for increased activity in merging
  clusters, including A1914 and A1758
• Same results for 1.5 r500, or considering
  X-ray/lensing density profiles (flatter for
  mergers)
• Galaxies already stripped of gas before clusters
  merge ?

18
LoCuSS Test of the infall scenarioCan we
identify the MIR-bright star-forming galaxies as
the infalling galaxy population, yet to be
quenched by interaction with the cluster ?

• Simple toy-model SF is quenched in infalling
  galaxies on first passage through the cluster
  core
• Take 30 most massive DM halos in Millennium
  simulation (M2001015 M?)
• Follow orbits of each KltK1.5 galaxy from Bower
  et al. SAM within 5Mpc of cluster, and identify
  those infalling into cluster for first time
• Calculate finfalling as function of projected
  cluster-centric radius
• Not all infalling galaxies would be classed as
  star-forming so normalize by observed fSF(field)
  to give fSF(r)

19
LoCuSS Radial population gradientsSplit
clusters into three redshift bins (zlt0.05,
0.15ltzlt0.23, 0.23ltzlt0.30), and estimate composite
radial gradients in fSF (with evolving LIR limit)

• Steady increase in fSF from zero in cluster
  core to 7-13 by 2r500
• Enhanced SF activity at rgtr500 in high-z
  subsample ?
• Consistent to first-order with simple infall
  model

20
LoCuSS Confirmation of LIR evolutionObtained
redshifts for gt90 of 24µm sources in 11 LoCuSS
clusters

• In the two zlt0.2 clusters maximum LIR is
  1x1011 L?
• In the 0.25ltzlt0.30 clusters many galaxies with
  1-3 x1011 L?
• Apparent evolution in LIR validates use of
  evolving LIR limits
• Previous cluster-to-cluster scatter in N(LIRGs)
  simply due to evolution
• ULIRGs extremely rare, except as BCGs (excluded
  from BO analysis)

21
LoCuSS Conclusions

• mid-IR Butcher-Oemler study in 30 massive
  clusters over 0ltzlt0.4
• Identifying galaxies as star-forming from 24µm
  data as LIRgt5x1010L? see rapid increase in fSF
  from 3 at z0 to 10 at z0.3
• Rapid evolution largely due to cosmic decline in
  SFRs in field galaxies which are accreted into
  the clusters at a constant rate
• Within r500, little cluster-to-cluster scatter
  in fSF, with no apparent dependency on either LX
  or cluster dynamical state
• See radial population gradients with fSF
  increasing from zero at the cluster core to
  7-13 at 2r500, consistent with a simple infall
  model in which star-forming galaxies are
  infalling galaxies which havent yet been
  quenched by cluster environment

22
LoCuSS Phase II spectroscopic survey
23
LoCuSS Galaxy evolution in cluster
environmentsNumber density and shape of LF(IR)
allow us to distinguish between transformation
mechanisms (Zhang 2008), e.g. mergers gt cluster
SFR dominated by a few ULIRGs, while
starvation models gt no ULIRGs and few LIRGs
Is global cluster SF dominated by quiescent
spirals whose SF have yet to be quenched, or SF
triggered by galaxy interactions ?
3.6, 8.0, 24µm
UV, HST, IR
24
Shapley supercluster project Spitzer analysis

• Post-BCD data analysis 24µm data reaches 300µJy
  (LIR109L?)
• Similar to Coma, A1367 but for 5 clusters, gt20x
  deeper than LoCuSS
• 24µm/r-band flux ratios gt mass normalized SFR,
  get bimodality!
• Passive galaxies lie along red sequence
• Also find significant contamination by normal
  star-forming galaxies