KIAS Workshop , October 27-28, 2008

Cosmic Voids and Void Galaxies

Michael S. VogeleyDepartment of Physics

Drexel Universityvogeley@drexel.edu

SDSS galaxy map

Voids as Cosmic Laboratories• Voids form by gravity through expansion of initially

underdense regions Voids grow more spherical with time Matter falls outward onto dense filaments and walls Velocity field is radial and regular (smooth)

• CDM theory predicts many low mass dark matter clumps (halos) in voids. Are these found?

• Galaxy formation connects mass to light Dark matter halos collapse Baryons cool and fall onto halos Stars form and cause energy feedback (supernovae) Black holes form, feed on gas, and cause feedback Galaxy interactions both stimulate and disrupt SF

Finding Voids in Galaxy Redshift Surveys

Hoyle & Vogeley 2002, ApJ 566, 641 (PSCz, CfA)

Hoyle & Vogeley 2004, ApJ 607, 751 (2dfGRS)

Pan, Hoyle, & Vogeley 2008, in prep (SDSS)

The Void Finder Algorithm

Procedure:

• Initial classification of galaxies as wall/void galaxies

• Detection of empty cells

• Growth of maximal spheres

• Unification of overlapping voids

• Calculation of void underdensity, profile

Hoyle & Vogeley 2002, ApJ, 566, 641

Goal: identify large voids that are dynamically distinct elements of large-scale structure = “bucket-shaped” voids with flat density profiles and sharp edges, <-0.8

Initial Classification of Void vs. Wall Galaxies

7h -1 Mpc

7h -1 Mpc

Void Galaxy

Wall Galaxy

If d3>7h-1 Mpc, then /<-0.6

Then grow maximal spheres bounded by wall galaxies

VoidFinder Works in Simulations

VoidFinder applied to “galaxies” in simulations

Radii of voids match sharp boundaries of voids seen in both galaxies and dark matter

Density within voids nearly constant, reaches mean at nearly twice void radius

Benson, Hoyle, Torres, & Vogeley 2003, MNRAS 340, 160

Voids in the PSCz SurveyYellow – voidsRed – void centersBlue – wall galaxies

Largest void = 17.85 Mpc/h

Average (of rmin = 10 Mpc/h) = 12.4 +/- 1.7

Nearly identical results for UZC.

Same voids found in IR-selected PSCz and optically-selected UZC.

Hoyle & Vogeley 2002, ApJ, 566, 641

Voids in the 2dFGRS

Red – void centersBlack – wall galaxies

289 voids in total (zmax=0.1)

Largest void radius =19.85 Mpc

Average(r > 10) =12.4 +/- 1.9 Mpc

Similar to PSCz and UZC results

Hoyle & Vogeley 2004, ApJ, 607, 751

QuickTime™ and aGIF decompressor

are needed to see this picture.

Voids in SDSS: intersection with 10 h-1 Mpc Slices

Pan, Hoyle, Vogeley 2008

Density Profiles

Linear Theory (Sheth & van de Weygaert)SDSS Voids

Observed voids show “bucket shape” profile with sharp edges, as predicted by linear gravitational theory.

Voids in SDSS DR5

Parent galaxy sample:

r<17.77, z<0.107, area = 5000 sq deg, 394,984 galaxies (after boundary cuts)

Volume-limited sample:

density field from 61,084 galaxies, M<-20.0

Results of voidfinder:

617 voids R>10 Mpc/h

40,635 void galaxies r<17.77 (10% of galaxies are in voids)

Pan, Hoyle, Vogeley 2008

Filling Factor of Different Void Sizes

Voids with effective radii ~14-22 Mpc/h fill most of void volume. Total filling factor is 50% for voids with average underdensity =-0.8 (20% of cosmic mean).

Void volume per bin of void radius

Shapes of Voids: Oblate or Prolate?

Fit triaxial ellipsoid to each void, with axes a, b, c (largest to smallest)

More very prolate voids (c/b>0.75) than very oblate. Very prolate voids are found to be caused by overlap of pairs of nearly spherical sub-voids.

Same behavior in mock surveys from Park & Kim and no significant variation from real to redshift space.

Very Prolate

Very Oblate

Alignment of Voids

Simulations: Real Space Redshift Space

We find no statistically significant alignment of the principal axis of voids with the line of sight in observations or simulations (neither real nor redshift space).

SDSS voids

Mock surveys by Park & Kim

Summary of Void Properties Void distributions of 2dFGRS, PSCz, UZC, and SDSS agree; void properties

are robust No detected evolution of void size with redshift in nearby universe Voids are on average ~12 h-1 Mpc in radius, but largest void larger in larger

surveys Density profiles plateau in center to = -0.95 Density profiles reveal sharp void edges, as predicted from gravitational

instability Filling factor 40% at density contrast for -0.8 Voids with effective radii 14-22 h-1 Mpc occupy most of void volume Oblate or prolate: very slight tendency towards prolate Alignment: not statistically significant Evidence of redshift space effects? Not so far… Voidfinder appears to detect dynamically distinct elements of large-scale

structure (peaks in initial gravitational potential, outflows in velocity, sharp boundaries in density)

Earlier papers on photometric and spectroscopic properties of SDSS void galaxies found that void galaxies are fainter, bluer, more disk-like, and have higher specific star formation rates. See

Rojas, Vogeley, & Hoyle 2004, ApJ, 617, 50 (Photometric Properties)

Rojas, Vogeley, & Hoyle 2005, ApJ, 624, 571 (Spectroscopic Properties)

Hoyle, Rojas, & Vogeley 2005, ApJ, 620, 618 (Luminosity Function)

Properties of Void Galaxies

Newer evidence places these trends in the context of a Morphology-Luminosity-Local Density Relation that extends from clusters deep into voids and reveals a few residual environmental effects peculiar to voids.

Park, Choi, Vogeley, Gott, & Blanton 2007, ApJ, 658, 898

Morphology-LuminosityLuminosity-Local Density Relation

At fixed L, morphology is a strong function of density

At fixed density, morphology is a strong function of L

Color-magnitude relations

• Early type “red sequence” shifts blueward by 0.025 mag from high to low density

• Late type “blue sequence” shifts blueward by 0.14 mag at low density

Size and environment

Small monotonic dependence of size on local density for all galaxies except the brightest E’s.

Galaxies in voids are slightly smaller: M=-19.7 galaxies are 8% smaller at the lowest density, for both early and late types (but possible sky subtraction error?).

Star formation rates (H)

At fixed morphology (early vs. late) and luminosity, very small dependence of SFR on local density: log[W(H)]=1.466-0.046 log(1+) for M=-18.9 galaxies, weaker for brighter galaxies

Early Late

Results on Environmental Dependence Strong local density-morphology-luminosity

relation. Environment matters down to very low density!

At fixed luminosity and morphology, properties of M<-19 galaxies show only weak dependence on density

Residual effects at low density Color-magnitude shifts blueward, particularly for late

typesSizes of galaxies are smaller Star formation in late types is higher

Morphology-density relation steepens for faint galaxies (for smoothing ~12 h-1 Mpc)

Mergers and Interactions in Voids: rare events in voids are clean test of galaxy

formation/interaction models

• Close Encounters of the First Kind (possible tidal interaction)

• Close Encounters of the Second Kind (close tidal interaction)

• Close Encounters of the Third Kind (merged)

White, Pan, Hoyle, & Vogeley (2008)

Interacting Void Galaxies

1. Tidal interaction

2. Overlap

3. Merged

Close Encounters of the First Kind

Close Encounters of the Second Kind

Strong star formation in tidally-disrupted void galaxies

Close Encounters of the Third Kind

Recently merged void galaxies show evidence of nuclear activity (LINER, Transition spectral types?)

Yes, there are Active Galactic Nuclei in voids!

Constantin & Vogeley 2006, ApJ 650, 727

Constantin, Hoyle, & Vogeley 2008, ApJ 673, 715

Finding AGN among SDSS galaxies

Seyferts

H II

nuclei

LINERs

Transition objects

emission-line

20% of all galaxies are strong line-emitters

52% H II20% LINERs7% Transition obj’s

5% Seyferts

(Constantin & Vogeley 2006)

AGN of all types in voids and walls

Constantin, Hoyle, & Vogeley 2008 “AGN in Void Regions”

AGNAGN in Voids: Populations

AGN of all types exist in voids:

(Constantin, Hoyle, & Vogeley 2007)

Fraction in voids

Fraction in walls

Seyferts 1.5% 1.5%

LINERs 2.0% 4.1%

Transition Objects

5.3% 6.4%

H IIs 32.8% 20.8%

• No AGN in bright (L>>L*) void galaxies• Seyferts more frequent among Mr ~ -20 mag void galaxies• Otherwise very similar AGN occurrence rate for S’s, L’s, and T’s (but not HII’s!)

fractions

in voidsin walls

Compare at fixed brightness

AGNAGN in voids: accretion activity

•Accretion rates may be lower in void AGN…lower fueling rate?

Log L[O I]/4 Log L[O III]/4

in wallsin voids

• Void galaxies have higher SFR per unit mass (Rojas, Vogeley, Hoyle 2005)

Gas available for forming stars but not driven efficiently to nucleus?

(Constantin, Hoyle, & Vogeley 2007)

Seyferts

LINERs

AGNAGN in voids: local environment*

In voids: HII’s have the closest nn, LINERs have the farthest nn LINERs most isolated

In walls: LINERs have closest nnHI’s have the farthest nn HII’s most isolated

*as measured by distance to nearest bright (M<-19.5) galaxy in volume-limited sample

(Constantin, Hoyle, & Vogeley 2007)

NN distance

voids

NN distance

walls

LINERs 5.9 0.6 1.3 0.05

Seyferts 5.8 0.9 1.6 0.1

Transition Objects

4.8 0.4 1.8 0.05

H IIs 4.4 0.2 1.9 0.03

HII’s in poor groups?

LINERs not driven by close interactions

AGN Clustering

H IIs: s0= 5.7 0.2 h-1

Mpc

Seyferts: s0= 6.0 0.6

LINERs: s0= 7.3 0.6

All galaxies: s0= 7.8

0.5

Absolute Magnitude limited samples, -21.6 <M < -20.2

higher peaks in the density field are more clustered

Seyferts & H II’s: - less massive Dark Matter halos

LINERs: - more massive Dark Matter halos

MBH Seyferts < MBHLINERs

MDM halo ~ MBH

(Ferraresse 2002, Baes et al. 2003)

(Constantin & Vogeley 2006)

Results on Void AGN

• All types of AGN are found in voids, though 50% fewer LINERs and 50% more HII’s

• No AGN in the brightest void galaxies• Excess of Seyferts in ~L* void galaxies• Possible lower accretion rate in void AGN• Local environments (nearest neighbor) of AGN

types are opposite in voids/walls• LINERs more clustered than Seyferts. HII’s least

clustered of all.

Does the Void Phenomenon indicate an intrinsic failing of the CDM model?

• Puzzle: Suppression of star formation in voids is required to match CDM theory to observation, but void galaxies actually have higher star formation rates than elsewhere.

• Hypothesis: Higher star formation rate in void galaxies occurs because void galaxies retain a reservoir of baryons that continues to feed star formation long after it shuts down in denser regions.

• No evidence for “failed” galaxies in voids. Extensive searches for HI emission find gas-rich optically-detected void galaxies but no evidence for dark objects at the masses predicted by CDM.

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

ALFALFA HI Survey

Optical

HI-only

Both

Haynes (2008)

Publications about nothing

Methods for void findingVoidFinder (Hoyle & Vogeley 2002, 2004)Comparison of void finders (Colberg et al. 2008)

Statistics and Properties of VoidsVoid Probability Function (Hoyle & Vogeley 2004)Void shapes and alignments (Pan, Hoyle, Vogeley 2008)

Properties of observed void galaxies Photometry (Rojas, Vogeley, Hoyle 2004)Spectroscopy (Rojas, Vogeley, Hoyle 2005)Luminosity Function (Hoyle, Rojas, Vogeley 2005)Mass function (Goldberg et al. 2005)AGN in voids (Constantin, Hoyle, & Vogeley 2008)Environmental dependence (Park, Choi, Vogeley, Gott, & Blanton 2007)Mergers in voids (White, Vogeley, Pan, Hoyle 2008)

Simulations of void galaxiesN-body + Semi-Analytic Models (Benson et al. 2003)Specialized void simulations (Goldberg & Vogeley 2004)