Large-Scale Winds in Starbursts and AGN

David S. Rupke University of Maryland

Collaborators:Sylvain VeilleuxD. B. Sanders

v = -1550 km s-1

Rupke, Veilleux, & Sanders 2005a,b,c, submitted

Density Temperature

Cooper, Bicknell, & Sutherland 2005, in prep., and . . .Veilleux, Cecil, & Bland-Hawthorn 2005, ARAA, in press

Sample

• 78 starbursts– 50% luminous infrared galaxies (LIR > 1011 L)

– 50% ultraluminous infrared galaxies (LIR > 1012 L)

• 26 AGN– mostly Seyfert 2 ULIRGs– a few Seyfert 1s

• 104 total galaxies!

the largest superwind survey to date at z < 3

Method

• Spectroscopy of the Na I D doublet– Moderate resolution 65-85 km s-1

– Data from Keck II (10m), MMT (6.5m), KPNO (4m)

• Fit absorption lines– multiple velocity components– Gaussians in optical depth– yields velocity, Doppler parameter, optical depth, covering

fraction

• Compute– Mass, Momentum, Energy, and their outflow rates– *Simple model of constant-velocity, mass-conserving wind

11221f yr M

s km 200

Δv

cm 10

(H)

kpc 5

rC

0.4

4π / Ω 5

t

M

N

d

d

(* Model assumes thin shell, time-averaged outflow rates)

Detection rates• Comparison of different subsamples

– starburst LIRGs 45% (± 10%)

– Seyfert 2 ULIRGs 45% (± 10%)

– starburst ULIRGs 70 – 80% (± 10%)

• Differences in detection rate appear to reflect GEOMETRY– wind opening angle for moderate starbursts and Seyfert

2s in local universe is CΏ ~ 0.4-0.5

there are winds in all starburst LIRGs and Seyfert 2 ULIRGs!

• Same conclusion applies to ULIRGs

Velocities

V(LIRG) < V(SB ULIRG) < V(Sy2 ULIRG) 100 km/s < 170 km/s < 220 km/s

V(H II) < V(LINER) < V(Sy2)120 km/s < 230 km/s < 310 km/s

Spectral Type

Star Formation rate / Nuclear Activity

} significant

AGN- vs. Starburst-driven winds?

• Can we demonstrate that AGNs help drive winds in Seyfert ULIRGs?– No! The statistics don’t convincingly indicate it.

– But there are hints . . .

• Seyfert 2 ULIRGs are statistically indistinguishable from SB ULIRGs– however, the SFRs and outflow detection rates are

lower in Seyfert 2 ULIRGs than in SB ULIRGs

– So differences between Sy2 ULIRGs and LIRGs may imply an AGN contribution.

Circular velocity Star formation rate

Ou

tflo

w v

eloc

ity

isothermal escape speed

= Starbursts = Seyfert 2s

Outflow properties vs.Galaxy properties

dwarf galaxies from Schwartz & Martin 2004

Murray et al. 2004, Martin 2005

Circular velocity Infrared luminosity

Mas

s ou

tflo

w r

ate

Starburst99 prediction(tSB > 40 Myr, Z = Z)

= Starbursts = Seyfert 2s

Circular velocity Infrared luminosityMom

entu

m o

utf

low

rat

e SB99

= Starbursts = Seyfert 2s

Radiation pressure

Circular velocity Infrared luminosity

En

ergy

ou

tflo

w r

ate

SB99

= Starbursts = Seyfert 2s

Slopes of Correlations

• Strong dependence on galactic mass– velocity, mass, momentum, and energy all increase

sharply with circular velocity

– Power-law slopes of 3-5 !• Linear dependence of mass, momentum on SFR

– but energy increases more sharply as SFR increases increase in thermalization efficiency with SFR?– v SFR0.2

• Metallicity effects– incorporating the K-band L–Z relationship (Salzer et al.

2005) does not change these conclusions significantly– I.e. metallicity is not driving these trends! (preliminary)

Correlations at SFR 100 M yr-1

• We observe a statistically significant flattening in dependence of outflow properties on galaxy properties.

• Possible explanations:– depletion of gas reservoirs– decrease in thermalization efficiency– velocity ceiling (Murray et al. 2004, Martin 2005)

– saturation in star formation surface density (Strickland et al. 2004a,b)

[O III]

[N II]

Na I D

model

Vmax(ionized) Vmax(neutral)

Neutral/Ionized CorrelationsF

WH

M [

O I

II] 5

007

Na I D velocity Na I D velocity

All galaxies Galaxies with BELA

Mrk 231 – AGN + Starburst winds?

• A Seyfert 1 ULIRG• Small-scale AGN outflow observed in Na I D

– broad, high-velocity (v = 4000 – 8000 km s-1) absorption (e.g., Boksenberg et al. 1977)

– highest velocity component is variable

• A large-scale outflow observed– blueshifted emission lines (Hamilton & Keel 1987)

– blueshifted absorption lines (v 2000 km s-1)

• Jet or wide-angle outflow?– disk geometry and measured velocities favor wide-angle outflow– however, jet could still inject energy . . .

blueshifted Na I D

vsys

Nu

clea

r of

fset

(k

pc)

4 kpc

4 kpc

Heliocentric velocity

Summary/Outlook• Winds occur in ~ all LIRGs and ULIRGs• Can’t yet convincingly demonstrate influence of AGNs on

large-scale outflow– need more data on Seyfert 2s– there are hints

• Outflow and galaxy properties are strongly correlated– flattening at high SFR, galaxy mass

• Ionized and neutral gas is correlated– Vmax(ionized) Vmax(neutral)

• Mrk 231 is a good example of a galaxy with both a small- and a (spatially-resolved!) large-scale wind.