O VI Absorbers at z=2-3Photoionized by Quasars or Tracers of

Hot Gas?

Andrew Fox (ESO-Chile)Jacqueline Bergeron & Patrick Petitjean (IAP-Paris)

H I–H II Si III-Si IV C III-C IV He II-He III N IV-N V O V-O VI

Metal lines as tracers of ionization level13.6 eV 33.5 eV 47.9 eV 54 eV 77.9 eV 113.9 eV

O VI advantages:• O VI is most highly ionized line available in rest-frame UV• Oxygen is most abundant metal in Universe• O VI doublet at 1031, 1037 Å is intrinsically strong

O VI disadvantage:• O VI falls in Ly-a forest blending/contamination. Only detectable at z2-3.

Energy

O VI absorbers have power-law column density distribution (Bergeron & Herbert-Fort 2005)

“Associated” or “proximate” absorbers (at dv<5000 km s-1 from QSO) often removed from sample affected by ionization conditions close to QSO. This talk: Examine this practice (Fox, Bergeron, & Petitjean 2008, MNRAS)

VLT/UVES, Keck/HIRES studies• Schaye et al. 2000• Bergeron et al. 2002• Carswell et al. 2002• Simcoe et al. 2002,2004,2006• Levshakov et al. 2003• Reimers et al. 2001, 2006• Bergeron & Herbert-Fort 2005• Lopez et al. 2007• Gonçalves et al. 2008

O VI probes IGM ionization and enrichment

Is there a proximity effect in O VI?

VLT/UVES Large Program 20 QSOs, high resolution

(FWHM 6.6 km s-1) and high S/N (~40–60)

Searched for O VI absorbers within 8000 km s-1 of zQSO.

zQSO is determined from several QSO emission lines, allowing for systematic shifts (Tytler & Fan 1992)

35 proximate O VI systems detected:

- 26 weak systems- 9 strong systems

UVES Spectra

-200 0 km/s 200 -200 0 km/s 200

Two Populations of Proximate O VIWEAK

◦ log N(O VI)≤14.5◦ Weak N V and C IV◦ 1 or 2 components◦ Velocities < zQSO◦ No evidence for partial

coverage

STRONGo log N(O VI) ≥ 15o Strong N V and C IVo Multiple componentso Velocities clustered around zQSOo Occasional evidence for partial coverage of continuum source.o Truly intrinsic: inflow/outflow near AGN central engine (several mini-BALs)

“Proximity Effect”: change in dN/dz at 2000 km s-1

Proximity zone extends over ~2000 km s-1, not 5000 km s-1.

Intervening systems(Bergeron & Herbert-Fort 2005)

Weak O VI absorbers: trends with proximity

At 2000 km s-1, see change in N(H I) and in N(C IV) but not in N(O VI)

Furthermore, O VI/HI offsets are observedSignificant velocity centroid offsets between O VI and H I are seen in

~50% of the weak O VI absorbers two ions are not co-spatial. (similar fraction of low-z O VI absorbers show offsets; Tripp et al. 2008)

Median b-values • O VI <2000 km s-1 from QSO: b=12.3 km s-1

• Intervening O VI: b =12.7 km s-1 T <1.6x105 K

• Intervening N V: b =6.0 km s-1 (Fechner & Richter 2009)

O VI and N V trace different regions

O VI Component Line Width Distribution

b=(2kT/m + b2non-thermal)

O VI absorbers (even narrow) may not be photoionized; can be formed in non-equilibrium cooling gas

Results of Gnat & Sternberg (2007)

Frozen-in ionization can lead to O VI being present in gas down to ~104 K if the metallicity is close to solar

Are there any physical reasons why such hot gas should exist at z=2-3?

YES: Galactic Winds

YES: Hot-mode accretion

Simulations from Kawata & Rauch (2007)

Simulations from Dekel & Birnboim (2007)See also Fangano, Ferrara, & Richter (2007)

Comparison of high-ion ratiosObservations vs theory (Gnat & Sternberg)

Cooling gas models can explain data if elemental abundance ratios are non-solar:Need -1.8<[N/O]<0.4 -1.9 <[C/O]<0.6

Single-phase photoionization models for IGM O VI absorbers are too simplistic, because1. O VI-H I velocity offsets imply O5+ and H0 occupy different regions2. O5+ may be collisionally- rather than photo-ionized3. Don’t know EGB shape above 100 eV that well

Use caution when combining O VI/H I ratio + CLOUDY IGM metallicity

Implications for O VI absorbers in general

H0, O5+, T~104 K H0, T~104 KO5+, T~105 K

O6+, O7+ T≥106 KEGB

What you see in H I

What you see in O VI

2000 km/s ProximityWarm plasma photoionized as you approach z(QSO), not hot plasma

QSO

N(H I)~1015 N(O VI)~1013.5 N(H I)~1014 N(O VI)~1013.5

Almost 1 dex uncertainty in Jn at 113.9 eV!!!

Simcoe et al. (2004)

Sawtooth modulation by He II Lyman series exacerbates the situation

Madau & Haardt (2009)

We don’t really know what’s happening out here!

In 20 high-quality QSO spectra from UVES, we search for O VI within 8000 kms-1 of zQSO, finding◦ 9 strong absorbers (truly intrinsic, gas near AGN)◦ 26 weak absorbers

Among weak O VI absorbers: dN/dz increases by factor of 3 inside 2000 km s-1

dN/dz in range 2000-8000 km s-1 matches intervening. N(H I) and N(C IV) show a proximity effect (dependence on Dv), N(O

VI) does not. O VI-H I velocity centroid offsets imply at least half the absorbers

are multiphase.

Cannot use O VI absorbers to probe high-energy tail of EGB: too many systematic uncertainties. Narrow O VI can form in radiatively-cooling hot gas, in interface regions that result from galactic winds/hot-mode accretion

Survey for Proximate O VI: Summary

Partial Coverage of Continuum Source

O VI absorber size is <200 kpc, based on lack of Hubble broadening.Simcoe et al. 2002

Strong O VI: ◦ Yes, we see strong O VI clustered around zQSO

Weak O VI: ◦ Yes, we see dN/dz increase by a factor of three within

2000 km s-1 (but galaxies are clustered near quasars).◦ No, the internal properties (b-values, log N) of the O

VI absorbers do not depend on Dv, unlike H I and C IV

Is there a line-of-sight proximity effect in O VI?

Is there observational evidence (at z=2-3) for an extended (~10 Mpc) QSO proximity zone of E>100 eV photons?•No: properties of weak O VI do not require photons at E>100 eV (you can create the O VI with [cooled] hot gas)

C IV ionization fraction in hot gas

Proximity zone extends over ~2000 km s-1, not 5000 km s-1.

1. convert QSO B-magnitude and zQSO to L912 (Rollinde et al. 2005)2. Determine size of “Stromgren Sphere” where QSO radiation density

exceeds estimated EGB radiation density at z=2.5 ( ~10 Mpc)3. Convert size to velocity assuming Hubble Flow and H(z=2.5)=250

km s-1 Mpc-1 (1500-2500 km s-1)