Ultraluminous X-ray Sources in Nearby Galaxies

Q. Daniel Wang (Univ. of Massachusetts, Amherst)

In collaboration with Yangsen Yao, David Smith, Yu Gao,

etc.

M51: X-ray sources & H image (Terashima & Wilson 2003): Large, medium, and small circles: L(0.5-8 keV) > 1039, (5-10) × 1038, and (1-5) × 1038 erg/s

Ultraluminous X-ray sources (ULXs) are extra-nuclear persistent point sources, each with isotropic Lx > (1-3) x 1039 erg/s, or > the Eddington luminosity of a ~10 Msun object.

Not seen in Local Group galaxies (probably except for GRS 1915+105, Lx~1039 erg/s; MBH ~ 14 Msun; Grener et al. 2001).

Why are ULXs interesting?

• The brightest X-ray sources in galaxies (except for AGNs)

• Potentially intermediate-mass black holes (IMBHs)– a link between stellar and supermassive

BHs– probably with a cosmic mass density > that

of supermassive BHs – Remnants of Pop III stars and/or formed in

star cluster?

• Impacts on the ISM– associated with very energetic structures– Acceleration of cosmic rays?

Outline • Brief history • Where to find ULXs?• Nature of ULXs: stellar mass BHs or IMBHs?• X-ray Properties

– Temporal– Spectral: Comptonized multi-color disk (CMCD)

modeling

• Evidence for IMBHs• How to form IMBHs?• ULXs and their environs• Summary and Future Recent Review papers:

•Miller & Colbert (2003)•Van der Marel (2003)

Brief History of ULX study

• Discovered with Einstein X-ray Observatory (Long et al. 1983; Fabbiano 1998)

• A few were characterized with ROSAT and ASCA (e.g., Colbert & Mushotzky 1999; Makishima et al. 2000)

• Chandra accurate positioning for IDs• XMM-Newton good S/N for spectral and

timing analysis• Recent extensive multi-wavelength

observations and theoretical studies

Where to find ULXs?

• The ULX rate (Bregman & Liu 2004):

– 0.29±0.08 ULXs per 1010 Lo,sun for spirals

– 0.02±0.05 ULXs per 1010 Lo,sun for ellipticals

• Tend to be associated with SF regions• Brighter ULXs tend to be found in

outskirts of galaxies:– e.g., M81 X-9 (Wang 2002), Cartwheel galaxy (Gao

et al. 2003), and NGC4559 X-7 (Soria et al 2003).

– low metallicity effect?• Lower mass-loss rate more massive BHs• Longer Roche-Lobe filling phase

The Antennae

Fabbiano et al. (2003)18 ULXs!

Cartwheel galaxyGao et al. (2003)

WFPC2 B-band image and

0.3-7 keV intensity contours

0.3-1.5 keV image and

1.5-7 keV contours

D=122 Mpc

• At least, 10 ULXs in the ring• ULXs are close to, but

typically not right on, optical peaks (too much extinction?)

• Lifetime of the ULX phase is < 107 yr

• Total number of dead ULXs ~ 300/bdb – beaming factord – duty cycle

• Assuming one IMBH formed from a ~3x105 Msun cluster, a total > 108 Msun/ cluster mass is need - efficiency to form a ULX,

e.g., capturing a companion.

King (2004)

Alternatives are probably fine:

•IMBHs are from Pop III stars

•IMBHs powered by the SN fallback (Wang 2002; Li 2003)

•X-ray binaries with Stellar-mass BHs and with strong beaming

•Very young SNRs

Difficult to explain with the IMBH X-ray binary scenario

3x108 yr

~107 yr

Cartwheel-X7

• L(0.5-10 keV) = 1.3 x 1041 erg/s

• Might be a composite of multiple sources

Nature of ULXs• Background AGNs (~<10%)

– Normally optical, IR, and/or radio bright (e.g., Foschini et al. 2002)

• Very young SNRs– With Lx up to ~1041 erg/s (SN1988Z; Fabian &

Terlevich 1996), easily IDed in optical and radio

– However, some may contain bright X-ray compact sources, e.g., NGC 6946 MF16:

• Bright radio and optical nebula• age ~ 3.5 x 103 yr• Variable in X-ray on both short and long scales

(Roberts & Colbert 2003) • Hard X-ray spectrum similar to most other ULXs

• Most of ULXs appear to be accreting BHs

Stellar-mass or intermediate-mass?

• Truly super-Eddington– E.g., accretion disks with radiation-driven

inhomogeneity (Begelman 2002). But the limit is probably less than a factor of 10 higher.

• Beamed or jetted toward us (King 2002; Markoff et al. 2001)– Similar to Galactic microquasars– Strong temporal variability expected

• Several ULXs do show such variability• But most ULXs remain steady

– Perpendicular to the disks, thus no eclipsing• A couple of ULXs do show possible orbital periods

X-ray temporal variability

• Mostly persistent (within a factor of < 2).

• Strong aperiodic variability in a few ULXs, e.g., M101-P098 (Mukai et al. 2003).

• A few with apparent periodic variability.• PDS of some ULXs show a low frequency

break:– E.g., 0.028 mHz for NGC4559-X7 (Cropper

et al. 2004) 103 Msun, interpolated from the break frequency and mass relationship between stellar and supermassive BHs.

ULX M101-P098

(Mukai et al. 2003)

beamed emission or changing photo-

sphere?

QPO of ULX M82-X41.4+60QPO – mostly a disk phenomenon

o = 54 mHz consistent with the IMBH, compared to o ~ 1 Hz for stellar mass BH

•Narrow QPO peak (fwhm=10 mHz) and large amplitude, ruling out multiple scattering

XMM-Newton/EPIC > 2 keV dataStrohmayer, &

Mushotzky (2003)

Circinus galaxy X-1

Bauer et al. (2001)

•Lx ~ 4 x 1039 erg/s

•Apparent period ~ 7.5 hr

•An eclipsing binary?

M51-TW#69• Apparent 2.1 hr period• Very broad dips• Drastic spectral

steepening with decreasing flux.

PN+MOS

Eclipsing?

Terashima & Wilson (2003)

ACIS

Smith & Wang 2004

M51-TW#69: PN+MOS spectrum of

• L(0.5-8)=1.3x1039 erg/s

• Power law with a photon index = 1.8

• Consistent with being completely Comptonized

X-ray Spectra of ULXs:Accretion disk structure

Log

Log

*F

Total disk spectrum

Annular BB emission

Comptonization of MCD

Problems with MCD+PW model:

• Nonphysical extension of PW to low energies

• No radiation transfer

• Little insight to the properties of the corona and its relation to the disk (e.g., incl. angle)

Log

Log

*F

MCD spectrum

CMCD spectrum

Implementation of a CMCD model, based on Monte-Carlo

simulations• Spherically symmetric corona with a thermal

electron energy distribution

• Parameters: Te, , Rc, , plus Tin and normalization (Rin/D)2.

• Assuming that Rin (after various corrections) is the last stable circular orbit radius, the BH mass M=c2Rin/G.

Yao et al. (2004)

Wang et al. (2004)

Test examples: LMC X-1 and X-2

• Independently estimates of , MBH, and NH

• Data from PeppoSAX – Broad-band coverage– No pile-up– Spectral change

LMC X-1 spectrum

Model Comparisons

LMC X-1 spectrum

Corrected for absorption

Comparisons of key measurements

LMC X-1

Incl. angle (deg) M (Msun) NH (1020 cm-2) Tin (keV)

Indep. Est. 24 < < 64 4 < M < 12.5 --CMCD 23 (< 43) 6.7 (?-?) 50(49 – 51) 0.93MCD+PW 79(74 – 84) 0.93

LMC X-3Indep. Est. < 70 deg > 7 3.8(3.1 – 4.6)a

CMCD 59 (< 69) 6.9 (?-?) 4.5(4.2 - 4.7) 0.98MCD+PW 7.6(6.7 – 8.5) 1.02

a from X-ray absorption edge study

Spectral evolution of LMC X-1

early part

Tin=0.91 keV

= 0.5

late part

Tin=0.99 keV

= 2

No Rin changes is needed!

ULX Spectral Fits

M81-X9

Wang et al. 2004

Notice the effect of the incl. angle

XMM-Newton Observations of Six ULXs in nearby galaxies

Source Galaxy typeD(Mpc)• NGC1313 X-1/X-2 SB(s)d 3.7• IC342 X-1 Scd 3.3• M81 X-9 Im 3.6 • NGC5408 X-1 IB(s)m 4.8• NGC3628 X-1 Sbc 10.0

Wang et al. (2004)

ULX spectral analysis

PN+MOS spectra fitted with the CMCD model

ULX Spectral Fit Results

• Satisfactory fits to the spectra.• Tin (~0.05-0.3 keV) values consistent with the IMBH

interpretation.• Constraints on accretion disk properties such as incl.

angle, etc.

Inferred Parameters from Spectral Fits

• BH mass on the order of ~ 103 Msun each.

• Accretion at a fraction of their Eddington rates.

Wang et al. (2004)

Evidence for IMBHs

• No unambiguous detections of individual IMBHs yet, only observational hints (van der Marel 2002):– ULXs

• High X-ray luminosities• Low frequency QPO or PDS breaks• A few possible eclipsing binaries, thus no jet

boosting

• Spectra consistent with MCDs of low Tin (~0.2 keV) plus Comptonization

• Some show hard/low-soft/high transitions, typical of BH candidate binaries.

– microlensing events– Optical kinematics of centers of nearby

galaxies and globular clusters.

How to form IMBHs?• Remnants of Pop III stars (Madau & Rees 2001)

– A couple of 102 Msun each is predicted.

– Grow by capturing stars in star clusters.– Induce SF in GMCs around them?

• Young star clusters– Formed in a runaway core collapse and merger of MS

stars (Portegies Zwart & McMillan 2002; Miller & Hamilton 2002)

– Fed by Roche lobe overflow from a tidally captured stellar companion (circularized without being destroyed by tidal heating; Hopman et al. 2004).

– Accreting IMBHs may outlive the host clusters.

• Globular clusters (Taniguchi et al. 2000)

Multi-wavelength counterparts

• Rarely radio-bright– Only known candidates:

• NGC5408-2E1400 (0.26 mJy at 4.8 GHz; Kaaret et al. 2003)

• M81-X6 (0.095 mJy at 8.3 GHz; Swartz et al. 2003)

– But consistent with Galactic micro-quasar radio luminosities.

• Optical/UV counterpart – Few ULXs have relatively firm IDs– E.g., NGC 5204 ULX –B0 Ib supergiant plus

NV emission line (Liu et al. 2004), predicting ~ an orbit period of 10 days.

NGC 4565• Edge-on Sb

galaxy• Low SF rate• The ULX is on the

side with little disk absorption.

• The Galactic foreground NH ~ 1.2x1020 cm-2.

Measurement of the intrinsic absorption in the ULX

NGC4565-X4

Wang 2004ACIS-S contours on optical

ULX NGC4565-X4

• Tin = 0.190 (0.191-0.271) keV

• L(0.5-10 keV) = 7 x 1039 erg/s

• M ~ 103 Msun

• Incl. angle = 18 (17-41) deg

• NH = 2.5 (1.9 – 2.7 ) x 1021 cm-2

– In contrast to the Galactic value of 1.3 x 1020 cm-2

– A warm absorber? Similar to the IMBH (M ~ 104 - 105 Msun) AGN of NGC4395 (Shih et al. 2003)

OVII K

NVI K

ACIS-S spectrum

ULX NGC4565-X4

•The optical counterpart as a globular cluster (Wu et al. 2002)

•An IMBH formed in a globular cluster (Taniguchi et al. 2000)?

Impacts of ULXs on Environments

M81-X9

Wang (2002)

Nebula Size ~ 260x350 pc

Shock-heating

Wang 2002

Pakull & Mirioni 2002NGC1313-X2 nebula

• Size ~ 570 x 400 pc• V ~ 100 km/s• n ~ 0.2 cm-3

• E ~ 1.0 x 1053 erg, assuming an 1-D wind bubble

E

W

HoII X-1: an X-ray-ionized nebulae

• Abnormally high [OIII]/H ratio (Remillard, Rappaport & Macri 1995)

• Strong He++ recombination line 4686

• Requiring He+ Lyman continuum (~ 54 -200 eV) ~0.3-1.3 1040 erg/s

• Agreeing with the observed Lx.

• Excluding significant non-isotropic X-ray beaming

Pakull & Mirioni 2002

Nature of the ULX and energetic shell associations

• Superbubble?– Timescale mismatch:

• Dynamic time of such a shell (~ R/v) is too short (~< 106 yr).

• Ionization of the shells is primarily due to shock heating age of the OB association ~> 107 yrs.

– Too much energy is required:• Typically 1052 – 1053 erg, or 1039 – 1040 erg/s (or 1 SN

per 104 -105 yr), energetically similar to 30 Doradus.

• Hypernova remnant?– Shell - interstellar remnant– ULX – stellar remnant, accreting from

• Fallback of the ejecta• Accreting binary with an original or captured

companion (Is the timescale too short?)Wang (2002)

• Shell powered by an X-ray binary?– Available binding energy (~GMBHMc/rBH ~

1054 Mc erg; rBH MBH)

– Required mechanical energy output ~ radiation luminosity

• Consistent with other accreting systems (microquasars or AGNs).

• Wind probably at a speed of ~ c.• Disk winds are observed in X-ray spectra of

binaries and AGNs.

• UV/soft X-ray ionization of nebulae– High electron temperature (H/H ~105 K for

M81-X9)

– Diffuse boundaries (due to long X-ray absorption path-length)

Summary and Conclusions• ULXs represent a heterogeneous population

– Very young SNRs– Stellar mass BHs with beamed and/or mildly super-

Eddington X-ray emission– IMBHs accreting from HN/SN fallbacks or companions,

though no conclusive evidence yet

• A self-consistent Comptonized MCD spectral model has been developed and tested– Satisfactory fits to several best-observed IMBHs

estimates of BH masses, plus constraints on disk incl. angle, etc.

• ULXs are often associated with highly-ionized and/or very energetic nebulae.– Clues to their origins– Constraints on outflows from accreting systems

Future• Longer exposures with Chandra/XMM-Newton:

– Variability: power spectrum break, QPO, and orbital period

– High S/N spectra for more sources diversity and spectral state changes.

• Astro-E2: – high resolution spectrometer for study both emission

and absorption lines– Sensitivity to higher energy photons better

constraints on Comptonization

• Multi-wavelength follow-up:– IR/Optical/UV ID nature of source, dynamic mass,

etc.– Nebulosity beam effect, energy output, and origin