1. Ultraluminous X-ray Sources Andrew King, University of Leicester ² L x (apparent) > 10 39 erg s -1 = L Edd (10 M ¯ ) ² do ULXs contain intermediate—mass black holes, M » 10 2 – 10 4 M ¯ (IMBH) ? Penn State 22.5.04

2. major constraint: ULX – star formation connection, e.g. Antennae

3. Using IMBH to make ULXs in star-forming galaxies 1. If IMBH are primordial (Pop III), new star clusters must `light up’ accretion: -- unclear how a primordial IMBH acquires a companion star

4. IMBH formation in dense star clusters? either 2. merge stars, t merge << t MS and build up large M (Gurkan et al. 2003; Portegies Zwart et al., 2004) problem: mass loss in merger? or 3. merge black holes  IMBH (Miller & Hamilton, 2002) problem: GR reaction: merged BH lost from cluster with low M

5. in all 3 cases, any ULX is formed in a cluster ² most ULXs are observed near but outside clusters -- must eject (with companion star?) ² make at most 1 ULX per cluster, i.e. > 10 5 M ¯ needed to make each ULX

6. could ULXs instead be an unusual phase of X-ray binary evolution? (King et al., 2001)

7. (Grimm, Gilfanov & Sunyaev, 2003) no break at 10 39 erg s -1 : most ULXs are HMXBs

8. likely candidates: 2 types (1) high—mass X—ray binaries thermal—timescale mass transfer rate M dot (tr) » M donor /t KH » 10 -4 - 10 -3 M ¯ yr -1 nuclear-timescale mass transfer rates comparable: black hole mass can grow significantly

9. star formation MS evolution of massive stars, < 10 8 yr high-mass X-ray binary (wind-fed) » 10 4, 5 yr star fills Roche lobe, very high M dot, ULX phase, » 10 3, 4 yr ULX phase reached in < 10 8 yr after SF

10. ² present in star-forming regions ² found near but outside clusters – SNe kicks ² thermal—timescale phase is like SS433 viewed `from the side’ high—mass X—ray binaries:

11. low-mass donorblack hole with unstable accretion disc (cool edges) (2) bright, long-lived soft X-ray transient outbursts (SXTs) ² present in both ellipticals and spirals ² long outbursts like GRS 1915+105 (on since 1992)

12. ² L Edd = 4.4 £ 10 39 erg s -1 (20 M ¯ BH, hydrogen-depleted accretion) two ways of increasing this: (1) ² GRS 1915+105 has L > 6 £ 10 39 erg s -1 with BH mass 14M ¯, i.e. > 3 L Edd ² with mild anisotropy apparent luminosity can reach » 4 £ 10 40 erg s -1 How does an X—ray binary appear so luminous?

13. luminosity (2) ² extremely high mass transfer rates M dot (tr) » 10 3 – 10 4 M dot (Edd) ² outer disc `unaware’ of this until radius R Edd where GMM dot (tr) /R Edd » L edd_

14. ² then total disc luminosity is L disc = L edd [1 + ln(M dot (tr)/M dot (Edd)] » 10L Edd

15. ² thus expect L » 1 – 4 £ 10 40 erg s -1 for 20M ¯ BH with hyper-Eddington accretion ² characteristic blackbody radius R » 10 9 cm ² cf ultrasoft components in ULXs e.g. NGC 1313 (Miller et al 2003: – if instead R is assumed to relate to BH size, get M » 10 3 M ¯ )

16. Outflows from ULXs ² M dot >> M dot (Edd), so most mass expelled ² optically thick outflow with M dot (out)v » L Edd /c ² outflow momentum sweeps up ISM  nebula

17. E out »  M 2 c 2 » 10 52 erg » hypernova energy ² ULX nebulae larger than SNR ² supermassive BH analogue  M-  relation for galaxies:

18. Gao et al., 2003

19. star formation ring began expanding t * = 3 £ 10 8 yr ago, but takes < 10 7 yr to pass any radius ULXs live t life < 10 7 yr, so number of `dead’ ones inside ring is N > (n/bd)(t * /t life ) > 300/bd where b is anisotropy and d is duty cycle (both <1) (King, 2004)

20. ² mass transfer lifetime ~ M 2 /L of ULX < 10 7 yr ² companion star’s MS lifetime < 10 7 yr, otherwise ULXs form after ring has passed ² consistent with 3000 super—Eddington HMXBs with M 2 > 15M ¯ ² but IMBH binaries transient (small disc) so duty cycle d << 1 ² requires > 3 £ 10 4 IMBH, and thus > 10 10 M ¯ in clusters, most mass not accreted

21. ² population properties of ULXs in star-forming galaxies similar to HMXBs, but incompatible with IMBH ² luminosities suggest HMXBs in super-Eddington phase ² outflows  nebulae most ULXs are HMXBs or SXTs

22. ² exception? M82 ULX : L > 10 41 erg s -1 too high for stellar-mass BH ? other sources possible too, but may be superpositions (check variability)

23. ² number of such `hyperluminous X—ray sources’ (HLXs) is very small – at most one per few galaxies ² Occam’s razor: try existing BH models – stellar—mass binaries or galactic nuclei ² not stellar—mass: galactic nuclei? (King & Dehnen, 2004)

24. ² hierarchical merging every large galaxy has 10 – 100 satellites ² most orbits miss host, but occasional collisions

25. ² if colliding satellite retains central BH and star cluster, tides trigger accretion, just like AGN ² satellite can have BH mass > 10 4 M ¯ ² accretion time << orbital timescale: HLX activity only close to galaxy plane ² passage of satellite stimulates star formation: HLX accompanied by stellar—mass ULXs

26. Summary ² most ULXs are stellar—mass XRBs rather than IMBHs (L < 10 41 erg s -1 ) ² HLXs (L > 10 41 erg s -1 ) may be captured satellite galaxy nuclei ² high L from large accretion rate, super—Eddington accretion or anisotropic emission ² ULX – star formation and HLX – galaxy formation links