Steinn Sigurdsson is the official blogger of the Clusters09 Conference: Read on for more...

Thurs, Jan 22, 2009: Afternoon discussion on runaway collisions and the formation of IMBHs.

Simon Portegies Zwart's presentation: Runaway collisions in young dense clusters and their consequences (with a slightly N-body-centric bent). Lively discussion. Apparently some people still believe IRS13E hosts an IMBH (ADS link to original paper), although the IRS13E stars may not be bound to anything at all (Trippe, et al. 2008). The story that SN 2006gy represents the collapse of a massive collisional stellar runaway (Portegies Zwart & van den Heuvel 2007) is mainly based on the existence of 1 M_sun of H inferred from the spectrum. Whether or not you believe that story the prediction is clear: a bright star cluster should appear at the position of the SN when the SN dims sufficiently. It may take years before this is the case, though. Making the jump from presence of an IMBH to ULX is fraught with uncertainty, as the fierce debate tonight confirmed. Unless one adopts optimistic tidal dissipation prescriptions, the present understanding is that it's very difficult to form an X-ray binary with an IMBH from standard few-body scattering (just ask Vicky!). Perhaps we'll hear more about that tomorrow morning during Pepi's coffee hour, though...

Stefan Umbreit's presentation: The effect of primordial mass segregation on the runaway collision process. More discussion about why, exactly, the mass of the collision product is generically 0.2% of the total cluster mass. Although this is the generic result of mass segregation, the claim is that it can be made up to a factor of 10 larger via special (unphysical, some might say) initial conditions. Punchline of Stefan's simulations is that the degree of initial mass segregation (parametrized by the S parameter; Subr, et al. 2008) doesn't make a huge difference for the "final" mass of the collisional runaway, but it can increase the critical cluster virial radius for a runaway by a factor of up to about 4 (clusters with smaller radii generically yield runaways).

Evert Glebbeek's presentation: Merger runaways as seen from the point of view of stellar evolution. Does wind mass loss from stellar evolution prevent you from getting a collision product more than ~100 M_sun? Took collision sequence from SPZ, et al. N-body run, applied Jamie Lombardi and Evghenii Gaburov's "Make Me a Massive Star" code for mergers. Very massive stars may lose a lot of mass in winds, but this is still very uncertain---for example there are no observations confirming this. Evert claims that if a collision product eventually collapses to form a BH, it must burn hydrogen (which is relevant for winds). However, the peanut gallery disagrees, citing old Rees papers on direct collapse of gas clouds to BHs. In any event, the results of Evert's post-processing calculations are clear: what grew to 1000 M_sun in N-body simulations ends up as only a pathetic, 10 M_sun, poor excuse for a star (although it makes it up to ~400 M_sun for a while). Although the radii of the collision product is huge during the runaway process, the 99.9% Lagrangian radius is much smaller. Fortunately, it turns out this is roughly the same as what the N-body models assumed to calculate the collision radius. Mass loss rates are extrapolated into LBV regime, but the claim is they are an underestimate. The winds yield interesting chemical abundances. Oxygen poor? Could be a runaway...

Remarks:


(Evert Glebbeek 27 Jan 2009 16:43): "Very massive stars may lose a lot of mass in winds, but this is still very uncertain---for example there are no observations confirming this." This is a somewhat misleading summary of what I said, or I may not have been very clear: hot (log Teff> 4, say) massive stars are known to lose mass, and their mass loss rates are fairly well known observationally and can be explained very well by current theoretical models (Vink & al (2001), Mokiem & al (2007)). The essential physics here is that of momentum transfer due to spectral lines of (in particular) iron, which is a temperature dependent effect (ie, it applies equally for all stars of a given effective temperature). What is not well known is how the mass loss rates change when a star comes close to its Eddington luminosity, corresponding to the observed Humphreys-Davidson limit; all that is known about mass loss of stars above this observed limit is that it appears to be huge (LBV-type behavior) and much larger than what can be explained by line-driven wind models.

(John Fregeau 30 Jan 2009 12:17): By "very massive" I meant >~ few x 100 M_sun. There are no observations of mass loss from such things, right? And some of your models get up to 400 M_sun temporarily, right? I agree that massive stars lose mass, as you say, and as the papers you mention also say. Didn't mean to imply otherwise.

(Evert Glebbeek 30 Jan 13:29): I reckoned that was what you meant, yes. :) I just thought the sentence might be misunderstood as written.