Monday, November 1, 2010

Session IV

Internal structure of galaxies during high SFR phase

  • dynamics/morphology
  • chemistry
  • dust and gas distribution: ISM structure, resolved studies
Invited Talks
Reinhard Genzel
Gas-rich Massive Disks at z~2
Leo Blitz
The Gas Consumption History Over Cosmic Times
Thorsten Naab
A Theoretical View on the Internal Structure of High Redshift Galaxies
Contributed Talks
Andrew Baker
Cold Gas inventories in Massive High-redshift Galaxies
Linda Tacconi
High Gas Fractions and the K-s Relation for “Normal” Star Forming Galaxies From z = 1-3
Discussion led by Nick Scoville


  1. Much of the differences between the inferred "star-formation efficiencies" between the BzKs and and SMGs is due to the adopted CO-to-H2 conversion factors (alpha~3-5 for spirals vs alpha~0.8 for ULIRGs). However, one could argue that we do not know the alpha value locally to better than a factor of 2, let alone at high-z. Any thoughts on the CO-to-H2 conversion factors at high redshift?

  2. I agree that it seems clear that the adoption of a lower Xco for SMGs as opposed to BzK's is driving the bulk of the shift in the normalization of the KS relation that Emanuele showed (though I think Fabian mentioned to me earlier that there is a shift in the SFE of SMGs even when using the Galactic Xco, and that the shift was only made bigger when using the ULIRG value).

    Where I think there's been little progress is in determining what physical conditions Xco depends on. Obviously it depends on both CO and H2 abundances. But maybe we can remove this effect by assuming in these high-z systems, the CO/H2 abundance is ~1e-4 in regions where the metallicity is roughly solar.

    Another physical effect that might affect Xco is the CO excitation (via density and temperature). If SMGs are denser and warmer in their ISM than typical BzKs, one might assume for a given amount of H2, that you get more CO flux simply from excitation arguments. Then you might expect a lower Xco for SMGs than BzKs qualitatively, though who knows how close it will be to the ULIRG value*.

    *[as a footnote to the "ULIRG value", I seem to remember in Downes and Solomon, there being a reasonably large dispersion in their ULIRG value, with some overlap with the Galactic value]

    I *think* on average, SMGs are more highly excited than BzKs which might lend some credence to this line of thought. This is certainly true in the simulations, and a few observational examples exist suggesting that SMGs peak around J=5 or 6 (Carilli's GN20 study comes to mind) where as I think Dannerbauer and Daddi found that BzK's peak around J=3 or so. Though I'm sure there are deviations from both of these (perhaps lower excitation SMGs are the inspiralling pairs that Chris Hayward showed yesterday, rather than the coalescing mergers?)

    Any thoughts on this?

  3. It is probably interesting at this point to mention that simulations do recover or predict a "double SF law" for disks vs mergers, without any assumption on the X_CO factor since they directly know the gas masses. A bit of this was shown in Teyssier et al 2010 ApJL. Recent simulations (in prep, but see: with improved resolution, SF schemes, feedback , etc, now find this over large samples.

    At the same time, these simulations predict that mergers compared to disks (hi-z mergers vs hi-z disks, or, low-z megers vs low-z disks) have an excess of very dense gas, because the density PDF is not a log-normal anymore. This is due to enhanced ISM turbulence or gas velocity dispersions in interactions, as already observeded by Irwin94 or Elmegreen95! A lower X_CO for mergers could well be justified by this excess of very dense cold gas.

    Of course, one needs models that explicitely capture supersonic turbulence in the ISM, cooling below 300K, and local densities up to 10^6 cm^-3. Oldish SPH models with 100pc smoothing length cannot show this.

    This effect is in fact the reserve of the morphological quenching effect, which corresponds to damped density fluctuations because of a low mach number, and which for the same reasons can be seen only by simulations modelling a cold turbulent phase and capable of resolving a large density range.

    To summarize:: simulations that capture very dense phases do predict a "double" SF law (at fixed local SF efficiency at the parsec scale). They do not assume anything on X_CO to predict this. But they offer a likely (although qualitative) explanation to a lower X_CO in mergers, at the same time, making the whole thing very consistent with Emanuele's interpretation.

    Thus, it is not clear at all that, physically, the adoption of a lower Xco for SMGs as opposed to BzK's is driving the bulk of the shift.


To post without an account, please select 'Name/URL' from the drop down box, enter your name and click 'continue'. Then write your comment and click 'Post Comment'. If you have any problems, please contact jeyhan [at] noao [dot] edu.