Notes taken by FL


  • Reservoir of energy : gravity and nuclear processes.
  • Interact at the scale of SF and starburst
  • Explain the low-rate of SF observed (low compared to what is observed in numerical simulations)
  • Feedback is invoked to explain why the many very low-mass and very massive galaxies formed in simulations are not observed. What is it ?
  • Peak of SF around z=2
  • Many knobs to turn in simulations (positive and neative feedbacks at stellar and AGN ends) to meet the observations…
  • Interaction of cold streams accreting onto pro-galactic disk shoudl be soruce of turbulence and shocks. Maby too tenuous gas for observations ?
  • Absorption spectroscopy of CH+ line in M82 against continuum of local ULIRGs
  • Strongly lensed submm high-z galaxies can be used to serve as lighthouses to probe the turbulent gas in absorption spectroscopy.
  • CH+ formation endothermic, very easily destroyed, so lifetime is 1 year / fH2 / n50 : it shines where it forms
  • ISM is turbulent : very large Reynolds numbers
  • Energy injection at large scales, destabilisation over a dynamical timescale L/v, cascade in k space to dissipation when advection and diffusion become similar in magnitude.
  • Dissipation scale propto n^(-3/4) while mean free path is propto 1/n and the two are about the same for the diffuse ISM (100 cm-3)
  • Bulk of the motion is in solenoidal modes, not compressive. More compressive in SF regions. Observations of Pety et al. and also simulations by PL.
  • Simulations of PL : half of dissipation is concentrated in 10% of the volume. Sign of intermittency.
  • Expriments of Douady 1991 : filamentary structures of dissipation, living for a turnover time scale of the large scales (rotating disks at top and bottom of the cylinder) so very long if compared to what is expected at their size.
  • Observations of CH+ absorption in high-z lensed starburst galaxies, but also very broad emission lines. Much broader than the CO line emission in the same objects.
  • Description of tasks : CH+ at high-z, turbulence in the local ISM, turbulent magnetic field using the Planck data, the LOFAR conection (how thermal electrons and dust are entwined)
  • Starburst galaxies lie above the main sequence of galaxies SFR vs stellar mass. About one order of magnitude above the MS.
  • 14 observations, 3 have no CH+ line. Those three are known to host AGNs.


  • Critical density at which emission starts dominating absorption in CH+ is about 1e4 cm-3. Critical density C21/A21 of CH+ is 1e7
  • Broad emission must come from the centre of the source. Absorption is much narrower.
  • Suprise is that the emission is much broader than in CO and H2O.
  • Other hydrides have similar cirticial densities (such as OH)
  • CH+ in emission requires nH>1e4 and T>1000 K
  • Eyelash : Source 1kpc with SFR 1000 Msun/yr so 1e4 to 1e5 times the UV flux of the MW. Heats the gas through photoelectric effect (1% of UV energy transferred).
  • PDRs would be everywhere and following the kinematics of the source : but why would you observe CH+ and not CO or H2O ?
  • Shocks fed by Galactic outflows ? Outflow interacting with environment, creates turbulent cascade, very many low velocity structures. OK to form molecules (not possible in high velocity shocks). Could be galacitc-scale wind related to high-pressure from combination of SN remnants.
  • Does not explain why we do not observe CO and H2O.
  • Irradiated shocks ? That would explain why CO and H2O are not observed. Models show that CH+ is on the contrary enhanced by an increase in G0. Also SH+, but for a different reason.
  • Observational diagnostics.
  • PDR physics : H2 and CO are dissociated only through lines, while C/C+ is in the continuum. H2 therefore self-shields.
  • Formation of a shock front. Distribution of the energy of a particle going through the discontinuity. Conversion from kinetic to thermal (e.g. viscosity) to internal (rovibration levels) to radiation
  • In the presence of a B field (depending on its strength and ionization fraction), some information is sent in advance because magneto-sonic speed in ions is larger than in neutrals, which irself is larger than the sonic speed in neutrals. If shock speed in between the two magneto-sonic speeds, decoupling the two species with a magnetic precursor.
  • Goes naturally from J shock to CJ shock to C shock at steady state. Ions are slowed down ahead of the shock, transferring some of this information to the neutrals through drag.
  • Increase of B → increase of width of magnetic precursor → neutrals slowed down more → ends in a C-type shock.
  • C-type shocks reach much lower temperatures than J-type shocks, so tracers may be different
  • Model of irradiated shocks : shock + PDR, in steady state, with B perpendicular to shock propagation.
  • Paris-Durham shock code version Lesaffre et al. (2013) includes irradiation with low G0.
  • Some reactions have a strong influence on the dynamics of the shock (e.g. dissociative recombination). Dynamical reduction of the network ? What is the effect of uncertainties in parameters (rates) ? Bistability ? Chaos ? Our experience is that it is quite robust.
  • Added radiative transfer in the UV, H2 self-shielding, dust ionization, possibility to treat C* and CJ shocks.
  • Chernoff (1987), Roberge & Draine (1989) description of trajectories of particles (ions and neutrals) across a shock. Forbidden regions of phase space in v_i and v_n. Trajectory can remain supersonic region (C-type shock). Other trajectories require a jump and just one can meet the condition to arrive at the trajectory going continuously through C’ (CJ shock). C* shocks are continuously going through C and C’ points. Do we actually necessarily reach a steady state ? Also there are steady state shock solutions that are actually unstable. Many trajectories go through C, but only one then goes through C’. Shooting technique varying slightly the position around C, exhibits the one trajectory going to C’.
  • End up with a phase space diagram of the different types of shocks as a function of nH, B0, G0, Av (« initial conditions »)
  • Increase of G0 increases ionization fraction and couplign of neutrals and ions. Reduce the size of the shock, increases the temperature, so does the sound speed, and some parts of the trajectory becomes subsonic. Numerical integration DVODE.
  • Drastic increase of CH+ abundance for irradiated shocks.
  • Should be able to observe CH+ ladder, with a ratio of 2-1 to 1-0 and 3-2 to 1-0 depending on shock velocity.
  • Excitation ladder of H2 has signatures of both C and J type shocks. In irradiated models, H2 intensity approximately constant whatever G0. The shock adapts to the type that allows for evacuating the kinetic energy through H2 radiation, because H2 is main coolant. But excitation diagram changes dramatically. Observing H2 will give some clues on shocks, but quite degenerate.


  • Self-irradiated shocks : shocks hot enough to self-irradiate.
  • Could be the population of shocks out of the cascade from a high-velocity outflow
  • AGN-type of outflow vs Galactic wind from more diffuse sources (combined SN explosions).
  • M82 : IR emission from dust at the edge of the outflow seen in X-ray (hot gas)
  • 500-1000 km/s shock at large scale is in part raiated in the X-ray, but also in part must cascade down to smaller scale lower velocity shocks, where energy radiates in H2 cooling lines.
  • Odd presence of H2 gas colocated with X-ray : we do not expect H2 to survive in such hot environments.
  • Distribution of shock velocities to model the cascade.
  • Cooling time overall several 100 Myr, much longer than the cycle WIM → WNM → Warm H2 → Cold H2
  • Mass flow rate from WIM to WNM much smaller (100 Msun/yr) than warm H2 → cold H2 (2e4 Msun/yr) so need for a process to replenish Warm H2 from cold H2. MHD shocks.
  • Fitting H2 rotational data has quite a strong degeneracy (2 gaussians at 5 and 20 km/s or a piecewise exponential both fit correctly)
  • « Slow shocks » are the case of B inclined with respect to the shock interface. « Fast shocks » have B parallel to interface. Lehmann et al. (2016) searches for distribution of shocks in a turbulent MC simulation. Velocities range to 5 km/s for slow shocks, 12 km/s for fast shocks.
  • Need to go to to high-J CO lines to disentangle fast and slow shocks.
  • At these speeds, temperatures reached are of 2e4 to 1.2e5 K, excitation of electronic levels of H
  • At 30 km/s Ly alpha radiation from these is comparable to ISRF. Then you have to take this into account (« like a PDR in the middle of your shock »)
  • UV photons can travel ahead of the shock, so quite complicated, needs to be solved iteratively. UV heats dust and gas ahead, and so changes « initial » state…. But not yet done.
  • LVG approximation not valid because the postshock region is not showing large gradients.
  • Implemented ALI+Ng for radiative transfer.


  • MISTy questions : Origin of molecules in dilute and violent media (CO in diffuse irradiated media, warm H2, CH+) ? Origin of the clumpy structure of the cold ISM (fragmentation) ? Origin and structure of the B field (link with matter) ?
  • Are these questions related to MHD turbulence dissipation ?
  • Large-scale energy transferred to small scales by turbulent cascade
  • Also cooling cascade leading to phase separation : how is kinetic energy shared between resulting phases ? What are the sizes of clumps ? How long does it take to form ?
  • Dissipation at small scales occurs in bursts : intermittency
  • from diffuse to stars : 30 orders of magnitude in density, but only two in velocity dispersion… Unclear what is meant here by the velocity dispersion (thermal ? macroscopic ?)
  • Numerical simulations by PL solve isothermal MHD equations.
  • Three parameters : Mach, viscosity, resistivity
  • Numerical resolution on a grid leads to extra dissipation. Need to increase a lot the physical ones to make them stand out.
  • Types of dissipation : Ohmic, viscous and ambipolar
  • Initial conditions such that RMS comparable to Alfven speed
  • Dissipation distribution log-normal
  • Dissipative structures appear as sheets
  • Have not looked at how these evolve in time.
  • Presumably, where there is dissipation, there is heating and this would be where molecules are produced and observed.
  • One dissipative structure is quite clearly either viscous or Ohmic in incompressible simulations, but more mixed in compressible simulations
  • B field preferentially parallel to dissipative structure in 3D
  • Distribution of dissipation vs convergence shows two branches, maybe more.
  • Computation of the norm of the gradient of the various quantities combined (rho, u, B). Local diagonalization, find the principal axes. Where gradient is large, one direction dominates the others, so isocontours close to plane-parallel. Can work in 1D. Looking for ways to decompose gradients as a propagating waves.
  • Paint the simulations with chemistry ? If shocks are identified, use the results of Andrew and BG to put chemistry.


  • CO factories in molecular clouds
  • Gould Belt clouds in the Solar neighborhood
  • 12CO(1-0) more diffuse than 13CO(1-0), which is closer to the morphology of dust emission (SPIRE 500µ)
  • If you stack spectra, even where you do not « see » CO, there appears CO after stacking (only was below noise level)
  • Below 1e15 or a bit above, there is CO, but not seen in individual pixels
  • Some chemical aspects : appreciable fraction (2/3) of hydrogen is atomic / Ortho-to-Para ratio / 1/3 of the oxygen not accounted for.
  • Formation of MC ? Through thermal instabilities ? Lifetime 10-20 Myr ?
  • Regulation of star formation : turbulence, magnetic field support and ambipolar diffusion
  • Herchel revealed the existence of very diffuse filaments
  • Differences between Taurus and Polaris : Polaris modre diffuse, less massive, more turbulent, not forming stars.
  • Observations of 12CO(2-1) : variations will depend on chemistry (existence of CO), excitation (to levels >0), and radiative transfer (observation of photons)
  • Idea of explaining the large velocity gradient : two large-scale HI+H2 flows counter-rotaing, where they meet, CO formation, then downstream CO-rich gas.
  • Structure function (computed on the centroid velocity) scaling very close to SL91 (incompressible turbulence)
  • Intermittency (large deviations) show up at high p, but careful that the higher p, the higher weight is given to the noise.
  • Taurus structure function scaling in between K41 and SL91, but this cannot be true (not incompressible, and there is intermittency)
  • Maps of centroid velocity increment extrema highly non-Gaussian (filamentary)


  • Power spectrum does not encode all the information in a given field, when it has non-Gaussian properties.
  • WST : scattering coefficients S0, S1, S2 : about 1000 coefficients. Based on translation and small deformations.
  • A second reduction is related to physics (isotropy, rotations) : ended up with about 70 coefficients
  • Power spectrum does not care whether a scale appears in conjunction with another or not.


  • Dynamical evolution of matter in the ISM is accompanied by a chemical evolution, and chemistry has in turn an impact on the dynamics (cooling)
  • Interest in the diffuse ISM : partly ionized, partly molecular, turbulent
  • Equipartition of energies
  • Theoretical approaches : dynamical simulations / PDR models and TDR models for chemical evolution. Scales to model : from large scales 50-100 pc to dissipation scales (not resolvable)
  • Numerical simulations : dissipation is numerical
  • Astrochemical models
  • Out-of-equilibrium chemically or thermally. Timescales for equilibria can be very different. Usually assuming chemical steady-state, driven by kinetics, not by thermodynamics.
  • Focus on H2 chemistry : formation time is particularly long, influence on dynamics through heating (exothermic formation) and cooling (lines, also from C+, O, Lyman alpha)
  • Photodissociation of H2 0.4 eV
  • Heating and cooling curve : net loss L=n^2\lambda(T)-n\Gamma —> thermal instability curve, depending on mostly G0 and abundances of PAHs
  • Simulations : perdiodic boundary conditions, isotropic turbulent forcing in Fourier space, no thermal conduction, no gravity. H2 formation and destruction computed on the fly.
  • Two fluids (H and H2) plus a prescription for C+, O, for cooling.
  • H2 treated out of equilibrium because timescale for reaching steady-state for H2 is long and determines the timescales for other species.
  • Large fraction of the gas is out of equilibrium thermally
  • N(H2) vs N_H shows bimodal distribution : transition H → H2 with self-shielding of H2 leas to phase transition
  • Different setups in density, G0, forcing amplitude, B field, resolution and box size.
  • Separation of voxels in <300K, 300K-3000K,>3000K
  • Influence of initial density : the larger, the more CNM in the « final » state.
  • G0 increase leads to decrease of f(H2) by increased photodissociation
  • Turbulence increase pushes more gas away from the thermal equilibrium curve.
  • Increase of B prevents the formation of dense structures.

Afterthoughts by BG

Starting with the idea that the ISM is a multiphase turbulent medium, several observations may help us to understand

  • Pierre's shear
  • Edith's starburst

I'm no observer so I'll just throw ideas even if they are ridiculous or not completely correct

• To trace the multiphase ISM

  1. - HI maps (mass of WNM+CNM)
  2. - ArH+ (tracer of purely atomic gas → even better tracer than HI)
  3. - CII and OI fine structure (set the global thermal balance of CNM → mass of CNM)
  4. - CII and OI metastable lines (trace the mass of WNM)
  5. - CI fine structure lines
  6. → gives a measure of the gas thermal pressure
  7. → may give the mass of unstable gas
  8. → its relation with CO gives strong constrain on chemical models
  9. - CS, C2H, OH, and H2O (tracers of PDR in diffuse ISM)
  10. - HF (tracer of H2)
  11. - CH (tracer of PDR, except maybe at small column densities)
  12. - OH+ and H2O+ (tracers of CR ionization and molecular fraction)

• To trace the dissipation of turbulent energy at small scales and / or the turbulent mixing of phases at all scales

  1. - CH+ and SH+
  2. - excited H2
  3. - HCO+ and CO
  4. - CH (anomaly with PDR predictions at small column densities)
  5. - SH (not sure)

• For the DENSE GAS in starburst galaxies, we could try

  1. - SH+
  2. - the rotational diagram of CH+ (as high as possible) → this would be new by the way
  3. - CO, OH or H2O which behave oppositely to CH+ in irradiated shocks.
  • ohp_meeting_september_2018.txt
  • Last modified: 2018/09/22 22:17
  • by admin-mist