On the structure of the neutral atomic medium

1
On the structure of the neutral atomic medium

• Patrick Hennebelle
• Ecole Normale supérieure-Observatoire de Paris
• and
• Edouard Audit
• Commissariat à lénergie atomique

2

• Important Physical Scales and Basic
  Understanding
• One static scale
• Fields length size of the thermal fronts
  connecting cold and warm phases
• In WNM 0.1 pc , in CNM 0.001 pc.
• gtSmallest structures of size 0.001 pc
• gtsmallest column densities 0.001 pc 100 cm-3
  3 1017 cm-2
• Three dynamical scales
• Cooling length of WNM scale at which WNM is
  linearly unstable
• (Hennebelle Pérault 1999, Koyama Inutsuka
  2000)
• Cs,wnm ?cool 10 pc
• Size of CNM fragments cooling length divided by
  phase density contrast (100)
• Cs,wnm ?cool/100 0.1 pc
• Size of shocked CNM framents fragments undergo
  collision at Mach10
• Isothermal shock gt ?shock?cnm102104 cm-3,
  Size 0.1/M2 0.001 pc

3
Numerical
experiments Mandatory resolutions 104 -105
Can be done in 1D and approached in 2D 2D
numerical experiment 104104 pixels, 20 pc size
box, 0.002 pc of resolution 3D numerical
experiment 120012001200 pixels, 15 pc size
box, 0.01 pc of resolution Initial conditions
compromise between large scale (cooling length of
WNM) and small scales (Hennebelle Pérault 1999,
Koyama Inutsuka 2000, 2002, Audit Hennebelle
2005) Impose a converging and turbulent flow
of WNM from left and right face (pm 1.5
Cs,wnm) The flow can leave the box through the
other faces WAIT UNTIL A STATISTICALLY
STATIONARY STATE IS REACHED (takes cpu) Better
forcing can be achieved if larger box are used gt
see Vazquez-Semadenis talk (Vazquez-Semadeni et
al. 2003, Gazol et al. 2000)

4
25002
20 pc
5
104104 pixels
20 pc
6
First Zoom
5 pc
-the flow is very fragmented, the structures are
well defined (Koyama Inutsuka 2002, Audit
Hennebelle 2005, Heitsch et al 2005)
-the structures are in pressure equilibrium with
the surrounding gas 2-phase model
-there are large fluctuations in density and
pressure
7
Second Zoom
Density 104 cm-3 Pressure 105 K cm-3 Size 400
AU- TSAS ?
0.2 pc
Converging flow
8
Extracting the individual structures
(achieved by simple clipping algorithm,
density gt 30 cm-3)
Mass powerspectrum of structures (weighted in
mass)
The mass is equally distributed from the largest
structures down to 100-1000 times smaller
structures
Even with higher resolution no strict numerical
convergence  Kind  of convergence is reached
for dx lt 0.01 pc
9
Fluid statistics density and pressure PDF
Pressure PDF
N50005000
Density PDF
N1000010000
gtsignificant fraction of the gas at high density
(2) but depends on numerical resolution (and
on thermodynamics)
10
Energy spectrum (2D)
Fluid statistic velocity powerspectrum and
energy spectrum

• Energy spectrum flat (k-1) gt energy equally
  distributed in k-space

Velocity powerspectrum (2D)
11
Energy spectrum (2D simulations)
gtFlat energy spectrum due to flat density
spectrum
density spectrum (2D simulations)
12
3D simulations 12003
50,000 cpu hours
13
(No Transcript)
14
Energy spectrum (3D simulations)
density spectrum (3D simulations)
gtBehaviour  seems  similar to 2D but
resolution is crude
Velocity powerspectrum (3D simulations)
15
(No Transcript)
16

• Conclusions
• -small scale structures are produced naturally
• -large density and pressure fluctuations (TSAS ?)
• Natural consequences of 2-phase fluid
• (ratio of sound speeds in the 2 phases 10)
• -there is a kind of  duality , coexistence
  between discrete structures (2-phase models) and
  classical turbulent behaviour
• -energy seems to  pile  up at small scales
  cascade a priori different from Kolmogorov. Bulk
  energy of cnm not easily removed.

17

Questions Are the molecular clouds multiphase
objects as well ? Indeed molecular
clouds -form by contraction of HI which is a
2-phase medium -are turbulent meaning than there
is a mixing with the surrounding HI halo The key
question seems to be Can WNM survive at high
pressure ? gtIs there a heating mechanism more
efficient than UV ? Hennebelle Inutsuka
(2006, ApJ in press) propose Heating due to
dissipation of MHD waves by ambipolar diffusion
(proposed by Falgarone Puget 86 and Ferrière
et al. 87 in different context)
18
(No Transcript)
19
Influence of thermal conduction ? - standard 
ISM conduction -pas de conduction - pure  WNM 
conduction (no variation with T) -conduction 10
times the ISM conduction
gt Effect of thermal conduction seems to be weak
20

Questions -structure of the flow Which
description turbulence / static 2-phase
description ? -presence of small scale
structure ? Density fluctuations ? Mass
Distribution ? -Turbulent flow
description Powerspectrum ? -Influence of the
numerical resolution. Have we reached convergence
? -Influence of thermal conduction ? -2D
versus 3D