The turbulent life of dust grains in the supernova-driven, multi

The turbulent life of dust grains in the supernova-driven,
multi-phase interstellar medium
Thomas Peters
Max-Planck-Institut für Astrophysik
Svitlana Zhukovska, Thorsten Naab, Philipp Girichidis,
Stefanie Walch, Simon Glover, Richard Wünsch, Ralf Klessen,
Paul Clark, Daniel Seifried, Dominik Derigs
Thomas Peters
The turbulent life of dust grains in the ISM
Introduction
The multi-phase interstellar medium results from the complex
interplay of
gravitational collapse,
radiative cooling,
magnetic fields,
formation and destruction of molecules,
stellar feedback in the form of
ionizing and non-ionizing radiation,
winds and outflows,
supernova explosions.
Thomas Peters
The turbulent life of dust grains in the ISM
Introduction
The multi-phase interstellar medium results from the complex
interplay of
gravitational collapse,
radiative cooling,
magnetic fields,
formation and destruction of molecules,
stellar feedback in the form of
ionizing and non-ionizing radiation,
winds and outflows,
supernova explosions.
We use idealized simulations of the supernova-driven, multi-phase
ISM to study the life cycle of dust grains.
Thomas Peters
The turbulent life of dust grains in the ISM
SImulating the LifeCycle of molecular Clouds (SILCC)
Gatto+’15,’16, Walch+’15, Girichidis+’16a,b, Peters+’15
Thomas Peters
The turbulent life of dust grains in the ISM
Simulation setup
box 0.5 kpc×0.5 kpc×2.5 kpc at 3.9 pc resolution
Σgas = 10 M pc−2 (solar neighborhood)
SFR from Kennicutt-Schmidt relation
assume constant supernova rate (15 SNe per Myr)
inject 1051 erg thermal energy per supernova
how to choose the locations of SN explosions?
stars form in molecular clouds (dense gas)
but: delay between SF and SN, early feedback creates cavities
contributions from field and runaway stars
random distribution of SN locations:
uniform in disk plane, Gaussian in vertical direction
Walch+’15, MNRAS 454, 238: chemistry
Girichidis+’16, MNRAS 456, 3432: dynamics
Thomas Peters
The turbulent life of dust grains in the ISM
Lagrangian tracer particles
t = 0.0 Myr
face-on
t = 0.0 Myr
0.2
edge-on
100
0.2
0.1
z (kpc)
y (kpc)
0.1
0.0
10−2
10−3
0.0
−0.1
−0.1
−0.2
−0.2
10−4
10−5
−0.2
−0.1
0.0
x (kpc)
0.1
0.2
−0.2
−0.1
0.0
x (kpc)
0.1
0.2
10−6
1 million Lagrangian tracer particles are inserted randomly
with a number density proportional to the initial gas density.
During the simulation, information on local density,
temperature, chemical composition, UV field, etc. is recorded
for all particles at 10 kyr time resolution.
Thomas Peters
The turbulent life of dust grains in the ISM
column density (g cm−2)
10−1
Phases of the ISM
107
106
T (K)
105
104
103
102
101
0
10
20
30
40
t (Myr)
50
60
70
80
We consider a four-phase medium defined by temperature cuts:
molecular phase (T < 50 K),
cold phase (50 ≤ T < 300 K),
warm phase (300 ≤ T < 3 × 105 K),
hot phase (T ≥ 3 × 105 K).
Thomas Peters
The turbulent life of dust grains in the ISM
Phase evolution and transition rates
random driving
10
10−6
molecular
cold
warm
hot
10−2
10
Ẋ (yr−1)
fpart
10−1
−5
molecular → cold
cold → warm
molecular → warm
molecular → hot
cold → hot
warm → hot
inverse transitions
10−6
10−7
10−8
10−7
10−9
10−8
−3
10−10
10−9
10−4
0
10
20
30
40
t (Myr)
50
60
70
80
10−10
0
10
20
30
40
t (Myr)
50
60
Molecular phase forms after ∼10 Myr.
High intermittency of transition rates.
Molecular-cold, molecular-warm and cold-warm transitions
dominate.
Neighboring phases are nearly in detailed balance.
Thomas Peters
The turbulent life of dust grains in the ISM
70
10−11
80
Σ̇d (M Gyr−1 pc−2)
random driving
100
Mass transfer scheme from simulation
stellar outflow, SN,
extragalactic material
warm
randomdriving
20
11
1 2
15
1
hot
12
1
140
cold
molecular
129
star formation
Weingartner & Draine ’99 used observed Ti depletion to infer
transition rates of their mass transfer scheme
comparison with averaged transition rates (60 ± 10 Myr)
order of WD ’99 transition rates similar to our values
in addition, we observe intense mass transfer from molecular
to cold phase
Thomas Peters
The turbulent life of dust grains in the ISM
Model predictions for Ti depletion
Phase
warm
cold
molecular
Observed log δj
−1.3
−3.0
—
Random driving
Model A Model B
−1.6
−1.3
−2.8
−2.7
−2.6
−2.4
Use our rates as input to Weingartner & Draine ’99 model
and compute depletion δ (ignoring hot phase).
Compare with observations compiled by Jenkins ’09.
Model A and B differ in grain size distribution and destruction
timescale, has only small impact.
Reasonable agreement with observations in warm and cold
phase, but depletion in molecular phase under-predicted.
Thomas Peters
The turbulent life of dust grains in the ISM
Residence times
random driving
106
105
random driving
1.0
molecular
cold
warm
hot
0.8
104
Φ
Npart
0.6
103
0.4
102
100
0
molecular
cold
warm
hot
0.2
101
10
20
30
40
50
tres (Myr)
60
70
80
0.0
0
10
20
30
40
50
tres (Myr)
60
70
We measure the residence times of all particles at 40 Myr.
Residence times have broad distributions.
Median values are 15, 5, 44 and 2 Myr in the molecular, cold,
warm and hot phase, respectively.
Caution! Residence time in molecular phase is not cloud
lifetime since particles can circulate between several clouds
without leaving molecular phase.
Thomas Peters
The turbulent life of dust grains in the ISM
80
Consequences of broad residence time distributions
0.007
0.006
0.005
D
0.004
0.003
0.002
τcl
τcl
τcl
τcl
0.001
0.000
0
2
4
6
t (Gyr)
8
10
=
=
=
=
10 Myr
20 Myr
30 Myr
50 Myr
12
Zhukovska+’08 model predicts variations of dust-to-gas ratio
at the percent level for different cloud lifetimes.
Broad distributions could contribute to observed scatter of
element depletion.
Thomas Peters
The turbulent life of dust grains in the ISM
Conclusions
First direct measurements of transition rates between and
residence times in ISM phases in a hydrodynamical simulation.
Find more sizeable transitions than considered in the literature.
Averaged transition rates produce good agreement with
observations of Ti depletion in warm and cold phase, but
under-predict depletion in molecular phase.
Residence time distributions are broad, particularly for the
molecular, warm and hot phase.
This contrasts with single values used in simplified models.
Broad residence time distributions could contribute to observed
scatter of element depletion.
Simulations with a more complete and self-consistent treatment of
star formation and stellar feedback are already underway.
Thomas Peters
The turbulent life of dust grains in the ISM