Gas Disk Migration - UCF Planetary Sciences Group · Gas Disk Migration Phil Armitage (Colorado)...
Transcript of Gas Disk Migration - UCF Planetary Sciences Group · Gas Disk Migration Phil Armitage (Colorado)...
Gas Disk Migration
Phil Armitage (Colorado)
• Regimes of gas disk migration (review)
• 2 emerging areas where migration may require major
changes to understanding of planet formation
- new work on “Type I” migration [core formation]
- stochastic migration [planetesimal formation]
Recent references:
- Paardekooper et al. (2010, MNRAS in press)
- Ida, Guillot & Morbidelli (2008)
What is gas disk migration?
Planet within gas disk exchanges angular momentum with
the gas due to gravitational torques
Origin of the torque can be understood semi-quantitatively
using impulse approximation (Lin & Papaloizou 1979):
r
b
gas
relative
velocity v
v
2GM
bv
v
1
2v
2GM
bv
2
Gas exterior to the planet has gained specific angular
momentum:
j 2G
2M
2a
b2v
3
Integrate over the disk outside the planet from (a+bmin)
to large distance, get total (exterior) torque on planet:
dJ
dt
8
27
G2M
2a
2b
min
3
…scaling with the square of the planet mass
Numerically, for an Earth mass at 5 AU with = 102 g cm-2,
bmin = h, migration rate due to one sided torque:
J
dJ / dt~ 1 Myr
Elementary argument: unless torques unexpectedly
cancel to high precision, angular momentum transfer
to the gas will lead to significant orbital migration
More rigorous analysis compute the torque as the sum of
partial torques exerted at resonant locations in the disk:
• interior and exterior Lindblad resonances, where
waves are excited
• corotation resonances (co-orbital for circular planet)
Goldreich & Tremaine (1979)… through to Tanaka, Takeuchi
& Ward (2002)… obtain qualitatively same result
Regimes of gas disk migration
For a low mass planet, torques do not affect disk structure
significantly: planet remains embedded within gas
For a high mass planet, torques repel gas from near the
planet, forming a gap in the disk surface density
QuickTime™ and a
YUV420 codec decompressor
are needed to see this picture.
Dividing line depends upon the
angular momentum transport
efficiency and vertical thickness
of the disk: normally a Saturn
mass planet suffice to open a
partial gap…
Type 1 migration
Mp = 1 - 10 Mearth, planet remains fully
embedded in gas
Said to undergo “Type 1” migration
Interaction is weak, standard estimate is linear calculation of
torque at Lindblad + co-orbital corotation resonances
Tanaka, Takeuchi & Ward (2002) find that migration in this
regime is always inward, irrespective of surface density
gradients in the disk.
Numerical simulations of the
same model (isothermal disk
with given structure) agree
with analytic prediction
(Bate et al. 2003)
For any reasonable disk
model, migration of giant
planet cores is very rapid,
faster than their formation
time scale!!!
Critical problem for giant planet formation
models: standard “solution” is to either ignore
migration or to scale the rate down by a large
(~102) and arbitrary factor…
Type 2 migration
Once gap open, planet migrates at rate
that gas tries to flow back into the gap
Orbital evolution occurs on the viscous
time scale of the disk
Alexander & Armitage (2009)
This is “good” - standard
way to end up with hot
Jupiters (Lin et al. 1996)
starting from formation
sites at several AU
Still no direct evidence
for this process
New ideas on Type 1 migration
Main uncertainty has always been computation of the torque
due to co-orbital material
Ward (1991) proposed a
different way to compute
this torque not based on
linear response of the gas
at corotation
Idea is to consider torque
from gas in closed region
where there are horseshoe
orbits
As gas executes horseshoe turns,
changes in density will affect the
torque on the planet
Magnitude of the instantaneous
torque will vary depending upon the
disk’s thermal and structural
properties, e.g.:
Barotropic: p = p()
D
Dt
v
0
depends on gradient
of vortensity in the disk
Adiabatic:
D
Dt
p
0
entropy is conserved along
streamlines, but may vary if
the gas can cool on time scale
of horseshoe turns
On longer time scale viscosity also matters: horseshoe
region is closed so unless viscosity couples to rest of disk
torque must vanish at sufficiently late times
Paardekooper et al. (2010)
If the torque is unsaturated:
2.5 1.7 0.1
1.13
2
7.9
A strong entropy gradient
(declining temperature or
positive ) means the
horseshoe drag beats
Lindblad torque and leads
to outward Type 1 migration
All still quite preliminary - still need to understand better 3D
effects, whether torque remains unsaturated in turbulent disk
BUT does not support the idea that Type 1 migration rate
for ~10 MEarth giant planet cores is slow… adding two large
contributions of opposite sign does not generally give zero!
• cores may migrate outward (rapidly) in optically thick,
almost adiabatic inner disk
• cores migrate inward (fast) in almost isothermal outer
disk, as in “classical” Type 1 migration
Intermediate zone where migration is slow, whose radius
evolves slowly as disk is depleted - perhaps cores accumulate
and stall there to form multiple giant planets?
Stochastic migration
Recall: dJ / dt ~ M2, so Type 1 ~ 1 / M
Negligible angular momentum transfer between very low
mass bodies (“planetesimals”) and a laminar gas disk
QuickTime™ and a
YUV420 codec decompressor
are needed to see this picture.
In a turbulent disk, stochastic
forces from transient surface
density perturbations can:
• excite random motions
of all bodies, even mass
less tracers
• result in random-walk
migration
Nelson & Papaloizou (2004);
Laughlin et al. (2004)
Why does this matter for planetesimal formation?
Conventionally, e / i of planetesimals is set by balance
between:
• excitation by planetesimal / planetesimal scattering
• damping by aerodynamic gas drag
Leads to low equilibrium
eccentricities
Collisions have specific energy
too low to fragment targets, even
km scale planetesimals would
experience runaway growth
Much stronger excitation by
gravitational coupling to
turbulence can change this
picture
Ida et al. (2008)
Using best estimates of the
properties of disks in which
magnetorotational instability
generates turbulence…
Find accretion occurs only
for ~100 km scales and
above, collisions in the
outer disk can be highly
erosive
Preliminary: suggests connections between astrophysics,
planetesimal formation, Solar System planetary science
Astrophysics: can we calculate the strength of disk
turbulence from first principles? Need simulations with
non-ideal MHD (Ohmic + Hall terms)…
Planetesimal formation: if 1 km planetesimals are
destroyed in collisions, must form planetesimals directly
at larger sizes - gravitational collapse. Streaming
instabilities, collection of particles in turbulence etc…
Solar System constraints: is there evidence for large
primordial planetesimals in the size distribution of
asteroids (Morbidelli et al. 2009)? Stochastic migration
would also erase compositional gradients - is that
consistent with observations?