The Early Evolution of Protostars
Neal J. Evans II
with much help from
Mike Dunham, Lori Allen, Melissa Enoch,
Jeong-Eun Lee, Joel Green, Hyo-Jeong Kim
Questions





How long do various stages of the process take?
Do any theories explain the data?
How are the star and disk built over time?
Have we found the “missing link”?
What can astrochemistry tell us about
evolution?
 What are we seeing in early Herschel
spectroscopy?
What We Need
 The key is to have a large, uniform sample
 “Blind” surveys at range of wavelengths
 Complete coverage of the SED





Millimeter wave (mass and structure)
Far-infrared (energy for embedded stages)
Mid-infrared (disks)
Near-infrared (inner disk and star)
Visible, UV, X-ray (star and accretion)
 Spectroscopic diagnostics to follow up
Some Star Formation Surveys
for Low-mass Stars
 Taurus Legacy project
 Nearly complete survey of Taurus
 Cores to Disks (c2d) Legacy Project
 Surveys of 7 nearby “large” clouds and many small ones
 Complementary molecular line and dust continuum maps
 Gould Belt Legacy Project
 Surveys of 13 nearby “large” clouds to complete census
 Herschel Surveys
 Gould Belt Herschel Survey
 Dust, Ice, and Gas In Time (DIGIT)
 Water In Star-forming regions with Herschel (WISH)
 JCMT Gould Belt survey (SCUBA2 and lines)
Surveys of Nearby Clouds and
Clusters
20 nearby molecular
clouds (blue circles)
35 young stellar clusters
(red circles)
90% of known stellar
groups and clusters
within 1 kpc
(complete to ~ 0.1 MSun)
+ Several massive sf
complexes at 2-3 kpc
(complete to ~1.0 MSun)
5
Questions





How long do various stages of the process take?
Do any theories explain the data?
How are the star and disk built over time?
Have we found the “missing link”?
What can astrochemistry tell us about
evolution?
 What are we seeing in early Herschel
spectroscopy?
Evolution
 Various Stages in the evolution
 Associated with Classes based on SED
 Durations in Classes inferred from
numbers
 Previous studies based on small numbers
 Typically 50 to 100 objects
 Fewer in early classes
 Estimates of durations differed by large
factors
Standard Evolutionary Scenario
Single isolated low-mass star
n~104-105
outflow
n~105-108 cm-3
T~10-300 K
cm-3
Stages
T~10 K
infall
Factor 1000
smaller
Classes
Core collapse
Protostar with disk
t=0
Class I
Class 0
Formation planets
t=105 yr (?)
t=106-107 yr
Solar system
t>108 yr
8
Note axis change!
Stages
Scenario for star- and planet formation
outflow
Formation planets
Classes
Cloud collapse
t=106-107 yr
t=0
Solar system
t>108 yr (?)
Protostar with disk
t=105 yr
Disk
Class II
Formation planets
Star
t=106-107 yr
Class III
Solar system
t>108 yr
9
Spitzer probes dust at temperatures between 100 and 1500 K.
20 clouds, 3124 YSO
c2d + GB
I
F
III
Class I lasts 0.54 Myr
Flat lasts 0.35 Myr
(longer than most previous
estimates)
II
I:
Flat:
II:
 III:

  0.3
0.3   0.3
1.6    0.3
  1.6
IF Time is the only variable
AND
IF star formation continuous
for t > t(II)
AND
IF Class II lasts 2 Myr,
THEN
14%
9%
47%
30%
Caveats:
Class III census incomplete
Class III not included in timescale
Depends on how  is calculated
Class 0 mixed with Class I
Timescales for earlier classes
 For 3 clouds with millimeter maps (Enoch08)




Use Tbol to separate Class 0 from I
Absence of IR source to say starless
Largest sample, B.H.: 200 cores
N(0) = 0.44N(I), so t(0) = 0.16 Myr
 Not consistent with fast, early infall (Andre et al.)
 Except Oph: 0.04 Myr, Oph was basis of low t(0)
 Oph has faster evolution or not continuous
 N(SL) = 0.8 N(0+I), so t(SL) ~ 0.43 Myr
 After <n> > 2 x 104 cm–3
 t(SL) ~ 3 tff; between predictions of fast and slow
Starless Core Lifetimes
Enoch et al. 2008
Questions





How long do various stages of the process take?
Do any theories explain the data?
How are the star and disk built over time?
Have we found the “missing link”?
What can astrochemistry tell us about
evolution?
 What are we seeing in early Herschel
spectroscopy?
Comparison to Shu model
 Assume inside-out collapse at 0.19 km/s
 Sound speed at 10 K
 In 0.54/2 Myr, rinf = 0.054 pc
 Consistent with some sizes
 Mean separation in clusters 0.072 pc (Gutermuth)
 At dM/dt = 1.6 x 10–6 Msun/yr, M* ~ f 0.86 Msun
 If f ~ 0.3, get 0.26 Msun ~ modal mass
 Consistent with assumptions, most data
 Picture holds together, except…
The Luminosity Problem!
M. M. Dunham et al. 2010
Most are under-luminous
Predicted L = GM(dM/dt)/R= 1.6 Lsun for standard (Shu) accretion onto
M = 0.08 Msun, R = 3 Rsun. Most (59%) are below this. M. M. Dunham et al. 2010
Questions





How long do various stages of the process take?
Do any theories explain the data?
How are the star and disk built over time?
Have we found the “missing link”?
What can astrochemistry tell us about
evolution?
 What are we seeing in early Herschel
spectroscopy?
Episodic Accretion
 Accretion from envelope to disk is not
synchronized with disk to star
 Kenyon and Hartmann (1995) suggested
this to solve luminosity problem (IRAS)
 Exacerbated by Spitzer data
 Simulations show it (Vorobyov and Basu
2005, 2006)
Direct Evidence for Episodic
Accretion
 Luminosity Variations (e.g., FU Orionis)
 VeLLOs (L<0.1 Lsun), much less than
prediction for standard accretion onto BD/star
 Outflow morphologies suggesting multiple
ejection events (e.g., HH 211)
 Comparison of L(now) with <L(t)>
 Outflows trace history of ejection, hence accretion
 Careful analysis of several sources gives strong
evidence for L(now) < <L(t)>
 Dunham et al. 2006, 2010
Episodic Jet
HH 211 Jet shows
series of bow
shocks. Time
between estimated
at 15-44 yr
Lee et al. 2007
A Toy Model of L(t)
1 Msun
3 Msun
M. M. Dunham et al. 2010
Assume no accretion most
of the time. Mass collects
in disk. When Md/M* = 0.2,
accretion at
dMacc/dt = 10–4 Msun/yr
until Md = 0. Then the disk
rebuilds from ongoing
infall.
Extreme version as proof
of concept.
With 2D, Outflows, Episodic
Accretion
M. M. Dunham et al. 2010
and the BLT
M. M. Dunham et al. 2010
Consequences of Episodicity
 The connection between Classes and Stages
becomes tenuous
 The luminosity is not an indicator of stellar
mass until nuclear burning dominates (Lacc ~
M*dMacc/dt)
 Stellar ages from tracks may be way off
(Baraffe et al. 2009)
 The initial conditions for planet formation may
be determined by time since last episode of disk
instability
Questions





How long do various stages of the process take?
Do any theories explain the data?
How are the star and disk built over time?
Have we found the “missing link”?
What can astrochemistry tell us about
evolution?
 What are we seeing in early Herschel
spectroscopy?
The First Hydrostatic Core
(Stage -1?)
 Long predicted phase of star formation
 Larson (1969)
 The FHSC is an H2 core
 Contracts slowly until H2 dissociates (2000K)
 Then the second (protostellar) core forms
 Has never been seen
 Short duration, very low luminosity
Theory
 Boss and Yorke (1995) predicted SED
 Distinguished from prestellar core by slight
excess in FIR (L < 0.1 Lsun)
 Omukai (2007) lifetime is short
 103 to 3 x 104 yr
 Expect one per 540 to one per 18 Class 0/I
 Zero to 8 in c2d sample, Zero to 23 in GB
First Core in Theory
First Core 500 yr after
formation.
“Fast flow” (2km/s)
driven by magnetic
pressure (weak fields)
“Slow flow” driven by
magneto-centrifugal
force (strong fields)
carries 10x more
mass and ang. mom.
Tomida et al. 2010
Standard Evolutionary Scenario
Single isolated low-mass star
n~104-105
outflow
n~105-108 cm-3
T~10-300 K
cm-3
Stages
T~10 K
infall
Factor 1000
smaller
Classes
Core collapse
Protostar with disk
t=0
Class I
Class 0
Formation planets
t=105 yr (?)
t=106-107 yr
Solar system
t>108 yr
29
Note axis change!
Candidates
 Chen et al. (2010) astro-ph 1004-2443
 L1448 IRS2E
 But faster outflow than expected
 Enoch et al. (2010)
 Per-Bolo 58, NE of NGC1333
 On astro-ph 1009.0536
L1448 IRAS2E
Detected by SCUBA
and Bolocam surveys
Not detected by c2d
Spitzer observations,
but a jet structure at
5.8 micron.
Detected by SMA
in continuum and in
redshifted CO lobe
(up to 25 km/s)
X. Chen et al. 2010
Per Bolo-58
Enoch et al. (2010)
The SED fits
Enoch et al. 2010
Questions





How long do various stages of the process take?
Do any theories explain the data?
How are the star and disk built over time?
Have we found the “missing link”?
What can astrochemistry tell us about
evolution?
 What are we seeing in early Herschel
spectroscopy?
Dramatic Changes during SF





Evolution of Dust (bigger, icier, …)
Depletion of molecules from gas onto dust
Sublimation when heated
Repeated “freeze-thaw” cycles if episodic
Further processing in outflows
Molecules freeze out as ices
Ice inventory
Boogert et al. 2004,
2007
- Abundances of some species similar within factor of 2 (e.g., CO2)
- Significant variations (>10) for other species (e.g., CH3OH, NH3, OCN-)
- Evidence for NH3 with high abundances (>10%) in some objects
- First detection of CH4 ice toward low-mass YSO’s
Chemical Memory
 Chemical timescales differ from
dynamical timescales
 Desorption of ices, photodissociation, …
essentially instantaneous
 Freeze-out, some chemical reactions
depend on density, can be long
 Chemistry may trace history
Combined Modeling of Dust and Gas
nd(r), L

Radiative
Transfer
TD(r)
Gas to
dust
S
I(b)
Dust
PhysicalModel
Model
Physical
n(r),v(r)
v(r)
n(r),
Iterate
TD(r) to
TK(r)
n(r), T K(r)
v(r)
Simulate
Observations
Observations
Observations
Gas
Chemistry
X(r)
Monte
Carlo
nJ(r)
Simulate
Obs.
TA (v,b)
Chemical Effects of Episodic
Accretion
CO
Blue line shows abundance at
4.1x104 yr, just before third
flare and during flare (red
line).
CO evaporation radius moves
out.
Magenta lines show evolution
every 1000 yr after third flare
up to 5 x 104 yr.
J-E. Lee (2007)
Chemical Effects
N 2H +
Blue line shows abundance
at 4.1x104 yr, just before
third flare and during flare
(red line).
N2H+ destruction radius
moves out.
Magenta lines show
evolution every 1000 yr after
third flare up to 5 x 104 yr.
J-E. Lee (2007)
Applied to CB130-1 IRS1
J, H, K
Kim et al. 2010
CB130 thought to
be starless.
Spitzer revealed
two sources, one a
Class 0 object with
L ~ 0.15 Lsun.
Obtained IRS
spectrum and
extensive mm line
spectroscopy.
SED Modeling
Red line is IRS
spectrum.
Black line is
best fit 2D
model with
envelope and
disk and 50o
orientation
angle. Green
is for 70o.
Note deep
CO2 feature.
Kim et al. 2010
Three models of L(t)
Model 1:
Standard Shu
(VERY young)
Model 2:
Extended FHSC
Model 3:
Episodic
accretion
Kim et al. (2010)
Abundance Profiles: Three
Models
Episodic models have had
6 episodes and 3000
years since the last burst.
Resulting Spectra
Need to convert
CO to CO2 to
match emission
lines. Also
matches the
observed CO2
absorption!
Herschel Early Results
Imaging surveys to complete census of earliest
stages (Monday’s talks)
Spectroscopy of known objects (talks by van
Dishoeck and others)
PACS Spectroscopy is Rich
DK Cha observations
van Kempen et al. (2010)
CO from levels up to J ~30,
E/k ~ 3000 K
Warm and Hot Gas
7 of 30 embedded
targets observed
All show many
lines, similar Trot:
<Twarm> = 390 K
<Thot> = 1380 K
N(warm)/N(hot)
~3-20
OH and H2O at
lower T ~ 160 K
van Kempen et al. (2010)
UV, shocks along outflow walls
Models indicate a
combination of UV
excitation and
shocks along the
outflow lobes.
Results on HH46 from
PACS spectroscopy on
Herschel (WISH)
van Kempen et al. (2010)
Open Questions
 What sets mass of stars/brown dwarfs?
 CMF, or feedback
 Have we found a FHSC?
 Can we get a census?
 Is mass accretion episodic?
 If so, what sets the cycle times?
 Can we constrain this with chemistry?
Summary
 Timescales for Class 0 and I longer
 But Class is not the same as Stage!
 Shu inside-out collapse consistent, except
 Luminosity distribution
 Accretion from disk to star appears to be
episodic
 Complex chemical changes throughout
 May provide constraints on episodic accretion
 Herschel spectra are rich
 CO, OH, H2O reveal information on outflow