Download ppt - Institute for Astronomy

The Formation of Single and
Multiple Stars
Bo Reipurth
Institute for Astronomy
Astrobiology Seminar
University of Hawaii Manoa
20 September 2004
Overview
•
•
•
•
The Formation of Binary and Multiple Stars
High Velocity Jets
Disk Accretion Events
Clustered Star Formation
Evolution of Young Low Mass Stars
* Class 0 objects are deeply embedded protostars only about 10, 000 yr old
surrounded by large infalling envelopes
* Class I objects are partly embedded protostars less than 100, 000 yr old
surrounded by massive circumstellar disks and a remnant envelope
* Class II objects are T Tauri stars typically a few million years old with disks
* Class III objects are T Tauri stars with less circumstellar material
Historically, there has been
a strong bias towards
looking at young stars as
single stars.
Courtesy of Pat Hartigan
Young Binary Stars: Some History
• The study of young binaries began with Joy & van Biesbrock (1944) who found
5 visual binaries among the newly recognized T Tauri stars.
• In 1962, George Herbig added another 24 T Tauri binaries, demonstrating that
binaries among young stars are not uncommon.
• Observations of young binaries remained decidedly out of main stream
astronomy until 1993, when within one month three major surveys appeared.
• By now, studies of young binaries are leading a paradigm shift in star formation.
IT IS A KEY FACT THAT 2/3 OF ALL STARS ARE BINARIES
The RZ93 Survey - I
•
Reipurth & Zinnecker (1993) [A&A
278, 81] imaged 238 young stars at
0.9 mm and detected 37 binaries with
separations between 1” and 12”.
•
Taking into account distances to the
binaries, this implies a PMS binary
frequency of 16% in the projected
separation range of 150 AU to 1800
AU.
The RZ Survey - II
•
The distribution of separations for
PMS binaries is complete to the
vertical dotted line.
•
The separation distribution function
shows a steep rise towards smaller
separations.
The RZ93 Study - III
•
•
•
•
The left figure shows a comparison between the distribution of PMS binary
separations and the distribution of G dwarf MS binaries from Duquennoy & Mayor
(1991), who found a binary frequency of 53% (curves are shown for 60, 80 and
100%).
The right figure shows the same for K dwarfs studied by Mayor et al. (1992), who
find a binary frequency for these stars of 45%.
An excess of PMS binaries relative to MS binaries is evident in the three bins where
the survey is complete.
At 130 AU the peak of the DM91 separation distribution function of 30 AU is at
0.2”-0.3”.
Binary Formation through Capture
•
Capture was the earliest suggestion for forming a binary.
•
All variations on this theme require some additional medium that can dissipate
excess kinetic energy so that two unbound stars can become bound during a close
passage.
•
This medium can be a third star, a circumstellar disk, or the stars themselves if the
encounter is close enough to raise tides.
•
For field stars this concept is now discredited, but under certain circumstances the
idea is still valid, e.g. in newly formed clusters.
•
Most importantly, if one or both stars have disks, the disk matter exterior to the
minimum separation of the encounter will become unbound and take away the
excess kinetic energy.
Fragmentation and the Formation of Wide Binaries - I
•
Fred Hoyle noted in 1953 that if collapse is isothermal then, as the density of
the cloud is increased, the Jeans Mass [Mj propto T^(1/2) r^(-1/2)] will
decrease, and smaller parts of the cloud will become gravitationally unstable.
•
Hoyle envisioned a repetitive process known as “hierarchical fragmentation”.
•
Although the detailed physics of fragmentation still poses problems,
fragmentation in various forms is still the most widely accepted formation
mechanism for wide binaries.
Fragmentation and the Formation of Wide Binaries - II
Accretion and Binary Evolution
•
The existence of close binaries, e.g. spectroscopic binaries with periods of
days or months, is not easy to explain.
•
Such binaries do not appear to result from our current understanding of star
formation.
•
It is likely that such close binaries evolved from wider pairs through
dynamical evolution.
•
The main driver of this evolution is the circumstellar or circumbinary gas
disks.
Disks in Young Binary Systems
•
Young binary stars show a great diversity
of disk properties, with a large part of the
variation accounted for by binary
separation.
•
Many of the closest binaries (with
separations less than a few AU) could
harbor circumbinary disks.
•
Binaries with separations larger than about
100 AU tend to have massive circumstellar
disks around the individual components.
•
But in the intermediate range, while disks
do exist, they are strongly affected by the
stellar companion, and consequently have
limited mass.
Mass Transfer in Eccentric Binaries
•
It is now recognized that the typical eccentricity of a binary star (whether
young or old) is considerable (e ~ 0.2 – 0.6) and not small or zero as has often
been assumed in early theoretical work.
•
The non-zero eccentricity allows for substantial interactions between the stars
and their circumstellar environments.
•
Simulations of eccentric binaries show that a circumbinary disk can transfer
mass onto the central binary through non-axisymmetric structures (streamers
or gas arms).
•
This resupply mechanism may explain the common presence of small
circumstellar disks which should otherwise have short life-times.
Matthew Bate’s Simulation of An Accreting Binary
QuickTime™ and a
YUV420 codec decompressor
are needed to see this picture.
The Case of DQ Tau - I
•
DQ Tau is a normal classical T Tauri star,
except that it is one of only two CTTS that
are known to be a close double-lined
spectroscopic binary (all other PMS
double-lined spectroscopic binaries are
WTTS).
•
DQ Tau is a short period, eccentric binary
with roughly equal-mass components.
•
The orbital solution shows that the
components are extremely close,
~8
Rsun, at periastron.
The Case of DQ TAu - II
•
For such close separations it is difficult to understand how the components can continue to
accrete, since circumstellar disks should be destroyed.
•
However, light curves and spectral variability studies show that DQ Tau undergoes periodic
accretion events.
•
The star brightens and shows increased Ha emission and veiling with the binary period,
synchronized with periastron of the binary. This is interpreted as gas channeled along the
streamers from a circumbinary disk to the individual components.
HERBIG-HARO JETS AND
MULTIPLE STARS
•
The very youngest stars sometimes
drive powerful jets. Such jets
represent a fossil record of the
accretion events in the source.
•
Examples are HH 111, HH 1, L1551NE, and Haro 6-10.
The Case of HH 111
•
Near-infrared images of the HH 111 source
region shows two sources, A and B. A is
driving the major jet and is heavily
extincted, B shows little extinction.
•
Both sources are detected by the VLA in
the radio continuum, so both are young
and lose mass.
•
Analysis of the brighter source A shows
that it has two radio jets coming out at
large angles to each other. It follows that A
is a close binary, so the whole system is a
hierarchical triple.
The Case of HH 1
•
•
•
•
The HH 1 jet and a neighboring flow HH 144 emanate from
two deeply embedded source radio continuum sources.
One of the two sources is extended along the axis of the HH
1 jet, and it shows very large proper motions.
Optical/infrared HST images have revealed a third tiny jet
flow. Proper motion studies show that it comes from the
extended radio source, indicating that it is a binary.
In other words, the source region contains a hierarchical
triple system.
The Case of L1551-NE
L1551-NE is a deeply embedded
source in a dense dust ridge. It
drives a series of HH knots and a
giant HH flow. In the IR it shows
an illuminated outflow cavity. VLA
observations reveal the source to
be a binary. Mm data have found
a third source nearby
The Case of Haro 6-10
•
Haro 6-10 drives a giant HH flow. Near the
source the flow breaks up into two components
with different directions. It appears as if there is
precession in play.
•
VLA observations have revealed that there are
three sources, forming a not quite hierarchical
triple system.
Disintegration of Multiple Systems
•
Detailed analysis of giant jet energy sources show
that 80-85% are observed to be binaries or
multiples, i.e. without corrections for
incompleteness [Reipurth 2000, AJ 120,3177].
•
Since much fewer stars are binaries and multiples
on the main sequence, it follows that some of them
must disintegrate.
•
Non-hierarchical systems oscillate between two
phases: interplay and close triple approach, and the
latter can lead to ejection.
•
Disks go through serious warping and truncation
during close triple approaches, leading to accretion
events, and thus outflow activity. Giant jets are
presumably formed in this way.
Burrau’s Problem
•
The motion of a non-hierarchical
triple system is chaotic and can
only be studied numerically.
•
In a few simple cases, one can
find an analytical description.
•
Burrau’s problem considers three
objects of 3, 4, and 5 mass units,
with separations of 3, 4, and 5
length units, located in a rightsided triangle.
•
The motion illustrates the three
classes of interplay, close triple
approaches, and ejection
(simulation by Michael Sterzik)
QuickTime™ and a
YUV420 codec decompressor
are needed to see this picture.
The Formation of Brown Dwarfs
•
The decay of a small non-hierarchical
system occurs statistically after 100
crossing times, a time unit that is
determined by the masses and separations
of the components. For newborn systems,
this typically amounts to 10,000 or 20,000
years.
•
At such early times, the stars are still busy
building up their masses.
•
Stars that are slightly more massive will
occupy the center of infall more often, and
thus grow even faster.
•
Such competitive accretion leads to delays
for the smallest objects. If these “runts of
the litter” are ejected before reaching 0.08
Msun, they will forever remain as brown
dwarfs.
[Reipurth & Clarke AJ, 122, 432, 2001].
The Bate, Bonnell, Bromm 2003 Simulation - 1
•
Matthew Bate, Ian Bonnell, and
Volker Bromm [MNRAS 339, 577,
2003] have calculated the collapse of
a 50 Msun cloud core with a diameter
of 0.375 pc.
•
The cloud is turbulent.
•
The simulation runs for 266,000
years.
•
It was carried out on the UK
Astrophysical Fluid Facility, and
required 100,000 CPU hours, about
10% of the total time available during
one year.
The Bate, Bonnell, Bromm 2003 Simulation - 2
•
As the simulation proceeds, the
turbulent motions in the cloud form
shock waves that slowly damp the
supersonic motions.
The Bate, Bonnell, Bromm 2003 Simulation - 3
•
When enough energy has been lost
in some regions of the simulation,
gravity can pull the gas together to
form a dense sub-core.
The Bate, Bonnell, Bromm 2003 Simulation - 4
•
The formation of stellar embryos
begins in the densest sub-core.
•
The embryos rapidly accrete gas
and grow in mass.
The Bate, Bonnell, Bromm 2003 Simulation - 5
•
As the stars and brown dwarfs
interact dynamically with each
other, many are ejected from the
cloud.
The Bate, Bonnell, Bromm 2003 Simulation - 6
•
At the same time that star
formation continues in the dense
sub-cores, members that have been
ejected scatter throughout and
beyond the cloud.
The Bate, Bonnell, Bromm 2003 Simulation - 7
•
The following slides show
enlargements of specific regions of
interest.
•
They are 16 times smaller than the
previous images, which showed the
entire cloud.
•
Each of the following squares are
5100 AU on the side.
•
The first star formation event in the
cloud starts with the birth of a binary.
The Bate, Bonnell, Bromm 2003 Simulation - 8
•
Both gas filaments and disks form
stars and brown dwarfs. Some of
the remaining gas falls in around
these protostars to form
circumstellar disks, helping to build
up the masses of the protostars.
The Bate, Bonnell, Bromm 2003 Simulation - 9
•
Stars and brown dwarfs fall into a
cluster.
•
At this stage the objects range in
mass from nearly the mass of the
Sun down to as small as the mass
of Jupiter.
•
A star with an edge-on disk is
ejected towards the left of the field.
The Bate, Bonnell, Bromm 2003 Simulation - 10
•
An unstable system of 5 stars
(lower right) breaks up and ejects
stars from the cloud in three
different directions.
The Bate, Bonnell, Bromm 2003 Simulation - 11
•
After a pause, star formation begins
again with the gathering together of
more gas (center).
•
The disks of gas that form around
the protostars also contain a lot of
mass, and gravity can also cause
protostars to form within these
disks.
The Bate, Bonnell, Bromm 2003 Simulation - 12
•
The orbits of these objects formed
in circumstellar disks are unstable,
and they are quickly ejected from
the system.
•
Because they are ejected out of the
cloud before they have assembled
their final masses, many of these
objects will remain substellar, i.e.
as brown dwarfs.
The Bate, Bonnell, Bromm 2003 Simulation - 13
•
The gas flows in the complex
potential of the cloud are highly
structured.
•
Many of the stars and brown
dwarfs are surrounded by disks
which are truncated during
encounters.
The Bate, Bonnell, Bromm 2003 Simulation - 14
•
A brown dwarf with a large disk is
ejected to the lower left.
The Bate, Bonnell, Bromm 2003 Simulation - 15
•
The simulation was terminated
after 266,000 years.
Cluster Formation
[Bate, Bonnell, Bromm 2003,
MNRAS, 339, 577].
The simulation runs for
266,000 years, i.e. a second
of the movie corresponds to
10,000 years.
QuickTime™ and a
BMP decompressor
are needed to see this picture.
FUORS IN EARLY STELLAR EVOLUTION
•
FUors are eruptive variables in star forming regions, brightening by 5-6 mag
•
They have F- or G-type supergiant spectra without emission lines (Herbig 1977).
•
In the near-IR they have strong broad CO bandheads in absorption (Reipurth & Aspin 1997).
•
Only about a dozen FUors or FUor-like objects are known.
•
FUor events are likely to be repetitive, but too few are known for good statistics
Decay Times and
Activity
•
The two best studied FUors are FU Orionis itself,
which erupted in 1936, and V1057 Cyg, which
exploded in 1970.
•
The two objects have similar amplitudes, but
whereas FU Ori is still near maximum light,
V1057 Cyg has decayed dramatically, as seen in
the lightcurve here.
•
Even near minimum, V1057 Cyg has extreme
mass loss, with dramatic P Cygni profiles that is
highly variable. The figure shows spectra
between 1996 and 2001, from Herbig et al.
(2001).
Herbig et al. 2001
Triggering Mechanisms
•
Several triggering mechanisms have been proposed:
[1] A throttle mechanism in which the accretion rate through a disk varies around
the more stable infall rate from an envelope (Hartmann & Kenyon 1996)
[2] A large-scale thermal ionization instability may affect the disk and lead to
occasional emptying-out of the inner disk regions (Bell & Lin 1994)
[3] A companion star on an elliptical orbit may perturb the inner disk regions
during periastron passage (Bonnell & Bastien 1992)
Binary FUors
•
Several FUors have companions, including the
famous Z CMa and L1551 IRS5.
•
In two cases, both components in a binary are
known to be FUors: RNO 1B/C (Kenyon et al.
1993) and AR6A/B (Aspin & Reipurth 2003).
•
Given the rarity of FUors, this would be extremely
unlikely if the events were not related, i.e.,
whatever triggered the eruption in one should be
connected to what triggered the eruption in the
other.
•
The separations of the above FUor pairs are so
large that it would take several thousand years
from a periastron passage to reach their present
locations. FUor eruptions probably do not last that
long.
Orbital Evolution of Newly Bound Binaries
•
An ejection will in some cases lead to a bound system
and sometimes to an escape.
•
A quadruple non-hierarchical system will often break
up into two binary systems, which sometimes detach
from each other, and sometimes form a bound
hierarchical quadruple system.
•
The dynamic evolution is a stochastic process that
cannot be predicted, only studied statistically.
•
In the process of being bound, such newly formed
binaries will have highly eccentric orbits (e~0.9).
•
These newborn binaries will evolve with significant
viscous interactions, and as a result start to spiral
together.
Courtesy P. Artymowicz
Herbig-Haro Jets and FUors
•
As the components spiral towards each other on highly eccentric
orbits, they trigger disturbances at periastron leading to increased
accretion and outflow, eventually forming the highly collimated
Herbig-Haro jets.
•
Eventually they get so close (<10AU or so) that they disturb the
inner regions where magnetic fields are anchored, and the stars
lose the ability to produce jets, even though they still have major
mass loss events.
•
The final periastron passages before the individual stellar disks
are converted to a circumbinary disk will appear as major
accretion events, i.e. as FUors.
In this picture, FUors occur right
after the period of jet formation.
Some FUors like L1551 IRS5
still drive small jets.
HH 34
L1551 IRS5
Binary FUors and the Time Scale Problem
•
•
•
The key to the disintegrating multiple
star scenario is that it can solve the
time scale problem: how does it
happen that two stars next to each
other can both turn into FUors but
being too far apart to affect each
other?
The solution is that FUor pairs are
really quadruple systems which
transformed from an unstable nonhierarchical configuration to an
orderly hierarchical configuration.
The quadruple then evolves slowly
through viscous interactions until the
components are so close that they
erupt in FUor outbursts at about the
same time.
1
2
3
Simulations of Disk-Disk
Encounters - I
•
Pfalzner and collaborators have
explored the encounters of two
stars with disks using SPH
simulations.
•
In the present simulation, a 1 Msun
star with a disk with 50 AU radius
moves past a similar object with a
velocity of 10 km/sec in a grazing
collision. The encounter generates
strong spiral structures and leads to
severe disk disturbance and
truncation of the outer disk regions.
Pfalzner, Henning, Kley 2000
Simulations of Disk - Disk Encounters - II
Courtesy Anthony Whitworth and Henri Boffin
The Companion to FU Orionis
•
Wang et al. (2004) discovered a star 0.5” (230AU)
from FU Ori and suggested it could be a companion.
•
We used Subaru to do photometry and spectroscopy
and indeed found the new companion to be a pre-main
sequence star, most likely a K-type star with IR excess
(Reipurth & Aspin 2004)
Is the FU Ori Companion related to the Eruption?
•
If the companion was ejected in 1936 when FU Ori erupted, then it must have a
tangential velocity of 15 km/sec. This is highly unlikely, since we do not see T
Tauri stars zooming around in star forming regions.
•
But the companion could have been ejected a few thousand years ago.
FU Ori must in this picture be a close binary with a separation of roughly 10
AU, i.e. the system should be a triple. In the intervening years the newborn
binary has had time to spiral together, reaching now the period of FUor
outbursts.
•
We should be able to detect this close binary spectroscopically, since the mean
velocity would be around 5 km/sec. However, in a highly eccentric orbit almost
all the velocity change occurs near periastron, which took place in 1936.
Currently little velocity variability is expected.
Should FUor Eruptions be Periodic?
•
In principle, FUor outbursts should take place at regular intervals at each
periastron passage.
•
However, other factors play in, including how long it takes to replenish and
reconfigure the disturbed disks following the previous closest encounter.
•
If a disk requires much longer to reconstitute than the time interval to the next
periastron passage, then there will not be another eruption for a while.
•
In this picture, FUors should be a very heterogeneous group, a veritable zoo of
objects with broad similarities, but distinct differences in specific
characteristics.
Has the Sun always been Single?
• Statistically, the Sun is more likely to have been born in a
small multiple system
• The obliquity of the Solar System: the mean plane of the
planetary orbits is tilted 7 degrees from the Sun’s rotation
axis
• This could be understood if the Sun was ejected from a
small group of newborn stars
Dynamical Interactions
in Pre-Main Sequence Evolution
Dynamical interactions between multiple components and their stochastic nature
are essential for understanding a variety of phenomena in early stellar evolution:
1 - FUor outbursts
2 - The formation of brown dwarfs
3 - The jet phenomenon
4 - The diversity of disk properties in CTTS and WTTS
5 - The binary separation distribution function
6 - The obliquity of the solar system
Attempting to understand star formation through the birth of single stars is no
longer a sufficient approach
STAR FORMATION IN CLUSTERS
• Isolated star
formation in dark
clouds is not so
common.
• Most stars are born
in clusters of stars
similar to the Orion
Nebula cluster,
containing a few
thousand stars.
Newborn Stars in Clusters are Irradiated
HH 502 in
Orion
ACS/HST
Courtesy
John Bally
Astrobiological Implications
• Most stars in our Galaxy are born in rich clusters of a
hundred to a thousand stars
• Most stars in our Galaxy are binaries, and perhaps all stars
are born in small multiple systems
• Whenever we think about conditions for other life forms in
other planetary systems, it is important to not employ a
heliocentric bias.
• The Sun most probably was not born in isolation, but rather
in a large cluster, and possibly/probably in a small multiple
system that disintegrated shortly after birth.
Upcoming Meetings
•
CHONDRITES AND THE PROTOPLANETARY DISK
November 8-11 2004 Kaua’i, Hawaii
http://www.lpi.usra.edu/meetings/chondrites2004
•
PROTOSTARS AND PLANETS V
October 24-28 2005 Hilton Waikoloa Village, The Big Island, Hawaii
http://www2.ifa.hawaii.edu/CSPF/ppv/ppv.html