ALMA and AGB starsHans Olofsson

Onsala Space Observatory & Stockholm Observatory

- The central stars- The circumstellar envelopes- The circumstellar molecules/isotopologues- Circumstellar structure: dust and gas- The termination of the AGB, early post-AGB

The importance of AGB stars

- The final stellar evolutionary stage for stars in the mass range ≈0.8-8 Msun

=> the majority of all stars that have died in our universe have done this as AGB stars

- They have a large mass return and so are important for the cosmic gas/dust cycle: - they produce heavy elements (3α, s-process) - they produce dust particles - they produce complex molecules (PAHs, ...)

The importance of AGB stars- They are very luminous and old => probes of galaxy structure, kinematics, and SF history

- Their CSEs provide excellent astrophysical and astrochemical laboratories

- They are intricate objects => a full description requires a complex interplay between different physical/chemical processes with different time scales

The structure of an AGB-object

1011 in size107 in temp.1029 in density

nucleosynthesis region

mass loss determined

The stellar mass lossThe mass loss is the process that determines it all:

- the lifetime on the AGB - the luminosity reached - the gas/dust return - the chemical composition of the returned gas

The probes of this process are:

- circumstellar atomic/molecular line emission - circumstellar dust continuum emission

AGB stars are dynamical objectsThe angular resolution of ALMA fits well the important time scales of AGB stars:

- stellar pulsation 1 yr- super-period effects 5 yr- clump ejection 1 yr- gas-dust interaction < 100 yr- thermal pulsing 102-3 yr and 104-5 yr- termination of AGB < 100 yr

!t !

vet

D! 0.3

!

t

100 yr

" !

1 kpc

D

"

arcs

ο Ceti visual lightcurve

AGB stars are dynamical objects

This applies even more to post-AGB objects:

- fast winds - bipolar outflows - jets - shocks - ionization fronts - dissociation fronts

ALMA and the central starsObservational space:

Observational space (1 hour, 5σ = 85 µJy, 230 GHz):

L=4000 Lsun, Te=2800KS!,lowL ! 0.5!

!

230

"2#

1 kpc

D

$2

mJy

S!,highL ! 4!

!

230

"2#

1 kpc

D

$2

mJy

D!,lowL ! 3 kpc

D!,highL ! 8 kpc

L=15000 Lsun, Te=2200K

Note: S350 ≈ 2 S230, and N350 ≈ 2 N230

ALMA and the central stars

Size (tens of convection cells per star is expected):

D > 50 pc => θ* < 0.1 arcs

0.02 arcs, 1 hour, 230 GHz,5σ = 5 K

!∗,highL ! 15

!

1 kpc

D

"

mas

!!,lowL ! 5

!

1 kpc

D

"

mas simulation

ALMA and the central starsA “radio photosphere” appears to exist, and it is larger than R* (≈2R*; Reid & Menten)

Lim et al. Nature 392, 575

α Ori, VLA, 7 mm

size ≈ 0.1”S ≈ 28 mJyTb ≈ 3500 K

Excellent probe ofthe complicated,

inhomogeneous stellar atmosphere

ALMA and the CSEs

vexp(t,θ,φ,MMS,Z,P, ...)

M(t,θ,φ,MMS,Z,P, ...) & dN/dM

ALMA and the CSEsFlux density of a dusty CSE (optically thin):

Observational space, 1 hour (5σ = 85 µJy, 230 GHz):

SCSE,d ! 15

!

M

10!6

"

#

!

230

$3%

1 kpc

D

&2

mJy

DCSE,d ! 13

!

M

10!6

"0.5

kpc Note: S350 ≈ 4 S230, while N350 ≈ 2 N230

- 10-7 Msun/yr in GC in 4 hours- 10-5 Msun/yr in LMC in 2 hours

ALMA and the CSEs

Note that

S! ! !2

while SCSE,d ! !3

≈ 80 GHz at 10-7 Msun/yr

≈ 6 GHz at 10-5 Msun/yr

S*

SCSE

SCSE

but the brightness distributions are very different

ALMA and the CSEs

Flux density of a gaseous CSE:

SCO(3!2) ! SCO(2!1) at high M

SCO(3!2) ! 2 SCO(2!1) at low M

(based on detailed modelling by Ramstedt et al., in prep.)

Note: N345 ≈ 2 N230

SCO(2!1) ! 6

!

M

10!6

"1.2#

15 km s!1

ve

$1.6 #

fCO

10!3

$0.7 #

1 kpc

D

$2

Jy

ALMA and the CSEsObservational space (1 hour, 2 km/s, 5σ = 6 mJy):

- 10-7 Msun/yr in GC in 1 hours- 10-6 Msun/yr in LMC in 4 hours- 10-5 Msun/yr in M31 in 40 hours

DCO(2!1) ! 30

!

M

10!6

"0.6#

15 km s!1

ve

$0.8 #

fCO

10!3

$0.4

kpc

ALMA and the CSEs!! Iso-distance samples are very important !!

M-stars C-stars

M-stars C-stars

dN/dM

M vs P(MMS)

M vs vexp

ALMA and the CSEsSizes of the emitting regions:

!mol ! 0.1 !CO

normally sharply peaked at the centre

sharply peaked at the centre, but extend much further than molecule-CSEs, and is optically thin even at high mass-loss rates

!CO ! 5

!

"0.1 + 0.9

#

M

10!6

$0.7%

&

'

1 kpc

D

(

arcs

BCSE,d(p) ! p!1.4

!CO(8 kpc, 10!6 M"/yr) ! 0.5 arcs

!CO(50 kpc, 10!5 M"/yr) ! 0.5 arcs

photodiss.radius

CO brightness distributions

V Cyg CO(1-0) TX Cam CO(1-0) TX Cam CO(2-1)

R Cas CO(1-0) V Hya CO(1-0) X Her CO(1-0)

30” to60”in

size

PdB large project, syst. vel. maps, Castro-Carrizo et al.

Circumstellar molecules

unique to post-AGB, ∑ = 11CH, CH+, CO+, H2, N2H+,

OCS, HC4H, HC6H, CH3C2H, CH3C4H, C6H6

∑ = 69 circ. species

= only IRC+10216* !!!!

*IRC+10216: the most nearby C-star, and it has a high mass-loss rate

Circumstellar molecules, brightness distributions

IRAM PdB data towards IRC+10216

(Guelin et al. 1996)

Some molecules are of photospheric origin (e.g., SiS),

some are photodissociation products (e.g., CN), and

some are due to circumstellar chemistry (e.g., HNC)

For chemical reasons, these emissions should not coincide (Guelin et al. 1993).

Circumst. SiO abund. of 43 M- and 17 C-type stars

Circumstellar SiO abundance estimates

Schöier et al. , A&A 454, 247

IRC+10216 is not rep-resentative of C-stars

IRC+10216

abundance in C-stars>> eq. value, while in M-stars it is << eq. value

254 F. L. Schöier et al.: SiO in C-rich circumstellar envelopes of AGB stars

Fig. 6. Best-fit models (solid lines; parameters given in Table 3) for R Lep and R For overlayed on observed spectra (histograms).

Table 5. Importance of IR pumping in the excitation of SiO.

I(2! 1) I(3! 2) I(5! 4) I(8! 7)Model with [K km s"1] [K km s"1] [K km s"1] [K km s"1]

CW Leono (star + dust) 11.2 22.8 41.1 74.3star 13.1 28.0 46.8 81.0star + dust 16.5 47.3 98.9 159.2510 K BB 20.4 51.5 100.3 172.6R Lepno (star + dust) 0.27 0.60 1.1 2.0star 0.34 0.88 1.7 2.8star + dust 0.34 1.08 2.6 4.5

From the results reported in Table 3 it is clear that the frac-tional abundance of SiO varies substantially between the carbonstars in the sample, from as low as 1 # 10"7 up to 5 # 10"5.As illustrated in Fig. 7 (solid circles) there is a clear trend thatSiO becomes less abundant as the density, M/ve, in the windincreases. Even more intriguing, when compared to the distri-bution of SiO fractional abundances for M-type AGB stars de-rived by González Delgado et al. (2003) (Fig. 7; filled squares),there appears to be no way of distinguishing a C-rich chemistryfrom that of an O-rich based on an estimate of the circumstel-lar SiO abundance alone. Also, the puzzling division of high-mass-loss-rate stars, into distinct low- and high-SiO fractionalabundance groups, found by González Delgado et al. (2003) forM-type stars is reenforced by the excitation analysis of the car-bon stars.

The non-detection of SiO maser emission (e.g., from the v =1, J = 2! 1 transition) towards carbon stars (e.g., Lepine et al.1978) is interesting in light of the high circumstellar SiO abun-dances found. Since SiO maser emission detected from M-typeAGB stars emanates very close to the photosphere and wellwithin the acceleration region (e.g., Cotton et al. 2004) this couldsuggest that in carbon stars the SiO molecules form somewhatfurther out in the wind where pumping by IR photons, and henceexcitation of vibrationally excited states, is less e!ective.

Fig. 7. SiO fractional abundance ( f0) obtained from an excitation anal-ysis, as a function of a density measure (M/ve), for carbon stars (filledcircles) and M-type (O-rich) AGB stars (filled squares). The horizontallines mark the abundances predicted from equilibrium chemistries. Thedashed line shows the expected f ($) (scaled to 5 # 10"5, roughly theexpected fractional abundance from stellar equilibrium chemistry whenC/O < 1) for a model including adsorption of SiO onto dust grains (seeGonzález Delgado et al. 2003, for details).

4.2. Condensation onto dust grains

The importance of dust grains in regulating the circumstel-lar fractional abundance of SiO is supported by the observedtrend that the SiO fractional abundance decreases with increas-ing density (M/ve) of the wind (Fig. 7). This trend is expectedfrom a classical treatment of the condensation process, as de-scribed in González Delgado et al. (2003), where the condensa-tion e"ciency strongly depends on the dust mass-loss rate. InFig. 7 a depletion curve based on the simple model presented inGonzález Delgado et al. (2003) is shown. Since the theoretical

Only detailed study of a species in a large sample of stars

3 orders of magnitude spread in abundance

fSiO

Rad. transf. modelling of interferometric SiO data

Schöier et al. A&A 422, 651, 2004

red line is the best-fit model using: i) a compact high-abundance (4x10-5) region (eq. value) ii) an extended low-abundance (3x10-6) region (SiO depleted)

only compactcomp.

single-dishmodel

compact and extended comp.R Dor

Circumstellar SiO abundance estimates

Circumstellar isotope ratios

Isotope ratios can be estimated for the highly obscured objects

Many of these ratios are difficult to determine from optical/IR data

nX/mX as a fcn of time

Even a simple analysis is expected to give reliable results as long as optically thin lines are used, provided

no chemical/excitation differences

ALMA and the CSEsDetectability of molecular species:

Example:H2CO(312-211) at 226 GHz4 km/s, 1 hour => S(5σ) = 4 mJyfX ≈ 1x10-8, 10-5 Msun/yr, out to 1 kpc(IRC+10216 lies at ≈120 pc)

S ! 6 guAul

!

fX

10!8

"

#

M

10!6

$

!

Re

1016 cm

" !

15

ve

"2e!El/kTx

Q(Tx)

!

1 kpc

D

"2

Jy

ALMA and the circumstellar structure

Clump characteristics (not well known):

size ! 1013

cm

n ! 1011

cm!3

! Mcl " 10!6

M"

!cl ! 0.5

!

1 kpc

D

"

mas

}

The detailed density distribution is very important for deriving mass-loss rates, abundances, ..., and

for comparison with chemical models

ALMA and the circumstellar structureClump optical depths and flux densities:

Note that Smasers >> SCO,cl

Sd,cl ! 0.006

!

Mcl

10!6

" !

Td

1000

"

#

!

230

$3!

1 kpc

D

"2

mJy

!d,cl ! 0.1

!

Mcl

10!6

" !

1013

rcl

"2#

"

230

$

!CO,cl ! 100

!

Mcl

10!6

" !

1013

rcl

"2 !

1000

Tx

"2

2 " 1 line

SCO,cl ! 0.03

!

Mcl

10!6

" !

1000

Tx

" !

1 kpc

D

"2

mJy 2 " 1 line

ALMA and the circumstellar structure

Example, IRC+10216, θcl ≥ 0.01 arcs:

Sb,d ≈ 0.4 mJy at 230 GHz 5σ ≈ 0.085 μJy in 1 hour, 0.02 arcs res.

SCO(2-1),l ≈ 2 mJy 5σ ≈ 4 mJy in 1 hour, 0.02 arcs and 2 km/s res.

Time-resolved chemistry of both dust and gas, including their interaction

Imaging of molecules and dust

IRC+10216 inscattered interstellar light

molecular line intensity peakson scattered light

Mauron & Huggins

R Scl HST/ACS image

f606w

Comparisons of gas and dust distributions

shell diameter ≈ 38”shell age ≈ 1800 yr

TT Cyg PdB mapshell diameter = 68”shell age ≈ 8000 yr

CO(1-0)

Termination of the AGB phaseMst.env/M < 10-100 yr at the end

M(θ,φ,t)ve(t)

; drastic changes?

The emergence of jetsThe shaping of planetary nebulae

NGC6543<- HST

NOT ->

Termination of the AGB phaseIRC+10216AGB starCO(3-2)

CRL618post-AGBCO(2-1)

HD101584, post-AGB(?)12CO(2-1)

13CO(2-1)

at least three

kinematicalfeatures

bipolar outflow

400 km/s 300 km/s

Termination of the AGB phase

He2-47, young PNHST image

Sahai & Trauger

W43A, OH/IR starVLBA imageImai et al.

8. Attachments (Figures)

Fig. 1: Illustration of the magnetic field collimating the 22 GHz H2O maser jet of W43A. The toroidal field isindicated by the ellipses around the precessing jet. The ellipses are scaled with magnetic field strength. Thesolid black circles denote the observed H2O masers and the vectors are the direction of the magnetic field asdetermined from the maser linear polarization.

Fig. 2: The biconical distribution of the 43 GHz SiO masers close to the stellar photosphere of W43A.

- 4 -

8. Attachments (Figures)

Fig. 1: Illustration of the magnetic field collimating the 22 GHz H2O maser jet of W43A. The toroidal field isindicated by the ellipses around the precessing jet. The ellipses are scaled with magnetic field strength. Thesolid black circles denote the observed H2O masers and the vectors are the direction of the magnetic field asdetermined from the maser linear polarization.

Fig. 2: The biconical distribution of the 43 GHz SiO masers close to the stellar photosphere of W43A.

- 4 -

Magnetic collimation (H2O, 22 GHz, pol)

Vlemmings et al.

Post-AGB circumstellar chemistry11 species are unique to the post-AGB circumstellar medium:

CH, CH+, CO+, H2, N2H+, OCS, HC4H, HC6H,

CH3C2H, CH3C4H, C6H6

- Increased temperature - Increased UV flux - Presence of shocks

stimulates a rich and complex chemistry

ALMA and AGB stars

252 F. L. Schöier et al.: SiO in C-rich circumstellar envelopes of AGB stars

Fig. 3. Best-fit models, derived from the dust radiative transfer mod-elling (parameters given in Table 3), of the observed SEDs for V821 Herand V1965 Cyg.

where I is the integrated line intensity and ! the uncertainty inthe measured value (usually dominated by the calibration uncer-tainty of ±20%), and the summation is done over N independentobservations. The sensitivity of the line emission to variations ofthe two adjustable parameters re and f0 is illustrated in Fig. 4.As can be seen reasonable estimates can be found for all six ob-jects, also the quality of the best-fit models are good, typically"2

red = "2tot/(N ! 2) " 1. The 1! range of acceptable values for re

and f0 are reported in Table 4. However, note that the range ofacceptable values does not follow a rectangular shape as seen inFig. 4.

Given the relatively low number of carbon stars where re canbe determined we have assumed it to scale with density as

log re = 19.2 + 0.48 log!

Mve

", (5)

where M is the mass-loss rate and ve is the expansion velocityof the wind. This scaling law was found by González Delgadoet al. (2003) for a larger sample of M-type AGB stars. As shownin Fig. 5 this scaling law can, within the uncertainties, accountfor the estimated envelope sizes derived for the carbon stars inour study.

The radial distribution of SiO molecules in the gas-phaseis most probably determined by a combination of condensa-tion and photodissociation processes. Thus, one can imagine an

Table 4. SiO abundance distribution.

f0a re

a N "2red

Source [cm]

S Cep 1.4 ± 0.8 # 10!5 4.4 ± 1.6 # 1015 5 0.6V Cyg 1.3 ± 0.7 # 10!5 3.6 ± 1.3 # 1015 6 2.0R For 2.2 ± 1.8 # 10!5 5.4 ± 3.6 # 1015 4 0.3CW Leo 2.8 ± 1.1 # 10!7 1.9 ± 0.7 # 1016 9 0.8R Lep 3.1 ± 1.4 # 10!6 4.4 ± 1.6 # 1015 4 0.9RW LMi 6.0 ± 4.0 # 10!6 7.7 ± 2.5 # 1015 5 1.1

a The abundance distribution is assumed to be Gaussian (see Eq. (3)).

initial (pre-condensation) SiO abundance determined by the stel-lar atmosphere chemistry, possibly modified by non-LTE chem-ical processes due to shocks in the inner part of the wind out toa few stellar radii (e.g., Willacy & Cherchne! 1998). In terms ofclassical evaporation theory (see González Delgado et al. 2003,and references therein) the region over which the SiO abun-dance is expected to change significantly due to condensationonto dust grains is confined within approximately 1015 cm. Forintermediate- to high-mass-loss-rate objects this region is sig-nificantly smaller than the post-condensation region (see Fig. 5),which is limited by photodissociation, and it does not contributenotably to the observed single-dish fluxes. Thus, a single com-ponent is adequate for describing the SiO abundance structurein this case. It should be pointed out that this also means thatno constraints can be put on the pre-condensation abundancefrom the present observations. This requires high spatial resolu-tion observations (Schöier et al. 2004). In the case of low-mass-loss-rate objects the post-condensation region is smaller, but,since the freeze out is less e!ective, the contrast between pre-and post-condensation SiO abundances is small. Hence, a singlecomponent description is adequate also in this case.

3.4. SiO abundances

Using the scaling relation in Eq. (5) it is possible to estimate thecharacteristic size of the envelope (re) containing SiO moleculesfor each of the sample sources. Once the size is known only onefree parameter remains, the fractional abundance of SiO ( f0).The derived sizes and abundances are reported in Table 3. Thefits are generally excellent with reduced "2 values of $1–2. Asexamples we show the best-fit models for R For and R Lep inFig. 6.

From the modelling it is noted that the SiO J = 3 % 2line observed using the IRAM 30 m telescope by Bujarrabalet al. (1994) is consistently lower by about a factor of two com-pared to model predictions. A possible explanation for this ispointing o!sets on the order of $8&&. Due to the larger beam theJ = 2 % 1 observations are less a!ected, typically on the samelevel as the absolute calibration uncertainty $20%. Howevera calibration error can not be ruled out. In the present analy-sis the SiO J = 3 % 2 line intensities reported by Bujarrabalet al. (1994) were not included, except for UU Aur which is onlydetected in this line.

Making an error estimate of the derived abundance is di"-cult as discussed in González Delgado et al. (2003). We havealso tested the e!ect of the assumed abundance distribution byreplacing the adopted Gaussian distribution (Eq. (3)) by an expo-nential decline of the abundance, f (r) = f0 exp (!r/re), as wouldbe the case for a simple photodissociation model without anydust shielding. In the case of a high-mass-loss-rate object such