the helium abundance in nearby stars and in globular clusterscemw10/talks/portinari.pdf ·...
TRANSCRIPT
Casagrande , Flynn, Portinari, Girardi & Jimenez 2007 (MNRAS 382, 1516)
Portinari, Casagrande & Flynn 2010 (MNRAS in press; arXiv:1004.1539)
The Helium abundance in nearby stars
and in Globular Clusters
Laura Portinari
Tuorla Observatory
University of Turku, Finland
Luca Casagrande (MPA)
Chris Flynn (Tuorla Observatory)
Big Bang Nucleosynthesis
75% H
25% He
+ traces of Li, Be, B
BBN + Stellar Nucleosynthesis
71% H
27% He
1–2% metals
X
X
Y YZ
Stellar populations have produced
all the heavier elements (metals Z)
and some extra helium, in a typical
ratio:
ΔY/ΔZ ≈ 2
YP=0.248 ± 0.003
ΔY/ΔZ=1.7 (Peimbert 2008)
YP=0.252 ± 0.001
ΔY/ΔZ=2.2 (Izotov et al. 2007)
YP=0.256 ± 0.002
ΔY/ΔZ=1.6 (Izotov & Thuan 2010)
1. ΔY/ΔZ ~2 from HII regions (direct He abundance measurements)
2. ΔY/ΔZ ~2 is the value expected from models of stellar
nucleosynthesis and chemical evolution (Chiosi & Matteucci 1982;
Maeder 1992; Carigi & Peimbert 2008; Casagrande 2008)
3. From BBN to the Sun:
(Y☼-YP)/Z ☼ = 1.6 YP = 0.248 (WMAP+BBN)
Y☼ = 0.27, Z ☼ = 0.014 (Asplund et al. 2009)
So far, so good…
ΔY/ΔZ in Globular Clusters
Discovery, in (at least 2) Globular Clusters, of super
helium rich sub-populations with Y~0.4 and ΔY/ΔZ > 100
ω Cen
Multiple Main Sequences
The blue MS is more
metal rich than the red MS
→ must be highly He
enriched, ΔY~0.15
Bedin et al. (2004)
Norris (2004); Piotto et al. (2005); Sollima et al. (2007)
Usually the reddening effect of Z
dominates, but if Y/Z >> 1,
the bluing effect of He opposes
the reddening and may even
invert the trend
In ω Cen the split between
the blue MS ([Fe/H]= –1.3)
and the red MS ([Fe/H]= –1.6)
implies ΔY = 0.15
Assuming YrMS ~ YP ~ 0.25
→ YbMS ~ 0.40
Norris (2004); Piotto et al. (2005);
Sollima et al. (2007)
Piotto et al. (2007)
Another case : NGC 2808
The 3 Main Sequences have the same metallicity within the errors :
[Fe/H] = –1.1 ± 0.03
→ the split is purely due to helium enrichment (ΔY/ΔZ→∞)
YP = 0.248 red MS
Y ~ 0.30 (ΔY=0.05) middle MS
(15%)
Y ~ 0.37 (ΔY=0.12) blue MS
(13%)
In these clusters, the helium enriched subpopulation(s) also provide
an excellent account of the morphology of the Horizontal Branch
Helium is one classic candidate for the 2nd parameter, and for NGC 2808
the helium abundance distribution derived from the HB morphology
corresponds very well to that of the multiple MS.
D’Antona & Caloi (2004)
ω Cen (Lee et al. 2005)
MS broadening detected also in 47 Tuc and in NGC 6752 (a 2nd parameter cluster)
47 Tuc (Anderson et al. 2009) NGC 6752 (Milone et al. 2010)
Some reasons for cautionY~0.4 is expected to leave some other marks in the HR diagram;
one of these is the luminosity of the RGB bump.
In ω Cen, the RGB bumps in the multiple RGB’s are consistent
with an upper limit ΔY < 0.1 (the multiple MS require ΔY = 0.15)
(Sollima et al. 2005)
In ω Cen, RR Lyrae stars have been observed with metallicity close
to that of the blue MS, but normal helium content
→ the Metal Intermediate population is split into a helium-rich
and a helium-normal sub-components?
Sollima et al. (2006)
In NGC 6752 (candidate host of a helium rich population due to its
blue HB and a broad, possibly multiple MS), the helium abundance
directly measured in (not too) hot blue HB stars is normal
Villanova et al. (2009)
In the next few years, much effort dedicated to find independent
confirmation (or not) of the extreme helium enrichment scenario
Theoretical explanationProgenitors challenge: which stars produce Y/Z > 100 ?
AGB (and sAGB stars) : suppress CNO production (Karakas et al. 2006;
Choi & Yi 2007, 2008; Renzini 2008; Ventura & D’Antona 2008)
Massive stars : rotation + meridional circulation + He–rich slow
equatorial winds (Maeder & Meynet 2006; Decressin et al. 2007)
Population III stars of 100—1000 M☼ : Y/Z >> 100; but how to mix
with He-normal population ? (Marigo et al. 2003; Choi & Yi 2007, 2008)
Quantitative problem : the He–rich population is 30%
of the cluster, and the progenitors are highly selected
we need a lot of them in the first generation of stars
Peculiar IMF strongly peaked on very specific objects
The cluster was originally ~ 10−100 times larger
OK for ω Cen, but other GC?
...and why only the ejecta of highly selected progenitors
end up in the second generation?
FINE TUNING REQUIRED
Yi (2009)
”The extreme helium abundance inferred by the blue MS
population would be* an exciting discovery to observers
but a desperate – to – forget nightmare to theorists.”
” I would almost feel happy if someone would come up
to say : It was all a mistake from the start, there is no
such extreme helium population.
Gorge Meynet disagreed [...] the problem is so enigmatic
that we are greatly challenged and excited”
*…the helium abundance was never directly measured
but inferred from the MS fitting
Nearby starsA concomitant increase of Z and Y have opposite
effects on the location of the low MS (ZAMS).
Z > 0 will have its maximum (reddening) effect
if Y=0. Since the corresponding Y > 0, the
effect of metallicity is reduced.
The larger Y/Z , the closer the two ZAMS
the broadening of the low MS depends on Y/Z.
This phenomenon has been used to estimate Y/Z
in nearby low MS stars since the 60´s (Faulkner 1967).
This argument applies within the luminosity range
5.4 ≤ Mbol ≤ 7 (K dwarfs; 0.7 – 0.85 M☼)
Brighter Mbol : evolutionary effects
(location on the HRD depends on age)
Fainter Mbol : largely to fully convective objects
(location on the HRD not so sensitive
to chemical composition)
The effect of Y/Z is a differential effect acting on top of the Z effect
need of accurate positions of stars in the HRD
accurate absolute magnitudes, accurate distances
A major dividing line in these studies was the Hipparcos satellite.
Before Hipparcos, low MS stars of all metallicities appeared to lie
on the same MS within the errors
Y/Z = 5±3 : at Y/Z = 5, theoretical ZAMS overlap around Z☼
Perrin et al. (1977) Fernandes et al. (1996)
After Hipparcos, the separation of the low MS of different metallicity
became apparent
Y/Z = 2–3 (Pagel & Portinari 1998; Jimenez, Flynn et al. 2003)
in agreement with HII regions studies and chemical evolution models
A step forward
In Casagrande et al. (2007) we aimed at improving upon previous results
A much larger homogeneous sample :
86 stars (previous works based on ≈ 30 stars);
homogeneous accurate data and analysis crucial to highlight
the small differential effects of Y/Z
Multi–band optical+NIR photometry and InfraRed Flux Method to
reconstruct the fundamental properties of stars : Mbol and Teff
Casagrande, Portinari & Flynn (2006)
more direct comparison to stellar models in the theoretical HRD
the effects of helium are more prominent in the theoretical HRD
(optical colours strongly metallicity dependent)
Castellani et al. (1999)
So we hoped to narrow down the errorbars...
Casagrande et al. (2007)
Y/Z ~ 2 for Z > 0.015
Y/Z ~ 10 for Z < 0.015
Y/Z ~ 10 is at odds with chemical evolution models and with
results from HII regions, but it is unacceptable especially because
it implies Y << YP at low Z the problem lies with stellar models.
Not noticed in previous Hipparcos–based samples due to sample size
and use of the observational (optical) HRD.
The problem lies with the stellar models : the real ZAMS as a function
of Z are closer to each other than theoretical isochrones predict
(the broadening of the low MS is less than expected)
Gennaro, Prada-Moroni
& Degl’Innocenti (2010)
It’s really worth going to the
theoretical HR diagram !!!
This result does not depend on the adopted set of stellar models
(Padova vs. Yale vs. Teramo vs. MacDonald)
If real ZAMS are closer, is it then ”easier” to revert them ?
The problem reminds, on a milder scale, that of the multiple MS in GC.
If low–MS stellar models over-
estimate Y/Z for local low Z stars,
is Y/Z overestimated in GC too ?
How can we ”quantify” this ?
Our local low MS sample
superimposed on the MS
of ω Cen with (m–M)=13.7
Homology relationsHomology relations hold for entirely radiative structures
— reasonable first approximation for K dwarfs —
and a number of other simplifying assumptions.
Given the location of a reference
ZAMS with composition (Yr, Zr),
homology relations predict where
a second ZAMS of composition (Y, Z)
is located with respect to the first.
Applying homology relations to ω Cen ...
Maximum colour MS split at
R=20.5 (MR≈Mbol=6.5) and B–R=1.2
(B–R) = 0.1 log Teff = 0.0185±0.0015
(colour-Teff-[Fe/H] relations from
Casagrande et al. 2006)
Sollima et al. (2007) The observed Teff split is explained
by means of homology relations
with Y/Z ~ 200, or Y=0.144
If YrMS=YP=0.246 YbMS=0.39
This is in perfect agreement with
isochrone analysis
(Piotto et al. 2005; Sollima et al. 2007)
... and to NGC 2808...
Maximum colour split at F814W=21.3
and F475W–F814W=1.72
(de-reddened B–I=1.74)
(B–I) = ±0.11 log Teff = ±0.01
Piotto et al. (2007)As ΔY/ΔZ → ∞ in NGC2808, homology
relations are recast in terms of Y.
The observed Teff splits imply Y=0.07
for the middle MS and Y=0.14 for the
blue MS.
If YrMS=YP=0.25 YmMS=0.32, YbMS=0.39
This compares well with isochrone analysis
(Fig. 2 of Piotto et al. 2007)
... and to 47 Tuc
Homology relations agree very well with detailed stellar models
in the interpretation of the multiple MS of GC.
Homology relations describe very well the behaviour of theoretical
isochrones as a function of Y and Y/Z (verified by comparing
homology relations with isochrones with a wide range of Y and Z)
Maximum colour split at F606W=20.5 and F606W–F814W=0.9
(F606W–F814W)=0.013 (V-I)=0.017 log Teff =0.003 Y=0.023
vs. Y=0.026 from isochrone analysis (Anderson et al. 2009)
N.B.: The helium dependent term depends only on ΔY, not on the
absolute values of Y: it holds for fixed Y☼ and Y<YP at low Z
(models by Leo Girardi for Casagrande et al. 2007) or for fixed
YP and large Y(Z>0) (Bertelli et al. 2008)
Theoretical stellar models and homology relations agree and
they are both wrong !!! as they return Y<<YP for local low Z stars
We define empirical homology relations, calibrated to return
Y/Z = 2 and reasonable helium abundances for local low Z stars.
We will then use them to reassess the multiple MS of GC.
Empirical homology relations
simplified homology relations
1st term : includes the global
dependence on Y
2nd term (opacity): depends only on Z
We can calibrate the empirical homology relations acting on the first
or on the second term.
Acting on the second term means to change the dependence of the
theoretical MS as a function of Z no effect on the GC multiple MS,
since in GC the metallicity difference is minimal or vanishing.
Acting on the first term means to maximize the change of the model
response to Y maximize the effect on the GC multiple MS.
Calibration of the first term
theoretical empirical
Consequences for globular clusters
ω Cen
Y/Z ~ 75, or Y=0.05 YbMS=0.3
old : Y/Z ~ 200, or Y=0.15 YbMS=0.4
NGC 2808
Y(mMS)=0.02 YmMS≈0.27
Y(bMS)=0.04 YbMS≈0.29
old : YbMS=0.32 and YbMS=0.39
Calibration of the second term
theoretical empirical
Note :
large scatter
No consequences for
Globular Clusters
Summary
The exercise with homology relations is meant to draw the attention of
theorists of stellar evolution on the problem of local low Z MS stars:
Low MS stellar models at low Z are wrong because they predict
unecceptably low Y < YP (Lebreton et al. 1999)
With our exercise we highlight the possible connections with
the puzzle of the extreme helium enrichment in GC
If the fault of current stellar models lies in their (opacity) dependence
on metallicity, the interpretation of GC will not change.
If the fault lies in the response of stellar models to the helium fraction,
the helium rich populations in GC could be far less rich (Y≤0.3 rather
than Y≈0.4 — within reach of ”reasonable” chemical models)
Solutions ? (for local stars)
Is there a problem with the data? Are the ”observed” Teff too cool?
The IRFM Teff scale by Casagrande et al. (2006, 2010) is
(the best on the market and) one of the hottest around : other Teff
scales show similar or even larger mismatch with the models.
All the Teff scales should be off by > 200 K at low Z
The problem of sub-cosmological Y
is found also for K dwarf binaries of
low Z (Casagrande et al. 2007)
Metallicity–dependent mixing length : α(Z☼) = 1.68, α=1.0 at low Z
The dependence of the mixing length on mass or other parameters
is still much debated in literature
Convection: since the convective envelope is thinner at lower Z,
any change in the convection scheme (extra-mixing, undershooting
etc.) shall affect solar metallicity models more than low Z ones
→ this may alter the relative location of isochrones as a function of Z
Casagrande et al. (2007)
Diffusion: it makes the MS models slightly redder and (most
important) sedimentation of metals lowers the surface metallicity of
the stars. The effect per se is not enough to solve the problem.
Casagrande et al. (2007)
One needs to combine diffusion with additional errors on the
spectroscopic metallicities due to NLTE effects (Lebreton et al. 1999)
BUT both effects need to be maximal and recent literature does not
favour this scenario (Gratton et al. 2001; Chaboyer et al. 2001;
Richard et al. 2002, 2005; Korn et al. 2007)
Boundary conditions (gray atmospheres vs. solar-scaled T-τ relation
vs. blanketed atmospheres). The effects are small once each set of
tracks and isochrones is consistently calibrated on the Sun, (adopts
the corresponding mixing length; Vandenberg et al. 2008).
Boundary conditions are expected to become relevant at lower
masses and cooler temperatures (Teff<4500 K) due to the deeper
convective envelopes; we detect no trend of ΔY/ΔZ with Mbol
(Casagrande et al. 2007)
Opacity: wrong dependence on Z ? ”tempting” since the break in
ΔY/ΔZ occurs around Z=0.01, i.e. where κff≈κbf
Is there room, with the recent OP and OPAL opacity databases,
to advocate major changes in κ (Z)? (cf. increase in opacity
suggested by recent Solar models)
Useful solutions (for GC)?
Differential diffusion/sedimentation for helium and metals ?
Helium dependence of opacity; maybe via H─ ?
Rotation? The star looks brighter and bluer, just like for an increase
in Y; but all our stars are slow rotators, and we see no systematic
trend with rotation velocity
Any of the previous suggestions (mixing length, boundary conditions
etc.) recast in terms of He dependence
Most of the solutions listed so far advocate, more or less directly,
a change in the metallicity dependence of isochones
→ not useful for GCs
This is the standard way we think of stellar models! but for the super-
helium rich MSs, we need a change in the dependence on helium.
Preserve the good agreement between MS split and HB morphology
→ act on the luminosity of the models (at fixed mass)
...and the problem may extend to FG dwarfs!
The Geneva-Copenhagen survey highlighted sustematic offsets
in Teff between FGK dwarfs and the location of theoretical MS
Clausen et al. (2010) studied a slightly metal poor F dwarf binary
([Fe/H]= ─ 0.25) and concluded that the best fit is obtained with
Y=0.23-0.24, i.e. formally sub-cosmological (with a set of models
where Y☼ =0.266).
More thinking is needed on how stellar models depend on
their helium content, or on physical effects that mimick He
THE END