Lithium patterns, Globular Clusters
formation
and the Big Bang Lithium abundance
Francesca D’Antona
plus my coworkers
INAF-OAR
Paolo Ventura
Roberta Carini
INAF-OAR
and…
Annibale D’Ercole
INAF-OABo
Enrico Vesperini
Indiana State University
Francesca Matteucci
University of Trieste
Chemical evolution in the Universe: the next 30 years
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
A short historical summary
• Fundamental discovery of the “Lithium plateau” by Spite
& Spite (1982) at A(Li)~2.3 (mass fraction ~10-9)
• Is this the “Big Bang” Lithium abundance, or is it the
solar system value, 10 times larger?
• (at that time, the problem was the difference between
pop. II results -2.3- and the solar system value -3.3:
no hint that WMAP would provide an “intermediate” Big
Bang Lithium abundance)
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
A short historical summary
• 1982: discovery of the “Lithium plateau” by Spite & Spite (1982)
already at that time, and in following years, three possibilities were
listed:
2) BB Li = solar system
1) BB Li = Spite’ Li 
Li =  depletion of Li
production of Li during
due to surface events
galactic chem. evol.
in pop. II
3) intermediate BB Li 
depletion of Li due to
surface events in pop. II +
production due to chem.ev.
0.0
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Modelling chemical evolution of Li
D’Antona & Matteucci 1990
1990 data of Li
chemical evolution
our best model included novae
massive AGB strong production
by Hot BottomBurning
my best
collaborator…
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Modelling chemical evolution of Li
… after D’Antona & Matteucci 1990…
many models were published by others: all of
them included “plausible” but “invented” and
very large yields for the Lithium
manifactured in the AGB phase
Romano, Matteucci, Ventura &
D’Antona (2001) take into account
Ventura’s new Li-yields for AGBs,
and show that AGBs are poor
contributors!
Romano et al (2010)
(unpublished) take into account
Ventura’s new Li-yields including
super-AGBs, and show that it
might work!
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
The Cameron-Fowler (1971) model and HBB
◊ Li produced by the chain
3
He( ,  )7Be (e  , )7Li
in a convective hot region, so that 7Be is transported to cooler
regions before it turns into Li. Convective mixing brings Li back to
the hot region where it can be burned, but it temporarily survives
in the envelope and in the atmosphere. Production of Li is linked to
the 3He abundance in the region (remnant of incomplete p-p chain)
and lasts until there is 3He
◊ Very luminous AGB stars 
THBB >4 x 107 K
a hydrostatic, slow process:
the bottom of the convective envelope reaches the H-shell burning
region and nuclear reaction products are transported to the
surface by convection
Iben 1973,
Sackman, Smith & Despain 1974,
Scalo, Despain & Ulrich 1975  hot bottom convective
envelopes  Hot Bottom Burning (HBB)
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Hot Bottom Burning : Li, C and O
3He
+ 4He -> 7Be -> 7Li
(Cameron & Fowler 1971)
T>4 x 107 K
In these same envelopes, Carbon also burns, so that
the Carbon star features disappear
12C
->
14N
T>6.5 x 107 K
A third possible processing occurs at even larger T:
H- burning through the ON cycle (this may occur in
the low metallicity massive AGBs and is possibly at
the basis of the self-enrichment process in GCs):
16O
->
14N
T>8 x 107 K
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
HBB temperatures in pop. II
ON processing is at the basis of the O reduction in second generation
stars in Globular Clusters!
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
The O-Na
anticorrelation in GCs
GC
E
I
P
Existence of “anomalous”
stars well known from the
’80s
Carretta et al. 2009 A&A
450, 523
grand total of 1958 individual red
giant stars in the 19 GCs of the
project. [Na/Fe] and [O/Fe] ratios
from GIRAFFE spectra are shown as
open (red) circles; abundance ratios
obtained from UVES spectra (Paper
VIII) are superimposed as filled
(blue) circles and show no offset
from the GIRAFFE sample. Arrows
indicate upper limits in oxygen
abundances.
halo field stars
halo stars should “never” be second generation
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
GC formation is in two steps
THE NEW PARADIGM OF GC FORMATION:
after a first generation (FG) of stars is born, forming a
globular cluster –or even something very different, like a
dwarf galaxy-, a second star formation event, based (at
least partly) on matter processed within FG stars (same Z!)
gives origin to the chemically anomalous second generation
(SG) stars. Loss of the ‘environmental’ dwarf galaxy, or of
MOST of the FG stars from the former GC, leaves out a
globular cluster as it is today
where is matter processed?
FRMS (Decressin et al. 2007…)
AGB-SuperAGB stars (Ventura et al. 2001 & following)
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Models provide O depletion, Na production,
but…
??
models!!
anticorrelation
this is the range of
abundances in low
metallicity field stars
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Successful models: dilution
AGB ejecta
+
pristine gas
provide the mixture
of composition that
can explain the
observations
a “dilution” model
is necessary in
order to obtain the
observed
anticorrelation
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
“Strong” and “mild” O-Na anticorrelations
successful models should be able to reproduce
different cases, as exemplified by NGC 2808 and M4
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
HST Treasury
project GO13297
(Piotto)
First data
reduction of
NGC2808 data
Chemical evolution models
NGC 2808
data from Carretta et
al. 2008 -2009
D’Ercole et al. 2012,
yields from Ventura&
D’Antona 2011
0) FG in place
1) a ‘pure
ejecta’ SG from
super-AGBs, high
Y
2) an intermediate
SG formed by
ejecta of massive
AGBs plus pristine
matter reaccreted
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Shortly: model for clusters with “strong”
chemical anomalies
1: FG in place,
standard C.C.
Na-poor, Orich
2: extreme SG
first phases of
cooling flow:
C.C. of superAGB ejecta:
Na-rich,
O-deprived
3: intermediate
SG: mixture of
massive AGB
ejecta and pristine
gas:
C.C. moderately
Na-rich and
O-poor
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Strong dilution in clusters with ‘mild’ O-Na
data from Marino et
al. 2008
M4
1) the ‘pure ejecta’
SG from super-AGBs,
high Y, must be
inhibited
2) mixing occurs with
a larger % of
pristine gas, SG
formation provides
small O-depletion
and Y spread, some
Na enhancement
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Shortly: model for clusters with “mild”
chemical anomalies
1: FG in place,
standard C.C.
Na-poor, Orich
2: extreme SG
first phases of
cooling flow:
C.C. of superAGB ejecta:
Na-rich,
O-deprived
2: SG is born directly
from pristine gas
contaminated by
massive AGB –
superAGB ejecta:
C.C. moderately Narich and scarcely
deprived in O
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
What we expect for Lithium?
Melendez & Ramirez
Li plateau for
-3<[Fe/H]<-1
(slope for Fe/H<-2.5
not included)
BUT GC dwarfs seem
to have a larger
dispersion: is it due to
the presence of SG
stars?
Lind et al. 2009
Pasquini et al. 2007
Bonifacio et al. 2007
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
What we expect for Lithium?
6
5
4
higher M 
larger T_bottom
 stronger and
faster Li
production
1.5
5
e
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Metallicity dependence at M=6Msun
0.0006
0.001
0.004
higher Z 
smaller T_bottom
 smaller and
slower Li
production
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
But production does not mean yield!
two ingredients are important:
how much Li is made,
and how long it lasts 
so that mass loss can recycle it in the i.s.m.
AND mass loss is another
big unknown parameter
(both absolute values and
dependence on the
luminosity)
Ventura et al. 2002 A&A 339,215
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
the yield depends dramatically on mass loss!
here we see the
role of mass loss
at Z=10-3
Center:
‘standard’ Mdot
Upper models:
Mdot*2
Lower: Mdot from
Vassiliadis & Wood
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
In VD models, mass loss rate calibrated on
Li in the MC O-rich luminous giants
A(Li)
decreases
with initial
mass due
to the
faster He3
destruction
in spite of
larger
abundances
reached in
the
envelope
A(Li) re-increases due
both to the stronger
production and huge
mass loss
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Observational results in some clusters
M4
D’Orazi & Marino 2010 giants shifted
Monaco et al. 2012 turnoff stars
Na-rich (SG) stars may have
just 0.1 dex smaller Li than
Na-poor (FG) stars!
Similar results in other
clusters: NGC 6397, e.g.
Is there an influence of AGB Li
production? Or not?
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Observational results in some clusters
Use the D’Ercole et al. 2012 chemical evolution model
We have a further ambiguity: Which is the Lithium content of the
diluting “pristine” matter? Is it the”observed” value A(Li)=2.2-2.3,
or we have to make the hypothesis that the true value is the Big
Bang abundance, now given at A(Li)=2.6-2.7 ??? In the latter
case, we have also to assume that the same amount of Li surface
depletion takes place in SG stars…
THIS CHOICE MAKES A DIFFERENCE IN THE MODEL RESULTS
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Possibilities (schematically)
1: OBSERVED:
FG stars
A(Li)=2.2-2.3
2: but in the gas
forming the FG
and diluting the
SG:
a) A(Li)=2.2-2.3
b) A(Li)=2.6-2.7
3: SG GAS:
mixture of gas having
either a) or b) plus Li of
relevant AGB ejecta
4: SG STARS:
a) the Li at the surface is that
resulting from the mixture above
b) Li observed is that resulting from
the mixture, but REDUCED by the
same mechanism which brings pop.II
Li from 2.6-2.7 to 2.2-2.3
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Modelling M4
model M4-3
A(Li)pristine=2.2
A(Li) pristine=2.7
A(Li)=2. in ejecta of 7.5 and 8Msun
A(Li)pristine=2.2
A(Li)pristine=2.7
result is OK also if
we assume NO Li in
the AGB ejecta, as
done in the above
figure…
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Result to bring home, from modelling M4
The super-AGB yields seem to be very overestimated to
be consistent with M4 data for Li
If there is Li in the ejecta, or not, does not make any
difference in the final result: We mainly need that the
dilution of Na-rich gas with pristine gas is strong
enough that the Lithium abundance is dominated by the
primordial gas abundance
The choice a) vs. b) for the primordial Li is not
discriminating (unless we take seriously the super-AGB
yields)
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
The case of clusters with strong anomalies
IF the yields of super-AGBs by VD2011 are reliable, we should find
a few very Li-rich stars (A(Li)=2.7-3) in the clusters having an
extreme SG, born directly from super-AGB ejecta, e.g. in wCen,
NGC 2808! (No evidence so far: these yields are too large?)
A(Li)pristine=2.3
A(Li)pristine=2.6
extreme SG
FG
intermediate SG
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Role of intermediate SG
A(Li)pristine=2.6
Slope = 0.6
extreme SG
FG
intermediate SG
A(Li)pristine=2.3
Slope=0.37
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
The case of NGC 6752
no extreme anomalies inthe O-Na anticorr.
Y spread relatively small (from MS)
slope 1
Shen et al. (2010) show
that Li and O are
correlated, with a slope
much smaller than 1
(corresponding to dilution
with matter devoid of both
O and Li)
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
The case of NGC 6752
DA(Li)/D[O/Fe]=0.62±0.05
A(Li)pristine=2.6
DA(Li)/D[O/Fe]=0.27±0.05
A(Li)pristine=2.3
if we had 1) better oxygen – lithium data; 2) better confidence in
the chemical evolution model  we could use the data as a TEST for
Big Bang Lithium!
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
Summary
◊ Still we are not sure about how to interpret the observed
abundance of Li in pop. II stars
◊ Galactic evolution of Lithium requires massive AGB – superAGB as
source of Lithium
◊ The Hot Bottom Burning producing Lithium is also the source of
further nucleosynthesis in AGB – superAGB, and the most plausible
source of the processed matter forming second generation stars in
Globular Clusters
◊ Modelling of GC abundance patterns (e.g. O – Na) can be
successfully achieved. This allows to predict Li distribution among GC
stars
◊ The result of simulations depend on the Big Bang Lithium abundance
assumed to be available in the matter diluting the AGb ejecta.
Observations of Li in GCs may in the end constrain the Original Li.
Meeting in honour of Francesca Matteucci, Castiglione, Sept. 16-20 2013
A MESSAGE FROM ITALO:
Hi Francesca:
what can I say? I’m proud of you!
But I’m proud of me as well, for
I could help in pushing ahead
somebody who is BETTER,
and not WORSE than I am!
So: to the next 30 yr of research
…(my art. 10)!
Francesca,
best wishes for your life,
and good work for the future 30yr
(is this enough?)
from myself and Italo,
thank you, you know why
M4
D’Orazi & Marino 2010