Lithium Abundances in Asymptotic Giant Branch Stars
Department of Physics & Astronomy, Michigan State University, East Lansing, MI 48824-1116
When stars undergo helium shell burning, they are subject to many different mixing processes
which contribute to unusual elemental abundances found in these stars. 7Lithium burns atrelatively low temperatures; however, it is found in these asymptotic giant branch (AGB) stars. This should not be possible, except through the production of lithium via hot bottom burning andthe Cameron-Fowler mechanism. In this study, 122 AGB candidates were analyzed for possiblelithium production. Lithium abundances (or upper limits) were determined for these stars usingMOOG, as well as [Fe/H] and rotational velocity estimates.
Subject headings: stars: Abundances –stars: Asymptotic Giant Branch Stars
dran 2000, Luck & Lambert 1982, Lebre et al. 2006); however, the majority of the stars in these
The understanding of the abundance of lithium
studies have been on the red giant branch, as op-
in stars is an important key to understanding the
posed to the AGB. One of the intentions of this
chemical nature of the universe. Lithium is one
study is to populate a previously under-sampled
of the three elements that were produced in the
Big Bang; lithium is also an integral part of the
The AGB is a fairly inaccessible region of the
proton-proton chain. Lithium is continually de-
H-R diagram because it’s difficult to ascertain that
stroyed throughout the lifetime of a star. Even
stars are indeed on the AGB. Luminosities must
before a star enters the main sequence, convection
be known, which is not an easy task due to the un-
dilutes the primal lithium from the surface of the
certainty that arises in distance calculations. De-
star where it is eventually burned deeper inside
spite the difficulties, it is important to understand
the star through the proton-proton (PP) chain.
these stars, as they provide observable clues to the
Lithium is also destroyed in PPII as the final stage
inside of their cores given their highly convective
of helium production. As a star ascends the RGB,
envelope. This convection allows for material in
what little lithium is left at the surface is con-
the core of the star to be brought to the surface
vected away from the surface, making these stars
and often causes hot bottom burning where the
highly lithium deficient. By the time a star enters
bottom of the convective envelope of the star be-
the AGB, there should essentially be zero lithium;
gins to undergo hydrogen to helium fusion. It also
however, this is not necessarily the case.
works to enable the Cameron-Fowler mechanism
There have been many studies about the abun-
which is responsible for the production of lithium
dance of lithium in stars because its primary fea-
ture is at an accessible area of the spectrum and
In the PPII and PPIII, 7Be is one in a step of
can be easily identified. The studies usually in-
many to form alpha particles. 7Be can gain a pro-
volve dwarf stars or stars on the red giant branch;
ton to become 8B and complete PPIII, or it can
extensive surveys of the lithium abundance in
gain an electron to form 7Li and complete PPII.
AGB stars have not really been conducted to
In order for 7Be to gain a proton, the temperature
date. Studies of the lithium content of giant stars
must be sufficiently high. In the Cameron-Fowler
have been conducted (Charbonnel & Balachan-
transport mechanism 7Be is taken from a hot re-
gion in the star to a cooler region where it’s only
Atmospheric models were created using Kurucz
able to capture protons. This facilitates the pro-
duction of 7Li and serves as an explanation of the
tive temperature, surface gravity, and microtur-
bulence. They also required the overall metallic-
In this analysis, 122 AGB stars are studied
ity of the star in the form of [Fe/H]. Kurucz mod-
to attain metallicities, v sin i, and the lithium
els were used for stars without significant TiO. In
abundance. Thirty-seven positive detections were
general, the coolest stars with Kurucz models were
around 3950 K. They were also made specifically
for each star. MARCS models were much moreappropriate for cooler stars, as they accounted for
line blanketing caused by the TiO, whereas the
122 stars were observed between three observa-
Kurucz models did not. MARCS models were not
tories. 92 stars were observed at McDonald Obser-
made to be star specific; rather a matrix of models
vatory on the 2.1m telescope using the Sandiford
Cassegrain Echelle Spectrometer. 16 stars were
observed at the European Southern Observatoryusing the Fiber-fed Extended Range Optical Spec-
trograph (FEROS) Instrument on the 1.52m tele-scope. 16 stars were also observed at the Haute-
The lithium line is notoriously temperature sen-
provedce Observatory using the 1.52m telescope
sitive. It is crucial to attain a correct value for the
effective temperature of the star. First guess ef-fective temperatures were based on the Ramirez
& Melendez (2005) paper for a temperature scalefor FGK stars. Temperatures were determined by
the V-I, V-J, and V-K colors, with the median be-
IRAF using the echelle, rv, and onedspec pack-
ing taken as the value. V and I magnitudes were
ages. Aurelie data were reduced using MIDAS.
known for these stars and the J and K colors were
Instrument independent data are necessary to
taken from the 2MASS catalog. An extensive liter-
calculate abundances, as well as to enable the ob-
ature search was also conducted to find other effec-
server to compare data taken with different in-
tive temperature estimates for the program stars.
struments. In echelle spectroscopy, instrument in-
When available, the literature temperatures were
dependence is maintained by removing the Blaze
used if the calculated temperatures were clearly
function of the echelle, which differs from instru-
ment to instrument; this process is known as nor-
Reddening is essential to photometric deter-
malization. Normalization of the spectra was car-
mination of effective temperatures. It provided
ried out by the splot routine for all the McDonald
a problem with many of the stars. The Schel-
and FEROS spectra. Aurelie data were reduced
gal (1998) dust maps were used as an indicator
using MIDAS. A fit of a hot star was made for
of reddening; however, given its notable problem
each instrument for each night. Cool stars with
with the overestimation of the reddening of disk
strong TiO bandheads were often fit using the the
stars, Neckel (1980) or Savage (1985) reddening
hot star fits because the continuum was engulfed.
The final determination of effective temper-
ature was conducted through iteration using
MOOG. For hotter stars, the exact temperature
to perform abundance calculations of lithium.
calculated (or quoted from the literature) was used
MOOG requires a parameter file which includes an
as a first guess for the effective temperature; for
atmospheric model, a line list, and the normalized
cooler stars, a matrix of models was used, with
the temperature being forced to the closest avail-
able in the matrix. Five iron lines in the lithium
region were used as an indicator of the tempera-
Microturbulence is a way to account for the tur-
ture. They all had differing oscillator strengths,
bulent stellar atmosphere. Line broadening that
which corresponded differently to incorrect tem-
cannot be attributed to other factors (e.g. rota-
tional velocity) is attributed to the microturbu-
The uncertainties associated with effective tem-
lence of the star. Microturbulence was derived
perature were assigned to be 100K given that the
from the relation of microturbulence and surface
reddening was not always correct. This temper-
ature effect corresponded to the largest source oferror determined in the lithium abundance.
The imprecise nature of the determination of
The surface gravity, log g, was derived from first
the microturbulence reflects the relative impor-
tance of the parameter. Surface gravity was moreimportant than microturbulence; however, tem-
perature was much more important than surface
gravity and was proved to be so in error analysis. Errors were assumed to be 0.5 km/s.
Metallicities were taken from a literature
search. When no metallicity was available, [Fe/H]
was initially assumed to be solar. Metallicity was
adjusted through MOOG, using the New Abun-dances option. For changes in [Fe/H] greater than
This equation (2) can then be substituted in the
or equal to 0.1 dex, another model was used with
the corresponding change in metallicity. Lithium
is a metallicity sensitive line; there is also a strong
iron feature near the lithium line with which it is
often blended. It is therefore essential to correctly
Surface gravity, much like [Fe/H], is really com-
puted in comparison to the solar value, so con-
matching several nearby Fe I lines in the spec-
stants G, π, σ, and 4 drop out of the equation
Given the parameters required to run MOOG,
synthetic spectra were created. This was the vehi-
cle through which the lithium abundance was de-
termined. The fit of the lithium line determined
In the final step of the derivation, surface grav-
the abundance. However, the stellar parameters
ity is actually taken as a logarithm in the atmo-
must first be correct and the abundance of C, Fe,
spheric models given its large values. The final
and Ti are crucial to correctly fitting the region.
The shift in velocity must be accounted for as well. This is trivial in MOOG because there is an op-
tion that allows the user to shift the normalized
spectrum to match the rest wavelengths of the fea-
tures of the synthetic spectrum. However, calcu-
The uncertainties associated with the surface grav-
lating the rotational and radial velocities of the
ity relied most heavily upon the uncertainties in
star, which necessitate the shift, is a nontrivial
the luminosity. This produced an error of 0.4.
process that has been performed on only a subset
tion and computation of these stars is to be com-
Iron is important to correctly synthesize be-
cause it accounts for the overall metallicity of the
star as well as shaping the lithium region. TiO
Of the 122 stars, 37 positive lithium detections
molecules swamp the area in cool stars, beginning
were found. Log values ranged between 1.22 and
at about 3800 K, and essentially destroy the con-
-1.10 dex. As effective temperature goes up, the
tinuum. Adjusting the Ti is important so that
ability to detect lithium and establish upper limits
the features are well matched. Carbon also plays
goes up. This is due to the temperature sensitivity
an important role. In hot stars, the C2 feature
of the lithium line; as the temperature increases,
can dominate the lithium region. In cooler stars,
the line weakens. High S/N with high resolution
the carbon must be adjusted to ensure proper TiO
is required to determine abundances in hot stars
abundances. High carbon abundances in cool stars
with weak lithium, but this is extremely difficult
creates CO, which depletes the available oxygen to
to do, given that high resolution reduces the S/N.
The largest source of error was due to effec-
tive temperature. Errors in effective temperaturecaused the abundances to vary by +0.17 and -0.2
Rotational velocities were computed for all of
in the population of stars hotter than 3900. For
the stars, whilst radial velocities were computed
the cooler stars, errors of +0.25 and -0.3 were more
for the McDonald stars. The rv package in IRAF
appropriate because the temperatures were even
was utilized. The fxcor routine was used to as-
more uncertain given the difficulty fitting the TiO
certain full width half maximums (FWHM), helio-
continuum. Changes in surface gravity and mirco-
centric velocities, and error estimates on the helio-
turbulence did not change the value of the lithium
centric velocities. The rotational velocities, which
line. This should be expected because changes in
correspond to the rotation of the star itself, were
surface gravity really only effect the comparison
calculated using the smallest FWHM per each in-
of ionized species of elements. The lithium here
was 7Li. The resolution of the data was not highenough to really distinguish between 6Li and 7Li.
Microturbulence was also not expected to play a
huge role in error analysis because of its relatively
small contribution to the synthetic spectrum.
In recent papers (Lebre et al. 2006) it has been
is the FWHM of the star. This method does not
suggested that rotational velocity and the lithium
account well for rotational velocities below about
abundance are linked. However, in this data set,
5 km/s. The vrot is obviously wrong for the small-
there are no correlations to be found in rotational
est FWHM of the set and is better understood as
an upper limit for all vrot less than 5 km/s.
abundances having high and low rotation. The
Radial velocities were computed by using the
upper limits were similarly scattered. Though this
rvcorrect routine. It was only performed on the
does not support the studies linking rotation and
McDonald data because those were the only stars
lithium abundances, this also does not negate the
with complete header files which included right as-
work. Errors are sufficiently high enough on the
cension, declination, and the UT at which the ex-
rotational velocities to question the validity of the
posure was taken. Without this information it is
numbers. A link between metallicity and lithium
not possible to complete the calculation. The val-
abundance has also been suggested. However, this
ues computed by rvcorrect must be added to the
heliocentric velocity to account for the earth’s mo-tion around the sun. This is the actual radial ve-
locity computed for the star. Aurelie and FEROSdata did not include the aforementioned informa-
The objective of this project was to analyze 122
AGB stars for lithium abundance; this has been
accomplished with 37 stars yielding positively forthe detection of lithium. An increase in the popu-lation of a previously undersampled region of theH-R diagram has been achieved. Future work onthis project could include calculations of the car-bon isoptic ratio of these AGB stars, as well asthe completion of the radial velocity calculationfor the Aurelie and FEROS data.
Fig. 1.— H-R Diagram of the program stars. Open stars are abundances and the other symbol indicates anupper limit.
Fig. 2.— Plot of the lithium abundances and upper limits versus the effective temperature. Abundancesappear as open stars.
Fe value. There is no obvious correlation between the
two parameters. Abundances appear as open stars.
Fig. 4.— Plot of the log Li value versus the rotational velocity value. Again, there is no obvious correlationbetween the two parameters. Abundances appear as open stars.
Cameron, A. G. W.; Fowler, W. A., 1971, Ap. J., 164, 111C.
Charbonnel, C., Balachandran, S. C., 2000, AandA 359, 563C. do Nascimento, J. D., Jr., et al., 2000, AandA, 357, 931D. Karakas, A., Ph.D. Thesis 2003. Lebre, A., et al. 2006, AandA, 450, 1173L. Luck, R. E., and Luck, D.L. 1982, Ap.J., 256, 189.
LIST OF CHANGES – V1.5.2.375 to V1.5.2.386 Miscellaneous Added the day and date to the status bar. Added a "Last visit" column to the Pregnancy list. An SA Maternal Screen button is now available on the Obstetric page when the practiceReminders due are now removed from the list on the patient window when marked asAdded "Show 'Non visits' in Past Visits list" to Setup/