arXiv subjects

T. C. Beers

Publications and source records attributed to T. C. Beers.

At least 55 records · Page 3Linked to original sources

arXiv2015

New Abundances From Very Old Stars

Metal-poor stars provide the fossil record of Galactic chemical evolution and the nucleosynthesis processes that took place at the earliest times in the history of our Galaxy. From detailed abundance studies of low mass, extremely metal-poor stars ([Fe/H] < -3), we can trace and help constrain the formation processes which created the first heavy elements in our Galaxy. Here we present the results of a ~25-star homogeneously analysed sample of metal-poor candidates from the Hamburg/ESO survey. We have derived abundances for a large number of elements ranging from Li to Ba, covering production processes from hydrostatic burning to neutron-capture. The sample includes some of the most metal-poor stars ([Fe/H] < -4) studied to date, and also a number of stars enhanced in carbon. The so called CEMP (carbon enhanced metal-poor) stars, these stars make up ~20% of the stars with [Fe/H] < -3, and 80% of the stars with [Fe/H] < -4.5. The progenitors of CEMP stars are still not fully constrained; they could be a result of binary mass transfer or high-mass explosive events in the early universe.

astro-ph.SR↗

arXiv2014

Chemical abundances of the metal-poor horizontal-branch stars CS 22186-005 and CS 30344-033

We report on a chemical-abundance analysis of two very metal-poor horizontal-branch stars in the Milky Way halo: CS 22186-005 ([Fe/H]=-2.70) and CS 30344-033 ([Fe/H]=-2.90). The analysis is based on high-resolution spectra obtained at ESO, with the spectrographs HARPS at the 3.6 m telescope, and UVES at the VLT. We adopted one-dimensional, plane-parallel model atmospheres assuming local thermodynamic equilibrium. We derived elemental abundances for 13 elements for CS 22186-005 and 14 elements for CS 30344-033. This study is the first abundance analysis of CS 30344-033. CS 22186-005 has been analyzed previously, but we report here the first measurement of nickel (Ni; Z = 28) for this star, based on twenty-two NiI lines ([Ni/Fe]=-0.21$\pm$0.02); the measurement is significantly below the mean found for most metal-poor stars. Differences of up to 0.5 dex in [Ni/Fe] ratios were determined by different authors for the same type of stars in the literature, which means that it is not yet possible to conclude that there is a real intrinsic scatter in the [Ni/Fe] ratios. For the other elements for which we obtained estimates, the abundance patterns in these two stars match the Galactic trends defined by giant and turnoff stars well. This confirms the value of horizontal-branch stars as tracers of the chemical properties of stellar populations in the Galaxy. Our radial velocities measurements for CS 22186-005 differ from previously published measurements by more than the expected statistical errors. More measurements of the radial velocity of this star are encouraged to confirm or refute its radial velocity variability.

astro-ph.SR↗

arXiv2014

Deep SDSS optical spectroscopy of distant halo stars I. Atmospheric parameters and stellar metallicity distribution

We analyze a sample of tens of thousands of spectra of halo turnoff stars, obtained with the optical spectrographs of the Sloan Digital Sky Survey (SDSS), to characterize the stellar halo population "in situ" out to a distance of a few tens of kpc from the Sun. In this paper we describe the derivation of atmospheric parameters. We also derive the overall stellar metallicity distribution based on F-type stars observed as flux calibrators for the Baryonic Oscillations Spectroscopic Survey (BOSS). Our analysis is based on an automated method that determines the set of parameters of a model atmosphere that reproduces each observed spectrum best. We used an optimization algorithm and evaluate model fluxes by means of interpolation in a precomputed grid. In our analysis, we account for the spectrograph's varying resolution as a function of fiber and wavelength. Our results for early SDSS (pre-BOSS upgrade) data compare well with those from the SEGUE Stellar Parameter Pipeline (SSPP), except for stars with logg (cgs units) lower than 2.5. An analysis of stars in the globular cluster M13 reveals a dependence of the inferred metallicity on surface gravity for stars with logg < 2.5, confirming the systematics identified in the comparison with the SSPP. We find that our metallicity estimates are significantly more precise than the SSPP results. We obtain a halo metallicity distribution that is narrower and more asymmetric than in previous studies. The lowest gravity stars in our sample, at tens of kpc from the Sun, indicate a shift of the metallicity distribution to lower abundances, consistent with what is expected from a dual halo system in the Milky Way.

astro-ph.SR↗

arXiv2014

The Alpha-element knee of the Sagittarius Stream

We employ abundances from the Sloan Digital Sky Survey (SDSS) and the Sloan Extension for Galactic Understanding and Exploration (SEGUE) to study the alpha-element distribution of the stellar members of the Sagittarius stream. To test the reliability of SDSS/SEGUE abundances for the study of Sagittarius, we select high-likelihood samples tracing the different components of the Milky Way, and recover known literature alpha-element distributions. Using selection criteria based on the spatial position, radial velocity, distance and colours of individual stars, we obtain a robust sample of Sagittarius-stream stars. The alpha-element distribution of the Sagittarius stream forms a narrow sequence at intermediate metallicities with a clear turn-down, consistent with the presence of an alpha-element "knee". This is the first time that the alpha-element knee of the Sagittarius dwarf galaxy has been detected. Fitting a toy model to our data, we determine that the alpha-knee in Sagittarius takes place at [Fe/H]=-1.27pm0.05, only slightly less metal-poor than the knee in the Milky Way. This indicates that a small number of Sagittarius-like galaxies could have contributed significantly to the build-up of the Milky Way's stellar halo system at ancient times.

astro-ph.GA↗

arXiv2014

Exploring the Origin of Lithium, Carbon, Strontium and Barium with four new Ultra Metal-Poor Stars

We present an elemental abundance analysis for four newly discovered ultra metal-poor stars from the Hamburg/ESO survey, with $\mathrm{[Fe/H]}\leq-4$. Based on high-resolution, high signal-to-noise spectra, we derive abundances for 17 elements in the range from Li to Ba. Three of the four stars exhibit moderate to large over-abundances of carbon, but have no enhancements in their neutron-capture elements. The most metal-poor star in the sample, HE~0233$-$0343 ($\mathrm{[Fe/H]} = -4.68$), is a subgiant with a carbon enhancement of $\mathrm{[C/Fe]}= +3.5$, slightly above the carbon-enhancement plateau suggested by Spite et al. No carbon is detected in the spectrum of the fourth star, but the quality of its spectrum only allows for the determination of an upper limit on the carbon abundance ratio of $\mathrm{[C/Fe]} < +1.7$. We detect lithium in the spectra of two of the carbon-enhanced stars, including HE~0233$-$0343. Both stars with Li detections are Li-depleted, with respect to the Li plateau for metal-poor dwarfs found by Spite \& Spite. This suggests that whatever site(s) produced C either do not completely destroy lithium, or that Li has been astrated by early-generation stars and mixed with primordial Li in the gas that formed the stars observed at present. The derived abundances for the $α$-elements and iron-peak elements of the four stars are similar to those found in previous large samples of extremely and ultra metal-poor stars. Finally, a large spread is found in the abundances of Sr and Ba for these stars, possibly influenced by enrichment from fast rotating stars in the early universe.

astro-ph.SR↗

arXiv2014

Extinction Maps towards the Milky Way Bulge: 2D and 3D Tests with APOGEE

Galactic interstellar extinction maps are powerful and necessary tools for Milky Way structure and stellar population analyses, particularly toward the heavily-reddened bulge and in the midplane. However, due to the difficulty of obtaining reliable extinction measures and distances for a large number of stars that are independent of these maps, tests of their accuracy and systematics have been limited. Our goal is to assess a variety of photometric stellar extinction estimates, including both 2D and 3D extinction maps, using independent extinction measures based on a large spectroscopic sample of stars towards the Milky Way bulge. We employ stellar atmospheric parameters derived from high-resolution $H$-band APOGEE spectra, combined with theoretical stellar isochrones, to calculate line-of-sight extinction and distances for a sample of more than 2400 giants towards the Milky Way bulge. We compare these extinction values to those predicted by individual near-IR and near+mid-IR stellar colors, 2D bulge extinction maps and 3D extinction maps. The long baseline, near+mid-IR stellar colors are, on average, the most accurate predictors of the APOGEE extinction estimates, and the 2D and 3D extinction maps derived from different stellar populations along different sightlines show varying degrees of reliability. We present the results of all of the comparisons and discuss reasons for the observed discrepancies. We also demonstrate how the particular stellar atmospheric models adopted can have a strong impact on this type of analysis, and discuss related caveats.

astro-ph.GA↗

arXiv2014

High-resolution abundance analysis of very metal-poor r-I stars

Moderately r-process-enriched stars (r-I) are at least four times as common as those that are greatly enriched in r-process elements (r-II), and the abundances in their atmospheres are important tools for obtaining a better understanding of the nucleosynthesis processes responsible for the origin of the elements beyond the iron peak. The main aim of this work is to derive abundances for a sample of seven metal-poor stars with classified as r-I stars, to understand the role of these stars for constraining the astrophysical nucleosynthesis event(s) that is(are) responsible for the production of the r-process, and to investigate whether they differ, in any significant way, from the r-II stars. We carried out a detailed abundance analysis based on high-resolution spectra obtained with the VLT/UVES spectrograph. The OSMARCS LTE 1D model atmosphere grid was employed, along with the spectrum synthesis code Turbospectrum. We have derived abundances of light elements Li, C, and N, alpha-elements, odd-Z elements, iron-peak elements, and the trans-iron elements from the first peak, the second peak, the third peak, and the actinides regions. The results are compared with values for these elements for r-II and normal very and extremely metal-poor stars reported in the literature, ages based on radioactive chronometry are explored using different models, and a number of conclusions about the r-process and the r-I stars are presented. Hydrodynamical models were used for some elements, and general behaviors for the 3D corrections were presented.

astro-ph.SR↗

arXiv2014

Chemodynamics of the Milky Way. I. The first year of APOGEE data

We investigate the chemo-kinematic properties of the Milky Way disc by exploring the first year of data from the Apache Point Observatory Galactic Evolution Experiment (APOGEE), and compare our results to smaller optical high-resolution samples in the literature, as well as results from lower resolution surveys such as GCS, SEGUE and RAVE. We start by selecting a high-quality sample in terms of chemistry ($\sim$ 20.000 stars) and, after computing distances and orbital parameters for this sample, we employ a number of useful subsets to formulate constraints on Galactic chemical and chemodynamical evolution processes in the Solar neighbourhood and beyond (e.g., metallicity distributions -- MDFs, [$α$/Fe] vs. [Fe/H] diagrams, and abundance gradients). Our red giant sample spans distances as large as 10 kpc from the Sun. We find remarkable agreement between the recently published local (d $<$ 100 pc) high-resolution high-S/N HARPS sample and our local HQ sample (d $<$ 1 kpc). The local MDF peaks slightly below solar metallicity, and exhibits an extended tail towards [Fe/H] $= -$1, whereas a sharper cut-off is seen at larger metallicities. The APOGEE data also confirm the existence of a gap in the [$α$/Fe] vs. [Fe/H] abundance diagram. When expanding our sample to cover three different Galactocentric distance bins, we find the high-[$α$/Fe] stars to be rare towards the outer zones, as previously suggested in the literature. For the gradients in [Fe/H] and [$α$/Fe], measured over a range of 6 $ < $ R $ <$ 11 kpc in Galactocentric distance, we find a good agreement with the gradients traced by the GCS and RAVE dwarf samples. For stars with 1.5 $<$ z $<$ 3 kpc, we find a positive metallicity gradient and a negative gradient in [$α$/Fe].

astro-ph.GA↗

arXiv2013

Chemical Analysis of the Ninth Magnitude Carbon-Enhanced Metal-Poor Star BD+44 493

We present detailed chemical abundances for the bright carbon-enhanced metal-poor (CEMP) star BD+44 493, previously reported on by Ito et al. Our measurements confirm that BD+44 493 is an extremely metal-poor ([Fe/H]=-3.8) subgiant star with excesses of carbon and oxygen. No significant excesses are found for nitrogen and neutron-capture elements (the latter of which place it in the CEMP-no class of stars). Other elements that we measure exhibit abundance patterns that are typical for non-CEMP extremely metal-poor stars. No evidence for variations of radial velocity have been found for this star. These results strongly suggest that the carbon enhancement in BD+44 493 is unlikely to have been produced by a companion asymptotic giant-branch star and transferred to the presently observed star, nor by pollution of its natal molecular cloud by rapidly-rotating, massive, mega metal-poor ([Fe/H] < -6.0) stars. A more likely possibility is that this star formed from gas polluted by the elements produced in a "faint" supernova, which underwent mixing and fallback, and only ejected small amounts of elements of metals beyond the lighter elements. The Li abundance of BD+44 493 (A(Li)=log(Li/H)+12=1.0) is lower than the Spite plateau value, as found in other metal-poor subgiants. The upper limit on Be abundance (A(Be)=log(Be/H)+12<-1.8) is as low as those found for stars with similarly extremely-low metallicity, indicating that the progenitors of carbon- (and oxygen-) enhanced stars are not significant sources of Be, or that Be is depleted in metal-poor subgiants with effective temperatures of ~5400K.

astro-ph.SR↗

arXiv2012

First stars XVI. STIS/HST abundances of heavy-elements in the uranium-rich star CS 31082-001

Detailed abundances of the elements produced by r-process nucleosynthesis in various circumstances are our best observational clues to their origin, since the site(s) of r-element production is(are) still not known with certainty. A small fraction of extremely metal-poor (EMP) stars exhibit excesses of heavy neutron-capture elements produced in the r-process, and CS 31082-001 is among the 4 well-known r-process-enhanced EMP stars. Observations with HST/STIS provide abundances for elements observable only from the UV region. Here we aim to supplement the optical data with abundances from near-UV spectroscopy of the first and second peak of the r-elements, which are crucial to giving insight into the nucleosynthesis of the elements beyond iron. The UVES spectrum provided additional measurements, thereby improving the previous results. The spectra were analyzed with the OSMARCS LTE model atmosphere and with a consistent approach based on the spectrum synthesis code Turbospectrum to derive abundances of heavy elements in CS 31082-001, using updated oscillator strengths from the recent literature. We computed synthetic spectra for all lines of the elements of interest, checking for proper intensities and possible blends. We combined the abundances of heavy elements derived in previous works with the derivation of abundances from all reliable new list of lines, for the first and second peaks of r-elements. We were able to derive new abundances for 23 n-elements, 6 of them - Ge, Mo, Lu, Ta, W, and Re - were not available in previous works, making this star the most complete r-II object studied, with a total of 37 detections of n-capture elements. We also present the first NLTE+3D lead abundance in this star. The results provide improved constraints on the nature of the r-process.

astro-ph.SR↗

arXiv2012

A New CEMP-s RR Lyrae Star

We show that SDSS J170733.93+585059.7 (hereafter SDSS J1707+58), previously identified by Aoki and collaborators as a carbon-enhanced metal-poor star (with s-process-element enhancements; CEMP-s), on the assumption that it is a main-sequence turn-off star, is the RR Lyrae star VIII-14 identified by the Lick Astrograph Survey. Revised abundances for SDSS J1707+58 are [Fe/H] = -2.92, [C/Fe] = +2.79, and [Ba/Fe] = +2.83. It is thus one of the most metal-poor RR Lyrae stars known, and has more extreme [C/Fe] and [Ba/Fe] than the only other RR Lyrae star known to have a CEMP-s spectrum (TY Gru). Both stars are Oosterhoff II stars with prograde kinematics, in contrast to stars with [C/Fe] < +0.7, such as KP Cyg and UY CrB, which are disk stars. Twelve other RR Lyrae stars with [C/Fe] >= +0.7 are presented as CEMP candidates for further study.

astro-ph.GA↗

arXiv2012

Stellar Archaeology in the Galactic halo with the Ultra-Faint Dwarfs: VI. Ursa Major II

We present a B, V color-magnitude diagram (CMD) of the Milky Way dwarf satellite Ursa Major II (UMa II), spanning the magnitude range from V ~ 15 to V ~ 23.5 mag and extending over a 18 {\times} 18 arcmin2 area centered on the galaxy. Our photometry goes down to about 2 magnitudes below the galaxy's main sequence turn-off, that we detected at V ~ 21.5 mag. We have discovered a bona-fide RR Lyrae variable star in UMa II, which we use to estimate a conservative dereddened distance modulus for the galaxy of (m-M)0 = 17.70{\pm}0.04{\pm}0.12 mag, where the first error accounts for the uncertainties of the calibrated photometry, and the second reflects our lack of information on the metallicity of the star. The corresponding distance to UMa II is 34.7 {\pm} 0.6 ({\pm} 2.0) kpc. Our photometry shows evidence of a spread in the galaxy subgiant branch, compatible with a spread in metal abundance in the range between Z=0.0001 and Z=0.001. Based on our estimate of the distance, a comparison of the fiducial lines of the Galactic globular clusters (GCs) M68 and M5 ([Fe/H]=-2.27 {\pm} 0.04 dex and -1.33 {\pm} 0.02 dex, respectively), with the position on the CMD of spectroscopically confirmed galaxy members, may suggest the existence of stellar populations of different metal abundance/age in the central region of UMa II.

astro-ph.GA↗

arXiv2012

Lithium abundances in CEMP stars

Carbon-enhanced metal-poor (CEMP) stars are believed to show the chemical imprints of more massive stars (M > 0.8 Msun) that are now extinct. In particular, it is expected that the observed abundance of Li should deviate in these stars from the standard Spite lithium plateau. We study here a sample of 11 metal-poor stars and a double-lined spectroscopic binary with -1.8 <[Fe/H]< -3.3 observed with VLT/UVES spectrograph. Among these 12 metal-poor stars, there are 8 CEMP stars for which we measure or constrain the Li abundance. In contrast to previous arguments, we demonstrate that an appropriate regime of dilution permits the existence of "Li-Spite plateau and C-rich" stars, whereas some of the "Li-depleted and C-rich" stars call for an unidentified additional depletion mechanism that cannot be explained by dilution alone. We find evidence that rotation is related to the Li depletion in some CEMP stars. Additionally, we report on a newly recognized double-lined spectroscopic binary star in our sample. For this star, we develop a new technique from which estimates of stellar parameters and luminosity ratios can be derived based on a high-resolution spectrum alone, without the need for input from evolutionary models.

astro-ph.SR↗

arXiv2011

Carbon-Enhanced Metal-Poor Stars in the Inner and Outer Halo Components of the Milky Way

(Abridged) Carbon-enhanced metal-poor (CEMP) stars in the halo components of the Milky Way are explored, based on accurate determinations of the carbon-to-iron ([C/Fe]) abundance ratios and kinematic quantities for over 30000 calibration stars from the Sloan Digital Sky Survey (SDSS). Using our present criterion that low-metallicity stars exhibiting [C/Fe] ratios ("carbonicity") in excess of [C/Fe]$ = +0.7$ are considered CEMP stars, the global frequency of CEMP stars in the halo system for \feh\ $< -1.5$ is 8%; for \feh\ $< -2.0$ it is 12%; for \feh\ $<-2.5$ it is 20%. We also confirm a significant increase in the level of carbon enrichment with declining metallicity, growing from $<$[C/Fe]$>$ $\sim +1.0$ at \feh\ $= -1.5$ to $<$[C/Fe]$>$ $\sim +1.7$ at \feh\ $= -2.7$. The nature of the carbonicity distribution function (CarDF) changes dramatically with increasing distance above the Galactic plane, $|$Z$|$. For $|$Z$|$ $< 5$ kpc, relatively few CEMP stars are identified. For distances $|$Z$|$ $> 5$ kpc, the CarDF exhibits a strong tail towards high values, up to [C/Fe] $>$ +3.0. We also find a clear increase in the CEMP frequency with $|$Z$|$. For stars with $-2.0 <$ [Fe/H] $< -$1.5, the frequency grows from 5% at $|$Z$|$ $\sim 2$ kpc to 10% at $|$Z$|$ $\sim 10$ kpc. For stars with [Fe/H] $< -$2.0, the frequency grows from 8% at $|$Z$|$ $\sim 2$ kpc to 25% at $|$Z$|$ $\sim 10$ kpc. For stars with $-2.0 <$ [Fe/H] $< -$1.5, the mean carbonicity is $<$[C/Fe]$>$ $\sim +1.0$ for 0 kpc $<$ $|$Z$|$ $<$ 10 kpc, with little dependence on $|$Z$|$; for [Fe/H] $< -$2.0, $<$[C/Fe]$>$ $\sim +1.5$, again roughly independent of $|$Z$|$.

astro-ph.GA↗

arXiv2011

The Southern Proper Motion Program IV. The SPM4 Catalog

We present the fourth installment of the Yale/San Juan Southern Proper Motion Catalog, SPM4. The SPM4 contains absolute proper motions, celestial coordinates, and (B,V) photometry for over 103 million stars and galaxies between the south celestial pole and -20 deg declination. The catalog is roughly complete to V=17.5 and is based on photographic and CCD observations taken with the Yale Southern Observatory's double-astrograph at Cesco Observatory in El Leoncito, Argentina. The proper-motion precision, for well-measured stars, is estimated to be 2 to 3 mas/yr, depending on the type of second-epoch material. At the bright end, proper motions are on the International Celestial Reference System by way of Hipparcos Catalog stars, while the faint end is anchored to the inertial system using external galaxies. Systematic uncertainties in the absolute proper motions are on the order of 1 mas/yr.

astro-ph.GA↗

arXiv2011

First stars XIII. Two extremely metal-poor RR Lyrae stars?

The chemical composition of extremely metal-poor stars (EMP stars; [Fe/H]<~ -3) is a unique tracer of early nucleosynthesis in the Galaxy. As such stars are rare, we wish to find classes of luminous stars which can be studied at high resolution. We aim to determine the detailed chemical composition of the two EMP stars CS30317-056 and CS22881-039, originally thought to be red horizontal-branch (RHB) stars, and compare it to earlier results for EMP stars as well as to nucleosynthesis yields from various supernova (SN) models. In the analysis, we discovered that our targets are in fact the two most metal-poor RR Lyrae stars known. Our detailed abundance analysis, taking into account the variability of the stars, is based on VLT/UVES spectra (R~ 43000) and 1D LTE OSMARCS model atmospheres and synthetic spectra. For comparison with SN models we also estimate NLTE corrections for a number of elements. We derive LTE abundances for the 16 elements O, Na, Mg, Al, Si, S, Ca, Sc, Ti, Cr, Mn, Fe, Co, Ni, Sr and Ba, in good agreement with earlier values for EMP dwarf, giant and RHB stars. Li and C are not detected in either star. NLTE abundance corrections are newly calculated for O and Mg and taken from the literature for other elements. The resulting abundance pattern is best matched by model yields for supernova explosions with high energy and/or significant asphericity effects. Our results indicate that, except for Li and C, the surface composition of EMP RR Lyr stars is not significantly affected by mass loss, mixing or diffusion processes; hence, EMP RR Lyr stars should also be useful tracers of the chemical evolution of the early Galactic halo. The observed abundance ratios indicate that these stars were born from an ISM polluted by energetic, massive (25 - 40M*) and/or aspherical supernovae, but the NLTE corrections for Sc and certain other elements do play a role in the choice of model.

astro-ph.SR↗

arXiv2011

First Stars XIV. Sulfur abundances in extremely metal-poor (EMP) stars

Sulfur is important: the site of its formation is uncertain, and at very low metallicity the trend of [S/Fe] against [Fe/H] is controversial. Below [Fe/H]=-2.0, [S/Fe] remains constant or it decreases with [Fe/H], depending on the author and the multiplet used in the analysis. Moreover, although sulfur is not significantly bound in dust grains in the ISM, it seems to behave differently in DLAs and in old metal-poor stars. We aim to determine precise S abundance in a sample of extremely metal-poor stars taking into account NLTE and 3D effects. NLTE profiles of the lines of the multiplet 1 of SI have been computed using a new model atom for S. We find sulfur in EMP stars to behave like the other alpha-elements, with [S/Fe] remaining approximately constant for [Fe/H]<-3. However, [S/Mg] seems to decrease slightly as a function of [Mg/H]. The overall abundance patterns of O, Na, Mg, Al, S, and K are best matched by the SN model yields by Heger & Woosley. The [S/Zn] ratio in EMP stars is solar, as found also in DLAs. We obtain an upper limit on the abundance of sulfur, [S/Fe] < +0.5, for the ultra metal-poor star CS 22949-037. This, along with a previous reported measurement of zinc, argues against the conjecture that the light-element abundances pattern in this star, and, by analogy, the hyper metal-poor stars HE 0107-5240 and HE 1327-2326, are due to dust depletion.

astro-ph.SR↗

arXiv2010

The metal-poor end of the Spite plateau. 1: Stellar parameters, metallicities and lithium abundances

We present VLT-UVES Li abundances for 28 halo dwarf stars between [Fe/H]=-2.5 and -3.5, 10 of which have [Fe/H]<-3. Four different T_eff scales have been used. Direct Infrared Flux Method (IRFM) has been used on the basis of 2MASS infrared photometry. H_alpha wings have been fitted against synthetic grids computed by means of 1D LTE atmosphere models, assuming different self-broadening theories. Finally, a grid of H_alpha profiles has been computed by means of 3D hydrodynamical atmosphere models. The Li I doublet at 670.8 nm has been used to measure A(Li) by means of 3D hydrodynamical NLTE spectral syntheses. An analytical fit of A(Li)(3D, NLTE) as a function of equivalent width, T_eff, log g, and [Fe/H] has been derived and is made available. A(Li) does not exhibit a plateau below [Fe/H]=-3. A strong positive correlation with [Fe/H] appears, not influenced by the choice of the T_eff estimator. From a linear fit, we obtain a strong slope of about 0.30 dex in A(Li) per dex in [Fe/H], significant to 2-3 sigma, and consistent among all the four T_eff estimators. A significant slope is also detected in the A(Li)--T_eff plane, driven mainly by the coolest stars in the sample which appear Li-poor. Removing such stars does not alter the behavior in the A(Li)-[Fe/H] plane. The scatter in A(Li) increases by a factor of 2 towards lower metallicities, while the plateau appears very thin above [Fe/H]=-2.8. The meltdown of the Spite plateau below [Fe/H]\sim-3 is established, but its cause is unclear. If the primordial A(Li) is the one derived from WMAP, it appears difficult to envision a single depletion phenomenon producing a thin, metallicity independent plateau above [Fe/H]=-2.8, and a highly scattered, metallicity dependent distribution below. The fact that no star below [Fe/H]=-3 lies above the plateau suggests that they formed at plateau level and underwent subsequent depletion.

astro-ph.GA↗