The Remarkable Chemical Compositions of

Blue Metal-Poor Stars

George Preston

on behalf of

Chris Sneden

friends & collaborators

George Preston (Carnegie Observatories)

John Cowan (University of Oklahoma)

Ian Thompson, Steve Shectman (Carnegie Observatories

Outline of the talk

What are blue metal-poor (= BMP) stars?

Use of binary fractions to resolve BMPs into blue straggler and intermediate-age populations

Fundamental differences between blue stragglers in globular clusters and the halo field

Chemical compositions: general results

A new spectroscopic study: binaries versus single stars

Present & future observational/theoretical opportunities

Identifying field BMP stars: Galactic disk star color-color relation

BMP domain can be populated by metal-poor MS

stars.

It is almost empty in the solar

neighborhood

Preston et al. 1994

Data points are the B8-F0 stars in the Bright Star Catalog

[Fe/H] ~ 0

[Fe/H] ~ 1

de-blanketing (blue) vectors

[Fe/H] ~ 3

BMP domain is a region in which isochrones for a wide range of ages and metallicities overlap in a tangled mess.

Preston & Sneden 2000

Isochrones of various [Fe/H] values and ages overlap in the BMP star domain

Isochrones in the U-B versus B-V plane are from Green et al. (1987)Revised Yale Isochrones

MS isochrones

SG isochrones

Turnoffs for: [Fe/H] = 2.2 ages 3 7,10 Gy

BMP stars were readily identified by UBV photometryof stars found in the HK objective-prism survey

Preston et al. 1994

BM

P st

ars

“HK” Survey: Beers et al. 1985, 1992

MS [Fe/H]=0

MS [Fe/H]= 1

RHB

Metal poor stars near turnoff

BHB

Blue metal-poor stars of the halo field have many physical characteristics of the blue stragglers first identified in globular clusters, but there is a problem:

THERE ARE FAR TOO MANY OF THEM IN THE HALO FIELD!

NGC 288(Kaluzny 1996)

Common distance makes identification easy

MS

RG

B

BH

B

RHB

AGB

MS turnoff

BMP

Resolution of the problem lay in a radial velocity survey:a surprisingly high proportion of BMP stars are members

of spectroscopic binaries

Preston & Sneden 2000

1.5 km/s

< 1.5 km/s: 1/18 stars with orbits

> 1.5 km/s: 41/43 stars with orbits

1.5 km/s is an observational limit; more low-amplitude binaries may be hidden in the RV errors.

>60% of BMP’s are binaries with atypically long orbital periodsand small mass functions.

Standard deviation of a RV measurement (km/s)

Typical orbital solutions based on radial velocity variations in the BMP sample

Preston & Sneden 2000

>60% of BMP’s are binaries with

unusually long orbital periodsand unusually small mass functions.

f(m) (K1)3P

We use binary fractions to resolve blue metal-poor stars into blue straggler and intermediate-age components.

xxxxxxxxxxxxxxx

n

To estimate the BS fraction of BMP n(BS)/n(BMP) fBMP, fBS, and fIA must be “good numbers”.

COMMENTS

fBMP Reliability limited only by accuracy of RV’s & duration of survey

fBS Adopt Duquennoy & Mayor n(P) for primordial blue stragglers.

Assume that 13% of primordial blue straggler binaries with P < 5 d have merged.

Remainder (87%) must be mass-transfer binaries.

fIA Adopt 0.15 as “universal binary fraction” for P<4000 d from:

fDisk = 0.15 Duquennoy & Mayor 1991

fHalo = 0.14 Latham et al 1998

Only this 13% of binaries with P4000d can merge in a Hubble time (Vilhu, O. 1982, A&Ap, 109, 17)

This is why we adoptfBS 0.87radial

velocitybinaries

visualbinaries

c.p.m.binaries

{

To estimate the BS fraction of BMP n(BS)/n(BMP) fBMP, fBS, and fIA must be “good numbers”.

COMMENTS

fBMP Reliability limited only by accuracy of RV’s & duration of survey

fBS Adopt Duquennoy & Mayor n(P) for primordial field blue stragglers.

Assume that 13% of primordial blue straggler binaries with P < 5 d have merged.

Remainder (87%) must be mass-transfer binaries.

fIA Adopt 0.15 as “universal binary fraction” for P<4000 d from

fDisk = 0.15 Duquennoy & Mayor 1991

fHalo = 0.14 Latham et al 1998

the blue straggler fraction of BMP stars is

fBMP fIA fBS

Binary Fraction (f) 0.60 0.15 0.87

nBS/nBMP = (fBMP fIA)/(fBS fIA) = 0.62

More than half of the blue metal-poor stars are blue stragglers.

BUT

Inserting our adopted binary fractions for the three populations

Halo field blue stragglers (FBS) are a different breed.

S = Specific frequency = (BMP)/(HB)

BMP = 300 kpc-3

SBMP = 6.7

HB = 45 kpc-3

BS~108yM(parent pop)

SHFBS = 0.62*6.7 = 4.2

Specific frequency of halo field blue stragglers exceeds

specific frequency of blue stragglers in globular clusters by

factor 10:

Specific frequency of blue stragglers in globular clusters (~0.4) is one order-of-magnitude smaller than value in halo field (~4).

increasing cluster mass

Blue stragglers occur more frequently in less massive, loosely-bound clusters

4.5

4.0

halo field blue stragglers

Mateo, Harris, Nemec, & Olszewski

1990, Astronomical Journal, 100, 469

Mateo et al (1990) used estimates of merger time-scale (~5E+8 y)and blue straggler lifetime (7E+9 y)to conclude that the specific frequency of blue stragglers in NGC 5466 can be explained entirely by mergers of the cluster population of W Uma systems.

Mapelli et al. (2004) simulations of 47 Tuc data confirm that mergers produce the observed SGCBS

outside of the core.

Observed distribution(Ferraro et al 2004)

Collisional formation only

Collisions plus binary mergers

TO SUMMARIZE

We use binary fractions to resolve BMPs into two populations:40% intermediate-age, metal-poor stars (IA)60% old metal-poor blue stragglers (FBS)

A small fraction of HFBS are formed by merger of close pairs. The rest must be formed by McCrea mass transfer, because there are no collisions in the halo.

GCBS are formed primarily by collisions (in core) and mergers (everywhere) of the small portion (10%) of primordial binaries that survive disruption by encounters.

By this reasoning we understand why the specific frequency of HFBS exceeds that of GCBS by an order-of-magnitude.

Abundance analysis of Las Campanas high resolution spectra (R ~ 25,000) of BMP stars

We analyzed summed spectra.

Individual frames used to search for velocity variations have too small S/N to be employed in abundance work.

Preston & Sneden 2000

Vsini=40 km/s

[Fe/H]=-2.30

The abundance analysis

Use Fe-peak lines to derive atmosphere parametersInterpolated model atmospheres from Kurucz’s ATLAS gridStandard LTE analysis with the MOOG code Teff from Fe I abundances with excitation potential log g from Fe I versus Fe II abundances vt from Fe I abundances with EW

Basic results: (1) overall metallicities, (2) Teff’s in 6700-7500K range, (3) main-sequence gravities

Limited # of lines → abundances of 8 elements

Abundances from the Las Campanas BMP high resolution survey

Normal results () compared to other Pop II halo stars:

(1) Mg, Ca, Ti ↑

(2) Mn ↓(3) Sr, Ba: large

at lowest [Fe/H]

Preston & Sneden 2000

Open circles denote stars with vesini > 24 km/s

lower accuracy

Orbital parameters for ordinary binary stars

... and for Carbon-rich & s process-rich binaries

Preston & Sneden 2000

The fraction of binaries with P > 25d & e < 0.15 is small.

The fraction of binaries with P > 25d & e < 0.15 is

larger.

Mass transfer during post-MS evolution circularizes orbits

25 d

Orbital parameters for ordinary binary stars

… and for BMP stars

Preston & Sneden 2000

The fraction of binaries with P > 25d & e < 0.15 is small.

The fraction of binaries with P > 25d & e < 0.15 is larger.

Mass transfer during post-MS evolution circularizes orbits

25 d

Followup high resolution BMP study

5 BMP binaries, 5 BMP RV-constant stars

Las Campanas echelle, with new CCD detector → higher S/N data

Original goal: a comparative Li abundance study

Oops!: Li undetected in all 10 stars!

Much more interesting: look at the 3/5 stars in each group with [Fe/H] < -2

Radial velocities of the low metallicity BMP star sample:Some of them are binaries and others are not.

RV-constant stars Binary stars

Individual and mean spectra of low metallicity RV-constant and binary stars

High excitation O I lines are somewhat stronger in the binaries.

-capture elementsare ~ normal in the whole BMP sample

Original BMP sample: Preston & Sneden 2000

neutron exposure ~ constantUpper-envelope forn-capture elements

declines in the binaries as if neutron

exposure is ~ constant for all [Fe/H].

Carbon species in the spectra of BMP RV-constant stars and binary CS 29497-030

CH & C I respond VERY differently to changes in temperature & gravity!

The “non-variable” spectrum is the mean of three stars

Neutron-capture species in BMP RV-constant stars and binary CS 29497-030 (another lead-

rich star)

Note similarity of lines for

Fe-peak elements

Preston & Sneden 2000

The “non-variable” spectrum is the mean of three stars

Mean abundances in the low-metallicity binary and RV-constant groups

Large values for C, Sr, & Ba

in the binaries indicate real star-to-star

differences

Abundances in CS 29497-030 and the RV-constant stars

Abundances of C, O, and n-capture

elements are new; other abundances are from

Preston & Sneden 2000

We know of several very lead-rich stars

Abundances are normalized to CS 29497-030

at Ba, or La, or both

Normalizations are simple vertical shifts

Sneden et al. 2003

s-process predictions versus abundances in lead-rich stars

Mean observed abundances are computed after normalizations

Neutron/seed ratio is the main variable in the theoretical computations

Arbitrary normalization between theory and observation at Ba & La

Pb-rich stars: a unique abundance signature?

Domain of the enhanced r-process metal-poor stars

Domain of the large s-process, lead-rich stars

Halo sample w [Fe/H]<-1.5

Evolutionary states of known lead-rich stars

Here is a new one:CS 22881-071, an

accidental discovery in a survey of 25 metal-poorred horizontal-branch

(RHB) stars

Preston et al. 2004

Most detailed n-capture abundance pattern of any lead-rich star?

Preston et al. 2004

Other stuff about CS 22881-071:

(1) It is relatively carbon-rich (like all other lead stars)

(2) It is an RR Lyrae star (P=0.59 d) for heaven’s sake!

(3) Is it in a binary? It ought to be! Time will tell.

Recapitulate

BMP stars are in the “wrong” HR diagram place for metal-poor main sequence stars2/3 of BMP stars are BS binaries. 5/7 very metal-poor BMP binaries are rich in

s-process products → AGB mass transferCompanion stars must now be compact objectsPb discovered in one star; others must existThe Pb-rich turn-off stars must have experienced AGB mass transfer? How does this happen?Question: How can mass transfer be so efficient in (now) widely separated pairs?

The End