Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~pgreen/Papers/turkey.ps Äàòà èçìåíåíèÿ: Sun Dec 22 01:38:25 1996 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 00:17:13 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: äèñïåðñèÿ ñêîðîñòåé

UNDERSTANDING THE ELUSIVE DWARF CARBON STAR
Invited Review Talk, 1996, IAU Symp. 177: The Carbon Star Phenomenon
PAUL J. GREEN
Smithsonian Astrophysical Observatory
Harvard­Smithsonian Center for Astrophysics
60 Garden St., Cambridge, MA 02140, USA
Abstract. Most stars in our Galaxy with photospheric C/O ? 1 (carbon
stars) are not giants but dwarfs. The newly­recognized class of dwarf carbon
stars joins the growing family of stars with peculiar abundances that are
now recognized as products of mass­transfer binary (MTB) evolution. The
dozen examples now known span a wide range of evolutionary histories,
ages, and abundances. These stars can already provide some much­needed
constraints on the formation of AGB C stars in the disk and spheroid
populations, and on the parameters characterizing binary evolution there.
A larger sample, with some bright members, would hasten our progress.
1. Introduction
The `C' and the star on the Turkish flag make this conference a good place
to set the stage for a new perspective on carbon stars. Just so we get it
straight, the cool, luminous AGB stars that dominate C­star theory and
observation (and the rest of this conference) are in all likelihood atypical.
If a carbon star is defined as a star showing carbon molecular features (C 2
bands, with CH, CN and/or s­process enhancements often associated), then
the vast majority of C stars are not giants at all but dwarfs. Recognition of
this paradigm shift is at first glance troublesome, both for observers, who
can no longer assume that a C­star spectrum implies a giant luminosity, and
for theorists, who must explain how a dwarf can show C/O ? 1 when no
carbon is produced by main­sequence hydrogen burning. I'll explain briefly
how we came to recognize that dwarf carbon (dC) stars are not some rare
freak of nature, but constitute a whole class of stars that fall naturally into
the family of post mass­transfer binary systems.

2 PAUL J. GREEN
2. Finding Faint High­Latitude Carbon Stars
Our initial search for Faint High Latitude Carbon (FHLC) stars was moti­
vated by the impression that models of the chemical and dynamical proper­
ties of the Galactic spheroid (the `halo') are still rather weakly constrained.
In the grand scheme, did a monolithic protogalaxy undergo rapid collapse
and enrichment (Eggen, Lynden­Bell & Sandage 1962), or did many smaller
dwarf galaxies merge together (Searle & Zinn 1978)? Both processes may
contribute, with mergings ongoing today. More modestly, how can we best
characterize the mass­to­light ratio, the velocity ellipsoid and systemic ro­
tation of the outer halo? Intrinsically bright stars visible to large galacto­
centric distances (10 \Gamma 100 kpc) provide the best opportunity. Faint C stars
have been sought as excellent tracers of the outer halo because they were
thought to be distant giants, and because they are readily recognizable from
their strong C 2 and CN absorption bands. As tracers, they do not suffer as
do globular clusters from selection effects complicated by tidal interactions
with the disk.
Objective­prism photography with wide­field Schmidt telescopes has
yielded low­dispersion spectra for thousands of objects over substantial
portions of the sky. At high galactic latitudes, we find mostly CH stars,
and possibly some R stars. (Unless stated explicitly, it is to these warmer
types that I refer here.) Fewer than 1% of the 6000 stars in Stephen­
son's (1989) catalogue are the faint, high­latitude carbon (FHLC) stars
(V ? 13; jbj ? 40 ffi ) most useful as dynamical probes of the outer halo. The
two most prolific sources of published FHLC stars, the Case low­dispersion
survey (CLS; Sanduleak & Pesch 1988) and the University of Michigan --
Cerro Tololo survey (UM; MacAlpine & Williams 1981) appear to be com­
plete to about V = 16 and have provided about 30 FHLC stars. Emission­
line objects, not FHLC stars, were the primary goal of these photographic
surveys, and known FHLC stars were not examined to help predefine selec­
tion criteria or estimate completeness. The surface density of FHLC stars
from objective­prism surveys is low, about one per 50 deg 2 to V ¸ 16. The
challenge was then to go deeper than the objective­prism surveys, while
still covering a wide area with CCD imaging. We sought to expand shal­
lower objective­prism survey samples to provide a significant­sized sample
of distant halo C stars. The combined sample could be used for tests of
the spheroid density law, and for dynamical analysis. We used a 3­filter,
two­color photometric technique to distinguish C stars from other late­type
stars via intermediate band (ú 200 š A FWHM) filters (e.g. Cook & Aaron­
son 1989; Palmer & Wing 1982). One (``77'') is centered on a region of TiO
absorption near –7752 š A, and the other (``81'') is on a CN absorption band
near –8104 š A. The 77 \Gamma 81 color thus separates C stars from other stars of

INVITED TALK, IAU SYMP. 177: THE CARBON STAR PHENOMENON 3
similar effective temperature (and similar V \Gamma I) because C stars appear
particularly faint in the 81 filter relative to the 77 filter, while the converse
prevails in M stars. By the end, our CCD survey covered 52 deg 2 of high
galactic latitude sky to a depth of about V = 18: Only one highly ranked
V = 17 candidate was found to have strong carbon and CN bands. That
star, (ff 1950 = 03 h 11 m 44 s :08; ffi = +07 ffi 33 0 38:0 00 ), is of course my favorite
in the whole sky. So perhaps the hubris of using one star to estimate a
surface density will be pardoned. To a depth of V = 18, the surface den­
sity of FHLC stars in our CCD survey is about 0.02 deg \Gamma2 (Green et al.
1994), the same as the surface density from objective­prism surveys. We
went deeper, so why didn't we find more? First, there might well be more
FHLC stars to V = 18 than to V = 16, but we can't tell within the large
uncertainties. If the apparent difference is real, it could be that we begin
to ``run out of Galaxy'' at these magnitudes. Also remember that most
photographic surveys pick up Swan bands of C 2 , while our survey picks
up CN, so that slightly different trends or normalization may obtain with
metallicity and/or galactocentric radius.
3. Stumbling Over Dwarfs
While taking follow­up optical spectra of candidate FHLC stars from our
survey at the KPNO 2.1­m telescope, we also obtained numerous spectra
of previously known FHLC stars (Green & Margon 1990). One of these
objects, CLS 96, also bore the comment ``LP 328­57?'' in the Sanduleak &
Pesch (1988) list, indicating a possible identification with a high­proper­
motion (p.m.) star. Bidelman and MacConnell had independently noticed
this association of CLS 96 with the Luyten p.m. star at around the same
time. When Peter Pesch alerted us to this, we immediately checked the
p.m. catalogues (via SIMBAD) for other faint C stars and found another
probable association, that of LHS 1075 and C*22 from the UM survey.
Astrometry using the 30­year baseline between the POSS and the HST
Quick V Survey plates yielded yet another high p.m. C star, CLS 31. Large
proper motions could only occur at small distances for a star bound to the
Galaxy, thus mandating that these faint C stars must be dwarfs.
Up until then, the only known flaw in the otherwise promising technique
of using FHLC stars as halo tracers had been the existence of one lone
star, G77--61, a V = 13:9 dC also with a high proper motion (Dahn et al.
1977). The object was subsequently shown to be a single­lined spectroscopic
binary of period 245 d and parallax ú abs = 0:017 \Sigma 0:003 00 (Dearborn et al.
1986). These authors argued that, since main­sequence stars can't dredge
up carbon, the most reasonable explanation for the prominent C 2 bands in
the dwarf's spectrum is photospheric deposition of mass from a now unseen

4 PAUL J. GREEN
companion during the companion's second ascent of the giant branch. Even
given such different origins, we found that the optical spectra of dCs at
resolution – 1 š A are strikingly similar to those of Pop II carbon giants,
from their wide distribution of C 13 /C 12 ratios to their enhanced s­process
abundances (Green & Margon 1994). `Faint' doesn't guarantee `distant'
anymore; C stars are nearly on the same footing as other late­type stars.
How could we know if there remained dCs of lower p.m. in the FHLC
star sample? Since IR colors are commonly used to determine C giant lu­
minosities, I took the time to plot the published IR colors of FHLC stars
(e.g. Bothun et al. 1991; Mould et al. 1985). The three dwarfs were redder
in H \Gamma K than most other C stars, but so were two or three other stars.
A more thorough p.m. survey of all known FHLC stars (Green et al. 1992)
showed that the other odd­colored stars were also moving! This brought
the total up to five and strongly suggested that dCs may have JHK colors
distinct enough to identify them as dwarfs.
4. Reign of the Dwarfs
Our Monte Carlo simulations indicated that the proper motion survey, sen­
sitive to p.m. ? 0:1 00 =year, could detect Pop II dCs brighter than V = 18
98% of the time. Therefore, since we detected proper motions for 5 of 39
FHLC stars in our survey, we know that at least 13% of FHLC stars to that
magnitude are dCs. Presumably, deeper surveys will find higher fractions,
because they will begin to probe beyond the Galaxy for giants. But as it
stands, how does the space density of dCs locally compare with other types
of C stars? Simply taking 13% of FHLC stars, with their a surface density
of 0.02 deg \Gamma1 , yields about 100 dCs. All dCs with absolute magnitude esti­
mates to date have M V ¸ 10. Assuming this holds for all dCs, a survey limit
of V = 18 corresponds to a sphere of 400 pc radius. Within this volume, the
number of dCs easily surpasses the sum of all other types of carbon stars
combined, including N, R, CH, Ba and sgCH (or dBa) stars (Green et al.
1992). My simplistic calculation is nevertheless quite conservative since all
these latter types will be brighter by several magnitudes than dCs within
the same volume. Furthermore the estimated C dwarf to C giant fraction
must be a lower limit, because it concerns only high­proper­motion dCs.
The binary mass transfer explanation for photospheric carbon in dCs pre­
dicts that they should exist in the disk as well. Disk dCs would tend to
have small proper motions. If, for example, 2 disk dCs in our sample were
counted incorrectly as giants, then the true fraction is closer to 20%.
Soon after our p.m. survey, Warren et al. (1992) found two faint dCs
in the south by means of their high proper motions. Heber et al. (1993)
found a composite spectrum DA/dC binary system, PG0824+289, with a

INVITED TALK, IAU SYMP. 177: THE CARBON STAR PHENOMENON 5
60,000 K white dwarf. The prototype dC G77--61 was also known to have a
faint companion (T eff ! 6000 K from an IUE upper limit), but PG0824+289
is truly the smoking gun, white hot evidence for the mass­transfer hypoth­
esis. In addition, since PG0824+289 had no detectable proper motion and
disk kinematics, it may represent the first known disk dC. Only a year
passed before a similar DA/dC composite, CBS 311, was found (Liebert et
al. 1994). Neither could have been found via proper motion selection. How
many such dCs are out there?
5. Predicting the Space Density of Dwarf Carbon Stars
If the space density of disk dCs scales with the ratio of disk/halo space
densities (at least ¸ 500: Bahcall & Soneira 1984; Morrison 1993), then we
would expect there to be about 1.3 deg \Gamma2 to our nominal limiting magnitude
of V = 18. But this exceeds the TOTAL surface density of FHLC stars by
about a factor of 6. This back­of­the­napkin calculation shows that the dC
fraction must actually be much lower in the disk than in the halo.
Age and metallicity differences offer a likely explanation for the larger
fraction of dwarfs that are dCs in the spheroid compared to the disk. Other
effects such as the means of accretion (e.g. Roche lobe overflow, or wind)
and mixing efficiency in the accreting star currently preclude simple ana­
lytic predictions. DeKool & Green (1995) have constructed simulated sam­
ples of dC stars to determine whether reasonable assumptions lead to dC
space densities compatible with observations, and to investigate how these
assumptions affect the expected properties of dCs. A simulated population
of dCs is constructed by following the evolution of a large number of bi­
naries using simple analytic fits to detailed evolutionary calculations, and
determining which ones would presently contain a dC star. The zero­age
parameters of the sample are chosen randomly from observed distributions
of unevolved binaries. The space density of halo dC stars that we predict
(¸ 2\Gamma4\Theta10 \Gamma7 pc \Gamma3 ) is in agreement with current observational constraints.
The predicted local space density of disk dC stars (¸ 1 \Theta 10 \Gamma6 pc \Gamma3 ) may
be a bit high, since it still predicts nearly as many disk dCs as there are
FHLC stars observed. The fraction of binaries that produces dCs depends
strongly on initial metallicity, and virtually no dCs are formed in systems
with an initial metallicity of more than half solar. Thus all disk dCs are
predicted to be in binaries that formed in the very early phases of disk
star formation, and their number depends strongly on assumptions about
the age­metallicity relation during this epoch. The predictions for the halo
are much less model­dependent. In either population, we may expect that
dCs on average will exhibit lower than average metallicities. We also pre­
dict that dCs in the disk, for instance, will eventually be shown to have a

6 PAUL J. GREEN
scale­height consistent with old or thick­disk ages. The simulated dC or­
bital period distributions are bimodal, with one peak between 10 3 and 10 5
days and another peak between 10 2 and 10 3 days. The shorter­period com­
ponent is caused by systems that have gone through a common envelope
phase, while most have accreted from an AGB wind. The simulated period
distributions bear a strong resemblance to the observed orbital period dis­
tribution of barium and CH giants, which may be the evolved descendants
of the disk and halo dC populations we modeled.
6. Joining the Mass Transfer Binary (MTB) Family
Because they are considerably more luminous, barium (Ba), CH, and S stars
have been better studied and characterized than dCs. These giants show pe­
culiar abundances, with C/O – 1 and a strong overabundance of s­process
elements, thought to be produced during shell burning on the AGB. Some S
stars are indeed on the AGB. Those suspected of being MTB products (the
`extrinsic' S stars) show no sign of the unstable element technetium (Tc) in
their spectra, unlike their AGB (or `intrinsic') S analogs. Ba, CH and ex­
trinsic (non­Tc) S stars are red giants that have not undergone the thermal
AGB pulsations necessary to produce their observed peculiar photospheric
abundances. Observations are consistent with the MTB hypothesis as an
explanation for all of them. McClure & Woodsworth (1990) present orbital
periods for CH stars; Jorissen & Boffin (1992) have collected orbital pa­
rameters and abundances for Ba stars, and Jorissen & Mayor (1992) for
S stars. The mass functions derived for systems with detected periods are
consistent with white dwarf companions. A fair fraction have no period yet
detected, which almost certainly means that it is very long. This implies
wind accretion, as does the non­zero eccentricity of many orbits. Diagrams
of e vs. log P can reveal the typical Roche lobe radius of the former AGB
primaries in a sample, since tidal effects or Roche lobe overflow will circu­
larize an orbit with separation near that radius.
CH stars are halo giants whose unevolved precursors could be halo dCs.
Similarly, the Ba giants, with old­disk kinematics and near­solar metallici­
ties, are likely to represent the high­mass end of the population of disk stars
that have experienced mass transfer from an AGB companion. The CH sub­
giant stars (Smith, Coleman & Lambert 1993), maybe better described as
Ba dwarfs (dBa), are main­sequence (MS) counterparts, or perhaps pre­
cursors of Ba giants (North, Jorissen & Mayor, this conference). Extrinsic
S stars probably represent somewhat cooler, lower mass MTB disk giants
(Jorissen & Mayor 1992) whose precursors have yet to be postulated. The
spectrum of any post­MTB giant (CH, Ba, or extrinsic S star) may differ
substantially from that of its unevolved dwarf precursor, particularly if the

INVITED TALK, IAU SYMP. 177: THE CARBON STAR PHENOMENON 7
mass of the convective zone changes greatly during evolution.
Lower limits to ages of white dwarf companions (e.g. Johnson et al. 1993;
Smith et al. 1993; Bohm­Vitense et al. 1984) generally exceed the lifetime of
the giant phase, so that the mass transfer episode must have occurred while
the contemporary giant was still on the main sequence. Since MS lifetimes
are much longer than giant lifetimes, the MS precursors to Ba, CH, and
extrinsic S giants should abound, but are clearly more difficult to recognize
or detect. We may explain the unique existence of dCs in both disk and
halo as a consequence of their mass range, the lowest of any post­mass­
transfer objects so far discussed (mostly near 0.5M fi ). Still, the undetected
companion of G77--61 in contrast to the very hot DA companions to the
dCs PG0824+289 and CBS 311 reveals a truly wide range of ages since
mass transfer among dCs.
7. Finding More about dC Stars
I'll just summarize four areas that can be clarified by observations in the
near future: (1) Long­term radial velocity monitoring is needed to prove
duplicity, and to determine the mass function and eccentricity. These must
be correlated with measured abundances. A trend predicted in our simula­
tions, and observed in Ba giants, is a small but significant anti­correlation
between s­process overabundance and orbital period. (2) Good abundance
determinations are needed, particularly of Tc, heavy to light s­process ele­
ment ratios, metallicity, C/O and carbon isotope ratios. More rapid progress
in our understanding will also be made when model atmospheres for dCs
have been matched to UV/optical/IR spectrophotometry and to luminosi­
ties derived from trigonometric parallax. (3) Does enriched accreted mate­
rial mix beyond the shallow convective zones of higher­mass dwarfs? Proffitt
& Michaud (1989) argued that the higher mean molecular weight of the ac­
creted material leads to instability and mixing even into radiative layers. If
there is no such mixing, the predicted number of disk dCs increases by an
order of magnitude, and extends to higher­mass dwarfs. (4) More FHLC and
more dC stars must be found. There may still be low p.m. dCs lurking even
in the sample of currently known FHLC stars. We need to know whether
JHK colors are a good luminosity discriminant for disk dCs, and why they
are a good discriminant for halo dCs. Ancient generations of AGB stars,
the physics of mass transfer, and important parameters describing binary
mass ratios and separations may be probed by such measurements.
Some of these suggestions for dCs mimic hard work already done on
peculiar red giants, which have led to enormous leaps in our understanding
of binary evolution. I hate to point out that the work on dCs may be even
harder, because to date all are fainter than about 14th visual magnitude.

8 PAUL J. GREEN
There are few surveys specifically designed to net new C stars, but we now
know that deeper surveys should reveal more dCs locally. As an example, I
have initiated a deep CCD multicolor Schmidt survey, and a survey using
the CCD grism transit scans of Schmidt, Schneider & Gunn (1995). Mul­
ticolor searches for CH and dC stars in globular clusters have also begun,
from which relative distance and age uncertainties are largely removed. We
hope that the Sloan Digital Sky Survey will be a major source of new dis­
coveries of FHLC and dC stars in the field. A large, well­quantified sample
of dC stars will go a long way toward a better understanding of how evolu­
tion in mass transfer binary systems is affected by age, metallicity and other
factors. Let's just say there's a lot of physics packed into these dwarfs, at
once the most elusive and most common type of carbon star in the Galaxy.
References
Bahcall, J. N. and Soneira, R. M. 1984, ApJS, 55, 67
Bohm­Vitense, E., Nemec, J. and Proffitt, C., 1984, ApJ, 278, 726
Cook, K. and Aaronson, M., 1989, AJ, 97, 923
Dahn, C. C., Liebert, J., Kron, R. G., Spinrad, H. and Hintzen, P. M. 1977, ApJ, 216,
757
Eggen, O. J., Lynden­Bell, D. and Sandage, A. R. 1962, ApJ, 136, 748
Dearborn, D. S. P., Liebert, J., Aaronson, M., Dahn, C. C., Harrington, R., Mould, J.
and Greenstein, J. L. 1986, ApJ, 300, 314
deKool, M. and Green, P. J. 1995, ApJ, 449, 236
Green, P. J. and Margon, B. 1990, PASP, 102, 658
Green, P. J., Margon, B. and MacConnell, D. J. 1991, ApJ, 380, L31
Green, P. J., Margon, B., Anderson, S. F. and MacConnell, D. J. 1992, ApJ, 400, 659
Green, P. J., Margon, B., Anderson, S. F. and Cook, K. 1994, ApJ, 434, 319
Green, P. J. and Margon, B. 1994, ApJ, 423, 723
Heber, U., Bade, N., Jordan, S. and Voges, W. 1993, A&A, 267, L31
Johnson, H. R., Ake, T. B. and Ameen, M. 1993, ApJ, 402, 667
Jorissen, A. and Boffin, H. M. J. 1992, in Binaries as Tracers of Stellar Formation, ed.
A. Duquennoy and M. Mayor (Cambridge University Press), p. 110
Jorissen, A. and Mayor, M. 1992, A&A, 260, 115
Liebert, J., Schmidt, G. D., Lesser, M., Stepanian, J. A., Lipovetsky, V. A., Chaffee, F.
H., Foltz, C. B. and Bergeron, P. 1994, ApJ, 421, 733
MacAlpine, G. M. and Williams, G. A. 1981, ApJS 45, 113
Markarian, B. E., Stepanian, J. A. and Erastova, L. K. 1985, Afz, 23, 439
McClure, R. D. and Woodsworth, A. W. 1990, ApJ 352, 709
Morrison, H. L. 1993, AJ, 106, 578
Mould, J. R., Schneider, D. P., Gordon, G. A., Aaronson, M. and Liebert, J. W. 1985,
PASP, 97, 130
Palmer, L. G., Wing, R. F. 1982, AJ, 87, 1739
Proffitt, C. R. and Michaud, G. 1989, ApJ, 345, 998
Sanduleak, N. and Pesch, P. 1988, ApJS, 66, 387
Searle, L. and Zinn, R. 1978, ApJ, 225, 357
Schmidt, M., Schneider, D. P. and Gunn, J. E. 1995, AJ, 110, 68
Smith, V. V., Coleman, H. and Lambert, D. L. 1993, ApJ, 417, 287
Stephenson, C. B. 1989, Pub. Warner and Swasey Obs., 3, No. 2
Warren, S. J., Irwin, M. J., Evans, D. W., Liebert, J., Osmer, P. S. and Hewett, P. C.
1993, MNRAS, 261, 185