SPECTROSCOPIC MONITORING OF CARBON-RICH LONG PERIOD VARIABLES

                          T. Lloyd Evans
                              SUPA
                  School of Physics and Astronomy
                     University of St Andrews

A sample of bright long period variables with carbon-rich spectra was 
observed with spectroscopic resolutions of 1-2 A from 1986 to 2001. 
Visual observations from the AAVSO and occasional near infrared photometry 
establish phases and reveal the long term photometric behaviour of these 
stars. Previous papers have reported: (1) the appearance of emission in the 
C_2 bands and the lines of neutral Na and K in the Mira R Lep during a 
rare faint episode and in the semiregular V Hya when at minimum in the long 
(19 year) cycle (Lloyd Evans 1997); (2) the occurrence of emission in the 
(1,0) 4737A band but not the (0,0) 5165A band of C_2 around the faint part 
of the regular 360d pulsation in the semiregular R Scl (Lloyd Evans 2007); 
(3) the appearance of strong emission at the (1,0) band and weak emission 
at the (0,0) band of C_2, accompanied by a severe weakening of absorption 
lines, during faint episodes of the semiregular T Mus (Lloyd Evans 2007). 
The latter observations were interpreted as resulting from the formation of 
dust within the atmosphere of T Mus and of R Scl, causing the weakening of 
the absorption lines and preferential emission of the C_2 band at shorter 
wavelength as the absorption coefficient of C-rich dust increases strongly 
to the violet. It was suggested that the long period fadings of both V Hya 
and T Mus result from internal processes in the envelope and atmosphere.
Lloyd Evans (1997) also showed from near infrared and IRAS mid-infrared 
photometry that V Hya and R Lep suffered greater circumstellar absorption 
during these faint episodes, and that carbon stars which showed similar 
faint episodes in their long term visual light curves were redder than 
those stars which were of constant mean light.

A widely discussed alternative explanation for these periodic fadings is 
that they result from eclipses in a binary system whose orbital period is 
that of the long term variation. This can be tested in the case of V Hya, 
as the secondary is visible in the violet region of the spectrum. The 
spectrum is not that of a normal star, but resembles those of systems with 
accretion disks. There is a continuum, with a colour temperature suggesting 
type F or possibly earlier, with emission lines of H, CaII, [SII] and [FeII]. 
The forbidden lines are displaced by about 160 km/s to the blue and most 
probably result from the bow shock where a jet collides with circumstellar 
clouds (Lloyd Evans 1991). The spectrum has two states, characterised by H 
absorption and emission lines, respectively, similar to those of the high 
and low states of dwarf nova spectra; spectrophotometric observations show 
that this presumed accretion disk is about three times brighter in the former 
state. The secondary must be accreting material from the carbon star; it is 
not clear whether it is a degenerate star, but the resemblance of some 
spectral features to those of T Tauri stars suggests that it is not. The M-
type semiregular SY For has a similar blue secondary spectrum, which is 
always seen in the high state (Lloyd Evans, unpublished). 

The spectrophotometric observations of the secondary allow a long term light 
curve to be drawn up: it matches the time of deep minimum and the slow 
recovery of the carbon star. This is not compatible with a model where the 
variations of the carbon star result from eclipses by a dust cloud associated 
with the secondary. Additionally, blue-displaced absorption lines of CaI 
4226A and SrII 4215A appear during the faint phase, along with similarly-
displaced absorption at KI 7699A. These suggest that increased mass loss 
from the carbon star and the consequent formation of dust is responsible for 
the simultaneous fading of both components of the binary. Fading of a star in 
a binary with a circumbinary, rather than a circumstellar, dust ring might 
occur, but if this resulted from purely geometrical circumstances it is hard 
to account for the sumultaneous fading of both components: one would expect 
the component on the near side of its orbit to suffer most extinction, and it 
is not clear why both components should simultaneously show displaced 
absorption lines. Explanations involving periodic puffing up of the material 
in a circumbinary disk, or the precession of such a disk, would also struggle 
to account for these lines. The case of T Mus seems to point even more 
strongly to phenomena in the envelope and atmosphere of the carbon star, as 
opposed to a binary star explanation. The range of phenomena observed in the 
spectra of the carbon stars requires further study.


References.

Lloyd Evans, T. 1991, MNRAS, 248, 479

Lloyd Evans, T. 1997, MNRAS, 286, 839

Lloyd Evans, T. 2007, in Why Galaxies should care about AGB Stars, ed. 
R. F. Wing & F. Kerschbaum, in press.