Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://star.arm.ac.uk/~csj/papers/conference/2005ehb_ahmad_sdss.ps Äàòà èçìåíåíèÿ: Fri Oct 21 17:54:38 2005 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 13:10:56 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: asteroid

Baltic Astronomy, vol. 14, XXX--XXX, 2005.
SPECTRAL ANALYSIS OF sdB­He STARS FROM THE SDSS
A. Ahmad, C. Winter and C. S. Je#ery
Armagh Observatory, College Hill, Armagh BT61 9DG. N. Ireland.
Received 2005 July 28
Abstract. We present spectral classification and physical parameters of a sam­
ple of ``helium­rich'' sdB­He stars from spectra obtained from the SDSS archive.
The spectral classification was carried out using an automated neural network
and the physical parameters were derived using LTE model atmospheres. The
results indicate that most of these stars are not typical He­sdB stars but rather
are normal sdB stars with slight helium enrichment. This is most likely a result
of the use of a di#erent definition of ``helium­rich'' in the initial SDSS classifi­
cation to that used more widely in the field.
Key words: stars: chemically peculiar ­ stars: early­type ­ subdwarfs ­ stars:
fundamental parameters
1. INTRODUCTION
Subluminous B stars form the dominant population of faint blue stars in our
Galaxy down to a limiting magnitude of B # 16. These so­called subdwarf B (sdB)
stars are thought to be low­mass core helium burning stars with a thin hydrogen
envelope. The surfaces of sdB stars are predominantly helium­deficient due to
di#usion and gravitational settling. However, a small number have extremely
helium­rich atmospheres. The evolution of these ``helium­rich subdwarf B'' (He­
sdB) stars has recently been the subject of much debate involving both single and
binary star evolution.
Only a very small fraction (#5%) of sdB stars identified in previous surveys
of faint blue stars like the Palomar Green survey (Green et al. 1986) and the
Edinburgh Cape survey (Kilkenny et al. 1997) are helium­rich. A small number of
stars discovered amongst the many thousand hot subdwarfs in the recent Quasar
survey -- the Sloan Digital Sky Survey (SDSS) have been reported to show strong
helium lines and labelled `sdB­He' (Harris et al. 2003).
In this study we used an artificial neural network (ANN) to classify spectra of
sdB­He stars from the SDSS and to derive fundamental atmospheric parameters.
The aim was to determine whether sdB­He stars are similar to He­sdB stars. This
would increase the number of known helium­rich subdwarfs for further studies.

2 A. Ahmad et al.
2. DATA MINING
Where available, reduced spectra were manually extracted from the SDSS Data
Release server using coordinates listed in Harris et al. (2003). The spectra were
then normalized using the continuum provided in the fits file. The normalized
spectra were then classified and parameterised. It should be noted that the spectra
analysed here are moderate resolution (#3 š A) and have a typical signal­to­noise
(S/N) ratio of #40
3. CLASSIFICATION
We have classified the SDSS sdB­He sample onto the MK­like system defined
by Drilling et al. (2000). As the hot subdwarfs do not fall within the scope of the
original MK system, Drilling et al. (2000) have extended and refined the earlier
work of Drilling (1996) and Je#ery et al. (1997) to construct a three­dimensional
MK­like classification scale for these stars.
Table 1. Classifications of the SDSS sdB­He
sample as determined by the ANN. Error esti­
mates are #2 subtypes for spectral type, #1 sub­
class for luminosity, and #4 subclasses for the
helium class.
Name [SDSS J+] n He ANN Classification
09 40 44.08+00 47 59 0.16 sdB0VIII:He23
11 38 40.69-00 35 31 0.01 sdB3V:He1
12 43 46.38+00 25 34 0.05 sdB1V:He23
12 54 10.86-01 04 08 0.01 sdB3III:He5
13 17 45.80+01 04 50 0.01 sdB0VI:He3
13 45 45.24-00 06 41 0.15 sdO9VII:He21
13 46 35.68-00 18 04 0.09 sdA2IV:He0
13 57 07.35+01 04 54 0.36 sdO6VII:He30
14 15 56.68-00 58 14 0.21 sdB8VI:He14
14 39 17.64+01 02 51 0.01 sdB6V:He3
14 45 14.93+00 02 49 0.02 sdB1VII:He11
15 27 08.31+00 33 08 0.45 sdO9VIII:He35
15 29 05.62+00 21 37 0.06 sdO9VII:He10
15 42 38.43-00 37 58 0.07 sdA2III:He2
This scale is based upon
a sample of spectra from a
number of sources, covering
the wavelength region 4050 --
4900 š A at a resolution of 2.5 š A.
It defines a spectral type run­
ning from sdO1 to sdA, anal­
ogous to MK spectral classes,
and uses luminosity classes
IV -- VIII, where most hot
subdwarfs have a luminosity
class #VII. A helium class
has been introduced, which
runs from `He0' to `He40',
based on H, HeI, and HeII
line strengths.
As our intention is to clas­
sify large quantities of spec­
tra obtained from digital sky
surveys such as the SDSS, we
have trained an artificial neu­
ral network (ANN) to per­
form classifications onto the
Drilling et al. (2000) scale.
The ANN is a feed­forward
back propagation network with an input layer of 901 nodes, two hidden layers
of 5 nodes each, and an output layer of 3 nodes from which is obtained the spec­
tral type, luminosity class, and helium class values determined by the network.
The ANN was trained for 700 iterations on the same set of hot standards used by
Drilling et al. (2000), with the spectra having been velocity corrected and resam­
pled onto a uniform wavelength grid of 4050 -- 4950 š A at a dispersion of of 1 š A per
pixel.

Spectral analysis of sdB­He stars from the SDSS 3
The results of applying the ANN to our sdB­He sample are presented in Table 1.
Each spectrum was velocity corrected and resampled onto the same wavelength
grid as was used for training the ANN. Error estimates (1#) for each of the pa­
rameters determined by the ANN are #2 subtypes for spectral type, #1 subclass
for luminosity, and #4 subclasses for the helium class.
The spectra from Ahmad & Je#ery (2003) were also classified with the ANN
to check for consistency as these have previously been manually classified by J.S.
Drilling. The ANN classification with the the manual classification, within the
above errors.
4. SPECTRAL ANALYSIS
Fig. 1. Optical spectra of sdB­He stars (thick
line) along with best fit model. The spectrum
of the He­sdB star prototype -- PG1544+488 is
plotted on the top for comparison.
The physical parameters ef­
fective temperature (T e# ), sur­
face gravity (log g) and helium
abundance (n He ) were measured
from the optical blue (4200 --
5000 š A) spectra (Fig. 1) using
the latest version of the spec­
tral fitting code SFIT2 and grid
of high­gravity LTE models (cf.
Ahmad & Je#ery 2003).
Note that the blue ends of
the SDSS spectra are incorrectly
normalized when corrected using
the continuum provided by the
SDSS therefore the region from
3900 -- 4200 š A was not consid­
ered in the model fit. Given the
low quality of the SDSS spectra
the errors in T e# are ±1 000K,
in log g are ±0.4, and n He are
±0.05. The sdB­He stars are plot­
ted on the log g - T e# diagram
using the derived parameters in
Fig. 2. The respective helium
abundances are listed in Table 1
as number fraction. From their
position on the log g - T e# di­
gram (Fig. 2), half of our sdB­
He stars are too luminous to be
subdwarfs, the others have a dis­
tribution typical of He­sdB stars.

4 A. Ahmad et al.
Fig. 2. Position of sdB­He stars on the log g - T e# diagram.
5. CONCLUSIONS
We have classified and parameterised a set of spectra of stars identified as
sdB­He in the SDSS. It is clear from both studies that most of these stars show
very little helium enrichment. Half of the stars in our sample are too luminous
to be subdwarfs. Out of the remaining subdwarfs only a handful are helium­rich
(i.e. having n He # 0.10 or He class > 20), again pointing out the need for a
homogeneous classification scheme for hot subdwarfs.
REFERENCES
Ahmad, A., & Je#ery, C. S. 2003, A&A, 402, 335
Drilling, J. S. 1996, in ASP Conf. Ser. 96: Hydrogen Deficient Stars, 461
Drilling, J. S., Moehler, S., Je#ery, C. S., Heber, U., & Napiwotzki, R. 2000, The
Kth Reunion, 49
Green, R. F., Schmidt, M., & Liebert, J. 1986, ApJS, 61, 304
Harris, H. C., Liebert, J., Kleinman, S. J. et al. 2003, AJ, 126, 1023
Je#ery, C. S., Drilling, J. S., Harrison, P. M., Heber, U., Moehler, S. 1997, A&AS,
125, 501
Kilkenny, D., O'Donoghue, D., Koen, C., Stobie, R. S., & Chen, A. 1997, MNRAS,
287, 867