Probing the circumstellar structure of pre-main sequence stars

Abstract

We present Halpha spectropolarimetry of a large sample of pre-main sequence (PMS) stars of low and intermediate mass, and argue that the technique is a powerful tool in studying the circumstellar geometry around these objects. For the intermediate mass (2 -- 15 Msun) Herbig Ae/Be stars we find that 16 out of 23 show a line effect, which immediately implies that flattening is common among these objects. Furthermore, we find a significant difference in Halpha spectropolarimetry behaviour between the Herbig Be and Ae groups. For the Herbig Be stars, the concept of an electron scattering disc is shown to be a useful concept to explain the depolarizations seen in this spectral range. At lower masses, more complex Halpha polarimetry behaviour starts to appear. The concept of a compact source of Halpha emission that is formed close to the stellar surface, for instance by hot spots due to magnetospheric accretion, is postulated as a working hypothesis to qualitatively explain the Halpha spectropolarimetry behaviour around Herbig Ae and lower mass (M < 2 Msun) T Tauri stars. The striking resemblance in spectropolarimetric behaviour between the T Tauri star RY Tau and the Herbig Ae stars suggests a common origin of the polarized line photons, and hints that low and higher mass pre-main sequence stars may have more in common than had hitherto been suspected.

Full text

Page 1

arXiv:astro-ph/0306094v1 4 Jun 2003
PROBING THE CIRCUMSTELLAR STRUCTURE
OF PRE-MAIN SEQUENCE STARS
Jorick S. Vink, Janet E. Drew, Tim J. Harries, and Rene D. Oudmaijer
Abstract
We present Hα spectropolarimetry of a large sample of pre-main se-
quence (PMS) stars of low and intermediate mass, and argue that the
technique is a powerful tool in studying the circumstellar geometry
around these objects.For the intermediate mass (2 – 15 M⊙) Her-
big Ae/Be stars we find that 16 out of 23 show a line effect, which
immediately implies that flattening is common among these objects.
Furthermore, we find a significant difference in Hα spectropolarimetry
behaviour between the Herbig Be and Ae groups. For the Herbig Be
stars, the concept of an electron scattering disc is shown to be a useful
concept to explain the depolarizations seen in this spectral range. At
lower masses, more complex Hα polarimetry behaviour starts to appear.
The concept of a compact source of Hα emission that is formed close
to the stellar surface, for instance by hot spots due to magnetospheric
accretion, is postulated as a working hypothesis to qualitatively explain
the Hα spectropolarimetry behaviour around Herbig Ae and lower mass
(M < 2 M⊙) T Tauri stars. The striking resemblance in spectropolari-
metric behaviour between the T Tauri star RY Tau and the Herbig Ae
stars suggests a common origin of the polarized line photons, and hints
that low and higher mass pre-main sequence stars may have more in
common than had hitherto been suspected.
1.Introduction
One of the most intriguing open issues in star formation concerns the
formation of massive stars (e.g. Zinnecker; these proceedings). Although
there is a well-established paradigm for the formation of low mass T
Tauri stars, namely via magnetospheric accretion, it is as yet unclear
whether such a scenario would also apply to the more massive stars.
To be able to answer the question whether Nature allows the scaling-up
of the formation mechanism of the Sun to the most massive stars, it
first needs to be established whether the conditions known to prevail
in the lower mass pre-main sequence (PMS) T Tauri stars, such as the

Page 2

2
presence of circumstellar discs and stellar magnetic fields, persist up to
the intermediate mass (2-15 M⊙) PMS Herbig Ae/Be stars.
The question as to whether Herbig Ae/Be stars in general are em-
bedded in circumstellar discs, is still under debate. Although there are
clear indications for flattening from millimeter imaging on larger spa-
tial scales (a few hundred AU) for at least some objects (Mannings &
Sargent 1997), other studies, probing smaller spatial scales, yield results
that seem contradictory. For instance, the IR interferometry of Millan-
Gabet et al. (2001) probes scales of only a few AU, and in this regime
the geometry is found to be rather more spherical. Nonetheless, to be
able to study the circumstellar geometries around PMS stars at the clos-
est spatial scales, one needs to resort to the tool of spectropolarimetry,
as this is the only technique that may probe the geometry on scales of
stellar radii (equivalent to ∼ 0.05 AU) compared to the > 1 AU scales
probed by other methods.
2.The Tool of Linear Spectropolarimetry
In principle, the detection of linear polarization of ∼ 2 %, would teach
us directly that a specific source is non-spherically symmetric on the sky.
However, such a level of polarization may also be due to polarization
by dust grains in the interstellar medium operating between the source
and the observer. Unfortunately, properly correcting for this interstellar
contribution has been proven to be a difficult task (e.g.
Clarke 1979). This is one of the prime reasons as to why spectrally-
resolved polarization changes across emission lines are so valuable, as
the interstellar polarization affects the continuum and the line in exactly
the same way: any observed change in the polarization across the line
has to be intrinsic to the source.
Although spectropolarimetry has widely been applied to more evolved
early-type stars, such as classical Be stars (e.g Poeckert 1975), the tech-
nique has only recently been applied to pre-main sequence stars (Oud-
maijer & Drew 1999, Vink et al. 2002, 2003). For classical Be stars,
the dominant effect is known to be due to unpolarized line emission in
the presence of intrinsic continuum polarization (e.g. Clarke & McLean
1974). This ‘depolarization’ effect across emission lines occurs because
the line photons are formed over a larger volume (in the circumstellar
disc) than the continuum photons and are therefore scattered to a lesser
extent off free electrons in the disc than are the continuum photons.
Consequently, a drop in the polarization percentage is seen (see Fig 1a
for an example).
McLean &

Page 3

The circumstellar structure of PMS stars
3
1a)1b)
Figure 1: Triplots of the observed polarization spectra at Hα of the Herbig Be
star BD+40 4124 (LHS) and the Herbig Ae star XY Per (RHS). On both plots, the
Stokes I spectrum is shown in the lowest panel, the %Pol is indicated in the middle
panel, while the position angle, θ, is plotted in the upper panel. The data have been
rebinned to constant errors of 0.05 % for BD+40 4124 and 0.12 % for XY Per, as
calculated from photon statistics. The data are taken from Vink et al. (2002).
In certain circumstances however, it is feasible that the converse oc-
curs: a proportion of the line photons originate from a compact source
and are scattered and polarized themselves (McLean 1979; Wood et al.
1993). Observationally, such effects have only recently been detected
in intermediate and low mass Herbig Ae and T Tauri stars (Vink et
al. 2002, 2003) using medium/high resolution (R ≃ 8000) spectropo-
larimetry. Here, the Hα line is believed to be polarized by scattering in
a rotating non-spherically symmetric medium, most likely an accretion
disc. Examples of both types of line effect, i.e. depolarization versus
line polarization, are presented for respectively Herbig Be and Ae stars
in Sects. 1.3 and 1.4 below.
3.The Herbig Be Stars
A polarization spectrum for the Herbig Be star BD+40 4124 is shown
in Fig. 1(a), and the observed behaviour across the Hα line profile in
both polarization percentage (%Pol) and position angle (PA) is consid-
ered to be consistent with depolarization. The reason the PA shows a

Page 4

4
change across the line as well is attributed to the vector addition of the
interstellar polarization contribution. Note that the smooth and broad
depolarization effect is represented in the QU diagram of Fig. 2(a) by a
more or less linear excursion of the line points out from the dense knot
representing the continuum at (Q,U) = (−0.3,1.25). The angle between
this knot and the linear line excursion is directly related to the direc-
tion of the flattening of the presumed electron scattering disc around
BD+40 4124.
1a)1b)
Figure 2: QU representations of the observed polarization spectra of the same
data as in Fig. 1. The arrow denotes the sense of increasing wavelength. The more
or less linear excursion of the Hα line data for the Herbig Be star (LHS) is consistent
with depolarization. The Herbig Ae data (RHS) is represented by a loop in the QU
diagram. Note that the plot axis directions +Q, +U, −Q, −U correspond to sky PAs
of respectively 0o, 45o, 90o, and 135o(i.e. ∆U/∆Q = tan 2θ).
In the event that all Herbig Be stars are embedded in electron scatter-
ing discs, one would not expect a 100% detection rate of Hα depolarisa-
tions, as at least some of the sources would have their discs too “pole-on”
with respect to the observer to yield a %Pol drop large enough to be de-
tectable. To estimate the expected fraction of depolarization detections,
we turn to a comparison with classical Be stars (Poeckert & Marlbor-
ough 1976), objects for which the presence of a circumstellar disc is
well-established. Applying the same detection threshold to the Poeckert
& Marlborough sample as in ours, we expect a detection rate of about
54%. Returning now to our Herbig Be star sample, we find that 7 out
of 12 (i.e. 58%) show a detectable depolarisation (Oudmaijer & Drew
1999, Vink et al. 2002). We conclude that, given both the statistics, as
well as the smooth and broad depolarization behaviour in the Herbig Be

Page 5

The circumstellar structure of PMS stars
5
Hα data, that all early Herbig Be stars are likely embedded in electron
scattering discs.
4.The Herbig Ae Stars
When observing the later spectral type Herbig Ae stars, one may
expect to see a sharp decrease in the frequency of line effect detections,
as the circumstellar ionization as well as the amount of free electrons that
can scatter and polarize, are expected to drop among later type stars.
Another reason to expect a decrease in the frequency of line effects going
from Herbig Be to the later Ae stars is that there appears to be a general
absence of Hα polarization changes in the even later type PMS T Tauri
stars (Bastien 1982; but see Sect. 1.6).
However, this turns out not to be the case at all. The number of
line effects in Herbig Ae stars is found to be particularly high: 9 out
of 11 Herbig Ae stars show a significant line effect (Vink et al. 2002).
XY Per is included here as an example, as represented in Figs. 1(b)
and 2(b): not only is there a line effect, it is noticeably different from
the depolarization behaviour seen in the Herbig Be stars. First, the drop
in the %Pol is not as broad as it is for the Herbig Be stars. Second, the
behaviour in PA is not smooth. Instead, a line-center flip of the PA is
clearly noticeable in the upper panel of the triplot in Fig. 1(b). This
PA rotation translates into a “loop” in the equivalent QU diagram of
Fig. 2(b).
The interpretation of these QU loops in Herbig Ae stars is still a
matter of ongoing investigation, but in the next section (Sect. 1.5), we
show that photons arising from a compact source of hot spots (a nat-
ural consequence of the magnetospheric accretion model) on the stellar
surface that are subsequently scattered in a rotating circumstellar disc
may explain the observed Hα spectropolarimetry data of Herbig Ae (and
possibly T Tauri stars; see Sect. 1.6).
5.Polarimetric Line Profiles from a Spotty Star
We employ a 3D Monte Carlo scattering model torus (Harries 2000)
to simulate a compact source of photons arising from two diametrically
opposed accretion hot spots onto a circumstellar disc. The disc has an
inner hole of 3 stellar radii, and is assumed to be flat. As the Hα emis-
sion is compact in this model, we may expect intrinsic line polarization
to occur. Figures 3(a) and 3(b) represent the two extreme cases for an
almost edge-on (LHS) and an almost pole-on (RHS) disc respectively.
As expected for an edge-on disc, there is a significant level of continuum
polarization of a few percent. Due to the asymmetry in the velocity field,