The close classical T Tauri binary V4046 Sgr: Complex magnetic fields & distributed mass accretion

Dip. di Fisica, Univ. di Palermo, Piazza del Parlamento 1, 90134 Palermo, Italy
Monthly Notices of the Royal Astronomical Society (Impact Factor: 5.11). 10/2011; 417(3):1747 - 1759. DOI: 10.1111/j.1365-2966.2011.19366.x
Source: arXiv
We report here the first results of a multi-wavelength campaign focusing on magnetospheric accretion processes within the close binary system V4046 Sgr, hosting two partly convective classical T Tauri stars of masses ≃0.9 M⊙ and age ≃12 Myr. In this paper, we present time-resolved spectropolarimetric observations collected in 2009 September with ESPaDOnS at the Canada–France–Hawaii Telescope (CFHT) and covering a full span of 7 d or ≃2.5 orbital/rotational cycles of V4046 Sgr. Small circularly polarized Zeeman signatures are detected in the photospheric absorption lines but not in the accretion-powered emission lines of V4046 Sgr, thereby demonstrating that both system components host large-scale magnetic fields weaker and more complex than those of younger, fully convective classical T Tauri stars (cTTSs) of only a few Myr and similar masses.
Applying our tomographic imaging tools to the collected data set, we reconstruct maps of the large-scale magnetic field, photospheric brightness and accretion-powered emission at the surfaces of both stars of V4046 Sgr. We find that these fields include significant toroidal components, and that their poloidal components are mostly non-axisymmetric with a dipolar component of 50–100 G strongly tilted with respect to the rotation axis; given the similarity with fields of partly convective main-sequence stars of similar masses and rotation periods, we conclude that these fields are most likely generated by dynamo processes. We also find that both stars in the system show cool spots close to the pole and extended regions of low-contrast, accretion-powered emission; it suggests that mass accretion is likely distributed rather than confined in well-defined high-contrast accretion spots, in agreement with the derived magnetic field complexity.


Available from: Antonio Maggio
arXiv:1109.2447v1 [astro-ph.SR] 12 Sep 2011
Mon. Not. R. Astron. Soc. 000, 000–000 (0000) Printed 13 September 2011 (MN L
X style file v2.2)
The close classical T Tauri binary V4046 Sgr: Complex
magnetic fields & d i stributed mass accretion.
J.-F. Donati
, S.G. Gregory
, T. Montmerle
, A. Maggio
, C. Argiro
G. Sacco
, G. Hussain
, J. Kastner
, S.H.P. Alencar
, M. Audar d
, J. Bouvier
F. Damiani
, M. G¨udel
, D. Huenemoerder
, G.A. Wade
IRAP–UMR 5277, CNRS & Univ. de Toulouse, 14 Av. E. Belin, F–31400 Toulouse, France
California Institute of Technology, MC 249-17, Pasadena, CA 91125, USA
Institut d’Astrophysique de Paris, 98bis Bd Arago, F–75014 Paris, France
IPAG–UMR 5274, CNRS & Univ. J. Fourier, 414 rue de la Pisc ine, F–38041 Grenoble, France
INAF, Osservatorio Astronomico di Palermo, Piazza del Parlamento 1, 90134 Palermo, Italy
Dip. di Fisica, Univ. di Palermo, Piazza del Parlamento 1, 90134 Palermo, Italy
Center for Imaging Science, Rochester Institute of Technology, 54 Lomb Memorial Drive, Rochester, NY 14623, USA
ESO, Karl-Schwarzschild-Str. 2, D-85748 Garching, Germany
Departamento de F`ısica ICEx U FMG, Av. Antˆonio Carlos, 6627, 30270-901 Belo Horizonte, MG, Brazil
ISDC Data Center for Astrophysics, University of Geneva, Ch. d’Ecogia 16, 1290 Versoix, Switzerland
Department of Astronomy, Universi ty of Vienna, Trkenschanzstr. 17, 1180 Vienna, Austria
Massachusetts Institute of Technology, Kavli Institute for Astrophysics and Space Research, 70 Vassar St., Cambridge, MA 02139, USA
Department of Physics, Royal Military College of Canada, PO Box 17000, Station Forces, Kingston, Ontario K7K 7B4, Canada
2011 April, MNRAS submitted
We report here the first results of a multi-wavelength campaign focussing on mag-
netospheric accretion process es within the c lose binary system V4046 Sgr, hosting
two partly-convective classical T Tauri stars of masses 0.9 M
and age 12 Myr. In
this paper, we present time-resolved spectropolarimetric observations collected in 2009
September with ESPaDOnS at the Canada-France-Hawaii Telescope (CFHT) and cov-
ering a full span of 7 d or 2.5 orbital/rotational cycles of V4046 Sgr. Small circularly
polarised Zeeman signatures are detected in the photospheric absor ption lines but
not in the accretion-powered emission lines of V4046 Sgr, thereby demonstrating that
both system components host large-scale magnetic fields weaker and more complex
than those of younger, fully-convective cTTSs of only a few Myr and similar masses.
Applying our tomographic imaging tools to the collected data set, we reconstruct
maps of the lar ge-scale magnetic field, photospheric brightness and accretion-powered
emission at the surfa ces of both stars of V4046 Sgr. We find that these fields include
significant tor oidal components, and tha t their poloidal components are mostly non-
axisymmetric with a dipolar component of 50-100 G strongly tilted with respect to
the rotation axis; given the similarity with fields of partly-convective main-se quence
stars of similar masses and rotation periods, we conclude that these fields are most
likely generated by dynamo processes. We also find that both stars in the system show
cool spots c lose to the pole and extended regions of low-contrast, accretion-powered
emission; it suggests that mass accretion is likely dis tributed rather than confined in
well defined high-contrast accre tion spots, in agreement with the de rived magnetic
field complex ity.
Key words: stars: magnetic fields stars: formation stars: imag ing stars: rotation
stars: binary stars: individual: V4046 Sgr techniques: polarimetric
From the collapse of giant molecular clouds to their frag-
mentation into ind ividual stars and their planetary systems,
Page 1
2 J.-F. Donati et al.
from the dissipation of the initial cloud angular momen-
tum content to the generation of outflows and jets, mag-
netic elds have a strong impact on most physical processes
involved in the formation of stars and planets and are n ow
clearly recognized as one of the few main ingredients (along
with gravitation and turbulence) in Nature’s recipe to build
new worlds (e.g., And r´e et al. 2009; Donati & Landstreet
2009). In the latter formation phases in particular, low-mass
Sun-like stars young enough to be still surrounded by, and
accreting mass from, a gaseous disc (the classical T Tauri
stars or cTTSs) are apparently capable of generating and
sustaining a large-scale magnetic field through dynamo pro-
cesses, which in turn manages to disrupt the inn er disc re-
gions and drastically brake the rotation of the central pro-
tostar (see, e.g., Bouvier et al. 2007a, for a review).
The presence of intense magnetic elds at the surfaces
of cTTSs, and more generally of all T Tauri stars, was
first derived through indirect proxies (i.e., continuum or line
emission throughout the whole electromagnetic spectrum,
from the radio to X-rays, e.g., Feigelson & Montmerle 1999;
Favata & Micela 2003; udel & Naz´e 2009, for reviews),
then demonstrated directly (through the Zeeman broaden-
ing of spectral lines) about 2 decades ago ( e.g., Johns-Krull
2007, for an overview). However, the actual large-scale mag-
netic topologies of cTTSs - a crucial parameter to eluci-
date the way such elds manage to couple the protostars
to their discs, t o funnel th e accreted disc material to the
stellar surface, and to strongly brake the rotation of the
protostar - remained rather elusive until recently. Through
sp ectropolarimetric observations consisting of time-series of
Zeeman signatures from accreting an d non-accreting regions
at t he surfaces of cTTSs, new studies demonstrated that
very young low-mass stars indeed host large-scale fi elds of
dynamo origin, whose topologies are reminiscent of those of
main-sequence stars (once differences in the internal stellar
structures are taken into account).
This new opportunity offers the option of investigat-
ing magnetospheric accretion processes of cTTSs in a much
more global and consistent way, by carrying out simultane-
ously observations of selected prototypical cTTSs with as
wide a spectral coverage as possible, and including in par-
ticular the X-ray and optical domains. While optical lines
provide information on the large-scale magnetosphere and
on the location and rate at which mass is accreted from
the disc to the surface of the protostar, high resolution X-
ray spectra (and in particular the softer component) yields
key material on the shock that occurs when th e accretion
flows collide with the high atmosphere of the protostar. The
present study takes place in the framework of an interna-
tional multi-wavelength, multi-site campaign organised on
V4046 Sgr, one of the few known close cTTS binaries and one
of the best cases where the softer component of th e X-ray
sp ectrum (attribut ab le to magnetospheric accretion shocks)
can be recorded with sufficient accuracy (G¨unther et al.
2006). This campaign was t riggered by a Large Program
with XMM-Newton, aimed at obtaining phase-resolved X-
ray observations of V4046 Sgr covering about 2 consecutive
orbital/rotation cycles of the cTTS binary (for a total ex-
posure time of 370 ks).
Spectropolarimetric (optical) observations of V4046 Sgr
were also scheduled in conjunction with (though slightly be-
fore than) the main X-ray p rogram, and were carried out in
a way similar to those achieved in the framework of the inter-
national MaPP (Magnetic Protostars and Planets) project.
Through a survey of a dozen cTTSs, MaPP investigates how
the large-scale magnetic fields of low-mass protostars de-
pend on stellar parameters such as mass, age, rotation and
accretion rates (e.g., Donati et al. 2010, 2011a); MaPP also
includes a th eoretical component aiming at describing con-
sistently (through analytical modelling and numerical sim-
ulations) how magnetic fields of cTTSs couple to their sur-
rounding accretion disc, how they channel accretion into dis-
crete funnels and what is the resulting angular momentum
evolution (see, e.g., Gregory et al. 2010, for an extensive re-
view on th e subject). At this stage, however, MaPP has
only focussed on single (or distant binary) stars and has
not addressed the specific issue of magnetospheric accre-
tion processes in close binary stars; the present campaign
on V4046 Sgr thus appears as a worthwhile and timely com-
plement to the core MaPP program.
The complete results of the V4046 Sgr project will
be published in a series of papers, starting with a global
overview of th e campaign and a sum mary of the main X-
ray results (Argiroffi et al. 2011, submitted). The present
paper belongs to this series and aims at obtaining, for
both cTTSs forming the close V4046 Sgr binary, observa-
tional constraints similar to those recently derived for sin-
gle cTTSs; in particular, it aims at simultaneously deriving
the large-scale magnetic fields of both stars of V4046 Sgr,
their photospheric brightness maps and t heir chromospheric
distributions of accretion-induced hot spots. Companion pa-
pers (e.g., Maggio et al. 2011, Sacco et al. 2011, in prepara-
tion) describe ad ditional data sets collected during the same
After summarising the main characteristics of the close
V4046 S gr cTTS binary (Sec. 2), we describe the new spec-
tropolarimetric observations we collected and outline the
associated temporal variability and rotational modulation
(Secs. 3 and 4). We then d etail the modelling of these data
with our magnetic imaging code (Sec. 5) and briefly compare
our results with those already published for single cTTSs
and with the predictions of the latest numerical simulations
of accretion from circumbinary discs in close cTTS binaries
(Sec. 6).
2 V4046 SGR (HDE 319139, HBC 662, AS 292)
Located at a distance of about 73 pc, V4046 Sgr is a likely
member of the nearby loose β Pic association (Torres et al.
2008). It is one of t he few known close cTTS binaries, show-
ing signs of accretion typical of single cTTSs (e.g., large
and variable Hα emission). It is a doub le line spectroscopic
binary with a circular orbit, a rather short orbital period
(2.42 d) and a prominent excess at IR and radio fre-
quencies indicating the presence of a circumbinary accretion
disc (with an inner radius of 0.2–0.4 AU and extending out
to several hundred AUs, Quast et al. 2000; Rodriguez et al.
2010). The photometric and spectroscopic periods are equal
and demonstrate that the sy stem is synchronised, as ex-
pected for a binary system as close as V4046 Sgr. Given
that V4046 Sgr has already completed both circularisation
and synchronisation, we can logically assume that the or-
Page 2
Magnetism & accretion of V4046 Sgr 3
Figure 1. Observed (open squares and error bars) location of the
primary (red) and secondary (green) components of V4046 Sgr
in the HR diagram, as derived from their respective tempera-
tures and radii (see text). The best match to these positions
when further imposing a primary-to-secondary mass ratio of 1.06
(as derived from the velocity curves) is also indicated (open cir-
cles). The PMS evolutionary tracks and corresponding isochrones
(Siess et al. 2000) assume solar metallicity and include convective
bital and rotation axes are aligned, following predictions of
the tidal theory (e.g., Hut 1981; Zahn & Bouchet 1989).
Two different detailed spectroscopic studies of
V4046 Sgr are available from the literature (Quast et al.
2000; Stempels & Gahm 2004). From the photometric
colors, velocity amplitudes and the wavelength-dependent
primary-to-secondary magnitude contrast, these studies
show in particular that both system components have
similar masses and surface temperatures of about 0.9 M
and 4250 K, the primary being about 6% more massive and
250 K hotter than the secondary. The system is seen mostly
pole-on, with an angle between the orbital axis and the line
of sight of 35
(Quast et al. 2000; Rodriguez et al. 2010).
The line-of-sight-projected equatorial rotation velocities
v sin i that we estimate for both system components (equal
to 13.5 ± 0.5 and 12.5 ± 0.5 km s
for the primary and sec-
ondary stars resp ectively) agree with previous estimates and
indicate respective radii of 1.12 ± 0.05 and 1.04 ± 0.05 R
The position of both stars in the HR diagram (as derived
from the above temperatures and radii) an d their relative
locations with respect to the theoretical evolutionary t racks
and iso chrones of Siess et al. (2000), shown in Fig. 1,
suggest an age of 15 Myr; this is in agreement with recent
age estimates of the β Pic moving group as a whole (ranging
from 10 to 20 Myr, e.g., Mentuch et al. 2008; da Silva et al.
2009; Yee & Jensen 2010) and in p articular with the widely
accepted and quoted age of 12 Myr (Torres et al. 2008). It
indicates that V4046 Sgr is approaching the end of the TTS
stage; at this age, both stars should be partly convective,
with the convective zone occupying the ou ter 50% of the
stellar radius (Siess et al. 2000).
The distance between both components as derived from
the semi-amplitude of the velocity curves (found to be
51.4 ± 0.2 and 54.3 ± 0.2 km s
from our data) is 8.8 R
0.041 AU, in good agreement with the results of Quast et al.
(2000). However, the system velocity that we derive (5.7±
0.2 km s
, see Sec. 5) is significantly different from the for-
mer estimate (e.g., 6.94 ± 0.16 km s
Quast et al. 2000)
and suggests th at the system also hosts a third distant com-
ponent. The very precise radial velocity (RV) of the cir-
cumbinary (CO disc) material (equal to 6.21±0.01 k m s
in the heliocentric frame, Rodriguez et al. 2010) confirms
that this is likely the case.
X-ray spectra of V4046 Sgr with Chandra-HETGS
clearly show He-like triplets with line strength ratios in
agreement with the predictions of magnetospheric accre-
tion models, confirming that V4046 Sgr is still accreting
mass from the surrounding disc (G¨unther et al. 2006 ). From
the equivalent widths and the corresponding line fluxes of
the optical emission lines usually considered as good ac-
cretion proxies (and in p articular the He i D
line and
the Ca ii infrared triplet, I RT) and using the pub lished
empirical correlations between lines and accretion fluxes
(Fang et al. 2009), we can derive an estimate of the average
logarithmic mass accretion rate (in M
) at the sur-
face of each star of V4046 Sgr, that we find to be equal to
9.3 ± 0.3 (assuming both stars have equal accretion rates)
in agreement with independent estimates from optical p rox-
ies (Curran et al. 2011). As usual, this is larger than the
estimate derived from X-ray data (equal to 9.84 ± 0.11 for
V4046 Sgr, Curran et al. 2011). Optical veiling, i.e., the ap-
parent weakening of the photospheric spectrum (presumably
caused by accretion) is apparently weak for V4046 Sgr (e.g.,
Stempels & Gahm 2004).
V4046 Sgr shares similarities with another memb er of
the β Pic moving group, the young pre-main-sequence close
binary HD 155555 (Dunstone et al. 2008; Yee & Jensen
2010), whose stars are slightly more massive (1.1–1.2 M
larger (1.3 R
), warmer (5500 K) and closer to each
other (0.036 AU); both systems are thus structurally alike
though different regarding their accretion prop ert ies and re-
lated issues, HD 155555 having dissipated its accretion d isc
already. HD 155555 may thus serve as a convenient check
to identify what in V4046 Sgr is typical of young partly-
convective Sun-like binaries and what is more specific to
cTTS binaries.
Spectropolarimetric observations of V4046 Sgr were col-
lected from 2009 September 03 to 09 using ESPaDOnS on
the CFHT. ESPaDOnS collects stellar spectra spanning the
whole optical domain (from 370 to 1,000 nm) at a resolv-
ing power of 65,000 (i.e., 4.6 km s
) and with a spectral
sampling of 2.6 km s
, in either circular or linear polarisa-
tion (Donati 2003). A total of 8 circular polarisation spec-
tra were collected over a timespan of 7 consecutive nights.
All polarisation spectra consist of 4 indiv idual subexposures
each lasting 390 s and taken in different polarimeter con-
figurations to allow the removal of all spurious polarisation
signatures at first order.
All raw frames are processed with Libre ESpRIT,
a fully automatic reduction package/pipeline available at
CFHT. It automatically performs optimal extraction of ES-
PaDOnS unpolarized (Stokes I) and circularly polarized
(Stokes V ) spectra grossly following the procedure described
Page 3
4 J.-F. Donati et al.
Table 1. Journal of observations collected in 2009 September.
Columns 14 respectively list the UT date, the heliocentric Julian
date and UT time (both at mid-exposure), and the peak signal
to noise ratio (per 2.6 km s
velocity bin) of each observation
(i.e., each sequence of 4 × 390 s subexposures). Column 5 lists
the rm s noise level (relative to the unpolarized continuum level
and per 1.8 km s
velocity bin) in the circular polarization
profile produced by Least-Squares Deconvolution (LSD), while
column 6 indicates the orbital/rotational cycle associated with
each exposure (using the ephemeris given by Eq. 1).
Date HJD UT S/N σ
(2,455,000+) (h:m:s) (10
) (3336+)
Sep 03 77.77383 06:33:30 180 2.9 0.755
Sep 04 78.81648 07:35:02 170 2.6 1.186
Sep 05 79.72103 05:17:41 200 2.1 1.560
Sep 06 80.72065 05:17:14 200 2.1 1.972
Sep 06 80.81321 07:30:32 210 1.9 2.011
Sep 07 81.72152 05:18:36 160 2.5 2.386
Sep 08 82.72171 05:18:60 200 2.0 2.799
Sep 09 83.72107 05:18:11 210 1.8 3.212
in Donati et al. (1997). The velocity step corresponding t o
CCD pixels is about 2.6 km s
; however, thanks to the fact
that the spectrograph slit is tilted with respect to the CCD
lines, spectra corresponding to different CCD columns across
each order feature a different pixel sampling. Libre ES-
pRIT uses this opportunity to carry ou t optimal extraction
of each spectrum on a sampling grid denser than the origi-
nal CCD sampling, with a spectral velocity step set to about
0.7 CCD pixel (i.e. 1.8 km s
). All spectra are automati-
cally corrected of spectral shifts resulting from instrumen-
tal effects (e.g., mechanical exures, temperature or pres-
sure variations) using telluric lines as a reference. Though
not perfect, this procedure provides spectra with a relative
RV precision of better than 0.030 km s
(e.g., Donati et al.
The peak signal-to-noise ratios (S/N, per 2.6 km s
velocity bin) achieved on the collected spectra (i.e., the se-
quence of 4 subexposures) range between 160 and 210 de-
pending on weather/seeing conditions, with a median of 200.
Following Stempels & Gahm (2004), orbital/rotational
cycles E are computed from h eliocentric Julian dates ac-
cording to the eph emeris:
HJD = 2446998.335 + 2.4213459E (1)
We however note that t he time of first conjunction (with
the primary component in front) in our data is significantly
shifted with respect to the ephemeris predictions and oc-
curs at phase 0.681
instead of phase 0.75 (i.e., 0.069 cy cle
or ab ou t 4 hr ahead of time). The origin of this shift is
not fully clear yet. It can reflect a slightly overestimated
orbital period; if this is the case, the orbital period need s
to be updated to 2.421296±0.000001 d, i.e., a value smaller
than those of Quast et al. (2000) and of Stempels & Gahm
(2004) by 9 and 12 of their σ (0.000004 d) respectively. It
may also potentially reflect (at least partly) temporal fluc-
The epoch of first quadrature as derived from our observations
occurs on HJD = 2455078.199±0.001 d, with one of our spectra
collected very shortly after second conjunction (on Sep 04).
Figure 2. LSD circularl y-polarized (Stokes V ) and unpolarized
(Stokes I) profiles of V4046 Sgr (top/red, b ottom/blue curves
resp ectively) coll ected on 2009 September 06 (cycle 1.972), i.e.,
slightly after first quadrature (occurring at phase 0.931, see
Sec. 3). Zeeman signatures are clearly detected in the LSD profiles
of both the primary (right) and secondary (left) system compo-
nents. The mean polarization profile is expanded by a factor of
20 and shifted upwards by 1.03 for display pur poses.
tuations of the orbital period, either caused by a light-time
effect (e.g., that induced by the distant third body causing
the RV changes mentioned in Sec. 2) and/or by changes in
the quadrupolar moments of both components of V4046 Sgr
(resulting from activity cy cles of one or both system compo-
nents and known as the Applegate effect, Applegate 1992;
Lanza 2006). This issue will be specifically addressed in more
details in a forthcoming paper.
Least-Squares Deconvolution (LSD, Donati et al. 1997)
was applied to all observations. The line list we employed
for LSD is computed from an Atlas9 LTE model atmo-
sphere (Kurucz 1993) and corresponds to a K5V spectral
type (T
= 4, 250 K and log g = 4.5) appropriate for
V4046 Sgr. Only moderate to strong atomic spectral lines
(with line-to-continuum core depressions larger than 40%
prior to all non-thermal broadening) are included in this list;
sp ectral regions with strong lines mostly formed outside the
photosphere (e.g., Balmer, He, Ca ii H, K and IRT lines)
and/or heavily crowded with telluric lines were discarded.
Altogether, about 8,000 spectral features (with about 40%
from Fe i) are used in this process. Expressed in units of the
unpolarized continuum level I
, the average noise levels of
the resulting Stokes V LSD signatures are ranging from 1.8
to 2.9×10
per 1.8 km s
velocity bin.
The full journal of observations is presented in Table 1.
Before applying ou r tomographic imaging tools to the
V4046 Sgr data, it is usually worthwhile to examine how
sp ectral lines (and in particular the equivalent widths, RVs
and the average magnetic fluxes) vary with rotation cycle
throughout the observing run; at the very least, it provides
a rough, intuitive understanding of how the large-scale field
is structured and oriented, and where the cool photospheric
Page 4
Magnetism & accretion of V4046 Sgr 5
Figure 3. Rotational modulation of the RV (top row), residual RV (second row), equivalent width (third row) and longitudinal field
(bottom row) derived from LSD photospheric profiles (left panels) and the average Ca ii IRT li ne (right panels) and for both the primary
(red circles) and secondary (green squares) system components of V4046 Sgr. First and second orbital conjunctions respectively occur
at phases 0.681 and 0.181 in the ephemeris used here (see Eq. 1). Data collected at near-conjunction epochs and at which spectral
contributions of both components overlap and cannot be easily separated from one another (i.e., cycles 1.186 and 3.212) were excluded
from this plot. Fits with sine/cosine waves are included (and shown as dashed lines) to outline the amount of variability attributable to
rotational modulation (whenever significant), but are not used in the subsequent imaging process. ±1 σ error bars on data points, based
on photon noise only, are also shown whenever larger than symbols.
and hot accretion spots are likely to be located - to be com-
pared later on with the detailed results of the full imag-
ing process (described in the next section). This is what
we present in this section; note that the 2 spectra recorded
near second conjunction (at cycles 1.186 and 3.212) were
excluded from part of this preliminary analysis, since the
contributions of both system components cannot be easily
separated from one another.
The main modulation observed in LSD profiles of pho-
tospheric lines of V4046 Sgr is the RV shifts resulting from
Page 5
6 J.-F. Donati et al.
the orbital motion of both system components; Fig. 2 shows
an example profile recorded at cycle 1.972, i.e., slightly after
the first quadrature (occurring at phase 0.931, see Sec. 3)
and when both components are clearly separated from one
another, while Fig. 3 (top left panel) shows the phased RV
curves of both components. The orbital elements derived
from these profiles (as a by-product of the imaging proce-
dure described in Sec. 5) agree with those of Quast et al.
(2000) (except for the significant shift in the phase of first
conjunction, see Sec. 3, an d for the change in th e system
velocity, see Sec. 2) but do not support the newer estimates
of Stempels & Gahm (2004).
Zeeman signatures are detected at all times in Stokes
V LSD profiles and in close association with the spectral
lines of both components (see Fig. 2 for an example). In
most spectra, Stokes V signatures are complex, with a typ-
ical peak-to-peak amplitudes of 0.2%; they feature several
reversals throughout the line profile, suggesting that the par-
ent eld topology is not simple. The line-of-sight projected
compon ent of the eld averaged over the visible stellar hemi-
sphere and weighted by brightness inhomogeneities (called
the longitudinal field and estimated from the first moment of
the Stokes V profile, e.g., Donati et al. 1997) is rather weak
(always smaller than 70 G and most of the time smaller
than 50 G, with typical error bars of 10 G) and poorly
informative about the large-scale fi eld (see Fig. 3 lower left
LSD Stokes I profiles also show a small level of rota-
tional modulation (in ad dition to the RV shifts caused by or-
bital motion
) as a likely consequence of the presence of cool
sp ots on the stellar surface of both system components (also
causing the reported photometric modulation). The line
equivalent widths of both system components are observed
to vary in anticorrelation and directly reflect the fluctua-
tions in the secondary to primary brightness ratio (around
a m ean of about 0.64, see Fig. 3 third left panel); the primary
star is brightest and/or th e secondary star is faintest by a
few % around the first orbital conjunction (i.e., phase 0.681)
while the opposite holds around the second orbital conjunc-
tion (i.e., phase 0.181). Once corrected from the spectral
dilution of the companion, the LSD profiles of both system
compon ents show equivalent widths of about 3.9 km s
fully compatible with those of non-accreting young stars of
similar spectral types (e.g., V410 Tau); this demonstrates
that veiling was weak (i.e., <5%) for V4046 Sgr at the time
of our observations.
The residual RV s of both system components (in the
rest frame of each star, i.e., with the orbital motion removed)
is also showing a low-amplitude modulation of < 1 km s
peak to peak, see Fig. 3 second left panel). As the v sin i’s
are much larger t han the amplitude of this modulation, this
suggests in particular the presence of high-latitude spots,
around phase 0.2 on the primary star and phase 0.7 on
the secondary star, in full agreement with the conclusions
The RV shifts of spectral lines resulting from the orbital motion
of both system components were estimated as part of the imaging
process (see Sec. 5) and are thus free of contamination fr om cool
surface spots (as opposed to those of Quast et al. 2000); they are
thus the most logical ones to use when looking for potential RV
signatures of cool surface spots.
Figure 4. Double Lorentzian fit (bottom blue line) with addi-
tional double Gaussian fit (top green line) to the observed mean
IRT line profile of V4046 Sgr (red circles) on 2009 Sep 06 (cy-
cle 1.972). The bottom blue line is the background pr ofile model
that we subtract fr om the observed profile to recover the emission
profiles of both system components.
derived from the rotational modulation in t he equivalent
widths of LSD profiles of both compon ents (see above).
We note that LSD Stokes I and V profiles of both
system components repeat well from one rotation cycle to
the next, furth er supporting the conclusion that rotational
mod ulation largely dominates over intrinsic variability in
V4046 Sgr, even more than in the moderately accreting
cTTS V2129 Oph (Donati et al. 2011a).
As in previous papers, we use core emission in Ca ii
IRT lines as a proxy of surface accretion even though it
usually features a significant (and often dominant) contri-
bution from the non-accreting chromosphere. Previous work
(Donati et al. 2010, 2011a) demonstrated in particular that
rotational modulation in Stokes I and V Ca ii IRT profiles
correlates well with t hat derived from the more conventional
He i D
accretion proxy, even for mass accretion rates as low
as (or even lower than) that of V4046 Sgr (see below). In
the particular case of V4046 Sgr, the emission contributions
of both system components are narrow enough to be well
separated in velocity during most of the orbital cycle, of-
fering the option of characterizing the accretion behaviour
of both stars indep endently from one another. In practice,
we start by constructing a LSD-like weighted average of the
3 IRT lines; we then subtract the underlyin g (much wider)
photospheric absorption profiles, with a single Lorentzian fit
to the far line wings when the emission peaks of both sys-
tem components overlap (i.e., at cycle 1.186 and 3.212, see
Sec. 5), and with a double Lorentzian fit to the far wings
and the line center when the emission peaks of b oth system
compon ents are well separated in velocity (see Fig. 4 for an
The RVs of the emission peaks are fully compatible
with the orbital solution derived from photospheric lines
(see Fig. 3 top right panel). In th e stellar rest frame of
each system component, both emission peaks are slightly
shifted redwards by about 0.7 km s
(see Fig. 3 second
right panel), as usual in cTTSs and confirming that it traces
Page 6
Magnetism & accretion of V4046 Sgr 7
slowly-moving regions of the p ost- shock accretion funnels
close to the surface of t he star. The equ ivalent widths of
both emission peaks are roughly constant with phase and
equal to 15 km s
or 0.040 nm (once corrected from the
sp ectral dilution) and vary by less than 2 km s
over the
rotation cycle (see Fig. 3 third right panel). It suggests a
rather low level of accretion and accretion regions either dis-
tributed over the whole visible hemisphere or located very
close to the pole (hence producing a very small level of mod-
ulation); this is further evidenced by the low-amplitude RV
mod ulation of the emission peaks of both components (once
the orbital motion is removed, see Fig. 3 second right panel).
We detect no Zeeman signatures in conjunction with
Ca ii IRT emission, with error bars on the correspondin g
longitudinal elds most of the time smaller than 100 G
(see Fig. 3 bottom right panel). This is fairly unusual in
cTTSs, even in those with low mass accretion rates (e.g.,
Donati et al. 2011a) and suggests that the large-scale fields
of both system stars have rather weak dipole components.
We nevertheless use the corresponding Stokes V profiles as
an additional constraint for the magnetic modelling (see
Sec. 5).
Although usually considered as the most reliable ac-
cretion proxy, emission in t he He i D
line at 587.562 nm
(not shown here) is hardly usable in the particular case of
V4046 Sgr. This is mostly th e consequence of the relative
weakness of this line in the spectrum of V4046 Sgr and of its
significant width (with the contribution of both components
overlapping most of the time) - all of this occurring within
a relatively crowded spectral region. The best we can ob-
tain is that the total line equivalent width (for both system
compon ents as a whole), equal t o 17.5 km s
or 0.034 nm
in average, is roughly constant with phase throughout the
orbital cycle (in agreement with what the Ca ii IRT lines
show) except for a (presumably sporadic) emission burst
detected at cycle 2.386 (also detected in Balmer lines, see
below, but not in Ca ii IRT lines). No Zeeman signatures are
detected in association with the He i D
, consistent with the
non-detection in Ca ii IRT lines.
Balmer lines are strongly in emission in V4046 Sgr, with
widths large enough to blend the contribution of both sys-
tem components into a single profile (at least for the first few
lines of the series). In the particular case of Hα and Hβ (see
Fig. 5), the total line equivalent widths of the time-averaged
profiles are equal to 2,600 km s
(5.7 nm) and 475 km s
(0.77 nm), in line with previous reports (Quast et al. 2000;
Stempels & Gahm 2004). The profiles are clearly variable
with time but are dominated by intrinsic fluct uations and
poorly repeat between consecutive rotational cycles (e.g.,
cycle 1.186 and 3.212). During our run, we observed in par-
ticular a strong short-lived, flare-like burst of emission (e.g.,
at cycle 2.386), with a conspicuous red-sh ifted emission com-
ponent at velocities of up to 350 km s
suggesting that th is
event may reflect a short episode of enhanced mass accre-
tion from the disc to the stars. This emission burst is also
detected in the He i D
line but not in the Ca ii IRT nor
in the LSD profile of photospheric lines (see Fig. 3 third
panels). The width of Balmer lines (200 km s
for the
full width at half maximum of Hα, Quast et al. 2000, and
400 km s
for the full width at 10% height, see below)
suggests that it traces free-falling material accelerated from
large distances to the star, possibly from the inner rim of
the circumbinary disc.
Although small, we note potential signatures of rota-
tional modulation in both Hα and Hβ, like for instance
the recurrent pseudo-absorption episode detected in the red
wing of both lines, at a mean velocity of abou t 200 km s
and close to phase 0.8 (at cycles 0.755 and 2.799). Similar
(though usually d eeper) absorption events are commonly ob-
served in Balmer lines of cTTSs (e.g., Bouvier et al. 2007b;
Donati et al. 2008b, 2010, 2011a) and are usually interpreted
as evidence for accretion veils an chored within the surround-
ing disc and periodically intersecting the line of sight as the
star rotates. In the particular case of V4046 Sgr h owever,
these episodes are only tracing pseudo-absorption (i.e., a
lack of emission with respect to the mean profile) rather
than true absorption (below th e continuum level) and could
also potentially be attributed to projection effects in a rotat-
ing non-axisymmetric magnetospheric accretion structure.
Data with denser phase coverage are necessary to address
this issue in more details.
From the average equivalent widths of the Ca ii IRT,
He i and Hβ lines, we derive logarithmic line fluxes (with
respect to the luminosity of the Sun L
) for each star equal
to 5.0, 5.0 and 3.8 respectively
, implying logarithmic
accretion luminosities (with respect to L
) of 2.2, 2.0
and 1.9 respectively (using empirical correlations from
Fang et al. 2009). As estimates from the two main accre-
tion proxies (Ca ii IRT and He i D
lines) agree with each
other and with the overall mean, we can safely conclude that
the average logarithmic accretion luminosity of each compo-
nent of V4046 Sgr is 2.0 ± 0.3 and thus that the average
logarithmic mass accretion rate of each star in V4046 Sgr (in
) is equal to 9.3 ± 0.3, with only limited fluctua-
tions around this mean. Our estimate is in good agreement
with that of Curran et al. (2011), equal to 9.22 ± 0.23 and
also derived from optical lines (though from fully indepen-
dent data an d a different analysis).
Mass accretion rates can in principle also be estimated
(though less accurately) through the full width of Hα at
10% height (e.g., Natta et al. 2004; Cieza et al. 2010). In the
case of V4046 Sgr, Hα shows a full width of 420 k m s
average, implying a logarithmic mass accretion rate estimate
of 8.8 ± 0.6 (in M
). This is larger (though not very
significantly) than the estimate derived from emission uxes
and in agreement with the results of Curran et al. (2011); at
this stage, the origin of this potential discrepancy is unclear.
To derive line fluxes from normalized equivalent widths, we ap-
proximate the continuum level by a Pl anck f unction at the tem-
perature of the stellar photosphere. Results are found to be com-
patible with those in the published literature (e.g., M ohanty et al.
2005) within better than 0.1 dex. In the particular case of
V4046 Sgr, we further assume that both stars exhibit more or less
the same emission flux, i.e., that the total emission fluxes mea-
sured in the spectra reflect (within a factor of 2) that of each
system component (in rough agreement with what Ca ii IRT lines
show, see Sec 5).
Page 7
8 J.-F. Donati et al.
Figure 5. Variation of the Hα (left) and Hβ (right) lines of V4046 Sgr. To emphasize variability, the average profile over the run is
shown in red. Orbi tal/rotation cycles (as listed in Table 1) are mentioned next to each profile.
5.1 Overview of the method
As for the previous papers, we aim at recovering the large-
scale field topology of V4046 Sgr, as well as the distribu-
tion of surface cool spots and of chromospheric accretion
regions, from the collected set of LSD and Ca ii IRT pro-
files presented and described in Secs. 3 and 4. We again
apply our new modelling technique (detailed extensively in
Donati et al. 2010, 2011a), with t he essential difference that
we are now aiming at mapping the two stars of the binary
system simultaneously.
Following the principles of maximum entropy, our
code automatically and simultaneously derives the sim-
plest magnetic top ologies, photospheric brightness images
and accretion-powered Ca ii emission maps compatible with
the series of rotationally modulated Stokes I and V LSD
and Ca ii IRT profiles. The reconstruction process is itera-
tive and proceeds by comparing at each step the synthetic
Stokes I and V profiles corresponding t o the current im-
ages with those of the observed data set. The magnetic eld
of each star is described through its poloidal and toroidal
compon ents expressed as spherical-harmonics (SH) expan-
sions (Donati et al. 2006). The spatial distributions of pho-
tospheric brightness (with respect to the q uiet photosphere)
and those of accretion-p owered Ca ii emission (in excess of
and with resp ect to that produced by the quiet chromo-
sphere) are mo delled as series of independent pixels (typi-
cally a few t housand) on a grid covering the visible surfaces
of the stars (with spots in the brightness images assumed to
be darker/cooler than the quiet photospheres and spots in
the accretion-powered Ca ii emission maps supposed to be
brighter than the quiet chromospheres).
Synthetic profiles are computed by summing up the el-
ementary spectral contributions from all image pixels over
the visible hemispheres of both stars, taking into account
the relevant local parameters of the correspond in g grid cells
(e.g., brightness, accretion-powered excess emission, mag-
netic eld strength and orientation, radial velocity, limb an-
gle, projected area). Since the problem is partly ill-posed, we
stabilise the inversion process by using an entropy criterion
(applied t o the SH coefficients and to the brightness/excess
emission image pixels) aimed at selecting the field topolo-
gies and images with minimum information among all those
compatible with the data. The relative weights attributed
to the various SH modes can be imposed, e.g., for purposely
producing antisymmetric or symmetric field topologies with
respect to the centre of the star (by favouring odd or even
SH modes, Donati et al. 2007, 2008b ). More details concern-
ing the specific description of local profiles used in the model
can be found in Donati et al. (2010).
5.2 Modelling V4046 Sgr
Given the relatively low level of intrinsic variability in the
LSD photospheric and Ca ii IRT emission lines of V4046 Sgr
(see Sec. 4), we applied our imaging model directly to the
original profiles without passing them t hrough our usual fil-
Page 8
Magnetism & accretion of V4046 Sgr 9
tering procedure (aimed at retaining rotational modulation
only, see, e.g., Donati et al. 2010, 2011a). We assume that
i = 35 (see Sec. 2), with values of i ranging from 30
to 40
yielding virtually identical results. We further assume that
both stars rotate as solid bodies given the relatively lim-
ited coverage and sparse sampling of our data set; assuming
Sun-like differential rotation (both in sign an d strength, i.e.,
inducing max imum phase delays of ±3% for the equator and
pole with respect to the orbital cycle over the full time span
of our observations) generates almost identical images, as
expected from the limited spatial resolution available (see
The parameters describing the local line profiles on both
stars of V4046 Sgr are assumed to be very similar to those
used in the previous studies (Donati et al. 2010, 2011a). In
particular, the emission profile scaling factor ǫ, describing
the emission enhancement of accretion regions over the quiet
chromosphere, is again set to ǫ = 10. The local filling factor
ψ, describing the relative proportion of magnetic areas at
any given point of the stellar surface and set here to ψ = 1,
has negligible impact on the result given the weak fields
detected on V4046 Sgr.
The magnetic, brightness and accretion maps we recon-
struct for both stars of V4046 Sgr are shown in Fig. 6, with
correspondin g fits to the data shown in Fig. 7. The SH ex-
pansion describing the field was limited to terms with 5;
first attempts with = 8 indicate that little power is recon-
structed in higher order modes ( 6), reflecting essentially
the limited spatial information accessible to Doppler tomog-
raphy at moderate v sin i values. For this modelling, we as-
sumed (as in previous studies) that the fields of both stars
are antisymmetric (with respect to the stellar centres); t his
assumption however has little impact on the reconstructed
magnetic maps over the visible regions of the stellar sur-
faces. Our ts reproduce the data down to the noise level,
i.e., to a unit reduced chi-square χ
starting from an initial
of 3.2.
As a by-product, we obtain accurate estimates of the
system orbital elements, and in particular of the phase
of first conjunction φ
= 0.6809 ± 0.0004, of t he veloc-
ity amp litude of the primary and secondary components
= 51.4 ± 0.2 and K
= 54.3 ± 0.2 km s
, and of the sys-
tem radial velocity γ = 5.7 ± 0.2 km s
. These estimates
are in agreement with those of Quast et al. (2000) except for
the latter (γ) whose significant variation suggests the pres-
ence of a third body in the system. Note in particular that
our estimates of K
and K
are mostly free of contamination
from cool surface spots (as opposed to those of Quast et al.
2000) since they are derived as part of t he imaging process;
we su ggest that the small difference between our estimates
and those of Quast et al. (2000) (about 0.5 km s
) mostly
reflects the RV signatures of cool surface spots (see Sec. 4
and Fig. 3 left panel second row).
We also find that Ca ii lines are redshifted by about
0.7 k m s
with respect to rest frames of both stars, a typi-
cal value for cTTSs. In addition, we derive estimates of the
projected equatorial rotation velocities v sin i for both stars,
equal to 13.5± 0.5 and 12.5±0.5 km s
for the primary and
secondary component respectively, as well as a measurement
of the magnitude contrast between both system components
at the average wavelengths of the LSD an d Ca ii IRT pro-
files (620 and 850 nm), equal to 0.52 ± 0.10 and 0.22 ± 0.15
5.3 Modelling results
The reconstructed large-scale magnetic topologies of both
cTTSs in the V4046 Sgr system are weak and multipolar,
as expected from the low amplitude and complex shape of
the detected Zeeman signatures. On the primary star, the
magnetic field has an average surface intensity of 230 G,
with the azimuthal field compon ent dominating the others
and reaching maximum strength (500 G) in conspicuous
arc-like structures around the pole; as a consequence, the
toroidal field comp on ent is found to be very significant, to-
talling as much magnetic energy as the poloidal component.
The poloidal field component is mostly non-axisymmetric,
with less than 10% of the energy concentrating in modes
with m < ℓ/2; the dipole (i.e., = 1) component has a po-
lar strength of 100 G and is highly tilted (at about 60
with respect to the rotation axis (pointing towards phase
0.80). On the secondary star, the magnetic field has a com-
parable average surface intensity (of 170 G), with the ra-
dial and azimuthal field components of roughly compara-
ble strengths and topologies. The toroidal field component
is weaker than in the primary star though still significant,
totalling only about 15% of the reconstructed magnetic en-
ergy; the poloidal field component is once more mostly non-
axisymmetric, with again less than 10% of the energy con-
centrating in modes with m < ℓ/2; the dipole component is
even weaker (70 G) than on the primary star and is roughly
perpendicular to the rotation axis (pointing towards phase
0.1) and to the dipole moment of the primary star.
The photospheric brightness maps we derive for both
stars of V4046 Sgr feature one dark spot close to the pole
and each covering 2% of the stellar surface. Both spots are
slightly off-centered (by about 10
) towards phase 0.2
for the primary star and phase 0.7 for the secondary star,
i.e., towards the hemisphere that faces the companion star
(best viewed from Earth at the orbital phases of the second
and fi rst conjunctions, respectively equal to 0.181 and 0.681,
see Sec. 3). This is fully consistent with t he observed mod-
ulation of the LSD profiles of photospheric lines as reported
in S ec. 4, showing that th e primary is brighter/fainter than
the secondary around phases 0.7/0.2 respectively. No obvi-
ous correlation between the magnetic and brightness maps,
apart from the fact that the (incomplete) counterclockwise
ring of azimuthal/toroidal field detected at high latitudes on
the p rimary star is more or less surrounding the correspond-
ing cool polar spot (see Fig. 6 upper panels).
The reconstructed maps of accretion-powered excess
emission of both stars mainly show an extended, low-
contrast region of enhanced emission centred onto th e pole,
each covering 1% of the stellar surface. Again, this is in
agreement with the moderate strength and small (largely
insignificant) modulation of the Ca ii emission profiles of
V4046 Sgr as reported in Sec. 4. As for the brightness maps,
We speculate that the smaller contrast observed in the Ca ii
IRT lines reflects the fact that the co oler secondary star is com-
paratively more active and/or generating more accretion-powered
emission than the primary star, by typically 35%.
Page 9
10 J.-F. Donati et al.
Figure 6. Maps of the radial, azimuthal and mer idional components of the magnetic field B (first and third rows, left to r ight panels
resp ectively), photospheric bri ghtness and excess Ca ii IRT emission (second and fourth rows, first and second panels respectively) at
the sur face of the primary (top two rows) and secondary (bottom two rows) components of V4046 Sgr. Magnetic fluxes are labelled in
G; local photospheric brightness (normalized to that of the quiet photosphere) varies from 1 (no spot) to 0 (no light); local excess Ca ii
emission vari es from 0 (no excess emission) to 1 (excess emission covering 100% of the local grid cell, assuming an intrinsic excess emission
of 10× the quiet chromospheric emission). In al l panels, the star is shown in flattened polar projection down to latitudes of 30
, with
the equator depicted as a bold circle and parallels as dashed circles. Radial ticks around each plot indicate phases of observations. The
Page 10
Magnetism & accretion of V4046 Sgr 11
Figure 7. Maximum-entropy fit (thin red line) to the observed (thick black line) Stokes I and Stokes V LSD photospheric profiles
(top panels) and Ca ii IRT profiles (bottom panels) of V4046 Sgr. The light-blue curve in the bottom left panel shows the (constant)
contribution of the quiet chromosphere to the Stokes I Ca ii profiles. Orbital/rotational cycles and 3σ error bars (f or Stokes V profiles)
are also shown next to each pr ofile. First and second orbital conjunctions occur at phases 0.681 and 0.181 respectively.
Page 11
12 J.-F. Donati et al.
we see no obvious connection b etween these features and
the corresponding magnetic m ap s. We suspect that these
extended, low-contrast regions of excess emission centred on
the pole may be tracing even more extended features poten-
tially covering most of the star, but with low latitudes trun-
cated by tomographic imaging - the maximum entropy crite-
rion naturally filtering out areas of lowest visibility (i.e., low-
est latitudes) when phase sampling and sp atial resolution are
both moderate (like in the present case). Simulations involv-
ing fake stars with properties similar to those of V 4046 Sgr
(regarding v sin i and i in particular) and with low-contrast
excess emission evenly distributed over the whole surfaces
confirm th at the reconstructed emission features (centred
on the pole) extend no further than latitude 45
(i.e., only slightly larger than what we recover in the case
of V4046 Sgr) when the quality and sampling of the spectra
are similar to those of our observations. Our maps there-
fore suggest that excess emission likely exten ds over a large
area on both stars of V4046 Sgr, with no more th an a small
residual excess in the polar regions; new observations with
improved phase sampling are needed to validate more firmly
this conclusion.
Our paper presents new results regarding magnetospheric
accretion processes in cTTSs; here we concentrate on the
bright cTTS close binary V4046 Sgr, comprised of two
12 Myr-old 0.9 M
stars separated by 0.041 AU in a
2.42 d-period circular orbit, and observed within the frame-
work of the international multi-wavelength multi-site mon-
itoring campaign organised for this object in 2009 Septem-
ber (involving in particular X-ray ob servations with XMM-
Newton; Argiroffi et al 2011, submitted). Spectropolarimet-
ric data were collected with ESPaDOnS@CFHT over 2.5
orbital/rotation cycles. From these time-resolved spectropo-
larimetric data and using the latest version of our magnetic
tomographic imaging code (extended to the case of close
double-line binaries), we reconstruct maps of the large-scale
magnetic field, of the photospheric brightness and of the
accretion-powered Ca ii IRT excess emission at the surface
of both the primary and secondary stars of V4046 Sgr.
The large-scale fields of both stars are found to be weak
and complex (with respect to those of other cTTSs of sim-
ilar masses in particular), with average field strengths of
typically 200 G, a highly non-axisymmetric poloidal com-
ponent, and a significant toroidal component; in particular,
the large-scale dipole field components are weak (100 and
70 G for the primary and secondary star respectively), highly
tilted with respect to the rotation axis and perpendicular to
each other. This is radically different from the strong, mainly
poloidal, mostly axisymmetric magnetic topologies of proto-
typical cTTSs such as BP Tau or AA Tau (Donati et al.
2008b, 2010), and similar to eld configurations of more
massive cTTSs (e.g., Hussain et al. 2009). The field config-
urations of both cTTSs of V4046 Sgr are also quite similar
to those found in HD 155555 (another slightly more mas-
sive close binary of the β Pic moving group, see Sec. 2),
also showing a weak and complex field with a highly non-
axisymmetric poloidal component and a significant toroidal
compon ent (Dunstone et al. 2008).
By analogy with dy namo-generated magnetic topolo-
gies of main-sequence stars, we speculate that this con-
trast reflects a drastic difference in the internal structure
of the star, with fully-convective stars (those n ot too far
below the full-convection limit at least) showing strong,
mostly axisymmetric and poloidal magnetic configurations
and partly-convective stars exhibiting much weaker and
more complex poloidal fields and a significant toroidal com-
ponent (Donati & Landstreet 2009). Given th eir likely age
of 12 Myr, th e components of V4046 Sgr should belong to
the second category while BP Tau and AA Tau belong to
the fi rst, furth er strengthening the conclusion that magnetic
fields of cTTSs have a dynamo origin (e.g., Donati et al.
Our images also show that both stars of V4046 Sgr
host cool spots close to the pole, similar to (thou gh smaller
in size than) those map ped at the surface of HD 155555
(Dunstone et al. 2008); we also find t hat, for each star in
the system, t hese spots are slightly off-centred towards th e
hemisphere that faces the companion star, in agreement with
the primary star being brighter and fainter than the sec-
ondary star (as judged from the p rimary to secondary line
depth ratio, see Sec. 4 and Fig. 3) at orbital ph ases of rst
and second conjunction respectively (i.e., phases 0.681 and
0.181 respectively). We find no clear correlation between the
reconstructed brightness maps and corresponding magnetic
topologies, despite both being reconstructed simultaneously
in a fully self-consistent way; in particular, the cool polar
sp ots are apparently not coincident with one pole of a mainly
axisymmetric magnetic topology as in most other cTTSs
(Donati et al. 2008b, 2010, 2011a). The only connection we
can report between magnetic topologies an d photospheric
brightness maps is that the (incomplete) unipolar rin g of
counterclockwise azimuthal field detected on the primary
star is encircling the cool polar spot; this is very similar to
what is reported for rapidly -rotating partly-convective solar-
type stars (e.g., Donati et al. 2003; Dunstone et al. 2008).
The maps of accretion-powered excess emission we re-
cover for both stars from sets of Ca ii IRT profiles show a
low-contrast, extended feature centred on the pole, directly
reflecting the low emission and small modulation that the
Ca ii IRT lines exhibit (see Sec. 4 and Fig. 3). In particu-
lar, these maps are less contrasted that those of most other
cTTSs imaged to date (Donati et al. 2010, 2011a) despite
a similar mass accretion rate (of 9.3 ± 0.3 in logarithmic
scale and in M
, see Sec. 4) ; they also show no ob-
vious resemblance with the corresponding magnetic maps.
Tomographic imaging simulations suggest that the parent
distributions of excess emission at the surfaces of both stars
of V4046 Sgr may potentially cover the whole surfaces (with
signal from low latitudes being filtered out by the maximum
entropy criterion) and no more than a low-contrast residual
emission excess close to the pole.
We suggest th at this reflects that accretion onto each
star of V4046 Sgr does not concentrate into a single specific
region of the visible hemisphere (as it does on other cTTSs,
e.g., Donati et al. 2008b, 2010, 2011a) but rather occurs si-
multaneously at several places over the whole stellar surface;
in each of the individual accretion sites, mass accretion is
presumably not high enough to produce distinct features in
the m ap s of accretion-powered emission excesses, especially
given the limited spatial resolution achievable at the surfaces
Page 12
Magnetism & accretion of V4046 Sgr 13
of both stars of V4046 Sgr. Any residual low-contrast excess
emission close to the pole may indicate that chromospheric
emission is not homogeneous on both stars of V4046 Sgr (as
originally assumed in the model) but is slightly stronger at
high latitudes, where cool spots are observed to cluster at
photospheric level.
Distributed accretion is consistent with the fact that the
dipolar components of t he large-scale fields of both stars are
weak and t hat the field topology within the m agnetospheric
gap (exten ding no further than 0.5–1 R
above the surface
of both components of V4046 Sgr) is complex, dominated
by higher-order terms of the magnetic spherical expansions
and thus featuring multiple poles over the visible hemi-
sphere. The red -shifted pseudo-absorption episodes poten-
tially observed in Balmer lines at specific phases (rather than
throughout the whole rotation cycle, see Sec. 4) may provide
more information about the spatial intermittency of accre-
tion funnels, if further data (with much denser phase cover-
age) demonstrate that these episodes are reliable probes of
accretion funnels crossing the line of sight in V4046 Sgr as
Recent 2D and 3D numerical hydro dynamical
simulations of mass accretion onto close cTTS bi-
naries (Kaigorodov et al. 2010; Fateeva et al. 2011;
de Val-Borro et al. 2011) suggest that accretion may not
directly proceed from the circumbinary disc to the stars;
more specifically, they find that local small-size discs
may develop around both stars, with material flowing
from the circumbinary disc to the small discs through the
circumbinary gap (and in particular through the co-rotating
outer Lagrange points), and forming a double-armed spiral
structure (with one arm joining each local disc to the wider
circumbinary disc) as well as a bridge connecting both local
discs. The local discs have predicted sizes similar to those of
the corresponding Roche lobes, with radii of 3 R
in the
particular case of V4046 Sgr. Observations of optically-thin
Balmer lines however suggest that only the accretion
streams actually exist in V4046 Sgr and not the local discs,
as a possible result of them being magnetically disrupted
by th e large-scale fields of both stars (de Val-Borro et al.
Self-consistent simulations including the effects of mag-
netic fields are not available yet. As of now, our results sug-
gest that the large-scale fi elds of both stars, and in particu-
lar their dipolar comp on ents, are likely not strong enough to
disrupt the predicted local discs around both system stars
beyond a typical distance of 1 R
above the stellar su rfaces.
However, our data can neither directly confirm nor invali-
date the existence of local accretion discs. For instance, the
Ca ii IRT emission components, although a useful tracer of
the hot footpoints at the base of accretion funnels, show
no evidence of accretion spots located beyond the surfaces
of the system stars, e.g., where the spiral accretion streams
supposedly meet the putative local accretion d iscs; it is h ow-
ever not clear from the existing simulations whether such
sp ots truly exist and, if they do, whether they are bright
enough to be detected.
The time-resolved contemporaneous (though not ex-
actly simultaneous) X-ray observations of V4046 Sgr col-
lected with XMM-Newton in the framework of our interna-
tional campaign can provide useful constraints onto magne-
tospheres and accretion processes in a close cTTS binary like
V4046 Sgr. They confirm in particular that th e X-ray emis-
sion measure distribution of V4046 Sgr is similar to that of
the evolved cTTS TW Hya, featuring in p articular a hard X-
ray compon ent of coronal origin (with plasma temperatures
of up to 10 MK) and a soft X-ray one (with an average tem-
perature of about 3 MK) presumably probing th e accretion
shocks at the base of accretion funnels (Argiroffi et al. 2011,
submitted). That TW Hya hosts a strong and mostly ax-
isymmetric large-scale magnetic field (Donati et al. 2011b)
drastically different from those of both stars of V4046 Sgr
makes this similarity very intriguing and even more inter-
esting to stud y.
At least three major flares in the hot X-ray compo-
nent were monitored (including both the rise and decay
phases) during these observations, suggesting flaring loops