A&A 459, L33-L36 (2006)
DOI: 10.1051/0004-6361:20066391
LETTER TO THE EDITOR
J. Gorosabel1 - V. Larionov2 - A. J. Castro-Tirado1 - S. Guziy1,3 - L. Larionova2 - A. Del Olmo1 - M. A. Martínez1 - J. Cepa4 - B. Cedrés4 - A. de Ugarte Postigo1 - M. Jelínek1 - O. Bogdanov3 - A. LLorente5
1 - Instituto de Astrofísica de Andalucía (IAA-CSIC),
Apartado de Correos, 3.004, 18080 Granada, Spain
2 -
Astronomical Institute of St. Petersburg University, Petrodvorets,
Universitetski pr. 28, 198504, Russia
3 -
Nikolaev State University, Nikolskaja 24, Nikolaev 54030, Ukraine
4 -
Instituto de Astrofísica de Canarias, La Laguna, Tenerife, 38200 Canary Islands, Spain
5 -
XMM-Newton Science Operations Centre, European Space Agency,
Villafranca del Castillo, PO Box 50727, 28080 Madrid, Spain
Received 13 September 2006 / Accepted 29 September 2006
Abstract
Aims. We have performed optical polarimetric observations of the SN 2006aj associated to the X-ray flash (XRF) of February 18, 2006, XRF 060218 that provide information on its expansion geometry.
Methods. The data were acquired in the R-band with the 0.7 m telescope of Crimea, 2.5 m Nordic Optical Telescope and the 2.2 m of Calar Alto.
Results. We report the detection of linear polarization between 3 and 39 days after the gamma-ray event (t-t0). This represents the first polarization detection of a Ic supernova (SN) associated to an XRF. Our data exhibit a degree of linear polarization (P) around
at
days, followed by a constant polarization phase with
at
days. Our data suggest a decay in P, and more interestingly, show a position angle (
)
rotation of
comparing data taken before and after the R-band lightcurve peak.
Conclusions. The reported polarization measurements can be explained by the evolution of an asymmetric SN expansion. We discuss on several ingredients that could account for the observed
rotation.
Key words: gamma rays: bursts - supernovae: general - techniques: polarimetric
The Swift satellite detected a long-soft gamma-ray burst (GRB) at t0 = Feb. 18.149 2006 UT, (Cusumano et al. 2006), showing both a low redshift (z=0.033, Mirabal & Halpern 2006) and a peak energy (Campana et al. 2006), typical of X-ray flashes (XRF 060218 hereafter). Further observations associated it to a Ic-type supernova, SN 2006aj (Masetti et al. 2006; Soderberg et al. 2006; Pian et al. 2006). This finding strengthened even more the already solid long-duration GRB/Supernova link (Hjorth et al. 2003; Stanek et al. 2003).
Since 1999 (Covino et al. 1999) several GRB afterglows have
shown optical polarization at a level of 1-3% (see Gorosabel et al. 2004 and references therein). Optical supernovae (SNe),
including typical Ic-types not related to GRBs, do not tend to show
optical polarization above
1%. However, some counterexamples
exist (Leonard et al. 2006). The very few polarimetric
studies of Ic SNe related to GRBs do not allow yet to infer clear
global polarimetric differences with normal Ic SNe. However, as we
report in this Letter, some normal Ic SNe like SN 2002ap
(Mazzali et al. 2002) share some similarities with
SN 2006aj; polarization levels above
1% and, more
interestingly, long-term rotations in the polarization angle
(Kawabata et al. 2002).
Most of the SN intrinsic polarization detections are attributed to
the break of the symmetry around line of sight, due to a Thompson
photon scattering through an aspherical SN expansion (Kasen et al.
2003). This polarization is expected to decay as t-2, as the atmosphere expands and the optical depth (and hence
the polarization) falls (Leonard et al. 2006). In most
cases the afterglow is so strong that overshines the SN polarization
component (for instance in GRB 030329; Greiner et al.
2003). In this Letter we report the first polarization
detection to date of a SN associated to an XRF, not coming from the
GRB ejecta forward shock emission (Fan et al. 2006; Soderberg
et al. 2006). Section 2 reports the
observations carried out for SN 2006aj. Section 3
presents our results and Sect. 4 discusses the
implications of our measurements. Section 5 summarises
the final conclusions.
Table 1:
Log of the observations. For the NOT and CAHA the
observation are cycles of 4 images, (
)
with
the plate rotator at 0, 22.5, 45, 67.5 degrees. The R-band
magnitudes are calibrated using the standard stars
given by Hicken et al. (2006).
Table 1 displays the log of our observations. All the measurements reported in this paper were carried out in the R-band. The observations were performed, chronologically ordered, with the 0.7 m telescope of Crimea (AZT-8), the 2.5 m Nordic Optical Telescope (NOT) and the 2.2 m of Calar Alto (CAHA).
The AZT-8 observations were done with two Savart plates; swapping them
the observer can obtain either the Stokes Q (two images are split
diagonally) or the U parameter (split horizontally). The field of
view (FoV) covered by the AZT-8 (+ST-7 CCD) is
and the pixel scale is 1.3
/pix. Since
the GRB field is substantially crowded in the AZT-8 images, we
performed PSF photometry (Stetson 1987).
The NOT observations were based on ALFOSC equipped with the FAPOL
unit. In order to obtain Q and U with ALFOSC, the GRB field was
imaged through a calcite and a 1/2 wave plate. Four images of the GRB
field were acquired, rotating the 1/2 wave plate at ,
22.5
,
45.0
and 67.5
.
The calcite plates
of FAPOL produce a vignetted field of about 140
in
diameter with a pixel scale of 0.19
/pix. The CAHA
observations were carried with CAFOS. The CAFOS polarization unit is
similar to FAPOL (is based on a Wollaston prism instead of a calcite
plate), but uses a strip mask on the focal plane to avoid accidental
overlapping on the CCD. The total FoV in CAFOS is composed by 14 strips of
each, containing stars
enough for a satisfactory interstellar medium (ISM) polarization
correction. For the NOT and CAHA data the determination of P and
was carried out fitting the
function with the
corresponding ISM normalization factor (di Serego Alighieri
1997; see Fig. 1). For both data sets aperture
photometry was performed, ranging the aperture radii from 1 pixel to 2.5 times the FWHM (Full Width at Half Maximum). After checking that
the magnitudes were independent of the apertures used, the ones
yielding minimum errors were adopted.
![]() |
Figure 1:
Upper panel: the Q, U plane of the stars
in the field (circles) of XRF 060218 (square) imaged from
CAHA on March 8.845 UT (
t-t0 = 18.7 days). The size of each
circle is proportional to the distance of the corresponding star to
the center of the CCD. This is used to check that the instrumental
polarization, if present, does not vary severely across the FoV.
Lower panel: the satisfactory fit of the ![]() ![]() |
Open with DEXTER |
The Galactic ISM polarization correction was done using 9 (13) field stars on the AZT-8 (CAHA) images. Due to the reduced ALFOSC FoV only one unsaturated bright field star was available in the NOT data. Thus for the NOT images zero (G191B2B) and high (HD 25443) polarization standards allowed us to infer the SN 2006aj Stokes parameters, and for completeness, also the ones of the unique unsaturated field star (which agreed with the ISM law by Serkowski et al. 1975).
Table 1 shows the position angle ()
and the degree of
linear polarization (P) of the afterglow once they were corrected by
the ISM polarization and also by the fact that P is a positive
quantity (multiplying P by
;
see di Serego Alighieri 1997). The AZT-8 results displayed in
Table 1 supersede the ones reported previously (Larionov &
Larionova 2006). The four AZT-8 data sets yield P marginal
detections, reaching a
confidence level at
days. The apparently large P variations (from P=2.43 to 4.53) in
the first four nights are within P errors, so we they are likely
statistical fluctuations. However, we note that for the four epochs
SN 2006aj is systematically placed at the same Stokes plane
area, once the ISM correction is included. In order to reinforce the
P detection, the four AZT-8 observing epochs were co-added and PSF
photometry performed. The fifth line of Table 1, shows the
result when the four AZT-8 epochs are combined. The estimate of the
P and
errors in the AZT-8 images were obtained modelling
with a Montecarlo method the flux error distribution of
SN 2006aj and the 9 field stars used for the ISM correction.
Figure 2 shows the Stokes plane when all the AZT-8 data are
combined. The cloud of dots representing the SN 2006aj (Q,U) distribution is clearly off the center defined by the field stars.
The P value inferred from the combined image is
,
detected at a 3.1
level (including the
term).
![]() |
Figure 2: The Stokes plane of 9 field stars and SN 2006aj on the combined AZT-8 image when the Galactic ISM correction is applied. The clouds of dots were created altering with a Montecarlo method the fluxes of the objects according to their photometric errors. The cloud representing SN 2006aj is shifted with respect to the origin defined by the field stars, assumed to be intrinsically unpolarized in average. |
Open with DEXTER |
Both the NOT and first CAHA data show conclusive P detections with
confidence levels of
and
respectively. The
values of both epochs agree with a stable angle around 50
,
but interestingly it appears rotated by
with respect to the AZT-8 images. Our data points
reject a stationary P value at a 3
level, so we conclude
that P likely fades at
5.6 <t-t0< 13.7 days. Such a P decay
resembles the late evolution of the asymmetric SN 2004dj
(Leonard et al. 2006), which showed an abrupt P peak at
days due to the sharp appearance of the inner ejecta,
followed by a late
t-2 decay.
We point out that the result of our R-band polarimetry is a
weighted mean of the continuum and the spectral features (in
principle each one with a different polarization) falling inside the
wavelength range of the filter transmittance (centred at 6500 Å).
With no spectro-polarimetric data it is not possible a priory to
quantify if the measured polarization evolution is dominated by
spectrally localized large polarization fluctuations. We note
however that the SN 2006aj spectra around 6500 Å are smooth. Thus
the
rotation measured at
5.5 <t-t0 < 13.7 days, is not
easily explainable by colour evolution, which occurs at wavelengths
bluer than 5000 Å in that epoch range (Mazzali et al. 2006).
![]() |
Figure 3:
Our polarization results, once the the Galactic
ISM correction was considered. Upper panel: as shown ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
The most natural framework to explain our polarization measurements is
a non spherical SN expansion, consistent with the scenario proposed
for XRF 060218 by recent studies (Pian et al. 2006). It is interesting to note that other Ic SNe
exhibiting non-spherical expansions have been linked to long-duration
GRBs (Höflich et al. 1999). Furthermore there is growing
evidence that an important fraction of Ic SNe are triggered by bipolar
ejections (Granot & Ramirez-Ruiz 2004). It is widely
accepted that non-zero SN polarization measurements demand some degree
of expansion asphericity. If the projection of the SN along the line
of sight is elliptical, a non canceled linear polarization should be
present parallel to the minor axis (Kasen et al. 2003). An
ellipsoidal expansion is fairly consistent with the measured P (see
t-2 curve of Fig. 3). However, a single ellipsoidal
model can not account for the measured
rotation.
The decay in P seems to be accompanied with a decay of the
photospheric expansion velocity (notice the similarity between our
Fig. 3 lower panel and Fig. 2 of Pian et al. 2006).
This possible correlation between the photospheric expansion speed and
the polarization, might imply the existence of a fast (
v > 20 000 km s-1, Pian et al. 2006) highly asymmetric ejecta
responsible of the high polarization measured 2.6-5.6 days after the
GRB. This rapid asymmetric ejecta might be aligned along the GRB
jet. Several days later a slower (
v < 15 000 km s-1, Pian et al.
2006) SN bulk ejecta would circularice the geometry, which
would explain the lower P of our CAHA and NOT measurements. We
could speculate that this slower ejecta might be in a toroidal-shape
equatorial expansion, so the geometric transition from the rapid to
slow ejecta might naturally explain the detected rotation of
.
These results partially resemble the jet-like
geometry proposed for SN 2002ap, which exhibited lower Pvalues than SN 2006aj, but with a
rotation similar
to it (
;
Kawabata et al. 2002). However,
other ingredients can also induce a rotation in
:
i) the
host galaxy ISM; ii) a non radial distribution of 56Ni; and iii) a dusty circumstellar medium (CSM). These effects can be
combined producing a complex picture.
As discussed by Klose et al. (2004) the polarization
measured in afterglows are sensitive to the host galaxy ISM, so it
might well affect our observations. Thus, the
stability at
t-t0>13.7 days could be the result of the P induced by the host
ISM, so the SN 2006aj intrinsic polarization (
)
would dominate just before our NOT data point. According to the
Serkowski's law
would be required in the host
line of sight to account for the
NOT and CAHA
measurements. This E(B-V) value agrees fairly well with the host
extinction estimated by Modjaz et al. (2006;
), Sollerman et al. (2006;
),
and Campana et al. (2006;
E(B-V)=0.2). We note however
the lower estimate by Guenther et al. (2006;
E(B-V)=0.042
), so the extinction although present, might not
account totally for the late polarization. In this Framework the
change of
would just obey to the chance
alignment between the asymmetric expansion and the linear
polarization induced by the host ISM. Assuming that our last P detection (
days) is purely due the host ISM, the
SN 2006aj intrinsic polarization at
days would be
higher (
,
see Fig. 3 lower panel).
An asymmetric distribution of 56Ni could alter the SN radiation
field, specially after the lightcurve peak when the 56Ni radioactive decay powers the optical lightcurve. The collapsar model
predicts a subrelativistic disk wind of neutron and protons, which
after cooling would form 56Ni (Woosley & Bloom 2006).
A picture to explain the rotation of
,
would require an equatorial deposition of synthesized 56Ni,
which would alter the
measured after the lightcurve peak.
Another possible mechanism to explain the observed
change
could be delayed CSM scattered light. This process was proposed to
interpret the SN 1987A polarization lightcurve (Wang & Wheeler
1996). In this scenario the polarization would consist of
two components, the one due to photons propagating directly from the
(possibly aspherical) SN, and the second one from scattered light by
a dusty CSM region. The epochs of our observations would push the
CSM scattering region further than
2.3 light-days from the
progenitor. However, this process should have originated a clear
distortion of the spectra at
Å after the
lightcurve peak, which is not observed (Modjaz et al. 2006;
Pian et al. 2006).
We report the detection of R-band linear polarization for
XRF 060218/SN 2006aj in observations done since
to 39
days. These observations represent the first detection of optical
polarization in an XRF. The measured high degree of polarization
gives further credence to the XRF/GRB off-axis scenario, since a
face-on SN would not likely produce any intrinsic polarization. As
Patat et al. (2001) noted there is a degeneracy between the
viewing angle and the asymmetry level, so a detailed geometric
picture based on only our data is not possible. Our data show a
value in the first
3-5 days after the GRB,
followed by a constant
polarization phase at
days, suggesting a P decay.
shows a
rotation at
5.6 < t-t0 <13.7 days. We propose that
the P evolution can be explained by an asymmetric SN. We discuss
several ingredients which separately, or combined, can explain the
observed
rotation (in decreasing order of relevance); i) a
highly asymmetric high-velocity SN ejection, followed by a less
asymmetric SN bulk; ii) polarization induced by the host ISM; iii) asymmetric distribution of 56Ni; and iv) scattering by a dusty
region placed a few light-days from the progenitor.
Acknowledgements
This study is supported by Spanish research programmes ESP2002-04124-C03-01 and AYA2004-01515. JG thanks the hospitality of the Donostia International Physic Center (DIPC). We thank our anonymous referee for helpful comments.