Dust and Gas in the Local Environments of Gamma-Ray Bursts

Abstract

Using a sample of gamma-ray burst (GRB) afterglows detected by both the X-Ray and the UV/Optical Telescopes (XRT and UVOT) on Swift, we modelled the spectral energy distributions (SEDs) to determine gas column densities and dust extinction in the GRB local environment. In six out of seven cases we find an X-ray absorber associated with the GRB host galaxy with column density (assuming solar abundances) ranging from (0.8 - 7.7)x10^{21}cm^{-2}. We determine the rest-frame visual extinction A_V using the SMC, LMC and Galactic extinction curves to model the dust in the GRB host galaxy, and this ranges from A_V = 0.12\pm 0.04 to A_V = 0.65^{+0.08}_{-0.07}. The afterglow SEDs were typically best fit by a model with an SMC extinction curve. In only one case was the GRB afterglow better modelled by a Galactic extinction curve, which has a prominent absorption feature at 2175angstrom. We investigate the selection effects present in our sample and how these might distort the true distribution of A_V in GRB host galaxies. We estimate that GRBs with no afterglow detected blueward of 5500angstrom have average rest-frame visual extinctions almost eight times those observed in the optically bright population of GRBs. This may help account for the ~1/3 of GRBs observed by Swift that have no afterglow detected by UVOT.

Full text

Page 1

arXiv:astro-ph/0702122v2 9 Aug 2007
Mon. Not. R. Astron. Soc. 000, 000–000 (0000) Printed 5 February 2008(MN LATEX style file v2.2)
Dust and Gas in the Local Environments of Gamma-Ray
Bursts
P. Schady1,2, K.O. Mason3,1, M.J. Page1, M.De Pasquale1, D.C. Morris2, P. Romano4,
P.W.A. Roming2, S. Immler5and D.E. Vanden Berk2
1The UCL Mullard Space Science Laboratory, Holmbury St Mary, Dorking, Surrey, RH5 6NT, UK.
2Department of Astronomy and Astrophysics, Pennsylvania State University, 525 Davey Laboratory, University Park, PA 16802, USA.
3The Particle Physics and Astronomy Research Council, Polaris House, North Star Avenue, Swindon, Wiltshire, SN2 1SZ, UK.
4INAF-Osservatorio Astronomico di Brera, via E. Bianchi 46, 23807 Merate (LC), Italy.
5NASA/Goddard Space Flight Center, Greenbelt, MD 20771, USA.
Received:
ABSTRACT
Using a sample of gamma-ray burst (GRB) afterglows detected by both the X-Ray
and the UV/Optical Telescopes (XRT and UVOT) on Swift, we modelled the spectral
energy distributions (SEDs) to determine gas column densities and dust extinction in
the GRB local environment. In six out of seven cases we find an X-ray absorber asso-
ciated with the GRB host galaxy with column density (assuming solar abundances)
ranging from (0.8 − 7.7) × 1021cm−2. We determine the rest-frame visual extinction
AV using the SMC, LMC and Galactic extinction curves to model the dust in the
GRB host galaxy, and this ranges from AV = 0.12 ± 0.04 to AV = 0.65+0.08
afterglow SEDs were typically best fit by a model with an SMC extinction curve. In
only one case was the GRB afterglow better modelled by a Galactic extinction curve,
which has a prominent absorption feature at 2175˚ A. We investigate the selection ef-
fects present in our sample and how these might distort the true distribution of AV in
GRB host galaxies. We estimate that GRBs with no afterglow detected blueward of
5500˚ A have average rest-frame visual extinctions almost eight times those observed
in the optically bright population of GRBs. This may help account for the ∼ 1/3 of
GRBs observed by Swift that have no afterglow detected by UVOT.
−0.07. The
Key words: gamma-rays: bursts - gamma-ray: observations - galaxies: ISM - dust,
extinction
1 INTRODUCTION
There is now a wealth of observational evidence link-
ing long duration gamma-ray bursts (GRBs) (prompt
emission lasting >2 s) with the collapse of a mas-
sive star (collapsar model; e.g. Woosley 1993). This in-
cludes the underlying supernova features in the after-
glow of some GRBs [e.g.GRB 980425 (Kulkarni et al.
1998), GRB 030329 (Hjorth et al. 2003) and GRB 060218
(Campana et al. 2006)], and the association between GRB
host galaxies and high-mass star formation (Tanvir et al.
2004). The γ-ray energy emission of GRBs is unaffected by
dust and gas in the intervening interstellar medium, which
combined with the vast energy released during the initial ex-
plosion, allows them to be detected out to very high redshifts
(e.g. Tagliaferri et al. 2005). These two facts give GRBs the
potential to be highly powerful tools with which to trace
the star formation history (SFH) in an unbiased way. Fur-
thermore, the longer lived, lower energy afterglows light up
their host galaxies, albeit for only a brief time (on the or-
der of weeks), providing invaluable insight into the chemical
makeup of these galaxies that would otherwise be unattain-
able in a majority of cases; certainly at high redshifts.
To fully maximise the potential that GRBs offer as cos-
mic probes the selection effects present in GRB studies need
to be known, such as the stellar populations that they trace
and the properties of their host galaxies. Infra-red observa-
tions of long GRB host galaxies taken with Spitzer indicate
that they are not strong starburst galaxies, nor are they
particularly dusty (Le Floc’h et al. 2006). This is in conflict
with expectations from the collapsar model, which predicts
GRBs to occur in regions of active star formation that are
heavily enshrouded by dust. This mismatch between obser-
vations and theory could very possibly be the result of selec-
tion effects, whereby the near infra-red (NIR) to ultra violet
(UV) afterglow of GRBs in dusty galaxies are extinguished,
thus reducing the chance of identifying the host. Further
detailed analysis of GRB environments is important to de-

Page 2

2
P. Schady, K.O. Mason, M.J. Page, et al.
termine the range in host galaxy properties, and, fundamen-
tally, to provide a better understanding of our observations
and the limitations that they present.
Due to the broadband power law spectral behaviour of
GRB afterglows, the effects of absorbing dust and gas in the
local environment on the GRB spectral energy distribution
(SED) can be well identified. In the analysis of X-ray and
optical afterglow spectra for eight GRBs, Galama & Wijers
(2001) found evidence for high column densities of gas in the
GRB local environment that were comparable with those ob-
served in giant molecular clouds. Furthermore, they found
the optical extinction to be 10-100 times smaller than ex-
pected given the column densities. Stratta et al. (2004) ex-
tended this sample and found the ratio of host galaxy col-
umn density to visual extinction to be an order of magnitude
larger that that observed in the Milky Way (MW), and also
greater than that in the Small Magellanic Clouds (SMC)
and Large Magellanic Clouds (LMC). Prompted by theoret-
ical studies that indicate that the intense radiation emitted
by a GRB should destroy small dust grains out to radii of
around 10 pc [e.g. Fruchter, Krolik & Rhoads (2001); Perna
& Lazzati (2002); Perna, Lazzati & Fiore (2003)], the large
gas-to-dust ratio was taken to be evidence of the destruc-
tion of dust in the surrounding vicinity of the burst, thus
reducing the visual extinction observed.
Stratta et al. (2004) also found little evidence of the
strong 2175˚ A Galactic absorption feature in their optical-
NIR spectral analysis. Instead they found the SED to be
best fit by a model in which the host galaxy has an SMC
or starburst galaxy dust extinction law, which has no such
absorption feature. More recently Kann, Klose & Zeh (2006)
analysed the afterglow spectral energy distributions in the
optical and NIR bands on a larger sample of 30 pre-Swift
GRBs, and also found the SMC extinction curve to provide
a better fit to the spectra than the MW or LMC extinction
curves.
The simultaneous observations taken with the X-Ray
Telescope [XRT; Burrows et al. (2005a)] and UV/Optical
Telescope [UVOT; Roming et al. (2005)] on-board the
Swift spacecraft (Gehrels et al. 2004) provide GRB after-
glow SEDs without the need to extrapolate data to the same
epoch. This capability is unique to Swift and the accurate
broadband spectral modelling that is possible with this pro-
vides an important data set to compare to previous multi-
wavelength GRB samples. In this paper we model the SEDs
of a sample of Swift GRBs with afterglows detected by both
the XRT and UVOT to investigate the rest-frame visual ex-
tinction and soft X-ray absorption at the GRB host galaxy.
This provides an indication of the dust and gas content in
the local environment of the GRB, which we compare to the
environment of the MW, the LMC and the SMC.
The effects of the selection bias that is introduced by
analysing only those GRBs with X-ray and optical after-
glows is addressed by using our results to examine the dust
in the local environments of GRBs that have no UVOT de-
tected afterglow. Our results have implications for the role
that dust plays in accounting for the lack of an UV/optical
afterglow detection in ∼ 1/3 of GRBs observed by the
UVOT.
In section 2 we present the GRB sample and describe
the X-ray and UV/optical data reduction and analysis and
in section 3 we describe the models that we used to fit the
data. We present the results from our spectral modelling
and discuss their implications in sections 4 and 5, and also
investigate how GRBs that have no UV/optical afterglow
detected by UVOT may be affected by dust. Our conclusions
are summarised in section 6. Throughout the paper temporal
and spectral indices, α and β respectively, are denoted such
that Ft ∝ t−αand Fν ∝ ν−β, and all errors are 1σ unless
specified otherwise.
2 DATA REDUCTION AND ANALYSIS
SEDs at a single-epoch were produced for a total of seven
bursts. This epoch was chosen to be 1 hour after the onset
of the initial prompt emission (time T) for four of the bursts
(GRB 050318, GRB 050525, GRB 050802 and GRB 060512).
For the remaining three GRBs (GRB 050824,GRB 051111
and GRB 060418) the SEDs were produced at a time T+2hrs
due to a lack of data at T+1hr, resulting from spacecraft
observing constraints.
The GRBs in our sample were chosen on the basis that
they had afterglows detected by both the XRT and UVOT
instruments and a spectroscopic redshift with z < 1.75.
Bursts for which the photometry was considered to be too
poor to provide useful constraints on the spectral fitting
were not used. This included those with UVOT detections
in fewer that three filters, or bursts that did not have well
enough sampled light curves with which to obtain reliable
multi-band photometry at a single epoch. To avoid any con-
tamination from absorption caused by the Lyα forest only
UVOT data with a rest-frame wavelength λ > 1215˚ A were
used in the spectral analysis. The redshift limit was, there-
fore, a necessary requirement to ensure that there were suf-
ficient optical and UV data points in the afterglow SED to
constrain the spectral fitting. At the redshift of our sample
the afterglow spectra cover the wavelength range where the
redshifted 2175˚ A extinction bump is expected to lie, and
those GRBs closer to the redshift upper limit also probe the
rest-frame far-ultraviolet (FUV) spectra, where the diver-
gence between extinction curves is greatest.
2.1 UVOT Data
The UVOT contains three optical and three ultra-violet
lenticular filters, which cover the wavelength range between
1600˚ A and 6000˚ A. Photometric measurements were ex-
tracted from the UVOT imaging data using a circular source
extraction region with a 6′′radius for the V, B and U optical
filters and a 12′′radius for the three UVOT ultra-violet fil-
ters to remain compatible with the current effective area cal-
ibrations1. Where possible the background rate was taken
from an annulus with 12′′inner radius and 20′′outer ra-
dius centred on the source. In the cases where there were
nearby sources that contaminated this extraction region, the
background was taken from a source-free region close to the
target with radii ranging between 10 and 20′′.
For each of the UVOT lenticular filters the tool
uvot2pha (version 1.1) was used to produce spectral files
1http://heasarc.gsfc.nasa.gov/docs/heasarc/caldb/swift/docs/uvot/

Page 3

Dust and Gas in the Local Environments of Gamma-Ray Bursts
3
Table 1. UVOT Data
GRB
z
Galactic E(B-V)1
αopt
050318
050525
050802
050824
051111
060418
060512
1.442
0.6063
1.714
0.835
1.5496
1.497
0.4438
0.017
0.095
0.021
0.034
0.162
0.224
0.014
0.949
1.56,0.6210
0.8211
0.5512
0.9613
1.25
0.84
1Schlegel et al. (1998)
2Berger & Mulchaey (2005);3Foley et al. (2005);4Fynbo et al.
(2005a); Fynbo et al.(2005b);
Dupree et al. (2006);8Bloom et al. (2006)
9Still et al. (2005)
10Light curve best fit by double power law (Blustin et al. 2006)
11Oates et al. (2006)
12Halpern & Mirabal (2005)
13Butler et al. (2006)
56
Prochaska (2005);
7
Table 2. X-Ray Data
GRB Data
Mode
Gal. NH
(1020cm−2)
Start
Time (s)1
Exp.
Time (s)
050318
050525
050802
050824
051111
060418
060512
PC
PC
PC
PC
PC
WT
PC
2.8
9.1
1.8
3.6
5.0
9.2
1.4
3280
7056
480
6096
5552
448
3680
23431
3560
4835
17157
1583
1841
58966
1Measured from the BAT trigger time.
compatible with xspec (version 12.2.1), and response ma-
trices were taken from version 102 of the UVOT calibration
files. To create an SED at an instantaneous epoch the count
rates in these files were set to correspond to the count rate
of the GRB at the appropriate epoch. These count rates and
the associated errors were determined from power law model
fits to the light curve in each filter where the GRB afterglow
is assumed to decay at the same rate in the UV and optical
bands. This is justified by the small variations present in the
decay rate between filters for the GRBs in our sample, which
have a typical variance of ∼ 0.2 and are consistent within
1σ errors. Furthermore, the colour evolution expected when
the cooling break migrates through the optical bands is typ-
ically observed at later times than the epochs we deal with,
on the order of ∼ 104s (e.g. Blustin et al. 2006).
Where possible the UV/optical decay index was taken
from the literature where UVOT data were used in the anal-
ysis. Otherwise, this analysis was carried out by ourselves,
with the exception of GRB 050824 and GRB 051111, for
which there was not sufficient UVOT data to constrain the
fits to the light curves and, therefore, reported indices from
R-band light curve fits were used. For the analysis done on
UVOT data, both our own and that reported in the liter-
ature, the decay index was determined from the combined
UVOT light curve, which was produced by normalising each
Galactic
LMC
SMC
Figure 1. Galactic (solid), LMC (dashed) and SMC (dotted)
extinction curves. Parameterisations taken from Pei (1992)
filter to the V -band light curve. The count rate at an instan-
taneous epoch for each filter was then determined from the
best-fit temporal decay model using the appropriate normal-
isation. The light curves were modelled as a power law, with
the exception of GRB 050525, which required further com-
ponents to describe the more complex early time temporal
behaviour (Blustin et al. 2006). The decay rates determined
for each burst are listed in Table 1, as well as the references
used where appropriate.
2.2 X-Ray Data
The XRT covers the 0.2 – 10 keV energy range, and all data
were reduced using the xrtpipeline tool (version 0.9.9). In
most cases the data used were taken in photon counting
(PC) mode with the exception of GRB 060418, for which
the majority of the data taken at the time of interest are in
window timing (WT) mode (Hill et al. 2004). For PC data
source counts were extracted from a circular region centred
on the source with an outer radius ranging from 50′′to 95′′.
In the case of GRB 050802 the data suffered from pile-up
and, therefore, an annular extraction region was used to ex-
clude those pixels that were piled-up. The inner radius to
this was 9.44′′(4 pixels) and the outer radius was 106.2′′
(45 pixels). The background count rate was estimated from
a circular, source-free area in the field of view (FOV) with a
radius of 118′′(50 pixels). For WT mode data, the extrac-
tion regions used for the source and background count rates
were 94′′slits positioned over the source and in a source free
region of the FOV, respectively. xselect (version 2.4) was
used to extract spectral files from the event data in the en-
ergy ranges 0.3 – 10 keV for PC and WT mode data, which
is the recommended band to use for compatibility with the

Page 4

4
P. Schady, K.O. Mason, M.J. Page, et al.
current calibration files2. Corresponding effective area files
were created using the xrtmkarf tool (version 0.5.1) and
binned to at least 20 counts per energy bin. Response ma-
trices from version 8 of the XRT calibration files were used
for both WT and PC mode data.
A large proportion of GRBs observed by XRT have
shown spectral evolution in the first few thousand seconds of
emission, during X-ray flares and after early time temporal
breaks, which could be the transition from internal shock
to external shock dominated emission (e.g Burrows et al.
2005b; Zhang et al. 2006). All X-ray data were taken from
time intervals where spectral evolution was no longer ob-
served, and the spectra were normalised to the epoch cor-
responding to the SED by using the best-fit model to the
X-ray light curve, in the same way as for the UVOT data.
The data mode and time intervals used are listed in Table 2.
3 THE MODEL
To model the afterglow spectral continuum we tried both
a power law and broken power law fit, where in the latter
the change in spectral slope was fixed to ∆β = 0.5 to corre-
spond to the change in slope caused by a cooling break. In
addition to this a constraint was also imposed on the break
energy such that it was within the observing window (i.e.
0.002 keV < Eb < 10.0 keV). Any fit with a spectral break
outside this energy range would be equivalent to a power
law fit to the data. In both the power law and broken power
law models we included two dust and gas components to
correspond to the Galactic and the host galaxy photoelec-
tric absorption and dust extinction. The Galactic column
density and reddening in the line of sight were fixed to the
values taken from Dickey & Lockman (1990) and Schlegel,
Finkbeiner & Davis (1998), respectively. The second pho-
toelectric absorption system was set to the redshift of the
GRB, and the equivalent neutral hydrogen column density
in the host galaxy was determined assuming solar abun-
dances. The dependence of dust extinction on wavelength in
the GRB host galaxy was modelled on the empirical extinc-
tion laws corresponding to the MW, LMC and SMC.
The wavelength dependence on dust extinction ob-
served in these three environments is well reproduced by a
dust model composed of silicate and graphite grains, where
variations in the relative abundance and grain size distri-
bution produce the different extinction laws (Pei 1992).
The parameterisation of the dust extinction laws in the
MW, LMC and SMC are shown in Fig. 1, where RV =
AV/E(B − V ) = 3.08, 2.93 and 3.16 for the Galactic, SMC
and LMC extinction laws, respectively (Pei 1992).
The prominence of the 2175˚ A absorption feature and
amount of FUV extinction vary between the three curves.
The MW has the strongest extinction at 2175˚ A and small-
est amount of FUV extinction, whereas the SMC has the
greatest amount of FUV extinction, rising faster than 1/λ,
and an insignificant 2175˚ A feature. A further difference in
these three extinction laws is in the amount of reddening
per H atom (Draine 2000), which is observed to be greatest
in the MW and least in the SMC.
2http://heasarc.gsfc.nasa.gov/docs/heasarc/caldb/swift/docs/xrt/
0.1
1.0
10.0
100.0
SMC
0.1
1.0
10.0
100.0
NN,X/AV (10
21 cm
-2)
LMC
0.1
1.0
10.0
100.0
MW
GRB050318
GRB050525GRB050802
GRB050824 GRB051111GRB060418
GRB060512
Figure 2. Ratio of rest-frame NH,Xto AV for the three models
used in the spectral analysis; SMC model (top panel), LMC model
(middle panel) and Milky Way model (bottom panel). Filled cir-
cles correspond to power law spectral models and open circles to
broken power law fits. Dashed and dotted lines show the mean
NH,X/AV value determined from each spectral model for a power
law and broken power law fit, respectively.
We chose these three curves to model the rest frame
visual extinction in the GRB afterglows to identify the pre-
dominant extinction properties of the dust within the GRB
local environment. We refer to each of the spectral models
as the MW, SMC and LMC model, where the name cor-
responds to the extinction law used to describe the dust
extinction properties in the GRB host galaxy.
4 RESULTS
The value that we determine in our spectral analysis
for NH is an equivalent neutral hydrogen column den-
sity that results from the amount of soft X-ray absorp-
tion in the spectrum, where solar abundances are assumed.
This is dominated primarily by oxygen K-shell absorption
(Morrison & McCammon 1983). To distinguish between the
equivalent neutral hydrogen column density determined
from the X-ray absorption and the true neutral hydrogen
column density, we use the notation NH,X to refer to the
former.
The results from our spectral analysis are provided in
Table 3 and Table 4 for a power law and a broken power law
fit, respectively. For five GRBs a broken power law provided
little improvement to the fit with an F-test probability P
> 0.08. For GRB 050802 and GRB 060418 a spectral break
at 1–2 keV and ∼ 3 keV, respectively, improved the good-
ness of fit of the model with an F-test probability ranging
from 0.003 to 1.7 × 10−5for GRB 050802 and from 0.02 to
9 × 10−4for GRB 060418, depending on the model. How-

Page 5

Dust and Gas in the Local Environments of Gamma-Ray Bursts
5
Table 3. Results from simultaneous UV/optical and X-ray power law spectral fits, where only
UVOT data with rest-frame wavelength λ > 1215˚ Awere used.
GRB Model
NH,X
1021cm−2
a
βAV
a
χ2
Null Hypothesis
Probability(dof)
050318SMC
LMC
MW
1.56+0.44
−0.42
2.28+0.53
−0.51
1.90+0.60
−0.55
0.95 ± 0.03
1.04 ± 0.04
0.99+0.06
−0.05
0.53 ± 0.06
0.83+0.10
−0.09
0.91 ± 0.14
101 (76)
105 (76)
154 (76)
0.03
0.02
0.00
050525SMC
LMC
MW
0.78+0.32
−0.29
0.87+0.32
−0.29
0.78+0.33
−0.30
0.76 ± 0.01
0.79 ± 0.02
0.78 ± 0.03
0.16 ± 0.03
0.23 ± 0.04
0.23 ± 0.06
29 (33)
33 (33)
58 (33)
0.67
0.48
0.01
050802SMC
LMC
MW
0.91+0.41
−0.39
1.30+0.44
−0.42
1.98+0.50
−0.48
0.66 ± 0.02
0.70 ± 0.02
0.76 ± 0.03
0.24 ± 0.03
0.38 ± 0.04
0.65+0.08
−0.07
106 (81)
98 (81)
88 (81)
0.03
0.10
0.29
050824SMC
LMC
MW
1.08+0.42
−0.37
1.14+0.43
−0.38
1.20+0.45
−0.38
0.95 ± 0.02
0.96 ± 0.03
0.98 ± 0.04
0.12 ± 0.04
0.18 ± 0.05
0.27 ± 0.10
27 (24)
27 (24)
31 (24)
0.33
0.29
0.16
051111 SMC
LMC
MW
7.71+1.80
−1.55
8.50+1.99
−1.73
8.63+2.29
−2.01
1.10 ± 0.06
1.18+0.09
−0.08
1.19 ± 0.13
0.39+0.11
−0.10
0.64+0.19
−0.18
0.85+0.35
−0.34
13 (11)
14 (11)
22 (11)
0.28
0.23
0.03
060418 SMC
LMC
MW
0.97+0.69
−0.62
1.83+0.76
−0.68
0.72+0.76
−0.71
0.89 ± 0.01
0.97 ± 0.02
0.88 ± 0.04
0.17 ± 0.02
0.38 ± 0.05
0.15 ± 0.10
79 (75)
68 (75)
156 (75)
0.36
0.70
0.13
060512SMC
LMC
MW
< 0.34
< 0.33
< 0.27
0.99 ± 0.02
0.98 ± 0.02
0.96 ± 0.02
0.44+0.04
−0.05
0.44 ± 0.05
0.37 ± 0.04
33 (20)
37 (20)
44 (20)
0.04
0.01
0.002
aAt the redshift of the GRB
ever, the column density and the dust extinction at the host
galaxy do not change significantly between the two contin-
uum fits, remaining consistent to 2σ.
For the purpose of this paper, where we are interested
in the properties of the GRB local environment, the factors
of importance are the dust extinction law that best fits the
GRB afterglows and the relation between the column den-
sity and visual-extinction in the GRB local environment. We
find that the former of these is not affected by the models
that we use to fit the continuum, and there is no significant
change in the ratio between the host galaxy column den-
sity and rest-frame visual extinction. This is illustrated in
Fig. 2, which plots the value of NH,X/AV for the best-fit pa-
rameters determined from a power law and a broken power
law fit as solid and open circles, respectively. Beyond this
section we, therefore, refer only to the spectral results from
the power law fits. In Fig. 3 we show the SEDs and best-fit
power law models for each GRB. In these figures data points
at rest-frame wavelengths λ < 1215˚ A, which were not used
in the fits (see section 2), are shown as open circles.
The amount of intrinsic absorption and extinction de-
termined in our sample varies by over a factor of five in AV
and by more than an order of magnitude in NH,X. Typically
the SMC model results in the smallest AV in the host galaxy
and the MW model in the largest. This general trend is to
be expected given the difference in FUV extinction observed
in the three curves. The MW extinction law is the shallowest
of the three extinction curves (see Fig. 1) and, in particu-
lar, has the least amount of dust-absorption in the FUV for
a given AV. A spectral model with a MW extinction law,
therefore, requires a larger AV to fit the same data.
Although the X-ray column density and UV/optical ex-
tinction are independent components in the fit, the fitted
value of NH,X depends on the spectral index, which in turn
depends on the UV/optical extinction. The best-fit X-ray
column density will, therefore, vary between extinction mod-
els. Typically the MW model requires a steeper spectral
index to compensate for the reduced amount of FUV ex-
tinction, and this consequently results in larger absorption
in the soft X-ray. Regardless of these differences the hydro-
gen equivalent column density determined from our spectral
modelling is generally consistent at the 1σ level. In the case
of GRB 050802 the column density is consistent at the 2σ
level between dust models.
To further investigate the model dependence between
the amount of X-ray absorption and dust extinction in the
local environment of the GRB we show the confidence con-
tours of E(B−V ) vs. NH,X for the best fit power law models
to each GRB in Fig. 4. For the most part the contours are
fairly circular, indicating that there is no significant corre-
lation between NH,X and AV in our spectral modelling. We
also tried fitting the X-ray data alone to make sure that the
UV/optical data were not in any way skewing the column
densities. We find the best-fit NH,X value determined from
our SED spectral analysis to be compatible with the best-fit
NH,X value from spectral analysis on the X-ray data alone,
as shown in Fig. 5. This shows the robustness of our method.

Page 6

6
P. Schady, K.O. Mason, M.J. Page, et al.
Table 4. Simultaneous UV/optical and X-ray broken power law spectral fits. Only using UVOT
data with rest-frame wavelength λ > 1215˚ A.
GRB Model
NH,X
1021cm−2
a
β1
Eb
β2
AV
a
χ2
Null Hypothesis
Probability(keV) (dof)
050318SMC
LMC
MW
1.70+0.42
−0.34
2.28+0.41
−0.44
1.89+0.51
−0.48
0.47+0.03
−0.02
0.54 ± 0.04
0.50 ± 0.05
0.003 ± 0.0003
0.002+0.001
−0.0
< 0.0004
0.97+0.04
−0.03
1.04 ± 0.04
1.00 ± 0.05
0.56+0.03
−0.04
0.83+0.09
−0.08
0.91+0.13
−0.14
99 (75)
105 (75)
154 (75)
0.03
0.01
2e−7
050525 SMC
LMC
MW
1.39+0.48
−0.51
0.92+0.30
−0.37
1.36+0.57
−0.40
0.42+0.11
−0.10
0.30+0.01
−0.03
0.67+0.03
−0.04
0.029+0.104
−0.022
0.004+0.005
−0.001
1.048 ± 0.290
0.92+0.11
−0.10
0.80+0.01
−0.03
1.17+0.03
−0.04
0.26 ± 0.04
0.38+0.03
−0.07
0.24 ± 0.07
27 (32)
32 (32)
80 (32)
0.73
0.47
5e−6
050802 SMC
LMC
MW
0.89+0.42
−0.40
1.14+0.44
−0.43
1.73+0.51
−0.49
0.61 ± 0.02
0.64 ± 0.03
0.72+0.03
−0.05
2.970+0.286
−0.348
2.990+0.516
−0.246
3.960+0.538
−0.900
1.11 ± 0.02
1.14 ± 0.03
1.22+0.03
−0.05
0.18 ± 0.03
0.28+0.05
−0.05
0.55+0.08
−0.11
84 (80)
81 (80)
79 (80)
0.36
0.43
0.53
050824 SMC
LMC
MW
1.16+0.46
−0.37
1.27+0.45
−0.44
1.36+0.48
−0.36
0.47+0.05
−0.03
0.50+0.03
−0.04
0.53+0.03
−0.04
0.003 ± 0.001
0.003 ± 0.001
0.003 ± 0.001
0.97+0.05
−0.03
1.00+0.03
−0.04
1.03+0.03
−0.04
0.16+0.06
−0.04
0.25+0.05
−0.08
0.38+0.03
−0.18
26 (23)
27 (23)
30 (23)
0.32
0.28
0.14
051111 SMC
LMC
MW
10.04+2.70
8.50+1.72
−1.63
11.45+2.56
−2.79
0.82+0.24
−0.20
0.68+0.42
−0.07
0.95+0.18
−0.10
0.040+0.820
−0.034
< 0.772
< 0.883
1.32+0.24
−0.20
1.180+0.42
1.45+0.18
−0.10
0.42 ± 0.13
0.64+0.26
−0.18
0.50+0.41
−0.34
12 (10)
14 (10)
23 (10)
0.25
0.17
0.01
−0.07
−2.83
060418 SMC
LMC
MW
2.80+0.97
−0.85
2.05+0.78
−0.71
1.64+0.93
−0.89
0.85 ± 0.01
0.95 ± 0.02
0.79+0.02
−0.01
1.279+0.254
−0.187
2.100+0.684
−0.513
1.347+0.526
−0.150
1.35 ± 0.01
1.45 ± 0.02
1.29+0.02
−0.01
0.17 ± 0.02
0.37 ± 0.05
< 0.03
68 (74)
64 (74)
145 (74)
0.68
0.79
2e−6
060512SMC
LMC
MW
< 0.47
< 0.42
< 0.32
0.54+0.05
−0.01
0.53 ± 0.01
0.51 ± 0.01
0.003 ± 0.001
0.003 ± 0.001
< 0.001
1.04+0.05
−0.01
1.03 ± 0.01
1.010 ± 0.01
0.58+0.06
−0.03
0.57+0.06
−0.03
0.49+0.04
−0.003
32 (19)
35 (19)
44 (19)
0.04
0.01
0.001
aAt the redshift of the GRB
4.1GRB Host Extinction Laws
Six of the seven GRBs in our sample were best fit by the
SMC or LMC model, with the MW model rejected with
at least 97% confidence for four of these (GRB 050318,
050525, 051111 and GRB 060512). For GRB 050824 and
GRB 060418 the MW model is rejected with 84% and 87%,
respectively. The relatively small amount of extinguishing
dust in the circumburst environment of those two latter
GRBs, indicated by the best fit AV, is likely to be the cause
for the smaller distinction between the spectral models. Al-
though the distinction in the goodness of fit between the
SMC and LMC models is small, the SMC model provides
the best fit to five of the GRBs in the sample, and only
GRB 060418 is best fit by the LMC model; χ2= 68 for 75
degrees of freedom (dof) compared to χ2= 79 for 75 dof for
the SMC model.
The afterglow of GRB 050525 was detected in all
six lenticular UVOT filters (Blustin et al. 2006), and
GRB 050824 (Schady & Campana 2005) and GRB 060418
(Schady & Falcone 2006) had an afterglow detection in all
but the bluest UV filter (UV W2). GRB 051111 did not have
an afterglow detected in the two bluest filters (Poole et al.
2005), and in the case of GRB 050318 (Still et al. 2005) and
GRB 060512 (De Pasquale et al. 2006) no afterglow was de-
tected in any of the UV lenticular filters above the 3σ level.
AttheredshiftsofGRB
Berger & Mulchaey (2005)], GRB 050824 [z
050318[z=
=
1.44;
0.83;
(Fynbo et al. 2005b)], GRB 051111 [z = 1.549; Prochaska
(2005)] and GRB 060418 [z = 1.49; (Dupree et al. 2006)],
the lack of an optical afterglow detection in the bluest filters
could either be the result of Lyα blanketing or high levels
of dust extinction blueward of ∼ 3800˚ A. GRB 050318 and
GRB 051111 have AV values that lie at the higher end of the
distribution observed in the sample, with best fit parameters
AV = 0.53±0.06 mag and AV = 0.39+0.11
if the SMC model is used. GRB 050824 and GRB 060418, on
the other hand, have at least half this amount in rest-frame
extinction if the SMC model is used (AV = 0.12 ±0.04 and
AV = 0.17 ± 0.02 mag, respectively). The correlation be-
tween AV and the number of filters in which the afterglow
is detected could indicate that it is dust present in the local
environment of these GRBs that contributes to the observed
dimness of their UV afterglow.
GRB 060512 was a low redshift burst [z=0.4428;
Bloom et al. (2006)], eliminating neutral hydrogen absorp-
tion as the cause for the lack of an UV afterglow detec-
tion. The best fit is provided by the SMC model (χ2=
33 for 20 dof) yielding a host galaxy extinction of AV
= 0.44+0.04
GRB 050318 and GRB 051111. This, therefore, provides fur-
ther support to the hypothesis whereby dust in the local
environment of the GRB blocks a large fraction of the UV
flux emitted.
In contrast to the other bursts discussed in this section,
GRB 050525 had very small amounts of local absorption
−0.10mag, respectively,
−0.05mag, which is comparable to that observed in

Page 7

Dust and Gas in the Local Environments of Gamma-Ray Bursts
7
1015
1016
1017
1018
Frequency (Hz)
0.001
0.010
0.100
Flux Density (mJy)
1015
1016
1017
1018
Frequency (Hz)
0.0001
0.0010
0.0100
0.1000
Flux Density (mJy)
1015
1016
1017
1018
Frequency (Hz)
0.001
0.010
0.100
Flux Density (mJy)
1015
1016
1017
1018
Frequency (Hz)
0.0001
0.0010
0.0100
Flux Density (mJy)
1015
1016
1017
1018
Frequency (Hz)
0.0001
0.0010
0.0100
0.1000
Flux Density (mJy)
1015
1016
1017
1018
Frequency (Hz)
0.001
0.010
0.100
Flux Density (mJy)
1015
1016
1017
1018
Frequency (Hz)
0.001
0.010
0.100
Flux Density (mJy)
GRB050318GRB050525GRB050802
GRB060418GRB060512
GRB050824 GRB051111
Figure 3. SEDs for seven GRBs at an instantaneous epoch (see text) with best-fit models for each corresponding dust extinction curve
shown; SMC (dashed), LMC (solid) and Galactic (dotted). Open circles are data points at wavelength λ < 1215˚ A in the rest frame, and
therefore not used in the spectral fitting. The positions of 2175˚ A and 1215˚ A in the rest frame are indicated.
and extinction. However, good quality data resulting from
the proximity [z=0.606; Blustin et al. (2006)] and brightness
of this burst in the UV and optical bands constrained well
the spectral fits, and provided a distinction between them.
The afterglow SED was best fit by the SMC model (χ2= 29
for 33 dof).
GRB 050802 is the only burst in our sample where the
goodness of the spectral fit is improved with the MW model,
(χ2= 88 for 81 dof). The SED of GRB 050802 flattens
out at longer wavelengths (Fig. 3) and this is well fit by a
model with a dust extinction curve that contains the 2175˚ A
absorption feature (i.e. LMC and MW models). The χ2of
the LMC model fit is still acceptable, with χ2= 98 for 81
dof, although it is rejected at the 90% confidence level in
contrast to the MW model, which is only rejected at the
71% confidence level.

Page 8

8
P. Schady, K.O. Mason, M.J. Page, et al.
0.100.150.200.25
E(B-V)
0.0
0.1
0.2
0.3
0.4
NH,X (10
22 cm
-2)
0.020.030.040.05
E(B-V)
0.060.070.080.09
0.00
0.05
0.10
0.15
0.20
NH,X (10
22 cm
-2)
0.150.20 0.250.30
E(B-V)
0.0
0.1
0.2
0.3
0.4
NH,X (10
22 cm
-2)
0.000.020.04
E(B-V)
0.060.08
0.00
0.05
0.10
0.15
0.20
0.25
0.30
NH,X (10
22 cm
-2)
0.000.050.100.15
E(B-V)
0.200.25 0.30
0.6
0.8
1.0
1.2
1.4
NH,X (10
22 cm
-2)
0.08 0.10 0.12
E(B-V)
0.140.16 0.18
0.0
0.1
0.2
0.3
0.4
0.5
NH,X (10
22 cm
-2)
0.10 0.150.20
E(B-V)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
NH,X (10
22 cm
-2)
GRB060512
GRB050318GRB050525GRB050802
GRB050824GRB051111GRB060418
Figure 4. Confidence contours for E(B − V ) vs. NH,Xfrom the spectral model fits to the SEDs of the GRBs in the sample. For each
GRB, the confidence contours shown are taken from the dust extinction model that provided the best fit. The contours are drawn at
∆χ2= 2.3,4.61,9.21, corresponding to 68%, 90% and 99% confidence for two interesting parameters.
5 DISCUSSION
5.1 The 2175˚A Absorption Feature
The 2175˚ A absorption feature falls in the UVOT wave-
length range for the GRB sample used in this paper, and its
prominence in the GRB SEDs provides information on the
graphite content of small grains in the surrounding circum-
burst material. The origin of the 2175˚ A feature is most likely
to be carbonaceous material such as small spherical particles
of graphite (a ? 30 nm), which have a strong feature at this
wavelength and of a similar width (Draine & Lee 1984). Its
strength in the MW extinction curve would require ∼ 15 %
of the solar abundance in carbon to be present in small par-
ticles of this size (Draine 2003), whereas its absence in the
SMC extinction curve can be explained by a difference in
the relative abundances of graphite and silicate grains (Pei
1992).
The evidence for the 2175 ˚ A absorption dip in the
SED of GRB 050802 suggests a larger abundance of small
carbonaceous grains in the surrounding environment of
this burst than is the case for the other bursts. A few
other GRBs have also shown evidence for such a feature
[e.g. GRB 970508; (Stratta et al. 2004; Kann et al. 2006),
GRB 991216; (Vreeswijk et al. 2006)], although the usual
absence of this feature in the spectra of GRBs indicate it to
be rare in GRB host galaxies.
5.2X-ray Absorption vs. UV/Optical Extinction
The amount of dust extinction observed in the afterglow of
GRBs is a measure of the column density of dust grains re-
sponsible for the absorption of UV and optical photons, and
the ratio between the X-ray column density and extinction
gives an estimate of the gas-to-dust ratio in the surround-
ing environment of the GRB. In Fig. 6a we show the region
of AV and NH,X parameter space occupied by our sample
of bright GRBs for the best fit AV and NH,X values de-
termined from the spectral analysis when fitting the SMC
(top panel), the LMC (middle panel), and the MW models
(bottom panel) to the data.
The dashed lines in the figure are plotted as a point of
reference and correspond to the empirical relation observed
between AV and NH,X in the SMC (top), LMC (middle) and
MW (bottom). These are determined from the NH/AV val-
ues reported in the literature for each of these environments,
which are then converted to an NH,X/AV ratio relating to
the column density that would be measured from X-ray ob-
servations of the galaxy if solar abundances were assumed.
The parameterisation of these lines differ in each panel, and

Page 9

Dust and Gas in the Local Environments of Gamma-Ray Bursts
9

0.1
1.0
10.0
SMC

0.1
1.0
10.0
NH,X (SED) (cm
-2)
LMC
0.1 1.010.0
NH,X (X-ray) (cm
-2)
0.1
1.0
10.0
MW
Figure 5. Best-fit NH,Xfrom SED spectral analysis vs. best-fit
NH,X from spectral analysis to X-ray data alone. Top, middle
and bottom panel correspond to the results from the SED spec-
tral analysis using an SMC, LMC and MW model, respectively.
In each panel the dashed line corresponds to NH,X(X-ray) =
NH,X(SED). All data points lie on or very close to this line, il-
lustrating the robustness of our spectral analysis.
correspond to
NH.X
AV
NH,X
AV
NH,X
AV
(SMC) =
1
8
NH
AV(SMC) =1
8(1.6 × 1022) cm−2
(1)
(LMC) =
1
3
NH
AV(LMC) =1
3(0.7 × 1022) cm−2
(2)
(MW) =
NH
AV(MW) = 0.18 × 1022cm−2
(3)
where the factors of1
ties observed in the SMC and LMC (Pei 1992). The NH/AV
relations are taken from Weingartner & Draine (2000) and
Predehl & Schmitt (1995) for the SMC and MW, respec-
tively, and the average of the ratios found by Koornneef
(1982) and Fitzpatrick (1985) are used for the LMC.
The data points in Fig. 6 primarily lie to the right of the
lines of constant NH,X/AV. However, they are confined to
an area of the NH,X–AV parameter space much closer to the
dust-to-gas ratio observed in the MW and Magellanic Clouds
than the region of space occupied by previous data. This
is illustrated in Fig. 6b, where we include the host galaxy
NH,X and AV for two pre-Swift GRB samples taken from
Stratta et al. (2004) (SFA sample) and Kann et al. (2006)
(KKZ sample), which are shown as open triangles and open
squares, respectively. Despite apparent differences between
the Swift sample and previous samples, the distribution in
AV and NH,X between samples remain consistent within
errors. Stratta et al. (2004) did not use the LMC extinction
law to model the afterglow spectrum, and consequently there
are no triangles shown in the centre panel of Fig. 6b.
The X-ray data in the SFA sample are from BeppoSAX
8and1
3account for the lower metallici-
and were taken hours to days after the prompt outburst,
with the earliest observation beginning at ∼ T+4 hours. In
contrast to this, the GRBs in our sample typically have X-
ray data starting hundreds of seconds after the BAT trigger,
and the longest delay between the GRB prompt emission
and the first X-ray observations is 1.7 hours (GRB 050824;
Campana et al. 2005). The larger signal-to-noise provided at
early times when the afterglow is significantly brighter im-
proves the accuracy of our spectral analysis and reduces the
errors associated with the best fit parameters. Stratta et al.
(2004) point out that the quality of the X-ray data for
the majority of the bursts in their sample does not allow
for significant detections of host galaxy absorption in addi-
tion to Galactic, and in only the cases of GRB 990123 and
GRB 010222 were the presence of excess absorption robustly
detected. This is indicated in Fig. 6b, where the NH,X value
for all but two of the SFA sample are upper limits.
Kann et al. (2006) focused primarily on optical and NIR
data, where X-ray column densities used in their analysis
were taken from the literature. In their sample the earliest
X-ray observation was still only ∼ T + 4 hours and the av-
erage delay was nearly 50 hours from the time of initial out-
burst. However, for most of their sample the column density
was determined from higher quality X-ray data taken with
XMM-Newton or Chandra. For this sample the host galaxy
NH,X ranged from an undetectable amount (GRB 021004;
M¨ oller et al. 2002) to (12+7
Stratta et al. 2004), and the mean is (4.6+1.7
which is consistent within 2σ of our sample, which is
(2.2 ± 0.3) × 1021cm−2. Their optical and NIR analysis of
19 GRBs provided a distribution in AV that ranged from a
negligible amount up to AV= 0.80 ± 0.29 mag, which is in
good agreement with our results, for which AV ranges from
AV = 0.12±0.04 to AV = 0.65+0.08
tion is 0.21 ± 0.04 mags and 0.38 ± 0.02 mags for the KKZ
and our sample, respectively.
The high quality of the data in our sample and the
simultaneous spectral fitting of the X-ray and UV/optical
data is providing greater constraints on the spectral mod-
elling, and consequently reducing the errors on the best-fit
parameters. The results from this are that the gas-to-dust
ratios in the local environment of GRBs are in better agree-
ment with those observed in the Milky Way and Magellanic
Clouds than previous data suggest. The relatively large gas-
to-dust ratios in GRB local environments indicated by pre-
vious data were interpreted as evidence of dust destruction
by the GRB, which would cause the value of AV to decrease.
However, the results from the Swift data analysis show little
evidence of this.
Moreover, when interpreting the NH,X to AV ratio in
the GRB local environment compared with that of other en-
vironments, it is necessary to consider more than just the
effect of the GRB emission on AV. Photoionisation of the
gas in the surrounding environment by the GRB X-ray ra-
diation causes NH,X to decrease with time, and the extent
to which the GRB reduces the AV and NH,X will depend on
the properties of the GRB and its local environment, such
as the prompt and afterglow spectral and temporal indices,
and the density and density profile of the absorbing and
extinguishing material.
The lack of evidence for any colour evolution in our
UV/optical data (see section 2.1), provides an upper limit
−6) × 1021cm−2(GRB 010222;
−1.0)×1021cm−2,
−0.07mag. The mean extinc-

Page 10

10
P. Schady, K.O. Mason, M.J. Page, et al.

-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5

-1.5
-1.0
-0.5
0.0
0.5
1.0
Log AV
19 202122
-2)
23 24
Log NH,X (10
21 cm
-1.5
-1.0
-0.5
0.0
0.5
1.0

-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5

-1.5
-1.0
-0.5
0.0
0.5
1.0
Log AV
1920 2122
-2)
2324
Log NH,X (10
21 cm
-1.5
-1.0
-0.5
0.0
0.5
1.0
a)b)
Figure 6. Host galaxy AV vs. NH,X. In both plots solid circles represent the host galaxy AV and NH,Xdetermined from our spectral
analysis using the SMC model (top panel), the LMC model (middle panel) and the MW model (bottom panel). The dashed curves are
the NH,X/AV ratios for each corresponding environment. This was determined using the NH/AV ratios reported in the literature, where
NHis converted to the X-ray equivalent NH,Xvalue assuming a metallicity 1/8 and 1/3 solar for the SMC and LMC, respectively (see
text for details). In the right panel the open triangles and open squares are GRBs taken from the SFA and KKZ sample, respectively.
on the time by which the GRB no longer destroyed signif-
icant amounts of dust. i.e. the time at which colour infor-
mation is available with UVOT, which is typically within
∼ 103s. A limit on the time after which the GRB no longer
photoionises the surrounding environment can also be de-
termined by investigating the change in NH,X over time
during the early stages of the X-ray afterglow. For this pur-
pose, we applied spectral analysis on the early time X-ray
data of the five GRBs in our sample for which there were
data within the first few 100 seconds of the BAT trigger.
Of these GRBs only GRB 060418 shows evidence for evo-
lution in the column density at greater than the 3σ level.
The column density measured from T+84 s to T+114 s
was (1.6 ± 0.10) × 1022cm−2, whereas beyond T+400 s
this was (3.9+1.3
large X-ray flare that peaked at T+135 s (Falcone et al.
2006) and increased the X-ray flux by about an order of
magnitude in ∼ 15 s, which could have caused photoioni-
sation of the circumburst environment out to greater radii
from the source. However, it is possible to misinterpret in-
trinsic spectral evolution as a change in the column den-
sity (Butler & Kocevski 2006). The spectral index indeed
changed from Γ = 2.75 ± 0.05 to Γ = 2.15 ± 0.08 between
the two spectral epochs analysed, and the apparently larger
NH,X could be due to intrinsic curvature of the flare spec-
trum. The column density measured in GRB 060418 no
longer evolves beyond T+300 s, by which time the flare is
over and, therefore, we determine a typical upper limit of
a few hundred seconds on the time interval over which the
GRB photoionises its surrounding environment.
Perna & Lazzati (2002) simulated the effect of the GRB
X-ray and UV emission on NH,X and AV over time, where
−1.2) × 1021cm−2. GRB 060418 had a very
they assumed the same initial hydrogen column density
NH = 1022cm−2and optical extinction AV = 4.5 mag
in all cases, but varied the compactness of the absorbing
medium and, therefore, also the number density, nH. Their
simulations indicate that the intense radiation emitted by a
GRB is capable of photoionising and destroying all gas and
dust out to a radius of ∼ 3 pc within a few tens of seconds.
Further evolution in the gas and dust column resulting from
this is, therefore, not expected more than a few hundreds of
seconds after the peak of the emission, consistent with our
observations.
This, therefore, places a lower limit of a few parsecs
on the scale of the absorbing and extinguishing systems
detected at the redshifts of GRBs for six of the seven in
our sample. If the dust and gas were any closer it would
have been fully destroyed and photoionised. Furthermore,
Perna & Lazzati (2002)’s work shows that if the dust and
gas extends out to a few tens of parsecs the AV and NH,X
measured local to the GRB do not change significantly. This
would suggest that the host galaxy dust and gas systems
probed by our sample lie a few tens of parsecs from the
source, which is consistent with findings by Prochaska, Chen
& Bloom (2006), who placed a lower limit of∼
the locations of host galaxy absorption systems from GRBs.
This suggests that the dust-to-gas ratios measured are ef-
fectively unaltered by the GRB and a fair representation of
their local environment. It is, therefore, not so surprising
that there is no significant deviation in the GRB local envi-
ronment NH,X/AV ratio when compared to the Milky Way
or Magellanic Clouds.
To verify that the effect of the GRB on the dust and
gas in its local environment is of little significance, we in-
> 50 pc on

Page 11

Dust and Gas in the Local Environments of Gamma-Ray Bursts
11
49.550.050.551.0
log Eiso (10
51.5
52 ergs)
52.052.553.0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
AV (mag)
050318
050525
050802
050824
051111
060418
060512
Figure 7. Host galaxy visual extinction against isotropic equiva-
lent energy in γ-rays for a sample of 16 bursts. Solid data points
correspond to the GRBs studied in this paper. Open data points
are GRBs from a pre-Swift sample of GRBs where the Eisovalues
are taken from Bloom et al. (2001) and the AV values are from
Kann et al. (2006) (square) and Stratta et al. (2004) (triangle).
vestigate further alternative methods of indirectly detecting
the process of dust destruction.
5.3Alternative Indicators Of Dust Destruction
Dust destruction, either by sublimation or shattering is de-
pendent on both the energy of the photons absorbed and
on the radiation flux, where a less-luminous burst will be
less effective at destroying the smaller, UV-absorbing dust
grains. If the radiation emitted by a GRB destroys signifi-
cant levels of dust in its surrounding environment a correla-
tion should exist between the energy released by the burst
and the amount of dust destroyed, which consequently af-
fects the amount of extinction observed in the GRB after-
glow spectrum.
To test this correlation we compute the isotropic equiv-
alent energy emitted in the 15 – 150 keV γ-ray energy band
(Eiso(γ)) for our sample of GRBs. A k-correction is applied
using the method described by Bloom, Frail & Sari (2001)
to convert the prompt energy observed to the same comov-
ing rest frame bandpass (15 – 150 keV). Fig. 7 shows the
host galaxy AV against the k-corrected Eiso(γ) for the seven
GRBs in our sample (solid circles), as well as a further fif-
teen pre-Swift GRBs (open squares or triangles). All the
GRBs included in Fig. 7 have spectroscopically measured
redshifts and estimates of the host galaxy AV, where the
host galaxy AV for the pre-Swift sample are taken from
either Kann et al. (2006) (open squares) or Stratta et al.
(2004) (open triangles). A similar k-correction is applied to
the non-Swift GRB sample to determine Eiso(γ) in the same
rest-frame energy range as the Swift sample.
Twelve of the pre-Swift GRBs had no spectral informa-
tion and, therefore, the spectral shape had to be assumed.
Band et al. (1993) found the GRB prompt emission to be
well modelled by a smoothed, broken, power law with a soft
and hard spectral index, α and β, respectively, and spectral
break E0. Those GRBs with no spectral information are as-
sumed to have a Band spectrum, and their Eiso(γ) is taken
from the average of 54 values, each computed with a dif-
ferent set of (α,β,E0). The set of 54 (α,β,E0) combinations
are taken from fits to 54 BATSE GRBs done by Band et al.
(1993).
Based on the Spearman’s rank test the correlation in
Fig. 7 is not significant, with the log of Eiso and AV show-
ing a correlation coefficient of only rs = −0.21 with a null-
hypothesis probability of P=0.60. This is consistent with
the findings of Nardini et al. (2006), whose analysis showed
a sample of 23 pre-Swift GRBs to show no correlation be-
tween the isotropic γ-ray emitted energies and the opti-
cal luminosities. This, therefore, suggests that any dust-
destruction caused by the GRB intense radiation is not sig-
nificant enough to affect greatly their observed extinction
properties, as already indicated in Fig. 6 and discussed in
section 5.2.
5.4 Extension to Dark Bursts
By imposing the criteria that the selected GRBs have a
UV/optical afterglow detected by UVOT, a bias is intro-
duced that favours afterglows with low source-frame extinc-
tion. The determined range in host-galaxy extinction may,
therefore, represent the low end of the extinction distribu-
tion. Given the association of GRBs with star formation,
it is likely that the lack of an optical counterpart in some
bursts is due to large amounts of dust in the host galaxy,
which blocks out the UV/optical afterglow. In this case the
distribution of host galaxy AV may extend much further
than that which is observed in our sample. GRBs with no
associated optical counterpart are typically referred to as
‘dark’, although the various possible causes for the lack of
an optical counterpart make the term ambiguous (Rol et al.
2005; Roming & Mason 2006a). In this paper we shall use
the term dark to refer to those GRBs that have no after-
glow detected at the 3σ level above background in any of
the UVOT filters within one hour of the prompt emission.
There are seven Swift GRBs that satisfy both our def-
inition of dark and which have spectroscopically measured
redshifts, where z < 5. At higher redshifts than this a GRB
would always appear dark to UVOT due to the redshifting of
the Lyman break below the UVOT energy band. To estimate
the amount of dust in the local environment of this sample
of GRBs, we initially assume the relation between NH,X
and AV to be linear, and thus use the X-ray column den-
sity, NH,X, of the GRB host galaxy to trace the amount of
dust-extinction affecting the GRB UV/optical afterglow. We
compare this with the dust extinction estimated for a sam-
ple of 34 optically bright Swift detected GRBs. This sample
is made up of those GRBs with an UV/optical afterglow de-
tected by UVOT and a corresponding spectroscopic redshift,
up until GRB 060912; this includes our original sample of
seven UVOT bright GRBs listed in table 3. All data were
taken from epochs where no spectral evolution is observed,
and it was reduced in the same way as described in sec-

Page 12

12
P. Schady, K.O. Mason, M.J. Page, et al.
tion 2.2. The intrinsic column density was then determined
from a single power law fit with two absorbers; one at z = 0
with the column density fixed at the Galactic value, and a
second one at the redshift of the burst with the column den-
sity and X-ray spectral slope, βX, left as free parameters.
The resulting distribution in NH,X within the two popula-
tions of bursts is shown in Fig. 8, where the solid histogram
corresponds to the dark bursts, and the UVOT bright bursts
are represented by the dashed histogram. Fourteen bright
GRBs had negligible absorption at the host galaxy, repre-
sented in the smallest bin of NH,X. The mean logarithm of
the X-ray column density, over and above the Galactic col-
umn is < 20.54 cm−2and 22.2 ± 0.1 cm−2for the bright
and dark population of GRBs, respectively, suggesting that
dark GRBs reside in denser environments.
Assuming GRB host galaxies to have an NH,X and AV
relation similar to that observed in the SMC (Eqn. 1), we use
the measured NH,X to determine AV. Doing this we estimate
a mean visual extinction of ?AV(dark)? = 10.1 mags for the
dark population of bursts shown in Fig. 8. The most ab-
sorbed GRB in the sample of dark GRBs was GRB 060510B,
which had a rest-frame column density NH,X= (3.54+0.24
1022cm−2and, therefore, an estimated rest-frame visual ex-
tinction of AV≈ 17.7 mags. However, GRB 060510B had
an R-band detection of R ∼ 21 at around 12 minutes af-
ter the prompt emission, making the inferred extinction
rather unrealistic, which suggests that the NH,X/AV ratio
of 2 × 1021cm−2is likely too small for the most absorbed
of GRBs.
Instead, we use the host galaxy gas-to-dust ratios de-
termined in our spectral analysis to acquire a mean value
of NH,X/AV, where we use the values determined from the
SMC model. This corresponds to:
−0.23)×
?NH,X
AV
? = 6.7 × 1021cm−2
(4)
Using this relation we estimate a range in rest-frame AV of
0.4 < AV < 5.3, with a mean of ?AV? = 3.0 mags. The
overlap in rest-frame visual extinction in the UVOT dark
and bright population of GRBs suggests that several factors
may contribute to the lack of an optical afterglow. However,
the mean value of ?AV? = 3.0 mags is almost eight times
the mean extinction observed in the bright GRBs listed in
Table 3, indicating that the local environments of UVOT
dark GRBs are much dustier than those with UVOT de-
tected optical counterparts. This could explain, or at least
contribute, to the lack of an afterglow blueward of 5500˚ A.
6 CONCLUSIONS
In this paper we use the SEDs of seven optically bright
GRBs, covering the optical, UV and X-ray energy bands, to
determine the amount of dust extinction and soft X-ray ab-
sorption present in the local environments of the GRB. We
use the Milky Way (MW) and the Large and Small Mag-
ellanic Cloud (LMC and SMC) extinction laws to model
the host galaxy dust extinction dependence on wavelength.
These show a decreasing prominence in the strength of the
2175˚ A feature, and increasing levels of far-ultraviolet ex-
tinction, respectively.
All GRBs but one were best-fit by the SMC or LMC ex-
20.0 20.5 21.0
Log [NH,X] (cm
21.522.0
-2)
22.523.0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Number of GRBs
Bright GRBs
Dark GRBs
Figure 8. Distribution of the X-ray column density at the red-
shift of the GRB for a sample of GRBs with (dashed), and with-
out (solid) a UVOT detected afterglow, where all GRBs have a
spectroscopically measured redshift. The smallest NH,Xbin rep-
resents those GRBs with no measurable X-ray column density
above the Galactic absorption.
tinction law, with the SMC extinction law typically giving a
better fit. Only in the case of GRB 050802 did a model with
a Galactic extinction law at the host galaxy provide a better
fit, although SMC and LMC extinction laws were rejected
with only 97% and 90% confidence, respectively. This, there-
fore, suggests that small graphite grains responsible for the
2175˚ A feature are not predominant in GRB host galaxies.
For six of the seven GRBs in our sample an absorption
and extinction system was detected at the redshift of the
GRB, which must be located at least several parsecs from
the source to have survived the intense radiation emitted by
the GRB. However, the gas-to-dust ratios measured in the
host galaxies of our sample of bursts are lower than previ-
ously suggested from analysis of pre-Swift GRBs, and consis-
tent with those observed in the Milky Way and Magellanic
Clouds. There is, therefore no evidence of dust destruction
by the GRB in its circumburst environment provided by the
spectral analysis alone. However, this does not rule out the
destruction and photoionisation of dust and gas within a few
parsecs of the GRB, which is expected to take place within
the first few tens of seconds of the GRB prompt emission.
The host galaxy visual extinction observed in our sam-
ple ranges from AV = 0.12 ± 0.04 to 0.65+0.08
ever, the requirement that the GRBs analysed have a de-
tected optical counterpart introduces selection effects that
favour GRBs with a low AV. Using the amount of host
galaxy NH,X as an indicator of the amount of visual extinc-
tion undergone in the GRB local environment, we estimate
that those GRBs with no optical counterpart have, on aver-
age, values of AV almost a factor of eight larger than those
observed in the sample of bright GRBs studied in this pa-
−0.07mags. How-

Page 13

Dust and Gas in the Local Environments of Gamma-Ray Bursts
13
per. This would suggest that the host galaxies of dark GRBs
are intrinsically dustier than those of GRBs with UVOT de-
tected optical counterparts, which could account for the lack
of an optical afterglow.
ACKNOWLEDGEMENTS
We gratefully acknowledge the contributions of all members
of the Swift team. PS acknowledges the support of a PPARC
Studentship. PS would also like to thank David Morris and
Claudio Pagani for all their patience and help regarding the
reduction of XRT data.
REFERENCES
Band, D., et al. 1993, ApJ, 413, 281
Berger, E., & Mulchaey, J. 2005, GCN 3122
Bloom, J. S., Frail, D. A., & Sari, R. 2001, AJ, 121, 2879
Bloom, J. S., et al. 2006, GCN, 5217
Blustin, A. J., et al. 2006, ApJ, 637, 901
Burrows, D. N., et al. 2005a, Space Sci. Rev., 120, 165
Burrows, D. N., et al. 2005b, Science, 309, 1833
Butler, N. R., et al. 2006, (astro-ph/0606763)
Butler, N. R., & Kocevski, D. 2006, submitted to ApJ
(astro-ph/0612564)
Campana, S., et al. 2005, GCN, 3872
Campana, S., et al. 2006, Nature, 442, 1008
De Pasquale, M., & Cummings, J. 2006, GCN 5130
Dickey, J. M., & Lockman, F. J. 1990, ARA&A, 28, 215
Draine, B. T. 2000, ApJ, 532, 273
Draine, B. T. 2003, Annual Rev. of A&A, 41, 241
Draine, B. T., & Lee, H. M. 1984, ApJ, 285, 89
Dupree, A. K.., et al., 2006, GCN, 4969
Falcone, A. D., et al. 2006, GCN, 5009
Fitzpatrick, E. L. 1985, ApJ, 299, 219
Foley, R. J., Chen, H.-W., Bloom, J., & Prochaska, J. X.
2005, GCN 3483
Fruchter, A., Krolik, J. H., & Rhoads, J. E. 2001, ApJ, 563,
597
Fynbo, J. P. U., et al. 2005a, GCN 3749
Fynbo, J. P. U., et al. 2005b, GCN 3874
Galama, T. J., & Wijers, R. A. M. J. 2001, ApJ, 549, L209
Gehrels, N., et al. 2004, ApJ, 611, 1005
Halpern, J. P., & Mirabal, N. 2005, GCN, 3907
Hill, J. E., et al. 2004, Proceedings of the SPIE, 5165, 217
Hjorth, J., et al. 2003, Nature, 423, 847
Kann, D. A., Klose, S., & Zeh, A.(KKZ) 2006, ApJ, 641,
993
Koornneef, J. 1982, A&A, 107, 247
Kulkarni, S. R., et al. 1998, Nature, 395, 663
Le Floc’h, E., Charmandaris, V., Forrest, W. J., Mirabel,
I. F., Armus, L., & Devost, D. 2006, ApJ, 642, 636
M¨ oller, P., et al. 2002, A&A, 396, L21
Morrison, R., & McCammon, D. 1983, ApJ, 270, 119
Nardini, M., Ghisellini, G., Ghirlanda, G., Tavecchio, F.,
Firmani, C., & Lazzati, D. 2006, A&A, 451, 821
Oates, S. R., et al. 2006,submitted to MNRAS
Pei, Y. C. 1992, ApJ, 395, 130
Perna, R., & Lazzati, D. 2002, ApJ, 580, 261
Perna, R., Lazzati, D., & Fiore, F. 2003, ApJ, 585, 775
Poole, T. S., et al. 2005, GCN 4263
Predehl, P., & Schmitt, J. H. M. M. 1995, A&A, 293, 889
Prochaska, J. X. 2005, GCN 4271
Prochaska, J. X., Chen, H.-W., & Bloom, J. S. 2006, ApJ,
648, 95
Rol, E., Wijers, R. A. M. J., Kouveliotou, C., Kaper, L., &
Kaneko, Y. 2005, ApJ, 624, 868
Roming, P. W. A., et al. 2005, Space Sci. Rev, 120, 95
Roming, P. W. A., & Mason, K. O. 2006a,
Conf. Proc. 838: Gamma-Ray Bursts in the Swift Era,
838, 224
Schady, P., & Campana, S. 2005, GCN 3878
Schady, P., & Falcone, A. D. 2006, GCN 4978
Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ,
500, 525
Still, M., et al. 2005, ApJ, 635, 1187
Stratta, G., Fiore, F., Antonelli, L. A., Piro, L., & De
Pasquale, M.(SFA) 2004, ApJ, 608, 846
Tagliaferri, G., et al. 2005, A&A, 443, L1
Tanvir, N. R., et al. 2004, Astronomical Society of the Pa-
cific Conference Series, 312, 275
Vreeswijk, P. M., et al. 2006, A&A, 447, 145
Weingartner, J. C., & Draine, B. T. 2000, Bulletin of the
American Astronomical Society, 32, 1466
Woosley, S. E. 1993, ApJ, 405, 273
Zhang, B., Fan, Y. Z., Dyks, J., Kobayashi, S., M´ esz´ aros,
P., Burrows, D. N., Nousek, J. A., & Gehrels, N. 2006,
ApJ, 642, 354
AIP