Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~kgordon/papers/PS_files/SMC_ext.ps.gz Äàòà èçìåíåíèÿ: Thu Apr 24 02:49:35 2003 Äàòà èíäåêñèðîâàíèÿ: Tue May 27 07:53:34 2008 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http www.astronomy.ru forum index.php topic 4644.0.html

Starburst­like Dust Extinction in the Small Magellanic Cloud
Karl D. Gordon and Geoffrey C. Clayton
Department of Physics & Astronomy, Louisiana State University
Baton Rouge, LA 70803
email: (gordon,gclayton)@fenway.phys.lsu.edu
ABSTRACT
The recent discovery that the UV dust extinction in starburst galaxies is similar
to that found in the Small Magellanic Cloud (SMC) motivated us to re­investigate
the ultraviolet (UV) extinction found in the SMC. We have been able to improve
significantly on previous studies by carefully choosing pairs of well matched reddened
and unreddened stars. In addition, we benefited from the improved S/N of the
NEWSIPS IUE data and the larger sample of SMC stars now available. Searching the
IUE Final Archive, we found only four suitable early­type stars that were significantly
reddened and had well matched comparison stars. The extinction for three of these
stars is remarkably similar. The curves are roughly linear with – \Gamma1 and have no
measurable 2175 š A bump. The fourth star has an extinction curve with a significant
2175 š A bump and weaker far­UV extinction. The dust along all four sightlines is
thought to be local to the SMC. There is no significant Galactic foreground component.
The first three stars lie in the SMC Bar and the line­of­sight for each of them passes
through regions of recent star formation. The fourth star belongs to the SMC Wing
and its line­of­sight passes though a much more quiescent region. Thus, the behavior
of the dust extinction in the SMC supports a dependence of dust properties on star
formation activity. However, other environmental factors (such as galactic metallicity)
must also be important. Dust in the 30 Dor region of the LMC, where much more
active star formation is present, does not share the extreme extinction properties seen
in SMC dust.
Subject headings: dust, extinction -- galaxies: individual (SMC) -- galaxies: ISM --
galaxies: starburst -- ultraviolet: ISM
1. Introduction
The interstellar dust in the Small Magellanic Cloud (SMC) has gained new importance
with the discovery that its ultraviolet (UV) extinction is uniquely similar to extinction found in
starburst galaxies (Calzetti et al. 1994; Gordon, Calzetti, & Witt 1997). Dust in the Galaxy and
the Large Magellanic Cloud does not show starburst galaxy­like extinction. Starburst galaxies are

-- 2 --
the only type of galaxy that has been detected from low to high (z ? 2:5) redshifts (Kinney et
al. 1993; Steidel et al. 1996; Lowenthal et al. 1997; Trager et al. 1997; Franx et al. 1997). These
galaxies serve as an excellent probe of galaxy evolution through the study of their star formation
and metal enrichment rates at different redshifts. The derivation of these rates for high redshift
galaxies is sensitive to the adopted UV dust extinction (Pettini et al. 1997; Madau, Pozzetti,
& Dickinson 1997). Thus, understanding the physical processes responsible for producing the
starburst­like dust seen in the SMC is important to the modeling of starburst galaxies.
In addition, the study of the UV extinction curve in Local Group galaxies promises to help
in determining the sizes, shapes, and materials which make up dust grains. Ultraviolet extinction
curves have been determined in the Milky Way, the Large Magellanic Cloud (LMC), the SMC,
and (tentatively) M31. The extinction curves in these four galaxies paint a complex picture of
the environmental dependence of dust properties. In the Milky Way, Cardelli, Clayton, & Mathis
(1989, hereafter CCM) found that the infrared to UV extinction curves can be described fairly
well by a relationship which depends on only one parameter, R V = A V =E(B \Gamma V ) which is a
measure of the overall dust grain size. There are significant deviations from the CCM relation
in different Galactic environments (Mathis & Cardelli 1992). In the LMC, the UV extinction
curves show a distinctly different behavior between the 30 Dor region (a mini­starburst [Walborn
1991]) and the rest of the LMC (Clayton & Martin 1985; Fitzpatrick 1985, 1986). The 2175 š A
bump is weaker and the far­UV rise is stronger in the 30 Dor region than in the rest of the LMC
which have strengths similar to the average Galactic extinction curve. In the SMC, the average
extinction curve is characterized by a roughly linear rise (versus – \Gamma1 ) increasing toward shorter
wavelengths without a 2175 š A bump (Pr`evot et al. 1984; Thompson et al. 1988). Yet, there is one
sightline which has an extinction curve with a significant 2175 š A bump (Lequeux et al. 1982).
In M31, the extinction curve is consistent with that of the average Galactic extinction within
the associated uncertainties, although the 2175 š A bump may be weak (Bianchi et al. 1996). The
complex behavior in these four galaxies implies that the physical properties of dust grains may
be dependent on a multitude of environmental parameters. Two of these are metallicity and star
formation activity both of which may affect the overall composition and size distribution of dust
grains (Clayton & Martin 1985; Fitzpatrick 1985; CCM; Gordon et al. 1997).
2. Previous Work
There have been a number of papers based in part or entirely on the derivation of SMC UV
extinction curves (Lequeux et al. 1982; Pr`evot et al. 1984; Thompson et al. 1988; Rodrigues et al.
1997). These studies used the pair method to derive extinction curves (see x3.1). The extinction
curves were derived by comparing a reddened SMC star to an unreddened SMC star or (more
commonly) a group of unreddened SMC stars. Using unreddened stars of the same temperature
and luminosity as the reddened stars is crucial to determining accurate extinction curves. All of
the previous work suffered from poor spectral matching between reddened and unreddened stars

-- 3 --
due in part to the paucity of accurate optical and/or UV 2D spectral types. Recent work has
provided 2D spectral types in both the optical (Garmany, Conti, & Massey 1987; Massey et al.
1995; Lennon 1997) and UV (Neubig & Bruhweiler 1997). The previous studies also suffered from
low signal­to­noise spectra due to the faintness of the stars in the SMC. The signal­to­noise ratio
and absolute calibration of all the spectra taken by the International Ultraviolet Explorer (IUE)
have been greatly improved through the NEWSIPS reduction routines (Nichols et al. 1994). Also,
there is now a larger sample of SMC stars observed with IUE than were available a decade or more
ago when the previous studies were done. Thus, the UV extinction curves in the SMC can be
greatly improved using star pairs that are better spectral matches and have higher S/N spectra.
We list the reddened stars used in previous works in Table 1 along with comments on the quality
of the derived extinction curves.
Most of the previous work has been done by three groups, one based in France, one in
England, and one in Brazil. The final results of the French group are presented in Lequeux et
al. (1982) and Pr'evot et al. (1984) which supersede earlier work (Rocca­Volmerange et al. 1981;
Lequeux et al. 1984). In Pr'evot et al. (1984), this group determined the best (and often used)
average UV extinction curve for the SMC from only three reddened stars (AZV 18, AZV 398,
& SK 191). One drawback to this paper is that the individual extinction curves are not shown,
just the average which makes it difficult to assess the accuracy of the individual curves. The
star, SK 191, is an essentially unreddened star, making its extinction curve useless. In addition,
Pr'evot et al. (1984) derived extinction curves for two other stars, but excluded them as having
``anomalous'' extinction curves. These other two stars were AZV 393 (Lequeux et al. 1984) and
AZV 456 (Lequeux et al. 1982). AZV 393 is an unreddened star with a UV spectral type of B3 Ia
(Neubig & Bruhweiler 1997). The extinction curve for AZV 456 is real and significantly different
from the average SMC extinction (Lequeux et al. 1982). This is the only curve in the SMC which
shows the presence of the 2175 š A bump. However, the extinction curve, derived by this group for
AZV 456, suffers from a spectral mismatch which shows up in an asymmetric 2175 š A bump. This
is due to a mismatch between the reddened and comparison stars in the Fe III lines around 2000
š A which are luminosity and temperature sensitive (Cardelli, Sembach, & Mathis 1992; Neubig
& Bruhweiler 1997). The validity of the Pr'evot et al. (1984) average extinction curve is called
into question through their inclusion of a highly uncertain extinction curve (SK 191) and their
exclusion of the ``anomalous'' extinction curve of AZV 456.
The final results of the English group are contained in Thompson et al. (1988) which
supersedes earlier work (Nandy et al. 1982; Bromage & Nandy 1983). In Thompson et al. (1988)
a sample of four reddened stars (AZV 18, 56, 373, & SK 191), when compared to the unreddened
star AZV 264, produced a UV extinction curve consistent with the Pr'evot et al. (1984) average
curve. This was done instead of calculating individual extinction curves as all of the reddened
stars, except for AZV 18, are only lightly reddened (\Delta(B \Gamma V ) ! 0:13) making their extinction
curves highly uncertain.
The Brazilian group's work is presented in Rodrigues et al. (1997). They calculate UV

-- 4 --
Table 1. Previous Extinction Work
AZV a SK b ref c quality d comments
18 e 13 P84,T88 good good comparison
20 14 R97 bad spectral type too late (A0Ia)
56 31 T88 bad too bright for reddened star
126 R97 bad unreddened star
211 74 R97 bad spectral type too late (A0Ia)
373 119 T88 bad very low E(B \Gamma V )
398 e P84,R97 good good comparisons
456 e 143 L82,R97 poor spectral mismatch
191 P84,T88 bad unreddened star
a Azzopardi, Vigneau, & Macquet 1975; Azzopardi & Vigneau
1979; 1982
b Sanduleak 1968, 1969
c L82 = Lequeux et al. 1982; P84 = Pr`evot et al. 1984; T88 =
Thompson et al. 1988; R97 = Rodrigues et al. 1997
d The quality of the extinction curve was good (reddened star
and at least one good comparison star), poor (reddened star and
no good comparison stars), or bad (star unsuitable for extinction
curve work)
e Used in this study

-- 5 --
extinction curves for five stars -- AZV 20, 126, 211, 398, & 456. Two of these stars, AZV 20 &
211, have A0 Ia spectral types. Stars with spectral types later than about B5 are normally not
used for UV extinction work. This is due to their lower UV fluxes and the rapid change in their
intrinsic spectra as a function of spectral type (Rodrigues et al. 1997). These problems resulted
in extinction curves for these two stars (shown in Fig. 2 of Rodrigues et al. [1997]) which are
very noisy (AZV 20) or have odd changes of slope (AZV 211). The extinction curve for AZV 126
presented by Rodrigues et al. (1997) has a very shallow slope which is a result of comparing the
unreddened star (AZV 126) to other unreddened stars of earlier spectral types. The UV spectral
type of AZV 126 is B1 II and the UV spectral types of the comparison stars, AZV 61, 317, and
454, are O5 III, O7 Ia, and O9 V,respectively (Neubig & Bruhweiler 1997). The extinction curves
for AZV 398 and 456 presented by Rodrigues et al. (1997) reproduce the work of Pr'evot et al.
(1984).
From a careful analysis of the previous work on the extinction curves in the SMC, only three
reddened stars (AZV 18, 398, & 456) emerge as good candidates for extinction curve work.
3. Extinction Curves
We collected all the stars with spectral types between O9 and B3 in the SMC with IUE low
dispersion spectra in order to have the largest possible sample of reddened and unreddened stars.
Optical and UV spectral types were taken from the literature (Bouchet et al. 1985; Garmany et
al. 1987; Massey et al. 1995; Lennon 1997; Neubig & Bruhweiler 1997). From this list, candidate
reddened stars were identified as having red (B \Gamma V ) colors when compared to other stars with
similar optical and UV spectral types.
For each candidate reddened star, we attempted to identify a comparison star which satisfied
the three Fitzpatrick criteria (Fitzpatrick 1985). In addition, we required the \Delta(B \Gamma V ) between
the reddened and comparison star to be greater than 0.15. The first Fitzpatrick criterion requires
that \Delta(U \Gamma B)=\Delta(B \Gamma V ) has a value appropriate for reddening due to dust. The average value
of \Delta(U \Gamma B)=\Delta(B \Gamma V ) for the SMC is 0:81 \Sigma 0:11 (Bouchet et al. 1985). The second Fitzpatrick
criterion requires that \DeltaV o , which is the difference between the dereddened V magnitude of the
reddened star and the V magnitude of the comparison star, is less than 0.8 magnitudes. This
criterion insures the luminosities of the two stars are comparable as all stars in the SMC are
located at approximately the same distance. The reddened star's V magnitude was dereddened
using R V = 2:72 (Bouchet et al. 1985) and the \Delta(B \Gamma V ) between the reddened and comparison
stars. The third Fitzpatrick criterion requires a good luminosity and temperature match (i.e. UV
spectral type) between the detailed spectra of the reddened and comparison stars. The errors
in the the UV spectral types are ¸1 temperature subclass and ¸1 luminosity class (Neubig &
Bruhweiler 1997).
We applied these criteria to the candidate reddened stars and found only one star, AZV 214,

-- 6 --
in addition to the three stars identified in x2, which satisfied all three of the Fitzpatrick criteria.
Thus, we are left with a small sample of 4 reddened stars with which we can study the UV
extinction in the SMC. Table 2 lists the spectral and photometric data for the four reddened and
four comparison stars. The UV spectral types are from Neubig & Bruhweiler (1997). The optical
and infrared photometry was taken from Bouchet et al. (1985), except for AZV 70 which was taken
from Ardeberg & Maurice (1977). The uncertainties in the optical and infrared photometry are
0.02, 0.045, 0.02, 0.028, 0.032, and 0.027 for V , (U \Gamma V ), (B \Gamma V ), (J \Gamma V ), (H \Gamma V ), and (K \Gamma V ),
respectively (Ardeberg & Maurice 1977; Bouchet et al. 1985; Bouchet 1997). Table 3 displays the
\Delta(B \Gamma V ), \Delta(U \Gamma B)=\Delta(B \Gamma V ), and \DeltaV o values for the reddened and comparison pairs.
3.1. Calculation of Extinction Curves
We calculated extinction curves using the standard pair method (Massa, Savage, & Fitzpatrick
1983) which uses a reddened star and an appropriately chosen unreddened comparison star. The
extinction curves were calculated using
E(–) = \Delta(– \Gamma V )
\Delta(B \Gamma V ) = m(– \Gamma V ) r \Gamma m(– \Gamma V ) c
(B \Gamma V ) r \Gamma (B \Gamma V ) c
(1)
where the subscripts r and c refer to the reddened and comparison stars, respectively. Individual
short (SWP) and long (LWR and LWP) IUE spectra were coadded resulting in a single spectrum
extending from 1300 š A to 3200 š A. The spectra were coadded using the nearest neighbor method
with bad points (as defined by the NEWSIPS quality vector) excluded. The individual spectra
were weighted by their exposure times. The resulting long and short wavelength spectra were
binned to the instrumental resolution of ¸5 š A. The two spectra were then merged at the maximum
wavelength in the short wavelength spectrum. Table 4 tabulates the IUE spectra we used.
Uncertainties in the extinction curves were calculated using the method of Massa, Savage, &
Fitzpatrick (1983) and Cardelli, Sembach, & Mathis (1992). It is no longer necessary to estimate
the uncertainties in the IUE fluxes as these are calculated in the NEWSIPS reduction. The
uncertainties in the extinction curve were calculated using
oe[E(–)] = E(–)
s `
oe[\Delta(– \Gamma V )]
\Delta(– \Gamma V )
' 2
+
`
oe[\Delta(B \Gamma V )]
\Delta(B \Gamma V )
' 2
(2)
where
oe[\Delta(B \Gamma V )] =
q
oe[(B \Gamma V ) r ] 2 + oe[(B \Gamma V ) c ] 2 ; (3)
oe[\Delta(– \Gamma V )] =
q
oe[m(–) r ] 2 + oe[V r ] 2 + oe[m(–) c ] 2 + oe[V c ] 2 ; (4)
oe[m(–)] = \Gamma2:5
2 log
`
F (–) \Gamma oe[F (–)]
F (–) + oe[F (–)]
'
; (5)
oe[F (–)] =
q
oe[F (–) NEWSIPS ] 2 + [oe repeat F (–)] 2 ; (6)

-- 7 --
Table 2. Stellar Data
UV Optical
type a AZV SK sp. type sp. type ref b V (U­V) (B­V) (J­V) (H­V) (K­V)
r 18 13 B3 Ia B2 Ia L97 12.40 ­0.75 0.01 ­0.03 ­0.05 ­0.08
c 462 145 B2 Ia B1.5 Ia L97 12.54 ­1.04 ­0.14 0.35 0.40 0.41
r 214 B2 Ia B3 Iab G87 13.35 ­0.72 0.05 0.02 0.00 ­0.03
c 380 120 B1 1a: B0.5: B85 13.50 ­1.00 ­0.11 0.27 0.29 0.37
r 398 O9 Ia: O9.7 Ia B85 13.86 ­0.70 0.08 ­0.22 ­0.25 ­0.26
c 289 103 O9 Ia B0 I B85 12.37 ­1.09 ­0.15 0.24 0.29 0.34
r 456 143 O8 II O9.7 Ib B85 12.90 ­0.69 0.07 ­0.08 ­0.10 ­0.10
c 70 35 O9 Ia O9.5 Iw H83 12.39 ­1.13 ­0.16 \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta \Delta
a r = reddened star, c = comparison star
b H83 = Humphreys 1983, B85 = Bouchet et al. 1985, G87 = Garmany et al. 1987, L97 = Lennon 1997
Table 3. Fitzpatrick Criteria
reddened comparison \Delta(B \Gamma V ) \Delta(U \GammaB)
\Delta(B \GammaV ) \DeltaV o
AZV 18 AZV 462 0:15 \Sigma 0:03 0:93 \Sigma 0:42 \Gamma0:55 \Sigma 0:09
AZV 214 AZV 380 0:16 \Sigma 0:03 0:75 \Sigma 0:31 0:59 \Sigma 0:09
AZV 398 AZV 289 0:23 \Sigma 0:03 0:70 \Sigma 0:26 0:86 \Sigma 0:10
AZV 456 AZV 70 0:24 \Sigma 0:03 0:75 \Sigma 0:23 \Gamma0:13 \Sigma 0:10

-- 8 --
F (–) is the IUE measured flux; oe[F (–) NEWSIPS ] is the uncertainty NEWSIPS calculates;
oe repeat ¸ 5% is the estimated uncertainty in the relative IUE calibration (the difference between
two spectra of the same object taken by IUE [Garhart & Nichols 1994]). We have not included
an error term for temperature or luminosity mismatch as they are not easily quantifiable. Also,
Cardelli et al. (1992) have shown that these uncertainties are smaller than those calculated
using equation 2. The uncertainties (as calculated in equation 2) in the UV extinction curves
can be divided into two parts. One part arises from random uncertainties in the UV data and
the magnitude of this uncertainty can be reduced by binning the UV data. The other part
arises from random uncertainties in the optical photometry and relative calibration of IUE which
cannot be reduced by binning the UV data. Thus, a lower limit to the uncertainty in the UV
extinction can be calculated by assuming oe[F (–) NEWSIPS ] = 0 in equation 6. The resulting
minimum uncertainties at all UV wavelengths are 22, 19, 13, & 13% for AZV 18, 214, 398, & 456,
respectively. This illustrates the drawback of using moderately reddened stars for extinction curve
work.
We fit each extinction curve with the Fitzpatrick & Massa (1990, hereafter FM)
parameterization of the UV extinction curve. This parameterization has a functional form with 3
terms. The first is a linear term (c 1 [y­intercept] and c 2 [slope]). The second is a Lorentzian­like
``Drude'' profile for the 2175 š A bump (c 3 [strength], x o [bump center], and fl [bump width]). The
third is a curvature term for the far­UV (x ? 5:9 ¯m \Gamma1 , c 4 [strength]).
3.1.1. AZV 18
The extinction curve for AZV 18 has been determined previously (Pr`evot et al. 1984;
Thompson et al. 1988). The comparison star which best satisfies the Fitzpatrick criteria is
AZV 462. This star was one of the three comparison stars used in Pr'evot et al. (1984). We only
used the extinction curve from the comparison star with the best spectral match. Averaging
extinction curves made with multiple comparison stars degrades the final extinction curve since
the most accurate extinction curve has then been averaged with less accurate extinction curves.
Figure 1a displays the spectra of AZV 18 and AZV 462. The extinction curve for AZV 18 is shown
in Figure 1b. The FM fit parameters are tabulated in Table 5.
3.1.2. AZV 214
The UV extinction curve for the star AZV 214 has never been calculated previously. This
is likely due to the presence of another early­type star only 7: 00 5 away. Both stars were included
within the IUE observing aperture. In order to separate the spectrum of AZV 214 from the
nearby star, we used the MGEX and NEWCALIB routines provided in the IUEIDL package to
extract and calibrate the spectrum for AZV 214 and the nearby star. Neubig & Bruhweiler (1997)

-- 9 --
Fig. 1.--- The spectrum of reddened star AZV 18 and its comparison star AZV 462 are plotted in
(a). The extinction curve derived for AZV 18 is displayed in (b) without any binning (lower) and
with bins of 0.05 ¯m \Gamma1 and shifted by \DeltaE(–) = 3 (upper). Plot (b) also shows the FM fit for this
extinction curve.

-- 10 --
determined the UV spectral type using the combined spectrum of the two stars. This resulted
in a UV spectral type too late for AZV 214 since the nearby star had the effect of decreasing
the intensity of the spectral lines. The optical and infrared photometry of AZV 214 are likely
unaffected by the nearby star as the flux from this star, relative to AZV 214, is already small at
3000 š A and decreasing to the red. The nearby star is bluer than AZV 214 and is probably an
unreddened O type star.
The comparison star which best satisfies the Fitzpatrick criteria is AZV 380. Figure 2a
displays the spectra of AZV 214 and AZV 380. The extinction curve for AZV 214 is shown in
Figure 2b. The FM fit parameters are tabulated in Table 5.
3.1.3. AZV 398
The extinction curve for AZV 398 has been determined previously (Pr`evot et al. 1984;
Rodrigues et al. 1997). The comparison star which best satisfies the Fitzpatrick criteria is
AZV 289. This star was one of the eight comparison stars used in Pr'evot et al. (1984) and one
of the five stars used in Rodrigues et al. (1997). Again, by using only the comparison star with
the best spectral match we were able to determine a more accurate extinction curve. Figure 3a
displays the spectra of AZV 398 and AZV 289. The extinction curve for AZV 398 is shown in
Figure 3b. The FM fit parameters are tabulated in Table 5.
3.1.4. AZV 456
The star AZV 456 is the only star in the SMC to show the signature of the 2175 š A extinction
bump in its spectrum (Figure 4a). We have chosen a different comparison star than previous
authors (Lequeux et al. 1982; Rodrigues et al. 1997) specifically to remove mismatches in the C IV
and Si IV lines which are sensitive to both luminosity and temperature (Neubig & Bruhweiler
1997). Figure 4a displays the spectra of AZV 456 and AZV 70. The extinction curve for AZV 456
is shown in Figure 4b. The FM fit parameters are tabulated in Table 5.
4. Discussion
Any discussion of the behavior of the dust in the SMC based on only four extinction curves
is obviously severely hampered by the small sample size. Yet, interesting trends can be seen even
in this small sample. The four extinction curves are plotted together in Figure 5 and their FM
fit parameters are tabulated in Table 5. The extinction curves for AZV 18, 214, & 398 are very
similar and are all roughly linear with – \Gamma1 . The extinction curve for AZV 456 is much different as
it has a significant 2175 š A bump and a weaker far­UV extinction. The similarity of the extinction

-- 11 --
Fig. 2.--- The spectrum of reddened star AZV 214 and its comparison star AZV 380 are plotted in
(a). The extinction curve derived for AZV 214 is displayed in (b) without any binning (lower) and
with bins of 0.05 ¯m \Gamma1 and shifted by \DeltaE(–) = 3 (upper). Plot (b) also shows the FM fit for this
extinction curve.

-- 12 --
Fig. 3.--- The spectrum of reddened star AZV 398 and its comparison star AZV 289 are plotted in
(a). The extinction curve derived for AZV 398 is displayed in (b) without any binning (lower) and
with bins of 0.05 ¯m \Gamma1 and shifted by \DeltaE(–) = 3 (upper). Plot (b) also shows the FM fit for this
extinction curve.

-- 13 --
Fig. 4.--- The spectrum of reddened star AZV 456 and its comparison star AZV 70 are plotted in
(a). The extinction curve derived for AZV 456 is displayed in (b) without any binning (lower) and
with bins of 0.05 ¯m \Gamma1 and shifted by \DeltaE(–) = 3 (upper). Plot (b) also shows the FM fit for this
extinction curve.

-- 14 --
Fig. 5.--- The four extinction curves for AZV 18, 214, 398, & 456 are plotted.
curves for AZV 18 & 214 (lightly reddened) to that of AZV 398 (more reddened) gives confidence
that the extinction curves for AZV 18 & 214 are real.
Since the extinction curve for AZV 456 looks more like Milky Way dust than SMC dust, one
wonders if a large fraction of this extinction is due to foreground Galactic dust. Velocity resolved
H I measurements toward AZV 456 show that 90% of the H I along the line­of­sight is located
in the SMC (Lequeux et al. 1982; McGee & Newton 1981, 1982). Also, the star AZV 454, which
is 1: 0 5 from AZV 456, shows very little evidence of reddening. These observations strongly imply
that most of the interstellar medium along the line of sight toward AZV 456 lies in the SMC.
In the Milky Way, most of the differences between extinction curves can be explained by the
variation in the single parameter R V = A(V )=E(B \Gamma V ). This parameter measures the average
dust grain size with R V increasing with increasing average grain size (CCM). Following Bouchet

-- 15 --
Table 4. IUE Data
AZV IUE Spectra
18 SWP 8295/10321/18014/18015
LWR 7241/14207
70 SWP 16621/18830--LWP 12387
214 SWP 22372--LWR 17263
289 SWP 16049--LWR 12345
380 SWP 10319--LWR17265
398 SWP 18911/22361--LWR 14963
456 SWP 16051/45199
LWR 12347--LWP 23556
462 SWP 10316--LWR 17259
Table 5. FM fit parameters
AZV c1 c2 c3 xo fl c4
18 \Gamma5:68 \Sigma 0:28 2:53 \Sigma 0:05 0:76 \Sigma 1:38 4:56 \Sigma 0:23 1:68 \Sigma 1:52 0:60 \Sigma 0:26
214 \Gamma3:72 \Sigma 0:29 2:03 \Sigma 0:05 0:09 \Sigma 3:03 4:98 \Sigma 1:97 5:12 \Sigma 19:24 \Gamma0:20 \Sigma 0:26
398 \Gamma5:16 \Sigma 0:29 2:27 \Sigma 0:05 0:14 \Sigma 11:29 4:41 \Sigma 17:06 \Gamma6:48 \Sigma 211:22 0:80 \Sigma 0:28
456 \Gamma0:96 \Sigma 0:09 1:18 \Sigma 0:02 2:57 \Sigma 0:22 4:71 \Sigma 0:01 1:00 \Sigma 0:05 0:10 \Sigma 0:15

-- 16 --
et al. (1985), the R V values for the four SMC extinction curves were calculated from
R V = 1:10 \Delta(V \Gamma K)
\Delta(B \Gamma V ) : (7)
The intrinsic colors of the reddened stars were assumed to be either the colors of their respective
comparison stars or the intrinsic colors of Galactic stars of the same spectral types as tabulated in
Johnson (1966) for (B \Gamma V ), and Koornneef (1983), for (V \Gamma K). The values of R V calculated both
ways are tabulated in Table 6. The two R V values are equivalent within the uncertainties and
the adopted values are contained in the last column of Table 6. The four extinction curves have
roughly similar values of R V . If the SMC dust followed a CCM­like relationship then AZV 456
would have the largest R V . This does not seem to be the case.
In order to investigate the dependence of the extinction in the SMC on environment, we
plotted the positions of the four stars on an Hff image of the SMC (Figure 6). The Hff intensities
trace star formation activity. The lines­of­sight toward all four stars are associated with the known
H II regions (Davies, Elliott, & Meaburn 1976; Caplan et al. 1996). The three stars with roughly
linear extinction curves (AZV 18, 214, & 398) are located in regions of high Hff intensities (SMC
Bar). The one star (AZV 456) with a more Galactic type extinction curve is also located in an
H II region but one with much weaker star formation (SMC Wing). The H II region associated
with the line­of­sight to AZV 214 is noteworthy because it is associated with the cluster NGC 346
which is the most massive star formation region in the SMC. NGC 346 contains 33 known O type
stars (Massey, Parker, & Garmany 1989). The Hff flux from this cluster is 10% of that seen from
30 Dor in the LMC (Massey, Parker, & Garmany 1989). The Hff fluxes from the H II regions along
the lines­of­sight toward AZV 18, 398, & 456 are are 1, 1, and 0.1%, respectively, of that seen for
30 Dor (Kennicutt & Hodge 1986; Caplan et al. 1996). The dust along the AZV 456 sightline has
likely been exposed to a less harsh environment than the other three sightlines.
It is known that processing of Galactic dust near regions of active star formation results in
changes in the UV extinction curve (Mathis & Cardelli 1992). A similar behavior is seen in the 30
Dor region in the LMC (Fitzpatrick 1986). It is clear from this work that the harsh radiation and
Table 6. R V Values
star comparison Galactic adopted
AZV 18 3:60 \Sigma 0:73 2:78 \Sigma 0:34 3:60 \Sigma 0:73
AZV 214 2:75 \Sigma 0:55 2:36 \Sigma 0:21 2:75 \Sigma 0:55
AZV 398 2:87 \Sigma 0:40 3:05 \Sigma 0:17 2:87 \Sigma 0:40
AZV 456 \Delta \Delta \Delta 2:66 \Sigma 0:16 2:66 \Sigma 0:16

-- 17 --
Fig. 6.--- The positions of the four reddened stars are plotted on a Hff image of the SMC (Bothun
1997). This image is displayed in the North Celestial Pole projection (Staveley­Smith et al. 1997).
The image was provided by G. Bothun and L. Staveley­Smith.

-- 18 --
shock environment associated with star formation regions is modifying the dust in the SMC. In all
three galaxies, star formation modifies nearby dust by altering the 2175 š A bump and increasing
the strength of the far­UV extinction. Yet, the extent of the modification does not seem to be well
correlated with the level of star formation activity. Figure 7 plots the average extinction curves
for the LMC (30 Dor and rest of LMC, [Fitzpatrick 1986]), the SMC (Bar and Wing), and the
Milky Way (R V = 3:1). The extinction curve for ` 1 Ori D, a star in the Orion H II region, is also
plotted. The average extinction curve for the SMC Bar is similar to the previous SMC average
extinction curve (Pr`evot et al. 1984). From this figure, it is obvious that the extinction in the
SMC Bar is unlike that which has been found anywhere except starburst galaxies (Gordon et al.
1997). On the other hand, the extinction in the SMC Wing is similar to that found in the LMC
(excluding the 30 Dor region) and the Milky Way open cluster Trumpler 37 (not shown, Clayton &
Fitzpatrick 1987; FM). The large difference between the extinction curve in the Orion H II region
and the SMC Bar implies that an Orion­like H II region is not enough to produce starburst­like
dust. Something more is needed.
This raises the question: Why are the extinction properties of the dust in the SMC Bar so
much more extreme than that found in the 30 Dor region of the LMC? The largest star formation
region in the SMC (NGC 346) has only 10% the activity of 30 Dor as measured by Hff (Caplan
et al. 1996). The extinction curve for AZV 456 shows that dust, very similar to Galactic and
LMC dust, exists in the SMC. Therefore, there must be some significant environmental difference
between the LMC 30 Dor and SMC Bar regions which affects the extent to which star formation
activity can modify dust. One known difference between the LMC and SMC is metallicity. The
metallicities of the LMC and SMC are 0.2 and 0.6 dex lower, respectively, than the that of the
local Galactic ISM (the ISM is 0.1 dex lower than solar). However, the relative abundances of the
elements in the LMC and SMC are similar to that found in the local interstellar medium (Russell
& Dopita 1992). Metallicity is correlated with the dust­to­gas ratio in galaxies (Issa, MacLaren,
& Wolfendale 1990). So, the amount of dust in the SMC is significantly lower than that found
in the LMC. This could affect the ability of dust grains in the SMC to shield themselves from
the radiation and shocks present near star formation regions. This contrasts with the finding
that starburst galaxies all possess dust with an extinction curve like that found in the SMC Bar
even though they have metallicities between 0.1 and 2.0 solar (Gordon et al. 1997). It is possible
that the much higher level of star formation present in starburst galaxies (10\Theta that of 30 Dor)
overwhelms other environmental factors and always produces SMC Bar­like dust. The starburst
galaxies were UV selected biasing the sample toward intrinsically bright, nearby starbursts with
at least one lightly reddened starburst region. Thus, the combination of a small column of dust
and the more intense star formation possibly accounts for the presence of SMC­like dust in all
starburst galaxies studied in Gordon et al. (1997).

-- 19 --
Fig. 7.--- The extinction curves for the SMC, LMC, and Milky Way are plotted. The curves plotted
are those calculated from the FM fits except for the Milky Way which was calculated from the CCM
relationship for an R V = 3:1. The extinction curve for the SMC Bar is the \Delta(B \Gamma V ) weighted
average of the curves for AZV 18, 214, & 398. The SMC Wing and ` 1 Ori D extinction curves have
been multiplied by 0.83 and 1.3, respectively, to allow easier comparison to the other four curves.

-- 20 --
5. Summary
ffl We have greatly improved the UV extinction curves for the SMC through improvements in
the S/N of the IUE spectra and careful choices of reddened and comparison star pairs.
ffl Four reddened SMC stars possess the Fitzpatrick criteria needed for accurate extinction
calculations. Three of the four stars possess a roughly linear (with – \Gamma1 ) extinction curve. The
lines­of­sight toward these three stars (AZV 18, 214, & 398) pass through active star formation
regions. The fourth star (AZV 456) has an extinction curve with a 2175 š A bump and a weaker
far­UV rise. Its sightline also passes through a star formation region, but one which is much less
active.
ffl Processing of dust near regions of star formation results in variations in UV extinction.
However, there is no simple correlation between the strength of the variations and the amount of
star formation activity. Other parameters such as galaxy metallicity must also play an important
role.
ffl As this work is based on only four sightlines, more observations of reddened stars in the
SMC are needed for to confirm these results. Of special need are observations outside the SMC
Bar to confirm the extinction curve of AZV 456.
We thank G. Bothun and L. Staveley­Smith for giving us the Hff image. This work was
supported by NASA grant NAG5­3531.
REFERENCES
Ardeberg, A. & Maurice, E. 1977, A&AS, 30, 261
Azzopardi, M. & Vigneau, J. 1979, A&AS, 35, 353
Azzopardi, M. & Vigneau, J. 1982, A&AS, 50, 291
Azzopardi, M., Vigneau, J., & Macquet, M. 1975, A&AS, 22, 285
Bianchi, L., Clayton, G. C., Bohlin, R. C., Hutchings, J. B., & Massey, P. 1996, ApJ, 471, 203
Bothun, G. 1997, private communication
Bouchet, P., Lequeux, J., Maurice, E., Pr'evot, L., & Pr'evot­Burnichon, M. L. 1985, A&A, 149,
330
Bouchet, P. 1997, private communication
Bromage, G. E. & Nandy, K. 1983, MNRAS, 204, 29P

-- 21 --
Calzetti, D., Kinney, A. L., & Storchi­Bergmann, T. 1994, ApJ, 429, 582
Caplan, J., Ye, T., Deharveng, L., Turtle, A. J., & Kennicutt, R. C. 1996, A&A, 307, 403
Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 (CCM)
Cardelli, J. A., Sembach, K. R., & Mathis, J. S. 1992, AJ, 104, 1916
Clayton, G. C. & Fitzpatrick, E. L. 1987, AJ, 92, 157
Clayton, G. C., et al. 1996, ApJ, 460, 313
Clayton, G. C. & Martin, P. G. 1985, ApJ, 288, 558
Davies, R. D., Elliott, K. H., & Meaburn, J. 1976, MmRAS, 81, 89
Franx, M., Illingworth, G. D., Kelson, D. D., van Dokkum, P. G., & Tran, K.­V. 1997, ApJ, 486,
L75
Fitzpatrick, E. L. 1985, ApJ, 299, 219
Fitzpatrick, E. L. 1986, AJ, 92, 1068
Fitzpatrick, E. L. & Massa, D. 1990, ApJS, 72, 163 (FM)
Garhart, M. P. & Nichols, J. S. 1994, in IUE Newsletter No. 55, 1
Garmany, C. D., Conti, P. S., & Massey, P. 1987, AJ, 93, 1070
Gordon, K. D., Calzetti, D., & Witt, A. N. 1997, ApJ, in press (1 Oct)
Humphreys, R. 1983, ApJ, 265, 176
Issa, M. R., MacLaren, I., & Wolfendale, A. W. 1990, A&A, 236, 237
Johnson, H. L. 1966, ARA&A, 4, 193
Kennicutt, R. C., Jr. & Hodge, P. W. 1986, 306, 130
Kinney, A. L., Bohlin, R. C., Calzetti, D., Panagia, N., & Wyse, R. F. G. 1993, ApJS, 86, 5
Koornneef, J. 1983, A&A, 128, 84
Lennon, D. J. 1997, A&A, 317, 871
Lequeux, J., Maurice, E., Pr'evot­Burnichon, M.­L., Pr'evot, L., Rocca­Volmerange, B. 1982, A&A,
113, L15
Lequeux, J., Maurice, E., Pr'evot, L., Pr'evot­Burnichon, M.­L., Rocca­Volmerange, B. 1984, in
Structure and Evolution of the Magellanic Clouds, IAU Symp. 108, eds. S. van den Bergh
& K. S. de Boer (Dordrecht: Reidel) 405

-- 22 --
Lowenthal, J. D., et al. 1997, ApJ, 481, 673
Madau, P., Pozzetti, L., & Dickinson, M. 1997, ApJ, in press
Massa, D., Savage, B. D., & Fitzpatrick, E. L. 1983, ApJ, 266, 662
Massey, P., Lang, C. C., DeGioia­Eastwood, K., & Garmany, C. D. 1995, ApJ, 438, 188
Massey, P., Parker, J. W., & Garmany, C. D. 1989, AJ, 98, 1305
Mathis, J. S. & Cardelli, J. A. 1992, ApJ, 398, 610
McGee, R. X. & Newton, L. M. 1981, Proc. Astro. Soc. of Australia, 4, 189
McGee, R. X. & Newton, L. M. 1982, Proc. Astro. Soc. of Australia, 4, 308
Nandy, K., McLachlan, A., Thompson, G. I., Morgan, D. H., Willis, A. J., Wilson, R.,
Gondhalekar, P. M., & Houziaux, L. 1982, MNRAS, 201, 1P
Neubig, M. M. S. & Bruhweiler, F. C. 1997, AJ, in press
Nichols, J. S., Garhart, M. P., De La Pe~na, M. D., & Levay, K. L. 1994, New Spectral Image
Processing System Information Manual: Low Dispersion Data ­ Version 1.0, IUE Newsletter
No. 53
Pettini, M., Steidel, C. C., Adelberger, K. L., Kellogg, M., Dickinson, M., Giavalisco, M. 1997, in
ORIGINS, ASP Conference Series, eds. J. M. Shull, C. E. Woodward, & H. Thronson, in
press (astro­ph/9708117)
Pr`evot, M. L., Lequeux, J., Maurice, E., Pr`evot, L., Rocca­Volmerange, B. 1984, A&A, 132, 389
Rocca­Volmerange, B., Pr'evot, L., Ferlet, R., Lequeux, J., & Pr'evot­Burnichon, M. L. 1981, A&A,
99, L5
Rodrigues, C. V., Magalh~aes, A. M., Coyne, G. V., & Piirola, V. 1997, ApJ, 485, 618
Russell, S. C. & Dopita, M. A. 1992, ApJ, 384, 508
Sanduleak, N. 1968, AJ, 73, 246
Sanduleak, N. 1969, AJ, 74, 877
Staveley­Smith, L., Sault, R. J., Hatzidimitriou, D., Kesteven, M. J., & McConnell, D. 1997,
MNRAS, 289, 225
Steidel, C. C., Giavalisco, M. Pettini, M., Dickinson, M., & Adelberger, K. L. 1996, ApJ, L17
Thompson, G. I., Nandy, K., Morgan, D. H., & Houziaux, L. 1988, MNRAS, 230, 429

-- 23 --
Trager, S. C., Faber, S. M., Dressler, A., & Oemler, A., Jr. 1997, ApJ, 485, 92
Walborn, N. R. 1991, in Massive Stars in Starbursts, eds. C. Leitherer et al. (Cambridge:
Cambridge Univ. Press), 145
This preprint was prepared with the AAS L A T E X macros v4.0.