far-UV extinction curves have reduced depletions com- pared to those with steep curves. The Orion Nebula in front of h1 C Ori shows Ñat far-UV extinction. &.
THE ASTROPHYSICAL JOURNAL, 480 : 272È282, 1997 May 1 ( 1997. The American Astronomical Society. All rights reserved. Printed in U.S.A.
ABUNDANCES AND DEPLETIONS TOWARD THE ORION NEBULA R. YOUNG SHUPING AND THEODORE P. SNOW Center for Astrophysics and Space Astronomy, University of Colorado, Boulder, CO 80309-0389 Received 1995 December 21 ; accepted 1996 November 25
ABSTRACT This paper presents an absorption-line study of h1 C Orionis (HD 37022), based on IUE data. The line of sight, as determined by OÏDell et al. (1993), is thought to consist of two major cloud complexes : the ““ neutral lid ÏÏ and some foreground material. A key challenge was to improve the S/N for IUE spectra, and this is discussed in some detail. Column densities for several species are derived in two ways. First, we Ðt equivalent widths of the interstellar lines on two semiempirical curves of growth (COGs). These COGs are derived from high-resolution ground-based data that resolve the cloud component structure on the line of sight (LOS), and it is assumed that the atomic species follow this structure. Second, we Ðt each species independently to its own COG, yielding approximate b-values and column densities. Both methods yielded similar results. The Lya line yields a column density for hydrogen of log N(H I) \ 21.60 ^ 0.1 cm~2. Abundances and depletions are calculated for the clouds and are compared to emission-line studies of the Orion nebula itself. The discrepancies between the two regions imply grain core destruction in the H II region. Also, the depletions are compared to other sight lines with di†erent far-ultraviolet extinction behavior. It is found that depletions are not strongly dependent on far-UV extinction except for Mg and possibly Si and P. These elements seem slightly less depleted for sight lines with Ñat far-UV extinction. Ionization balance gives an electron density of n D 5 ] 10~3 e of n [ 9 cm~3 for the foreground clouds, which in turn yields a lower limit to the hydrogen density H cm~3. Subject headings : ISM : abundances È ISM : individual (Orion Nebula) È ultraviolet : ISM 1.
INTRODUCTION
From the abundances it will be possible to calculate the depletions from ““ cosmic ÏÏ (assumed to be solar) region for each species. These depletions are important to theories regarding interstellar dust (see Snow 1992 ; Mathis 1990 ; Whittet 1992). The Orion region is a unique environment in which to study dust : recent star formation, intense stellar winds, and a higher than normal UV Ñux all contribute to the dust chemistry (see Sorrell 1992). It is well known that UV extinction curves that are due to dust vary from cloud to cloud in the ISM (see, e.g., Bless & Savage 1972 ; Fitzpatrick & Massa 1986, 1988, 1990). The parameter R (ratio of selective to total extinction) seems V uniquely the shape of the extinction curve, to characterize including the bump at 2175 Ó and the steepness of the far-UV portion (Cardelli, Clayton, & Mathis 1988, 1989). The general shape of the extinction curve indicates the size distribution of grains : Flat far-UV extinction (R [ 3) results from a lack of smaller grains. It is thoughtV that atomic species that are depleted from the gas phase of these clouds reside in the dust. Recent work has shown that gasphase depletions are enhanced for clouds with steep far-UV extinction (Joseph et al. 1986) ; speciÐcally, there may be an anticorrelation between R and depletions. A complemenV tary study for a LOS with abnormally low UV extinction (p Sco) was less conclusive (Allen, Snow, & Jenkins 1990), so it is still somewhat unclear as to whether sight lines with Ñat far-UV extinction curves have reduced depletions compared to those with steep curves. The Orion Nebula in front of h1 C Ori shows Ñat far-UV extinction (Fitzpatrick & Massa 1990), and it is our intention to use the depletions from this abundance study to test whether high-R sight V lines are less depleted than their low-R counterparts. V Forthcoming work will continue to test sight lines with steep far-UV extinction. Recent studies also suggest that as di†use clouds collapse,
The Orion Nebula (M42, NGC 1976), excited by the star h1 C Ori, is perhaps the most studied H II region in the sky (for reviews see Genzel & Stutzki 1989 and OÏDell 1994). While there are many emission-line studies of the glowing gas, there are relatively few addressing absorption features, the most notable exceptions being OÏDell et al. (1993) and Hobbs (1978). These studies focused primarily on the velocity structure of the interstellar gas on the line of sight (LOS) to h1 C Ori using lines of Na I, Ca II, and He I. Franco & Savage (1982) also detected many UV absorption features toward h1 C Ori but only analyzed the higher ionization species (e.g., Si IV and C IV). In this paper we present an absorption-line study in the ultraviolet (UV) using International Ultraviolet Explorer (IUE) data (both ours and archival). IUE does not have the spectral resolution required to measure thermal velocity widths of lines from cold interstellar material directly. Equivalent widths, however, are independent of resolution and can be used in an accurate curve-of-growth study to derive column densities for di†erent species, so long as blends can be taken into account properly. Of primary importance in this study will be the gas-phase atomic abundances. Narrow absorption lines that are due to atoms and Ðrst ions form in the cooler portions of the interstellar medium (ISM) on the LOS to h1 C Ori. The abundances found there will be compared to abundances determined by emission-line studies of the glowing nebula itself. The hydrogen Lya transition at 1215.7 A will also be measured and compared to previous results. Because molecular hydrogen is known to be negligible in this sight line (Savage et al. 1977), the measurement of neutral hydrogen will allow the determination of absolute abundances relative to hydrogen rather than abundances relative to some other standard (S or Zn, for example). 272
ABSORPTION-LINE STUDY OF h1 C ORI
273
dust grains preferentially coagulate, eliminating the small grain population and producing high-R sight lines (Jura V molecular cloud 1980 ; Whittet 1992, p. 222). It is then in the stage that the coagulated grains begin to form icy mantles (Whittet 1992, p. 224). The Orion region should display this sort of evolution and is a good testing ground for these theories as well. Section 2 will describe the LOS toward h1 C Ori, including the extinction curve. In ° 3 we will cover the IUE data and reductions performed, focusing on the ““ Ðxed pattern noise ÏÏ problem. Section 4 will detail the various analyses used to obtain the gas-phase atomic abundances and depletions. The Ðnal section will brieÑy address dust in Orion and in the general ISM. 2.
THE LINE OF SIGHT
h1 C Ori (HD 37022, V \ 5.13) is the O6(p) star responsible for exciting the Orion nebula, roughly 460 pc from the Sun. Intense photoionization and evaporation is carving a hole into the neighboring Orion A molecular cloud, producing the spectacular Ñower-shaped H II region we see from Earth in the visible range. The LOS to h1 C Ori is moderately reddened, with E(B[V ) \ 0.34 and R \ V 5.50 (Cardelli, Clayton, & Mathis 1989). The observed extinction curve can be found in Bless & Savage (1972), Bohlin & Savage (1981), and more recently in the IUE extinction atlas presented by Fitzpatrick & Massa (1990). The curve has a broad 2175 A bump and abnormally Ñat far-UV behavior, which implies an absence of smaller dust grains. The velocity structure on the LOS to h1 C Ori has been thoroughly studied by Hobbs (1978) and extended by OÏDell et al. (1993), using Na I, Ca II, and He I (all in absorption). Foreground cloud structure has also been studied using 21 cm H I absorption by van der Werf & Goss (1989). Of primary importance is the ““ neutral lid,ÏÏ an H I cloud complex just overlying the H II region. The Lid is composed of three distinct velocity components (V \ 15, _ I, and 20, and 23 km s~1, OÏDell et al. 1993) seen in H I, Na Ca II absorption. The cloud complex is warm (T \ 75 K) on the side facing the Orion nebula and cooler (T \ 25 K) toward the side facing away (van der Werf & Goss 1989). There is also evidence (from Na I and Ca II absorption) for three foreground clouds which lie in front of i Ori and 42 Ori with V \ 7, 31, and 39 km s~1 (OÏDell et al. 1993). Figure 1 is _a cartoon summary of the LOS, as detailed in OÏDell et al. (1993). Because of the high electron and radiation temperatures (T \ 8500 K and T D 37,000 K ; Genzel & Stutzki 1989), e H II region very radnear h1 C Ori is highly ionized and, the consequently, devoid of almost all neutral and most Ðrst ionized species (see the models of Baldwin et al. 1991 and Osterbrock, Tran, & Veilleux 1992, and other emission-line studies). Because of the proximity of the Orion region to the Sun, we would not expect much absorption from intervening interstellar material. Also, there is some strong evidence that many local supernovae have cleared out the region in front of Orion, producing a hot, thin environment (Cowie, Songalia, & York 1979). This further reduces its impact on our absorption study. In summary, the work of OÏDell et al. (1993) strongly suggests that most of the absorption lines that are due to cold IS material should arise in the neutral lid cloud complex and the three foreground clouds. Note that IUE does not have the resolution
FIG. 1.ÈCloud components in the LOS to h1 C Ori, based on OÏDell et al. (1993).
to distinguish between these components, and any abundances determined will represent an average through the clouds. 3.
DATA AND REDUCTIONS
The wide UV spectral coverage and relatively high dispersion of IUE make it well suited for a multi-ion absorption-line study. All the data for this project came from the IUE archives. A list of all the images used and any pertinent information are given in Table 1. Small-aperture (3ÏÏ aperture) images were used to reduce nebular emission from around h1 C Ori, which could confuse some of the absorption features. Of course, only high-dispersion spectra are of interest. The observations from 1993 are part of a program conducted by T. P. Snow and M. M. Hanson. One way to improve the limited signal-to-noise ratio (S/N, usually 10È20) of IUE is to co-add individual spectra. This can be dangerous, however, because ““ Ðxed pattern noise ÏÏ can begin to dominate the co-added spectrum even after only a few images are summed. Luckily, there are a few ways around this problem. We used a cross-correlation technique along with new software developed by T. Ayres at the University of Colorado. The new software helps to suppress Ðxed pattern noise by applying an accurate ““ ÑatÐeld ÏÏ correction before extraction. The result is signiÐcantly improved S/N (compared to IUESIPS extractions), reduced Ðxed pattern noise, and a photometric error for each extracted Ñux value (Wood & Ayres 1995). TABLE 1 IUE IMAGES Camera and Image Number
Observation Year
Observation Day
Exposure Time (s)
SWP 09991 . . . . . . SWP 48991 . . . . . . SWP 48992 . . . . . . LWR 08698 . . . . . . LWR 08714 . . . . . . LWP 26605 . . . . . . LWP 26606 . . . . . .
1980 1993 1993 1980 1980 1993 1993
246 296 296 246 248 296 296
200 300 180 600 270 240 360
274
SHUPING & SNOW
FIG. 2a
Vol. 480
Figure 2. The error bars in Figure 2a are from the Ayres extraction routines : the IUESIPS extractions for the LW region have no corresponding error bars. Note also the complexity of the continuum in the SWP sample (Fig. 2a) which is typical of O stars. One of the goals of the study was to measure the hydrogen Lya absorption feature at 1215.7 A . The section of data including this mammoth line was reduced independently using only SWP 48991 and SWP 48992. These data were reduced identically to the LW spectra. The SWP 09991 images were not included because of some discrepancies in the Ñux levels : it appears as if h1 C Ori was not entirely in the slit for this observation. The omission of the 09991 image does not signiÐcantly a†ect the S/N at Lya. 4.
FIG. 2b FIG. 2.ÈSample portions of IUE co-added spectrum. Note the echelle break in (a).
Fixed pattern noise can also be detected (as a sharp spike) in a cross-correlation between two spectra intended to be co-added. No pixel pattern noise showed up in any of the cross-correlations, probably because of the short exposure times (3 to 5 minutes). The cross-correlation also provides a measure of the shift in pixel space between like features in two di†erent spectra. Hence, spectra can be optimally registered in pixel space, thereby minimizing artiÐcial broadening of lines in co-addition. Hanson, Snow, & Black (1992) and Welty et al. (1992) provide useful discussions on the Ðxed pattern noise problem and strategies in overcoming it. Once obtained from the archives, the data were reduced two di†erent ways. All the LWP and LWR images were extracted using the IUESIPS routines available through GSFC. At Colorado, the extracted spectra were resampled, cross-correlated, and shifted accordingly in pixel space. Once shifted and checked, the spectra were co-added with weights according to the square root of exposure time (which is proportional to the S/N of each spectrum). The Ðnal co-added LW spectrum covers 2100È3000 A , with S/N D 22. The SWP spectra were extracted using the new routines written by T. Ayres. Again, the extracted spectra were crosscorrelated, shifted, and co-added by the square root of the exposure time. The Ðnal SWP spectrum covers 1238È1538 A , with S/N D 25 (note that S/N D 16 for IUESIPS extractions of the same SWP data). Note that there are two regions of the SWP cameras not covered by the Ayres routines. The sections covering 1190È 1240 A and 2030È2096 A were reduced identically to the LW portions. These portions contain some important absorption features and could not be abandoned, even at reduced S/N. Sample portions of the co-added spectrum are shown in
ANALYSIS AND RESULTS
4.1. L ine Measurements After reduction, the majority of interstellar features were identiÐed and measured with the IDL program MSLAP (written by C. L. Joseph). Atomic transition data come from Morton (1991), except for Si II, which has been updated in Snow et al. (1996). Table 2 lists all interstellar lines identiÐed and measured. The 1 p errors are, in general, calculated from noise in the continuum surrounding the line. Unusually large errors reÑect difficulty in continuum placement. Most weak lines are listed as upper limits as a result of a noisy or confusing continuum. Some strong lines were discarded because of poor data quality or contamination by reseau marks. Some resolved but blended (wing overlap) lines were deblended and measured using IRAF or by Ðtting Voigt proÐles using a program called FITS, written by Vidal-Madjar et al. (1977). Some lines, however, were obviously blended but not resolved ; these were ignored because of ambiguities in Ðtting. The only exception is the C I system at 1560 A : these unresolved features were Ðtted using FITS and assumed to be weak. Because of the ambiguities in velocity shift for each component, the corresponding equivalent widths should be regarded as rough estimates. Only measurements of the total equivalent width are secure since the resolution of IUE is too poor to determine individual components. 4.2. High-Resolution Data A curve of growth (COG) allows one to derive column densities from the measured equivalent widths. A semiempirical COG can be calculated from high-resolution data on the same LOS (see, e.g., Morton 1975). First, the highresolution absorption data must be proÐle-Ðtted to determine the column density (N), b-value, and velocity of each absorption component. This information is then used to reconstruct the multicomponent line at di†erent total column densities (preserving the relative N values between each component). The reconstruction is then integrated for an equivalent width, producing a curve of growth of total column density versus equivalent width of the entire line. This semiempirical COG is then assumed to model the component structure along the entire LOS. Even though the IUE data are not resolved into the many components, they can still be Ðtted on this COG since equivalent width is independent of instrument resolution. From this, it is possible to derive column densities for each species if we assume that the species in question follow a similar component structure as the species that was used to generate the COG (which may not necessarily be the case).
ABSORPTION-LINE STUDY OF h1 C ORI
No. 1, 1997
275
TABLE 2 LINE DATA AND COLUMN DENSITIES
Species C I......... C I* . . . . . . . C I** . . . . . . C II . . . . . . . . N I ........ O I ........ O I* . . . . . . . O I** . . . . . . Mg I . . . . . . Mg II . . . . . . Si II . . . . . . . . Si II* . . . . . .
P II . . . . . . . . S I ......... S II . . . . . . . . Cr II . . . . . . . Mn II . . . . . . Fe II . . . . . . .
Ni II . . . . . . . Zn II . . . . . .
j rest (Ó) 1560.31 1560.68 1560.71 1561.34 1561.44 1334.53 1200.71 1200.22 1355.60 1302.17 1304.86 1306.03 2852.96 2026.48 2803.53 2796.35 1239.93 1808.01 1526.71 1304.37 1533.43 1309.28 1265.00 1264.74 1532.53 1301.87 1807.31 1253.81 2066.16 2062.23 2056.25 2606.46 2594.50 2576.88 2600.17 2586.65 2382.76 2374.46 2344.21 2260.78 2249.88 1608.45 1741.55 1317.22 2062.66 2026.14
f
Equivalent Width (Ó)
Error (Ó)
log N (cm~2)
Error
0.0805 0.0603 0.0201 0.012 0.0675 0.1278 0.0442 0.0885 1.246e-6 0.0489 0.0488 0.0488 1.8311 0.1154 0.3057 0.6129 0.0003 0.00208 0.1304 0.0884 0.1298 0.0881 0.1158 1.046 0.0076 0.0172 0.1107 0.0109 0.0698 0.1049 0.1403 0.1927 0.2710 0.3507 0.2239 0.0646 0.3006 0.0282 0.1096 0.0037 0.0025 0.0619 0.1035 0.1460 0.2527 0.5144
0.090 0.067 \0.024 \0.015 0.078 0.5847 0.1587 0.1634 \0.0324 0.3048 0.0920 0.1057 0.2721 0.0250 0.5435 0.6121 \0.042 0.1157 0.2918 0.2067 0.1544 0.0999 0.0937 0.2152 \0.0261 0.0307 \0.0067 0.2218 \0.0404 0.067 0.0420 0.1381 0.1674 0.1534 0.3887 0.3233 0.3988 0.2227 0.2909 0.0670 0.0759 0.1545 0.0433 \0.032 0.099 0.123
... ... ... ... ... 0.0177 0.0350 0.0295 ... 0.0180 0.0124 0.0205 0.0191 0.0040 0.0198 0.0200 ... 0.0081 0.0058 0.0102 0.0082 0.0071 0.0120 0.0180 ... 0.010 ... 0.0271 ... 0.01 0.0068 0.0203 0.0470 0.0181 0.0280 0.0264 0.0196 0.0299 0.0227 0.0355 0.0200 0.0110 0.0144 ... 0.02 0.01
D14 13.9
... ...
T T
13.8
...
T
\18.35 15.2
... ]1.5, [0.5
\18.5 15.5
T S
17.9
]0.2, [0.1
17.9
T
14.4 14.5 12.8
^0.2 ^0.2 ^0.1
14.5 14.5 12.8
S S
16.2
]0.1, [0.2
16.1
T
15.6
^0.1
15.5
14.2
^0.1
14.0
14.1
^0.1
14.2
\12.3 17.5 D13.2
... ^1.0 ...
... 17.5 ...
13.2
^0.1
13.2
14.7
]0.2, [0.4
14.8
13.2
^0.1
13.2
T
...
13.1
T
[ 13.1
log N (cm~2) (indep)
Notesa
T S, T T
a S : all lines saturated. T : see ° 4.3 for explanation.
We Ðtted the Na I D2 line (5889.951 A ) from Hobbs (1978) (Yerkes Observatory, *v \ 0.9 km s~1) and the Ca II K line (3933.663 A ) from OÏDell et al. (1993) (KPNO, *v \ 3.3 km s~1) using the FITS program written by Vidal-Madjar et al. (1977). The results of the proÐle Ðtting are shown in Table 3 and Figures 3 and 4 (hyperÐne structure does not a†ect the COGs signiÐcantly and was not included). We used the Gaussian Ðts of OÏDell et al. (1993) as initial values, but the Ðts presented here are otherwise independent of theirs. The Ðnal results are very similar, except that our Voigt proÐle Ðts yield real column densities and b-values for each component. In addition, we could not justify the Ðfth component (V \ 13.9 km s~1) in the OÏDell et al. (1993) Ca II _ left it out. Ðt and have Semiempirical COGs were then constructed for both Na I and Ca II for both the highest and lowest possible values of N in each component. It was found that the resulting di†er-
FIG. 3.ÈProÐle Ðt to the Na I D2 line (5889.951 A ) from Hobbs (1978)
276
SHUPING & SNOW
Vol. 480
TABLE 3 PROFILE FITS Component
V _ (km s~1)
b (km s~1)
% of N(tot) (best Ðt)
N Range (cm~2)
N(tot) (cm~2)
(1.75È3.25) ] 1010 (4.25È5.30) ] 1011 (0.75È1.10) ] 1012 (0.75È1.10) ] 1012 (1.30È1.80) ] 1011 (1.50È2.80) ] 1010
2.41 ] 1012
(2.90È3.50) ] 1011 (4.00È4.80) ] 1011 (3.70È4.70) ] 1011 (3.50È5.50) ] 1010
1.24 ] 1012
Na I D2 1 ............ 2 ............ 3 ............ 4 ............ 5 ............ 6 ............
[15.72 6.77 18.32 22.26 31.69 38.48
1.17 5.52 2.10 3.31 2.19 1.105
0.95 19.3 37.3 35.3 6.1 0.84 Ca II K
1 ............ 2 ............ 3 ............ 4 ............
7.61 20.99 31.19 39.20
5.66 4.79 1.63 2.50
ences in COG shape were negligible in relation to the errors accompanying the equivalent width measurements. Therefore, the COGs were composed from the best-Ðt percentages for each component. 4.3. Curve-of-Growth Analysis The observed absorption-line data were Ðtted to the resulting semiempirical COGs to determine the corresponding total column density, N, through both cloud complexes (see, e.g., Spitzer 1975). There are, however, some serious problems with this technique, which will be addressed in ° 4.5. In general, the Ðts were quite good, resulting in errors of 0.1 in log N and suggesting that the proÐle Ðts to the Na I and Ca II lines are in fact accurate. Each species was Ðtted either to the Na I or Ca II COG based on its ionization structure. The COG Ðts are shown in Figures 5, 6, and 7 and the corresponding column densities are quoted in Table 2. Errors in N are based on scatter of the absorption-line data about the COG. It is important to note that some species (N I, for example) have only saturated lines. If these species do not exactly follow the component structure used to construct our semiempirical COGs, then the column density could be very di†erent. Species with only saturated lines are marked as such in Table 2. The values of N and the corresponding errors for these cases should be considered with care. Most other species, however, have at least one fairly weak line, which helps to pin down the column density regardless of the COG. Some species deserve special
FIG. 4.ÈProÐle Ðt to the Ca II K line (3933.663 A ) from OÏDell et al. (1993).
25.3 35.5 35.1 4.1
attention : C I : All the C I lines (including those from the Ðrst and second Ðne-structure levels) were measured from the system at 1560 A using FITS. The EWs Ðt the COG very well (Fig. 7), but since the Ðts are poorly constrained (making EWs hard to measure exactly and errors impossible to estimate), the column densities should be considered as estimates only. It seems clear, however, that C I is negligible compared to C II. C II : The equivalent width for the single C II line at 1335 A places it on the damping portion of the COG. Since the damping constants for Ca II and C II are slightly di†erent, the column density derived for C II will be slightly in error
FIG. 5.ÈCurve-of-growth analysis for the ions. The semiempirical Ca II COG (solid) was used to Ðt line data for the di†erent species shown. Many of the weaker lines are upper limits (see Table 2).
FIG. 6.ÈCurve-of-growth analysis for the excited ions. None of the points are upper limits.
No. 1, 1997
ABSORPTION-LINE STUDY OF h1 C ORI
277
damping proÐle. All techniques yielded a value of log N(H I) \ 21.6 ^ 0.1 cm~2 (integrated through both cloud complexes), which is substantially larger than older values (Bohlin 1975 ; Savage et al. 1977 ; Savage & Jenkins 1972), yet agrees with the more current IUE study by Fitzpatrick & Massa 1990 [log N(H I) \ 21.60 cm~2]. From Copernicus observations, Savage et al. (1977) found the upper limit for molecular column density of hydrogen to be log N(H ) \ 17.55 cm~2, which is negligible compared to 2 N(H I). Since small grains are very efficient at creating molecular hydrogen, this result further supports the idea that a population of small grains is missing from the dust. FIG. 7.ÈCurve-of-growth analysis for the neutral species. Upper limits are marked.
(see ° 4.5 for further discussion). More importantly, damping wings are not readily apparent in the proÐle (Fig. 2a). It is highly possible that the C II line is blended with the P III line at 1335 A . There appears to be a weak absorption feature at 1750 A that coincides with a weak transition of P III. Since the P III line at 1335 A would almost surely be saturated, it is nearly impossible to disentangle it from the C II feature. We therefore retain the full, probably blended, equivalent width for the 1335 A feature and quote the column density for C II as an upper limit. Mg II and O I : The column densities for both of these species are constrained by upper limits on weak lines. Since the stronger lines are saturated, it is possible for N to be somewhat smaller. S I : The S I line at 1807.311 A resides in a portion of the spectrum with high S/N and relatively simple continuum. Nonetheless, it is very weak and should be considered as an upper limit. S II : Only one of the three strong S II transitions was usable, the other two being corrupted by an IUE spectral order break and blending with other lines. The 1253.81 A line is saturated, and hence the column density and error should be viewed cautiously. Cr II : The problem with the Cr II lines is that the EWs and f-values do not make sense : One of the lines with a high f-value actually has a smaller equivalent width than the others. The line at 2066.161 Ó su†ers from poor continuum placement (although this should be reÑected in the error), and the 2062.234 A line was deblended from its neighboring Zn II feature with FITS. It is not clear to us which point or points are the problem, but we have estimated an N anyway because they all fall in roughly the same region of the COG. Ni II : The line at 1317.2 A is considered as an upper limit because of confusion in the continuum. Zn II : Both of these lines were deblended from neighboring features using FITS. A lower limit for N is quoted in Table 2 based on the weak line limit, as the lines do not seem to Ðt the Ca II COG very well (Fig. 5). 4.4. H I Column Density The H I column density was measured from the Lya line. It is possible for relative extinction to skew this line because of its spectral extent, so the extinction was calculated explicitly at Lya from Cardelli, Clayton, & Mathis (1989). The e†ect was found to be negligible. We measured the column density from Lya three separate ways : First, by continuum reconstruction (see, e.g., Bohlin 1975), then by Ðtting a Voigt proÐle, and Ðnally by Ðtting just the wings with a pure
4.5. Independent COG Analysis As stated before, there are some problems with the semiempirical COG method. In general, COG analyses of saturated lines are not unique : one can always come up with alternative explanations for a particular value of equivalent width (i.e, by adding multiple components which are not saturated). In addition, it is not clear a priori whether all species will follow the velocity structure of Ca II or Na I (or any other species). Furthermore, damping constants di†er from ion to ion. Any species with lines even close to damping will deÐnitely not Ðt the semiempirical COGs based on Ca II and Na I. To address some of these problems, we have done a more basic COG analysis independent of the semiempirical method. First, we compared each species to its own COG, Ðtting various b-values. Mg I, Fe II, and Si II (species with both weak and slightly saturated lines) all pointed to a b-value of roughly 10 km s~1 on the LOS. The species with saturated lines only were then Ðtted to their COG assuming a b-value of 10 km s~1 (which may be incorrect). The results are shown in Table 2. C I and S I were not reanalyzed because both species have only weak lines or upper limits. Note that the column densities for both methods are almost exactly the same, but this does not mean that one method or the other is ““ correct ;ÏÏ the nonuniqueness of COG analyses in general precludes this deduction. The similarity does, however, lend weight to the values derived and probably means that IUE data are too poor to di†erentiate between these methods. In conclusion, we will proceed with the column densities derived from the semiempirical method, with the caveat that there are some inherent problems with both the independent and semiempirical COG analyses. We feel it is the best we can do with the given data. Obviously, proÐle Ðtting of higher resolution and higher S/N data would alleviate many of these difficulties. 4.6. Ionization Stages Before we can state the total abundances for the neutral lid and foreground clouds, it is necessary to analyze the ionization balance along the LOS. The primary assumption here is that all photons more energetic than the Lyman limit TABLE 4 DETERMINATION OF n
e
Element
FIP (eV)
! (s~1)
a (90 K) r s~1) (cm3
Mg . . . . . .
7.646
7.9([11)
6.3([12)
NOTE.Èx(y) \ x ] 10y.
N
/N neu ion 3.98(-4)
n (90 K) e (cm~3) 5.0([3)
278
SHUPING & SNOW
Vol. 480
TABLE 5 TOTAL ABUNDANCES RELATIVE TO H IN ORION Element
h1 C Ori LOS (This Study)
Error
BFM91a
OTV92b
C ........ N........ O........ Mg . . . . . . Si . . . . . . . . P ........ S ........ Caf . . . . . . Cr . . . . . . . Mn . . . . . . Fe . . . . . . . Ni . . . . . . . Zn . . . . . .
\8.75 5.60 8.30 6.60 6.00 4.50 7.9 2.5 D3.60 3.60 5.10 3.60 [3.50
... ]1.5, [0.5 ]0.2, [0.1 ]0.1, [0.2 ^0.2 ^0.1 ^1.0 ^0.1 ... ^0.1 ]0.2, [0.4 ^0.1 ...
8.33 7.94 8.58 6.50
7.72 8.49
CL92c
WDH92d
RDW93e 8.45
8.65
8.78 7.73 8.90
6.65 7.12
6.97
6.62
5.97 4.75
NOTE.ÈAbundances are given as 12 ] log (N /N ). X H a Baldwin et al. 1991. b Osterbrock, Tran, & Veilleux 1992 with update by Lucy 1995 for Fe and Ni. c Cunha & Lambert 1992. d Walter, Dufour, & Hester 1992. e Rubin, Dufour, & Walter 1993. f From proÐle Ðts ; see Table 3 and ° 4.2.
(13.6 eV) are completely absorbed in the H II region surrounding h1 C Ori and that the clouds are optically thin to UV radiation with E \ 13.6 eV. Therefore all the hydrogen in these cold clouds will be neutral and ionization of other elements will depend on the ionization potentials for any particular stage (we used ionization potentials found in Lang 1980, p. 246). Both N and O have Ðrst ionization potentials that are greater than 13.6 eV ; hence, they should be predominantly neutral. All other elements in this study have Ðrst ionization potentials less than 13.6 eV and can therefore be ionized. The second ionization potential, however, is greater than 13.6 eV in every case, and thus we expect to Ðnd only neutral and Ðrst ionized stages for elements other than N and O. So how are the elements distributed between neutral and ionized species ? One simple way to Ðnd out is via the ionization equilibrium equation (see Spitzer 1975 or Hanson, Snow, & Black 1992). If we assume the cloud to be cold (T \ 100 K), then we can neglect collisional ionization, and the ionization balance for any two species in the same TABLE 6 DEPLETIONS FROM SOLAR Element
Solara
h1 C Ori LOS
Depletionb
Error
C ........ N........ O........ Mg . . . . . . Si . . . . . . . . P ........ S ........ Ca . . . . . . Cr . . . . . . . Mn . . . . . . Fe . . . . . . . Ni . . . . . . . Zn . . . . . .
8.56 8.05 8.93 7.58 7.55 5.45 7.21 6.36 5.67 5.39 7.67 6.25 4.60
\8.75 5.60 8.30 6.60 6.00 4.60 7.9 2.50 D3.60 3.60 5.10 3.60 [ 3.50
\0.19 [2.45 [0.63 [0.98 [1.55 [0.95 ]0.7 [3.9 D[2.07 [1.79 [2.57 [2.65 [ [1.10
... ]1.5, [0.5 ]0.2, [0.1 ]0.1, [0.2 ^0.2 ^0.1 ^1.0 ^0.1 ... ^0.1 ]0.2, [0.4 ^0.1 ...
NOTE.ÈAbundances : same as Table 5. a Anders & Grevesse 1989. b DeÐned as log (N /N )[log (N /N ) . X H X H_
region is given by : Sn TN ! e i`1 \ , (1) N a i r where the NÏs are the column densities of the two ionization stages, Sn T is the average electron density (cm~3), ! is the e photoionization rate (s~1), and a is the recombination rate r (cm3 s~1). From elements for which we have column densities of both the neutral and Ðrst ionized stages, we can derive Sn T by adopting reasonable values for ! and a and assuminge that the neutrals and ions coexist in the rsame region (which is not necessarily true). From Sn T, we can estimate the quantity of neutrals present givene only the column density of the ions. Table 4 shows the value of Sn T e based on the column densities for Mg (which has the most reliable data). Values for ! were taken from van Dishoeck (1988) based on the standard UV radiation Ðeld of Draine (1978) (which is surely too weak since the clouds of the neutral lid are probably close to h1 C Ori), and the values for a come from Pequignot & Aldrovandi (1986) (except for r C, which comes from Weisheit 1973). The quantity a depends primarily on temperature within the cloud, so wer followed the spin temperature analysis done by van der Werf & Goss (1989). Instead of using the Savage & Jenkins (1972) value for N(H I), we used our own (which is roughly 4 times greater). Hence, we arrived at T D 90 K for the neutral lid. The temperature in the foreground clouds is unknown, but we will assume it to be around 90 K as well (which is a reasonable approximation for neutral clouds in the ISM). One of the problems with this ionization equilibrium treatment is that one must assume the two stages under discussion to be cospatial. Since this is probably not the case, the derived Sn T must be handled with caution. e With the above word of caution in mind, we can use the same equation to calculate the ratio of N /N for the neutral ion other species (using the same assumptions and references for ! and a ), given the Sn T just calculated. In every case, r orders of magnitude e N is 4È5 smaller than N ; hence, neutral ion
ABSORPTION-LINE STUDY OF h1 C ORI
No. 1, 1997
the total abundance for all the species except N and O will be based on the Ðrst ionized stage. Recombination data are not available for Cr and Zn, but we will simply assume they are singly ionized like the rest. Even though most material in the H II region is between h1 C Ori and the ionization front (i.e., behind the star ; OÏDell et al. 1993), it is possible that there is some nebular contribution to the lines that would cause an overestimation of column density in the foreground clouds. Baldwin et al. (1991) estimate a blister radius for the Orion Nebula H II region of r \ 3 ] 1017 cm~3. From S II emission they Ðnd a central density of n D 6000 cm~3, which e decreases with increasing distance from h1 C Ori. Using the central n as an upper limit on density and assuming that all e the hydrogen is ionized, we calculated a conservative upper limit for the hydrogen column density in the H II region of N \ 1.8 ] 1021 cm~2. Baldwin et al. (1991) and H Osterbrock, Tran, & Veilleux (1992) both calculate model ionizations for many elements within the H II region. Based on these fractions and our upper limit for N , we calculated H column densities for many of the Ðrst ionized species we observed. In every case except for Fe, Ni, and Si, we found that the model nebular column density was at least 1 order of magnitude less than that of the foreground clouds. The model H II region upper limits for Fe II and Ni II column densities are 14.3 dex and 13.9 dex, respectively. In addition, we estimated an H II region upper limit column density of 15.2 dex for Si II using the theoretical models of Galactic H II regions found in Stasinska (1990). Again, the true values are surely much smaller, but even if the upper limits are accurate, they would only a†ect the foreground column densities by 0.1 or 0.2 dex. Note that we could not check for nebular contamination of P, Cr, Mn, or Zn, but since there are no glaring problems with the other species, we will carefully assume these to be una†ected as well. 4.7. Abundances and Depletions The total abundances relative to hydrogen (as measured in this paper) and errors are listed in Table 5, along with values from other recent studies (to be discussed later). The abundance is given as 12 ] log (N /N ), according to conX study H vention. The abundances from this reÑect the total column densities through both cloud complexes on the LOS to h1 C Ori (Fig. 1). The solar values for each element (Anders & Grevesse 1989) and the corresponding depletions are listed in Table 6, where depletions are deÐned as
A B
5.
region itself (Goudis 1982). Baldwin et al. (1991) show, however, that the column density of hydrogen in the H II region must be roughly 1.6 ] 1020 to 2 ] 1021 cm~2. This is signiÐcantly less than the column density of H I we Ðnd in the foreground. If hydrogen traces dust in any reliable way, then we must conclude that most of the dust resides in the foreground, not in the H II region. Sorrell (1992) gives an informative analysis of the dust in the Orion Nebula, but it is unclear whether the dust in the neutral lid is the same as that in the H II region itself. One of the primary conclusions of Sorrell (1992) is that the dust is primarily silicate in natureÈwith a Si depletion from the gas phase of 33%. Our data for the foreground imply a much higher depletion (see Fig. 8 and Table 6). Most other abundance studies for the Orion nebula are based on emission lines from the hot gas in the inner portions of the H II region close to h1 C Ori. There is no reason a priori to believe that the abundances from this study, which sample clouds outside of the nebula itself, should match those of studies based on emission lines. In fact, we might expect enhanced depletions in the clouds compared to the nebula that is due to dust grain destruction in the H II region (which returns atoms to the gas phase). All the studies listed in Table 5 are emission-line studies except for Cunha & Lambert (1992), which is based on O II lines in the atmosphere of h1 C Ori. Note that there may be a nebular extinction problem a†ecting the emission-line measurements (see Bautista, Pogge, & DePoy 1995), conÐrming that there is at least some dust in the H II region. The comparison of values in Table 5 should be thought of as a comparison of the two di†erent environments. Some elements in the comparison deserve special attention : S : Our high abundance for S just barely agrees with other works within errors. It is not prudent to draw any hard conclusions for S since the abundance is ultimately based on one saturated line (see ° 4.3). N : Our low N abundance nearly matches other works within errors. Both the N I lines are saturated, however, and again no strong conclusions should be drawn. O and Si : The abundances for O and Si are slightly lower than for the other studies, suggesting that O and Si are weakly depleted in these clouds compared to the H II
A B
N N X [ log X . (2) N N H H _ As usual, solar abundances are taken as being typical of abundances everywhere (““ cosmic ÏÏ), even though there is no strong evidence that this is so. Current work by Snow & Witt (1996) addresses this issue. Errors in depletion are due both to errors in our abundances and those from Anders & Grevasse (1989). Note that some species have abundances without errors, as discussed in ° 4.3. Also, the abundance for Ca is based on the proÐle Ðt of Ca II (assumed to be dominant over Ca I) found in Table 3. d \ log
279
DISCUSSION
5.1. Dust in Orion Early analysis of the Orion nebula concluded that all of the dust responsible for the extinction resides in the H II
FIG. 8.ÈDepletions for various lines of sight. See Table 8 for a summary of each LOS and reference. The ““ cosmic ÏÏ value for each element is at the top of the Ðgure (see Table 6). h1 C Ori is given as three points : A central best Ðt bracketed by errors. p Sco has a best Ðt and error bars. The values for f Oph are bracketed by two points. The steep extinction curve mean is taken from Table 4 of Joseph et al. (1986), and the error bars are 1 p errors in the mean. Finally, the depletions for the H II region around h1 C Ori are included for completeness (from Baldwin et al. 1991 ; Osterbrock, Tran, & Veilleux 1992 ; Lucy 1995).
280
SHUPING & SNOW
region. The column density could be even lower, as discussed in ° 4.3. Mg : The abundance of Mg agrees well with Baldwin et al. (1991), implying that Mg is not depleted in these clouds relative to the H II region. Fe and Ni : Both species are roughly an order of magnitude more depleted than in the H II region, implying some di†erence between the composition of the dust in the H II region and that in the neutral lid. Note that we may have slightly overestimated the Fe and Ni abundance, as discussed in ° 4.6. We can understand the general trends of these depletions if we adopt the dust grain model found in Whittet (1992, p. 224), although our data are not nearly good enough to prove or disprove any model in particular. Since the foreground appears to have high R (and it seems that the dust in the H II region is negligible),Vthe grains in the neutral lid and foreground clouds are most likely associated with the Orion molecular cloud (although we cannot prove this decisively). We will make the assumption that the grains in the Orion Nebula had the same structure as those in the foreground before the star formation near h1 C Ori ““ turned on :ÏÏ i.e., the foreground grains represent initial conditions for the Orion Nebula. For molecular cloud regions, it is thought that grains have a core composed of refractories (silicates and species with high condensation temperatures like Fe, Mg, Ni, Cr, etc.), an inner mantle of water and ammonia ices and an outer mantle of molecular oxygen and nitrogen, and CO (Whittet 1992, p. 224). These outer mantles are volatile in general, and when exposed to energetic conditions (like those found in the Orion Nebula), they liberate C, N, and O readily. The greater abundance of N and especially O in the H II region, as opposed to the foreground, supports this with no conclusion drawn for C (remembering that N and C su†er from line saturation and blending problems, respectively). More importantly, the depletion of Si, Fe, Ni, and O (in the form of silicates) in the foreground implies dust core destruction in the H II region as well. Mg should also show depletion in the foreground since it too is a grain core element. We have no immediate explanation for the similarity of the Mg abundances in the H II region and the foreground, except perhaps that our data and analysis and/or the other emission studies are inaccurate for Mg. We draw no conclusions regarding S and Zn since they are generally little depleted anyway. The recent study by Meyer et al. (1994) found O abundances toward two di†erent stars in Orion, i Ori and i Ori (see Table 7). i Ori is important to this study because it lies in between the two cloud complexes on the LOS to h1 C Ori (OÏDell et al. 1993 ; also see Fig. 1). Hence we would expect the column density of O to be less toward i Ori than h1 C Ori, which is indeed the case : i Ori has roughly 6% the amount of O as h1 C Ori, implying that the foreground TABLE 7 OTHER SIGHT LINES IN ORION LOS
log N (cm~2)O
i Ori . . . . . . . . . . i Ori . . . . . . . . . h1 C Ori . . . . . .
16.7 17.0 17.9
12 ] log (N /N ) O H 8.59 8.48 8.30 (]0.2, [0.1)
NOTE.Èi and i Ori data are from Meyer et al. 1994
Vol. 480
clouds are somewhat thinner than the neutral lid. Notice, however, that the abundances (relative to H) are similar. From the electron density calculated in Table 4, we can estimate a total density of hydrogen for the LOS toward h1 C Ori. It is assumed that all of the H is neutral, and hence the primary source of electrons in the cloud is C (which is far more abundant than anything else except N and O, which are neutral anyway). Therefore, Sn T D Sn T and e C n \ n (N /N ). Using the n value based on Mg (best H C H C e data), n (T \ 90 K) [ 9 cm~3, remembering that the value H an upper limit. This should be a good lower limit for N is C for the hydrogen density since, as stated earlier, molecular hydrogen is D4 orders of magnitude less abundant than H I. Our lower limit for n is 10È1000 times lower than H values found in van der Werf & Goss (1989), based on H I (21 cm) column densities and some assumptions regarding the size of each absorbing system. Note that our value for n is an average over the whole LOS ; thus, the discrepancy H between our lower limit and values found by van der Werf & Goss (1989) is not necessarily cause for alarm. The values for n shown in Table 5 of van der Werf & Goss (1989), H seem high ; especially for component C, which is at however, the inner edge of the neutral region and is quoted with n \ H 2000È6500 cm~3. As discussed in ° 4.6, Baldwin et al. (1991) derive an upper limit of n D n \ 6000 cm~3, which H h1e C Ori. We feel that the descends rapidly with radius from densities in van der Werf & Goss (1989) have been systematically overestimated, probably because the size assumptions for each component are too small. We feel our n is a H good lower limit for the typical density in the foreground. The total column density of H on the LOS is (4.2È 6.0) ] 1021 cm~2, based on N(H I) for the foreground and the estimate of N(H) in the H II region calculated by Baldwin et al. (1991) (note again that H is negligible). The 2 h1 C Ori is (1.2È ratio of N(H) to E(B[V ) for the LOS to 1.8) ] 1022 cm~2 per magnitude. This is much higher than the nominal interstellar value (from various sight lines) of 5.9 ] 1021 cm~2 per mag (see Spitzer 1975, p. 156). The ratio calculated for h1 C Ori implies that there is ““ not enough ÏÏ dust to account for the column density of hydrogen. This Ðts well with the absence of a small grain population and may imply some dust grain destruction. In summary, it seems clear that most of the dust on the LOS to h1 C Ori is in the cold foreground (the neutral lid and foreground clouds), with a negligible amount in the H II region. The extinction curve for h1 C Ori also shows an absence of small grains in this dust population, which is probably due to coagulation during the collapse of the Orion molecular cloud. This conclusion is supported by the corresponding absence of molecular hydrogen (which forms most efficiently on small grains) and the high ratio of H I to reddening (which may also imply some grain destruction). In addition, the di†erences in abundances between the H II region and the foreground are consistent (except for Mg) with a core and mantle grain structure and imply core destruction in the H II region. The data are not good enough, however, to examine this structure in detail. 5.2. Other Sight L ines Figure 8 compares the depletions toward h1 C Ori to those from other sight lines. Ultimately, we would like to see if there is some sort of correlation between gas-phase depletion and the steepness of the far-UV extinction for a particular LOS. p Sco (Allen, Snow, & Jenkins 1990) has
ABSORPTION-LINE STUDY OF h1 C ORI
No. 1, 1997 TABLE 8
OTHER SIGHT LINES LOS h1 C Ori . . . . . . . . HD 37903a . . . . . . p Scob . . . . . . . . . . . f Ophc . . . . . . . . . .
A
E(B[V )
V 1.87 1.44 1.52 0.99
0.34 0.35 0.40 0.32
R
V 5.50 4.11 3.8 3.09
Far-UV Extinction Flat Flat Steeper Steep
a Joseph et al. 1986. b Allen, Snow, & Jenkins 1990. c Morton 1974.
slightly steeper far-UV extinction than h1 C Ori. Both f Oph (Morton 1974) and the mean depletions from Joseph et al. (1986, Table 4) represent sight lines with steep far-UV extinction. The star HD 37903 excites the reÑection nebula NGC 2023, is embedded in the Orion molecular cloud, and has Ñat far-UV extinction (Joseph et al. 1986). The Orion nebular depletions are included as well for comparison (from Baldwin et al. 1991 ; Osterbrock, Tran, & Veilleux 1992 ; Lucy 1995). Parameters for each LOS are shown in Table 8. Any similarities in depletion, from extinction curve to extinction curve, indicate the same probability of sticking for any particular element regardless of mean grain size (Snow 1992). Although contrived, we must also consider the possibility that the grains responsible for depletion are not the same as those responsible for extinction curve behavior. There do not appear to be any strong trends in depletion for the various extinction curves. Note that p Sco (with Ñat far-UV extinction) is systematically less depleted than all other sight lines for nearly all elements. The closest match in depletions to h1 C Ori is HD 37903, which implies that depletion is more dependent on local environment than extinction curve behavior (both are situated in Orion molecular cloud star forming regions). It does seem from Figure 8, however, that sight lines with Ñat far-UV extinction may be slightly less depleted for Mg and maybe Si and P. There is deÐnitely no gross trend for the heavier elements. It is possible that these data are not good enough to detect whatever systematic di†erences there are in depletions from one extinction curve to another. As usual, we would like to test more sight lines, especially those with Ñat far-UV extinction, with higher quality data. Other lines of sight in the Orion region could provide a test for whether or not depletions are really a strong function of local environment as opposed to grain size population. Another interesting result is the depletion of N. At Ðrst, we thought our unusually low abundance and depletion for N toward h1 C Ori was highly questionable (see ° 4.3). HD 37903, however, has a similarly low depletion value for N, which forces us to wonder if the result is indeed correct. Better data are needed toward h1 C Ori before any conclusion can be drawn. 6.
CONCLUSION
It was critical for this study to have the highest S/N possible with IUE. The Ayres reduction routines and some
cross-correlation analysis now make it possible to extract IUE spectra with higher S/N than previous reductions ; proving that IUE archive data is still tremendously useful for wide band spectral analysis, even after 18 years. Based on the work of OÏDell et al. (1993), the LOS toward h1 C Ori has two primary components which generate interstellar absorption : the neutral lid cloud complex, and some thin foreground clouds. In this study, we have measured equivalent widths of interstellar lines for many species on the LOS to h1 C Ori. These widths (along with transition information) were then Ðtted to a semiempirical curve of growth (COG) to determine the column density for each species. The COGs were derived from high-resolution spectra on the LOS to h1 C Ori. Assuming that the other species follow the same component structure as the high-resolution data, the COG Ðts yield column densities for the integrated LOS (see Table 2). These column densities matched well with those found from Ðtting each species independently to its own COG. In addition, the hydrogen Lya line was also measured, yielding a column density of log N(H I) \ 21.60 ^ 0.1 cm~2. Based on some simple assumptions regarding the environment of these clouds, an electron density was calculated and corresponding ionization structures were determined. Finally, we calculated the abundances (relative to hydrogen) and depletions (from ““ cosmic ÏÏ) for the LOS (see Tables 5 and 6). From these abundances and depletions came a few interesting results : 1. Abundance discrepancies (except for Mg) between the Orion Nebula and foreground are consistent with a core and mantle grain model and imply some core destruction in the H II region (° 5.1). 2. Using some simple assumptions, a lower limit for the total hydrogen density of n [ 9 cm~3 was calculated for H the foreground. 3. The depletions toward h1 C Ori were compared to depletions for other sight lines (Fig. 8), and it seems that there is no gross correlation between far-UV extinction and depletion, except in the cases of Mg and possibly Si and P, which seem slightly less depleted for Ñat far-UV cases. 4. The depletions for h1 C Ori are very similar to those for HD 37903, implying that perhaps depletions are more dependent on local environment than grain size population. Further work should include more studies of Ñat far-UV extinction sight lines (especially more sight lines in Orion) with higher quality data. We would like to Ðrst thank M. M. Hanson for all her help throughout this project. We would also like to thank T. Ayers for his reduction programs and assistance. Thanks go to L. Hobbs and C. R. OÏDell for graciously supplying their data to us for this analysis. Finally, R. Y. Shuping would like to thank D. E. Schutz for many useful conversations and support. This research has been supported by NASA grants NSG-5300 to the University of Colorado and GSRP NGT-51380 to R. Y. Shuping.
REFERENCES Allen, M., Snow, T., & Jenkins, E. 1990, ApJ, 327, 377 Anders, E., & Grevesse, N. 1989, Geochim. Cosmochim. Acta, 53, 197 Baldwin, J. A., Ferland, G. J., Martin, P. G., Corbin, M. R., Cota, S. A., Peterson, B. M., & Slettebak, A. 1991, ApJ, 374, 580 Bautista, M. A., Pogge, R. W., & DePoy, D. L. 1995, ApJ, 452, 685 Bless, R., & Savage, B. 1972, ApJ, 171, 293
281
Bohlin, R. C. 1975, ApJ, 200, 402 Bohlin, R. C., & Savage, B. D. 1981, ApJ, 249, 109 Cardelli, J., Clayton, G., & Mathis, J. 1988, ApJ, 329, L33 ÈÈÈ. 1989, ApJ, 345, 245 Cowie, L. L., Songalia, A., & York, D. G. 1979, ApJ, 230, 469 Cunha, K., & Lambert, D. L. 1992, ApJ, 399, 586
282
SHUPING & SNOW
Draine, B. T. 1978, ApJS, 36, 595 Fitzpatrick, E., & Massa, D. 1986, ApJ, 307, 286 ÈÈÈ. 1988, ApJ, 328, 734 ÈÈÈ. 1990, ApJS, 72, 163 Franco, J., & Savage, B. D. 1982, ApJ, 255, 541 Genzel, R., & Stutzki, J. 1989, AR&A, 27, 41 Goudis, C. 1982, in The Orion Complex : A Case Study of Interstellar Matter, Vol. 90 (Dordrecht : Reidel), 38 Hanson, M. M., Snow, T. P., & Black, J. H. 1992, ApJ, 392, 571 Hobbs, L. M. 1978, ApJS, 38, 129 Joseph, C. L., Snow, T. P., Seab, C. G., & Crutcher, R. M. 1986, ApJ, 309, 771 Jura, M. 1980, ApJ, 235, 63 Lang, K. R. 1980, Astrophysical Formulae (Berlin : Springer) Lucy, L. B. 1995, A&A, 294, 555 Mathis, J. S. 1990, AR&A, 28, 37 Meyer, D. M., Jura, M., Hawkins, I., & Cardelli, J. A. 1994, ApJ, 437, L59 Morton, D. C. 1974, ApJ, 193, L35 ÈÈÈ. 1975, ApJ, 197, 85 ÈÈÈ. 1991, ApJS, 77, 119 OÏDell, C. R. 1994, Ap&SS, 216, 267 OÏDell, C. R., Valk, J. H., Wen, Z., & Meyer, D. M. 1993, ApJ, 403, 678 Osterbrock, D. E., Tran, H. D., & Veilleux, S. 1992, ApJ, 389, 305 Pequinot, E., & Aldrovandi, S. M. V. 1986, A&A, 161, 169
Rubin, R. H., Dufour, R. J., & Walter, D. K. 1993, ApJ, 413, 242 Savage, B. D., Bohlin, R. C., Drake, J. F., & Budich, W. 1977, ApJ, 216, 291 Savage, B. D., & Jenkins, E. B. 1972, ApJ, 172, 491 Snow, T. P. 1992, Australian J. Phys., 45, 543 Snow, T. P., Black, J. H., van Dishoeck, E. F., Burks, G., Crutcher, R. M., Lutz, B. L., Hanson, M. M., & Shuping, R. Y. 1996, ApJ, 465, 245 Snow, T. P., & Witt, A. N. 1996, ApJ, 468, L65 Sorrell, W. H. 1992, MNRAS, 255, 594 Spitzer, L., Jr. 1975, Physical Processes in the Interstellar Medium (New York : Wiley) Stasinska, G. 1990, A&AS, 83, 501 van Dishoeck, E. F. 1988, in Rate Coefficients in Astrochemistry, ed. T. J. Millar & D. A. Williams (Dordrecht : Kluwer), 49 van der Werf, P. P., & Goss, W. M. 1989, A&A, 224, 209 Vidal-Madjar, A., Laurent, C., Bonnet, R. M., & York, D. G. 1977, ApJ, 211, 91 Walter, D. K., Dufour, R. J., & Hester, J. J. 1992, ApJ, 397, 196 Weisheit, J. C. 1973, ApJ, 185, 877 Welty, D. E., Thorburn, J. A., Hobbs, L. M., & York, D. G. 1992, PASP, 104, 737 Whittet, D. C. B. 1992, Dust in the Galactic Environment (New York : AIP) Wood, B. E., & Ayres, T. R. 1995, ApJ, 443, 329