Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/~lt/preprints/oribar/oribar.ps.gz Äàòà èçìåíåíèÿ: Tue Sep 11 16:56:49 2007 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 07:29:05 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 17

astro­ph/9710051
4
Oct
1997
A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
08(09.03.1; 09.13.2; 13.19.3)
ASTRONOMY
AND
ASTROPHYSICS
4.10.1997
Near Infrared Spectra of the Orion Bar
A. Marconi 1;3 , L. Testi 1;4 , A. Natta 2 , and C.M. Walmsley 2
1 Dipartimento di Astronomia e Scienza dello Spazio, Universit`a degli Studi di Firenze, Largo E.Fermi 5, I­50125 Firenze, Italy
2 Osservatorio Astrofisico di Arcetri, Largo E.Fermi 5, I­50125 Firenze, Italy
3 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218
4 Division of Physics Mathematics and Astronomy, Caltech, MS 105­24, Pasadena, CA 91125, USA
Received date; accepted date
Abstract. We have used the LONGSP spectrometer on
the 1.5­m TIRGO telescope to obtain long slit spectra in
the J, H, and K wavelength bands towards two positions
along the Orion bar. These data have been supplemented
with images made using the ARNICA camera mounted on
TIRGO as well as with an ESO NTT observation carried
out by Dr A. Moorwood. We detect a variety of transitions
of hydrogen, helium, OI, FeII, FeIII, and H 2 . From our
molecular hydrogen data, we conclude that densities are
moderate (3 \Gamma 6 \Theta 10 4 cm \Gamma3 ) in the layer responsible for
the molecular hydrogen emission and give no evidence for
the presence of dense neutral clumps. We also find that
the molecular hydrogen bar is likely to be tilted by ¸10
degrees relative to the line of sight. We discuss the relative
merits of several models of the structure of the bar and
conclude that it may be split into two structures separated
by 0.2­0.3 parsec along the line of sight. It also seems likely
to us that in both structures, density increases along a line
perpendicular to the ionization front which penetrates into
the neutral gas.
We have used the 1.317¯m OI line to estimate the
FUV radiation field incident at the ionization front and
find values of 1 \Gamma 3 \Theta 10 4 greater than the average inter­
stellar field. From [FeII] line measurements, we conclude
that the electron density in the ionized layer associated
with the ionization front is of order 10 4 cm \Gamma3 . Finally, our
analysis of the helium and hydrogen recombination lines
implies essential coincidence of the helium and hydrogen
Str¨omgren spheres.
Key words: interstellar medium: HII regions -- interstel­
lar medium : Orion -- IR: interstellar medium : lines and
bands
Send offprint requests to: C.M. Walmsley
1. Introduction
The properties of the Orion nebula are the starting point
for many of our ideas on high mass stars and their in­
teractions with the environment. A symposium held in
1981 (Glassgold et al. 1982) summarises much of what was
known at that time. More recent work has been reviewed
by Genzel & Stutzki (1989). In general, the aim has been
to understand the ionization structure and dynamical evo­
lution of the nebula. In recent years, much attention has
been paid to the hot neutral gas adjacent to the ionization
front known as a PDR or Photon Dominated Region.
Work on the ionized nebula has tended to concen­
trate upon determinations of the ionization structure
and elemental abundances (see Peimbert 1982, Simpson
et al. 1986). More recent optical and infrared studies
have been carried out by Osterbrock et al. (1990, 1992),
Baldwin et al. (1991), Peimbert et al. (1992), Pogge et
al. (1992), De Poy & Pogge (1994), Bautista et al. (1995),
Rubin et al. (1993), and Rodriguez (1996). A review of the
results has been made by Peimbert (1993) and discussions
of the methods employed are given by Mathis (1995) and
by Peimbert (1995). These studies in general indicate that
the major fraction of elements such as C,N,O,S are in the
gas phase within the ionized nebula whereas species such
as Si and Fe appear to be depleted by roughly an order
of magnitude relative to abundances either in the Sun or
nearby B stars.
Radio work on Orion has revealed an immense variety
of structures in the emissions of the ionized gas ( e.g. Felli
et al. 1993, Yusuf--Zadeh 1990). Particularly striking is the
bar--like structure situated roughly 2 arc minutes (0.25
parsec) to the south--east of the Trapezium stars which is
the subject of this article. ``The Bar'' is also observed in
molecular line emission (see below) and clearly marks an
ionization front where Lyman continuum photons from the
O6 star \Theta 1 C Ori are absorbed. The dynamical behavior
of the ionized gas can be studied in radio recombination
lines (Pankonin et al. 1979, Wilson & J¨ager 1987, Wil­
son & Filges 1990, Wilson et al. 1997) from which one

2 A. Marconi et al.: NIR Spectra of the Orion Bar
concludes that much of the ionized material in Orion is
streaming towards the observer.
Infared studies of HII regions sample not only the ion­
ized gas but also the adjacent neutral material or PDR. A
recent review of the properties of these regions is that of
Hollenbach & Tielens (1997) (see also the discussions of
Genzel 1992, and Walmsley 1997). Modelling studies have
been carried out by Tielens & Hollenbach (1985), Hollen­
bach et al. (1991), Sternberg & Dalgarno (1989; 1995),
Fuente et al. (1993), Jansen et al. (1995a,b), Bertoldi &
Draine (1996) and Draine & Bertoldi (1996). Much of
this activity has centred on attempts to understand the
properties of ''The Bar'' mentioned above. Recent obser­
vational studies using a variety of molecular tracers have
been carried out by Tielens et al. (1993), by Tauber et
al. (1994; 1995), by Hogerheijde et al. (1995), and by van
der Werf et al. (1996). These show a stratification along
the direction of the perpendicular to the bar in the plane
of the sky. This is in the sense expected for gas heated
by the Trapezium stars and consistent according to the
models with attenuation by a gas of density 5 10 4 cm \Gamma3 .
However, the data also seem to show that the gas in the
bar is far from homogeneous and that clumps of density
as high as 10 6 cm \Gamma3 are embedded in the filament. Such
high density condensations presumably either have been
or will be soon overrun by the ionization front and will
give rise to dense ionized globules within the HII region
(see e.g Lizano et al. 1996, Dyson et al. 1995). Under­
standing the characteristics of such high density clumps
may thus be of critical importance for the evolution of the
HII region.
One of the most useful tracers of PDR's has turned
out to be the near infrared lines of molecular hydrogen.
For example, van der Werf et al. used the FAST camera
on the ESO/MPI 2.2 m telescope to image the H 2 v =
1 ! 0 S(1) (2.122 ¯m) and v = 2 ! 1 S(1) (2.248 ¯m)
lines towards the bar with 1.5 00 resolution. These show
that the transition from atomic to molecular hydrogen
in the bar occurs 15 arc seconds (0.03 pc at a distance
of 450 pc) to the SE of the ionization front (i.e. away
from the ionizing stars). Van der Werf et al. also find that
the ratio R 12 of the intensities of the 1 ! 0 and 2 !
1 lines varies between a value of 8:1 \Sigma 0:7 at the peak
of the H 2 emission to a value of 3:4 \Sigma 1:9 30 00 from the
ionization front on the side shielded from the radiation
of the Trapezium stars. The latter value is characteristic
of UV­pumped fluorescent emission in a low density gas
(Sternberg and Dalgarno 1989).
The present study had as its aim to obtain near IR
spectra of the gas in the vicinity of the bar in order to
verify and extend understanding of the physical conditions
on both sides of the ionization front. We were partly moti­
vated by the idea that there is a link between the ionized
and PDR components in that the former is mainly sen­
sitive to the radiation just shortward of 912 š A while the
latter is basically a measure of the radiation longward of
this limit (see Bertoldi & Draine 1996 for a discussion).
It is thus of considerable interest to compare the two us­
ing the same instrument. We therefore used the TIRGO
telescope on the Gornergrat (Switzerland) to obtain slit
spectra in the J, H, and K bands at 3 positions in the
vicinity of the bar, shown in Fig. 1. As supplementary in­
formation, we also made use of unpublished observations
carried out using the IRSPEC spectrometer on the ESO
NTT telescope by Dr A.Moorwood.
We summarize in the next section the techniques used
for observations and data reductions. The results are pre­
sented in Sect.3 and discussed in Sect. 4. We summarize
our conclusions in Sect. 5.
2. Observations
2.1. ARNICA Observations
The Orion Bar was observed during two observing runs in
January 1996 and February 1997 using ARNICA (ARcetri
Near Infrared CAmera) mounted on the 1.5m TIRGO 1
telescope. ARNICA is equipped with a 256x256 NIC­
MOS3 array, the pixel size with the optics used at TIRGO
is 0:96 00 ; for a complete description of the instrument
and of its performance, see Lisi et al. (1996) and Hunt
et al. (1996). The Bar was imaged in the three J, H,
and K broad band filters (centered at 1.25, 1.65, and
2.2 ¯m, respectively) and in the Brfl narrow band filter
(– = 2:166 ¯m, \Delta–=– ¸ 1%, Vanzi et al. 1997b). The
seeing was approximately 2­3 00 and the observed field was
approximately ¸ 4 0 :5 \Theta 4 0 :5, covering all the Bar region.
Data reduction was carried out using the IRAF 2 and AR­
NICA 3 (Hunt et al. 1994) software packages. Photomet­
ric calibration in the J, H and K bands was performed
by observing photometric standard stars from the list of
Hunt et al. (1997) ; the calibration accuracy is estimated
to be ¸ 5%. The Brfl image was continuum subtracted
and calibrated using the K band image. We show in Fig. 1
(left panel) an image obtained combining the J, H, and K
images and (right panel) the continuum subtracted Brfl
image.
2.2. LONGSP Observations
J (1.25¯m), H(1.65 ¯m), and K (2.2¯m) band spectra of
the Orion Bar were obtained using the LonGSp (Longslit
Gornergrat Spectrometer) spectrometer mounted at the
Cassegrain focus on the TIRGO telescope. The spectrom­
eter is equipped with cooled reflective optics and a grat­
1 The TIRGO telescope is operated by the C.A.I.S.M.I.­
C.N.R Firenze, Italy
2 IRAF is made available to the astronomical community by
the National Optical Astronomy Observatories, which are op­
erated by AURA, Inc., under contract with the U.S. National
Science Foundation
3 A description of ARNICA can be obtained from the Arcetri
Observatory at ftp://150.217.20.1/pub/arnica/

A. Marconi et al.: NIR Spectra of the Orion Bar 3
Composite J, H and K Continuum subtracted Brg
Fig. 1. On the left: composite J, H, and K image of the Orion Bar region showing the slit positions used for the LONGSP and
IRSPEC observations. Coordinates are offsets in right ascension and declination relative to the position of the star \Theta 2 A Ori
(R.A(1950)= 5 h 32 m 55. s 5 , Dec(1950)= ­5 ffi 26 0 51 00 ). We show our three slit positions (1,2,and 3) as well as the four positions
for which we tabulate line intensities (A,B,C,CS). On the right: Brfl image of the Orion Bar region.
ing in Littrow configuration. The detector is a 256\Theta256
engineering grade NICMOS3 array (for detector perfor­
mances see Vanzi et al. 1995). The pixel sizes are 11.5 š A
(first order) and 1: 00 73 in the dispersion and slit directions,
respectively. LONGSP operates in the range 0.9­2.5 ¯m
achieving a spectral resolution at first order of R ' 550
in J, 700 in H and 950 in K. For a more comprehensive
description of the instrument, refer to Vanzi et al. (1997a).
Observations were conducted in two runs in January
and March 1996 under non­photometric conditions. The
slit used had dimensions 3: 00 5\Theta70 00 and was oriented N­S.
The seeing during the observations was in the range 2 00 ­4 00 .
The Orion Bar was observed at three slit positions labeled
as 1, 2, 3 and shown in Fig. 1 superimposed on a NIR
image obtained by co­adding the J, H and K ARNICA
observations discussed in the previous section. Position 1
and 2 were chosen in order to study the variation of line
intensities along a cut encompassing all the bar. Position
3 was subsequently chosen to be coincident with the CS
peak discovered by van der Werf et al. at R.A.= 5 h 32 m
58.5 s and Dec.= ­5 ffi 26 0 25 00 (B1950.0). This high density
(10 6 cm \Gamma3 ) clump appears to be illuminated directly from
the Trapezium and we thought it useful to examine di­
rectly the relative variations in line intensities across the
clump. The center of the slits were offset by ­35 00 , ­23 00
(Pos. 1) ­35 00 , 23 00 (Pos. 2) and 42 00 , 26 00 (Pos. 3) in R.A.
and Dec. with respect to the star ` 2 A Ori.
At each grating position we performed 5 ABBA cycles
(A=on source, B=on sky) with an on­chip integration time
of 60 sec, for a total of 10 min integration on source. At
the beginning or at the end of the five cycles on the object
we performed 1 ABBA cycle on the O6 star BS 1895 (\Theta 1 C
Ori).
Data reduction was performed with the ESO package
MIDAS, within the context IRSPEC, modified to take into
account LonGSp instrumental characteristics. The frames
were corrected for bad pixels, flat­fielded, sky subtracted
and wavelength calibrated using the OH sky lines present
on all the frames (Oliva & Origlia 1992). After a direct
subtraction, sky removal was optimized by minimizing the
standard deviation in selected areas where the OH sky
lines were poorly subtracted but no object emission was
present. The wavelength calibration was performed to bet­
ter than 1/5 of a pixel ('2 š A).
The spectra were then corrected for telluric absorption
by dividing by the featureless spectrum of \Theta 1 C Ori. For
more details on LonGSp data reduction, see Vanzi et al.
1997a.
Flux calibration of the spectra in the J, H, and K
bands was achieved by rescaling the observed flux dis­
tribution along the slit to match that obtained from the
ARNICA images at the positions of the slits. We consider
such calibration accurate to '20% when comparing the
fluxes of lines measured in two different bands. Indeed, the

4 A. Marconi et al.: NIR Spectra of the Orion Bar
Table 1. Spectra in the J, H and K bands
Line – A B C CS
(¯m)
[OI] 2p 3 3D \Gamma 2p 3 3P 1.129 3.6 4.7 6.8 12
[PII] 3p 2 D2 \Gamma 3p 2 P2 1.189 2.4 3.5 3.2 6.6
HeI 5 3 D \Gamma 3 3 P ffi 1.197 5.1 5.4 6.0 5.4
HeI 4 3 P ffi \Gamma 3 3 S 1.253 6.9 5.9 5.3 2.3
[FeII] 4sa 4 D 7=2 \Gamma 4sa 6 D 9=2 1.257 4.8 8.3 6.7 18
?? 1.268 2.3 1.5 !1.8 5.7
HeI 5 3 F ffi \Gamma 3 3 D 1.278 18 16 18 26
HI 5­3 1.282 370 400 450 430
HeI 5 3 P ffi \Gamma 3 3 D 1.298 2.2 2.3 1.6 !2.6
+[FeII] 4sa 4 D 3=2 \Gamma 4sa 6 D 1=2
[OI] 2p 3 4S \Gamma 2p 3 3P 1.317 3.3 6.8 7.4 8.4
[FeII] 4sa 4 D 7=2 \Gamma 4sa 6 D 7=2 1.321 1.1 2.3 !1.6 4.7
HeI 5 1 S \Gamma 3 1 P ffi 1.341 1.5 1.3 1.6 2.3
HI 21­4 1.514 3.3 2.7 2.7 3.0
HI 20­4 1.520 3.6 3.6 3.6 3.9
HI 19­4 1.526 4.4 4.3 4.5 4.4
HI 18­4 1.534 6.4 6.2 6.6 8.7
+[FeII] 4sa 4 D 5=2 \Gamma 3d 7 a 4 F 9=2
HI 17­4 1.544 6.5 5.9 6.3 6.2
HI 16­4 1.556 8.2 7.3 7.2 7.2
HI 15­4 1.570 9.4 9.0 8.2 9.4
HI 14­4 1.588 11 10 10 13
[FeII] 4sa 4 D 3=2 \Gamma 3d 7 a 4 F 7=2 1.600 0.5 0.6 0.9 3.0
HI 13­4 1.611 14 13 13 16
HI 12­4 1.641 18 17 17 19
[FeII] 4sa 4 D 7=2 \Gamma 3d 7 a 4 F 9=2 1.644 5.9 11 8.7 24
[FeII] 4sa 4 D 5=2 \Gamma 3d 7 a 4 F 7=2 1.677 0.8 1.1 !1.1 2.3
HI 11­4 1.681 21 21 23 24
HeI 4 3 D \Gamma 3 3 P ffi 1.701 9.1 8.5 8.6 6.4
HeI 10 1 P ffi \Gamma 4 1 D 1.732 0.5 1.0 2.1 1.2
HI 10­4 1.737 26 28 29 30
HeI 7 3 P ffi \Gamma 4 3 S 1.746 0.9 1.8 3.0 3.9
+H2(1,0)S(7)
H2(1,0)S(6) 1.788 1.5 1.1 1.6 2.2
H2(1,0)S(2) 2.034 0.6 .... .... 4.9
HeI 2 1 P ffi \Gamma 2 1 S 2.058 99 110 74 95
H2(2,1)S(3) 2.073 !0.5 1.2 3.6 4.2
HeI 4S \Gamma 3 P ffi 2.113 4.2 3.7 2.8 2.1
H2(1,0)S(1) 2.122 1.5 5.1 17 15
H2(2,1)S(2) 2.154 !0.1 0.2 1.3 1.3
HeI 7F ffi \Gamma 4 1 D 2.162 3.0 2.5 2.4 1.9
HI 7­4 2.166 100 100 100 100
??+H2(3,2)S(3) 2.199 0.5 0.9 1.4 3.5
[FeIII]3d 6 G5 \Gamma 3D 6 H6 2.219 1.4 2.0 1.3 2.3
H2(1,0)S(0) 2.223 0.5 1.6 6.6 6.5
[FeIII]3d 6 G4 \Gamma 3d 6 H4 2.242 0.7 0.6 0.6 0.4
H2(2,1)S(1) 2.248 0.4 1.2 4.1 3.8
??+H2(3,2)S(2) 2.286 0.7 0.6 1.0 1.4
Note: Intensity 100 corresponds to:
A ) 2:65 \Theta 10 \Gamma3 erg cm \Gamma2 s \Gamma1 sr \Gamma1
B ) 2:08 \Theta 10 \Gamma3 erg cm \Gamma2 s \Gamma1 sr \Gamma1
C ) 0:61 \Theta 10 \Gamma3 erg cm \Gamma2 s \Gamma1 sr \Gamma1
CS) 0:50 \Theta 10 \Gamma3 erg cm \Gamma2 s \Gamma1 sr \Gamma1
comparison between the flux distributions of H 2 (1,0)S(1)
–2.12¯m in our observations and in those by van der Werf
et al. shows only a 10% discrepancy in the absolute flux
level.
2.3. IRSPEC Observations
IRSPEC (Moorwood et al. 1991) observations of the bar
using the ESO NTT telescope were carried out in 1991 by
Dr A. Moorwood. The detector was a SBRC 62x58 InSb
array with pixels of '5š A (H band ) along the dispersion
and 2: 00 2 along the slit direction. The slit, 4: 00 4\Theta120 00 in
size, was oriented NE­SW as shown in Fig 1. The data
were uncalibrated and in this paper, we merely make use
of the profiles of line intensity along the slit.
3. Results
In Figs. 2, 3, 4, we show sample spectra averaged over
three portions of slit positions 1 and 2 (see Fig. 1). Based
on the profiles of line intensity along the N­S direction,
we decided to divide the combined slit into three sections
which we named A (a 28 pixel section to the north), B (14
pixels in a central region), and C (28 pixels to the south).
These sections are displayed in the left panel of Fig. 1. The
hydrogen and helium recombination lines peak in position
A and become weaker to the south whereas the molecular
hydrogen lines become stronger and reach their maximum
intensity in position C. However, there is clearly ionized
gas towards the region C and vice­versa. In Figs. 2, 3, 4,
we show also the spectra summed over 40 pixels in slit
position 3 (the ``CS peak'').
In Table 1, we give intensities corresponding to the
spectra shown in Figs. 2, 3 4. They have been averaged
over the respective apertures. Line intensities, in each re­
gion, are given relative to Brfl (put equal to 100). Typical
uncertainties vary from ' 10% for strong lines (I ? ¸ 10)
to ' 20% for lines with 1 ! ¸ I ! ¸ 10 and about 50% for
the others. When comparing lines in two different bands,
a 20% error due to spectrophotometric calibration must
also be taken into account.
The line intensities can be corrected for reddening
using the Cardelli et al. (1989) prescription A – /A V =
0:48– \Gamma1:61
¯ , where we have adopted R=5.5, as appropriate
for the Orion region. ?From the ratio Pafi/Brfl we derive
A V ¸ 2 mag (Sect.3.2). Note that this value applies only
to lines forming in the ionized gas.
The variation in intensity along the amalgamation of
slit positions 1 and 2 of a variety of interesting line tracers
is shown in Figs. 5 to 7. Figure 8 shows intensity varia­
tions in a number of lines from the IRSPEC data. Figure
9 plots the intensity profile of selected lines at slit posi­
tion 3. ?From Figs. 5­7, we can see that region A coincides
roughly with a peak in the lines of HI, which have a second
peak in B, while the emission of molecular hydrogen has a
strong peak in C and a weaker one in B. Lines of FeII and

A. Marconi et al.: NIR Spectra of the Orion Bar 5
Fig. 2. J band spectrum obtained towards the three slit sections A, B and C and in the CS position shown in Fig.1. The upper
spectrum in each panel has been multiplied by 15 to emphasize weak features.
OI have a sharp peak in B and a secondary one in A, but
are absent in region C. In the following, we will use the
definition A, B and C (see caption to Fig. 5) both to iden­
tify the peaks in the intensity profiles and with reference
to Table 1.
We now summarize our results starting with the IR­
SPEC data which have the advantage that the slit is ori­
ented perpendicularly to the bar. We then discuss in turn
the hydrogen and helium recombination lines (which we
presume form in the ionized gas), the two oxygen transi­
tions, the collisionally excited iron lines which may form
close to the ionization front, and the molecular hydrogen
lines which are thought to form in hot neutral gas close to
the ionization front.
3.1. IRSPEC cut
The observations made using IRSPEC provide a useful
introduction to the TIRGO results which have a wider
spectral coverage. Figure 8 compares profiles in the Br12
line from the ionized gas, in the FeII 1.644¯m (4sa 4 D 7=2 ­
3d 7 a 4 F 9=2 ) line which traces gas close to the ionization
front (see below), and in the molecular hydrogen v=1­0
S(1) line from hot (T? 1000K) molecular gas. Figure 8
demonstrates the fact that the molecular hydrogen peak is
offset ¸16 00 (0.035 pc) from the ionization front as marked
by FeII (or by the fall­off in Br12). The simplest expla­
nation of this (see Tielens et al. 1993, van der Werf et al.
1996) is that one is observing an edge­on PDR and that
the observed offset corresponds to the difference between
the ionization front where Lyman continuum photons are
absorbed and the H 2 dissociation front where photons ca­

6 A. Marconi et al.: NIR Spectra of the Orion Bar
Fig. 3. H band spectrum.
pable of dissociating molecular hydrogen are absorbed. We
note also that an offset of 16 00 in the IRSPEC data corre­
sponds (given the orientation NE­SW of the bar) to 23 00
in the TIRGO slit oriented N­S.
3.2. HI lines
In the LONGSP data, we detect many recombination lines
of atomic hydrogen, 13 lines of the Brackett series (Brfl in
the K band and 12 lines, from (10­4) to (21­4) in the H
band), and Pafi and Pafl in the J band.
We use the 13 lines in the Brackett series to check the
accuracy of our data. Figure 10 plots the ratio (n­4)/(13­
4) as a function of the quantum number n and compares
our results with the prediction of recombination theory
(Storey & Hummer 1995). The agreement is quite good,
well within our estimate of 10­20% for the observational
errors.
The ratio of Pafi/Brfl provides a value of the extinc­
tion A V ¸2 mag, assuming the reddening curve of Cardelli
et al. (1989) and R=5.5. We detect a slight variation along
slit 1+2, from 2.3 mag in A to 1.4 mag in C. This vari­
ation, however, is within our estimated 25% error in this
line ratio. The value in the CS position is A V =1.6 mag.
Figure 10 shows as a dashed line the theoretical line ratios
corrected for A V = 2 mag.

A. Marconi et al.: NIR Spectra of the Orion Bar 7
Fig. 4. K band spectrum.
3.3. He lines
One of the aims of our observations was to examine the ex­
tent to which helium is neutral within the zone of ionized
gas. Estimates of the helium abundance based upon mea­
surements of either radio or optical recombination lines
often assume that the helium and hydrogen Str¨omgren
spheres are coincident with one another (see e.g. Mezger
1980). Our profiles along the slit allow us to make a direct
comparison of the HeI and HI line intensities which can
then in principle be transformed into the abundance ra­
tio [He + ]=[H + ] in the immediate vicinity of the ionization
front of the Bar.
The chief obstacle in doing this is the uncertainty in
helium line intensities which results from collisional ex­
citations from the metastable 2 3 S and 2 1 S states. Smits
(1996) has computed helium line intensities in an approx­
imation where collisions (and self­absorption) out of the
metastable levels into n=3 and 4 are neglected although
collisions between the n=2 levels are considered. We have
compared our observed intensities with Smits predictions
for electron density 10 4 cm \Gamma3 and temperature 10 4 K.
We normalise for this purpose to the 1.701¯m 4 3 D­3 3 P ffi
transition which has the same upper level as the 4471 š A
line often used in optical analyses. It is expected that this
transition (see Osterbrock et al. 1992) is only affected at
the 1­2 percent level by the collisional effects mentioned
above and we neglect such effects in the following. One
finds then that, relative to 4 3 D­3 3 P ffi , lines such as 5 3 D­
3 3 P are in good agreement with the Smits predictions but
4 1 S­3 1 P ffi and 5 1 S­3 1 P at 2.113 and 1.341¯m respectively
are factors of roughly 3 stronger. The reason for this may
be the neglect of the collisional effects and trapping dis­
cussed above (see e.g. Robbins & Bernat 1973; Peimbert
& Torres­Peimbert 1977). We in any case have assumed

8 A. Marconi et al.: NIR Spectra of the Orion Bar
Fig. 5. Variation with declination offset (relative to decli­
nation (1950) =\Gamma05 ffi 24 0 51 00 ) of line intensities measured in
the amalgamation of slit positions 1 and 2. The lines are H
(7­4) (Brfl; solid) , H 13­4 (dotted), and the He line at 1.701
¯m (dashed). The vertical scale is arbitrary. We note that re­
gions A, B, and C discussed in the text are defined as follows
: A, \Deltaffi ? 16 00 , B, \Gamma10 00 ! \Deltaffi ! 16 00 , C, \Deltaffi ! \Gamma10 00 .
that the 1.701 ¯m line behaves essentially as predicted by
the Smits models and can be used to estimate the He +
abundance. It is natural to compare the He 1.701¯m in­
tensity with that of the adjacent Br10 line. With the above
assumptions, we find that :
F (He; 1:7¯)=F (H;10 \Gamma 4) = 3:61 [He + ]
[H + ] (1)
In Fig. 11, we show the profiles along the slit of the
abundance ratio [He + ]=[H + ] deduced from Eq.(1) as well
as the ratio of the 2.06 (2 1 P ffi ­2 1 S) to the 1.7¯m lines. The
Brfl profile along the slit is shown for comparison. We
derive an abundance ratio of 0:093 \Sigma 0:005 over region A
consistent with He abundance estimates from other au­
thors (see e.g. Baldwin et al.1991 who find 0.088 based
on optical measurements and the review of Mezger 1980
(his Fig.5) who shows that radio estimates at positions
less than 100 00 from \Theta 1 C Ori are in the range 0.083­0.09).
It appears that at the position of our slit, the helium and
hydrogen Str¨omgren spheres are close to being coincident.
We note with interest the decrease of [He + ]=[H + ] to val­
ues of ¸ 0:075 in the southern part of the slit or effectively
in zone C, where we seem to observe different ionization
Fig. 6. Variation with declination offset of H (7­4) (dashed
line) and H2(1­0)S(1) (solid line), measured in the amalgama­
tion of slit positions 1 and 2. The regions defined as A, B, and
C are shown.
conditions in this (presumably) foreground ionized mate­
rial. One notes also the fact that the degree of ``enhance­
ment'' of the 2.06 micron line seems to increase slightly
(by a factor of 1.3) in zone C.
3.4. OI lines
We detect two OI lines in the J band, the 2p 3 3D­2p 3 3P at
1.129¯m and the 2p 3 4S­2p 3 3P at 1.317¯m, with compara­
ble intensity. These lines are produced in the neutral gas
by excitation of OI to the upper level of the transitions by
UV photons, at 1027 and 1040 š A respectively, followed by
radiative decay. The fact that the two lines have similar
intensity (though not at the CS peak, see Fig.2) suggests
that the contribution of Ly\Gammafi photons to the excitation
of the 2p 3 3D level (the upper level of the 1027 š A line) is
negligible. That the excitation mechanism is fluorescence
is confirmed by the spatial variation of the line intensities,
which peak at the edge of the ionized region as marked by
the HI recombination line emission (cf. Fig. 7). In other
words, the oxygen lines are an excellent marker of the ion­
ization front. More precisely, one can say that due to the
rapid charge exchange of OI with H + and the fact that
the hydrogen and oxygen ionization potentials are close
to being identical, the oxygen lines trace neutral gas close
to the ionization front.

A. Marconi et al.: NIR Spectra of the Orion Bar 9
Fig. 7. Top panel: variation with declination offset of H (7­4)
(dotted line), H2 1­0S(1) (dashed line) and FeII 1.644 ¯m(solid
line). Bottom panel: variation with declination offset of H (7­4)
(dotted line), H2 1­0S(1) (dashed line) and OI 1.317 ¯m(solid
line). In both cases, the variations are measured in the amal­
gamation of slit positions 1 and 2.
It is possible to use the OI lines to derive a measure of
the UV radiation field G 0 at the edge of the bar. If fluores­
cence is the dominant excitation mechanism, the number
of photons emitted in the IR line equals the number of
photons absorbed by the UV line. Since the optical depth
of the UV lines is always very large (Ü 0 ¸ 5 N=10 18 for the
1040 š A line and ¸ 10 N=10 18 fo the 1027 š A where N is the
atomic H column density and we have assumed a velocity
dispersion \Deltav=3 km s \Gamma1 and O/H=6\Theta10 \Gamma4 ), the number
of absorbed UV photons is proportional to the line equiv­
alent width in the ``flat'' portion of the curve of growth.
This is given by : W – =–UV ¸ 1:2\Deltav=c F (Ü 0 ) ¸ 3:6 \Theta 10 \Gamma5
(Spitzer 1978, p53, where Ü 0 is the UV line center optical
depth and F(Ü 0 ) ¸ 3 for optical depths of order 1000).
Since both UV lines are in fact triplets with separation
larger than W – , the total number of UV photons absorbed
and re­emitted in each IR line is given by:
I UV
š = 4ú sin ` t
3
– IR –UV
cW –
I(IR) (erg cm \Gamma2 s \Gamma1 Hz \Gamma1 ) (2)
where I(IR) is the observed intensity of the IR line in
erg cm \Gamma2 s \Gamma1 sr \Gamma1 and ` t is the angle between the PDR
and the line of sight (` t = 90 deg in a face­on PDR; see
Fig. 8. IRSPEC cuts: HI (12­4) (dashed line), H2 1­0S(1)
(dotted line) and FeII 1.644 ¯m(solid line).
Appendix). Note that the three components of the IR lines
are not resolved in our spectra.
Equation 2 gives I UV
š ¸ 1:2 \Theta 10 \Gamma13 sin ` t for I(IR) =
2:6 \Theta 10 \Gamma4 erg cm \Gamma2 s \Gamma1 sr \Gamma1 , as observed in the main
peak of the 1.317 ¯m line. Here, we have corrected for
2 magnitudes of visual extinction (see Sect. 3.2). The in­
ferred UV intensity can be compared to the flux from \Theta 1 C
Ori at the projected distance of the bar (I p ¸ 4 \Theta 10 \Gamma14
erg cm\Gamma2 s \Gamma1 Hz \Gamma1 for T ? =40000K, L ? =2.5\Theta10 5 L fi ).
The main uncertainty is the appropriate value for ` t but
the Orion Bar is known to be close to edge on (see e.g
Hogerheijde et al. 1995). A plausible value is sin ` t ¸ 0:2
(` t ¸ 10 \Gamma 15 deg; see Sect. 3.6), and one finds then that
the physical distance of the bar from \Theta 1 C Ori is about
(I p =I UV
š ) 0:5 or 1.3 times the projected distance.
The UV intensity can be expressed relative to the inter­
stellar diffuse field taken here to be 3 \Theta 10 7 photons cm \Gamma2
s \Gamma1 . The normalized UV intensity is then 1:3 \Theta 10 5 sin ` t
or G 0 ¸ 2:6 \Theta 10 4 (for sin ` t =0.2), similar to the value
used by Tielens et al. (1993).
The second, weaker peak of emission A at \Deltaffi ¸23
arcsec has an OI intensity about two times lower than the
main peak. This may be due to a different orientation of
the front with respect to the line of sight. The OI intensity
on the CS peak is about 1/4 that of the main peak which
again may be due to an orientation effect although it could
also imply dust extinction between the CS peak and the

10 A. Marconi et al.: NIR Spectra of the Orion Bar
Fig. 9. Variation with declination offset along the slit cen­
tered on the CS peak of selected lines: HI (7­4) (dashed line,
top panel), H2 1­0S(1) (solid line, top panel and dotted line,
bottom panel), and FeII 1.257 ¯m(solid line, bottom panel).
Trapezium. In general, our OI results imply G 0 values at
the ionization front in the range 6000 \Gamma 3 \Theta 10 4 .
3.5. Iron lines
Model calculations suggest (see Baldwin et al. 1991, Rubin
et al. 1991, Osterbrock et al. 1992) that iron is mainly in
the form FeIV in the Orion nebula. It follows that one ex­
pects to see FeII and FeIII emission predominantly at the
edge of the Str¨omgren sphere close to the ionization front.
In Fig. 7, we show profiles along our amalgamated slit of
the 1.644 ¯m [FeII] transition compared with the 1.317¯m
OI line discussed in the previous section. One sees that the
two lines show rather similar behavior with a peak slightly
offset from one of the maxima observed in the H recombi­
nation lines. It seems plausible that this approximate co­
incidence denotes the presence of an ionization front and
the OI data which we discussed above confirm this idea.
It is notable also that the iron lines show no evidence for a
coincidence with the molecular hydrogen 1­0 S(1) C peak
at \Deltaffi = ­27 00 which is also shown on Fig. 7. In fact, our
data suggest that both the FeII and FeIII emission lines
form in ionized (or partially ionized but not neutral) gas
close to the ionization front. It is worth noting here that
extinction estimates which we have made using the ratio
of the 1.644 4sa 4 D 7=2 ­3d 7 a 4 F 9=2 to the 1.26¯m 4sa 4 D 7=2 ­
Fig. 10. Ratio of the H­band hydrogen recombination lines
(n­4)/(13­4) as a function of the quantum number n. Filled
circles refer to position A, triangles to B, squares to C, and
diamonds to position CS (the CS peak). The line (18­4) is
blended with an [FeII] line. The solid line shows the predictions
of recombination theory (Case B). The dashed line shows the
Case B ratios corrected for a reddening of AV =2 mag.
4sa 6 D 9=2 transitions is consistent with a visual extinction
of 2:7 \Sigma 0:9 magnitudes at all positions.
We can estimate the conditions required to explain the
relative intensities of the FeII lines. FeII line ratios can
be used as indicators of electron density (see Oliva et al.
1990, Pradhan & Zhang 1993). Our most useful indicator
appears to be the ratio of the 1.600 4sa 4 D 3=2 ­3d 7 a 4 F 7=2
line intensity to that of the 1.644 4sa 4 D 7=2 ­3d 7 a 4 F 9=2 .
We find that this ratio varies in the range 0.06­0.1 (see
table 1) over the regions covered by our slit. Based on
the collisional rates of Pradhan & Zhang(1993) (see also
Oliva et al. 1990), we conclude that this corresponds to an
electron density n e of 4000 cm \Gamma3 in the region A), 6000
cm \Gamma3 in region C, and 3000 cm \Gamma3 in region B. At the CS
peak position, the observed 1.600/1.644 ratio is 0.12 cor­
responding to n e = 10 4 cm \Gamma3 . Thus, we can exclude high
density clumps with n e = 10 6 cm \Gamma3 of the type discussed
by Bautista et al.(1994). On the contrary, the FeII data
seem consistent with ionized gas of electron density ¸ 10 4
cm \Gamma3 or less in the vicinity of the ionization front. It is
worth stressing here that this line ratio converges to the
LTE value at electron densities of roughly 10 5 cm \Gamma3 and
so the lack of high density clumps is significant. More­

A. Marconi et al.: NIR Spectra of the Orion Bar 11
Fig. 11. The figure shows the He abundance [He + ]/[H + ] and
the ratio of the two HeI lines at 2.058 and 1.701 ¯m as a
function of the declination along the amalgamation of slits 1+2,
compared with the Brfl profile. The data have been smoothed
over three pixels.
over, the ionization degree in the layer where the FeII lines
are formed is unlikely to be much smaller than unity and
hence the hydrogen density is also likely to be of order 10 4
cm \Gamma3 .
3.6. H 2 lines
We measured the intensity of 8 H 2 lines in the K band, cov­
ering an excitation range from 6472 to 18089 K (see Table
1). Using the data from table 1 and transition probabilities
from Turner et al. (1977), we have determined upper level
column densities at positions A, B, C and CS. In Fig. 12,
we plot the column densities per sub--level derived in this
manner against excitation energy. One sees that there are
clear departures from an LTE distribution suggesting ei­
ther that fluorescence is playing a role in determining level
populations or that there is a sharp gradient of tempera­
ture along the lines of sight sampled. The ``best fit tem­
peratures'' derived from fitting a Boltzmann population
distribution to the data in Fig. 12 are moreover rather
high with values ranging from ¸2500 K in region C to
3000 K in region A.
The high excitation temperatures as well as the (prob­
able) detection of lines from levels as high as v=3 suggest
that we are detecting extended fluorescent emission (seen
Fig. 12. Column density per sub--level against excitation en­
ergy of the level at positions A, B, C and on the CS Peak. The
dotted lines show the best fit to a single temperature popula­
tion. The two v=3 transitions are doubtful and have not been
considered on the fit. Filled squares represent ortho transitions
and circles para transitions.
on larger scales by Luhman & Jaffe, 1996) in addition to
a ``thermal'' layer.
The evidence that some fluorescent emission is present
is strengthened by the behaviour of the intensity ratio
of the v=1­0 S(1) (2.12¯m) and v=2­1 S(1) (2.25 ¯m)
H 2 lines along our amalgamated slit (A,B,C), shown in
Fig. 13. In a pure fluorescent model, this ratio is predicted
to be ¸2, whereas an admixture of collisional excitation
leads to higher values (12 for pure thermal emission at
2000 K). Our results are consistent with those of van der
Werf et al. (1996) in that at the main molecular hydrogen
intensity peak C, the ratio is ¸ 6­7. A similar value is also
found at the secondary peak B, which is about coincident
with the position of the ionization front as traced by the

12 A. Marconi et al.: NIR Spectra of the Orion Bar
Fig. 13. Ratio of the 1­0S(1) to 2­1S(1) H2 lines as a func­
tion of the declination offset along our amalgamation of slits
1+2 (solid line). To compute the ratio, the profiles have been
smoothed over three pixels. The horizontal bar shows the value
averaged over declination offset 0­50 arcsec. We show for com­
parison the profile of the 1­0S(1) line (dotted line).
OI lines. Over the rest of the slit the ratio is ! ¸ 4. Thus the
fraction of fluorescent emission is smaller at the peaks of
v=1­0 S(1) emission.
These measurements, as well as the observed intensity
of the v=1­0 S(1) line, can be compared to the PDR model
calculations of Hollenbach & Natta (1995) (at steady
state). Figure 15 shows the predicted intensity of the 1­0
S(1) line in a face­on PDR as a function of the density
for different values of G 0 , and in Panel (2) the ratio of
the 1­0 S(1) to the 2­1 S(1) line. In the main molecular
peak C, we observe a peak intensity of the 1­0 S(1) line
of ¸ 2:3 \Theta 10 \Gamma4 erg cm \Gamma2 s \Gamma1 sr \Gamma1 , and a ratio of ¸7. If
we assume that the H 2 lines are produced in a PDR with
G 0 ¸ 3 \Theta 10 4 (see Sect. 3.3), this corresponds to a density
6 \Theta 10 4 cm \Gamma3 .
These results assume a face--on PDR and there is good
reason to suppose that the Orion Bar is seen edge on (see
Jansen et al. 1995b for a discussion of the geometry). The
effect of a slant of the Bar on the H 2 lines is complicated
by the effects of dust extinction and a brief discussion is
given in the appendix. The results are different for the vi­
brationally excited H 2 lines at 2¯m and for lines at longer
wavelengths, for which the extinction is negligible. Par­
Fig. 14. Same as Fig.13 for the CS position.
mar et al. (1991) have observed the J=3­1 (17¯m) and
4­2 (12.3¯m) v=0 H 2 lines toward the bar and find that
their observations are compatible with a hydrogen column
density of 3 \Theta 10 21 cm \Gamma2 at a temperature of order 500
K. Figure 15 shows in Panel (3) the model­predicted ra­
tio for the two lines observed by Parmar et al. and in
Panel (4) the ratio of the v=2­1 S(1) to the J=3­1 (17¯m)
lines. These ratios can be well reproduced by a model with
G 0 ¸ 3 \Theta 10 4 and n ¸ 6 \Theta 10 4 cm \Gamma3 having a moderate
enhancement of the intensity of the 12 and 17 ¯m lines
due to an inclination of the Bar with respect to the line
of sight ` t ¸10deg. These values are also consistent with
the inclination required to explain the intensity of the OI
lines, discussed in Sect. 3.4.
The density in the other regions where molecular lines
are measured can be estimated in a similar way from Fig.
15. In the secondary molecular peak B, assuming the same
inclination ` t ¸ 10 deg, we obtain n ¸ 4 \Theta 10 4 cm \Gamma3 .
These densities are a factor of ¸5­10 larger than that of
the ionized gas (see Sect. 3.5) and roughly consistent with
the density required to explain the stratification of the bar
(i.e the offsets between cold molecular gas and ionization
front, see Tielens et al. 1993), which is of order 5 \Theta 10 4
cm \Gamma3 .
Similar densities can be derived from our data at slit
position 3 , the ``CS peak'' (CS). We show the 1­0 S(1) in­
tensity in Fig. 9 and the v=1­0/2­1 line ratio in Fig. 14.
We see a peak of H 2 emission at \Deltaffi ¸ 49 00 with an intensity

A. Marconi et al.: NIR Spectra of the Orion Bar 13
Fig. 15. Panel (1): model predictions for the intensity of the H2 1­0 S(1) line as a function of hydrogen number density for
different values of G0 . Panel (2): ratio of the v=1­0 S(1) line at 2.12 ¯m to the v=2­1 S(1) line at 2.24 ¯m as a function of
density. Panel (3): ratio of the v=0,J=4­2 line at 12 ¯m to the v=0,J=3­1 at 17 ¯m. Panel (4): ratio of the v=1­0 S(1) line to
the v=0,J=4­2. Each curve is labelled by the logarithm of G0 . Triangles indicate G0 = 10 5 , squares with G0 = 10 4:5 , circles
with G0 = 10 4 , and crosses with G0 = 10 3 . The observed value at peak C is shown by the arrow in each panel.
of ¸ 1:6 \Theta 10 \Gamma4 erg cm \Gamma2 s \Gamma1 sr \Gamma1 , and a broader emission
between 0 00 and 30 00 , with a peak intensity of ¸ 8 \Theta 10 \Gamma5 erg
cm \Gamma2 s \Gamma1 sr \Gamma1 . In both cases the ratio of the two H 2 lines
is about 5 implying densities ! ¸ 10 5 cm \Gamma3 for G 0 ¸ 10 4 .
Neither of the peaks coincide with the ``ionization front''
as defined by FeII (Fig. 9). One would expect such a coin­
cidence if the densities were considerably above 10 5 cm \Gamma3
as implied by the CS data (see van der Werf et al.1996).
Thus the general conclusion is that the PDR as seen in
molecular hydrogen appears to be at densities below 10 5
cm \Gamma3 in contrast to the millimeter results which suggest
the existence of clumps at densities well above 10 5 cm \Gamma3 .
A caveat to much of the above discussion is that com­
parisons which we have made between the predictions of
the Hollenbach & Natta code used by us and other results
(in particular, the results of St¨orzer & Hollenbach 1997)
shows that there can be substantial differences in both the
H 2 line intensities and in some line ratios. The main rea­
son for this is the extreme sensitivity of H 2 line intensities
to the temperature structure although there are more mi­
nor effects which also play a role (note for example that
the Hollenbach Natta models neglect the effects of the car­
bon oxygen chemistry upon H 2 ). This has the consequence
that rather minor errors in the treatment of thermal equi­
librium can affect our conclusions. In this respect, mid
infrared intensity ratios such as the 12/17 micron ratio
measured by Parmar et al. are important because they
afford a direct measure of temperature. Thus if the tem­
perature structure can be adjusted to fit the mid infrared
observations, one may have some confidence in the predic­
tions for other lines. For the moment, we conclude that the
fact that we have been able to fit the 12/17 ratio suggests
that the model which we have used is reliable.

14 A. Marconi et al.: NIR Spectra of the Orion Bar
4. Discussion
Our analysis of molecular hydrogen in the previous sec­
tion has assumed implicitly that the H 2 emission comes
at all positions from material heated by the stellar radia­
tion field (PDR). The two peaks C and B we observe must
then come from two different structures. In fact, there are
indications in the molecular hydrogen images of van der
Werf et al. that in H 2 lines there are two bars. One of these
(which one might call the main H 2 bar) corresponds to the
feature seen in our Fig. 6 at position C (\Deltaffi = \Gamma24 00 ). The
``second H 2 bar'' (less well defined in the van der Werf et al.
images) is close in projection to the main ionization front
and corresponds to the peak in Fig. 6 at \Deltaffi =­2 00 (roughly
coincident with the OI peak). We propose that these two
bars are separated along the line of sight and that the shift
in position (15 00 accounting for our slit orientation) is due
to a tilt in the bar of ¸10 degrees as discussed above.
Thus, the bar is split along its length (essentially along
the line of sight) into two sections separated by 0.2­0.3
parsec. Each half of the bar in this scenario has a length
of order 0.1 parsec and thus the total length along the
line of sight is of order 0.3­0.4 parsec or somewhat smaller
than proposed by Jansen et al. ( 1995b).
There are problems with this model however. The prin­
cipal difficulty is that our data do not show evidence for
ionization front indicators (OI and FeII) roughly coinci­
dent (one expects a shift of order 6 arc second for a den­
sity of 5 10 4 cm \Gamma3 ) with the main (\Deltaffi = \Gamma24 00 on our slit)
molecular hydrogen peak. One explanation for this might
be that the density in the layer of gas between the ion­
ization and H 2 dissociation front is a factor of 2­3 lower
(of order 2 \Theta 10 4 cm \Gamma3 ) and thus that the main H 2 peak
is shifted 15 00 (20 00 in our NS oriented slit) relative to the
main ionization front (i.e., peak C in H 2 corresponds to
peak B in OI ). Equally, the ionization front emission seen
in Fig. 7 at \Deltaffi =23 00 might correspond to the H 2 emis­
sion at \Deltaffi = 3 00 . This lower density between ionization
and photo--dissociation fronts implies a density increase
between the atomic and molecular regions by at least a
factor of 4 since the general bar stratification requires an
average density of at least 5 \Theta 10 4 cm \Gamma3 in order to account
for the observed offsets of ionization front and molecular
lines (see also Wyrowski et al. 1997).
Thus, one possible interpretation (see also Simon et
al. 1997) of the H 2 data is that there is a sharp density
gradient perpendicular to the line of sight such that the
molecular gas has much higher density than the partially
ionized atomic medium adjacent to it. This has the at­
tractive feature that it helps explain one of the puzzles
concerning the bar which is the discrepancy (more than
an order of magnitude) between the hydrogen column den­
sity derived by Parmar et al. (1991) and that inferred by
Hogerheijde et al. (1995) from their C 18 O data. The Par­
mar et al. data refer essentially to the main H 2 bar whereas
Hogerheijde et al. preferentially sample the fully molecu­
lar gas to the SE. A density gradient may also cause the
offset between the ionization front and the main molecular
hydrogen peak to be larger than that estimated using a
homogeneous model whereas molecular hydrogen and car­
bon radio recombination lines would become closer to one
another.
This last aspect is particularly interesting in view of
the coincidence found by Wyrowski et al. (1997) between
the bar seen in C91ff emission and the main H 2 bar. The
difficulty in explaining this result stems from the fact that
one expects the H 2 emission to come from gas with tem­
perature above 2000 K while the carbon line is thought to
be formed at temperatures which are considerably lower.
In fact, there is a firm upper limit of 1600 K on the temper­
ature of the gas responsible for the C91ff emission (based
on the line width). The proposed density gradient dis­
cussed above may cause the offset between the photodis­
sociation front (i.e H 2 emission) and carbon line emission
to diminish.
A different interpretation of our observations is in prin­
ciple possible. Our main molecular hydrogen peak C at
\Deltaffi = \Gamma24 00 could be produced by a low velocity shock pre­
ceding the ionization front. This would explain the lack of
ionization front indicators coincident with molecular hy­
drogen. However, the kinematics of the emission seen in
C91ff by Wyrowski et al. are difficult to explain in this
scenario . One needs a shock velocity of at least 3 km s \Gamma1
to excite H 2 (see Tielens et al. 1993) and then the ob­
served C91ff line widths become difficult to understand.
It is intriguing that our H 2 observations do not show
any evidence of high density gas. This is in contrast with
the fact that the molecular line data (e.g. Tauber et al.
1995, Simon et al. 1997, van der Werf et al. 1996) give
evidence for a considerable fraction of the gas being in
clumps with density AE 10 5 cm \Gamma3 . Such clumps can be
expected to affect our results because most of the lines
observed by us are sensitive to high emission measure and
high density. High density PDRs are expected to be hot­
ter and hence considerably more intense in H 2 v=2­1 and
1­0 emission than lower­density PDRs (cf. Fig. 15). Nev­
ertheless, our data show no evidence for gas with density
above 10 5 cm \Gamma3 even towards regions where Simon et al.
(see also van der Werf et al.) estimate molecular hydro­
gen densities of order 2 10 5 cm \Gamma3 . We see no reason on
the other hand why clumps should dissipate on a short
timescale when traversing the H 2 dissociation front and
suggest therefore that clumps, while possibly present, are
a secondary phenomenon.
More important in our opinion is the density gradient
mentioned above. It is worth noting that in the scenario
which we are advocating, the thermal pressure may be
constant (at a value of the order of 10 8 cm \Gamma3 K) along a
line perpendicular to the bar and we conclude that isobaric
models of the Bar are worth examining. This incidentally
would suggest relatively low values for the magnetic pres­
sure and hence magnetic field (below 0.5 mG).

A. Marconi et al.: NIR Spectra of the Orion Bar 15
5. Conclusions
This study has presented NIR slit spectra of the Orion Bar
region. Our main result is based on the molecular hydro­
gen line intensities and is that the densities derived from
these tracers are of order 3 \Gamma 6 \Theta 10 4 cm \Gamma3 and thus consis­
tent with estimates of the mean density derived from the
observed stratification of the bar (e.g. Tielens et al. 1993,
Wyrowski et al. 1997). Comparison with the longer wave­
length H 2 data of Parmar et al.(1991) implies a tilt for
the bar relative to the line of sight of ¸10 degrees. Our
data suggest also that the bar may be split into two por­
tions along the line of sight which are separated by 0.2­
0.3 parsec. It seems plausible that the density is lower in
the ``atomic'' region (perhaps 2 \Theta 10 4 cm \Gamma3 ) than in the
molecular gas (of order 10 5 cm \Gamma3 ). We conclude that mod­
els with constant thermal pressure should be examined in
future studies of the bar region.
We have also derived densities for the ionized gas in
the vicinity of the ionization front using [FeII] line ratios
and find values of order 10 4 cm \Gamma3 . This together with the
molecular hydrogen data has convinced us that high den­
sity neutral clumps play a rather minor role in determining
the observed characteristics of the bar.
A by­product of these observations was that we were
able to use the observed OI line at 1.317¯m as an esti­
mator for the ultraviolet radiation field incident upon the
bar. We estimate the normalised FUV intensity G 0 on the
bar to be 0:6 \Gamma 3:0 \Theta 10 4 using this tracer.
Finally, we have used the 1.701¯m line of He to ex­
amine the degree of coincidence of helium and hydrogen
Str¨omgren spheres. To within our errors, we find that He +
and H + coexist and hence that He abundance estimates
using these tracers should be reliable.
Acknowledgements. We are indebted to D.P. Smits, who pro­
vided us with the results of his He level population calculations
and J.H. Black for making available to us his H2 transition
probabilities. Paul van der Werf allowed us to use his H2 im­
age of the Bar and Alan Moorwood gave us his IRSPEC data.
Special thanks are due to Tino Oliva, for his help and com­
ments on this projects. This work was partially supported by
ASI grant 94­RS­152 and GNA grant 96/00317 to the Osser­
vatorio di Arcetri. A.M. acknowledges partial support through
GO grant G005.44800 from Space Telescope Science Institute,
which is operated by the Association of Universities for Re­
search in Astronomy, Inc., under NASA contract NAS 5­26555.
A. The effects of inclination on the H 2 lines
The intensity of a given line as a function of the angle ` t
can be approximately estimated from the expression (cf.
Fig. 16):
I = I 0 =Ü 0 (1 \Gamma e \GammaÜ )e \GammaÜ n (A1)
where I 0 is the face­on intensity (` t = 90) ignoring ex­
tinction due to dust, Ü the dust optical depth at the line
Fig. 16. Sketch of the geometry for molecular hydrogen emis­
sion discussed in the Appendix. The symbols used in Eq.(A1)
are defined. The observer sees the PDR at an angle ` t . The
layer of H2 emission has continuum optical depth Ü0 and the
foreground atomic layer optical depth Ü0;n .
frequency across the layer of the PDR which contributes
to the H 2 emission, Ü n the optical depth of the HI re­
gion along the line of sight. Ü 0 and Ü 0;n are the optical
depths in the H 2 and HI regions, respectively, in the di­
rection perpendicular to the PDR. To first approximation,
Ü ¸ Ü 0 = sin ` t , Ü n ¸ Ü 0;n = sin ` t . For ` t ! ¸ l=L, the line of
sight does not intercept the neutral region of the PDR
(Ü n ¸0) and the line intensity tends to I=I 0 ¸ 1=Ü 0 . When
Ü 0 = Ü 0;n = 0, then I=I 0 = 1= sin ` t .
Figure 17 shows the ratio I=I 0 for the 1­0 S(1) line
(solid curve), which has been derived using Ü 0 ¸ 1 \Theta
A – =A v , Ü 0;n ¸ 1 \Theta A – =A v , A – =A v =0.145. Here we assume,
based upon models, 1 magnitude of visual extinction in
the H 2 emitting layer and 1 magnitude of extinction in
the foreground HI layer. For the 1­0 S(1) line, I=I 0 ¸ 1
with an accuracy better than 30% for ` t ? ¸ 8 deg . The
figure shows also the variation with ` t of the ratio of the
1­0 S(1) line at 2.12 ¯m to the v=0,J=3­1 17¯m line (for
which Ü 0 ¸ Ü 0;n ¸ 0), with respect to the same ratio for a
face­on PDR.
References
Baldwin J.A., Ferland, G.J., Martin, P.G. et al. 1991, ApJ 374,
580

16 A. Marconi et al.: NIR Spectra of the Orion Bar
Fig. 17. The solid line shows, as a function of the inclination
angle ` t , the intensity of the 1­0 S(1) line normalized to the
value in a face­on PDR where dust extinction is neglected. The
dashed line shows the ratio of the two lines 1­0 S(1) and v=0,
J=3­1 at 17 ¯m, normalized in an analogous faction.
Bautista M.A., Pradhan A.K., Osterbrock D.E. 1994, ApJ 432,
L135
Bautista M.A., Pogge R.W., DePoy D.L. 1995, ApJ 452, 685
Bertoldi F., Draine B. 1996, ApJ 458, 222
Cardelli J.A. , Clayton G.C., Mathis J.S. 1989, ApJ 345,245
DePoy D.L., Pogge R.W. 1994, ApJ 433, 725
Draine B.T., Bertoldi F. 1996, ApJ 468, 269
Dyson J.E., Williams R.J.R., Redman M.P., 1995, MNRAS
277,700
Felli M., Churchwell E., Wilson T.L., Taylor G.B. 1993, A&AS
98,137
Fuente A., Martin--Pintado J., Cernicharo J., Bachiller R.,
1993, A&A 276,473
Genzel R. 1992 , in ``The galactic Interstellar Medium'', p275,
(Burton W.B., Elmegreen B.G., Genzel R.), Springer
Genzel R., Stutzki J. 1989, ARA&A 27,41
Glassgold A.E., Huggins P.J., Schucking E.L. (Editors) 1982
Symposium on the Orion Nebula to honor Henry Draper,
Annals of the New York Academy of Sciences, Vol. 395
Hogerheijde M.R., Jansen D., van Dishoeck E.F. 1995, A&A
294, 792
Hollenbach D., Takahashi T., Tielens A. 1991, ApJ 377, 192
Hollenbach D., Natta A. 1995, ApJ 455,133
Hollenbach D., Tielens A. 1997, ARA&A (in press)
Hunt L.K., Lisi F., Testi L., et al., 1996, A&AS 115, 181
Hunt L.K., Migliorini S., Testi L., et al., 1997, AJ (submitted).
Hunt L.K., Testi L., Borelli S., Maiolino R., Moriondo G., 1994,
Technical Report 4/94, Arcetri Astrophysical Observatory
Jansen D.J., van Dishoeck E.F., Black J.H., Spaans M., Sosin
C. 1995a, A&A 302, 223
Jansen D.J., Spaans M., Hogerheijde M., van Dishoeck E.F.
1995b, A&A 303, 541
Lisi F., Hunt L.K., Baffa C., et al., 1996, PASP 108, 364
Lizano S., Canto J., Garay G., Hollenbach D. 1996, ApJ
468,739
Luhman M.L., Jaffe D.T., 1996, ApJ 463,191
Mathis J.S. 1995, Rev Mex AA (Serie de Conferencias) 3, 207
Mezger P.G. 1980 pp81­97 in Radio Recombination Lines,
P.A.Shaver (ed), publ. D.Reidel , Vol. 80, Astrophys. and
Space Science Library
Moorwood A.F.M., Monet A., Gredel R., 1991, The Messenger
63, 77
Oliva E., Moorwood A.F.M., Danziger I.J. 1990 A&A 240,453
Oliva E., Origlia L., 1992, A&A 254, 466
Osterbrock D.E., Shaw R.A., Veilleux S. 1990, ApJ 352, 561
Osterbrock D.E., Tran H.D., Veilleux S. 1992, ApJ 389, 305
Pankonin V., Walmsley C.M., Harwit M. 1979, A&A 75,34
Parmar P.S., Lacey J.H., Achtermann J.M. 1991, ApJ 372, L25
Peimbert M. 1982 in Symposium on the Orion Nebula, edited
Glassgold A.E. et al., Annals of New York Academy of Sci­
ences Vol 395, 24
Peimbert M. 1993 , Rev. Mexicana de Astron. y Astrof. 27,9
Peimbert M. 1995, p163 in The Analysis of Emission Lines,
edited R.E.Williams and M.Livio, Space Tel. Science Inst.
Series, 8
Peimbert M., Torres--Peimbert S. 1977, MNRAS 179, 217
Peimbert M., Torres--Peimbert S., Ruis M.T. 1992, RevMexAA
24,155
Pogge R.W., Owen J.M., Attwood B. 1992, ApJ 399, 147
Pradhan A.K., Zhang H.L. 1993, ApJ 409, L77
Robbins, R.R., Bernat, A.P. 1973, Mem.Soc.R.Sci. Liege 6th
ser., 5, 263
Rodriguez M. 1996 A&A 313,L5
Rubin R.H., Simpson J.P., Haas M.R., Erickson E.F. 1991,
PASP 103, 834
Rubin R.H., Dufour R.J., Walter D.K., 1993, ApJ 413,242
Simon R., Stutzki J., Sternberg A., Winnewisser G. 1997, A&A
(in press)
Simpson J.P., Rubin R.H., Erickson E.F., Haas M.R. 1986 ,
ApJ 311, 895
Smits D.P. 1996, MNRAS 278, 683
Spitzer L. 1978, Physical Processes in the Interstellar Medium,
publ. J.Wiley & Sons
Sternberg A., Dalgarno A. 1989, ApJ 338,197
Sternberg A., Dalgarno A. 1995 , ApJS 99, 565
Storey P.J., Hummer D. 1995, MNRAS 272,41
St¨orzer H., Hollenbach D. 1997, ApJ in press.
Tauber J.A., Tielens A.G.G.M., Meixner M., Goldsmith P.F.
1994, ApJ 422,136
Tauber J.A., Lis D.C., Keene J., Schilke P., B¨uttgenbach, T.H.
1995, A&A 297, 567
Tielens A., Hollenbach D. 1985, ApJ 291, 722
Tielens, A.A.G.M., Meixner, M.M., van der Werf, P.P., Breg­
man, J., Tauber, J.A., Stutzki,J., Rank, D. 1993, Science
262,86
Turner J., Kirby--Docken K., Dalgarno A. 1977 ApJS 35, 281

A. Marconi et al.: NIR Spectra of the Orion Bar 17
van der Werf P., Stutzki J., Sternberg A., Krabbe A. 1996,
A&A 313, 633
Vanzi L., Sozzi M., Marcucci G., et al. 1997a A&AS 124, 573
Vanzi L., Gennari S., Ciofini M., Testi L., 1997b, Experimental
Astronomy, submitted
Vanzi L., Marconi A., Gennari S., 1995, in ``New Developments
in Array Technology and Applications'', eds. A.G. Davis
Philip et al., p. 231
Walmsley M. 1997, in Proceedings if the 1 st Arecibo Workshop
on ``Molecular Spectroscopy with the upgraded Arecibo
telescope in the 1­10GHz range'', editors L.Olmi, W.Baan.
, in press
Wilson T.L., J¨ager B. 1987, A&A 184,291
Wilson T.L., Filges L. 1990 in Radio recombination lines ; 25
years of investigation, ed. M.A.Gordon, R.L.Sorochenko.
Wilson T.L., Filges L., Codella C., Reich W., Reich P. 1997,
A&A (in press)
Wyrowski F., Schilke P., Hofner P., Walmsley M. 1997, ApJ
487, L171
Yusuf--Zadeh F. 1990 ApJ 361, L19
This article was processed by the author using Springer­Verlag
L A T E X A&A style file L­AA version 3.