Документ взят из кэша поисковой машины. Адрес оригинального документа : http://star.arm.ac.uk/preprints/402.ps Дата изменения: Wed Sep 17 19:22:18 2003 Дата индексирования: Mon Oct 1 21:34:47 2012 Кодировка: Поисковые слова: m 63

Multi-wavelength spectroscopy of the bipolar out ow from
Cepheus E 1
Michael D. Smith
Armagh Observatory, College Hill, Armagh BT61 9DG, Northern Ireland
Dirk Froebrich, Jochen Eisloel
Thuringer Landessternwarte Tautenburg, Sternwarte 5, D-07778 Tautenburg, Germany
ABSTRACT
Cepheus E is the site of an exceptional example of a protostellar out ow with a very young
dynamical age and extremely high near infrared luminosity. We combine molecular spectroscopic
data from the submillimeter to the near infrared in order to interpret the rotational excita-
tion of CO and the ro-vibrational excitation of H 2 . We conclude that C-type shocks with a
paraboloidal bow shock geometry can simultaneously explain all the molecular excitations. A
dierence in the extinction between the red and blue-shifted out ow lobe accounts
for the measured ux dierence and the deviation of the column densities from local
thermodynamic equilibrium. The out ow is deeply embedded in a quite dense clump
(density greater than 10 5 cm 3 ), yet a good fraction of atomic hydrogen, about 40%, is
required to explain the excitation and statistical equilibrium. We propose that this atomic com-
ponent arises, self-consistently, from the dissociated gas at the apex of the leading bow shocks and
the relatively long molecule reformation time. At least 20 bow shocks are required in each lobe,
although these may be sub-divided into smaller bows and turbulent shocked regions. The total
out ow mechanical power and cooling amounts to over 30 L , almost half the source's bolometric
luminosity. Nevertheless, only about 6% of the clump mass has been set in outward motion by
the out ow, allowing a collapse to continue.
Subject headings: Shock waves { Molecular processes { ISM: jets and out ows { ISM: kinematics and
dynamics { ISM: molecules { stars: mass-loss
1. Introduction
Twin collimated lobes of molecular gas stream
away from newly forming stars (Bachiller 1996).
These bipolar out ows possess particularly high
power and thrust during the main phase of in ow
onto the protostar. The driving agents are often
recognised as pulsating supersonic jets, originat-
ing from near the protostellar surface. The ex-
tended environment is pushed and shocked, pro-
1 Based on observations with ISO, an ESA project with
instruments funded by ESA Member States (especially the
PI countries: France, Germany, the Netherlands and the
United Kingdom) and with the participation of ISAS and
NASA.
ducing bow-shaped structures called Herbig-Haro
(HH) objects (Reipurth & Bally 2001). The HH
objects are often optically invisible because the
protostar is deeply enshrouded in a dusty cloud.
We analyse here one such out ow, originating
from the Class 0 source CepheusE { MM (Le och
et al. 1996), through combined near-infrared data,
mid- and far-infrared (FIR) ISO spectra and sub-
millimeter maps. The numerous emission line
strengths at these long wavelengths constrain and
relate the gas components.
Concerning the high mass protostellar
envelope (Le och et al. 1996; Chini et al.
2001) and the source and out ow lumi-
1

nosity, Froebrich et al. (2003) conclude
that, according to an evolutionary model,
the out ow is powered by an intermedi-
ate mass young stellar object. This con-
trasts with other well-known text-book examples
of jet-driven Class 0 out ows, such as HH 211 and
HH 212, which are powered by low-mass or solar-
mass stars. Cepheus E is a powerful out ow from
a powerful source at a distance of  730pc. A CO-
derived kinetic luminosity of 0.2 L (Moro-Martn
et al. 2001), a FIR line luminosity of 2.8 L (Gi-
annini et al. 2001) and an H 2 ro-vibrational line
luminosity of 0.7 L (Froebrich et al. 2003) have
been derived (based on simple assumptions). The
driving protostar has a bolometric luminosity of 
80 L (Le och et al. 1996; Froebrich et al. 2003).
The out ow is apparently driven by jets or bul-
lets with radial velocities of -120 and +80km s 1
(Smith, Suttner, & Yorke 1997), as derived from
CO spectra (Le och et al. 1996; Hatchell et al.
1999). The proper motion of an optical knot at
the edge of the southern blue-shifted lobe, HH 377,
is 10714km s 1 (Noriega-Crespo and Garnavich
2001) with a radial velocity of -7010km s 1 (Ay-
ala et al. 2000). This implies that the jet is
more dense than the environment through which
it propagates. Reversing the well-known formula
for thrust balance yields a jet-ambient density ra-
tio of  = 1=(v jet =v bow 1) 2 = 2.0 (taking the
radial velocity components and assuming the gas
ahead of the bow shock is stationary). The inclina-
tion angle to the line-of-sight is  tan 1 (107=70)
= 57 ф 7 ф and the bow speed is 12812km s 1 .
We feature CepE in this study because of its
extreme youth and high luminosity. Several other
properties have attracted attention to this out ow.
 The low excitation. Both near-infrared H 2
and optical H lines yield remarkably low
excitation. Despite the strong emission
uxes stemming from vibrationally excited
H 2 states, the excitation temperature is low,
with its value sensitive to the measured ex-
citation levels (Eisloel et al. 1996). The
shock speed of HH 377 derived from H and
H uxes are under 20 km s 1 (Ayala et al.
2000).
 The constant molecular excitation. The
H 2 line ratios vary remarkably little over
the entire out ow on small and large scales
(Eisloel et al. 1996).
 The evidence for a precessing underlying
ow Eisloel et al. (1996)
We here analyse the radiative shock waves, si-
multaneously modelling the many strong emis-
sion lines from carbon monoxide and hydro-
gen molecules through both rotational and ro-
vibrational transitions. To accomplish this we
combine ISO and ground-based data. We thus
hope to go a stage further than previous stud-
ies of K-band spectra and ISO data, analysed in
isolation. Secondly, we model with bow shock dy-
namics and ambipolar diusion (C-shock) physics,
both of which have not yet been applied to recent
Cepheus E spectroscopic data (although Ladd &
Hodapp (1997) compared Smith's (1995) tabu-
lated C and J-type planar models to K-band spec-
tra).
Using refurbished shock codes, tested by Froe-
brich et al. (2002a) for the out ows from CepA
and L 1448, we model the out ow as planar and
bow shocks with a symmetric shape Z/R s (in
cylindrical coordinates). The main aim of this
process is to draw conclusions about the physics,
chemistry and dynamics of the shocks and the
properties of the surrounding gas. This analysis
will also provide hints concerning the early evolu-
tion of an out ow and possibly of the source itself.
In Sect. 2 we describe the observations and data
reduction. The main results are presented in
Sect. 3. The results of the extinction modelling
(Sect. 4), and modelling of the H 2 and CO line
uxes (Sect. 5) are then given.
2. Observations and Data Reduction
2.1. Near infrared data
The NIR spectra were obtained from 26{29 Au-
gust 1996 with the KSPEC spectrograph on the
UH 2.2-m telescope at the Mauna Kea Observa-
tory in Hawaii. This is a cross-dispersed Echelle
designed to provide medium-resolution spectra of
the 1 { 2.5 m region, optimized for 2.2 m. A
HAWAII 1024x 1024 detector array was used. We
observed at three dierent positions in the out ow
(as shown in Fig. 1). Data reduction, including
2

RA (J2000.0)
23: h 03: m 16: s 23: h 03: m 12: s 23: h 03: m 08: s
Dec
(J2000.0)
61:
ф
42
0
20
00
61:
ф
43
0
00
00
South
North
CepE LWS
........................................................................................... S1
........................................................................................... Na
........................................................................................... Nd
Fig. 1.| Positions of the apertures for LWS ob-
servations (circles) in Cepheus E superimposed on
an image in the 1 { 0 S(1) line of molecular hydro-
gen at 2.122 m. The slit positions of the three
KSPEC spectra are also indicated.
atelding, sky-subtraction and extraction of the
spectra, was done using our own MIDAS routines.
Absolute ux calibration was not possible due to
non-photometric weather conditions. The relative
uxes at wavelengths larger than  >
 2:4 m (1 {
0 Q() lines) are unreliable because the ateld was
quite poor. Since the H 2 emission lines were de-
tected only in the H- and K-band, the wavelength
calibration had to be performed only in these two
bands. Therefore we employed the OH-night-sky
emission lines and the tables of Rousselot et al.
(2000).
2.2. ISO data
We use three sets of data acquired with the
ISO satellite (Kessler et al. 1996). CepE was ob-
served at two dierent positions (the apertures
are indicated as circles in Fig. 1) with the Long
Wavelength Spectrometer (Cesarsky et al. 1996)
in LWS 01 grating mode. In this mode a spectrum
from 43 m to 197 m was obtained with a spec-
Table 1: Observation log for Cep. E ISO data.
TDT Object  (J2000) ф (J2000) AOT
56600912 CepE S 23 03 13 +61 41 56 LWS01
56601113 CepE N 23 03 13 +61 42 59 LWS01
79200740 CepE 23 03 13 +61 42 27 CAM04
tral resolution of about 200. (See the ISO Hand-
book, Volume IV: LWS|The Long Wavelength
Spectrometer 2 and Cesarsky et al. (1996) for in-
strument and Astronomical Observing Template
details).
In addition, the region was imaged with the
ISOCAM instrument (Cesarsky et al. 1996) and
the CVF-lters in a wavelength range from 5 m
to 17 m and a pixel size of six arcseconds. We re-
fer to the work of Moro-Martn et al. (2001), where
these data were rst presented. The observed eld
of the observations is larger than the NIR image
shown in Fig. 1. (See the ISO Handbook, Volume
III: CAM { The ISO Camera 3 and Cesarsky et al.
(1996) for instrument and AOT details.) A log of
the ISO observations used here is given in Table 1.
Data reduction was performed using the pro-
vided ISO software (ISAP 1.6a and LIA 7.3 for
the LWS data and CIA for the ISOCAM-CVF
data), and the data from standard pipeline 8.7
(LWS) and 10 (ISOCAM-CVF). In the LWS spec-
tra glitches due to cosmic ray hits are removed
as well as the heavy fringing which occurred in
the spectra. Flux measurements of the lines were
done using a Gaussian ts to the lines plus sec-
ond order polynomials for the background. Lines
with a FWHM signicantly smaller than the in-
strumental prole are not considered as real, and
are excluded from further analysis.
2.3. sub-mm data
We took submillimeter data of the CepE out-
ow in the 12 CO(3 { 2), line with the JCMT at
Manua Kea in Hawaii in the period 18 { 21 June,
1997. We employed the B3 and C heterodyne re-
ceiver. Data reduction was done using the SPECX
software. A spectrum was taken on each point of
a 7x11 grid with a grid point separation of six
2 http://www.iso.vilspa.esa.es/manuals/HANDBOOK/IV/lws hb/
3 http://www.iso.vilspa.esa.es/manuals/HANDBOOK/III/cam hb/
3

arcseconds. We also obtained some data for the
lines 12 CO(4 { 3), CS(7 { 6) and SiO(8 { 7) which
we brie y describe below.
3. Results
3.1. Near infrared data
Almost 30 ro-vibrational transitions of molec-
ular hydrogen were detected in the H and K bands.
These include the 1 { 0 S branch up to the 1 {
0 S(9), the 1 { 0 Q branch up to 1 { 0 Q(8), the
2 { 1 S branch and even a few 3 { 2 S branch lines.
Some lines were not detected because they are sit-
uated in atmospheric absorption bands. Table 2
provides the complete list of all relative uxes de-
tected with KSPEC. The line ratios are quite sim-
ilar in all three positions. As we will nd, the K-
band extinction is quite high and might also have
signicant in uence on the measured excitation of
the molecular hydrogen.
There are now 4 independent determinations
of the ux ratio R21 = I(2{1 S(1))/I(1{0 S(1)),
at various locations. For location Na, we nd
R21 = 0.12, in agreement with Ayala et al. (2000)
(their North location). We estimate R21  0.1
from Ladd & Hodapp (1997), while Eisloel et al.
(1996) found R21 = 0.0780.005 at location ND
(roughly a 2 00 box). Hence the very low excitation
reported in the latter work is not conrmed for
this or the other locations. For the location in the
southern lobe, S1, we nd R21 = 0.13, while Ayala
et al. (2000) nd R21 = 0.11. This is consistent
with typical 10% errors in the 2{1 measurements.
A subset of the NIR emission lines detected here
were also measured by Ayala et al. (2000) for two
positions, Na and S1, close to ours. This pro-
vides an indirect means of calibrating our data.
The main problem, however, is to determine the
total ux of the K-band lines within the ISO
apertures in order to directly compare and model
the emission levels. The integrated 1{0 S(1)
lobe uxes are 1.4  10 15 W m 2 (North) and
2.7  10 15 W m 2 (South) (Eisloel et al. 1996).
The ux ratio is reasonably close to the value of
0.44 measured at the location of the peak uxes
by Ayala et al. (2000).
Table 2: Relative uxes of H 2 lines in the H- and K-
band measured with KSPEC in the three observed
slits (see Fig. 1). The uxes are normalised to the
1 { 0 S(1) line. The errors are about 10% and 25 %
(for  >
 2:4 m).
Transition  0 CepE CepE CepE
[m] Na Nd S1
1 { 0 S(9) 1.6873 0.025 0.014 0.035
1 { 0 S(8) 1.7143 < 0.015 0.012 0.026
1 { 0 S(7) 1.7475 0.097 0.093 0.144
1 { 0 S(6) 1.7876 0.063 0.061 0.090
1 { 0 S(5) 1.8353 0.261 0.293 0.322
1 { 0 S(4) 1.8914 0.099 0.109 0.351
1 { 0 S(2) 2.0332 0.328 0.323 0.349
1 { 0 S(1) 2.1213 1.000 1.000 1.000
1 { 0 S(0) 2.2226 0.339 0.309 0.281
1 { 0 Q(1) 2.4059 1.101 1.010 1.106
1 { 0 Q(2) 2.4128 0.417 0.379 0.341
1 { 0 Q(3) 2.4231 1.301 1.160 1.215
1 { 0 Q(4) 2.4368 0.367 0.323 0.340
1 { 0 Q(5) 2.4541 0.584 0.648 0.779
1 { 0 Q(6) 2.4749 0.276 0.164 0.209
1 { 0 Q(7) 2.4993 0.548 0.430 0.464
1 { 0 Q(8) 2.5270 < 0.044 0.081 0.136
2 { 1 S(7) 1.8523 < 0.033 < 0.010 0.051
2 { 1 S(4) 2.0035 < 0.016 0.015 0.040
2 { 1 S(3) 2.0729 0.084 0.085 0.095
2 { 1 S(2) 2.1536 0.043 0.041 0.048
2 { 1 S(1) 2.2471 0.108 0.103 0.117
2 { 1 S(0) 2.3550 0.019 0.016 0.027
3 { 2 S(6) 2.0130 < 0.011 < 0.010 < 0.007
3 { 2 S(5) 2.0650 0.016 < 0.010 0.015
3 { 2 S(4) 2.1274 < 0.016 < 0.010 < 0.008
3 { 2 S(3) 2.2008 < 0.010 0.023 0.025
3 { 2 S(2) 2.2864 < 0.010 0.009 0.015
3 { 2 S(1) 2.3858 < 0.010 0.013 0.027
3.2. ISO LWS data
The ISO-LWS spectra of the two out ow lobes
show a wide variety of atomic ne structure and
molecular lines. We detected rotational CO tran-
sitions from J up = 14{ 21 and some water lines in
both lobes. The CO emission is on average 1.5
times stronger in the southern out ow lobe, sug-
gesting a higher lling factor of the beam or a
higher CO abundance. The same applies for the
H 2 O lines. All detected lines and uxes in the
LWS spectra are listed in Table 3. The LWS spec-
tra for both out ow lobes can be found in Fig. 2 of
Moro-Martn et al. (2001). We nd no particular
anomalies with their derived uxes.
4

Table 3: Observed lines in Cepheus E North and
South. The H 2 uxes measured in the ISOCAM
data are co-added in the LWS apertures. The LWS
uxes are in 10 16 Wm 2 , the ISOCAM H 2 uxes
are in 10 15 Wm 2 .
Element Transition  0 CepE Cep E
[m] North South
H2 0{0 S(7) 5.510 4.20.8 3.40.7
H2 0{0 S(6) 6.107 2.20.4 2.00.4
H 2 0{0 S(5) 6.908 8.31.7 5.31.1
H 2 0{0 S(4) 8.023 2.80.6 3.60.7
H 2 0{0 S(3) 9.662 6.11.2 6.61.3
H2 0{0 S(2) 12.275 4.60.9 5.81.2
[OI] 3 P1 { 3 P2 63.184 93.45.0 1296.5
o{H2O 321 { 212 75.380 17.73.4 <10.0
o{H2O 505 { 414 99.492 11.21.0 <10.0
CO 25{24 104.445 <10.0
o{H 2 O 2 21 { 1 10 108.073
CO 24{23 108.763
o
<15.0
o
<16.0
CO 23{22 113.458
o{H 2 O 4 14 { 3 03 113.537
o
<22.4
o
<12.9
CO 22{21 118.581 <10.0 <13.2
CO 21{20 124.193 4.42.0 <13.4
p{H 2 O 4 04 { 3 13 125.353 <8.0
p{H2O 331 { 322 126.713 <6.0 <6.5
CO 20{19 130.369 6.51.1 10.31.8
CO 19{18 137.196 9.21.0 13.32.5
p{H2O 313 { 202 138.527 <2.5 5.93.3
CO 18{17 144.784 12.31.5 16.81.6
[OI] 3 P0 { 3 P1 145.525 <2.0 <3.0
CO 17{16 153.267 11.92.2 19.81.0
[CII] 2 P 3=2 { 2 P 1=2 157.741 72.41.1 86.12.0
CO 16{15 162.812 15.71.2 20.72.2
CO 15{14 173.631 19.93.0 30.95.8
o{H2O 303 { 212 174.626 8.32.1 14.00.9
o{H 2 O 2 12 { 1 01 179.527 26.32.6 29.12.2
CO 14{13 185.999 23.83.2 28.93.5
3.3. ISOCAM CVF data
Pure rotational lines of the ground vibrational
level of H 2 were detected (0 { 0 S(2) .. S(7)) with
ISOCAM. Images in these lines as well as spectra
of selected pixels are presented in Moro-Martn et
al. (2001). To achieve our aim, to simultaneously
model the CO and H 2 lines, we co-add the spectra
of all pixels of the ICOCAM images which are sit-
uated in the LWS beam. Thus we compare uxes
measured in the same aperture for both species,
CO and H 2 . The co-added uxes of the 0 { 0 S
lines in the two LWS apertures are given in Ta-
ble 3.
Fig. 2.| Gray-scale and contour map of the in-
tegrated ux in the CO 3 { 2 line of the Cepheus E
out ow.
3.4. sub-mm data
In the obtained CO(3 { 2) map (see Fig. 2)
one clearly can identify the two out ow lobes.
Note the S-shaped structure within the CO(3 { 2)
greyscale image. At the brightest points, the line
shape shows signs of being optically thick (self ab-
sorption and re-emission). Also detected are high
velocity bullets with v rel  100 km s 1 . In the
CS(7 { 6) map we found weak emission only near
the source position and extremely weak emission
at the three brightest knots of the out ow (S1, Na
and Nd). Nevertheless, the out ow structure is
visible in this line. At the three positions where
the CO(4 { 3) line is observed we see the same
line structure as for the CO(3 { 2) line: optically
thick in the line centre and very fast CO bullets
(v rel  100km s 1 ). The northern positions ob-
served in SiO(8 { 7) show no sign of a line. Only
at the two positions in the south very weak
emission is detected.
5

4. Extinction
4.1. Methods
To determine the luminosity and excitation of
a deeply-embedded out ow it is necessary to re-
move the extinction. Extinction reduces observed
uxes from their intrinsic values, especially for
those from the shorter wavelength KSPEC data.
Hence, in order to test shock models, we rst ad-
just the KSPEC uxes.
Since the CepE out ow is embedded in the
parental cloud of the source (only the southern-
most part is visible at optical wavelengths as
HH 377), the extinction is high. Moreover, the ex-
tinction may vary not only over the eld of the
LWS aperture but also through a lobe. That is,
the K-band data will be more representative of
emission arising from low-excitation regions while
ISO and CO data will sample more evenly. Here,
however, since we do not have an extinction map
and cannot distinguish components of contrasting
extinction, we determine the extinction for each
KSPEC location and then apply one value per
LWS and ISOCAM beam.
Three means are at our disposal to estimate
the extinction. First, 1.25 mm continuum emis-
sion from cool dust is spatially coincident with the
out ow (Le och et al. 1996). This can be inter-
preted as an H column of 6  10 22 cm in the lobe
locations, corresponding to 3{4 mag of K-band ex-
tinction (incorrectly converted by Le och et al.
(1996)). If half of this emission lies on average in
front of the lobe, an extinction of 1.5{2 mag would
result.
Second, the Q-branch 1{0 lines beyond 2.4m
arise from the same upper energy levels as the S-
branch 1{0 lines within the K-band. Hence, dier-
ential extinction between these wavelengths can,
in theory, be determined exactly since the number
of emitted photons must be proportional to their
Einstein coeфcients (radiative decay rates). Ab-
solute extinction can then be derived from the dif-
ferential value provided the properties of the dust
are known.
The extinction is estimated through the NIR
KSPEC spectra. A dierential extinction is deter-
mined here as dlog(A  ) = 0.4 AK ((2.12m)/) 1:7 -
1.0) Using just the 1 { 0 S(1) and 1 { 0 Q(3) lines
(both originating from the same upper energy
level), we determine K-band extinctions of AK 
2.6{3.2 mag for the three KSPEC slit positions.
In contrast, the Q(2)/S(0) ratio yields a uniform
extinction of just 1.10.1 mag.
The Q-branch lines are, however, greatly ef-
fected by transmission and are often inaccurate.
One good test for accuracy is to inspect the ra-
tio of the Q(3)/Q(1) lines which should be within
the range [0.91,1.12] if lter or transmission dis-
tortions are low and if the gas is excited within
the range [1500K,3000K] (Smith 1995). This was
not case for the previously derived values in Ladd
& Hodapp (1997) and Moro-Martn et al. (2001).
Moro-Martn et al. (2001) nd values of 0.58 and
0.86 for locations in the two lobes and Ladd &
Hodapp (1997) nd 0.78 and 0.88. Dierential ex-
tinction would only exacerbates the inconsistency.
Here, we derive Q-branch Q(3)/Q(1) ratios of 1.18,
1.16 and 1.10 for the 3 slits, which are comfort-
ably consistent when some dierential extinction
is taken into account. Nevertheless, even in this
case, extinctions derived from the S(1)/Q(3) and
S(0)/Q(2) ratios do not agree.
Here, we plot the columns from all the levels
across the H and K bands and then determine the
extinction range which best correlates the accu-
mulated data. To achieve this result, however,
one must apply the normalised Column Density
Ratio (CDR) method. Plotting the derived ab-
solute columns of H 2 , N j against the upper en-
ergy level, E j hides the information in the data
since columns are spread out over three orders of
magnitude while error bars are only 10% on many
data points. In the CDR method, the columns
are normalised to the 1{0 S(1) ux and divided
by the equivalent LTE column of a gas at 2000K
(see Froebrich et al. (2002a) and Eisloel, Smith,
& Davis (2000) for details). The results are shown
in Fig. 3.
Fig. 3 conrms that high extinction is present.
Extinction is responsible for the distribution of the
data points along the edges of a rhombus (top pan-
els), which collapses to a line on applying the ex-
tinctions indicated. The four sides of the rhom-
bus consist of (i) the K-band 1{0 S-branch, (ii)
1{0 Q-branch, (iii) 2{1 S-branch and (iv) H-band
1{0 S-branch. We note no evidence for UV excita-
tion and uorescent emission (as usually identi-
able from rotational transitions within the higher
vibrational levels). Hence, the CDRs from the 2{1
6

Fig. 3.| Column density ratios for the near infrared H 2 ro-vibrational data, before applying extinction (top
panels) and after adjustment for extinction (lower panels). The slit locations and extinction are indicated on
each panel. Upper bounds are displayed as vertical lines. Squares represent 1{0 lines, with lighter squares
denoting the less-reliable columns derived from the Q-branch, diamonds indicated 2{1, and triangles indicate
CDRs from 3{2 lines.
lines should lie level or below those from the 1{0
lines. However, we also require that the Q-branch
lines are not too far below the S-branch lines. In
this manner, a good approximation for extinction
is reached in each case.
4.2. Extinction removed
The higher K-band extinction in the northern
lobe by  0.7 mag. implies that the northern lobe
is not intrinsically weaker than the southern lobe
in molecular hydrogen emission. The north-south
1{0 S(1) ratio was found by Eisloel et al. (1996)
to be 0.54. This suggests that the out ow may
be much more symmetric than appears. Since the
northern lobe is redshifted, this is also consistent
with the lobe dynamics within a spherical cloud.
Noting the out ow size of 0.39 pc and an av-
erage column to the out ow of  7  10 22 cm 2 ,
yields a mean density estimate of 10 5 cm 3 .
The extinction also implies that the intrinsic H 2
luminosity from the 1{0 S(1) line alone is 0.33 L ,
and the total H 2 luminosity will probably lie in the
range 3{9 L , depending on the type and strength
of shock waves involved (Smith 1995). Notably,
this total is very close to that estimated for the
complete far-infrared cooling by Giannini et al.
(2001) of 2.8 L , which includes CO, H 2 O, O and
OH emission.
5. H 2 and CO Modelling
To simultaneously model the KSPEC, ISO-
CAM and LWS CO data, we require a calibration
of the KSPEC data set. We employ the integrated
1{0 S(1) K-band uxes of 1.45 10 15 W m 2
(North) and 2.68 10 15 W m 2 (South) tabulated
by Eisloel et al. (1996) since the associated areas
correspond quite closely to the eective apertures
of the ISO observations.
7

We have attempted to apply simple planar J-
shock and C-shock models, without success. We
have employed an updated shock code described
in detail by Smith et al. (2003). The code as-
sumes that a shock is stable and in a steady state,
the ion number is a conserved quantity and that
the H 2 dissociation rate is given by equilibrium
conditions. We take an ortho-para ratio for H 2 of
3. Single J-type shock waves predict high excita-
tion H 2 spectra and C-type shocks predict quite
constant excitation across a wide range of upper
energy levels.
Multiple shock waves are required, as demon-
strated in Fig. 4. Two carefully chosen C-type
shocks provide a reasonable t to the full set of
data. The top panel displays both the ISO H 2
data (stars at low T j ) and the KSPEC data. The
lower panel displays the CO rotational uxes. It
is not clear, however, how two shocks with almost
the same parameters can be found in the three
separate locations since the excitation produced
in planar shocks is very sensitive to the shock ve-
locity, eld strength and ionization fraction.
Next, we t a single paraboloidal C-type bow
shock. C-type and J-type bow shocks are modeled
according to a scheme illustrated in Fig. 5. Cylin-
drical coordinates (z,R,) are used, and the mag-
netic eld B, is dened via the density and Alfven
speed. The spectroscopic results presented here
are independent of the direction of the observer,
. The molecules are completely dissociated at the
bow cap provided the bow is moving faster than
the appropriate dissociation speed limit. The data
provides a large number of constraints. For exam-
ple, the CO uxes and excitation determine not
only the CO abundance but also the density. The
excitation state of the vibrationally excited H 2
also determines the density as well as the shape of
the bow. The rotationally-excited H 2 determines
the atomic hydrogen fraction.
The nal results are displayed in Fig. 6 for Cep
E South and Fig. 7 for Cep E North (location Nd
for the KSPEC data). Note that bow congura-
tions are strongly supported by two independent
observations. Firstly, the H 2 images display nu-
merous bow-shaped structures in both lobes (e.g.
Fig. 1 and Ladd & Hodapp (1997) and, secondly,
many locations within these 1{0 S(1) bows pos-
sess double-peaked line proles, as predicted by
bow shock models (Eisloel 1997) and numerical
simulations Suttner, Smith, Yorke, & Zinnecker
(1997).
The displayed CBOW model ts are further ev-
idence that these are bow shocks. Critical param-
eters which are determined during the modelling
procedure are (i) the density, (ii) the molecular
fraction, (iii) the bow shape and (iv) the CO abun-
dance. Non-critical parameters to the gas excita-
tion state are (i) the bow speed, (ii) the Alfven
speed (i.e. the magnetic eld), (iii) the ion frac-
tion and (iv) the oxygen abundance. The latter
parameters in uence the location of the molecular
emission along the bow surface but, provided the
location exists and is not at the apex, the exci-
tation is not strongly aected. Perhaps the most
signicant result is the atomic hydrogen fraction of
0.4. Additionally, the only dierence between the
models is the derived CO abundances, 1.2 10 4 at
Nd, 1.2 10 4 at Nd and 1.5 10 4 at S1. Given the
large and slightly overlapping LWS beams, this is
not a signicant dierence.
Finally, we consider the type of J-type bow
shock that could t the spectroscopic data. Since
J-shocks heat the gas impulsively, higher vibra-
tional levels are relatively well populated. Hence,
to t the data, a bow with longer cooled wings
than a paraboloid is needed to compensate. After
varying the critical parameters, Fig. 8 shows that
a bow with shape s = 1.5 where z/L = (1/s) _ (R/L) s
ts the data extremely well.
A major reason that quite blunt shapes are pre-
dicted here, in comparison to those derived for
Cepheus A, is that lower densities are involved.
A lower density is deduced from the relatively low
CO uxes in Cepheus E. The lower density then
necessitates a higher fraction of hydrogen atoms
to maintain a small divergence from local thermo-
dynamic equilibrium between the populations of
the H 2 vibrational levels.
6. Further diagnostics
To distinguish between J and C-type bow mod-
els, we examine other signatures. We present the
contributions to the cooling, calculated from the
two best tting models to Cep E S1, in Table 4.
The luminosities are calculated by assuming a suf-
cient number of bow shocks to explain the H 2
1{0 S(1) luminosity. Indeed, a reasonable number
of distinct bow shocks are inferred. We nd from
8

Table 4: Contributions to the intrinsic cooling of
the ow in Cep E South and the bow shock pre-
dictions. The total cooling estimates are extracted
from Giannini et al. (2001).
Coolant Observed C-Bow J-Bow
L L L
H 2 1 { 0 S(1) 0.33 0.33 0.33
H 2 dissociative { 0.11 0.92
H2 rot+vibrat. { 6.77 8.32
CO rote > 0.51 4.76 56.56
CO rote (J > 10) 0.51 1.26 1.38
CO vibe { 0.01 0.03
H 2 O > 0.67 1.53 3.71
H2O (FIR, T>800K) 0.67 0.54 1.39
OH 0.17 0.01 0.01
O I 63m > 0.21 1.72 21.36
C I 370m { 0 0.17
the above model that a single C-type bow of speed
v bow = 120 km s 1 and scale size L = 10 15 cm
emits 0.016 L in the H 2 1{0 S(1) line. There-
fore, 20 such bows in each lobe would suфce to
provide the observed luminosity. Note that these
bows would appear larger than L in size, with H 2
being thermally dissociated until the normal com-
ponent of velocity falls below 45 km s 1 . For a
paraboloid, it is straightforward to show that this
occurs at R = 2.47 L and z = 3.05 L.
A single J-type bow with s = 1.5, v bow =
60 km s 1 and scale size L = 10 15 cm emits
0.02 L . Hence, 16 such bows would be needed to
account for the H 2 emission. The apparent bow
size is, however, larger. With a dissociation speed
limit of 25 km s 1 , we nd molecules survive only
for bow locations at R > 4.76 L and z > 4.36 L.
The [O I] 63m line is potentially a good shock
diagnostic. As shown in Table 1, however, this line
ux is overpredicted by factors of a few. Given gas
columns of 4{10 10 22 cm 2 yet an optical depth
at line centre of just 5 10 20 cm 2 (Hollenbach &
McKee 1989), the observed [OI] emission may well
not have a shock origin. Furthermore, we also de-
tected the [CII](158 m) line in the LWS spectrum.
The measured ratio [OI](63 m)/[CII](158 m) is
about 1.3 (1.5) for the north and south lobes, re-
spectively. In shocks, the [CII](158 m) line is usu-
ally several orders of magnitute fainter than the
[OI](63 m) Hollenbach & McKee (1989). hence
these lines may have a PDR origin, as already
discussed by Moro-Martn et al. (2001). The ob-
served OH luminosity is, according to Giannini et
al. (2001), to be treated with caution since it may
well be due to continuum pumping.
We conclude from Table 4 that the long wings
of the J-type bow produce extremely high uxes
in the lines from the cooler molecular gas. This is
clearly inconsistent with all expectations, with the
total cooling then exceeding the bolometric lumi-
nosity of the protostar. Furthermore, in a shock-
dissipative momentum-driven out ow, we expect
the mechanical power of the ow to be very close
to the total cooling. We estimate here the mechan-
ical power through CO J=2{1 observations by rst
noting that most of the emission is from low-speed
CO but most of the kinetic energy lies in the CO
moving with radial speeds of 15{25 km s 1 (Smith,
Suttner, & Yorke 1997) (a result of the very at
line prole found for Cepheus E). Hence, given a
blue lobe mass of 0.16 M Ladd & Hodapp (1997),
we derive a mechanical energy of  5  10 45 erg
and a mechanical power  20 L . Although this
estimate is subject to considerable uncertainty, it
is consistent with the C-type bow model.
A further diagnostic is provided by the inte-
grated intensity of the CO(3 { 2) emission shown
in Fig. 2. The C-bow model predicts a CO(3 {
2)/CO(18{ 17) ratio of 6.0 in contrast to the J-
bow model (276) for the southern out ow lobe.
The measured ratio is 4 (north) and 2 (south),
much nearer to the predictions of the C-type bow
model.
The columns in the 0-0 S(3) and 0-0 S(5) ap-
pear lower than predicted in the southern lobe.
This suggest that the ortho-to-para ratio of H 2
may be under three, the LTE value for high tem-
peratures. This can occur in C-shocks since the
gas is gradually heated from a low pre-shock tem-
perature. Therefore, the gas maintains the pre-
shock ortho-para ratio. After some heating, how-
ever, atomic hydrogen can induce ortho-para con-
version (Smith, Davis, & Lioure 1997).
Note that the distribution of shocks generated
by a supersonic turbulent velocity eld can pro-
duce an excitation that mimics that from a bow
shock surface (Eisloel, Smith, & Davis 2000). It
can be shown that a power law number distribu-
tion of shock speeds (with power law index = -
2s/(s-1) where N / v  ) is required. This also
assumes that there is no systematic magnetic eld
eects although, for bow shocks with moderate
9

Alfven speeds, we nd that the magnetic eld di-
rection has an insignicant in uence on the exci-
tation.
7. Conclusions
We have analysed the molecular out ow from
the Cepheus E-MM source over a broad wave-
length range, from the near infrared into the
sub-millimeter regime. We combined KSPEC
NIR data, ISO mid- and far-infrared spectra and
JCMT sub-mm observations. We have concen-
trated on the properties of the vibrationally and
rotationally excited H 2 and CO since suфcient
data points exist to permit a simultaneously in-
terpretation in terms of shock models using J-
and C-type physics. We investigated planar and
curved shocks.
Cepheus E is quite exceptional in its strong ra-
diative power (above 10 L ) and short dynamical
age (under 700 yr given an advance speed of over
100 km s 1 ). Our main results are as follows:
1. Extinction is most accurately derived by sat-
isfying multiple constraints on the complete
NIR data set rather than from specic line
ratios. The southern blueshifted lobe has
a mean K-band extinction of 1.4 mag and
the northern redshifted lobe has extinction
in the range 2.1 {2.4 mag. This is consis-
tent with the extinction being internal to a
spherical cloud, with a density of 10 5 cm 3
and radius 3  10 17 cm 3 . Remarkably, the
extinction-corrected H 2 columns, including
the ISO data, demonstrate a state close to
local thermodynamic equilibrium.
2. We interpret all the measured H 2 and CO
line uxes simultaneously in terms of shock
models. The best tting models are obtained
using shock distributions in the form of bow
shocks.
3. C-type physics is favoured here mainly be-
cause J-type physics predicts extremely
strong emission from the low-excitation
anks of a bow, which is not observed.
4. At least 20 bow shocks, close to paraboloidal
in shape, are predicted for each lobe.
5. A small dierence in the models for the
North and South lobes can be interpreted
through CO abundance variations in the
range 1.0 { 2.0 10 4 . Alternatively, given
the large and overlapping apertures, there is
probably no signicant dierence in the CO
abundances or excitation properties.
6. We also nd no signicant dierences in the
H 2 excitation when analysing the integrated
spectra of a whole out ow lobe or a 6 00 x 6 00
pixel-sized part. This is in agreement with
Eisloel et al. (1996) and suggests that the
large bows are built up of smaller unresolved
bow shocks, generated by ow instabilities.
7. The pre-shock medium is not fully molecu-
lar in these models. We have found a mean
atomic fraction of n(H)/(n(H)+2n(H 2 )) = 0.4.
With bow speeds of 120 km s 1 , exten-
sive bow apices are predicted within which
molecules are completely destroyed. The
reformation time is of order 10 17 /n c (H),
where n c (H) is the atomic density in the
compressed layers. Hence, a reformation
time of  3000 yr may be achieved for
n c (H) = 10 6 cm 3 .
8. The total mass set into motion by the out-
ow is low, 0.25 M (Ladd & Hodapp 1997).
We estimate a total cloud mass of 5 M
given a density of 10 5 cm 3 and radius
2.2  10 17 cm. With an out ow half-opening
angle of , a fraction f c = 1 - cos  of the
cloud would be disturbed by two lobes. Tak-
ing  = 20 ф , yields f c = 0.06, and thus pro-
vides a consistent basic model.
High out ow powers have also been uncovered
from many other Class 0 protostars. Mechanical
luminosity exceeds 50% of the bolometric luminos-
ity for all of the Perseus Class 0 sources conrmed
by Barsony, Ward-Thompson, Andre, & O'Linger
(1998). On the other hand, a few Class 0 proto-
stars such as L1527 and B335 possess relatively
weak out ows as measured by total far-infrared
luminosities (Giannini et al. 2001) and CO mo-
mentum ow rates (Bontemps, Andre, Terebey, &
Cabrit 1996). In this respect, the Cepheus E out-
ow could represent a powerful but abrupt evolu-
tionary phase, about to be brought to a halt as
underlying jets exit a small 0.1 pc cloud.
Jochen Eisloel and Dirk Froebrich received -
10

nancial support from the DLR through Verbund-
forschung grant 50 OR 9904 9. The ISO Spectral
Analysis Package (ISAP) is a joint development
by the LWS and SWS Instrument Teams and Data
Centers. Contributing institutes are CESR, IAS,
IPAC, MPE, RAL and SRON.
REFERENCES
Ayala, S., Noriega-Crespo, A., Garnavich, P.M.,
Curiel, S., Raga, A.C., Bohm, K.-H., Raymond,
J. 2000, AJ, 120, 909
Bachiller, R. 1996, ARA&A, 34, 111
Barsony, M., Ward-Thompson, D., Andre, P., &
O'Linger, J. 1998, ApJ, 509, 733
Black, J.H., van Dishoeck, E.F. 1987, ApJ, 322,
412
Bontemps, S., Andre, P., Terebey, S., & Cabrit, S.
1996, A&A, 311, 858
Cesarsky, C.J., Abergel, A., Agnese, P., et al.
1996, A&A, 315, L32
Chini, R., Ward-Thompson, D., Kirk, J.M., Niel-
bock, M., Reipurth, B., Sievers, A., 2001, A&A,
369, 155
Clegg, P.E., Ade, P.A.R., Armand, C., et al. 1996,
A&A, 315, L38
Davis, C.J., Smith, M.D. 1996, A&A, 310, 961
Eisloel, J., Smith, M.D., Davis, C.J., Ray, T.
1996, AJ, 112, 2086
Eisloel, J. 1997, IAU Symp. 182: Herbig-Haro
Flows and the Birth of Stars, 182, 93
Eisloel, J., Smith, M. D., & Davis, C. J. 2000,
A&A, 359, 1147
Froebrich D., Eisloel, J., Smith, M.D. 2001, AG
Abstr. Ser., 18, 59
Froebrich D., Smith, M.D. Eisloel, J. 2002a,
A&A, 385, 239
Froebrich D., Smith, M.D., Hodapp, K.-W.,
Eisloel, J. 2002b, A&A, submitted
Giannini, T., Nisini, B., Lorenzetti, D. 2001, ApJ,
555, 40,
Hatchell, J., Fuller, G.A., Ladd, E.F. 1999, A&A,
346, 278
Hollenbach, D. & McKee, C. F. 1989, ApJ, 342,
306
Kessler, M.F., Steinz, J.A., Anderegg, M.E., et al.
1996, A&A, 315, L27
Ladd, E. F. & Hodapp, K.-W. 1997, ApJ, 474, 749
Le och, B., Eisloel, J., Lazare, B. 1996, A&A,
313, L17
Moro-Martn, A., Noriega-Crespo, A., Molinari,
S., Testi, L., Chernicharo, J., Sargent, A. 2001,
A&A, 555, 146
Noriega-Crespo, A., Garnavich, P.M. 2001, AJ,
122, 3317
Noriega-Crespo, A., Garnavich, P.M., Molinari, S.
1998, AJ, 116, 1388
Reipurth, B. & Bally, J. 2001, ARA&A, 39, 403
Rousselot, P., Lidman, C., Cuby, J.-G., Moreels,
G., Monnet, G. 2000, A&A, 354, 1134
Smith, M.D., Brand, P.W.J.L. 1990, MNRAS,
245, 108
Smith, M. D. 1995, A&A, 296, 789
Smith, M. D., Suttner, G., & Yorke, H. W. 1997,
A&A, 323, 223
Smith, M. D., Khanzadyan, T., Davis, C.J., 2003,
MNRAS, in press
Smith, M. D., Davis, C. J., & Lioure, A. 1997,
A&A, 327, 1206
Suttner, G., Smith, M. D., Yorke, H. W., & Zin-
necker, H. 1997, A&A, 318, 595
This 2-column preprint was prepared with the AAS L A T E X
macros v5.0.
11

Fig. 4.| Planar shock models for CepE South.
In addition to the symbols described in Fig 3,
the ISOCAM data are represented by stars.(a)
A hot C-shock component with 32 km s 1 pro-
vides a t to the vibrationally-excited columns
but does not contribute to the CO rotational
lines (other critical parameters are a density of
10 5 cm 3 , an Alfven speed 1.5 km s 1 , ion frac-
tion  10 6 , a transverse magnetic eld and
n(H 2 )/(n(H)+2n(H 2 )) = 0.3). The maximum
temperature is 3160 K. (b) A cool C-shock with
speed 11 km s 1 provides a t to the rotationally-
excited H 2 and CO but does not contribute to the
vibrational lines (other critical parameters are as
above except n(H 2 )/(n(H)+2n(H 2 )) = 0.49998).
The maximum temperature in the shock is 950 K.
Also modelled are abundances of (O) = 4 10 4
and (C) = 1:5 10 4 .
Fig. 5.| The geometrical parameters associated
with a fast-moving bow shock, as originally em-
ployed by Smith and Brand (1990).
12

Fig. 6.| A C-type bow shock model for
CepE South 1. A C-type bow model with
speed 120km s 1 , a density of 10 5 cm 3 , an
Alfven speed 1.5 km s 1 , ion fraction  10 6
and n(H 2 )/(n(H)+2n(H 2 )) = 0.3, abundances of
(O) = 3 10 4 and (C) = 1:5 10 4 (initially
in CO) and a magnetic eld aligned with the bow
axis.
Fig. 7.| A C-type bow shock model for CepE
North (location Nd). A C-type bow model
with speed 120km s 1 and critical parameters
as in Fig. 6, except for the initial abundance
(CO) = 1:2 10 4 .
13

Fig. 8.| A J-type bow shock model for
CepE North, position Nd, with shape parame-
ter s = 1.5. A speed of 70 km s 1 , a density
of 4 10 4 cm 3 , an Alfven speed 2 km s 1 , ion
fraction, n(H 2 )/(n(H)+2n(H 2 )) = 0.44, and abun-
dances of (O) = 4 10 4 and (C) = 1 10 4
(initially in CO) and a magnetic eld aligned with
the bow axis were taken.
14