Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/~lt/preprints/g962/g962.ps.gz Äàòà èçìåíåíèÿ: Tue Sep 11 16:56:44 2007 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 06:53:28 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: photosphere

A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
08 (09.08.1; 09.13.2; 09.10.1; 08.06.2; 13.09.6; 09.09.1 G9.62+0.19)
ASTRONOMY
AND
ASTROPHYSICS
30.7.1997
HII and hot dust emission around young massive stars in
G9:62 + 0:19
Leonardo Testi 1;2 , Marcello Felli 3 , Paolo Persi 4 and Miguel Roth 5
1 Dipartimento di Astronomia e Scienza dello Spazio, Universit`a degli Studi di Firenze, Largo E. Fermi 5, I­50125 Firenze, Italy
2 California Institute of Technology, Division of Physics, Math and Astronomy, 105­24, Pasadena, CA 91125, USA
3 Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, I­50125 Firenze, Italy
4 Istituto di Astrofisica Spaziale, C.N.R., C.P. 67, I­00044 Frascati, Italy
5 Las Campanas Observatory, Casilla 601, La Serena, Chile
July 30, 1997
Abstract. In this paper we present new near infrared
(NIR) observations (J, H, and K broadbands), of the
G9:62 + 0:19 star forming complex.
Comparison of our observations with similar resolution
centimetric continuum, millimeter continuum and molec­
ular emission show that the mm continuum source F, not
detected in the cm wavelength free­free radio continuum
and associated with a high density molecular peak, is de­
tected at 2:2 ¯m, while the ultracompact HII region D,
one of the youngest of the HII regions in the complex, is
not detected in the near infrared.
We propose a simple model that explains why, in the
first stages of evolution of a young massive star, the source
may be observable at K but not in the cm radio contin­
uum. When the size of the UC HII region is Ü 10 \Gamma3 pc
the hot dust present around the YSO strongly emits at
K band, but the radio emission will be strongly self­
absorbed. At later stages, when the size of the UC HII be­
comes greater than ?
¸ 10 \Gamma3 pc, the dust temperature goes
down and the K band dust emission strongly decreases;
at the same time the cm radio continuum becomes de­
tectable. As the UCHII expands the extinction drops and
the K band emission rises again due to the stellar and
ionized gas free­free and free­bound emission.
We propose an evolutionary sequence of the different
sites of star formation in the complex, based on the ra­
dio continuum--infrared morphology and on the associa­
tion with H 2 O masers.
Key words: HII regions ­ molecules ­ jets and outflows
­ star formation ­ infrared sources: stars ­ individuals:
G9.62+0.19
Send offprint requests to: Testi: Caltech, lt@astro.caltech.edu
1. Introduction
Although originally associated with advanced stages of
massive star formation (such as diffuse HII regions), H 2 O
masers are now believed to be associated with the ear­
liest phases of massive star formation. Several pieces of
evidence (Palla et al. 1993; Codella et al. 1994) suggest
that the masers are associated with massive stars that
have not even had the time to develop an ultra compact
(UC) HII region. High resolution radio continuum and
maser line observations (Forster & Caswell 1989; Tofani
et al. 1995; Hofner & Churchwell 1996) showed that the
majority of the H 2 O maser spots, although generically as­
sociated with high mass star forming regions (i.e. HII re­
gions), are not closely related to compact radio continuum
sources, suggesting that they must have a close­by exciting
source, independent from the ones that power more dif­
fuse HII regions in the same complex. On the other hand,
the masers are always associated with dense hot molec­
ular cores and compact mm wavelength and far infrared
continuum sources (Cesaroni et al. 1991; 1994, hereafter
CCHWK; Codella et al. 1997; Jenness et al. 1996; Cesa­
roni et al. 1997), indicating that the powering source for
the masing gas should be a Young Stellar Object (YSO)
still deeply embedded in its parental molecular clump.
To unveil the short wavelength tail of the spectral en­
ergy distribution (SED) of the sources associated with the
masers, and to better constrain their age and evolutionary
status, we have undertaken a survey at near infrared wave­
lengths (1 \Gamma 2:4 ¯m) of galactic H 2 O masers sources. In a
preliminary sub­sample (Testi et al. 1994) observed with
moderate sensitivity (although the survey is much deeper
and with higher resolution than any one done so far), we
were able to find a NIR source reliably associated with the
maser source in almost 90% of the objects. The infrared
characteristics of these sources, namely a very large in­
frared excess (H--K)? 2, suggested that they are YSOs

2 L. Testi et al.: The SFR G9.62+0.19
deeply embedded in molecular cores. In particular, the
NIR colours of the objects could not be explained in terms
of extincted stellar photospheres, but a large fraction of
the observed radiation had to be produced by dusty en­
velopes heated to high temperatures (up to 1000 K) by the
YSO radiation field. Given the expected mass of the ob­
jects and the lack of radio continuum emission the sources
were suggested to be in a pre­UC HII region phase.
A few sources analysed in greater detail (Hunter
et al. 1995; Palla et al. 1995; Persi et al. 1996; Felli et
al. 1997; Cesaroni et al. 1997) confirmed this view, and
emphasized the power of coordinated multi­wavelength
observations with high resolution and sensitivity.
In this paper we will focus our attention on the star
forming complex G9:62 + 0:19. This is quite an interest­
ing case, since a cluster of HII regions placed at a dis­
tance of 5:7 kpc (Hofner et al. 1994) in different evolution­
ary phases (i.e. with different sizes and mean densities) is
present.
Radio continuum observations made at several wave­
lengths with different resolutions showed the presence of
five components, labelled from A to E (Garay et al. 1993;
Kurtz, Churchwell & Wood 1994; CCHWK). From west
to east: component A is a diffuse and extended (¸ 0:5 0
diameter) HII region, with mean density 2 \Theta 10 2 cm \Gamma3 ;
component B is a less extended ( !
¸ 0:3 0 , mean density
2 \Theta 10 3 cm \Gamma3 ) cometary (or blister shaped) HII region,
with the tail toward the south­west. Toward the north­
east there are three compact radio continuum components
C, D, and E, classified as UC HII regions. The radio com­
ponents detected by CCHWK are labelled in Fig. 2 and 3.
Note that component A lies outside the area shown in the
figures (see Fig. 3 of Garay et al. 1993).
Low resolution single dish molecular observations of
Cesaroni et al. (1991) and Cesaroni, Walmsley & Church­
well (1992), revealed a molecular core associated with the
cluster, but could only rule out the association of the hot
molecular material with radio continuum component A
(the more evolved), without being able to indicate which
of the other components was likely to be associated with
the molecular peak.
The region shows maser emission of several in­
terstellar high­excitation molecules. High resolution
VLA observations of the OH (Forster & Caswell 1989;
Forster 1993), CH 3 OH (Norris et al. 1993), H 2 O (Forster
& Caswell 1989; Forster 1993; Hofner & Churchwell 1996)
and NH 3 (5,5) (Hofner et al. 1994) masers showed that
the maser spots are concentrated in a filamentary struc­
ture extending from component D to E. In particular the
H 2 O masers are found close to components D, E, and
F. The comparison of the number, positions and inten­
sities of the spots detected by Forster & Caswell (1989)
and those detected by Hofner & Churchwell (1996) show
some distinct changes which reflect the variability of the
maser activity. Here we will use the observations of Hofner
& Churchwell (1996) because they are more sensitive and
with higher resolution.
Interferometric observations of the NH 3 inversion lines
(CCHWK, Hofner et al. 1994) showed that there is a
molecular clump associated with the maser activity, close
to component D, but a few seconds of arc to the north
of the radio continuum region. A fainter ammonia clump
was found associated with component E.
Interferometric NH 3 (4,4) (CCHWK) and millimeter
continuum and molecular line observations (Hofner et
al. 1996) revealed the presence of a sixth component (F)
¸ 4 00 north of component D which does not have de­
tectable centimeter wavelength radio continuum emission,
but it is the strongest source in the field at mm contin­
uum and in the NH 3 (4,4) and CH 3 CN(J=6\Gamma5) lines. The
strongest H 2 O maser spots are located between compo­
nent F and component D. The C 18 O(1\Gamma0) interferomet­
ric spectra taken at the position of component F show
a compact molecular component with broad wings, with
asymmetric structure, which was interpreted as a local­
ized bipolar molecular outflow associated with component
F (Hofner et al. 1996). From their mm continuum and
molecular line data Hofner et al. concluded that compo­
nent F is the youngest object in the region and that it is
probably a high mass star still in the accretion phase.
In this paper we present J,H, and K near infrared im­
ages of the whole star forming region. The observations are
described in Sect. 2. The results are presented in Sect. 3.
In Sect. 4 we discuss the evolutionary status of the sources
in the region, focusing our attention on the two sources F
and D, and presenting a simple model that can explain
the occurrence of NIR and radio emission during the evo­
lution of a massive star. Finally in Sect. 5 we summarize
the main conclusions of this work.
2. Observations
2.1. Near infrared observations
The near infrared images were obtained in the framework
of an extensive project of infrared observations of galac­
tic maser sources (Testi et al. 1994; 1995; 1997). Broad
band near infrared observations were taken on June 4 th
1992 with a NICMOS3 (256 \Theta 256 HgCdTd pixels) cam­
era (Persson et al. 1992) mounted at the 1 m telescope of
the Las Campanas Observatory. The scale on the detec­
tor was 0:92 00 =pix. The source was observed at J, H and K
(1:25, 1:65 and 2:2 ¯m respectively) using a dithering tech­
nique: five partially overlapping frames were taken in each
band; the flat field image was calculated by clipping and
median averaging the five frames taken in each band. Af­
ter flat fielding, the images were registered and mosaiced
together in order to increase the signal to noise ratio in the
common area (centered on the nominal H 2 O maser posi­
tion). All the data reduction and analysis was performed

L. Testi et al.: The SFR G9.62+0.19 3
Fig. 2. a K­band greyscale image of the ¸ 1 0
\Theta 1 0 region centered on the H2O maser position. b J­band greyscale of the
same region. In both images the H2O maser spots detected by Hofner & Churchwell 1996 are indicated with filled triangles
(symbols are larger than the position uncertainties). On the K­band image the VLA D­array radio continuum (thin contours)
and NH3(4,4) emission (thick contours) observed by CCHWK are reported; the cm and mm components B, C, D, E, and F
are labelled; component A lies to the west (outside the plotted area), see Fig. 3 of Garay et al. (1993). The contour values are:
from 10 to 90 by 10 mJy/beam and from 10 to 70 by 10 mJy/beam, for the continuum and line emission respectively. The large
ellipse is the error box of IRAS18032 \Gamma 2032. Coordinates are at B1950.
Fig. 1. K­band greyscale image of the ¸ 2:5 0
\Theta 2:5 0 observed
region. North is at top and east to the left.
using the IRAF 1 and ARNICA 2 (Hunt et al. 1994) soft­
1 IRAF is made available to the astronomical community by
the National Optical Astronomy Observatories, which are op­
ware packages. The K­band image of the observed region
is shown in Fig. 1.
Photometric calibration was achieved by observ­
ing some of the UKIRT faint standards (Casali &
Hawardeen 1992). The photometric accuracy is estimated
to be ¸ 5%. The limiting magnitudes in the overlapping
region (3oe in 4 arcsec aperture) are: 16:3, 15:4 and 14:4,
at J, H and K respectively.
Astrometric calibration was performed using a set of
optical stars extracted from the Digitized Sky Survey,
made available by the Space Telescope Science Institute.
We estimate the calibration accuracy to be better than 1
arcsec (see Testi 1993, for a complete description of the
astrometric calibration procedure).
2.2. H 2 O maser observations
The water maser emission in G9.62+0.19 has been ob­
served in four different epochs (3/87, 6/87, 1/91, 1/94)
with the 32 m Medicina 3 VLBI antenna (observational
procedures and instrumental setup are described in Brand
et al. 1994). The observations are part of the Arcetri H 2 O
masers variability monitoring program.
erated by AURA, Inc., under contract with the U.S. National
Science Foundation
2 ARNICA can be obtained from the Osservatorio Astrofisico
di Arcetri at ftp://150.217.20.1/pub/arnica/
3 The Medicina telescopes are operated by the Istituto di
Radioastronomia (CNR), Bologna, Italy.

4 L. Testi et al.: The SFR G9.62+0.19
3. Results
In Fig. 2a the K band image of a region 1 0 \Theta 1 0 wide cen­
tered on the maser position is shown. Fig. 2b shows the J
band image of the same region. In both figures the maser
spots detected by Hofner & Churchwell (1996) are also
shown. It can be clearly seen that the masers are well
separated from the diffuse K­band emission (and the dif­
fuse radio continuum emission). All the maser spots lie
towards the north­east of the infrared diffuse emission,
approximately at the position of the components D, E,
and F.
As also found in other regions (see e.g. Testi et
al. 1994), the IRAS source is not coincident with the H 2 O
maser and compact radio continuum components, but is
associated with the extended K­band and radio continuum
emission. Consequently the IRAS fluxes are indicative of
the global infrared emission of the area and the luminosity
of 4:3 \Theta 10 5 L fi given by CCHWK refers to all the sources
in the area.
The comparison between our near infrared images and
the radio continuum maps of Kurtz et al. (1994) and
CCHWK show that at 2:2 ¯m we detect emission from
the radio continuum sources B, C, and E, but not from A
and D. Radio source A is probably so diffuse that it has
a surface brightness at K­band too low to be detected.
More difficult to explain is the lack of K­band emission
from the UC HII region D, which is associated with a star
of spectral type earlier than O9:5 (CCHWK). In Fig. 3,
the NH 3 (4; 4) and radio continuum emission are overlaid
on the K­band image. From the figure, it can be clearly
seen that a NIR source is found coincident with compo­
nent F, which is defined by the ammonia clump, some of
the strongest H 2 O maser spots and the mm continuum,
but it does not have cm­wavelength radio continuum emis­
sion. Component F is detected only at K­band, with upper
limits at J and H, indicating that it is heavily extincted
and has a large NIR excess.
In Fig. 4 (upper panel) the (J--H,H--K) colour­colour
diagram of the near infrared sources detected in a 2:5 0 \Theta
2:5 0 area around the maser position is presented. About
12% of the detected NIR sources show infrared excess, and
are probably young stellar objects surrounded by hot dust,
either in circumstellar disks or envelopes. The NIR sources
associated with radio components C, E, and F and the
point source inside component B (called NIR--B? to avoid
confusion between diffuse emission and the pointlike ``stel­
lar'' component) are marked. All the sources show NIR
excess, source F being the reddest of the whole field. In
Fig. 4 (lower panel) a diagram showing the position of the
NIR sources on the sky is also presented. It is worth noting
that the infrared excess sources (filled diamonds) are not
confined inside a well defined cluster as usually found in
similar regions (see e.g. Cesaroni et al. 1997 for the case of
IRAS20126+4104 or Persi et al. 1997 for G35.20\Gamma1.74).
Fig. 3. Greyscale: enlargement of the K­band image of Fig. 2.
Contours: VLA C­array radio continuum at 1:3 cm (thin con­
tours) and NH3(4; 4) (thick contours) emission observed by
CCHWK. Filled triangles: H2O maser spots detected by Hofner
& Churchwell (1996). Open squares: OH maser spots from
Forster (1993). Contour levels are: 10, 30, 50, 70, 90 mJy/beam
and 8, 13, 18 mJy/beam, for radio continuum and NH3(4,4)
respectively. The cm and mm components D, E, and F, as well
as the NIR sources NIR­B?, NIR­C, NIR­E and NIR­F, are
labelled (see text). Coordinates are at B1950.
In Table 1, the observed parameters of the NIR sources
associated with the radio and mm continuum components
are shown (note that the value of ffi for radio continuum
source C in Table 4 of CCHWK is incorrect). For each
source the first column gives the identifier, the second and
third the coordinates at 1950 epoch, and the fourth, fifth
and sixth the magnitudes measured in the J, H and K fil­
ters, respectively. When no error is reported, and a symbol
``?'' is indicated, the source has not been detected in that
filter, and the quoted magnitude is a lower limit (i.e. the
corrensponding flux is an upper limit).
3.1. Radio components A and B
In the radio continuum both components A and B are dif­
fuse and have an optically thin spectrum. Radio source
A is the largest source of the complex. Its surface bright­
ness at K is too low to be detected by these observations.
Radio source B has a cometary or blister morphology,
with the ``tail'' directed to the south­west. Component B
is detected in the near infrared as a diffuse component,
detected only at K, and a point source near to the ra­
dio continuum peak (see Table 1). From the radio con­

L. Testi et al.: The SFR G9.62+0.19 5
Table 1. Observed parameters of the NIR sources associated with the cm and mm components
Name ff (1950) ffi (1950) mJ mH mK
NIR­B? 18 : 03 : 15:39 \Gamma20 : 32 : 05:4 ? 16:7 14:7 \Sigma 0:1 12:3 \Sigma 0:05
NIR­C 18 : 03 : 15:49 \Gamma20 : 31 : 47:5 ? 16:7 14:4 \Sigma 0:1 12:1 \Sigma 0:04
NIR­E 18 : 03 : 15:94 \Gamma20 : 31 : 53:6 15:5 \Sigma 0:07 15:0 \Sigma 0:2 14:0 \Sigma 0:2
NIR­F 18 : 03 : 16:15 \Gamma20 : 32 : 0:0 ? 16:7 ? 15:4 12:9 \Sigma 0:08
tinuum observations, CCHWK found that the source re­
sponsible for the ionization should be an O7:5 zero age
main sequence (ZAMS) star. We believe that the point
source inside the diffuse K­band emission coincides with
the star responsible for the ionization of the HII region.
The colours and magnitudes of this source (NIR--B?) are
compatible with the magnitudes and colour of an O7:5
star at 5.7 kpc and with an extinction of AV ' 30, as­
suming a Rieke & Lebofsky (1985) extinction law and in­
trinsic colour and magnitudes from Koornneef (1983) and
Schmidt­Kaler (1981). In fact, an O7:5 main sequence star
should have (H\GammaK)= \Gamma0:05, and MH = \Gamma4:2 (we use the
magnitude measured in the H­band, instead of that mea­
sured at K, because at H the contribution of the infrared
excess should be much lower than the photospheric emis­
sion), which are consistent with the observations, assum­
ing AV ' 30, and the presence of a moderate infrared
excess at K.
3.2. Radio component C
At K band, this source is almost pointlike and its colour
and magnitude cannot be simply explained as a reddened
B0 ZAMS star (required to explain the radio continuum
emission). From the NIR colours and magnitudes (see Ta­
ble 1 and Fig. 4) we expect that a large fraction of the
K­band emission is not due to the (reddened) stellar pho­
tosphere, but is probably coming from hot circumstellar
dust.
3.3. Radio component D
This source is not detected in our near infrared images,
hence we can only quote upper limits for its magnitudes
(m J ? 16:7; mH ? 15:4; mK ? 14:5). Considering that
the ionizing star should be earlier than O9:5 (CCHWK),
we derive an extinction toward the source greater than 40
magnitudes in the visual.
3.4. Radio component E
This is the only source detected in all the three NIR broad
bands (see. Table 1). The source has a near infrared excess
as shown in Fig. 4, and it is probably more evolved than
the other sources which are associated with H 2 O masers.
Assuming a spectral type B1 (required to explain the radio
continuum emission, see CCHWK and Hofner et al. 1996)
we find that the infrared colours of the source are compat­
ible with those of a ZAMS star extincted by more than 30
magnitudes (in the visible) and with a substantial infrared
excess.
3.5. The mm and molecular component F
Taking into account the upper limits in the J and H emis­
sion, this is the source with the largest NIR excess in the
field (H--K?2.5, see. Fig. 4 and Table 1). This source is
close to the strongest H 2 O maser source (located mid­
way between this component and component D) and as­
sociated with a molecular outflow observed in C 18 O (1--0)
(Hofner et al. 1996). The large infrared excess, the associ­
ation with the maser source, the presence of a molecular
hot core, the mm dust emission, the lack of a radio UC HII
region, and the molecular outflow, all point to the conclu­
sion that this is the youngest source of the complex. From
the inferred temperature and size of the molecular conden­
sation Hofner et al. 1996 derive an expected luminosity of
the central star of ¸ 1:8 \Theta 10 4 L fi which would correspond
to a B0--B0.5 ZAMS star. Assuming that CCHWK do not
detect radio continuum emission from component F due
to beam dilution, and assuming that the free­free emission
is optically thick, one can derive an upper limit on the size
of the (expected) HII region embedded inside the molec­
ular clump. We derive a (3oe) upper limit on the radius of
1:5 \Theta 10 \Gamma2 arcsec, which corrensponds to 4 \Theta 10 \Gamma4 pc at
5.7 kpc.
4. Discussion
There are several models that fit the far infrared emis­
sion of embedded high­mass young stars (see e.g. Natta &
Panagia 1976; Churchwell, Wolfire & Wood 1990). How­
ever all of these models are focused on the interpretation of
the far infrared emission, which originates from the large
scale dusty envelope around the star and is dominated by
the cooler dust. Accurate modelling of the near infrared
emission from young stars has been performed for low and
intermediate mass stars (Beckwith et al. 1990; Natta et
al. 1993; Calvet, Hartmann & Strom 1997) where no ion­
izing radiation is present.
Since we are primarily interested in the near infrared
emission during the first stages of evolution of high mass
stars, we have developed a qualitative evolutionary model
focused on hot circumstellar dust to explain the present
observational results.

6 L. Testi et al.: The SFR G9.62+0.19
Fig. 4. Upper panel: (J--H,H--K) colour­colour diagram of the
near infrared sources detected in a ¸ 2:5 0 \Theta 2:5 0 region around
the maser position (shown in Fig. 1). The line labelled MS
represents the colours of main sequence stars (Koorneef 1983).
The two dashed lines define the region in which reddened main
sequence stars should lie (assuming Rieke & Lebofsky 1985 in­
terstellar extinction law). Sources with infrared excess are indi­
cated with filled diamonds. Lower panel: positions of the NIR
sources plotted in the colour­colour diagram. The offsets are
calculated from the position of source NIR­F (see text). Filled
diamonds represent infrared excess sources; filled circles rep­
resent the positions of sources NIR­B?, NIR­C, NIR­E, D and
NIR­F.
Such a model should be able to explain why source F
is observed in K band, but not in the cm radio continuum
down to ¸ 4 mJy at 22 GHz, while source D is observed
in the cm radio continuum but not in the near infrared.
4.1. A simple evolutionary model
We will consider three contributions to the near infrared
emission: the stellar photosphere, the free­free and free­
bound emission from ionized gas and the hottest layer
of the dusty envelope which surrounds the YSO. We will
assume a spherically symmetric model, thus neglecting
anisotropies introduced by an accretion disk and/or an
outflow. Similarly, the contribution of radiation escaping
through optically thin paths of the envelope and reflected
into the line of sight of the observer will not be considered.
The photospheric emission in the near infrared bands
has been estimated using the calibration of Schmidt­
Kaler (1981) and Koornneef (1983). The free­free and free­
bound ionized gas emission in the NIR has been computed
using the emission coefficients of Sibille et al. (1974). The
dusty envelope is assumed to emit in the near infrared as
a black body with the temperature and size of the hottest
dusty layer (the one closest to the star). Such an assump­
tion is valid if only the hottest layers of the dust shell
contribute significantly to the NIR emission, while the
external colder layers of the large scale dusty molecular
cloud provide most of the extinction but negligible NIR
emission.
The contribution of each type of emission may change
during the evolution of the source. When the extinction is
large it is very unlikely that the stellar photosphere and
the free­free and free­bound emission from the ionised gas
give the main contribution to the observed emission at
2:2 ¯m, this may be the case for component F. On the
other hand, in more evolved cases, such as component B,
the pointlike infrared component (NIR­B?) can be easily
explained in terms of a stellar photosphere of an early type
star surrounded by an optically thin HII region, extincted
by optically thick (AV ' 30) cold dust.
A young source, which has not even had time to de­
velop an observable UC HII region or if the expansion
of the ionized gas is quenched by a strong accretion rate
(Walmsley 1995), may still be surrounded by a hot dust
envelope, with the dust as close to the star as permitted
by the sublimation temperature. As the HII region ex­
pands, the dust close to the star will be destroyed by the
ionizing radiation and pushed outward by the radiation
pressure and the stellar wind. Thus, as the time increases
and the UC HII region expands, we expect the tempera­
ture of the dust that emits in the near infrared to drop
approximately as r \Gamma1=2 , where r is the radius of the UC
HII region. The effective increase in the emitting surface
(as r 2 ) of the hot dust shell does not balance the decrease
in its temperature (due to the exponential dependence on

L. Testi et al.: The SFR G9.62+0.19 7
Fig. 5. Dependence of the K­band and 22 GHz emission on
the three main parameters of the model: AV (0) (top panel), the
index of the density power­law (central panel) and the spectral
type of the ionizing star (lower panel). In all plots the K­band
emission is presented as a solid line (the scale is that on the
left axis), and the 22 GHz flux is presented as a dotted line
(the scale is that on the right axis). The models have been
calculated assuming a distance of 5.7 kpc.
Fig. 6. K­band emission of the three components of the model,
for an O8 (upper panel) and a B0 (lower panel) ZAMS star.
The solid line represents the total emission, the dotted line the
``hot layer'', the short dashed line the continuum emission of
the ionized gas, and the long dashed line the contribution of
the stellar photosphere.
T in the Wien region of the black­body spectrum), and
the NIR dust emission drops rapidly with r.
On the other hand, the cm radio continuum emission
of the ionized gas will be optically thick and self­absorbed
in the first stages, when the radius of the HII region, r, is
small. At a given frequency, as r increases, the radio con­
tinuum emission of the HII region increases (until even­
tually becomes optically thin) making the ionized gas de­
tectable.
To develop a simple model, we assume that the ionized
region is spherical and the density is uniform within the
HII region at any time, but variable with the radius r
of the HII region (which in turn depends on the age of
the source) in a way to always balance the ionizing stellar
radiation, which is kept constant. The hot dust is present
only outside the HII region and it emits like a black body
at temperature T (r) defined by the thermal balance with
the stellar radiation field. The external cooler parts of the
envelope provide only the extinction in the NIR, and emit
the bulk of the far infrared and submillimeter radiation
with negligible contribution in the NIR. The extinction is
computed assuming a power law density distribution in
the neutral part of the clump and a radius of the clump
R = 0:5 pc. Hence it has a maximum at the beginning and
decreases as r increases. In the calculation we parametrize

8 L. Testi et al.: The SFR G9.62+0.19
Fig. 7. K­band (upper panel), H­band (central panel) and
J­band (lower panel) emission of the three components of the
model, for an O9 ZAMS star. The three different components
are represented as in Fig. 6.
AV (the equivalent extinction in the V band) in terms of
AV (0), the extinction at the beginning of the simulation,
and the power law index of the density distribution. For
all the model simulations we have assumed a distance of
5:7 kpc.
Under these assumptions the K­band emission and
the free­free radio continuum emission (for instance at
22 GHz) depend only on the assumed spectral type of the
central star, the extinction provided by the cool dust, and
the radius of the ionized region.
In Fig. 5 the observed magnitude at K and the radio
continuum flux at 22 GHz are presented as a function of
the radius of the ionized region. In the three panels the
dependence of the K­band and 22 GHz continuum on the
three main parameters of the model is shown. In the upper
panel of Fig. 5 three different simulations for an O9 zero
age main sequence (ZAMS) star and a density profile n /
r \Gamma1:25 are given for A V (0) = 40, 60 and 80 mag. The cm
radio continuum emission, which, under our assumptions,
depends only on the number of ionizing photons available,
is the same in all the three models and depends only on the
size of the HII region. The K­band emission is dominated
by the ``hot dust layer'' for r !
¸ 5 \Theta 10 \Gamma4 pc, then there is
a minimum in the emission for increasing r when the dust
becomes too cold to emit significantly at 2:2 ¯m and the
stellar and the free­free and free­bound emission from the
Fig. 8. Infrared K­band and radio continuum observations of
components C, D, E, and F are compared with the predictions
of some models. Open symbols represent the radio continuum
emission measured at 22 GHz by CCHWK (note that for com­
ponent F we report an upper limit of 1 mJy). Filled symbols
represent our infrared measurements (see Table 1). Upper lim­
its are indicated with arrows. We use the sizes of sources C
and E reported by CCHWK (note that we use the radius). An
external ambient extinction of AV =30 mag has been added.
ionized gas are still too extincted. Finally, the emission
rises again as the extinction decreases. In practice the hot
dust produces a peak of K­band emission at small radii
in spite of the high extinction; then there is a minimum
in the K emission due to the drop of the ``hot dust layer''
temperature; while at large radii the contribution of the
star and ionized gas raises the K emission again when the
extinction drops.
In the central panel of Fig. 5 we present four simu­
lations for an O9 ZAMS star, A V (0) = 60 mag and for
density profiles with power law indices 1.1, 1.25, 1.50, and
2.0. The qualitative behaviour of the models is very similar
to that of the previous models.
In the lower panel of Fig. 5 three simulations for
A V (0) = 60 mag, n / r \Gamma1:25 and different spectral types
B0, O9 and O8 ZAMS stars are presented. Also in these
cases the qualitative behaviours of the curves are similar
to the previous simulations.
The general trend shown by all the simulations is that
when r is small (say less than 100 AU), i.e. when the region
is very young, the near infrared emission is high due to the
hot dust near the stellar photosphere, while the radio con­
tinuum emission is strongly self absorbed. As the radius
increases (i.e. at later times) the NIR emission drops and

L. Testi et al.: The SFR G9.62+0.19 9
the radio emission rises toward the optically thin value.
At large radii (r ?
¸ 10 \Gamma3 pc) only the star and the free­free
and free­bound radiation from the ionized gas contribute
to the emission at K, which increases as the extinction de­
creases due to the erosion of the inner part of the molecular
cloud by the expanding HII region. The ``height'' of the
peak of the K emission at small radii (i.e. the difference
in magnitudes between the peak and the minimum in the
emission at log(r) ' \Gamma3:5 pc) depends on the model pa­
rameters, but, in the range of parameters that we have
explored, is between 1 and 2 magnitudes.
In Fig. 6 the contributions of the three emissions at
K­band are shown for two different central stars (O8 and
B0 ZAMS). Note that the contribution of the free­free and
free­bound emission from the ionized gas is smaller than
the contribution of the stellar photosphere for a B0 ZAMS
star, whereas the opposite holds for an O8 ZAMS star. In
Fig. 7 the three contributions at J, H and K band are
shown for an O9 ZAMS star. The ``hot layer'' of dust in
the surrounding envelope dominates the infrared emission
at small radii only in the K and H bands; at J the dust
emission is always very low and the source should not be
detectable in the early stages of evolution.
This model may explain, at least qualitatively, the dif­
ferences observed between sources D, E, and F. In fact,
source F is probably in the stage of very small r, when
the radio continuum emission is below the mJy level, and
the K­band emission is near its peak; source D is in the
phase in which: 1) the radio continuum emission has al­
most reached the peak, 2) the dust shell is too large and
too cold to emit at NIR wavelengths and 3) the extinction
too high to detect the stellar and ionized gas free­free and
free­bound emission in the NIR. Source E is in the phase
in which the NIR emission becomes detectable again and
the radio continuum emission is optically thin. Source C
is similar to E but in a more evolved stage, since it is not
associated either to molecular or mm continuum emission.
In Fig. 8 our near infrared and CCHWK radio con­
tinuum measures for sources C, D, E, and F are com­
pared with the predictions of the model. Open dots refer
to the radio continuum measurements, while filled dots
are the NIR measurements. An external (constant) am­
bient extinction of AV = 30 mag (as measured toward
source NIR­B?) has been added to the model. The sizes
of sources C, D and E are those measured by CCHWK
and Hofner et al. (1996). The upper limit on the size of
source F has been derived assuming that CCHWK does
not detect the source due to beam dilution (see Sect. 3.5).
We positioned it on the graph at 10 \Gamma4 pc. The agreement
of the prediction of the model in the radio and NIR bands
and the observations is generally rather good, considering
the crudeness of the model.
It is worth pointing out some obvious limitations of
our model and the proper range in which it can be used:
1) at very small radii (r !
¸ 6 \Theta 10 \Gamma5 pc) the near infrared
emission decrease is due to the higher values of the ex­
tinction in the first stages of the simulations. Clearly, a
full radiative transfer calculation should be performed at
small radii. 2) other processes not taken into account in
this simple model may contribute to the NIR emission,
such as the hot dust inside the HII region, an accretion
disk close to the star and line emission. 3) finally, most of
the sources observed in detail show non­spherical geome­
tries. For example in W75N (Moore et al. 1991; Hunter et
al. 1994) and in IRAS20126+4104 (Cesaroni et al. 1997)
the NIR morphology of the source is highly asymmetric
and there are clear indications that most of the infrared
radiation is actually starlight plus circumstellar emission
reflected in large scale lobes. However, these cases can be
clearly distinguished from source F since the NIR emission
does not coincide with the molecular and mm­continuum
peak.
4.2. The maser activity
In the Medicina single dish observations, the water maser
is variable (as in almost all known star formation H 2 O
masers) but not as much as observed in other sources as
for instance W75N (Hunter et al. 1994) and S235B (Felli et
al. 1997). From the limited data that we have, we can see
that the shape of the spectrum does not change much dur­
ing the years, and the components between 0 and 10 km/s
have raised their intensity by a factor of two in 8 years.
The fact that the masers are associated with the earli­
est evolutionary phases of a YSO fits nicely with the qual­
itative model described above. In fact, the maser sources
are expected to be pumped efficiently only close to the
YSO, between 10 14 cm and 10 16 cm. This corrensponds ex­
actly to the region close to the K­band peak due to the hot
shell NIR emission. YSOs equally bright at K but with a
more extended UCHII region (i.e. with the NIR emission
coming from stellar and ionized gas contributions) do not
show associated H 2 O maser emission.
5. Conclusions
In this paper we have presented new NIR observations
in the J, H and K broad bands of the G9.62+0.19 star
forming region. The NIR observations have been com­
pared with published high resolution radio continuum and
molecular line observations in the centimetric and milli­
metric bands.
The infrared observations enabled us to the detect
the photospheric emission of the ionizing stars associated
with the more evolved UC HII regions as well as the free­
free and free­bound emission from the ionized gas (compo­
nents E, C, and B). The most compact (and presumably
younger) of the UC HII regions (component D) has not
been detected at NIR wavelengths. The mm­continuum
and molecular component F (which is not associated to
any detected centimetric radio continuum emission) has
been detected at K­band.

10 L. Testi et al.: The SFR G9.62+0.19
We propose a simple evolutionary model which is able
to explain the infrared and radio morphology of the young
massive stars in the complex. A young massive star is sur­
rounded by a spherical ionized region and a dusty enve­
lope, as the HII region expands the envelope is eroded. In
the near infrared the emission is produced by: 1) the stel­
lar photosphere, 2) the free­free and free­bound radiation
from the ionized gas and 3) the hottest layer of the dusty
envelope. The cooler parts of the envelope provide the ex­
tinction. When the source is young and the HII region
is compact and self­absorbed, the dusty envelope is close
to the star and the dust, heated by the stellar radiation,
emits strongly in the K band. In spite of the high value of
the extinction in these early phases, the hot dust emission
is observable at K, while the cm radio continuum from
the HII region is too self­absorbed to be detectable (com­
ponent F is expected to be in this evolutionary phase).
As the HII region expands the radio free­free emission be­
comes detectable, but the dust cools down rapidly and the
K­band emission drops since the star and ionized gas are
still too extincted to be detectable (component D). The K­
band emission rises again, due to the contributions of the
stellar photosphere and free­free and free­bound radiation
from the ionized gas, when the exctinction drops due to
the erosion of the molecular cloud by the UC HII region.
In this picture the H 2 O maser activity is associated
with the earliest evolutionary phases, in which the ion­
ized gas is confined in a sub­milliparsec region around the
young star.
Acknowledgements. We thank Riccardo Cesaroni for provid­
ing the VLA maps of CCHWK and for carefully reading the
manuscript. Support from C.N.R.--N.A.T.O. Advanced Fellow­
ship program and from NASA's Origins of Solar Systems pro­
gram (through grant NAGW--4030) is gratefully aknowledged.
References
Beckwith S., Sargent A.I., Chini R.S., G¨usten R., 1990, AJ,
99, 924
Brand J., et al. 1994, A&AS, 103, 541
Calvet N., Hartmann L. & Strom S.E., 1997, ApJ, 481, 912
Casali M., & Hawardeen T., 1992, JCMT­UKIRT Newsletter,
No. 4, p. 33
Cesaroni R., Walmsley C.M., K¨ompe C., Churchwell E., 1991,
A&A, 252, 278
Cesaroni R., Walmsley C.M., Churchwell E., 1992, A&A, 256,
618
Cesaroni R., Churchwell E., Hofner P., Walmsley C.M., Kurtz
S., 1994, A&A, 288, 903, CCHWK
Cesaroni R., Felli M., Testi L., Walmsley C.M., Olmi L., 1997,
A&A, in press
Churchwell E., Wolfire M.G., Wood D.O.S., 1990, ApJ, 354,
247
Codella C., Felli M., Natale V., Palagi F., Palla F., 1994, A&A,
291, 261
Codella C., Testi L., Cesaroni R., 1997, A&A, in press
Felli M., Testi L., Valdettaro R., Wang J.­J., 1997, A&A, 320,
594
Forster J.R. 1993, in ``Astrophysical Masers'', eds. A.W. Clegg
& G.E. Nedoluha (Heidelberg, Springer), p. 108
Forster J.R., & Caswell J.L., 1989, A&A, 213, 339
Garay G., Rodeiguez L.F., Moran J.M., Churchwell E., 1993,
ApJ, 418,368
Hofner P., & Churchwell E., 1996, A&AS, 120, 283
Hofner P., Kurtz S., Churchwell E., Walmsley C.M., Cesaroni
R., 1994, ApJ, 429, L85
Hofner P., Kurtz S., Churchwell E., Walmsley C.M., Cesaroni
R., 1996, ApJ, 460, 359
Hunt L.K., Testi L., Borelli S., Maiolino R., Moriondo G., 1994,
Technical Report 4/94, Arcetri Astrophysical Observatory
Hunter T.R., Taylor G.B., Felli M., Tofani G., 1994, A&A, 284,
215
Hunter T.R., Testi L., Taylor G.B., Tofani G., Felli M., Phillips
T.G., 1995, A&A, 302, 249
Jenness T., et al., 1996, MNRAS, 276, 1024
Koornneef J., 1983, A&A, 128, 84
Kurtz S., Churchwell E., Wood D.O.S., 1994, ApJS, 91, 659
Moore T.J.T., Mountain C.M., Yamashita T., McLean I.S.,
1991, MNRAS, 248, 377
Natta A., Palla F., Butner H.M., Evans N.J.II, Harvey P.M.,
1993, ApJ, 406, 674
Natta A. & Panagia N., 1976, A&A, 50, 191
Norris R.P., Whiteoak J.B., Caswell J.L., Wieringa M.H.,
Gough R.G., 1993, ApJ, 412, 222
Palla F., Cesaroni R., Brand J., Caselli P., Comoretto G., Felli
M., 1993, A&A, 280, 509
Palla F., Testi L., Hunter T.R., Taylor G.B., Prusti T., Felli
M., Natta A., Stanga R.M., 1995, A&A, 293, 521
Persi P., Felli M., Lagage P.O., Roth M., Testi L., 1997, A&A,
in press
Persi P., Roth M., Tapia M., Marenzi A.R., Felli M., Testi L.,
Ferrari--Toniolo M., 1996, A&A, 307, 591
Persson S.E. West S.C., Carr D.M., Sivaramakrishnan A., Mur­
phy D.C., 1992, PASP, 104, 204
Rieke G.H. & Lebofsky M.J., 1985, ApJ, 288, 618
Sibille F., Lunel M., Bergeat J., 1974, Astr. and Space Science,
30, 173
Schmidt­Kaler Th. 1981, in Landolt­B¨ornstein, Gruppe VI,
Bond 2, 1 (publ. Springer)
Testi L., 1993, Technical Report 10/93, Arcetri Astrophysical
Observatory
Testi L., Felli M., Persi P., Roth M., 1994, A&A, 288, 634
Testi L., Felli M., Persi P., 1995, Mem. S. A. It., v. 66, p. 677
Testi L., Felli M., Persi P., Roth M., 1997, A&AS, submitted
Tofani G., Felli M., Taylor G.B., Hunter T.R. 1995, A&AS,
112, 299
Walmsley C.M. 1995, RevMexAA SC, 1, 137
This article was processed by the author using Springer­Verlag
L A T E X A&A style file L­AA version 3.