Astron. Astrophys. 332, 1035-1043 (1998)

4. Interpretation

4.1. Unidentified infrared bands

Seven years after the first observation of the UIBs in the spectra of astrophysical sources, Léger & Puget (1984) and Allamandola et al. (1985) suggested that these bands may be attributed to free polycyclic aromatic hydrocarbon (PAH) molecules heated by single photons. Another suggestion was that hydrogenated amorphous carbon (HAC) particles are the carriers of the UIBs. In this case, weakly linked aromatic islands in these solids are responsible for the UIB emission (Duley, 1989). The aromatic islands have to undergo large temperature excursions. However, experimental and theoretical studies contradict the island model (Boutin et al., 1995). Recent experiments on nanoparticles produced IR spectra similar to the UIBs in position, width, and shape of the bands (Herlin et al. 1997; see also Schnaiter et al. 1997). This nanometre-sized carbon dust is a quite promising model for the carriers of the UIBs.

In recent studies of UIB spectra, small or larger PAH molecules (Mattila et al., 1996) over PAH clusters (Molster et al., 1996) up to coal grains (Guillois et al., 1996) are proposed as carriers of the features (and the continuum if present). In order to radiate in the wavelength region in question (3 to 16 µm) the particles must be heated transiently by single or multiple photon heating.

An exciting result of the ISO mission is the presence of UIBs in environments with weak UV radiation fields and the relative similarity of UIB spectra in these cases (see, e.g., Boulanger et al. 1996). Visual or even IR photons can excite the emission of the UIB carriers. To distinguish between PAH molecules or very small hydrocarbon particles one has to consider in detail the excitation and fluctuating temperatures based on the different heat capacities of the possible band carriers. Such a study has to be relied deeply on new laboratory experiments and is beyond the scope of this paper.

In case of strong radiation fields, the UIBs show a much wider variation in intensity ratios and shapes (Roelfsema et al., 1996; Beintema et al., 1996). The spectrum of M 17-North belongs more to this category. Furthermore, the strong rise of the continuum for wavelengths larger than 15 µm points to the emission from VSGs transiently heated by the radiation field (Desert et al., 1990).

Geballe (1997) divided the UIB spectra in four different classes. The spectrum of M 17-North falls into the most common class "A" because of its four principal UIBs at 3.3, 6.2, 7.7, and 11.3 µm .

In M 17-SW several positions from the H II -region to the molecular cloud have been observed with ISO-SWS (Verstraete et al., 1996). Comparing the SWS spectrum of M 17-North to the SWS spectra of M 17-SW our spectrum is similar to the spectrum of the interface region in M 17-SW concerning the relative band intensities and the high band-to-continuum ratio. Verstraete et al. (1996) stated that the intensity of most of the UIBs is not correlated to the 16 µm continuum which is characteristic for single photon heating. Towards the H II region the 8.7 µm band becomes more intense (higher than the 7.7 µm band) and correlates to the 16 µm continuum suggesting that the band originates from VSGs which dominate the continuum at these wavelengths. The band at 13.5 µm is also assigned to VSGs by Verstraete et al. (1996). A similar investigation of M 17-SW with ISOCAM led to the same results (Cesarsky et al., 1996). They suggested that these UIBs are carried by small aromatic grains or large PAHs being in an out-of-equilibrium thermal state. The strong continuum in the H II region is dominated by the emission of VSGs which are bigger than those particles responsible for the UIBs. From the inspection of our SWS spectrum (see Sect.  3.1), we conclude that in M 17-North the physical state of the UIB and continuum carriers should be similar to those in the interface region of M 17-SW observed by Verstraete et al. (1996) with SWS and at the position 3 observed by Cesarsky et al. (1996) with ISOCAM.

4.1.1. Hydrogen coverage of the polycyclic aromatic particles

The bands from 11 to 14 µm are formed by C-H out-of-plane vibrations in an aromatic system. The positions of those bands depend on the number of hydrogen atoms bound to one aromatic ring. There are single, two, three or four adjacent hydrogen atoms (solo, duo, trio, quarto bonds) resulting in features with ascending wavelengths. In the spectrum of M 17-North , these features may be ascribed to the bands located at 11.3, 12.0, 12.7, and 13.5 µm . Therefore, this spectral region provides a specific signature of the location of hydrogen on the aromatic carbon skeleton. For example, when coal matures from semi-anthracite to anthracite the duo, trio, and quarto bands almost vanish because of the release of hydrogen (Guillois et al., 1996). The same is expected if PAH molecules dehydrogenate and Schutte et al. (1993) have proposed a method to estimate the hydrogen coverage from the C-H out-of-plane bands. They derived the number of hydrogen in solo, duo, and trio bonds depending on the hydrogen coverage neglecting the quarto bonds. The ratio of the different bond types in dependence of the total hydrogen coverage can be derived using statistical arguments only. Schutte et al. (1993) just assumed that every bond type is equally likely which is true for large PAHs (). If this assumption holds also for carbonaceous nanoparticles, this method should be equally applicable to VSGs. Assuming that the hydrogen atoms are randomly distributed over the available sites, the number of single, two, and three adjacent C-H bonds as a function of the hydrogen coverage are given by

where is the number of solo, duo, trio C-H bonds () and is the total number of C-H bonds. The quantity is the fraction of the used bonding sites for hydrogen or in other words the relative hydrogen coverage.

The 11 to 14 µm band intensities in the M 17-North spectrum are derived by subtracting a straight baseline at the bottom of each band also if the band sits on a broad plateau. This avoids the difficulty of the uncertain heights of the plateaus. In the next step, the band intensities are divided by the intrinsic band strengths also given by Schutte et al. (1993) to obtain a value proportional to the number of bonds. To determine the ratios of the three bond types to the total number of C-H bonds, the value for each of the three bands is divided by the sum of the three. In Fig. 6 the asterisks display the relative number of bond types derived from the observations at the value for the hydrogen coverage where they fit best the functions. The result is a hydrogen coverage of about 70%.

Fig. 6. The number of the three C-H bond types relative to the total number of C-H bonds as a function of fraction of the used bonding sites for hydrogen (Schutte et al., 1993). The stars indicate the ratio of bond types found from the band intensities in the spectrum of M 17-North . The star for the duo C-H mode is set a little apart to separate the error bars.

4.2. Fine structure lines

Table 1 lists all detected fine structure lines (and Br ) with the measured wavelengths. The ground state of the respective atoms and ions displays fine structure splitting due to spin-orbit-coupling. All the fine structure lines result from forbidden transitions between the levels within the ground state.

Table 1. Fine structure lines in the M 17-North spectrum. The given wavelengths are the observed ones.

Electron collisions excite the atoms/ions. We expect low excitation conditions because of the absence of lines from highly ionized species such as Ne V or O IV. Due to the small excitation energy, the fine structure lines are relative insensitive to the actual electron temperature. Only at low densities the temperature becomes more important. The species, where two lines could be observed, are good indicators for the electron temperature and density (Tab.  2) because the radiation is emitted by the same species and comes from the same area which means equal density of emitting atoms/ions and equal excitation conditions. The ratio of these line intensities is equal to the ratios of the respective emissivities. We use the calculations by Simpson (1975) for the emissivity coefficients of forbidden IR lines.

The intensity of the two [O III ] lines are almost equal, which indicates an electron density of 100-200 cm^-3 independent of the temperature. The intensity ratios of the [S III ] transitions are more sensitive to the temperature. A temperature between 7000 K and 5000 K ¹ and a density between 50 and 200 cm^-3 reproduce the observed ratio. Sulphur is ionized more easily than oxygen (the ionization potential for the second ionization step of sulphur and oxygen are 23.33 eV and 35.12 eV, respectively), so that the radiation in the [S III ] lines could originate from a less ionized region with a lower electron density. As an average, we estimate that the electron density amounts to 100 cm^-3 and the electron temperature is about 5000 K around M 17-North .

The intensity ratios of the transitions of [Ne III ] and [O I ] given in Tab. 2 are at least by factors 3 and 4 smaller than predicted by the theory (Simpson, 1975) for and , respectively.

Table 2. Intensity ratios of the species with two transitions in the IR. The errors represent the uncertainty in the line fits and do not include systematic errors form the flux calibration.

The same considerations for the lines in the off-source spectrum lead to a density of the order of 10 cm^-3 at a low temperature. This value can be deduced from the [S III ] and [O III ] transitions. The ratio of the [O I ] lines almost reaches 10 which is the correct value for low densities and temperatures. The [N II ] line is not intrinsic to the spectrum of M 17-North as it is as bright in the off-source spectrum as in the on-source spectrum.

4.3. Radiative transfer for the continuum

We fitted the SED from the NIR up to the millimetre region by a spherically symmetric RT model. We used the RT code described in Manske et al. (1997). With this code it is now also possible to calculate the contribution from transiently heated VSGs as described by Manske & Henning (1997).

Gatley et al. (1979) pointed out that the FIR luminosity ( ) of M 17-North can be explained by the radiation coming from the centre of M 17 taking into account the distance of and the extent of M 17-North of . We should stress again that the total extent of M 17-North is larger than just the extent of the compact core seen in the millimetre emission. Therefore, our model includes an outer radiation field which provides almost all the luminosity of the cloud core and heats its outer regions. The outer radiation field is treated in an approximate way by a black body radiation field of the temperature . We placed a heating source in the centre of the model cloud to explain the MIR flux. The central heating source is surrounded by a dust shell, which has a constant density up to an radius . From this radius on, the density decreases as . The dust shell ends at the radius . With the parameters given in Tab.  3 the dust shell has a visual extinction of mag and the gas mass of the model cloud including the extended envelope is 6000 . The mass of the core is 100 (integrated up to an radius of ) in agreement with the rough estimates from the millimetre map. The parameters of the internal heating source are the temperature T and its luminosity L. The temperature T should only be taken as a characteristic temperature for the central heating region, it is not the effective temperature of an embedded star.

Table 3. These parameters give a fairly good fit to the observations.

The dust model consists of amorphous carbon, graphite, and silicate particles following a size distribution . The size of the transiently heated graphite and silicate particles ranges from 1 to 10 nm. The size of the amorphous carbon particles ranges from 0.015 to 0.120 µm and of the silicate particles from 0.030 to 0.240 µm . The optical properties for amorphous carbon are taken from Preibisch et al. (1993), for graphite from Draine (1985), and for silicate from Dorschner et al. (1995). A model with the parameters compiled in Tab.  3gives a fairly good fit to the observations.

Fig. 7 shows the observed SED and the model SED. The total luminosity emitted by the model cloud is . In general, the observations do not reflect the total flux but refer to the beamsize. The asterisks denote the flux density which should be received in the different apertures of the observations.

Fig. 7. Comparison of observations with predictions by the RT model.

In Fig. 7 the ISO spectrum (dotted line) is rebinned to a low resolution (R=100) and the off-source spectrum is subtracted. The solid line is the prediction of the model for the ISO observation. The overall structure of the spectrum is reproduced by the model but we were not able to produce such a big change of flux density due to the different apertures of SWS and LWS. Our model produces a flux a little too high between 15 and 45 µm and a little too low for wavelengths longer than 45 µm . The high optical depth of the model also results in the rapidly vanishing flux towards shorter wavelengths suggesting a much higher colour index for IRS1 than observed if it was in the centre of the core. The KAO measurements by Gatley et al. (1979) are well reproduced. The total fluxes coming from a region of size are estimated from IRAS HiRes maps. These fluxes are an order of magnitude higher than the ISO fluxes demonstrating the relatively large contribution from the molecular cloud envelope. Our model fails to explain the 12 µm emission from the envelope which is due to the simplicity of the assumed outer radiation field. The outer radiation field should be produced by the stars powering the H II  region M 17 . Their radiation will be processed by intervening material resulting in a broader SED than just a black-body curve. A "fine-tuning" of the shape of this radiation field makes no sense without additional observational constraints about its nature. Furthermore, the radiation hits M 17-North only from the south, whereas in the model an isotropic radiation field is assumed.

The observed fluxes at 1.3 mm are the peak flux density and the total flux density of 3.5 Jy emitted by the core (cf.  3.2). The asterisks again denote the predictions by the model which fit the observed data very well. In general, we can conclude that the RT code with VSGs reproduces the broad-band spectral energy distribution of M 17-North . The expected reddening from the model is higher than observed for IRS1, which points to a more complicated geometry which would especially influence the NIR part of the SED.

4.4. The central source

The core of M 17-North contains the embedded infrared source IRS1 which probably contributes to the total luminosity of M 17-North . Evidence for ionizing radiation and therefore for the presence of an early-type star comes from radio observations (Wilson et al., 1979), and our detection of the relevant fine structure lines and Br emission.

However, the luminosity of the central core region ( in the millimetre map) is only 20% of the total luminosity. It was not possible to produce the relatively bright extended envelope emission seen in the comparison of the IRAS/ISO data and the aperture effects in the ISO data without external heating in the RT model. Furthermore, the colour index of IRS1 had to be much higher than observed according to the RT calculations, if it would really be the central object of the cloud. The intensity of the Br line and the fine structure lines may also be explained by the ionizing radiation field emitted from the central cluster in M 17 . The intensity of this radiation field could reach times the standard interstellar intensity around M 17-North taking into account the total luminosity of the OB cluster in M 17 and the geometric dilution.

© European Southern Observatory (ESO) 1998

Online publication: March 30, 1998
helpdesk@link.springer.de