Observations of the 3.3-μm UIR band in the Red Rectangle: relation to unidentified optical emission

Abstract

The biconical Red Rectangle nebula exhibits very strong unidentified infrared (UIR) emission bands, a subset of the optical diffuse interstellar bands (in emission) and extended red emission (ERE). A key question is the extent to which the carriers of these spectroscopic signatures may be related. In a new study of the 3.3-μm emission, CGS 4 spectra were recorded at UKIRT, which give information on the spatial distribution of the 3.3-μm carrier in the nebula and on the width, peak wavelength and profile of the feature as a function of offset from the central star, HD 44179. Both Type 1 (Λ₀∼ 3.289 μm, full width at half-maximum FWHM ∼ 0.042 μ m) and Type 2 (Λ₀∼ 3.296 μm, FWHM ∼ 0.020 μm) 3.3-μm features, as defined by Tokunaga et al., are found within the nebula. Type 2 is seen predominantly towards the central star, at the bicone interfaces and east and west of the star in the nebula. The broader Type 1 feature appears in the nebula 5 arcsec south of the central star, whereas the 3.3-μm band at 5 arcsec north appears to be a blend of Type 1 and Type 2. We find that there is no significant correlation between the intensity of the 3.3-μm feature and that of either the unidentified optical (diffuse) emission bands or ERE. This result suggests that there is at most an indirect link between the carrier(s) of the 3.3-μm band and this subset of diffuse bands. Such a link could arise, for example, if these diffuse band carriers were formed by chemical erosion or photodissociation of polycyclic aromatic hydrocarbon material.

stars: individual: Red Rectangle, ISM: abundances, ISM: molecules, infrared: ISM: lines and bands

1 Introduction

One of the major challenges in molecular astrophysics is the identification of the carriers of the unidentified infrared (UIR) emission bands, extended red emission (ERE) and the diffuse interstellar absorption bands (DIBs). The UIR bands and ERE are prominent in a wide range of objects that are associated with sources of ultraviolet excitation, but have also been observed in the general interstellar medium (ISM) where the radiation field is relatively low. The diffuse interstellar absorption bands are found in near-ultraviolet, visible and near-infrared spectra recorded towards stars that are reddened by interstellar dust. The widespread appearance of these emission and absorption spectroscopic signatures is indicative of ubiquitous interstellar material but the exact nature of the carriers is uncertain. In each case there is debate as to whether the carriers are in the gaseous or condensed phase, or in some intermediate ‘state’ such as clusters. A key question is the extent to which the carriers of these spectra may be related chemically, spectroscopically and spatially.

The UIR bands exist throughout the infrared, the most prominent of which occur near 3.3, 6.2, 7.7, 8.6 and 11.3 m. They are most commonly attributed to polycyclic aromatic hydrocarbon (PAH) molecules or PAH-type material from which infrared (IR) emission occurs after the absorption of ultraviolet (UV) radiation (see e.g. Allamandola & Tielens 1989; Sellgren 1994; Molster et al. 1996; Joblin 1998 and references therein) although other possible excitation mechanisms have been put forward (Guillois et al. 1998). Some of the bands have been detected spectroscopically by the Infrared Space Observatory (ISO) in the diffuse ISM (Mattila et al. 1996). The UIR bands are generally thought to arise from a superposition of transitions in different species, the 3.3-μm band being attributed to the fundamental C—H stretch motion of PAH molecules or material. This assignment and the photoexcitation mechanism have been explored in the laboratory for moderate-sized neutral gas-phase PAH molecules by UV laser excitation followed by detection of IR emission (Cherchneff & Barker 1989; Cook et al. 1998 and references therein).

ERE is found in a wide range of objects (Witt & Boroson 1990; Sivan & Perrin 1993) and in the diffuse interstellar medium (Gordon, Witt & Friedmann 1998). It is ascribed to fluorescence or phosphorescence of condensed-phase PAH/HAC (hydrogenated amorphous carbon) (Duley 1985; Duley & Williams 1988; Duley, Seahra & Williams 1997) as detected in laboratory experiments (Watanabe, Hasegawa & Kurata 1982; Furton & Witt 1993), to quenched carbonaceous composites (Sakata et al. 1992), or to emission from gas-phase molecules (d'Hendecourt et al. 1986; Miles & Sarre 1993; Mattila et al. 1996).

The diffuse interstellar absorption bands appear largely in the visible part of the spectrum and arise from unidentified carriers in diffuse clouds (Herbig 1995). Among the many proposals for their origin, electronic transitions in PAH molecules have been discussed in some detail (see Salama et al. 1996 and references therein). There is evidence that some of the bands that belong to ‘family 3’ in the classification of Krelowski & Walker (1987) are present in emission in some parts of the Red Rectangle nebula (Sarre, Miles & Scarrott 1995) and in the R CrB star V854 Cen at minimum light (Rao & Lambert 1993). In this paper we report new spectroscopic observations of the 3.3-μm band in the Red Rectangle (AFGL 915) and focus on the carrier distribution in the nebula. Comparison is made with the distribution of the unidentified optical emission bands and ERE in the Red Rectangle as determined by Schmidt & Witt (1991).

2 The Red Rectangle Nebula and the Central Star Hd 44179

The Red Rectangle nebula (Cohen et al. 1975) has a remarkable biconical geometry, a schematic diagram of which is given in Fig. 1. At the heart of the nebula lies an iron-deficient star (Waelkens et al. 1992), classified as B9—A0 by Cohen et al., and firmly established as a spectroscopic binary (Van Winckel, Waelkens & Waters 1995; Waelkens et al. 1996). Recent ISO spectra have revealed oxygen-rich material including crystalline silicates in a circumbinary disc (Waters et al. 1998). In contrast, the X-shaped biconical nebula, which is of principal interest in this work, extends over 1 arcmin (Cohen et al. 1975) and contains carbon-rich material. These unusual characteristics have led to many studies of the morphology, nature and evolution of the star and its associated circumstellar and nebular material (Grasdalen, Sloan & LeVan 1992; Sloan, Grasdalen & LeVan 1993; Bregman et al. 1993; Lopez, Mékarnia & Lefèvre 1995; Roddier et al. 1995; Jura, Balm & Kahane 1995; Cruzalèbes et al. 1996; Hora et al. 1996; Jura, Turner & Balm 1997; Lopez et al. 1997). The nebula is unique in that strong UIR, ERE and unidentified optical emission bands are all observed in one object. Because it has a well-defined geometry and the location of the exciting source is known, the Red Rectangle offers a good opportunity for the investigation of possible relationships between the carriers of UIR, ERE and some of the diffuse bands. A brief introduction to the observations of each of these spectroscopic features in the nebula is now given.

Figure 1.

The hollow bicone geometry of the Red Rectangle centred on the star HD 44179.

In a survey of the UIR 3.3-μm feature in a range of sources including the Red Rectangle, Tokunaga et al. (1991) found that it could be classified into two types. Type 1 is found to occur commonly in extended objects including planetary nebulae and H II regions with a central wavelength (Λ₀) of 3.289 m and a full width at half-maximum (FWHM) of approximately 0.04 m. Type 2 is narrower (FWHM ∼ 0.02 m) with a peak wavelength of Λ₀= 3.296 m and is present in a number of stellar sources. It is less common and apparently occurs in environments with T< 10⁴ K. Early observations of the 3.3-μm feature in the Red Rectangle revealed a complex picture with evolution from Type 2 near the central star (HD 44179) to the broader Type 1 feature at larger distances from the star (Geballe et al. 1989; Tokunaga et al. 1991). This was inferred from recordings with small (2.7 arcsec) and larger (5 arcsec) apertures, as well as from observations 5 arcsec north of HD 44179 in the nebula. These results implied that excitation conditions, compositional changes, or a combination of these, may affect the width of the feature (Tokunaga et al. 1991).

The unidentified visible emission bands occur in three main groups near 5800, 6380 and 6615 with wavelengths that lie close to those of the prominent narrow ΛΛ5797, 6376/9 and 6614 diffuse absorption bands (Sarre 1991; Fossey 1991; Scarrott et al. 1992). By examining the evolution of the optical emission features as a function of offset from the central star, Sarre et al. (1995) showed that these bands arise from a subset of the diffuse interstellar absorption bands. The behaviour of the bands from hot to cooler regions is reminiscent of that for rotational band contours in molecular electronic spectra and is therefore consistent with the hypothesis of an origin in free gas-phase molecules rather than in grains. This is supported by the results of a computational contour modelling study of the Λ6614 diffuse band observed at ultrahigh resolution in absorption (Kerr et al. 1996). In an earlier study of the Red Rectangle, Schmidt & Witt (1991) showed that the intensity maxima of the optical diffuse band emission feature near 5800 (labelled F1 in their paper) lies along the north-east—south-west (NE—SW) and north-west—south-east (NW—SE) interfaces of the nebula. They proposed that the carriers of the group of 5800 emission features are produced by erosion of the nebular material.

Finally, a detailed study of ERE in the Red Rectangle has been made (Witt & Boroson 1990; Schmidt & Witt 1991). It was found that the ERE-producing material was apparently more efficient than in other ERE-emitting regions, and that it was confined largely to the whiskers, as for the 5800- feature. However, although the spatial distributions of ERE and the 5800- feature were similar, they were not identical. In particular the ERE had its maximum intensity in the walls of the biconical structure. It was concluded that the relatively broad ERE and the sharper features near 5800 have distinct but possibly related origins.

The UKIRT observing programme described here was designed to allow comparison between the new 3.3-μm data and the published data of Schmidt & Witt (1991) on the spatial distribution of ERE and the optical bands. The slit positions and orientations were therefore selected to match those adopted in their observations. In this paper 3.3-μm observations are described in Section 3 and the results are presented in Section 4. A discussion including the distribution of the UIR emission in comparison with the optical bands is presented in Section 5 and conclusions are given in Section 6.

3 Observations

Observations were made at the United Kingdom Infrared Telescope (UKIRT) on Mauna Kea, Hawaii, using the cooled grating spectrometer CGS 4. The short-focal-length camera was employed (F = 150 mm) with the 75 line mm⁻¹ grating and a 256 × 256 InSb array to give a resolving power (Λ/ΔΛ) at 3.3 m approaching 1300. The wavelength coverage of ∼ 0.65 m enabled a detailed examination of profile shape, Λ₀ and FWHM to be made. Spectra were taken with the 90 arcsec long slit (width 1.2 arcsec) at various position angles (PA) centred on the central star (PA = −85), 5 arcsec and 10 arcsec east (PA = 5), north and south of HD 44179 (PA = −85), and also along the NE—SW and NW—SE ‘whiskers’ (PA = 43) or ‘bicone walls/interfaces’ (see Figs 1 and 2). A 32 × 256 subarray was used, limiting the effective slit length to 40 arcsec. The telescope was nodded 60 arcsec along the slit in order to account for telluric features; such a distance also took the sky beam out of the nebula itself. Both early-type and F-type standard stars were observed and a log of the observations is presented in Table 1.

Figure 2.

A sketch of the geometry of the Red Rectangle showing some of the adopted slit positions 5 arcsec east, 5 arcsec west (PA =5) and along the north-east—south-west whisker (PA =43). See Table 1 for full details of slit alignments.

Table 1.

Observing log.

Wavelength calibration was achieved by comparison with an argon lamp and is considered accurate to ∼ 0.001 m. The spectra were divided by an appropriate spectrum of a nearby standard star and then multiplied by a Planck function appropriate to the temperature of the standard. In order to measure equivalent widths (W_Λ) of the 3.3-μm feature, the spectra were normalized with reference to a second- or third-order polynomial continuum, determined between the wavelengths 3.0 to 3.2 m and 3.35 to 3.7 m (avoiding the 3.4-m emission feature when present). In all cases, the optically thick and variable telluric CH₄ line at 3.33 m was removed.

4 Results

4.1 3.3-μm band characteristics at specific positions in the Red Rectangle

The measured peak wavelengths, FWHM, equivalent widths (where possible) and deduced Type 1/2 classification for the 3.3-μm band for a number of positions both towards the central star and in the nebula are presented in Table 2. In each case the table entry pertains to an aperture of c. 1.2 × 1.2 arcsec². It is found that the Type 2 feature is the predominant form both towards the central star and in the nebula, exceptions being at 5 arcsec south where the broader Type 1 is present (see Fig. 3 for comparison), and 5 arcsec north where a blend of Type 1 and Type 2 is suspected (see Fig. 4). For both 5 arcsec north and south a feature at 3.4 m is evident.

Table 2.

Profile parameters and Type 1/2 classification for the 3.3-μm UIR band towards HD 44179 and at a range of positions in the nebula.

Figure 3.

Spectra of Type 2 towards the central star HD 44179 (dots), and of Type 1 in the nebula 5 arcsec south of the star (small crosses). Note the 3.4-m feature in the nebular spectrum.

Figure 4.

The 3.3-μm UIR band recorded 5 arcsec north of HD 44179. Although Λ₀ is consistent with a Type 2 profile, the width is larger than usual, and the feature appears to consist of a blend of Type 1 and Type 2.

At 5 arcsec south the maximum band intensity of the 3.3-μm feature occurs at a relatively short wavelength (3.286 m) and it is much broader (FWHM ∼ 0.041 m). There is a possibility of some contribution at 3.298 m (the Λ₀ of the band towards HD 44179 and elsewhere in the nebula) from a Type 2 feature. It should be noted that this wavelength is also that of the PfΔ hydrogen recombination line, and although this line is very weak or absent in this nebula (Geballe et al. 1989), its contribution to the UIR band cannot be totally ruled out, especially at the higher resolution of the spectra presented here. At 5 arcsec north, the characteristics of the 3.3-μm UIR band agree well with those reported by Geballe et al. (1989). The central wavelength Λ₀ remains unchanged from that towards HD 44179, but the feature is broader because of an apparent extra contribution to the short-wavelength wing. It should be noted that this spectral region suffers greatly from contamination by telluric water and CH₄ lines. These are difficult to remove completely, and some effect from these features cannot be entirely dismissed, despite careful attempts at their cancellation.

A summary of possible interfering stellar and telluric features in the 3.3-μm spectral region is given with respect to the 5 arcsec south data in Fig. 5. Telluric features marked include those of H₂O and CH₄ and also many hydrogen (Humphrey series) and He II recombination lines; these could influence the shape and fine structure within the spectra.

Figure 5.

The 3.0–3.7 m spectrum taken at 5 arcsec S, with many of the possible telluric (lower marks), stellar and interstellar lines marked.

4.2 3.3-μm band characteristics and evolution along the biconical interfaces

Spectra of the 3.3-μm spectral region obtained along the north—east and south—west interfaces are shown in Figs 6(a) and (b) (see also Figs 1 and 2). The 3.3-μm feature extends beyond the 6 arcsec offsets shown but it is very weak beyond these points, and little information about the profile can be obtained. However, a measurement of the approximate equivalent width could be made. Inspection of Table 3 shows that Λ₀ remains effectively constant along the interfaces and, except for a reduction in the width and peak flux, the 3.3-μm band is essentially unchanged from that observed towards the central star. Little or no emission at 3.4 m is seen along the interfaces, in contrast to a previous observation by Geballe et al. (1989). They had indicated that this feature becomes relatively more prominent in the Red Rectangle with increasing offset as the 3.3-μm feature weakens. However, Geballe et al. observed the feature 5 arcsec directly north of HD 44179, and their study therefore did not include the interfaces. This has also been commented on briefly by Sloan, Bregman and Woodward (1994).

Figure 6.

The 3.3-μm spectra along the north-east and south-west interfaces out to approximately 6 arcsec, with the spectra towards HD 44179 for comparison. The spectra in (b) (right column) are normalized to a common peak intensity in order to allow comparison of the band profiles.

Table 3.

Profile parameters for the NE and SW interfaces.

The lack of variation in Λ₀ along the bicone interfaces is striking, and extends out to at least 10 arcsec from the central star. The optical emission bands reported by Scarrott et al. (1992) in the Red Rectangle, and shown to be related to a subset of the diffuse interstellar bands (Sarre et al. 1995), show a systematic change in both Λ₀ and FWHM with increase in offset from HD 44179. On extrapolation they converge towards the diffuse interstellar band absorption wavelengths and widths in the limit of large offset. However, if the wavelength shift here scaled in proportion to the observed wavelength (c. 1 part in 10³) this would not be easy to detect in the IR spectra at the current resolving power. In contrast to the invariance of the peak wavelength, there is some evidence for a reduction in FWHM with increase in offset along all four whiskers (see Table 3). As Fig. 6(b) shows, telluric contamination is of increasing importance as the signal weakens. However, we consider that the effect is real and that it is unlikely to arise from incomplete cancellation of telluric features, or from possible Pf Δ emission as the Hα emission line is evident only very close to the central star. If the IR profile arises from gas-phase molecular carriers, a decrease in width can be rationalized in terms of a reduction in the number of rotational lines that contribute to the feature as the temperature declines with increase in offset from the star.

A notable difference between the 3.3-μm and optical/ERE data is that the IR (C—H bond) feature is about twice as strong along the NW—SE interface compared with the NE—SW interface, whereas the opposite is the case for the optical emission features (see Tables 3 and 4; Schmidt & Witt 1991). A possible interpretation is that the optical emission carriers (and the carrier(s) of ERE) are less hydrogenated relative to the PAH IR carriers.

Table 4.

Profile parameters for the NW and SE interfaces.

4.3 3.3-μm spectra recorded with slit offset 5 arcsec north, south, east and west from HD 44179

Some of the most impressive results from the study of the spatial distribution of ERE and the 5800- optical emission and other features by Schmidt & Witt (1991) were achieved by aligning the slit 5 arcsec east and west of the central star (see Fig. 2). We adopted the same slit alignments for this IR study, and also recorded data 5 arcsec north and south of HD 44179. The 3.3-μm UIR band intensities measured along the four slit axes are shown in Fig. 7, where the vertical lines indicate the points of intersection between the slit and the bicone interfaces. The 3.3-μm distributions are highly symmetrical about the point of closest approach to HD 44179, and, with the exception of 5 arcsec south, are smooth profiles. The continuous lines are least-squares fits for an r⁻³ function, which holds for isotropic (stellar) irradiation of material emanating from the central source (HD 44179), and described by an r⁻² density fall-off law (Mauron 1997). Omission of the data points between the bicone axes in the 5 arcsec south data leads to a fit that is very similar to those in the other three quadrants. From the high quality of the fits it may be inferred that the model of a photoexcitation mechanism and an r⁻² spatial distribution of the 3.3-μm emitting material is reasonable. It is notable that the principal discrepancy occurs for the data 5 arcsec due south (between the bicone axes), which is the region that exhibits the change in spectral character to Type 1 with an associated 3.4-m feature. The slit positions used in obtaining our spectra at 5 arcsec offset (north, south, east and west) are on the edge of the region reported in an imaging study by Hora et al. (1996) in which evidence for slight enhancement along the walls of the bicone was found for UIR features between 8 and 13 m. Our results may also be compared with the relatively symmetrical 3.3-μm image described by Bregman et al. (1993) in which the data extend about 3.5 arcsec from HD 44179 and show a central maximum and a slight elongation in the north—south direction.

Figure 7.

Integrated intensities of the 3.3-μm UIR band along the slit at positions 5 arcsec west, south, east and north. The vertical lines mark the points of intersection of the slit with the bicone interfaces. The solid curves were obtained by least-squares fitting to an r⁻³ model distribution (see text), where r=(h²+r₀²)^1/2, where h is the distance along the slit direction measured from the point of nearest approach to the central star (in arcsec) and r₀=5 is the value of the slit offset (in arcsec). The difference between the distributions of integrated intensity and peak intensity of the 3.3-μm band is negligible.

The lack of a direct connection between the 3.3-μm band and the unidentified optical emission is clearly illustrated in Fig. 8 where the IR and optical data for ERE, the 5800-Å (F1) and Na I features are plotted. The optical emissions are very pronounced near to the intersections between the slit and the bicone interfaces, whereas the 3.3-μm distribution is very different, being centred symmetrically about the point of closest approach to the star. It appears, therefore, that there is no direct link between the carriers of the 3.3-μm UIR band and the carriers of the optical emission bands.

Figure 8.

Comparison between the intensity distribution of the 3.3-μm band (this work) and the ERE, F1 (5800 Å) and Na I emission recorded by Schmidt & Witt (1991).

5 Discussion

In this section the spectroscopy and spatial distribution of the 3.3 and 3.4-m emission bands, and their relation to the unidentified optical emission bands and ERE, are discussed. Summarizing the IR data, the narrower (Type 2) 3.3-μm band is seen towards the central star, at the bicone interfaces and east and west of HD 44179. The broader Type 1 feature appears 5 arcsec south of the central star and the 3.3-μm feature at 5 arcsec north can be described as a blend of Type 1 and Type 2. Over a range of Galactic and extragalactic sources the Type 1 profile is much more common, and, at least across the Orion Bar region, the Type 1 shape does not appear to vary with the degree of ionization of the medium (Sloan et al. 1997). The Red Rectangle is unusual in that Type 2 is found to predominate. There appears to be a link between the occurrence of the 3.4-m feature and the Type 1 3.3-μm profile in the Red Rectangle as the 3.4-m band is present 5 arcsec south and (weakly) 5 arcsec north of HD 44179, whereas it is absent in those regions with a Type 2 profile. In contrast to the IR feature, the unidentified optical emission bands and ERE occur in a well-defined region along the bicone interfaces.

The 3.4-m feature is most commonly ascribed to the C—H stretch of aliphatic side-groups on the PAH frame (Jourdain de Muizon, d'Hendecourt & Geballe 1990; Joblin et al. 1995, 1996) and/or superhydrogenated PAHs (Schutte, Tielens & Allamandola 1993; Bernstein, Sandford & Allamandola 1996; Sloan et al. 1997). Other proposals include alkanes (Pinho & Duley 1995), nitrogen-containing or nitrogen-related species (Roche et al. 1996) and transitions between higher-lying vibrational levels (Barker, Allamandola & Tielens 1987) although these would be expected to appear at longer wavelength (3.43 m) than is observed here (Geballe et al. 1994). Given the lack of spatial correlation between the 3.4-m (and Type 1) features and the unidentifed optical bands, and the result of Section 4.2 where the relative strengths of the C—H IR emission and the optical emission were seen to be anticorrelated, it is inferred that the optical emission is unlikely to arise from strongly hydrogenated molecules or material. This is consistent with the very widespread occurrence of the Type 1 and 3.4-m features in other objects for which the optical diffuse bands are absent in absorption or emission. However, it represents the first such information derived from study of a single object.

Although the carriers of a Type 2 profile and the optical bands coexist along the bicone axes, there can be no direct correspondence between them because the UIR Type 2 emission is also seen east and west of HD 44179 whereas the optical bands are absent from these regions. Schmidt & Witt (1991) have suggested that the occurrence of ERE and the optical (F1) emission requires chemical erosion of the grains as well as optical excitation by the star. Assuming that Type 2 arises from less strongly hydrogenated material than Type 1, could additional dehydrogenation by reactions at the bicone interfaces leave (relatively) bare carbon molecules as the optical emission (and therefore diffuse band) carriers? This would be consistent with the general non-observation of diffuse band spectra in UIR-emitting regions, but would retain a link with carbon-rich and most probably PAH material. Complete loss of hydrogen from the carbon frame would of course mean that the 3.3-μm emission would not appear. In fact the optical emission extends to far greater offset along the whiskers than do the UIR bands. It should be emphasized that, in contrast to the IR spectra of PAH molecules (material) in which, for example, the C—H stretch appears at a similar frequency whatever the details of the molecular structure, electronic spectra vary enormously according to the size, shape and charge state of the molecular carriers. This means that the optical bands are probably carried by rather few ‘special’ molecules.

We now consider the possible identity of molecules arising from PAH molecule/material decomposition. In a recent experiment by Scott, Duley & Pinho (1997), the masses of the gaseous molecular ions produced by UV laser irradiation of a film of hydrogenated amorphous carbon (HAC) were recorded by time-of-flight mass spectrometry. Although not expected to be identical, the ion mass spectrum should also be a good representation of the gaseous neutral molecules generated. The molecular distributions were found to be strongly dependent on the number and fluence of the laser pulses, and on the irradiation history of the surface. In addition to the initial release of low-mass molecules with up to 20 carbon atoms, which occurred during the first few laser pulses, a wide range of higher-mass molecules with over 30 carbon atoms and, under certain conditions, clusters with up to several hundred carbon atoms were detected. We suggest that the most resilient molecules (and by implication those present in the Red Rectangle) would be those found after substantial irradiation. Although the closest astrophysical parallel would be with shocked or photon-dominated regions, the results may nevertheless indicate what happens in the Red Rectangle where erosion, possibly by chemical reactions, is combined with UV irradiation. This may also be similar to the conditions under which diffuse band carriers form in diffuse clouds.

Under conditions of high fluence and after three laser pulses the mass distribution was found to be characterized by three narrow peaks corresponding to molecules containing 10, 14 (strong) and 18 carbon atoms, and a broader distribution of masses with between c. 60 and 110 carbon atoms each separated by two-carbon-atom units (m=24), but with no favoured mass number. It was suggested that the low-mass peaks arose from molecules such as naphthalene (C₁₀H₈), anthacene (C₁₄H₁₀), etc. Another possibility is that these peaks arise from pure monocyclic ring molecules, which are particularly stable 4n+2 π-electron systems with no dangling bonds. We have discussed their attributes as potential carriers of some of the diffuse bands in an earlier paper (Kerr et al. 1996). Monocyclic carbon rings have also been observed in the laser vaporization of graphite via the photoelectron spectra of their anions (Yang et al. 1988), mass spectrometrically in a laser vaporization/photolysis experiment on graphite in which the C₁₀⁻ ion was found to be particularly persistent (Wakabayashi et al. 1997), and in photodissociation experiments where their cross-section for fragmentation was observed to be low (Geusic et al. 1987). A second result of the experiment of Scott et al. (1997) under multiple but low-fluence laser pulses was the discovery of prominent mass peaks with c. 38–44 carbon atoms and described as ‘protographitic clusters’ following earlier IR and scanning tunnelling microscopy studies (Scott & Duley 1996). This is quite a surprising result as they do not appear to be ‘magic’ clusters, although we note that they do come at the lower end of the set of fullerenes observed in laser vaporization experiments.

6 Conclusions

We have observed the 3.3-μm UIR band in several regions of the Red Rectangle, including the central star (HD 44179), the bicone interfaces and the nebula, in order to investigate the variations of the band throughout the region. We find that there is no significant correlation between the appearance of Type 1 or Type 2 (3.3-μm) or 3.4-m features with the optical emission bands or ERE. Although there is little evidence for a direct link between the carriers of the 3.3-μm UIR band and the optical emission bands, it is suggested that the results are consistent with the diffuse band carriers in the Red Rectangle being substantially if not wholly dehydrogenated carbon molecule products arising from the breakdown of PAH molecules or material. The Red Rectangle results are discussed in connection with data from recent laser experiments on the vaporization of HAC. This provides evidence for low-mass products possibly corresponding to monocyclic rings, as well as for more massive products with 40 or more carbon atoms. For any of these potential carriers, the combination of these new data from the Red Rectangle with the recent mass spectrometric experiments indicates a clear need and motivation for laboratory (electronic) spectroscopy on the molecular species released from PAH/HAC samples.

Acknowledgments

We thank the UK Panel for the Allocation of Telescope Time for the award of observing time on UKIRT, Dr T. Geballe for assistance, the UK Particle Physics and Astronomy Research Council (PPARC, ref. GR/J33357) and The Royal Society for research grants, PPARC for a Fellowship to JRM, and the UK Engineering and Physical Sciences Research Council for a studentship to MEH. We are most grateful to Professor A. N. Witt for providing the data from the optical study of the Red Rectangle.