OH masers at 1612 and 1720 MHz in star-forming regions Free

Abstract

Masers at the ground-state OH satellite transitions near 1612 and 1720 MHz are occasionally found in star-forming regions, accompanying the dominant maser of OH at 1665 MHz. The satellite lines can then be valuable diagnostics of physical conditions in star-forming regions if we can first ascertain that all maser species truly arise from the same site. For this purpose, newly measured satellite line positions with subarcsecond accuracy are reported here, and compared with masers of main-line OH at 1665 MHz, with methanol masers at 6668 MHz, and with ultracompact H ii regions. We confirm that most of the satellite-line OH masers that we have measured are associated with star-forming regions, but a few are not: several 1612-MHz masers are associated with late-type stars, and one 1720-MHz maser is associated with a supernova remnant. The 1720-MHz masers in star-forming regions are accounted for by a pumping scheme requiring high densities, and are distinctly different from those in supernova remnants where the favoured pumping scheme operates at much lower densities.

1 introduction

Molecular masers of OH and methanol delineate sites of massive star formation in their earliest stages. At these stages, the optical emission of the star is concealed by dust, and its presence must be inferred indirectly, from infrared emission near the 60-μm band (from heated dust around the star), or from radio continuum emission (from an enveloping ultracompact H ii region, hereafter ucH ii). However, the masers are often the most useful tracer, since the IRAS 60- and 100-μm surveys are badly confused in the Galactic plane, and compact radio H ii regions at low flux densities are difficult to recognize unless other clues, such as maser emission sites, are available to guide a targeted search. The masers also offer high-resolution spatial and dynamical information (including evidence of large-scale mass outflow and perhaps rotation), and magnetic field measurement (through Zeeman splitting) at the star-forming site, as well as immediately providing a systemic velocity and, hence, an estimate of the distance to the site.

Where masers are detectable in several species (e.g. OH and methanol) and on several transitions of each species, they can probe in further detail the physical conditions around newly born stars. Here, we investigate the presence of 1612- and 1720-MHz OH masers in some star-forming regions (SFRs). Masers at these transitions are better known for their occurrence in quite different contexts: the ground-state OH satellite transition at 1612.231 MHz is most commonly found in late-type stars (e.g. Sevenster et al. 1997a,b), and the satellite transition at 1720.530 MHz usually delineates shock fronts associated with supernova remnants (e.g. Green et al. 1997). It is only occasionally that one or other of these varieties of OH maser is found in an SFR, accompanying the dominant masers of OH at 1665 MHz and methanol at 6668 MHz. When the satellite-line masers do occur, they can be especially valuable for diagnosing physical conditions in SFRs.

Comparisons of ground-state OH maser positions with each other and with radio continuum emission were made for some northern maser sites by Gaume & Mutel (1987). For the southern sky, preliminary investigations of this kind have been reported (Caswell 1998) using a sample of SFRs defined by a comprehensive catalogue of southern OH masers which contained new precise positions of OH 1665-MHz masers and 6668-MHz methanol masers. These positions were compared with published (mostly lower precision) data on the OH satellite lines. The comparisons suggested that maser sites displaying the 1665-MHz transition can also show either 1612- or 1720-MHz emission; indeed, it seems unlikely that both could arise from precisely the same region. Furthermore, it appears that the sites with 1720-MHz emission also host methanol masers, whereas those with 1612-MHz emission frequently do not host methanol masers. Close association of 1720-MHz OH with 6668-MHz methanol masers was also remarked on for the source W3(OH) by Masheder et al. (1994).

Methanol masers may be most prominent at a stage in massive star evolution somewhat earlier than the OH-dominant phase (Caswell 1997). If the methanol masing requirements are similar to those of 1720-MHz OH, then the demise of the methanol and 1720-MHz OH may result from the same changes that favour the onset of 1612-MHz emission. A brief exploration of these implications will be considered, but the present objective is chiefly to ascertain which of the apparent associations are real, since this is a prerequisite before any interpretation is justified.

The investigation is especially timely in view of several recent developments: most notably, the renewed investigation of 1720-MHz masers in supernova remnants, which we argue in Section 6 are a variety of maser quite distinct from those in SFRs, and extensions of OH maser pumping theory by Pavlakis & Kylafis (1996).

2 observations and data reduction

The majority of the 200 known southern SFR OH maser sites have been searched for 1612- and 1720-MHz masers with low-precision single-dish measurements, and the objects studied here include all of those for which a detection has been reported.

Observations were made on 1997 November 27 with the Australia Telescope Compact Array (ATCA). The six antennas yielded 15 baselines between 153 and 6000 m. The correlator provided a 2048-channel spectrum across a 4-MHz bandwidth for each of two orthogonal linear polarizations. A 2-h observing sequence, with target fields and calibrators interspersed and observed for about 4 min each, was repeated (with minor variations) seven times. The uv-coverage on each target was thus well distributed over 12 h, and yielded a total integration time of 28 min on each field. In the 1721.0-MHz data, weak interference was occasionally encountered at 1720.5 MHz, and in the 1612.0-MHz data some interference from Glonasssatellites was present, fortunately confined to the edge of the band, near 1614 MHz.

The aips package was used for data reduction. Hanning smoothing was applied to the full data set, to give a frequency resolution of 3.9 kHz (= 0.68 km s⁻¹ at 1720 and 0.73 km s⁻¹ at 1612 MHz). After calibration, cvel was used to correct for Doppler shifting caused by the Earth's motion. After inspection of each spectrum, uvlsf was used to remove any underlying continuum emission, and total-intensity maps (the sum of the two linear polarizations) were then made of the channels showing maser emission. The synthesized beam has a RA width to half-power of 7 arcsec, and a declination width larger by the factor cosec (Dec.). The rms noise level is typically 30 mJy on any individual channel map. Positions of strong maser spots on the maps were measured using imfit, and have rms uncertainties of 0.4 arcsec, arising chiefly from residual calibration errors. For sources as weak as 0.3 Jy, the uncertainty is increased to 0.6 arcsec owing to a comparable error introduced by noise.

3 results

The data yielded positions for 17 masers: 10 at 1720 MHz (one of them previously measured to similar accuracy, and observed as a check source) and seven at 1612 MHz (four of them previously positioned to similar accuracy). The positions of the strongest feature at each of the maser sites are given in Tables 1 and 2, with typical rms uncertainties of 0.4 arcsec. Table 1 includes only the masers believed to arise in SFRs; and for the convenience of the subsequent discussion, it also lists positions of 11 other OH satellite-line masers of similar type that are already known to coincide to arcsecond accuracy with the corresponding main-line OH masers. For each source, we list the positions of the satellite line (1612 or 1720 MHz) and the main line (1665 or 1667 MHz) counterpart. Where a new measurement confirms an earlier satellite-line position of similar accuracy (Gaume & Mutel 1987; Sevenster et al. 1997a,b), a reference is given (in parentheses) to the earlier measurement. The peak flux density of any methanol maser at the same site, and the flux density of any associated ucH ii region (at a frequency of typically 6 GHz - see Caswell 1998) are also listed. Table 2 lists positions of 1612-MHz masers not of the SFR variety. Spectra from the present observations are shown in Fig. 1 for 12 sources in Table 1 and in Fig. 2 for four sources in Table 2. The fortuitous measurement of a 1720-MHz maser associated with a supernova remnant is discussed at the end of Section 4.

4 sfr maser sites with oh satellite-line emission

This section first summarizes present and previous data on the newly measured maser sites.

306.322-0.334. The 1720-MHz maser was first reported by MacLeod (1997), with flux density of 1 Jy. The accurate position measured here lies close to the 1665-MHz maser, with a nominal offset to smaller RA by 0.047 s (= 0.32 arcsec), and north by 2.0 arcsec. The 1665-MHz emission is weak and, as a consequence, its position (Caswell 1998) has an uncertainty that is larger than for most masers in the survey, and thus the 2-arcsec discrepancy is not significant. A methanol maser apparently coincides, although its position estimate, from the Parkes telescope, has low precision (∼10 arcsec rms uncertainty).

328.809+0.633. The only previously reported 1720-MHz detection is from MacLeod (1997), but with no detailed information presented. Our 1720-MHz observation shows it to be strong (10-Jy peak) and with a position offset from the 1665-MHz emission by only 0.5 arcsec - well within the combined uncertainties. The methanol emission, the 6035-MHz OH emission and the H ii continuum overlap these positions, and are discussed by Caswell (1997).

331.512-0.103. The strong 1612-MHz emission at this site was positioned by Sevenster et al. (1997b), with independent confirmation in the present observations. Earlier data show the 1612-MHz emission to be highly polarized, and its classification as an SFR with (surprisingly weaker) 1665- and 1667-MHz emission is discussed by Caswell (1998).

338.925+0.557. The 1720-MHz maser was first reported by MacLeod (1997). The 1720-MHz position derived here confirms its coincidence with the 1665-MHz maser. Note that our peak is only 0.8 Jy, whereas MacLeod measured 5 Jy; significant variability must have occurred, although some of the reduction is caused by our lower velocity resolution. Of additional interest is the 1612-MHz maser (35-Jy peak: Robinson, Caswell & Goss 1974) which also lies at the same site, as confirmed by the precise position from Sevenster et al. (1997b). Note that a methanol maser of 71-Jy peak is offset 30 arcsec, but coincident with the OH site there is a weaker methanol maser of 5-Jy peak (Caswell 1998).

339.622-0.121.MacLeod (1997) reported a 1.5-Jy 1720-MHz maser in this direction, and our accurate position confirms its coincidence with the 1665-MHz site. Note that there is no 1612-MHz emission here, contrary to earlier reports based on positions of lower accuracy (see Section 5).

340.785-0.096.Caswell & Haynes (1983a) reported a 1-Jy 1720-MHz maser with high circular polarization, chiefly left-handed (LHCP). The present measurements confirm that it is coincident with the 1665-MHz SFR maser.

345.003-0.224. The 1720-MHz maser at this site was studied by Gaume & Mutel (1987). The observation made here is in agreement with their position measurement, and usefully confirms the reliability of both data sets. Note that the associated methanol maser with peak of 73 Jy is the weaker one of a close pair (Caswell 1997).

347.628+0.148. Our new data confirm that a 1612-MHz maser essentially coincides with OH masers at 1665 and 6035 MHz, and with a ucH ii region. It is one of the examples in which there is not only 1612-MHz emission, but also a methanol 6668-MHz maser.

351.775-0.536.MacLeod (1997) briefly referred to his unpublished work on the 1720-MHz maser at this site, but did not list its intensity or derive an accurate position. Our data show a clearly detected source of several-Jy peak flux density, with position contained within the area of several arcseconds diameter over which the 1665-MHz maser spots are scattered. Note that there is also broad absorption at 1720 MHz (Caswell & Haynes 1983a), but the emission was not detected in 1980 and must therefore have increased significantly in flux density since that epoch.

354.615+0.472. The 1612-MHz maser is the weakest detected in the current observations. Our measurement of its position confirms its association with the SFR masers of OH at 1665 MHz and methanol at 6668 MHz.

10.473+0.027. A weak 0.4-Jy feature at the 1720-MHz transition was reported by Braz & Sivagnanam (1987), but with very uncertain position. Their report prompted the present search which has yielded a clear detection and allowed a position measurement confirming coincidence with the 1665-MHz emission. The velocity of our single 1720-MHz feature is +64.7 km s⁻¹, and differs from the value (+67.7 km s⁻¹) reported by Braz & Sivagnanam- perhaps a real change, or possibly just a typographical error in their table.

11.034+0.062. A previous inconclusive report of 1720-MHz emission is discussed by Caswell & Haynes (1983b). The present observation has yielded a clear detection, with position coincident with that of the 1665-MHz emission.

For sources in Table 1 that were not observed in the present measurements, the 1612-MHz information is taken from Sevenster et al. (1997a,b) and the 1720-MHz information from Gaume & Mutel (1987). The latter paper cited B1950 coordinates of several features at each site, from which we have selected only the strongest, and listed them with coordinates precessed to J2000. The coincidence of these 1720- and 1612-MHz masers with 1665-MHz masers of SFRs has been noted and discussed extensively by Caswell (1998), and they are re-listed here for completeness so as to assess the statistics of satellite-line occurrence in southern SFRs.

In addition to the successful detections, we searched for previously reported strong 1612-MHz emission towards 343.127-0.063 and found nothing (< 0.4 Jy) at this SFR site, but detected emission at a position offset by 20 arcsec at 343.119-0.067. We regard the 1612-MHz maser at this offset position as unrelated to SFRs and describe it fully in Section 5. Our 1720-MHz maser search at the position of the main-line OH maser 12.03-0.04 (RA 18^h 12^m 03.6^s, Dec. −18° 31′ 53″), where MacLeod (1997) had previously detected emission, was unsuccessful (< 0.4 Jy within a 5-arcmin radius). 1720-MHz emission was also searched for unsuccessfully (< 0.4 Jy within a 5-arcmin radius) towards 309.921+0.479, 323.459-0.079 and 337.705-0.053. However, in the latter field, an unrelated maser was discovered by chance at 337.802-0.053. Its position measured in the present observations is (J2000) RA 16^h 38^m 52.15^s, Dec. −46° 56′ 16.1″. It has been previously reported (with no flux density, velocity or precise position measurement) by Green et al. (1997), who detected it with the Parkes telescope in a search towards the supernova remnant G337.8-0.1. At the epoch of our observation, the 1720-MHz maser had a peak intensity of 1.5 Jy (corrected for its offset of nearly 6 arcmin from our field centre) at velocity −45 km s⁻¹, and width approximately 1 km s⁻¹. The present ATCA measurement confirms that it is associated with the supernova remnant, as surmised by Green et al. (1997).

5 Other varieties of 1612-MHz maser emission

The most common variety of maser with 1612-MHz emission is the class of OH/IR late-type stars, characterized by two spectral peaks separated in velocity by typically 15-40 km s⁻¹. Several hundred are listed in a recent sensitive southern survey by Sevenster et al. (1997a,b). Three such objects (331.594-0.135, 331.646-0.259 and 347.395+0.394) lay, by chance, in the fields that we observed, and the positions that we derived are listed in Table 2, with spectra shown in Fig. 2. The observations agree well with those of Sevenster et al. (1997a,b) [Note also that the stronger of two peaks of a 1612-MHz maser, 353.536+0.605, of the OH/IR variety was presumably responsible for the 1612-MHz emission once thought to be associated with 1665-MHz SFR maser 353.464+0.562 (Caswell & Haynes 1983a); the postulated association is ruled out by the more precise 1612-MHz position of Sevenster et al. (1997a).]

The present study also included the 1612-MHz maser 343.119-0.067 (not reported by Sevenster et al. 1997b) which is a single spectral feature that we originally expected to be at the position of the SFR maser 343.127-0.063 (as noted in Section 4) but found to be unrelated, with an offset of 20 arcsec. A comparable 1612-MHz maser with a single spectral feature is 339.621-0.126 (Sevenster et al. 1997b), similarly displaced by nearly 20 arcsec from the nearest SFR maser site (339.622-0.121), and we cite its properties in Table 2. What are the other properties and appropriate classification of these two unusual 1612-MHz masers? Early spectra from Parkes (Caswell & Haynes 1975; Caswell et al. 1981) established that neither of them displays significant circular polarization, and the new, precise, ATCA positions now show that they do not coincide with detectable maser emission from any other OH transition, or from methanol. Since, in the near vicinity (20 arcsec away) of these masers, there are SFR sites with similar velocity, it is possible that each might be an unusual member of a star-forming cluster. Alternatively, since all properties, other than the single (rather than double) peak in velocity, are compatible with an OH/IR interpretation, it seems most likely on current evidence that they are akin to OH/IR stars rather than to SFRs.

Two other objects that may belong to this class are 337.321-0.206 and 349.963-0.025. They are single, unpolarized spectral features at 1612 MHz, not accompanied by emission at any other OH transition, and their velocities lie outside the range of typical SFRs at their respective Galactic longitudes, and thus they resemble the OH/IR late-type stars in this characteristic. The first one has a peak flux density of 1.9 Jy, at velocity −152 km s⁻¹. It was reported (as 337.3-0.2) by Caswell & Haynes (1975), and a precise position was measured by Sevenster et al. (1997b). Note that its spectrum at the discovery epoch of 1974 is similar to that taken 20 years later by Sevenster et al. (1997b), and the apparent negative feature in the latter spectrum is a sidelobe response to the strong OH/IR star 337.539+0.131. The other similar object, first listed as 349.96-0.03 by Caswell et al. (1981), has also been accurately positioned by Sevenster et al. (1997b). It has a peak flux density of 6 Jy at velocity +39 km s⁻¹.

6 discussion

Among the 200 southern SFR maser sites at 1665 MHz, there are now well-measured counterparts for 12 1720-MHz masers and 11 1612-MHz masers, as listed in Table 1. (Previously there were accurate positions for only four of the 1720- MHz masers and nine of the 1612-MHz masers.) It thus appears that, in the near vicinity of 1665-MHz maser sites, conditions conducive to 1720- or 1612-MHz masers are equally common (or rare, depending on one's perspective), in agreement with the inference based on limited statistics and poorer positions of the earlier data compiled by Caswell et al. (1980) and by Caswell & Haynes (1983a, 1987). At only two sites (338.925+0.557 and 351.775-0.536) do we find both 1612- and 1720-MHz masers present, and even here, not surprisingly, they do not coincide precisely in position and velocity (since one expects 1612- and 1720-MHz masers to be mutually exclusive insofar as the same volume of gas cannot be emitting on both transitions). However, it is interesting that it is rare (i.e. with only two examples) for a maser site to have in its vicinity both the conditions favouring 1612-MHz emission and those favouring 1720-MHz emission.

For SFR maser sites as a whole, the ratio of 1667- to 1665-MHz maser intensity is generally slightly less than unity, and this is equally true for the subset showing maser emission at 1720 and/or 1612 MHz.

Full interpretation of the physical conditions at those SFR sites that display either 1612- or 1720-MHz masers will require further theoretical development. A basic understanding of conditions yielding 1612-MHz maser emission, specifically its requirement of quite low densities and a field of infrared radiation at wavelengths of 35 and 53 μm, has long been documented, but the 1720-MHz requirements have recently been reassessed. Pavlakis & Kylafis (1996) agree with earlier work suggesting that there is a low-density regime where collisional excitation is adequate to cause weak 1720-MHz masers, and this is the mechanism commonly invoked for masers around supernova remnants, and recently developed further by Lockett et al (1999). However, in addition, Pavlakis & Kylafis suggest that a different excitation mechanism for 1720-MHz masers can occur at a higher density regime in which the infrared radiation field, and large velocity gradients that lead to non-local line overlap, play a principal role. This is confirmation of a process that was earlier identified by Gray et al (1992). The mechanism can yield high maser brightness temperatures and seems likely to correspond to the conditions prevailing in the SFRs that exhibit 1720-MHz masers (see also Masheder et al. 1994).

In support of the conclusion that the pumping schemes are quite different, there are several corresponding differences between the observed properties of 1720-MHz masers in SFRs and those in supernova remnants.

• (i)

  In SFRs, 1665- and 1667-MHz masers accompany the 1720-MHz maser emission (not so in supernova remnants).

• (ii)

  The 1720-MHz masers in SFRs are quite variable (with further evidence in the present data, as discussed earlier), whereas there has been no evidence of variability from those in supernova remnants.

• (iii)

  The magnetic fields in SFR maser sites are commonly several mG, approaching 10 mG (Gaume & Mutel 1987; Caswell & Vaile 1995), whereas, in supernova remnants, the magnetic fields estimated from Zeeman splitting of the 1720-MHz maser line are commonly an order of magnitude weaker (Claussen et al. 1997).

• (iv)

  The 1720-MHz masers in SFRs are of very small angular size (e.g. Masheder et al. 1994) whereas those in SFRs appear several orders of magnitude larger (Claussen et al. 1997); the latter may be the net result of a cluster of many smaller diameter features, but the consequence remains that the brightness temperature of each spot must be much lower in the case of the supernova remnant masers, with no evidence for brightness temperatures exceeding 10⁸ K (to be contrasted with values of 10¹² K for 1720-MHz masers in SFRs).

Further observations of the supernova remnant masers would help to verify the last three points of distinction.

Returning to consideration of the SFR maser sites, one might expect that the properties of an associated ucH ii region would influence any preference for 1720- or 1612-MHz masing. However, the initial investigation is disappointing. At the sites of 1720-MHz masers, the ucH ii regions show properties ranging from the barely detectable (3 mJy at 351.775-0.536) to the very strong (2800 mJy at 351.417 + 0.645), for masers at distances of several kpc. Likewise, towards the 1612-MHz masers, the ucH ii flux densities range from the barely detectable (again, 3 mJy at 351.775-0.536) to the very strong [6000 mJy at 5.885-0.392; note the misprint in Table 1 of Caswell (1998) where it was incorrectly listed as 6 mJy]. It is unlikely that better data for all the sources will lead to any discernible correlations.

A correlation that does show promise is one relating to 6668-MHz methanol maser emission. At every site of 1720-MHz emission, there is methanol emission; but at nearly half the sites of 1612-MHz emission there is no detected methanol emission (the methanol search was not biased in this respect). These curious facts are worth further study. The first suggests that conditions favourable to methanol and 1720-MHz emission are similar, perhaps with both maser varieties preferentially occurring at an early stage in the evolution of a ucH ii region. The second suggests that the 1612-MHz emission is most favoured under conditions that are inimical to methanol emission; supporting this inference is the failure to detect methanol masers from any of the OH/IR stars, the primary OH maser-emitting transition of which is at 1612 MHz. Where 1612-MHz emission does occur in an SFR, this may indicate that the SFR is at a well-developed stage in its evolution. Detailed comparison of masing conditions for different species is fraught with uncertainty, in view of assumptions needed with regard to the relative abundances, and parameters that may affect one species much more than the other, such as the velocity gradients and radiation field. None the less, if we accept that the SFR variety of 1720-MHz masers require densities higher than needed for 1612-MHz masers (Pavlakis & Kylafis 1996), and recognizing that methanol is also most effectively pumped at similarly high densities (Sobolev et al 1997), this may simply explain why the methanol is more commonly associated with the 1720- than the 1612-MHz masers. The correlation of methanol and 1720-MHz OH masers may also be regarded as corroboration that the relevant 1720-MHz pumping scheme in SFRs is the high-density one with large velocity gradients, rather than the longstanding low-density scheme, now favoured for supernova remnant environments.

MacLeod (1997) used a sample of 29 1720-MHz masers (with preliminary positions) to search for correlations with 11 4765-MHz and 31 6035-MHz emitters. The latter correlation can now be further explored for sources with improved positions, albeit for a smaller sample: of the 12 1720-MHz masers that we list in Table 1, all except 10.473+0.027 have now had a 6035-MHz counterpart detected. One can conservatively conclude that the required environments are not markedly different. In assessing the significance of this correlation, one should recall, however, that the 6035-MHz masers are quite common, occurring towards more than 30 per cent of the ground-state OH SFR maser sites (Caswell & Vaile 1995). Furthermore, no reliable model for the 6035-MHz pumping scheme is yet available (Pavlakis & Kylafis 1996), and thus the significance of a correlation is unclear.

The satellite-line data for the 200 southern SFR masers are still incomplete. Although most sites have been searched to a level of below 1 Jy with single dishes, a search to deeper levels is desirable, and any variability necessarily aggravates the incompleteness of the search. It will therefore be worthwhile to monitor the known sources and search for new ones, although one might expect the statistics to remain largely unchanged.

7 conclusion

The present observations have established the coincidence, to arcsecond accuracy, of OH satellite-line emission at 1720 or 1612 MHz with 1665-MHz main-line emission in more than 20 SFRs. Thus it is now legitimate to use the variety of satellite lines detected as an indicator of the physical conditions prevailing at these different SFR sites. Where 1612-MHz masers occur, this most likely implies quite low densities, with the generally accepted excitation by far-infrared radiation. 1720-MHz masers most likely imply higher densities and the presence of significant velocity gradients.

Observed differences between 1720-MHz masers in SFRs and those in supernova remnants argue that the pumping scheme in supernova remnants is quite different and occurs at low densities, with no requirement of large velocity gradients.

The observations have also revealed that two candidate associations of 1612-MHz masers with SFRs were spurious. These rejected associations involve strong 1612-MHz emission confined to a single velocity feature, with negligible polarization, and these, we suggest, may belong to a class akin to the OH/IR stars.

Acknowledgment

I am indebted to an anonymous referee for several valuable comments.

References