Ks-band, 2.14-μm Imaging of Southern Massive Star Formation Regions Traced by Methanol Masers

Keywords: Galaxy: stellar content, masers — infrared: stars — stars: early type — stars: formation

Stellar populations towards young star formation regions are deeply embedded within large column densities of molecular gas and dust. Observations at IR wavelengths are particularly powerful for peering into these natal cocoons as they suffer much less from extinction than optical wavelengths and the young stars may also be intrinsically cool. Several excellent large-area IR surveys exist [e.g. 2MASS (Skrutskie et al. 2006), GLIMPSE (Benjamin et al. 2003)], the results from which have significantly advanced our understanding of early phases of cluster formation. However, these surveys are limited to ∼2″ resolution which may be insufficient to resolve the closest cluster members (e.g. Longmore et al. 2006) and the 2MASS Ks-band (2.14-μm) observations are not deep enough to detect the most heavily embedded/reddest objects seen in the GLIMPSE images.

In this work we present 2.14-μm images and photometry for 87 fields associated with massive star formation regions. These were originally targetted due to their IRAS colours in a search for methanol maser emission (Walsh et al. 1997, 1998) and most contain hot molecular cores (Longmore et al. 2007a; Purcell et al. 2006, 2009). By matching source positions with those in existing near-IR (2MASS) and mid-IR (GLIMPSE) catalogues, we have created composite 2.14- to 8-μm catalogues which are typically 3–4 magnitudes deeper at 2.14 μm than 2MASS. Detailed analysis of these regions requires consideration of completeness and confusion etc., which are specific to individual fields — a result of differing levels of differential extinction and source confusion as well as differing sensitivity across the fields. However, we calculate a number of metrics which describe the quality and quantity of the data sets. We provide these catalogues and calibrated images as a resource for the astronomical community.

The observations were made using IRIS2 ^{1 , 2} (Infrared Imager and Spectrometer 2) on the 3.9-m Anglo-Australian Telescope (AAT) at Siding Spring Observatory using the Ks-band (1.982–2.306 μm) filter. 87 regions were imaged, as listed in Table 1. The images are shown in figures 7 to 92. The observations were taken over several different periods between 2003 and 2006. For each source, a 3 × 3 image grid was created with ∼1′ offsets from the pointing centre. The integration time at each of the nine grid positions was around one minute, with 9 × 6-s exposures. The initial reduction was carried out using a data reduction pipeline provided by the Anglo-Australian Observatory. The detector bias was removed at readout time using the Double Read Mode method. The data were reduced using the in-house orac-dr pipeline with the ‘JITTER_SELF_FLAT_KEEPBAD’ orac-dr ³ recipe to correct for dark current, create flat-field images from star free pixels in the 9 jittered source fields and apply a bad pixel mask to each image. The pipeline then corrects for distortion at the outer edges of the image caused by the slightly curved focal plane. The 9 individual fields of 7.7′ × 7.7′ were finally aligned and mosaiced together to give a 9.7′ × 9.7′ field of view, although only the inner 5.7′ × 5.7′ are covered by all 9 fields. A typical sensitivity in the inner region of an uncrowded field would be expected to be ∼19.6 mag (5σ) in good conditions.

Table 1.  Summary of the data characteristics for each of the regions observed with IRIS2

Using two 1.3 m telescopes in Arizona and Chile, the 2MASS survey scanned the entire sky at the J, H and Ks bands (1.25, 1.65 and 2.17 μm) (Skrutskie et al. 2006). The survey has a 10-σ limiting magnitude of 15.8, 15.1, 14.3 ± 0.03 mag at J, H and Ks respectively with a pixel size of 2.0″ and 0.1″ pointing accuracy. We make use of this data to provide a reference frame for the images as well as to provide the flux calibration.

Using the infrared array camera (IRAC) on-board the Spitzer Space Telescope (SST ⁷ ), the GLIMPSE survey has observed the galaxy at 3.6, 4.5, 5.8 and 8.0 μm with a 1.2″ pixel size between 10° < |l| < 65° and |b| < 1° (Benjamin et al. 2003). The photometric accuracy is 0.2 and 0.3 mag for bands 1/2 and 3/4 respectively. The astrometric accuracy of the point source catalogue is ∼0.3″. We combine this with the AAT data to construct a source catalogue from 2.14 to 8.0 μm.

Figures 7 to 92 show the IRIS2 2.14-μm images towards each of the regions overlayed with the methanol maser emission from Walsh et al. (1998). Note that in most cases, regions of extinction are readily apparent. Rarely is there a 2 μm point source associated with a methanol maser. In many cases, extensive regions of 2-μm nebulosity are seen. Several targetted regions overlap within a few arcminutes (e.g. G14.99–0.70 and G15.03–0.68 in the vicinity of M17) but each field was imaged independently to ensure uniform coverage (within the observational systematics) in the 5.7′ × 5.7′ inner region of the mosaic with optimum sensitivity (see Section 2).

Table 2 shows an example extract from the first five lines of the IRIS2 Ks-band (2.14-μm) point source catalogue for a sample region. The first two columns give the source right ascension and declination in decimal degrees. Columns 3 and 4 give the magnitude and magnitude uncertainty. Sections 4.1 and 4.2 examine the issues of seeing and completeness in these Ks-band (2.14-μm) PSCs including photometry for these sources.

The FITS images, IRIS2 PSCs and IRIS2/GLIMPSE PSCs for each region are available through the Centre de Données astronomiques de Strasbourg (CDS) ⁸ . Photometric calibration for the FITS images is provided by the FITS header keyword, ‘MAGZERO’, which gives the magnitude in each image corresponding to a flux count of 1 ⁹ . This is the same value as that given in Column 6 of Table 1.

Figure 4 shows a histogram of the seeing conditions during observations of each of the regions. The majority of regions were observed in seeing of 1.2 to 1.4″, a noticeable improvement on the ∼2″ resolution of both 2MASS and GLIMPSE. The other two plots in Figure 4 show the effects of seeing on the brightest magnitude and number of sources detected. When the seeing is poor, the flux from bright stars is spread over more pixels and brighter stars can be recovered. Given the crowded nature of the observed fields towards the Galactic plane, it is unsurprising that the number of sources detected increases dramatically as the seeing decreases. The inverse linear relation between source counts and seeing suggests the difference in source counts resulting mostly due to the seeing present at the time of each observation.

Figure 4  Left, histogram of seeing conditions during observations of the regions; centre, brightest detected magnitude per region vs. seeing; right, number of sources detected per region vs. seeing.

Longmore et al. (2007b, hereafter L07B) used artificial star recovery to investigate the spatial variation in point source sensitivity as a function of wavelength for the field associated with G305.2 + 0.2. Having calculated the PSF for each image, a grid of artificial stars of the same magnitude separated by 30″ was placed across the image. The same automated finding technique outlined in Section 2 was then used to calculate how many of the artificial stars were recovered. By shifting the 30 ″ × 30″ grid of artificial stars in small steps through the image, it was possible to measure the completeness at 5 ″ intervals without the PSF of individual artificial stars overlapping. The process was repeated by increasing the artificial star magnitudes in steps of 0.5 mag until no more stars were recovered. By recording the largest recovered magnitude at each position and wavelength it was possible to build a three-dimensional picture of the point source sensitivity across the region. The relative completeness as a function of position was then calculated at each wavelength by subtracting the median completeness magnitude at that wavelength from every position. This method is similar to that used by Gutermuth (2005) in an analysis of completeness limits of Spitzer IRAC data. Through analysis of the results from the artificial star recovery method, L07B found a simpler, empirical method of estimating the 90% completeness limit by calculating the turnover in the histogram of the number of stars as a function of magnitude which yields approximately the same limit as the full analysis. This metric is provided in Table 1, but we stress that a comprehensive analysis of each field would require this completeness limit to be determined individually for each. Figure 5 presents the histogram of this distribution of turnover magnitude.

Figure 5  Histogram of estimated 90% completeness using a simple, empirical method of calculating the turnover in the histogram of the number of stars as a function of magnitude, following L07B.

As an example of what the data has the potential to provide once a full completeness analysis is conducted, we show the results towards G305.2+0.2, as reported by L07B. Figure 6 shows the slope of the spectral energy distribution between 2.14 and 4.5 μm (i.e. α _2.14–4.5) of all sources detected towards 1) a control field using a combined 2MASS/GLIMPSE PSC, 2) the target field using a combined 2MASS/GLIMPSE PSC and 3) the target field using a combined IRIS2/GLIMPSE PSC. The spectral index is related to the categories of young stellar objects (Class I, II, III), with younger sources tending towards larger spectral indexes. In this way the PSC provides a measure of the distribution of evolutionary states of the embedded cluster seen towards G305.2 + 0.2. The considerably more sensitive 2.14-μm IRIS2 images means much more deeply embedded (redder) sources are detected with the IRIS2/GLIMPSE PSC compared to the 2MASS/GLIMPSE PSC. This is obvious in the bottom panel of Figure 6 as a significant increase in the number of sources with α ≥ 0 compared to the middle panel.

Figure 6  Histogram of the distribution of spectral indices (α = d log λ F _λ/d log λ) for all sources matched at Ks (2.14 μm) and 4.5 μm towards the region G305.2+0.2 (from Longmore et al. 2007b). The top and central figures show the distribution in the control field and target region respectively, for sources matched in the 2MASS/GLIMPSE PSC. The bottom figure gives the distribution of sources matched in the IRIS2/GLIMPSE PSC of the target region. The considerably more sensitive 2.14-μm IRIS2 images means much more deeply embedded (redder) sources are detected, which is seen in the bottom figure as a significant increase in the number of sources with α ≥ 0.