SPITZER OBSERVATIONS OF IC 2118

We observed IC 2118 with Spitzer using both IRAC and MIPS. These observations were obtained as director's discretionary time (DDT) as part of the Spitzer Observing Program for Teachers and Students. Most of the data were part of programs 235 and 266 (PI: T. Spuck), obtained in Spring 2005 and 2006, respectively; additional observations of small portions of the map were obtained as part of program 462 (PI: L. Rebull) in Fall 2007 and Spring 2008 in an effort to increase the legacy value of the data set. The IRAC observation was broken into nine astronomical observation requests (AORs), and the MIPS observation into three; the associated AORKEYs¹² are given in Table 2. The IRAC observations used the 12 s high-dynamic-range (HDR) mode such that each exposure consists of a 0.6 s and 12 s frame. There were three dithers per position to minimize instrumental effects, stepping by 260'' for each portion of the map. The IRAC observation covers ∼1.6 deg² in all four IRAC bands (3.6, 4.5, 5.8, and 8 μm); Figure 2 shows a three-color IRAC mosaic made using 4.5, 5.8, and 8 μm. The outlines of our MIPS and optical observations are included for reference.

Zoom In Zoom Out Reset image size

The MIPS observations were fast scans, and each scan leg is offset by 160'' (half an array). There is complete coverage at 70 μm and two scan legs per position at 24 μm. We did not anticipate having viable 160 μm data, so we did not plan on complete coverage at this bandpass. However, the observations were conducted in cold MIPS campaigns, which results in viable data even if the coverage is incomplete. The MIPS observation covers ∼4 deg²; the 24, 70, and 160 μm full mosaics are shown in Figure 3.

Zoom In Zoom Out Reset image size

For the IRAC data, we started with the Spitzer Science Center (SSC) pipeline-processed basic calibrated data (BCDs), version S18.7 for program 235 and S18.5 for the two other programs; for our purposes, there is no difference between these pipeline versions. We further processed the BCDs with the IRAC artifact mitigation software written by S. Carey and available on the SSC Web site. Mosaics were constructed from the corrected BCDs using the SSC mosaicking and point source extraction software package MOPEX (Makovoz & Marleau 2005). The mosaics were computed to have a pixel scale of 122 pixel⁻¹.

We ran the APEX portion of MOPEX for source detection on the mosaics; we performed aperture photometry on the combined long- and short-exposure mosaics separately using the APEX output and the "aper" IDL procedure from the IDLASTRO library. We used a 2 pixel aperture radius and a sky annulus of 2–6 pixels. The (multiplicative) aperture corrections we used follow the values given in the IRAC Data Handbook: 1.213, 1.234, 1.379, and 1.584 for IRAC channels 1–4, respectively. Flux densities have been converted to magnitudes with the zero magnitude flux densities of 280.9 ± 4.1, 179.7 ± 2.6, 115.0 ± 1.7, and 64.1 ± 0.9 Jy for channels 1–4 respectively, as given in the IRAC Data Handbook.

To make a realistic estimate of the internal uncertainties, we obtained photometry from the individual BCD images for the candidate members. We found that the scatter within that distribution is generally of order 1%. Absolute uncertainties, arising from source variability and uncertainties in the instrument calibration (primary calibrators, aperture corrections, etc.), will be an additional error. This absolute calibration uncertainty is estimated to be at least ∼1.4%. For the longer integration times, we estimated average photometric errors in the four IRAC channels to be 0.02 mag for bright stars, increasing to 0.1 mag at 17.2, 16.6, 14.4, and 13.8 mag for channels 1–4, respectively.

The APEX source detection algorithm has a tendency to identify multiple sources within the point-spread function (PSF) of a single bright source, which can cause significant confusion at the bandmerging stage. Because our flux densities are determined from aperture photometry with a 2 pixel aperture radius that is not deblended, any object with a companion within 2 pixels is not actually resolved. The photometry lists were therefore purged of multiple sources prior to bandmerging; objects with an accidental companion within 2 pixels had the lower signal-to-noise source removed.

We extracted photometry from the long and short exposure mosaics separately for each channel, and merged these source lists together by position using a search radius of 2 pixels (244) to obtain a catalog for each channel. The magnitude cutoff transition, where we used the long exposure rather than the short exposure photometry, corresponds to magnitudes of 11, 10, 8.4, and 7.5 mag for channels 1 through 4, respectively. We obtained ∼40,000 sources in channel 1, ∼34,000 sources in channel 2, ∼7500 in channel 3, and ∼5000 sources in channel 4. These relative numbers are consistent with expectations from other star-forming regions and with the sensitivities of the channels.

For MIPS, we started with the SSC-pipeline-produced BCDs, version S16.1. The 24 μm data were processed differently from the 70 and 160 μm data.

For 24 μm, we used MOPEX to construct a mosaic from the BCDs, with the pixel scale set at 245, very close to native pixel size. We used the APEX 1-Frame routines and the point-response function (PRF) provided on the SSC Web site to perform point-source PRF-fitting photometry on the mosaics. We determined via visual inspection that 24 μm sources with an APEX-extracted signal-to-noise ratio (S/N) below 7 were not reliable; therefore, we did not include those sources in the following analysis. Moreover, we inspected individual sources and removed 13 more clearly false detections (arising from nebulosity or diffraction patterns around bright stars). This resulted in a catalog of 1082 point sources ranging from 0.9 mJy to 700 mJy. The zero point used to convert flux densities to magnitudes is 7.14 Jy, based on the extrapolation from the Vega spectrum as published in the MIPS Data Handbook. We took the error from the S/N returned by APEX as the baseline statistical error. However, the true uncertainty is affected by flat fielding issues, etc., and is estimated to be ∼4%. We added a 4% flux error in quadrature to the APEX-derived error to estimate our final 24 μm uncertainties.

At 70 and 160 μm, the individual unfiltered frames (more than 5400 BCD files) were combined using MOPEX, setting their pixel scales to their native values of 98 and 16'' for 70 and 160 μm, respectively. The final mosaic at 70 μm has a peak surface brightness of ∼100 MJy sr⁻¹, which is within the linear regime of the 70 μm array (Gordon et al. 2007); for the 160 μm mosaic, the value is ∼200 MJy sr⁻¹ which is also in the linear regime of the array (Stansberry et al. 2007), and therefore we expect at most a 20% absolute flux density uncertainty. The data were further processed to maintain the structure of the extended emission and preserve its calibration. For the 160 μm mosaic, we applied a two-dimensional (3 × 3 native pixel) median boxcar filter to interpolate across missing ("NaNs" or "not a number") data due to the incomplete coverage (see above). This step affects only those pixels near the "NaNs" by a few percent. For both Ge channels, we destriped the image using a ridgelet algorithm (J. Ingalls 2009, private communication), which conserves flux and reduces the noise. For the 70 μm point sources, we performed point-source photometry in the same fashion as we did for 24 μm, using APEX 1-Frame on the mosaic constructed as described, and the SSC-provided PRF. Sources with S/N below 8 were rejected, and we carefully removed obvious false detections by individual inspection. This resulted in a catalog of 84 point sources. The zero point used for conversion between flux densities and magnitudes for the 70 μm data is 0.77 Jy, again from the MIPS Data Handbook.

No point sources were detected in the 160 μm data.

We obtained UVR_cI_c images by observing with the United States Naval Observatory (USNO) 40 inch telescope (Flagstaff, AZ) over several epochs between 2006 November and 2007 January. An area of 2.8 deg² was covered in all 4 bands by mosaicking the ∼239 × 232 field of view. We acquired at least two images at two different exposure times for each pointing; see Table 3. On 2010 March 15, an additional epoch of relatively short (⩽60 s) integrations was obtained in VR_cI_c just of the head of the nebula (where our YSO candidates are located; see below) as a final check on our measurements. Aperture photometry was performed on the objects in each frame, magnitudes between ∼10 and 12 in the shallowest integrations and ∼20 and 22 in each band in the deepest integrations, subject to variations in the quality of each night. Absolute photometric calibration was achieved through observation of Landolt standards during each night.

The additional epoch in 2010 allowed us to check our calibration and assess intrinsic variability in these stars on ∼3 year timescales. Several of the YSO candidates are variable, some highly variable (>0.2 mag; see Table 4). The three most variable sources are all embedded in reflection nebulae.

In order to include the optical data in our spectral energy distributions (SEDs), we needed to convert the measured magnitudes to flux densities in Janskys. For the U and V filters, we used the Johnson zero points from Allen's Astrophysical Quantities (Arthur 2001), specifically 1823 and 3781 Jy, respectively. For the R_c and I_c filters, we used the zero points from Bessell (1979), which are 3040 and 2433 Jy, respectively.

We matched IRAC sources from adjacent channels to empirically estimate our astrometric uncertainty. The histograms of coordinate differences show a 1σ uncertainty of ∼03 in both directions. At this Galactic latitude and longitude (207.4,−27.9), the density of sources is relatively low (∼10 detections per square arcmin in the 3.6 μm images, 4 detections per square arcmin in R_c and I_c images). Via empirical tests across all bands, we determined that a 2'' radius (1.6 IRAC pixels) for source matching works for our sources across the optical, 2MASS, IRAC, and MIPS bands. For a matching radius less than ∼1'', not all of the real matches are found. The number of matches increases steeply with radius until 1''; after that, the number of matches increases slowly with radius until ∼4'' where the number of matches is dominated by false source associations. Using a matching radius of 2'' appears to provide the best possible merging of sources across bands, even for 24 μm; the centroiding for the 24 μm data is much better than 2'', even though the data have a spatial resolution of ∼6''.

Because the 3.6 μm observations are the most sensitive of our Spitzer observations, we started by constructing a catalog based on these detections. To be included this catalog, sources have to be detected in 3.6 μm and in at least one of the other bands. Out of the 39,817 sources in this catalog, 3365 (8.5%) have been detected in all 4 IRAC channels, 15% in the JHK_s bands, 50% in the I_c band (recall that the optical observations are deeper than the 2MASS observations), and just 12% in the U band. At 24 μm, 1% (426 sources) of the 3.6 μm-based catalog are detected. However, since the spatial coverage provided by our MIPS map is larger than that for IRAC, and since some very embedded young objects may not be detected in shorter wavelengths than 24 μm, we have also included in the catalog those sources with 24 μm detections even if they do not have 3.6 μm detections. Out of the 1082 sources detected in 24 μm, 48% are detected in the I_c band and 27% in the U band. However, the region mapped by MIPS is not exactly matched to the optical observations; considering just the region covered by both the optical and MIPS maps, these numbers become 69% and 39%, respectively. Similarly, 30% of the 24 μm sources are detected in all four IRAC bands; if one is restricted to the area covered by both MIPS and IRAC maps, 70% are detected at all four IRAC bands plus MIPS-24. Finally, 38% of the 24 μm sources are detected in JHK_s.

We first select sources using a set of IRAC color–magnitude and color–color diagrams. Our selection is based on the Gutermuth et al. (2008) method and is described in Guieu et al. (2009); in summary, this method uses cuts in several IRAC color–color and color–magnitude spaces to identify likely YSOs and likely contaminants such as active galactic nuclei (AGNs). Using this technique, we identified 459 sources having colors consistent with galaxies dominated by polycyclic aromatic hydrocarbon (PAH) emission and 765 more sources having colors consistent with AGN. Just 27 sources are not flagged as background contaminants and have colors compatible with YSOs with an IRAC excess in the [3.6] − [5.8]/[4.5] − [8] plane (where the bracket notation denotes magnitudes, e.g., [3.6] means magnitude at 3.6 μm). Figure 4 shows the IRAC color–color diagram for the sources that passed and failed these color tests. (We note for completeness that Gutermuth et al. 2009 have updated their YSO selection method with new rejection criteria; we have verified that these new criteria do not affect our list of YSO candidates.)

Zoom In Zoom Out Reset image size

Since we expect to find a lot of contamination here, the list of 27 sources not flagged as contaminants bears further scrutiny. These 27 sources consist of 9 sources brighter than [3.6] = 12, with the remaining 18 all fainter than this limit. The Spitzer Wide-area Infrared Extragalactic Survey (SWIRE; Lonsdale et al. 2003) observations of the ELAIS N1 extragalactic field¹³ consists of integrations of similar depth, but this sample is expected to be essentially entirely galaxies and foreground stars, and as such provides a guide to the range of values expected from these contaminants. The SWIRE survey shows that very few galaxies have [3.6] brighter than 12 mag. The IC 2118 objects that we have identified explicitly as contaminants have 3.6 μm flux densities comparable to the SWIRE contaminants, with [3.6] ranging from 12 to 18 mag. But, out of the YSO candidates, 18 also have magnitudes fainter than [3.6] = 12. As the brightness decreases, the chances of these objects being background contaminants increases.

Since we have a wealth of supporting data, we can use these data to additionally constrain our YSO selection. Figure 5 shows an optical color–magnitude diagram for these sources. The optical colors and magnitudes for these 18 faint sources all fall in the region occupied by main-sequence stars or background galaxies. In contrast, the 9 remaining YSO candidates with [3.6] < 12 all appear redder, near or above a 30 Myr Siess et al. (2000) isochrone in the V/V − I_c color–magnitude diagram. We strongly suspect that the faint ones are contaminants. The 9 brighter objects selected in [3.6] − [5.8]/[4.5] − [8] with [3.6] < 12 are included in our list of IRAC-selected YSO candidates. For reference, the 18 objects with [3.6] >12 are represented with a blue filled circle in most of the remaining figures in this paper. The restriction we have imposed that [3.6] < 12 means that we are effectively sensitive to masses greater than 0.1 M_☉ at a distance of 210 pc for an assumed age of ∼3–5 Myr.

Zoom In Zoom Out Reset image size

All of these nine IRAC-selected YSO candidates are seen at MIPS-24. This is in contrast to the catalog as a whole, where only 10% of the sources seen in all four IRAC bands are also seen at 24 μm. Since legitimate members of IC 2118 with infrared excesses located at ∼210 parsec should be detected at 24 μm by our observations, the fact that we see these objects gives us further confidence that these objects are truly YSOs. Three of them are also seen at 70 μm; similarly, legitimate YSOs with large excesses should be detected by our shallow 70 μm observations, so this also suggests that these objects are true YSOs. (Note, of course, that most of the sources seen in our observations at 24 or 70 μm are foreground stars or background galaxies, so detection at any MIPS band by itself is insufficient evidence for youth.)

We extracted and examined the POSS, 2MASS, and our optical images for each of these candidates to verify that they did not appear extended in any of these bands.¹⁴ While all extended objects may not be galaxies (many YSOs in Taurus at 140 pc appear as extended objects due to circumstellar reflection nebulae, edge-on disks, jets, etc.; see Rebull et al. 2010), extended objects seen here are much more likely to be galaxies than YSOs at 210 pc. All of our YSO candidates passed this check and appear to be point sources in all available bands.

This IRAC-based selection method does not allow us to detect young stars having an infrared excess only at wavelengths longer than 8 μm. We looked for potential sources with a [3.6] − [24] color excess but no IRAC excess, which would be intermediate between photospheric objects and Class II stars; these objects are Class III sources with weak excesses (possible transitional disks). Figure 6 shows the [3.6]/[3.6] − [24] color–magnitude diagram. In this parameter space, photospheres have zero color, and galaxies are red and faint. This figure also contains data from the SWIRE ELAIS N1 field. We have truncated the SWIRE catalog to fit our shallower IC 2118 24 μm detection limit, rejecting SWIRE sources with [24] > 9.4, and plotted the catalog here as black contours. The SWIRE sample is essentially entirely galaxies and foreground stars, and as such provides a visual guide to the locations where such objects appear in the diagram.

Zoom In Zoom Out Reset image size

The most striking thing about this diagram for IC 2118 is that there are three distinct groups of objects: objects of zero color (likely foreground/background stars), objects that are faint and red (likely galaxies), and objects that are bright and red (likely YSOs). No sources appear between the photospheric/Class III objects and the Class II objects. The bright objects we selected above based on IRAC colors are also bright here, strongly suggesting the presence of an infrared excess we interpret as circumstellar dust. The contaminants identified above are also identified as likely contaminants here; these objects nearly all fall on top of the SWIRE contours. The objects identified as having YSO-like IRAC colors but being as faint as the contaminants are also faint and have contaminant-like colors in this parameter space. Within the faint contaminant portion of the diagram, the YSO candidates are among the brightest or reddest objects, which is a likely result of our selection process. This lends further support to our assertion that the faint objects are likely background galaxies.

Despite the fact that Figure 6 does not reveal any new candidate YSOs near the IRAC-identified candidate YSOs, it does identify another source of particular interest. Figure 6 highlights a source (as a red triangle) that is red, with a [3.6] − [24] color of ∼6.5 and a moderately faint [3.6] ∼ 12, which is among the brighter values for the likely contaminants. This source (named 050721.9−061152) has not strictly been selected as a YSO candidate or background contaminant by our IRAC-based method, but it is located in the border of our YSO selection area in the [3.6] − [5.8]/[4.5] − [8] diagram; it has a clear [3.6] − [5.8] excess but a low [4.5] − [8] excess. Somewhat surprisingly, it is clearly detected at 70 μm. Since our 70 μm survey is shallow enough that we would only detect IC 2118 YSOs if they have a large infrared excess, we take that detection, plus the marginal shorter-band color selections, as evidence that this object may very well be a YSO candidate, and we add it to our list. We discuss this object again below.

We use [3.6] versus [3.6] − [24] to look for YSOs, whereas many papers in the literature (e.g., Rebull et al. 2010, and references therein) use the K_s versus K_s − [24] color–magnitude space to find YSOs. In that diagram, the K_s − [24] colors for M stars deviate significantly from zero (Gautier et al. 2007), so objects without disks but with a (possibly unknown) M spectral type will appear to have an infrared excess. The color departure from 0 for M stars using [3.6] − [24] happens at later types than for K_s − [24], which makes it a better choice to use for searches for small excesses in cases where the spectral type of the candidates is unknown. However, one can obviously only construct a [3.6] versus [3.6] − [24] diagram for those regions where one has both IRAC and MIPS data. Since our maps do not match exactly, and since there are previously identified T Tauris well off of the densest parts of the nebula, we also constructed a K_s versus K_s − [24] color–magnitude diagram using 2MASS and MIPS data to see if any YSO candidates appeared off the edge of our IRAC maps. Two new sources appear as potential YSOs here, but by inspection of the images, both of these sources are extended, and one (2MASX J05015386−0745321) is a known galaxy. We do not include either of these objects on our YSO candidate list.

One previously identified T Tauri star (RXJ 0502.4−0744) is located in our MIPS map but not in the IRAC map, and is detected at 24 μm; however this object does not show K_s − [24] infrared excess. Its measured flux density is 1.3 ± 0.12 mJy at 24 μm ([24] = 9.36 ± 0.1).

Coordinates and our measured magnitudes between U and 70 μm for our 10 YSO candidates appear in Table 4. Errors in the optical photometry for these 10 objects are generally dominated by the systematic errors, and as such are essentially all ∼0.1 mag at U, 0.03–0.05 mag at V, R_c, and I_c, 0.02–0.03 mag at JHK_s (where these values come from 2MASS), and 0.02 mag at all four IRAC bands. The 24 μm errors are typically 0.04 mag; the 70 μm errors vary enough between objects that they appear separately in the table. SEDs for the nine IRAC-selected YSO candidates appear in Figure 7. To guide the eye, we wished to add reddened stellar models to these plots, but spectral types are only known for four of the nine candidates. In order to provide a reference, for the remaining objects, we assumed an M0 type. Thus, for each object in Figure 7, a reddened model is shown, selected from the Kurucz–Lejeune model grid (Lejeune et al. 1997, 1998) and normalized to H band. Note that this is not meant to be a robust fit to the object, but rather a representative stellar SED to guide the eye such that infrared excesses are immediately apparent.

Zoom In Zoom Out Reset image size

Our 10th YSO candidate is the MIPS-selected object, 050721.9−061152. This object's SED appears in Figure 8 and shows that the strong infrared excess starts after ∼8 μm (which is why this object is not found via the IRAC-based selection method). This kind of SED is typical of a YSO seen edge-on. At wavelengths shorter than ∼24 μm, the thermal emission is absorbed by the disk itself and the source is mostly seen in scattered light; this object is indicated by the the triangles in Figure 5, where it is below the pre-main-sequence locus defined by the other members. At longer wavelengths, the disk becomes optically thin and the thermal emission dominates the SED. In Figure 8, we have compared the SED of 050721.9−061152 with two well-known edge-on disk SEDs: HH 30 (Burrows et al. 1996) and HV Tau C (Duchene et al. 2010). The flux densities have been normalized to match that of 050721.9−061152 in the J band. The figure illustrates that the global shape of the 050721.9−061152 SED, with a minimum at 8 μm, is compatible with an edge-on disk YSO. In our color–magnitude (Figures 5, 10, and 6) and color–color (Figures 4 and 9) diagrams, this source is plotted with a red triangle. Its location in the optical CMDs excludes its identification as simply a debris disk or transition disk. Since photospheric optical and NIR emissions are seen through the disk in scattered light, they are hard to predict and broadband observations cannot tell us much about the nature of the central object. If it is an edge-on disk, then it joins a relatively short list of similar objects; such objects can tell us much about the physical properties of the occulting edge-on disk. Further study of this object is warranted.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Out of our entire list of 10 YSO candidates, six are new discoveries. Four of our candidates are already known as young members of IC 2118; they were discovered by Kun et al. (2001, 2004) as classical T Tauri stars (cTTS) using Hα objective prism plates and confirmed by spectroscopy to be young. They are 050711.6−061510, 050653.5−061712, 050730.6−061060, and 050730.2−061016. For one of these, 050711.6−061510, the SED appears "disjoint" between the I_c- and J-band observations; this is not uncommon in T Tauri stars, and is usually explained by the star having undergone significant variation in between the epochs of observations. In fact, this star is one of three that appear in our optical observations to be nebulous and highly variable (>0.2 mag) even within the timescale of our observations.

In the spirit of the Lada (1987) and Wilking et al. (2001) Class I/Flat/II/III system, we fit a line to all of the available data between 3.6 to 24 μm. Eight of the IRAC-bright YSO candidates are Class II T Tauri stars and one (050729.4−061637) is a flat disk. These classifications did not change if we instead fit just 3.6 to 8 μm. We did not attempt a similar fit of the edge-on-disk candidate, since the object is seen through the disk, and the slope changes significantly depending on whether the MIPS points are included in the fit.

Since spectral types are not known for six of the ten YSO candidates, it is not easy to determine masses, ages, or degree of reddening. However, Figure 9 shows the J − H versus H − K_s diagram for the sample, with the same notation as earlier figures. By comparison to the ZAMS, it can be seen that all have moderate reddening. The source that does not show significant IR excess in the J − H versus H − K_s diagram is located on the 1 Myr model isochrone. Likewise in the H versus H − K_s diagram, Figure 10, the T Tauri stars appear well separated from the main sequence stars and the identified background contaminants. They all appear redder than the 1 Myr isochrone model mostly because they have infrared excess starting at H band. If one makes the simplifying and probably incorrect assumption that the offset is entirely due to A_v, and trace the objects back to the 1 or 3 Myr isochrone, the brightest objects are ∼2 M_☉, and the faintest are ∼0.1 M_☉.

We note that Kun et al. (2004) reported other classical T Tauri star (cTTS) candidates in this region (see Table 1), but they are either on the southern end of the molecular cloud or significantly off the nebular region, and are not covered by our survey.

Kun et al. (2004) reported a weak line T Tauri star (wTTS) candidate (2MASS J05060574−0646151) that may or may not be a true member of IC 2118. This object is within our surveyed region. This object does not show any infrared excess in the IRAC bands and is not detected at any of the MIPS bands. Therefore, our selection methods based on infrared excesses will not recover it. However, its position in our R_c/R_c − I_c and V/V − I_c CMDs is compatible with a pre-main-sequence star according to models and the position of our nine (IRAC-selected) YSOs in the same diagram. This object is represented with a × symbol in optical and NIR color–color and color–magnitude diagrams (Figures 5, 9, 10) and in Figure 11.

Zoom In Zoom Out Reset image size

All 10 of our YSO candidates, including the edge-on-disk candidate, are located in the head of the nebula, i.e., G206.4-26.0, the most massive molecular cloud (Kun et al. 2001) of IC 2118. Figure 11 shows the spatial distribution of the YSOs, using the same notation as in earlier figures. The distributions seen here lend further support to our assertions above that the green points are the contaminants (which are evenly distributed), the blue points (faint IRAC-selected objects) are likely galaxies (generally off-cloud and evenly distributed), and the IRAC- and MIPS-selected bright objects are likely YSOs. We expected to find YSO candidates further south in this cloud; however, evidently the conditions do not support substantial star formation in regions other than the "head." The surface brightness of the ISM at 8, 24, 70, and 160 μm is fainter in the southern portion of the nebula compared to the more northern portion. The CO maps from Kun et al. (2001) find that the head of the nebula, G206.4-26.0, at 85 M_☉, is at least 3 times more massive than any of the other clouds. G206.8-26.5, with 28 M_☉, houses only 2MASS J05060574−0646151, the wTTS candidate mentioned in the prior section, which may or may not be a true member of the cloud and does not have an infrared excess. At 24 μm, the peak surface brightness of G206.4-26.0 is ∼34 MJy sr⁻¹, and the peak surface brightness of G206.8-26.5 is not that different at ∼33 MJy sr⁻¹. At 160 μm, the peak surface brightnesses are ∼205 and ∼125 MJy sr⁻¹, respectively. The remaining clouds within our map, G207.3-26.5, G207.2-27.1, and G208.1-27.5, are 2, 2, and 6 M_☉, respectively. Kun et al. (2001) find them all to be fainter, cooler, and physically smaller on the sky than either of the other two clouds in the head of the nebula. At 24 μm, their peak surface brightnesses are ∼31, ∼31, and ∼27 MJy sr⁻¹, respectively; at 160 μm, the peaks are ∼77, ∼30, ∼80 MJy sr⁻¹, respectively. All of this suggests that the local conditions are different enough to significantly affect star formation in the two locations.

Ultraviolet (UV) excesses can also be an indication of youth (see, e.g., Rebull et al. 2000, and references therein). We have U-band measurements for nine of the known plus new candidate IC 2118 members, including the WTTS; the two objects without U measurements are 050721.9−061152 and 050650.1−061908 (see Table 4). Because these objects are subject to moderate reddening, which will significantly affect the U-band observations, a robust determination of the UV excess requires a spectral type and secure determination of A_v. However, the photospheres plotted to guide the eye in Figure 7 suggest that there are UV excesses in at least six of our nine IRAC-selected YSO candidates. Two of the largest apparent UV excesses are in two of the stars with known types, 050730.2−061016 and 050730.6−061060 (omitting 050711.6−061510 because of the evident stellar variation), suggesting that, when better spectral types are obtained for the remaining objects, UV excesses may be apparent in most of them.

At least two (perhaps) intertwined mysteries enshroud the Witch Head—its distance and the external source responsible for sculpting its surface, illuminating it, and possibly triggering star formation within it. There is at least a factor of two uncertainty in the distance to IC 2118: 210 pc versus 440 pc. Our process for identifying YSOs toward IC 2118 does not depend strongly on the distance; the same stars would have been selected for any distance up to the point where the faintest objects start to blend with the background galaxies, and all of our YSOs are considerably brighter than this limit. However, physical interpretation of the IC 2118 YSOs depends on accurate determination of the distance. For that reason, we return to the distance question now.

Two sources (at different distances) are listed as potential illumination and/or wind-sculpting sources for the nebula—the Trapezium and Rigel. Figure 3 includes two arcs from circles centered on the Trapezium (735 from IC 2118) and Rigel (448 from IC 2118). We discuss these more below.

4.5.1. Arguments in Favor of d∼ 400 pc

The most secure and accurate distance for any part of the Orion complex is the distance derived using the VLBA for several members of the Orion Nebula Cluster (ONC) by Menten et al. (2007), which is d = 414 ± 7 pc. As the most massive members of the ONC, we assume that the Trapezium cluster stars, and in particular θ¹ Ori C, are at this distance. Presumably, this also means that the Orion Nebula is at this distance. At slightly lower accuracy, the BN/KL complex just north of the Trapezium has been determined to be at 437 ± 19 pc via VLBI measurement of water masers. This puts the most active current star formation and the densest molecular gas in Orion at ⩾400 pc.

As noted by Bally et al. (1991) and expanded upon considerably by Ogura & Sugitani (1998, hereafter OS98), there is a large population of nebular structures surrounding the ONC, many of which have cometary shapes or other morphology suggestive of shaping by winds or UV photoionization. If arrows are attached to each structure pointing toward the apparent direction of the wind, these arrows form a centro-symmetric pattern pointing to an origin roughly between the Trapezium and the Orion belt stars (see Figure 2 of Bally and Figure 2 of OS98). Both Bally and OS98 include IC 2118 as a nebula that exemplifies this pattern. Figure 12, based on IRAS data, shows a wide-angle view of the gas and dust west of Orion (centered on IC 2118), illustrating that IC 2118 is simply one of a number of nebulae located many degrees from the ONC whose morphology appears "wind-blown" by a source roughly toward the ONC and belt stars. Detailed examination of the leading edge of the MIPS and IRAC images of IC 2118 (Figures 2 and 3) supports this assertion. The arcs in Figure 3 show that while the morphology seems to better follow the arc from Rigel, the direction of the "windblown" structures in the nebula suggests that the Trapezium is instead the exciting source, or perhaps a structure centered near it, such as the expanding Orion-Eridanus Bubble (see, e.g., Kun et al. 2001). These arguments therefore support a distance to IC 2118 of order 400 pc.

Zoom In Zoom Out Reset image size

4.5.2. Arguments in Favor of d∼ 210 pc

Rigel (V ∼ 0.3, spectral type B8I) is located only about 25 from IC 2118, compared to the Trapezium and belt stars at more than 7° distant. In the optical, IC 2118 is normally described as a reflection nebula due to its blue colors and lack of a bright rim. The most obvious photon source—and the one most frequently mentioned in the literature—for illuminating IC 2118 is Rigel. Since, according to Hipparcos, Rigel is at d≃ 264 ± 25 pc, it follows that IC 2118 should be at a distance of order 250 pc.

As noted by Kun et al. (2001), HD 32925 is located along the line of sight to IC 2118 and appears to have significant reddening (A_V = 1.36). This suggests it lies within or behind IC 2118. If one adopts the Walraven photometric distance for HD 32925 (Brown et al. 1994) of d∼ 260 pc, this places IC 2118 nearer than this. However, the distance to HD 32925 is not well established. The best spectral type available for HD 32925 is A0III (Houk, Michigan Spectral Catalog); given V = 9.2, M_V = 0.0 (Schmidt-Kaler 1982) and the A_V, this yields d∼ 370 pc. The original Hipparcos catalog parallax for HD 32925 was π = 1.69 ± 1.35 mas, hence supporting any distance more than about 250 pc. The new Hipparcos reduction yields π = 3.66 ± 1.2 mas, or d = 273 (+133, −66) pc. No firm conclusion can be drawn from these disparate estimates.

Figure 13 presents the M_V versus V − I CMD, comparing IC 2118 stars to YSOs in Taurus (with data from Rebull et al. 2010; Güdel et al. 2007, and references therein). Based on morphological grounds (e.g., the degree to which the YSOs are still embedded in their natal gas), we expect that the IC 2118 stars might be slightly younger than the often more physically dispersed Taurus stars. On the other hand, based on the ratio of Class I to Class II sources, the IC 2118 objects might be slightly older than Taurus. In Figure 13, the distribution of IC 2118 objects is broad, but assuming a distance of 440 pc, then IC 2118 appears to be slightly younger than Taurus. Assuming a distance of 220 pc, the bulk of the IC 2118 distribution is comparable to the bulk of the Taurus distribution at ∼3 Myr. Figure 13 thus weakly suggests that IC 2118 is closer rather than farther. A similar calculation with ONC members yields similar results.

Zoom In Zoom Out Reset image size

4.5.3. No Clear Answer

It is tempting to try to use the infrared colors to set some constraint on the distance, since after all the infrared is reprocessing the UV starlight, and the distance between a given source and the cloud clearly affects the radiation field bathing the cloud. The integrated (over the entire cloud) infrared spectral energy distribution is shown in Figure 14, where for each band, a corresponding background outside the cloud has been subtracted and the images have been smoothed to the lowest angular resolution before the measurements. The figure includes a fit performed using the DUSTEM model using the interstellar standard radiation field (ISRF; Mathis et al. 1983) In a nutshell, the DUSTEM model requires at least three different dust populations (PAHs, very small and big grains) to interpret the infrared dust emissivity (for details see Compiégne et al. 2008, 2010; Désert et al. 1990); plus the incident radiation field that is the sum of the mean ISRF and a stellar component diluted by the distance squared. The fact that the observations are well matched by the standard radiation field immediately indicates that neither θ Ori nor Rigel are dominant sources in illuminating the nebula, and therefore, no simple constraint arises from the mid-IR measurements.

Zoom In Zoom Out Reset image size

Unfortunately, it is our opinion that the evidence currently available pertaining to the distance to IC 2118 is not definitive. The upcoming GAIA mission could provide a definitive answer by deriving accurate parallaxes for HD 32925 and/or for the brighter YSOs of IC 2118. A spectral type for the diffuse optical emission from IC 2118 could establish whether the dominant photon source is Rigel or the hotter, earlier type stars of the ONC or belt region.

While constructing three-color images using Spitzer and POSS images, we noticed one object projected onto the nebulosity that appeared to have moved significantly between the epoch of the POSS image and the current epoch; see Table 5. This object, J05080318−0617141, moves at a steady rate of ∼0.2 arcsec yr⁻¹ between the 1955 and 2000 epochs; at a distance of 210 pc, this would be ∼180 km s⁻¹. The object does not appear to move much between the 2MASS and IRAC epochs (2000–2005); five years may not have been enough time to see significant motion. Our optical observations (from 2007) do not have precise enough astrometry for this calculation. However, this object is in the USNO-B catalog of objects with measurable proper motion; it is object 0837–0052579, having μ_α = 134 mas yr⁻¹ and μ_δ = −86.0 mas yr⁻¹, quite comparable with the net ∼0.2 arcsec yr⁻¹ noted above.

We were granted two nights at the Palomar 200'' in 2007 Jan to obtain follow-up spectroscopy of our YSO candidates. Unfortunately, due to high velocity winds, high clouds, and poor seeing on both of our nights, little of our data were usable. However, the spectrum we obtained of this relatively bright, high proper motion object reveals it to be an M star.

We do not find an IRAC excess for this object, and it is not detected at 24 μm. We have optical and near-infrared measurements: U = 18.77, V = 15.98, R_c = 14.62, I_c = 13.19, J = 11.68, H = 11.08, and K_s = 10.85 (where the near-infrared points come from 2MASS, and the error on all the optical points is ∼0.03 mag). Using the V/V − I_c diagram and the models from Siess et al. (2000), it is likely to be an M2 to M5 star between 20 and 70 pc away. Assuming those distances, its projected space motion is between ∼20 and 60 km s⁻¹.