AKARI NEAR-INFRARED SPECTROSCOPY OF THE EXTENDED GREEN OBJECT G318.05+0.09: DETECTION OF CO FUNDAMENTAL RO-VIBRATIONAL EMISSION

The Spitzer Galactic Legacy Infrared Mid-Plane Survey Extraordinaire (GLIMPSE: Benjamin et al. 2003) has discovered a new class of interesting objects that show extended bright emission at the IRAC band 2 (4.5 μm), called Extended Green Objects (EGOs, Cyganowski et al. 2008). EGOs are often discussed in terms of candidates of massive young stellar objects (YSOs) with outflows. The greenish color is attributed either to H₂ ($v=0-0$, S(9, 10, 11)) lines and CO ($v=1\mbox{--}0$) band heads excited by shocks, or scattered light in the outflow cavity (e.g., Noriega-Crespo et al. 2004; Smith et al. 2006; Cyganowski et al. 2008; Qiu et al. 2008; Tobin et al. 2008). Takami et al. (2010, 2012) make detailed analysis of the origin of the excess emission based on photometric data of EGOs and conclude that extra components other than H₂ emission are needed to explain the excess emission at 4.5 μm and that continuum emission of the scattered light in the outflow cavity is likely the primary origin of the excess, which is supported by Lee et al. (2013). Supernova remnants (SNRs) also show excess emission at 4.5 μm due to H₂ and CO molecular bands (Reach et al. 2006). While H₂ emission is thought to dominate in the line emission, the contribution from CO emission becomes comparable to that from H₂ for the H₂ density of a range ${10}^{6}\mbox{--}{10}^{7}$ cm⁻³ (Neufeld & Yuan 2008). CO ($v=1\mbox{--}0$) emission is not detected in the SNR IC443 by ground-based observations (Richter et al. 1995), but it is detected toward the Orion molecular cloud (OMC) (Geballe & Garden 1987, 1990; González-Alfonso et al. 2002), suggesting that CO emission could make a contribution to the excess emission.

It is indispensable to make spectroscopic observations in order to study the origin of the excess at 4.5 μm in EGOs unambiguously. However, no direct spectroscopy has so far been made for EGOs except for the 2 μm region (Caratti o Garatti et al. 2015) due to the lack of sensitive spectrometer in the 4 μm region. In this paper, we report the result of spectroscopy in 2.0–5.3 μm of one of the EGOs, G318.05+0.09, made with the Infrared Camera (IRC) onboard AKARI, which offered high-sensitivity spectroscopy in 2.0–5.3 μm even in its warm mission phase (Onaka et al. 2010).

The AKARI/IRC observation of G318.05+0.09 was carried out in the director time on 2010 February 18, which was one of the last observations of the AKARI mission (Observation I.D.: 5201570.1). The observation was made with the slit spectroscopy mode at the Ns slit of $5^{\prime\prime}$ width with the prism and grism dispersers (AOT IRCZ4 with c;Ns). The prism spectroscopy provides spectra of 2.0–5.3 μm with the spectral resolution of about 0.14 μm at 4 μm, while spectra of 2.5–5.0 μm with the resolution of about 0.03 μm are obtained with the grism. In the present observation, four exposure frames with the prism were first taken and then five exposures with grism were performed, which sandwiched a 3.2 μm-band imaging observation in-between that was used for the position determination (see Onaka et al. 2007; Ohyama et al. 2007, for more details of the IRC spectrsocpy). It provided long slit spectroscopy of about $40^{\prime\prime}$ long along the slit direction. The slit location is shown in Figure 1 superimposed on the Spitzer/IRAC artificial three-color image together with the positions of the known sources in the catalogs (see Table 1). The slit of the present observation runs from an infrared bright region at the western edge, which is most likely to be IRAS 14498–5856, although the catalog position is slightly different from the bright source (cross #3 in Figure 1). The slit covers a greenish blob at around its center. The two distinct sources are clearly seen in the long-slit spectrum image. We call the bright source S1 and the greenish blob S2 in the following. Spectra are extracted from $5^{\prime\prime} \times 8\buildrel{\prime\prime}\over{.} 7$ (6 pixels in the slit direction) areas indicated by the yellow and white boxes both for the grism and prism data. The former corresponds to S1 and the latter to S2. The slit area of S1 does not collect the entire emission from the bright source. EGO G318.05+0.09 is defined by a diamond-like shape area of $\sim 35^{\prime\prime} \times 10^{\prime\prime}$, whose center is located at slightly north of the bright source as marked by cross #2. It extends to the east and west directions symmetrically to enclose the bright source and cover the greenish regions, including S2 (Figure 3.14 of Cyganowski et al. 2008). The present AKARI/IRC spectroscopy covers only the eastern part of G318.05+0.09. The center position of EGO G318.05+0.09 is located in the region for which the spectrum of S1 is extracted. The methanol maser source MMB 318.050+0.087 (Casswell 2009; Green et al. 2012) is also located within the box of S1 (cross #1). Table 1 summarizes the central positions of the regions where the spectra are extracted together with the known sources in this region.

Zoom In Zoom Out Reset image size

Table 1.  Positions Of The Sources In The G318.05+0.09 Region

Source name	R.A. (J2000)	Decl. (J2000)	Note	Number in Figure 1
S1	14 53 42.2	−59 08 48.8	Present worka	⋯
S2	14 53 44.0	−59 08 54.3	Present worka	⋯
MMB318.050+0.087	14 53 42.67	−50 08 52.4	Methanol maserb	1
G318.05+0.09	14 53 42.6	−59 08 49	EGO ($\sim 35^{\prime\prime} \times 10^{\prime\prime}$)c	2
IRAS 14498–5856	14 53 41.6	−59 08 54	UCH IId	3
Source B	14 53 39.5	−59 08 52	X-ray sourcee	4
G318.2+0.1	14 54 50	−59 46	SNRf	Outside of Figure 1

Notes.

^aThe center position from which the spectrum is extracted in the slit. ^bCasswell (2009), Green et al. (2012). ^cCyganowski et al. (2008). ^dWalsh et al. (1997). ^eBocchino et al. (2001). ^fWhiteoak & Green (1996).

Download table as:  ASCIITypeset image

The data reduction was performed with the latest toolkit for the phase 3 (warm mission phase) of the IRC spectroscopy, IRC_SPEC_TOOLKIT_P3_20150331.² Since the observation was made in the very last period of the warm mission phase of AKARI, the effect of hot pixels became severe and the noise level was higher than typical values in the phase 3 spectroscopy (Onaka et al. 2010). To remove residual dark current and possible contribution from the surrounding background/foreground emission, we subtract the emission enclosed by the white rectangle in the eastern part of the slit labeled by BKG in Figure 1. There is a shift in the slit position on the detector array between the grism and prism spectroscopy due to the optical alignment. The amount of the shift is estimated by matching the bright region and the spectra of the same region are extracted within an accuracy of a half pixel ($\sim 0\buildrel{\prime\prime}\over{.} 8$) along the slit direction for the grism and prism modes. Details of the data reduction are summarized in Mori et al. (2014). Due to the relatively large number of hot pixels, a few residual hot pixels remain after the standard reduction process. They are masked manually. The masked pixels are at 4.457 and 4.5154 μm of the grism spectrum of S1 and at 3.9607–3.9801 μm in the grism spectrum of S2 and they do not affect the following discussion. It should also be noted that the absolute calibration may have a large uncertainty (∼20%) due to the elevated temperature of the detector array at the observing time (∼47.5 K) because the sensitivity is not well calibrated at warm temperatures, while no change in the spectral response due to the change in the temperature is recognized (Onaka et al. 2010). We have not applied correction for the contamination of the second order light for the grism spectra in the phase 2 (cold mission phase) recently reported by Baba et al. (2016) because there is no calibration of the contamination estimate available for the phase 3 data at present. Instead we discard the data in the spectral region where the contamination is expected ($\gt 4.9\,\mu {\rm{m}}$). The improved wavelength calibration provided by Baba et al. (2016) is applied to the present grism spectra. The correction for wavelengths longer than 4 μm is not significant ($\lt 0.006\,\mu {\rm{m}}$) compared with the instrumental spectral resolution.

Figure 2 shows the extracted spectra of S1 and S2. There is apparent discrepancy between the grism and prism spectra of S1 in 4–5 μm. We attribute the difference to the saturation effect of the prism spectrum since S1 is very bright for the prism spectroscopy in phase 3. The grism spectrum does not have a saturation problem at this brightness level and thus we regard it more reliable. The grism spectrum shows absorption features at around 3.0, 4.25, and 4.67 μm clearly, whose traces are also seen in the prism spectrum. The former two can be attributed to H₂O and CO₂ ice absorption, and the latter to CO gas or ice absorption. To estimate the absorption features quantitatively, we fit the continuum by a spline curve with anchor points at 2.6, 2.65, 3.8, 4.1, 4.4, and 4.9 μm (thin brown dotted line in Figure 2).

Zoom In Zoom Out Reset image size

For the H₂O ice absorption profile, the laboratory data taken at 10 K are adopted (Ehrenfreund et al. 1996). We find that an extra "red wing" component is needed to fit the 3 μm absorption feature in addition to the laboratory data profile. The red wing is seen in many objects, but its origin is not yet fully understood (e.g., Smith et al. 1993; Thi et al. 2006; Noble et al. 2013). In the present fit, it is approximated by a Gaussian profile with the central wavelength of 3.35 μm and the FWHM of 0.31 μm. Figure 3(a) shows details of the fit of the 3 μm absorption feature in units of the optical depth.

Zoom In Zoom Out Reset image size

We assume a Gaussian function for the CO₂ ice feature at 4.25 μm because it is narrow compared with the present spectral resolution and use the band strengths of CO₂ in Gerakines et al. (1995) to estimate the column density. The observed profile is well fitted by a Gaussian with the central wavelength of 4.25 μm and the FWHM of 0.036 μm. The FWHM is adopted from the study of AKARI/IRC Ns spectroscopy of Galactic star-forming regions (Mori et al. 2014). The central wavelength of the absorption peak of CO₂ ice is slightly shorter than the typical value of 4.26 μm obtained in the same study, but it is still within the uncertainty due to the large noise level and the spectral resolution of the present observation. The absorption feature at around 4.67 μm is broader than the CO ice feature (Gibb et al. 2004; Mori et al. 2014). We fit it with both CO gas and ice components. For CO ice, we adopt the central wavelength of 4.67 μm and the FWHM of 0.031 μm obtained from the Ns spectroscopy by Mori et al. (2014) and assume the band strength reported in Gerakines et al. (1995). The observed feature can be fitted by absorption of CO gas at 20 K and CO ice as shown in Figure 3(b). The contribution from CO gas dominates in the absorption. With the present spectral resolution, the P and R branches of CO gas absorption are not resolved. The red side is relatively well fitted, while the fit is not good at the blue side, where there might be an additional component at around 4.62 μm. Because of the large noise, however, it is difficult to make further discussion on the additional component. CO gas with 15–30 K gives similar fits and the gas temperature cannot be well constrained. As Figure 3 suggests, the contribution from CO ice also has a large uncertainty. Therefore, the present estimate of the column densities of CO gas and ice should be taken with caution. The derived column densities are $(26.3\pm 0.5)$, $(3.42\pm 0.03)$, $(2.0\pm 0.5)$, and ($62\pm 2)\times {10}^{17}$ cm⁻² for H₂O ice CO₂ ice, CO ice, and CO gas, respectively.

Source S2 shows a peculiar spectrum. It does not show emission features at 3.3–3.5 μm, which are generally seen in star-forming regions and are attributed to polycyclic aromatic hydrocarbons (Mori et al. 2014), nor ice absorption features seen in YSOs as in S1. It shows two broad peaks at around 4.5 and 4.7 μm and no significant emission is seen below 4 μm. Both the prism and grim spectra show similar characteristics and thus we conclude that these are real features despite the noisy spectra. A small difference between the grism and prism spectra is seen at wavelengths longer than 4.5 μm. It can be attributed partly to the large noise in the grism spectrum and/or to a possible difference in the spectrum extracted region due to the shift of the slit positions on the detector array (see Section 2).

The S2 spectrum can be accounted for by blueshifted optically thin emission of the fundamental ro-vibrational transitions ($v=1\mbox{--}0$) of CO molecules with red continuum. The two peaks are attributed to the P and R branches of CO gas emission. Since the grism spectrum is very noisy and does not have data longer than 4.9 μm, we use the prism spectrum to estimate the temperature and the velocity of the CO gas. We simply assume that the continuum increases linearly with the wavelength and the CO emission is approximated by the local thermodynamic equilibrium (LTE) condition. From the fit, the temperature and the velocity of the CO gas are estimated as ${1600}_{-400}^{+500}$ K and $-{4700}_{-1500}^{+1200}\,\mathrm{km}\,{{\rm{s}}}^{-1}$, respectively. The grism spectrum requires more steeply increasing continuum, but the best fit parameters of the CO gas (2000 K and −4100 km s⁻¹) are within the uncertainties estimated from the prism spectrum. The model spectra with the best fit parameters are shown in Figure 2(b) both for the prism (thick black line) and grism spectra (thin black line) together with the linear continuum (thick and thin dotted brown lines). The column density of the CO gas is estimated from the prism spectrum with the best fit parameters as $(1.2\pm 0.1)\times {10}^{10}$ cm⁻², which does not include the systematic error due to the calibration uncertainty at the time of the observation (Section 2).

The shift in the wavelength corresponding to the estimated blueshift velocity is $-0.06\sim -0.07$ μm, which is about a half of the resolution of the prism spectrum, but more than twice of that of the grism spectrum at 4.5 μm. The latest wavelength calibration of the grism spectroscopy has been applied (Section 2, Baba et al. 2016) and the accuracy is estimated to be better than 0.006 μm, which is much smaller than the shift. The wavelength calibration of the prism spectroscopy is less accurate and estimated to be better than 1 pixel or 0.04 μm at 4–5 μm (Onaka et al. 2009). The comparison of the prism and grism spectroscopy confirms that both wavelengths match with each other within this accuracy (Mori et al. 2014).

The shift is constrained by the profile shape of the CO emission in the fit and non-LTE effects could affect the shift estimate. To investigate the reliability of the suggested large blueshift velocity, we simply calculate the peak positions of the observed spectra and compare them with those of the LTE model spectra of different temperatures. The peak positions are estimated by a Gaussian fit in the ranges 4.3–4.6 μm and 4.65–4.9 μm for the R and P branches, respectively. We also vary the slope of the underlying continuum for 2–7 MJy sr⁻¹ μm⁻¹, which is determined by the range of the slopes obtained in the CO profile fit. The effect of the continuum slope is found to be smaller ($\lt 0.005$ and $\lt 0.01$ μm for the R and P branches, respectively) than the uncertainties in the Gaussian fits both for the prism and grism spectra. The peaks of the model CO emission are estimated in the same manner.

Figure 4 plots the difference between the P and R branch peak wavelengths against the estimated peak wavelength of the R branch both for the prism and grism spectra together with those of the model CO emission for the temperature range 500–4000 K with the various blueshift velocities. The peak wavelength is sensitive both to the temperature and the blueshift velocity, but the difference in the branch peaks is sensitive mostly to the temperature. The latter is used to separate the temperature effect from the velocity effect. The peak wavelength of the R branch of the prism spectrum agrees with that of the grism spectrum within the calibration uncertainty of the prism spectroscopy. The prism and grism spectra show compatible values for the difference in the R and P branch peak wavelengths within the uncertainties in the fit. Both data points suggest that the observed spectra can be accounted for by CO emission with blueshift velocities of −3000 ∼ −6000 km ⁻¹ and that it is difficult to explain the observed peak wavelengths by simple LTE models with velocities larger than −2000 km s⁻¹. The suggested temperatures of the CO emission range from 1200 to 3700 K, which are in agreement with the CO emission profile fit. While these analyses suggest a large blueshift velocity consistently, the present spectra do not have a high signal-to-noise ratio and good spectral resolution. Spectra with higher spectral resolution and better quality are needed to investigate the non-LTE effects and confirm the large blueshift velocity.

Zoom In Zoom Out Reset image size

4.1. Nature Of S1

4.2. Nature Of S2

We carried out near-infrared spectroscopy of the EGO G318.05+0.09 with AKARI/IRC. We have detected two distinct infrared sources in the region. One, S1, is associated with the UCH ii IRAS 14498−5856 and identified as an embedded massive YSO based on the presence of the absorption features due to H₂O and CO₂ ice species together with the associated methanol maser emission. The other greenish-color source, S2, shows a quite peculiar spectrum, which has double peaks at around 4.5 and 4.7 μm. We attribute the emission to highly blueshifted (−3000 ∼ −6000 km s⁻¹) fundamental ro-vibrational transitions ($v=1\mbox{--}0$) of CO molecules of temperatures of 1200–3700 K with red continuum.

We discuss two possibilities of the nature of S2: outflow associated with S1 and supernova ejecta. Detection of the CO fundamental ro-vibrational emission toward OMC-1 supports the association of S2 with outflow from the massive YSO S1, although the characteristics of the CO emission are different. From simple comparison with the ISO/SWS spectra of OMC-1, non-detection of H₂ in S2 suggests a more H₂-poor environment in S2 than in OMC-1. It also seems to be difficult to reconcile the large blueshift velocity with the outflow origin. The strong similarity of the S2 spectra to those toward the ejecta knots in the Cas A SNR suggests that S2 may be associated with SN ejecta. EGO G318.05+0.09 is located within the SNR G318.2+0.1, however, there is no supporting evidence for infrared emission associated with this SNR. There is no other evidence to relate S2 with SN ejecta. Observations with high spectral resolution and better quality are definitely needed to draw a clear conclusion on the nature of S2.

The present observation suggests that the greenish color of EGOs has various origins and that one EGO could contain more than one objects of different types. It demonstrates significance of infrared spectroscopy for the study of the nature of individual objects. There might be similar CO emission associated in EGOs.

The authors thank the referee W. Reach for very useful comments, which greatly improved the paper. This work is based on observations with AKARI, a JAXA project with the participation of ESA. The authors thank all the members of the AKARI project for their continuous support. They would also like to thank Fumihiko Usui for his help in the data reduction, Bon-Chul Koo for the information on the SNR G318.2+0.1, and François Boulanger for useful discussion. This research has made use of the NASA/IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. The laboratory data of the H₂O ice are obtained from the Leiden atomic and molecular database. TM received financial support from a Grant-in-Aid for JSPS Fellows. AA was supported by the University of Tokyo Research Internship Program (UTRIP) of 2014, which initiated this work. This work is supported by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (nos. 23103004 and 16H00934).

Facilities: AKARI - , Spitzer - Spitzer Space Telescope satellite.