SPITZER INFRARED SPECTROGRAPH OBSERVATIONS OF CLASS I/II OBJECTS IN TAURUS: COMPOSITION AND THERMAL HISTORY OF THE CIRCUMSTELLAR ICES

The optical depth, τ, and shape of the individual absorption features under consideration are extracted from the astronomical spectra using a mathematical continuum fit. We make a cubic spline fit to the ∼5–20 μm continuum of each spectrum using the following relatively feature-free regions: blueward of the 6 μm H₂O feature; between the unidentified 6.8 μm feature and the onset of the 9.7 μm silicate band; between the red wing of the silicate band (∼14.5 μm) and the 15.2 μm CO₂ feature; and redward of the CO₂ feature, while ensuring that none of the feature wings are included in the continuum. We calculate the optical depth in the spectral features using $F_\lambda =F_{\lambda,\,{\rm cont}}e^{-\tau _\lambda }$.

Because the 9.7 μm silicate feature is highly complex, and detailed modeling of the silicate mineralogy is beyond the scope of this paper, accurate extraction of ice features blended with the 9.7 μm silicate complex (e.g., those of CH₃OH, NH₃, and H₂O) is impractical. However, the 11–13 μm H₂O libration band and the 18 μm absorption feature in particularly silicate-rich spectra are much broader and shallower than the 15.2 μm ice absorption they overlap, so we can treat these shapes as pseudocontinua in order to isolate the CO₂ ice absorption.

Additional steps are needed to extract small features in the 7–8 μm range. The wide variation in 6.8 μm and silicate absorption profile wings requires that each spectrum be treated individually. For objects not affected by the silicate feature blueward of 8 μm, a straight-line continuum (∼7.1–8 μm) is imposed to isolate any features. For objects affected by both the 6.8 and the 9.7 μm features, two straight-line continua are imposed: one from ∼7.1–7.5 μm and one from ∼7.5–8 μm (Figure 3).

Zoom In Zoom Out Reset image size

We use laboratory data from the Leiden University ice analog databases⁶ (Gerakines et al. 1995, 1996; Ehrenfreund et al. 1996, 1999; van Broekhuizen et al. 2006) to identify the composition and apparent temperatures of the ices seen in our targets. In order to provide a direct comparison between the feature optical depths measured in the laboratory and those observed in the astronomical spectra, we removed a straight-line continuum from the laboratory data. For the wavelength ranges of interest, optical depths calculated using a straight-line continuum differ, on average, by less than half of a percent from those calculated using a more complicated cubic spline one. The wavelengths of the baseline ends are generally equal to the continuum points used in the astronomical spectra's spline fitting, with an occasional slight shift to ensure feature wings are excluded from the continuum. In Figure 4, we show the range of abundances contained in the Leiden databases' nonirradiated H₂O:CH₃OH:CO₂ mixtures (T = 10 − 185 K, all circles); note that this plot does not represent the fraction of the database comprising other molecules (e.g., CH₄ and H₂OH) or UV-irradiated mixtures.

Zoom In Zoom Out Reset image size

The temperatures given in this paper, except where otherwise noted, refer to laboratory conditions. Interstellar temperature barriers for crystallization, sublimation, and other physical processes are lower than their laboratory counterparts, due to the lower pressure and longer interaction timescales in interstellar space (e.g., Boogert et al. 2000). Assumptions about particle shapes have not been applied to the laboratory spectra used in this broad analysis, and such corrections would in any case be inappropriate for the annealed inhomogeneous mixtures used (Gerakines et al. 1999).

Water is the least volatile and most abundant of the ices typically found along protostellar lines of sight. The mid-infrared spectrum offers three strong features of water ice: narrow ones at 3.1 and 6.0 μm, and a broad (λ ∼11–20 μm) feature peaking at 13 μm. To turn these profiles into water–ice column density is somewhat problematic in each case. The 3.1 μm ice feature is generally regarded as the cleanest measure of H₂O ice column density (e.g., Tielens et al. 1984). However, inconvenient gaps in the spectrum transmitted by Earth's atmosphere make it difficult to determine the continuum on the short wavelength side of this feature in ground-based observations, which would be the source of most of the comparisons we could make to our data. In addition, rather significant extinction and scattering near 3 μm are inevitably associated with protostars and can affect the profile of the water–ice feature. Futhermore, this may lead to the bulk of the 3 μm background continuum taking a significantly different path through the protostellar envelope than that which gives rise to the longer wavelength features. The latter effect would be rather more serious in systems viewed closer to edge-on, such as IRAS 04169+2702 and DG Tau B.

Extinction at 6 μm is similar to that at 13 μm, and both are significantly smaller than that at 3 μm (extinction in magnitudes smaller by factors of ∼2–3; Mathis 1990; Indebetouw et al. 2005). Thus, the optical paths sampled at 6 and 13 μm, and the ice column densities derived, should be similar and may be more characteristic of the line of sight to the embedded protostar than is the case at 3 μm. In observed spectra, the 6 μm feature is the easiest of the three to analyze, as there are fewer problems in determination of the continuum. However, the depth and profile of the 6 μm feature are possibly influenced by absorption from species other than water ice, as shown, for example, by Boogert et al. (2008). The 13 μm water–ice libration feature is broad and overlaps many other absorptions, including the loosely constrained 9.7 μm silicate feature, so its optical depth and feature shape are hard to measure precisely.

Taking account of all of these difficulties, we choose to measure water–ice column density with the 6 μm feature, and check against the possibility of substantial "contamination" from other species using the 13 μm feature. To calculate the H₂O column density, we use the 6.0 μm feature fit results along with Equation (2) (summed over λ = 5.5–7 μm), with a band strength of $A_{{\rm H_2O}}=1.2\times 10^{17}\,{\rm cm}\,{\rm molec}^{-1}$ for the T = 50 K water–ice components and $A_{{\rm H_2O}}=1.3\times 10^{17}\, {\rm cm}\,{\rm molec}^{-1}$ for the T = 120 K one. These values pertain to pure H₂O ice, though they are also suitable for apolar dilutant concentrations of ≲ 30% (Gerakines et al. 1995). We note that the derived column densities N(H₂O) (Table 5) may in some cases represent upper limits, as the overall composition of the 6 μm feature is still a source of discussion (e.g., Gibb et al. 2004; Boogert et al. 2008). However, as we will see below, the consistency of these results with those from 13 μm features in our spectra indicates that the contribution of unaccounted-for species at 6 μm is rather small compared to that of water ice. In general, our values of water–ice column density are somewhat larger than those derived for the same objects from the 3.1 μm feature (e.g., Boogert et al. 2008) and tend to depart more the closer the object is to an edge-on view, so the differences may in part be due to the expected differences between the light paths viewed at the two wavelengths.

To look for signs of additional ice components with features near 6 μm that are not resolved spectrally from the water–ice feature, we have measured the 11–20 μm solid H₂O absorption in the nine targets for which the 10 μm silicate feature is most deeply in absorption. To do so, we followed a procedure similar to that by Boogert et al. (2008). From a low-order polynomial fit to the continuum, constrained to follow the spectrum at 5.3–5.6 and 30–35 μm, we determined the total optical depth in the 8–30 μm range. To this we "fit" the profile of interstellar silicate absorption, exemplified by the galactic center source GCS 3 (Kemper et al. 2004) and constrained to match the observed spectrum at 9.9–10.1 μm and 23–30 μm. Since this interstellar silicate profile may not be fully applicable to circumstellar silicates, this step introduces additional uncertainty. Subtraction of the resulting profile from the optical depth spectrum reveals the water–ice libration absorption, peaking at ∼13 μm, and interrupted at 15.2 μm by the CO₂ ice feature. We measured from such results the peak optical depth in this water–ice feature around 13 μm. By small variation of the wavelengths at which the continuum and silicate profile were constrained to follow the spectrum, we estimated the uncertainty in the peak optical depth. Because the shape of this very broad water feature is more sensitive to both the continuum and silicate profile fits than is the peak optical depth, the corresponding water–ice column density is considerably more uncertain. In Figure 7, the values for peak optical depth are plotted against the column density N(H₂O) derived from the 6 μm feature as described above. Evident in Figure 7 is a good correlation between the 13 μm and 6 μm water signatures, with Pearson's r = 0.74. That the correlation is tight indicates that, if an additional ice component is significant at 6 μm, its abundance relative to water ice must be essentially constant.

Zoom In Zoom Out Reset image size

Many of the objects in our sample have additional absorptions centered around ∼5.8 μm that can be well fit by a combination of H₂CO (formaldehyde) and HCOOH (formic acid) ices. Solid H₂CO is observed in many protostellar systems (e.g., Keane et al. 2001; Gibb et al. 2004) and plays an important role in the production of complex organic molecules (Schutte et al. 1993). H₂CO is difficult to detect independently in YSO Spitzer spectra; its 3.47 μm C–H stretch feature is beyond the wavelength range of the IRS, and its weaker 6.68 μm C = O stretch feature is blended with the 6.8 μm absorption, itself not well understood. The H₂CO column densities (Table 5) are derived using the observed integrated optical depth of the 5.8 μm feature (as determined from the spectral fits) and a band strength of $A_{H_2CO}=9.6\, \times 10^{-18}\,{\rm cm}\,{\rm molec}^{-1}$ (Schutte et al. 1993).

Solid and gas-phase HCOOH has been detected in both low- and high-mass star-forming regions (e.g., Cazaux et al. 2003; Schutte et al. 1999; Bottinelli et al. 2004; Remijan et al. 2004; Keane et al. 2001). Within the IRS wavelength range, there are solid-state HCOOH absorptions at 5.8 and 7.2 μm, and several in the 8–11 μm range (Bisschop et al. 2007). The latter ones are blended with the 9.7 μm silicate complex, but the 7.2 μm band, though intrinsically ∼4 times weaker than the 5.8 μm resonance, is detectable and indeed is tentatively identified in 6 of our 16 objects (Section 5.3). We have derived HCOOH column densities (Table 5) by integrating the observed optical depth of the 5.8 μm feature (as determined from the spectral fits) and using a band strength of A_HCOOH = 6.7 × 10⁻¹⁷ cm molec⁻¹ (Maréchal 1987).

Solid NH₃ (ammonia) has mid-infrared features at 6.15 μm and 9.3 μm (e.g., Sandford & Allamandola 1993), blended with the 6 μm water–ice and 9.7 μm silicate features, respectively. The effect on the 6 μm H₂O feature is negligible until the NH₃ concentration reaches ∼9% (Keane et al. 2001). At higher concentrations, an excess appears around 6.15 μm on the red side of the 6 μm band in laboratory spectra, which fits well the broad 6 μm feature observed in many of our YSO spectra. Analyses of the 9 μm inversion mode resonance by Lacy et al. (1998) and Gibb et al. (2000) indicate that circumprotostellar NH₃ exists primarily in an H₂O-dominated matrix. Therefore, an H₂O:NH₃ (1:0.2) mixture, rather than pure NH₃, is used in the spectral fits. Because no satisfactory band strength measurements for NH₃ in such a mixture exist, the upper limit NH₃ column densities (Table 5) are estimated by using the band strength of pure water ice for the H₂O:NH₃ mixture and extrapolating the column density of NH₃ based on its fractional abundance in the final fit.

After contributions due to H₂O, H₂CO, HCOOH, and NH₃ are subtracted, small residuals in the 6.1–6.4 μm range, varying in strength, shape, and peak position, are observed in 11 of the sample's 16 YSOs (Figure 8). PAH absorption has been suggested as a candidate for the 6.2–6.3 μm feature observed in many massive protostars (Schutte et al. 1996). Using the average interstellar PAH spectrum compiled by Hony et al. (2001) as a reference, we examined the peak position and widths of the YSO residuals. Of the objects within our sample with excess absorptions peaking between 6.2 and 6.3 μm, only three (IRAS 04361+2540, DG Tau B, and HL Tau) have an excess with a width similar to that seen in the PAH spectrum, and none of these three match the PAH feature in shape and peak position. Though the PAH relative band strengths often depend upon the molecular environment and ionization level, the broad 7.5–9 μm complex and the 6.2 μm feature are consistently observed at comparable strength in astronomical environments (Hony et al. 2001), yet we find no correlation in the YSO spectra between the 6.2 μm and 7.7 μm optical depths–the ratios are scattered between 0 and 7.5. Moreover, the prominent but narrow PAH peak at 11.3 μm (consistently ∼1.4× as strong as the 6.2 μm peak; Chan et al. 2000; Hony et al. 2001) and the lesser ones at 12.7, 13.5, and 14.3 μm are not observed in the YSO data. Thus, in agreement with Keane et al. (2001), we find it unlikely that PAHs contribute significantly around 6.2–6.3 μm.

Zoom In Zoom Out Reset image size

The position of the 6.8 μm feature coincides with the C–H deformation mode of alcohols, and of these, methanol is the most likely to be found in interstellar abundance. Theoretical models of ice processes predict the creation of solid methanol through the sequential hydrogenation of CO ice, that has been accreted from the gas phase onto dust grains (Tielens & Charnley 1997). The presence of methanol ice in interstellar and protostellar environments is observationally supported by detection of its own resonances—e.g., the C–H stretching mode at 3.54 μm (Grim et al. 1991) and the C–O stretch at 9.8 μm (Gibb et al. 2000)—as well as by signatures of its interactions with other molecules (especially CO₂; e.g., Ehrenfreund et al. 1998).

Our sample spectra do not contain the near-infrared wavelengths of CH₃OH's multiple stretching modes, and the 9.8 μm resonance is blended with the silicate absorption. However, constraints on the CH₃OH abundance can be derived using the 15.2 μm CO₂ band, as the interaction of CH₃OH with CO₂ changes the feature shape near 15.4 μm (Sections 5.4 and 6.1). The relative strength of the 15.4 μm shoulder implies that at most 50% of the 6.8 μm feature is due to CH₃OH in a CO₂ matrix, although CH₃OH unmixed with CO₂ could be present to provide additional absorption at 6.8 μm. Since CH₃OH is thought to form in a hydrogenated environment, the assignment of at least part of the 6.8 μm absorption to this molecule is then supported by the close correlation of the 6.0 μm and 6.8 μm optical depths (Section 4).

The NH⁺₄ ion was proposed as the carrier of the 6.8 μm band by Grim et al. (1989), and since then, this proposal has achieved varying levels of success (e.g Schutte et al. 1996; Keane et al. 2001; Schutte & Khanna 2003). The ion can be produced in the laboratory by the acid-base reaction HNCO + NH₃ → OCN⁻+ NH⁺₄ (Keane 1997) or in UV-irradiated H₂O:CO:NH₃ mixtures (as in Figure 8).

NH⁺₄ has expected counter-ion features at 4.62 μm (OCN⁻) and 6.3 μm and 7.41 μm (HCOO⁻, produced by photolysis). The 4.62 μm feature does not fall within the IRS wavelength coverage, and the 7.41 μm signature is too weak to be conclusively identified in our data. However, some of our sample objects do have a small excess absorption near 6.3 μm (Section 5.1.4), which is unlikely to be attributable to PAHs. It may instead be due to HCOO⁻, indicating photolysis and subsequent warming of a polar ice containing NH₃. We use the ratios of the laboratory 6.3 μm (HCOO⁻) and 6.8 μm (NH⁺₄) peaks (τ_6.8/τ_6.3 ∼ 1.7 after the H₂O feature is removed), along with the equivalent widths of the YSO and laboratory features, to estimate the NH⁺₄ contribution to the 6.8 μm feature. Using this method, we find that, within our sample, 6%–58% of the 6.8 μm feature can be attributed to NH⁺₄ (HL Tau is an outlier whose particularly large 6.3 μm excess indicates ≲91% of its 6.8 μm feature may be due to NH⁺₄ using this method).

Figure 8 shows a comparison of the 6.8 μm features, after subtraction of the 6 μm contributors, with a variety of 6.8 μm candidate laboratory spectra, all rebinned to the resolution of the observations. The bottom two lab spectra (Labs B and C) have CH₃OH/H₂O abundance ratios of 100 and 0.1, and the gray vertical lines indicate the peak positions of these mixtures. The relative amount of methanol present affects the peak substructure. While in some cases, the YSO peak positions and relative strengths closely resemble the laboratory data (e.g., IC 2087 IR and IRAS 04239+2436), in others there is no correlation at all (e.g., IRAS 04169+2702). The abundance of CH₃OH (relative to H₂O) has been found to vary by almost an order of magnitude among different protostellar lines of sight (e.g Pontoppidan et al. 2003; Gibb et al. 2004), which may partly explain the observed variety of peak substructures.

The top laboratory spectrum in Figure 8 (Lab A) is an H₂O:CO:NH₃ deposit, irradiated to create NH⁺₄ and subsequently warmed from 10 K to 165 K. The irradiated spectrum at 10 K peaks 0.1 μm blueward of our sample's bluest feature; in order to align with any of the observed profiles, the irradiated mixture must be heated to near the H₂O sublimation temperature, as the sublimation appears to cause the redward peak shift (Schutte et al. 1996). Moreover, the width of the NH⁺₄ feature is too narrow (∼0.25 μm) to account for the entire 6.8 μm feature widths observed toward each YSO (average ∼0.4 μm). Even if multiple NH⁺₄ laboratory spectra at various temperatures are combined in order to reproduce the observed feature width, the resultant peak position remains too blue to match the YSO data.

Noting that many of the YSO 6.8 μm features share characteristics with the profiles of both CH₃OH (triple peaks, 7 μm shoulder) and NH⁺₄ (asymmetric peak near 6.75 μm), we attempted fitting them with a combination of the laboratory components. While no remarkably good fits were achieved, there emerged a general trend that equal contributions from NH⁺₄ and from CH₃OH in a water-dominated matrix, with varying amounts of nearly pure CH₃OH intermixed, came closest to reproducing many of the YSO feature width-to-depth ratios (i.e., mixtures of irradiated NH₃, CH₃OH/H₂O∼0.1, and CH₃OH/H₂O∼100 in the approximate ratios from 1:1:0 to 1:1:1). However, the overall 6.8 μm astronomical feature shapes could not be well matched in terms of either peak alignment or subpeak relative strengths.

Based on this work, we conclude that we cannot rule out NH⁺₄ and CH₃OH as contributors each of at most ∼50% to the 6.8 μm band, but we emphasize that even the 50/50 combinations do not reproduce the overall feature shapes. Studies of this feature performed with higher-resolution ISO–SWS observations (e.g., Keane et al. 2001; Schutte et al. 1996) find a similar variety of structure and peak positions, confirming that it may be difficult or physically impossible to adequately explain the feature using the same set of components for all sight lines.

The 7.25 μm HCOOH band is intrinsically ∼4 times weaker than the 5.85 μm one (Bisschop et al. 2007). Based on the amount of HCOOH in the fitted spectra in Section 5.1, the expected 7.25 μm optical depths in the YSO spectra are ≲0.02. Indeed, we do observe small features at 7.25 μm in six of our 16 objects. It should be noted, however, that the strength of these do not correlate with the HCOOH column densities derived from the spectral fits and the stronger 5.85 μm feature (r = 0.06); there are also other species with weak signatures very close to this wavelength (e.g., HCOO⁻ and HCONH₂; Schutte et al. 1999). As the optical depth profiles here are neither strong nor resolved enough to conclusively distinguish among these species' signatures, these HCOOH detections must remain tentative.

In 13 of our 16 objects, we observe a feature near 7.7 μm, where the (ν₄) C–H deformation mode of CH₄ is expected. Laboratory studies have shown that this resonance's position and width are strongly dependent on the molecular environment, becoming blueshifted (as far as 7.64 μm) and broadened when mixed with a variety of other ices, including H₂O, CO, CH₃OH, and CO₂ (Boogert et al. 1997). In addition, there is the possibility of blending with the nearby SO₂ feature, whose 7.58 μm S–O stretching band has been shown to vary its peak position from 7.45 to 7.62 μm, depending on the temperature and the molecular environment (Boogert et al. 1997).

In order to disentangle absorption due to SO₂ and CH₄, we fit each YSO's 7.45–8 μm wavelength range with two Gaussian curves, allowing the peak of one (representing SO₂) to shift between 7.45 μm and 7.62 μm and constraining the other peak (CH₄) to 7.62 μm or larger. A Gaussian approximation was chosen because both the SO₂ and CH₄ profiles are too narrow, compared with the IRS resolution, to evaluate any substructure. In many cases, a single Gaussain was sufficient, indicating a single, distinct feature of either SO₂ or CH₄, depending on peak position. Of our 16 YSOs, seven show only a 7.7 μm feature with little or no asymmetrical excess, three only have a distinct feature peaking within the SO₂ range, four show both distinct features, and two show a broad absorption with additional peaks within the SO₂ range (i.e., fit by overlapping Gaussians). Using the range of laboratory feature FWHMs measured by Boogert et al. (1997), we find that in four of the nine objects with either distinct features or a Gaussian fit in the 7.45–7.62 μm range, the feature width is too narrow (Δλ≲0.06 μm) to be attributed to solid SO₂. These may be due to random noise creating an apparent peak or to CH₄ gas lines found between ∼7.4 μm and 8 μm. SO₂ also has gas lines in this region, but these are very weak (Boogert et al. 1997). For the five remaining features, we use Equation (2) with a band strength $A_{\rm SO_2}=3.4\times 10^{-17}\,{\rm cm}\,{\rm mol}^{-1}$ (Khanna et al. 1988) to determine the SO₂ column density; for the other objects, an upper limit to N(SO₂) is determined (Table 5).

Of the 13 objects with a distinct 7.7 μm CH₄ feature or Gaussian fit, seven have FWHMs comparable to those observed in laboratory mixtures (Boogert et al. 1997), particularly in those mixtures where CH₄ and either H₂O, CH₃OH, or NH₃ exist in roughly equal abundance. The remaining six objects have wider features, which may be explained by variations in temperature or ice mantle composition along the line of sight. To determine the column density of CH₄ toward each YSO, we use Equation (2) with a band strength $A_{\rm CH_4}=7.3\times 10^{-18}\,{\rm cm}\,{\rm molec}^{-1}$ (pure CH₄; discussion in Boogert et al. 1997) and integrate the optical depth between 7.6 μm and 7.9 μm. Upper limit estimates for the three nondetections are provided by the noise in the spectrum (Table 5).

The column densities observed in our sample of Class I/II YSOs yield N(CO₂)/N(H₂O) ∼0.12(±0.04) (Table 5, Figure 12), which is consistent with the values previously derived for many different lines of sight, averaging N(CO₂)/N(H₂O) ∼0.17(±0.03) (Gerakines et al. 1999; Nummelin et al. 2001; Gibb et al. 2004; Knez et al. 2005; Alexander et al. 2003), but significantly lower than the value of N(CO₂)/N(H₂O) ∼0.3, with large scatter, reported by Boogert et al. (2004) and Pontoppidan et al. (2008). Our study is unique in the sense that we use the band strength $A_{\rm CO_2}$ of an H₂O:CO₂ mixture. If we instead use the band strength for pure CO₂, we find N(CO₂)/N(H₂O) ∼0.16(±0.04), nearly equivalent to the results of the majority of previous studies, which include high- and low-mass YSOs, background field stars sampling molecular clouds, and galactic center sight lines sampling diffuse interstellar material and molecular clouds.

Zoom In Zoom Out Reset image size

The dashed line in Figure 4 identifies the locus of compositions with CO₂/H₂O = 0.12, which are clearly under-represented in the heated samples from the Leiden University database. Good profile fits can only be achieved by combining ice mixtures, and this gap in composition space may explain some of the difficulties experienced in fitting 6 μm features. Further laboratory exploration of the water-rich mixtures around the observed ratio with CO₂ may help to improve constraints on the column density of H₂O ice based on the 6 μm feature.

When considering the Class I/II sample in conjunction with the Class 0 objects, we find N(H₂O) to be well correlated (N(CO₂) less so) with the spectral index α, where more embedded sources (higher α) show higher column densities, as is expected (Figure 13; left 2 panels). The ratio N(CO₂)/N(H₂O) increases as the objects evolve toward lower α-values (Figure 13; the right panel).

Zoom In Zoom Out Reset image size

The sublimation temperature of solid H₂O is ∼40 K higher than that of pure CO₂ ice (Gerakines et al. 1995). Thus, with increased exposure to thermal radiation, N(CO₂)/N(H₂O) is expected to go down, seemingly contradicting our observations. Although CO₂ molecules embedded in a polar matrix may become "trapped" and able to survive to temperatures nearing the sublimation limit of H₂O, the thermal destruction of H₂O before CO₂ is unlikely.

An explanation of the high column density of CO₂ in more exposed environments may be found in the formation of additional CO₂ in Class I/II objects. Photodissociation of H₂O leads to the production of radicals (e.g., OH) which can react to form CO₂ on the grain surface (e.g., CO+OH→CO₂; Allamandola et al. 1988). Some other reaction pathways to form CO₂ in molecular clouds and star-forming regions have activation barriers which become less obstructive in warmer, more exposed environments (e.g Ruffle & Herbst 2001; van Dishoeck & Blake 1998). Indeed, in UV-irradiated H₂O:CH₃OH:CO₂ laboratory samples, CO₂ eventually reaches a constant abundance level due to an equilibrium between destruction and formation (Ehrenfreund et al. 1999), while the abundance of H₂O further decreases due to photodissociation.

The values measured toward the background stars Elias 16, Elias 13, and CK 2 cannot be extrapolated from this trend. The N(CO₂)/N(H₂O) ratios for these lines of sight (Table 8) are not lower than the ratios observed toward the YSOs. This discrepancy may be due to the fact that the interstellar lines of sight probe a variety of environments and compositional structure, while protostellar sight lines are dominated by the circumstellar material. Possibly, the lower temperatures in interstellar regions stifle CO₂ formation mechanisms with activation barriers, while areas of increased density favor H₂O production.

Apart from one outlier at N(H₂O) ∼1.1 ×10¹⁹ cm⁻² (L1551 IRS 5), the column densities of HCOOH and H₂O are strongly related (r = 0.61, Figure 14). This supports laboratory experiments demonstrating that the most likely formation environments of solid HCOOH are those that also produce H₂O ice, through either radiative processing or cold-surface reactions (Bisschop et al. 2007). We find N(HCOOH)/N(H₂O) ∼0%–2%, with an average of 0.6%, well within the ≲5% abundance ratio observed in lines of sight toward massive and low-mass protostars (Bisschop et al. 2007; Boogert et al. 2008).

Zoom In Zoom Out Reset image size

The observed N(CH₃OH)/N(H₂O) ratios in our sample do not show a clear trend (Figure 14). It is known that the N(CH₃OH)/N(H₂O) abundance varies greatly from source to source (e.g Boogert & Ehrenfreund 2004), and it may be better to analyze the CH₃OH abundance with respect to CO₂.

CH₃OH can be detected in the spectrum in three places. When it is intermixed with CO₂, a shoulder at 15.4 μm due to CH₃OH can be observed in the 15.2 μm feature; and CH₃OH also shows 6.8 and 9.8 μm bands (the latter is blended in the 9.7 μm silicate complex). In Figure 15, we show the ratio of the optical depth at 15.4 μm (vertical axis) and 6.8 μm (horizontal axis), both with respect to the optical depth at 15.2 μm. Filled circles represent the astronomical measurements, while the plus signs indicate the ratios measured in the laboratory spectra. The laboratory spectra show a correlation between the two features, whose exact position depends on the temperature. The 15.4/15.2 μm τ ratios observed in the astronomical data are within the range measured in the laboratory, but the corresponding 6.8 μm optical depths are significantly greater than the laboratory values. This could be understood if only a small fraction of the CH₃OH along the line of sight is embedded in a CO₂ matrix or if there are significant contributions to the 6.8 μm feature from other components.

Zoom In Zoom Out Reset image size

Using the laboratory fits to the 15.2 μm complex in the high-spectral-resolution data, we estimate the column density of solid CH₃OH intermixed with CO₂ as 30%–56% with respect to CO₂, corresponding to $N({\rm CH}_{3}{\rm OH})_{{\rm CO_2}}/N({\rm H}_{2}{\rm O}) \sim 2 \% \hbox{--}9 \%$ (Table 5). Assuming that ≲50% of the 6.8 μm feature is due to CH₃OH (Section 5.2.3), we estimate N(CH₃OH)_total/N(H₂O) ≲ 5%–17%. There are two outlying values, for IRAS 04108+2803A and IRAS 04181+2654B, at 32% and 28%. Although we do not place high confidence in these as absolute values because of the 6.8 μm uncertainties, they are consistent with the detection of solid CH₃OH with abundances (relative to H₂O) in this range up to ∼25% toward low-mass YSOs (as well as the overall scatter in abundance; Pontoppidan et al. 2003; Boogert et al. 2008). Thus, our data support the presence of a significant CO₂-poor H₂O:CH₃OH ice component.

We observe a range of N(CH₄)/N(H₂O) ratios from ∼0.3% to 8% in our sample (Figure 16), with an average of 2.6% and a single outlier (IRAS 04154+2823) at 23%. Our measurements largely overlap with the ratios of ∼0.5%–4% reported in the literature (e.g., Boogert et al. 1996; Gibb et al. 2004), though ratios up to 13% are reported by Öberg et al. (2008). The higher values in our sample (>4%, dashed line in Figure 16) may be partially ascribed to additional absorptions unresolvable at the SL-module resolution, such as gas-phase CH₄ lines (Boogert et al. 1997), or to matrix-dependent variations in band strength. Hudgins et al. (1993) found that the band strength for the 7.7 μm CH₄ feature may vary by as much as a factor of 2.8 between H₂O-rich and CO-rich matrices. We use the value for pure CH₄, which lies approximately midway between the two extremes.

Zoom In Zoom Out Reset image size

Laboratory studies of solid CH₄ indicate that it can be formed by a variety of processes-including UV photolysis of CH₃OH ices (Allamandola et al. 1988) and grain–surface hydrogenation of carbon (Tielens & Hagen 1982)—and that the resulting absorption profile has characteristics unique to its formation pathway and embedding matrix (Boogert et al. 1997). Unfortunately, the low signal-to-noise ratio of the CH₄ detections, and the low spectral resolution, prevent a detailed analysis of the spectral profiles; however, the broad range of peak positions and widths implies that the CH₄ embedding matrices vary across the sample.

All three interstellar cloud sight lines show single-peaked 15.2 μm profiles with very little substructure; CK 2 and Elias 16 are remarkably similar in shape (Figure 11), despite indications that ice mantles in Serpens may be fundamentally different from those in Taurus (e.g., Eiroa & Hodapp 1989; Whittet et al. 2007). These results give support for the cold-grain origin of at least some of the CH₃OH observed toward protostars; the presence of the unidentified 6.8 μm band in background star sight lines (in particular CK 2 and Elias 16; Knez et al. 2005) is consistent with the assignment of CH₃OH as a carrier.

Interestingly, the spectrum of W3(OH), which has the reddest SED of our entire sample, shows a clearly double-peaked CO₂ feature, indicating pure CO₂ along the line of sight, which could have been formed by thermal segregation (Figure 20). The segregation fraction is ∼26%, comparable to the Class I/II average of ∼30%. Given the difference in segregation with the pure molecular cloud lines of sight, the most likely explanation is that this line of sight samples not only the cold foreground W3 cloud but also the warmer, thermally processed CO₂ ices around the massive protostars in the W3(OH) region. The strong similarity of the 15.2 μm absorption profile toward W3(OH) to those of the low-mass Taurus objects is curious, as the line of sight toward W3(OH) is highly diverse, containing an ultra-compact Hii region, strong outflows, and H₂O/OH maser emissions (Turner & Welch 1984; Alcolea et al. 1993; Argon et al. 2003). Future spectroscopic observations with higher angular resolution will resolve the spatial structure and provide useful information on the complexity of the early stages of CO₂ heating in embedded objects.

Zoom In Zoom Out Reset image size