A&A 521, L18 (2010)

Herschel/HIFI: first science highlights

LETTER TO THE EDITOR

[CII] observations of H₂ molecular layers in transition clouds

T. Velusamy - W. D. Langer - J. L. Pineda - P. F. Goldsmith - D. Li - H. W. Yorke

Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109-8099, USA

Received 31 May 2010 / Accepted 15 July 2010

Abstract
We present the first results on the diffuse transition clouds observed in [CII] line emission at 158 m (1.9 THz) towards Galactic longitudes near 340 (5 LOSs) & 20 (11 LOSs) as part of the HIFI tests and GOT C+ survey. Out of the total 146 [CII] velocity components detected by profile fitting we identify 53 as diffuse molecular clouds with associated ¹²CO emission but without ¹³CO emission and characterized by  mag. We estimate the fraction of the [CII] emission in the diffuse HI layer in each cloud and then determine the [CII] emitted from the molecular layers in the cloud. We show that the excess [CII] intensities detected in a few clouds is indicative of a thick H₂ layer around the CO core. The wide range of clouds in our sample with thin to thick H₂ layers suggests that these are at various evolutionary states characterized by the formation of H₂ and CO layers from HI and C⁺, respectively. In about 30% of the clouds the H₂ column densities (``dark gas'') traced by the [CII] is 50% or more than that traced by ¹²CO emission. On the average 25% of the total H₂ in these clouds is in an H₂ layer which is not traced by CO. We use the HI, [CII], and ¹²CO intensities in each cloud along with simple chemical models to obtain constraints on the FUV fields and cosmic ray ionization rates.

Key words: ISM: structure - ISM: molecules -- ISM: atoms - submillimeter: ISM

1 Introduction

The Herschel key program GOT C+ (Galactic Observations of Terahertz C⁺) is designed to study the diffuse interstellar medium (ISM) by observing with the HIFI instrument the [CII] ²P_3/2 ²P_1/2 fine structure line emission and absorption at 1.9 THz (158 m) over a volume weighted sampling of 500 lines of sight (LOSs) throughout the Galactic disk. The GOT C+ project is described by Langer et al. (2010a) and the use of [CII] emission to detect diffuse warm ``dark gas'' (H₂ molecular gas not seen by CO observations) by Langer et al. (2010b). C⁺ is a major ISM coolant, and its 158 m line is an important tracer of the properties of the diffuse atomic and diffuse molecular gas clouds. The [CII] line thus enables us to trace an important but to date poorly-studied stage in cloud evolution - the transition clouds going from atomic to molecular: HI to H₂ and C⁺ to CI and CO (Snow & McCall 2006). These clouds have a large molecular hydrogen fraction in which carbon exists primarily as C⁺ rather than as CO (van Dishoeck & Black 1988; Tielens & Hollenbach 1985). Transition clouds are difficult to study using the standard tracers (HI or CO) but [CII] can trace this gas.

There is growing evidence that a substantial amount of interstellar gas exists as molecular H₂, not traced by CO, for example: from gamma-ray data from EGRET (e.g. Grenier et al. (2005)) and Fermi-LAT (e.g. Abdo et al. 2010); and, the infrared continuum in diffuse clouds (Reach et al. 1994). Goldsmith et al. (2010) detected warm H₂ in emission beyond the CO extent of Taurus. Wolfire et al. (2010) have modeled the molecular cloud surfaces to estimate the amount of ``dark gas'' in the form of molecular H₂ in the H₂/C⁺ layers and find it contributes about 30% of the total mass in clouds with total  mag. Here, we present direct observational evidence for the H₂/C⁺ layer in a number of transition clouds through the detection of an excess [CII] line emission in them. We use a sample of 53 transition clouds characterized by  mag. and the presence of both HI and ¹²CO emissions but no ¹³CO. We analyze the observed [CII] intensities combined with HI and CO data to obtain an inventory of the total molecular H₂ in different layers in transition clouds and then constrain the physical conditions by applying simple models for CO formation and photodissociation.

2 Observations and data analysis

The observations reported here were made as part of the HIFI performance verification and priority science phases. We observed the [CII] line at 1900.5469 GHz towards 16 LOSs in the galactic plane with the HIFI (de Graauw et al. 2010) instrument on the Herschel Space Observatory (Pilbratt et al. 2010). The [CII] spectra were obtained using the wide band spectrometer (with 0.22 km s^-1 velocity resolution, over 350 km s^-1 range) at band 7b and using integration of 800 s to 1800 s (with rms of 0.1 K to 0.2 K on data smoothed to 1 km s^-1). For each target we used the load chop (HPoint) with a sky reference offset by 2 in latitude. The data were processed in HIPE version 3.0 using the standard pipeline for HIFI. Using a fringe fitting tool within HIPE we were able to mitigate the standing waves in band 7b (Higgins & Kooi 2009) to sufficiently low levels to provide good baselines in the [CII] spectra (Boogert, priv. comm.). The data presented here are in the Galactic plane at l = 337.8, 343.04, 343.91, 344.78, 345.65, 18.3, 22.6, 23.5& 24.3; out of the plane at b=0.5 for l = 24.3and b=1 at l=22.6& 24.3; at b=-0.5& -1at & 23.5. Table 1 summarizes all the observational data used in our analysis.

Table 1:   Observational data.

An example of the [CII] spectrum is shown in the top panel in Fig. 1. The [CII] intensities were corrected for main beam efficiency (0.63). For comparison the HI and the CO spectra are shown in the lower panel. The [CII] spectra show many velocity resolved features. All [CII] emission features show an overall correlation with the HI, though not all HI features show corresponding [CII] emission. Many [CII] features are also correlated with CO features. To separate the individual velocity components we used multiple Gaussian fitting. In the case of complex (overlapping) velocity features we used both [CII] and HI profiles together to identify the individual components. We identified a total of 146 velocity components in all the LOSs. As seen in Fig. 1 as well as in the examples shown in Langer et al. (2010b) & Pineda et al. (2010) in each spectrum we detect many velocity components. However, their identity as clouds is somewhat uncertain as the decomposition itself is not very unique and may not be reliable (e.g. Falgarone et al. 1994). Though we use the HI profile as an independent check on the features, the beam sizes (Table 1) are not modeled into the decomposition. For simplicity, here we refer to them as clouds, but in reality some of them may be for example, isolated turbulent clumps, transient fluctuations of larger structures, or superposition of extremely narrow velocity components. In view of the uncertainties, for all our quantitative analysis we do not use all of the Gaussian fit parameters. Instead we use the fitted to locate a parcel of the gas at a certain velocity and width. The I(CII), I(HI), I(¹²CO) intensities for each cloud were then obtained, in a consistent manner, by integrating the intensities ( ) over the velocity width (V) centered at the respective (except in a few cases which are confused by the adjacent component).

We identified 58 [CII] components as dense molecular clouds traced by their ¹³CO emission (e.g. the red Gaussian fits in Fig. 1) and these are discussed in a separate paper by Pineda et al. (2010). We regard the remaining 88 components without ¹³CO counterparts (e.g. the black Gaussian fits in Fig. 1) as diffuse clouds, envelopes or transition clouds. We examined these 88 diffuse [CII] clouds by correlating them with the ¹²CO spectra. We found that 53 components have associated ¹²CO emission while the remaining 35 have no ¹²CO counterparts. These 35 clouds are labeled diffuse atomic clouds of which 29 are discussed by Langer et al. (2010b). To place our [CII] cloud samples in the context of the general interstellar clouds, in Fig. 2 we identify them in an A - FUV parameter space. We use our 3- detection limits (Table 1) for [CII], ¹²CO, ¹³CO, and C¹⁸O to estimate the corresponding thresholds of and FUV based on the calculations by Visser et al. (2009). The Visser et al. calculations use  K, and cm^-3 similar to what we use below in our analysis. Here we present results on 50 transition clouds excluding 3 for data quality and other issues.

3 Results and discussion

3.1 [CII] sample of transition clouds

In Fig. 2 we find that our [CII] sample of transition clouds are diffuse having -4 mag. for reasonable interstellar FUV, in the range of 1-10 (the average FUV intensity  erg cm^-2 s^-1sr^-1 Draine 1978). In Fig. 3 we show a schematic of the diffuse cloud layers. In the dense cores with ¹³CO emission, the conversion of C⁺ to CO is more complete while it is partial in these diffuse transition clouds due to lack of sufficient self-shielding. All clouds contain some quantity of HI. As seen in Fig. 3, the observed [CII] emission originates from the purely atomic HI layer along with a contribution from the H₂/C⁺ layer, while the ¹²CO emission originates in the H₂/CO core. Thus estimates of H₂ column densities using ¹²CO intensity alone entirely misses the H₂ in the H₂/C⁺ (``dark gas'') layer. Therefore, a complete inventory of molecular H₂ in the cloud requires both the [CII] and ¹²CO intensities.

3.2 [CII] in the HI/C⁺ layer

In Fig. 4a we plot the [CII] intensities against the HI intensities for all 50 transition clouds. The error bars in Figs. 4a,b represent the 1- uncertainties in the respective measured intensities. In spite of the large scatter we note a lower bound to I(CII) that increases gradually with I(HI) which is consistent with the [CII] emission expected from a HI/C⁺ layer (Fig. 3). For quantitative analysis of [CII] emission from the HI/C⁺ layer we use the following steps (see Langer et al. 2010b):

i)

The observed I(CII) is regarded as the total I(CII) , where I(CII) and I(CII) are the emissions originating from the HI/C⁺ and H₂/C⁺ layers respectively with no [CII] emission from the ¹²CO emitting core (Fig. 3).

ii)

Use the HI intensity, I(HI) to estimate the HI column density, N(HI) = I(HI) cm^-2.

iii)

Use N(HI) to estimate the C⁺ column density in the HI/C⁺ layer, N(C X(C⁺)N(HI), where X(C⁺) = n(C⁺)/n(HI) is assumed to be .

iv)

Using the N(C above we calculate I(CII) f((n(HI), T_K) N(C , where the function f accounts for the excitation conditions for density n(HI) and temperature T_K (see Langer et al. 2010b). Then we can express it in terms I(HI), as I(CII) f(n(HI), T_K)I(HI).

We find that on average the form of I(CII) versus I(HI) can be fitted by a straight line obtained for density n(HI)  200 cm^-3 at temperature  K as shown in Fig. 4a. In more massive clouds with HI intensities greater than 1000 K km s^-1 Wolfire et al. (2010) estimate n(HI) 50-150 cm^-3 and -80 K. However, all our [CII] clouds are less massive with HI intensities <600 K km s^-1. In the present analysis we assume n(HI) 200 cm^-3 and  K which seem to describe best the contribution to the [CII] intensity from the HI/C⁺ layer.

3.3 [CII] in the H₂/C⁺ layer

Having estimated I(CII) arising from the HI/C⁺ layer we can now calculate the [CII] excess arising from the H₂/C⁺ layer as I(CII) - I(CII). In Fig. 4b we show this excess plotted against I(¹²CO). Since all the carbon in the ¹²CO emitting region is converted to CO we do not expect to see any correlation. However, in spite of the large scatter we do note a lower bound to the excess I(CII) that increases gradually with I(¹²CO) as seen by the straight line fit in Fig. 4b. This suggests the presence of a C⁺ layer surrounding the ¹²CO emitting core as shown in the schematic in Fig. 3. The straight line in Fig. 4b is an approximate fit to the lower bound to the [CII] intensities as a function of I(¹²CO) and it corresponds roughly to [CII] intensities for clouds with a H₂/C⁺ layer containing a H₂ column density 15% of the H₂ in the CO core; that is, N(H₂) (see below). Therefore this line may be regarded as representing the ``nominal'' diffuse CO clouds which contain a small H₂ envelope around the CO core. However, the clouds with large [CII] excess well above this line could represent a sample of clouds in transition with larger H₂ envelopes and relatively smaller CO cores. We this excess I(CII) and the observed I(¹²CO) to estimate the H₂ column densities in the H₂/C⁺ and ¹²CO layers respectively. For ¹²CO we use the phenomenological relationship (cf. Dame et al. 2001):

In the H₂/C⁺ layers the C⁺ excitation is by H₂ molecules and we can use the [CII] excess shown in Fig. 4b to derive the N(H₂) column density as follows:

i)

Use the I(CII) to calculate the C⁺ column density, N(C⁺) in the H₂/C⁺ layer, as a function of density n(H₂) and temperature (T_K). Here we assume a higher density of n(H₂) 300 cm^-3 than in the HI layer and a temperature  K.

ii)

Use this N(C⁺) column density to estimate N(H₂)= N(C⁺)/2X(C⁺), where X(C . Thus we get H₂ column density as a function of excess I(CII) for the above assumed n(H₂) and T_K (see Langer at al. 2010b),

Table 2:   Selected transition cloud parameters.

In Fig. 5 we show the distribution of the ratios of the H₂ column density traced by [CII] to that traced by ¹²CO. In Table 2 we list a few diffuse clouds showing a large H₂ layer around the ¹²CO emitting core. A majority of the clouds have N < N . In 15 clouds the is 50% or greater than N . In this sample of 50 transition clouds, on average, 24% of the total H₂ column density is in the H₂/C⁺ layer which is not traced by ¹²CO. Although these estimates are only approximate, they show a likely scenario in the transition cloud structure. Lower densities (100 cm^-3) and/or lower temperatures (50 K) will increase the N(H₂) in the H₂/C⁺ layer (required to account for the observed I(CII)) by factors of 2-3, while higher density (500 cm^-3) will decrease the N(H₂) by a factor of 2. However, at higher densities the temperature is likely to be <100 K and the required N(C⁺) and N(H₂) will be larger.

We can use N(H₂) in the H₂/C⁺ layer and N(HI) in the HI layer derived from I(CII) and I(HI) to evaluate A in the cloud up to the C⁺/CO transition layer. We define the C⁺/CO transition layer as an inner cloud boundary where X(C⁺) X(CO) = X(C )/2. We can now solve for the ratio of external FUV to density (₀/n(H₂)) balancing the photodissociation and the CO formation rates at the C⁺/CO transition layer. We derive analytical photodissociation rates for the attenuation and the self-shielding which are consistent with those given by Lee et al. (1996). For simplicity in the warm regions (in all our chemical modeling and analysis we use  K) we consider only the H⁺ + OI chemistry; however, detailed modeling should also include C⁺ + H₂ radiative association. Therefore, for the CO formation rates we use a simple chemical network incorporating the H⁺ + OI chemistry by extending the approach discussed by Nelson & Langer (1997) for a CO core surrounded by a warmer tenuous C⁺ envelope. In our calculation we use the reaction rates given by Glover et al. (2010). The results for a few clouds are listed in Table 2; the last three columns list A up to the C⁺/CO transition layer and the external FUV, , in units of Draine radiation field ( ). (Though the solution to the chemical modeling was obtained as /n(H₂) here we give only as we have assumed n(H₂)= 300 cm^-3 in our analysis of [CII] intensities). In nearly half of our sample the clouds have very low FUV, in the range of 0.01 to 0.1 . Such low values seem less likely in the ISM; it has been suggested that the [CII] in the ISM originates from clouds exposed to FUV, (Cubick et al. 2008). Using PDR models (Pineda et al. 2010) find that in a [CII] sample of dense molecular clouds the majority have - . Furthermore it may be noted that the H⁺ + OI chemistry used here is sensitive to the cosmic ray (CR) ionization rate. Above we used the standard CR ionization rate  s^-1. However, there is recent evidence of much higher rates in the outer layers (low ) of clouds (Indriolo et al. 2007,2009; Shaw et al. 2008). We find that using 40 increases the derived value of the FUV substantially as shown in the last column in Table 2. At least two of the clouds with high FUV values (G337.82+0.00V-127 and G337.82+0.00V-118) are near the supernova remnant G337.8-0.1, about a shell radius from its boundary at  km s^-1 (Caswell et al. 1975), and thus may be consistent with our results for higher value for CR ionization. However, for the cloud G345.65+0.00V-120, though this LOS passes near a HII region (G345.645+0.010), no enhanced radiation feature is observed at this (Caswell & Haynes 1987).

Our preliminary analysis assumes optically thin HI and C⁺ emission. We do not take into account the different beam sizes used in the observations. We do not include the gas traced by CI in the C⁺/CO transition zone. Nevertheless, the results of our simplified approach show a definite statistical trend for the presence of a majority of ``nominal'' diffuse clouds with a thin H₂ layer and a significant fraction of clouds with a thick H₂ layer without any accompanying CO.

4 Conclusions

We have observed [CII] line emission in 16 LOSs towards the inner Galaxy and detected 146 velocity resolved [CII] components. We identify 53 of these components that are characterized by the presence of both HI and ¹²CO but no ¹³CO emission as transition clouds in which the conversion of C⁺ to CO is partial and a large fraction of carbon exists as C⁺ mixed with H₂ in a ``dark gas'' layer surrounding the ¹²CO emitting core. Our results show that [CII] emission is an excellent tool to study transition clouds in the ISM, in particular as a unique tracer of molecular H₂ which is not easily observed by other means. In about 10% of the clouds the H₂ column density traced by the [CII] emitting layer is greater than that traced by ¹²CO emission. On average 25% of the H₂ in these clouds is in the H₂/C⁺ layer which is not traced by CO. Finally our estimates of the FUV field indicate the CR ionization is likely much larger than the standard value in the outer layers, consistent with recent determinations from chemical abundances in diffuse regions.

Acknowledgements

We thank the referee for suggestions. This work was performed by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. The Mopra Telescope is managed by the Australia Telescope, and funded by the Commonwealth of Australia for operation as a National Facility by the CSIRO.

References

Footnotes

... clouds

Herschel is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA.

All Tables

Table 1:   Observational data.

Table 2:   Selected transition cloud parameters.

All Figures

---

Copyright ESO 2010