© ESO, 2011

1. Introduction

Since the first detection of a deuterated isotopologue in space (DCN, Jefferts et al. 1973), molecular deuteration has been considered as a powerful diagnostic tool for studying the physical conditions and past history of the interstellar gas (see for example the reviews by Roueff & Gerin 2003 and Ceccarelli et al. 2007). Among the various molecules with high degrees of deuteration, methanol is a particularly interesting one, for several reasons. First the detection of the (gaseous) triply deuterated methanol with an abundance of a few percent with respect to the hydrogenated counterpart makes it the most deuterium-enriched molecule so far known (Parise et al. 2004). Second, theoretical and experimental studies predict that methanol is formed on the grain surfaces rather than in the gas-phase (Garrod et al. 2006). Thus, the measurements of methanol deuteration permit us to study, almost unambiguously (see below), the process(es) on the grain surfaces. The link with observations is however complicated. This is firstly because there are at least two ways to form methanol on the surfaces: either by addition of hydrogen atoms on iced CO (e.g. Nagaoka et al. 2005) or by the photolysis or radiolysis of an iced mixture of methane, CO, and water (e.g. Hudson & Moore 1999; Wada et al. 2006). Deuteration is expected to be very different in the two cases and, specifically, much higher in the first case. Secondly, the alteration of the deuteration degree can occur on the grain surfaces via H/D atom exchanges both because of thermal activation (Ratajczak et al. 2009) or, again, photolysis (Weber et al. 2009). In addition, when solid methanol sublimates, it can undergo chemical reactions in the gas phase, which may again alter the original deuteration degree set on the grain surfaces (Charnley et al. 1997; Osamura et al. 2004). Despite the complications and thanks to them, deuterated methanol is a tool rich in information.

Previous studies toward low mass protostars have shown that the singly deuterated methanol CH₂DOH can be almost as abundant as CH₃OH (e.g. Parise et al. 2006), a property observed in several other deuterated molecules in low mass protostars, although to a lesser extent (Ceccarelli et al. 2007). This large deuteration very likely implies that, in low mass protostars, methanol was formed during the pre-collapse, cold, and dense phase. Surprisingly, however, the other form of singly deuterated methanol, CH₃OD, is much less abundant, by more than a factor of 10. According to the grain chemistry statistical model of Charnley et al. (1997), if the deuterated isotopologues of methanol are formed by H/D atoms addition to CO on the grain surfaces, the abundance of CH₂DOH should always be only a factor of ~3 larger than the CH₃OD one. To make the situation even more puzzling, the same [CH₂DOH]/[CH₃OD]  ratio is close to 1 in the high-mass protostar Orion IRc2 (Jacq et al. 1993). This difference between high- and low-mass protostars might reflect different deuteration degrees and/or ice compositions, as suggested by Ratajczak et al. (2009). On the other hand, no other measurement of the [CH₂DOH]/[CH₃OD]  ratio exists in high-mass protostars other than Orion IRc2, so that the measured low [CH₂DOH]/[CH₃OD] ratio may be peculiar to this source, itself peculiar in many other aspects.

In this Letter, we present new measurements of the [CH₂DOH]/[CH₃OD] ratio in two additional high-mass protostars, which confirm the low value measured in Orion IRc2. In addition, we present observations of deuterated methanol in intermediate mass protostars. As shown by Crimier et al. (2010), intermediate mass protostars are in several aspects also intermediate in the characteristics between low- and high-mass protostars. Their case can therefore help us to understand why the [CH₂DOH]/[CH₃OD]  ratio is so different at the two extremes of the mass range.

2. Observations and data reduction

The observations were carried out using the IRAM-30 m telescope in June 2010. Two frequency ranges were observed simultaneously with the new broad-band EMIR receivers, one centered at 110.2 GHz, the other at 134.0 GHz. At these frequencies, the 30 m telescope HPBW is, respectively, 22″ and 18″ and the main beam efficiency is close to 0.78. Each receiver was equipped with two channels corresponding to orthogonal linear polarizations and showed image rejections higher than 20 dB when operating in single sideband mode. The broad-band autocorrelator WILMA, divided into four parts, each providing a spectral resolution of 2 MHz and a frequency coverage of 3.7 GHz, was associated with the four channels. In addition, a high resolution autocorrelator, VESPA, was divided into two parts, one centered at 110.2 GHz covering 140 MHz with a spectral resolution of 80 kHz, the second centered at 134.0 GHz with a coverage of 290 MHz and a resolution of 310 kHz. The system temperatures were between 95 K and 115 K at 134 GHz and between 155 K and 180 K at 110 GHz. Observations were performed in wobbler-switch mode, with a 90″ throw at a rate of 0.5 Hz. The temperature scale was calibrated every 15 min using two absorbers at different temperatures and the calibration uncertainty (1σ) is about 15% of the line flux at both frequency ranges. The pointing and focus were checked every two hours on planets and continuum radio sources close to the targets. Effective on-source integration times were 45 min for G24.78, 140 min for OMC2, and 210 min for CepE and W3(H2O), where both polarisation channels were added (after checking that the line intensities were identical).

The deuterated methanol lines detected in the high resolution spectra were fitted with Gaussian profiles using the CLASS tool¹ and the frequencies given in Parise et al. (2002). For the CH₃OH lines detected in the broad-band spectra, we used the CASSIS package² and the spectroscopic data from the CDMS database (Müller et al. 2001, 2004, and references therein) to perform the line identification and the Gaussian fit. In both cases, only linear baselines were subtracted from the spectra.

3. Data analysis

Table A.1 presents the results of Gaussian fits to our observations for CH₃OH and its two isotopologues CH₂DOH and CH₃OD. The spectra of the CH₂DOH 3₀₃−2₀₂e₀ and the CH₃OD 1₁₋−1₀₊ transitions are plotted in Fig. 1.

Fig. 1 CH2DOH and CH3OD spectra towards the four sources. The observed positions and the assumed vLSR rest velocities are the following: OMC2 FIR4 (05:35:26.7, –05:09:59), 10.5 km s-1; CepheusE (23:03:12.7, +61:42:27), –11.4 km s-1; W3(H2O) (02:27:04.6, +61:52:26), –48.0 km s-1; G24.78+0.08 (18:36:12.6, −07:12:11), 111.0 km s-1. The line intensities are expressed in antenna temperature scale. The curves are Gaussian fits to the detected lines and the dashed lines indicate the expected frequencies of the lines. The small horizontal segment has a width of 5 km s-1. The broad feature right to the CH3OD line in Cep E includes the CH2DOH line (321−220 e1) at 133930.2 GHz. The CH3OD line in G24.78+0.08 appears at a significantly velocity shift so that its identification must be considered as uncertain.

Assuming that the spatial distribution of both isotopologues is the same in the region probed by the observations, the [CH₂DOH]/[CH₃OD] abundance ratio is simply equal to the column density ratio. In addition, if the lines are optically thin and the populations of all molecular energy levels in thermodynamical equilibrium are at the same rotational temperature T_rot, then for each isotopologue (i) the total column density N_i can be derived from the line flux W_i of any of its transitions by the relation: ${\mathit{N}}_{\mathit{i}}\mathrm{=}\left(\frac{\mathrm{3}{\mathit{k}}_{\mathrm{b}}{\mathit{W}}_{\mathit{i}}}{\mathrm{8}{\mathit{\pi }}^{\mathrm{3}}{\mathit{\nu }}_{\mathit{i}}\mathrm{(}{\mathit{\mu }}^{\mathrm{2}}\mathit{S}{\mathrm{)}}_{\mathit{i}}}\right){\mathit{Z}}_{\mathit{i}}\mathrm{\left(}{\mathit{T}}_{\mathrm{rot}}\mathrm{\right)}\exp \left[\frac{{\mathit{E}}_{\mathrm{u}\mathit{,i}}}{{\mathit{k}}_{\mathrm{b}}{\mathit{T}}_{\mathrm{rot}}}\right]\mathit{,}$(1)where Z_i(T_rot) is the partition function and ν_i the frequency of the transition from the upper level of energy E_u,i, corresponding to the effective line strength (μ²S)_i. Thus, the [CH₂DOH]/[CH₃OD] ratio can be derived directly from any line flux ratio using $\frac{{\mathit{N}}_{\mathrm{1}}}{{\mathit{N}}_{\mathrm{2}}}\mathrm{=}\left(\frac{{\mathit{W}}_{\mathrm{1}}}{{\mathit{W}}_{\mathrm{2}}}\right)\frac{{\mathit{Z}}_{\mathrm{1}}\mathrm{\left(}{\mathit{T}}_{\mathrm{rot}}\mathrm{\right)}{\mathit{\nu }}_{\mathrm{2}}\mathrm{(}{\mathit{\mu }}^{\mathrm{2}}\mathit{S}{\mathrm{)}}_{\mathrm{2}}}{{\mathit{Z}}_{\mathrm{2}}\mathrm{\left(}{\mathit{T}}_{\mathrm{rot}}\mathrm{\right)}{\mathit{\nu }}_{\mathrm{1}}\mathrm{(}{\mathit{\mu }}^{\mathrm{2}}\mathit{S}{\mathrm{)}}_{\mathrm{1}}}\exp \left[\frac{{\mathit{E}}_{\mathrm{u}\mathit{,}\mathrm{1}}\mathrm{-}{\mathit{E}}_{\mathrm{u}\mathit{,}\mathrm{2}}}{{\mathit{k}}_{\mathrm{b}}{\mathit{T}}_{\mathrm{rot}}}\right]\mathit{,}$(2)where Eqs. (1) and (2) refer respectively to CH₂DOH and CH₃OD. The rotational partition functions were computed within the rigid-rotor approximation, assuming that the rotational temperature is much higher than the largest rotational constant (here A(CH₃OH) ~ 6 K). The corresponding expression for CH₃OH and CH₃OD, neglecting the energy difference between the A and E symmetry species, is simply twice the normal asymmetric rotor function $\mathit{Z}\mathrm{\left(}\mathit{T}\mathrm{\right)}\mathrm{=}\mathrm{2}\sqrt{\frac{\mathit{\pi }}{\mathit{ABC}}{\left(\frac{{\mathit{k}}_{\mathrm{b}}\mathit{T}}{\mathit{h}}\right)}^{\mathrm{3}}}\mathrm{·}$(3)We thus obtain Z(T) = 1.23T^3/2 for CH₃OH and Z(T) = 1.42T^3/2 for CH₃OD. In the case of CH₂DOH, the energy difference between the e₀, e₁, and o₁ symmetry species cannot be neglected and the partition function is given by (Parise 2004): $\mathit{Z}\mathrm{\left(}\mathit{T}\mathrm{\right)}\mathrm{=}\mathrm{0.75}{\mathit{T}}^{\mathrm{3}\mathit{/}\mathrm{2}}\left[\mathrm{1}\mathrm{+}\exp \mathrm{(}\mathrm{-}\mathrm{13.4}\mathit{/}\mathit{T}\mathrm{)}\mathrm{+}\exp \mathrm{(}\mathrm{-}\mathrm{18.3}\mathit{/}\mathit{T}\mathrm{)}\right]\mathit{.}$(4)Thus, the estimate of the abundance ratio depends on the assumed rotational temperature. In practice, however, this ratio was found to be essentially insensitive to rotational temperatures in the range 20−220 K. As the number of detected lines for the deuterated species is too small to get a reliable estimate of T_rot using the rotational diagram method, the rotational temperature was derived from the lines of the main isotopologue CH₃OH, assuming that the rotational temperature is similar for all three isotopologues. For each source, the rotational temperature and the CH₃OH column density are reported in Table 1.

Once the rotational temperature of methanol isotopologues is known, the [CH₂DOH]/[CH₃OD] ratio can be computed via two methods: i) total column densities of CH₂DOH and CH₃OD can be estimated from each of the detected lines using Eq. (1). The ratio of the averaged column densities provides a first measure of the [CH₂DOH]/[CH₃OD] ratio. This method however requires us to take into account the calibration uncertainty in each line, since data coming from both frequency ranges are used. The results are shown in Col. 7 of Table 1. ii) Using Eq. (2), we can obtain as many abundance ratios as there are pairs formed with CH₂DOH and CH₃OD lines. However, some of these pairs allow us to achieve a better precision than others. They have to satisfy two criteria: the upper levels must be close in energy and the lines must belong to the same frequency band to be able to ignore the calibration uncertainty. In practice, in our data, a single pair of lines satisfies both criteria and the ratios derived by this method are given in column 6 of Table 1. It is observed that both methods give essentially the same result, within the estimated uncertainties, suggesting the reliability of the derived ratios. These results clearly show that in the two observed intermediate-mass protostars, the [CH₂DOH]/[CH₃OD] ratio is larger than 3 and much higher than the ratios in high-mass protostars. For these latter, only upper limits of the ratio were derived because no line of CH₂DOH was detected. Implications of these results are discussed in detail below.

4. Discussion and conclusions

In Fig. 2, we have plotted the [CH₂DOH]/[CH₃OD] ratio as a function of the luminosity of the protostars for all the sources (except G24.78+0.08) where this ratio has been measured so far. In addition to the present observations, we have thus included a sample of low-mass protostars (Parise et al. 2006), as well as the hot-core Orion IRc2 (Jacq et al. 1993). Our results confirm that the [CH₂DOH]/[CH₃OD]  ratio is substantially lower in massive hot cores than in (low-mass) hot-corinos, by typically one order of magnitude. Furthermore, they suggest that intermediate-mass protostars have similar properties to low-mass protostars. We have also found that in all sources, the observed [CH₂DOH]/[CH₃OD] ratio significantly differs from the value of 3 predicted by the grain chemistry models of Charnley et al. (1997) and Osamura et al. (2004). In low- and intermediate- mass protostars, this ratio is much greater than 3, while in high-mass protostars it is significantly lower. As a result, whatever the mass of the source, the observed gas-phase [CH₂DOH]/[CH₃OD] ratio is inconsistent with the current theory of methanol deuteration in the grain mantles, that is via additions of H/D atoms on solid CO. Alternative or additional processes in solid-state and/or in the post-evaporative gas-phase must be at play in the hot cores/corinos. We now discuss the different possibilites based on theoretical and laboratory works.

Fig. 2 [CH2DOH]/[CH3OD] ratio as a function of the protostar luminosity. Low-mass data (IRAS 16293, IRAS 4a, IRAS 4b and IRAS 2) were taken from Parise et al. (2006), while the data for Orion IRc2 is from Jacq et al. (1993). The upper limit reported in Table 1 for G24.78+0.08 is not plotted because of the uncertain identification of the CH3OD line at 133.9254 GHz. The horizontal dashed line refers to the value predicted by grain chemistry models (Charnley et al. 1997).

For low- and intermediate-mass protostars, various mechanisms have been suggested to explain the much lower abundance of CH₃OD compared to CH₂DOH. First, CH₃OD could be selectively destroyed in the gas-phase via protonation reactions because the dissociative recombination (DR) of CH₂DOH⁺ reforms CH₂DOH, while the DR of CH₃ODH⁺ leads to both CH₃OD and CH₃OH (Charnley et al. 1997; Osamura et al. 2004). Second, laboratory experiments have suggested that H/D exchange reactions in solid state might contribute to reducing the CH₃OD abundance. Thus, in the experiment of Nagaoka et al. (2005), the deuterium enrichment was found to proceed via H/D substitution in solid CH₃OH leading to the formation of methyl-deuterated methanol isotopologues, but no CH₃OD. An alternative solid-state depletion mechanism was proposed by Ratajczak et al. (2009): in their experiment, rapid H/D exchanges between CD₃OD and H₂O ices were found to occur at T ~ 120 K on the hydroxyl (-OD) functional group of methanol, resulting in CD₃OH and HDO molecules.

All the previous mechanisms can thus qualitatively explain a much lower abundance of CH₃OD compared to CH₂DOH, as observed in low- and intermediate-mass protostars. The reverse behavior, as observed unambiguously in W3(H₂O) and Orion IRc2, is more difficult to understand. We note, however, that the gas temperature in massive hot cores may be significantly higher than in low- and intermediate-mass protostars, as suggested for example by the CH₃OH rotational temperatures reported in Table 1 (with the exception of the value measured in IRAS 2). Temperature effects might thus explain the difference in the measured [CH₂DOH]/[CH₃OD] ratios. Following the pioneering work of Charnley et al. (1997), Osamura et al. (2004) investigated the impact of increasing the gas temperature from 50 to 100 K, using a hot core chemistry network. The [CH₂DOH]/[CH₃OD] ratio was found to decrease from greater than 3 at 50 K to ~1 at 100 K. This effect was attributed to the reaction between CH₃OH and H₂DO⁺ (which is the major deuteron donor at 100 K) ${\mathrm{H}}_{\mathrm{2}}\mathrm{D}{\mathrm{O}}^{\mathrm{+}}\mathrm{+}\mathrm{C}{\mathrm{H}}_{\mathrm{3}}\mathrm{OH}\mathrm{\to }\mathrm{C}{\mathrm{H}}_{\mathrm{3}}\mathrm{OH}{\mathrm{D}}^{\mathrm{+}}\mathrm{+}{\mathrm{H}}_{\mathrm{2}}\mathrm{O}\mathit{,}$(5)followed by dissociative recombination. However, an unrealistically large initial abundance ratio [HDO]/[H₂O] ~ 0.1 was necessary to decrease the [CH₂DOH]/[CH₃OD] ratio below 3. Isomerization processes were finally proposed by Charnley et al. (1997) to explain the finding that [CH₂DOH] ~ [CH₃OD] in Orion IRc2. The quantum calculations of Osamura et al. (2004) demonstrated, however, that internal rearrangements of energized methanol (in the neutral, ionized, or protonated forms) cannot occur efficiently.

We are thus compelled to conclude that the deuteration of methanol in massive hot cores, where the abundance of CH₂DOH is at most similar to that of CH₃OD, remains enigmatic. It is possibly related to the lower deuterium enrichment of massive protostars compared to low-mass sources (Ceccarelli et al. 2007). We also note that since the distances of the high-mass protostars are significantly larger than those of low- and intermediate-mass sources, the region probed by the observations is much larger in the high-mass objects. This may affect the measurements of the [CH₂DOH]/[CH₃OD] ratio if the (large-scale) spatial distribution of both isotopologues is different. In any case, additional laboratory and theoretical studies are needed, in particular on H/D exchange processes in ices. A revision of the deuterium chemistry in hot cores also seems necessary. In this context, we note that Thi et al. (2010) demonstrated that a high [HDO]/[H₂O] ratio (>10^-2) can be obtained in dense (n_H > 10⁶ cm^′3) and warm gas (T = 100−1000 K) owing to photodissociation and neutral-neutral reactions. In addition to modeling efforts, new observations are also crucial to assessing whether the methanol deuteration is peculiar or whether a similar selective deuteration affects other interstellar species with various functional groups, e.g. HCOOCH₃, HCONH₂, and CH₃NH₂. The detection of new multiply deuterated species (e.g. CH₂DOD) is also fundamental to helping us to discriminate between the different classes of fractionation/depletion processes.

[Page is available in the electronic edition of the journal at http://www.aanda.org]

Online material

Acknowledgments

We are deeply thankful to the IRAM staff for assistance during remote observations at the IRAM-30 m telescope. We thank an anonymous referee for constructive comments, which have contributed to improve this manuscript. This work has been supported by l’Agence Nationale pour la Recherche (ANR), France (contract ANR-08-BLAN-022).

References

All Tables

All Figures

Fig. 1 CH2DOH and CH3OD spectra towards the four sources. The observed positions and the assumed vLSR rest velocities are the following: OMC2 FIR4 (05:35:26.7, –05:09:59), 10.5 km s-1; CepheusE (23:03:12.7, +61:42:27), –11.4 km s-1; W3(H2O) (02:27:04.6, +61:52:26), –48.0 km s-1; G24.78+0.08 (18:36:12.6, −07:12:11), 111.0 km s-1. The line intensities are expressed in antenna temperature scale. The curves are Gaussian fits to the detected lines and the dashed lines indicate the expected frequencies of the lines. The small horizontal segment has a width of 5 km s-1. The broad feature right to the CH3OD line in Cep E includes the CH2DOH line (321−220 e1) at 133930.2 GHz. The CH3OD line in G24.78+0.08 appears at a significantly velocity shift so that its identification must be considered as uncertain.

In the text

Fig. 2 [CH2DOH]/[CH3OD] ratio as a function of the protostar luminosity. Low-mass data (IRAS 16293, IRAS 4a, IRAS 4b and IRAS 2) were taken from Parise et al. (2006), while the data for Orion IRc2 is from Jacq et al. (1993). The upper limit reported in Table 1 for G24.78+0.08 is not plotted because of the uncertain identification of the CH3OD line at 133.9254 GHz. The horizontal dashed line refers to the value predicted by grain chemistry models (Charnley et al. 1997).

In the text