EDP Sciences
Free Access
Issue
A&A
Volume 558, October 2013
Article Number A132
Number of page(s) 9
Section Interstellar and circumstellar matter
DOI https://doi.org/10.1051/0004-6361/201321349
Published online 17 October 2013

© ESO, 2013

1. Introduction

Asymptotic giant branch (AGB) stars come in different chemical flavours ranging from oxygen-rich, over S-type, to carbon-rich stars. This sequence, which reflects progressing evolution, is characterised by an increasing abundance of atmospheric carbon (C), where the oxygen-rich chemistry (C/O < 1) is assumed to correspond to the initial composition of AGB stars.

From several surveys towards carbon-rich evolved stars, it has been established that the prevailing chemistry in their circumstellar envelopes (CSEs) is very rich, with molecules varying from the simple carbon monoxide (CO) to large cyanopolyynes and hydrocarbons, including polycyclic aromatic hydrocarbons (PAHs; Boersma et al. 2006; Cherchneff 2011). The carbon-rich AGB star most studied is without a doubt the nearby (120 − 150  pc; e.g. Loup et al. 1993; Crosas & Menten 1997; Groenewegen et al. 2012; Menten et al. 2012) and high mass-loss rate (1 − 4 × 10-5  M  yr-1; e.g. Crosas & Menten 1997; Groenewegen et al. 1998; Cernicharo et al. 2000; De Beck et al. 2012) object IRC +10 216, towards which many spectral line surveys have been executed in several different wavelength ranges (e.g. Cernicharo et al. 2000, 2010; He et al. 2008; Tenenbaum et al. 2010; Patel et al. 2011).

The chemistry of oxygen-rich stars has been much less the subject of such surveys at high resolution and high sensitivity. Also, the CSEs of these stars seem to have a poorer chemical content than those of carbon-rich evolved stars (Olofsson 2005). This might, however, be largely due to an observational bias, caused by the lack of an O-rich star with a combination of mass-loss rate and distance comparable to those of IRC +10 216. Only high-sensitivity unbiased surveys of several oxygen-rich stars will provide the much needed insight into their chemical content and the chemical evolution from oxygen-rich to carbon-rich in the latest phases of low and intermediate-mass stars. This paper focuses on the CSE of the oxygen-rich AGB star IK Tau. This star loses mass at a rate of ~10-5  M  yr-1 in a wind that expands at a terminal velocity of ~18.5  km  s-1 (Maercker et al. 2008; De Beck et al. 2010; Decin et al. 2010). The CSE can be traced out to a distance of ~27 000 AU from the central star, or slightly exceeding 100′′ in radius, assuming the star is located at a distance of 250 − 265 pc (Decin et al. 2010). The great similarity of its fundamental parameters (luminosity, effective temperature, mass-loss rate) to those of IRC +10 216 allows for a direct comparison of our results with those reported for the latter.

With a solar elemental abundance of P/H ≈ 3 × 10-7 (Grevesse & Sauval 1998), phosphorus is an element only rarely observed in astrophysical environments. Thus far, HCP, PH3, the CP and CCP radicals, PO, and PN are the only P-bearing molecules that have been identified around evolved stars. All of these, except PO, were observed towards the CSE of the carbon-rich AGB star IRC +10 216 (Agúndez et al. 2007, 2008; Guélin et al. 1990; Halfen et al. 2008; Milam et al. 2008). HCP and PN were also observed in the envelope of the carbon-rich post-AGB object CRL 2688 (Milam et al. 2008), while only PO and PN have thus far been shown to be present in the CSE of an oxygen-rich evolved star, namely the red supergiant (RSG) VY CMa (Milam et al. 2008; Tenenbaum et al. 2007; Ziurys et al. 2007; Kamiński et al. 2013b).

We report on the PO and PN emission from the oxygen-rich AGB star IK Tau detected in observations with the IRAM 30 m telescope, and in an unbiased spectral line survey with the Submillimeter Array (SMA) in the spectral range 279 − 355 GHz.

2. Observations

2.1. IRAM 30 m telescope: targeted observations

Targeted line observations of PN J = 2 − 1,3 − 2 and PO J = 5/2 − 3/2,7/2 − 5/2 were performed on 2009 May 13, 14, and 15 using the EMIR heterodyne receiver system on the IRAM 30 m telescope. The observations were carried out using wobbler switching with a 120′′ beam throw, which resulted in flat spectral baselines.

The VESPA and WILMA autocorrelator backends were used, with spectral resolutions of 0.23 − 0.39 MHz and 2.0 MHz, respectively. Due to the large spread in the frequency domain of the hyperfine structure present in the rotational transitions of PO (see Sect. 4), we limited our analysis of PO to the WILMA data. For the analysis of PN we used the high-resolution VESPA data.

Data processing was done using the GILDAS/CLASS package1. Vertical and horizontal polarisation signals were averaged in order to reduce the noise. First-order polynomial baselines were subtracted from the spectra after the emission lines were masked, and the data were scaled by the main-beam efficiencies in order to obtain the main-beam temperature scale. The half-power beamwidth (HPBW) for the PN J = 2 − 1,3 − 2 observations are 26′′ and 17′′, and 22′′ and 16′′ for PO J = 5/2 − 3/2,7/2 − 5/2, respectively.

2.2. SMA: unbiased survey covering 279-355 GHz

Between 2010 January 17 and February 2, we carried out a spectral survey in the range 279 − 355 GHz of IK Tau using the SMA in the extended configuration. The spectral range of this survey covers two rotational transitions of PN (J = 6 − 5,7 − 6) and two of PO (J = 13/2 − 11/2,15/2 − 13/2). The tracks covering these transitions were taken on 2 February 2010 (~280 GHz) and 22 January 2010 (~328 GHz). Per observing track two windows of 4 GHz separated by 8 GHz were covered. The observations of IK Tau were carried out in the same setup as and contemporaneously with the observations of VY CMa. Kamiński et al. (2013a) reported on some first results for the latter. Further results and more detailed descriptions of both surveys will be described in upcoming publications.

thumbnail Fig. 1

Coverage of the uv-plane (left) and shape of the synthesised beam (right) in the SMA survey of IK Tau, at 281.91 GHz. The contours in the right panel are at 10% spacings and start at the 10% level. The 50% contour is in green, all other positive contours are blue. The negative contours, marked in red, are at − 10%.

Open with DEXTER

The shortest projected baseline in the extended configuration is 37 m, the longest 225.8 m, resulting in a characteristic spatial resolution at 345 GHz of ~09 and to the observations not being sensitive to emission structures larger than about 5′′. The uv-coverage of IK Tau for the PN(J = 6 − 5) line at 281.9 GHz is shown in the left panel of Fig. 1. It is clear that, although the tracks were shared between IK Tau and VY CMa, and the hour-angle coverage was hence limited, the uv-coverage was still sufficient to result in a nice synthesised beam, as shown in the right panel of Fig. 1.

thumbnail Fig. 2

PN emission from the CSE of IK Tau. Upper two panels:IRAM 30 m data, lower two panels: SMA survey data. The vertical dashed lines indicate the rest frequency of the respective rotational transition. For J = 2 − 1 and 3–2 we also indicate the position of the HFS components with dash-dotted lines. For J = 6 − 5 we indicate the rest frequency of NaCl ν= 2,J = 22 − 21, and the range over which was integrated to obtain the moment-zero map in Fig. 3.

Open with DEXTER

The coordinates used for the phase tracking on IK Tau are . Self-calibration could not be performed for the IK Tau observations, because the central stellar continuum source was not strong enough for this purpose. Peak continuum fluxes in the survey vary around 0.05 − 0.10 Jy beam-1 (per 2GHz window). Complex gain calibration for the IK Tau observations was performed using observations of the quasar 0423-013. The spectral bandpass was calibrated using observations of 3C 273 and flux calibration was obtained with observations of Vesta and Titan for the frequency ranges around 280 GHz and 330 GHz, respectively. For the 280 GHz range the average system temperature, Tsys, was 190 K. For the 330 GHz range Tsys averaged 140 K, except for one antenna which had a Tsys of about 200 K, on average. The correlator mode was configured with a uniform spectral resolution of 0.8 MHz per channel.

Table 1

Observed PN lines.

3. PN

Low-lying rotational transitions in the vibrational ground state of the closed-shell molecule PN (X1Σ+) show hyperfine structure (HFS) which is resolvable in laboratory measurements. This hyperfine splitting is caused by the interaction of the nuclear quadrupole moment of 14N with the electric field gradient at the nucleus (Raymonda & Klemperer 1971; Hoeft et al. 1972). HFS in PN was neglected in the present study, because the component separation is much smaller than the intrinsic line widths in the source (Cazzoli et al. 2006). Spectroscopic parameters such as frequencies and Einstein-A coefficients were obtained from the standard entry in the Cologne Database for Molecular Spectroscopy (CDMS; Müller et al. 2001, 2005).

We present the detection of PN emission in four rotational transitions. The J = 2 − 1 and 3 − 2 transitions were observed in the targeted line observations with IRAM 30 m, while the J = 6 − 5 and 7 − 6 emission were measured in the SMA survey. Table 1 lists frequencies and energies above the ground state of these transitions, and the integrated fluxes and noise levels of the observations. Figure 2 shows all four observed transitions. We indicate the positions of the HFS components of the J = 2 − 1 and 3 − 2 transitions in Fig. 2. Several of the observed narrow peaks coincide very well with the catalogue frequencies of the HFS components. Interpreting the line shape is significantly complicated by this behaviour, and it is not unambiguous what type of profiles (parabolic, flat-topped, two-horned) these transitions exhibit. The IRAM 30 m observations hence do not set straightforward constraints on the spatial extent or the optical thickness of the PN emission. The displayed SMA spectra for J = 6 − 5 and 7 − 6 were integrated over a square aperture of in size, centred on the star. The J = 7 − 6 emission is detected with a signal-to-noise ratio (S/N) of ~3 at a velocity resolution of 5  km  s-1, while the three lower-lying lines are detected with S/N in the range 5 − 11 at velocity resolutions of 2.5 − 3.0  km  s-1. Only in the case of PN J = 6 − 5 is there potential contamination, owing to a narrow NaCl ν= 2,J = 22 − 21 line at 281.932 GHz. The total flux of this narrow line (full width at zero power FWZP ≈ 3.5  km  s-1), however, represents only ~3% of the integrated flux in the range 281.895 − 281.935 GHz. This was not further taken into account.

3.1. Emitting region

We cannot directly constrain the size of the PN J = 2 − 1 and J = 3 − 2 emission regions from the IRAM 30 m observations. The only available size estimate for the PN emitting region of an oxygen-rich evolved star is for VY CMa (see Sect. 3.3). Based on this, we adopt an upper limit of 40   R for the radial extent of the J = 2 − 1 emission region, corresponding to a 065 source size. From the rotational-diagram analysis presented in the next section we conclude that this estimate is reasonable. Additionally, comparing the width of the J = 3 − 2 line to the velocity profile of IK Tau’s wind presented by Decin et al. (2010), we derive a radial extent of 25   R, equivalent to a 040 source size for the gas emitting in this transition.

Figure 3 shows the integrated-intensity map (moment-zero map), integrated over the velocity range [νLSR − 22,νLSR + 20]   km  s-1, of the J = 6 − 5 emission measured in the SMA survey. Here νLSR is IK Tauri’s LSR velocity of 33.8  km  s-1 (e.g. De Beck et al. 2010).

At the line’s frequency, the restoring beam has a major axis full width at half maximum (FWHM) of 0960 and a minor axis FWHM of 0707, at a position angle of 96.3°. From a 2-dimensional Gaussian fit to the moment-zero map, using the JMFIT task in the NRAO Astronomical Image Processing System (AIPS), we find that the emission distribution is unresolved. We derive an upper limit on the deconvolved source size of with a position angle of . This corresponds to a maximum diametrical extent of about 1.4 × 1015 cm ≈ 87 stellar radii (R), assuming the stellar parameters listed by De Beck et al. (2010, see their Table 6). Comparing the detected velocity ranges to the velocity profile of Decin et al. (2010), we however derive much smaller upper limits to the emission-region sizes. For the J = 6 − 5 and J = 7 − 6 transitions we find, respectively, maximum radial extents of ~4 × 1014 cm (~12   R) and ~3 × 1014 cm (~9   R), i.e. maximum source sizes of 020 and 015.

thumbnail Fig. 3

Moment-zero map of the PN J = 6 − 5 transition. The emission was integrated over the velocity range [νLSR − 22,νLSR + 20]   km  s-1. Contours start at 2σ (1.0  Jy  beam-1  km  s-1), with 1σ increments. The synthesised beam FWHM is represented by the filled ellipse in the bottom left corner.

Open with DEXTER

3.2. Rotational temperature diagram

From the construction of a rotational temperature diagram (Snyder et al. 2005, see their Appendix) based on the four emission lines presented in Fig. 2 we can derive a rotational temperature of 30.5 ± 14.2 K and a column density of 2.2 ± 1.2 × 1015 cm-2 for PN (see Fig. 4). This method assumes that the emission is optically thin and excited under conditions of local thermodynamic equilibrium (LTE). We point out that the rotational diagram was constructed assuming different source sizes for the different transitions, in accord with the constraints presented in Sect. 3.1. We note that the four observed lines cover only a small range in excitation energy (~5 − 65 K) and as such leave large uncertainties on the derived values. Assuming a mass-loss rate of 1 × 10-5  M  yr-1, a constant expansion velocity of 15  km  s-1, and an abundance cut-off at rmax = 1.25 × 1015 cm ≈ 40   R, we obtain an abundance X(PN/H2) = 2.8 ± 1.5 × 10-7, suggesting that up to half of all available P could be locked into PN.

The assumption of optically thin emission might hold, but that of LTE probably does not. Kinetic temperature profiles for the CSE of IK Tau, such as those presented for the radiative transfer models by e.g. Decin et al. (2010, see their Fig. 4) and Maercker et al. (2008, see their Fig. 4) give gas kinetic temperatures Tkin ≳ 300 K for r ≲ 40   R. Based on the large difference we find here with the derived Trot, we suggest that PN is not excited under LTE conditions, although the rotational temperature diagram leads to an unambiguous fit. The PN lines are hence subthermally excited, since Trot ≪ Tkin. The large dipole moment of PN (μPN = 2.75 Debye; CDMS; Raymonda & Klemperer 1971; Hoeft et al. 1972) could explain this discrepancy between Trot and Tkin, in the sense that PN could be strongly affected by radiative excitation to its first vibrational state (ν= 1) at ~7.5 μm. To say more about this, however, observations of higher-J transitions in the ground and excited vibrational levels are needed, and a detailed radiative transfer analysis of the PN emission needs to be carried out. This is outside the scope of this paper.

3.3. Comparison to IRC +10 216 and VY CMa

Patel et al. (2011) identified emission from the PN J = 7 − 6 transition in the SMA survey of IRC +10 216 as line 251, while the PN J = 6 − 5 line was not covered in their survey (294 − 355 GHz). The line flux in the velocity range [νLSR − 20,νLSR + 20]   km  s-1 is ~5 Jy. Assuming a distance of 150 pc and a mass-loss rate of 2 × 10-5  M  yr-1 for IRC +10 216, and a distance of 260 pc and a mass-loss rate of 1 × 10-5  M  yr-1 for IK Tau, we find that the IK Tau J = 7 − 6 emission (Table 1) is approximately 2.5 times as strong.

Based on the analysis of three lines (one of which was blended with a 30SiC2 emission line) Milam et al. (2008) derived a column density of 1 × 1013 cm-2 for IRC +10 216, corresponding to an abundance X(PN/H2) ≈ 3 × 10-10 for a mass-loss rate of 3 × 10-5  M  yr-1. Agúndez et al. (2007) found a column density of 1.6 × 1012 cm-2, i.e. almost an order of magnitude lower. Milam et al. (2008) derived an e-folding radius of 4 × 1016 cm (≈36′′) for a Gaussian PN abundance profile, while Agúndez et al. (2007) derived a distribution of PN peaking in a shell at ~20 − 30′′or ~1000 − 1500   R (see their Fig. 3). Agúndez et al. (2007) proposed that the formation of PN in the CSE of IRC +10 216 follows from neutral-neutral reactions in the outer wind. These reactions involve CP and CN – photodissociation products of HCP and HCN – and are hence probably not as significant in an O-rich CSE. Moreover, the SMA observations of IK Tau support the formation of PN in the inner regions of the CSE, at a few R, as opposed to what was derived for IRC +10 216.

thumbnail Fig. 4

Rotational temperature diagram for PN emission. The red and blue fits only take into account the IRAM 30 m or the SMA data, respectively, while the grey fit is based on all four measurements.

Open with DEXTER

Combining the single-dish data of PN in VY CMa of Milam et al. (2008) with the SMA data obtained in the survey mentioned earlier (Kamiński et al. 2013b) and from Fu et al. (2012), we derive a column density of 3 × 1015 cm-2. Assuming a mass-loss rate of 10-4  M  yr-1 and a constant expansion velocity of 35  km  s-1, this leads to an abundance X(PN/H2) ≈ 1 × 10-8. The PN emitting region in the case of VY CMa covers a radial extent out to ~41   R.

Our derived abundance X(PN/H2) ≈ 3 × 10-7 for IK Tau and the newly derived X(PN/H2) ≈ 1 × 10-8 for VY CMa imply that PN is by roughly two to three orders of magnitude more abundant in O-rich environments than in C-rich environments. Since there is no reason to assume a different atmospheric abundance of P, and given the similar stellar and wind parameters of IK Tau and IRC +10 216, we conclude that there must be a fundamental difference in the phosphorous chemistry between the respective chemical types, which influences both the abundance and the distribution of PN in the CSE. In Sect. 5 we further discuss the observed and predicted abundances of different P-bearing species.

Table 2

PO components within the observed frequency ranges, ordered according to increasing frequency.

thumbnail Fig. 5

Parts of the SMA survey spectrum corresponding to transitions of PO, ordered according to increasing frequency from top to bottom. All components are indicated as they appear in the last column of Table 2. Their theoretical intensities, relative to the strongest within the given spin component of the rotational transition, are indicated by the height of the blue bars. Note that the model predictions for some components are too weak to be plotted. The frequency ranges integrated to obtain the line fluxes used in the rotational diagram and listed in Table 3 are indicated with the grey-shaded areas. The Λ components presented in Fig. 6 are labelled with e and f in the top panel. Component 1 of the transition is not shown since it falls in a high-noise region of the spectrum. The top panel shows the contribution of the SO2(160,16 − 151,15) transition at 283.465 GHz.

Open with DEXTER

thumbnail Fig. 6

Integrated intensity maps of the two strongest PO () emission features. Contours start at 0.5  Jy  beam-1  km  s-1 (1.7σ), with 0.5  Jy  beam-1  km  s-1 increments. The synthesised beam FWHM is represented by the filled ellipse in the bottom left corner.

Open with DEXTER

thumbnail Fig. 7

IRAM 30 m observations of PO, ordered according to increasing frequency from top to bottom. All components are indicated as they appear in the last column of Table 2. Their theoretical intensities, relative to the strongest within the given spin component of the rotational transition, are indicated by the height of the blue bars. Note that the model predictions for some components are too weak to be plotted. The frequency ranges integrated to obtain the line fluxes used in the rotational diagram and listed in Table 3 are indicated with the grey-shaded areas.

Open with DEXTER

4. PO

The rotational spectrum of the vibrational ground state of PO (X2Πr) in the submillimetre range is characterised by fine and hyperfine structure. The lower and upper spin components, and , respectively, of every rotational transition J′ − J are separated by ~320 K in energy. The HFS components correspond to the transitions F′ − F that belong to either the e or fΛ component. The intrinsically strongest components for any rotational transition are those with F′ − F = 1 in the component. See Table 2 for the rest frequencies, excitation energies and Einstein-A coefficients of the observed transitions (as taken from CDMS, following Kawaguchi et al. 1983; Kanata et al. 1988; Bailleux et al. 2002).

PO was for the first time detected in space by Tenenbaum et al. (2007) who measured the Λ components for and at ~240 GHz and ~283.5 GHz, respectively, towards VY CMa. We now present the first detections of PO towards an O-rich AGB star.

Figure 5 shows the PO and spectra of IK Tau measured with the SMA, extracted for a 1″ × 1″ area around the central star. The F′ − F = 1 components of with an excitation energy of ~50 K are clearly detected and moment-zero maps of these components are shown in Fig. 6. In these maps, both components are detected with a S/N ≈ 4.5 and we find that the emission is completely unresolved by the SMA beam. The detection of the components at ~66 K is uncertain owing to high noise at this frequency. The higher-energy and transitions, at 370.8 K and 386.7 K, respectively, were not detected.

The IRAM 30 m observations of the , 7/2 − 5/2 transitions (see Fig. 7) bear further evidence for the presence of PO in the CSE of IK Tau. Analogous to what was seen for the SMA observations, the lower-energy components are detected, while the components are not detected with the present sensitivity.

Owing to the low S/N we cannot firmly establish the velocity range spanned by the PO emission lines in the SMA data, but we find velocities not reaching more than ≈7  km  s-1, implying that the observed emission most likely does not arise from the fully accelerated wind. The IRAM 30 m spectra, on the other hand, indicate that the emission could arise from a part of the wind that is fully accelerated. However, the interpretation of the data is not unambiguous, owing to the blends of the HFS components, the rather low S/N values, and the low velocity resolution (3.9 − 5.5  km  s-1).

Based on the interferometric maps and the comparison of the measured velocities with the velocity profile for IK Tau’s wind presented by Decin et al. (2010), we derive a maximum radial extent of the PO emission distribution of 1.25 × 1015 cm ≈ 40   R.

4.1. Rotational temperature diagram

In Fig. 8 we show a rotational temperature diagram for the PO emission measured in the IRAM 30 m and SMA data. For all transitions we assumed the same spatial extent of the emitting region. For every transition, we treat the two Λ components e and f separately. When blends are involved, we split the line intensity over the different HFS components according to their theoretical line strengths. Since the first HFS component of falls in the spectral window heavily contaminated by the atmospheric H2O absorption line at ~325 GHz, we assume that it has the same integrated intensity as component number 6 (at 327.550 GHz) of this rotational transition. These HFS components, belonging to the f and e Λ components, respectively, have identical theoretical line strengths. We also assume the same noise characteristics. Relevant line fluxes, noise values, and frequency resolutions are stated in Table 3. The ranges integrated to obtain these line fluxes are indicated in Figs. 5 and 7. Considering all of the above, we wish to be cautious in interpreting the obtained results. For the IRAM 30 m data we find Trot = 9.7 ± 6.7 K and Ncol = 7.3 ± 7.3 × 1014 cm-2, while the SMA data lead to Trot = 10.8 ± 5.7 K and Ncol = 1.3 ± 3.5 × 1015 cm-2. Combining the two data sets leads to Trot = 13.1 ± 1.5 K and Ncol = 6.5 ± 2.2 × 1014 cm-2. Omission of the J = 15/2 − 13/2 data points leads to an insignificant change of the rotational temperature and column density in the combined fit. Assuming a mass-loss rate of 1 × 10-5  M  yr-1, a constant expansion velocity of 15  km  s-1, and an abundance cut-off at 1.25 × 1015 cm, analagous to our assumptions for PN, we obtain a tentative abundance X(PO/H2) in the range 5.4 × 10-8 − 6.0 × 10-7.

As for PN, we find that Trot ≪ Tkin and hence, that PO is subthermally excited. Also here, a detailed radiative transfer analysis would be needed to say more about this.

thumbnail Fig. 8

Rotational temperature diagram for PO emission measured in the IRAM 30 m and SMA data. The error bars represent 1σ uncertainties in the data. The red and blue fits only take into account the IRAM 30 m or the SMA data, respectively, while the grey fit combines these.

Open with DEXTER

4.2. Comparison to VY CMa

Tenenbaum et al. (2007) reported an abundance X(PO/H2) ≈ 9 × 10-8 for VY CMa, comparable to what they found for PN. In a new study of PO, combining the single-dish observations of Tenenbaum et al. (2007) and the SMA survey data of VY CMa, we find a column density of ~2 × 1016 cm-2, corresponding to X(PO/H2) ≈ 7 × 10-8. The size of the emitting region could be constrained to a radial extent of ~32   R. For both IK Tau and VY CMa, the derived abundances and radial extents of the PO and PN-bearing gas are comparable. Observations at higher spatial resolution will, however, be needed to put stronger constraints on this, as the emission from both molecules is not spatially resolved in the observations with the SMA.

Table 3

Overview of the observed PO transitions.

5. Discussion

5.1. Current modelling status on phosphorous chemistry

From the LTE calculations of Tsuji (1973) it follows that PS should be the major carrier of P in the gas phase of an O-rich CSE, and that PO, PN, and PH are of much less importance. In case of very high pressures PH3 would be the major P-sink in the gas phase2.

Willacy & Millar (1997) presented a model for the chemistry in the CSE of O-rich AGB stars at radii larger than R0 = 2 × 1015 cm, corresponding to about 60   R in the case of IK Tau. At the inner radius of their model, they injected PH3 as a parent molecule, with an abundance of 3 × 10-8. Atomic P, P+, PH, and PH2 are the most abundant phosphorous species in their model, while PO reaches an abundance peak of ~3 × 10-10, and only has an abundance above 10-10 for a narrow shell in the range 1 − 3 × 1016 cm, i.e. 700−2100   R in the case of IK Tau.

MacKay & Charnley (2001) described the chemistry at radii larger than R0 = 1 × 1015 cm, corresponding to about 30   R in the case of IK Tau. The chemical processes in this region of the envelope are strongly linked to photodissociation of the CSE material. They assumed PS to be the dominant P-bearing species at R0 in the case of O-rich stars. The abundance of PS injected into the outer layers of the CSE (r > R0) was assumed to correspond to 10% of the cosmic abundance of P, where all the rest is locked up in solid dust. They found that only low abundances (peaking at ~10-10) of PO could be produced in the outer envelope and that essentially no other P-bearing species would be formed. They report a high atomic P abundance in the photodissociation region, which coincides with the presence of reactive species such as the OH radical. Neutral-neutral processes could in this case, if they are sufficiently fast, contribute to the formation of P-bearing molecules.

Agúndez et al. (2007) presented thermochemical equilibrium models for the region 1−9   R for O-rich and C-rich AGB stars (see their Fig. 2). In the O-rich model PO, PS and P2 reach abundances up to ~10-7 at radii below 4   R. The abundances of PN and PH3 do not exceed 10-10 and 10-12, respectively. At radii 4   R < r < 9   R, P4O6 is the dominant carrier of phosphorus, with an abundance ~10-7, while all other P-bearing species become insignificant. Unfortunately, Agúndez et al. (2007) did not present modelling results regarding nonequilibrium chemistry at larger radii in O-rich envelopes.

Cherchneff (2012) modelled in detail the chemistry in the inner envelope of the C-rich IRC +10 216 as it is influenced by pulsational shocks passing through these layers. She found that PN can be formed very efficiently in the post-shock gas, reaching an abundance of 4.3 × 10-7 at 5   R. The formation path of PN is in this case dominated by reactions involving CN and CP. Cherchneff (2006) reported that the abundance of CN in the post-shock gas (at 1   R) is comparable in O-rich and C-rich stars. Phosphorous species were, however, not included in this older model. It is hence difficult to assess the potential efficiency of PN formation in the inner wind of a shocked O-rich environment. Cherchneff (2012) did not include PO, PS, or PH3 in her recent study of IRC +10 216’s inner wind, all molecules that might be potentially important in the O-rich case.

From the above mentioned models, it is clear that PH3, P4O6, PS, and PO are considered the main sinks for gas phase phosphorus. Predicted fractional abundances (with respect to H2) are limited to ~10-9, leaving all species but PO barely detectable3. The main uncertainties for the P-network in the current chemical models are the depletion factor of P, and the rate coefficients and activation energies for many of the involved reactions. Often, the isovalence of P and N is assumed to derive these rates and energies, but the validity of this assumption has been questioned (Cherchneff 2012).

5.2. Comparison to observations

In contrast to the model calculations, we established that PN and PO are about equally abundant in the winds of the O-rich supergiant VY CMa and of the O-rich AGB star IK Tau. The emission of both species arises from a region with a maximum radial extent of 32 − 41   R in the case of VY CMa and an estimated 40   R in the case of IK Tau. For VY CMa the high abundances of PO and PN, with respect to the available chemical models, were tentatively explained by the fact that the CSE might be more active than that of the average O-rich AGB star assumed in the chemical models. However, as pointed out by Milam et al. (2008) and confirmed by the SMA observations, PO and PN are both present in the more quiescent spherical part of VY CMa’s CSE, which is quite probably comparable to the CSE of an O-rich AGB star. Since IK Tau is believed to be a rather well behaved Mira-type O-rich AGB star, and we can now firmly establish the presence of PO and PN with much higher than predicted abundances in its CSE, it is clear that there is a strong need for an updated treatment of the chemical networks involving P-bearing species.

We estimate that PO and PN are produced at radii r ≲ 40   R in the CSEs of both IK Tau and VY CMa. Very close to the star, shock chemistry (e.g. Cherchneff 2006, 2012) could play an important role. Further out, where dust is forming or has formed, dust-gas chemistry could become important. To date, however, this type of chemical modelling has not been presented for AGB stars, owing to the many uncertainties involved in the interpretation of dust-gas interaction. Photodissociation chemistry typically becomes important at radii r ≈ 1016   cm (a few 100 stellar radii). The latter is thus, in first instance, not as urgent to understand in terms of phosphorous chemistry as the first two. It is, however, clear from the literature and from this work, that the chemical network involving phosphorus is still very uncertain and that further efforts are needed to understand the phosphorous chemistry throughout the circumstellar environments of evolved stars of all chemical types.

6. Conclusions

We present the first observations of PN and PO emission towards an oxygen-rich AGB star. We analysed dedicated line observations with the IRAM 30 m telescope and data obtained in the course of an unbiased SMA survey which covers the range 279 − 355 GHz. Emission is detected in four rotational transitions for each species: J = 2−1,3−2,6−5, and 7−6 for PN, and J = 5/2−3/2,7/2−5/2,13/2−11/2, and 15/2−13/2 for PO. We detected 13 different spectral features corresponding to different fine and hyperfine components of the PO transitions. From the observed lines we estimate an abundance X(PN/H2) = 2.8 ± 1.5 × 10-7 and we derive X(PO/H2) to lie in the range 5.4 × 10-8−6.0 × 10-7. A detailed radiative transfer study is needed to further constrain these abundances, but in order to successfully do this, observations with higher S/N are needed, as well as collision rate coefficients. Observations at higher spatial resolution will also help constrain the spatial distribution of both molecules in the CSE. Full science capabilities of the Atacama Large (sub)Millimetre Array (ALMA) will lead to a spatial resolution in the range , corresponding to about 0.7−5   R for IK Tau and VY CMa. We should hence be able to trace where the molecules are produced and destroyed, directly constraining the potential chemical formation/destruction routes.

Our abundance estimates for PO and PN exceed by several orders of magnitude the abundances predicted by models of phosphorous chemistry in O-rich CSEs. At the same time, no lines of PS or PH3 are detected, although these molecules are often predicted or assumed to be the more abundant P-bearing species in these CSEs, and the parent molecules in the P-network. We suggest that PO and PN are the major gas-phase reservoirs of phosphorus in O-rich environments like IK Tau’s and VY CMa’s CSEs.

The influence of shocks on the inner-wind chemistry of O-rich evolved stars should be further studied and extended to include chemical networks involving P-bearing species. Reaction rates and activation energies of chemical reactions involving P-bearing species should be (better) determined. Another large uncertainty in the current models that treat P-bearing species is the depletion factor of P throughout the CSE.

It will be important to attempt to trace the parent molecules in the P-network. For O-rich stars there should be a focus on further observing PO and PN, but also on PS and PH3, since the latter two are often put forward as major P-carriers. For C-rich stars, there should most likely be a focus on HCP, CP, PH3, and PN. It would hence be important to try and trace P-bearing species in the CSEs of a substantial sample of evolved stars, covering the different chemical types (O, S, C), but also the different pulsational types, in order to probe the influence of pulsation/shock strength on the phosphorous chemistry.


2

Further details on or explanation of this were not given by Tsuji (1973).

3

We do not consider P4O6 in the further discussion. It has a tetrahedral symmetry, and due to the lack of a permanent dipole moment pure rotational transitions in the (sub)millimeter domain are not observable (Carbonnière & Pouchan 2008).

Acknowledgments

The Submillimeter Array is a joint project between the Smithsonian Astrophysical Observatory and the Academia Sinica Institute of Astronomy and Astrophysics and is funded by the Smithsonian Institution and the Academia Sinica.

References

All Tables

Table 1

Observed PN lines.

Table 2

PO components within the observed frequency ranges, ordered according to increasing frequency.

Table 3

Overview of the observed PO transitions.

All Figures

thumbnail Fig. 1

Coverage of the uv-plane (left) and shape of the synthesised beam (right) in the SMA survey of IK Tau, at 281.91 GHz. The contours in the right panel are at 10% spacings and start at the 10% level. The 50% contour is in green, all other positive contours are blue. The negative contours, marked in red, are at − 10%.

Open with DEXTER
In the text
thumbnail Fig. 2

PN emission from the CSE of IK Tau. Upper two panels:IRAM 30 m data, lower two panels: SMA survey data. The vertical dashed lines indicate the rest frequency of the respective rotational transition. For J = 2 − 1 and 3–2 we also indicate the position of the HFS components with dash-dotted lines. For J = 6 − 5 we indicate the rest frequency of NaCl ν= 2,J = 22 − 21, and the range over which was integrated to obtain the moment-zero map in Fig. 3.

Open with DEXTER
In the text
thumbnail Fig. 3

Moment-zero map of the PN J = 6 − 5 transition. The emission was integrated over the velocity range [νLSR − 22,νLSR + 20]   km  s-1. Contours start at 2σ (1.0  Jy  beam-1  km  s-1), with 1σ increments. The synthesised beam FWHM is represented by the filled ellipse in the bottom left corner.

Open with DEXTER
In the text
thumbnail Fig. 4

Rotational temperature diagram for PN emission. The red and blue fits only take into account the IRAM 30 m or the SMA data, respectively, while the grey fit is based on all four measurements.

Open with DEXTER
In the text
thumbnail Fig. 5

Parts of the SMA survey spectrum corresponding to transitions of PO, ordered according to increasing frequency from top to bottom. All components are indicated as they appear in the last column of Table 2. Their theoretical intensities, relative to the strongest within the given spin component of the rotational transition, are indicated by the height of the blue bars. Note that the model predictions for some components are too weak to be plotted. The frequency ranges integrated to obtain the line fluxes used in the rotational diagram and listed in Table 3 are indicated with the grey-shaded areas. The Λ components presented in Fig. 6 are labelled with e and f in the top panel. Component 1 of the transition is not shown since it falls in a high-noise region of the spectrum. The top panel shows the contribution of the SO2(160,16 − 151,15) transition at 283.465 GHz.

Open with DEXTER
In the text
thumbnail Fig. 6

Integrated intensity maps of the two strongest PO () emission features. Contours start at 0.5  Jy  beam-1  km  s-1 (1.7σ), with 0.5  Jy  beam-1  km  s-1 increments. The synthesised beam FWHM is represented by the filled ellipse in the bottom left corner.

Open with DEXTER
In the text
thumbnail Fig. 7

IRAM 30 m observations of PO, ordered according to increasing frequency from top to bottom. All components are indicated as they appear in the last column of Table 2. Their theoretical intensities, relative to the strongest within the given spin component of the rotational transition, are indicated by the height of the blue bars. Note that the model predictions for some components are too weak to be plotted. The frequency ranges integrated to obtain the line fluxes used in the rotational diagram and listed in Table 3 are indicated with the grey-shaded areas.

Open with DEXTER
In the text
thumbnail Fig. 8

Rotational temperature diagram for PO emission measured in the IRAM 30 m and SMA data. The error bars represent 1σ uncertainties in the data. The red and blue fits only take into account the IRAM 30 m or the SMA data, respectively, while the grey fit combines these.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.