Main

In the Outer Galaxy, roughly defined as more than 16 kpc from the Galactic Centre, molecular clouds are typically small and sparse30, resulting in a lower star formation rate24. Supernovae are also extremely rare in this part of the Galaxy, with none known beyond 13 kpc (ref. 26). Elemental abundance gradients estimated for carbon, oxygen and nitrogen support a decline in star formation in the Outer Galaxy. The gradients suggest a decrease in abundances by factors of 10, 3 and 5, respectively, for C, O and N, from 8.5 to 23 kpc (ref. 31). However, these studies do not typically characterize the metallicity of the Galaxy beyond 13–16 kpc (refs. 31,32), such that extrapolation beyond these distances is highly uncertain.

Recent observations of so-called Galactic edge clouds have demonstrated that dense gas at large galactocentric distances (approximately 13–24 kpc) contains unusual amounts of gas-phase carbon-bearing molecules such as methanol33. Therefore, we conducted a search for the phosphorus-bearing molecules PO and PN in one of these star-forming regions, WB89-621 (right ascension (α) = 05 h 17 min 13.3 s, declination (δ) = 39°22′14″ (J2000.0)), located at a galactocentric radius of 22.6 kpc (ref. 24). PN and PO had previously been found in molecular clouds, including Orion-KL, AFGL 5142 and W51 (refs. 25,34,35,36), as well as in circumstellar envelopes of evolved stars. However, these molecular observations have been limited to less than 10 kpc of the Galactic Centre (Fig. 1). Millimetre-wave observations of WB89-621 were conducted with the Arizona Radio Observatory (ARO) 12 m telescope and the Institut de Radioastronomie Millimétrique (IRAM) 30 m telescope. The J = 3 → 2 and J = 2 → 1 transitions of PN at 2 and 3 mm were identified, as well as the four hyperfine/lambda-doubling components of the J = 2.5 → 1.5 transition of PO (Table 1).

Fig. 1: Currently known Galactic distribution of phosphorous.
figure 1

A graphic illustrating the Galactic distribution of phosphorus in the range 0–25 kpc from the Galactic Centre, as identified from stellar and interstellar observations. Detections by atomic transitions in stellar photospheres are denoted by green stars (from ref. 2 and references therein), in diffuse clouds by purple circles (ref. 16 and those therein17,18,19), and in planetary nebulae by blue triangles6,7,8,9,10,11,12,13,14,15, while magenta circles and cyan diamonds mark the identifications in molecular clouds and circumstellar envelopes from molecular lines25,34,35,36,40. The red square shows the position of WB89-621. The orange X indicates the solar system. The APOGEE survey detections are not shown per possible contamination. WB89-621 is the only known source of phosphorus in any form beyond 12 kpc.

Full size image
Table 1 Observations of PN (X1Σ+), PO (X2Πr) and other molecules towards WB89-621
Full size table

The measured spectra of PN and PO in WB89-621 are shown in the top six panels in Fig. 2. The spectral profiles typically exhibit narrow linewidths of approximately 1.5 km s−1 and have a velocity with respect to the local standard of rest (VLSR) near −25.4 km s−1, characteristic of molecular material in WB89-621, as illustrated in the lower two panels of Fig. 2. Here spectra of the J = 1 → 0 transition of N2H+ and the Jτ = 80 → 71, A of CH3OH are shown. In the N2H+ features, nitrogen quadrupole hyperfine components are resolved. Note that aside from first-order baseline subtraction, no other data reduction or model dependent analysis has been done, unlike with stellar spectra2. Line parameters for all species, determined from Gaussian fits to the line profiles, are summarized in Table 1. The linewidths and LSR velocities are consistent among these molecules. Note that the PO and PN detections are separate measurements, and each give an independent assessment of the phosphorus abundance.

Fig. 2: Spectra of molecular rotational transitions of PO, PN and other species in molecular cloud WB89-621.
figure 2

Spectra are plotted as intensity (TA*), in mK, versus VLSR, in kilometres per second. The top two panels show the J = 2 → 1 and J = 3 → 2 transitions of PN, followed by the four hyperfine/lambda-doubling components of the J = 2.5 → 1.5 transition of PO. The bottom two panels display the N2H+ J = 1 → 0 and CH3OH Jτ = 80 → 71, A transitions. Each panel includes the molecular species and quantum numbers corresponding to the rotational transition shown. The J = 2 → 1 line of PN (156 kHz spectral resolution) was measured with the 12 m antenna of the ARO, while all others were observed using the IRAM 30 m telescope (200 kHz spectral resolution).

Full size image

The abundances of PN and PO relative to H2 were calculated to be 3.0 ± 1.6 × 10−12 and 2.0 ± 1.1 × 10−11, respectively. These values are comparable to those found in the solar neighbourhood, as shown in Fig. 3. A linear fit to the PO and PN abundances with respect to distance from the Galactic Centre (RGC) suggests a slight gradient—roughly a factor of 1.5 and 2.3 decrease in abundance from 8.5 to 22.6 kpc for PO and PN, respectively. The abundances of phosphorus, derived from limited observations of atomic lines in stellar photospheres (see ref. 2 and those therein), diffuse clouds16,17,18,19 and planetary nebulae6,7,8,9,10,11,12,13,14,15 (Fig. 1) are also shown in Fig. 3, plotted on the same galactocentric distance scale. These measurements cluster primarily between 6 and 10 kpc, with no values beyond 12 kpc (ref. 7). Extrapolating from a relatively small galactocentric range to across the Galaxy is problematic; a simplistic analysis suggests a drop in abundance by a factor of 10 between 8.5 and 22.6 kpc, but such an estimate may be misleading. Note that the solar abundance of phosphorus is [P/H] approximately 2.6 × 10−7 (ref. 37).

Fig. 3: Abundances of PN and PO, as well as atomic phosphorus, as a function of distance from the Galactic Centre.
figure 3

Molecular abundances, relative to H2, are plotted with respect to RGC (kpc) for AFGL 5142, G + 0.693-0.03, W3(OH), W51, L1157, Orion-KL25, B1-a, NGC 1333-IRAS 3, Ser SMM1, L723 (ref. 36) and WB89-621. Atomic abundances from sources in Fig. 1 are also displayed. Abundances of PO, PN and atomic P are shown in red, black and blue, with 3σ estimated uncertainties. From 8.5 to 22.6 kpc, abundances of PO and PN decrease by a factor of 1.5 and 2.3, respectively, showing little decrease in the Outer Galaxy.

Full size image

Because phosphorus is thought to be produced by supernovae3, and there are few, if any, supernovae beyond 13 kpc (ref. 26), another form of elemental enrichment must be occurring. One possibility is a ‘Galactic Fountain’, where supernovae eject gas into the halo/disk-halo interface, where it cools and subsequently ‘rains’ down on the disk27. Such fountains, however, are not known to operate beyond 15 kpc. Furthermore, the infall from a fountain creates distinct clouds38 with ‘forbidden velocities’ that do not follow the galactic rotation curve27. Given the location of WB89-621, the rotation curve27,38 would indicate VLSR ≈ 0 km s−1, as opposed to the actual LSR velocity of −25.4 km s−1. Velocities less than 30 km s−1 can be accounted for by turbulence38. WB89-621 also resides roughly 300 pc (galactic latitude b = 0.8197°) above the Galactic midplane, while most fountain clouds have galactic latitudes greater than 3° and are clearly separated from the disk27. The origin of this phosphorus is also unlikely to be from extragalactic sources, such as the Large Magellanic Cloud. Although the outer regions of the Milky Way are thought to have interacted with the Large Magellanic Cloud, such dwarf galaxies are characterized by low metallicity28 and therefore are unlikely to cause phosphorus enrichment.

The source of phosphorus in the Outer Galaxy may be alternative routes in stellar nucleosynthesis. Because of the failure of the galactic chemical evolution models, it has been postulated that non-explosive massive stars or thermal-pulsing asymptotic giant branch stars (AGB) stars may be possible sources of this element29. The lower mass population of stars in the Outer Galaxy39 suggests an AGB origin. In such stars, a neutron excess can be produced in the H/He shell interface, where a pocket of 13C is generated from the addition of protons to 12C. An excess of neutrons is then created through the reaction 13C(α,n)16O that leads to overproduction of 31P and other elements. Models have been speculative about the phosphorus yields from AGB stars, ranging from ‘negligible’ to enhancements of 4–20, depending on stellar mass and metallicity3,29. The surprising number of phosphorus-bearing molecules observed in the circumstellar envelopes of AGB stars provide some additional evidence for this hypothesis40. There may be other possible reactions leading to phosphorus in AGB stars as well, such as proton addition to 30Si and α-capture on 27Al (ref. 20). Furthermore, it has been suggested that phosphorus may exhibit primary element behaviour over the metallicity range for [Fe/H] between −0.9 and 0.3 (ref. 3), meaning that it is generated by reactions directly from H and He. Extrapolating the gradient measured out to 20 kpc to 23 kpc (ref. 41), the [Fe/H] ratio in WB89-621 is −0.61. If primary behaviour is indeed attributable to this element, then its existence in the Outer Galaxy is perhaps expected.

AGB stars, which generate roughly half of all 12C (ref. 21), may also be responsible for the high abundances of C-containing molecules in the Outer Galaxy. Methanol (CH3OH), for example, has been observed in twenty edge clouds in the range 13–23 pc. The abundance of CH3OH has been found to remain fairly constant from 7 to 23 kpc (ref. 33), despite the predicted 10-fold decline in carbon over the same distances31. However, this elemental gradient is extrapolated from measurements in H ii regions that extend only out to approximately 12 kpc. Across the distance from the Sun to 12.4 kpc, the maximum extent of the measurements, the carbon abundance drops at most by a factor of two. Given the lack of data points beyond approximately 12 kpc, the true carbon gradient may substantially differ from the direct extrapolation. Studies of abundances of N, O and Fe in H ii regions in other spiral galaxies have shown a ‘flattening’ of the metallicity gradient in the outer regions42. If such dampening of the metallicity decline applies to the Milky Way, as studies of open clusters suggest42, it would certainly change the extrapolated gradient. Detailed modelling of the gas-phase elemental abundances in the outer disks of galaxies has yet to be developed42 and warrants more comprehensive investigation, as underscored by our molecular observations in WB89-621.

Methods

Observations

All molecular spectra were obtained using the ARO 12 m telescope or the IRAM 30 m telescope, located on Kitt Peak, Arizona, and Pico Veleta, Spain, respectively. Observations at the ARO 12 m telescope were taken with a dual-polarization receiver with sideband-separating mixers. Image rejection was typically greater than 18 dB. The intensity scale (TA*) for both the ARO 12 m telescope and IRAM 30 m telescope, determined by the chopper wheel method, is related to the main-beam brightness temperature, TR, by TR = TA*/ηB, where ηB is the main-beam efficiency (0.82 and 0.81, respectively). The ARO 12 m telescope observations were conducted at 2 mm with the ARO Wideband Spectrometer as the backend, configured to a frequency resolution of 156 kHz and 1 GHz bandwidth, per polarization. The IRAM measurements were taken with the Eight Mixer Receiver at 3 mm in dual-polarization mode, with the FTS 200 (fast Fourier transform spectrometer with a resolution of 200 kHz). The line parameters are provided in Table 1. Total integrations times required for the new identifications were 28.6 h at the ARO 12 m telescope and 17.4 h at the IRAM 30 m telescope, respectively.

Analysis

The column densities were calculated using the non-local thermodynamic equilibrium radiative transfer code RADEX43. The program employs the Sobolev approximation to produce line profiles to compare with measured spectra. RADEX varies the molecular column density (Ntot), gas kinetic temperature (Tk) and the H2 gas density (n(H2)). The data files, which consist of the energy levels, transitions and collisional rate information for each molecule, were obtained from the Leiden Atomic and Molecular Database44. Because only two rotational transitions of PN were measured, only two parameters could be varied at a time. The gas kinetic temperature was set to 25 K, a value determined by other molecules that had multiple transitions, such as CH3OH and CH3CN. The H2 density was varied between 1 × 105 and 1 × 107 cm−3 and the column density was varied from 1 × 109 to 1 × 1014 cm−2. Based on the two transitions of PN that were measured, a gas density of n(H2) of approximately 1.5 × 105 cm−3 was determined, which was subsequently used to model the single rotational line observed for PO. The ‘best fit’ was determined through a reduced χ2 analysis.