We have studied four complex organic molecules (COMs), methyl formate ($CH_3OCHO$), dimethyl ether ($CH_3OCH_3$), formamide ($NH_2CHO$), and ethyl cyanide ($C_2H_5CN$), towards a large sample of 39 high-mass star-forming regions representing different evolutionary stages, from early to evolved phases. We aim to identify potential correlations between the molecules and to trace their evolutionary sequence through the star formation process. We analysed spectra obtained at 3, 2, and 0.9 mm with the IRAM-30m telescope. We derived the main physical parameters for each species by fitting the molecular lines. We compared them and evaluated their evolution, also taking several other interstellar environments into account. We report detections in 20 sources, revealing a clear dust absorption effect on column densities. Derived abundances are ~$10^{-10}-10^{-7}$ for $CH_3OCHO$ and $CH_3OCH_3$, ~$10^{-12}-10^{-10}$ for $NH_2CHO$, and ~$10^{-11}-10^{-9}$ for $C_2H_5CN$. The abundances of $CH_3OCHO$, $CH_3OCH_3$, and $C_2H_5CN$ are very strongly correlated (r>0.92) across ~4 orders of magnitude. $CH_3OCHO$ and $CH_3OCH_3$ show the strongest correlations in most parameters, and a nearly constant ratio (~1) over a remarkable ~9 orders of magnitude in luminosity for a wide variety of sources: pre-stellar to evolved cores, low- to high-mass objects, shocks, Galactic clouds, and comets. This indicates that COMs chemistry is likely early developed and then preserved through evolved phases. Moreover, the molecular abundances clearly increase with evolution. We consider $CH_3OCHO$ and $CH_3OCH_3$ to be most likely chemically linked: they could e.g. share a common precursor, or be formed one from the other. We propose a general scenario for all COMs, involving a formation in the cold, earliest phases of star formation and a following increasing desorption with the progressive heating of the evolving core.

Phosphorus nitride (PN) is the P-bearing species with the highest number of detections in star-forming regions. Multi-line studies of the molecule have shown that the excitation temperature of PN is usually lower than the gas kinetic temperature, suggesting that PN is likely in conditions of sub-thermal excitation. We present an analysis of PN which takes the possible sub-thermal excitation conditions into account in a sample of 24 massive star-forming regions. We observed PN (2-1), (3-2), (4-3), and (6-5) with the IRAM-30m and APEX telescopes and detected PN lines in 15 of them. Together with 9 similar sources detected in PN in previous works, we have analysed the largest sample of star-forming regions to date, made of 33 sources with 24 detections in total (among which 13 are new detections). Hence, we have increased the number of star-forming regions detected in PN by more than a factor 2. Our analysis indicates that the PN lines are indeed sub-thermally excited, but well described by a single excitation temperature. We have compared line profiles and fractional abundances of PN and SiO, a typical shock tracer, and found that almost all objects detected in PN have high-velocity SiO wings. Moreover, the SiO and PN abundances with respect to H2 are correlated over several orders of magnitude, and uncorrelated with gas temperature. This clearly shows that the production of PN is strongly linked to the presence of shocked gas, and rules out alternative scenarios based on thermal evaporation from iced grain mantles.

We present multi-transition observations of PN towards a sample of nine massive dense cores in different evolutionary stages. Using transitions with different excitation conditions, we have found for the first time that the excitation temperatures of PN are in the range 5-30 K. To investigate the main chemical route for the PN formation (surface-chemistry vs. gas-phase chemistry), and the dominant desorption mechanism (thermal vs. shock), we have compared our results with those obtained from molecules tracing different chemical and physical conditions (SiO, SO, CH3OH, and N2H+). We have found that the PN line profiles are very well correlated with those of SiO and SO in six out of the nine targets, which indicates that PN may be released by sputtering of dust grains due to shocks. This finding is corroborated by a faint but statistically significant positive trend between the PN abundance and those of SiO and SO. However, in three objects the PN lines have no hints of high velocity wings, which indicates an alternative origin of PN. Overall, our results indicate that the origin of PN is not unique, as it can be formed in protostellar shocks, but also in colder and more quiescent gas through alternative pathways.