PDRs4All VI Probing the photochemical evolution of PAHs in the Orion Bar using machine learning techniques

Context. Extraordinary observations of the Orion Bar by JWST have shown, for the first time, the incredible richness of polycyclic aromatic hydrocarbon (PAH) emission bands and their variation on very small scales. These variations are the result of photochemical evolution of the PAH carrier.
Aims. We aim to probe the photochemical evolution of PAHs across the key zones of the ideal photodissociation region (PDR) that is the Orion Bar using unsupervised machine learning.
Methods. We used JWST NIRSpec IFU and MIRI MRS observations of the Orion Bar from the JWST Early Release Science programme PDRs4All (ID: 1288). We levered bisecting k-means clustering to generate highly detailed spatial maps of the spectral variability in the 3.2–3.6, 5.95–6.6, 7.25–8.95, and 10.9–11.63 μm wavelength regions. We analysed and subsequently described the variations in the cluster profiles and connected them to the conditions of the physical locations from which they arise. We interpreted the origin of the observed variations with respect to the following key zones: the H II region, the atomic PDR zone, and the layers of the molecular PDR zone stratified by the first, second, and third dissociation fronts (DF 1, DF 2, and DF 3, respectively).
Results. Observed PAH emission exhibits spectral variation that is highly dependent on the spatial position in the PDR. We find the 8.6 μm band to behave differently than all other bands, which vary systematically with one another. Notably, we find a uniform variation in the 3.4–3.6 μm bands and 3.4/3.3 intensity ratio. We attribute the carrier of the 3.4–3.6 μm bands to a single side group attached to very similarly sized PAHs. Further, cluster profiles reveal a transition between characteristic profile classes of the 11.2 μm feature from the atomic to the molecular PDR zones. We find the carriers of each of the profile classes to be independent, and reason the latter to be PAH clusters existing solely deep in the molecular PDR. Clustering also reveals a connection between the 11 .2 and 6.2 μm bands and that clusters generated from variation in the 10.9–11.63 μm region can be used to recover those in the 5.95–6.6 μm region.
Conclusions. Clustering is a powerful and comprehensive tool for characterising PAH spectral variability on both spatial and spectral scales. For individual bands as well as global spectral behaviours, we find ultraviolet processing to be the most important driver of the evolution of PAHs and their spectral signatures in the Orion Bar PDR.