Disentangling multiple high-energy emission components in the Vela X pulsar wind nebula with the Fermi Large Area Telescope

Vela X is a pulsar wind nebula in which two relativistic particle populations with distinct spatial and spectral distributions dominate the emission at different wavelengths. An extended $2^\circ \times 3^\circ$ nebula is seen in radio and GeV gamma rays. An elongated cocoon prevails in X-rays and TeV gamma rays. We use 9.5 years of data from the Fermi Large Area Telescope (LAT) to disentangle gamma-ray emission from the two components in the energy range from 10 GeV to 2 TeV, bridging the gap between previous measurements at GeV and TeV energies. We determine the morphology of emission associated to Vela X separately at energies < 100 GeV and > 100 GeV, and compare it to the morphology seen at other wavelengths. Then, we derive the spectral energy distribution of the two gamma-ray components over the full energy range. The best fit to the LAT data is provided by the combination of the two components derived at energies < 100 GeV and > 100 GeV. The first component has a soft spectrum, spectral index $2.19\pm0.16^{+0.05}_{-0.22}$, and extends over a region of radius $1.36^\circ\pm0.04^\circ$, consistent with the radio nebula. The second component has a harder spectrum, spectral index $0.9\pm0.3^{+0.3}_{-0.1}$, and is concentrated over an area of radius $0.63^\circ\pm0.03^\circ$, coincident with the X-ray cocoon that had already been established to account for the bulk of the emission at TeV energies. The spectrum measured for the low-energy component corroborates previous evidence for a roll-over of the electron spectrum at energies of a few tens of GeV possibly due to diffusive escape. The high-energy component has a very hard spectrum: if the emission is produced by electrons with a power-law spectrum the electrons must be uncooled, and there is a hint that their spectrum may be harder than predictions by standard models of Fermi acceleration at relativistic shocks. (Abridged)