ATLASGAL-selected massive clumps in the inner Galaxy. V. Temperature structure and evolution

Context: Observational identification of a solid evolutionary sequence for high-mass star-forming regions is still missing. Spectroscopic observations give the opportunity to test possible schemes and connect the phases identified to physical processes. Aims: We aim to use the progressive heating of the gas caused by the feedback of high-mass young stellar objects to prove the statistical validity of the most common schemes used to observationally define an evolutionary sequence for high-mass clumps, and characterise the sensitivity of different tracers to this process. Methods: From the spectroscopic follow-ups carried out towards submillimeter continuum (dust) emission-selected massive clumps (the ATLASGAL TOP100 sample) with the IRAM 30?m, Mopra, and APEX telescopes between 84 GHz and 365 GHz, we selected several multiplets of CH3CN, CH3CCH, and CH3OH emission lines to derive and compare the physical properties of the gas in the clumps along the evolutionary sequence, fitting simultaneously the large number of lines that these molecules have in the observed band. Our findings are compared with results obtained from optically thin CO isotopologues, dust, and ammonia from previous studies on the same sample. Results: The chemical properties of each species have a major role on the measured physical properties. Low temperatures are traced by ammonia, methanol, and CO (in the early phases), the warm and dense envelope can be probed with CH3CN, CH3CCH, and, in evolved sources where CO is abundant in the gas phase, via its optically thin isotopologues. CH3OH and CH3CN are also abundant in the hot cores, and we suggest that their high-excitation transitions are good tools to study the kinematics in the hot gas associated with the inner envelope surrounding the young stellar objects that these clumps are hosting. All tracers show, to different degrees according to their properties, progressive warming with evolution. The relation between gas temperature and the luminosity-to-mass (L/M) ratio is reproduced by a simple toy model of a spherical, internally heated clump. Conclusions: The evolutionary sequence defined for the clumps is statistically valid and we could identify the physical processes dominating in different intervals of L/M. For L/M ? 2 L?M?-1 a large quantity of the gas is still accumulated and compressed at the bottom of the potential well. Between 2 L?M?-1 ? L/M ? 40 L?M?-1 the young stellar objects gain mass and increase in luminosity; the first hot cores hosting intermediate- or high-mass ZAMS stars appear around L/M ~ 10 L?M?-1. Finally, for L/M ? 40 L?M?-1 Hii regions become common, showing that dissipation of the parental clump dominates