Wasteomics: a quantitative, molecular-level workflow for complex reaction media, waste, and realistic feeds. We describe the complete reaction medium at the molecular level, using realistic feeds instead of model compounds, and we convert that description into reaction networks and kinetics. The workflow covers the gas, the liquid, and the solid, including the species that remain trapped on the catalyst.
Problem statement. In heterogeneous catalysis, the reaction medium is a mixture of reactants, intermediates,s and products. Part of it is adsorbed or trapped on the catalyst and elsewhere in the reactor. These media are multiphase, contain unstable species, es and are difficult to analyze. Some species also act as catalysts or as deactivating agents for the conversion of others.
The usual practice is to study model molecules or surrogates to resolve mechanisms, microkinetics, or spectroscopy. A model compound, even an abundant one, does not reproduce the kinetics of the mixture [110] . Analytical techniques keep improving, but each one targets a fraction or a type of species. No single technique resolves the whole mixture.
The gap is largest for the feeds that a circular economy depends on: plastic waste, tires, sewage sludge, biomass, algae and seaweed, and refinery streams of low value.
Our approach, scale by scale. The -omics fields inspire our approach: we characterize and quantify the full pool of molecules and relate it to structure, function,n and dynamics. The video follows one sample from the process to the reaction network, and each paragraph below matches one scene.
1. Process: real feeds. We work with the feeds as they are. These include waste polyolefins [86] , scrap tires [87, 98, 107] , oxygenated plastics such as PMMA and PET [143, 172] , marine heavy-fuel waste oils [123] , crude oil [125] , bio-oils from biomass [99, 100, 186] and algae [193] . Each one is a mixture of thousands of molecules. We study two upcycling routes:
2. Reactor: every phase is sampled. We sample every phase that leaves the reactor: the gas, the liquid, and the spent catalyst. The spent catalyst holds the trapped and deactivating species, which are usually left out of the mass balance [138, 196] .
3. Analysis: several techniques, one dataset. No single instrument reads the whole sample. We combine high-resolution mass spectrometry with chromatographic, spectroscopic,c and thermal techniques, and we merge their results into one quantitative dataset per experiment [186, 194] . The same protocols apply to the coke extracted from the catalyst [138] .
4. Molecules: the mixture is mapped, not averaged. Each signal becomes a molecule with formula, class, s, s, and concentration. This molecular map shows where heteroatoms (O, N, S) go during upgrading [125, 172] and how coke precursors form and grow on zeolites with time on stream [109, 138, 196] . In some cases, coke is not only a problem: on a hierarchical zeolite, partial coking blocks the sites that drive side reactions [122] .
5. Network: from molecular data to kinetics. Molecules and lumps are linked into reaction networks and kinetic models, including the species that deactivate the catalyst [82, 194, 195] . This is the step that turns analysis into a design tool for catalysts and processes.
Goals. Our goals are grouped in these lines of work:
Applications. We apply this competency to:
