The academic field of Origins of Life seeks to explain abiogenesis: how abiotic matter can be transformed to living systems. Recent decades have seen a substantial development of prebiotic scenarios: hypotheses on the place, chemistry and physical mechanisms of abiogenesis. In this thesis, we introduce rigorous frameworks, for the systematic study of physical aspects of abiogenesis. These frameworks build upon recent insights in nonequilibrium thermodynamics, chemical reaction networks and group selection. They stress thermodynamic consistency and the fundamental structure of chemistry. In Ch.1, a critical introduction to the field of origins of life is given, highlighting what we truly know, what popular scenarios assume and historical developments that have shaped the current thinking. In Ch.2, the theoretical framework of chemical networks is introduced. We extend the framework with new tools and a criterion for a universal description of chemistry: nonambiguity. In Ch.3, we describe ways to make networks ‘open’: chemostats, CSTR, serial transfer, diffusive coupling to compartments. The concept of chemostats is extended to ‘composite chemostats’, which chemostats combinations of species. We correct the zeroth law of thermodynamics, which is shown to be violated for conserved integer quantities in stochastic thermodynamics. In Ch.4, we illustrate the concept of information in chemical networks, using a single-molecule information engine that also works in the macroscropic limit, and that extracts work from the racemization of enantiomerically pure molecules. We then move to the opposite process: purifying compositions. We illustrate a variety of chemical networks that achieve purification, and we discuss their tradeoffs. In Ch.5, we derive universal criteria for catalysis and self-replication for unambiguous chemical networks. The addition of exchange processes between phases and compartments (diffusion, evaporation, partitioning etc.) leads to emergent new forms of multicompartment autocatalysis. In Ch.6, we review the concept of chemical evolution and some of the frameworks developed for it. These frameworks focus on specific chemistries and network structures, and we show that their interpretation critically hinges on the level of coarse graining. These approaches, often treated as mutually exclusive, are united, extended and encompassed by our general framework for autocatalysis. We study structural and thermodynamic aspects of autocatalytic evolution in a single reactor, which occurs by branching processes built up from microscopic rates. The extension to multicompartment autocatalysis leads to new emergent ecological behavior (syntrophy, parasitism), favoring cooperation and spatial confinement. In Ch.7, we study the thermodynamics of making long polymers in various out-of-equilibrium situations (adsorption, recombination reactions, chemically activated ligation). We derive thermodynamic costs for the dissipative generation of random copolymer sequences. This allows to place energetic bounds on scenarios that rely on the appearance of rare structures. In Ch.8, we set up a statistical framework to study transient compartmentalization. This new form of multilevel selection has no lineages: surviving compartments vanish after growth and selection, which means contents may multiply by more than a factor 2 (experimentally: > 10^6). This mechanism is shown capable of overcoming parasite invasions, induce cooperation and lower error thresholds. Compositional noise is derived from growth kinetics. Polymerization can drastically decrease such noise, which can improve selection. On this level of description, a new parasite catastrophe emerges: a complexation catastrophe. In Ch.9, we formulate a mechanism-based scenario. The scenario is based on the structural features of chemistry, multicompartment autocatalysis and multi-level evolution, but does not specify any molecules: they can be introduced a posteriori. We provide a foundation on which rigorous future scenarios can be built.