Mechanisms me reviewed for thermal decomposition of selected hydrocarbons that serve as models for processing of coal. The hydrocarbon classes emphasized are aromatics, alkylaromatics, a,o-diarylalkanes, and hydroaromatics, especially tetralin. Perturbations by added sources of hydrogen, especially H2 and hydroaromatics, are also covered. Free-radical pathways dominate. Emphasis is placed on quantitative understanding of elementary radical processes for bond breaking, bond making, and hydrogen transfer and on the response of the competitions among these to hydrocarbon structure and reaction conditions. Selected implications for coal processing are drawn.

Gaps in the voluminous data on pyrolysis of polyethylene that impede mechanistic understanding are highlighted, especially for H:C material balances, product distributions at varying conversions in isothermal closed systems, and inconsistencies between GC and FIMS analyses of products with lower volatility. Thermochemical kinetic estimates are made for the rates of various initiation processes; these suggest that molecular disproportionation contributes to chain initiation, along with homolysis, at lower Mn. Simulations indicate that the standard statistical test for random scission, a linear relationship between log Nc and c for volatile products, is not generally valid in open systems, and its empirical observation implies restrictions on the dependence of the rate of volatilization on c.Simulations of the initial distributions of volatile products and residue functionalities were performed based on propagation rate constants estimated by thermochemical kinetic procedures. The practice of deducing the amount of backbiting from positive deviations at lower c from the linear log N c-c relationship established for higher c values is shown to underestimate the fraction of backbiting because the latter will give some products at every c so long as the 1,5-shift is not exclusive (k15 . k16 > k14). Compositions of volatiles at finite conversions were simulated by formal superposition of (a) backbiting and unzipping products predicted from this kinetic model and (b) random scission products, which form only after multiple bond cleavages, predicted from a statistical model. Comparisons with experimental data showed some success but were limited by data gaps and inconsistencies.

No abstract

Empirical structure-reactivity correlations are developed for log k(298), the gas-phase rate constants for the reaction (Cl(•) + HCR(3) → ClH + CR(3)(•)). It has long been recognized that correlation with Δ(r)H is weak. The poor performance of the linear Evans-Polanyi formulation is illustrated and was little improved by adding a quadratic term, for example, by making its slope smoothly dependent on Δ(r)H [η ≡ (Δ(r)H - Δ(r)H(min))/(Δ(r)H(max) - Δ(r)H(min))]. The "polar effect" ((δ-)Cl---H---CR(3)(δ+))(++) has also been long discussed, but there is no formalization of this dependence based on widely available independent variable(s). Using the sum of Hammett constants for the R substituents also gave at best modest correlations, either for σ(para) or for its dissection into F (field/inductive) and R (resonance) effects. Much greater success was achieved by combining these approaches with the preferred independent variable set being either [(Δ(r)H)(2), Δ(r)H, ΣF, and ΣR] or [η, Δ(r)H, ΣF, and ΣR]. For 64 rate constants that span 7 orders of magnitude, these correlation formulations give r(2) > 0.87 and a mean unsigned deviation of <0.5 log k units, with even better performance if primary, secondary, and tertiary reaction centers are treated separately.

No abstract