Résumé
Macromolecules 2020, 53, 7, 2568-2581 In order to characterize the geometrical mesh size$\xi$ , we simulate a solution of coarse-grained polymers with densities ranging from the dilute to the concentrated regime and for different chain lengths. Conventional ways to estimate$\xi$rely either on scaling assumptions which give$\xi$only up to an unknown multiplicative factor, or on measurements of the monomer density fluctuation correlation length$\xi_c$ . We determine$\xi_c$from the monomer structure factor and from the radial distribution function, and find that the identification$\xi=\xi_c$is not justified outside of the semidilute regime. In order to better characterize$\xi$ , we compute the pore size distribution (PSD) following two different definitions, one by Torquato et al. (Ref.1) and one by Gubbins et al. (Ref.2). We show that the mean values of the two distributions,$\langle r \rangle_T$and$\langle r \rangle_G$ , both display the behavior predicted for$\xi$by scaling theory, and argue that$\xi$can be identified with either one of these quantities. This identification allows to interpret the PSD as the distribution of mesh sizes, a quantity which conventional methods cannot access. Finally, we show that it is possible to map a polymer solution on a system of hard or overlapping spheres, for which Torquato's PSD can be computed analytically and reproduces accurately the PSD of the solution. We give an expression that allows$\langle r \rangle_T$to be estimated with great accuracy in the semidilute regime by knowing only the radius of gyration and the density of the polymers.