Coagulation of Agglomerates Consisting of Polydisperse Primary Particles

The ballistic agglomeration of polydisperse particles is investigated by an event-driven (ED) method and compared to the coagulation of spherical particles and agglomerates consisting of monodisperse primary particles (PPs). It is shown for the first time to our knowledge that increasing the width or polydispersity of the PP size distribution initially accelerates the coagulation rate of their agglomerates but delays the attainment of their asymptotic fractal-like structure and self-preserving size distribution (SPSD) without altering them, provided that sufficiently large numbers of PPs are employed. For example, the standard asymptotic mass fractal dimension, D_f, of 1.91 is attained when clusters are formed containing, on average, about 15 monodisperse PPs, consistent with fractal theory and the literature. In contrast, when polydisperse PPs with a geometric standard deviation of 3 are employed, about 500 PPs are needed to attain that D_f. Even though the same asymptotic D_f and mass-mobility exponent, D_fm, are attained regardless of PP polydispersity, the asymptotic prefactors or lacunarities of D_f and D_fm increase with PP polydispersity. For monodisperse PPs, the average agglomerate radius of gyration, r_g, becomes larger than the mobility radius, r_m, when agglomerates consist of more than 15 PPs. Increasing PP polydispersity increases that number of PPs similarly to the above for the attainment of the asymptotic D_f or D_fm. The agglomeration kinetics are quantified by the overall collision frequency function. When the SPSD is attained, the collision frequency is independent of PP polydispersity. Accounting for the SPSD polydispersity in the overall agglomerate collision frequency is in good agreement with that frequency from detailed ED simulations once the SPSD is reached. Most importantly, the coagulation of agglomerates is described well by a monodisperse model for agglomerate and PP sizes, whereas the detailed agglomerate size distribution can be obtained by scaling the average agglomerate size to the SPSD.