Experimental studies have suggested that soot particles formed along different streamlines in a diffusion flame may experience different morphologies. However, the uncertainties linked to the sampling and their analysis makes difficult to quantify those differences. In this work, this problem is explored from a numerical point of view by coupling macroscopic simulations of the flame by using typical CFD codes (CoFlame, Eaves et al., 2016) with microscopic simulation of their aggregation (MCAC). This innovative coupling is one-way and consists in extracting the macroscopic Lagrangian trajectories of particles based on CoFlame simulations by considering the related temporal evolution of the flame temperature and the soot particles-gas mass transfer rates. This information is subsequently used by the Monte Carlo Discrete Element Model (MCAC) which then simulate the Brownian motion of the particles, their aggregation, their surface growth as well as nucleation and oxidation (including fragmentation) all along the residence time in the flame. Simulations show remarkably larger and more locally compact aggregates formed near the wings of the flame as compared to the centerline of the flame. Analyzing extreme morphological parameters such as anisotropy, monomers overlapping and coordination numbers reveals the complex morphology of soot including the occurrence of chain-like aggregates along the different considered streamlines. These different morphological parameters lead to the identification of considerably different soot aggregates structures. Maximum effective numbers of primary spheres are observed near the wings of the flame with values as high as 59. Maximum anisotropy coefficients as large as 15 have been observed for aggregates having a quasi-linear chain-like structure. Primary particle overlapping as high as 90% can exist locally, while population values are not larger than 50%. These aggregates show larger local compacity than those studied in previous premixed flames (Morán et al., 2021).