The def_diam and def_diam_g parameters have been assumptions within the CMAQ code for a long time, and have changed little over the years. They came originally from the Binkowski and Roselle paper as you point out, based approximately on the box model inputs in Table 4. There are some differences though, that I believe are partially from constraining the contemporary CMAQ model with available data back in the early 2000’s and from modest changes over the years to accommodate model spinup scenarios that seemed overly sensitivie to, for example, large standard deviation in the coarse mode.
Note however, that CMAQ in general is not very sensitive to the values of these parameters. They are important for describing the ratio in magnitude among the minimum values of number, surface area and mass, and in setting the default size distribution if flawed data are provided by the boundary condition files. They are not expected to play a significant role in the large majority of calculations determining the evolution of aerosol moments within the domain.