The direct radiative forcing of aerosols remains uncertain in global climate models. In this study, we investigate the inter-model spread of direct aerosol radiative forcing (DARF) across 14 global models within the Coupled Model Intercomparison Project Phase 6 (CMIP6) using unified radiative transfer calculation and aerosol optical parameter assumptions. The global mean DARF for clear sky in 2014 with respect to 1850 is estimated as -0.77 ± 0.52 W/m² assuming externally mixed state and -0.68 ± 0.53 W/m² for internally mixed state, whereas it is -0.49 ± 0.44 W/m² and -0.38 ± 0.44 W/m² under all sky condition. We further conduct a quantitative analysis of the contribution of DARF uncertainty by the simulated aerosol optical depth (AOD), single scattering albedo (SSA), asymmetry factor (g) and vertical distribution. It is found that for the external mixing assumption, AOD is the dominant factor, whose spread across models results in 36% of the total DARF spread. For the internal mixing assumption, SSA becomes the major factor, the spread of which also leads to 36% of the total DARF spread. g and vertical distribution combined contribute ~30% of the DARF uncertainty. The optical parameters calculated under the external mixing assumption agree better with satellite and ground observation in general. Some regional differences are noted in the results, however, the overall conclusion is that the uncertainty of simulated SSA contributes the most to the DARF uncertainty, which is closely followed by the uncertainty of AOD. Aerosol vertical distribution is also an important factor for South America and South Africa. Our results highlight the importance of absorbing and scattering aerosol fraction in DARF estimation. Besides, we find the inter-model difference in aerosol vertical distribution accounts for ~20% of the total DARF uncertainty. 

Direct aerosol radiative forcing uncertainty, aerosol optical properties