Spectral induced polarization (SIP) has become an increasingly popular geophysical method for hydrogeological and environmental applications. These applications include for instance the non-intrusive characterization of the textural and interfacial physicochemical properties of bentonites used as permeability barriers in landfills or to store various types of contaminants including radioactive wastes. Bentonites are mainly constituted of smectites, which have very high specific surface areas (SSA) and cation exchange capacities (CEC). Therefore, these minerals have very high electromigration and polarization current densities responsible for very high in phase and quadrature conductivities, respectively. In addition, in diluted water, the diffuse layer of smectites occupies a large fraction of the pore space and may be therefore considered as part of the pore space. In our approach [1], complex electrical conductivities of saturated unconsolidated bentonite and bentonite-sand mixtures are modeled at different salinities (NaCl) of the bulk pore water using a Donnan equilibrium model coupled to the revisited SIP model of Leroy and Revil [2]. Our complex surface conductivity model considers the DC contribution of the diffuse and Stern layers as well as the electrochemical polarization of the Stern layer coating the grains with different sizes. The macroscopic SIP model is based on the differential effective medium theory and considers the complex surface conductivity of the sand and smectite grains and the complex conductivity of the pore space. In our model, the diffuse layer of quartz sands occupies a small fraction of the pore space and is considered therefore as part of the surface of the grains. Our SIP model predicts very well the low frequency (0.1 Hz - kHz) complex electrical conductivities of bentonite and bentonite-sand mixtures, except for very low frequencies (