Basically, higher the zeta potential, the more the particles will repel each other in suspensionand therefore the more it will stay stable and not aggregate.
I had observed some delay at publication of “answer”. I expect that some part of below be relevant. In your question there are three or four classical concepts that would be addressed to reach a further understanding of the terminology applied to this question. Please, consider words and concepts mentioned here as a base to a future and deeper bibliography revision.
Despite of question be correlated to the nanoparticles, the proper origin of phenomena stemming of suspensions or dispersion of colloidal particles in aqueous medium. See, now, all theory about this tem has been gradually extended to non-aqueous medium with excellent results. In a broad sense, colloidal particles are particles with diameter lower than 1 micrometer, only that when the diameter tend to nanoscale of size, problems undergoes a increasing of magnitude in an inverse proportion.
The surface of colloidal particles is electrically charged. This charge is highest at small particles tending to very low values at particles with tens of micrometers, as can be verified below. This phenomenon is an intrinsic phenomenon of solids materials, a priori the magnitude this phenomenon is variable depending on the nature of solids, atomic arrangement, chemical bonds,… By convenience, let’s suppose an ordered and crystalline material that exhibits a crystalline lattice that at surface has an interface solid/gas. The crystalline lattice has all chemical bond satisfied at the bulk, at surface the absence of a continuous lattice leads to distortions of chemical bonds, creation of pontual defects, etc,…, etc. As a whole, all kind of distortion and defects at surface of particle generate a charge density at this surface. This charge density gives origin to an electric potential, in fact an electrostatic potential at surface of the particle. Nanoparticles has a great defect concentration at surface and high level of chemical bond deformation at surface, being expected major electrostatic potential at surface.
As above mentioned, “the surface” exhibits a density of charge that can be positive or negative. In aqueous medium, “the surface charge density” attracts charge of contrary sign from ions and molecules or both of the medium forming the called diffuse-layer. The opposite side of the diffuse layer exhibits a continuous decreasing of charge density, in this sense after diffuse layer the medium is electroneutral. The set described form “the diffuse electrical double-layer”, see the first layer is the charged surface (electrostatic charge), while the second layer is the diffuse layer.
The magnitude of the “charge density” determines of degree of interaction with the medium and the stability of particle in this medium. The potential decay in a exponential way from charged surface, as mentioned in another answer a potential at around 25 mV is sufficient to think in some stability degree, taking in account water and proper ionic species.
The potential zeta can be defined as the potential that exert a maximum particle electromobility, being a direct function of medium viscosity and inversely proportional to the vacuum permittivity multiplied by the static dielectric permittivity, does not considering a multiplication factor equal to 3/2. Further analysis of the magnitude of potential zeta should be carried out by Smoluchowski equation that provides evidence for that zeta potential is correlated to the mobility of the particle and independent of size of particle or electrolyte concentration.
The dependence of the potential zeta with parameters above mentioned is easy being its demonstration commonly present at chapters or texts approaching colloid physical-chemistry.
When a colloidal suspension is stabilized mean that coagulation is not expected, since the magnitude of repulsive electrostatic forces between charged particles is quite similar. Otherwise, if repulsion is very low or absent, as well as surface charge was neutralized, attractive forces can be operational leading after each particle collision the growthing of aggregates (seems that the term aggregate is preferable under the term agglomerate) from a specific size this kind of structure settle out from liquid. In fact, the topic flocculation can be useful here. As a whole, the attractive forces belong to the secondary bonds class being classical the Van der Waals force and Lennard-Jones potentials. In this sense, the DLVO theory of colloid stability emerges.