While my first question qould be "What dielectric medium are you using?", aluminum iron oxide is a relatively general-purpose, good-performance nanoabrasive. Enclosed below, please find a paper detailing the abrasive properties of the system:
Article Abrasive Properties of Aluminum Iron Oxide Nanoparticles
Thank you for the details. The main requirements I see here would be insolubility (easy for kerosene oil, far more difficult for deionised water) and stability of the nanoparticles.
As such, the best nanoparticles would be as non-ionic as possible, so as to not have the dielectric be contaminated with an "electrolyte" over the course of its use. With that in mind, I would recommend the following (apart from the aluminum iron oxide, which is relatively easily acquired and should prove at least decent):
cubic boron nitride (insoluble in water, and an rather harsh abrasive)
silicon (virtually totally insoluble, chemically resistant and even if oxidised, will produce a nanoparticle that is both dielectric and abrasive)
silica (unless you will be dealing with fluorine contmaination, SiO2 will remain as it is; the SiO2 NPs, however, tend to aggregate over time)
copper(II) oxide (very cost-efficient, but is feasible for kerosene and may be unsuitable for deionised water)
If you do not need a particularly harsh abrasive, I would recommend the following review (table 3 is of most interest, IMHO):
Article Mechanical properties of nanoparticles: Basics and applications
An interesting variation on silicon-based NP abrasives:
Article Preparation of spherical ceria coated silica nanoparticle ab...