Abstract : This study deals with the formulation of oxide suspensions for liquid processing of ceramic matrix composites. Those processes are more advantageous than conventional ones regarding: cycle time, quality, flexibility and cost. To properly impregnate the fiber reinforcement, the suspension must be stable, well dispersed and have a low viscosity (< 1 Pa.s). Regarding those requirements, the dispersion of an alpha-alumina ultrafine powder (AKP50, d50 = 0.3 µm, Sumitomo Chemical) in aqueous suspension, using an ammonium polymethacrylate (Darvan® C-N, Vanderbilt Minerals) as dispersant, was investigated by zeta potential, sedimentation and rheological measurements. The dispersant concentration minimizing the viscosity was found to be 0.26 wt.%. Moreover, this concentration permits to shift the isoelectric point (IEP) from pH = 9.5 (without dispersant) to pH = 5.1. The influence of powder concentration on suspension viscosity is well described by a
Quemada model. The maximum volume fraction was found to be equal to 40.16 vol.%, thanks to the viscosity. No significant sedimentation was observed, regarding the operating time of the process. This work comprises also the study of a commercial suspension of colloidal silica (Ludox AS-40, 40 wt.% suspension in water, W.R. Grace & Co.-Conn.). The zeta potential is negative from pH = 2 to pH = 11. The impact of dilution on viscosity was studied. For all powder loadings (from 1 to 40 wt.%), the viscosity is lower than 1 Pa.s (Newtonian) and the variation is also well described by a Quemada model. No sedimentation was observed. Furthermore, the mixture of both suspensions in a stoichiometric ratio of 3Al2O3-2SiO2, corresponding to mullite, was investigated. The natural pH and the IEP were measured at pH = 9.5 and pH = 3.0, respectively. The viscosity, as a function of solid concentration, is well described by a Quemada model. No significant sedimentation was observed.