(653c) DEM Simulations and Experiments: Study of Dry Impregnation and Its Dependence on Particle Morphology: Spheres Cylinders, Trilobes and Quadrilobes

The impregnation of metal solutions onto porous catalyst supports is one of the key steps for preparing industrial metal-supported catalysts. In this process, metal salts or complexes are first dissolved in an aqueous solution. The volume of the metal solution applied is the same as or less than the total pore volume of the catalyst support. When conducting dry impregnation in a rotating vessel, metal solution is typically sprayed over a granular bed containing a porous catalyst support such as alumina (Al₂O₃) or silica (SiO₂). During spraying, capillary action draws the metal solution into the pores and metal complexes are adsorbed onto the high surface area support particles.

A problem may arise as excess liquid forms liquid bridges between particles, which causes wet cohesive forces and disrupts flow and further affects granular mixing processes. As a consequence, this can also adversely affect the homogeneous dispersion of the metal precursor in the particle.

Due to the large volumes and high cost associated with producing heterogeneous catalysts, several open questions regarding the optimization of this process remain: (i) how mixing and flow are affected when the particles have a certain degree of moisture or are saturated with fluid, (ii) how to improve fluid distribution to and within the granular bed (iii) the extent and distribution of dead zones for a given impregnator configuration, (iv) how does the morphology of the particles affect the uniformity of the liquid in the granular bed, (v) can the mixing and content uniformity be improved by using alternate geometries, such as baffles, spray pattern, or systems, (vi) how does the rotation pattern, nozzle and spray pattern, as well as the morphology of the catalyst support affect mixing profiles within the granular bed, and finally, (vii) what are the rules of thumb for process scale-up?

In this work, we performed DEM simulations to study impregnation in different porous support morphologies, spheres cylinders, trilobes, quadrilobes of different nominal diameters and aspect ratios.

We found that the time to achieve liquid uniformity in the particle bed for quadrilobes and trilobes of a given nominal diameter converges to the time for uniformity of a sphere with the same nominal diameter. (Aspect ratio (AR) equal to 1). Quadrilobes and Trilobes behave very similarly to cylinders of the same nominal diameter and for a given nominal diameter, the higher the AR the smaller the time to achieve uniformity. We observe that the time to achieve uniformity for particles, being cylinders, trilobes and quadrilobes with Aspect Ratio greater than 1 is always shorter than for spheres. We believe that higher probability of collisions correlates well with shorter times to uniformity. The length of the extrudate does not have a strong influence in the S/V ratio after a nominal diameter of 6mm. This ratio converges to a fixed number for a given diameter after 6mm, however, for shorter lengths it matters.

The effect of flow and operating efficacy in impregnation (adsorption of liquids, adsorption of chemicals) through any extrudate bed would be influenced not only by the Surface to volume (S/V) ratio but also by the packing fraction of the extrudate and the tap density of the material and both would be influenced by the length of the cut extrudate pieces

It is all these factors and not only S/V ratio that will determine the final operation functionality of the impregnation of the material in a bed. In addition, we performed 18 experiments with metal solutions, 3 supports and 2 different concentrations. Regardless of the type of support, titania or alumina, we observe that particles with larger aspect ratios always fall below spheres as predicted by simulations. Trilobes are much longer particles and more uniform aspect ratio. This is consistent with simulations: particles of larger AR reach uniformity faster: Trilobes show better uniformity than cylinders. Also, for cylinders, we observe that larger cylinders show uniformity faster as expected since they have smaller S/V. The time to achieve uniformity follows the same trends predicted by simulations.