||Synthesis of nano-particles by flame-spray-pyrolysis (FSP) is a process where the precursor is released to the gas phase by combustion of fuel-droplets. These droplets consist of precursor and liquid fuel. This process is different from conventional synthesis of particles in flames, where the precursor already is present in the gas-phase from the beginning. The advantage of FSP compared to conventional flame-synthesis processes is, that it is possible to use a wide range of precursors and liquid fuels. Furthermore, it is also possible to include a carrier-material in the liquid, which can be useful in the manufacture of heterogeneous catalyst. In this study, fuels as iso-octan and 1-butanol have been studied together with the precursor titan-tetra-isopropoxide (termed precursor from now). When the droplet burns, the precursor is released to the gas-phase where it will decompose to monomers. These monomers will coagulate followed by fast sintering resulting in spherical primary particles. The spherical particles will collide with each other followed by slow sintering, which finally will produce an aggregate with large specific surface area (SSA). If the temperature is too high the aggregates will sinter and reduce the SSA, which is unwanted and can be prevented by quench-cooling of the gas. The present process for FSP consists of an expansion of oxygen in a small valve, where small droplets of fuel and precursor is produced. Further upstream the small fuel-droplets are ignited by means of a hydrogen flame.
The combustion of liquid droplets is considered to follow a two-zone model, in which the droplet is surrounded by a thin flame-front where oxidizer and fuel reacts very fast. This two-zone model is the basis of a mathematical model to describe the mass-flow-rate from the droplet. A simple PFR-model to describe the combustion of fuel-droplets of variable diameters was developed. This model includes radiation, where variable emissivity of the gas-phase is included. The variable emissivity is primarily due to the species H2O and CO2, which is produced under the combustion. The resulting model is solved using a developed Fortran-routine, resulting in very different lifetimes for different droplet sizes. The maximum temperature was found to increase as the droplet diameter decreased, which was due to much faster combustion, and hence shorter time to transport the energy away by radiation. Furthermore, it was found, that droplets consisting of iso-octan has a shorter lifetime than those consisting of 1-butanol.
Fundamentally, Computational Fluid Dynamics (CFD) consists of equations representing conservation of momentum, energy and mass. These equations are solved in Fluent using a finite-volume-method in which the discretised equations is solved and linearised. The used turbulence models is the realizable k--model and the Reynolds Stress Model, where the latter has a better description of the turbulence, but is also more difficult to converge.
CFD-simulations have been carried out in 2D and 3D. Regarding the 2D-simulations, the round holes for injection of hydrogen and cooling-air was approximated by a slit with equivalent area. This is not a good approximation, but it certainly decreases the number of cells in the geometry. To reduce the number of cells in the 3D-calculations a rotational periodic boundary on 60° was employed. Preliminary CFD-simulations of the FSP-process without radiation and quench-cooling resulted in temperatures too high, due to the lack of heat-transfer by radiation. The estimated maximal temperatures were approximately 3000 K. Further CFD-simulations including radiation lowered the maximal temperature to approximately 2000 K, which is more realistic. The Discrete-Ordinates model is used to model radiation because it can model the penetration at the outer glass-wall, and it has influence on the evaporating liquid droplets. It was found that the effect of the quench-cooling-device was not sufficient in the original geometry, because a cold tongue escaped in the center above the cooling-ring. Furthermore it was found, that recirculation around the burners for hydrogen reduced the temperature in the center between these.
To avoid recirculation and to gain a better effect of the quench-cooling-device, new geometries were suggested. To make some preliminary tests of these new geometries, new CFD-simulations were employed, which solved the above problems was solved. Furthermore it was found that the new geometry was much less susceptible to changes in sizes of the droplets than the original geometry and gave higher maximum temperatures.
A survey of literature on the combustion/evaporating of liquid droplets consisting of several species showed that these can behave very differently than mono-component liquid droplets. A hypothesis was developed to explain the difference in SSA that different fuels give at small quench heights : Occurrence of microexplosions inside the droplet creates small fuel fragments that burn very fast, thus producing large SSA at small quench-cooling heights. Contrary to normal burning, in which the less volatile component usually precursor - is kept back, while fuel combusts leaving the precursor as a compact salt with low SSA. This hypothesis is followed up by empirical calculations, which compared to experimental results supported the thesis. To be certain about this conclusion, more experiments should be performed. To illuminate the complex mathematical situation of evaporation/combustion of multi-component droplets, mathematical models are developed to describe mass-fractions and temperature inside the droplet. To solve this model, the gas-phase and the rate of combustion need to be coupled. Furthermore, detailed physical properties for titan-tetra-isopropoxid, such as diffusion-coefficient, thermal conductivity etc., are not available at present.