Modified Nukiyama–Tanasawa and Rizk–Lefebvre models to predict droplet size for microconcentric nebulizers with aqueous and organic solvents

Abstract

Aerosols characteristic of several organic solvents (hexane, acetone, xylene, toluene, methanol, and ethanol) produced by a direct injection high efficiency nebulizer (DIHEN) are measured using a phase Doppler particle analyzer (PDPA) over a wide range of operating conditions (nebulizer gas = 0.2–1.0 L min⁻¹, solvent flow rate = 10–500 μL min⁻¹). The Sauter mean diameter, defined as the volume-to-surface area ratio (D_3,2), and cumulative count percent of organic aerosol are measured and compared with those of aqueous droplets. These parameters are chosen because the performance of inductively coupled plasma (ICP) as an excitation and ionization source is affected by the size of the introduced droplets, particularly in the case of direct injection nebulizers where the primary aerosol is directly introduced to the plasma without being filtered by the spray chamber. The size distribution of the droplets demonstrates a notable shift toward smaller droplets, and is generally narrower when organic solvents are used instead of water. This effect is more pronounced for hexane and acetone, having a considerably lower surface tension and viscosity, respectively, compared with water. A D_3,2 of 4.6 μm is obtained for hexane, compared with 7.2 μm for aqueous solutions at a nebulizer gas flow rate of 0.2 L min⁻¹ and a solution uptake rate of 50 μL min⁻¹. This decrease in droplet size is less significant for ethanol, methanol, toluene and xylene. Experimental results are also compared to D_3,2 values calculated by the Nukiyama–Tanasawa (N–T) equation and Rizk–Lefebvre (R–L) model. While the cited models correctly predict the trend in size variation as a function of nebulizer gas flow rate, and to some extent solvent characteristics, an overestimation and an underestimation of D_3,2 is observed for all tested solvents at low nebulizer gas flow rates for the N–T model and the R–L model, respectively. Modified equations are proposed which are capable of predicting D_3,2 values for several solvents with greater accuracy.