European Drying Conference - EuroDrying'2011 Palma. Balearic Island, Spain, 26-28 October 2011 EXPERIMENTAL STUDY OF SPRAY DRYING AND ATOMISATION WITH A TWO-FLUID NOZZLE TO PRODUCE FINE PARTICLES Ian C Kemp, Robert Wadley, Thoralf Hartwig, Ugo Cocchini, Yoong See-Toh, Kim Fordham, Francois Ricard Glaxo SmithKline, R&D Gunnels Wood Road, Stevenage, SG1 2NY, United Kingdom Tel.:+44 1438 768433, E-mail: ian.c.kemp@gsk.com Abstract: A series of 36 experiments was performed to investigate particle size and yield from a Buchi laboratory spray dryer with a two-fluid nozzle. First principles theory suggests that the main parameters affecting final particle size should be atomization gas velocity and solution concentration, and the experiments confirmed this. Hybrid ratios such as the ALM (atomization gas to liquid flow ratio) did not correlate well to measured final particle size. Particles of 2-3 microns were obtained at high atomization gas flows, indicating initial droplet sizes of 4-7 microns, while lower flowrates gave larger particles. Keywords: particle size distribution, yield, atomiser, respirable, inhaled particles, droplet size INTRODUCTION Spray dryers have conventionally given particles in the size range from about 20 to 500 microns. In the last 20 years, manufacturers have produced dryers capable of producing particles down to 1 micron in size, suitable for inhaled pharmaceutical products, using two-fluid or ultrasonic atomisation. These dryers can also be very compact, as the tiny droplets dry out before they hit the dryer wall, allowing operation at very small scale (1g product or below). Only limited information is available on how to control the dryer to achieve given product properties, particularly particle size distribution. This study looked systematically at the final particle size produced from an aqueous solution by a Buchi B-290 spray dryer with a two-fluid nozzle under a wide range of operating conditions. THEORY In spray drying, liquid is atomized into fine droplets and these then dry out in a hot gas stream. The droplets shrink as liquid is evaporated, and a crust may form, particularly with solution feeds. The atomization step is clearly key. Most information is available for rotary atomizers and pressure nozzles, which are unsuitable for fine particles. Some correlations have been suggested for two-fluid nozzles, e.g. Thybo et al (2008): 1 5 . 1 2 2 1 k g g L L atm L sm D U k W W D d ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ + = ρ ρ σ μ (1) The ratio of the mass flowrates of the atomizing gas and liquid feed, W atm /W L (sometimes entitled ALM) has been suggested as a characterisation parameter. However, from first principles, two-fluid atomization is caused by the high gas velocity at the nozzle (typically 100-500 m/s), unlike pressure nozzles and rotary atomizers which use high speed liquid flow. Hence one would expect droplet size to depend more on gas flowrate than on liquid flowrate. The initial droplet size and the final particle size are related by a mass balance on the solids fraction, assuming no coalescence, agglomeration or attrition: p p p p p d d d d d S S d S m S d S m m ρ π ρ π 3 3 6 6 = = = = (2) Hence the final particle size can be estimated from the initial droplet size as: 3 1 ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ = p p d d d p S S d d ρ ρ (3) Thus, for a given droplet size, low solids concentration and high particle density are expected to give smaller final particles. EXPERIMENTAL PROCEDURE The experiments were performed using a Buchi B- 290 mini spray dryer with standard Buchi two-fluid nozzles and a high efficiency cyclone to collect the dried particles. Recycled nitrogen was the carrier gas (oxygen <6%) with solvent condensed in the B-295 inert loop. Atomisation gas flow was measured by a digital mass flow meter. The material dried was egg albumin in aqueous solution, mixed with 32% excipients (mainly sugars) to produce a strong matrix on drying. The product was collected in the glass jar at the bottom of the cyclone, which was then weighed to give the yield.