Aerosol Science, 1973, Vol. 4, PP. 235 to 243. Pergamon Press. Printed in Great Britain.

T H E C O L L I S O N N E B U L I Z E R : D E S C R I P T I O N ,

P E R F O R M A N C E A N D A P P L I C A T I O N

K. R. MAY

Microbiological Research Establishment, Porton, Wiltshire

(Received 23 October 1972)

Abstract--The Collison nebulizer is widely used to produce fine aerosols from a liquid supply. Details of its design and operating characteristics are given, including air and liquid consumption, aerosol output rate and droplet size distribution. An adaptor for the outlet of the nebulizer is also described. This is used when monodispersed aerosols are being generated and enables the output of particles to be increased.

I N T R O D U C T I O N

THE FOLLOWING account of the Collison nebulizer is intended to provide useful information on the device for those who are not fully familiar with it, or who are using it for the first time.

W. E. Collison introduced the "inhaler" which bears his name at a meeting of the British Medical Association in 1932. Its employment was described in his book Inhalation Therapy Technique (1935) but no technical details were given. During the 1939-1945 war W. H. Walton at Porton used it to nebulize a methylene blue solution for the testing of the penetration of chemical respirator filters. Later it was adapted to the "sodium flame" test apparatus for nebulizing sodium chloride solutions. A British Standard specification (1955) describes the methylene blue apparatus and gives some details of the dimensions and use of the nebulizer, without actually naming it. The first mention of the Collison device in connection with spraying bacterial suspensions is by RosEmmY (1947) who refers to Henderson's use of it in 1942. Later HENDERSON (1952) described his apparatus, which has since become known as the "Henderson Apparatus" or "Piccolo" and was adapted from Walton's methylene blue system. Henderson gives a few brief details of the functioning of his Collison nebulizer which was similar to that shown in Fig. 1 of the present note. GREEN and LANE (1964) refer to the Fig. 1 model and they present some performance figures for dibutyl phthalate spray fluid. These figures were extracted from an unpublished report by LANE and EDWARDS (1950), which was a fairly detailed account of the device, though aqueous spray fluids were not studied. More recently DRuu~ (1959, 1969) has modified the device so that it can be used with small volumes of precious or highly toxic fluids. Druett's modified design is the one shown in Fig. 2 and this is the type which is principally used in this establishment for microbiological work and which has been supplied by us extramurally.

Readers interested in the finer details of the functioning of this type of nebulizer from a more theoretical standpoint should consult MnRcnR, TILL~RY and CHOW (1968). Their paper discusses nebulizers similar in general principle to the Collison but not, unfortunately, the Collison itself.

233

236 K. R. MAY

Compressed air

FIG. 1. British Standard Specification nebulizer.

It may be noted here that the terms "nebulizer" and "atomizer" seem to be used synonym- ously in the literature. In the Oxford English Dictionary both words have the same definition and are attributed to late nineteenth century medicine. The present author proposes and uses the term "nebulizer" as being specific to the refluxing, baffled spray cloud-producing device. Semantically it seems preferable to "atomizer" which has an optimistic inference to say the least and which is commonly applied for example to simple two-fluid (air, water) spraying systems and even to single fluid sprays. Furthermore Collison himself used the term "nebulizer".

DESCRIPTION Figure 1 represents the "methylene blue" and British Standard form of the Collison

nebulizer in which the spray bottle was originally a "Kilner" jar designed to hold the large volume of several hundred ml of fluid so that on prolonged running the change in concen- tration of solute due to the evaporation of the solvent would be slow. As the only essential difference between Figs. 1 and 2 is in the volume and shape of the spray bottle the following description is confined to the Druett design of Fig. 2. Nevertheless, for long period nebul- izing such as in inhalation therapy or filter testing a large liquid-holding capacity, as in Fig. I, is essential. A very simple and cheap arrangement is to use an ordinary glass honey pot to hold the spray fluid. The lid holds the spray head and aerosol outlet port.

In Figure 2 the thick-walled glass vessel (1) is blown with a lip at the top so that it can be sealed into the head (4) by the rubber gaskets (5) and the coarsely threaded pieces (2) and (3). Compressed air is introduced into the tube (6), to the bottom end of which screws the stainless steel nebulizing head (9). (It may be mentioned here that the specified and com- monly used compressed air pressure of 26 psi shown at the top of the figure is merely an

The Collison nebulizer 237

^r r psl

~ S c r e w 0-25 dia. x40 f.pJ. whir. form

~/ ~Drill 0.35 mm dia. { No. 80-0"01:55 dia.] 14

Holes-1"6mm dTa. (0.063 dio. N0.52)

ia. Tolerance + 0.005

1 ~ 7 2 ~ . ~ Single jet

Alfernaflve ~ -~ fittings

~.u- J, Triple X jet

3-Holes equi- spaced

on o 0"344 RC.D.

• ,, 2' in.

FIG. 2. Druett (MRE) design of nebulizer.

evolutionary relic dating from the original use of rubber tubing with that safe limit. In fact the device can function well with air pressure at least within the range 15-50 psi). An enlarged detail of the nebulizing head is seen at the right of Fig. 2. Its central hole is blocked at the lower end by the screw (13) which can be removed when jet cleaning is necessary. The side hole (7), is either single or triple as indicated by B at the bottom right of Fig. 2. Either one or three jet holes, as the case may be, of 0.0135 in dia. (14) are drilled between the central hole and the side holes. The jets (14) connect to the spray nozzles (15) drilled through the side wall of the head (9). At the lower end of the head is the well-fitting sliding extension piece (8) which dips into the liquid (11) in the teat blown into the bottom of the vessel (1). The initial volume of fluid which can be used is anywhere between 4 and 100 ml. The piece (8) must dip well into the fluid so that its lower end is never uncovered during spraying. An extension rubber tube can be fitted for very small starting volumes and it is essential that the sliding face of (8) must be a good fit or leaks will compromise the working of the system. Some workers prefer to remove (8) using rubber tubing instead. The final component of Fig. 2 is the outer protective shield (10) which screws into (2). Normally this may be used to protect the glass part during carriage etc. and it is as well to remove it when spraying so that one can see that all jets are functioning properly. The fineness of the holes (14) gives a tendency to blockage unless the compressed air supply is thoroughly clean and filtered. A convenient way of mounting the nebulizer is to rest the rim under (3) on the ring of a tripod.

A.s. 4/3----D

238 K.R. MAY

MODE OF OPERATION

When compressed air expands from (14) into the nozzle (15) the reduction of static pressure sucks fluid up the tube (7) from the reservoir (11) (hence the need for a good seal between (8) and (9)). This fluid is then broken up by the air jet into a dispersion of droplets of very wide size distribution. Most of the droplets are blown on to the internal wall of the glass vessel as indicated at (12) in Fig. 2. In fact with water, about 99.92 per cent of that sucked up refluxes down again. With the triple-jet head, about 200 ml/min is drawn up and refluxes so that, for example, a fill of 20 ml of water in the flask will recirculate every 6 sec. The remaining 0-02 per cent of liquid which escapes impact comprises only the finest tail of the drop-size distribution and these droplets are carried up and out of the nebulizer by the spent compressed air flow. I f the drops in this emerging aerosol are aqueous, or of some other volatile liquid, they will evaporate very rapidly on admixture with unsaturated air. For example a 10/zm water drop in air at 20°C and 80 per cent R.H. has a wet lifetime of 0.6 sec. (the lifetime is proportional to diameter squared at a given temp and r.h.). Figure 4 shows that only 1 per cent of the mass of emerging aerosol is in

FIG. 3. Impactor for limiting the size of emitted droplets.

I 0.1

m 5 o

//2° Single-jef collison, 25psi, .-¢I" Di-oct"yl sebacafe / .~ lo

. / , , , , , , / , , , ,1 I 2. .5 [0 20 50 50 70 90 95 99 99"9 9999

Per cent undersize

FIG. 4. Size distributions from Collison nebulizer.

Whole aerosol, X I000 Ordinary illumination

Af'l'er impacfor, XiO00 Phase contrast

Elec'l'ron microscope, X 2000

FIG. 5. Thermal precipitat, or samples f rom aerosols o f B. globigii spores, nebulized f rom a concent ra t ion o f 2 × 10 l° m1-1.

[A.S.f.p. 238]

The Coilison nebulizer 239

drops larger than 10/zm so the bulk of a water aerosol will dry offin a very small fraction of a second after dilution.

Temperature changes during spraying During the time between formation of drops and their escape from the aerosol outlet at

the top of the nebulizer, a period which averages about 3 see in the 3-jet model, the droplets come to equilibrium with the air and their evaporation cools it. For pure water, aqueous suspensions or dilute aqueous solutions, saturation of the working air by evaporation from the droplets is complete and the fall in temperature is almost entirely due to this evaporation. The temperature fall from the adiabatic expansion of the compressed air is largely absorbed by the thermal energy released by break-up and impact of the liquid.

In practice the temperature in the vicinity of a head spraying water (using dry air) is about 6°C below ambient but by the time the aerosol emerges its temperature initially rises by exchange with the ambient air via the hardware of the nebulizer to about 3°C below ambient. Because of the heat capacity of the system further temperature changes are slow and after about 30 min of operation the aerosol and hardware reach an equilibrium at about 5-6°C below the ambient. Hence we would expect small changes in the ratio of liquid droplets/vapour in the emerging aerosol to take place during this period. When solutions of any substance which lowers the vapour pressure are sprayed the relative humidity and temperature of the emerging aerosol will of course depend on the nature of the solute and its concentration. With involatile liquids, no significant change in temeprature can be expected.

Air consumption. Over the normal working range of the device, 15-50 lb in -2 (1-3.5 kg cm-2) the following formulae apply:

for a 3 jet nebulizer, F ----- 0-216P + 2.8

for a 1 jet nebulizer, F = 0.075P + 0.9.

Where F is free air in 1. min- 1 and P is lb in- 2 gauge. With P in kg cm- 2 we have F = 3-07 + 2.8 and 1.07P + 0.9 respectively. Small differences in machining the jets account for the ratio between the two flow rates not being exactly 3:1 and for the same reason other units to the same design cannot be expected to have identical performances.

Liquid consumption and aerosol output. Table 1 applies to a 3-jet model spraying water with dry compressed air in an ambient air temperature of 19°C.

TABLE 1

Air pressure lb in- 2 gauge 15 20 25 30 40 50 Free air consumption I min- 1 6-1 7-1 8"2 9"4 11"4 13"6 Water loss, droplets + vapour ml h r - t 7.8 8.7 9.5 10-4 12-0 14.0 Approx. water vapour output ml h r - 1 4"6 5"4 6.2 7-1 8-6 10"2 Approx. droplet output ml h r - 1 3"2 3"3 3-3 3"3 3"4 3-8 Total water cone. in outlet port g m - a 21.3 20.4 19.3 18.4 17.5 17.2 Droplet cone. in outlet port g m - a 8"7 7"7 6.7 5-9 5'0 4"7

The table shows that the vol/vol ratio of air consumed to output of water as droplets is roughly 105, or on the mass basis 120: 1. Many two-fluid atomizers operate with a mass ratio of about 2:1, so that the Collison value emphasizes the extremely low overall efficiency of the device, which is the price paid for allowing only the very finest droplets to escape. There is no secondary shatter of liquid into fine droplets where the main jet strikes the glass

240 K.R. MAY

wall (GREEN and LANE, 1964). The second line and the last two lines of Table 1 show that the increase in volume of the aerosol with air pressure is accompanied by only a slow decrease in the concentrations of total water and droplets. It will be shown in the next section that the size distribution of the emerging aerosol is not very sensitive to pressure, if at all, so that a great latitude in the setting of the nebulizer can be tolerated, unless the actual volume of aerosol emitted is critical. The reason for this self-limiting performance must be that the number of small droplets formed is nearly proportional to the air con- sumed, any excess of droplets as the pressure is increased being absorbed by their increased chance of striking the wall with the increase of velocity.

The average upflow velocity within the flask is about 5.8 cm sec- 1 with three jets totalling 7 1. min -1 (20 lb in-2). If this flow were laminar (which it is not) it would just support a droplet of about 43/~m. The 1.9 cm sec- 1 upflow from a single jet at the same pressure would just support a 25/zm droplet.

Prehumidification of the compressed air would increase the total water output in droplet form to values similar to those of the third and sixth lines in Table 1. The average emerging droplet size would also be increased and the temperature change within the nebulizer would become negligible. If, however, the nebulizer were being used to produce a cloud of residual dried particles from a solution or suspension by allowing the emergent droplets to dry completely before being used, neither the particle size distribution nor the total particle output would be affected by the humidity of the compressed air.

For comparison with Table 1, Table 2 shows some measurements obtained with in- volatile liquids over a range of viscosities (centistokes) in a 3-jet nebulizer.

TABLE 2. LIQUID LOSS, ml hr-1. IN BRACKETS, AEROSOL CONCENTRATION, g m - a

Pressure lb- 2 Liquid and viscosity 15 20 25 30 40 50

Dioctyl sebacate, 20 cs 8 (30) 9 (24) 11 (26) 13.2 (28) 13 (23) 13 (19) Dibutyl phthalate, 18 cs 11 (26) 12 (25) 12.8 (22) Silicone fluid, 100 cs 9 (19) 13.4 (24) Silicone fluid, 500 cs 3"4 (7)

The table shows that liquids of quite high viscosity can be successfully nebulized. The size distribution appears to remain substantially unchanged at the higher viscosities but the recirculation rate is much reduced. At 20 centistokes this rate was half of that with water.

SIZE DISTRIBUTION

Since the nebulizer is commonly used for inhalation experiments, the range of droplet or particle sizes it produces is an important parameter. The upper limit of drop size escaping is much larger than is generally realized and is in the region of the theoretical maximum, derived in the previous section from the maximum drop size which the upflow of air within the nebulizer can support. Thus with water the absolute maximum of drop diameter found, by impact on magnesium oxide coated slides, was about 55/~m with no obvious dependence on pressure, while drops of about 43/zm were more commonly found. With the single jet model the absolute maximum drop size was about 30 with 23/zm more commonly found. Increasing the viscosity of the spray fluid decreased these figures a little. Thus with fluid of 100 centistokes the absolute maximum was about 35 /zm with 28 /zm more commonly

The Collison nebulizer 241

found, over the whole pressure range of 15-50 psi. These large drops are, of course, exceed- ingly infrequent compared with drops at the mass median dia., for example.

When one nebulizes a solution and collects a sample of the evaporated aerosol on a high efficiency impaction stage, enormous numbers of very fine particles are seen, down to the limit of microscope resolution (see Fig. 5). An accurate microscope analysis of the entire droplet range from say 30-0-25/~m with the latter about 2 × 107 times more abundant than the former is a daunting task which has not been attempted. Instead, use was made of a nine-stage cascade impactor capable of covering adequately the range of drop sizes emerging from the nebulizer.

This impactor, which is still under development, deposits the size fractions of the indrawn aerosol on standard microscope slides, convenient for either optical or bulk assessment. It operates at 5 1 min- 1 and uses a multiple jet system so that comparatively large quantities of material can be collected on each stage. The Dso or "effective cut-offdiameters" (MERCER, 1965) were designed to be, for each stage in succession, 32, 16, 8, 4, 2, 1, 0.5, 0.25 and 0.1 /~m respectively. In the present work, to assess the mass of aerosol droplets collected by each stage, dioctyl sebacate (DOS) was tagged with the fluorescent dye Uvitex and the quantities washed off were estimated in a high quality fluorimeter. Then, by plotting the percentage cumulative mass below a given stage against that stage's Dso, the mass distri- bution curve B in Fig. 4 was obtained. Curve D in Fig. 4, the estimated number distribution of the initial cloud, was derived from curve B by calculation.

For this calibration it was greatly preferable to use an aerosol of permanent droplets rather than one of dried residue particles because this is thought to eliminate problems of "bounce off" of particles from their proper stages, and complete washing-off of collected liquid is easy. Also the larger size of the unevaporated droplets compared to dried residue particles greatly aids size classification by the impactor method in such a fine aerosol.

DOS is virtually involatile and yields nearly the same total output from the nebulizer as does water (see Tables 1 and 2). There is no reason to think that the size distribution of the DOS droplets, sampled as they emerge from the nebulizer, is significantly different from that of water droplets initially generated within the nebulizer and whose subsequent history is to be carried upwards and evaporate to dryness after emergence.

The size distribution of residual particle aerosols left after complete evaporation of a solvent phase can readily be obtained from Fig. 4 by multiplying the drop sizes by the cube root of the fractional concentration by weight of the solute and by the cube root of the ratio of densities of solution/residual particles, should the latter be appreciably different from unity. Often, however, the dry particle density is uncertain.

GREEN and LANE (1964) quote the mass below 2, 5 and 15 tzm as 32, 79 and 99.5 per cent respectively. These figures were obtained by use of the conifuge sampler which cannot give an accurate sample of the largest particles and does not collect particles smaller than about 0.6 tzm. If, however, in the present work, we ignore drops smaller than 1 t~m, curve A in Fig. 4 results and this is in good agreement with the quoted figures from Green and Lane.

Effect of operating pressure on the size distribution. To study this a "Royco" automatic 13-channel particle counter was employed. Because this senses light scattered from particles, its response depends on the shape and surface properties of the particles, so that it is of no value for particle sizing without prior calibration. Thus, in the present work it was found that there were many-fold differences in indicated particle size for aerosols of differ- ing substances whose size ranges were in fact known to be closely similar. Also the counter ignored great numbers of particles which were demonstrably present by impaction. How-

242 K.R. MAY

ever the counter should be suitable for detecting relative changes in particle size distribution as they occur from a given liquid as the pressure is changed. When this was done with an aqueous 4 per cent solution of Fluorescein, the Royco indicated number frequency modal values of 1-8/zm at 15 psi, 2.2/zm at 25 psi and 2.7/zm at 40 psi. These figures (which in absolute terms were about an order of magnitude too high) show the opposite tendency to the result of LANE and EDWARDS (1950) who found a small decrease in mass median diameter as pressure was increased.

Whichever result is correct it is clear that changes in size distribution are small and that the main effect of varying the operating pressure is to vary the total liquid output as shown in Table 1.

OBTAINING MONODISPERSED PARTICLES

A common use of the nebulizer is to obtain monodisperse particle aerosols of small and uniform particle size. This is done by spraying dilute suspensions of particles, usually Dow Corning polystyrene latex spheres (which may be obtained in a range of carefully graded monodisperse suspensions) or bacterial cells and the dilution must be such that the chance of any droplet containing more than one particle is small. HENDERSON (1952) recommends a cell concentration not exceeding 109 ml-1, when 95 per cent of the resulting dried aerosol particles consist of single cells. This concentration is equivalent to an average of 1 cell per 12.5/~m drop and is consistent with the experimental finding that of the drops greater than 1/~m (i.e. those which are capable of containing a single cell) about 95 per cent are below 2.7/zm initial dia. (Curve C, Fig. 4), and so have a very remote chance of containing more than 1 cell.

At 109 ml- 1 the majority of drops capable of carrying a cell are of course vacant and wasted, e.g. of the 2/~m drops, on average only about 1 in 230 would be occupied. The 25 per cent of the total mass of liquid which resides in droplets below 1/~m, and so is incapable of being occupied, is also inevitably wasted. The concentration of monodispersed particles may be increased considerably by spraying more concentrated liquids and fitting to the outlet of the nebulizer an impaction device designed to trap droplets greater than a specified diameter. A further advantage of this system is that the occasional large rogue cluster is eliminated. The following equation (MERCER et al., 1970) is convenient for calculating the required parameters of such an impactor: Dso = 1.22 x 10 a (Wa/F) ~, where W is the jet dia. in cm and F the flow rate in cm a rain- 1. The largest drop penetrating the impactor is about 1.5 x Dso.

A suitable outlet impactor is shown in Fig. 3. For easy cleaning the two halves screw apart on a quick thread. The left side of the device is connected directly to the outlet tube of the nebulizer so that only unevaporated drops enter. The impaction jet plate (1) is made symmetrical so that it may be used either way round and it is drilled with the ring of holes (2). The air jets issuing from these holes impact the large droplets on the flat surface of the right hand half of the device, and the collected liquid drains into the well (3) which may be enlarged if necessary for prolonged running. In one device of this kind there were 30 holes of 1 mm dia. which at 7 1. min -1 total flow gives a value of 2.3 tzm for Dso.

The effect of this outlet impactor is demonstrated by the photographs of Fig. 5. A suspension of spores of B. globigii (BG) was sprayed at 25 lb in-2 and the resulting evaporated aerosol was sampled by a thermal precipitator. The cell concentration was 2 x 101° ml-1 (20 times higher than recommended by Henderson). In the left hand photo- graph, taken without the impactor, fewer than 60 per cent of the particles consist of single

The Collison nebulizer 243

cells. The rest are multi-cell clusters up to 5 t~m in the photograph and larger elsewhere in the sample.

For the right hand photograph everything was unchanged except that the impactor was in place. In a scan across the whole of this thermal precipitator trace 96.5 per cent of the bacterial particles were found to be single cells and none of the remaining particles appeared to contain more than two cells. This is also illustrated by the lower photograph in Fig. 5, the left side of which has been printed so that the BG cells show in strong contrast to the background of unoccupied particle residues from the whole culture suspension that was sprayed. The vast number of these residual particles is indicated by the heavier printing of the right hand strip which has been included to give an idea of the great range of the fine particle end of the distribution. These particles are not present when a thoroughly washed suspension is sprayed. Comparing sprays at 109 without the impactor and 2 × 10 l° with it and taking into account the mass of suspension trapped in the latter case the samples indicated a net gain of 4- or 5-fold in concentration in the latter case, with an aerosol of superior uniformity.

R E F E R E N C E S British Standard 2577 (1955) Specification for Methylene Blue Particulate Test for Respirator Canisters.

British Standards Institution, London. COLLmON, W. E. (1935) Inhalation Therapy Technique. Heinemann, London. DRUETT, H. A. (1959) Apparatus for Generating and Holding Aerosols in Controlled Conditions. Micro-

biological Research Establishment Development Note No. 45. DRUETI, H. A. (1969)./. Hyg. Camb. 67, 437. GREEN, H. L. and LANE, R. (1964)Particulate Cloud& Spon, London. HENDERSON, D. W. (1952) d. Hyg. Camb. 50, 53. LANE, W. R. and EDWARDS, J. (1950) A Study of the Performance of the Coilison Atomizer. Porton Technical

Paper No. 177. MERCER, T. T. (1965) Am. Ind. Hyg. Ass. d. 26, 236. MERCER, T. T., TILLERY, M. I. and Chow, H. Y. (1968) Am. Ind. Hyg. Ass. d. 29, 66. MERCER, T. T., TILLERY, M. I. and NEWTON, G. J. (1970) d. Aerosol Science 1, 9. ROSEaURY, T. (1947) Experimental Airborne Infection, Williams & Wilkins, Baltimore.