centrated solution of sulfur dioxide for the first series of runs was stoppered with a rubber stopper for a short time; this may have been the source of contamination. The first section of the reaction tube was accidentally broken between each series of runs and had to be replaced. This may account for some of the scatter in the results, although no definite trend could be established conclusively. Only the results from the second and third series of runs are presented. Comparison of Results from the Batch System and Flow System. Figure 5 shows the results of the batch system and the flow system plotted on the same coordinates. M'hen the line representing the batch system is extrapolated, it joins the curve for the flow system. This is quite remarkable when one considers the widely different methods used to obtain the data. I n general, the batch results appear to be consistent and agree very well with previous work found in the literature. Although the flow results show some scatter, the relationship between the rate of reaction and the concentration of manganous sulfate is fairly well established.
literature Cited
(1) Basset, H.. Parker, W.G., J . Chem. SOL.1951, p. 1540. (2) Coughanowr, D. R., "Oxidation of Sulfur Dioxide in Drops." Ph.D. thesis in chemical engineering, University of Illinois. 1956, (University Microfilms. Inc.. Ann Arbor. Mich.), (3) Firket. J.? Trans. Faraday Soc. 32, 1192 (1936). (4) Gerhard. E. R., Johnstone, H . F., Ind. E n q . Cheni. 47, 9'2 (1955) (5) Gerhard, E. R., Ph.D. thesis in chemical engineering. University of Illinois, 1953. (6) Hoather, R. C.. Goodeve. C. F.. Trans. Faraday SOC.30, 626-9. 630-5, 1149-56, 1156-61 (1934). (7) Johnstone. H. F.. Combustion 5 , No. 2 , 19 (1933). (8) Johnstone. H . F., Ind. Eng. Chem. 23, 559 (1931). (9) Johnstone, H . F., Coughanowr. D. K.. Ihid.. 50, 1169 (1958). (10) Krause. F. E., "The Reaction of Sulfur Dioxide and Oxygen in Aqueous Solution Containing Manganous Sulfate as a Catalyst," M.S. thesis i n chemical Engineering, Purdue Uni\ ersity, 1959. (1l ) , Schrenck, H. H., Heimann. H., Clayton. G . D.. Gerafer. 1%. M.. U.S. Public Health Serv. Bull. 306, 162 (1949). (12) [Ceissberger. A , Friess. S.L.. eds., "Techniques of Organic Chemistry," Vol. 3, pp. 674-8. Interscience. New York. 1953. RECEIVED for re\iew October 5. 1964 ACCEPTEDDecember 14, 1964
ELECTRICAL N EUT RA LIZAT IO N AND PARTICLE SIZE MEASUREMENT OF DYE AEROSOLS K E N N E T H T. W H l T B Y AND C A R L M . P E T E R S O N University of Minnesota, Minneapolis, M i n n . Data are presented on the magnitude of the natural charge on aerosol particles formed by the evaporation of atomized solutions of dye and on the residual charge after neutralization by mixing with a bipolar ionic atmosphere. Measured values of the charge after neutralization are compared with those predicted by assuming a Boltzmann distribution of the energy states in the mixture of ions and particles. The importance of this residual charge to such aerosol applications as filter testing and particle classification is discussed. Under certain conditions a size representative fraction of the dye particles smaller than 0.2 micron may carry a unit charge. Two mobility analysis techniques for measuring the particle size distribution of such aerosols are described and the results compared with electron microscope data.
some time it has been known that solid aerosol particles formed by the evaporation of liquid suspensions of particles or of solutions of solids may be electrically charged. However. little has been published on the characteristics of such charges or on the performance of systems for their electrical neutralization. T h e purpose of the work described here was to study the characteristics of the natural charge on dye aerosols immediately after generation and the residual charge after electrical neutralization by mixing with bipolar ion atmosphere. O u r interest in these charge characteristics resulted from the discovery that under some conditions the maximum natural charge on dye aerosols is nearly equal to the maximum charge that the aerosol can carry. Charges of this magnitude make the aerosol unsuitable for many experimental uses. such as filter testing. Lundgren (75) showed that under some conditions the natural charge on 1-micron particles decreases the penetration through a felt filter by a factor of 10 or more. Langer (72) found that a high initial charge on aerosol particles influences the charge received when they pass through a corona charger. OR
66
I&EC
FUNDAMENTALS
Origin of Natural Charge
When droplets of dye solution are formed in an atomizer, they have an electric charge, the ratio of positive to negative particles apparently depending on the particular solute-solvent system. As the droplets evaporate, the charge is conserved until it reaches some maximum value. From this point on, the charge is removed by some mechanism, probably the physical ejection of highly charged very small drops from the surface. Thus a droplet of small charge-mass ratio ma)- become a small particle having high charge-mass ratio. If the maximum charge is reached before the conclusion of evaporation. surface forces may cause the droplet to dry to a particle of dumbbell or pear shape, or in some cases actually to disintegrate. Electrical Neutralization
T h e electrical neutralization of aerosols is ordinarily accomplished by mixing the charged aerosol with a mixture of positive and negative small ions. generated in the gas phase of the aerosol either by ionizing radiation or by mixing with a separate stream of ion-carrying gas
T h e equilibrium charge distribution of aerosol particles in air containing a bipolar mixture of small ions has received extensive theoretical but only limited experimental study. Several-for example, Bricard ( I ) , Gunn (5),and Junge (70)treated the problem as one of diffusion of ions to the particles. Natanson (76) treated the case for particles less than 0.02 micron by a kinetic method. Lisovskiy ( 7 4 , as reported by Fuks ( 3 ) , and Keefe, Nolan, and Rich ( 7 7) assumed that because of the frequent collisions between the particles and small ions, the Boltzmann distribution should describe the charge as well as the thermal equilibrium T h e result of this approach, as given by Fuks ( 3 ) , is given in Equation 1, where f ( n ) is the fraction of particles having a given number of elemental chargrs, n (n may be either f or -), D , is the particle diameter, e is the charge on an electron, k is Boltzmann's constant, and T is absolute temperature.
T h e fraction of particles, f(n), with a small number of charges for particle sizes from 0.01 to 1 micron is shown in Table I. T h e last column, giving average charge (including particles of zero charge), is also plotted in Figure 1 I. Comparisons of the various theoretical models with previously available data [Nolan and Keenan (77) and Gillespie and Langstroth ( 4 ) ]and with our results suggest that Equation 1 is a reasonable model over the size range from 0.01 to 2 microns. Several characteristics of the equilibrium charge are evident from Table I. Less than 1% of the particles below 0.01 micron carry a charge. Most charged particles less than 0.1 micron in size carry only a single charge--e.g., only 9% of 0.1-micron particles carry two charges. Most particles larger than 1 micron are multiply charged.
Figure 1.
IO
Distribution of charges on particles
Langer (72), E = 30 kv./cm. 6. Hewitt (6),E = 10.8 kv./cm. C. Lundgren (151, E = 14.5 kv./cm. D. Penney and Lynch (191, E = 2.3 kv./cm. E. Hewitt (61,E = 0.6 kv./cm. F . Drozin and La Mer ( 2 ) G. Unipolar charging Flow system In room (21) H. Neutralized aerosol 0 Mass median size 0 Number median size I. Equation 1 (3, II ) J . Unneutralized aerosol A This study A Lundgren (75) K. Nolan and Keenan (17) I. Gillerpie and Langstroth (4). Equilibrium for 2 0 0 minutes
A.
From Equation 1 it can be seen that the equilibrium charge distribution does not depend on either the small ion or particle concentration. However, the concentrations d o determine the time to reach equilibrium. Although a theoretical calculation of the neutralization time is difficult because of the ion concentration gradients existing in practical neutralizers and because of the distribution of charges existing a t any given time, an equation suitable for an order of magnitude estimate can be derived from the differential equation for ion combination.
dt
Table I.
0.1 I.o Particle Diameter -Microns
0.01
Distribution of Charges on Particles in Equilibrium with a Bipolar Ion Atmosphere According to Equation 1 ~.
D,,Micron
0
1
2
3
0.01 0.015 0.02 0.03 0.06 0.1 0.3 1. o
0.093 0.955 0 900 0.763 0,550 0.424 0.241 0.133
0.007 0.045 0.100 0.236 0.430 0.48 0.41 0.253
0,001 0,020 0.09 0,232 -0.214
0.006 0.093 0.162
iVumber of Charges on Particle 4 5 6
0.024 0.109
0.005 0.065
0.035
7
8
9
10
0.017
0.007
0.003
0,001
VOL. 4
NO. 1
Av. 0.007 0.045 0.10 0,238 0.470 0.677 1.247 2.36
FEBRUARY 1 9 6 5
67
/ -7
M L,-
"r"! ! I
,../
n
"It
d = 1.59 m m r.2.38 m m
R P M Measuring C i r c u i t (A) (6) (C) (Dl (E) Aerosol Generator
,Aerosol
Outlet
Ion G e n e r a t o p
Compressed Air S y s t e m
Figure
1 2.5cfrn o f
a t 40 psig
Apparatus
1% here .VI = the concentration of charged particles, 7 = the small ion particle combination coefficient, and n = the small ion concentration. If n is assumed constant and Equation 2 is integrated from .VIOto .VI and t = 0 to t , then.
As a first approximation LV1/.l-lo may be taken as the fraction of charged particles at equilibrium. Gunn (5) from a consideration of the diffusion equilibrium also derived Equation 4 for the time to reach equilibrium, in terms of the small ion mobility. 2.and concentration, n.
For ion concentrations on the order of lo6 per cc. and for particle sizes in the range of 0.02 and 1 micron, both Equations 3 and 4 predict equilibrium times of the same order of magnitude---e.g., from 0.1 to 1 second. For neutralizer arrangements in which the ions are injected and mixed \vith the aerosol a t a single point, n in Equations 3 and 4 decreases with time because of the recombination of the small ions according to Equation 5.
(5) where no is the initial concentration and a is the recombination coefficient of the small ions. O n substituting Equation 5 in 2 and solving, the equilibrium time becomes LI
a no
68
I&EC
FUNDAMENTALS
Ion Generator Circdit
Spinning disk generator
Prlssure Regulator
Air
Figure 2.
3.
Feed System Satellite Removal Orifice Plate Liquid Feed Needle Disk Air Motor Satellite Air (0-3 0 c f m ) Ion Generatw Aerosol Discharge Main Air (0-150cfm) Absolute Filter Pressure Gages Air Line Oiler Pressure Regulators Air Line Filter 3.5 KVAC Transformer WOO ohm Earphone Collet of Air Motor Transformer 10 VAC Meter
Substitution of reasonable values of a and no into Equation 6 shows that when no in Equation 6 equals n in Equation 3, the equilibrium time from Equation 6 will be three to four times that calculated from 3 if a = q , and approximately the same if q is much greater than a as Gunn (5) suggests. Thus it is clear that if effective charge neutralization of aerosol streams is to be accomplished a t reasonable flow velocities, and within reasonable apparatus dimensions, initial ion concentration of a t least 106 and preferably from 10' to 109 must be used.
Experimental
Aerosol Generators. Uranine aerosols in the 0.01- to 0.26-micron size range were generated by means of the generator shown in Figure 2. T h e water solution of uranine is atomized by the Collison atomizer. In the impactor the resulting 1- to 10-micron spray is classified so that only droplets less than about 2 microns pass on to the mixing tee consisting of a 3:r-inch pipe tee cut as shown in Figure 2. In the mixing tee the droplets are mixed with a high concentration of bipolar ions generated by a sonic jet ion generator described by Whitby (20). The ions are transported into the tee in a free stream jet of about 2.5 cu. feet per minute issuing from a 1, 16-inch orifice. After neutralization and evaporation in the holding chamber. the solid dry particles leave at the aerosol outlet. Aerosols in the size range between 0.5 and 3.0 microns were produced with the spinning disk generator shown in Figure 3. T h e dye solution. fed onto the center of disk H a t about 5 cc. per minute, is spun off as a mixture of homogeneous primary droplets and heterogeneous satellites. T h e primary droplets pass upward xvith the main air stream, are evaporated and mixed with the ions, and leave at I as solid particles. In addition to the ion generator location shown in Figure 3. neutralization experiments were also run with a single generator located in an elbow 40 inches downstream from the spinning disk and with t\vo generators located 40 and 76 inches downstream from the spinning disk. as shown in Figure 6. The dyes. dye concentrations. solvents, and aerosol particle sizes for these experiments are showm in Table 11. A mixture of 1 part of uranine ivith 4 parts of methylene blue dye was used with the spinning disk because it made better aerosols than uranine alone.
the front edge of the plate, q is the particle charge, and E i? the electric field. A fluorometric analysis of the uranine on the foil-covered plates and afterfilter was used to determine the concentration of dye on each, T h e mass distribution of dye on each plate and the afterfilter was obtained with respect to x or y and converted to either a cumulative mass or charge distribution, with respect to D,,by Equation 7.
1N-d-'
AEROSOL IN CLEAN AIR
PERFORATED PLATEL
MICROMANOMETEF;
F E L T FILTER
I
\
iJ
I
I /Ix
FLOWMETER
'I1
)VALVE
9
VALVE
F'LTER\~y)
0
4
8
1
'HREE S T A G E BLOWER
2
Figure
4. Spectrometer
M e a s u r e m e n t of Particle C h a r g e , Particle charge and the particle size distributions of some of the aerosols were measured with a charge spectrometer constructed by Lundgren ( / 5 ) ,similar to that described by Langer (72). .4s shown in Figure 4, the spectrometer consists of a n aerosol inlet tube mounted mid\\ ay between two collecting plates carrying potentials u p to 10 kv. (+) and ( - ) relative to ground. T h e perforated plate and felt filter provide a laminar flow field. By proper adjustment of the two valves it is possible to control both the main air flow and the static pressure inside the precipitator and consequently the aerosol sampling rate. Charged fluorescent aerosol particles leaving the inlet tube are deposited either on the aluminum foil-covered plates or on the glass filter according to Equation 7, derived by equating the Stokes drag to the electrostatic force on the particle.
Here u. is the air velocity, y is one half the plate spacing or the distance from the center of the afterfilter, x is the distance from
Table II.
Aerosol Generator
Atomizer impactor
Spinning" disk
DY e Ilranine
1 part U b
4 parts MB
a
Disk diameter, 4.7 cm.; r.p.m., 42,000.
M e a s u r e m e n t of Particle Size a n d Size Distribution. T h e particle size distributions of the spinning disk aerosols were measured by light field microscope measurement of membrane filter-collected samples or in some cases by electron microscopy. However, accurate size distribution measurement of the atomizer impactor aerosols was more difficult. Consequently three different techniques were used, a t least two being used as a check on each aerosol. T h e 8, 1, and O.lyoaerosols were sized from electron micrographs of grids sampled with a low pressure impactor and with a thermal precipitator. The 1, 0.1, and 0.01% aerosols were also measured with the charge spectrometer shown in Figure 4, and with a modified Wesix-type parallel plate ion counter (7). Modifications to the ion counter include operation at from 0.2 to 1 cu. foot per minute, addition of a special flowstraightening entrance section containing several screens to ensure a flat, laminar velocity profile in the plate section, and a battery supply for the polarizing plates, which provide voltages from 0.3 to 575 volts. Collector plate currents were read with a 1O-13-ampere full scale micromicroammeter connected to a recorder. Current-voltage curves were analyzed graphically by the method of tangents to obtain the mobility distribution for those aerosols which had a representative fraction of the whole aerosols singly charged. This was found to be true only for the natural charge on the 0.01, 0.1, and 17,uranine in water aerosols. Figure 5 compares the size distribution of 1% uranine in water weighted by mass as determined by electron microscopy, charge spectrometer, and ion counter. T h e agreement is good, considering the difference in principle between the three methods. Since the charge spectrometer measures the size distribution weighted by mass directly, whereas the distribution obtained from the electron microscope count and the ion counter is weighted by number, analysis using both the charge spectrometer and ion counter together can provide more accurate size distribution data than either alone. A recent paper by Hurd and Mullins (8) provides a good discussion of the use of an ion counter for particle sizing of aerosols. They also obtained good agreement between the electron microscope and mobility size analysis of a salt aerosol made by evaporating a 0.1% salt solution. However, these electrical sizing techniques are accurate only if a representative fraction of aerosols is singly charged.
Particle Size and Charge Data for Aerosols
Solvent
Water
Denatured alcohol
DY e Concn., W e i g h t yo 8 1 0.1 0.01 0.1 0.01 0.002 0.0084
8 parts alc. 2 parts water U, uranine; M B , methylene blue.
Particle Size, Microns GeoNumber Mass metric median median std. deu. 0.09 0.26 1.67 0.054 0.103 1.41 0.028 0.049 1.49 0.016 0.028 1.45 2.6 2.6 1. I .
1.2 0.75 1.0
1.2 0.075 1 .o
Disk not grounded.
1.1 1.1 1.09
Au. N o . U n i t Charges on Aerosol UnneuNeutraltralized ired 3.62 1.42 0.75 0.59 0.80 0.24 0.75 0.13 304c 7.7 139d 180~ 4.7 46c 3.5 300d~
Disk grounded.
6
Unipolar charging (-) 8.7 2.92 1.01 0.28 75.2
29.6
Q5yoof particles negatiue
( 15).
VOL. 4
NO. 1
FEBRUARY 1 9 6 5
69
95
1 ION GENERATOR
80 W
E tn z I
60 40
I-
$ 20 W -I
B
A
10
0.I
IJ. 0.01 Figure 5.
0.I
0.2
PARTICLE DIAMETER -MICRON uranine in water
Size distribution of 1 %
Results and Discussion
Natural Charge. T h e unneutralized or natural charge w,is measured with the compressed air to the ion generators on, but with the ion generator voltage off. T h e measured average number of unit charges per particle irrespective of sign is given in Table I1 and plotted in Figure 1J against the number median size for the 0.01 6-, 0.028-, and 0.054-micron aerosol and against the mass median size for the rest. Also shown on Figure 1 for comparison are the unipolar charges obtained by a number of investigators. Langer’s ( 7 2 ) , Lundgren’s ( 7 5 ) , and Hewitt’s ( 6 ) field charging data using strong fields represent approximately the maximum charges that can be placed on particles. Drozin and La Mer’s (2) and Hewitt’s (6) low field data represent typical diffusion charges. A comparison of the natural aerosol charges with the above data shows that it may approach the maximum charge which may be placed on the particles in a corona charger. I t was found that the natural charge on the atomizer-impactor aerosols was about equally distributed between positive and negative, whereas the charge division for the spinning disk aerosols depended on the solvent-solute combination and whether the disk was grounded or not. From Table I1 it may be noted that grounding the disk decreased the average charge on the 2.6-micron aerosol from 304 to 139. With the disk grounded the ratio of negative to positive was about 2 to 1 and with it ungrounded the ratio was about 1 to 1. T h a t the charge could be equalized by insulating the disk was also noted by Jordan ( 9 ) . However, he did not measure the charge magnitude. Lundgren (75) also observed a very high nearly unipolar negative charge of 300 electrons on a 1-micron aerosol when using 8 parts of alcohol to 2 parts of water as the solvent. From Table 11 and Figure 1 J it can be seen that the average charge on the atomizer-impactor aerosols. for which D, < 0.1 micron, is constant and about equal to 0.78. I n reality, 70
I&EC FUNDAMENTALS
D
C
Figure 6.
Configurations
this should be 1, the difference between 0.78 and 1 being a measure of the absolute error of the charge measurements. T h e fact that the natural charge on these aerosols is constant for D, < 0.1 micron means that a representative fraction of the aerosol is singly charged and that the electrical particle sizing techniques described earlier can be used with acceptable accuracy to size these aerosols. Charge after Neutralization. Average measured charge irrespective of sign after neutralization is shown in Table I1 and Figure IH. T h e data for the atomizer impactor aerosols were obtained with a single ion generator located as shown in Figure 2. This arrangement had been previously determined to be optimum. Effective neutralization in this case is fairly easy because the small total air flow, 0.4 cu. foot per minute from the impactor and 2.5 cu. foot per minute from the ion generator, permits the aerosol to be thoroughly mixed with the high concentration of ions immediately outside the ion generator. However, effective neutralization of the spinning disk aerosols is not quite so simple because of the greater air flow (50 to 100 cu. feet per minute), and longer evaporation time of the larger droplets (about 25 microns in these experiments) makes it difficult to expose all of the aerosol particles to a high ion concentration for a sufficient time. The design of the original configuration (Figure 6 A ) , which had been based on the configuration used on the Model I generator (22),proved to be ineffective because the droplets were not yet completely evaporated at the point of ion injection. Though some reduction in charge was obtained by moving the ion generator 2 feet further downstream (Figure 6B and Table 111), the most effective neutralization was obtained with the configuration of Figure 6C. T h e data shown in Figure 1 and Table I1 were obtained with this arrangement. Apparently the first ion generator reduces the charge on the droplets just before the completion of evaporation, thereby preventing drop distortion,
I
ODOOl
I
D, greater than 0.1 micron the experimental charge is greater.
1.118
@
For Dpt0.16, ep=I; For D,,~Oo16,ep=l027D,
@
For D,0.032,e$425 Dk76
@
For D,
