T h e individual cyclones of t h e system were tested and calibrated in the laboratory under conditions similar to those frequently encountered in field tests: gas temperatures of 25, 93, and 204 “C, flow rates of 7.1, 14.2, and 28.3 L/min, and particle densities of 1.05, 1.35, and 2.04 g/cm3. T h e Djo’s for the cyclone system a t various operating conditions are given in Table IV. For laboratory test conditions the small cyclones have sharp collection efficiency curves, and thus t h e system should function adequately as a particle sizing device. Data from this study (wherein different particle densities, p , were used) tend t o support the D50 vs. p-’I2 relationship suggested by several theories (4, 17, 18).On t h e other hand, the experimental results indicating that the D ~ o ’were s directly proportional to the gas viscosity are in opposition t o most theories ( 4 , 5, 17, 18). Also, it was found in this study and others t h a t the D50’s of small cyclones are not inversely proportional to the square root of the flow rate as some theories predict (7). I t appears t h a t establishment of a definitive relationship between D50 and flow rate will require further investigation.
Acknowledgment T h e mechanical design was done by David Hussey and part of the experimental data were taken by Don Johnson.
Literature Cited (1) McCain, J. D., McCormack, J. E., Harris, D. B., paper presented
a t the 70th Annual Meeting of the APCA, Toronto, Ontario, Canada. 1977. Paaer 77-35.3. (2) Felix, L. G.; Clinard, G. I., Lacey, G. E., McCain, J. D., EPA600/7-77-060 ( N T I S P B 276583), U.S.Environmental Protection Agency, Research Triangle Park, N.C., June 1977,89 pp.
(3) Smith, W. B., Cushing, K. M., Lacey, G. E., McCain, J. D., EPA650/2-74-102-a ( N T I S No. P B 2451841, US.Environmental Protection Agency, Research Triangle Park, N.C., 1975, 132 pp. (4) Lapple, C. E., Chem. Eng., 58, 144-51 (1951). (5) Leith, D., Licht, W., “The Collection Efficiency of Cyclone Type Particle Collectors-A New Theoretical Approach”, AIChE Symposium Series, New York, 1971, pp 196-206. (6) Leith, D., Mehta, D., Atmos. Enuiron., 7, 527-49 (1973). (7) Chan, T., Lippmann, M., Enuiron. Sei. Tpchnol., 11, 377-82 (1977). (8) Hochstrasser, J. M., Ph.D. Thesis, University of Cincinnati, 1976. (9) Cushing, K. M., Farthing, W., Felix, L. G., McCain, J. D., Smith, W. B.. EPA-600/7-78-009 ( N T I S P B 279170). U S . Environmental Protection Agency, Research Triangle Park, N.C., 1978, 174 pp. (10) Smith, W. B., Wilson, R. R., Jr., EPA-600/7-78-008 ( N T I S P B 2790841. U.S.Environmental Protection Aeencv. Research Triangle Park, N.C., 1978,66 pp. (11) Berelund. R. N.. Liu. B. Y. H.. Enciron. Sci Technol.., 7.147-53 . (1973): (12) Cushing, K. M., Lacey, G. E., McCain, J. D., Smith, W.B., EPA-60012-76-280 ( N T I S No. P B 262849), U.S. Environmental Protection Agency, Research Triangle Park, N.C., 1076, 94 pp. (13) Supplied by Dow Diagnostics, the Dow Chemical Co., Indianapolis, Ind. (14) Calvert, S.,Lake, C., Parker, R., EPA-600/2-76-118 ( N T I S No. P B 262656). U.S. Environmental Protection Aeencv. Research Triangle Park, N C., 1976 (15) Weast. R. C.. E d . “Handbook of Chemistrv and Phvsics”. 44th ed., The Chemical Rubber Co., Cleveland, Ohio, 1961 , p p 2264-5. (16) Lippmann, M., Chan, T., A m . Ind. Hyg. Assoc. J . , 35, 189-200 (1974). (17) Muschelknautz, E., Staub, 30(5), 1 (1970). (18) Sproull, W. T.,“Air Pollution and Its Control”, Exposition Press, New York, 1970, Chapter 6, pp 70-75.
- -
-
Kccc~ivedfor reuieu J u l y 10, 1978. Accepted July 27, 1979. This uork ri’as supported by the L’.S. Encironrnental Protection Agency under (’ontract No. 68-02-2131.
Particle Bounce in Cascade Impactors Yung-Sung Cheng’ and Hsu-Chi Yeh Inhalation Toxicology Research Institute. Lovelace Biomedical and Environmental Research Institute, P.O. Box 5890, Albuquerque, N.Mex. 871 15
Particle bounce in inertial impactors has been studied. A semiempirical model t h a t relates impact velocity a t the inception of bounce, particle diameter, and particle density is proposed. Impact velocity on the center line of both round jets a n d rectangular jets was calculated by solving the equation of particle motion. Experiments were performed to evaluate t h e retention efficiency of the Sierra radial slit jet impactor using latex particles collected on uncoated stainless steel. The critical value of the product of calculated impact velocity, particle diameter, and square root of specific gravity was found and 4.64 X mn/s a t the incepto be between 2.50 X tion of bounce. Values larger than this predict particle bounce. Experimental results show that effects of particle bounce are reduced collection efficiencies, increased wall losses, and altered collection distribution among stages. Criteria are established to minimize these effects by adjusting the operating conditions in both round and rectangular jet cascade impactors. Cascade impactors are useful instruments for fractionating airborne particles according to their aerodynamic size and subsequently measuring the size distribution. One of the inherent problems is particle bounce off the collecting plates. 1392
Environmental Science & Technology
Many investigators have found that particle bounce increases wall losses and also gives biased size distribution (1-5). In practice, the collecting surface is usually coated with grease to reduce bounce effects. However, grease coating is not suitable in sampling of many aerosols such as those from coal combustion because it could interfere with chemical and gravimetric analyses and would be ineffective in high-temperature sampling. Reduction of collection efficiency in cascade impactors due t o particle bounce was reported by Rao and Whitby (5). Esmen et al. (6) investigated particle bounce in a single stage impactor in conjunction with adhesion theory. A semiempirical criterion of inception of bounce in cascade impactors is established in this report. A Sierra radial slit jet impactor (SRSJ) (Sierra Instruments, Model 216) was used in this study. T h e impactor was calibrated, and the reduced collection efficiency as the result of particle bounce was measured for each stage. T h e critical impact velocity a t inception of bounce was calculated. Semiempirical criteria for both rectangular and round jet impactors were established, and the critical condition relating size and density of particles and impaction velocity was derived from Dahneke’s adhesion theory (7). Effects of particle bounce on wall losses are also reported.
0013-936X/79/0913-1392$01,00/0
@ 1979 American Chemical Society
Table 1. Operation Parameters of the SRSJ Impactor (Model 216) (Flow Rate = 21 Llmin, Temperature = 25 OC, Pressure = 760 mmHg) stage no.
0 1 2 3 4 5 a
Re = 2pg V, W i p
total slit length, L , cm
Slit width, W , crn
0.3556 0.1956 0.1118 0.0610
5.156 5.141 3.942 3.863 3.889 2.291
0.0330 0.0279
vo
1
Re a
m/s
868 870
1.91 3.48
1134 1161 1152 1953
7.97 14.9 27.3 54.7
50% cut-off diameters of unit density given by manufacturer.
D50,
*
D50r
Crn
m
16.0 8.6
12.1 6.4
3.9 2.4 1.2 0.61
3.1 1.7 1.o 0.57
5 0 % cut-off diameters of unit density calculated from calibration
curves.
Diluting AV
Dilufmg Ail
Vaive
Compressed
Ftifer
OPTICAL
Figure 1. Schematic diagram of the experimental setup
Calibration T h e SRSJ impactor has six stages and a backup filter. Dimensions were rneasured and are given ,in Table I. T h e impactor was assemhled with all stages except those helow the test stage and backup filter. Collection efficiency curves of each stage were measured for both coated (Dow-Corning Anti-Foam A) and uncoated stainless steel plates. T h e experimental setup is shown schematically in Figure 1. Aerosols of monodisperse latex particles were generated by a Retec nebulizer modified for constant feed with a syringe p u m p (Harvard Model 3050) to give steady particle concentrations. Liquid feed rate was 0.59 mL/min using a 50-mL syringe a t pump gear position 14. T h e variation of concentration was within &5% over 1-h continuous operation. Concentrations of particles penetrating impactor stages were measured using a Climet Model 208A optical counter (Climet Instruments). T h e sampling rate was 7 L/min for 1 min. T h e collection efficiency was derived from the relation E = 1 n/no, where no and n are the upstream and downstream concentrations of the test stage, respectively. Volumetric flow rates through t h e impactor were varied f'rom 7 to 40 L/min a t an atmospheric pressure of 620 mmHg and 23 "C. Collection efficiency curves for stages 0 to 5 for coated surfaces are plotted in Figure 2 as a function of Stokes number. Also shown in Figure 2 are Marple's theoretical efficiency values for rectangular jets for Reynolds numbers of 500 and 3000. T h e 50% cutoff points in all stages are close to theoretical predictions. As shown in Table I, the 50% cut-off diameters for normal temperature and pressure are lower than values given by the manufacturer (Bulletin No. 175-216). Ef'ficiency curves are not as sharp especially for the first three stages. Efficiency curves level off a t a maximum of about 89%' efficiency. This may be due to bounce between airborne latex
Figure 2. Collection characteristics of SRSJ impactor, stages 0 through 5: (0)coated stainless steels plates; ( A )uncoated plates; ( 0 )50% cut-off point of Marple's theory: the dotted lines are hand-fitted curves and solid lines are Marple's theory for rectangular jet for Re = 500 and 3000
particles and collected particles, since subsequent experiments using DOP oil droplets showed that essentially 100%collection efficiency was obtainable.
I'urticle Bounce According to t h e bounce theory proposed by Dahneke ( 7 ) , the critical impact velocity, u,, when particle bounce starts can be related to t h e particle surface properties:
where 11 and p" are particle density and unit density, respectively, d,, = particle diameter, A = Hamaker constant, e = coefficient of restitution, and 2 0 = the equilibrium distance between particle and surface. T h e right-hand side of Equation 1 would remain the same for a fixed particle surface system. T h e particle impact velocity a t the impaction surface stagnation point can be calculated by solving t h e following dimensionless equation ( 2 ) of particle motion: Volume 13, Number 11, November 1979
1393
Table II. Parameters at Inception of Bounce for the SRSJ Impactor particle dlameter,
particle density,
stage
dP. wm
glcm3
mls
1 2 3 4 5
5.7 3.40
1.05 1.027 1.027 1.05 1.05
3.90 6.44 10.37 20.01 26.40
P
2.02 1.011 0.822
vo
9
0.455 0.467 0.486 0.486 0.501
0.111 0.137 0.177 0.177 0.209
2.50X IOp6 3.03X lop6 3.76X 3.67X 4.64 X
0.433 0.881 1.84 3.55 5.51
797 755 667 697 774
I
I
Andersen lmpucfor
'.,
f
SRSJ Impactor
1
2
I 4
6
PARTICLE D I A M E T E R . ( p m )
Figure 4. Comparison of particle bounce conditions for impactors with round and rectangular jets. Data for the Andersen impactor are derived from Rao and Whitby (5)
RECIPROCAL OF STOKES NUMBER,
1 -Vi = I - - + - +1( ) VO 4St 96St2
&
Figure 3. Impact velocity at stagnation point of collection surface (Equation 4). The curve representing 1 - (1143)is asymptotic at large Stokes number
du 2St - = -v dt
+ ug
where u and u g are particle and gas velocities normalized by the average gas velocity V Oa t jet throat, St = C p d P z V&pDh, c = the slip correction, p = viscosity of gas, and Dh = hydraulic diameter = W (diameter) for round jets and 2W for rectangular jets. As a first approximation, the flow fields proposed by Mercer and Chow (8)and Mercer and Stafford ( 9 ) ,in the stagnation streamline for both round and rectangular jets, are used, which can be written as: ug =
1
-1 -y
y > l O l y I I
(3)
where y is the distance from the impaction surface normalized by Dhl4. Substituting Equation 3 into 2, the resulting equation can be solved analytically with initial conditions at t = 0:
T h e result of the impact velocity a t stagnation point, Le., the particle velocity evaluated a t y = 0, is: S t I 118
For larger Stokes number (St > l), the asymptotic solution is given by: 1394
Environmental Science & Technology
(k3)
Figure 3 shows the impact velocity as a function of l/St. From efficiency curves (Figure 2), the critical impact velocities, V,, a t the inception of particle bounce were calculated according to Equation 4 and tabulated in Table 11. The values to 4.64 X m2/s for of u c d p mrange from 2.50 X the SRSJ. In Figure 4,these results were compared t o those derived from a study by Rao and Whitby ( 5 )for the Andersen cascade impactor, which gave values of 3.40 X to 4.17 X m2/s. In their study, the Andersen impactor was tested with methylene blue-uranine particles. Although the data are scattered, they do indicate the order of magnitude for solid particles bouncing off dry surfaces. Wall Losses Effects of particle bounce on wall losses were studied using the same impactor. T h e experimental setup is similar to Figure 1, except that the Bergland-Liu vibrating orifice aerosol generator (Thermosystems Inc.) was used instead of a nebulizer. Monodisperse fluorescent particles ranging from 1.7'7 to 7.02 pm were produced using an isopropyl alcohol solution of Eosin-Y. The SRSJ impactor was assembled including all stages and backup filter. A sample was taken for 5 to 10 min a t flow rate of 20 L/min. Particles collected on collection plates, backup filter, and stages were washed off with water (pH 10) and the mass of Eosin-Y was determined by the standard fluorometric technique. T h e mass distributions of collected particles for both coated (Anti-Foam A) and uncoated stainless steel surfaces are shown in Figure 5. On uncoated plates, particle bounce occurred on stages where decreased collection efficiency was observed. T h e calculated values of V,dp\/plp* a t plates where particles start bouncing are 4.50 X 10-6 to 9.2 X lop6 m2/s. T h e values are slightly higher than previously obtained using latex particles, and may indicate t h a t Eosin-Y particles are less prone to bounce than are latex particles. Particles bouncing off uncoated surfaces of early stages stay
Table 111. Wall Losses in the SRSJ Impactor oarticle diameter, dp, Pm
staoe no. where bounce begins
coated
7.02
1
10.0
5.05 3.04 2.93 1.77
2 2 2 3
9.7
12.5 13.7
3.6 2.9 5.7
3.8 6.1 5.5
upper part uncoated
wall losses. a % lower part coated uncoated
5.3
45.0
1.9 4.6 0.8 0.5
40.0 38.7 32.5 14.3
total coated
uncoated
15.3 11.6 8.2 3.7
57.5 53.7 42.5 38.6 19.8
6.3
a Wall losses are separated into two parts, the upper part includes stage 0 down to the stage where bounce begins: the lower part includes the rest of the stages. Wall losses are expressed in percent of total mass collected by an impactor. Volumetric flow rate is 20 Llrnin.
STAGE
BA~KUP
NUMBER
L i 1
d p 7.02 p m j
Uncoated Plates Cooted Plates 5 4
0
0
I
~
/ I
-
L
O0
0
I
Coaled Plafes
0 I
2
I
I
PARTICLE 4 DIAMETER, 6 (pm)
I
8
9
Figure 6. Wall losses of SRSJ impactor
0
Figure 5. Mass distribution of fluorescent particles collected on stages and backup filter of SRSJ impactor
airborne until they are either lost on the walls or collected on the backup filter. Most of the large particles were lost on the wall while small particles were found in both places. Wall losses were separated into upper losses containing stage 0 down to the stage where particle bounce starts and t h e lower losses including the remaining stages as shown in Table 111. I t is shown that the increases of wall losses on uncoated plates are in the lower part, Le., after t h e stage where bounce starts. Total wall losses for both collecting surfaces as a function of particle size are shown in Figure 6.
Discussion a n d Conclusion As reported by other investigators, experimental results indicate t h a t particle bounce does occur on uncoated dry surfaces of impactors. Effects of particle bounce are reduced collection efficiencies, increased wall losses to the subsequent stages, and shifted distribution among stages. A semiempirical relationship indicat.ing t h e inception of particle bounce is given: -
I
Vid,,
I/4 2R X P
mn/s
(6)
where H has values ranging from 2.5 to 9.2 depending on the material and size of particles, type of impactor, etc. As a general criterion For particle bounce in impactors, a value of 5 is chosen for R . Thus, to minimize particle bounce in impactors, the following operating condition should be satisfied for a unit density particle: V,d,, i 5 x lo-" m'/s
(7)
In designing a cascade impactor with minimum effects of particle bounce, it is desired that an impactor stage must not only be able to collect particles a t its own 50%cut-off diameter, dso, but also it must collect particles as large as t h e d;,o of the
preceding stage. If impactors are designed to operate a t optimal conditions according to Marple and Willeke ( 1 0 )where v ' G = 0 . 4 9 3 and 0.464 for rectangular and round jet impactors, respectively, then:
Vi,nds0,n-1
where the slip correction was neglected, Q = volumetric flow rate (L/min), W = nozzle diameter or slit width (cm),L = total length of slit in a stage (cm), N = number of nozzles in a stage, and subscripts n and n - 1 denote stage n and its preceding stage n - 1. Equations 7 and 8 not only can be used as a guide to design a cascade impactor with minimum particle bounce, they also can be used to determine the optimal operating conditions for a given cascade impactor. Cascade impactors are usually designed such t h a t values of L and N W decrease with subsequent stages. Therefore, values of Vi,nd:o,n-l have a maximum in the last stage. For a given impactor the volumetric flow rate can be reduced to minimize the particle bounce. For instance, the SRSJ impactor has to be operated below 5 I,/min in order to prevent bounce. However, these criteria are not general, but are for particles with bounce properties similar to latex or fluorescent particles. T h e bounce characteristics of other particles may be different.
A c k n O I C 1e dg me nt s T h e authors gratefully acknowledge the assistance of Dr. S. J. Rothenberg, Dr. R. L. Carpenter, and Mr. G . J. Newton for technical review of the manuscript, Mr. E. E. Goff for illustrations, and Mrs. CT. Miller for typing t h e manuscript. Volume 13, Number 11, November 1979
1395
(6) Esmen, N. A,, Ziegler, P., Whitfield, R., J . Aerosol Sci., 9,547-56 119781. -, (7) Dahneke, B., J . Colloid Interface Sci., 37,342-53 (1971). (8)Mercer, T. T., Chow, H. Y., J . Colloid Interface Sci., 27,75-83 (1968). (9) Mercer, T. T., Stafford, R. G., Ann. Occup. H j g . , 12, 41-8 1 1969). ( 10) Marple, L'. A,, Willeke, K., Atmos. Enciron., 10,891-6 (1976).
Literature Cited
~~~
i 1) Cushing, K. M., McCain, J. D., Smith, FV. B., Enuiron. Sei.
Tcchnol., 13, 726-31 (1979). ( 2 ) Dzubay, T. G., Hines, L. E., Stevens, R. K., Atmos. EnL'iron., 10, "9-34 11976). (3) Peele, E. R., Newton. G. J., Yeh, H. C., "Effect of Collection Substrates on Performance and Wall Losses in Cascade Impactors", Inhalation Toxicology Research Institute Annual Report, LF-58, 1976-1957, p p 264-6. (4) Rao, A. K., Whitby, K. T., J . Aerosol Sci., 9,77-86 (1978). ( 5 ) Rao, A. K., Whithy, K. T., J . Aerosol Sci., 9,87-100 (1978).
I k c , i i e d for reuieio May 21, 1979. Accepted Jul? 27, 1979. Research pcjr/ormed under I.'.S. Department of Energy Contract No. E Y 76-('-04I O 1 9.
Removal and Recovery of Organic Pollutants from Aquatic Environment. 1. Vinylpyridine-Divinylbenzene Copolymer as a Polymeric Adsorbent for Removal and Recovery of Phenol from Aqueous Solution Nariyoshi Kawabata" and Kozo Ohira Department of Chemistry, Faculty of Polytechnic Science, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606, Japan
Vinylpyridine-divinylbenzene copolymer was found t o have an excellent capacity for removing phenol from aqueous solution. T h e breakthrough capacity of t h e copolymer with