( 8 ) Burns, W. G., Bernstein, H. J.. J . C h e m . Phys., 18, 1669
(1950). (9) Eberhardt, W. H.. Burke, T. G.. ibid.. 20.529 (1952). (IO) Robbins, R. C., Cadle, R. D., Eckhardt, D. L., J Meteorol, 16.53 (1959). (11) Cadle, R. D., Robbins, R. C., Discuss. Faraday S O C ,30, 155 (1960). (12) Calvert, J. G., Pitts, J. N., Jr., “Photochemistry,” p 230, Wiles & Sons. New York, N.Y., 1966. (13) Cotton, F . A . , Wilkinson, G., “Advanced Inorganic Chemistry.” p 355, Interscience, New York, Y.Y., 1966. (14) Beckman. L. J., Fessler, W. A , , Kise, M . A , , Chem ReL 48, (3), 319 (1951). (15) U.S. Department of Commerce, Xational Bureau of Stan~
dards. Circ. 500-Part I. “Selected Values of Chemical Thermodynamic Properties,” 1952. (16) Stull, D. R., Westrum, E . F., Jr., Sinke, G . C., “The Chemical Thermodynamics of Organic Compounds,” pp 224, 232, Wiley & Sons, New York, N.Y., 1969. (17) Latimer, W. M.. “The Oxidation States of the Elements and Their Potentials in Aqueous Solutions,” 2nd ed., Prentice-Hall, New York, N.Y., 1952. Receiced for recieu Jul? 20, 1973. Accepted April 22, 1974. This Lcork U Q S made possible b3 grants from the National Science Foundation (GP-34238)and the Air Pollution Control Office. EPA (AP-00357-07).
Aerosol Filtration by Means of Nuclepore Filters Filter Pore Clogging Kvetoslav R. Spurny* Institute o f Aerobiology, 5949 Grafschaft, Germany
Jarmila Havlova J. Heyrovsky Institute of Physical Chemistry and Electrochemistry, Czechoslovak Academy of Sciences, Prague, Czechoslovakia
James
P. Lodge, Jr., Evelyn R . Ackerman, and David C. Sheesley
National Center for Atmospheric Research, Boulder, Colo. 80303
Boris Wilder Institute for Medical Research and Occupational Health, Zagreb, Yugoslavia
Experimental Experiments and results are described to help clarify the filtration kinetics and pore clogging mechanisms during filtration or sampling of aerosols with Nuclepore filters (NPF). Monodisperse latex aerosols (MLA) of different sizes were used to clog the filter. The curves of pressure drop, L p , and collection efficiencies, E, as functions of time, t , have very similar shapes; they are thus an aid to identifying the clogging mechanisms and to determining the equations L p = f ( t ) and E = F ( t ) . (A subsequent paper will deal with the equations.) Electron microscopy makes the NPF surfaces visible a t the different clogging phases and helps one to imagine how particles of different sizes settle inside and outside the pores. In previous papers (1-6), the filtration properties of clean NPFs were studied and equations given to describe the dependence of the filter pressure drop, Lp, on the gas filtration face velocity, the gas pressure, and the gas temperature. In addition, the dependence of the filter collection efficiency, E, on the previous parameters, as well as on the aerosol particle radius r, was presented, verified, and discussed. Although some preliminary considerations and measurements were also presented in those previous papers ( I , 4, 7), NPF clogging had not been investigated systematically. In this communication the equipment, measurement technique, and experimental results of such an investigation are described and discussed. Mathematical description of the clogging theory and comparison with experimental results will be the subject of a subsequent communication. 758
Environmental Science & Technology
Filters. American analytical NPFs (Kuclepore Corp., Pleasanton, Calif.) with pore diameters from 0.5-0.8 pm were used for the investigation. Structural properties (such as pore radius, porosity, thickness) were measured according to methods described previously (3, 7). Aerosols. Monodisperse latex aerosols (MLA) were used to produce clogging. Suspensions of monodisperse latex (Dow Chemical Co., USA and Serva, Heidelberg, Germany) with particle sizes, 2 r, of 0.088 pm, 0.268 pm, 0.312 pm, and 0.796 pm were used for aerosol generation. The aerosols were generated by atomizing a dilute water suspension of the particles in a glass nebulizer (Jouan, Paris; similar to the U.S.-made DeVilbiss nebulizer); the suspensions were diluted from their original concentrations (10%) according to the statistical theory of dilution ( 8 ) using nebulizer characteristics determined by Mercer et al. (9). Suspensions of the particles in bidistilled and filtered water with concentrations from 0.01-0.2% were the most practical according the particle size. The aerosol concentration (cme3 or pg m - 3 and so forth) and the number of aggregates of ML particles were measured by electron microscopy after the aerosol had been sampled in a thermal precipitator (Casella). Table I shows some of the results. Agreement with the theory of Raabe ( 8 ) for the smallest particles ( r = 0.044 pm) was not satisfactory; use of much lower concentrations brought no improvement. Probably agglomerates are already present in the water suspension (Table I). Equipment. Measurement of E during the clogging process was accomplished with practically the same equip-
ment as reported earlier ( 4 ) . The equipment for the measurement of I p during clogging is depicted in Figure l . MLA was produced by atomizing a dilute suspension of latex particles in glass atomizer, G. Pressurized nitrogen gas, filtered by means of a high efficiency glass fiber filter, F, was used for atomizing the suspension and, if necessary, for diluting the aerosol. Flow rates and nitrogen pressures for the atomizer (R1 and M I ; MI = 0.9 atm) were measured; these parameters were also measured (R2 and M 2 ) when gas dilution was used. The Plexiglas chamber, K1, had a volume of about 0.3 m3. The electrical charge of MLA was decreased by irradiation with a Krypton source (S5Kr,5 mCi) (10). The MLA was then passed through a tared filter, NPF. The gas pressure was measured in front of and behind the filter, -443 and Mq, manometers in each pair were filled with dioctylphthalate and with mercury, respectively. Flow rates through the K P F were measured by means of rotameters, R3. In the chamber, K2, it was possible to extract samples for measurement of the aerosol concentration in front of the NPF and for observation of particles and agglomerates by electron microscopy. Samples were collected by means of a thermal precipitator, Casella T. Procedure of I p Measurement. The glass atomizer was filled with a dilute ML suspension. Once the atomizer's output was consistent (the conditions for atomizing a defined MLA were determined for each ML before the clogging experiment began), the pump, P, was started in order to draw the aerosol through the NPF (the initial value of I p was therefore low). The N P F was weighed on a Cahn electrobalance ( 1 2 ) before and after the experiment. The weight difference allowed the number of latex particles per pore, P,, at the end of the experiment to be approximated. Latex aerosol concentrations were also measured by meanb of a Royco particle counter. Parameters that could be changed included the pore sizes of NPF, R, the particle sizes of MLA, r, and the face gas filtration velocities, q The values of I p therefore could be measured as a function of time and as a function of the P, values for different values of R, r, q , and p The dependence of -Ip on the gas pressure, p , has already been published (3). After each experiment the surfaces of NPFs contaminated with MLA particles examined under an electron microscope or a scanning electron microscope. These techniques have been described elsewhere ( 2 1-23). Procedure for Measuring E . As noted above, equipment for this measurement has been described previously ( 4 ) . Since the Royco counter can measure only relatively low particle concentrations, the equipment included two MLA generators and was operated in short time steps. The NPF was weighed before and after contamination and was carried out over a short time interval with a high aerosol concentration (e.g., lo6 ~ m - ~ Then ) . followed a second time interval during which the value of E was measured by means of a Royco counter a t a lower aerosol concentration (e.g., lo2 ~ m - ~Altogether ) . about 10 intervals were used to measure one NPF. During each interval the value of -Ip was measured as well.
Good reproducibility was obtained in the measurement of Ip and E through three to five repetitions, with a fresh NPF of the same pore size used each time.
Results and Discussion In some of our previous papers ( 4 , 6) the various separation mechanisms were delineated (such as diffusion separation, direct interception, impaction of aerosol particles on the filter surface). The main parameters determining the collection efficiency of an N P F are r, R, arid q . These parameters are also important for determining the different clogging mechanisms. Therefore, the mathematical description that will follow this communication can only approximate each individual separation mechanism. If diffusion separation is the prevalent separation mechanism (e.g., 0.1 < q < 10 cm sec-I, r/R
