An acterization of an aerosol with relatively high precision, and most importantly, without prior assumptions about size distributions. accurate picture of real size distributions should emerge. This is particularly useful for measurements of aerosols which are rapidly aging. Refractive index measurements can be useful for studies of aerosols in which composition changes may occur during formation and aging. Solvent volatilization can be followed during aerosol aging, and component partitioning during aerosol formation can be recognized. Extension to other size aerosols have not yet been completed, but initial efforts with 0.5 μm particles show no barriers for that size range. Visual comparisons of calculated curves indicate that the present instrument will allow sizing of 0.3 μm particles to within about two percent. On the other end of the scale, the apparatus can be set to size 20 μm particles to within at least one percent, and it is probable that the range be extended to somewhat larger particles without great difficulty. REFERENCES [1] Gucker, F. T. and Egan, J. J., J. Colloid Sci. 16, 68-84 (1961). and Rowell, R. L., Disc. Faraday Soc. 30, 185-191 J., and Berkman, R. M., J. Colloid [3] Phillips, D. T., Wyatt, P. Interface Sci. 34, 159-162 (1970). [4] Breuer, Breuer, H., Gebhart, J., and Robock, K., Staub-Reinh. Luft 30, No. 10, pp. 25-31 (English Translation) (1970). [5] Blau, H. H., McCleese, D. J., and Watson, D., Appl. Optics 9, 25222528 (1970). [6] Tuma, J. and Gucker, F. T., Proc. 2nd Int. Clean Air Congr., p. 463 (Academic Press, New York, N.Y., 1971). [7] Gucker, Gucker, F. T., Toma, J., Lin, H.-M., Huang, C.-M., Ems, S. C., and Marshall, T. R., J. Aerosol Sci. 4, 389 (1973). DISCUSSION MILTON KERKER: I noticed that your polystyrene distribution was negatively skewed. It was asymmetric with relatively more particles of the smaller size than you get from a Gaussian distribution. We found the same thing. This is opposite to what people normally accept as natural distribution, i.e., a positively skewed logarithmic distribution. It will be interesting to see how that holds up further work and whether the asymmetry has some statistical information to impart concerning the formation of these particles. up under PARMENTER: We would, of course, feel much more secure had we measured as many particles as you did. However, these results from about 100 particles, so I'm sure the trend is correct. I've no reason to believe that this distribution is in error. KERKER: The other thing is that we also tried to fit the size with the refractive index, and I would have to go back and look at our data again. I think they were within the same ballpark. We did not consider this a successful way of getting the refractive index. That is, we found also that if you perturb the refractive index by about .01, you got a size that was different by an amount greater than you would normally be willing to accept. This is something, of course, that would be interesting to pursue. feel that PARMENTER: We are finding the opposite now. We feel generally the computer can make a distinct choice of combinations. KERKER: We were strictly using visual estimates. One could use a method of steepest descent or something to come into the well and find out where you are, but we really didn't do anything like that. We just kind of eyeballed these .... PARMENTER: We were unable to do this by eye alone. The index of refraction is more difficult to see by eyeballing, so your experience is consistent with ours. KERKER: It's worth looking at more precisely. Finally, what criterion did you use as a measure of fit between experimental and computer results? We just used the least squares type. ROBERT KNOLLENBERG: It appears that once your particle size is as you say up to about 20 microns, you have several more points to fit in your set-up. Is this not correct? It appears that you have a problem in resolution if you are to acquire all the data counts in 20 milliseconds. PARMENTER: We have 5° resolution in the optics, and the resolution on the data acquisition is about 1°. 55 KNOLLENBERG: It is true your number of data goes up as you' particle size increases? PARMENTER: It would have to go up as the particle size increases. That's clear. EDGAR ETZ: When you do these computer fits, I believe you said yc. analyze the sum of the squares of the deviations of the experimental values from the calculated curves. You must have an estimate of tre uncertainty in the experimental scattering intensities because you might come up with an equation that fits the experimental curve much better than is actually warranted. PARMENTER: We fit the noise in other words? ETZ: Yes. Consequently, you get so many different solutions--some are real and some are not real. You come up with more than one mini-a value in this standard deviation. PARMENTER: We haven't yet made a careful analysis of it. I might say that we have worried about it and made a preliminary exploration of it in the following way. You saw the four curves which came from the same particle. That represents very typical noise--at least in this particle range. We let the computer look at each one of those curves individually to tell us what the particle was. The computer decided that the particle was particle was identical in three of those cases and in tre fourth case the computer decided that the particle was different in size by 5/1000 of a micron--about 0.4 percent which is a neighboring size increment. So this indicates that noise is not going to introduce a serious error in this analysis. GREG ROSASCO: You said you used a logarithmic amplifier? Would pulse counting techniques and then a computer conversion to logarithmic output help you in your sensitivity? PARMENTER: I think it probably would. We are in the process now of considering how to build a better instrument, and this is one of tre things we have to consider. The problem that we would encounter with pulse counting techniques is pulse pile-up. I don't know if this is going to be severe or not. We haven't yet gone through it carefully. It would be great if it works because that would eliminate several other problems. SINNOTT: For about $500 you can buy a phototube that has a two microsecond pulse pair resolving time. m NATIONAL BUREAU OF STANDARDS SPECIAL PUBLICATION 412 ACTIVE SCATTERING AEROSOL SPECTROMETRY Robert G. Knollenberg Particle Measuring Systems, Inc. 5469 Western Avenue Boulder, CO 80301 ABSTRACT An active scattering aerosol spectrometer is one that uses the active open cavity of a laser as the source of particle illumination. The interferometric aspects of the oscillating radiation illuminating the particles produces both forward and backward scattered radiation at all collecting angles. This, coupled with the fact that the collecting optics solid angle can be considerably greater than one steradian, and extremely intense source, results in a system with an extremely high sensitivity, fully capable of sizing particles several hundred Angstroms diameter using solid state silicon detectors. Key words: aerosol light scattering; aerosol sizing; aerosol spectrometer; cloud droplet measurements; interferometer; laser imaging of particles; laser light scattering by aerosols; particle size measurements. I've decided to change the content of my presentation considerably in light of the presence of the FDA and their interest in aerosol aerosol can products, but I will cover the active scattering spectrometry at the end of the presentation. What I'm going to do is outline several techniques that we use to design instruments which we build for various customers. These range from imaging systems to single particle extinction. With particle scattering we have two types of techniques: one we call classical scattering and the other we call active scattering. Let me mention briefly that the type of imaging system we use essentially involves a collimated light field where we project the shadows of particles and onto a photodetector array (fig. 1). At one time we did use fiber optics with photomultiplier tubes; today we use photodiode arrays. This type of system is one that we designed for use in sizing droplets from aircraft and has a size range at aircraft speeds of about 5 to 5,000 micrometers. Not within one instrument mind you, the instrument itself typically has 32 elements so that we can resolve part in 32. The system has been designed in two standard probes, one covering 10 to 320 micrometers and a larger probe that covers about 200 to 4500 micrometers. I'm going to pass around some photographs so that Figure 1. Schematic of imaging system designed to size droplets with airborne equipment. you can see what the instrument looks like (fig. 2). These instruments have been mounted and used on 17 different aircraft. I would say that even in the case of aerosol particles if we could use imaging techniques we would. In fact, one of the things we looked at for the Johnson Space Center recently was a possibility of extending Figure 2. Photograph of imaging system instrument used on aircraft. |