History of Impactors—The First 110 Years

Abstract

This history of impactors detailed in this article covers the time span from 1860 to 1970. The origin of the inertial impactor was in the time period from 1860 to 1870, and the very first impactors were essentially the same as they are today—a jet of particle-laden air impinging on a plate. Impactors have a history of going through spurts of development. In the 1800s, impactors allowed researchers for the very first time to quickly collect particles and inspect them under a microscope to study the relationship between dust and disease. This was an exciting time for researchers, and these early impactors were used extensively. The next spurt of development was about 1920, when impactors were rediscovered in the form of konimeters, dust counters, and impingers, which were considerably more advanced than the impactors of the 1800s and were used extensively for determining the dust concentrations to which industrial workers were exposed. These instruments were used in different forms until the mid 1940s when the cascade impactor was developed. The discovery of the cascade impactor, along with extensive theoretical analysis of jet impaction in the early 1950s, led to much activity in cascade impactor development. In 1970, with the help of finite difference techniques and high-speed computers, a parametric study of impactors was made which unlocked nearly all of the secrets of how to design an impactor. From 1970 forward, the design of impactors became rather routine, and numerous impactors of all shapes and sizes for all manner of purposes were designed and built.

Introduction

You are driving along a highway on a nice sunny day and all of a sudden a large bug goes “splat” right on your windshield. You have just witnessed impaction first hand and in its simplest form. As you were driving along, the air was flowing around your car. The bug, was large enough to have sufficient inertia that it slipped across the air streamlines and impacted on the windshield. Small bugs however never seem to hit the windshield; only the large ones hit. However, if you examine a smaller cross-sectional area component of the car, such as the radio antenna or some of the thin members of the grill, you will see smaller bugs impacted.

An even more informative scenario is that you are riding in a car during a snowstorm and watching snowflake streak lines. This was my first experience with impaction. I can remember as a child riding in the front seat of my father's 1942 Nash at night during a snowstorm and watching snowflakes in the headlights of the car. If the snowflakes were light and fluffy, they would head straight for the car and then at the last moment flow over the car. However, if the snowflakes were wet and heavy, the streak lines would show tracks directly into the windshield.

Both of these examples illustrate the basic parameters of particle impaction; that particle collection is somehow related to the particle size (the bug or snowflake size), to a characteristic dimension (the width of the car or the width of the antenna), and to the relative velocity between the particle and the collector (the speed of the car). The above experiences are examples of body impactors, one of the classes of impactors shown in Figure 1.

FIG. 1 Three types of inertial impactors: (a) plate (or jet) impactor, (b) virtual impactor, and (c) body impactor.

Impactors are simple devices, consisting of air flowing around a body or an impaction plate. Particles with sufficient inertia will slip across the air streamlines and impact on the impaction surface. Particles with less inertia will not slip across the streamlines sufficiently to strike the surface and will instead follow the airflow away from the impaction area.

Traditionally the impactor, sometimes called a jet impactor, consists of a nozzle directing air at a flat impaction plate, as shown in Figure 1a. The other impactors shown in Figure 1—the body impactor and the virtual impactor—are special cases. The body impactor, as illustrated in the bug/windshield example, is when the impaction surface is moved through the air and particles impact on the moving surface. In the virtual impactor, a receiving tube replaces the flat impaction plate where particles that normally would impact on the plate now enter the receiving tube. There is a small flow through the receiving tube to sweep the particles out of the tube. However, the air in the tube is nearly stagnant, creating a virtual impaction surface at the tube inlet.

Of the above-mentioned impactor groups, the nozzle/flat plate impactor illustrated in Figure 1a is by far the most common and widely studied and will be the primary focus of this article. Because of the simplicity of the impactor, consisting of a jet of air directed at a flat plate, there have been numerous impactors designed, built, and described in various reports and in the open literature. Many impactors that have been built for special studies, such as particle bounce studies, interstage loss studies, etc., or simply for collecting particles from some source for evaluation, have not been, or are only vaguely, described in the literature. I, for one, have several impactors in my laboratory that I have built for various purposes that have not been reported in the literature.

Many impactors have been made commercially available, as evidenced by the rather long list of impactors in the various editions of the Air Sampling Instruments reference handbook published by ACGIH (ACGIH 1960, 1962, 1967, 1972, 1978, 1983, 1989, 1995, 2001), but only a few of the impactors have been widely distributed. In the early editions of the ACGIH handbook, there were even sections on “Home-Assembled Instruments.” Of the impactors that have been widely used, there are many writings in the literature describing variations and modifications made to these more “standard” impactors. This article will not attempt to describe all of these numerous impactors and variations, or reference all the many researchers that have used and reported on impactors, but instead will describe the more significant developments, starting with a search for the origin of the impactor. However, along the path of the impactor's history, there have been some interesting and unconventional, although maybe not significant, designs that I thought may be of interest, and some of these are included in this aritcle.

Figure 2 shows the progression of impactor development since its origin in 1860. The impactor appears to have undergone spurts of development driven by the need to solve particular problems, especially in the early years. In the period from 1860 to 1880, the impactor allowed researchers, for the first time, to collect particles from the air and inspect and study them under a microscope. This was an exciting time for researchers, and their primary concern appeared to be to investigate the relationship between airborne particles and disease. They were not overly concerned about particle concentration but were primarily interested only in what types of particles were airborne.

FIG. 2 The first 110 years of impactor development (1860–1970).

There was little activity in impactor development for the next 35 years, and then a burst of development of impaction instruments for studying industrial aerosols, especially in mines. Here they became more interested in the quantity of airborne particles. The developers of these instruments appear to have been unaware of the 1800s work, and their instruments replaced the rather crude instruments that collected particles in filters and wash bottles of various forms which were used in the time prior to 1915. The developers of the instruments around 1920 describe these earlier collection devices that their impactors replaced.

There was development work in the period from 1920 to 1945, but this work was primarily refinement of three instruments developed around 1920. The next spurt of development was in 1945 with the May cascade impactor, which provided information on particle size distribution as well as concentration. This work was related to military studies of chemical dispersion on the battlefield.

After the development of the cascade impactor in 1945, there were many impactors developed; many of them were variations of the original May cascade impactor. Also in this period, serious work began on understanding fluid and particle motion in impactors through theoretical analysis.

In 1970, the application of high-speed computers and finite difference methods enabled a thorough understanding of the flow field and particle trajectories within impactors. The theoretical work of 1970 showed that impactors are able to provide very sharp particle size classifications, with predicable aerodynamic cut sizes, if specific guidelines are followed. Furthermore, the work showed from first principles the reasons that impactors have sharp cutoff characteristics. With this understanding, the impactor became considered by many to be as close to a standard instrument for particle aerodynamic size analysis as we will get in the field of aerosol particle size classification. Thus, in 1970 the design of impactors entered a new era, and from 1970 forward the design of impactors was simply the systematic application of guidelines that resulted from the theoretical work. If the design follows these guidelines, the impactor is known a priori to provide a sharp classification at the correct particle size. For this reason, this article follows the impactor development from its origin in 1860 to 1970, which I consider to be the period of time encompassing the fascinating development of this basic and important instrument.

Origin of Impactor

The search for the origin of the impactor has been most interesting. I initially reported (Marple 1995) that the impactor was first developed in England in 1870 by Maddox (Maddox 1870, 1871), who gave a detailed description of his sampling device, the aëroconiscope, which included a nozzle/impaction plate stage. However, recent investigation led me to a French report of the April 16, 1860 Sessions of the Academy of Sciences, where M. F. Pouchet described a sampling instrument, the Aéroscope, that also included an impaction stage (Pouchet 1860). Since this was an oral presentation, there were no sketches or figures of his device in the paper. However, the description given leaves little doubt that an impactor nozzle and impaction plate was used. Furthermore, P. Miquel in 1879 describes a particle sampler based on Pouchet's device and indicates that he believes that Maddox's device is a modification of Pouchet's sampler (Miquel 1879). Maddox makes no mention of Pouchet's sampler in his 1870 paper, and it is possible the impactors were developed independently. Nevertheless, I now believe that the origin of the impactor should be credited to M. F. Pouchet in 1860.

1860 to 1899

Pouchet's Aéroscope

In a presentation to the Academy of Sciences in Paris on April 16, 1860, Pouchet stated in his opening paragraph:

With the help of a very simple instrument, I have come to be able to concentrate on an infinitely small surface all of the solid and normally invisible corpuscles that float in the air, in such a way as to allow a strict appreciation of their nature and to enumerate them. When we choose to do so, we concentrate on a glass slide and in a space of two square millimeters all of the solid corpuscles that are found disseminated in one cubic meter of atmosphere or even much more. (Pouchet 1860, p. 748)

Although Pouchet provided no sketch of his aéroscope, we can get an idea of the instrument from his description:

Here is how the instrument of which we made use for the concentration of the atmospheric corpuscles is constructed. It is made up of a crystal tube hermetically sealed at its two ends by copper sleeves. The upper sleeve, which is fixed, receives a copper tube, that comes to an end on the outside with a very small funnel, and on the inside by a very finely stretched extremity whose opening is no more than 0.5 mm in diameter. Through the lower sleeve is introduced into the device a flat, circular piece of glass, which is placed 1-millimeter from the tapered end of the tube. The device is closed and its interior is connected with a pump by means of a tube that goes through the lower sleeve. (Pouchet 1860, pp. 748–749)

This paragraph is somewhat confusing (possibly my translation), but his next paragraph is more clear:

When the pump starts, the surrounding air is blown into the tube and leaves by its tapered end, and comes to strike the plate of glass and deposits on its surface all of the atmospheric corpuscles which it contains, absolutely in the same way that Marsh's device spreads on a porcelain plate the metal particles that emerge from it. The most voluminous corpuscles are gathered together in a little central mound, which is no more than a millimeter in diameter; and the others spread out a little further.

By carefully removing the glass plate which has received the jet of air, and by examining it under a microscope, one thus finds there concentrated, on an infinitely small surface, the entire group of corpuscles that swam invisibly in a proportionately immense space of atmosphere, and one perfectly determinable thanks to the capacity of the pump which is itself strictly known. (Pouchet 1860, p. 749)

This passage is interesting in that there may have been an earlier impaction device used by Marsh to deposit metal particles on a porcelain plate. No reference was given and a search for Marsh publications at about this time (1860) yielded no results. It appears that the work by Marsh was not relative to an aerosol sampling instrument and I believe that the first use of an impactor as an instrument was by Pouchet. At the very least, some work that Marsh had done had inspired the impactor developed and used by Pouchet. Also, this passage indicates that the aéroscope was used to determine the particle concentration values because the capacity of the pump was known.

In the next two paragraphs, Pouchet gives an indication that he had a good understanding of impactors. He described the use of a sticky coating on the impaction plate to prevent particle bounce, and then describes a modification that makes his aéroscope the first multiple nozzle impactor for the purpose of spreading out the deposit for easier examination:

To bring even greater precision to the device to make sure that no corpuscle escape, not even the slightest or lightest of them, one can smear the surface of the piece of glass with an adhesive substance. As a result all of these [finest corpuscles] come to stick to the surface at the very spot where the current applies them.

One can also, if preferred, disseminate the corpuscles on the sheet of glass by closing the tube not with a single tiny hole but with a little flat diaphragm, like the head of a watering can. (Pouchet 1860, p. 749)

Pouchet then in his final statement gives an indication of how he intends to use his aéroscope: “I propose, with the help of the instrument of which this Note has been concerned itself, to concern myself with the air of hospitals, swamplands and mountains. It will be my honor to inform the academy of my experiments if, as I hope, they produce results worthy of knowing” (Pouchet 1860).

In a latter document, Pouchet (1870) describes using the aéroscope for inhalation studies: “In all of our aéroscope experiments, we have always collected corpuscles from the air at the height where humans breathe, such that one could say that all bodies that we identify in the atmosphere hurry, by incalculable numbers, into our lungs, at each moment of inhaling, and go back out in part by the opposite movement” (Pouchet 1860).

Maddox's Aëroconiscope

Although the origin of the impactor may not be the apparatus developed by R. L. Maddox in 1870, he was a pioneer in impactor use and development, and his publication gives the first rather detailed sketches (Maddox 1870, 1871). In addition, Maddox's sampler was apparently the inspiration for the early design of the May impactor 70 years later, a very important and widely used cascade impactor.

Maddox's aëroconiscope (Figure 3) not only used a nozzle/ impaction plate arrangement of a conventional impactor, but also was a very versatile particulate air-sampling instrument. In one configuration the impactor is imbedded in a wind vane mounted on a pivot of a tripod. As the wind blows, “cone a” is turned into the wind and “cone b” is facing downwind. This creates a flow through the impactor nozzle with particles being collected on a glass slide cover at “c.” In another configuration, the impactor is mounted vertically at “f.” The heat from an oil lamp creates, by a chimney effect, a flow through the impactor nozzle. Maddox also indicates that flow could be obtained by the use of a water aspirator.

FIG. 3 R. L. Maddox's Aëroconiscope (Maddox 1870).

The circumstances driving the development of this device was an attempt “To establish fully any relationship between ‘dust ad disease’ …” (Maddox 1870, p. 286). However, it appears that the dust particles that he was interested in were germs and spores. The reason for the vertical configuration was so the sampler could be used “over a cesspool in any nook or corner, in an ordinary room, in a cow-shed or stable, or near a patient suffering from any infectious disease” (Maddox 1870, p. 288).

Although there have been many impactors designed, built, evaluated, calibrated, and used, and the fluid mechanics of impactors have been studied in great detail, the basic conventional impactor has remained unchanged for 140 years. Maddox recognized even the ease of use of the impactor:

It is not pretended that this form is the only useful one or the most convenient that can be adopted, but as it has now been in use some days, I find it answer its chief purpose very well, and is exceedingly easy to manipulate. The advantages claimed are, ready application at any spot, the collection of the atmospheric particles into a small space in such a manner as to be at once microscopically examined with a 1/16th or 1/20th objective, placed on a growing slide, or mounted permanently. (Maddox 1870, pp. 289–290)

It is humbling to realize that the attributes of the inertial impactor were so well understood by researchers at its very inception 130 to 140 years ago. Many of us today believe that the inertial impactor is as close to a standard as we have for determining the aerodynamic diameter size distributions of aerosol particles. Possibly it is the fact that the basic conventional impactor is, and always has been, so simple and its attributes so well understood, that it qualifies as a standard. The robustness of this early impactor is evidenced by the fact that in one reported study (Maddox 1871) the aëroconiscope was used for 155 days, sampling from 8 to 11 hours per day.

Maddox ends his paper (Maddox 1870, p. 290) with a challenge of giving the apparatus a fair trial or find something better in his final statement: “I believe it will be only by constant, varied, and multiplied research, we shall ever obtain any answer to the important question of ‘dust and disease;’ hence my excuse for trespassing on the pages of this Journal, in the hope others may be induced to give the apparatus a fair trial or suggest something more useful.”

Maddox can rest assured that the impactor has been given a fair trial as evidenced by the large variety of impactors that have been designed and used in countless studies. The remainder of this study will describe some of the numerous impactors that make the historical record of the impactor.

Miquel's Aéroscope

P. Miquel (1879) reported on an impactor sampler that combined the features of both the Pouchet and Maddox instruments. Miquel provided some insight into the early impactor development and also indicated that “so called aéroscopic methods” are replacing previous methods which included dust fall samplers, analyzing pollutants in snow, frost, rain, and dew, or use of cotton tufts that are then dissolved in a mixture of alcohol and ether:

To the methods of experimentation which came before have been advantageously substituted the so called aéroscopic methods. The primitive instrument of which Pouchet made use consisted of a cylindrical crystal tube fitted on one of its extremities with a tubing that drew in air and on the other with a diaphragm pierced with one or several holes, immediately following which was fitted a plate of glass either coated or not with a sticky substance. Dr. Maddox modified this device and gives it the complicated name “aëroconiscope.” He fitted it with a weathercock and a conical diaphragm, to allow it to be ready to work by the force of the wind; the air intake was left out. This instrument as a result lost much of its precision. It was no longer possible to calculate the quantity of air passed during each experiment, and the statement of the number of germs collected had only a vague meaning. (Miquel 1879, p. 36)

Miquel had two versions of his impactor, as shown in Figure 4. One version had a cone with a nozzle at the apex. An impaction plate was suspended in a carriage that could be adjusted up and down relative to the nozzle (an adjustable jet-to-plate distance). After the air passed through the impactor, it flowed into a counter receiver that measured the flow exactly.

FIG. 4 Miquel's Aéroscope: (a) Aéroscope with aspiration and (b) Aéroscope with wind vane (Miquel 1879).

The second design (Figure 4) employed the same impaction stage but used wind power to pull air through the impactor, similar to Maddox An anemometer placed next to the device provided an estimate of the volume of flow sampled.

In 1888 John Aitken reported on an apparatus where water was condensed on atmospheric particles and the droplets allowed to settle on a polished silver plate to be counted (Aitken 1888). Aitken described an improved apparatus in great detail the next year (Aitken 1889), which was used to measure the particle concentration in a wide variety of settings of atmospheric conditions and locations. Although this was not an inertial device, Owens would use the condensation of water on droplets some 30 years later in the Owens dust counter, which was an impaction device.

A final device developed in the late 1800s was an impactor stage incorporated in an experimental apparatus for determining the efficiency of a respirator for dust particles (Michaelis 1890). This device, shown in Figure 5, appears to be a breathing simulator. The dust was aerosolized into a bell jar (T) and then passed through the respirator (R). Any dust particles penetrating the respirator were collected by impacting the particles onto the surface of a liquid (L). The liquid was undoubtedly used to eliminate particle bounce and re-entrainment problems.

FIG. 5 Apparatus for testing dust penetration through a respirator (R) which includes an impaction stage (L) for impacting dust particles into a liquid (Michaelis 1890).

1900 to 1939

In the early 1900s, the impactor was being discovered as an instrument that could be applied as a desired solution for a variety of particle sampling applications in industrial atmospheres. It almost appeared to be a period when the impactor was being reinvented. Three important instruments that evolved in this period were the Kotzé konimeter, The Owens jet dust counter, and the Greenburg-Smith impinger. The importance of these three devices is probably said best by Green and Watson (1935, p. 9) in a publication looking specifically at instruments to study the disease of silicosis:

Since 1925, the number of types of instruments in use in various parts of the world has been reduced and reliance has been placed on a few well-tried designs. Of these, the Greensburg-Smith impinger, konimeters of various kinds, and the Owens jet dust counter and its modifications are the best known, and, indeed the only ones that are being used to any great extent in the field.

Essentially the same thing was said by Rowley and Jordan (1942, p. 3) in a report describing a comparative study of dust counters, except that they identified them by their method of particle collection:

Separation of dust from the air by impingement has probably been more widely used than any other method for dust counters. In general the method has been applied in three different ways:

1. By impingement of the sample against a surface which is submerged in water or other liquid. [impingers]

2. By condensation of moisture on the dust particles and subsequent impingement against a dry plate. [Owens-type dust counters]

3. By impingement of the sample against a slide which is covered with a film of oil or other viscous material. [konimeters]

In the following sections we will examine the development of these three general, and important, methods for particle collection that comprise the history of the impactor in the early 1900s. In this discussion, you will note that the Kotzé konimeter was the first of these three instruments developed and was widely used. The Owens jet dust counter was developed next to correct perceived shortcomings of the Kotzé konimeter. The Greenburg-Smith impinger was a combination konimeter and a popular nonimpactor water spray dust collector.

Impingement on a Coated Plate (Konimeter)

It appears that two konimeters, Hill and Kotzé, were developed independently in the 1916–1917 time period. The Hill and Kotzé units were developed for testing air scrubbers and dust in mining atmospheres, respectively.

In 1917, E. Vernon Hill, Chief of the Division of Ventilation for the Chicago Department of Health, reported on the development of the Hill dust counter (Hill 1917). Hill was investigating methods for determining dust concentrations in studies of dust

removal efficiencies of air-cleaning devices, such as air washers in ventilation systems. Although the origin of the impactor was in the 1860s to 1870s, he makes no mention of the earlier work in his paper. However, Hill does describe methods that were being used to determine particle removal efficiencies, and it is interesting to look at the methods that were available at that time. He puts these methods in five categories: visual, weighing, condensation, diffraction, and counting.

The visual method consisted of putting identical particle collection devices upstream and downstream of the air washer. In one case this consisted of cheesecloth stretched over one-square-foot frames, and in another case porcelain dishes were coated with white vaseline. In both cases “the degree of blackening or discoloration roughly approximated as a percentage ratio” (Hill 1917, p. 23). The weighing methods consisted of various collection devices upstream and downstream of the washer. The devices included filter bags made of fine mesh muslin, cotton filters that were sometimes covered with sand and sugar, and Drechsel wash bottles where air is bubbled through water, the water evaporated, and the dust weighed. The condensation method consisted of two instruments developed by John Aitken. The first instrument, a koniscope, consisted of a horizontal tube lined with a hygroscopic material. Dust was drawn into the tube and the pressure decreased, which caused water to condense on the dust particles causing a fog “the density of which by comparison with suitable charts determining the extent of air dustiness” (Hill 1917, p. 23). The second device, the Aitken dust counter, operated by the same principle as the first instrument, but instead of observing the fog, the droplets settled onto a plate where they were counted. Counting methods consisted of filtering out dust particles in sugar filters, dissolving the sugar in distilled water, and counting the particles with the Sedgwick-Rafter counting cell, or else collecting the particles in Drechsel wash bottles and counting the particles with the Sedgwick-Rafter counting cell. In another counting method, dust particles were blown through an anemometer toward a photographic plate where they were counted after development.

After testing, and rejecting, the above dust samplers, Hill developed his own dust counter, consisting of an air pump and an impaction stage. There were two designs. In the first design, “The writer's first experiments in this line were made by drawing a quantity of air into a syringe and forcing it out against a slide covered with adhesive material and counting the number of particles with a microscope” (Hill 1917, pp. 29–30). This, of course, was an impactor in its simplest form. Hill found, however, that particle loss inside the syringe was unacceptably large. Thus he changed his design, as shown in Figure 6, by placing a microscope cover slip inside the syringe, so that particles would be impacted on the cover slip as air was drawn into the syringe. The pump capacity was 4 cu. in. per stroke, and he found that 5 to 10 strokes were necessary in an office or school room, but 15 to 20 strokes were required for outside air to collect enough particles for counting.

FIG. 6 Hill dust counter, original version of impaction plate inside syringe (Hill 1917).

The Hill dust counter was evaluated for studying the dustiness of air in granite-cutting plants by Katz and Trostel (1922), and they describe the details of a slightly later version (Figure 7a) that was actually the Hill dust counter routinely used. This version has a capsule, which holds the impaction plate, attached to the inlet of the piston/cylinder syringe. Several capsules could be prepared in advance and carried in a case for ready use. Figure 7b shows six such capsules in a carrying case, along with the piston/cylinder hand pump.

FIG. 7 Hill dust counter. (a) Schematic drawing showing removable capsule attached to hand pump (Katz and Trostel 1922), (b) Hill dust counter in carrying case for hand pump and six capsules (Hill 1917), and (c) Hill dust counter being calibrated with six capsules in series (Hill 1917).

Hill realized that the impaction stage did not collect 100% of the particles, so he devised a method for calibrating the stage by drawing the air through six identical capsules (stages) in series, as shown in Figure 7c. By counting the particles on each cover slip and taking a ratio of the particles counted, he could calculate a collection efficiency. He found that for ordinary tests this efficiency was 62 %. Hill was very satisfied with this instrument:

During the past winter we have made several hundred tests with this instrument and have found the results, without exception, consistent and satisfactory. With very simple instructions any ordinarily intelligent person can collect samples and make the count in a satisfactory manner. Not only is it possible with the instrument to accurately determine the number of dust particles, but a study of the kind and character of the dust under observation is often instructive and valuable. This feature greatly enhances the value of the method and makes the work extremely interesting. (Hill 1917, pp. 31–32)

After the crudeness of some methods and the complexity of others, the development of the Hill dust counter (an impactor) must have been very exciting and the results very enlightening. The Hill dust counter impactor was indeed elegant in its simplicity. Even though the Hill dust counter was a breakthrough in instrumentation, I found little reference to it, except in the review paper by Katz and Trostel (1922), Heymann (1931), and Rowley and Jordan (1942).

In 1916, about the same time that Hill was working on his dust counter, Sir R. N. Kotzé was developing the Kotzé konimeter for use in South African mines (Kotzé 1919). This instrument operated on the same principle as the Hill dust counter, in that particles were collected on a coated plate, but appeared to be much more popular and widely used. The primary difference between the Hill and Kotzé instruments is that in the Hill dust counter air was drawn through the nozzle by pulling on the piston of a hand pump. In the Kotzé konimeter, the piston was driven by a spring. Between each sample, the piston must be cocked by hand and was held in position by a pin. A trigger then released the piston.

The Kotzé konimeter was simple to operate and evolved into two forms. In the original form (Figure 8a) the impaction plate was a microscope slide. In the other form (Figure 8b) the konimeter had a unique feature of a circular disk impaction plate, which could be rotated so that up to 29 dust deposits could be collected on the periphery of the disk. This allowed multiple samples to be taken without changing the impaction plate. Another unique feature was a built-in plunger in the handle that could be cocked against a spring and, when released by a trigger on the handle, would pull about 10 cc of air through the impactor. Multiple plunger pulls could be used to collect larger samples. Even in this latter form, the konimeter was compact, being only 6.5 in long and weighing 2.2 lbs. Mavrogordato (1923) reported on an extensive evaluation of the konimeter and even proposed a 50 sample circular konimeter.

FIG. 8 Kotzé konimeter: (a) original Kotzé konimeter (Heymann 1931) and (b) Circular konimeter with 29 spot impaction plate (Katz and Trostel 1922).

Konimeters were later fitted with microscopes (sometimes these versions were called koniscopes) so that the deposits could be inspected immediately after being collected. Two commercial versions were the Sartorius and the Zeiss konimeters. The Sartorius konimeter, shown in Figure 9a, could collect 36 samples on one disk. The deposit was then examined immediately by rotating the disk so that the deposit was at a position under the microscope (Kusnetz 1962). The Zeiss konimeter is shown in Figure 9b (Heymann 1931). This figure is interesting in that it shows the assembled unit (a), the condensor for the microscope (b), a cut-away view of the piston/spring/trigger assembly and the air flow path from the inlet into the piston suction chamber (c), and a cutaway view of the microscope (d). A special version of the Zeiss konimeter was the Lehmann-Lowe konimeter, which used a coarse filter at the inlet to remove large innocuous silica particles from the air stream and allow the smaller particles known to cause silicosis to be collected in the konimeter stage (Iron & Coal Trades Review 1934). This unit also used a DC electric motor-powered pump to pull a steady flow of air through the sampler instead of the hand pump.

FIG. 9 Commercial konimeters: (a) Sartorius konimeter [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments, First Edition, Copyright 1960. Reprinted with permission] and (b) Zeiss conimeter (Heymann 1931).

In 1923, Kotzé took note of the development of the impinger by Greenburg and Smith of the American Bureau of Mines. He combined the impingement of the dust into a liquid and his hand pump in what he called a hydro-konimeter, shown in Figure 10a (Kotzé 1923). The impinger nozzle is shown in Figure 10b. Note that there is an impaction plate directly below the nozzle. This impaction plate was inserted just below the surface of the water so that the wet impaction plate would be the same as a sticky-coated plate. The air stream dragged water off the impaction plate, carrying the dust particles with the water. He stated, “If the nozzle is merely inserted in water, the air balloons out in the water and the impinging effect is lost” (Kotzé 1923, p. 2). Kotzé believed that counting the number of particles was a better measure of potential health hazard than weight, so he devised a method for counting the number of particles in the liquid as follows:

A cell is made by placing a square of brass accurately ground flat to a thickness of 1m.m. or 2 m.m. with a central hole (3/4 inch in diameter), on a glass microscope slide. After the thinnest possible film of Vaseline, a drop of the turbid liquid is placed in the cavity and covered with a thick cover-glass. The particles immediately begin to settle on the bottom and can be counted. (Kotzé 1923, p. 4)

FIG. 10 Kotzé Hydro-Konimeter (Kotzé 1923) [Reprinted with permission]: (a) schematic diagram and (b) nozzle and impaction (baffle) plate.

The konimeter played an important role in sampling of dust in South African mines up to at least the early 1950s as is evidenced by a statement made by Beadle (1951, p. 265):

Dust sampling plays an important part in the control of dust in Witwatersrand gold mines, and has been practiced for over forty years. Many types of dust sampling instruments have been tested for their suitability under local conditions, but for the routine dust sampling work carried out by the staffs of the mines, and the Inspectors of the Department of Mines of the Union Government, only two main instruments have been used—the sugar tube from 1911 to 1938, when it was finally abandoned, and the konimeter from 1916 onwards.

Beadle (1951) did an extensive study of the konimeter and concluded that errors were caused by what he called “snap-sampling” (he stated that the sample time of a konimeter was about 1/4 sec.) and low collection efficiencies when compared to a thermal precipitator. He suggested that a photoelectric counter, which he was currently developing, would eliminate the human factor and provide more accuracy.

Condensation of Moisture on Particles (Owens Jet Dust Counter Apparatus)

Another interesting development in this period was the Owens' jet dust counter. Owens was Superintendent of the Air Ministry Meteorological Office, Atmospheric Polution Division, in London. He was investigating problems of much wider disciplines than mine-related dusts:

The object which the author had in view in making this investigation was to provide some simple and effective method for examining the quantity and nature of suspended dust, and its relation to visibility, fumes from industrial processes, dust produced in mines, fogs such as are experienced in our larger cities, and generally in the hope of throwing some light on problems which are concerned with the presence of fine suspended impurities in the air. (Owens 1922, p. 18)

Owens also reviewed other methods used at this time, including (1) filtering air through white filter paper, with the resulting mark being an indication of the dust present; (2) the Aitken dust counter, where moisture was condensed on the particles and they settled to the floor of a chamber where they were counted; (3) filtration through soluble filters, such as collodion wool which could be dissolved in ether or sugar which could be dissolved in water; (4) collecting particles on filters and weighing; (5) the Palmer apparatus, where particles are collected in a water spray and the particles either counted or weighed; or (6) exposing plates or dishes to the air and letting particles settle on their surfaces. He was not satisfied with any of these devices for a variety of reasons and proceeded to develop his own instrument, the Owens jet dust counter.

Owens' earliest form of this device, called a “jet apparatus,” (p. 20) consisted of a very small circular jet of air impinging on the surface of a microscope cover-glass. The surface above the cover-glass was made of a sheet of mica so that the particle impingement on the cover-glass could be observed in real time with a microscope. In this apparatus, the flow was upwards so that the impaction of the particles could be studied with a microscope while in operation.

When 0.15 to 0.18 cc of air was pulled through the circular nozzle jet apparatus, about 30 particles would be collected on the cover-glass. This was about the correct particle density for microscope counting, but he felt that it was not enough air to provide a representative sample of the particles. When a larger quantity of air (2 liters) was drawn through the apparatus, a cone of particles would be found on the cover-slip. Owens concluded that the collection characteristics on the apex and sides of the cone would not be the same as a flat plate and the test not satisfactory. In addition, the deposit was too dense for counting.

To solve the above problems, Owens built a second apparatus that used a rectangular slit (Figure 11a) instead of a round nozzle so as to spread the deposit out over a larger area. He could then increase the air flow from 0.15 cc to 50 cc and still have the same density of particles on the plate for counting. This appears to be the first use of a rectangular nozzle.

FIG. 11 Owens jet dust counter (Owens 1922): (a) single cell, (b) double cell, and (c) dust counter with cylindrical chamber containing moistened blotter paper at inlet.

The next development of the Owens' jet apparatus was the “double cell” unit (p. 22), shown in Figure 11b. Here two rectangular slits of identical dimensions (B and A) were used in two stages in series. It is of interest to note that the impaction plate of the first stage contained the nozzle of the second stage, a design feature used today to make compact, low internal loss impactors. The reason Owens developed this two-stage apparatus was for the purpose of determining the total number of particles in the air. Owens realized that the collection of the particles below the slit was not 100%. However, he reasoned that the efficiency of the two stages should be the same. Therefore, if C_B and C_A are the number of particles counted under slits B and A, respectively, then

where T is the total number of particles drawn through the apparatus. The value of T could then be calculated from:

(1)

Owens calculated from the data of six runs that the collection efficiency of a single stage was in the range of 53% to 83%. In further work, Owens discovered the following:

A curious phenomenon was noticed in the test for efficiency above described. Occasionally the second cell collected no particles whatever, that is all were trapped in the first; while at other times about 30 per cent of the particles penetrated to the second. A test on October 26, during a smoke fog in London, gave 100 per cent efficiency, i.e., all the particles were trapped in the first cell. There appears, therefore, to be some variable condition which affects the adhesion of the particles, and from subsequent experiments this condition appears to be the humidity of the air; an attachment was therefore made by means of which the air was made to approach the jet through a damp-walled chamber in order to raise the humidity and keep the air as nearly saturated as possible. (Owens 1922, p. 24)

This led to the development of the apparatus shown in Figure 11c. The air passed through a cylindrical chamber before entering the cell. A moistened blotter paper lined the inside surface of the cylinder, so that the humidity was somewhat controlled. He found that this created conditions favorable for particle adhesion to the impaction plate. He concluded that the fall of pressure in the jet resulted in moisture being condensed around the particles at the moment before impacting on the plate. He noted, “This is the same effect as is produced in the Aitken dust-counter, but the quantity of moisture so condenses is extremely small and, when the cover-glass is removed from the apparatus for counting, it evaporates rapidly; thus there need be no confusion over the counting of water drops as dust particles” (Owens 1922).

In a subsequent paper, Owens (1923) gave more details on the Owens dust counter and also indicated that he investigated impaction methods with sticky coatings on impaction plates. He noted that “if the velocity is high enough to make the jet efficient, the sticky material is blown away, while on the other hand, if the velocity is low enough to preserve the sticky film intact, the efficiency to so reduced as to make the method useless” (Owens 1923). He then described a comparative study between his dust counter and “a well-known instrument in which a jet, circular in section, is caused to impinge upon a glass surface coated with Vaseline” (Owens 1923), which was probably a konimeter. He concluded that the “jet and vaseline” (Owens 1923) instrument had far less collection efficiency, as low as 4 or 5% of the results from the Owens dust counter.

Two commercial versions of the jet dust counters are shown in Figure 12. The Bausch and Lomb dust counter (Figure 12a) shows a bulb to pull air into the humidification chamber, where the dust was allowed to stay for a short period of time to equilibrate, and a hand pump to pull the air through the jet for impaction. Also, there was a built-in darkfield microscope for the examination of the deposits (Gurney et al. 1938; Rowley and Jordon 1942; Kusnetz 1962).

FIG. 12 Commercial jet dust counters (Kusnetz 1962) [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments, Second Edition, Copyright 1962. Reprinted with permission.]: (a) Bausch & Lomb dust counter and (b) Casella jet dust counter.

The Casella jet dust counter (Figure 12b) shows the high humidity chamber inlet to the right and the pump to the left. In this unit, the cock was set so that the pump plunger could pull air into the damping chamber. Then the cock was turned so that the saturated air was pulled through the nozzle and the particles impacted on the plate. The plate could be turned to collect up to six deposits on one impaction plate (Kusnetz 1962).

Impingers (Greenburg-Smith)

The third of these three important instruments, introduced at about the same time as the Kotzé konimeter and the Owens jet dust counter, was the Greenburg-Smith impinger (Greenburg and Smith 1922). The U.S. Public Health Service (Greenburg) and the Bureau of Mines (Smith) developed this instrument because they were “cognizant of the importance of dust in the causation of pulmonary disease” (Greenburg and Smith 1922, p. 1). They too described the instruments that were being used for dust collection at the time. They were the sugar tube sampler, the Palmer water-spray apparatus, and the Kotzé konimeter.

The sugar tube and Palmer water-spray apparatus were not impaction devices but were granular bed filtration and scrubbing devices, respectively. The sugar tube consisted of a 2 3/8 in diameter by 5 1/2 in long tube with a 1 1/2 in deep (100 g) bed of granulated sugar. As air was pulled through the tube at a rate of about 1 cfm, particles were filtered out in the granulated sugar bed. The particles could be recovered by dissolving the sugar, leaving the particles. The Palmer water-spray apparatus consisted of a pear-shaped glass bulb with a water trap at the lower end containing about 40 cc of water. Air was drawn through the water trap, creating a spray and washing the particles from the air.

Katz et al. (1925, p. 41) stated that it occurred to Greenburg and Smith “that it might be possible to combine principles of collecting dust by impingement with a water-washing or bubbling method, and so to make an apparatus possessing the advantages of both principles.” This apparatus was thus a combination of the impactor section of the Kotzé konimeter and the Palmer water-spray apparatus.

The first model of the impinger that occurred to Greenburg and Smith as reported by Katz et al. (1925, pp. 41–42) “consisted of four glass tubes about one-fourth inch in diameter drawn to small nozzles at the lower end, and inserted through holes in a rubber stopper which fitted a 500 cc conical form beaker containing about 200 cc of water.” These tubes extended to within 0.5 mm of the bottom of the flask. A fifth tube inserted through the rubber stopper lead to a suction source (Katz et al. 1925).

This rather crude impinger was soon replaced with an improved design shown in Figure 13a, consisting of a 5/8 in diameter tube 9 1/2 in long inserted into a 500 cc flat-bottom flask or precipitating jar containing 300 cc of distilled water. At the lower end of the tube were 15 holes, each 0.6 mm in diameter, and on the exit side of this nozzle plate were three 0.8 mm high lugs that would keep the 15 nozzles 0.8 mm above the bottom of the flask (Greenburg and Smith 1922). Katz et al. (1925) tested a very similar impinger, shown in Figure 14a, that had a smaller inner tube and less water.

FIG. 13 Greenburg-Smith impingers (Katz et al. 1925): (a) impaction on bottom of flask and (b) impaction on suspended bronze plate.

FIG. 14 Impinger for personal dust sampling (Greenburg and Bloomfield 1932): (a) all-glass impinger and (b) impinger in leather holster with inlet at inhalation height.

Hatch et al. (1932, p. 301) indicated that this was a very important instrument, stating that:

Since that time (when the impinger was introduced) this instrument has almost entirely replaced other sampling devices in field investigation of the industrial dust hazard in the United States. The impinger possesses certain advantages which in no other method of dust sampling are combined in a single instrument. These advantages may be summarized briefly as follows:

1. High dust collecting efficiency.

2. No limitation with respect to dust concentration.

3. Simple construction and cheapness.

4. Quantitation of the sample by chemical, gravimetric, or microscopic means.

Because of its importance, the impinger was developed into a number of different configurations. The first change was to use only one nozzle of a larger diameter and to control the distance from the nozzle to the impaction plate by suspending a bronze plate from the end of the nozzle as shown in Figure 13b (Katz et al. 1925). The next change was to remove any metal from the impinger, since acid was sometimes used as the collection fluid, and replace the metal with three glass rods fused to the tube and to a glass impaction plate.

To enhance further the usefulness of the impinger, the glass flask was replaced with a glass cylinder, 30 cm long by 5 cm in diameter, as shown in Figure 14a (Greenburg and Bloomfield 1932). This impinger could then be placed in a cylindrical leather holster (“for protection from the impact of large pieces of flying material”) and strapped to a workers chest to serve as a personal dust sampler (Figure 14b). Note that a baffle was installed at the exhaust tube to prevent any carryover of liquid caused by the higher velocities in the small diameter tube. This personal sampling technique of attaching an impinger to the chest was first used by Badham et al. (1927) with an impinger tube 12 in long by 2 in in diameter, but without the leather holster.

An even more portable impinger, the midget impinger, was developed in 1932 (Hatch et al. 1932) and used as a standard sampling instrument in the United States and other countries for many years. The midget impinger body, as shown in Figure 15, was a flat-bottomed flask 5 cm in diameter by 21 cm long. The bottom of the flask was used as the impaction surface, and a mark 5 mm from the bottom of the flask indicated the correct depth to which the nozzle was to be inserted. Only 75 cc of water was required in this impinger. Note that the inlet and outlet tubes were now combined in one compact fused assembly and that a rubber ring was used at the upper end of the flask to prevent entrainment of water in the exhaust air.

FIG. 15 Original midget impinger (Hatch et al. 1932).

1940 to 1949

Milestones in impactor development historically were driven by need. Pouchet's aéroscope and Maddox's aëroconiscope were developed to study disease-causing airborne particles. The developments of the Hill dust counter, the Kotzé konimeter, the Owens' jet dust counter, and the Greenburg-Smith impinger in the early 1920s were mainly used to study respirable diseases caused by industrial activity.

One of these devices, the Hill dust counter, was further developed in the 1940s by Rowley and Jordan (1942) into the Minnesota dust counter. This instrument, shown in Figure 16, had some interesting features that one might expect to be added to a basic dust counter as the instrument begins to mature. It was designed to spread out the deposit on the impaction plate for easy counting and had the capability to take multiple samples on one impaction plate. In the Minnesota dust counter, a ¹ < eqid3 > ₂ in × 2-in rectangular impaction plate was held in a moving slide carriage. The operation was described as follows:

As the pump handle is pulled back drawing air through the impingement orifice, the linkage between the slide carrier and the pump handle slowly moves the impingement slide past the orifice. The rate of this movement may be varied between 0.3 and 4.2 inches per second by changing the setting of the lever…. By varying the relative positions of the slide and the orifice, as many as five or six samples may be obtained on a single oiled slide. (Rowley and Jordan, 1942 pp. 7–10)

FIG. 16 Minnesota dust counter (Rowley and Jordon 1942) [Reprinted with permission]: (a) detailed drawing and (b) photograph.

There was also a need in the 1940s that led to the development of the first widely used cascade impactor. This need was described by May in his 1977 paper on the history of the cascade impactor (May 1977, 1982, p. 37):

Among other jobs I soon became involved in the assessment of the efficacy of droplet clouds produced by exploding charges of various unpleasant liquids contained in high explosive shells.

The test procedure was astonishingly crude. The assessment team huddled in a semi-dugout shelter on the firing range with the artillery established some 4 miles away. Shells were aimed at a point about 100 yards upwind of the shelter. After a shell burst, sampling men in gas masks plus protective clothing had to rush out and chase the fast disappearing mist with the object of standing in it to hold up glass “quarter plates” facing the wind, so catching some of the droplets. Sometimes, depending on the nature of the dispersed agent, unprotected “volunteers” would stand in the mist to test its incapacitating effects. The droplets on the plates were counted and sized in the laboratory. All of this was obviously highly ineffective, also the smug gunners seemed to have an exaggerated idea of the accuracy of their weapons. A six-inch H.E. shell landing 20 yards behind one instead of 100 yards in front inspired little confidence among the gallant scientists! Amazingly, as far as I know, no one was actually injured in these trials but a shell splinter burying itself in wood just by my ear encouraged thoughts of better methods.

The “better methods” that May devised led to the development of the May cascade impactor, the first true cascade impactor (May 1945). However, there were several steps taken between running out on the firing range with a glass plate and the development of the May cascade impactor.

The origin of the development of the May cascade impactor appears to have been the Maddox aëroconiscope: “Early in 1941, H. H. Watson suggested to me that some sort of funnel might be fitted in front of the glass plate to direct a jet of wind-powered air onto it and so increase the collection efficiency rather in the manner of the ancient (and probably useless) ‘aëroconiscope’ (Maddox, 1871).” (May 1982, p. 37)

The first step was the “swinging impactor” shown in Figure 17. A wind vane directed the inlet into the wind and airflow through the nozzle was wind powered. The cut size was about 30 μ m for wind speeds greater than 8 mph. This device was developed in 1941 and May stated:

This device was easily made and was soon deployed in large numbers in the field. It was suspended from a wooden gantry by a string attached to the point of balance so that the tail fin directed the air intake slot (3 ¹ < eqid4 > ₂ × ³ < eqid5 > ₄) into the wind. Arcs of these gantries were easily set up downwind of the functioning point of weapons and the ludicrous chasing of clouds was done away with. The term “impactor” was coined by J.H. Gaddum to provide a distinction from “impingement” of particles into liquid. (May 1982, p. 37)

FIG. 17 Swinging impactor (1941) for sampling particles above about 30 μm diameter in the open air with wind speeds exceeding 8 mph (May 1982) [Reprinted from Journal of Aerosol Science, Vol. 13, K. May, A Personal Note on the History of the Cascade Impactor, 37–47, Copyright 1982, with permission from Elsevier].

Much work was done in 1941 to understand this impactor, including some flow visualization of particle sampling in a wind tunnel (Figure 18). Studies of the original design (upper photo) showed particles settling in the inlet nozzle, and secondary impaction caused interstage losses in the internal surfaces of the impactor. The lower photo shows an attempt to capture these internal losses in a second stage by using two stage hemispherical impaction plates. This was further refined in 1941 by Druett with the “microimpactor” (Figure 19), which was powered by an ejector. This device also attempted to collect the particles impacted on the wall with second-stage glass plates.

FIG. 18 (upper) Streak photographs of 30 μm particles in a swinging impactor with a wind speed of 4 mph. Note the deposition of particles under gravity in the inlet and by impaction from the flow spilling off the impaction plate. (lower) Swinging impactor with two-stage hemispherical impaction plates. (May 1982) [Reprinted from Journal of Aerosol Science, Vol. 13, K. May, A Personal Note on the History of the Cascade Impactor, 37–47, Copyright 1982, with permission from Elsevier].

FIG. 19 Microimpactor (1941) powered by air ejector. Glass impaction plates at edge of first-stage impaction plate provide second-stage collection (May 1982) [Reprinted from Journal of Aerosol Science, Vol. 13, K. May, A Personal Note on the History of the Cascade Impactor, 37–47, Copyright 1982, with permission from Elsevier].

It appears at this point that May understood that the cut size could be controlled by jet velocity, and the cascade impactor as we know it today was developed. The first cascade impactor had cut sizes that represented various sections of the respiratory tract:

My contemporary note book shows the genesis of the cascade impactor from a jet impacting on a flat plate behind a curved tube “Goddum nose” with a third stage of collection in a glass bubbler. This was in March 1942 and was followed in April by a device consisting of a straight glass tube “nostril” impinging an air jet on to a microscope slide “turbinal” then a curved glass tube “trachea,” followed by a “lung” consisting of a fine jet impaction stage backed by a filter. (May 1982, p. 38)

Because of the various components of this device it was difficult to use in the field, and May made the final step to the May impactor design (May, 1945) for the following reasons (May 1982, pp. 38–45):

What was wanted was something which could be married to the capabilities of the ejector and which would be reasonably easy to handle in considerable numbers in respect of transportation, deployment and changing of sampling slides between tests, often by forceps wielded by frozen or sweaty fingers (according to the season) with or without protective gloves in a wind, on rough grassland, as viewed through a gas mask! It also had to have a good collection efficiency for both large and small airborne particles.

By 20 July 1942, the first four-stage cascade impactor was ready. It was substantially the same as described in my 1945 paper and as later manufactured by Casella of London until they changed the design to one which has proved less popular. The 1942 design featured minimal frontal area and light weight so that the tail fin could give sensitive response to wind directional changes. Wind tunnel and other studies showed that the particle capture was satisfactory and the device replaced the swinging impactor in the field.

With this understanding of the background circumstances at the time of the development of the May cascade impactor, those of us that have had the privilege of using this impactor can appreciate the genius of the design for the purpose it was to serve. The rugged glass slide impaction plates can be easily removed and replaced with just the removal of rubber end caps; no disassembly of the impactor is required, or even possible.

My first impactor sampling experience was with this impactor in 1963. I used it to sample particles in the exhaust of a solid fuel rocket. Unfortunately, the impactor met its demise, as I had placed the impactor too close to the rocket nozzle; after the test the impactor was just a pile of disconnected tubes and nozzles. All parts of the impactor were held together with a soft solder that melted in the heat of the exhaust.

The May cascade impactor indeed was the first “cascade impactor” of significance. Figure 20 shows an X-ray photograph and a schematic with dimensions of the impactor, and Figure 21 shows the impactor with a wind vane when it was used as a swinging cascade impactor.

FIG. 20 The original 1942 May cascade impactor (May 1982) [Reprinted from Journal of Aerosol Science, Vol. 13, K. May, A Personal Note on the History of the Cascade Impactor, 37–47, Copyright 1982, with permission from Elsevier]: (a) X-ray photograph and (b) schematic and dimensions.

FIG. 21 A later type of swinging [May] cascade impactor for use in the open air. It is provided with a duct before the inlet that decreases losses (May 1982) [Reprinted from Journal of Aerosol Science, Vol. 13, K. May, A Personal Note on the History of the Cascade Impactor, 37–47, Copyright 1982, with permission from Elsevier].

May calibrated his impactor by sizing and counting the particles on the stages with a microscope. He apparently felt very strongly that using the microscope was the only accurate method of calibrating an impactor:

To return to the original calibration of the impactor, this had to be done by extensive counting and sizing of sampled droplets under the microscope. This is tedious and needs mental application but I find no sympathy with the horrified revulsion with which such work is regarded by so many. May I suggest that laziness or perhaps the prospect of a Ph.D. has inspired much lengthy and highly expensive research to automate the rapid evaluation of a particle deposit (with dubious accuracy?). It could well be that 100 man-hours of research have been spent to save one person-hour at the microscope. Why spend $50,000 on a pile of black boxes which often go wrong when for the same money you can get 10 years out of a prettier and replaceable assistant who can also make coffee? (May 1982, p. 45)

May was such a strong believer in using the microscope for sizing and counting particles for the impactor that he developed an eyepiece graticlule specially designed for counting and sizing particles and felt that it should be widely used for this type of work. He felt that it gave “considerable and reproducible accuracy without excessive eyestrain” (May 1982, p. 45).

Shortly after May introduced his impactor, the design was modified by Sonkin (1946) with higher velocity jets so that smaller particles could be collected. This impactor, shown in Figure 22, used glass tubes with the ends of the tubes drawn into rectangular slits. The basic layout, however, was the same as the original May impactor.

FIG. 22 Sonkin modification of the May cascade impactor, using flattened glass tube nozzles to collect smaller particles. (Sonkin 1946) [Reprinted with permission].

A more robust version of the May impactor was designed by Laskin (1949) for sampling uranium dust, where the impactor was machined out of solid brass instead of being made from sheet metal, the construction material of the original May impactor. This allowed for threaded end caps on the stages, greater precision of the nozzle dimensions, and interchangeability of stage nozzles between units. In addition, machining from brass provided highly polished interior surfaces to reduce interstage particle losses. Finally, an afterfilter was incorporated after the fourth stage to collect particles that would normally penetrate through the May impactor.

Because of the sturdy design of the Laskin (1949) version of the May impactor, Wilcox (1953) chose it for the basic of his own version, shown in Figure 23. He elected to use the first, second, and fourth stages of the Laskin design as the first three stages of his design and then added two more stages with lower cut sizes. The last stage was designed for sonic flow to obtain the minimum cut size and to serve as flow control. A special feature of his design was impaction plates that allowed for collecting samples directly on thin films for electron microscope examination (Wilcox 1955).

FIG. 23 Wilcox modification of the May cascade impactor, with five stages, the last of which was sonic for flow control (Wilcox 1953) [Reprinted with permission].

A further modified version of the May impactor was made commercially available in 1960 as the Casella Mk. 2 impactor, shown in Figure 24 (Casella Instruction Leaflet; Kusnetz 1962). Although the Casella impactor is of the same general configuration as the May impactor, there are several notable differences. The location of the end caps are relocated so that round 1 in diameter glass disks are used for the impaction plates instead of the rectangular microscope slides that are inserted from the side, as in the May impactor and the Laskin version. The body and end caps were cast aluminum and also included an afterfilter, and construction is such that fewer than four stages can be used if desired.

FIG. 24 Casella Mk. 2 impactor, a commercial version of the May impactor (Casella Instruction Leaflet) [Reprinted with permission].

The Casella Mk. 2 impactor was modified for sampling from ducts by First (1952) and later modified by Gussman and Gordon (1966). The Gussman and Gordon version is shown in Figure 25. The first stage was located inside the duct with an extension tube leading to the rest of the cascade impactor outside the duct. An interchangeable nozzle was added to the inlet so as to achieve isokinetic sampling, and a filter adapter was added at the exit to accommodate a 2 in filter. The deposits on the impaction plates were examined by microscope so the deposits had to be light, which could be difficult if the dust concentration in the duct was large. Thus, the impactor had to be run for a very short period of time. This was achieved by a very unique feature; clean, dry, compressed air was supplied to the first stage cavity inside the duct at a flow rate larger than was sampled by the impactor. When the compressed air was turned on, clean air flowed through the impactor and also out the inlet nozzle as a reverse flow. This effectively shut off the impactor. When the compressed air flow was turned off, duct air was sampled through the impactor. Sampling times as short as 0.6 s could be achieved if a timer was used with a solenoid valve.

FIG. 25 Casella Mk. 2 impactor modified for duct sampling (Gussman and Gordon 1966) [Reprinted with permission]: (a) sectional view of impactor and sampling system in duct and (b) photograph of modified impactor.

The Casella Mk. 2 impactor was thoroughly evaluated and calibrated by Wells (1967) with unit density spheres and oxides of uranium and plutonium. Wells found that the impactor worked well for nonspherical particles over a wide density range. This work also investigated which parameter—the effective drop size, the mass median diameter, or the effective 50% cutoff size—should be used to characterize the aerosol collected by an impactor. He concluded that the 50% cutoff size was best because it was independent of the incident particle size distribution.

May (1956) made a very interesting modification to his basic impactor by developing a version with moving impaction plates. This involved using a rack and pinion at each stage, as shown in Figure 26a, and rotating all of the pinions with one motor by use of a chain and sprockets, as shown in Figure 26b. His discussion on using this rather complex mechanism rather than some design that has all the slides on one plane leads to some insight for the reasoning of the layout of the original May impactor:

When it was decided to design a moving-slide impactor, much thought was given to the geometric layout of the proposed instrument. There was obviously no very neat way of moving four slides in different planes simultaneously, so experiments were done with an instrument in which the air jets impinged successively on the rims of drums arranged one behind the other. Consideration was also given to the use of rotating disks as impingement surfaces and to various other arrangements, e.g. with all slides in one plane. It soon appeared from the experiments that the layout of the original instrument is the only feasible one when it is required to obtain good fractionation of particles larger than .5 μ m–10 μ m, without high internal losses. The reason for this is that in the original layout, stages where coarse particles remain in the cloud have the jet chamfered to force the air stream straight to the next jet, which is reached without any further change in direction after the impingement. All other practical layouts which could be envisaged necessitated 90° or, more frequently, 180° turns to the particle-laden air stream between successive impactions. With coarse particles in the cloud, this proved to give an inevitable loss of a seriously high proportion, as particles are thrown out onto the internal wall of the instrument at the bends by inertial forces. It has not been possible to envisage a design embodying a turn between successive impactions where this loss would not take place. (May 1956, pp. 482–483)

FIG. 26 May cascade impactor with moving slides (May 1956) [Reprinted with permission]: (a) sectional view and (b) impactor with sprockets and drive chain.

This shows that the basic layout of the original May impactor was to eliminate any turns of the air flow between the air coming off of the impaction plate to the next nozzle, for the purpose of reducing interstage losses as much as possible.

May's main purpose for developing the moving slide impactor was to not overload the impaction plates: “adhesive coating of the slides essentially used for dry particles, becomes overloaded as soon as sufficient particles strike the narrow impaction area to give then an appreciable chance of striking others already impacted. Particles then tend to bounce off instead of being retained by the adhesive.” (May 1956, p. 481)

He also states that this feature could be useful for obtaining a time history of the aerosol cloud or to collect several separate samples on a single set of slides by operating the motor that moves the slides remotely between samples.

Lippmann (1959) describes the HASL (Health and Safety Laboratory) modified casella cascade impactor, which was a modification to the original commercial Casella cascade impactor. The aim was to eliminate air leaks from the press-on rubber caps that sealed the tube ends and the rubber bands at the seams. These modifications consisted of bonding wide-mouth Polyethylene bottlenecks onto the tube ends, which could be sealed with screw bottle caps, and sealing the seams with plastic electrical tape as shown in Figure 27. A filter was also added after the final stage. Lippmann performed an extensive calibration and evaluation of the modified Casella impactor since, up to this time, “Its [Casella cascade impactor] usefulness has been questioned since the various stage calibrations that had been proposed for it were not in agreement. With the same mass distribution data, each gave a distribution curve different from the others and different from the sizing of simultaneous membrane filter samples.” (Lippmann 1959, p. 415)

FIG. 27 Health and Safety Laboratory modified Cassella impactor (Lippmann 1959) [Reprinted with permission].

After calibration of the Casella impactor, modified to eliminate leaks, and comparing the resulting particle size distributions to size distributions determined from examination of particles simultaneously collected on membrane filters, Lippmann concluded that:

The results obtained using this calibration have been found to check very closely with the results obtained from equivalent and simultaneous membrane filter samples. With the use of this new calibration, a Casella cascade impactor, modified to include a filter paper stage and to prevent air leakage, should be capable of providing a reasonably accurate estimate of the particle size distribution in the usual industrial atmosphere. (Lippmann 1959, p. 416)

This work by Lippmann is important in that it signifies the start of noted scientists beginning to look at cascade impactors as serious samplers for general-purpose sampling of aerosol particles. Lippmann (1961) went on to design a cascade impactor that was more suitable for general field studies of industrial aerosols, and later Lundgren (1967) designed an impactor for general field studies of atmospheric aerosols (see below section, “1960 to 1969”).

Although it falls outside of the time period covered by this article, I feel that I must mention May's final contribution to impactor design, the “ultimate” cascade impactor shown in Figure 28 (May 1975). As with the original May impactor, rectangular nozzles were used with microscope impaction plates.

FIG. 28 May's “Ultimate” Impactor (1975) (May, 1982) [Reprinted from Journal of Aerosol Science, Vol. 13, K. May, A Personal Note on the History of the Cascade Impactor, 37–47, Copyright 1982, with permission from Elsevier].

May was not the first to design an impactor with moving impaction plates. Bourdillon et al. (1941) described an impactor that became the Casella slit sampler, which used a rotating agar culture plate below a slit for time-resolved collection of airborne bacteria. There were two versions of this sampler (Figure 29). The small version operated at 1 cfm and had one slit. The larger version was used for sampling where the bacteria concentrations were lower and operated at 25 cfm. This unit had four slits, 90° apart above a 6 in diameter impaction plate, about four times the effective area of the smaller unit. The plate rotated through a gear train and a synchronous motor with three speeds provided (Kusnetz 1962).

FIG. 29 Casella Slit Sampler for Airborne Bacteria (Kusnetz 1962) [From American Conference of Governmental Industrial Hygienists (ACGIH®), AirSamplingInstruments , SecondEdition, Copyright 1962. Reprintedwithpermission]: (a) smallmodel (1 cfm) mountedonpumpbox (forconcentrationsfrom 1–100,000 bacteriumpercubicfoot) and (b) largemodel (25 cfm) (forconcentrationsdownto 1 bacteriumper 100 ft ³$).

Another bacteria sampler (Figure 30) with a moving impaction plate, described by Kusnetz (1962), was unique in that the deposit was spread over a distance of 484 in. This was accomplished by rotating a 5 ½ indiameterby 3 5/8 inlongdrumimpactionplateinsideofa 3 ½ quart cylindrical can that had an impactor nozzle penetrating through the can wall. As the drum impaction plate rotated, it advanced axially 1/8 in per revolution for a total of 28 revolutions. Timing motors from 1 rpm to 1 rph allowed for sampling times of 28 min to 28 h. Flow rates ranged from 100 ml/min to 3 l/min and nozzles of different diameters could be selected to achieve the desired cut size. Preparing the impaction plate was not simple in that a sheet metal band was tightened around the drum, and melted agar was injected into the cavity between the cylinder and the band. After the agar solidified, the band was removed and the impaction plate was ready for use.

FIG. 30 Monitor for airborne bacteria (Kusnetz 1962) [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments, Second Edition, Copyright 1962. Reprinted with permission]: (a) sampling can with nozzle on side and (b) drum impaction plate.

1950 to 1959

Although there were some theoretical analyses of particles on body collectors prior to 1950, in the early 1950s there was much activity in using theoretical analysis to understand the flow field and resulting particle collection efficiency curves of impactors.

One of the simplest theories was one devised by Baurmash et al. (1949) and reported by Wilcox (1953). This theory assumed that the flow leaving the nozzle made a 90° turn at the impaction plate. The particles were thrown to the impaction plate by centrifugal force as they traveled the 90^ˆ turn with a radius of curvature equal to the half width of the nozzle.

Davies and Aylward (1951) used the method of conformal mapping to obtain solutions to the Euler's equation for the flow of an ideal frictionless fluid through a rectangular nozzle. Particle trajectories were calculated in these flow fields in a step-by-step calculation and a determination made as to whether or not a particle of a certain size will strike the impaction plate. This theory was then applied for determining efficiency curves for the May cascade impactor, the midget impinger, the Bausch and Lomb dust counter, and the Owen's jet dust counter. The results were compared to experimentally determined efficiency curves of the same samplers (Davies et al. 1951).

Ranz and Wong (1952a) developed a theory for both round and rectangular nozzle impactors using approximate flow fields for both cases. The flow fields were assumed to be frictionless stagnation flow in the vicinity of the stagnation point. The resulting equation was similar in form to the corresponding equations for free vibration with viscous damping. Thus, the known solutions for free vibration were applied to determining if the particle would strike the impaction plate. The theoretical curves were compared to experimental curves by Ranz and Wong (1952b). The work of Ranz and Wong (1952a), however, did not account for the effects of jet-to-plate distance on the flow field. This ability was added many years latter by Mercer and Chow (1968) and Mercer and Stafford (1969) for round and rectangular nozzles, respectively. Their changes were only in the specification of the flow field; the determination of whether a particle impacted was the same as used by Ranz and Wong (1952a).

The extensive theoretical work being performed on impactors in the early 1950s was followed by the design and development of a number of cascade impactors for a variety of purposes. Thus, the 1950s can be considered the time when cascade impactors were being recognized as very useful instruments for many types of studies related to aerosols, and the knowledge gained through the theoretical results encouraged diversification in impactor design.

In 1954, the Brink cascade impactor was developed at Monsanto Chemical Company to sample particles from industrial gas streams that were saturated with vapor (Brink 1958). This impactor was part of a sampling train consisting of a cyclone inlet and glass-wool–packed filter tubes as afterfilters, all packaged in a lightweight environmentally controlled box as shown in Figure 31. The Brink cascade impactor consisted of five single-nozzle stages with low wall cups as impaction plates and operated in a flow rate range of 2.7 to 3.7 L/min and cut sizes in the range of 0.3 to 3 μm.

FIG. 31 Brink Cascade Impactor (Brink 1958) [Reprinted with permission from Industrial & Engineering Chemistry, Vol. 50, J. Brink, Cascade Impactor for Adiabatic Measurements, 645–648, Copyright 1958, American Chemical Society]: (a) system for sampling from a process line and (b) section of five-stage single nozzle impactor.

Another impactor developed specifically to sample industrial aerosols was the Battelle No. 6 cascade impactor (Pilcher et al. 1955; Mitchell and Pilcher 1959), shown in Figure 32. Their concern was to develop an impactor that could be used to characterize the droplet sizes from various spraying apparatuses. They realized there would be different optimum droplet sizes for insect sprays, paint sprays, fruit sprays, medical sprays, and other commercial aerosols. Thus, they required a cascade impactor with sharp cuts, a wide range of cut sizes, and low interstage losses for sampling. The result was a six-stage cascade impactor with an afterfilter using single round nozzles at each stage. The upper cut size they believed to be useful for their work was 16 μm, and the lower practical cut size was 0.5 μm. Much effort was given to determining the distance between the edge of the impaction plate and the wall of the impactor body that would give them minimum losses of particles in the interstage passages.

FIG. 32 Battelle No. 6 Cascade Impactor (Pilcher et al. 1955; Mitchell and Pilcher 1959): (a) schematic [Reprinted with permission from Industrial & Engineering Chemistry, Vol. 51, R. Mitchell and J. Pilcher, Improved Cascade Impactor for Measuring Aerosol Particle Sizes in Air Pollutants, Commercial Aerosols and Cigarette Smoke, 1039–1042, Copyright 1959, American Chemical Society] and (b) photograph [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments, Second Edition, Copyright 1962. Reprinted with permission].

The annular impactor was used for sampling radioactive elements, such as plutonium, without collecting the coexisting particles containing radon and thoron daughter products. The cut size was controlled by the slit width and/or the flow rate. The collection efficiency could be determined experimentally by connecting two annular impactors with the same cut sizes in series, as shown in Figure 33c. Using the basic equations applied by Owens with his two-stage impactor to determine the total challenge aerosol concentration, the annular impactor solved for the stage collection efficiency using the radioactive activity on each of the two impaction plates:

(2)

where T is the total challenge aerosol activity and A and B are the activity on the stages plates A and B, respectively (ACGIH 1960).

FIG. 33 Annular impactor air sampler. (a) Annular impactor mounted on Staplex high-volume sampler [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments , Second Edition, Copyright 1962. Reprinted with permission], (b) cross−section view of annular impactor [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments , First Edition, Copyright 1960. Reprinted with permission], and (c) calibration attachment on annular impactor [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments , First Edition, Copyright 1960. Reprinted with permission].

Some impactors were designed specifically to collect radioactive particles. The annular impactor (Hoy and Croley 1956; Kusnetz 1962) was one of the more unconventional impactors. As shown in Figure 33a, the annular impactor was connected to the entrance of a Staplex high-volume sampler that could operate up to 50 cfm. The air stream entered the base of the cone (see Figure 33b) and passed through an annular slit at the end of the cone created by the cone and the central tubular passage leading to the Staplex sampler. A grease-coated impaction plate called a “planchet” was placed on the small end of the cone and was held in place by the suction of the Staplex sampler. At the end of a run, an open envelope was held below the planchet and the Staplex sampler tuned off so that the planchet would fall into the envelope.

In 1956 a cascade impactor, the Andersen impactor, was developed at Dugway Proving Ground for the specific purpose of determining the size distribution of viable airborne particles (Andersen 1958). This impactor, shown in Figure 34a and 34c, was traditionally referred to as the Andersen viable impactor. Each of the six stages consisted of 400 nozzles and impaction plates which were petri dishes containing agar culture medium. There were six stages, with the smallest cut size being 1 μm when operated at 1 cfm. The original prototype design used 340 nozzles and only four stages (Andersen 1956).

FIG. 34 Andersen cascade impactors: (a) six-stage viable sampler (Andersen 1958) [Reprinted with permission], (b) six-stage dust sampler (Andersen 1966) [Reprinted with permission], and (c) Andersen samplers. The sampler on the left is the viable sampler and the sampler on the right is the dust sampler. (Kusnetz 1962) [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments, Second Edition, Copyright 1962. Reprinted with permission.]

After a sampling period, the petri dish impaction plates were incubated and the viable deposits counted. Two methods were used to determine the number of viable particles. In one method, the number of 400 deposits that had viable growth were counted and a conversion made by using a “positive hole count” table to correct for the probability that multiple viable particles were in one deposit. This technique could be used for up to 1200 to 1500 particles per stage. The second method used a microscope to count the number of microcolonies in a number of the 400 deposits and calculating the total viable particles. This technique could be used for counts as high as 40,000 to 50,000 (Andersen 1958).

May (1964) did a redesign of the Andersen viable impactor after doing an extensive calibration and evaluation of stage deposits. He found that the deposits of the first three stages were not uniform across the impaction plates and attributed this to the fact that the pressure drop from the center to the exterior of the stage was about equal to the pressure drop across the nozzles. This meant that the static pressure was larger near the center of the plate than near the outer edge, with the result that more flow passed through the exterior nozzles as evidenced by larger deposits under the outer nozzles. He also found that the cross-flow of spent air would deflect the outer jets from impinging on the impaction plate in certain areas. He ascertained that if there were channels from the center to the exterior where there were no nozzles, the air jets were not blown off the impaction plate as readily as areas were these channels were obstructed with air jets from nozzles. He called these channels “streets.”

After his study of the Andersen impactor, May made three changes to the upper stages. First, he reduced the number of nozzles in the first two stages from 400 to 200. This increased the pressure drop across the nozzles so that the airflow through the nozzles was uniform across the plate. For his second change, he arranged these nozzles in a radial pattern so as to form “streets” for the spent flow to escape. He stated that the inner air jets in a radial pattern would shield the outer jets from cross-flow effects. Finally, he machined conical inlets to the nozzles of the upper stages to reduce losses at the nozzle inlets.

The basic Andersen viable impactor has been used in several configurations. A preseparator was developed for sampling in areas where there were considerable quantities of particles larger than the cut size of the first stage (Lidwell and Noble 1965).

Where particle concentrations were light, Solomon and Gilliam (1970) proposed using the Andersen viable impactor with the impaction plate petri dishes removed from the upper five stages of the impactor and collected all particles at the sixth stage. This reduced the sampling time before enough particles could be collected and reduced the number of impaction plates that had to be analyzed. This was further simplified by using the sixth stage only, creating a single-stage Andersen viable impactor (Jones et al. 1985).

In 1966 the Andersen viable impactor was modified for respiratory health hazard assessment of any airborne particles by substituting stainless steel impaction plates for the petri dishes, as shown in Figure 34b and 34c (Andersen 1966). The number and size of the nozzles per stage remained unchanged from the original viable sampler; only the bodies of the stages were redesigned to accommodate the stainless steel plates. This version was known as the nonviable Andersen impactor.

An MkI version of the impactor was introduced in about 1970 with the addition of two upper stages to make an eight-stage impactor with a larger cut size range than the original. Later, the top two stages were modified and a preimpactor was added in the MkII version following the work of McFarland et al. (1977).

The development of the Andersen impactor was a milestone in impactor evolution in that it was the first multiple nozzle cascade impactor. More copies have been manufactured, sold, and used than any impactor before or since. As a result of its popularity, the Andersen impactor has been the subject of many papers describing its use or evaluations of the impactor itself.

Couchman (1965) adjusted the jet-to-plate distance of the Andersen impactor to about 3 for all stages and operated it at 18 lpm to simulate the human breathing rate. Flesch et al. (1967) showed the difficulties of impactor calibration by calibrating the Andersen impactor with two different particles, methylene blue and polystyrene latex, and obtained efficiency curves that did not match unless he adjusted the methylene blue density. In the process of their calibrations, they also noted “there is considerable departure from equal mass collection at each hole on stages 1 and 2,” the same observation made by May (1964).

Other calibration studies were made and reported by Lundgern (1967), Riediger (1974), Rao and Whitby (1979), Franzen and Fissan (1979), Boulaud et al. (1980), Tanaka et al. (1983), Mitchell et al. (1988), and Vaughan (1986, 1989).

1960 to 1969

In the 1960s, cascade impactors became recognized as a versatile instrument for the sampling and classification of aerosol particles and became the driving force for the development of impactors for the “general” field sampling of aerosols. The original cascade impactor, the May (Casella) impactor, was still an inspiration for new impactor development. In 1961, Lippmann (1961) designed the HASL (commercially known as the Unico) impactor after the calibration and modifications he performed on the original Casella impactor (Lippmann 1959). The Unico impactor still used rectangular nozzles and glass slide impaction plates as did the Casella impactor but the design was very much simplified, as shown in Figure 35. It was a four-stage cascade impactor with an afterfilter, and all of the stages were incorporated into one block. Two 1 in × 3 in glass slides covered two sides of the block and became the impaction surfaces, with two stages per slide. Note that the impaction characteristics of the Unico impactor would be different than the Casella impactor since all stages of the Unico impactor impacted into a right angle corner, forcing the flow all to one side. This type of flow field was only in stage one of the Casella impactor. A novel feature of this impactor was that the two slides were mounted in a frame that could slide relative to the nozzle block. In this manner, the impaction plates could be moved to nine positions, allowing much more sample to be collected before the plates became overloaded than in the original Casella impactor. Although there was the Casella impactor with moving slides, it was more complicated and expensive to build than the Unico impactor.

FIG. 35 The HASL (Health and Safety Laboratory) Cascade impactor (Lippmann 1965): (a) assembled and disassembled [Reprinted with permission] and (b) flow path [From American Conference of Governmental Industrial Hygienists (ACGIH®), Air Sampling Instruments, Second Edition, Copyright 1962. Reprinted with permission].

In 1967, another four-stage rectangular nozzle cascade impactor was designed and built by Lundgren (1967). This impactor used rotating drums for impaction plates, as shown in Figure 36 for sampling very dusty atmospheres. Because the design of this impactor allowed the air at the impaction surface to flow in both directions (no impaction into right angle corners), the Lundgren impactor would be expected to provide sharper cutoff characteristics than for rectangular nozzle impactors, where the flow is forced in one direction at the impaction plate. The rotating drum feature allowed for 10 in² of impaction surface for each stage. Although the large surface area of the impaction drums could be used to collect large samples, the Lundgren impactor was more commonly used to obtain time-resolved particle concentration data. For example, the impaction plate drum speed could be set for one revolution in 24 h and inspection of the deposits on the drums would provide a 24 h time-resolved history of the aerosol. In addition to presenting the details of the new impactor design, Lundgren did an extensive evaluation of the interstage losses in the Andersen, Casella, Unico, and Lundgren impactors.

FIG. 36 Lundgren impactor with rotating drum impaction plates (Lundgren 1967) [Reprinted with permission].

As investigators were still modifying the original May cascade impactor, May was developing a new cascade impactor of an entirely different design, the multistage liquid impinger (May 1966). This impactor was designed for sampling viable airborne particles into a liquid. As May indicated in his paper, there were two philosophies for the sampling of viable particles. In one method, the particles were impacted directly into an agar and the viable colonies were counted. In the second method, the particles were impacted into a liquid where particles were separated into individual component cells.

Figure 37a shows sectional views of the multistage liquid impinger and reveals some interesting features. First, the impinger was made of Pyrex® glass and there were three chambers, each containing a nozzle/impaction plate stage. The impaction plates of the first two stages were porous frits. Liquid was put in these chambers so that the levels were just below the upper surface of the frits. Due to evaporation, liquid had to be added approximately every 45 min. The third-stage nozzle is interesting in that it was not only used to impact particles but the tangential component of the jet created a vigorous swirl in the liquid, which insured that the impaction surface was always wet. Also note that the exhaust port was at the center of the third chamber so that water would not be expelled with the air. There were attachments for different sampling conditions. For example, Figure 37a shows a shield on the inlet for sampling in a crosswind. The air was brought to rest and the impinger sampled from the stagnated flow. Finally, Figure 37b shows that there were three sizes of the impinger, 50, 20, and 10 lpm.

FIG. 37 Multistage liquid impinger (May 1966): (a) cross-sectional views [Reprinted with permission from Bacteriological Reviews, Crown Copyright 1966] and (b) three models of the multistage liquid impinger (A-50 L per min, B-20 L per min, and C-10 L per min) [Reprinted with permission from Her Majesty's Stationary Office, Crown Copyright 1966].

There were some special purpose impactors being developed in the 1960s, such as a large volume impactor collector for sampling particles from the upper atmosphere (Zeller 1965, p. 1). This impactor had a unique purpose as stated by the author:

At the present time there is a need for sampling systems which can operate at altitudes above 30 km and which can provide basic data concerning particle size, size distribution, and particle composition. In addition to basic studies involving micrometeorites and naturally occurring debris, this requirement is of special importance because of the potential hazards to man arising out of the re-entry burnup of reactor-powered satellite vehicles and the need for information relative to high altitude nuclear weapons tests.

Although the impactor was rather standard, being a two-stage multiple-round nozzle cascade impactor, the entire sampling system, shown in Figure 38, was unique since it had to be carried to high altitudes by balloon. Battery-powered pumps were too heavy, so an air ejector powered by compressed nitrogen was developed. The nitrogen was contained in a 76 l spherical titanium tank at 3600 psi. On one flight, this operated the impactor for 80 min at a flow rate of 500 cfm at an altitude of 34 km, and on another flight, the impactor was operated for 150 min at a flow rate of 450 cfm at an altitude of 31.9 km. In the later flight, two nitrogen tanks were used.

FIG. 38 Large-volume high-altitude cascade impactor (Zeller 1965) [Reprinted with permission]: (a) assembled sampling system in a frame to be carried to high altitudes by a balloon and (b) exploded view of cascade impactor.

The cascade impactor consisted of nineteen 1.8 cm diameter nozzles and forty nine 0.694 cm diameter nozzles for the first and second stages, respectively. Because the sampling was at 34 km (a pressure of about 0.009 atm), the impactor was operating at low-pressure, one of the techniques employed to collect small particles. Indeed, the cut sizes were about 0.3 μm and 0.03 μm, respectively. This was not the first work with impactors operating at low pressures. Stern et al. (1962) did a rather extensive study of impactors operating at low pressures to verify the effect of the Cunningham correction factor on the particle collection efficiency curves. Later, Buchholz (1970a, b) developed a multijet low-pressure impactor.

Another unique special purpose impactor, called the “Stamp-Licker” air sampler, was one developed for the collection of respiratory viruses where the virus particles were to be collected on a wet surface of the impaction plate (Lidwell 1966). This impactor, as shown in Figure 39, has two 2.5 cm diameter stainless steel cylinders rotating below two 5 cm-long slit nozzles. The cylinders were in a horizontal orientation and the cylindrical surfaces kept moist by rotating through a liquid bath.

FIG. 39 “Stamp-Licker” air sampler for virus aerosols. Sampler contains two rectangular slit nozzles impacting on two drum impaction plates that are rotating with the lower edges rotating through a liquid (Lidwell 1966) [Reprinted with permission from Bacteriological Reviews, Crown Copyright 1966].

A particularly clean cascade impactor design was one by Cohen and Montan (1967). The impactor was designed for a very specific purpose of sampling radioactive particles, as stated by the authors:

Among the most potentially useful projects of the Plowshare Program for peaceful uses of nuclear explosives are those involving large-scale excavation. In these projects, the resulting dust cloud may contain significant quantities of radioactive material, including fission products and products of neutron activation of the device components and medium in which the detonation occurred. To evaluate properly the extent of hazard that might be created by these operations, it is essential to know not only the quantities but also the inertial characteristics of the released radioactive material. This information is necessary for predicting the fate of these materials in the biosphere. (Cohen and Montan 1967, p. 95)

For this purpose they needed an impactor that could operate over a wide range of flow rates and a large range of particle sizes because the quantity and size of the radioactive particles could vary widely. Thus they designed a cascade impactor that had greater flexibility than other impactors available commercially. As shown in Figure 40, the impaction plate was a disk with a hole in the center where the air and particles not collected can pass to the next stage. This arrangement of nozzle and impaction plates greatly reduced particle interstage losses as compared to a design where the impaction plate was a circular disk and the particles must flow between the edges of the disk and the impactor wall to pass to the next stage. The design was also very flexible, in that it was easy to put different stages into the impactor to obtain the desired number and size of nozzles at each stage, as well as the number of stages.

FIG. 40 Cohen and Montan cascade impactor (Cohen and Montan 1967) [Reprinted with permission] (nozzles in annular ring around central hole in the impaction plate leading to next stage: this configuration used in several subsequent cascade impactors): (a) cross-sectional view, (b) assembled impactor, and (c) nozzle plates and impaction plates.

Another specifically designed impactor for sampling radioactive particles was the Lovelace low flow rate cascade impactor (Mercer et al. 1970). Since this was used for sampling radioactive particles of high specific activity in animal exposure chambers, it was designed for a low flow rate in the range of 50 to 150 cc/min. The cascade impactor, as shown in Figure 41, had seven stages with single round nozzles and was a very compact design. The sampler was only 3 in long by 1 5/8 in in diameter.

FIG. 41 Lovelace low flow rate cascade impactor (Mercer et al. 1970) [Reprinted from Journal of Aerosol Science, Vol. 1, T. Mercer, M. Tillery, and G. Newton, A Multi-Stage, Low Flow Rate Cascade Impactor, 9–15, Copyright 1970, with permission from Elsevier].

In the late 1960s and early 1970s impactors became popular for sampling from flue gas ducts or stacks. Due to high temperatures, the impactors were made of stainless steel and were cylindrical in shape so that they could be inserted in a 3 in diameter access sampling port. One such impactor was the University of Washington Mark 1 six-stage source test cascade impactor (Pilat et al. 1970). The Mark 2 impactor followed. It included an integral afterfilter and a first stage consisting of a single nozzle, to make a seven-stage impactor. Both impactors had disk impaction plates with rather large interstage particle losses. A Mark 3 impactor was then developed with the flow pattern of the Cohen and Montan (1967) impactor, which reduced interstage losses. This version has been widely used for stack sampling. The Mark 2 and Mark 3 source test cascade impactors are shown in Figure 42.

FIG. 42 Pilat (UW) source test cascade impactors (Pilat et al. 1970). (a) Mark 2 source test cascade impactor [Reprinted from Atmospheric Environment, Vol. 4, M. Pilat, D. Ensor, and J. Bosch, Source Test Cascade Impactor, 671–679, Copyright 1970, with permission from Elsevier] and (b) Mark 3 source test cascade impactor [Reprinted with permission].

Also in the late 1960s and early 1970s, impactors were being developed that could provide real-time readout of the mass of particles collected on the impaction plates. One technique involved using piezoelectric crystals for impaction plates in the QCM cascade impactor (Chuan 1970, 1976). As the particles were collected on the piezoelectric crystals, the mass of the crystals would increase, changing its natural vibration frequency. This frequency was monitored and printed out as the mass increased on each stage. Figure 43a shows a six-stage (a ten-stage was also available) QCM Cascade. The impactor is shown on the left and the readout module on the right. Figure 43b shows a stage of the impactor with the piezoelectric crystal located in the center of the stage. The rather large cavity, relativity small flow rate (100 ml/min), and the presence of electronics in the cavity, resulted in up to 50% interstage losses (Fairchild and Wheat 1984; Horton et al. 1992).

FIG. 43 QCM cascade impactor (Chuan 1976) [Reprinted from Fine Particles Aerosol Generation, Measurement, Sampling and Analysis, B. Liu, ed.; R. Chaun, Rapid Measurement of Particulate Size Distribution in the Atmosphere, 763–775, Copyright 1976, with permission from Elsevier]: (a) impactor (left) and data recorder (right) and (b) one stage with quartz crystal impaction plate in center.

Another method of sensing the change in the mass of the particle deposit is to apply beta attenuation technology, as in the 2 L/min single-stage impactor of the Respirable Dust Monitor (RDM) by GCA (Lilienfeld and Dulchinos 1972). The beta source (C¹⁴) is located at the inlet of the nozzle and the beta detector is directly beneath the impaction plate. Figure 44 shows the respirable dust monitor with the inlet removed and inverted to show the impactor nozzle. The impaction plate, shown in the monitor body, is a circular Mylar disk that was treated with a sticky coating to keep the impacted particles in a tight deposit in the line of sight between the beta source and the beta counter. This impaction plate could be indexed, so 95 samples could be taken on one impaction plate. Evaluation of the RDM by Marple and Rubow (1978) showed that it could provide mass concentration measurements for coal, rock, silica, potash, and Arizona road dust up to about 20 mg/m³, if there was not appreciable mass below approximately 0.7 um aerodynamic diameter.

FIG. 44 GCA respirable dust monitor (inlet removed and inverted to show impactor nozzle and impaction plate) (Lilienfeld and Dulchinos 1972) [Reprinted with permission].

1970

Exactly 110 years after the origin of the impactor, Marple (1970) published the results of his PhD dissertation, which marked a milestone in the way in which researchers would henceforth design impactors. In the late 1960s, Marple received a preprint of a book written by Gosman et al. (1969) that described the fascinating procedure of finite difference numerical solution for solving the dimensionless form of the Navier-Stokes equations, the governing equations for defining the flow field, in terms of the vorticity and stream function. Since this finite difference technique required no simplifying assumptions to the Navier-Stokes equations, it provided the possibility of defining the streamlines in the impaction region of an impactor with great detail and precision. Also in the late 1960s, Marple had the availability of a high-speed (not high speed by today's standard, but sufficiently fast to provide a solution to flow fields in a 24 h period) digital computer at the University of Minnesota. This was used to determine the flow fields in rectangular and round nozzle impactors.

Since the application of the numerical solution to the Navier-Stokes equations was new, comparisons of the numerically determined streamlines in rectangular nozzle impactors were compared to streamlines in a water flow visualization study. Flow visualization of the streamlines through the impactor consisted of flowing an electrolytic solution through an impactor model made of Plexiglas® in a 20-gallon aquarium. The solution contained a pH indicator (thymol blue) whose color was changed locally by means of metal wire electrodes held at some negative potential. This solution was titrated to the acid side of neutral, which gave it a light orange color. Negative potential wire electrodes in the flow changed the pH of the fluid locally, and the pH indicator in the water changed to a dark blue, allowing the streamlines to be observed and photographed.

The agreement of the streamlines was good, and the numerical technique was considered valid. Since the Navier-Stokes equations were in dimensionless form, the flow fields for impactors of any size would be similar if the Reynolds number of the flow through the nozzle, Re, were the same. Thus, Re became a parameter that defined the flow field in the impactor.

The next step was to solve for particle trajectories in this flow field by numerically solving the equation of motion for the particles, F = ma. When this equation was expressed in dimensionless form, the Stokes number, St, became the defining parameter. Particles of specific values were started at a specific location at the tapered inlet to the nozzle and the trajectory numerically solved for its journey through the impactor nozzle and through the impaction region between the nozzle and the impaction plate. The value of the Stokes number for the particle that just touched the plate was considered the “critical” value to impact on the plate for a particle starting at that location. By starting particles at a number of positions along the entrance, a particle collection efficiency curve could be defined. Again, to verify that the numerical particle trajectory technique was giving realistic results, an experimental rectangular nozzle impactor was designed so that particles could be introduced at various points across the inlet. Particles of various sizes (various Stokes numbers) were introduced at specific locations and the position of the particles impacting on the impaction plate determined with a microscope. These positions were compared to the impaction positions predicted by the numerical program and found to be in good agreement.

A parametric study was then performed to determine the effects of the nozzle Reynolds number, the jet-to-plate distance, and the nozzle throat length on the flow field for both rectangular and round nozzle impactors. It was concluded that an impactor would provide sharp and predicable collection efficiency curves if the value of the Reynolds number was between 500 and 3000, and if the jet-to-plate distance was greater than 0.5 for round nozzle impactors and greater than 1.0 for rectangular nozzle impactors. By using these guidelines, impactor stages could be designed with good confidence that the designed particle cut size would be achieved. The main variables that were left to impactor designers were the design of the passages between stages for low interstage particle losses, design of impaction plates to collect large deposits without particle bounce, and manufacturing techniques.

With this understanding, the design of impactors entered a new era and many impactors for a variety of purposes were designed in the 1970s, 1980s, and 1990s, but this will have to be the subject of another article.

Conclusion

The impactor has had a long history and has been a significant instrument in solving important problems of the day. In doing the research for this article, I felt that one scientist, Ken May, stood out above all others in his understanding of impactors and his contribution to impactor development. I would call him an “impactorholic,” which I would like to be known as one day. Ken May worked with, and developed, impactors for at least 30 years, from 1945 to 1975. He was the first to develop a true cascade impactor (1945) for the purpose of determining a particle size distribution; he made significant improvements to the Andersen impactor; he designed an impactor with moving impaction plates; he developed the multistage liquid impinger; and, finally, he developed the “Ultimate” impactor (1975). To be sure, there were other very significant developments, such as the work by Pouchet and Maddox in the late 1800s, and the later developments by Kotz, Owens, and Greenburg and Smith in the early 1900s, but few made a lifetime study of impactors as May did.

This article covered what I consider the interesting discovery period of impactors from 1860 to 1970. After 1970, a new era unfolded, where the mystery of exactly how an impactor could classify particles so sharply in different size classifications was solved. There were indeed many impactors developed from 1970 to the present time, but I classify them as impactors that are the product of this new enlightened era, and not yet historical.

Epilogue

Although I have chosen to cover only the history of impactors up to 1970, reviewers of the manuscript have suggested that some of the significant impactors developed between 1970 and the present should be mentioned. Although many impactors were designed and tested in this period, three general areas of impactor development seem to stand out. These are: (1) extending the cut sizes of impactors to very small sizes, (2) further development in impactors to provide near real-time indication of the particle mass collected, and (3) designing impactors that provide precise particle size characterization of medical aerosol inhalers for regulatory purposes. Some of these were under development in the late 1960s but not reported in the literature until after 1970, and some are just recently developed.

Some work had been done on low-pressure impactors in the 1960s. However, Hering and coworkers were the first to apply the 1970 impactor theory to the design of a 1 L/min low-pressure cascade impactor to achieve small cut sizes and sharp cuts (Hering et al. 1978, 1979). This was an eight-stage impactor with a sonic orifice between the upper four and the lower four stages. Thus, the upper stages would operate at about atmospheric pressure but the lower four stages would operate at pressures ranging from about 0.2 atmospheres for a cut size of 0.26 μ m at stage 5 to 0.01 atmospheres and a cut size of 0.05 μ m at stage 8.

The Berner impactor (Berner et al. 1979) was a cascade impactor that could also collect particles of very small sizes at the lower stages. This impactor (Figure 45) was designed such that the pressure drop through the impactor stages had sufficient pressure drop to result in the lower stages being at “low pressure” to aid in the collection of small particles. With this arrangement a special pressure drop orifice, commonly used in low pressure impactors, was not required. The cut size at the lowest stage, as calibrated and reported by Wang and John (1988), was 0.082 μ m (0.06 μ m theoretical) at a flow rate of 30 L/min. This was achieved with 130 nozzles of 0.25 mm diameter at an absolute pressure of 33 kPa. A low pressure version of the Berner impactor was evaluated by Hillamo and Kauppinen (1991), and they found that it could collect particles as small as 0.032 μ m (0.015 μ m theoretical) at a pressure of 8.34 kPa. The impactor stage configuration was similar to that of the Cohen and Montan (1967) impactor.

FIG. 45 Berner impactor (Wang and John 1988) [Copyright 1988 from Characteristics of the Berner Impactor for Sampling Inorganic Ions by H. Wang and W. John. Reproduced by permission of Taylor & Francis, Inc., http://www.routledge-ny.com].

Another method to achieve small cut sizes, but at a higher pressure than low-pressure impactors, is to use small nozzles. This was employed in the micro-orifice uniform deposit impactor (MOUDI) (Marple et al. 1991). In this impactor the final stages used 2,000 nozzles, 52 μm in diameter, to collect particles as small as 0.056 μm at an absolute pressure of about 53 kPa, considerably higher than pressers in a low pressure impactor. Figure 46a shows a schematic drawing of a MOUDI impactor body assembly, which contain the impaction plate for the stage above and the nozzles for the stage below. A device, shown in Figure 46b, rotated alternate stages of the MOUDI so that each impaction plate was rotated relative to the nozzles of that stage. The nozzles at each stage were in specific radial patterns so as to provide generally uniform particle deposits on the impaction plates.

FIG. 46 Micro-orifice uniform deposit impactor (MOUDI) (Marple et al. 1991) [Reprinted with permission]: (a) schematic of a body with impaction plate for stage above and nozzles for stage below and (b) photograph of impactor in rotator mechanism.

Further development in near real-time readout impactors took the form of detecting the number of charged particles collected on an impaction plate with an electrometer, as in the electronic cascade impactor (ECI) described by Troop et al. (1980). The five-stage ECI operated at 1.5 L/s with cut sizes from 0.26 to 3 μm. A similar impactor, the electrical low-pressure impactor (ELPI), developed by Keskinen et al. (1992), utilized low pressure to go to smaller particle cut sizes. Figure 47a shows schematically how the impactor stages are electrically isolated, with spring-loaded contacts leading to a multichannel electrometer (Figure 47b). The impactor stage design was based on the Berner low-pressure impactor and thus it also is a derivative of the Cohen and Montan (1967) impactor. A 13-stage commercial version of the ELPI, evaluated and calibrated by Marjamäki et al. (2000), had a cut size as small as 0.029 μm at a pressure of 10 kPa.

FIG. 47 Electrical low pressure impactor (ELPI) (Keskinen et al. 1992) [Reprinted from Journal of Aerosol Science, Vol. 23, J. Keskinen, K. Pieterinen, and M. Lehtimaki, Electrical Low Pressure Impactor, 353–360, Copyright 1992, with permission from Elsevier]: (a) typical stages with electrical contacts and (b) impactor with particle charger and multichannel electrometer.

Recently, an impactor has been designed specifically to characterize the particle size distribution being emitted from medical aerosol inhalers for regulatory purposes. This has resulted in the development of the Next Generation Pharmaceutical Impactor (NGI) shown in Figure 48 (Marple et al. 2003). A special requirement of this impactor was that it had to be easily automated. This led to a design where: (1) all of the impaction plates were incorporated into one assembly that could be quickly and easily removed as a single component for analysis, and (2) the traditional afterfilter was replaced with a fine cut impaction stage, so that all particle deposits of the impactor could be analyzed in the same matter.

FIG. 48 Next generation pharmaceutical impactor (NGI) (shown with lid open to expose nozzles in lid and impaction cups in base) (Marple et al. 2003) [Reprinted with permission].

Acknowledgments

The author would like to thank Dr. Bernard Olson for the preparation and scanning of figures for this publication.