This article was downloaded by: [Stony Brook University] On: 29 October 2014, At: 16:17 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of the Air Pollution Control Association Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/uawm16
Particle Size Distributions of Kraft Paper Mill Aerosols Obtained by Airborne Sampling a
Frank E. Augustine & Richard W. Boubel a
b
The Aerospace Corporation , El Segundo , California , USA
b
Oregon State University , Corvallis , Oregon , USA Published online: 13 Mar 2012.
To cite this article: Frank E. Augustine & Richard W. Boubel (1975) Particle Size Distributions of Kraft Paper Mill Aerosols Obtained by Airborne Sampling, Journal of the Air Pollution Control Association, 25:6, 617-621, DOI: 10.1080/00022470.1975.10470117 To link to this article: http://dx.doi.org/10.1080/00022470.1975.10470117
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Particle Size Distributions of Kraft Paper Mill Aerosols Obtained by Airborne Sampling
Frank E. Augustine The Aerospace Corporation, El Segundo, California
Richard W. Boubel
Downloaded by [Stony Brook University] at 16:17 29 October 2014
Oregon State University, Corvallis, Oregon
Size distributions of particles at several downwind points in a Kraft paper mill plume have been determined by means of airborne sampling. Size distributions from samples close to the stack were found to have a log normal frequency distribution, but significant deviations from the log normal were found farther downwind. Several possible physical mechanisms are postulated as causes for this behavior. Plume dilution with background particles appears to be the most likely mechanism. The airborne sampling system is described, and electron micrographs of sampled particles are presented.
The most widely used process for the manufacture of pulp and paper in the United States is the Kraft process. Two components of this process, the recovery furnace and the lime kiln, emit particles which contribute to air pollution problems in the areas where plants are located. Although most Kraft plants control over 90% by weight of their particulate emissions, the remaining particles in the submicrometer size range, can constitute a significant air pollution tion problem. It is well known that particles in this size range contribute to loss of visibility and possible health effects. The contribution of a particular atmospheric aerosol to visibility and health problems can only be evaluated and understood through detailed study of aerosol physical and chemical properties as it exists in the atmosphere. Objectives
The objectives of this research were (1) the study of certain physical properties of Kraft aerosols, such as size distribution and morphology, (2) chemical identification of as many of the various chemical species as possible and to relate, if possible, chemical identity to visual identification through morphological characteristics, and (3) the development of sampling and analytical techniques which would permit the accomplishment of tasks 1 and 2. The results of task 1 and portions of task 3 are described here. The results of task 2, chemical identification and related analytical techniques, will be described in a subsequent paper. Experimental Methods Airborne Sampling Systems
A light aircraft equipped with a high-volume sampling system was the primary means of sampling. The airborne system served two functions. The primary one was the collection of particulate on a membrane filter, and the second was to indicate when the aircraft entered and exited the plume. June 1975
Volume 25, No. 6
The sampling probe protruded outside the window of the aircraft, roughly aligned with the airstream. It introduced the sampled air into the integrating nephelometer mounted inside the aircraft, and the nephelometer served to indicate entry and exit from the plume. Not only did this instrument give a more precise indication than could be made visually, but it also located the plume at distances far downwind, or under lighting conditions impossible for the naked eye to discern. From the nephelometer the sampled stream passed through approximately 2 ft of tubing to a standard 4-in. filter holder modified to accommodate a 90 mm Nuclepore or Millipore filter. The flow next passed through the orifice flowmeter, then through the high-volume blower powered by a xk hp electric motor. The blower delivered a flow rate of approximately 5 cfm through a 1 /xm. pore size Nuclepore filter. The power supply for the system was a 12 V storage battery and a DC-AC inverter. Flow rate, and thus velocity at the filter face, was controlled with a powerstat in the circuit. The sampling probe was equipped with a fixed-inlet nozzle which yielded sampling velocity/airspeed ratios from 0.4 to 1.0 depending on filter pore size. Since the typical duration of a single pass through the plume was a matter of seconds, many passes were normally required to obtain a sample sufficiently dense for analysis. It was possible to obtain the total sample volume gathered during the multiple passes by timing each pass with a stopwatch and controlling flow to the filter face by use of the valving and filter bypass built into the system. At the end of each sampling period, exposed filters were removed from the holder and placed in containers to prevent contamination during transfer to the laboratory. Nuclepore Filters Both Nuclepore and Millipore filters were considered for use in this project. Nuclepore filters were selected as the primary sampling filters, because (1) flow rate through the Nuclepore filters was approximately four times greater 617
than through the Millipore at a given pore size and pressure drop, and (2) they were considerably easier to prepare for transmission electron microscopy than Millipore. Nuclepore filters also are an excellent substrate for scanning electron microscopy. Nuclepore filters have poorer retention characteristics than Millipore, but this disadvantage is offset by the fact that their retention efficiency is quantifiable and therefore particle losses can be estimated.
Downloaded by [Stony Brook University] at 16:17 29 October 2014
Sampling Errors
There are several possible sources of sampling error associated with the system described. These are (1) errors due to anisokinetic sampling, (2) particle deposition in the sampling train, (3) angularity of flow relative to the sampling probe due to propeller wash, (4) filter retention characteristics, and (5) filter contamination by engine exhaust. Examination of these errors led to the conclusion that the first three error sources listed above would not seriously affect the results and that appropriate corrections could be made for the fourth. The fifth source was eliminated as a possibility at the beginning of the sampling program. Errors due to anisokinetic sampling, particle deposition, and angularity of flow were found to be negligible, based on the work of Nelson,1 Sehmel,2 and May.3 This was principally due to the fact that over 95% of the sampled particles had diameters less than 1 jum. Correction factors were applied for filter retention efficiency as a function of particle size based on equations given by Spumy, et al.4 and as modified by the data of Spumy and Madelaine.5 The possibility of filter contamination by aircraft engine exhaust was eliminated by neutron activation analysis of samples taken at the engine exhaust stack and at the normal sampling point. The latter sample was found to be completely free of bromine and vanadium, which were found in relatively heavy concentrations on the former. Stack Sampling
Several stack samples were taken at the Kraft mill in order to make a comparison of particle characteristics in the stack with those of atmospheric samples. Nuclepore filters were placed in a standard 47 mm filter holder fitted to a %-in. stainless steel probe with a ^-in. nozzle. Flow through the sampling system was induced by an air eductor in the plant high-pressure air stream. With this system it was possible to achieve the velocity necessary for isokinetic sampling. The sampling system was allowed to remain in the stream several minutes before sampling was begun, in order to bring the probe up to temperature and minimize condensation problems. Analytical Methods Electron Microscopy
The principal analytical tool in this research was the electron microscope. Light microscopy was not adequate for these studies, since the practical limit of resolution with such instruments is about 0.5 /j.m, greater than the median size of the particles being studied. Electron microscopes, on the other hand, have resolution capabilities on the order of angstroms (10~4 urn), and their much greater magnification permits accurate sizing and morphological studies of submicrometer particles.
morphological studies. The scanning electron microscope (SEM) was used for morphological studies. The latter instrument forms its images by bombarding the specimen with an electron beam, causing the emission of secondary electrons which are, in turn, used to form an image. By virtue of the beam geometry the image formed has a threedimensional perspective. This added perspective was very useful in supporting and confirming conclusions reached concerning particle morphology with the TEM. Two transmission instruments were used, a Hitachi HU11B-3, a 125 kV instrument with magnification capability up to 150,000X in the Department of Mechanical and Metallurgical Engineering at Oregon State University, and a Phillips 300 instrument in the University's Botany Department. The latter was a 100 kV instrument with magnification up to 250,000X. The scanning instrument used was a Cambridge Stereoscan Mark II at the University of Washington. Specimen Replication
Since transmission microscopy requires a very thin substrate for transmission, a specimen replication process is required. In this experiment the technique of Frank, et al.6 was adopted. This consisted of the vacuum evaporation of a thin (1000 A) silicon monoxide film at vertical incidence on the exposed filter surface. A small piece of this preparation was placed on an electron microscope speci-
2.0
1.0 0.8 £0.6 a. or 0.4
o Filter A Filter D Filter v Filter
019, 0.35 mi D.W 020, 1.7 mi D.W. 021, 2.7 mi D.W. 022, 4.7 mi D.W.
0.2
0.1
1
5 10 20 3040 60
80 90 95 98 99
99.9
Percent of particles less than the stated size Figure 1. Particle size distribution at selected downwind points in a Kraft mill plume (Filters 019, 020, 021 and 022).
men grid, and the filter material dissolved by placing the specimen on a small piece of foam saturated in a chloroform bath in a covered petri dish. A multiple wash of several minutes each was necessary to dissolve the filter completely. Following the filter dissolution process, the specimen grid with the thin film of silicon monoxide entrapping the particles was ready for microscopic examination. Specimen preparation for scanning microscopy was considerably simpler. It was not necessary to remove the filter, since transmission was not required. It was necessary only to evaporate a thin film of conducting materials (in this case, palladium) on the specimen surface in order to minimize charging of the particles.
Instrument Description
Two types of electron microscopes were used in this research. The transmission electron microscope (TEM), which forms images by transmission of electrons through the specimen, was used for high magnification, sizing, and 618
Particle Sizing
In this paper particle size is defined in terms of equivalent area diameter. This diameter is defined as the diameter of a circle with a projected surface area equal to that of Journal of the Air Pollution Control Association
Downloaded by [Stony Brook University] at 16:17 29 October 2014
the particle. Sizing was accomplished by use of a Porton graticule adapted from one used in a light microscope and properly calibrated. Following the practice of industrial hygienists as reported by Corn7 approximately 200 particles per sample were sized. In some cases many more (up to 1200) were sized to confirm the results. Size distributions determined in the manner described were identified for a number of filters in several downwind sequences. Each sample (filter) in a sequence was taken by making multiple crosswind passes through the plume at a given distance downwind. The initial point in a sequence ranged from 0.2 to 0.4 mi. downwind of the stack. Differences in frequency distribution of particles and median size of particles were noted with distance downwind. A means was required to establish that samples taken close in to the stack were from different populations than those taken downwind from the stack. One method of accomplishing this involved comparison of measures of central tendency, in this case the medians, of the size distributions. The statistical test used for such comparison was the "t" test. Discussion of this testing procedure can be found in any elementary text on statistical methods. The median of the distribution, m, and the geometric deviation of the distribution, ag, were determined graphically from the log probability plots. It was assumed that the medians and geometric deviations were also the optimum estimators of the sample means and standard error of the
2.0
1.0 0.8
o Filter 046, 0.3 mi D.W. A Filter 047, 1.3 mi D.W. a Filter 048, 2.8 mi D.W.
E 0.6 g0.4 CO
0.2 0.1 1
5 10 20 3040 60 80 90 95 98 99 99.8 Percent of particles less than the stated size
Figure 2. Particle size distribution at selected downwind points in a Kraft mill plume (Filters 046, 047 and 048).
means. This is a satisfactory approximation if the sample sizes are large, which was the factor in all cases. The large samples also allowed the test to be applied to distributions other than log normal, since, by the Central Limit Theorem, the distribution of their means approaches normality. Results
The numerical size distribution of particles in a Kraft mill plume was determined. Numerical size distribution (cumulative) is defined as the fraction of particles less than a stated size and is given as a function of size. The size measured for these distributions was the equivalent area diameter defined above and will be referred to hereafter as particle diameter or size. Size distributions for two downwind sequences are presented in Figures 1 and 2. Both were taken from the plume of the same mill, the American Can Company, Halsey, Ore. At this plant the effluent from both the recovery furnace June 1975
Volume 25, No. 6
and the lime kiln is combined and emitted to the atmosphere from a stack 300 ft above the ground. The distributions were obtained at several points in each plume, from as near to the stack as possible, to points several miles downwind. The curves reveal marked changes in the size distribution of the particles with distance downwind. There is a significant decrease in median particle size, and the frequency distribution deviates from log normal. The statistical significance of the decrease in median particle size is established by data in Table I, where the difference in medians is tested. The null hypothesis formulated for each case was that "the count median particle size in the plume was invariant with distance downwind from the source." This hypothesis was rejected for all cases at a significance level of less than 0.1%, as indicated in Table I. In two cases there was evidence that median particle size achieved a minimum at about 2 mi downwind, increasing somewhat at points further downwind. This increase in size was statistically significant at the 0.1% level for filter pairs 021 and 022 taken at 2.7 and 4.7 mi downwind, respectively, and at the 1% level for 047 and 048 taken at 1.3 and 2.8 mi downwind, respectively. In all sequences it appears that the sampling distribution obtained close to the stack can be well represented by a log normal distribution, as indicated by the curve fit on a log probability scale. All these samples were taken within 400 to 800 yards of the source where the plume was quite concentrated. Downwind of the source the size distribution curves were distinctly nonlinear, possibly indicating a heterogeneous size distribution. There were several possible physical mechanisms which could have contributed to the observed changes in size distribution with distance downwind. Among these are (1) a breakup of agglomerated particles after exit from the stack, (2) loss of water from particle droplets due to evaporation, (3) dilution of the plume with particles of a smaller average size, and (4) downwind agglomeration of particles, resulting in a larger percentage of particles in the high end of the size range. Most probably all of these mechanisms were in operation to a greater or lesser degree. The particle morphology in the electron micrographs of Figures 3 through 6 lends credence to these observations. Figure 3 is a scanning electron micrograph of a sample taken in the stack of the Kraft mill. Particle agglomeration is evident, probably due to charging by the electrostatic precipitator through which they have just passed. Figure 4 is a micrograph of a sample taken in the plume about 0.25 mi downwind from the stack. Clearly there has been deagglomeration at this point, and this process may continue downwind. Figures 5 and 6 are micrographs of samples taken by flying up the centerline of the plume from a point about 5 mi downwind to a point about 0.25 mi from the stack. They are representative of all the different types of particles observed during this experimentation. The large flaky particles proved to be calcium carbonate, and the larger smeared particles were shown to be hydrated sodium sulfate. The latter particles appeared to be in a semi-fluid state when sampled and flattened out when they impacted the filter. The large smeared particles were quite often in evidence on the upwind samples. However, downwind samples were dominated by smaller, more regularly shaped particles. This could be the result of evaporation, with the smaller particles retaining their shape due to greater surface tension. Considerable additional research is required to establish definitely the effects of relative humidity in the plume on particle size. The possible agglomeration of particles downwind in the plume is indicated by both the nonlinearity of the size distributions and the increase in median particle size after a 619
Downloaded by [Stony Brook University] at 16:17 29 October 2014
Table I.
Significance levels for changes in count median particle size with distance downwind based on the t statistic.
Downwind Filter Sequence
Distance Downwind from the Stack (mi)
Count Median Particle Size
019 020
0.35 1.7
021 022
Geometric Deviation (erg)
Particle Counts (N)
0.36 0.20
1.83 1.90
2.7 4.7
0.16 0.18
034 035
0.35 1.5
036 046 047 048
t Statistic
Significance Level (%)
197) 304/
10.4
Journal of the Air Pollution Control Association Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/uawm16
Particle Size Distributions of Kraft Paper Mill Aerosols Obtained by Airborne Sampling a
Frank E. Augustine & Richard W. Boubel a
b
The Aerospace Corporation , El Segundo , California , USA
b
Oregon State University , Corvallis , Oregon , USA Published online: 13 Mar 2012.
To cite this article: Frank E. Augustine & Richard W. Boubel (1975) Particle Size Distributions of Kraft Paper Mill Aerosols Obtained by Airborne Sampling, Journal of the Air Pollution Control Association, 25:6, 617-621, DOI: 10.1080/00022470.1975.10470117 To link to this article: http://dx.doi.org/10.1080/00022470.1975.10470117
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions
Particle Size Distributions of Kraft Paper Mill Aerosols Obtained by Airborne Sampling
Frank E. Augustine The Aerospace Corporation, El Segundo, California
Richard W. Boubel
Downloaded by [Stony Brook University] at 16:17 29 October 2014
Oregon State University, Corvallis, Oregon
Size distributions of particles at several downwind points in a Kraft paper mill plume have been determined by means of airborne sampling. Size distributions from samples close to the stack were found to have a log normal frequency distribution, but significant deviations from the log normal were found farther downwind. Several possible physical mechanisms are postulated as causes for this behavior. Plume dilution with background particles appears to be the most likely mechanism. The airborne sampling system is described, and electron micrographs of sampled particles are presented.
The most widely used process for the manufacture of pulp and paper in the United States is the Kraft process. Two components of this process, the recovery furnace and the lime kiln, emit particles which contribute to air pollution problems in the areas where plants are located. Although most Kraft plants control over 90% by weight of their particulate emissions, the remaining particles in the submicrometer size range, can constitute a significant air pollution tion problem. It is well known that particles in this size range contribute to loss of visibility and possible health effects. The contribution of a particular atmospheric aerosol to visibility and health problems can only be evaluated and understood through detailed study of aerosol physical and chemical properties as it exists in the atmosphere. Objectives
The objectives of this research were (1) the study of certain physical properties of Kraft aerosols, such as size distribution and morphology, (2) chemical identification of as many of the various chemical species as possible and to relate, if possible, chemical identity to visual identification through morphological characteristics, and (3) the development of sampling and analytical techniques which would permit the accomplishment of tasks 1 and 2. The results of task 1 and portions of task 3 are described here. The results of task 2, chemical identification and related analytical techniques, will be described in a subsequent paper. Experimental Methods Airborne Sampling Systems
A light aircraft equipped with a high-volume sampling system was the primary means of sampling. The airborne system served two functions. The primary one was the collection of particulate on a membrane filter, and the second was to indicate when the aircraft entered and exited the plume. June 1975
Volume 25, No. 6
The sampling probe protruded outside the window of the aircraft, roughly aligned with the airstream. It introduced the sampled air into the integrating nephelometer mounted inside the aircraft, and the nephelometer served to indicate entry and exit from the plume. Not only did this instrument give a more precise indication than could be made visually, but it also located the plume at distances far downwind, or under lighting conditions impossible for the naked eye to discern. From the nephelometer the sampled stream passed through approximately 2 ft of tubing to a standard 4-in. filter holder modified to accommodate a 90 mm Nuclepore or Millipore filter. The flow next passed through the orifice flowmeter, then through the high-volume blower powered by a xk hp electric motor. The blower delivered a flow rate of approximately 5 cfm through a 1 /xm. pore size Nuclepore filter. The power supply for the system was a 12 V storage battery and a DC-AC inverter. Flow rate, and thus velocity at the filter face, was controlled with a powerstat in the circuit. The sampling probe was equipped with a fixed-inlet nozzle which yielded sampling velocity/airspeed ratios from 0.4 to 1.0 depending on filter pore size. Since the typical duration of a single pass through the plume was a matter of seconds, many passes were normally required to obtain a sample sufficiently dense for analysis. It was possible to obtain the total sample volume gathered during the multiple passes by timing each pass with a stopwatch and controlling flow to the filter face by use of the valving and filter bypass built into the system. At the end of each sampling period, exposed filters were removed from the holder and placed in containers to prevent contamination during transfer to the laboratory. Nuclepore Filters Both Nuclepore and Millipore filters were considered for use in this project. Nuclepore filters were selected as the primary sampling filters, because (1) flow rate through the Nuclepore filters was approximately four times greater 617
than through the Millipore at a given pore size and pressure drop, and (2) they were considerably easier to prepare for transmission electron microscopy than Millipore. Nuclepore filters also are an excellent substrate for scanning electron microscopy. Nuclepore filters have poorer retention characteristics than Millipore, but this disadvantage is offset by the fact that their retention efficiency is quantifiable and therefore particle losses can be estimated.
Downloaded by [Stony Brook University] at 16:17 29 October 2014
Sampling Errors
There are several possible sources of sampling error associated with the system described. These are (1) errors due to anisokinetic sampling, (2) particle deposition in the sampling train, (3) angularity of flow relative to the sampling probe due to propeller wash, (4) filter retention characteristics, and (5) filter contamination by engine exhaust. Examination of these errors led to the conclusion that the first three error sources listed above would not seriously affect the results and that appropriate corrections could be made for the fourth. The fifth source was eliminated as a possibility at the beginning of the sampling program. Errors due to anisokinetic sampling, particle deposition, and angularity of flow were found to be negligible, based on the work of Nelson,1 Sehmel,2 and May.3 This was principally due to the fact that over 95% of the sampled particles had diameters less than 1 jum. Correction factors were applied for filter retention efficiency as a function of particle size based on equations given by Spumy, et al.4 and as modified by the data of Spumy and Madelaine.5 The possibility of filter contamination by aircraft engine exhaust was eliminated by neutron activation analysis of samples taken at the engine exhaust stack and at the normal sampling point. The latter sample was found to be completely free of bromine and vanadium, which were found in relatively heavy concentrations on the former. Stack Sampling
Several stack samples were taken at the Kraft mill in order to make a comparison of particle characteristics in the stack with those of atmospheric samples. Nuclepore filters were placed in a standard 47 mm filter holder fitted to a %-in. stainless steel probe with a ^-in. nozzle. Flow through the sampling system was induced by an air eductor in the plant high-pressure air stream. With this system it was possible to achieve the velocity necessary for isokinetic sampling. The sampling system was allowed to remain in the stream several minutes before sampling was begun, in order to bring the probe up to temperature and minimize condensation problems. Analytical Methods Electron Microscopy
The principal analytical tool in this research was the electron microscope. Light microscopy was not adequate for these studies, since the practical limit of resolution with such instruments is about 0.5 /j.m, greater than the median size of the particles being studied. Electron microscopes, on the other hand, have resolution capabilities on the order of angstroms (10~4 urn), and their much greater magnification permits accurate sizing and morphological studies of submicrometer particles.
morphological studies. The scanning electron microscope (SEM) was used for morphological studies. The latter instrument forms its images by bombarding the specimen with an electron beam, causing the emission of secondary electrons which are, in turn, used to form an image. By virtue of the beam geometry the image formed has a threedimensional perspective. This added perspective was very useful in supporting and confirming conclusions reached concerning particle morphology with the TEM. Two transmission instruments were used, a Hitachi HU11B-3, a 125 kV instrument with magnification capability up to 150,000X in the Department of Mechanical and Metallurgical Engineering at Oregon State University, and a Phillips 300 instrument in the University's Botany Department. The latter was a 100 kV instrument with magnification up to 250,000X. The scanning instrument used was a Cambridge Stereoscan Mark II at the University of Washington. Specimen Replication
Since transmission microscopy requires a very thin substrate for transmission, a specimen replication process is required. In this experiment the technique of Frank, et al.6 was adopted. This consisted of the vacuum evaporation of a thin (1000 A) silicon monoxide film at vertical incidence on the exposed filter surface. A small piece of this preparation was placed on an electron microscope speci-
2.0
1.0 0.8 £0.6 a. or 0.4
o Filter A Filter D Filter v Filter
019, 0.35 mi D.W 020, 1.7 mi D.W. 021, 2.7 mi D.W. 022, 4.7 mi D.W.
0.2
0.1
1
5 10 20 3040 60
80 90 95 98 99
99.9
Percent of particles less than the stated size Figure 1. Particle size distribution at selected downwind points in a Kraft mill plume (Filters 019, 020, 021 and 022).
men grid, and the filter material dissolved by placing the specimen on a small piece of foam saturated in a chloroform bath in a covered petri dish. A multiple wash of several minutes each was necessary to dissolve the filter completely. Following the filter dissolution process, the specimen grid with the thin film of silicon monoxide entrapping the particles was ready for microscopic examination. Specimen preparation for scanning microscopy was considerably simpler. It was not necessary to remove the filter, since transmission was not required. It was necessary only to evaporate a thin film of conducting materials (in this case, palladium) on the specimen surface in order to minimize charging of the particles.
Instrument Description
Two types of electron microscopes were used in this research. The transmission electron microscope (TEM), which forms images by transmission of electrons through the specimen, was used for high magnification, sizing, and 618
Particle Sizing
In this paper particle size is defined in terms of equivalent area diameter. This diameter is defined as the diameter of a circle with a projected surface area equal to that of Journal of the Air Pollution Control Association
Downloaded by [Stony Brook University] at 16:17 29 October 2014
the particle. Sizing was accomplished by use of a Porton graticule adapted from one used in a light microscope and properly calibrated. Following the practice of industrial hygienists as reported by Corn7 approximately 200 particles per sample were sized. In some cases many more (up to 1200) were sized to confirm the results. Size distributions determined in the manner described were identified for a number of filters in several downwind sequences. Each sample (filter) in a sequence was taken by making multiple crosswind passes through the plume at a given distance downwind. The initial point in a sequence ranged from 0.2 to 0.4 mi. downwind of the stack. Differences in frequency distribution of particles and median size of particles were noted with distance downwind. A means was required to establish that samples taken close in to the stack were from different populations than those taken downwind from the stack. One method of accomplishing this involved comparison of measures of central tendency, in this case the medians, of the size distributions. The statistical test used for such comparison was the "t" test. Discussion of this testing procedure can be found in any elementary text on statistical methods. The median of the distribution, m, and the geometric deviation of the distribution, ag, were determined graphically from the log probability plots. It was assumed that the medians and geometric deviations were also the optimum estimators of the sample means and standard error of the
2.0
1.0 0.8
o Filter 046, 0.3 mi D.W. A Filter 047, 1.3 mi D.W. a Filter 048, 2.8 mi D.W.
E 0.6 g0.4 CO
0.2 0.1 1
5 10 20 3040 60 80 90 95 98 99 99.8 Percent of particles less than the stated size
Figure 2. Particle size distribution at selected downwind points in a Kraft mill plume (Filters 046, 047 and 048).
means. This is a satisfactory approximation if the sample sizes are large, which was the factor in all cases. The large samples also allowed the test to be applied to distributions other than log normal, since, by the Central Limit Theorem, the distribution of their means approaches normality. Results
The numerical size distribution of particles in a Kraft mill plume was determined. Numerical size distribution (cumulative) is defined as the fraction of particles less than a stated size and is given as a function of size. The size measured for these distributions was the equivalent area diameter defined above and will be referred to hereafter as particle diameter or size. Size distributions for two downwind sequences are presented in Figures 1 and 2. Both were taken from the plume of the same mill, the American Can Company, Halsey, Ore. At this plant the effluent from both the recovery furnace June 1975
Volume 25, No. 6
and the lime kiln is combined and emitted to the atmosphere from a stack 300 ft above the ground. The distributions were obtained at several points in each plume, from as near to the stack as possible, to points several miles downwind. The curves reveal marked changes in the size distribution of the particles with distance downwind. There is a significant decrease in median particle size, and the frequency distribution deviates from log normal. The statistical significance of the decrease in median particle size is established by data in Table I, where the difference in medians is tested. The null hypothesis formulated for each case was that "the count median particle size in the plume was invariant with distance downwind from the source." This hypothesis was rejected for all cases at a significance level of less than 0.1%, as indicated in Table I. In two cases there was evidence that median particle size achieved a minimum at about 2 mi downwind, increasing somewhat at points further downwind. This increase in size was statistically significant at the 0.1% level for filter pairs 021 and 022 taken at 2.7 and 4.7 mi downwind, respectively, and at the 1% level for 047 and 048 taken at 1.3 and 2.8 mi downwind, respectively. In all sequences it appears that the sampling distribution obtained close to the stack can be well represented by a log normal distribution, as indicated by the curve fit on a log probability scale. All these samples were taken within 400 to 800 yards of the source where the plume was quite concentrated. Downwind of the source the size distribution curves were distinctly nonlinear, possibly indicating a heterogeneous size distribution. There were several possible physical mechanisms which could have contributed to the observed changes in size distribution with distance downwind. Among these are (1) a breakup of agglomerated particles after exit from the stack, (2) loss of water from particle droplets due to evaporation, (3) dilution of the plume with particles of a smaller average size, and (4) downwind agglomeration of particles, resulting in a larger percentage of particles in the high end of the size range. Most probably all of these mechanisms were in operation to a greater or lesser degree. The particle morphology in the electron micrographs of Figures 3 through 6 lends credence to these observations. Figure 3 is a scanning electron micrograph of a sample taken in the stack of the Kraft mill. Particle agglomeration is evident, probably due to charging by the electrostatic precipitator through which they have just passed. Figure 4 is a micrograph of a sample taken in the plume about 0.25 mi downwind from the stack. Clearly there has been deagglomeration at this point, and this process may continue downwind. Figures 5 and 6 are micrographs of samples taken by flying up the centerline of the plume from a point about 5 mi downwind to a point about 0.25 mi from the stack. They are representative of all the different types of particles observed during this experimentation. The large flaky particles proved to be calcium carbonate, and the larger smeared particles were shown to be hydrated sodium sulfate. The latter particles appeared to be in a semi-fluid state when sampled and flattened out when they impacted the filter. The large smeared particles were quite often in evidence on the upwind samples. However, downwind samples were dominated by smaller, more regularly shaped particles. This could be the result of evaporation, with the smaller particles retaining their shape due to greater surface tension. Considerable additional research is required to establish definitely the effects of relative humidity in the plume on particle size. The possible agglomeration of particles downwind in the plume is indicated by both the nonlinearity of the size distributions and the increase in median particle size after a 619
Downloaded by [Stony Brook University] at 16:17 29 October 2014
Table I.
Significance levels for changes in count median particle size with distance downwind based on the t statistic.
Downwind Filter Sequence
Distance Downwind from the Stack (mi)
Count Median Particle Size
019 020
0.35 1.7
021 022
Geometric Deviation (erg)
Particle Counts (N)
0.36 0.20
1.83 1.90
2.7 4.7
0.16 0.18
034 035
0.35 1.5
036 046 047 048
t Statistic
Significance Level (%)
197) 304/
10.4
