Module 4: Introduction to Aerosols


Hello, and welcome to the module Introduction
to Aerosols. My name is Gurumurthy Ramachandran, and I’m
a faculty member at the University of Minnesota School of Public Health in the Division of
Environmental Health Sciences. By the end of this module, learners should be
able to do the following: Define the term aerosol and other related terms,
interpret summary properties of particle
distributions, predict the relationship between aerosol sources and the resulting particle sizes,
and finally, describe various approaches to
measuring airborne particles. In everyday usage, the term aerosol refers to a
product such as a perfume, paint, or a
pharmaceutical delivered in the form of spray or mist from a container with a liquid and
a propellant inside it, as shown in this diagram. However, the word aerosol has a much broader
definition. We can define an aerosol as a suspension of
liquid or solid particles in a gas. The sizes of
aerosol particles span a wide range. They can be as small as one-thousandth of a
micrometer or 1 nm. To put this in perspective, 1 nanometer is
roughly two to five times larger than the size of
small gas molecules. At the upper end, particles can be as large as
one hundred to two hundred micrometers. There are a number of sources that generate
aerosol particles, both natural and manmade. This picture shows some of the sources of
naturally occurring aerosols, including volcanic
emissions on the right hand side, smoke from forest fires, volatile organic
chemical emissions from plants, and these
emissions then undergo chemical reactions smoke from forest fires, volatile organic
chemical emissions from plants, and these
emissions then undergo chemical reactions to form particles, dust from desert sand storms,
and emissions of dimethyl sulfide from the
plankton in oceans that then undergo further reactions to form sulfate particles or
sulfate aerosol. Aerosols are responsible for the beautiful red
sunsets that we have all seen, the annual forest
fires in the Western United States. They cause farther away hills to appear lighter
than nearer hills due to the scattering of light by
aerosol particles. Rainbows are observed due to the interaction of
sunlight with water droplets suspended in air, and of course, the smoke and ash from volcanic
eruptions as we have mentioned before. Aerosols are also created by human activities. For example, the very fine particles in fumes
that are generated during welding at high
temperatures, the much larger dust particles that are generated when we cut a piece of
concrete using a mechanical saw, the smoke
from industrial smokestacks, the fine mist that is generated when we spray a
field with pesticides from an airplane, the smoke
when we grill a piece of meat, and, of course, cigarette smoke. It is useful to get a sense of the range of particle
sizes of aerosols from different sources. As we discussed before, aerosol sizes can
range from one-thousandth of a micrometer or 1
nm to roughly 100 to 200 µm. So, this covers around five orders of magnitude To put this in perspective, we can relate this
order of magnitude range to the range of sizes
from a pea to a very large building. Windblown dusts are created by mechanical
processes such as erosion of rocks over
geological time scales, and these have particles in the range of roughly
1 µm to several hundred micrometers. Mining dusts are also created by mechanical
processes such as the crushing of ore bearing rocks, and these have roughly the
same size range: slightly less than 1 µm to
around 100 µm. The lengths of asbestos fibers are also between
1 µm to around 100 µm. Biological spores are also roughly in this size
range. The sizes of airborne bacteria cover a smaller
range from around half a micrometer to around
10 µm. What is commonly referred to as the urban
aerosol is made of emissions of particles from
vehicles of various kinds, emissions from industrial sources such as
power plants, and particles created by chemical reactions between gases such as
ozone and volatile organic chemicals. The range of these particles is from around 50
nm, or 0.05 µm, to about 5 µm. Metal smelting fumes range from around 10 nm
to around 500 nm. Vehicular exhaust, also a result of combustion
reaction, is roughly in the same size range. Viruses range from around 10 nm to 100 nm in
size. We commonly refer to particles less than 1 µm
or 1000 nm as fine particles. Particles less than
100 nm are called ultrafine. This is also the range of what we call
nanoparticles, although this term typically refers only to particles that have been intentionally
created in an engineering process. To get a perspective of these small sizes, it is
useful to remember that the width of the human
hair is about 70 to 75 micrometers. The wavelengths of colors in the visible
spectrum of light range from around 380 to 780
nm. In addition to size, particles come in a wide
variety of shapes, ranging from fibers where we
have the length of the particles much greater than the width of the fibers, to chain
agglomerates that are made of chains of small
particles. Particles can be spherical, in various crystalline
shapes. Pollen come in a variety of shapes with
interesting surface features. Water droplets are roughly spherical And all these shapes where the length and width
and height of the particles are roughly the same, although not exactly the same, are called
isometric, unlike fibers that have one dimension such as
the length much larger
than another such as the height or width. This slide shows the array of shapes associated
with nanoparticles with the same composition. All these particles, even though they have very
different shapes, have the same composition
chemically. They are all made of zinc oxide, and this
illustrates the range of shapes that
nanoparticles can take. Another important property of particles is their
density. Depending on their composition, particle density
can vary widely. Water droplets have a density of 1000 kg per
cubic meter. The metal lead has a density of 11,300 kg per
cubic meter. It is useful to compare these with the density of
air, which is roughly 1 kg per cubic meter. So water has a density that is almost 1000
times greater than that of air. The physical and chemical processes that
generate aerosol particles determine the range
of particle sizes. In mechanical processes, energy is supplied to
a bulk material to break it into smaller particles. For example, during cutting, blasting, or
crushing for solid particles, or spraying in the
case of liquid particles, the sizes of the particles depend on the ratio of
the amount of energy supplied and the energy of the adhesive bonds holding
the parent material together. Now we will watch three videos that show
aerosol generation from various mechanical
processes. The three processes that we will see now are
cutting off the concrete block, a coal mining
operation, and finally spray painting. In this video, the worker is using a saw to cut
through a piece of concrete and generating dust
in the process. Here, we see a diesel operated vehicle dumping
crushed coal onto a pile, so some of the
particles are mechanically generated coal dust, but we can also see the black diesel
combustion smoke being emitted by the vehicle. This smoke is made of much smaller particles
than the coal dust, and this is an example of a
mixed aerosol. In this video, we see a car being sprayed with a
liquid paint. The hose supplies high pressure air to the liquid
in the container and breaks it up into very small
spray droplets. Particles can also be created by molecular
processes such as those shown in this slide. Molecules of a vapor or gas can come together
by themselves to form small clusters in a
process called nucleation. This can happen when a gas that is saturated
with a vapor undergoes rapid cooling and, thus,
creates conditions for these nuclei to form. These nuclei are roughly the size of a
nanometer. An example is the formation of primary soot
particles from the tailpipe emissions of vehicles. Once these nuclei are formed, they can
continue to increase in size by colliding and
sticking to each other, a process called coagulation. They may
increase in size by condensation – that is
molecules of the vapor condense onto the surface of particles, or
decrease in size by evaporation, that is
molecules leave the particle surface. Thus once an aerosol is created, either
mechanically or by nucleation, its particles can
change in size by these mechanisms. These phenomena play an important role in the
exposure to aerosols that human beings face. Depending on where people are located with
respect to these phenomena, they would be exposed to different aerosols sizes and
concentrations. To summarize what we have learned so far,
chemical processes such as combustion, welding and those in nanoparticle reactors lead
to particle sizes less than 1 µm in size and frequently less than one-tenth of a
micrometer, or 100 nm, in size. These are
referred to as ultrafine or nanoparticles. and frequently less than one-tenth of a
micrometer, or 100 nm, in size. These are
referred to as ultrafine or nanoparticles. Mechanical processes on the other hand, such
as grinding or sanding, lead to much larger
sizes, ranging from 1 µm to several 100 µm. The concept of an equivalent diameter is very
important in describing particle sizes. It is easy to see that two spherical particles with
the same diameter, for example an air bubble and a lead particle, will behave very differently in
air. Likewise, particles with different shapes could
behave similarly under some conditions. How can we describe the sizes of particles
when they come in different shapes, and
chemical and physical properties? How can we describe the sizes of particles
when they come in different shapes, and
chemical and physical properties? In defining an equivalent diameter, we first
imagine a spherical particle and then we keep
some property unchanged, that is invariant between the real particle and the imaginary
sphere, and then we describe the real particle in
terms of the diameter of this imaginary sphere. In other words, the equivalent diameter is the
diameter of the sphere that has the same value of a physical property as that of the particle in
question. In this slide, we have an actual particle that is
irregularly shaped and is made of a material with
a density rho sub P. How can we describe the size of this irregularly
shaped particle? We can imagine several different physical
properties that we can use as the basis for
defining an equivalent diameter. For example, we can imagine a sphere with the
same projected surface area as our actual
particle. We think of the projected surface area of a
particle as the area of the shadow cast by the
particle on a projection screen, and then the diameter of this sphere is the
projected surface area equivalent diameter. Alternatively you can imagine a sphere with the
same total surface area as the surface area of
the actual particle The diameter of this sphere is the surface area
equivalent diameter. We can also imagine a sphere with the same
volume as our actual particle The diameter of this sphere is the volume
equivalent diameter. As a final example, we can imagine a spherical
water droplet with the same speed of falling through air, also known as its terminal settling
velocity, as our actual particle. The diameter of this water droplet is called the
aerodynamic equivalent diameter, or simply the
aerodynamic diameter. We will learn about the aerodynamic diameter in
greater detail in a later module. Next we discuss the topic of particle size
distributions. A monodisperse aerosol is one in which all the
particles are of the same size. It is almost never found in nature, and it is
artificially constructed or engineered in very highly controlled circumstances, and it’s very
difficult to make monodisperse aerosols. highly controlled circumstances, and it’s very
difficult to make monodisperse aerosols. A polydisperse aerosol, on the other hand, is
one in which there is a range of particle sizes Almost all natural and workplace aerosols are
polydisperse. Now, we will discuss the statistics of particle
size distributions. At a very basic level, if we have a collection of a
number of particles with different diameters, then we can calculate the mean and standard
deviation of the particle sizes. The arithmetic mean diameter is a measure of
the central tendency of the distribution of
particle sizes. If N is the total number of particles in a sample,
or collection of particles, and n sub i is the
number of particles with size d sub i, then the arithmetic mean is calculated using
this formula, where we multiply each size d sub
i by the number of particles with that size n sub i, and then summing this over all particle
sizes. This sum of the products is then divided by the
total number of particles, N, or summation of n sub i, to yield the average diameter, or the
arithmetic mean diameter, denoted by d hat. The median diameter is the value below which
50% of the particle diameters are, and above
which 50% of the particles lie. It is, thus, the 50th percentile of particle
diameters. The standard deviation of particle sizes is
calculated using this formula. We find the difference between each particle
size from the arithmetic mean diameter, square
the difference, multiply this by the number of particles of that size, n sub i, and
then sum this across all the size ranges, and take the sum and divide this by the total
number of particles minus one. After that, we take the square root of this
quantity. The standard deviation is a measure of the
variability in particle size. Let us consider an aerosol where we have been
able to determine some equivalent diameter for a large number of particles using some
aerosol measurement instrument. The instrument in this example counts and
classifies the particles into seven size ranges or
size bins, as shown here: 10 to 50 nm, 50 to 80 nm, and so on all the way
up to 890 to 1260 nanometers. 10 to 50 nm, 50 to 80 nm, and so on all the way
up to 890 to 1260 nanometers. The total number of particles counted is 832. We want to calculate the arithmetic mean,
median, and standard deviation of this collection
of particles, but from this table, it looks like we only know
the ranges within which the particles sizes lie,
and not each individual particle’s size. Therefore, we will make a simplifying
assumption. We will assume that within each
size bin, all the particles are of the same size. Therefore, we will make a simplifying
assumption. We will assume that within each
size bin, all the particles are of the same size. We will further assume that this is the midpoint
of the size bin. Although this is an assumption, it is not too far
from the truth, and so it is a justifiable
assumption. We can now add a third column to our table that
lists the midpoint diameter for each size range. This allows us to calculate the various terms
needed to calculate the arithmetic average. Now, the fourth column is obtained by
multiplying the second column n sub i, or
particle count, with the third column d sub i, or the midpoint of the size interval to obtain the
product n sub i times d sub i, and we do this for
each size range. At the bottom of this column, we can add up all
the n sub i times d sub i terms. We can now go to the next slide, where we use
this information to calculate the arithmetic mean
d hat. Using formula that I described earlier, the
arithmetic mean can be calculated as 142 nm. The median diameter can also be calculated
very simply. As we see, there are 832 particles in all, and the
median diameter is the middle particle. If we can line up these 832 particles in
ascending order, then in fact there are two
middle particles, the 416th and 417th particle. So we should take the average of these two
middle particles, however, since we don’t know the sizes of the
individual particles, we can at best say that the
median is somewhere between 50 and 80 nm. Next, we have to calculate the standard
deviation. We can now look at the fifth column. This
column shows the product of n sub i and (d sub
i-d hat) squared. At the bottom of this column, we can add up all
these product terms. Now we go to the next slide again and look
where we use this summation in the formula for
the standard deviation. We take the summation, divide it by N-1 or
832-1, and then take the square root of this
whole quantity. This gives us a standard deviation of 186 nm. This gives us a standard deviation of 186 nm. While the calculations we did for the mean,
median, and standard deviation are useful to some extent, they do not give us a feel for the
shape of the distribution. We can construct a histogram of these data as
shown in the figure. Here, the particle diameter is on the horizontal x
axis, and the particle count is on the vertical y
axis. This histogram gives us a better sense of how
the particles are distributed among the various size ranges, that is, the shape of the
distribution. However, this type of a plot has an inherent
limitation in that the shape of the histogram depends very much on the size
ranges that we choose. For example, we could combine the ranges 140
– 270 nm and 270 – 560 nm to form a new range
140 – 560 nm with 96 + 53, or 149 particles. For example, we could combine the ranges 140
– 270 nm and 270 – 560 nm to form a new range
140 – 560 nm with 96 + 53, or 149 particles. I’m showing the original and the modified
histogram here, and we see the height of this
new size interval that we have created is I’m showing the original and the modified
histogram here, and we see the height of this
new size interval that we have created is much greater than either of the original two size
intervals. Our histogram’s shape now looks different. The shape of our distribution should not depend
on the size intervals in our instrument; it should
not depend on the instrument design. This is not acceptable. To avoid this kind of a problem, we can calculate
the particle count per nanometer as shown in
the fourth column. We are dividing the count in each size range by
the width of that size range, so, we end up with
a count per micrometer column. So, this is our second try at drawing a useful
histogram. The advantage of this type of histogram is that
the interval heights are independent of size
interval. Additionally, since the height of each block or
rectangle is n sub i divided by delta d, which is
the interval size, and the width is the interval size delta d, the area of each block of the
histogram is equal to the number of particles, or
n sub i, in that size range. Therefore, the total area of all the blocks of the
histogram is equal to the total number of
particles, that is, the sum of n sub i, or N. . Well, I am still not satisfied. One drawback to
this type of a histogram is that the heights of the intervals are still dependent on the number of particles collected in that
particular sample. So, if we had collected twice the number of
particles, but with the same distribution in the
various bins, as shown in the histogram on the bottom, the
heights of our intervals would have been
doubled, leaving the shape unchanged. So, let us make one more adjustment. We can calculate the fraction of particles per
nanometer in each size interval instead of the
absolute number of particles per nanometer. The fifth column in this table shows the fraction
of particles per nanometer in each size interval. We obtain this by dividing the fourth column,
which shows the count per nanometer, by the
total number of particles, that is, 832. So, now let us try the histogram one more time. In this histogram, the area of each rectangular
block is equal to the fraction of particles in that
size interval. Therefore, the total area of all the blocks
combined is equal to 1.0. This histogram can be said to be independent of
the type of sampling instrument we use, that defines the size of our bins, and also the
number of particles we happened to collect. It is more representative of the distribution of
particles in the environment, independent of our
instrument and sampling conditions. If we want, we can do one more thing. We can draw a smooth curve through the
midpoints of the tops of the rectangular blocks
to obtain a particle size distribution curve. This is an approximation to what is called the
probability density function. Just like the sums of the areas of all the
rectangles in the histogram is 1.0, Just like the sums of the areas of all the
rectangles in the histogram is 1.0, the area under the curve of the probability
density function is also 1.0. We can now take a closer look at this
histogram. We see that the histogram has a
skewed shape. Particles appeared to be distributed in a skewed
manner, and there are more particles in the
smaller size ranges than in the larger size ranges, and it is important to note that the x
axis, the horizontal axis, is a linear scale. However, if we change the horizontal axis from a
linear to a logarithmic scale, the particle
distribution appears to be more symmetric and we can fit a symmetrical looking curve to
the tops of the histogram. This brings us to what we call lognormal
distributions. Aerosol size distributions are seldom
symmetrical. They are typically positively
skewed with a long tail to the right. While the frequency distributions of particle
sizes have a skewed shape, the log of the particle sizes often have a
symmetrical, or Gaussian, or normal
distribution. Thus, the particle sizes are said to be log-
normally distributed. Thus, the particle sizes are said to be log-
normally distributed. The lognormal distribution is good way to
describe particle size distributions in workplaces
as well as in ambient environments. The statistics of lognormal distributions are very
similar to what we have done before. The lognormal distribution is described by the
geometric mean and geometric standard
deviation. This is analogous to the arithmetic mean and
standard deviation that we have learned to
calculate earlier. The geometric mean diameter, d sub g, can be
calculated as shown in this formula. You can see that this is similar to the formula for
the arithmetic mean except that the diameter is
replaced with the log of diameter. The geometric standard deviation, sigma sub g,
is calculated as shown in this formula. You can see that this is similar to the formula for
the standard deviation, except that the diameter
is replaced with the log of diameter. We can further modify this equation using one of
the rules of logarithms, that is, that the difference between the logarithms of two
quantities is the logarithm of the ratio of those
two quantities.
0:30:20.066,0:30:20.000
Log di minus log dg is equal to the log of di
divided by dg. Log of di minus log of dg is equal to the log of di
divided by dg. Thus, both of these expressions are similar to
that for the arithmetic mean and standard deviation, except that the diameter is
replaced with the log of diameter. The geometric mean and geometric standard
deviation completely describe a lognormal
distribution. The geometric mean is a central measure of the
size of the aerosol and the geometric standard deviation is a measure of the variability
of particle sizes in the aerosol. While the geometric mean has units of diameter,
the geometric standard deviation is a ratio of two
diameters, and, therefore, it is dimensionless. It cannot take a value of 0 since particle size
cannot be zero. It cannot take a value of 0 since the particle size
cannot be zero. We can use what we just learned to calculate
the geometric mean and geometric standard
deviation for our data set. The first three columns are the same as before.
The third column shows the midpoints of the
size intervals, but since for lognormal calculations we need to
use the log of diameter, the fourth column is the
logarithm of the midpoint diameters, or log (di). The fifth column is obtained by multiplying the
second column n sub i with the fourth column log (di) to obtain the product n sub i times log
(di). At the bottom of this column, we can add up all
the n sub i times log (di) across all the size
bins. We can now go to the next slide where we use
this information to calculate log of (d sub g), and for our data set, we calculate log of (d sub
g) to be equal to 1.96. Then, we can exponentiate it, or, in other words,
calculate ten raised to the power of 1.96, and
this comes out to be 92. So our geometric mean diameter in 92 nm.
We now go back to the previous slide and look
at the sixth column. This column shows the product of ni and (log di-
log dg) squared, and we calculate this for each
size interval. At the bottom of this column, we obtain the sum
of all these terms across each size interval. We now go back to the next slide where we use
this summation in the formula for the standard
deviation as shown below. We take the summation, divide it by N-1, or
832-1, and then take the square root of this
quantity. This gives us a standard deviation of 2.29. This gives us a standard deviation of 2.29. At this point, we should remember that for this
very data set, we had calculated an arithmetic
mean of 142 nm, and now we see that the geometric mean is 92
nm. This is a defining characteristic of lognormal
distributions that are skewed, in that their arithmetic means are always greater
than their geometric means We can also a plot the cumulative distribution of
the particles Here, the first two columns in the table are the
same as in previous slides. and the third column shows the cumulative
count up to a given particle size. So, we can
read this column in the following way: there are 120 particles less than 50 nm; there
are 120 plus 380 or 500 particles less than 80 nm; there are 120 plus 380 plus 146 or 646
particles less than 140 nm and so on. Finally there are 832 particles, that is, all the
particles, are less than 1260 nm. We can now construct a fourth column which
shows the cumulative percentage. Here, 14% of the particles, that is 120 out of the
832 particles, are less than 50 nm, 60% of the particles are less than 80 nm and so
this corresponds to the 500 out of the 832
particles, and finally 100% of the particles, all 832
particles, are less than 1260 nm. These data can be plotted as shown in the figure
on this slide. On the horizontal axis, we again have particle
diameter on a linear scale, and the vertical axis contains the cumulative percentage, again on a
linear scale. contains the cumulative percentage, again on a
linear scale. When the x axis is a linear scale, the
cumulative distribution rises and then levels off
at 100%. Nothing remarkable. When the x axis is a linear scale, the
cumulative distribution rises and then levels off
at 100%. Nothing remarkable. When the x axis is a linear scale, the
cumulative distribution rises and then levels off
at 100%. Nothing remarkable. This is to be expected. However, there is a
different way of plotting the same data. We can plot the diameter on a logarithmic scale
and the cumulative fraction on a probability
scale. This is called a log probability plot. The figure on this slide is an example of the log
probability plot. The particle size on the vertical
axis is on a logarithmic scale. The particle sizes go from 10 to 100 to 1000 to
10000 nm in equal intervals. The horizontal axis
is on a probability scale, and the intervals on this scale are wider at either
end and more compressed in the middle of the
range. The lower end does not go to zero but
approaches zero, and the upper end does not
reach 100% but approaches 100%. If we plot the cumulative fraction data on such a
graph, the data approximately fall on a straight
line. We can draw a best fit line through the data
points. The fact that all the data fall on a straight line
indicates that the data come from a log normal
distribution. That is the characteristic of the log probability
plot, that if we have a log normal distribution, then the cumulative percentage plot will fall on a
straight line. From this plot, we can read off important
statistical parameters of the log normal
distribution. The median diameter is at the 50th percentile,
that is 50% of the particles are less than this
diameter, and this corresponds to the 50th percentage point on the cumulative
percentage axis.
0:38:21.066,0:38:21.000
This corresponds to the geometric mean
diameter. This corresponds to the geometric mean
diameter. The geometric standard deviation, which is a
measure of the variability or the spread of the distribution, is defined as the ratio of the 84th
percentile to the 50th percentile. For this particular data set, the geometric mean
diameter, which corresponds to the 50th
cumulative percentage point, is 95 nm, as shown here, and the geometric standard
deviation is the ratio of the 84th percentile
diameter to the 50th percentile diameter. This is equal to 240 nm divided by 95 nm, and
this is equal to 2.5. So, the geometric standard
deviation is 2.5 in this case. This is equal to 240 nm divided by 95 nm, and
this is equal to 2.5. So, the geometric standard
deviation is 2.5 in this case. If you remember, we had also calculated the
geometric mean and geometric standard
deviation using a table of calulations. We obtained a value of 92 nm for the geometric
mean and 2.29 for the geometric standard
deviation. As we see, the two sets of values, one obtained
through tabular calculations and the other one obtained graphically, are very similar
to each other, as we should expect. But the graphical method is far simpler,
provided, of course, that you have a log
probability graph on a spreadsheet. We had discussed earlier that the geometric
standard deviation is a dimensionless number
as it is the ratio of two diameters. In the graphical method, we determine it by the
ratio of the 84th to the 50th percentile. Since the 84th percentile can never be less than
the 50th percentile, it follows that the geometric
standard deviation can never be less than one. In fact, the smallest value that the geometric
standard deviation can take is one, and this
happens when all the particles in an aerosol have the same exact diameter. That is, it is a
monodisperse aerosol . In this case, the 84th percentile is the same as
the 50th percentile, and every other percentile
for that matter. The larger the geometric standard deviation, the
more variability there is in particle size. The concentration of an aerosol can be
expressed in several different ways. First, we can express it as the number of
particles per cubic meter of air, or N, as shown
in the first row of this table. If we have particles of diameter d expressed in
units of nanometers, we can also calculate the
total surface area of all the particles with this diameter as N times pi times the square of the
diameter, where pi times the square of diameter is the surface area of one particle assuming that
the particles are spherical. This is the surface area concentration in
nanometers squared per cubic meter of air. We can calculate the volume of these particles
as N times pi times the cube of the diameter
divided by six. This is the volume concentration in nanometer
cubed per cubic meter of air. We can also multiply the volume concentration
by the density of the particles in the appropriate
units to get mass concentration in milligrams per cubic
meter of air. In this instance, the density of the particles is
expressed in milligrams per nanometer cubed
for the units to come out right. We can apply the relationships described in the
previous slide to each size bin in the particle
count table we encountered earlier The top histogram is the same as before, except
that instead of plotting particle count in each
size bin on the vertical axis, we are showing the count fraction in each
particle size range divided by the width of that
size range on a log scale. We can take the midpoint of each size interval
to be the representative diameter for all the
particles in that size interval. This diameter can then be used to convert count
in that size range to either surface area or volume, using the relationships shown in the
previous slide. Even though the size bin 50 to 80 nm has the
most number of particles, these particles contribute very little to the surface area and
almost none to the volume. contribute very little to the surface area and
almost none to the volume. Even though there are very few number of
particles in the size bin 890 to 1260 nm, these particles contribute a disproportionate
amount to the surface area and volume. This is because the surface area is proportional
to the diameter squared and the volume is
proportional to the diameter cubed, and so we see that the histograms change in
their shape as we go from a histogram based on particle count, to one based on surface area, to
one based on volume of the particles. This slide shows the type of size distribution by
mass that we can expect from some of the
activities we have discussed earlier. For instance, we saw a video of a worker using a
saw to cut a concrete block. This is a mechanical operation, and as we have
discussed earlier we expect to see a coarse
dust being generated. The histogram on the left shows the particle
mass distribution as a function of aerodynamic
diameter. Particles between roughly 2 to 20 µm
contributed most to the mass of the aerosol. The histogram on the right is typical of the
scenario when diesel operated machinery is used to haul crushed coal as we saw on the
second video. The diesel exhaust contributes particles
between roughly 0.1 to 0.5 µm while the coal dust contributes the most mass from
particles between roughly 1 to 20 µm. This is referred to as a bimodal distribution, and
reflects contributions to particle mass from two
aerosol sources. In the next segment of this module, we will
discuss the various elements of an aerosol sampling system and various sampling
strategies. At the bottom of the slide, we can see
schematically the components of an aerosol
sampling system. A particle laden air is drawn through a sampling
inlet. Unlike when we sample for gases and vapors,
the design of the aerosol sampling inlet ensures that only a specific size range of particles is
sampled. The aerosol then passes through a transport
section where some of the aerosol may be lost. This could be something as simple as a length
of duct or a tube where particles may deposit
and may not make it to the measurement zone. The aerosol then enters the sensing zone or
collection medium which captures the particles
or measures some relevant property. This can be a filter that collects airborne dust
that can then be analyzed by microscopy,
gravimetry, or other chemical means. It can also be a zone through which the aerosol
moves and interacts, for example, with a light beam, and in the process some
optical property gets sensed. Other elements of a sampling system include a
flow measurement device and the pump for
forcing the air through the system. Even though there are a wide array of aerosol
measuring instruments, they all share this basic
overall configuration. Depending on the reasons for sampling and
instrument design, several sampling strategies
can be employed: active vs. passive sampling, area vs. personal
sampling, and grab sampling vs. integrated
sampling. In active sampling the aerosol is drawn in
through a pump onto a collection medium, that is, an external source of energy is needed
to move the air through the sampling train. The photograph on the left shows aerosol
samplers on a rooftop. Air is drawn through the sampling inlet using an
electrical pump, and we can see electrical wires coming out of the pumps being connected to
outlets for power supply. The photograph on the right shows a person
wearing a sampler connected to a battery
operated pump. The battery is typically low power and, hence,
only a small flow rate of air can be drawn using
this pump, compared to the pumps on the left. In passive sampling, aerosol particles are
transported by gravity, diffusion, and inertia onto
the sampling surface. There is no external power source. The passive sampler shown in this photograph is
just a metal stub onto which particles deposit. The deposited particles can then be analyzed by
various analytical techniques, for example,
microscopy. Area sampling is done to determine general
background environmental conditions. These samplers on a rooftop sample the air in
the Twin Cities and provide a measure of the background aerosol concentration in the Twin
Cities metro area. In contrast, personal sampling is done to
determine the personal exposure of workers to aerosol by monitoring their breathing zone
concentration. In this photograph, you see an aerosol sampler
which is attached to a pump, and the sampler is
drawing air from the same general region of air from which the person is breathing, and
this volume is called the breathing zone of this
person. Another technique that is commonly being used
is to videotape a worker or a person carrying out
a series of tasks, and then overlay on it the real time measurements made by an instrument. In this photograph, we see a worker on the left
hand side doing various tasks and on the right
hand side we see the time trace of the exposure measured by a real time instrument. By focusing on the peaks, we know which tasks
contribute more to the exposure, and this By focusing on the peaks, we know which tasks
contribute more to the exposure, and this helps us focus our resources on addressing
those particular tasks for added attention. Time averaging is an important concept. Samples are collected over a period of time, and
represent an average over that time interval. In this graph, we show a time trace, that is, the
concentration as a function of time over a time
interval from zero to capital T. If we obtain a sample over a very short time
interval, as from t1 to t2, we refer to that as a
grab sample, and the grab sample is an average over that time interval t1 to t2. It is defined as
the integral of that concentration C as a function
of time over the time interval t1 to t2, over that time interval t1 to t2. It is defined as
the integral of that concentration C as a function
of time over the time interval t1 to t2. Then, we can divide this integral by the length of
the time interval, t2 minus t1.
0:51:23.066,0:51:23.033
Thus, it represents an average concentration
over a very small time interval, for example, just
a few minutes Thus, it represents an average concentration
over a very small time interval, for example, just
a few minutes In integrated sampling, the contaminant is
collected over a much longer period of time,
zero to capital T, for example. It could be an entire work shift or an entire day.
It represents an average concentration over this
entire time period. The integral in both these equations is the area
under the curve over the specific time interval: t1 to t2 in the case of grab sampling, and zero
to capital T in the case of integrated sampling. The concentration can also be expressed as the
mass of contaminant collected divided by the
sampling flow rate times the sampling time T. In the case of grab sampling, the amount of
mass collected is small and, hence, grab
sampling can only be used for identifying the contaminants rather than detailed quantification
of the mass. In integrated sampling, however, we can
typically collect a sufficient amount of
contaminant for quantitative assessment.