Hello, and welcome to "Air Cleaners for
Nanoparticles". My name is Pete Raynor. I'm a faculty member
at the University of Minnesota School of Public
Health. The learning objectives for this lesson are that,
by the end of the module, learners should be
able to list the main technologies used to remove particles from moving air, identify air
cleaning technologies that are effective for
nanoparticles, describe the mechanisms by which fibrous air filters capture particles, and
explain why fibrous air filters can capture
nanoparticles with high efficiency. There are a variety of factors that influence air
cleaner selection. They can be categorized into characteristics of
the moving air from which a pollutant must be
removed, characteristics of the pollutant, and the required outlet conditions for the air
leaving the air cleaner. The airflow characteristics include the flow rate
of the air, temperature, humidity, and the
pressure, especially if the air in the system is either at a vacuum or over-pressured relative to
the air outside the system.
The pollutant characteristics include the form of
the pollutant (particles, gases, or vapors), as
well as the concentration of the pollutant. If particles are the pollutant of concern, what is
the size distribution of the particles in the air? Are the particles solid or liquid? Are they sticky?
Are they viscous? Are they abrasive? Also, is
the pollutant corrosive? Required outlet conditions might be set by
standards, regulations, or permits, and might
include the concentrations of the pollutants, the size distribution in case there are different
requirements for different sizes of particles, and requirements that might be more stringent if
the air will be recirculated back into a
workplace.
Important parameters to consider when
evaluating different air cleaners include, first and
foremost, the efficiency of the air cleaner. All else being equal, we want efficiency to be as
high as practicable. The flip side of efficiency is
penetration. While efficiency is how much of a material
coming into an air cleaner is removed,
penetration is how much gets through, which can be more important from the
standpoint of exposure. Another important parameter is pressure drop.
The pressure drop across a collector is related
to how much energy is required to move air through the collector, which, in turn, is related to
the cost of operating the collector. A higher
pressure drop is more costly. Other parameters include the reliability of the
collector – is it going to work each and every
time?; the maintenance requirements – are there significant costs associated with
having to change out collection media like filters,
or do elements of the collector break down frequently?; the size of the collector – can the
collector fit into a suitable space inside or outside of the work space?; and capital expense
– how much will it cost to install the collector? Particle collectors that use a variety of operating
principles are available.
They include settling chambers, inertial
collectors, cyclones, particle scrubbers, electrostatic precipitators, fabric filters, and
fibrous filters. While we will talk at least briefly about all of
these approaches, we will focus most on fibrous
filters. In a settling chamber, a particle-laden airstream
enters a larger plenum and slows down because
the cross-sectional area is so large. Velocity is equal to the airflow rate divided by
the cross-sectional area. So, if the area
increases, the velocity must slow down. As the velocity slows, particles have a better
opportunity to fall out of the air flow due to
gravity. In a settling chamber, as the image on the right
shows, particles fall due to gravity into a hopper
and are collected there. They have low pressure drop and can be used
with high-temperature air streams. However, settling chambers are generally useful
only for particles with diameters larger than
about 50 micrometers. Additional collectors using different operating
principles are required for smaller particles.
Therefore, this is not a technology that is
effective for nanoparticles. Inertial collectors capture particles as they
deviate from air streamlines that follow a
tortuous route through the collector. These tortuous routes require the air and the
particles to make several sharp turns. Larger particles are unable to follow the air
streamlines due to their inertia, and they may be
collected on surfaces within the collector. A couple of examples are shown in the
illustrations on the right. At the top is a curtain collector, in which the air
encounters a series of elements and must make a greater than 90° turn in order to get through
each element.
Collected particles build up on the inside of the
elements, and they can be rinsed off later using
a cleaning system. Another example is a louvre collector, where the
air comes in from the top and makes a turn
greater than 90°. Particles can potentially be collected on louvres
and then later be rinsed off. Care must always be taken with inertial
collectors to ensure that the particles do not
build up too much and block the airflow. So, a cleaning approach is required. For inertial collectors, the 50% cutpoint – the size at which 50% of the particles will be collected – varies by the design, but this technology is only
effective for particles with diameters larger than
about 5 micrometers. Pressure drops required to achieve a smaller
50% cutpoint would make these collectors
completely impractical. Therefore, inertial collectors are not an approach
that is suitable for nanomaterials. Cyclones collect particles by centrifugal force. The diagram on the right shows how dirty air
containing particles comes in and swirls around
the inside wall, kind of like a tornado. Large particles cannot follow the air streamlines
due to their inertia, impact the inner wall, and fall toward the bottom of the cyclone where
they can be collected.
When the cleaned air reaches the bottom of the
cyclone, it moves up through the middle of the
cyclone and out of the collector. There are large, industrial-scale cyclones, and
ones that are smaller that can be used in a
laboratory or a small manufacturing facility. Cyclones have many advantages including that
they are good for air flows at extreme
temperatures and pressures, they can handle high flow rates of air, they are suitable for high
particle mass concentrations, they are
inexpensive and relatively easy to make, and they work well for mist droplets in addition
to solid particles.
Cyclones perform poorly with sticky or
hygroscopic dusts because the particles tend to
cling to the inner wall rather than falling down into the bottom of the collector and into a
hopper. The pressure drop across cyclones can
vary widely depending on the specific design. This is a very effective technology for particles
larger than about 5 micrometers in diameter. Unfortunately, they are not effective for smaller
particles. So, much like settling chambers and inertial
collectors, cyclones are not suitable for
removing nanoparticles from air streams.
Particle scrubbers collect particles by impaction
onto falling drops. There are a couple of different
types. As shown in the image, a spray tower releases
drops, much like rainfall, into an upward-flowing,
dirty air stream. These collectors have low pressure drop, and
they capture most particles larger than about 10
micrometers in diameter. However, they are not effective for nano-scale
particles. As shown in this second image, a venturi
scrubber brings particles and drops together in a
high velocity jet. The air stream is compressed as it passes
through a region where there is a frothy liquid. The particles impact onto that frothy liquid, and
they leave this region encapsulated in drops. Venturi scrubbers are considerably more
efficient than spray towers, with the ability to capture some, but certainly not all, sub-
micrometer particles.
The cost of this greater
efficiency is a very high pressure drop. The large drops leaving a venturi scrubber that
contain the incoming particles can be removed
from the air stream very effectively by a cyclone. Both types of particle scrubbers perform well
with sticky particles. However, the liquid which contains the collected
particles may need to be treated before being
reused or released. The bottom line is that, while particle scrubbers
– especially venturi scrubbers – can be more
efficient than settling chambers, inertial collectors, and cyclones, they are not
highly effective for nanoparticles. Electrostatic precipitators, or ESPs, operate by
charging particles in an electric field and then collecting them on electrically-grounded metal
plates in an electric field. Particles can be
charged either positively or negatively in ESPs. Wires at very high voltage, generally greater
than 10,000 volts or 10 kilovolts, produce a corona that generates a flow of ions that attach
to particles in a moving air stream. The voltage is limited on the upper end by the
requirement that the corona must not lead to sparking, which reduces the effectiveness of an electrostatic precipitator.
Due to the high voltages and the risk of
sparking, an ESP is a poor choice for flammable
air streams. Particles collected on the plates in an
electrostatic precipitator are later either shaken off or washed off using a liquid in order to
maintain effectiveness over long periods of time.
Sticky particles are a concern for this cleaning. Therefore, ESPs may not be a suitable choice
for sticky dusts. Electrostatic precipitators can be highly efficient
for particles greater than several hundred
nanometers.
They can be up to 95% efficient for 100-
nanometer particles and up to 90% efficient for
10-nanometer particles. While this efficiency is higher than for the other
types of collectors we've discussed, even higher
collection efficiency would be desirable. Because ESPs have low pressure drop, they are
not expensive to operate. However, they are
expensive to purchase and install. Electrostatic precipitators typically use one of
two primary design approaches: single-stage
and two-stage collectors. This slide shows diagrams of both types of ESP
configurations. On the left, single-stage ESPs have two
grounded plates that surround high-voltage
wires. The wires and plates are coming out of the
screen, perpendicular to the screen's surface. The dirty air flows in from the left and clean air
leaves from the right. The high voltage on the wires ionizes the air –
here we're depicting the creation of positive ions – and the ions then move toward the grounded
plates. Some of these ions will attach themselves to
particles moving with the airflow, causing the
particles to become charged.
The charged particles will then also move with
the electrical field toward the grounded plates.
It's a relatively simple, but effective, design. In a two-stage ESP, high voltage wires in the
first stage charging section work very much like those in a single-stage ESP except that the
charging section is much shorter. The second stage is a collection section in
which a steady electrical field has been established between highly-charged plates and
grounded plates. These plates are much closer than those in
single-stage collectors leading to more intense
electrical fields and higher collection efficiency.
Fabric filters are cylindrical bags of woven or
felted fabrics that collect particles mostly on the
surface of the filter material. The filters are located in a housing in which the
velocity of the air entering the housing slows
down dramatically. The bags can be either inside or outside
collectors, meaning that particles may be collected on either the inside or the outside of
the cylindrical bags. Inside collectors are typically about 30 feet long,
1 foot in diameter, and have a 1 to 4 cubic foot
per minute per square foot air-to-cloth ratio, which is equivalent to a velocity of 1 to 4 feet per
minute at the filter face. Outside collectors are shorter, typically about 8
feet long, about four-and-a-half inches in diameter, and have an air-to-cloth ratio of 5 to 15
cubic feet per minute per square foot of cloth. The pressure drop across a fabric filter depends
on the air-to-cloth ratio, the properties of the
fabric, the amount of particles loaded onto the fabrics, and particle properties, including their
size.
The filters are cleaned cyclically. We might have tens of even hundreds of fabric
filters in a single collector, usually referred to as
a baghouse. One group of filters in a baghouse can be taken
off-line briefly and cleaned while others continue
to operate. Inside collectors are typically cleaned by
mechanically shaking them or using a reverse
gas flow. Outside collectors are cleaned by a pulse jet. Nozzles above the filter outlets release very
short bursts of pressurized air that flex the fabric outward to knock off dust cakes that have built
up on the outside of the filters.
Fabric filters have relatively high efficiency. They capture particles larger than 1 micrometer
in diameter with nearly 100% efficiency once a semi-permanent dust layer builds up on the
fabric. Initially, the fabrics are less efficient. Because some time is required for the particle
layer to build up naturally, particles are often released artificially into the airflow upstream of
new filters to increase the efficiency faster. It is difficult to achieve greater than 95%
efficiency for nano-scale particles because some of these particles will get around or
through the dust layer. Fabric filters are compatible with high particle
concentrations, but they perform poorly with
sticky or hygroscopic particles because these particles are difficult to remove from the fabrics
during cleaning cycles. These drawings show the air flows through an
inside collector.
In the diagram on the left, the dirty air enters
near the bottom of the baghouse, and flows
upward through the cylindrical fabric filters. The dirty air passes through the filters and is
cleaned. The captured particles build up on the
inside of the bags. As shown in the diagram on the right, the bags
can be cleaned by a reverse gas flow that flexes the fabric filters inward and dislodges the
captured particles in clumps that then fall into a
hopper.
An alternative to the reverse flow is to clean the
bags by mechanically shaking them. In outside collectors, the dirty air enters the
baghouse, as shown in the diagram on the left, and flows through the bags from the outside,
with particles collecting on the fabric surface. The cylindrical bags are supported by a cage or
frame to keep them from collapsing inward. The clean air emerges on the inside of the bags
and flows upward and out at the top. The diagram on the right shows what happens
when these bags are cleaned by pulse jets. Pressurized air flows through a blowpipe to
venturi nozzles that, when opened for fractions
of a second, create a very strong pulse of air that flexes the bags rapidly outward, knocking
collected particles off in clumps that fall into a
hopper. The last air cleaning technology that we'll
discuss is fibrous filtration, the technology that
can be most efficient for nanoparticles, and the one used most commonly to remove
nanoparticles from airflows. Fibrous filters are beds of nonwoven fibers that
collect particles by both surface and depth
filtration.
This means that, while particles can collect on
the surface of fibrous filters, many more particles are captured within the
depth of the filters. Due to this depth filtration,
fibrous filters cannot be cleaned in most cases. Therefore, once they build up too much
resistance to the air flowing into the filter, they
must be discarded. Because they are not able to be cleaned, fibrous
filters are not intended for high particle
concentrations. So, for mixtures of nanoparticles and larger
particles, many air cleaning systems use a
combination of technologies such as a cyclone or an electrostatic precipitator that capture the
larger particles followed by fibrous filters to
collect the remaining nanoparticles. This combination approach extends the useful
life of the fibrous filters.
There are a wide range of options for fibrous
filters. They are available in many, many different
configurations with any efficiency level that is
needed. Fibrous filters can be highly, highly efficient,
even for nanoparticles, and they can be more efficient than any other type of particle collector,
especially for nanoparticles. For example, one class of filters is called High
Efficiency Particulate Air, or HEPA, filters. These filters are defined as having greater than
99.97% efficiency for 300 nanometer diameter
droplets using a standard test protocol. The pressure drop across a fibrous filter
depends on the velocity of the airflow, the
diameters of the fibers used in the filter, the solidity or packing density of the filter (the
fraction of the volume occupied by the filter that is solid material), and the filter thickness in the
direction of air movement. The surface area of the face of the filters can be
increased with pleats or pockets which can
improve the collection of nanoparticles by slowing down the velocity of the air passing
through the filter medium.
The seating of a fibrous filter within a filter
housing, or, similarly, the fit of a particle respirator to a wearer's face, must be proper for
that high efficiency to be effective. If the particles can flow around a filter, then there
is no point in having a filter that is highly
efficient. This is a scanning electron micrograph of a
fiberglass filter. We get a sense for the depth of
the filter. We also see that the fibers are nonwoven; they
are random in their orientation in at least two
dimensions. The fibers have a range of diameters. Manufacturers of certain types of filters will
include fibers of many diameters to provide
optimal collection of different-size particles. How do fibrous filters capture particles? Let's
talk first about three mechanical filtration
mechanisms. The first is interception. Interception is the
capture of particles that follow air streamlines. The central circle in this diagram represents a
fiber with a circular cross-section coming out of
the screen.
A spherical particle represented by a blue circle
follows the air streamline moving around the fiber
from left to right. The blue circles show the path of the particle
along the streamline. If the particle is large enough, it will touch the
fiber and be collected by interception. This is a physical process; particles bigger than
a certain size cannot pass by a fiber because they come in contact with it even if they are
following the air streamlines. The second mechanism, impaction, relies on
the inertia of particles. Particles that are relatively large and/or heavy
will deviate from air streamlines, especially if the
velocity through the filter is high.
In the second diagram, the movement of a
particle with a lot of inertia from left to right is
illustrated by the red circles. The particle deviates from the air streamline and
impacts on the forward surface of the fiber. This mechanism is sometimes referred to as
inertial impaction. The third mechanical filtration mechanism is
diffusion. In this case, a small particle will have difficulty
following an air streamline because it wanders away from the air streamline due to Brownian
motion.
If the particle moves far enough in the right
direction, it may impact the fiber surface and be
captured. In the third diagram, this diffusion capture
mechanism is illustrated by the motion of a
particle represented by the green circles. Interception, impaction, and diffusion are the
three most important mechanical filtration
mechanisms. Equations have been developed to predict the
efficiency of particle capture by single fibers for
these three mechanisms. Interception efficiency – the Greek letter eta,
sub R – can be predicted by the equation 1 plus
R divided by 2 times something called the Kuwabara number times the long, complicated
expression in brackets. This expression contains three variables. R is
the interception parameter, which is the ratio of
the particle diameter to the fiber diameter. Ku is called the Kuwabara number, which is a
property of the flow field that is predicted for the
air flowing around the fiber. The Kuwabara number is a function of alpha,
which is the solidity or packing density of the filter, the fraction of the filter volume that is
occupied by fibers as opposed to air.
The
solidity also appears in the efficiency equation. So, interception efficiency can be predicted from
the interception parameter, the Kuwabara
number, and the filter solidity. The predictive equation for impaction efficiency –
eta, sub I – includes three factors: the
parameter J, a non-dimensional parameter called the Stokes number (Stk), and the
Kuwabara number. The Stokes number, related to the particle size,
the air velocity, and other parameters, increases
as particle inertia increases. It is equal to 1 over 18 times the particle density
times the particle diameter squared times the
velocity of the air flowing across the face of the filter times the Cunningham slip correction factor
divided by the air viscosity and the fiber
diameter. The Cunningham slip correction factor accounts
for the fact that particles smaller than about 1
micrometer are affected by individual air molecules rather than being affected by air as
though it is a continuous fluid.
The equation for J
was developed empirically. It is a function of the filter solidity and the
interception parameter. Diffusion efficiency – eta, sub D – can be
predicted from two factors: the Kuwabara number and a non-dimensional parameter called
the Peclet number (Pe). The Peclet number relates particle diffusion due
to Brownian motion to particle advection due to
air flow. The Peclet number is equal to the air velocity
times the fiber diameter divided by the diffusion
coefficient of the particle (capital D). The diffusion coefficient is equal to the
Boltzmann constant times the absolute
temperature times the slip correction factor divided by three times pi, the air viscosity, and
the particle diameter. How do we use these three single-fiber
efficiency equations for individual mechanisms to predict an overall single-fiber efficiency and
total filter efficiency? We add the single-fiber efficiencies for the three
mechanisms together to estimate the overall
single-fiber efficiency. So, the single-fiber efficiency is equal to the
interception efficiency plus the impaction
efficiency plus the diffusion efficiency.
Using assumptions about the distribution of
fibers within a filter, we calculate total filter
efficiency from the single-fiber efficiency as being equal to one minus the mathematical
constant e to the power minus 4 times the filter
solidity times the thickness of the filter (L) times the single-fiber efficiency divided by pi and the
fiber diameter. We also have a formula for predicting the
pressure drop across a fibrous filter. This is an empirical formula based on
measurements of many different fibrous filters. The pressure drop is set equal to the velocity of
air entering the filter times the air viscosity times
the thickness of the filter divided by the fiber diameter squared, times a function of the
solidity of the filter.
When we use the equations to predict single-
fiber efficiency, we can look at how efficiency changes as a function of particle diameter for
each mechanism and for a single fiber overall. For a filter made from fibers that are 5
micrometers in diameter, with a solidity or packing density of 0.05 or 5%, a face velocity of
10 centimeters per second, and assuming that particles have a density of 1 gram per cubic
centimeter, we get the curves shown in this
figure. For the interception and impaction mechanisms, single-fiber efficiency increases as particle diameter increases. On the other hand, for the diffusion mechanism, efficiency increases as the particle diameter decreases because smaller particles move more
by Brownian motion than larger particles. When we add these three curves together, we
have the overall single-fiber efficiency
represented by the solid, dark curve. The overall single-fiber efficiency curve has a
minimum between about 0.2 to 0.3 micrometer,
or 200 to 300 nanometers. Particles larger than this minimum are collected
more effectively by interception and impaction, whereas particles smaller than the minimum are
captured more effectively by diffusion.
This minimum efficiency at 200 to 300
nanometers is why HEPA filters are tested for efficiency using 300-nanometer diameter
droplets. These curves are typical for filters that collect
particles by mechanical filtration mechanisms. The predictions have clear implications for the
fibrous filtration of nanoparticles because they
tell us that, due to diffusion, fibrous filters can be highly efficient at removing nanoparticles from air
streams. When we use the single-fiber efficiency to
predict total filter efficiency using the same
parameters as on the previous slide, with a filter thickness of 2 millimeters, we end
up with this curve for efficiency as a function of
particle diameter. The efficiency is high for large particles and very
small particles with a minimum efficiency at a diameter between 0.2 and 0.3 micrometer or 200
and 300 nanometers. An important filtration mechanism that we have
not talked about yet is electrostatic attraction. Electrostatic charges can be built into fibrous
filters to enhance their efficiency. Four different scenarios can be considered
related to fiber and particle charge conditions. So far, we've been considering scenarios in
which neither fibers nor particles carry charge. However, we can have scenarios in which just
the fibers are charged but not the particles, in which the particles are charged but not the
fibers, and in which both the fibers and the
particles carry charge.
When considering filtration by electrostatic
attraction, we are generally interested in filters
made from fibers that carry charges and particles from workplace or outdoor
environments. Usually, there is a distribution of charges on
these particles, so some will carry positive or
negative charges and some will be uncharged. Thus, we will focus most on scenarios (b) and
(d) from this table. Let's look at electrostatic filtration mechanisms
a little more closely. A charge on a fiber can attract oppositely-
charged particles by Coulombic forces. As illustrated in this diagram, a negative charge
on a fiber will enhance collection of a particle
carrying a positive charge. The same will be true for positive charges on
fibers and particles carrying negative charges. Because most electrostatically-charged filters
carry both negative and positive charges on their fibers, the capture of both positively- and
negatively-charged particles is enhanced. If a fiber carries a negative charge but a particle
is neutral so that it does not carry a net charge,
the charge on the fiber can induce a dipole – a polarization of charge – within the particle that
causes the neutral particle to be drawn toward
the fiber.
If a fiber does not carry a charge but a particle
does, the particle – if it gets close enough to the
fiber – will induce an image force on the fiber that enhances the fiber's ability to attract the
particle. In essence, the particle helps itself be
collected. Fibrous filters that carry electrostatic charges
are often referred to as "electret" filters. Electret
filters have been around since about 1930. Different processes are used to create these
filter materials. When fibers made from two different materials
are placed together and rub against each other, negative charges can build up on one material
while positive charges build up on the other. This process, referred to as triboelectric
charging, can create a filter with stable charges
made from two different types of fibers. The earliest electret filters, called Hansen filters
or resin-wool filters, were made using
triboelectric charging. Today, triboelectrically-charged filter materials
are produced by layering one type of fiber on top
of another and then carding them together using needles to mix the fibers together and cause
them to rub against each other, creating the
stable charge.
Corona-charged electret filter materials are
created using an ionization source that causes one side of a fiber to be charged positively and
the other side to be changed negatively. These can include split-fiber materials, where a
polymer sheet is charged in this manner and
then the sheet is sliced into thin fibers, and melt-blown materials that are charged after
being formed into filters. Induction charging, which occurs during the
electrostatic spraying of polymers, applies electrostatic charges to polymer fibers
as they are created. This is a particularly good way to create electret
filters because fibers with very small diameters
are created, even ones with diameters smaller than 1 micrometer, that can be very effective at
capturing particles. As mentioned previously, both positive and
negative charges are generally present on
electret filters. Why should we consider using electret filters? The electrostatic attraction mechanisms allow for better filter performance. Electret filters can achieve lower pressure drop
through the filter with the same efficiency as purely mechanical filters, or higher efficiency for
the same pressure drop.
Applications include large building heating,
ventilation, and air conditioning, or HVAC,
filtration; home air filters; respirator filters, most of which are electret filters; and some air
pollution control filters. Equations are shown here for predicting single-
fiber efficiency by the electrostatic attraction
mechanism when both the fibers and particles carry charge and when the fibers carry charge
but the particles are neutral. For the charged fiber and particles, the single-
fiber efficiency is equal to the Cunningham slip
correction factor times the fiber charge per unit length times the charge on a particle divided by
the product of 3 times pi times the permittivity of
free space (a measure of how easy it is to form an electric field in a vacuum) times the air
viscosity times the particle diameter times the fiber diameter times the velocity of the air flowing
through the filter.
For the collection of neutral particles by charged
fibers, the expression is more complicated. It uses many of the same terms, but also
includes the dielectric constant of the particle material, which indicates whether a particle is
electrically conductive or insulating. What is the influence of electrostatic attraction
on the penetration of particles through filters? This figure from Romay and co-authors shows
the penetration of particles through a filter, on a
logarithmic scale, as a function of particle diameter for the four different combinations of
fiber and particle charging. If we compare curve (2) to curve (4), we can see
the effect of charging a filter on the collection of
uncharged, or neutral, particles. The efficiency is enhanced, the penetration is
reduced, more effectively for particles larger than 10-to-the-minus-1 micrometer or 100
nanometers than for nano-scale particles. Larger particles are affected more than smaller
ones because the strength of the induced dipole
is proportional to the volume of the particle.
If we next compare curve (1) to curve (3), we are
considering the effects of filter charging on particles that each carry a single charge
regardless of their diameter. In this comparison, we see broad decreases in
penetration, or increases in efficiency, across all particle sizes when we use a charged filter
versus an uncharged filter. One thing to keep in mind, however, is that
many particles larger than about 80 nanometers are likely to carry multiple charges, allowing
them to be captured more effectively. In addition, particles can carry considerably
higher levels of charge as size increases, making electrostatic attraction much more
effective than if the particles carry only a single
charge. The effects of electrostatic enhancement on filter
efficiency can be substantial when we consider
particles with an equilibrium charge distribution. These graphs from a paper by Rengasamy and
co-authors show penetration through respirator
filters as a function of particle size.
The authors measured penetration for a variety
of respirator filters that carry electrostatic
charge. Then, they rinsed the filters with isopropyl
alcohol, which removes the charges from the
filters. In each of the four figures, the two higher curves
are for filters that have been discharged and no
longer carry electrostatic charges. The lower curves are for the charged filters.
The filters that carry electrostatic charge have
much lower penetration, or much higher efficiency, than the ones that have been
discharged. While the effects of electrostatic charging are
more pronounced for particles larger than 100
nanometers than for particles smaller than 100 nanometers because large particles can carry
many more charges, the improved capture due to electrostatic effects is still very significant for
many nanoparticles. The American Society of Heating, Refrigerating,
and Air-Conditioning Engineers, or ASHRAE, publishes standard test protocols to measure
filter performance. Over the years, there have
been three different tests. One is called arrestance, which was included in
ASHRAE Standard 52.1-1992, a standard that has since been withdrawn, but is an addendum
to the current standard 52.2-2007. The arrestance test measures the percentage of
a standard test dust captured by a filter on a
mass basis.
This is a single measure of efficiency integrated
across all sizes of particles in the test dust. Another test, once very common but no longer
used, was the dust spot efficiency test. This test was a little odd: it measures the percentage of staining atmospheric dust collected. It involved optical measurement of atmospheric
dust collected on target papers upstream and
downstream from the filter being evaluated. Because the results depended on the properties
of the dust in the atmosphere at a test facility, it
was not a very consistent or reliable test, and it did not provide efficiency data for different
particle sizes. In ASHRAE Standards 52.2-1999 and -2007, the
Minimum Efficiency Reporting Value, or MERV
rating, was introduced.
This test measures filter efficiency on a count
basis in three particle size ranges using salt
particles generated by standard methods. The three size ranges are 0.3 to 1 micrometer, 1
to 3 micrometers, and 3 to 10 micrometers. Notably, none of the size ranges in the test
protocol include nanoparticles. The protocol includes a conditioning test to
measure how efficiency may change as
particles are collected on a filter. MERV ratings are assigned based on the
results of the efficiency tests. As shown in this table, MERV ratings of 1 to 4
only consider the results of the arrestance tests.
Therefore, filters in these categories are only
effective for very large particles. As we move down the table to ratings of 5 to 8,
we start to see higher efficiency for particles in
the range of 3 to 10 micrometers in diameter. For MERV ratings of 9 to 12, the efficiency for
particles in the range of 1 to 3 micrometers
starts to increase. Finally, for ratings of 13 to 16, we start to see
high efficiencies for particles in the 0.3 to 1
micrometer range.
None of these standard tests consider the
effectiveness of filters against nanoparticles. While this may not be much of a problem for
purely mechanical filters that have a minimum
efficiency at around 0.3 micrometer or 300 nanometers, it is a considerable limitation
for electret filters that are likely to have a minimum efficiency at a diameter smaller than
100 nanometers. Filters do not capture air molecules because
individual molecules move so quickly that they bounce right off of a fiber if they come in contact
with one.
This is referred to as "thermal rebound". At
some small diameter, nanoparticles may do the
same thing. In 1991, Wang and Kasper used a theoretical
approach to predict that thermal rebound would occur in filters for particles smaller than
about 10 nanometers. As shown in the figure, they predicted that
penetration would reach a minimum at about 10
nanometers and then increase for smaller particles rather than showing a continuous
decrease as predicted by diffusion theory. This would present an obvious concern for the
effectiveness of fibrous filters at capturing the
smallest nanoparticles. In 2005, Balazy and co-authors presented some
measurements at the European Aerosol
Conference that seemed to suggest that thermal rebound occurred for particles that were smaller
than about 20 nanometers. Their graphs, which show efficiency rather than
penetration as a function of particle diameter,
indicated that efficiency decreased dramatically as particle size decreased for particles smaller
than 20 nanometers in diameter for two different
filters. Another group of researchers, however, was able
to show that there were methodological errors related to the efficiency measurements in the
Balazy et al. study. Subsequently, three research groups tried to
verify the Balazy et al.
Findings, but they could
not. Let's take a look at their results. Heim and co-authors published a study that
measured filtration efficiency as a function of
particle diameter, and they did not observe any decrease in efficiency as a sign of thermal
rebound for particles down to about 2.5
nanometers. Their measurements largely followed filtration
theory. Kim and co-authors published a paper with this
figure showing penetration as a function of
particle diameter that indicated that efficiency increased as particle diameter decreased for
particles at least down to 2 nanometers in
diameter.
A group from the University of Minnesota with a
different lead author with the last name of Kim
showed that penetration decreased, in other words efficiency increased, down to
particles of about 3 nanometers in diameter. So, all three study showed that thermal rebound
will not occur in fibrous filters for particles larger
than 2-3 nanometers. The Kim et al. study from 2006 did show an
increase in penetration for particles smaller than about 2 nanometers, suggesting that thermal
rebound may occur for these very smallest
particles.
This may not be much an issue, however, in a
practical sense because we do not expect many 1 to 2 nanometer particles to remain as
individual particles for very long. According to coagulation theory, we expect
such small nanoparticles to rapidly form
agglomerates that function effectively as larger particles that are likely be collected by fibrous
filters very effectively. The bottom line is that fibrous filters are capable
of collecting nanoparticles with very high efficiency, if the filters are designed
appropriately. One caution we need to keep in mind, however,
is that we do not know very much about
changes in filter performance with time as the filters collect and load with engineered
nanoparticles. The evidence that we have with other kinds of
particles gives us reason for concern about
changes in fibrous filter performance with use. For instance, we visited 140 hospitals in the
state of Minnesota and tested filter efficiency
with a P-Trak Ultrafine Particle Counter, a condensation particle counter, to answer the
question, "Do filters work as well as they are
rated?" In each hospital, we entered an air handling unit
upstream and downstream from its filter bank to
measure particle number concentrations, or we inserted a probe through holes drilled
upstream and downstream of the filter bank
through which we made the measurements.
We compared the efficiency that we calculated
from the measurements to the dust spot
efficiency indicated on the filter itself. This plot of measured efficiency on the vertical
axis versus rated filter efficiency on the
horizontal axis indicated that, in almost all cases, the measured efficiency was either lower
or much lower than the efficiency that we
thought we would see. There could be a couple of reasons for these
discrepancies. One is that the filters might not have been
installed properly so that they were not seated
well against the filter mount. Improper filter seating can allow leaks around
the filters. However, it is unlikely that almost all
of the filters would be improperly seated. Another possibility is that electret filters used for
HVAC filtration did not perform as well as they were rated once they have been used for a
while. Regardless of the reason, it does not appear
that the filters performed as well as they were
rated.
We decided to investigate the performance of
electrostatically-charged filters with use more
deeply. Some laboratory tests have shown that
efficiency declines as electret filters load with dust, while other tests in labs show that
efficiency stays the same or increases. Therefore, our objectives in this study, which
ended up being published in 2004, were to
measure the efficiency and pressure drop for electrostatically-charged filters for several
months in working HVAC systems. We installed filters in a building and then
measured the efficiency over time. In addition, we did the same for filters that did
not carry electrostatic charge in a parallel HVAC
system. This image shows our test filters. On the left is a
filter made from uncharged glass fibers. Its face was 2 feet tall by 2 feet wide, and the
filter had 15 pleats. The pleats go back and forth through the depth
of the frame to provide more filter area. These filters were 90 to 95% efficient according
to the ASHRAE dust spot efficiency test, which
was the standard test at the time of the study. The filter on the right was made from synthetic
polymer fibers that carried electrostatic charge.
Like the other filters, these electret filters
measured 2 feet by 2 feet across the face, had 15 pleats, and were 90 to 95% efficient
according to the dust spot test. The test location was Hasselmo Hall, a
laboratory building on the University of
Minnesota campus. The two air handling units that were used were
nearly identical. They had adjacent air intakes
that used 100% outdoor air. The units had the same layout and equipment.
Each was capable of moving as much as 60,000
cubic feet per minute, or cfm, of air. The flows through the two systems were similar
but varied depending on the demands of the
spaces within the building. Each system included pre-filters that prevented
most particles larger than about 3 micrometers
in diameter from reaching the test filters. The humidification systems were turned off
throughout the test period so that water droplets
would never interfere with the measurements. The test filters included 30 fiberglass filters in Air
Handling Unit #1, in five rows by six columns, and 30 synthetic polyolefin fiber filters in Air
Handling Unit #3 in the same configuration.
Air was sampled through long probes that were
inserted either upstream or downstream of the
filter banks and could be moved to four different locations, each denoted by a red X on the
drawing, at which we measured size-
differentiated concentrations. We measured efficiency on many dates over
about 4½ months. We measured efficiency more frequently toward
the beginning of the test than near the end to document the rapid initial changes in filter
performance.
Nevertheless, we made regular measurements
throughout the test period of late June to early
November. For each test, we measured 32 size
distributions. These included measurements in the two air
handling units, at four locations per air handling
unit, at upstream and downstream points for each location, and with two instruments at each
point. We measured the size distribution for particles
between 0.5 and 3 micrometers using an
Aerodynamic Particle Sizer, or APS. The APS measures particle sizes using the time
of flight for particles in an accelerated air
stream. We measured particles between 0.15 and 0.5
micrometer, or 150 to 500 nanometers, with a
Differential Mobility Particle Sizer, or DMPS, an instrument that separates particles by size
according to their electrical mobility and then counts the particles in each size interval using a
condensation particle counter.
For pressure drop, we referred to data measured
and stored hourly by the computer that
controlled the HVAC systems. What changes in efficiency did we find over time
for the uncharged fiberglass and charged
synthetic filters? For the fiberglass filters, we saw that efficiency
did not change much over the 4½ month test
period. In this graph, which shows efficiency as a
function of particle diameter, the initial efficiency
data are shown by green data points with a curve fit through the data, and the efficiency data
for day 134 are shown in yellow. While the fitted curves do not overlap, the only
statistically significant differences between the two sets of data were small increases in
efficiency with time for the very smallest
particles. For the synthetic polyolefin fiber filters (the ones
that carried electrostatic charge), however, we measured reductions in efficiency over the
test period. In this case the initial data are in
green and the data for day 134 are in blue. Efficiency reductions were quite significant
except for particles that were 1.5 micrometers in
size and larger. This graph shows efficiency on the vertical axis
versus time on the horizontal axis for both the
uncharged fiberglass and charged polyolefin filters for particles that were about 1 micrometer
in diameter.
For the polyolefin filter in blue, we measured a
decrease that started to occur almost
immediately upon installation and reached a minimum efficiency around 12 to 13 weeks
into the test. After this minimum, we observed a small
increase in efficiency through the rest of the
test. The efficiency for the fiberglass filter, in yellow,
was almost constant throughout the 4½-month
test. The two kinds of filters had similar pressure
drops across them throughout the test. This suggested to us that the polyolefin fiber
filters that carry electrostatic charge were not
well-designed because, as we discussed earlier, electret filters should have higher efficiency for
the same pressure drop or lower pressure drop
for the same efficiency relative to purely mechanical filters, and this was not the case for
this polyolefin fiber filter. In this case, when the filters were new, we had
essentially the same efficiency and the same
pressure drop.
The electret filters should have exhibited a
performance advantage when they were clean,
but they did not. So, what's going on to cause these changes in
efficiency for electret filters? The efficiency of the synthetic fiber electret
filters fell substantially with use. At the minimum efficiency around 12 to 13
weeks into the test, the electret filters allowed 6 times more 1 micrometer particles to pass
through than the uncharged fiberglass filters. This is a significant difference that could lead to
increased exposures to particles. The scientific literature indicates that collected
particles likely shield the charges on the fibers
and render them less effective. It must be pointed out that these results
represent what happened during one summer in
Minnesota using two particular kinds of filters in two particular HVAC systems collecting
atmospheric particles. Are these results generalizable? To answer this
question, we did another study with better
electret filters. In this figure, we show pressure drop as a
function of time over our three month test period
in working HVAC systems. We see that the synthetic electret filter we used
this time had much lower pressure drop than the uncharged fiberglass filter, which should be the
case for a well-designed electret filter, considering that the two kinds of filters had
almost identical efficiency at the start of the
test.
When we look at the changes in efficiency over
the three month test period, we see on the left
that the fiberglass filter showed no change in efficiency, just as in the previous test, while, just
as in the previous test, the electret synthetic
filter exhibited a large decrease in efficiency. So, our test results do appear to be somewhat
generalizable. What implications do our findings have for the
filtration of engineered nanomaterials? We have not yet tested electret filter efficiency
as a function of time as the filters capture
engineered nanomaterials. This is critical information that occupational
hygienists do not yet have. A reasonable hypothesis might be that if
engineered nanomaterials are collected on
fibrous filters that carry electrostatic charge, we may observe decreases in efficiency similar
to reductions we've seen in the tests with atmospheric aerosols, which include nano-scale
particles. The data suggest that electret filters suffer a
greater loss in efficiency when they are
challenged with smaller particles.
Until we have data for engineered nanomaterials,
we should be cautious with the use of electret filters as a way to collect engineered
nanoparticles over extended periods of time. Additional recommendations when using fibrous
filters to control particle exposures include to
visually evaluate the seal of mounted filters in all applications because having a good seal is
critical to ensuring that particles do not flow
around the filters with leaking air. In critical applications, we might test leakage
between filters and filter mounts using a direct-reading particle instrument such as an
optical particle counter or a condensation
particle counter. Also, it might be valuable to establish a change-
out schedule that is based not only on elevated
pressure drop, which is how the time for replacing filters is usually determined so that air
flow does not become restricted, but also based
on efficiency if the filters being used carry electrostatic charge, and those responsible for
replacing the filters should stick to the change-
out schedule.
Let's consider penetration of particles through
respirator filters. These graphs from a paper by Rengasamy and co-authors show penetration as a function of particle diameter. The efficiencies of these respirator filters are
rated by the National Institute for Occupational
Safety and Health (NIOSH) using particles with diameters of 300 nanometers because that was
the typical most penetrating particle size for purely mechanical filters, which were once
predominant. However, these graphs show that, due to the
presence of electrostatic charges on almost all
respirator filters that are currently sold, the minimum efficiency and greatest penetration
occur for particles that are about 30 to 60
nanometers in diameter. Therefore, the NIOSH filter ratings may not
reflect the true minimum efficiency for all
particles. In the case of the N95 filters in the left figure, the
N95 rating reflects performance appropriately because the penetration is less than 5% for all
particle diameters.
For the P100 filters in the right figure, the
efficiency rating is suitable because the
penetration is less than 0.03%, meaning efficiency is greater than 99.97%, for
all particles. However, not every filter will maintain its rating at
300 nanometers for all nanoparticles. We can see from the upper curve in the figure on
the left, for example, that the efficiency is
between 95 and 96% at the most penetrating particle size; the N95 rating is just barely
appropriate. For those working with nanoparticles, a key
recommendation is to consider using P100
respirator filters that are likely to have very high efficiency for all particle sizes rather N95
respirator filters that may not be sufficiently
protective at critical particle diameters.
What do we know about changes in respirator
filter penetration with loading of particles? Moyer and Bergman performed a study in which
they looked at the loading of respirators with sodium chloride particles that included many
nano-scale particles. With steady loading, they observed that
penetration decreased (efficiency increased)
with particle loading, which is good. However, they also looked at intermittent
loading, and they found that, when they loaded
the respirator filter for short periods, left the filter out unprotected in the open air, and then came
back repeatedly and loaded that filter again, the
initial penetration each day gradually increased over time even though the penetration within a
single day would decrease with loading. The reason for this increase in daily initial
penetration has never been explained
adequately. Therefore, another key recommendation for
those working with nanoparticles is to consider using respirator filters for only a single work shift
before switching to a new, clean filter.
Here are a few key points related to respiratory
protection against nanoparticle exposures. Penetration through respiratory protection filters
is typically greatest for nanoparticles. For some respirator filters, penetration for
nanoparticles may exceed their NIOSH rating because the filters are rated using 300
nanometer particles rather than nanoparticles. The change interval for respiratory protection
filters should be short if the respirator filters are
actually collecting nanoparticles. If the respirators are being used for secondary
protection and are not collecting a significant amount of nanoparticles, then they can be used
repeatedly. However, if the filters are loading with
nanoparticles during a shift, then go ahead and
change the filters out at the end of the shift.
After all, filtering facepieces and the filters used
with elastomeric respirators are meant to be
disposable. We should also remember that respirator fit is
vitally important for those working with nanomaterials, just as it is when working with
any other contaminant. The fit to the face must be good, or those
wearing the respiratory protection will be
exposed. To summarize this module, fibrous filters can
collect airborne nanoparticles with high efficiency, and they are the best air cleaning
technology for nanoparticles.
For now, changes in filter performance with time
as filters collect and load with engineered
nanoparticles are hard to predict. Therefore, filter change-outs should be guided by
efficiency considerations as well as by pressure
drop considerations. Also, filters must be seated properly in filter
housings in order to work effectively, or, in the case of respirators, the respirators must fit well
to the face in order for respirator filters to be
effective. Thank you for your attention!.