United States
Environmental Protection
Agency
Office of Research and Development
National Homeland Security Research Center
-------
-------
EPA/600/R-09/009 | Feburary 2009 www.epa.gov/ord
HRTI
I I- K
FINAL REPORT ON
Resuspension of Fibers From Indoor
Surfaces Due to Human Activity
Prepared by
Jonathan Thornburg and Charles Rodes
Center for Aerosol Technology
RTI International
Post Office Box 12194
Research Triangle Park, NC 27709
Order Number: 3C-R321-NANX
Submitted by
Charles Rodes, PhD
Project Manager
Prepared for
Jacky Rosati, PhD
Project Officer
U.S. Environmental Protection Agency
National Risk Management Research Laboratory
Research Triangle Park, NC 27711
-------
RTI/08924 March, 2005
Notice
This report is submitted in fulfillment of Order Number 3C-R321-NANX by RTI International under
the sponsorship of the United States Environmental Protection Agency. This report covers a period
from October 2003 to March 2005, and work was completed as of March 4, 2005.
-------
Disclaimer
The U.S. Environmental Protection Agency through its Office of Research and Development funded
and managed the research described here under contract to RTI International. It has been subjected to
the Agency's review and has been approved for publication. Note that approval does not signify that
the contents necessarily reflect the views of the Agency.
-------
Abstract
This research was directed toward 1) determining the quantity of asbestos simulant fibers resuspended
(emitted) as normalized by the amount deposited, and 2) calculating asbestos simulant fiber emission
factors at two heights while walking on and vacuuming seeded carpet. The asbestos fiber simulant
selected for this research was calcium silicate, commonly known as Wollastonite. Three methods for
measuring the quantity available for resuspension were studied:
a. Micro Vac method following a modified version of ASTM D5755-95,
b. Ultrasonication method developed by Millete et al. (1993),
c. Individual carpet fiber analysis via scanning electron microscopy.
Wollastonite resuspension during walking and vacuuming was studied. Total quantity and size
dependent fractions of resuspended Wollastonite were measured gravimetrically and with real-
time aerodynamic particle instrumentation, respectively. Established experimental procedures were
followed to seed new and old carpet with Wollastonite, characterize the quantity and size distribution
of simulant fibers deposited on the carpet, resuspend the simulant fibers within an exposure chamber,
and collect representative samples of resuspended fibers. The research defined a fractional carpet
resuspension emission factor as ratio of Wollastonite resuspended and Wollastonite available for
resuspension on the surface. The best method for estimating the amount available for resuspension
was the modified Micro Vac technique. This simple method only collected Wollastonite from the upper
carpet fiber surfaces that were potentially available for resuspension; Wollastonite fibers embedded
deep in the carpet with low probability of being resuspended were not collected. SEM analysis of
individual carpet fibers worked only for new carpet fibers. Old carpet samples typically had too many
"background" particles that confounded the analysis. The Millete et al. ultrasonication method poorly
estimated the quantity available. The removal of a carpet plug and sonic bath released a very high
number of carpet material particles that completely overwhelmed the ability to detect Wollastonite.
Simulant fiber emission factors ranged from < 0.01 to 0.45, with the majority falling between 0.01 and
0.10. As expected, experimental conditions (primarily resuspension method, carpet age, and relative
humidity) affected the emission factors. The majority of Wollastonite fibers resuspended from carpets
were between 2 and 10 mm, with particles between 2 and 6 mm yielding the highest mass emission
factors. The vacuum beater bar did resuspend a significant number of sub-micrometer particles that did
not contribute much to the mass resuspended. Emission factor testing did not elucidate the influences
of electrostatic and surface tension adhesion forces between the Wollastonite and carpet fibers in
determining the amount available for resuspension. Further investigation of these mechanisms and
their influence on emission factors will provide the requisite data needed for robust modeling exposure
to resuspended particles.
-------
Table of Contents
Notice iv
Disclaimer v
Abstract vi
List of Figures viii
List of Tables ix
1.0 Introduction 1
1.1 Scope 1
1.2 Research Objectives 1
1.3 General Approach 1
1.4 Underlying Variables 2
1.5 Data Presentation 2
2.0 Experimental Methods 3
2.1 Instrumentation & Procedures 3
2.2 Tests Conducted 7
2.3 Emission Factor Calculations 7
3.0 Results & Discussion 11
3.1 General Results 11
3.2 Wollastonite Available Method Comparison 12
3.3 Resuspended Wollastonite Measurement Method Comparison 19
3.4 Wollastonite Size Distribution 19
3.5 Emission Factors 21
3.6K-Factors 24
4.0 Quality Assurance 31
4.1 URG Samples 31
4.2 Micro Vac Samples 31
4.3 Deposition Chamber Samples 32
4.4 Ultrasonication 32
4.5 SEM Image Analysis 34
4.6 Aerodynamic Particle Sizer 34
4.7 Temperature & Relative Humidity 34
5.0 Conclusions 35
6.0 References 37
Appendix A: SEM Image Analysis A-l
Appendix B: Relationship between Aerodynamic and Fiber Diameters B-l
-------
List of Figures
Figure 2-1. Aerosol generator and deposition chamber used to load Wollastonite on carpet 5
Figure 2-2. Relationship between mass injected into deposition chamber with mass
deposited on carpet surface 5
Figure 2-3. Schematic of Large Dynamic Chamber used in resuspension tests 6
Figure 2-4. Particle transport efficiency in APS isokinetic sample line in Large
Dynamic Chamber 7
Figure 3-1. Conversion of total Wollastonite mass to total number of Wollastonite fibers
per 1296 in2 of carpet 16
Figure 3-2. Relationship between total Wollastonite mass resuspended during a 5 minute
testandthe average number concentration 17
Figure 3-3. Relationship between mass resuspended estimates measured by URG and APS 20
Figure 3-4. Relationship between resuspended Wollastonite counts measured by URG
and APS 21
Figure 3-5. Average size dependent, mass based emission factors calculated from APS and
SEM Fiber data for high and low relative humidity tests 26
Figure 3-6. Average size dependent, count based emission factors calculated from APS and
SEM Fiber data for vacuuming and walking resuspension method tests 28
Figure A-l. Comparison of RTI and METI aerodynamic diameters A-2
Figure B-l Comparison of fiber aerodynamic (from APS) and fiber cross sectional diameters B-l
-------
List of Tables
Table 2-1. NyglosMS Wollastonite fiber size distribution 3
Table 2-2. Carpet characteristics 4
Table 2-3. Description of tests conducted in RTI exposure chamber 8
Table 3-1. Background particle mass found in tested carpets by multiple methods
for determining mass available for resuspension 11
Table 3-2. Experimental conditions recorded during each test 13
Table 3-3. General experimental results presented as particle mass 14
Table 3-4. General experimental results presented as particle count 15
Table 3-5. Correlation coefficients between different estimates of mass available
for resuspension 18
Table 3-6. Correlation coefficients between different estimates of counts available
for resuspension 18
Table 3-7. Size distribution of Wollastonite deposited on new and old carpet as measured
from ultrasonication filters and individual carpet fibers 22
Table 3-8. Size distribution of Wollastonite resuspended via walking and vacuuming
as measured by the APS 22
Table 3-9. Emission factors calculated on total mass resuspended for each test condition 23
Table 3-10. Table of p-values for each method of calculating mass emission factors (EmFa) 23
Table 3-11. Emission factors calculated on total count resuspended for each test condition 25
Table 3-12. Table of p-values for each method of calculating count emission factors (EmFa) 25
Table 3-13. Table of size dependent APS/SEM Fiber mass emission factors for each test
condition using new carpet 25
Table 3-14. Table of p-values for APS/SEM Fiber mass emission factors including particle
diameter as an independent variable 26
Table 3-15. Table of size dependent APS/SEM Fiber count emission factors for each test
condition using new carpet 27
Table 3-16. Table of p-values for APS/SEM Fiber count emission factors including particle
diameter as an independent variable 27
Table 3-17. Aggregate K-factors for each test 29
Table 3-18. Size specific K-factors for each test using new carpet. K-factors calculated
from emission factors in Table 3-15 29
Table 4-1. QA/QC criteria for measurements collected 31
Table 4-2. Quality control measures to implement during testing 32
Table 4-3. URG sample completeness, filter blank, precision, and accuracy statistics 33
Table 4-4. Micro Vac sample completeness, filter blank, precision, and accuracy statistics 33
Table 4-5. Deposition chamber filter sample completeness, filter blank, precision,
and accuracy statistics 33
Table 4-6. Gravimetric mass as measured by ultrasonication: sample completeness,
blank, precision, and accuracy statistics 33
Table 4-7. Size distribution from SEM images of ultrasonication filters: sample
completeness, blank, precision, and accuracy statistics 33
Table 4-8. Aerodynamic Particle Sizer sample completeness statistics 34
Table 4-9. QA/QC results for the HOBO H8 temperature and relative humidity data 34
-------
-------
1.0
Introduction
1.1 Scope
The events of 9/11 and the subsequent related activities
released an enormous particulateload into the environment
proximal to the WTC complex. This aerosol burden included
building material fragments that contained asbestos and
other mineral fibers known or suspected to be health hazards.
Although much of the aerosol subsequently deposited
outdoors, a significant portion penetrated into nearby
residences and businesses. The size distributions and quantity
of the aerosol relative to the outside environment were
strongly affected by the filtration ability of the ventilation
system and building envelope. The model of Thornburg et al.
(2001) provides a means of predicting the resultant aerosol
sizes and quantities that would have been found in these
buildings. A large fraction of the aerosols that penetrated
these buildings deposited onto horizontal and vertical
surfaces. Normal occupant activities can resuspend portions
of the deposited aerosol on horizontal surfaces. Walking
on medium pile carpeting is known to resuspend particles
in the 2 to 10 mm range (Rodes and Wiener, 2001). This
aerosol can result in unhealthful asbestos and other mineral
fiber aerosol concentrations within the general vicinity of
the activity and translocate the aerosol to other locations and
surfaces within the building.
The research described here seeks to relate concentrations of
asbestos-type fibers deposited upon carpeting by processes
similar to those that occurred after 9/11 to the airborne
concentrations that may have resulted from normal resident
activities, such as walking and vacuuming. The general
focus of this pilot effort was problem-identification rather
than phenomenon characterization, since a) it is clearly
retrospective, b) only simulants representing the real aerosol
are available, and c) only a limited number of samples will be
collected and analyzed within the resources available. This
work built upon dust resuspension characterization methods
for generic PM on carpeting studying the resuspension
of generic paniculate matter and metals (respectively)
from flooring surfaces due to human activity (Rodes and
Thornburg, 2004; Thornburg and Rodes, 2004). These pilot
efforts provided the fundamental experimental methodologies
and limited data to determine whether the resuspension of a
tracked-in particles and metals from soil dust deposited on
medium pile carpeting was significant and warrants further
study. An important outcome of this research was collection
of data that could be used in subsequent multi-pathway
exposure modeling. This focus involved an experimental
design that carefully considered how lifestyle factors such
as adult and children's activity levels, residence volumes,
fraction of flooring with carpeting, etc. could be used in
exposure models in conjunction with the more fundamental
data from this study.
1.2 Research Objectives
1. Determine quantity (mass and number) of simulant
fibers available for resuspension, as compared to
deposited mass, using a variety of techniques that
included:
a. Micro Vac method following a modified version of
ASTMD5755-95.
b. Ultrasonication method developed by Millete et al.
(1993).
c. Individual carpet fiber analysis via scanning
electron microscopy method (Thornburg et al.,
2006)
2. Calculate simulant fiber emission factors at two heights
while walking on and vacuuming seeded carpet.
Aggregate and size dependent emission factors were
calculated on a mass and count basis.
1.3 General Approach
The key elements of the research approach defined to address
the objectives given in Section 1.1 were:
1. Select an appropriate asbestos fiber simulant
2. Develop or identify a methodology to deposit and
embed (seed) known quantities of fiber simulant on
carpet that emulates probable post-9/11 activities.
3. Select a vacuum cleaner representative of a
homeowner quality unit rather than a HEPA unit that
would be used in remediation efforts.
4. Develop a project Quality Assurance Project Plan
before beginning experimental work (see: "Indoor PM
Resuspension Testing of Fibers - Quality Assurance
Project Plan For Basic Research Projects, EPA Order
No. 3C-R321-NANX, EPA/NRMRL under Contract
No. QT-OH-03-00572 RTI Project No. 08924)
5. Obtain new and old carpet for experimental use.
Characterize background loading of particles on carpet.
Condition carpet as necessary to obtain acceptable
background levels.
6. Use existing RTI standardized test methodologies,
incorporate other established procedures, and devise
new methods as necessary that consider fiber simulants
by particle size that define:
a. How to characterize the fiber simulant loadings on
carpet fibers by methods specified in the project
objectives that consider how much dust is available
for resuspension, as a function of particle size.
b. How to characterize carpeting in terms of pile
height (fiber length), age, soiling history, and
surface loading level.
-------
c. How to resuspend particles in a controlled
environment with minimal interference from
background particles, load carpeting with defined
areal loadings of a selected dust,
d. How to quantify air concentrations as function of
height, fully correcting for transmission line losses
between the measurement points and the sensing
zone, thus facilitating mass balance closure by
accounting for the (large) fraction of particle mass
often lost to internal surfaces
7. Conduct controlled walking and vacuuming
experiments on new and/or old carpet loaded
with up to three different quantities of fiber simulant.
1.4 Underlying Variables
Two variables that probably influence resuspension of
particles from carpets were not studied during this project
because of the limited scope of work. However, our
hypotheses and suppositions regarding these variables based
on our carpet research experience are presented below
because these concepts aid in the interpretation of the data
collected during these experiments.
1.4.1 Flooring Surfaces
Previous RTI resuspension data showed that as the pile
height decreased, the level of resuspension decreased
substantially from normal walking. No measurable dusts
were observed from dusts on bare flooring. It could be
conjectured that resuspending dust from bare floors requires
substantial turbulence from either stomping or very fast
walking to provide the energy to both release particles from
the surface and elevate them into the air sufficiently to add
to the air concentration. Low pile, indoor-outdoor carpeting
also provided essentially immeasurable air concentration
levels. Thus, the current work focused only on medium pile
carpeting (-70% of all new carpeting sold) which has been
shown to contribute significantly to resuspension.
1.4.2 Particle Adhesion
An important factor felt to be extremely important in
understanding resuspension of particles from carpet fibers is
adhesion. Adhesion of particles to carpet fibers is influenced
by relative humidity (Rodes and Thornburg, 2004), thought
to be controlled primarily by electrical charging of both the
fibers and the particles. Very low humidities are routinely
found to increase the charging of certain formulations of
carpet fibers. An additional adhesive force considered here
is surface tension: particle-to-particle and particle-to-fiber.
Surface tension forces bonding particles together or to the
fibers is felt to be very important at relative humidities above
-45%. At 45% Rh, sufficient water is present to increase
the surface adhesion force by increasing the contact angle
between the particle and a second surface (Ranade, 1987).
Both of these types of adhesive forces work together to bond
particles to surfaces. Undoubtedly, resuspension occurs when
sufficient energy is imparted to exceed the cumulative force
levels.
Our extensive scanning electron imaging of carpet fibers has
shown that as carpeting ages or becomes significantly soiled
(including coating of the fibers over time by grease aerosol
from cooking), the carpet fibers ability to generate static
charging appears to decrease substantially. This substantial
change in potential adhesion characteristics for particles to
fibers between new and old carpeting was addressed in the
current research by considering either new, unsoiled carpeting,
or soiled carpeting that was at least 1 year (or substantially
more) old as a binomial variable.
No efforts were made to measure either type of adhesion in
these experiments, but temperature and relative humidity
were recorded during all tests to determine if the influences
from these types of adhesions could be estimated categorically.
Successful modeling of particle resuspension will at some
point require more detailed investigation of the relationships
among these factors.
1.5 Data Presentation
The wide array of methods used to measure the asbestos
fiber simulant size distribution and concentration on the carpet
fibers and resuspended required selection of a reference metric.
Asbestos simulant size was measured via microscopy to yield
a projected area diameter, and via aerosol instrumentation
to yield an aerodynamic diameter. In addition, the bulk
asbestos fiber simulant size distribution was measured in
terms of optical diameters. For the purposes of this research,
aerodynamic diameter was selected to characterize the
asbestos simulant size distribution. Aerodynamic diameter is
the standard reference for all aerosol research. As such, factors
are available for converting other diameter measurements to
aerodynamic diameter. For example, projected area diameters
can be converted to aerodynamic diameters using established
conversion factors as described in Section 2.1.9. However,
a conversion between optical and aerodynamic diameters
does not exist because of the complexity surrounding
particle refractive indices. Therefore, a basis does not exist
for comparing the aerodynamic diameters reported from the
microscopy measurements and aerosol instrumentation with
the optical diameters used to characterize the bulk asbestos
simulant size distribution.
Similarly, microscopy measurements provided asbestos
simulant fiber count data, realtime aerosol instrumentation
provided either count or mass concentration data, and
gravimetric analysis yielded mass concentration data.
Gravimetric analysis determined the quantity of asbestos
simulant deposited on the test surfaces and one method
for measuring the quantity of simulant resuspended during
the experiments. Without a priori knowledge of the bulk
simulant aerodynamic size distribution (due to the lack
of an optical-aerodynamic conversion discussed above)
an established method for converting mass data to count
data was not available. Asbestos simulant count data from
microscopy and real-time aerosol instrumentation easily were
converted to a mass basis from the volume and density of the
measured particles. To provide an estimate of the deposited
and resuspended asbestos simulant count concentrations,
regression curves relating the mass to the count concentrations
are presented in Section 3.1.
-------
2.0
Experimental Methods
2.1 Instrumentation & Procedures
The instrumentation and procedures selected determined the
particle aerodynamic diameter and concentration, either mass
or number, of the particles resuspended from the carpet or
embedded on the carpet fibers.
2.1.1 Asbestos Fiber Simulant
Calcium Silicate (CaSiO3), commonly known as
Wollastonite, was the asbestos fiber simulant selected.
Wollastonite fibers were the simulant of choice for asbestos
occupational exposure studies because of their acicular
nature. Nyglos M3 (NYCO Minerals, Willsboro NY) was
used (Table 2.1). Wollastonite fibers had a mean aspect ratio
of 3.33 ± 1 and density of 2.9 g/cm3.
2.1.2 Carpet Characteristics
New and old carpets were obtained for the resuspension
experiments (Table 2-2). New carpet was beige, medium
pile carpet manufactured by Shaw Carpets or Beaulieu. The
carpet was purchased retail and was removed directly off the
main roll. New carpet was vacuumed multiple times with a
standard household vacuum to remove excess loose fibers
and dirt. Old carpet was obtained from the same retailer after
removal from unknown private residences. The age, use,
manufacturer, and history of the old carpet was unknown.
Old carpet exhibiting normal wear and soiling were selected;
carpets with excessive wear and obvious soiling were
avoided. Old carpet was cleaned thoroughly with a rental
hot water carpet cleaner and vacuumed multiple times with a
standard vacuum before experimental use.
2.1.3 Deposition Chamber
New and old carpet was seeded with Nyglos M33
Wollastonite fibers using a flow through deposition chamber
(Figure 2-1). Simulant dust was aspirated and dispersed into
a laminar flow transport pipe that carried the fibers to the
injection head at the top of the chamber. A mixing baffle at
the injection head evenly dispersed the fibers across the cross
section of the chamber. The velocity at the injection head
outlet was balanced with the downward air velocity generated
by a fan that pulled the fibers into the carpet.
A known mass of Wollastonite was injected into the chamber.
47 mm Teflo® filters placed in a 9-point grid pattern across
the surface of the carpet collected samples of deposited
Wollastonite to determine gravimetrically the total mass
deposited. Mass deposited on the carpet was calculated by
taking the average mass loading (mg/in2) measured per filter
and multiplying by the area of exposed carpet (Ac),
(2-1)
where F is the filter mass and d2 is the filter diameter.
m
Carpet loadings in the chamber were found to be linear with
injected dust mass, and predictable, having an R2 value of
0.96. Figure 2-2 illustrates the experimental utility of the
chamber, showing the linearity and strong correlation of the
regression of injected mass versus deposited mass. Without
demonstrating this linearity, it would have been inappropriate
to normalize the emission factors by loading. Note a leak in
the injection nozzle was identified after all experiments were
completed. This leak limited the percentage of Wollastonite
deposited to 2%. However, the percentage deposited was
linear with quantity injected. As a result, the quality of the
Wollastonite resuspension data was not compromised.
Table 2-1. Nyglos M3 Wollastonite fiber size distribution.
Median diameter = 2.5 urn by Cilas Granulometer
<0.5
0.09
0.09
0.11
0.20
0.22
0.42
0.16
0.58
0.14
0.72
0.09
0.81
0.08
0.89
10
0.07
0.96
0.04
-------
Table 2-2. Carpet characteristics
,V PUPIR
»i!Hj<'4!
Mi^iNii^rt!!
New#l
New #2
Old #1
Old #2
C*W«
ii'litaiU-v
Shaw
Beaulieu
unknown
unknown
j jpffwfvfl
i liifeiMNiiiiiilii!!^!!!!
Nylon 6,6
Nylon 6,6
unknown
Nylon 6,6
i f'HIPIi'H'I'f}!'!
1 jiiiiili NiiiiiiiiiiiiiiiiiNiiiiiiiiiiiiiiiiiiiH
4000
4000
3000a
2500a
i 'i'tliltlMfiWl
i iillliliiiiiiiil i'iitpiiiiiife;.^!!!!!!!!
10 mm
8 mm
10 mm
12 mm
flB
liNi!!:!!!:!!^!:! faiteM
Walking
Vacuuming
Walking
Walking&
Vacuuming
"Estimated by counting number of fibers per loop and number of loops per in2.
2.1.4 Exposure Chamber
All tests were conducted in the RTI Large Dynamic Chamber
(Figure 2-3). This positive pressure chamber equipped with
six pre-filters and six ultra low penetration air (ULPA) filters
can provide essentially particle-free air at flows ranging from
0.25 to 80 mVmin. For these tests, the flow was 13.2 m3/
min; equivalent to a linear face velocity in the test section
of 10 cm/s. A flow distribution baffle preceding the filters
creates a uniform velocity profile through the chamber. The
test space within the chamber is 4 ft wide x 7 ft long x 6 ft
high. Carpet (3 x 3 ft) samples were placed in this space.
A contraction of 12° over 2.4 meters with a mixing baffle
immediately upstream of the 20 cm diameter outlet pipe
uniformly concentrates the aerosol generated in the test
space. Mass samplers were installed in this section to collect
PM10 filter samples isokinetically for gravimetric analysis.
Sampling ports in the contraction allow direct sampling from
within the test space at the desired heights and with multiple
instruments. A Model 3321 Aerodynamic Particle Sizer (TSI
Inc., Minneapolis MN) was connected to sample ports in
the outlet pipe for isokinetic collection of resuspended PM
samples for size distribution measurements. Temperature and
relative humidity were monitored with a HOBO H8 Data
Logger (Onset Computer Corp., Bourne MA) installed in
the test space. Preliminary tests confirmed air entering the
chamber and activities conducted by test personnel wearing
cleanroom coveralls inside the chamber did not generate
a significant number of particles above background. As
an extra precaution the chamber test section surfaces were
vacuumed and wiped with clean, damp cloths prior to
each test.
2.1.5 Resuspension Methods
Walking and vacuuming were the resuspension methods
selected. A single volunteer walked randomly across the
carpet area for 5 minutes during an experiment. The volunteer
was a 73 inch, 170 Ib male with size 12 shoes. A particle free
cleanroom garment was worn for all tests. The same shoes
were worn during all walking experiments. A pedometer
worn by the volunteer determined the number of steps
taken by the volunteer during the experiment. The volunteer
walked across each test carpet for 3 minutes prior to seeding
with Wollastonite to determine the background resuspended
particle concentration. The subject attempted to apply a
constant foot pressure (energy) to the carpet during since
previous work showed emission factors varied with energy
level. A standard 13 amp Mach 2.1 Hoover (Model # U5330-
900) upright vacuum with beater bar was used. The entire
vacuum was cleaned thoroughly prior to the experiments. A
new Hoover Type Y Allergen® vacuum bag (99.98% filtration
efficiency) was installed for each experiment. The vacuum
was ON for approximately 5 minutes before an experiment
began to allow particles generated from the starting of the
motor to leave the chamber. A different volunteer, wearing
a particle free cleanroom garment, randomly moved the
vacuum across the carpet in all directions for 5 minutes
during a test. The volunteer did not walk on the carpet during
the tests.
2.1.6 Carpet Fibers
Individual carpet fibers were collected from two locations
within each piece of test carpet after seeding with
Wollastonite. Fibers were analyzed via SEM to quantify
the size distribution of the Wollastonite on the carpet fibers
available for resuspension. A dial micrometer was used to
estimate the carpet pile height at multiple locations for each
type of carpet.
-------
1 54 in. PVC ,
Top of dust
deposition
chamber
_ Aerosol and earner air
from aerosol generator
Figure 2-1. Aerosol generator and deposition chamber used to load Wollastonite on carpet
150
125
100
o>
£
«
H
ra
I 75
I
M
O
o.
O
Q
50
25
Deposited = 0.02 * Injected + 4.9, R2 = 0.96
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500
Injected Mass, mg
Figure 2-2. Relationship between mass injected into deposition
chamber with mass deposited on carpet surface.
-------
From chamber outlet plus
mate-up air (If needed)
Variable speed blower
3000 cfm, 8" H O
Air Plenum
Removable
i.i_:•-•', Bank
Test Chamber
71 X 4'W X 6,5'H
Return to fan
inlet of exhaust
Sample Ports
(not shown)
Figure 2-3. Schematic of Large Dynamic Chamber used in resuspension tests.
2.1.7 Carpet Vacuum Samples
The quantity of paniculate matter potentially available
for resuspension was estimated by collecting two vacuum
samples onto 47 mm Teflo® filters from an area adjacent to
where the carpet fibers were extracted per Section 2.1.6.
The vacuum used was custom designed by RTI. A modified
ASTM method D5755-95: Standard Test Method for
Microvacuum Sampling and Indirect Analysis of Dust by
Transmission Electron Microscopy for Asbestos Structure
Number Concentrations was followed for the sample
collection. The RTI modifications included: using a brush
on the vacuum nozzle to trap carpet fibers, reducing the
sampling velocity to 45 cm/s to minimize the aggressiveness
of the procedure, manual removal of obvious (visible
to naked eye) carpet fibers collected on the filter, and
gravimetric analysis of the filter instead of liquid extraction
for microscopic analysis.
2.1.8 Ultrasonication Method
Samples of carpet for ultrasonic extraction of particles
were collected following a modified version of Millette
et al. (1993). Approximate 20 cm2 pieces of carpet were
carefully removed from the test carpet following seeding
with Wollastonite fibers for the ultrasonic extraction. Carpet
pieces were immersed fibers down in a 1 L beaker containing
100 ml of a 0.002% solution of methyl cellulose surfactant
in particle free water. The entire beaker was placed in an
ultrasonic bath for 30 minutes. The carpet piece then was
removed and rinsed with 100 ml of particle free water,
raising the total volume to 200 ml. The entire solution was
shaken vigorously to disperse particles then allowed to sit
for 2 minutes. Only 50 ml aliquots of solution were extracted
from % to !/2 inch below the surface. Smaller aliquots did not
yield sufficient number of particles for accurate SEM image
analysis. The 50 ml aliquot was passed through a coarse
metal screen to remove large carpet material pieces placed
over a buchner funnel. The coarsely screened aliquot was
filtered through a 47 mm, 0.2 um pore polycarbonate filter,
backed by a 0.45 um pore cellulose ester filter placed inside
the funnel. The filters were equilibrated inside a temperature
(23 °C) and humidity (35%) controlled chamber for at least 24
hours before transfer to U.S. EPA for SEM imaging. A subset
of the ultrasonication filters underwent gravimetric analysis
to provide comparative data for the Wollastonite fiber mass
measured via SEM.
2.1.9 SEM Image Analysis
Mantech Environmental (METI) provided scanning electron
microscopy (SEM) images of individual carpet fibers and the
ultrasonication filters. The aerodynamic diameter of each
particle contained in the images was determined using
image processing software and established algorithms. The
algorithms disregarded particles with aspect ratios less than 2
(non-Wollastonite fibers) and converted the measured particle
projected area to an equivalent diameter then an aerodynamic
diameter using standard dynamic shape and volume shape
factors for fibers (Hinds, 1982). SEM images at 880x
magnification were provided, equivalent to 5.059 pixels per
micrometer, providing a particle minimum detection limit of
0.4 mm. Additional details are provided in Appendix A.
2.1.10 Aerodynamic Particle Sizer
The Aerodynamic Particle Sizer was used to obtain re-
suspended PM mass and count size distribution data from
0.5 to 20 um. The instrument was operated according to the
instructions provided by the manufacturer. Isokinetic samples
were collected through the sample probe installed in the
outlet pipe. Particle losses within the sampling tube, shown
in Figure 2-4, were estimated using standard aspiration
and transport equations for laminar flow accounting for
gravitational and inertial deposition losses (Brockman, 2001).
Counting efficiency errors were corrected using the data from
Peters and Leith (2003).
2.1.11 URG Mass Sampler
URG mass samplers (University Research Glassware,
Chapel Hill, NC) collected PM10 samples isokinetically at
16.7 Lpm within the test chamber. Two samplers were used
per test: one at 18" and the other at 36" above the floor.
However, mixing within the chamber probably eliminated
any gradient in resuspended PM concentration. Samples were
collected on 47 mm Teflo® filters. Filter mass was determined
gravimetrically.
2.1.12 Gravimetric Analysis
Aerosol mass collected on filters were weighed in RTI's
temperature and humidity controlled (23 °C, 35% Rh) weigh
chamber on a Mettler Toledo MT2 balance with 0.1 ug
resolution.
-------
2.1.13 Temperature & Relative Humidity
Temperature and relative humidity within the homes was
measured with a HOBO H8 Data Logger. Temperature and
relative humidity were recorded at 1-minute intervals.
2.1.14 Statistical Analysis
All statistical analyses were performed using SAS version
9.1 (SAS Inc., Gary NC). The General Linear Models module
was used to determine the significance of experimental
parameters on emission factors. Forward addition and
backward subtraction methods were used to identify the
statistical significance of the experimental parameters.
2.2 Tests Conducted
The experimental design selected provided Wollastonite fiber
emission factor data for various resuspension mechanisms,
carpet characteristics, and environmental conditions. Walking
and vacuuming were the resuspension mechanisms. Carpet
characteristics encompassed new/unmatted carpet with strong
static charge and old/soiled/matted carpet with little static
charge. Humidity ranged from typical, indoor air conditioned
values to extremely high simulating buildings without air
conditioning during precipitation events. Potential variations
in emission factors as a function of Wollastonite loading were
investigated as well. Total Wollastonite mass deposited on
the carpet surface varied from ~25 mg to -60 mg to -100
mg, representing low, medium, and high loading levels,
respectively.
Emission factors were calculated multiple ways.
Resuspended Wollastonite concentrations were measured
with the APS and URG instruments. The Wollastonite
available for resuspension was calculated from vacuum
samples, SEM images, and ultrasonication of carpet sections.
2.3 Emission Factor Calculations
Emission factors were calculated multiple ways.
Resuspended mass concentrations were measured with the
APS and URG instruments. Corresponding resuspended
Wollastonite count data were provided by the APS or
calculated from the URG gravimetric data. The mass
available for resuspension was calculated from vacuum
samples, SEM images, and ultrasonication of carpet sections.
SEM images provided available Wollastonite count data
directly. Wollastonite count data from gravimetric samples
were estimated. Emission factors were normalized by the
number of steps made during an experiment.
0.9
0.8
0.7
>s
o
c
-------
Table 2-3. Description of tests conducted
in RTI exposure chamber.
^fefel
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
y>;ifea/
New#l
New#l
New#l
Old#l
Old#l
Old #1
New#l
Old #2
New #2
New #2
Old #2
Old #2
New #2
Old #2
Old #1
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Walk
Mm%
Low
Medium
High
Low
Medium
High
Medium
Medium
Low
High
Low
High
High
High
Medium
1«%
Low
Low
Low
Low
Low
Low
High
Low
Low
Low
Low
Low
High
High
High
2.3.1 Wollastonite Resuspended by APS
Mass resuspended measured by APS was calculated using Eq.
2-2. The mass concentration in each size bin (CR) corrected
for the background resuspended particle concentration (CBkg)
and transport efficiency into the APS (m.) were summed to
yield a total concentration. This concentration was multiplied
by the flow through the chamber (Qc = 0.013 Lpm), the total
time particles were resuspended (t), and a correction factor
for APS counting efficiency (x =2). The total time particles
were resuspended was not constant across all tests because of
depletion of the particles available for resuspension.
Resuspended Wollastonite number concentrations (Cuga)
were estimated from the gravimetric mass and APS mass size
distribution data. The cumulative gravimetric mass (MURG)
was multiplied by the mass fraction in each APS size bin (f)
then converted to particle counts assuming each fiber was an
aerodynamic sphere.
M APS =
x n
(2-2)
The APS also provided resuspended Wollastonite number
concentration data. The total counts resuspended (CAPS)
were calculated using an equation similar to Eq. 2-2.
(2-5)
2.3.3 Wollastonite Available by MicroVac
Mass available by MicroVac was calculated from the
gravimetric mass collected on the filters, Mfilter divided by the
total area vacuumed (Avac = 9 in2) and multiplied by the area
of the test carpet piece (Ac = 1296 in2).
filler
1
x —
x A
(2-6)
2.3.2 Wollastonite Resuspended by Gravimetric Data
Mass resuspended as measured by the URG samplers was
calculated from the gravimetric mass collected on the filters,
Mfilter divided by the URG sample flow (Qma = 20 Lpm)
and multiplied by the flow through the chamber (Qc = 0.013
Lpm). Mass resuspended was not corrected for background
because concentrations either were not statistically different
from zero or contributed less than 5% of the total mass
collected (as measured by APS).
The number of Wollastonite fibers available for resuspension
was calculated from the MicroVac mass using an equation
similar to Eq. 2-5. The Wollastonite size distribution data
measured by the SEM images of the carpet fibers determined
the mass fraction in each size range.
, lf ,6
f*MVAC }—T P (2-7)
(2-4)
-------
2.3.4 Wollastonite Available by Carpet Fiber SEM
Images
Fiber SEM images provided cumulative and size dependent
estimates of the Wollastonite mass available for resuspension.
From the size distribution calculated in each image, the total
mass, MSEM, of particles on the 36" x 36" carpet section was
calculated based on the incremental mass in each size interval
(0 via the following equation,
M
SE.M
- x — xN
F,- N
xL-xA
(2-8)
where MFi is the incremental particle mass divided by the
total number of images processed, TV is the fiber length
visible per photo (101 um), TV is the number of millimeters
of fiber available for resuspension, Z, is the number of fibers
per unit area of carpet, and Ac is the area of carpet tested
(36" x 36"). A TV value for each carpet type was estimated
by scanning the entire length of the fiber and determining the
distance from the fiber tip that Wollastonite fibers were no
longer present. Maximum TV value was 4 mm, or 50% of the
fiber length.
The number of Wollastonite fibers available for resuspension
(CSEM) was determined directly from the incremental count
data in each size interval (CF.) provided by the SEM images.
1
(2-9)
2.3.5 Wollastonite Available by Ultrasonication
SEM images of ultrasonication filters provided a second
cumulative and size dependent estimate of the Wollastonite
mass available for resuspension. One photo per buchner
funnel hole was collected. Usually 5 to 10 holes were
photographed. The number of holes photographed depended
on the number of particles surrounding the hole. A sufficient
number of particles were desired to insure an accurate size
distribution. From particle size distribution calculated from
each image, the total mass, MSON, on the 36" x 36" carpet
section was calculated via the following equation,
Msov =
A.
N ,
(2-10)
where MF. is the incremental particle mass per size bin, A hoto
is the image area, TV hoto is the number of photos collected,
A. , is the area of a buchner funnel hole, TV. , is the total
hole ? note
number of buchner funnel holes (91), Ac is the carpet section
area tested, and A is the carpet area sonicated. Additionally,
? mm A •> ^
gravimetric analysis of the ultrasonication filters provided
another estimate of the Wollastonite mass available for
resuspension.
A similar equation estimated the total number of Wollastonite
particles available for resuspension using the particle count
data from the SEM images.
"
-
*
.
x
,
x <
x
photo
(2-11)
-------
-------
3.0
Results & Discussion
3.1 General Results
The background mass of particles available for resuspension
from the carpet and resuspended by walking and vacuuming
were measured for all four carpets tested. The background
mass does not contain Wollastonite particles, but may
contain other fibrous particles that affected the SEM analysis.
Background particle mass data affecting the mass available in
the carpet is presented in Table 3.1. All resuspension data for
a respective metric were corrected for the background mass.
As expected, the old carpets yielded higher background
particle mass, even after extensive cleaning. The Micro Vac
yielded a significant quantity of particles in addition to
carpet fiber fragments. The destructive ultrasonication
method yielded a very high mass of particles as measured
gravimetrically and via SEM. The high mass measured
via SEM for ultrasonication filters resulted from a large
number of fibrous type particles being collected on the filter.
Fiber SEM analysis provided a much smaller background
particle mass because it is a non-destructive technique. The
differences between the two types of old carpet was caused
by the presence of high concentrations of sawdust in Old #2
that presented as fibrous particles with an aspect ratio similar
to Wollastonite.
On the other hand, new carpets yielded much lower mass by
all methods. The small quantity collected by the Micro Vac
probably was carpet fiber fragments. The gravimetric analysis
of the ultrasonication filters from new carpets showed a
significant mass of carpet backing is generated during the
procedure. However, SEM analysis of these filters showed
these particles were not fibrous (aspect ratios < 2) such that a
mass and size distribution could not be determined. Similarly,
the lack of any particles on the new carpet fibers prevented
quantification of the mass and size distribution by the Fiber
SEM method.
Resuspension tests to characterize the background
concentration of particles generated during walking and
vacuuming were conducted. The background resuspended
particle mass was less than 0.05 mg, equivalent to a
concentration of 3 ug/m3. Background concentrations
were 10 times, minimum, lower than the concentration
resuspended during a test. Resuspended fiber mass data
measured by APS and URG systems were corrected for this
background mass.
General experimental conditions for all tests are presented in
Table 3-2. Note that depletion occurred during resuspension
tests with low Wollastonite mass loadings. Depletion of
the Wollastonite during these tests was accounted for in
the data analysis via Eqs. 2-2 and 2-3. Also, some of the
mass deposited estimates were invalid due to movement of
deposition coupons within the chamber during the carpet
loading.
The mass available and mass resuspended data from each test
are presented in Table 3-3. Overall, data quality objectives
for each metric were achieved except for the ultrasonication
SEM mass measurements. The total mass and number of
particles collected on the ultransonication polycarbonate
filters was sufficiently high that valid photographs over
buchner funnel holes could not be obtained. The particle
loading around the holes was sufficiently large for all
filters that it was impossible to accurately assess the size
distribution because of particle overlap. Only by moving
the image away from the buchner funnel holes could image
be collected of individually isolated particles so the size
distribution could be assessed.
Table 3-1. Background particle mass found in tested carpets by multiple
methods for determining mass available for resuspension. Mass data
presented in mg. Mass median diameter (MMD) and geometric standard
deviation (GSD) data are in mm.
New#l
9.0
482
New #2
4.6
238
Unable to Quantify
Unable to Quantify
Old#l
194
1161 5.05 2.01
7.29 1.88
Old #2
916
2049
3066 5.09 1.87
84
3.52
1.76
Ultrasonication gravimetric mass for Old #1 was invalid.
-------
Otherwise, only a few data points noted by asterisks,
collected during the experiments were invalid. Typical
reasons included movement of filters during Wollastonite
deposition on carpet, lost samples (Test 7a), accidental
destruction of polycarbonate filters during the ultrasonication
extraction procedure, or insufficient particles were measured
via SEM on carpet fibers for a statistically valid assessment
of mass and size distribution. Other samples, noted by
dashes, were collected but not analyzed because analytical
protocols were changed during sample analysis. The first
couple ultrasonication filters were not gravimetrically
analyzed until an adequate procedure for removal of
electrostatic charge was developed. Other ultrasonication
filters were not analyzed via SEM to expedite the data
analysis and limited resources.
Table 3-4 is similar to Table 3-3, except all data are
presented as Wollastonite counts. Because of the error in the
ultrasonication measurements, these data were not included
in the table.
As discussed in Section 1.5, the relationship between
mass and count is required to link the data collected by the
gravimetric and particle counter metrics. The Wollastonite
mass loaded on the carpet and the calculated number of
Wollastonite particles loaded onto the carpet is presented
in Figure 3-3. Similarly, the relationship between the
Wollastonite mass and number of Wollastonite particles
resuspended is shown in Figure 3-4. The regression
equations presented provide a conversion for comparison
of the data collected presented here with current occupational
exposure standards. The conversion from mass to counts
per 1296 in2 of carpet as measured by SEM images of carpet
fibers is shown.
3.2 Wollastonite Available Method Comparison
3.2.1 Mass Comparison
Table 3-5 presents correlation coefficients for the various
estimates of Wollastonite mass available for resuspension.
The primary correlation of interest is the comparison
between the mass deposited and the mass available. Direct
correlations between the various methods determining mass
available are shown for completeness. Since the background
particle and fiber masses associated with new and old carpets
were drastically different, correlations also are presented
separated by age of carpet.
The Micro Vac estimate of mass available correlated best
with the mass deposited on the carpet, regardless of carpet
age. Fiber SEM method correlated significantly only for
new carpet. The insignificant correlation for old carpet
indicates a large number of particles with aspect ratios
similar to Wollastonite were present. It was noted during
carpet preparation that Old #2 had a significant quantity
of wood chips, either sawdust or pet bedding, present.
These chips possibly degraded into micrometer size fibers
indistinguishable from Wollastonite by SEM. Gravimetric
mass measured on the ultrasonication filters were not
correlated with the mass deposited. The destructive nature
of the technique generated a very large quantity of carpet
backing particles with a total mass that easily exceeded the
quantity of Wollastonite deposited.
Based on these results, the Micro Vac provided the best
estimate of Wollastonite mass available for resuspension.
Therefore, these values will be used when calculating total
mass emission factors. Size dependent emission factors can
be calculated from the Fiber SEM estimate of mass available,
but only for the new carpet tests.
-------
Table 3-2. Experimental conditions recorded during each test. Asterisk (*) denotes invalid data collected.
ill
la
Ib
2a
2b
3a
3b
4a
4b
5a
5b
6a
6b
7a
7b
8a
8b
9a
9b
lOa
lOb
lla
lib
12a
12b
13a
13b
14a
14b
15a
15b
ill
New#l
New#l
New#l
New#l
New#l
New#l
Old#l
Old#l
Old#l
Old #1
Old#l
Old #1
New#l
New#l
Old #2
Old #2
New #2
New #2
New #2
New #2
Old #2
Old #2
Old #2
Old #2
New #2
New #2
Old #2
Old #2
Old#l
Old#l
!;\iUi||irt«BiM|i| if!
ii hi! i^ipilr^1*
1 1 ItttWMMf&l
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Walk
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Vacuum
Walk
Walk
i^iiiiiiS-l^^i^lil^lfe !$Mi 1%
*
18. 2 ±7.3
75.7 ± 13.2
65. 0± 18.2
99.2 ± 17.2
101. 3 ±25. 6
22. 7 ±8. 6
29.1 ±3.8
69. 2 ±8.4
71.4±3.9
101.6 ± 13.2
108. 3 ±8. 2
*
*
63. 7 ±6. 2
60. 9 ± 10.2
21.8 ± 1.0
20.5 ±2.1
98.1 ±9.7
100. 5 ±7.3
20. 3 ±2. 5
*
86.0 ± 12.2
91.2 ± 12.5
31.8 ±3.0
33. 9 ±4. 2
41. 2 ±6.2
33. 6 ± 18.1
21. 7 ±6.0
22.1 ± 1.0
i :Ai'ui,
40
45
37
44
45
45
51
47
51
47
48
50
80
90
48
48
41
40
41
40
41
40
41
40
90
90
90
90
90
90
1 §
72
70
73
70
70
70
71
71
71
71
71
71
72
69
70
70
71
72
71
72
71
70
72
72
70
70
70
70
70
70
1 ill I
240
250
230
240
260
220
250
230
250
220
270
270
250
260
220
270
-
-
-
-
-
-
-
-
-
-
-
-
225
245
, iy:J iuf :,•
Deposition mass data invalid.
Depletion @ 2 min
Depletion @ 4 min
Deposition mass & MicroVac
data invalid.
Deposition mass data invalid.
Depletion @ 2.5 min
Depletion @ 4 min
Depletion @ 4.5 min
Deposition mass invalid.
Depletion @ 2.5 min
-------
Table 3-3. General experimental results presented as particle mass. Asterisk (*) denotes invalid data.
Dash (-) denotes sample collected, but not analyzed. "NS" denotes no sample collected.
la
Ib
2a
2b
3a
3b
4a
4b
5a
5b
6a
6b
7a
7b
8a
8b
9a
9b
lOa
lOb
lla
lib
12a
12b
13a
13b
14a
14b
15a
15b
• |
*
18.2
75.7
65.0
99.2
101.3
22.7
29.1
69.2
71.4
101.6
108.3
*
*
63.7
60.9
21.7
20.5
98.1
100.5
20.3
*
86.0
91.2
131.8
133.9
141.2
133.6
71.7
72.1
1
10.1
18.7
65.1
40.0
74.7
123.8
72.1
48.6
218
166.7
262.9
250.3
*
52.4
211.2
159.7
21.7
53.4
110.2
139.9
75.1
64.9
242.4
307.2
73.9
101.8
190.8
389.9
99.2
168.2
I III '
PR?
*
-
-
926
587
117
4212
2779
2163
4837
6755
1454
*
-
409
4805
244
2953
1148
1236
2860
1306
2595
3497
ns
ns
ns
ns
ns
ns
HI
iWimn
11 1
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
ns
ns
ns
ns
ns
ns
;| I1
3'.i!.'t'i ii'lft:1!!!
;. ! I' M J 'fj,
*
4.1
5.8
-
4.0
-
5.4
6.4
4.7
7.2
5.5
6.6
*
4.8
5.4
5.7
4.5
4.8
4.0
5.2
7.0
6.2
6.1
5.5
ns
ns
ns
ns
ns
ns
ill 13
I..I Mill
*
1.7
1.9
-
1.7
-
1.9
1.9
1.7
2.1
1.8
1.9
*
1.8
1.8
1.8
1.7
1.8
1.7
1.8
1.9
2.0
1.8
1.8
ns
ns
ns
ns
ns
ns
I ?:W!f! I
1 I
fSa"; ,•-.•• • ii
107.6
53.3
69.8
91.4
101.5
217.2
182.8
154.5
116.1
177.2
52.2
356.2
*
45.7
81.6
22.1
48.5
104.3
189.0
101.0
*
*
*
*
ns
ns
ns
ns
ns
ns
P:W3 1
iisyjy i
5.7
3.1
3.6
4.7
2.9
4.6
8.2
5.9
7.1
5.5
4.9
6.6
*
3.2
5.4
3.0
4.6
6.0
8.6
5.1
*
*
*
*
ns
ns
ns
ns
ns
ns
ijl«««i i
Imffife 1
t rt,';^^ i
1.9
1.8
1.8
1.6
1.8
1.7
2.1
1.9
2.0
1.8
1.8
1.8
*
1.8
1.8
1.9
1.9
1.8
2.1
1.7
*
*
*
*
ns
ns
ns
ns
ns
ns
1
ill
0.2
1.6
2.9
0.5
3.5
8.9
17.8
27.4
28.5
40.5
70.9
52.1
18.5
21.2
39.5
36.8
0.3
3.0
0.9
5.8
5.2
3.2
4.0
7.7
5.8
7.3
7.3
10.1
27.2
51.7
.Wj^ilK . ,,,'M
;:\r %&•
1$ j| jp||
1|i
' m$j}s
0.2
0.3
0.5
2.0
1.7
1.5
5.2
7.1
17.1
11.6
28.5
17.3
2.8
1.8
14.8
18.9
0.2
0.1
1.6
0.6
0.3
0.2
1.2
1.4
0.5
0.5
0.8
1.4
7.2
19.8
•Sfp?"'**;)"?;;1
i Iliitlilll 1
4.0
3.3
4.0
3.9
3.6
3.5
5.0
4.2
4.5
3.7
4.3
4.2
3.7
3.9
3.7
3.9
4.8
3.9
3.8
2.5
3.2
3.3
3.0
3.1
2.5
1.9
1.8
3.1
3.6
3.7
irll&fi'^
1.6
1.5
1.6
1.6
1.5
1.6
1.7
1.6
1.6
1.5
1.6
1.6
1.6
1.5
1.6
1.5
1.9
1.7
2.1
1.8
2.4
1.9
1.7
1.7
1.7
1.7
2.1
1.6
1.5
1.5
-------
Table 3-4. General experimental results presented as particle count. MicroVac count data calculated from Fiber SEM or
Filter SEM size distribution and MicroVac mass. URG count data calculated from APS size distribution and URG mass.
CMD is the count median diameter. Asterisk (*) denotes invalid data. Dash (-) denotes sample collected, but not analyzed.
"NS" denotes no sample collected.
.
la
Ib
2a
2b
3a
3b
4a
4b
5a
5b
6a
6b
7a
7b
8a
8b
9a
9b
lOa
lOb
lla
lib
12a
12b
13a
13b
14a
14b
15a
15b
.!
*
2.37E+10
2.13E+11
3.45E+10
5.78E+11
1.91E+11
1.63E+10
2.09E+10
1.61E+11
1.27E+11
5.22E+11
2.22E+11
*
*
1.69E+11
9.34E+11
1.17E+10
2.71E+10
1.06E+11
1.86E+11
1.31E+10
*
2.73E+10
3.18E+10
-
-
-
-
-
-
if
1.47E+09
2.43E+10
1.83E+11
2.12E+10
4.35E+11
2.33E+11
5.18E+10
3.49E+10
5.08E+11
2.97E+11
1.35E+12
5.12E+11
*
1.56E+11
5.59E+11
2.45E+12
1.17E+10
7.07E+10
1.19E+11
2.59E+11
4.84E+10
6.66E+10
7.70E+10
1.07E+11
-
-
-
-
-
-
i will t
| ffr.)l.j il
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
ns
ns
ns
ns
ns
ns
ill
*
1.7
1.6
-
1.5
-
1.4
1.8
1.8
1.6
1.9
1.9
*
1.8
2.0
1.8
1.7
2.1
1.6
1.8
1.9
2.0
2.1
1.8
ns
ns
ns
ns
ns
ns
tlllii
*
1.9
2.2
-
1.9
-
2.1
2.3
1.9
2.2
2.0
2.2
*
1.9
2.0
2.0
2.0
1.8
1.8
1.9
2.4
2.2
1.9
5.0
ns
ns
ns
ns
ns
ns
|!1'
' lit
ill
1.56E+06
3.70E+06
3.02E+06
1.21E+06
7.92E+06
3.30E+06
l.OOE+06
1.50E+06
4.23E+06
7.53E+06
2.73E+06
2.08E+06
*
2.60E+06
1.19E+06
1.23E+06
1.20E+06
1.72E+06
1.88E+06
1.34E+06
*
*
*
*
ns
ns
ns
ns
ns
ns
! flllS
1 IP
1.5
1.0
1.2
1.7
0.9
1.7
1.3
1.5
1.5
1.5
1.2
2.1
*
1.1
16
1.0
1.4
1.8
1.2
1.5
*
*
*
*
ns
ns
ns
ns
ns
ns
11
III
2.1
2.0
2.0
2.2
2.0
1.9
2.9
2.2
2.5
2.4
2.4
2.1
*
2.0
2.2
1.8
2.1
1.9
3.2
2.3
*
*
*
*
ns
ns
ns
ns
ns
ns
ISA it j
if
3.34E+06
3.57E+07
4.39E+07
6.35E+06
6.98E+07
2.30E+08
2.24E+08
4.09E+08
3.97E+08
6.54E+08
1.01E+09
7.50E+08
3.18E+08
3.37E+08
5.22E+08
4.64E+08
4.13E+06
6.39E+07
2.91E+07
2.90E+08
3.81E+08
6.81E+07
1.30E+08
2.65E+08
1.47E+09
1.10E+09
6.17E+08
2.98E+08
4.64E+08
1.52E+09
!
Pfl[:||U|«f||f
II
II I
4.80E+04
2.49E+05
5.45E+04
1.84E+05
2.23E+05
5.75E+04
4.78E+05
7.66E+05
1.71E+06
1.36E+06
2.92E+06
1.80E+06
2.85E+05
2.13E+05
1.42E+06
1.72E+06
2.48E+04
2.38E+04
4.97E+05
6.12E+05
3.37E+05
1.15E+05
5.33E+05
4.63E+05
3.90E+05
5.16E+05
4.64E+05
2.94E+05
9.01E+05
1.80E+06
111
2.5
2.2
2.2
2.4
2.1
2.0
2.7
2.8
2.7
2.7
2.7
2.7
2.4
2.5
2.7
2.8
1.7
1.6
0.7
0.9
0.9
0.9
1.1
1.4
0.6
0.8
0.9
1.2
2.0
2.2
I*!!! i
1.7
1.6
1.6
1.6
1.5
1.5
1.7
1.7
1.7
1.7
1.7
1.6
1.6
1.6
1.6
1.6
2.1
2.2
1.9
1.9
1.9
2.6
2.0
1.8
1.7
1.7
1.9
2.0
1.8
1.7
-------
4.00x1 cr
3.00x106
CD
.o
Hi
o
2.00x106
1.00x106
0.00x10°
Carpet Type
+ Old Ca
* NewC
*
t
+ * +*
+
rpet
irpet
*
New Carpet: Y = 6431 * X + 1 .1 2E6, R2 = 0.28
/
/-- *
*
* +
t
--'"'"* 1
'
^-^"'''
__,,----''"
Old Carpet: Y = 3566 * X + 5.48E5, R* = 0.40
100 200
Wollastonite Mass, mg
300
400
Figure 3-1. Conversion of total Wollastonite mass to total number of Wollastonite
fibers per 1296 in2 of carpet. Wollastonite mass data calculated from SEM images
of carpet fibers.
-------
40
30
E
o
,2
"5
4-<
c
OJ
o
c
o
o
E
3
z
C/3
Q,
20
10
Y = 1.09 * X + 2.22, R2 = 0,86
t +
0 10 20
APS Mass Resuspended, mg
Figure 3-2. Relationship between total Wollastonite mass resuspended during
a 5 minute test and the average number concentration. Mass and number
concentration data collected by APS.
30
-------
Table 3-5. Correlation coefficients between different
estimates of mass available for resuspension.
Statistically significant coefficient (a = 0.95) are italicized.
1 ,;*•.«' |
Deposit
Micro Vac
Sonic. Grav.
Fiber SEM
I'll !/
Deposit
Micro Vac
Sonic. Grav.
Fiber SEM
Deposit
Micro Vac
Sonic. Grav.
Fiber SEM
r *X* ^
1
0.64
-0.11
0.36
j ",'.''.' i V1 '; I
1
-0.14
1
-0.39
*y\> !-
i
;,rf V4 ! |
1
1
Table 3-6. Correlation coefficients between different
estimates of counts available for resuspension.
Statistically significant coefficient (a = 0.95)
are italicized.
Deposit
Micro Vac
Fiber SEM
IWJ
Deposit
Micro Vac
Fiber SEM
Deposit
Micro Vac
Fiber SEM
".•q'K:l? 't A
-> ,.:'!.' I-'
1
0.55
0.31
l^j'^ib' I
1
0.65
0.45
1
0.52
0.15
!- ft.'.';.', |W
1
0.25
\t \" \ V
1
0.39
1
0.01
'''-•',' ''•?'"' ? !' ,'•''''
1
IJM y ! i
i
i
-------
3.2.2 Count Comparison
Table 3-6 presents the same correlation coefficients
calculated on a count basis. Because of the error associated
with the ultrasonication gravimetric mass measurement,
these data were not included in the analysis. Count based
correlation coefficients followed the same trend as mass
based coefficients. However, the magnitudes of the
coefficients were somewhat lower, probably due to the error
introduced by converting gravimetric data to count data.
3.3 Resuspended Wollastonite Measurement
Method Comparison
3.3.1 Mass Comparison
Figure 3-3 shows the relationship between the resuspended
Wollastonite mass as measured gravimetrically by the URG
sampler and aerodynamically by the APS. The relationship
between the two methods is statistically significant (a =
0.95). In general, the APS measured about 30% of the
total mass measured gravimetrically. The difference in
resuspended mass was caused by the difference between
aerodynamic and physical diameter resulting from how the
asymmetric Wollastonite fibers align within the sensing
volume of the APS. Appendix B provides a more thorough
explanation. Therefore, both methods provided a valid
estimate of mass resuspended allowing calculation of total
mass and particle size dependent emission factors.
3.3.2 Count Comparison
Figure 3-4 illustrates the relationship between the
resuspended Wollastonite particle counts as measured by
the URG sampler and the APS. Again, there is a strong
linear relationship between the two methods. However, the
resuspended counts calculated from the URG data were 200
times greater than the APS counts. This dramatic decrease in
the regression slope most likely resulted from the conversion
of gravimetric data to particle counts because of the cubic
influence of particle diameter, especially at smaller sizes.
3.4 Wollastonite Size Distribution
The Wollastonite size distributions deposited on the carpet
fibers and resuspended were analyzed to identify any
differences between experimental conditions.
The size distribution of the Wollastonite deposited was
measured via SEM of the ultrasonication filters and
individual carpet fibers (Table 3-7). The resuspended
Wollastonite size distribution was measured automatically
by the APS. Statistical analysis yielded differences in mass
median diameter (MMD) of Wollastonite deposited as a
function of carpet age by both techniques. The Wollastonite
MMD on old carpet was about 1.3 um greater than that
measured on new carpet. The larger size distribution more
than likely was caused by the presence of other fibrous
particles in the carpet not removed by the conditioning
procedure. The geometric standard deviation (GSD) of the
size distribution was consistent across all experimental
conditions. There were not any differences in size distribution
as measured by SEM analysis of ultrasonication filters
andindividual carpet fibers when carpet age was controlled.
The resuspended Wollastonite size distribution as measured
by the APS varied between resuspension methods (Table
3-8). Age of carpet did not influence the resuspended
size distribution, although the median diameter from
the older carpet was slightly larger (2.7 um vs. 2.3 um).
Walking generated a larger size distribution of particles
with a narrower range than vacuuming. The much smaller
resuspended CMD from vacuuming indicates many more
small particles were generated. The force imparted to the
carpet fibers by the vacuum beater bar probably provided
sufficient energy to dislodge more particles smaller than
1 um. This increase in resuspension of 1 um particles
was not evident from the mass data because of their small
contribution to the total mass resuspended.
As compared to the size distribution of Wollastonite
deposited on the carpet fibers, the resuspended particle
size distribution was smaller and narrower. This finding
is expected considering only a fraction of the deposited
Wollastonite becomes resuspended.
-------
30
20
O)
E
re
E
t/3
Q.
10
MAPS = 0.28 *MURG +0,71, R2 = 0.68
* *
* *
20
40 60 80
URG Mass, mg
100
120
Figure 3-3. Relationship between mass resuspended estimates measured by URG
and APS. APS measured approximately 30% of the mass measured gravimetrically.
-------
3.00x1 Q6
2.00x106
c
3
o
o
1.00x10
0.00x10°
Cflps = 0.002*
1.32E+05,R2 = 0.63
0.00x10°
4.00x10a
8.00x10*
URG Counts
1.20x10*
1.60x109
Figure 3-4. Relationship between resuspended Wollastonite counts measured by
URG and APS.
3.5 Emission Factors
3.5.1 Total Mass
Emission factors calculated on a total mass basis are
presented in Table 3-9. The average plus standard deviation
for each experimental condition is presented. Emission
factors in the APS/Fiber SEM and URG/Fiber SEM ratios
for tests using old carpet were not calculated. As discussed
earlier, the Fiber SEM mass available data for tests with
old carpet could not be related to the Wollastonite mass
deposited because of the presence of other fibrous particles.
Any emission factors calculated using this data would be
incorrect. Therefore, emission factors using Fiber SEM data
for old carpets were not analyzed.
The emission factors in Table 3-9 indicate differences
between experimental conditions. Table 3-10 shows the
p-values (a = 95%) for each experimental variable calculated
using linear regression models as described in Section 2.1.14.
However, mixing within the exposure chamber prevented
calculation of emission factors as function of height.
3.5.2 Total Counts
Emission factors calculated on a total count basis are
presented in Table 3-11. The magnitudes of the count
emission factors vary greatly between calculation methods
and in comparison with the mass emission factors. The
APS/SEM and URG/MicroVac emission factors probably
are the most representative of reality because these factors
were calculated from count:count or mass:mass raw data.
Calculation of emission factors where the numerator and
denominator were measured on a different basis (e.g.,
count:mass or mass:count) introduced significant error into
the values. As a result, extremely small emission factors were
calculated for the APS/Micro Vac. In the other extreme, URG/
Fiber SEM emission factors greater than 1 were calculated
These emission factors are impossible to achieve given the
low background Wollastonite fiber concentrations in the
carpet.
Statistical analysis of the APS/SEM and URG/MicroVac
count emission factors showed the same influence of the
independent variables as the mass emission factors. Table
3-12 shows the p-values (a = 95%) for each experimental
variable calculated using linear regression models as
described in Section 2.1.12.
-------
Table 3-7. Size distribution of Wollastonite deposited on new
and old carpet as measured from ultrasonication filters and
individual carpet fibers. Size distribution of bulk Wollastonite
shown for comparison.
New Carpet3
Old Carpet3
Bulk Wollastonite"
HI
MMD 4.
GSD 1.
CMD 1.
GSD 1.
MMD 5.
GSD 1.
CMD 1.
GSD 2.
MMD
GSD
IIil!ltt;Ui|*|! (IS |
yiiiKi i i
65 ± 0.64
76 ±0.07
70 ±0.19
94 ±0.13
98 ± 0.74
87 ±0.11
80 ±0.21
10 ±0.18
|
•illM&iNii^:!! i
4.74± 1.65
1.81 ±0.13
1.37 ±0.30
2. 15 ±0.37
5.83 ± 1.56
1.89 ±0.11
1.50 ±0.29
2. 30 ±0.26
2.5
2.2
"Size distribution presented as aerodynamic diameters
bSize distribution measured optically. Data provided by manufacturer.
Table 3-8. Size distribution of Wollastonite
resuspended via walking and vacuuming as measured
by the APS. Size distribution of bulk Wollastonite
shown for comparison.
Walking3
Vacuuming3
Bulk Wollastonite"
MMD
GSD
CMD
GSD
MMD
GSD
CMD
GSD
MMD
GSD
3. 93 ±0.40
1.57 ±0.06
2. 53 ±0.28
1.62 ±0.06
2.98± 1.06
1.90 ±0.30
1.15 ±0.36
2. 05 ±0.26
2.5
2.2
"Size distribution presented as aerodynamic diameters
bSize distribution measured optically. Data provided by manufacturer.
-------
Table 3-9. Emission factors calculated on total mass resuspended for each
test condition.
! iptliUl fi
-------
3.5.3 Interpretation of Emission Factors
Data from the tables show consistency in parameters
influencing emission factors between the different calculation
methods. Carpet age (when included as a variable),
resuspension method, and chamber relative humidity all
influenced the total mass emission factor. Interestingly, the
Wollastonite loading did not influence the emission factors.
A similar result was discovered for generic Arizona Test Dust
(ATD) (Rodes and Thornburg, 2004).
Based on these data, three different emission factor scenarios
were statistically different from each other.
1. New carpet versus old carpet at low Rh: With the
relative humidity constant at about 40%, emission
factors from new carpet were significantly lower than
those from old carpet. The lower emission factors
possibly were caused by greater adhesion of the
particles to the fibers due to electrostatic charging of
the fibers. This relationship was consistent with the
one between new-old carpet seeded with Arizona Test
Dust developed during earlier research.
2. Influence of relative humidity on new carpet: Similar
to the relative humidity emission factor influence
discovered using ATD, raising the relative humidity
within the chamber possibly weakens the static charge
on the fibers. As a result, the Wollastonite emission
factors from walking and vacuuming are on new
carpet are 2-4 times greater at relative humidity levels
approaching 90%. Relative humidity did not have
a statistically significant influence on walking and
vacuuming emission factors from old carpet.
3. Influence of resuspension method: Vacuuming new or
old carpet seeded with Wollastonite fibers resuspended
significantly less mass than walking. Even with a
beater bar, the vacuum efficiently sucked a majority
of the particles into the machine for collection in the
HEPA type bag and prevented the fibers from reaching
the surrounding air. Using a less efficient vacuum
bag probably will increase the emission factors as
more particles penetrate through and escape into the
room air. Also, a different vacuum with less efficient
collection of particles by the vacuum head probably
will yield higher emission factors.
3.5.4 Size Dependent Emission Factors
Size dependent, mass based emission factors were calculated
using the APS/Fiber SEM data for new carpet tests (Table
3-13). Emission factors for 0.5 um fibers of—10-9 were
obtained. However, these values were not statistically
different from zero. The size distribution and mass of
Wollastonite available for Test 13 was not determined.
A statistical analysis of these size dependent emission factors
did not yield any statistically significant relationships with
the test conditions (Table 3-14). However as shown in Figure
3-5, there seems to be a difference in emission factors as
function of particle size and relative humidity. At low relative
humidity, the emission factor generally decreases as the fiber
aerodynamic diameter increases. At high relative humidity,
the emission factor peaks at a larger fiber diameter and
remains elevated. Although statistically insignificant, these
general trends agree with the relationship between particle
diameter and emission factor found with ATD. Additionally,
the trends may support the validity of the influence of relative
humidity on emission factors discussed earlier. High relative
humidity possibly decreases influence of electrostatic forces
as well as increases the surface tension bond. As a result,
smaller particles will have lower emission factors at high
relative humidity because the dominant surface tension force
is increased by the presence of water. Conversely, the larger
particles which adhere to carpet fibers due to electrostatics
have lower emission factors because the humidity appears to
neutralize some or all of the charge on the carpet fibers.
A similar analysis was conducted for count based emission
factors (Table 3-15). Interestingly, vacuuming generated
statistically significant count-based emission factors. The
vacuum beater bar probably provides sufficient energy to
release a large number of sub-micrometer particles that do
not contribute significantly to the total mass resuspended.
Although the statistical analysis did not indicate any
statistically significant variables (Table 3-16), Figure 3-6
does indicate a relationship between count based emission
factor and particle diameter may exist due to the resuspension
of the submicrometer particles.
3.6 K-Factors
K-factors, ratio of resuspended Wollastonite number
concentration and Wollastonite surface loading, were
calculated for each test with valid APS and Fiber SEM
image data. K-factors provide a widely recognized means for
calculating airborne concentrations from surface loading data
and the type of activity being performed. In this research,
K-factors are a different method for expressing emission
factors. For these tests, K-factors were calculated as:
car
L #
carpel /
cur
Aggregate K-factors across all fiber aerodynamic diameters
and all test conditions are shown in Table 3-17. Average
size-dependent K-factors for new carpet only (Table 3-18)
were calculated from data presented in Table 3-15 and the
corresponding volumetric airflow and carpet area.
-------
Table 3-11. Emission factors calculated on total count resuspended for each test condition.
Sfflf'li I
|B!|B!JB|JB!JB!J! lillliii i
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
f'fjfflpyipfgpiH
Hlilliiti ^l^iiJiliiiiiiiteifell
2.1E-05± 1.6E-05
4.5E-06 ± 5.9E-06
3.8E-07 ± 1.9E-07
1.6E-06±8.9E-06
4.0E-06±8.6E-07
2.8E-06 ± 9.6E-07
1.4E-06
1.6E-06± 1.3E-06
1.2E-06± 1.3E-06
3.3E-06± 1.3E-06
4.3E-06±3.7E-06
5.6E-06± 1.8E-06
9.5E-07±2.4E-07
1.1E-07 ± 1.2E-07
1.2E-06±3.1E-07
ft ?f:« IIT
£iliJiifii&§lf ll
4.9E-02 ± 2.6E-02
8.5E-02 ± 9.5E-02
2.3E-02 ± 7.6E-03
*
*
*
8.2E-02
*
1.7E-02±4.8E-03
3.6E-01 ± 1.4E-01
*
*
ns
ns
ns
ifif-PPWfffPi
i::|!::||:|!ii!!::!!:l|^!|!::|!!:||:i!::!!:|::|!!|i::|:|!::!!:bl
1.9E-03±5.7E-04
2.7E-04±4.2E-05
5.7E-04 ± 5.9E-04
8.0E-03 ± 5.2E-03
1.5E-03± l.OE-03
1.1E-03±5.1E-04
2.2E-03
5.6E-04 ± 5.2E-04
6.3E-04±3.9E-04
6.8E-04 ± 6.2E-04
4.5E-03±4.8E-03
2.1E-03±5.6E-04
2.9E-03± 1.8E-03
1.5 E-04± 1.7E-04
7.9E-04±7.1E-05
1 liiil lii'li&lil
> i
> i
> i
*
*
*
> i
*
> i
> i
*
*
ns
ns
ns
* Emission factors for old carpet were not useable because mass available estimated by SEM could not be
related to mass deposited.
ns: No sample collected for analysis
Table 3-12. Table of p-values for each method of calculating count
emission factors (EmFa).
URG/MicroVac
0.254
0.0001
0.0002
0.0001
0.240
0.46
APS/Fiber SEM
0.604
0.0002
0.007
0.089
0.77
Table 3-13. Table of size dependent APS/SEM Fiber mass emission factors for each test condition using new carpet.
sii f f i
iti /I?
1 1
| »J ||
: i 10.0 o
\ 1 I 8-°
; ! 7.0 - o.o
i ! j 6.0 0.011 ±0.013 o.o;
; ! 5.0 o.oie ±0.021 o.oc
Jit 4.0 0.054 ±0.073 O.OC
§ 3.5 0.007 ± o.oos o.o;
3.0 0.022 ±0.028 0.0
2.5 0.019 ±0.019 0.05
2.0 0.012 ±0.010 0.0
1.5 0.006 ±0.004 O.Of
1.0 0.001 ±0.000 O.OC
0.5b 0
il'llfli
i::!j::j'4::!!::jj 1; :| 1
<» i H-ifl i f"« i 1%
!-, i <• i fa, i :.},
Hi!! i !!! i iiili! iiii i ililiili
- - -
- - o.ooi ±0.00;
L2± 0.002 0.001 - 0.001 ±0.00
>0± 0.019 0.008 ±0.010 0.055
39 ±0.001 0.021 ±0.029 0.034 0.001 ± 0.00
39 ±0.028 0.010 ±0.012 0.045 0.004 ± 0.00
>9± 0.019 0.008 ±0.009 0.080 0.002 ± 0.00
L8± 0.009 0.007 ±0.006 0.025 0.002 ± 0.00(
55 ±0.067 0.008 ±0.006 0.028 0.004 ± 0. 001
L5± 0.013 0.005 ±0.001 0.019 0.001 ± 0.00(
30 ±0.105 0.007 ±0.001 0.007 0.007 ± 0.00
35 ±0.006 0.001 ±0.000 0.001 0.003 ± 0. 001
0 000
Fi?1 II f? 1
]i i~ ilia! i i ! !a!ia!J!i!jaj i
0.001 ±0.001 ns
1 0.003 ns
1 0.001 ns
0.007 ± 0.004 ns
1 0.012 ±0.009 ns
1 0.016 ±0.002 ns
1 0.015 ±0.011 ns
D 0.016 ±0.006 ns
3 0.044 ± 0.024 ns
D 0.046 ±0.019 ns
1 0.092 ± 0.033 ns
3 0.168 ±0.018 ns
0 ns
- = No particles measured by SEM
ns = No SEM fiber sample collected
"Emission factors without standard deviations only
Emission factors for 0.5 mm particles were not sij
had one valid measurement per test condition
;nificantly different from zero
-------
Table 3-14. Table of p-values for APS/SEM Fiber mass emission factors
including particle diameter as an independent variable.
feo:-fe
APS/Fiber SEM
.\"^|?WrV"iTjrK%ifeiVs|TK
Vy?-iy
0.060
•'i* /,;/"."-;
-
fl':tsi'iK'/l
0.282
;:i}
0.167
|i;!f;':ff iX^i'-.ji
0.122
fe.
0.742
fn>
0.09
0.1
o
ra
o
E
LJJ
0.01
£ 0.001
0.0001
APS:SEM Fiber EmFa
+ High Rh
TJ(f Low Rh
OA 0-5 0.6 0.70.80.9
1
789
10
Wollastonite Aerodynamic Diameter, |im
Figure 3-5. Average size dependent, mass based emission factors calculated from
APS and SEM Fiber data for high and low relative humidity tests. Curves fitting
size distribution are shown for illustrative purposes only. Curves are not statistical
regressions.
-------
Table 3-15. Table of size dependent APS/SEM Fiber count emission factors for each test condition using new carpet.
i'XjY1'
HI.JRI
it
I
IS II
Si ; 10.0 o
|; \ 8.0
ii : 7.0 - 0.01
|; ! 6.0 0.031 ±0.004 0.01
|:!| 5.0 0.032 ±0.021 O.OC
lifill 4.0 0.054 ± 0.050 0.0^
I 3.5 0.012 ±0.004 0.01
3.0 0.023 ±0.012 0.01
2.5 0.028 ±0.012 0.0^
2.0 0.025 ±0.019 O.OC
1.5 0.026 ±0.028 0.0^
1.0 0.002 ±0.002 O.OC
0.5 0
:fe itei I siii ii
pri.ji : IjfHiiij i -!^%ii i i iP^'i
_ _ _
0.001 ±0.001
0± 0.010 0.001 - 0.001 ±0.001
3 ±0.002 0.005 ±0.006 0.097
)6± 0.003 0.015 ±0.020 0.040 0.002 ± 0.001
JO ±0.034 0.007 ±0.007 0.031 0.001 ± 0.001
8 ±0.017 0.005 ±0.006 0.060 0.001 ± 0.001
0± 0.008 0.004 ±0.004 0.015 0.001 ± 0.001
Jl± 0.040 0.005 ±0.003 0.014 0.0005 ± 0.000
)8± 0.008 0.003 ±0.001 0.010 0.001 ± 0.001
19 ±0.067 0.005 ±0.001 0.005 0.001 ± 0.001
)2± 0.003 0.000 ±0.000 0.001 0.004 ± 0.003
0 00 0.011 ±0.012
i'1ij*iii ii ;i\iN;
'its.' ii liR.1
'i« 0 ( (I**
0.0001 ± 0.0002 ns
0.004 ns
0.002 ns
0.003 ± 0.002 ns
0.006 ± 0.002 ns
0.009 ± 0.009 ns
0.006 ± 0.002 ns
0.009 ± 0.007 ns
4 0.016 ±0.003 ns
0.020 ± 0.002 ns
0.067 ± 0.064 ns
0.057 ±0.030 ns
0.080 ± 0.040 ns
- = No particles measured by SEM
ns = No SEM fiber sample collected
"Emission factors without standard deviations only had one valid measurement per test condition
Table 3-16. Table of p-values for APS/SEM Fiber count emission factors
including particle diameter as an independent variable.
APS/Fiber SEM
0.130
0.141
0.684
0.079
0.005
0.06
-------
0.1
(J
flj
LL
c
o
'at
at
E
UJ
in
CD
.o
iZ
UJ
CO
to
Q.
0.01
0.001
0.0001
APS:SEM Fiber Em Fa
•f- High Rh
•If Low Rh
0.4 0.5 0.6 0.70.80.9
Wollastonite Aerodynamic Diameter, (im
Figure 3-6. Average size dependent, count based emission factors calculated from
APS and SEM Fiber data for vacuuming and walking resuspension method tests.
Curves fitting size distributions are shown for illustrative purposes only. Curves are
not statistical regressions.
-------
Table 3-17. Aggregate K-factors for each test.
^fiifi
itoiiifeil
1
2
3
4
5
6
7a
8
9
10
11
12
13
14
15
j || JP {iWfWMjj
i | i^liiiii!!'^!^!^!!!!!^.-1^!!!!!
6.19E-04
2 1.07E-03
2.88E-04
*
*
*
1.03E-03
*
2.18E-04
4.55E-04
*
*
1 IfiBflififl}
i feii« iyisiii
3.26E-04
1.20E-03
9.59E-05
*
*
*
*
6.10E-05
1.72E-03
*
*
ns
ns
ns
a only 1 valid sample collected
* Emission factors for old carpet were not useable because
mass available estimated by SEM could not be related to
mass deposited.
ns: No SEM Fiber sample collected for analysis
Table 3-18. Size specific K-factors for each test using new carpet. K-factors calculated from
emission factors in Table 3-15.
I ii f Mr
li it Jill"
1 ^ 1 ''< (
, III
It 10-° °
III I 8-°
1 7-° - !-3C
HI) 6.0 3.94E-04 1.5?
1 5.0 4.07E-04 7.92
|j|j f 4.0 6.77E-04 3.82
P 3.5 1.55E-04 2.2y
3.0 2.89E-04 1.22
2.5 3.51E-04 3.9J
2.0 3.14E-04 9.76
1.5 3.34E-04 6.22
1.0 2.79E-05 3. If
0.5 0
13!; Hi
>!! ! ji^S : :i » | iii!!q|
!|5 i liii! i lii i iiilll 1
- - - -
1.22E-05
)E-04 1.27E-05 - 1.21E-05
iE-04 6.27E-05 1.23E-03
!E-05 1.89E-04 5.11E-04 1.90E-05
IE-04 8.21E-05 3.91E-04 1.45E-05
'E-04 6.19E-05 7.64E-04 1.24E-05
IE-04 5.29E-05 1.95E-04 1.26E-05
)E-04 6.27E-05 1.74E-04 6.11E-06
)E-05 4.34E-05 1.23E-04 9.33E-06
iE-04 6.26E-05 6.33E-05 1.29E-05
)E-05 3.47E-06 8.21E-06 5.22E-05
000 1.35E-04
I
! II ! 111! !
1.71E-06 ns
4.89E-05 ns
2.43E-05 ns
3.71E-05 ns
7.36E-05 ns
1.19E-04 ns
7.02E-05 ns
1.12E-04 ns
2. 06 E-04 ns
2. 50 E-04 ns
8. 52 E-04 ns
7.19E-04 ns
1.01E-03 ns
-------
-------
4.0
Quality Assurance
Quality assurance and quality control measures for the
project were outlined in the Quality Assurance Project Plan
for basic research projects - "Resuspension of Fibers from
Indoor Surfaces Due to Human Activity." Tables 4-1 and
4-2 summarize the QA/QC measures for the project. The
QA/QC results for each metric are presented in subsequent
sections. Note the detection limits for the URG samplers,
Micro Vac, and deposition chamber are the same because
all rely on gravimetric analysis of the filters collected to
determine whether the data quality objectives were achieved.
The Mettler Toledo MT2 balance was used for all gravimetric
analyses.
4.1 URG Samples
Gravimetric analysis of URG filters determined the
mass resuspended from the carpet during experiments.
The data quality indicators used to determine whether the
data quality objectives for this metric were achieved are
listed in Table 4-3.
The number of URG samples attempted and successfully
collected determined the completeness percentage. The
cumulative completeness percentage, including all blank and
collocated samples, was 100% and exceeded the data quality
objective. Field and lab filter blanks determined whether
handling the filters caused inadvertent contamination. Field
blanks were loaded and unloaded with the experimental URG
samples. Lab blanks were kept in the gravimetric analysis
chamber and were weighed with the experimental and field
blank filters. The mean masses collected on the filter blanks
were used as correction factors for the experimental samples.
Precision and accuracy were other quality assurance criteria.
Precision in the sample collection and gravimetric analysis
was assessed by collecting 4 sets of collocated URG
samples. Precision was calculated as the percent relative
standard deviation (%RSD). Collocated URG samples during
experiments could not be collected. Therefore, collocated
samples were collected by sampling laboratory air at height
of 36 inches for 8 hours. Accuracy of the gravimetric
analysis was assessed every session (both pre-weighing
and post-weighing) by weighing a 100 ug standard weight.
A gravimetric analysis session was not started until the
measured weight was within 5% of the stated value.
4.2 MicroVac Samples
Gravimetric analysis of MicroVac filters determined the
mass available for resuspension from the carpet. The
data quality indicators used to determine whether the
data quality objectives for this metric were achieved are
listed in Table 4-4.
The number of MicroVac samples attempted and successfully
collected determined the completeness percentage. The
cumulative completeness percentage, including all blank and
collocated samples, was 97.2% and exceeded the data quality
objective. Both MicroVac filters from Test 7a were invalid
because the samples were dropped during unloading from the
cassette after sample collection.
Field and lab filter blanks determined whether handling
the filters caused inadvertent contamination. Field blanks
were loaded and unloaded with the experimental MicroVac
samples. Lab blanks were kept in the gravimetric analysis
chamber and were weighed with the experimental and field
blank filters. The mean masses collected on the filter blanks
were used as correction factors for the experimental samples.
Precision and accuracy were other quality assurance criteria.
Precision in the sample collection and gravimetric analysis
was assessed by collecting duplicate MicroVac samples for
each carpet piece. The MicroVac mass collected on each
sample was normalized by the mass loaded in that section of
carpet to eliminate the variability caused by the Wollastonite
loading procedure. Precision was calculated as the percent
relative standard deviation (%RSD. Accuracy of the
gravimetric analysis was described in Section 4.1. The data
listed in Tables 4-3 and 4-4 are identical. The accuracy data
in Table 4-4 are included for completeness.
Table 4-1. QA/QC criteria for measurements collected.
!««««!!
: -V. : *.'.ij1!> 95%
> 95%
> 90%
> 90%
> 90%
> 95%
> 95%
> 95%
i(iii(iaiiii(ia(ii
0.1 ug
0.1 ug
0.1 ug
0.5 |jm
1°C
1%
I ||i|iiiiii{^Niai
0.5 ug
0.5ug
0.5ug
NA
NA
NA
i iB{|i{|B{jBaBJ!B{jB{
1-5 Mg
1.5ug
1.5ug
NA
NA
NA
NA = not applicable
" Determined from manufacturer's calibration certificate.
-------
Table 4-2. Quality control measures to implement during testing.
tollAlliiliililiali!
URG
Micro Vac
Deposition Chamber
Ultrasonication
SEM Image Analysis
APS
Temp/Rh
II RWHIflWiltlllfifl!
i 1 liii^iiiiiyilifc^t liiiiii^iiiyilsifey ililillfa^ill
Precision: Collocated samples 5% of filters collected
Accuracy: Asses for =5% filters
Field Blanks: =5% of filters collected
Lab Blanks: =5% of filters collected
Background: Collected prior to each test
Precision: Duplicate samples for each carpet
Accuracy: Asses for =5% filters
Field Blanks: =5% of filters collected
Lab Blanks: =5% of filters collected
Precision: 9 filters for each carpet
Accuracy: Asses for =5% filters
Field Blanks: =5% of filters collected
Lab Blanks: =5% of filters collected
Precision: Duplicate aliquots of =5% of extracts
Field Blanks: =5% of filters collected
Lab Blanks: =5% of filters collected
Precision: Duplicate analysis of =5% of images
Blanks: NA
Precision: Collocated instruments once/week
Zero: HEPA filter installed on inlet dailv
Background: Collected prior to each test
Precision: NA
Accuracy: Temp = 5%, Rh = 10%
4.3 Deposition Chamber Samples
Gravimetric analysis of deposition chamber niters
determined the Wollastonite mass per unit area deposited and
the variability in the mass loading for each carpet sample.
The data quality indicators used to determine whether the
data quality objectives for this metric were achieved are
listed in Table 4-5. The number of deposition chamber
niters attempted and successfully collected determined the
completeness percentage. Nine niters were collected for each
carpet sample. The cumulative completeness percentage,
including all blank and collocated samples, was 91.3%.
Three sets of filters (27 total) were invalid. Two sets were
not attached properly to the carpet and moved during the
Wollastonite loading. The third set of filters was placed in the
wrong locations on the carpet sample.
Field and lab filter blanks determined whether handling
the filters caused inadvertent contamination. Field blanks
were loaded onto the carpet for five minutes then returned
to the filter storage container. Lab blanks were kept in
the gravimetric analysis chamber and were weighed with
the experimental and field blank filters. The mean masses
collected on the filter blanks were used as correction factors
for the experimental samples.
Precision and accuracy were other quality assurance criteria.
Precision in the sample collection and gravimetric analysis
was assessed from the 9 filters deployed per carpet piece.
Precision was calculated as the percent relative standard
deviation (%RSD). The precision determined the variability
in the Wollastonite mass loaded across the surface area of
the carpet. The average variability in the mass deposited was
17.4%. Accuracy of the gravimetric analysis was described in
Section 4.1.
4.4 Ultrasonication
The ultrasonic extraction of particles was one method for
quantifying the amount and size distribution of Wollastonite
available for resuspension. The data quality indicators
for measurement of particle loading via gravimetric mass
using this method are listed in Table 4-6. The data quality
indicators for measurement of particle size distribution via
SEM image analysis are presented in Table 4-7. As described
in Section 3.1, the particle loading (mass or count) could
not be determined from SEM analysis of the Ultrasonication
samples because of the large number of particles collected
on the filter. Therefore, QA/QC data for this metric are not
presented.
The completeness percentage was 93% for gravimetric
and size distribution analyses. Four Ultrasonication filters
were invalid because filter handling errors. Note that
Ultrasonication samples were not planned for Tests 13 thru 15
because of budget limitations.
Field blanks were fresh polycarbonate filters that were
extracted according to the procedure described in Section
2.1.8. Lab blanks were filters removed directly from the
package and either weighed gravimetrically or imaged via
SEM to determine the particle size distribution. Both sets of
blanks did not show a significant increase in mass because
particles were not present on the filter surface. This finding
was verified by the SEM image analysis.
Variation in the paired Ultrasonication filter particle masses
and size distributions for the two samples removed from
a single 36" by 36" piece of carpet was measured for each
test. The mass extracted from carpet sections removed from
-------
different areas varied greatly (RSD = 33%) because of the
large amount of (backing material, glue, etc) collected on
the filter along with the Wollastonite particles. However, the
%RSD is still within the limit set by the author of the method
(Millette et al., 1993). Precision in the size distribution
measurements was better (RSD = 17.4%).
Variation in the mass and size distribution within a single
sample also was measured. The procedure specified removal
of one 50 ml aliquot. For three randomly selected samples, a
second 50 ml aliquot was collected and filtered. The variation
in mass and size distribution within a sample was much
smaller, indicating a single aliquot was representative of the
carpet extract.
Table 4-3. URG sample completeness, filter blank, precision, and accuracy statistics.
Valid Samples
Planned Samples
72
Number
72
Mean (ug)
Number
0.2 ±0.2
Mean (ug)
Number of Pairs
0.1 ±0.3
Mean %RSD
6.1%
Number
Mean (ug)
20
99.6
% Completed 100%
% Difference 0.4%
'Includes 5% field blanks, 5% lab blanks, and 5% collocated samples
Table 4-4. MicroVac sample completeness, filter blank, precision, and accuracy statistics.
Valid Samples
Planned Samples3
70
Number
72
Mean (ug)
Number
0.1 ±0.2
Mean (ug)
Number of Pairs
0.1 ±0.1
Mean %RSD
30
5.7%
Number
Mean (ug)
20
99.6
% Completed 97.2%
% Difference 0.4%
"Includes 5% field blanks, 5% lab blanks, and 5% collocated samples
Table 4-5. Deposition chamber filter sample completeness, blank, precision, and accuracy statistics.
Valid Samples
Planned Samples3
283
Number
310
Mean (ug)
13
Number
13
Number of Sets
-0.8 ±0.5
Mean (ug)
-0.1 ±0.6
Mean %RSD
22
17.4%
Number
Mean (ug)
20
99.6
% Completed 91.3%
% Difference 0.4%
"Includes 5% field blanks, 5% lab blanks, and 5% collocated samples
Table 4-6. Gravimetric mass as measured by ultrasonication: sample completeness,
blank, precision, and accuracy statistics.
Valid Samples
Planned Samples3
53
Number
57
Mean (ug)
Number
-0.3 ± 0.4
Mean (ug)
Number of Sets
0.0 ±0.3
Mean %RSD
22
33%
Number of Sets
Mean %RSD
3
3.4%
% Completed 93.0%
"Includes 5% field blanks, 5% lab blanks, and 5% collocated samples
Table 4-7. Size distribution from SEM images of ultrasonication filters: sample completeness,
blank, precision, and accuracy statistics.
Valid Samples
Planned Samples3
53
Number
Number
57
Mean (urn)
no particles
Mean (urn)
Number of Sets
no particles
Mean %RSD
22
17.4%
Number of Sets
Mean %RSD
3
7.3%
% Completed 93.0%
"Includes 5% field blanks, 5% lab blanks, and 5% collocated samples
-------
4.5 SEM Image Analysis
Completeness, precision, and accuracy were the QA/QC
metrics applicable to the SEM image analysis. Blanks were
not needed because particles were not present on clean filters
or fibers. Completeness was 100% (416 images) for the SEM
image analysis. The high quality of all photographs allowed
SEM image analysis to be performed on all. Precision was
assessed by repeating the image analysis for particle number
and size measurements for 5% (21 images) of the images
collected. The %RSD in the particle counts and diameters
were 1.3% and 2.8%, respectively. The difference in particle
concentration between the SEM image analysis used in this
project and the computer controlled SEM measurements
(selected as the reference standard) was less than 5%. There
was bias in the accuracy of the particle size measured by the
analysis of the SEM images. This issue is described in
Appendix A.
4.6 Aerodynamic Particle Sizer
The APS measured resuspended particle number and mass
concentration at the resuspension area boundary. The QA/QC
results for this metric are summarized below.
The number of APS samples attempted and successfully
collected is outlined in Table 4-8. The cumulative
completeness percentage (100%) exceeded the data quality
objective.
Precision and accuracy of the APS measurements were
other quality assurance criteria. Precision could not be
assessed because only one APS was available. Accuracy
was determined to be within specifications because the APS
manufacturer's calibration was still valid. The APS was
calibrated on December 15, 2004.
The quality control assessment was to make sure the APS
measured a concentration of 0 particles per cm3. A HEPA
filter was installed on the inlet to the APS once per day, for a
total of 24 samples. The APS always measured 0 particles per
cm3 when the HEPA filter was installed.
4.7 Temperature & Relative Humidity
The HOBO H8 measured the temperature and relative
humidity within each house during resuspension and tracking
data collection. Only one HOBO was used, so precision was
not assessed. Accuracy was measured by placing the unit in
a temperature and humidity controlled chamber for 2 hours.
The QA/QC results for this metric are summarized in Table
4-9. All data quality objectives were achieved.
Table 4-8. Aerodynamic Particle Sizer sample
completeness statistics.
Valid Samples
30
30
60
Planned Samples
30
30
60
% Completed
100%
100%
100%
Table 4-9. QA/QC results for the HOBO H8
temperature and relative humidity data
,'i,r,r,r,r,r,r,r,iv;>in"i W ,'n\] "",\\'t
';' ,':,','• <•,:;'
Valid Samples
Planned Samples
% Completed
ti \ i""'"";;'1,]',
#
24
24
100%
" ^'WiWiWiWi'*','! i )>*'
Temperature
Rh
'; ' 'i'j 'ijyyyyyyyy* ^':
[;,',, ,'„'„'„'„',„$••
%
98.8
97.1
-------
5.0
Conclusions
Asbestos Fiber Simulant Available Estimates
• Neither the Micro Vac, Fiber SEM, nor Ultrasonication
SEM methods are sufficiently robust to provide
accurate mass available measurements for calculation
of emission factors. Additional work is needed to
develop methods to accurately quantify the mass
available.
• The Micro Vac method (the RTI-modified version of
ASTM D5755-95) had the highest correlations of
collections with the Wollastonite mass deposited.
Typically, the Micro Vac method underestimated the
quantity deposited on new carpet. Conversely, the
Micro Vac method overestimated the amount deposited
on older carpet because additional non-Wollastonite
particles were collected. This suggests that any more
aggressive surface vacuuming method (e.g. HVS3)
would exhibit the same or more pronounced problems.
• The Fiber SEM method had a statistically significant
correlation with the amount deposited only for new
carpet. Fibrous particles originally present that were
not removed during the cleaning positively skewed the
mass estimate such that a significant correlation could
not be obtained. This method generally overestimated
the mass deposited.
• The Ultrasonication SEM method had the poorest
correlations with the Wollastonite mass deposited.
This destructive technique generated many additional
particles (fibrous and general). As a result, the method
created a positive artifact in the gravimetric and SEM
mass estimates that could not be distinguished from
Wollastonite mass.
Emission Factors
• Emission factors based on total mass varied from 0.005
to 0.45, depending on the experimental conditions
and calculation method. Count based emission factors
covered a similar range: 0.0001 to 0.30
Size dependent emission factors, for new carpet
only, varied from 0.0 to 0.17 for fiber aerodynamic
diameters from 0.5 to 10 um.
• Emission factor estimates can be improved by
determining the amount available for resuspension as
a function of the force applied to the fibers. Additional
research is required to account for the influence of
electrostatic and surface tension adhesion forces on
emission factors.
• Wollastonite emission factors increase with carpet
age, consistent with the trend observed for Arizona
Test Dust emission factors. Emission factors from
old carpet were 2 to 3 times greater than those from
new carpet, possibly due to the decreased electrostatic
adhesion of large Wollastonite particles (> 3 um) to
the fibers.
• Increasing the relative humidity from —45% to —90%
increased the Wollastonite emission factors from
new carpet 3 to 4 times. It is suspected that the extra
moisture neutralizes the electrostatic charge and
particles larger than 3.5 um became resuspended
easier.
• Walking generated larger emission factors than
vacuuming, especially for older carpet. The vacuum
suction efficiently collected particles resuspended by
the vacuum beater bar and prevented these particles
from becoming airborne.
• Quantifiable count based emission factors for sub-
micrometer particles were obtained from vacuuming
experiments but not from walking. The energy
imparted to the carpet fibers by the vacuum beater bar
was sufficient to dislodge these particles. Mass based
emission factors for sub-micrometer particles were
zero because the mass resuspended was insignificant
and difficult to quantify.
• Emission factors for all test conditions were constant at
all levels of Wollastonite loading. This finding suggests
emission factors can be used to estimate airborne
concentrations from the quantity available
for resuspension found on the carpet
via the Micro Vac technique.
Size dependent emission factors were not influenced
by the experimental variables. General trends indicate
that emission factors for particles less than 2 um
decrease as relative humidity increases, and emission
factors for particles greater than 3.5 um increase with
increasing relative humidity.
• Emission factors as function of height were not
calculated because of mixing within the exposure
chamber and the contraction of the chamber plenum
did not yield the expected concentration gradient.
General Findings
• Wollastonite was a suitable simulant for asbestos
fibers. The Wollastonite fibers had a consistent aspect
ratio of 1:3 over the range of cross sectional diameters
from 1 to 5 um. Aerodynamically, the Wollastonite
fibers covered the size ranges (1-10 um) that penetrate
into the lung during respiration. The Wollastonite
dispersed easily and agglomeration was not an issue
during seeding of carpets. The only issue may be that
Wollastonite aspect ratio was not as large as typical
asbestos fibers (> 1:5).
-------
K-factors ranged from 10~3 to 10~6, depending on
particle diameter and experimental conditions.
Depletion of the Wollastonite fibers resuspended
during an experiment was noticed, but did not affect
the validity of the experiments. The elapsed time
before depletion occurred was used as input into Eq.
2-2 to calculate the total mass or counts resuspended.
New and old carpets could be cleaned sufficiently
to minimize the mass resuspended from unseeded
carpets to almost the minimum detection limits of the
instrumentation. However, the cleaning process did
not completely remove all particles from the carpet
fibers, as evidenced by the ultrasonication and fiber
SEM data.
Background characterization of the new and old carpet
for fibrous particles on the fibers and resuspended
during walking/vacuuming significantly improved
data quality. Additional characterization samples from
old carpet are recommended for future studies to check
for spatial inhomogeneity in the background particle
loading.
Theoretically determined correction factors could be
applied to the APS mass concentration data to make
the values equivalent to the gravimetric data.
-------
6.0
References
Brockman, J.E., "Sampling and Transport of Aerosols", Chapter 6 in Aerosol Measurement:
Principles. Techniques, and Applications. 2nd editioa edited by P. Baron and K. Willeke, John Wiley
and Son, New York, September, 2001.
Leith, D., "Drag on Nonspherical Objects." Aerosol Sci. & Technol., 6:153-162 (1987).
Hinds, W.C., Aerosol Technology: Properties. Behavior, and Measurement of Airborne Particles.
Wiley Interscience, New York, New York. 1982.
Millette, J.R., Clark, P.I, Brackett, K. A., and Wheeles, R.K., "Methods for the Analysis of Carpet
Samples for Asbestos." Environmental Choices: Technical Supplement. 1:21-24 (1993).
Peters, T.M., Leith, D., "Concentration Measurement and Counting Efficiency of the Aerodynamic
Particle Sizer 3321." J. Aerosol Sci., 34:627-634 (2003).
Ranade, M.B. "Adhesion and Removal of Fine Particles on Surfaces." Aerosol Sci. & Technol.,
7:161-176, 1987.
Rodes, C.E., and Thornburg, J. 2004. "Study of Resuspension of Paniculate Matter on Flooring
Surfaces Due to Human Activity." Final Report, RTI Project 08886. U.S. EPA Contract 3C-R185-
NALX.
Rodes, C. E. and R. W. Wiener, "Indoor Aerosols and Exposure Assessment", Chapter 29 in Aerosol
Measurement: Principles. Techniques, and Applications. 2nd edition, edited by P. Baron and K.
Willeke, John Wiley and Son, New York, September, 2001.
Thornburg, J., Ensor, D.S., Rodes, C., Lawless, PA., Sparks, L.E., and Mosley; "Penetration of
Particles into Buildings and the Influence of Associated Physical Factors, Part I: Sensitivity Model."
Aerosol Sci. & Technol, 34:284-296, 2001.
Thornburg, J., and Rodes, C.E. 2004b. "Indoor PM Resuspension Testing for Metals." Final Report,
RTI Project 08931. U.S. EPA Contract 3C- R340-NALX.
Thornburg, J., Rodes, C.E., Lamvik, M., Willis, R., and Rosati, J., Image Analysis Method (IAM) for
Measurement of Particle Size Distribution and Mass Availability on Carpet Fibers. Aerosol Sci. &
Technol. 40:274-281, 2006.
-------
-------
Appendix A
SEM Image Analysis
Introduction
Determination of the total number of particles available
for resuspension from carpet fibers as a function of particle
size is important for accurate determination of resuspension
emission factors. As part of previous research for Department
of Housing and Urban Development, RTI developed a
procedure for counting and sizing particles from SEM
images. The HUD research focus and optimization of
the procedure was for Pb (lead) particles. The developed
procedure was modified for EPA funded projects for general
paniculate matter (PM), fibers, and metals. The modified
procedure for general PM is presented here. SEM images
and particle size distribution were provided by ManTech
Environmental (METI) under contract to EPA. The RTI
method for size distribution analysis was compared with the
particle size distribution measured automatically during the
computer controlled SEM imaging by METI. The results of
this comparison are presented in this report.
Procedure
The RTI procedure for measuring the particle size from
SEM images is attached. The image analysis procedure to
highlight particles from the background was optimized for 6
SEM photographs (500x) provided by METI in August 2004.
Several assumptions are required to convert project area
measurements from the RTI image analysis to aerodynamic
diameters (Hinds, 1982). A volume shape factor of 0.25 was
assumed to convert the projected area diameter (d ) to an
equivalent diameter (de). A dynamic shape factor of 1.12 was
used to convert the de to an aerodynamic diameter (dae). RTI
used a constant particle density of 2.5 g/cm3.
The procedure used by METI to calculate aerodynamic
diameter is unknown, but RTI assumed the actual particle
density measured by elemental analysis was used.
Results
One interesting finding between the RTI and METI
procedures was immediately evident. The RTI procedure
counted 2 to 3 times more particles per image than the METI
procedure. The extra particles counted by RTI always were
smaller than 2 um. These extra particles were ignored by
RTI. Only the largest RTI particles corresponding to the same
number of particles measured by METI were compared.
The RTI and METI aerodynamic diameters calculated from
the six 500x images are compared in Figure 1. The linear
regression results are shown as well. The slope less than 1
and the intercept greater than 1 indicate a difference in the
aerodynamic diameters. However, the high coefficient of
determination (R2) shows the bias is repeatable.
The RTI aerodynamic diameters showed a positive bias for
the upper end of the size distribution, and a negative bias
for the lower end of the size distribution. This bias is an
unavoidable artifact because of the type of particles being
measured from the SEM images. Particles containing silicon,
calcium, and potassium are more difficult to distinguish
from the image background in a backscatter image. Higher
molecular weight elements, like lead, are much easier to
distinguish and this bias is less pronounced. As a result, the
thresholding of the image to separate the particles from the
background tends to lose area for smaller particles composed
of lower molecular weight elements. Increasing the number
of pixels associated with a smaller particle to offset the
negative bias leads to the positive bias for the larger particles.
Additional pixels cannot be applied solely to the smallest
particles. The size of all particles is increased during the
procedure. The influence of differences in the RTI and METI
procedures for calculating aerodynamic diameters from
project areas on the bias needs to be determined too.
Even with the bias in particle size measurements, the
calculated size distributions were quite similar. There was
a 5% difference in count median diameter. This bias did
cause a larger difference in the geometric standard deviation,
about 15%.
Conclusions
1. METI should not change their procedure for obtaining
SEM images. The RTI image analysis procedure has
been optimized for the images being obtained by
METI. Changing the SEM image quality would require
substantial effort to recalibrate the RTI procedure.
2. Confirm METI and RTI are using the same conversion
factors for calculating aerodynamic diameters from
projected area measurements.
3. METI used individually calculated particle densities
based on the elemental composition of the particles.
The METI densities averaged 2.4 g/cm3 ± 0.2. The
difference in RTI and METI densities had a negligible
effect on calculated aerodynamic diameter.
4. It is unknown why the RTI procedure was able to
detect particles less than 2 um from 500x images
whereas METI did not. Is it possible METI was not
looking for particles less than 2 mm during their
particle measurements?
-------
Particle Size Measurement Comparison
6 images at 503x
6 images at IDDDx
RTI Size Dist: CMD = 0,828, GSD = 2.13
METI Size Dist: CMD = 1.38. GSD = 1.58
B.QQ-,
B.OO
S
a
I
2.00
1.00
RH Aeracfynamic Diameter (urn)
Figure A-l. Comparison of RTI and METI aerodynamic diameters
Particle Size Distribution Measurement
Using SEM Image Analysis
Jonathan Thornburg, PhD
RTI International
August 19,2004
This procedure was modified specifically for analysis of
METI generated SEM photographs of particles on carpet
fibers. Original procedure is located in RTI\CAT Standard
Operating Procedure 07061.001, SEM-EDS Analyses,
Version 1, 2000. This modified procedure is applicable to
RTI projects 08886, 08931, and 08924.
Procedure
1. Use NIH Image Beta V. 4.02, available from Scion
Corp. http://www.scioncorp.com/frames/fr_scion_
products.htm
2. Open desired photograph in NIH image
3. Select ANALYZE\SET SCALE from menu bar
a. Select units of scale (micrometers)
b. Specify # pixels covered by image in known
distance box. Move cursor to far right of image and
read # pixels in "Info" box
c. Enter corresponding scale of image in measured
distance box.
i. 500x = 177.8 urn
ii. 880x=101um
iii. lOOOx = 88.9 urn
4. Select PROCESSYENHANCE CONTRAST to
highlight particles
5. Select OPTIONS\THRESHOLD to highlight particles.
Grab threshold intensity bar in LUT window. Move
cursor up until threshold value in "Info" window reads
148. Threshold value of 130 is for METI produced
photographs. Other photographs may need different
threshold. Trial and error process to get best
6. Covert photo to binary image by selecting PROCESS\
BINARYVMAKE BINARY
7. Remove holes in particles by selecting PROCESS\
BINARYVERODE
8. Remove phantom particles by selecting PROCESS\
BINARYVDILATE
9. Remove new holes in particles by selecting PROCESS\
BINARYVERODE
10. Select EDITMNVERT to make particles black,
background white
11. Select ANALYZE\OPTIONS. Check boxes for:
a. Area
b. Ellipse major axis
c. Ellipse minor axis
d. Include interior holes
e. Heading
f. Max Measurements = 1000
g. Field Width =9
h. Digits = 3
-------
12. Select ANALYZE\ANALYZE PARTICLES. Check
boxes for:
a. Label particles
b. Ignore particles touching edge
c. Include interior holes
d. Reset measurement counter
13. Save image of labeled particles
14. Select ANALYZE\SHOW RESULTS. Projected area
data for each particle opens in new window.
15. Select EDIT\COPY RESULTS to copy data to
clipboard.
16. Copy results into Excel spreadsheet "SizeDist.xls".
Paste into Cell A2.
17. Spreadsheet uses Hinds equations to automatically
convert projected area measurements to:
a. Projected area diameter
b. Equivalent diameter
i. Use volume shape factor = 0.25
c. Aerodynamic diameter
i. Shape factor = 1.12
ii. Density = 2.5 g/cm3 for ATD. Use appropriate
density for other materials.
User specified size intervals can be used to calculate CMD,
GSD, and R2. Upper size of each interval needs entered in
Column
-------
-------
Appendix B
Relationship between Aerodynamic
and Fiber Diameters
Introduction
Particles with high aspect ratios, like the Wollastonite
used in these experiments, present a challenge when trying
to interpret mass data based on different measurement
techniques. The gravimetric mass measurements depend
on the physical volume and true density of the collected
particles. On the other hand, the mass measurements from the
APS are estimated from aerodynamic diameters and assume
unit density. APS mass measurements of fibrous aerosols
typically are underestimated because the crosssectional
diameter of the fiber is measured due to how the fibers align
in the flow field through the APS sensing volume.
To check the accuracy of the APS mass concentrations
measured, a theoretical exercise, presented below, provided
a correlation between the APS mass measurement (based on
aerodynamic diameter) URG mass measurement (based on
physical volume).
Procedure
1. Convert cross sectional fiber diameter to
aerodynamic diameter using equations from
Leith (1987). See Figure B-l.
0.25
L
2. Relate mass measurements based on aerodynamic
(APS) and physical (Gravimetric) data
M
APS
cL, =1.199cl
(I,
(B-l)
M
Results
URG
nd'fL
• = o.o:
df
(B-2)
Using Figure B-l and Eq. B-2, the mass measured by the
APS will be about 25% of the mass measured gravimetrically
(See Eq. B-3). This result agrees with the slope of the linear
regression equation shown in Figure 3-1.
iut-fL
= 0.25A/tSG
(B-3)
E
3, e
(U
to
Is
5
^
d>
(S
c
o
in
§2
O
= 0.64d,« - 0.78, R" = 0.99
3456789
Aerodynamic Diameter (urn)
10 11 12 13
Figure B-l. Comparison of Fiber Aerodynamic (from APS) and Fiber Cross Sectional Diameters.
-------
-------
-------
&EPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGES FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development
National Homeland Security Research Center
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
------- |