EPA/600/R-11/156
October 2011
Laboratory Study of Poly chlorinated Biphenyl (PCB) Contamination and
Mitigation in Buildings
Part 1. Emissions from Selected Primary Sources
Zhishi Guo, Xiaoyu Liu, and Kenneth A. Krebs
U.S. Environmental Protection Agency
Office of Research and Development
National Risk Management Research Laboratory
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
and
Rayford A. Stinson, Joshua A. Nardin, Robert H. Pope, and Nancy F. Roache
ARCADIS, US Inc.
4915 Prospectus Dr., Suite F
Durham, NC 27709
-------
NOTICE
This document has been reviewed internally and externally in accordance with the U.S. Environmental
Protection Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
-------
Forward
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the Nation's
land, air, and water resources. Under a mandate of national environmental laws, the Agency strives to
formulate and implement actions leading to a compatible balance between human activities and the ability
of natural systems to support and nurture life. To meet this mandate, EPA's research program is providing
data and technical support for solving environmental problems today and building a science knowledge base
necessary to manage our ecological resources wisely, understand how pollutants affect our health, and
prevent or reduce environmental risks in the future.
The National Risk Management Research Laboratory (NRMRL) is the Agency's center for investigation of
technological and management approaches for preventing and reducing risks from pollution that threaten
human health and the environment. The focus of the Laboratory's research program is on methods and their
cost-effectiveness for prevention and control of pollution to air, land, water, and subsurface resources;
protection of water quality in public water systems; remediation of contaminated sites, sediments and
ground water; prevention and control of indoor air pollution; and restoration of ecosystems. NRMRL
collaborates with both public and private sector partners to foster technologies that reduce the cost of
compliance and to anticipate emerging problems. NRMRL's research provides solutions to environmental
problems by: developing and promoting technologies that protect and improve the environment; advancing
scientific and engineering information to support regulatory and policy decisions; and providing the
technical support and information transfer to ensure implementation of environmental regulations and
strategies at the national, state, and community levels.
This publication has been produced as a continued effort to support the EPA's mission of protecting human
health and the environment. It is published and made available by EPA's Office of Research and
Development to assist the user community and to link researchers with their clients.
Cynthia Sonich-Mullin, Director
National Risk Management Research Laboratory
-------
Executive Summary
Background
In recent years, EPA has learned that caulking materials containing potentially harmful polychlorinated
biphenyls (PCBs) were used in many buildings, including schools, in the 1950s through the 1970s. On
September 25, 2009, EPA announced new guidance for school administrators and building managers with
important information about managing PCBs in caulk and tools to help minimize possible exposure. EPA
also announced additional research into this issue to address several unresolved scientific questions that
must be better understood to assess the magnitude of the problem and identify the best long-term solutions.
For example, the link between the concentrations of PCBs in caulking materials and PCBs in the air or dust
is not well understood. The Agency is also conducting research to determine the sources and levels of PCBs
in schools and to evaluate different strategies to reduce exposures. The results of this research will be used
to provide further guidance to schools and building owners as they develop and implement long-term
solutions (U.S. EPA, 2009). The EPA research on PCBs in schools is designed to identify and evaluate
potential sources of PCBs in order to better understand exposures to children, teachers, and other school
workers, and to improve risk management decisions. Specific research areas include characterization of
potential sources of PCB exposures in schools (caulk, coatings, light ballasts, etc.), investigation of the
relationship of these sources to PCB concentrations in air, dust, and soil, and evaluation of methods to
reduce exposures to PCBs in caulk and other sources (U.S. EPA, 2010).
As part of the EPA research effort, this report summarizes the test results for PCB emissions from primary
indoor sources, with emphasis on PCB-containing caulking materials and light ballasts, and the factors that
may affect the emissions. Subsequent reports will discuss the research results on PCB transport in buildings
and evaluation of selected mitigation methods.
Objectives
The main objectives of this study were to seek a general understanding of the behaviors of the primary PCB
sources in buildings, especially caulking materials and light ballasts, to support risk management decision
making by providing new data and models for ranking the primary sources of PCBs, and to support the
development and refinement of exposure assessment models for PCBs, such as the Stochastic Human
Exposure and Dose Simulation (SHEDS) model (Zartarian et al, 2008), by reducing uncertainty in the
models.
Methods
The rates of PCB congener emissions from caulking materials and light ballast were determined according
to the principles described in ASTM Standard Guide 5116 — Standard Guide for Small-Scale
Environmental Chamber Determinations of Organic Emissions from Indoor Materials/Products (ASTM,
2010). Caulk samples were tested in a micro-chamber system consisting of six 44-mL Silicosteel® coated
stainless steel chambers (Figure E. 1). Light ballasts were tested in 53-liter environmental chambers (Figure
E.2). During the test, clean air passed through the chamber at a constant rate. Air samples were collected
-------
from the outlet of the chamber. To test the ballasts with electrical load, one 53-liter chamber was modified
to allow the ballast inside the chamber to be connected to the lamps located outside the chamber.
Figure E.I. The micro chamber system with air sampling cartridges
Figure E.2. Two 53-liter environmental chambers in the temperature-controlled incubator
Findings
In this report, the word "caulk" is used as a generic term for all types of caulking materials and sealants
found in buildings. Among the thirteen caulk samples tested, twelve were from PCB contaminated buildings
and the remaining one was made in the laboratory. Eleven out of the 12 field caulk samples were determined
-------
to contain Aroclor 1254 and the remaining sample was determined to contain Aroclor 1260. The Aroclor
concentrations in the caulk ranged from <10 to 136000 ug/g with a mean of 50300 ug/g and a median of
42600 ug/g.
The experimentally determined emission factors (i.e., the emission rate per unit area) showed that, for a
given PCB congener, there is a linear correlation between the emission factor and the concentration of the
congener in the source (Figure E.3 and Equation E.I). Furthermore, the coefficient (a^ in Equation E. 1 is
related to the vapor pressure of the congener (Equation E.2).
1500
1000 2000 3000 4000 5000
Congener content in caulk (jig/g)
6000
Figure E.3. Emission factor for congener #52 as a function of congener content in caulk
(r2 = 0.9816; n = 8)
= a x
a, =1805/1
(E.I)
(E.2)
where E; = emission factor for congener i (ug/m2/h)
x; = content of congener i in caulk sample (ug/g)
^ = a constant specific to congener i [(ug/m2/h) / (ug/g)]
P; = vapor pressure of congener i (torr)
When compared to the congener profiles of caulk samples, the congener profiles of air samples
are skewed toward the congeners that are more volatile. A log-linear correlation exists between the
vapor pressure of the congener and the normalized emission factor (Equation E.3, Figure E.4), which is
defined as the emission factor for a congener when its concentration in the caulk is 1000 ug/g.
In 7VS = 14.02+ 0.976 In P,-
(E.3)
-------
where NEi = normalized emissions factor for congener i (|ig/m2/h)
Pj = vapor pressure for congener i (torr)
10000
2 iooo -
o
•c
o>
.-
"3
o
Z
100 -
10 -
1
l.OE-6
l.OE-5 l.OE-4
Vapor Pressure (torr)
l.OE-3
Figure E.4. Normalized emission factor as a function of vapor pressure for eight target congeners in
a caulk sample (r2 = 0.9748)
These correlations (Equations El to E3) provide a tool for predicting the congener emissions from caulk
once the congener concentrations in the caulk are determined. This tool can be used to rank the PCB sources
and to estimate the PCB concentration in air due to the contribution from PCB-containing caulk.
PCB fluids, such as Aroclor 1242, were once used as dielectric heat transferring liquids in the capacitor of
light ballasts for fluorescent lamps. Thus, PCB-containing light ballasts are a potential source of PCBs in
buildings. Nineteen light ballasts were tested. None of them were marked "PCB Free", "No PCBs", or "Non
PCB", and none of them had visible fluid leakage. These samples represent thirteen different models from
five manufacturers. Some of them are shown in Figure E.5. Three light ballasts were opened after the
emission test to collect the fluids in the capacitor. All three fluids were identified as Aroclor 1242. The PCB
emissions from light ballasts were relatively low with or without electrical load at or near room temperature.
However, the PCB emission rate increased significantly as the temperature increased. Given that most light
ballasts are located in enclosures and may operate at elevated temperature, the emission rate can be higher.
One ballast unit failed during a chamber test with electrical load, causing the release of the PCB fluid from
the capacitor (Figure E.6) and leaking of the potting material (Figure E.7). Such an event could cause severe
indoor environmental contamination. MacLeod (1981) reported that the concentrations of PCBs in the room
where a light ballast burned out were more than 50 times higher than normal (11600 versus 200 ng/m3) on
the day of burnout and that the concentrations remained elevated for three to four months afterward.
According to the literature, the failure rate for light ballasts increases drastically when they approach the end
of their designed life span (Philips, undated). Thus, the presence of PCB-containing light ballasts in
buildings may pose a potential risk to the occupants because most existing PCB-containing light
-------
ballasts have approached or exceeded their designed service life and because the decontamination
process is both difficult and costly.
Figure E.5. Part of the light ballasts tested; for comparison, a modern light ballast, marked
"PCB-free", is shown on the far right
Figure E.6. Condensation of fluids in the chamber outlet manifold after the failure of the light ballast
-------
Figure E.7. The light ballast that burst during the emission test with electrical load
Study Limitations
This study was conducted in a relatively short period of time and only a few samples were tested. It was not
our intention to collect and test samples that are statistically representative of the primary sources in U.S.
building stock or to link the test results to the buildings from which the samples were collected. Over a
dozen types of primary sources have been identified in PCB-contaminated buildings. Only caulk, light
ballasts, and ceiling tiles were tested in this study because of the unavailability of other types of samples and
time constraints.
References
ASTM (2010). ASTM D5 1 16-10 Standard guide for small-scale environmental chamber determinations of
organic emissions from indoor materials/products, ASTM International, West Conshohocken, PA.
MacLeod, K. (1981). Polychlorinated biphenyls in indoor air, Environmental Science & Technology, 15:
926-928.
Philips (undated). Ballast life calculations, Technical note TN 005, Philips.
http://www.lighting.philips.com/gl_en/global_sites/fluo-gear/dimming/download/pdf/technical-
notes/tn005.pdf
-------
U.S. EPA (2009). EPA news release — EPA announces guidance to communities on PCBs in caulk of
buildings constructed or renovated between 1950 and 1978 / EPA to gather latest science on PCBs in caulk.
http://yosemite.epa.gov/opa/admpress.nsf/6fa790d452bcd7f58525750100565efa/
28c8384eeaOe67ed8525763c0059342f!OpenDocument
U.S. EPA (2010). Research on PCBs in caulk, http://www.epa.gov/pcbsincaulk/caulkresearch.htm
Zartarian, V., Glen, G., Smith, L., and Xue, J. (2008). Stochastic human exposure and dose simulation
model for multimedia, multipathway chemicals, SHEDS-multimedia model, Version 3 technical manual,
U.S. Environmental Protection Agency, EPA 600/R-08/118.
http://www.epa.gov/heasd/products/sheds_multimedia/sheds_mm.html
-------
TABLE OF CONTENTS
Forward i
Executive Summary ii
List of Tables xii
List of Figures xv
Acronyms and Abbreviations xviii
1. Introduction 1
1.1 Background 1
1.2 Goals and Objectives 3
2. Experimental Methods 4
2.1 Test Specimens 4
2.1.1 Caulk 4
2.1.2 Ceiling Tile 6
2.1.3 Light Ballasts 6
2.2 Test Facilities 9
2.2.1 Micro Chamber 9
2.2.2 Standard 53-Liter Chamber 11
2.2.3 Modified 53-Liter Chamber 12
2.3 Test Procedures 14
2.3.1 Caulk and Ceiling Tiles 14
2.3.2 Light Ballasts 14
2.3.2.1 Screening Testing 15
2.3.2.2 Elevated Temperature Testing 16
2.3.2.3 Live Ballast Testing 16
2.4 Sampling and Analysis 17
2.4.1 Air Sampling 17
2.4.2 Extraction and Sample Preparation 18
2.4.3 Target Compounds 18
2.4.4 Instrument and Analytical Methods 21
3. Quality Assurance and Quality Control 26
3.1 Data Quality Indicator Goals for Critical Measurements 26
3.2 GC/MS Instrument Calibration 27
3.3 Detection Limits 30
IX
-------
3.4 Environmental Parameters 32
3.5 Quality Control Samples 32
3.6 Daily Calibration Check 33
3.7 Recovery Check Standards 33
3.8 Comparison of Extraction Methods 34
4. Results 36
4.1 Caulk 36
4.1.1 PCB Content in Caulk Samples 3 6
4.1.2 Summary of the Micro Chamber Tests 38
4.1.3 General Emission Patterns 38
4.1.4 Calculation of the Emission Rates and Emission Factors 40
4.1.5 Dependence of the Emission Factor on Congener Content in Caulk Samples 43
4.1.6 Dependence of Congener Emissions on Vapor Pressure (1) —the P-N
Correlation 44
4.1.7 Dependence of Congener Emissions on Vapor Pressure (2) — the P-S
Correlation 46
4.1.8 Temperature Dependence of the Emission Factor 47
4.1.9 The Difference between the Exposed and Freshly-cut Caulk Surfaces 50
4.1.10 Emission Factors for Aroclors 5 3
4.2 Ceiling Tiles 56
4.3 Light Ballasts 59
4.3.1 Test Summary 59
4.3.2 Method for Calculating the Emission Rate 60
4.3.3 Screening Tests 60
4.3.4 Live Ballast Tests 61
4.3.5 Effect of Ambient Temperature 63
4.3.6 Emissions from a Burst Light Ballast 64
4.3.7 Inside the Ballasts 70
4.3.7.1 Physical Descriptions 70
4.3.7.2 Analytical Results 75
5. Discussion 79
5.1 Predicting the Emission Factors for PCB-Containing Caulk 79
5.1.1 Using the x-E Correlation (Method 1) 79
5.1.2 Using the P-N Correlation (Method 2) 79
-------
5.1.3 Predictive Errors 79
5.1.4 Method Selection 80
5.1.5 Predicting the Emission Factors for Aroclor 1254 80
5.1.6 Estimating the Air Concentration Due to Emissions from Caulk 81
5.2 Using the Advanced Emission Models for Emissions from Caulk and Other Building
Materials 81
5.3 Using the Emissions Data for Light Ballasts 83
5.4 Expressing the PCB concentrations as Aroclors 84
5.5 Study Limitations 85
6. Conclusion 87
Acknowledgments 88
References 89
Appendix A Test Conditions for Caulk Samples and Determination of PCB Concentrations 94
Appendix B Test Conditions for Light Ballasts 99
Appendix C Simulating the Long-term PCB Emissions from Caulk 102
Appendix D Simulation of a Failed Light Ballast 105
XI
-------
List of Tables
Table 2.1. Summary of caulk samples 5
Table 2.2. Summary of light ballast samples 8
Table 2.3. Conditions and reasons for testing PCB emissions from light ballasts 15
Table 2.4. Chemical names and CAS Registration Numbers for the PCB congeners analyzed 20
Table 2.5. Chemical names and CAS Registration Numbers for the internal standards and
recovery check standards 21
Table 2.6. Operating conditions for the Agilent 6890/5973N GC/MS/ CTC PAL Auto Sampler
System for the analysis of PCB congeners in Aroclor 1254 22
Table 2.7. Operating conditions for the Agilent 6890/5973N GC/MS/ CTC PAL Auto Sampler
System for the analysis of PCB congeners in Aroclor 1242 and Aroclor 1248 23
Table 2.8. Operating conditions for the Agilent 6890/5973N GC/MS/ Agilent 7683 Auto Sampler
System for the analysis of PCB congeners in Aroclor 1254 23
Table 2.9. SIM acquisition parameters for the Agilent 6890/5973N GC/MS for the analysis of
PCB congeners in Aroclor 1254 24
Table 2.10. SIM acquisition parameters for the Agilent 6890/5973N GC/MS for the analysis of
PCB congeners in Aroclor 1242 and Aroclor 1248 25
Table 3.1. Data quality indicator goals for critical measurements 26
Table 3.2. Objectives for small chamber operating parameters 27
Table 3.3. Objectives for micro chamber systems operating parameters 27
Table 3.4. GC/MS calibration for PCB congeners from Aroclor 1254 28
Table 3.5. GC/MS calibration for PCB congeners from Aroclor 1242 and 1248 29
Table 3.6. IAP results for each calibration 30
Table 3.7. Instrument detection limits (IDLs) for PCB congeners for the PUF Soxhlet method 31
Table 3.8. Method detection limits (MDLs) of the PUF Soxhlet extraction method for PCB
congeners on GC/MS 32
Table 3.9. Average recoveries of DCCs for small chamber and micro chamber tests 34
Table 3.10. Comparison of extraction methods 3 5
Table 4.1. Concentrations of target congeners and Aroclors in caulk samples 37
Table 4.2. Calculated emission factors (E) and normalized emission factors (NE) at room
temperature 41
Table 4.3. Estimated constants (aO for the x-E correlation 44
Table 4.4. Vapor pressures for the target congeners in Aroclor 1254 45
Table 4.5. Estimated constants bi and b2 in Equation 4.5 46
XII
-------
Table 4.6. Constants di and d2 in the N-T correlation for caulk sample CK-11 and CK-13 50
Table 4.7. Emission factors (ug/m2/h) for the exposed surface (Es) and the newly cut surface (E0)
for caulk CK-01 52
Table 4.8. Emission factors (ug/m2/h) for the exposed surface (Es) and the newly cut surface (E0)
for caulk CK-02 52
Table 4.9. Emission factors (ug/m2/h) for the exposed surface (Es) and the newly cut surface (E0)
for caulk CK-12 52
Table 4.10. Aroclor 1254 concentrations in caulk samples (x) and chamber air (C) and the
calculated emission factors (E) 55
Table 4.11. Concentrations of target congeners in ceiling tile samples 58
Table 4.12. Congener emission rates for light ballasts at room temperature and without electrical
load 61
Table 4.13. Rates of congener emission from ballasts with electrical load 62
Table 4.14. Estimated constants (fi and f2) for the effect of ambient temperature on congener
emissions from light ballasts 65
Table 4.15. Concentrations of target congeners in chamber background (Co), during the live test (C)
and the calculated emission rates (R) for ballast BL-08 68
Table 4.16. Concentrations of target congeners in chamber air seven days after the burst of ballast
BL-08 and the calculated average emission rates (R) 68
Table 4.17. PCB content in the gel-like material and the tar-like resin collected from the chamber
floor 70
Table 4.18. Congener content in potting material in BL-02 76
Table 4.19. Congener content in potting material in BL-12 77
Table 4.20. Congener content in the potting material in the burst ballast (BL-08) 78
Table 5.1. Predictive error for the x-E and P-N correlations 80
Table 5.2. Variations of Aroclor concentrations in caulk and air samples calculated based on five
individual congeners 85
Table 5.3. Variations of Aroclor concentrations in air sample for light ballast BL-08 calculated
based on five individual congeners 85
Table A.I. Test conditions for PCB emissions from caulk at room temperature 94
Table A.2. Test conditions for PCB emissions from caulk at different temperatures 95
Table A.3. Test conditions for comparing the PCB emissions from different surfaces 95
Table A.4. Average congener concentrations in chamber air, relative standard deviations, and
number of valid data points 96
Table A.5. Air concentrations at different temperatures for field caulk CK-11 98
Table A.6. Air concentrations at different temperatures for laboratory-mix caulk CK-13 98
xm
-------
Table B.I. Summary of conditions for the screening tests 99
Table B.2. Summary of conditions for the live tests 99
Table B.3. Summary of test conditions for the effect of ambient temperature 100
Table B .4. Congener emission rates for four light ballasts at different temperatures 101
Table C.I. Content in caulk, partition and diffusivity coefficients for four congeners in
Aroclor 1254 104
Table D.I. Physical properties of the congeners used in the simulation 106
XIV
-------
List of Figures
Figure E. 1. The micro chamber system with air sampling cartridges iii
Figure E.2. Two 53-liter environmental chambers in the temperature-controlled incubator iii
Figure E.3. Emission factor for congener #52 as a function of congener content in caulk iv
Figure E.4. Normalized emission factor as a function of vapor pressure for eight target congeners
in a caulk sample v
Figure E.5. Part of the light ballasts tested; for comparison, a modern light ballast, marked
"PCB-free", is shown on the far right vi
Figure E.6. Condensation of fluids in the chamber outlet manifold after the failure of the light
ballast vi
Figure E.I. The light ballast that burst during the emission test with electrical load vii
Figure 2.1. Caulk samples as received 4
Figure 2.2. Five caulk samples provided by building owners 5
Figure 2.3. Ceiling tile sample CT-02 7
Figure 2.4. Seven of the light ballast samples tested; for comparison, a modern light ballast
(marked "PCB-free") is shown on the far right 9
Figure 2.5. Markes u-CTE system with polyurethane foam (PUF) sampling tubes 10
Figure 2.6. Diagram of a single micro chamber 10
Figure 2.7. Two small environmental chambers in the temperature-controlled incubator 11
Figure 2.8. Modified chamber faceplate for live ballast testing 12
Figure 2.9. Ballast system setup - overhead view 13
Figure 2.10. Ballast wiring diagram for BL-09 and BL-11 13
Figure 2.11. Caulk sample in one of the micro-chambers 14
Figure 2.12. Ballast orientation in the small chamber for screening tests 15
Figure 2.13. Live ballast with wiring connections 16
Figure 2.14. Lamp was powered on by the ballast in the chamber 17
Figure 2.15. Comparison of chromatograms of a field caulk sample and Aroclor 1254 standard
solution analyzed by GC/MS 19
Figure 4.1. Comparison of chromatograms (from top to bottom: Aroclor 1254 standard, caulk
CK-09, caulk CK-08, and Aroclor 1260 standard) 36
Figure 4.2. Comparison of chromatograms: Aroclor 1254, a caulk sample and an air sample 38
Figure 4.3. Relative abundances of the target congeners for Aroclor 1254 39
Figure 4.4. Concentration profiles for seven target congeners in chamber air for caulk CK-09 tested
at room temperature 39
-------
Figure 4.5. x-E correlation for congener #52 43
Figure 4.6. Correlation between the normalized emission factor and vapor pressure for eight target
congeners in caulk CK-10 45
Figure 4.7. Slope of the x-E correlation (a^ as a function of congener vapor pressure 47
Figure 4.8. Normalized emission factor (NE) as a function of temperature for five congeners in
caulk sample CK-11 49
Figure 4.9. Normalized emission factor (NE) as a function of temperature for five congeners in
caulk sample CK-13 49
Figure 4.10. Caulk samples for testing the PCB emission rates of different surfaces 51
Figure 4.11. Ratio of the emission factors for the exposed surface (Es) and the newly cut surface (E0)
as a function of vapor pressure 53
Figure 4.12. Emission factor for Aroclor 1254 as a function of Aroclor content in caulk sample 56
Figure 4.13. Comparison of chromatograms - from top to bottom: Aroclors 1254, 1260, 1262, and
1268 and ceiling tile CT-01 57
Figure 4.14. Relative abundances of the target congeners in three ceiling tile samples 57
Figure 4.15. Congener content in the top (with paint) and bottom layers of the ceiling tile 59
Figure 4.16. Normalized emission factor as a function of vapor pressure for sample CT-03 59
Figure 4.17. Dependence of congener emission rate on vapor pressure for light ballast BL-09C 63
Figure 4.18. Effect of ambient temperature on congener emissions from ballast BL-09C 64
Figure 4.19. Condensation of fluids in the chamber outlet manifold after the failure 66
Figure 4.20. Comparison of the PUF sampling cartridge for ballast BL-08 to a normal cartridge 66
Figure 4.21. Temperature profile for chamber air during the live test for ballast BL-08 67
Figure 4.22. PUF sampling from the sealed 53-L chamber containing the burst ballast 69
Figure 4.23. Light ballast CK-08 after the burst 69
Figure 4.24. Ballast BL-02 after the bottom metal plate was removed 71
Figure 4.25. Ballast BL-02 (top side) 71
Figure 4.26. Capacitor in ballast BL-02 72
Figure 4.27. Ballast BL-12 after removing the casing 72
Figure 4.28. Capacitor in ballast BL-02 73
Figure 4.29. Ballast BL-08 after removing the bottom metal plate 73
Figure 4.30. The capacitor in the burst ballast (BL-08) 74
Figure 4.31. Fluid collected from the ruptured capacitor in ballast BL-08 74
Figure 4.32. Comparison of chromatograms for Aroclor 1242 standard and fluids in light ballasts
BL-02, BL-08, and BL-12 75
XVI
-------
Figure 5.1. Predicted congener concentrations over a 50-year period 82
Figure 5.2. Percent of congener mass emitted over a 5 0-year period 83
Figure D.I. Predicted concentrations of "total PCBs" and congener #18 following light ballast
failure 106
XVll
-------
Acronyms and Abbreviations
ACH air changes per hour
ANZECC Australian and New Zealand Environment Conservation Council
ASHRAE American Society of Heating, Refrigerating and Air-Conditioning Engineers
ASTM American Society for Testing and Materials
ATSDR Agency for Toxic Substances and Disease Registry
AWG American wire gauge
CASRN Chemical Abstract Services Registry Number
DAS data acquisition system
DCC daily calibration check
DQI data quality indicator
EPA Environmental Protection Agency
GC gas chromatography
GC/MS gas chromatography/mass spectrometry
IAP internal audit program
IS internal standard
IUPAC International Union of Pure and Applied Chemistry
LCs laboratory controls
NIOSH National Institute for Occupational Safety and Health
PCB polychlorinated biphenyl
ppm parts per million
PQL practical quantification limit
psi pounds per square inch
PUF polyurethane foam
QSAR quantitative structure-activity relationship
RCS recovery check standard
RH relative humidity
RSD relative standard deviation
RTD resistance temperature detector
SIM selected ion monitoring
TMX tetrachloro-ra-xylene
UNEP United Nations Environment Programme
VOC volatile organic compound
WHO World Health Organization
xvm
-------
1. Introduction
1.1 Background
Polychlorinated biphenyls (PCBs) are a class of 209 organic compounds, known as congeners, with the
chemical formula of Ci2Hi0.xClx, where x is the number of chlorine atoms in the range of 1 to 10. Different
mixtures of these congeners were sold under many brands and trade names worldwide, among which
Aroclors marketed by Monsanto Company were the most common trade names in the United States.
Commercial production of PCBs started in 1929 and was banned by the U.S. Congress in 1978. According
to a report by the National Institute for Occupational Safety and Health (NIOHS), the domestic sales of
PCBs by Monsanto Company between 1957 and the first quarter of 1975 were 894 million pounds or
approximately 400,000 tons (NIOSH, 1975). The approximate PCB usage in the U.S. included 60% for
closed system and heat transfer fluids (e.g., transformers, capacitors, and fluorescent light ballasts), 25% for
plasticizers, 10% for hydraulic fluids and lubricants, and 5% for miscellaneous uses (EIP Associates, 1997).
PCBs were once used as plasticizers — substances for providing flexibility and elongation — in caulking
materials because of their compatibility with the base resin or binder such as polysulfide and polybutene
(Monsanto, undated). According to the U.S. Department of Commerce (2009), these caulking materials
could contain up to 30% PCBs. In 1974, the addition of PCBs to caulking materials was discontinued, but
the use of existing stocks that contained PCBs continued at construction sites until about 1980. Thus, all
buildings that have expansion joints and that were built or renovated between the 1940s and the late 1970s
(Some references cited between the 1950s and the 1970s — author) are likely to contain PCBs in the
caulking materials.
In the past two decades, a series of field measurements conducted in Europe and North America has shown
that PCB-containing caulk and sealant can be a significant source of PCBs in buildings (Europe: Benthe et
al, 1992; Balfanz et al, 1993; Piloty and Koppl, 1993; Fromme et al., 1996; Kohler et al, 2005; Priha et al.,
2005 and North America: Herrick et al., 2004, 2007; Newman, 2010, Robson et al., 2010). For example, in a
study conducted in Berlin (Fromme et al., 1996), the building blueprints and associated documents for
public utility buildings, especially schools and childcare centers, were scrutinized and some buildings were
investigated to determine whether they contained elastic sealants that contained PCBs. In the suspected
buildings, samples of sealant materials and samples of room air were analyzed for PCBs. The air analyses (n
= 410) in the community rooms of the schools and childcare centers showed that the average concentration
of PCBs was 114 ng/m3, the maximum concentration was 7,360 ng/m3 and the geometrical mean was 155
ng/m3. About 15% of the school buildings and 3% of the childcare centers had indoor air values of over 300
ng/m3, indicating need for precautionary measures. Five percent of the school buildings were found to have
concentrations exceeding 3,000 ng/m3, indicating the need for intervention according to the German
government.
In another study, Herrick and his co-workers (Herrick et al., 2004) investigated 24 schools and other public
buildings in the Greater Boston area. Eight of these buildings contained caulking materials with PCB
content exceeding 50 ppm, ranging from 70.5-36,200 ppm; the mean value was 15,600 ppm. In a university
building in which similar levels of PCBs were found in caulking material, the PCB levels in the indoor air
ranged from 111 to 393 ng/m3; in dust taken from the ventilation system of the building, the range was < 1
-------
ppm to 81 ppm. The authors also found that, in seven of the eight buildings with PCB-containing caulk, the
PCBs were identified as Aroclor 1254; the remaining sample contained Aroclor 1260.
Light ballasts for fluorescent lamps are also potentially important sources of PCBs in buildings. As the
primary electrical components of fluorescent light fixtures, light ballasts are generally located within the
fixture under a metal cover plate. A light ballast unit is composed of a transformer to reduce the incoming
voltage, a small capacitor (that may contain PCBs), and possibly a thermal cut-off switch and/or safety fuse.
A tar-like substance, known as the potting material, is used to surround these components to muffle the
noise that is inherent in the operation of the ballast. This substance covers the small capacitor in which
liquid PCBs in the ballast would be located. If PCBs are present in the capacitor, the amount ranges from
approximately 1 to 1.5 oz (30 to 45 mL) (U.S. EPA, 1993). Another estimate (UNEP, 1999) indicated that
the amount of PCBs in ballasts ranges from 50 to 100 grams, which is equivalent to 37 to 74 mL of Aroclor
1242. The ballasts for high intensity discharge (HID) lamps, often used in large facilities such as indoor
parking spaces and school gymnasiums, operate at much higher wattage than fluorescent lamps. The
capacitors in the HID units are considerably larger than those in a fluorescent fixture. Most HID ballasts
contain between 91 and 386 g PCBs (equivalent to 67 to 286 mL of Aroclor 1242) (Environment
Canada,1991).
Over the last thirty years, studies have shown that PCB-containing ballasts could be a significant source of
PCBs inside buildings. A recent field study involving three communities in New York State found
significant association between the presence of fluorescent lights and the total PCB concentrations in indoor
air in the study area (Wilson et al., 2011). When certain types of ballasts reach the end of their useful life,
spontaneous leaking and smoking may occur, and this is accompanied by a remarkably objectionable odor
that penetrates the area (Staiff et al., 1974; U.S. EPA, 1993; Funakawa et al., 2002; Hosomi, 2005). A study
by Staiff et al. (1974) reported PCB concentrations of 12,000 to 18,000 ng/m3 in room air after the burnout
of a ballast, and the concentration was still approximately 1,000 ng/m3 after three days. MacLeod (1979,
1981) reported that concentrations of PCBs in the rooms containing the burned-out light ballast were more
than 50 times higher than normal (11,600 versus 200 ng/m3) on the day of burnout and that the
concentrations remained elevated for three to four months afterward. According to a study conducted in
Japan, the PCB emission rate is highly dependent on temperature. The emission rate increased by a factor of
400 as the temperature increased from 30 to 50 °C (Funakawa et al., 2002; Hosomi, 2005). Therefore,
identification and proper removal of PCB-containing ballasts must be considered in any PCB mitigation
plan.
Researchers and others have raised concerns over the potential exposure to PCBs in buildings, including
schools, because of the high concentrations of PCBs in some buildings and the toxicological effects of
PCBs, including carcinogenicity and detrimental effects on the immune, reproductive, nervous and
endocrine systems (ATSDR, 2009). EPA's peer reviewed cancer reassessment concluded that PCBs are
probable human carcinogens (U.S. EPA, 2008a). On September 25, 2009, the U.S. EPA announced a series
of steps that building owners and school administrators should take to reduce exposure to PCBs that may be
found in the caulk used in many buildings that were constructed or renovated between 1950 and 1978 (U.S.
EPA, 2009). Also, at the present time, the Agency is conducting research to better understand the risks
posed by PCB-containing caulk. There are several unresolved scientific issues that must be better
understood to assess the magnitude of the problem and to identify the best long-term solutions. For example,
-------
the link between the concentrations of PCBs in caulk and PCBs in the air or dust is not well understood
(U.S. EPA, 2009). This research will guide EPA's decisions concerning further recommendations for long-
term measures to minimize exposure and decisions concerning the steps that must be taken to prioritize and
conduct actions, such as removing the caulk, to protect public health. This report is part of the Agency's
research effort. It complements and supplements a field study in school buildings currently conducted by the
National Exposure Research Laboratory (NERL, 2010).
1.2 Goals and Objectives
The main goal of this study was to conduct laboratory characterization of the PCB emissions from primary
sources in buildings (especially in schools), with a focus on PCB-containing caulk and light ballasts. In
addition to determining PCB emission rates, several factors that may have affected the emission rates were
evaluated. This laboratory study supplemented and complemented the field measurements in buildings by
providing a better understanding of the emission process and by establishing a direct link between the
sources and the PCBs in the air. In addition to seeking a general understanding of the behaviors of primary
sources of PCBs, this study was designed to: (1) support risk management decision making by providing
new data and models for ranking the primary sources of PCBs, and (2) support the development and
refinement of exposure assessment models for PCBs, such as the Stochastic Human Exposure and Dose
Simulation (SHEDS) model (Zartarian et al, 2008; Stallings et al, 2008), by reducing the uncertainties in
PCB emission estimates.
-------
2. Experimental Methods
2.1 Test Specimens
27.7 Caulk
In this report, the word "caulk" is used as a generic term for all types of caulking materials and sealants
found in buildings. Thirteen caulk samples were tested. Unless indicated otherwise, all the samples were
provided by building owners on a voluntary basis through the offices of EPA Region 1 and Region 2. The
sample providers were instructed to wrap each caulk sample with aluminum foil and place it in a sealed
plastic bag. Then the samples were placed in a container with ice blocks (Figure 2.1) and shipped to the
authors by second-day delivery. Upon receipt, the packages were checked for damage. Then the samples
were stored in a freezer at -20 °C.
Figure 2.1. Caulk samples as received
Table 2.1 provides a brief description and identification number for each sample. Most samples were in
good or fair condition, and were approximately 15-centimeter long with width that varied from 3 to 12 mm.
CK-09 was the only sample that had deteriorated severely and was in the form of small pellets (Figure 2.2).
-------
Table 2.1. Summary of caulk samples
ID
CK-01
CK-02
CK-03
CK-04
CK-05
CK-06
CK-07
CK-08
CK-09
CK-10
CK-11
CK-12
CK-13
Description
interior building caulk
interior expansion caulk
exterior window caulk
interior window caulk
interior window sill caulk
interior window sill caulk
interior window sill caulk
interior window frame caulk
interior door frame caulk; deteriorated pellets
interior masonry joint caulk
interior masonry joint caulk
interior window sill caulk
laboratory mixed two-part polysufide caulk
Color
gray
off-white
gray
gray
light brown, translucent
brown
brown
brown
gray
light gray
brown
gray
gray
Notes
[a]
[a]
[a]
[b]
[a] This sample was collected by the authors from a pre-demolition public building.
M Two-part THIOKOL 223 5M industrial polysulfide joint sealant for concrete expansion joints. Aroclor 1254 (0.160 g)
was spiked into 2.66 g activator (part B), which was then mixed with 20 g polysulfide polymer (part A).
Figure 2.2. Five caulk samples provided by building owners (sample CK-09 on far right is in an
aluminum container)
-------
For emissions testing, field samples were prepared by cutting approximately 3.5 cm long sections from the
strip with a utility knife. The sides of the section were trimmed to form a rectangular cuboid. After the
weight and dimensions of the cuboid were determined, five sides of the sample were coated twice with an
oil-based primer (Sherwin-Williams), leaving one side exposed to air. The coated sample was placed in a
fume hood to allow the primer to cure before emissions testing. Several samples were too thin to create a
cubiod, but the exposed side was always a trimmed flat rectangle. Laboratory mixed caulk was prepared to
specified dimensions.
To prepare samples for determination of congener content in the caulk, two 1-cm pieces were cut from the
field caulk strip. Pieces were then cut into thin (<1 mm thick) slices, which were cross-cut into pellets no
larger than 2 mm x 2 mm in size. Duplicate samples, weighing approximately 0.2 g each, were placed in 20-
mL amber-glass extraction vials.
2.7.2 Ceiling Tile
Three ceiling tile samples were received and identified as CT-01, CT-02 and CT-03. Each sample was 15
cm by 15 cm in size. The sample storage and shipping procedures were the same as the procedures for the
caulk samples. The three samples looked identical and were most likely made from the same type of fiber.
They had densities of approximately 0.06 g/cm3. One side was painted (Figure 2.3).
For emissions testing, a 3.9-cm punch was used to cut a cylinder from the tile. All the surfaces except the
painted side (the side facing the room) were coated with a silicone rubber sealant (Silicone I, General
Electric), leaving only the painted side exposed to air. The sample was then placed in a fume hood to allow
the sealant to cure for four days.
To determine the PCB content in the ceiling tile, duplicate samples, weighing approximately 0.5 g each,
were prepared using scissors.
2.1.3 Light Ballasts
Nineteen light ballasts were received, representing thirteen different models from five manufacturers. Brief
descriptions and identifications of the samples are presented in Table 2.2. The ballasts were shipped to the
authors' laboratory at ambient temperature. Each unit was wrapped in dual sealed plastic bags. The samples
were inspected upon receipt and they showed no signs of damage or fluid leakage. None of the 19 ballasts
was marked "PCB free", "No PCBs", or "Non PCB" by the manufacturer. Four models (BL-10 through BL-
13) were marked "Class P", which indicated that integral protection was provided to prevent overheating of
the ballast. All but one of the labels on the ballasts were readable. Photographs of seven of the ballasts are
shown in Figure 2.4.
-------
Figure 2.3. Ceiling tile sample CT-02 (top: unpainted side; bottom: painted side)
-------
Table 2.2. Summary of light ballast samples
Ballast
ID
BL-01
BL-02
BL-03
BL-04
BL-05
BL-06
BL-07
BL-08
BL-09
BL-10
BL-11
BL-12
BL-13
Manufacturer / brand
Jefferson Electric Co.
General Electric
(Unreadable)
Universal Therm-O-Matic
General Electric
General Electric
Ad-Lite
General Electric
Universal Rapid Start
Universal Therm-O-Matic
Universal Therm-O-Matic
Universal Therm-O-Matic
Advance
Catalog #
234-983
59G276
263
446-LR-TC-T
8G1011
58G983
AD-240
89G347
598-L-STF
412-L-TC-P
443-LR-TC-P
458-L-TC-P
VQM-2S40-2-TP
Power (W)
118V1.3 A, 3x40W
118 V 1.3 A, 3x40 W
100 Watt
120 V 0.8 A, 2 x 40 W T12/RS lamps
120 V 1.4 A, 2 x40 WF96T12 orF72T12
118V0.8A2x40watt
118V0.8A
1 18 V 0.45 A, 1 lamp
265 V 0.37 A, 2 x 40 W T12RS
120 V 60 Hz; one 40 W rapid start lamp
277 V 60 Hz 0.36 A, 2 x 40 W T12/R.S. lamps
277 V 60 Hz; one 40 W lamp
277 V 60 Hz 0.35 A, 2 x 40 W rapid start lamps
Additional
Descriptions
Oct 1953D
1953D; 23 W power loss
rapid start
equip with coil
15.5 W power loss
1 1 W power loss
[a]
[b]
#of
Units
1
1
1
1
1
1
1
1
6
1
2
1
1
Mount lamp within i/2" of grounded metal reflector
Ground ballast and mount lamps within 1/2" of grounded metal reflector
-------
Figure 2.4.
Seven of the light ballast samples tested; for comparison, a modern light ballast (marked
"PCB-free") is shown on the far right
2.2 Test Facilities
2.27 Micro Chamber
The Markes Micro-Chamber / Thermal Extractor (u-CTE) (Markes International, United Kingdom) was
used to determine the PCB emissions from the samples of caulk and ceiling tiles. According to a study by
Schripp et al. (2007), (i-CTE shows good quantitative and qualitative correlation with conventional emission
test methods.
The u-CTE system (Figure 2.5) consists of six micro-chambers that allow surface or bulk emissions to be
tested from up to six samples simultaneously at the same temperature and flow rate. Each micro-chamber
consists of an open-ended cylinder (cup) constructed of Silicosteel® coated stainless steel measuring 25 mm
deep with a diameter of 45 mm and a volume of 44 mL. The system has temperature control that allows the
tests to be conducted at ambient temperature or at temperatures up to 120 °C. The chamber's flow
distribution system, shown in Figure 2.6, maintains a constant flow of air through each sample chamber,
independent of sorbent tube impedance and whether or not a sorbent tube is attached. The flow rate was
controlled by the source air pressure and the flow distribution device in the unit. For all of the PCB tests the
high flow rate option (50 mL/min to 500 mL/min) was selected. According to the vendor, surface air
velocities were roughly uniform across the surface of the sample and they ranged from approximately 0.5
cm/s at an inlet gas flow rate of 50 mL/min to approximately 5 cm/s at an inlet gas flow of 350 mL/min.
Planar materials can be lifted up within the micro-chambers using spacers until they reach the collar that
projects down from each micro-chamber lid. Samples of different thickness can be accommodated using
spacers that are appropriately sized.
-------
Figure 2.5. Markes ji-CTE system with polyurethane foam (PUF) sampling tubes
PUF sampling tube
O-ring specific to tube type
Detachable micro-
chamber sample top
Micro
chamber
-Heated air supply
low control
Figure 2.6. Diagram of a single micro chamber
The u-CTE system was set up in a fume hood. The air supply was from a clean air generation system
consisting of house-supplied high-pressure oil-free air, a pure air generator (Aadco model 737-11A, Cleves,
OH), a dryer (Hankinson model SSRD10-300, Canonsburg, PA), a Supelco activated charcoal canister, a
Supelco micro sieve canister and gross particle filters (Grainger Speedaire, Chicago, IL).
10
-------
2.2. 2 Standard 53-Liter Chamber
All of the emission tests for light ballasts were conducted in 53-liter stainless steel chambers that conformed
to ASTM Standard Guide D5116-10 — Standard Guide for Small-Scale Environmental Chamber
Determinations of Organic Emissions from Indoor Materials/Products (ASTM, 2010). These chambers had
nominal dimensions of 51 cm (width) by 25 cm (height) by 41 cm (depth). A stainless steel plate, fitted with
a Teflon-coated Viton O-ring, was used to seal the open side. Clean air, free of volatile organic compounds
(VOCs), was supplied to the chambers through the dedicated clean air system described in section 2.2.1.
Each chamber was equipped with inlet and outlet manifolds for the air supply, a K-type thermocouple for
temperature measurement in the chamber, and two RTD (resistance temperature detector) probes (HyCal
model HTT-2WC-RP-TTB, Elmonte, CA) for measuring the relative humidity at the air supply inlet and
inside the chamber. The relative humidity of the air supply to the chamber was controlled by blending dry
air with humidified air from a glass one-liter round-bottom flask with an impinger submerged in a
temperature-controlled water bath. All air transfer lines and sampling lines were made of glass, stainless
steel, or Teflon. An OPTO 22 data acquisition system (OPTO 22, Temecula, CA) continuously recorded the
outputs of the mass flow controllers, temperatures, and relative humidities. A I!/*" (3.8 cm) computer
cooling fan (RadioShack, Fort Worth, TX) was placed in the chamber to provide mixing for all of the small
chamber tests. The two chambers were housed in a temperature-controlled incubator (Forma Scientific,
model 39900), Figure 2.7.
Figure 2.7. Two small environmental chambers in the temperature-controlled incubator
The small environmental chambers were used with standard indoor parameters [23 °C, 50% RH, and one air
change per hour (ACH)] for all of the ballast screening tests. The temperature tests were operated at 50%
RH, as measured at 23 °C, and one ACH, with the temperature varying from 23 °C to 45 °C (at 5 °C
increments from 30 °C to 45 °C) at 24-h intervals. Special modifications were made to one of the chambers
11
-------
to accommodate live ballast testing (i.e., under electrical operation). Those details are presented in section
2.2.3.
2 2.3 Modified 53-Liter Chamber
To provide more realistic conditions for testing a ballast, one of the small chambers was modified to allow
the electrical input to the ballast through the appropriate lighting fixture. The faceplate of the chamber was
modified to support internal ballast wiring to an external 4-ft (122-cm) fluorescent light (Figure 2.8). Two
sealed electrical cord entrances were formed in the upper part of the faceplate. The right side contained a
3/C 14 AWG (American wire gauge) cable and the left side had a 9/C 16 AWG wire bundle. The 3/C
bundle was the inlet power supply and the 9/C bundle provided the power to the lamp. Immediately outside
the chamber, two "quick-disconnect" junctions were formed using locking plug and socket connectors on
each cord to maintain the reparability of the chamber and allow for its removal from the incubator without
disturbing the seal.
Figure 2.8. Modified chamber faceplate for live ballast testing
The ballasts that were evaluated during the screen testing were not identical. Some consisted of a 270-V,
2-lamp output; other ballasts included 120-V outputs, single lamp setups; a couple of the ballasts required a
starter. For the 270-V ballasts, 120-V power from the wall outlet was sent to a junction box nearby using a
3/C 14 AWG cable. The transformer inside the junction box boosted the voltage to a 270-V output which
was sent inside the chamber to the ballast via a second 3/C 14 AWG cable. The outgoing power from the
ballast was then sent via the 9/C 16 AWG bundle to the fluorescent light fixture. This general system setup
is shown in Figure 2.9. The setup for the 120-V ballasts was similar except that the junction box was not
needed and power from the wall outlet was routed directly to the ballast. An example of the ballast wiring
arrangements is shown in Figure 2.10. All the electrical wiring was done by a licensed electrician.
12
-------
Figure 2.9. Ballast system setup - overhead view
270V Power In from Junction Box
Grounding Block
270V Power Out to Lamp
B 14 gauge nlbck w re
W: 1"- gauge white wire
G: 14 gauge green were
V*: IS gauge yellow wire
R*: 16 gauge red wire
B*: 16 gauge blue wire
Figure 2.10. Ballast wiring diagram for BL-09 and BL-11 (270 V, 2 lamps)
13
-------
2.3 Test Procedures
2. 3.1 Caulk and Ceiling Tiles
PCB emissions from the caulk and ceiling tiles were tested in the micro-chambers. Prior to a test, each
chamber was cleaned with ultra grade or equivalent hexane (Fisher, Pittsburgh, PA) and then sonicated for
10 minutes. The inlet air pressure was set at approximately 55 psi (3.8* 105 Pa) to achieve the desired flow
rate of air through the chambers of approximately 500 mL/min. The temperature was set to the test
requirement. The system was allowed to equilibrate for several hours before a background sample was
collected from one of the chambers. A polyurethane foam (PUF) sampling cartridge (Supelco, pre-clean
certified) was attached to the outlet of the micro-chamber on the top of the lid covering the empty chamber
(See Figure 2.5, above). The outlet air flow through the PUF was measured using a Gilibrator™ diagnostic
calibration system (Sensidyne, Clearwater, FL). The background sample was collected over a 16-h period,
after which samples were placed in each of the chambers (Figure 2.11). Typical sampling schedule was five
PUF samples being collected over a two week period; the sampling duration was up to 16 hours.
Figure 2.11. Caulk sample in one of the micro-chambers
212 Light Ballasts
Three types of testing were conducted to measure the PCB emissions from the light ballasts in the 53-liter
environmental chambers. Table 2.3 summarizes the conditions and reasons. Test procedures are described
below.
14
-------
Table 2.3. Conditions and reasons for testing PCB emissions from light ballasts
Type of Test
Screening
Temperature
effect
Live
Temperature
Setting
23 °C; constant
23, 30, 35, 40 °C
23 °C; constant
Electrical
Load
No
No
Yes
Purpose
PCB emissions from ballasts without electrical load
Effect of ambient temperature on PCB emissions
from ballasts without electrical load
PCB emissions from ballasts with electrical load
2.3.2.1 Screening Testing
Prior to each test the selected chamber was cleaned by wiping all of the interior surfaces with isopropyl
alcohol wipes (Walgreens, Deerfield, IL) followed by washing with water with detergent. An inlet air flow
rate of 1 ACH and a 50% RH was set via the data acquisition system. The incubator temperature was
maintained at 23 °C. An empty-chamber background PUF sample was collected overnight at a sampling
flow rate of approximately 600 mL/min for 16 hours. The designated ballast was then taken from storage
and placed in the fume hood. The chamber was opened, and the ballast was placed on top of a sheet of
aluminum foil at the center of the chamber floor (Figure 2.12). After approximately 2 hours, an individual
PUF sample was collected at a sampling flow rate of approximately 600 mL/min overnight. After testing,
the ballast was removed and relocated to its secure location. Then, the chamber was cleaned in preparation
for testing the next ballast.
Figure 2.12. Ballast orientation in the small chamber for screening tests
15
-------
2.3.2.2 Elevated Temperature Testing
Elevated temperature testing of ballasts was conducted in the 53-L stainless steel chambers following all of
the same cleaning and setup procedures for the screening tests. The ballast was placed on top of a sheet of
aluminum foil on the chamber floor (Figure 2.12, above) after a background PUF sample was collected
overnight at the initial temperature setting of 23 °C. Then the chamber was sealed and a PUF sample was
collected overnight at 23 °C. After sampling, the incubator temperature was increased to 30 °C at a rate of
approximately 1 °C/h. Approximately six hours later, another PUF sample was collected overnight. This
process was repeated every day for 3 additional days increasing the temperature by 5 °C until the incubator
temperature reached 45 °C. Duplicate PUF samples were collected at 40 °C. For two tests, tandem samples
were collected at 35 °C and 45 °C to determine if PCB breakthrough had occurred.
2.3.2.3 Live Ballast Testing
Before each live ballast test, the modified chamber and internal wiring were prepared using the same
cleaning and set-up procedures detailed above. An inlet air flow with a rate of 1 ACH and 55% RH was
introduced to the chamber.
Prior to a test, a background sample was collected. Then the chamber was opened; the designated ballast
was connected to the electrical circuit (Figure 2.13) and placed on top of a sheet of aluminum foil on the
chamber floor. Then the power to the ballast was turned on by plugging the electrical plug into the wall
outlet, turning the lamp on to start the tests (Figure 2.14).
Figure 2.13. Live ballast with wiring connections
16
-------
Figure 2.14. Lamp was powered on by the ballast in the chamber
Power to the ballast was maintained for an hour before any sampling began, allowing the ballast to reach its
full operating temperature. PUF samples were collected at a flow rate of approximately 600 mL/min for
individual samples and 300 mL/min for duplicate samples. The general sampling schedule was to activate
the power to the ballast early in the morning, let it warm up for an hour, and then initiate the collection of an
individual PUF sample that continued throughout the workday. At the end of the day, the PUF sample was
removed, and duplicate PUFs were connected to the sampling manifold to collect air samples overnight. The
next morning, the duplicates were removed and the power to the ballast was turned off. The final inlet and
outlet flows were measured and then the ballast was removed from the chamber.
2.4 Sampling and Analysis
2.4.1 Air Sampling
Air samples from both the micro-chambers and small chambers were collected on polyurethane foam (PUF)
at approximately 500 mL/min for 16 hours. The sampling method was modified based on EPA Method TO-
10A (U.S. EPA,1999). The micro-chamber system has a flow distribution system that maintains a constant
flow of air through each sample chamber, independent of sorbent tube impedance and whether or not a
sorbent tube was attached. Thus, no pump or mass flow controller was used for micro-chamber tests. For the
small chamber tests, PUF samples were collected by drawing air from the small chamber outlet through
PUF cartridges with a mass flow controller and a vacuum pump. The sampling flow rate was set by the mass
flow controller and measured frequently by using the Gilibrator™ air flow calibrator before and during the
tests.
After collection, the sample and glass holder were wrapped in a sheet of aluminum foil, placed in a scalable
plastic bag, and stored in the refrigerator at 4 °C. The sample was extracted within seven days and analyzed
within 40 days. Sample information was recorded on labels affixed to the glass holder in which the sample
was stored and in the electronic sample log file. PUF samples and extracts were stored in the refrigerator at
17
-------
4 °C before extraction or analysis. Quality control samples such as chamber background, duplicates, and
field blanks were also collected. (See Section 3, below)
2.4.2 Extraction and Sample Preparation
To determine the PCB content in caulk and potting material in light ballasts, approximately 0.2 g sample
was extracted using a sonicator (Ultrasonic Cleaner FS30, Fisher Scientific, USA) with 10 mL of hexane
(ultra grade or equivalent, Fisher, Pittsburgh, PA) and approximately 100 mg of sodium sulfate (anhydrous
grade or equivalent, Fisher, Pittsburgh, PA) for 30 min in a scintillation vial. Before extraction, 100 |oL of 5
ng/mL recovery check standards, including 2, 4, 5, 6-tetrachloro-ra-xylene (TMX), 13C-PCB-77, and 13C-
PCB-206, were added to the extraction solution. After extraction, 990 \\L of the extract was placed in a 1-
mL volumetric flask containing 10 joL of 10 (ig/mL internal standards, including 13C-PCB-4,13C-PCB-52
and 13C-PCB-194, and then transferred to gas chromatography (GC) vials for analysis. The final
concentrations of each recovery check standard and each internal standard were 50 ng/mL and 100 ng/mL,
respectively. Because of their low density (0.06 g/cm3), ceiling tile samples were too bulky for the
sonication method. The Soxhlet extraction method was used. The typical sample weight was 0.5 g.
All PUF samples were extracted using Soxhlet systems by following EPA Method 8082A (U.S. EPA,
2007). The PUF samples were placed in individual Soxhlet extractors with about 250 mL of hexane. Fifty
microliters of 5 (ig/mL recovery check standards were spiked onto the PUF samples inside the Soxhlet
extractor. The samples were extracted for 16-24 h. The extract solution was concentrated to about 50 - 75
mL using a Snyder column. Then the concentrated solution was filtered through anhydrous sodium sulfate
into a 100-mL borosilicate glass tube and further concentrated to about 1 mL using a RapidVap N2
Evaporation System (Model 791000, LabConco, Missouri, USA). The 1 mL solution was cleaned up with
sulfuric acid (certified plus grade or equivalent, Fishser, Pittsburgh, PA) and brought up to 5 mL with the
rinse solution (i.e., hexane for rinsing the concentration tube) in a 5 mL volumetric flask. One milliliter of
the 5-mL solution was separated, and 10 (iL of 10-ng/(iL internal standards were added, after which the
extract was transferred to GC vials for analysis. The final concentrations of each recovery check standard
and each internal standard were 50 ng/mL and 100 ng/mL, respectively.
When the concentrations of PCBs in the samples were above the highest calibration concentration, the
extract solution was diluted with hexane. At that point, the recovery check standards were diluted with the
sample, but 10 joL of 10 (ig/mL internal standards were always added to the 1 mL of final solution before
GC/MS analysis.
2. 4.3 Target Compounds
PCBs can be analyzed and quantified either as an Aroclor mixture or as individual congeners. Aroclors can
be identified by recognition of Aroclor patterns (U.S. EPA, 2007). However, if the samples contain more
than one Aroclor or the Aroclors have undergone environmental degradation, such Aroclor mixtures may
have significant differences in peak patterns compared to those of Aroclor standards. The benefit of
analyzing congeners is that it allows a direct estimation of the risk of PCBs (Prignano, 2008). There are 209
PCB congeners, and analyzing all of them would be very complicated and time consuming. Thus, it was our
18
-------
intention to select certain PCB congeners as our target compounds for source characterization testing so that
the emissions of PCB congeners can be linked to their physical properties such as vapor pressure.
Selection of the target congeners was based on several factors: inclusion of some predominant congeners in
the source and in the emissions, inclusion of congeners with a wide range of vapor pressures and chlorine
numbers, and inclusion of at least one dioxin-like congener. By comparing the chromatographic peak
patterns of the Aroclor standards with the field caulk samples, we concluded that Aroclor 1254 was the
major component in the field caulk (Figure 2.15). Thus we selected 10 individual PCB congeners for the
source characterization study on caulk and ceiling tiles (i.e., PCB-52, PCB-66, PCB-101, PCB-154, PCB-
77, PCB-110, PCB-118, PCB-105, PCB-17, and PCB-187). Their identifications were based on the
literature (Frame et al., 1996; Rushneck et al., 2004) and comparison of retention times and mass spectra
with individual PCB congener standards. Among these compounds, PCB-52, PCB-66, PCB-101, PCB-154,
PCB-77, PCB-110, PCB-118, and PCB-105 are major PCB congeners in Aroclor 1254. Some of them
(PCB-52, PCB-101, and PCB-110) are also the major congeners in the emissions. PCB-154, PCB-77 and
PCB-110 co-elute but contain different numbers of chlorine atoms, so they can be quantified by GC/MS
with selected ion monitoring (SIM) mode. PCB-77, PCB-105 and PCB-118 are compounds listed by World
Health Organization (WHO) as dioxin-like congeners (Mydlova-Memersheimerova, 2009). PCB-17 (with 3
chlorines) and PCB-187 (with 7 chlorines) exist in Aroclor 1254 in small amounts. These compounds were
added to the analyte list to cover a wider range of vapor pressures.
2500000
2000000 -
O
O.
1500000 -
1000000 -
500000 -
lill H -A- A-
*. 1 ^ A.
jj
I J
i
JU
Ifc
1 It
ui
It II
Jn
Ifl
i Caulk Sample
ll.Ll.JL t I _A
, Aroclor 1254
1 ..L ..JL _1 i .
16 18 20 22 24
Retention Time (min)
26
28
30
Figure 2.15. Comparison of chromatograms of a field caulk sample and Aroclor 1254 standard
solution analyzed by GC/MS
19
-------
According to the literature, the PCBs used in the capacitor of light ballasts were either Aroclor 1242 and
1248 (Frame et al, 1996; Staiff et al, 1974; Hosomi, 2005). We compared the patterns of the
chromatographic peaks for the emissions from several light ballasts with the patterns for the emissions from
the Aroclor 1242 standard solution and concluded that the PCBs in those light ballasts were Aroclor 1242
(see chromatograms in Section 4.3.7.2). Nine individual PCB congeners were selected for ballast source
emission research. They were PCB-13, PCB-18, PCB-17, PCB-15, PCB-22, PCB-52, PCB-49, PCB-44, and
PCB-64. The selected PCB congeners did not have high peak responses, but they were the congeners that
can be separated with the GC/MS. PCB-13 and PCB-18 co-eluted, but they have different numbers of
chlorines, so they could be quantified by GC/MS in SIM mode. PCB-64 mainly existed in the gas phase of
Aroclor 1248. Chemical names and chemical abstract services registration numbers (CASRN) for the target
congeners, internal standards, and recovery check standards are presented in Tables 2.4 and 2.5.
Table 2.4. Chemical names and CAS Registration Numbers for the PCB congeners analyzed
Congener #
13
15
17
18
22
44
49
52
64
66
77
101
105
110
118
154
187
Short Name
PCB-13
PCB-15
PCB-17
PCB-18
PCB-22
PCB-44
PCB-49
PCB-52
PCB-64
PCB-66
PCB-77
PCB- 101
PCB-105
PCB-110
PCB-118
PCB-154
PCB- 187
IUPAC Name
3 ,4'-Dichlorobiphenyl
4,4'-Dichlorobiphenyl
2,2',4-Trichlorobiphenyl
2,2',5 -Trichlorobiphenyl
2,3,4'-Trichlorobiphenyl
2,2',3,5'-Tetrachlorobiphenyl
2,2',4,5 '-Tetrachlorobiphenyl
2,2',5,5'-Tetrachlorobiphenyl
2,3 ,4',6-Tetrachlorobiphenyl
2,3 ',4,4'-Tetrachlorobiphenyl
3,3',4,4'-Tetrachlorobiphenyl
2,2',4,5,5'-Pentachlorobiphenyl
2,3,3',4,4'-Pentachlorobiphenyl
2,3,3',4',6-Pentachlorobiphenyl
2,3',4,4',5-Pentachlorobiphenyl
2,2',4,4',5,6'-Hexachlorobiphenyl
2,2',3,4',5,5',6-Heptachlorobiphenyl
CASRN
2974-90-5
2050-68-2
37680-66-3
37680-65-2
38444-85-8
41464-39-5
41464-40-8
35693-99-3
52663-58-8
32598-10-0
32598-13-3
37680-73-2
32598-14-4
38380-03-9
31508-00-6
60145-22-4
52663-68-0
20
-------
Table 2.5. Chemical names and CAS Registration Numbers for the internal standards and
recovery check standards
Purpose
Internal
standard
Recovery
check
standard
Short Name
13C-PCB-4
13C-PCB-52
13C-PCB-194
TMX
13C-PCB-77
13C-PCB-206
IUPAC Name
2,2;'-Dichloro[13C12]biphenyl
2,2',5,5'-Tetrachloro[13C12]biphenyl
2,2',3,3',4,4',5,5',-Octachloro[13C12]biphenyl
l,2,3,5-Tetrachloro-4,6-dimethylbenzene
3,3',4,4'-Tetrachloro[13C12]biphenyl
2,2',3,31,4,41,5,51,6-Nonachloro[13C12]biphenyl
CASRN
234432-86-1
208263-80-3
208263-74-5
877-09-8
105600-23-5
208263-75-6
2 4.4 Instrument and Analytical Methods
The analytical method used for this project was a modification of EPA Method 8082A and EPA Method
1668B (U.S. EPA, 2008b). The analytical instruments used for quantitative analysis of PCBs congeners in
the project were the Agilent 6980/5973N GC/MS (Agilent, Santa Clara, CA) with CTC PAL Auto Sampler
(LEAP Technology, Carrboro, NC) and Agilent 6980/5973+ GC/MS with 7683 Agilent Auto Sampler
(Agilent, Santa Clara, CA). The operational conditions of the instruments are presented in Tables 2.6
through 2.8. The MSB selected ion monitoring (SIM) parameters were changed overtime during analysis to
achieve the best sensitivity, and they are presented in Tables 2.9 and 2.10. The instruments were calibrated
with PCB congeners in the range of 5 to 200 ng/mL. The GC/MS calibration and quantitation were
performed using the relative response factor (RRF) method based on peak areas of extracted ion profiles for
target analytes relative to those of the internal standard.
Certified PCB standards (in isooctane) and Aroclor standards (in hexane) were purchased from
AccuStandard Inc. (New Haven, CT). Certified 13C labeled internal standards and recovery check standards
(in nonane) were purchased from Wellington Laboratories Inc. (Guelph, Ontario, Canada). Certified TMX
standard (in acetone) was purchased from ULTRA Scientific (N. Kingstown, RI).
21
-------
Table 2.6. Operating conditions for the Agilent 6890/5973N GC/MS/ CTC PAL Auto Sampler
System for the analysis of PCB congeners in Aroclor 1254
Parameters
Injector
Injection volume
Inlet temperature
Inlet mode
Inlet Flow
Carrier gas
GC column
Oven temperature program
Transfer line temperature
Acquisition Mode
Solvent delay
Settings
CTC PAL
luL
250 °C
Splitless
1.9 mL/min measured at 100 °C
Helium
Restek RTX-5Sil ms, 30 m with 0.25 mm ID and 0.25 urn film thickness
100 °C for 2 min, to 150 °C at 25 °C/min, to 200 °C at 3 °C/min, to 280 °C at
8 °C/min, hold for 4 min, total time 34.67 min
280 °C
SIM
6 min
22
-------
Table 2.7. Operating conditions for the Agilent 6890/5973N GC/MS/ CTC PAL Auto Sampler
System for the analysis of PCB congeners in Aroclor 1242 and Aroclor 1248
Parameters
Injector
Injection volume
Inlet temperature
Inlet mode
Inlet Flow
Carrier gas and flow
GC column
Oven temperature program
Transfer line temperature
Acquisition Mode
Solvent delay
Settings
CTC PAL
luL
250 °C
Splitless
1.8 mL/min measured at 100 °C
Helium
SGE BPX5 30m with 0.25 mm ID and 0.25 urn film thickness
100 °C for 2 min, to 150 °C at 25 °C/min, to 200 °C at 3 °C/min, to 300 °C at
8 °C/min, hold for 4 min, total time 37. 17 min
280 °C
SIM
6 min
Table 2.8. Operating conditions for the Agilent 6890/5973N GC/MS/ Agilent 7683 Auto Sampler
System for the analysis of PCB congeners in Aroclor 1254
Parameters
Injector
Injection volume
Inlet temperature
Inlet mode
Inlet Flow
Carrier gas and flow
GC column
Oven temperature program
Transfer line temperature
Acquisition Mode
Solvent delay
Settings
Agilent 7683
luL
250°C
Splitless
1.0 mL/min measured at 100°C
Helium
SGE BPX5 30m with 0.25 mm ID and 0.25 urn film thickness
100 °C for 2 min, to 150 °C at 15 °C/min, to 200 °C at 3°C/min, to 280 °C at
8 °C/min, hold for 6 min, total time 38.00 min
280°C
SIM
8 min
23
-------
Table 2.9. SIM acquisition parameters for the Agilent 6890/5973N GC/MS for the analysis of
PCB congeners in Aroclor 1254
Analytes
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX (RCS) [a]
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
13C-PCB-4 (IS) w
13C-PCB-52 (IS)
13C-PCB-194 (IS)
Internal Standard
13C-PCB-4
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-4
13C-PCB-52
13C-PCB-194
-
-
-
Retention Time
(min)
16.6
21.0
25.2
26.4
26.5
26.7
24.3
27.4
28.2
29.2
10.2
23.7
31.0
10.2
17.8
30.2
Primary Ion
(m/z)
258
292
326
360
326
292
292
326
326
396
244
304
476
234
304
442
[a] TMX is tetrachloro-/w-xylene; RCS is recovery check standard.
M IS is internal standard.
24
-------
Table 2.10. SIM acquisition parameters for the Agilent 6890/5973N GC/MS for the analysis of
PCB congeners in Aroclor 1242 and Aroclor 1248
Analytes
PCB-13
PCB-18
PCB-17
PCB-15
PCB-22
PCB-52
PCB-49
PCB-44
PCB-64
TMX (RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
13C-PCB-4
13C-PCB-52
13C-PCB-194
Internal Standard
13C-PCB-4
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-52
13C-PCB-4
13C-PCB-52
13C-PCB-194
—
—
—
Retention Time
(min)
16.9
16.9
16.9
17.3
20.4
21.4
21.5
22.2
22.8
12.6
26.2
32.8
12.7
21.3
32.1
Primary Ions
(m/z)
222
258
258
222
258
292
292
292
292
244
304
476
234
304
442
25
-------
3. Quality Assurance and Quality Control
Quality assurance (QA) and quality control (QC) procedures were implemented in this project by following
guidelines and procedures detailed in the approved Category II Quality Assurance Project Plan (QAPP),
Poly chlorinated Biphenyls (PCBs) in Caulk: Source Characterization to Support Exposure/Risk Assessment
forPCBs in Schools. Quality control samples consisted of background samples collected prior to the test,
field blanks, spiked field controls, and duplicates. Daily calibration check samples were analyzed on each
instrument on each day of analysis. Results of QA/QC activities are described in the following subsections.
3.1 Data Quality Indicator Goals for Critical Measurements
Data quality indicator (DQI) goals for the measurement parameters and validation methods are listed in
Table 3.1.
Table 3.1. Data quality indicator goals for critical measurements
Measurement Parameters
Temperature
Relative humidity (RH)
Air exchange rate (ACH) for small chamber
Air flow rate
Weight of materials
GC/MS b calibration
GC/MS calibration
Recovery of spiked PCB standards [c]
Methods
Thermocouple, RID probe [a]
RTD Probe, thin film
capacitance sensor
Mass flow controller/meter
Mass flow controller
Gravimetric
Relative response factor
Internal audit program
GC/MS
Accuracy/Bias
±0.5°C
±5%RH
± 0.05 ACH
±10% of full scale
±2mg
Not applicable
75-125%
60-140%
Precision
±2°C
10%
10%
15%
±2mg
25%
25%
40%
js Resistance Temperature Detector.
M GC/MS is gas chromatography/ mass spectrometry.
[c] Recovery check standards are listed in Table 2.5.
In addition to the DQI goals for the critical measurement parameters, objectives established for the control
of operating parameters for the small chamber system and the micro-chamber system are shown in Tables
3.2 and 3.3.
26
-------
Table 3.2. Objectives for small chamber operating parameters
Operating Parameters
Chamber temperature
Chamber inlet air RH
Air exchange rate
Air velocity *
Individual PCB congener
Total PCB congeners
Control Methods
Incubator
Water vapor generator/dilution system
Mass flow controllers/meters
Fan
Clean Air System
Clean Air System
Typical set point
23 °C
45% RH
1ACH
10 cm/s
<10 ng/sample
<100 ng/sample
Bias
± 1.0 °C
±5%RH
± 0.05 ACH
Not defined
Not applicable
Not applicable
* Measured by hot wire anemometer 1 cm above source surface
Table 3.3. Objectives for micro chamber systems operating parameters
Operating Parameter
Chamber temperature
Low inlet air flow
High inlet air flow
Total PCB congeners
Control Method
Air supply temperature control
Gas tank regulator
Gas tank regulator
Clean air system
Typical Set Point
28-120 °C
10-70 mL/min
50-500 mL/min
<100 ng/sample
Accuracy
±0.5°C
± 10%
± 10%
Not applicable
3.2 GC/MS Instrument Calibration
The GC/MS calibration and quantitation of PCBs were performed by using the relative response factor
(RRF) method based on peak areas of extracted ion profiles for target analytes relative to those of the
internal standard. The calibration standards were prepared at six levels ranging from approximately 5 to 200
ng/mL in hexane. Three internal standards were added in each standard solution for different PCB
congeners. The calibration curve was obtained by injecting 1 joL of the prepared standards in triplicate at
each concentration level. Tables 3.4 and 3.5 summarize all GC/MS calibrations conducted for the project,
including the practical quantification limit (PQL) and the highest calibration concentration. The percentage
relative standard deviation (RSD) of average RRF meets the DQI goal of 25%.
27
-------
Table 3.4. GC/MS calibration for PCB congeners from Aroclor 1254
Date
Analytes
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX(RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
8/6/2010
RRF
1.07
1.56
1.28
1.41
1.58
1.34
1.39
1.27
1.12
0.83
0.62
1.30
1.61
%RSD
7.61
6.30
9.09
14.8
11.1
24.0
11.8
14.8
15.8
13.1
4.21
24.9
12.8
10/12/2010
RRF
0.90
1.23
1.18
1.20
1.52
1.54
1.40
1.42
1.32
0.93
0.40
1.15
1.01
%RSD
9.37
8.22
7.48
8.19
7.83
11.9
8.24
7.96
8.44
8.54
5.89
15.5
7.42
2/14/2011
RRF
0.69
1.05
0.90
0.90
1.18
1.21
1.07
1.03
0.95
0.68
0.40
1.12
1.08
%RSD
6.14
3.53
7.86
7.80
12.1
19.0
7.22
10.9
11.0
9.78
4.11
16.7
11.5
PQL
(ng/mL)
5.00
5.01
5.01
4.98
5.01
5.01
5.03
5.05
5.00
4.98
5.01
5.00
5.00
HiCal
(ng/mL)
200
200
200
199
200
200
201
202
200
199
200
200
200
[a] The DQI goal for %RSD was 25%.
28
-------
Table 3.5. GC/MS calibration for PCB congeners from Aroclor 1242 and 1248
Date
Analytes
PCB-13
PCB-18
PCB-17
PCB-15
PCB-22
PCB-52
PCB-49
PCB-44
PCB-64
TMX (RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
1/11/2011
RRF
0.91
0.58
0.73
0.92
0.79
0.81
0.82
0.69
1.09
0.41
1.04
0.93
%RSD
17.3
8.58
10.1
14.7
10.4
5.43
7.92
7.13
7.46
9.70
14.2
15.0
PQL (ng/mL)
5.03
5.03
5.00
5.03
4.95
5.01
5.02
4.98
4.98
5.01
5.00
5.00
Hi Cal (ng/mL)
201
201
200
201
198
200
201
199
199
201
200
200
1 The DQI goal for %RSD was 25%.
29
-------
The Internal Audit Program (IAP) standards that contain three calibrated PCB congeners were analyzed
after the calibration to evaluate instrument performance in terms of accuracy and precision. The IAP
standards were purchased from a supplier (ChemService,West Chester, PA) different from the standards
used for calibration and were certified as to their concentrations of PCB congeners.
Table 3.6 presents the results of the IAP standards analyzed for each calibration. The recoveries of IAP
ranged from 80% to 124% and percentage RSDs ranged from 0.13% to 3.34%. They all meet the criteria for
IAP analysis, which are 100 ± 25% recovery with percentage RSD of triplicate analyses within 25%.
Table 3.6. IAP results for each calibration
Calibration
8/6/2010
10/12/2010
1/11/2011
2/14/2011
Analyte
PCB-52
PCB-101
PCB-77
PCB-52
PCB-101
PCB-77
PCB-13
PCB-15
PCB-44
PCB-52
PCB-101
PCB-77
IAP Concentration
(ng/mL)
70.8
69.6
70.8
150
150
150
50.0
50.0
50.0
100
100
100
Avg. Recovery
%
114
90
93
92
86
80
97
116
124
104
93.5
79.9
%RSD
(n=3)
0.46
1.48
1.10
1.22
1.64
1.37
3.34
1.00
1.18
0.13
0.33
0.64
[a] The DQI goal for %RSD was 25%.
3.3 Detection Limits
After each calibration, the instrument detection limit (IDL) was determined by analyzing the lowest
calibration standard seven times and then calculating three standard deviations from the measured
concentrations of the standard. IDLs are listed in Table 3.7 for all calibrated PCB congeners.
30
-------
Table 3.7. Instrument detection limits (IDLs) for PCB congeners for the PUF Soxhlet method
Date
Analytes for
Aroclor 1254
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX (RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
8/6/2010
IDL
(ng/mL)
0.77
0.44
1.01
0.54
0.98
1.17
0.94
1.31
1.72
0.91
0.77
1.13
2.50
10/12/2010
IDL
(ng/mL)
0.48
0.44
0.43
0.17
0.25
0.21
0.42
0.35
0.44
0.33
1.05
0.34
1.36
2/2011
IDL
(ng/mL)
0.69
0.32
0.35
0.47
0.38
0.41
0.13
0.23
0.24
0.26
0.43
0.21
0.44
Analytes for
Aroclors 1242/1248
PCB-13
PCB-18
PCB-17
PCB-15
PCB-22
PCB-52
PCB-49
PCB-44
PCB-64
TMX (RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
-
1/11/2011
IDL
(ng/mL)
0.49
0.67
1.04
0.81
0.93
1.02
0.69
1.07
0.71
0.90
0.83
1.58
~
The method detection limit (MDL) was investigated for the PUF Soxhlet extraction method for PCB
congeners. Seven PUFs were prepared by spiking seven aliquots of the PCB standard (the final
concentration of which after extraction would be close to the PQL), and the recovery check standard
solution into the matrix. The PUFs were extracted by following the same extraction and analytical procedure
as for the samples. After analysis, the MDL was calculated by using three standard deviations from the
measured concentrations of those standards. The results are tabulated in Table 3.8.
31
-------
Table 3.8. Method detection limits (MDLs) of the PUF Soxhlet extraction method for PCB
congeners on GC/MS[a]
Analytes for
Aroclor 1254
PCB- 17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX (RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
MDL
(ng/mL)
2.32
1.65
2.54
2.38
2.67
2.28
1.97
o o o
J.JJ
3.90
3.85
1.69
1.79
1.44
MDL
(ng/PUF)
11.6
8.25
12.7
11.9
13.3
11.4
9.87
16.6
19.5
19.2
8.44
8.94
7.19
Analytes for
Aroclors 1242/1248
PCB-13
PCB-18
PCB- 17
PCB-15
PCB-22
PCB-52
PCB-49
PCB-44
PCB-64
TMX (RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
~
MDL
(ng/mL)
1.58
1.23
1.41
1.59
1.47
1.60
1.43
1.43
1.70
1.19
1.79
1.76
-
MDL
(ng/PUF)
7.91
6.16
7.05
7.93
7.36
8.02
7.15
7.15
8.48
5.95
8.94
8.81
~
1 To convert MDL to the air concentration unit: MDL (ng/m3) = MDL (ng/PUF) / sampling volume (m3).
3.4 Environmental Parameters
The temperature and RH sensors used to measure environmental conditions for the small chamber tests were
calibrated by the EPA metrology laboratory in July, 2010. The air flow and temperature of the micro-
chamber were manually measured before and after each sampling. Environmental data such as temperature
and RH in the small chambers were recorded by the OPTO 22 data acquisition system (DAS). The air
exchange rate of the small chamber was calculated based on the average flow rate of outlet air measured
with a Gilibrator at the start and end of each small chamber test. The measurement device was a primary
reference method calibrated by the EPA metrology laboratory.
3.5 Quality Control Samples
Data quality control samples discussed here included background, field blank and duplicates. Background
samples were collected from the outlet of the empty chamber for all tests. A typical background sample
showed the contribution of the contamination in the empty chamber, the sampling device, and the clean air
supply. Concentrations of all PCB congeners detected in all micro chamber background samples were less
than the PQL. The concentration of PCB-18 in 6 of 27 small chamber ballast tests was above the PQL,
possibly due to carryover from previous tests since all ballast tests were conducted in a relatively short
period of time, and there were some difficulties in cleaning up the PCB residues. These high backgrounds
were subtracted when calculating the emission rates.
32
-------
Duplicate samples were used to estimate the precision of the sampling and analysis methods. No duplicate
samples were collected from the micro chamber tests because there was only one outlet for each chamber.
Duplicate samples were prepared and analyzed for all bulk analysis of the solid sources. One duplicate
sample was collected during each of the live ballast tests. The data showed that the percent RSD of all
duplicate samples, except one pair, was less than 25%, meeting the data quality goal. Overall, the precision
of the sampling and analysis methods was very good for all target PCB congeners with concentrations above
the PQL.
Field blank samples were acquired to determine background contamination on the sampling media due to
media preparation, handling, and storage. Field blank samples were handled and stored in the same manner
as the samples. Seven field blank samples collected for micro-chamber tests and three for the ballast tests.
The target PCB congener concentrations in the field blank were below PQL for all samples.
3.6 Daily Calibration Check
On each day of analysis, at least one daily calibration check (DCC) sample was analyzed to document the
performance of the instrument. DCC samples were analyzed at the beginning and during the analysis
sequence on each day. Table 3.9 summarizes the average recovery of DCCs for the small chamber and
micro chamber tests. The recoveries meet the laboratory criterion of 75 to 125% recovery for acceptable
GC/MS instrument performance.
3.7 Recovery Check Standards
Three recovery check standards (RCSs), TMX, 13C-PCB-77, and 13C-PCB-206, were spiked in each of the
samples before extraction to serve as the laboratory controls (LCs). When the measured concentrations of
PCBs in the sample were above the highest calibration level, which mostly happened during bulk analysis,
dilution of the extract was performed to re-analyze the sample. In that case, recoveries of RCS were not
reported. The analytical results are considered acceptable if the percent recovery of laboratory controls was
in the range of 60-140% for at least two of the three recovery check standards.
33
-------
Table 3.9. Average recoveries of DCCs for small chamber and micro chamber tests
Test Type
Micro
Chamber
Tests
Small
Chamber
Tests
DCC
Compound
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX(RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
PCB-13
PCB-18
PCB-17
PCB-15
PCB-22
PCB-52
PCB-49
PCB-44
PCB-64
TMX(RCS)
13C-PCB-77 (RCS)
13C-PCB-206 (RCS)
Average
Recovery
101%
107%
101%
100%
104%
110%
102%
102%
102%
99.3%
101%
106%
97.4%
106%
103%
102%
105%
104%
97.3%
95.1%
94.3%
94.5%
99.6%
93.2%
94.1%
SD
0.051
0.064
0.052
0.065
0.058
0.062
0.056
0.054
0.062
0.080
0.048
0.053
0.032
0.081
0.066
0.061
0.086
0.095
0.019
0.023
0.029
0.031
0.034
0.081
0.042
%RSD
5.09
5.99
5.10
6.46
5.60
5.64
5.51
5.35
6.04
8.10
4.77
5.04
3.27
7.68
6.42
5.96
8.20
9.08
1.92
2.39
3.12
3.25
3.46
8.66
4.44
NW
98
98
98
98
98
98
98
98
98
98
98
98
98
44
44
44
44
44
44
44
44
44
44
44
44
[a] N is the number of DCCs analyzed.
3.8 Comparison of Extraction Methods
To ensure that the sonication method for extraction of caulk samples is comparable with the Soxhlet
extraction method, the extraction efficiencies of the two methods were evaluated. A field caulk sample was
chopped into small pieces to make six subsamples. Triplicate subsamples were extracted by the sonication
and Soxhlet methods, following the procedures for samples. The concentrations measured by the GC/MS
are listed in Table 3.10. The percentage RSD for all target PCB congeners above the PQL was less than
17%. The percent RSD for all target PCB congeners was less than 24%. The Soxhlet and sonication
methods are comparable for bulk analysis for this project.
34
-------
Table 3.10. Comparison of extraction methods (n=3 for each method) w (units: \iglg)
Analytes
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
Sum
Soxhlet[b]
4r3-7-[c]
322
660
69.1
694
4^82-
87.4
651
294
±7-74
2800
SonicationM
4r47-
372
838
77.6
856
244
98.2
745
320
24r4
3336
Mean[c]
4r42-
347
750
73.4
775
4^8
92.8
698
307
30r9
3068
%RSD
4.88
10.2
16.8
8.17
14.8
11.1
8.26
9.51
5.95
23.6
12.4
aJ XTiimhprc in ctriVpthrniiah fnnt nrp hplrvw POT
M Mean of three measurements.
[c] Average of the means for Soxhlet and sonication.
35
-------
4. Results
4.1 Caulk
4.1.1 PCB Content in Caulk Samples
The PCBs in 11 of 12 field samples were identified as Aroclor 1254. The remaining sample contained
Aroclor 1260 (Figure 4.1). The concentrations of the 10 target congeners and Aroclor 1254 are presented in
Table 4.1. Judging from their low PCB content, samples CK-4, CK-5, and CK-6 are likely contaminated
replacement caulk. It was noticed that the relative abundance of congener #52, the most abundant congener
in most air samples, varied significantly from sample to sample. Its percentage in the sum of 10 target
congeners ranged from 0.3% to 13.2% with a median of 6.8%, as compared to 15.6% for the laboratory-
mixed caulk (CK-13). This variation may reflect the different weathering conditions of the caulk samples.
For instance, among the caulk samples with low percentage of congener #52, CK-03 is an exterior window
caulk and CK-09 is severely deteriorated, (see Table 2.1).
L * . .1. Ji il ,.
n Ji id..
A . . » ... IA
1 1 1. 1.
0 25
L
LJ
1.
L
J
u
I.J
Aroclor 1254 standard
L »... . . . JL
Caulk CK-09
Ann n_fljuiAr * -
ILL
Caulk CK-08
Arpclor 1260 standard
fil. ...
30 3
Retention Time (min)
Figure 4.1. Comparison of chromatograms (from top to bottom: Aroclor 1254 standard, caulk
CK-09, caulk CK-08, and Aroclor 1260 standard)
36
-------
Table 4.1. Concentrations of target congeners and Aroclors in caulk samples (units: jig/g)[:
Sample
ID
CK-01
CK-02
CK-03
CK-04
CK-05
CK-06
CK-07
CK-08
CK-09
CK-10
CK-11
CK-12
CK-13
#17
Q QQ M
12$
Q QQ
n nn
VJ.UU
Q QQ
n nn
\j ,\j\j
Q QO
n nn
\J .W
Q QQ
223
Q gg
35*
3*6
#52
2790
2540
3^
615
Q Q^
045
0.41
8.49
269
4850
223
3140
330
#101
6400
5020
1401
3080
Q Qg
0.73
1.39
843
4570
9240
545
6420
509
#154
672
517
198
346
Q Qg
0.26
0.51
488
538
971
3:48
5^9
344
#110
6940
5260
2734
3970
Q Qf)
0.25
0.62
462
7330
9505
602
7090
540
#77
404
$m
35r5
444
Q QQ
n nn
U.VJVJ
Q QQ
231
Q QQ
45£
4r30
632
Q QQ
#66
549
510
63r4
247
04±
n m
U.VJ -J
Q ^Q
43:0
340
975
96.6
1160
78.4
#118
5780
4290
3434
3440
Q Qf)
043-
0.39
242
7330
7710
614
6470
499
#105
2370
1790
1813
1560
2.02 [d]
0.33
2.16
37-4
3180
3170
259
2650
192
#187
j^g
455
182
107
04i
0.30
0.61
2770
311
365
19.2
186
444
Aroclor [bl
96100
74300
52100
42600
W
7.14
29.0
39700 [f]
93300
136000
9128
103000
8280
LaJ Values are average of duplicate samples. Unless indicated otherwise, the RSD for all duplicates above the PQLs met the data quality goal of less than 25%.
M Aroclor 1254 unless indicated otherwise. Calculation method is described in 4.1.10.
[c] Values in strikethrough font is below the practical quantification limit.
[d] RSD for duplicate samples was greater than 25%.
[e] The Aroclor content was not calculated because most target congeners were below the practical quantification limit.
[f] Aroclor 1260.
37
-------
4.1.2 Summary of the Micro Chamber Tests
All of the 13 caulk samples listed in Table 4.1 were tested for PCB emissions at room temperature. Five
were tested in duplicate. Two caulk samples were tested at different temperatures to evaluate the
dependence of the emissions on temperature. Three samples were tested to compare the emissions from
freshly cut surfaces and previously exposed surfaces. Test conditions are summarized in Appendix A.
4.1.3 General Emission Patterns
Several studies (e.g., Balfanz et al., 1993) have recognized the significant difference in congener profiles
between air and solid samples. When compared to the congener profiles of caulk samples, the
congener profiles of air samples are skewed toward the congeners that are more volatile. As an
example, Figure 4.2 compares the chromatograms of the Aroclor 1254 standard, a caulk sample, and an air
sample taken from the emissions of the caulk. Similar patterns can also be seen by comparing the relative
abundances of the target congeners (Figure 4.3). For example, the most abundant congener in the caulk
sample was #110, which has vapor pressure of 1.7* 10~5 torr; its abundance in the air sample was 58% less.
On the other hand, congener #52, which has vapor pressure of 1.5 x 10"4 torr, was the most abundant
congener in the air sample, where there was three times as much of it as there was in the caulk.
Jl_
Jill
JL
Aroclor 1254
ULjlLil^
12
Air sample, CK-12
16
18
20
22
24 26
Time(mins)
28
30
32
Figure 4.2. Comparison of chromatograms: Aroclor 1254, a caulk sample and an air sample
38
-------
u
a
«
•a
=
s
60%
50%
40%
« 30%
J5 20%
0%
10%
rrfl
DAroclorl254
• Caulk CK-10
D Air sample
#17 #52 #66 #77 #101 #105 #110 #118 #154 #187
Congener ID
Figure 4.3. Relative abundances of the target congeners for Aroclor 1254
The air sample data showed that emissions remained stable over the test period (approximately two weeks).
All the target congeners had similar patterns (Figure 4.4).
100
M)
U
U
a
o
U
10 --
D.I
100 200 300
Elapsed Time (h)
#52
#66
-*-#105
-•-#110
-•-#118
O #154
400
Figure 4.4. Concentration profiles for seven target congeners in chamber air for caulk CK-09 tested
at room temperature
39
-------
4.1.4 Calculation of the Emission Rates and Emission Factors
In this study we used three terms to describe the emissions from caulk: emission rate, emission factor, and
normalized emission factor. Emission rate is in (|ig/h) and can be used for both area sources such as caulk
and non-area sources such as light ballasts. Emission factor is in ((ig/m2/h) and can only be used for area
sources (ASTM, 2010).
The caulk samples were treated as constant emission sources (see Figure 4.4) and the average air
concentration was used to calculate the emission rate and emission factor by using Equations 4. 1 and 4.2
(ASTM, 2010):
R=C
A (4.2)
where R = emission rate ((ig/h)
Q = air change flow rate (m3/h)
C = congener concentration in chamber air ((ig/m3)
E = emission factor (|ig/m2/h)
A = area of the source (m2)
The concept of normalized emission factor is new. The normalized emission factor is defined as
#*=£,— (4.3)
x
where NE = normalized emission factor (|ig/m2/h)
Ex = emission factor at congener content of x (|ig/m2/h)
x0 = a reference value for congener content in caulk sample ((ig/g)
x = actual congener content in caulk sample
A major advantage of this parameter is allowing for comparison of congener emission factors on an equal
basis (i.e., the same source strength). Throughout this report, x0 was set to 1000 (ig/g. Thus, the normalized
emission factor for a congener is the emission factor that corresponds to a content of 1000 (ig/g in the caulk.
The air concentration data and test conditions presented in Appendix A were used to calculate the emission
rates. The results are summarized in Table 4.2. Values in strikethrough font are below the practical
quantification limit. Tests for caulk CK-05, CK-06, and CK-07 were unsuccessful because of the low PCB
content in the samples. (See Table 4.1)
40
-------
Table 4.2. Calculated emission factors (E) and normalized emission factors (NE) at room temperature (ug/m2/h)
Sample
ID
CK-Ola
CK-Olb
CK-02a
CK-02b
CK-03
CK-04
CK-08 [c]
CK-09
CK-lOa
CK-lOb
Parameter
E
NE
E
NE
E
NE
E
NE
E
NE
E
NE
E
NE
E
NE
E
NE
E
Congener ID
#17
-
-
5.2
-
12.3
-
11.7
-
-
-
-
-
-
-
-
-
25.9
1153
27.5
#52
746
267
910
326
688
271
691
272
17.4
462-
106
172
4r05
424
424
464
1118
231
1310
#66
58.6
107
60.6
110
56.9
112
53.6
105
8.6
136
20.4
82.5
-
-
72.1
212
84.4
86.6
89.1
#101
-
-
-
-
299
59.5
326
64.9
114
81.6
158
51.5
22.8
27.0
575
126
441
47.8
480
#105
371
58.0
439.5
68.7
20.8
11.6
22.7
12.7
21.2
11.7
15.8
10.1
-
-
72.1
22.6
40.9
12.9
35.4
#110
25
10.6
29
12.2
158
30.0
172
32.8
109
40.0
113
28.5
6.77
14.7
451
61.6
237
25.0
248
#118
202
29.1
233
33.6
72
16.9
88
20.5
70.8
20.6
56.9
16.6
4r9S
8749
352-
344
121
15.7
117
#154
102
17.6
115.6
20.0
28.7
55.6
27.7
53.7
12.2
61.5
14.1
40.8
12.2
25.0
56.9
106
42.2
43.4
42.4
#187
33.9
50.4
42.3
62.9
-
-
-
-
-
-
-
-
9.66
3.49
336
437-7-
-
-
-
41
-------
Sample
ID
CK-lla
CK-llb
CK-12
CK-13a
CK-13b
Parameter
NE
E
NE
E
NE
E
NE
E
NE
E
NE
Congener ID
#17
12^10
-
-
-
-
28.8
-
1.23
32Q
1.39
360
#52
270
40.6
182
34.7
156
906
288
22.0
66.6
25.3
76.8
#66
91.3
3.79
39.2
3.25
33.6
74.0
64.0
2.03
25.9
2.54
32.4
#101
52.0
20.8
38.2
19.5
35.7
365
56.8
7.27
14.3
8.45
16.6
#105
11.1
-
-
-
-
25.8
9.74
-
-
-
-
#110
26.1
9.99
16.6
9.48
15.7
182
25.6
4.21
7.80
4.83
8.95
#118
15.2
5.56
9.05
5.29
8.62
85.5
13.2
1.94
3.88
2.28
4.57
#154
43.7
-
-
-
-
35.9
-
-
-
-
-
#187
-
-
-
-
-
1.26
6.78
-
-
-
-
w Values in strikethrough font were calculated from the concentration data below the PQL.
w Caulk CK-01, CK-02, CK-10, CK-11 and CK-13 were tested in duplicate.
[c] Aroclor 1260.
42
-------
4.1.5 Dependence of the Emission Factor on Congener Content in Caulk Samples
There is a linear correlation between the content of a congener in the caulk and its emission factor:
(4.4)
where
E; = emission factor for congener i (|ig/m2/h)
x; = content of congener i in caulk sample ((^g/g)
aj = a constant specific to congener I [((ig/m2/h) / ((ig/g)]
Figure 4.5 shows the correlation for congener #52, the most abundant congener in most air samples. The
estimated constant (a^, confidence of determination (r2), and sample number (n) are presented in Table 4.3.
For the convenience of discussion, Equation 4.4 is referred to as the x-E correlation. It can be used to
estimate the emission factors once the congener or Aroclor concentration is known.
The x-E correlation also exists for Aroclor concentrations. (See Section 4.1.10).
1500
900 O
1000 2000 3000 4000 5000
Congener content in caulk (jig/g)
6000
Figure 4.5. x-E correlation for congener #52 (r2 = 0.9816; n = 8)
43
-------
Table 4.3. Estimated constants (aO for the x-E correlation
Congener ID
#52
#66
#101
#105
#110
#118
#154
Slope (aO
0.268
0.0809
0.0557
0.0112
0.0280
0.0163
0.0459
r2
0.9816
0.8709
0.9652
0.9016
0.9568
0.9201
0.7897
n
8
7
8
6
8
8
6
4.1.6 Dependence of Congener Emissions on Vapor Pressure (1) — the P-N Correlation
According to mass transfer theories, the rate of pollutant emission from a solid material is mainly controlled
by two parameters: the pollutant's solid/air partition coefficient and the diffusion coefficient in the material.
The former is a function of the vapor pressure of the pollutant whereas the latter is a function of the size of
the pollutant molecule and the properties of the solid material. A sensitivity analysis using a mass transfer
model (Little et al, 1990) suggested that, for pollutants with low volatilities such as PCBs, the partition
coefficient is the most important parameter for the emission rate. Thus, there should be a link between the
emission rates and the vapor pressures for different congeners.
The test results showed that an excellent correlation exists between the normalized emission factor and the
vapor pressure:
In NEl = bi + b2 In P,
(4.5)
where NEi = normalized emission factor for congener i (|ig/m2/h)
Pj = vapor pressure for congener i (torr)
bi, b2 = constants
Several sets of vapor pressure data for PCBs are available in the literature (i.e., Foreman and Bidleman,
1985; Fischer et al., 1992). In this study we selected the values from Fischer at al.'s method B approach
(Table 4.4) because they used experimentally determined vapor pressures of specific congeners to
interpolate the vapor pressures, whereas their Method A and (Foreman's method) used vapor pressures of
alkanes to interpolate the other congener vapor pressures.
44
-------
Table 4.4. Vapor pressures for the target congeners in Aroclor 1254
Congener
#17
#52
#66
#77
#101
#105
#110
#118
#154
#187
Cl#
o
J
4
4
4
5
5
5
5
6
7
P (torr)
5.82xlO"4
1.50xlO'4
4.42xlO'5
1.43 xlO'5
2.99xlO'5
5.82xlO'6
1.68X10'5
8.42xlO'6
1.36X10'5
2.79xlO'6
Figure 4.6 shows the correlation for caulk CK-10. The calculated constants, bi and b2, coefficient of
determination (r2), and sample number (n) are presented in Table 4.5.
10000
-§ 1000
100 -
•o
D
o
Z
10 -
1
l.OE-6
l.OE-5 l.OE-4
Vapor Pressure (torr)
l.OE-3
Figure 4.6. Correlation between the normalized emission factor and vapor pressure for eight target
congeners in caulk CK-10 (r2 = 0.9748). The content of #17 in the caulk was below the
PQL.
45
-------
Table 4.5. Estimated constants bt and b2 in Equation 4.5
Caulk ID
CK-Ola
CK-Olb
CK-02a
CK-02b
CK-03
CK-04
CK-08
CK-09
CK-10
CK-lla
CK-lla
CK-12
CK-13
bt
14.19
14.37
14.14
13.73
16.45
14.17
12.59
14.37
14.06
13.98
13.73
13.28
13.14
b2
0.964
0.967
0.956
0.913
1.146
0.971
0.879
0.916
0.960
1.025
0.993
0.988
1.005
r2
0.9418
0.9344
0.9305
0.9493
0.9200
0.9223
0.8846
0.9304
0.9748
0.9531
0.9766
0.9867
0.9899
n
7
7
7
7
6
6
4
7
8
5
5
6
5
[a] Statistics: bj = 14.02±0.90; b2 = 0.976±0.065; n=13.
Both constants (bi and b2) were observed to be consistent among different caulk samples, which indicated
that a single correlation could be applied to all caulk samples.
In 7VS = 14.02 + 0.976 In Pf
(4.6)
For convenience of discussion, Equations 4.5 and 4.6 are referred to as the P-N correlation. This correlation
can be used to predict the emission rate for a congener once its content in the caulk sample is known. For
example, congener #77, a dioxin-like PCB, can be detected in caulk samples but it is difficult to measure in
air samples because of its low concentration. The P-N correlation can be used to estimate its emission factor.
4.1.7 Dependence of Congener Emissions on Vapor Pressure (2) — the P-S Correlation
The slopes of the x-E correlation (a^ in Table 4.3 differed significantly from congener to congener and the
more volatile congeners had greater slopes. A plot of the slopes against the vapor pressure showed an
excellent correlation (Figure 4.7). Equations 4.7 and 4.8 make it possible to predict the value of the slopes
for other congeners.
46
-------
0.40
0.30 -•
0.20 -•
0.10 -•
0.00
O.OE+0 5.0E-5 l.OE-4 1.5E-4
Vapor Pressure (torr)
2.0E-4
Figure 4.7. Slope of the x-E correlation (ai) as a function of congener vapor pressure
(r2 = 0.9925; n=7)
a,-= 0.00504+ 1753
r2 = 0.9925 (n=7)
(4.7)
or
a,•= 1*05 Pf
where
r2 = 0.9902 (n=7)
3i = slope of the x-E correlation for a given congener [((ig/m2/h) / ((ig/g)]
P; = vapor pressure of the congener (torr)
(4.8)
Equations 4.7 and 4.8 are designated the P-S correlation, where P represents vapor pressure and S represents
the slope in the x-E correlation. A combination of the x-E and P-S correlations can be used to predict
emission rate of a congener from its content in the caulk sample. Between the two correlations, Equation 4.8
is recommended.
4.1.8 Temperature Dependence of the Emission Factor
Seasonal variations of indoor PCB concentrations have been observed, suggesting a significant effect of
temperature on PCB emissions from caulk and other PCB sources (Minegishi et al., 2010). As described by
the x-E correlation, the emissions of PCB congeners from the primary sources are driven mainly by the
vapor pressure. Because the vapor pressure increases with increasing temperature (Paasivirta and
Sinkkonen, 2009), the congener emission rates are expected to increase as the temperature increases.
47
-------
To quantify the effect of temperature, two caulk samples (CK-1 1 and CK-13) were tested for emissions in
the micro-chambers at four different temperatures. The effect of temperature on vapor pressure can be
expressed by Equation 4.9, which is known as the Clausius-Clapeyron relation:
RT (4.9)
where P = vapor pressure (torr)
AH = enthalpy of vaporization (J moF1)
R = gas constant (8.3 14 J moF1 K'1)
T = temperature (K)
c = constant
In this study an equation similar to the Clausius-Clapeyron relation was used to determine the dependence
of the normalized emission rates on the temperature (Equation 4.10):
(4.10)
,
1 rri
1
where NE = normalized emission factor (|ig/m2/h)
T = temperature (K)
di and d2 = constants
Figures 4.8 and 4.9 show the correlations. In Figure 4.8, data for sample CK-1 1 at 40 °C were discarded
because of unexpected low concentrations possible due to a sampling leakage. Estimated constants (di and
d2) in Equation 4.10, confidence of determination (r2), and sample number (n) are presented in Table 4.6.
48
-------
10000
1000 -•
"oh 100
a
Z
10 -•
1
0.00320
0.00325
0.00330
0.00335
D#66
A#101
X#110
X#118
0.00340
Figure 4.8. Normalized emission factor (NE) as a function of temperature for five congeners in caulk
sample CK-11
1000
M)
a
Z
100 -•
10 -•
1
0.0031
0.0032
0.0033
1 / T (K)
0.0034
D#66
A#101
X#110
X#118
0.0035
Figure 4.9. Normalized emission factor (NE) as a function of temperature for five congeners in caulk
sample CK-13 (trend lines for congeners #66 and #101 are superimposed)
49
-------
Table 4.6. Constants dt and d2 in the N-T correlation for caulk sample CK-11 and CK-13[a
Congener ID
#52
#66
#101
#110
#118
CK-11 (n=3)
di
63.0
64.0
63.4
63.4
64.2
d2
1.70xl04
1.78xl04
1.77xl04
1.79xl04
1.83xl04
r2
0.960
0.973
0.967
0.962
0.964
CK-13 (n=4)
di
38.7
41.3
41.9
41.4
42.7
d2
l.OlxlO4
1.13xl04
1.15xl04
1.16xl04
1.21xl04
r2
0.986
0.978
0.989
0.988
0.987
LaJ Air concentrations are presented in Tables A.5 and A.6 in Appendix A.
The intercepts (di) and the slopes (d2) for different congeners in the sample were very close to each other,
but they differ between the samples. According to the coefficients for CK-11, the normalized emission
factor increases by a factor of 5.4 to 9 when the temperature increases by 10 °C in the temperature range of
10 to 50 °C. The coefficients for CK-13 predict an increase by a factor of only 3 to 4. It appears that the
composition of caulk has a significant effect on the temperature dependence of the emission rate. More tests
are needed to reduce the uncertainty in predicting the temperature effect.
4.1.9 The Difference between the Exposed and Freshly-cut Caulk Surfaces
Since there were continuous emissions from caulk, mass transfer models would suggest that a concentration
gradient may exist between the exposed surface and the interior of the source material. In other words, the
pollutant concentration is lower near the exposed surface than in the deep layers. Consequently, the
emission rate at the exposed surface is expected to be lower than that at the newly cut surface of the same
sample. To determine whether this difference is significant, three caulk samples were tested to compare the
emission rates from the two types of surfaces (Figure 4.10). These caulk samples, as received from the
buildings, had coatings on their exposed surfaces. There was a thin layer of clear coat on the exposed
surface of caulk CK-01. Caulk CK-02 had two coats. The top coat looked like plain latex paint and was
severely deteriorated. Caulk CK-12 had a thick layer of black gloss paint on the exposed side.
50
-------
Figure 4.10. Caulk samples for testing the PCB emission rates of different surfaces
From left to right: caulk samples CK-12, CK-01, and CK-02
(Samples with the exposed surfaces are in top row, and those
with newly cut surfaces are in bottom row.)
Comparisons of the emission factors are presented in Tables 4.7 to 4.9. For caulk CK-01, the emission
factors for the exposed surface were slightly higher than those for the newly cut surface, but the differences
were within the range of experimental error. For caulk CK-02 and CK-12, the emission factors for the
exposed surfaces were 36.7% and 25.6% lower than their respective emission factors for the newly cut
surfaces. Overall, the limited number of tests shows that the difference between the exposed and newly-cut
surfaces is 40% or less.
For caulk CK-12 there seemed to be a linear relationship between the ratio of the emission factors (ES/E0)
and the logarithm of the vapor pressure (Figure 4.11). Such a correlation simply means that, the more
volatile the congener, the greater the concentration gradient between the surface and the interior of the
source. However, such a trend is not apparent for caulks CK-01 and CK-02.
51
-------
Table 4.7. Emission factors (ug/m2/h) for the exposed surface (Es) and the newly cut surface (E0) for caulk CK-01
Emission factor
p M
J-^s
F M
•M)
ES/EO
p-Value
Congener ID
#52
439±21.7
481±22.3
91.3%
0.008
#66
35.3±1.68
34.6±2.28
102.0%
0.299
#101
257±13.4
244±7.83
105.2%
0.052
#105
20.5±3.23
17.7±2.50
116.3%
0.077
#110
142±13.4
130±8.75
109.1%
0.069
#118
76.7±8.56
65.6±5.10
117.0%
0.018
#154
24.4±1.57
23.9±0.96
101.9%
0.293
Mean ± SD (n = 5) for all emission factors.
Table 4.8. Emission factors (ug/m2/h) for the exposed surface (Es) and the newly cut surface (E0) for caulk CK-02
Emission factor
Es[a]
EoM
ES/EO
p-Value
Congener ID
#52
292±21.1
496±33.2
58.9%
<0.001
#66
29.6±2.16
48.5±1.94
61.0%
<0.001
#101
170±15.1
265±13.7
64.2%
<0.001
#105
14.0±2.10
21.3±3.10
65.8%
0.001
#110
93.3±10.1
143±9.4
65.2%
<0.001
#118
52.0±6.17
80.1±8.60
64.9%
<0.001
#154
16.1±1.47
25.6±1.59
63.0%
<0.001
1 Mean ± SD (n = 5) for all emission factors.
Table 4.9. Emission factors (ug/m2/h) for the exposed surface (Es) and the newly cut surface (E0) for caulk CK-12
Emission factor
Es[a]
E0[a]
ES/EO
p-Value
Congener ID
#17
15.7±0.77
28.8±1.21
54.5%
<0.001
#52
598±9.22
906±35.7
66.1%
<0.001
#66
58.6±3.71
74.0±7.25
79.2%
0.001
#101
281±7.21
365±16.6
76.9%
<0.001
#105
21.7±3.55
25.8±4.54
83.8%
0.072
#110
141±29.4
182±15.2
77.9%
0.013
#118
75.5±5.54
85.5±5.15
88.2%
0.009
#154
24.7±1.86
35.9±2.91
68.7%
<0.001
1 Mean ± SD (n = 5) for all emission factors.
52
-------
100%
80% -•
60% -•
40%
l.OE-6
l.OE-5 l.OE-4
Vapor Pressure (torr)
l.OE-3
Figure 4.11. Ratio of the emission factors for the exposed surface (Es) and the newly cut surface (E0)
as a function of vapor pressure (r2 = 0.746; n = 8)
4.1.10 Emission Factors for Aroclors
This study focused on individual congeners to a greater extent than on Aroclors because we expected the
transport rates to be dependent on the properties of individual congeners such as vapor pressures and
because we expected different transport rates for different congeners. We needed to quantify the individual
congeners to best characterize and model their transport through the indoor environment. However,
expressing the emission factors as Aroclors is of practical interest because most field measurements of PCB
concentrations in indoor air are given as Aroclors. This report accommodated these interests by following an
adaptation of Method 8082A, which calculates the Aroclor concentration in three steps:
Step 1: One-point calibration to determine the response of three to five major congeners to the amount of
Aroclor injected.
(4.11)
where Rt = response factor of congener /' per nanogram of Aroclor standard injected
Asi = area count for congener /
Ws = amount of Aroclor standard injected (ng)
53
-------
Step 2: Calculate the amount of Aroclor in sample x based on individual congener peaks
A
wx,=—
7?
(4.12)
where W^ = amount of Aroclor in sample x based on congener /'
A-a = area count for congener /' in sample x
Ri = response factor of congener /' in the Aroclor (from Equation 4.1 1)
Step 3: Calculate an average based on three to five major congener peaks
n
IX
n
(4.13)
where Wx = calculated amount of Aroclor in sample x
n = number of congener peaks used to calculate the Aroclor concentration (3 < n < 5)
In this study, we deviated from strict adherence to the 8082A method because we were quantifying
individual congeners. The method described below is equivalent to the original methods described above.
Step 1: One-point calibration to determine the content of three to five major congeners in the Aroclor.
w
77 — si
~
where Ft = weight fraction of congener /' in the Aroclor standard injected
wsi = content of congener /' in the Aroclor standard injected (ng)
Ws = amount of Aroclor standard injected (ng)
Step 2: Calculate the amount of Aroclor in the sample based on individual congener peaks
w •
W • = —2-
' ' VI
(4.14)
(4.15)
where W^ = amount of Aroclor in sample x based on congener / (ng)
wxi = amount of congener / in sample x (ng)
F,•= weight fraction of congener/' in the Aroclor (from Eq. 2.1)
54
-------
Step 3: Calculate an average based on three to five major congener peaks
2X,
W =—
n
where Wx = calculated amount of Aroclor in sample x
W-a = amount of Aroclor in sample x based on congener /'
n = number of congener peaks used to calculate the Aroclor concentration (3 < n < 5).
The calculated emission factors as Aroclors are presented in Table 4.10.
Table 4.10. Aroclor 1254 concentrations in caulk samples (x) and chamber air (C) and the
calculated emission factors (E)
(4.16)
Caulk
ID
CK-Ola
CK-Olb
CK-02a
CK-02b
CK-03
CK-04
CK-09
CK-lOa
CK-lOb
CK-lla
CK-llb
CK-12
CK-13a
CK-13b
x
(ng/g)
96100
74300
52100
42600
93300
136000
9128
103000
8280
C
Oig/m3)
74.9
82.6
57.2
62.3
10.1
19.9
46.0
86.5
70.6
8.36
7.87
154
8.01
9.14
E
(jig/m2/h)
5550
6920
5000
4990
1210
1900
5260
8030
5610
187
176
6450
179
207
Similar to individual congeners, the x-E correlation (Equation 4.4) is applicable to the emission factors for
Aroclor 1254 (Equations 4.17 and 4.18). Figure 4.12 shows the emission factor as a function of Aroclor
1254 content in caulk samples.
55
-------
E = -675 + 0.0672 x
r2 = 0.9454 (n = 9)
(4.17)
or
E = 0.0600 x
r2 = 0.9301 (n=9)
where E = emission factor for Aroclor 1254 (|ig/m2/h)
x = content of Aroclor 1254 in caulk sample (ug/g)
10000
(4.18)
0 30000 60000 90000 120000 150000
Aroclor Content in Caulk (ug/g)
Figure 4.12. Emission factor for Aroclor 1254 as a function of Aroclor content in caulk sample (r2 =
0.9301; n = 9)
The validity and usefulness of expressing concentrations of PCBs in air as Aroclors are debatable. More
discussion on this matter is given in Section 5.4.
4.2 Ceiling Tiles
The congener peak patterns in the three ceiling tile samples were similar but the Aroclor type could not be
positively identified (Figure 4.13). The three samples may be of the same products and may have
experienced similar weathering conditions because their congener profiles are similar (Figure 4.14). The
content of target congeners in the samples is presented in Table 4.11.
56
-------
llltlIIL
Aroclor 1254
. , ., Ill, I
•iiuli
Aroclor 1260
III . . .
Aroclor 1262
Aroclor 1268 1 |
. 1 II L, .1
., .uL
8 23 28
J]
Ceiling tile CT-01
33 3
Time (min)
Figure 4.13. Comparison of chromatograms - from top to bottom: Aroclors 1254,1260,1262, and
1268 and ceiling tile CT-01
•a
=
s
0)
o%
r-m
#52 #101 #154 #110 #66 #118 #105 #187
Congener Number
Figure 4.14. Relative abundances of the target congeners in three ceiling tile samples
57
-------
Table 4.11. Concentrations of target congeners in ceiling tile samples (jig/g)'
Congener
ID
#17
#52
#66
#77
#101
#105
#110
#118
#154
#187
Ceiling Tile ID
CT-01
Q QQg
0.124
0.125
Q Q]^
0.931
3.81
1.85
3.53
0.179
1.53
CT-02
Q QQg
0.059
0.082
Q Q]^
0.242
1.07
0.502
1.05
0.048
0.5
CT-03
Q QQ/I
0.117
0.108
Q 0^5
0.5
2.06
0.957
1.59
0.087
0.687
To convert the congener content to ug per cm2 paint, multiply the values in the table by the density of the ceiling tile
(0.063 g/cm3) and then by the height of the ceiling tile (2 cm).
To determine whether the PCBs were in the paint or fiber, a piece of the ceiling tile was split into two parts
at approximately % of the height from the top (i.e., the painted side). The two parts were extracted
separately. The results confirmed that the PCBs were mainly in the paint (Figure 4.15). The unevenness of
the paint (Figure 2.4) may have contributed to the difference in PCB content between the three samples
shown in Table 4.11.
The congener concentrations in air samples were all below the practical quantification limit, so the emission
factors were not reported. However, the data did show that the P-N correlation could be applied to ceiling
tiles (Figure 4.16).
58
-------
#52 #101 #154 #110 #118 #105 #187
Congener Number
Figure 4.15. Congener content in the top (with paint) and bottom layers of the ceiling tile
1000
l.OE-06
l.OE-05
Vapor Pressure (torr)
l.OE-04
Figure 4.16. Normalized emission factor as a function of vapor pressure for sample CT-03 (All
congener concentrations in air samples were below the practical quantification limit)
4.3 Light Ballasts
4.3.1 Test Summary
Three types of chamber tests were conducted for PCB emissions from light ballasts: screening tests, live
tests, and tests for temperature effect. The purpose of screening tests, which were conducted at room
temperature and without electrical load, was to identify leaking ballasts. The live ballast tests were
59
-------
conducted with the matching fluorescent lamps on. The tests for temperature effect were conducted at five
different temperatures. The test conditions are provided in Appendix B.
4.3.2 Method for Calculating the Emission Rate
The emission rate for a PCB congener was calculated from Equation 4.8, which is identical to Equation 4.1
for caulk:
R = QC (4.19)
where R = emission rate ((ig/h)
Q = chamber air flow rate (m3/h)
C = air concentration with the light ballast in the test chamber ((ig/m3)
During the screening tests the 53-L chambers were found to be more difficult to clean than the micro-
chambers. As a result, the background concentration of congener #18, the most abundant congener in the
emissions, was above the practical quantification limit in several screening tests. These high backgrounds
were subtracted during the rate calculations (Equation 4.20):
R=Q(C-C0) (4.20)
where C0 = background concentration of PCB in chamber air ((ig/m3)
4.3.3 Screening Tests
As shown in Table 4.12, the emissions were low for all the ballasts tested, indicating that the leakages were
either minor or insignificant. Congener #18 made the largest contribution to the emissions.
60
-------
Table 4.12. Congener emission rates for light ballasts at room temperature and without electrical
load (units: ug/h)[a]
Ballast
ID
BL-01
BL-02
BL-03
BL-04
BL-05
BL-06
BL-07
BL-08
BL-09A
BL-09B
BL-09D
BL-09E
BL-10
BL-11B
BL-12
BL-13
Congener ID
#15
-
-
-
-
-
-
-
-
-
-
-
-
-
0.0040
-
#17
-
-
-
-
-
-
-
0.0032
-
-
-
-
0.0031
-
0.0105
-
#18
0.0026
0.0050
0.0003
0.0040
0.0074
0.0051
0.0014
0.0089
0.0068
0.0053
0.0043
0.0026
0.0111
-
0.0307
0.0031
#52
-
-
-
-
-
-
-
-
-
-
-
-
-
-
0.0036
-
1 Emission rates for #13, #22, #44, #49, and #64 were all below practical quantification limit.
4.3.4 Live Ballast Tests
Six light ballasts were tested under conventional use conditions (i.e., with electrical load). The calculated
emission rates are presented in Table 4.13. Overall, the emission rates were roughly of the same order as
those from the screening tests. Ballast BL-08 burst unexpectedly during a test. Details are described in
Section 4.3.6. The custom-made test chamber became unusable after the ballast burst, which made it
impossible to conduct more live ballast tests.
The PCB emission rates reported in Table 4.13 may be lower than the emission rates of the same ballasts
had they been operated under realistic operating conditions. In the real world, the light ballasts are often
placed in enclosures causing higher ambient temperatures locally (Rensselaer Lighting Research Center,
2004). Because the PCB emission rate is highly sensitive to the temperature (see section 4.3.5 below), the
PCB emission rates from light ballasts under realistic use conditions could be much higher.
61
-------
Table 4.13. Rates of congener emission from ballasts with electrical load (jig/h)[al [bl
Congener
ID
#13
#18
#17
#15
#22
#52
#49
#44
#64
BL-09A
-
0.0056
Q QQ2
0 0007
0 0001
-
-
-
-
Light Ballast ID
BL-09C
0 0007
0.0093
0 0027
Q QQ13
0 0005
0 0006
0 0001
0 pop /|
-
BL-09D [c]
0 0003
0.0101
0 002S
Q QQ15
0 0009
0 0006
0 0003
0 0003
0 0001
EL-W
0.0140
0.0190
0.0010
-
-
-
-
-
-
BL-11A
~
0 002S
0 Qoi
0 0029
-
-
-
-
-
LaJ Data for the burst ballast (BL-08) are presented in Section 4.3.6.
M Values in strikethrough font were calculated from air concentrations below the PQL.
[c] No chamber background sample for this test.
Although most congeners were below the PQL in air samples, the results do show that the P-N correlation
applies to light ballasts as well as caulk (Figure 4.17). The normalized emission rate for a light ballast is
defined by Equation 4.21:
x
(4.21)
where NR = normalized emission rate for a congener ((ig/h)
R = emission rate for the congener
x0 = reference concentration for the congener in the liquid source; x0 = 1000 ((ig/g)
x = actual concentration for the congener in the liquid source ((ig/g)
In Figure 4.17, the liquid source is pure Aroclor 1242 and the congener content in the liquid source is from
Table 4A, data column 7 (G3) in Frame et al. (1996).
62
-------
l.OE-4
M)
l.OE-5
lnNR =-1.10 + 1.13 InP
(r2 = 0.961 ;n=7)
l.OE-04
l.OE-03
Vapor Pressure (torr)
Figure 4.17. Dependence of congener emission rate on vapor pressure for light ballast BL-09C
4.3.5 Effect of Ambient Temperature
A study by Hosomi (2005) showed that the rate of PCB emissions from PCB-containing light ballasts
increases as the ambient temperature increases. The same trend was observed in this study (Figure 4.18).
The linear model used for ballasts (Equation 4.22) was similar to the one for caulk (Equation 4.10) except
that the normalized emission factor was replaced by the normalized emission rate:
(4.22)
where
NR = normalized emission rate ((ig//h)
T = temperature (K)
fi and f2 = constants
According to the estimated values for constants fi and f2 (Table 4.14), every 10 °C increase in temperature
results in an increase of the emission rate by a factor 3 to 6. These results are roughly in agreement with the
data reported by Hosomi (2005). The variations of the fi and f2 values for different ballasts suggest that
ballast type or condition has an effect.
63
-------
l.OE-2
l.OE-3 -•
l.OE-4 -•
l.OE-5 -•
l.OE-6
0.0031
0.0032
0.0033
1 / T (K)
0.0034
0.0035
Figure 4.18. Effect of ambient temperature on congener emissions from ballast BL-09C
4.3.6 Emissions from a Burst Light Ballast
Light ballast BL-08 failed during a live test. The power to the lamp was shut off (safety design) and there
was a substantial amount of thick oily residue coating the interior surface of the sampling manifold (Figure
4.19). The PUF sample that was being collected had a dark yellow color, not generally seen during sampling
(Figure 4.20). Because of safety concerns, the test was immediately suspended and the chamber was sealed
and moved to a fume hood.
64
-------
Table 4.14. Estimated constants (fi and f2) for the effect of ambient temperature on congener
emissions from light ballasts
Sample
ID
BL-02
BL-05
BL-09C
BL-12
fi
f2
i2
n
fi
f2
o
r
n
fi
f2
0
r
n
fi
f2
o
r
n
Congener ID
#18
24.9
10100
0.9143
5
36.6
13600
0.8821
5
33.2
12700
0.9637
6
49.0
16400
0.9623
5
#17
27.2
10800
0.8997
5
38.5
14300
0.8711
5
34.4
13100
0.9694
5
48.0
16200
0.9605
5
#15
-
-
-
-
22.0
9350
0.9896
3
42.0
15500
0.9666
5
56.8
19000
0.9612
4
#22
-
-
-
-
23.7
10000
0.9702
4
37.7
14600
0.9326
5
-
-
-
-
#44
-
-
-
-
11.8
6520
0.9868
3
-
-
-
-
-
-
-
-
#52
-
-
-
-
14.6
7260
0.9692
4
-
-
-
-
-
-
-
-
#49
-
-
-
-
12.3
6650
0.9914
3
-
-
-
-
-
-
-
-
65
-------
Figure 4.19. Condensation of fluids in the chamber outlet manifold after the failure
Figure 4.20. Comparison of the PUF sampling cartridge for ballast BL-08 (right) to a normal
cartridge (left)
The temperature profile for the chamber air indicated that the ballast became overheated shortly after the test
was started (Figure 4.21). The temperature increase suddenly at approximately 10 elapsed hours, suggesting
the possible time when the ballast failed.
66
-------
-------
Table 4.15. Concentrations of target congeners in chamber background (C0), during the live test
(C) and the calculated emission rates (R) for ballast BL-08 [a]
Congener
ID
#13
#15
#17
#18
#22
#44
#49
#52
#64
Aroclor 1242
CoOig/m3)
-
Q QQ2
Q QQg
Q Q2/|
ND
ND
ND
ND
ND
-
C (jig/m3)
2^7±
106
244
766
100
48.5
43.6
63.2
2&2
3270
R
(Mg/h)
1.33
5.63
12.9
40.6
5.32
2.57
2.31
3.35
1.07
173
[a] Values in strikethrough font are below practical quantification limit.
Table 4.16. Concentrations of target congeners in chamber air seven days after the burst of ballast
BL-08 and the calculated average emission rates (R)[al
Congener
ID
#13
#15
#17
#18
#22
#44
#49
#52
#64
Aroclor 1242
Sample set 1 [b]
Oig/m3)
5.6
18.6
43.9
154
(18.2) [c]
8.08
6.78
9.61
2.74
605
Sample set 2 w
ftig/m3)
$m
21.9
46.4
159
21.7
9:47-
8.74
12.8
3^i
659
R
(Mg/h)
0.574
2.19
4.88
16.9
2.15
0.948
0.839
1.21
0.337
68.3
LaJ Values in strikeout font are below practical quantification limit.
M Sample sets 1 and 2 were taken sequentially (4 hours apart); duplicate samples for each set.
[c] Recovery check standard failed to meet all acceptance criteria for this sample. Emission rate was calculated by using C2
68
-------
Figure 4.22. PUF sampling from the sealed 53-L chamber containing the burst ballast
After air sampling, the burst ballast chamber was opened inside the fume hood. The fan and its supporting
wires were coated with a sticky black resin. The chamber walls were also darkened. These signs suggested a
smoldering period immediately after the failure. The ballast showed leakage of a tar-like resin and gel-like
material, as shown in Figure 4.23. Samples of these materials were collected and the analytical results are
presented in Table 4.17.
The failed ballast was opened later for examination. Details are described in Section. 4.3.6.
Figure 4.23. Light ballast CK-08 after the burst (the tar-like material is on the right and the gel-like
material is on the left)
69
-------
Table 4.17. PCB content in the gel-like material and the tar-like resin collected from the chamber
floor (units: jig/g)
Congener
ID
#13
#18
#17
#15
#22
#52
#49
#44
#64
Aroclor 1242
Material [al
"Gel"
330
5270
1830
1180
1530
924
695
854
364
31000
Tar
46r7
388
148
117
79.4
30
24^
23rS
9rS
1900
1 See Figure 4.23.
4.3.7 Inside the Ballasts
4.3.7.1 Physical Descriptions
Three ballasts (BL-02, BL-08, and BL-12) were opened to collect the fluids in the capacitor for Aroclor
identification. According to ANZECC (1997), each ballast has a capacitor that is cylindrical or rectangular,
encased in an aluminum container with a weld running all the way around the top edge with two quick-
connect terminals. The capacitors in the three light ballasts fit the description well except that the one in BL-
02 appeared to have three terminals.
There were no signs of fluid leakage in ballasts BL-02 and BL-12 (Figures 3.24 - 4.28). To collect the fluid
sample, a screwdriver was used to punch a hole (approximately 5 mm long and 1.5 mm wide) along the top
edges. The capacitor was almost full of liquid. Approximately 2.5 mL of fluid were collected from each of
the two ballasts with a glass pipette. After sampling, the hole was sealed with a silicone rubber sealant.
Samples of the potting material were taken from different locations and extracted by the method for caulk
samples. The analytical results are presented in Section 4.3.7.2.
70
-------
Figure 4.24. Ballast BL-02 after the bottom metal plate was removed (the entire casing was filled with
the potting material)
Figure 4.25. Ballast BL-02 (top side) (the capacitor is on the right)
71
-------
Figure 4.26. Capacitor in ballast BL-02
Figure 4.27. Ballast BL-12 after removing the casing (the capacitor is on the left)
72
-------
Figure 4.28. Capacitor in ballast BL-02
BL-08 was the ballast that burst during a live test. The potting material on the opposite side of the capacitor
showed signs of burning (Figure 4.29) and a total loss of elasticity. The capacitor in this ballast ruptured
(Figure 4.30), creating a small opening (approximately 1 mm in diameter) near one of the two wire
terminals. No fluid could be seen inside the capacitor. The attempt to collect a fluid sample with a glass
pipette was unsuccessful. The opening was then widened with a screwdriver. The capacitor was turned
upside down to allow any residual fluid to drip through the opening. Approximately 1.5 mL of fluid was
collected. Unlike the fluids in ballasts BL-02 and BL-012, which were clear, this fluid was yellow (Figure
4.31). A small amount of fluid could be seen on the potting material that was in contact with the capacitor.
Figure 4.29. Ballast BL-08 after removing the bottom metal plate (above) (there were signs of
smoldering on the left side; capacitor is on the right side)
73
-------
Figure 4.30. The capacitor in the burst ballast (BL-08) (note the expansion on both ends likely due to
the instantaneous high pressure)
Figure 4.31. Fluid collected from the ruptured capacitor in ballast BL-08
74
-------
4.3.7.2 Analytical Results
The fluids collected from the three capacitors were all Aroclor 1242 (Figure 4.32). The PCB content in the
potting material varied from ballast to ballast. For ballast BL-02, the material near the capacitor contained a
much higher concentration of PCBs than the opposite side (Table 4.18), suggesting early development of
leakage. Ballast BL-12 showed a similar trend, but the contamination was more modest (Table 4.19). As can
be expected, the potting material in the burst ballast (BL-08) was severely contaminated (Table 4.20).
ft . A 11 1" \ fl A.
A . . Jl . IV A
A . . ft 1" * /V ^
ft ^ »I . (UJ\ .
Aroclor 1242 standard
Jlj. .jK_lnM> -A JV .-«-- .
Ballast BL-02
LL . k HM. urt. Ml.
Ballast BL-08
JL,\I_^...AK J*.A» • «^n, t-
Ballast BL-12
10 12 14 16 18 20 22 24
Retention Time (min)
_«j(JL~jJL
26
Figure 4.32. Comparison of chromatograms for (from top to bottom) Aroclor 1242 standard and
fluids in light ballasts BL-02, BL-08, and BL-12
75
-------
Table 4.18. Congener content in potting material in BL-02 (units: jig/g)'
Congener
ID
#13
#18
#17
#15
#22
#52
#49
#44
#64
Aroclor 1242
Sampling Locations Ibl
A
0:4*
9.19
3.68
3.45
1.10
0.71
0:44
0.61
0:32-
42.3
B
11.5
179
70.9
111
38.1
9.25
6.37
5.95
3.00
861
C
93.1
1890
707
669
544
266
179
222
94.8
10800
D
379
5980
2180
1460
1410
1000
593
823
312
33400
[a] Numbers in strikethroughfont are below PQL; numbers in bold are average of duplicate samples withRSD greater
than 25%.
M Locations:
A = from the end of the ballast opposite to the capacitor
B = from the middle section of the ballast
C = from the wiring side of the capacitor
D = from underneath the capacitor
76
-------
Table 4.19. Congener content in potting material in BL-12 (units: jig/g)'
Congener
ID
#13
#18
#17
#15
#22
#52
#49
#44
#64
Aroclor 1242
Sampling Locations Ibl
A
0.86
10.1
4.00
5.36
2.53
1.00
0.77
0.65
Q gg
53.3
B
10.2
144
52.2
53.2
50.9
44.8
32.9
40.4
20.1
1036
C
47.3
714
252
190
260
335
230
348
153
6101
D
30.1
466
165
110
125
154
108
137
58.4
3200
Numbers in strikethrough font are below PQL.
Locations:
A = from the end of the ballast opposite to the capacitor
B = from the middle section of the ballast
C = from the wiring side of the capacitor
D = from underneath the capacitor
77
-------
Table 4.20. Congener content in the potting material in the burst ballast (BL-08) (units: jig/g)'
Congener
ID
#13
#18
#17
#15
#22
#52
#49
#44
#64
Aroclor 1242
Sampling locations [bl
A
57.8
898
308
262
220
102
72.5
79.0
40.1
4612
B
1090
18400
5700
2870
5660
4390
3100
4670
1840
117000
C
1680
28100
7940
3980
8370
7300
4950
7180
3000
176000
D
±723
27700
8560
3920
8300
6920
4790
6670
2780
175000
E
1080
17000
6070
2500
5610
4630
3230
4860
1980
118000
F
1050
17000
5410
2570
5180
4460
3060
4390
1740
110000
Numbers in strikethrough font are below PQL.
Locations:
A = from the end of the ballast opposite to the capacitor; burned; lost elasticity;
B = from the middle section of the ballast;
C = from the wiring side of the capacitor;
D = from underneath the capacitor; w/ leaked fluid;
E = from outside of the ballast;
F = next to the wiring side of capacitor; w/ leaked fluid.
78
-------
5. Discussion
5.1 Predicting the Emission Factors for PCB-Containing Caulk
When the congener content in the caulk is known, its emission factor can be estimated by using either the x-
E correlation (Equation 4.4) or the P-N correlation (Equation 4.6). Details are discussed below.
5.1.1 Using the x-E Correlation (Method 1)
The calculation includes two steps: (1) obtain the congener content in caulk [x; in (ug/g)] and (2) calculate
the emission factor using Equation 4.4. The coefficient, ab can be found in Table 4.3. For example, if a caulk
contains 5000 ug/g of congener #52, its estimated emission factor is:
E, = 0.268 x 5000 = 1340 (ug/m2/h) (5.1)
For congeners that are not listed in Table 4.3, the P-S correlation (Equation 4.8) can be used to estimate
coefficient a^
5.1.2 Using the P-N Correlation (Method 2)
As illustrated, the calculation includes three steps by using the value for congener #52 mentioned above:
Step 1: Obtain the congener content in caulk [x; in (ug/g)] and the vapor pressure (Pj in torr):
^ = 5000 (ug/g) (5.2)
P,= 1.497xlO'4 (torr) (5.3)
Step 2: Calculate the normalized emission factor (NE;) from Equation 4.6:
ln7Vs = 14.02 + 0.976 In 1.497 x 10'4 = 5.42 (5.4)
Nm = 226 (ug/m2/h) (5.5)
Step 3: Convert the normalized emission factor to the emission factor (E;):
Et = 226 - 1000 x 5000 = 1130 (ug/m2/h) (5.6)
5.1.3 Predictive Errors
The predictive errors for the two methods were calculated by using Equation 5.7, and the results are
presented in Table 5.1.
79
-------
s =
—F
^
xlOO%
(5.7)
where e = predictive error (%)
Ep = predicted emission factor (|ig/m2/h)
Em = measured emission factor (|ig/m2/h)
Table 5.1. Predictive error for the x-E and P-N correlations
Congener
ID
#52
#66
#101
#105
#110
#118
#154
Average
Correlation
x-E (Eq. 4.4) [bl
28.0%
39.7%
32.2%
13.3%
32.0%
26.8%
39.0%
30.1%
P-N(Eq. 4.6) [c]
30.0%
40.8%
32.3%
21.6%
31.2%
29.7%
59.4%
35.0%
LaJ Sample CK-13 (laboratory-mix caulk) was excluded.
w Coefficients from Table 4.3.
w Coefficients from Equation 4.6.
5.1.4 Method Selection
Method 1 is recommended for congeners that are listed in Table 4.3. Method 2 is recommended for other
congeners.
5.7.5 Predicting the Emission Factors for Aroclor 1254
The emission factor for Aroclor 1254 can be calculated from Equation 4.18. The average predictive error
was 32.1%, excluding the laboratory-mixed sample (CK-13). It is emphasized that the composition of the
congener mixture in air samples is significantly different from that in the Aroclor 1254 standard or that in
caulk samples. In general, there are proportionally more volatile congeners in the air. As a result, there is
greater uncertainty when the air concentration is expressed in Aroclor. See section 5.4 for more discussion.
80
-------
5.1.6 Estimating the Air Concentration Due to Emissions from Caulk
At steady-state conditions, Equation 5.8 (for a single source) or Equation 5.9 (for multiple sources) can be
used to estimate the contribution of emissions from caulk to the PCB concentration in the air:
Q
where C = congener or Aroclor concentration in room air ((ig/m3)
A = source area (m2)
E = emission factor for a congener or Aroclor (|ig/m3/h)
Q = air change flow rate (m3/h)
(5.8)
(5.9)
Q
where C = congener or Aroclor concentration in room air ((ig/m3)
n = number of sources
A; = source area for the ith source (m2)
Ej = emission factor for the 1th source for a congener or Aroclor ((ig/m2/h)
Q = air change flow rate (m3/h)
5.2 Using the Advanced Emission Models for Emissions from Caulk and Other Building Materials
The results presented above represent the current status of the caulk samples. To estimate their emissions in
the past or future, mathematical models must be used. Emissions of volatile and semi-volatile chemicals
from solid building materials have been studied extensively in the past two decades and, consequently,
many mass transfer models have been developed (Little et al, 1994; Huang and Haghighat, 2002; Xu and
Yang, 2003; Deng and Kim, 2004; Qian, et al., 2007). While differing in complexity and applicability, these
models have several features in common:
• They are all derived from the Pick's second law.
• They use the same set of parameters to describe the source, i.e., the content of the chemical in the
source, the solid/air partition coefficient, the diffusion coefficient of the chemical in the source, and the
area and thickness of the source.
• With one exception, they all require that a non-linear equation be solved.
81
-------
To apply the mass transfer models to PCB emissions from caulk, the partition and diffusion coefficients for
PCB congeners must be determined. Although several methods are available for experimental determination
of these two parameters (Bodalal et al., 2000; Cox et al., 2001, Haghighat et al., 2002), their applicability to
PCB congeners is questionable because of the low vapor pressures of these compounds. In this study,
chamber data and existing quantitative structure-activity relationship (QSAR) models were used to make
rough estimations of these two parameters. Technical details are described in Appendix C. The simulation
conditions and results are summarized below.
Room volume
300m3
Ventilation rate
Caulk area
1 air change per hour
0.5m2
Caulk density
1.5g/cm3
• Initial content of Aroclor 1254 in caulk 10% by weight
The time-concentration profiles and percent of mass emitted for congeners #52, #77, #154, and #187 over a
50-year period are shown in Figures 5.1 and 5.2, respectively. These results should be considered as semi-
quantitative.
100
f>
I
=
o
4*
u
=
ij
10 -
1 -
0.1 -
0.01 - ;•
0.001 -
0.0001
10 20 30 40
Elapsed Time (years)
50
Figure 5.1. Predicted congener concentrations over a 50-year period
82
-------
50%
10 20 30 40
Elapsed Time (years)
50
Figure 5.2. Percent of congener mass emitted over a 50-year period
5.3 Using the Emissions Data for Light Ballasts
The behavior of PCB-containing light ballasts as an emission source is difficult to predict. The test results
showed that ballasts with no fluid leakage or with small amounts of fluid leakage do not emit significant
amounts of PCBs. The rates of PCB emissions from light ballasts increase quickly at elevated temperature.
The live tests (i.e., with electrical load) conducted in this study may have been at lower temperatures than
the operating temperatures under realistic operating conditions. In addition, chamber walls may adsorb
PCBs and cause underestimation of the emission rate. Thus, the test results should be considered the lower
bounds for the emission rates. Because the PCB-containing ballasts that are currently in use have
approached or even exceeded their designed service life, leakage of PCB fluid will develop. More
importantly, the rate of capacitor failure increases drastically as the light ballasts age (Philips, undated).
PCB release from failed light ballasts have been reported in the United States (Staiff et al., 1974) and Japan
(Funakawa et al., 2002; Hosomi, 2005). In this study, ballast BL-08 burst during a live test and,
consequently, most of the Aroclor 1242 in the capacitor was ejected into the air inside the test chamber.
Equation 5.10 (Hosomi, 2005) is recommended for estimating the PCB concentration in room air at steady-
state conditions:
83
-------
(5.10)
Q
where C = congener or Aroclor concentration in room air ((ig/m3)
Rj = emission rate for ballast i ((ig/h)
n = number of ballasts in the room
Q = ventilation flow rate (m3/h)
For predicting the release of PCB fluid from a failed ballast, an existing liquid spill model is recommended.
Details are described in Appendix D.
This study did not test any light ballasts with apparent fluid leakages. Consequently, the test results may not
be representative of the large population of PCB-containing light ballasts that are currently in use in the
buildings in the United States.
5.4 Expressing the PCB concentrations as Aroclors
It has been a common practice to express the PCB concentrations in some environmental samples as
Aroclors (U.S. EPA, 2008c). The advantage of this approach is its simplicity. However, the uncertainties
associated with this method have never been fully addressed. Because the Aroclor concentration is a
calculated value based on the congener content in the Aroclor standards, the uncertainty of the calculation
depends on how similar the congener profile of the sample is to that of the Aroclor standards. For example,
the congener profile for caulk CK-10 (a weathered field sample in good condition) is similar to that for the
Aroclor 1254 standard (Figure 4.1), and the calculated Aroclor concentrations based on individual
congeners are very close to the average and the relative standard deviation (RSD) is only 10% (Table 5.2).
Caulk CK-02 (another field caulk) and C-13 (a laboratory-mix caulk) had similar variations. For caulk CK-
09, severely deteriorated, the variation was much greater. The variations for air samples were even worse
because, proportionally, there are more volatile congeners in the air than in caulk. Clearly, the greater the
variations between congeners, the more uncertainty there will be in the calculated Aroclor concentration. It
is also a critical factor to select the congener peaks, especially for air samples. As shown in Table 5.2, the
results were significantly different depending on whether congener #52, the most abundant congener in
caulk emissions, was included or not. A similar problem exists for air samples associated with emissions of
Aroclor 1242 from light ballast (Table 5.3).
The authors recommend further investigation into these issues. Standardization is needed for selecting
congener peaks for calculating the Aroclor concentrations in environmental samples, especially air samples.
84
-------
Table 5.2. Variations of Aroclor concentrations in caulk and air samples calculated based on five
individual congeners
Sample type
Caulk
(MS/g)
Chamber air
(ug/m3)
Caulk
ID
CK-02
CK-09
CK-10
CK-13
CK-02
CK-09
CK-10
CK-13
Calculated Aroclor 1254 concentrations
based on individual congeners [a]
#52
58500
6210
112000
7610
199
25.2
381
22.7
#101
75800
69000
140000
7680
61.4
76.0
91.4
4.91
#110
78400
109000
142000
8050
32.1
58.8
46.6
2.81
#118
80600
138000
145000
9370
20.6
41.4
27.6
1.63
#105
81100
144000
144000
8680
12.8
28.6
20.2
~
Concentration of
Aroclor 1254 w
Mean
74900
93300
136000
8280
65.2
46.0
113
8.01
RSD
13%
61%
10%
9.0%
118%
46%
134%
123%
[a] Calculated from Equations 4.14 and 4.15 in Section 4.1.10.
w Calculated from Equation 4.1.6 in Section 4.1.10.
Table 5.3. Variations of Aroclor concentrations in air sample for light ballast BL-08 calculated
based on five individual congeners (concentration units: ug/m3) [a][b]
Calculated Aroclor 1254 concentrations
based on individual congeners
C#17
1061
C#18
1202
C#22
434
C#44
146
C#52
182
Concentration of
Aroclor 1242
Mean
605
RSD
82.0%
LaJ The air sample was taken four days after the burst.
w See footnotes [a] and [b] for Table 5.2.
5.5 Study Limitations
This study was conducted in a relatively short period of time and only a few samples were tested. It was not
our intention to collect and test samples that are statistically representative of the primary sources in U.S.
building stock. The study was not intended to link the test results to the buildings from which the samples
were collected.
This study investigated several factors that may affect PCB emissions from caulk, including the PCB
content of the source, the properties of PCB congeners, temperature, and exposed versus unexposed surfaces
of the source. The effects of humidity and ventilation rate were not evaluated. The moisture content of the
air may have a significant effect on the emissions of hydrophilic pollutants such as formaldehyde, but PCBs
are highly hydrophobic, so the effect of humidity on PCB emissions is expected to be negligible. Although
85
-------
the effect of ventilation rate on air concentrations can be significant due to different levels of dilution, its
effect on the emission rate from a dry source (such as caulk) is rather small.
For similar reasons, we also chose not to evaluate the effects of humidity and ventilation rates on PCB
emissions from light ballasts. Although a light ballast that is leaking may be considered an evaporative
source, none of the 19 light ballasts we evaluated had any visible signs of PCB leakage.
Because of time constraint and technical difficulty, this study did not investigate the effects of caulk
composition and weathering conditions on PCB emissions. It should be a topic of future research because
understanding such effects will reduce the uncertainty in the QSAR models for PCB emissions such as the
x-E correlation.
Over a dozen types of primary sources have been identified in PCB-contaminated buildings (EH&E, 2011).
Our study tested only caulk, light ballasts, and very limited number of ceiling tile samples because of the
unavailability of other types of samples and short testing schedule.
86
-------
6. Conclusion
Among the 12 field caulk samples tested, 11 were determined to contain Aroclor 1254 and the remaining
sample was determined to contain Aroclor 1260. The Aroclor content ranged from less than 10 to 136000
(ig/g. A linear correlation, designated the x-E correlation, exists between the emission factor and Aroclor or
congener content in caulk. There are significant differences in the congener profiles between the caulk and
air samples; proportionally, more volatile congeners are found in the air samples. An excellent correlation
exists between the normalized emission factor and the vapor pressure of the congener on a logarithmic scale.
This correlation, which is designated the P-N correlation, allows the estimation of the emission factor for a
congener in the caulk as long as its content in the caulk and vapor pressure are known. These correlations
make it possible to estimate the emission factors for either congeners or Aroclor 1254, as long as their
content in the caulk is known. PCB emissions increase as temperature increases. The test of a field caulk
sample showed that, for every 10 °C increase between 10 and 50 °C, the emission factor increases by a
factor of 5.4 to 9. The PCB emissions from the exposed surface are less than the emissions from a newly cut
surface, but the difference is 40% or less based on a limited number of tests. Further study should include
developing methods for measuring the partition and diffusion coefficients for PCB congeners in caulk and
other building materials. These parameters can help further understand the sources, sinks, and mitigation
methods for PCBs in buildings. The effect of the composition of caulk and sealants on PCB emissions
should also be investigated.
The emissions from PCB-containing light ballasts are difficult to predict. Overall, the emission rates are
small at room temperature for light ballasts with no or little leakage of fluid. The emission rate increases
significantly at elevated temperature. The emission rates determined in the test chamber may be much lower
than those in realistic use conditions because, in the latter case, the ballasts are often located in enclosures
causing higher operating temperatures. This study did not test any light ballasts with visible fluid leakage
because of safety concerns. One ballast unit burst during a live test, causing release of Aroclor 1242 fluid
from its ruptured capacitor. Thus, the presence of PCB-containing light ballasts in buildings may
pose a potential risk to the occupants because most existing PCB-containing light ballasts have
approached or exceeded their designed service life and because the decontamination process is
both difficult and costly.
Overall, this study established a direct link between the PCB content in primary sources (caulk and light
ballasts) and PCB concentrations in room air by experimentally measuring the emission rates. The data and
empirical models reported above can be used to rank indoor PCB sources or as input for indoor contaminant
models and for exposure models. However, it is beyond the scope of this study to link the emissions to
potential health risks.
87
-------
Acknowledgments
The authors thank the building owners who provided field samples for this study; Kimberley Tisa of EPA
Region 1 and Dennis Santella and James Haklar of EPA Region 2 for facilitating sample acquisition;
Jacqueline McQueen of the EPA Office of Science Policy for assistance in communications; Kent Thomas
of the EPA National Exposure Research Laboratory for technical consultation; Dale Greenwell of the EPA
National Risk Management Research Laboratory, Russell Logan, Corey Mocka, and Aaron DeBlois of
ARCADIS for laboratory support; Robert Wright and Joan Bursey of the EPA National Risk Management
Research Laboratory and Libby Nessley of ARCADIS for QA support.
-------
References
(Website accessibilities were last verified on March 28, 2011)
ANZECC (1997). Identification of PCB-containing capacitors, Australian and New Zealand Environment
and Conservation Council (ANZECC).
http://www.environment.gov.au/settlements/publications/chemicals/scheduled-waste/pubs/pcbid.pdf
ASTM (2010). ASTM D5116-10 Standard guide for small-scale environmental chamber determinations of
organic emissions from indoor materials/products, ASTM International, West Conshohocken, PA.
ATSDR (2000). Toxicological profile for polychlorinated biphenyls (PCBs), Ch 5, Production,
import/export, use, and disposal, Agency for Toxic Substances and Disease Registry.
http://www.atsdr.cdc.gov/toxprofiles/tp.asp?id=142&tid=26
Balfanz, E., Fuchs, J., and Kieper, H. (1993). Sampling and analysis of polychlorinated biphenyls (PCBs) in
indoor air due to permanently elastic sealants, Chemosphere, 26(5): 871-880.
Bodalal, A, Zhang J. S., and Plett E. G. (2000). A method for measuring internal diffusion and equilibrium
partition coefficients of volatile organic compounds for building materials, Building and Environment, 35:
101-110.
Bodalal, A., Zhang, J. S., Plett, E. G., and Shaw, C. Y. (2001). Correlations between the internal diffusion
and equilibrium partition coefficients of volatile organic compounds (VOCs) in building materials and the
VOC properties, ASHRAE Transactions, 107: 789-800.
Cox, S. S., Zhao, D., and Little, J. C. (2001). Measuring partition and diffusion coefficients for volatile
organic compounds in vinyl flooring, Atmospheric Environment, 35: 3823-30.
Dean, R. B, and Dixon, W. J. (1951). Simplified statistics for small numbers of observations, Analytical
Chemistry, 23(4): 636-638.
Deng, B., and Kim, C. N. (2004). An analytical model for VOCs emission from dry building materials,
Atmospheric Environment, 38:1173-1180.
EH&E (2011). Literature review of remediation methods for PCBs in buildings, prepared for U.S. EPA by
Environmental Health & Engineering, Inc., Needham, MA (in press).
EIP Associates (1997). Polychlorinated biphenyls (PCBs) source identification, prepared for: Palo Alto
Regional Water Quality Control Plant.
http://www.cityofpaloalto.org/civica/filebank/blobdload.asp?BlobID=3772
Environment Canada (1991). Identification of lamp ballasts containing PCBs, Environment Canada Report
EPS 2/CC/2 (Revised). http://www.ec.gc.ca/Publications/default.asp?lang=En&xml=FlD91988-3B5E-
4956-A705-78D054685FFE
89
-------
Fischer, R. C., Wittlinger, R. and Ballschmiter, K. (1992). Retention-index based vapor pressure estimation
for polychlorobiphenyl (PCB) by gas chromatography, Fresenius' Journal of Analytical Chemistry, 342(4):
421-425.
Foreman, W. T. and Bidleman, T. F. (1985). Vapor pressure estimates of individual polychlorinated
biphenyls and commercial fluids using gas chromatographic retention data, Journal of Chromatography A,
330: 203-216.
Frame, G. M., Cochran, J. W., and B0wadt, S. S. (1996). Complete PCB congener distributions for 17
Aroclor mixtures determined by 3 HOC systems optimized for comprehensive, quantitative, congener-
specific analysis, Journal of High Resolution Chromatography, 19(12): 657-668.
Fromme H., Baldauf, A. M., Klautke, O., Piloty, M., and Bohrer, L. (1996). Polychlorinated biphenyls
(PCB) in caulking compounds of buildings — assessment of current status in Berlin and new indoor air
sources, Gesundheitswesen, 58(12): 666-672.
Funakawa, M., Takada, M., Niida, M., and Hosomi, M. (2002). Volatilization of PCB from PCB containing
ballast in fluorescent lamp and indoor PCB pollution, Journal of Environmental Chemistry, 12: 615-620.
Guo, Z. (2000), Simulation tool kit for indoor air quality and inhalation exposure (IAQX), Version 1.0
user's guide, U.S. EPA, National Risk Management Research Laboratory, Research Triangle Park, NC,
Report No. EPA-600/R-00-094 (NTIS PB2001-101221), 76 pp.
Guo, Z. (2002). Review of indoor emission source models — part 2. parameter estimation, Environmental
Pollution, 120:551-564.
Guo, Z. (2005). Program PARAMS user's guide, U.S. EPA, National Risk Management Research
Laboratory, Research Triangle Park, NC, Report No. EPA-600/R-05/066, 32 pp.
Haghighat, F., Lee, C-S., and Ghaly, W. S. (2002). Measurement of diffusion coefficients of VOCs for
building materials: review and development of a calculation procedure, Indoor Air, 12: 81-91.
Herrick, R. F., McClean, M. D., Meeker, J. D., Baxter, L. K., and Weymouth, G. A. (2004). An
unrecognized source of PCB contamination in schools and other buildings, Environmental Health
Perspectives, 112(10): 1051-1053.
Herrick, R. F., Meeker, J. D., Hauser, R., Altshul, L., and Weymouth, G. A. (2007). Serum PCB levels and
congener profiles among US construction workers, Environmental Health, 6(1): 25-32.
Hosomi, M. (2005). Volatilization of PCBs from PCB-containing ballast in fluorescent lamp and indoor
PCB pollution: odor of PCBs, Journal of Japan Association on Odor Environment, 36(6): 323-330.
Huang, H., and Haghighat, F. (2002). Modelling of volatile organic compounds emission from 3 dry
building materials, Building and Environment, 37: 1117-1126.
90
-------
Kohler M., Tremp, J., Zennegg, M., Seller, C., Minder-Kohler, S., Beck, M., Lienemann, P., Wegmann, L.,
and Schmid, P. (2005). Joint sealants: an overlooked diffuse source of poly chlorinated biphenyls in
buildings. Environmental Science & Technology, 39(7): 1967-1973.
Little, J. C., Hodgson, A. T., and Gadgil, A. J. (1994). Modeling emissions of volatile organic compounds
from new carpets, Atmospheric Environment, 28(2): 227-234.
MacLeod, K. (1979). Sources of emissions of polychlorinated biphenyls into the ambient atmosphere and
indoor air, U.S. EPA, Health Effect Research Laboratory, Research Triangle Park, Report EPA-600/4-79-
022.
MacLeod, K. (1981). Polychlorinated biphenyls in indoor air, Environmental Science & Technology, 15:
926-928.
Minegishi, T., Allen, J. G., Macintosh, D. L. (2010). Predicting seasonal indoor PCBs concentrations based
on fundamental equations, The 30th International Symposium on Halogenated Persistent Pollutants,
September 12-17, 2010, San Antonio, Texas, paper ID 1569.
http://www.xcdtech.com/dioxin2010/pdf/l 569.pdf
Monsanto (undated). Aroclor plasticizers, Monsanto Technical Bulletin O/PL-306.
Mydlova-Memersheimerova, J., Tienpont, B., David, F., Krupcik, J., and Sandra, P. (2009). Gas
chromatography of 209 polychlorinated biphenyl congeners on an extremely efficient nonselective capillary
column, Journal of Chromatography A, 1216(32): 6043-6062.
NERL (2010). A research study to investigate PCBs in school buildings, EPA 600R-10/074, National
Exposure Research Laboratory, http://www.epa.gov/pcbsincaulk/research-plan.pdf
Newman, D. M. (2010). PCBs in schools: what about school maintenance workers? New Solutions, 20(2):
189-191.
NIOSH (1975). Current Intelligence Bulletin 7: Polychlorinated biphenyls (PCBs), National Institute of
Occupational Safety and Health, http://www.cdc.gov/niosh/78127_7.html#background
Paasivirta, J. and Sinkkonen, S. I. (2009). Environmentally relevant properties of all polychlorinated
biphenyl congeners for modeling their fate in different natural and climate conditions, Journal of Chemical
Engineering Data, 34: 1189-1213.
Philips Lighting (undated). Ballast lifetime calculations, Philips Lighting Technical Note TN005.
http://lighting.philips.com/gl_en/global_sites/fluo-gear/dimming/download/pdf/technical-notes/tn005.pdf
Piloty, M. and Koppl, B. (1993). PCB caulking sealants — experiences and results of measures in Berlin and
decontamination of a school (Parti), Gesundheitswesen, 55(11): 577-81.
Prignano, A. L., Narquis, C. T., Hyatt, J. E. (2008). Generating the right PCB data: determination of
Aroclors versus PCB congeners, Waste Management 2008 Conference, February 24-28, Phoenix, AZ.
91
-------
Priha, E., Hellman, S., and Sorvari, J. (2005). PCB contamination from polysulphide sealants in residential
areas — exposure and risk assessment, Chemosphere, 59(4): 537-543.
Qian, K., Zhang, Y., Little, J. C., and Wan X. (2007). Dimensionless correlations to predict VOC emissions
from dry building materials, Atmospheric Environment, 41: 352-359.
Rensselaer Lighting Research Center (2004). Methods for the efficient operation of ceiling-mounted
fluorescent residential luminaries.
http://www.lrc.rpi.edu/programs/lightingtransformation/heatManagement/pdf/fmalThermalTesting.pdf
Robson M., Melymuk L., Csiszar S. A., Giang A., Diamond M. L., Helm P. A. (2010). Continuing sources
of PCBs: the significance of building sealants, Environmental International, 36: 506-513.
Rorabacher, D. B. (1991). Statistical treatment for rejection of deviant values: critical values of Dixon Q
parameter and related subrange ratios at the 95 percent confidence level, Analytical Chemistry, 63 (2): 139-
146.
Rushneck, D. R., Beliveau, A., Fowler, B., Coreen, H., Hoover, D., Kaye, K., Berg, M., Smith, T., Telliard,
W. A., Roman, H., Ruder, E., and Ryan, L. (2004). Concentrations of dioxin-like PCB congeners in
unweathered Aroclors by HRGC/HRMS using EPA Method 1668A, Chemosphere, 54(1): 79-87.
Schripp, T., Nachtwey, B., Toelke, J., Salthammer, T., Uhde, E., Wensing, M., and Bahadir, M. (2007). A
microscale device for measuring emissions from materials for indoor use, Analytical and Bioanalytical
Chemistry, 387(5): 1907-1919.
Staiff, D. C., Quiby, G. E., Spencer, D. L., and Starr, Jr., H. G. (1974). Fob/chlorinated biphenyl emission
from fluorescent lamp ballasts, Bulletin of Environmental Contamination and Toxicology, 12(4): 455-463.
Stallings, C., Zartarian, V., and Glen, G. (2008). Stochastic human exposure and dose simulation model for
multimedia, multipathway chemicals SHEDS-multimedia model, Version 3 user guide, EPA 600/R-08/118.
http://www.epa.gov/heasd/products/sheds_multimedia/sheds_mm.html
UNEP (1999). Guidelines for the identification of PCBs and materials containing PCBs, United Nations
Environment Programme. http://www.chem.unep.ch/Publications/pdf/GuidIdPCB.pdf
U.S. Department of Commerce (2009). PCBs in concrete structures.
http://www.docep.wa.gov.au/worksafe/content/safety_topics/hazardous_substances/Additional_resources/P
CBs_in_concrete_structures.html#3
U.S. EPA (1993). PCBs in fluorescent light fixtures, U.S. Environmental Protection Agency, Region 10, Air
and Toxics Division; U.S. Government Printing Office, Washington, DC.
U.S. EPA (1999). EPA method TO-10A (1999), Determination of pesticides and polychlorinated biphenyls
in ambient air using low volume polyurethane foam (PUF) sampling followed by gas
chromatographic/multi-detector detection (GC/MD), EPA Center for Environmental Research Information,
92
-------
Office of Research and Development, EPA/625/R-96/010b.
http://www.epa.gov/ttn/amtic/files/ambient/airtox/to-10ar.pdf
U.S. EPA (2007). EPA method 8082A, polychlorinated biphenyls (PCBs) by gas chromatography.
http://www.epa.gov/waste/hazard/testmethods/sw846/pdfs/8082a.pdf
U.S. EPA (2008a). Health effects of PCBs. http://www.epa.gov/epawaste/hazard/tsd/pcbs/pubs/effects.htm
U.S. EPA (2008b). EPA method 1668B, chlorinated biphenyl congeners in water, soil, sediment, biosolids
and tissue by high resolution gas chromatography/high resolution mass spectrometry (HRGC/HRMS).
http: //epa.gov/waterscience/methods/method/files/166 8 .pdf
U.S. EPA (2008c). Test methods for evaluating solid waste, physical/chemical methods, in EPA publication
SW-846, U.S. EPA, Government Printing Office: Washington, D.C.
U.S. EPA (2009). EPA news release — EPA announces guidance to communities on PCBs in caulk of
buildings constructed or renovated between 1950 and 1978 / EPA to gather latest science on PCBs in caulk.
http://yosemite.epa.gov/opa/admpress.nsf/6fa790d452bcd7f58525750100565efa/28c8384eeaOe67ed852576
3c0059342f! OpenDocument
Wilson, L. R., Palmer, P. M., Belanger, E. E., Cayo, M. R., Durocher, L. A., Hwang, S-A. A., and
Fitzgerald, E. F. (2011). Indoor air polychlorinated biphenyl concentrations in three communities along the
upper Hudson River, New York, Archives of Environmental Contamination and Toxicology, 61(3): 530-
538.
Xu, Y. and Zhang, Y. (2003). An improved mass transfer based model for analyzing VOC emissions from
building materials, Atmospheric Environment, 37: 2497-2505.
Zartarian, V., Glen, G., Smith, L. and Xue, J. (2008). Stochastic human exposure and dose simulation model
for multimedia, multipathway chemicals, SHEDS-multimedia model, Version 3 technical manual, U.S.
Environmental Protection Agency, EPA 600/R-08/118.
http://www.epa.gov/heasd/products/sheds_multimedia/sheds_mm.html
93
-------
Appendix A
Test Conditions for Caulk Samples and Determination of PCB Concentrations
Table A.I. Test conditions for PCB emissions from caulk at room temperature'"1
Caulk
ID
CK-01 w
CK-02 w
CK-03
CK-04
CK-05
CK-06
CK-07 w
CK-08
CK-09
CK-10 w
CK-11M
CK-12 w
CK-13 w
Airflow
rate (L/min)
0.449
0.455
0.455
0.454
0.481
0.452
0.428
0.450
0.444
0.441
0.472
0.446
0.459
0.473
0.473
0.451
0.451
0.492
0.498
Temperature
(°C)
22.2
22.2
22.2
22.2
22.2
22.2
22.2
22.2
22.2
22.2
22.2
22.2
22.8
22.8
21.2
21.2
21.2
21.8
21.8
Source area
(cm2)
3.63
3.26
3.12
3.4
2.25
2.85
1.46
3.79
2.96
3.31
0.21
2.34
2.97
3.56
6.64
6.66
6.45
13.2
13.2
[a] The ratio of air change rate to chamber loading factor ranged from 22 to 1350 m/h.
[b] This caulk sample was tested in duplicate.
94
-------
Table A.2. Test conditions for PCB emissions from caulk at different temperatures
Caulk
ID
CK-11
CK-13
Air flow
rate (L/min)
0.470
0.457
0.411
0.417
0.453
0.445
0.436
0.434
Temperature
(°C)
21.9
30.0
35.0
40.0
21.9
30.0
35.0
40.0
Source area
(cm2)
3.12
7.89
Table A.3. Test conditions for comparing the PCB emissions from different surfaces'
Caulk
ID
CK-01
CK-02
CK-12
Surface type
Previously exposed
Newly cut surface
Previously exposed
Newly cut surface
Previously exposed
Newly cut surface
Air flow
rate (L/min)
456
455
484
448
447
451
Source area
(cm2)
7.76
7.78
7.36
6.15
6.95
6.45
[a] The temperature was 22.6 °C for all six chambers.
95
-------
Table A.4. Average congener concentrations in chamber air, relative standard deviations, and number of valid data points W[bl
Caulk ID
CK-Ola
CK-Olb
CK-02a
CK-02b
CK-03
CK-04
CK-08
Mean (ug/m3)
RSD
n
Mean (ug/m3)
RSD
n
Mean (ug/m3)
RSD
n
Mean (ug/m3)
RSD
n
Mean (ug/m3)
RSD
n
Mean (ug/m3)
RSD
n
Mean (ug/m3)
RSD
n
#17
-
-
-
0.062
16.4%
4
0.141
3.4%
4
0.146
12.9%
4
-
-
-
-
-
-
-
-
-
#52
10.1
12.0%
5
10.9
12.4%
5
7.86
24.3%
4
8.63
14.5%
4
0.145
18.1%
4
1.11
28.5%
5
0 0230
22.9%
5
#66
0.791
21.9%
5
0.724
4.6%
4
0.651
19.1%
4
0.669
2.0%
3
0.0719
14.4%
4
0.214
24.6%
5
-
-
-
#101
-
-
-
-
-
-
3.42
20.7%
4
4.07
20.8%
4
0.950
12.6%
4
1.67
21.8%
5
0.500
16.9%
5
#105
5.01
13.0%
5
5.25
9.9%
5
0.238
26.2%
4
0.284
19.2%
4
0.176
25.4%
4
0.166
32.7%
5
-
-
-
#110
0.34
3.6%
4
0.34
6.6%
4
1.80
20.8%
4
2.15
21.9%
4
0.910
18.6%
4
1.19
26.4%
5
0.148
20.3%
5
#118
2.73
16.1%
5
2.78
10.5%
5
0.83
2.5%
o
5
1.09
18.6%
4
0.589
21.2%
4
0.599
31.6%
5
Q Q/l/l
20.2%
5
#154
1.37
4.9%
5
1.38
3.8%
4
0.329
20.2%
4
0.346
3.1%
3
0.101
12.0%
4
0.148
25.3%
5
0.267
19.0%
5
#187
0.457
0.0623
4
0.505
0.159
5
-
-
-
-
-
-
-
-
-
-
-
-
0.212
24.2%
5
[a] Values in strikethrough font are below the PQL; caulk CK-01, CK-02, C-10, CK-11 and CK-13 were tested in duplicate; results for CK-05, CK-06 and CK-07 were
not reported because all the conger concentrations were below the PQL.
[b] Outliers in air concentration measurements were discarded according to Dean and Dixon (1951) and Rorabacher (1991).
96
-------
Table A.4. Average congener concentrations in chamber air, relative standard deviations, and number of valid data points (continued)
Caulk ID
CK-09
CK-lOa
CK-lOb
CK-lla
CK-llb
CK-12
CK-13a
CK-13b
Mean (jig/m3)
RSD
n
Mean (ng/m3)
RSD
N
Mean (jig/m3)
RSD
N
Mean (jig/m3)
RSD
N
Mean (jig/m3)
RSD
N
Mean (ng/m3)
RSD
n
Mean (jig/m3)
RSD
n
Mean (jig/m3)
RSD
n
#17
-
-
-
0.279
8.1%
3
0.346
11.9%
2
..
..
..
..
..
..
0.692
4.2%
5
0.055
0.253
4
0.0613
13.7%
4
#52
4r09
25.2%
5
12.0
5.8%
5
16.5
9.4%
4
0.902
3.0%
5
0.818
3.9%
5
21.8
3.9%
5
0.983
24.6%
6
1.120
14.3%
5
#66
0.631
21.9%
5
0.909
14.5%
5
1.12
12.6%
4
0.084
11.6%
5
0.077
5.0%
5
1.8
9.8%
5
0.091
28.4%
6
0.112
12.2%
5
#101
5.03
21.2%
5
4.75214
10.4%
5
6.05
8.7%
4
0.463
8.7%
5
0.459
6.5%
5
8.79
4.6%
5
0.325
27.6%
6
0.373
18.5%
5
#105
0.630
18.3%
5
0.440
20.3%
3
0.445
32.9%
2
..
..
..
..
..
..
0.622
17.6%
5
-
-
-
-
-
-
#110
3.95
20.7%
5
2.56
13.4%
5
3.13
9.8%
4
0.222
10.8%
5
0.224
10.5%
5
4.37
8.4%
5
0.188
28.8%
6
0.214
19.0%
5
#118
2^30
14.3%
5
1.303
18.5%
5
1.47
17.1%
4
0.124
13.8%
5
0.125
13.0%
5
2.06
6.0%
5
0.087
32.4%
5
0.101
19.1%
4
#154
0.498
17.2%
5
0.454
12.2%
5
0.534
5.7%
4
..
..
..
..
..
..
0.865
8.1%
5
-
-
-
-
-
-
#187
Q Qgg
22.5%
5
-
-
-
-
-
-
..
..
..
..
..
..
0.030
20.2%
5
-
-
-
-
-
-
97
-------
Table A.5. Air concentrations at different temperatures for field caulk CK-ll[al
Temperature
CQ
21.9
30.0
35.0
Congener ID
#17
9.60-10^
2.72-10^
-W^W4
#52
6.20X10'1
1.95x10°
7.73x10°
#101
2.48 xlO'1
8.46 xlO'1
3.39x10°
#154
1.61X10'2
1.97-10^
346*46^
#110
1.32X10'1
4.44 xlO'1
1.88x10°
#66
5.39xlO'2
1.92X10'1
^4^^
#118
7.27 xlO'2
2.55 xlO'1
1.09x10°
#105
2.02-10^
7.13-10^
S^xiO-*
#187
0.00-10°
3.02-10^
±m*x^
Values in strikethrough font are below the PQL.
Table A.6. Air concentrations at different temperatures for laboratory-mix caulk CK-13[a
Temperature
CC)
21.9
30.0
35.0
40.0
Congener ID
#17
o
2
i
1.75-10^
#52
8.49X10"1
2.64x10°
3.87x10°
6.22x10°
#101
2.54X10"1
8.98X10"1
1.45x10°
2.45x10°
#154
2.42 xlO"2
7.30-10^
123-lQ-^
2.03-10^
#110
1.35X10"1
4.81X10"1
7.84X10"1
1.30x10°
#66
3.81xlO"2
1.44X10"1
2.20X10"1
3.51 vlff*
#118
6.11 xlO'2
2.33X10'1
3.99X10"1
6.55 xlO"1
#105
1.61^ Iff3
7.26-10^
i.ie^iff4
1.95-lff4
#187
3.27-lQ^
5.25-lQ^
2
LaJ Values in strikethrough font are below the PQL.
98
-------
Appendix B
Test Conditions for Light Ballasts
Table B.I. Summary of conditions for the screening tests
Ballast
ID
BL-01
BL-02
BL-03
BL-04
BL-06
BL-07
BL-08
BL-09a
BL-09b
BL-09d
BL-09e
BL-09f
BL-10
BL-llb
BL-12
BL-13
Duration
(hrs)
21.5
10.7
95.8
27.0
68.2
26.9
18.1
68.6
19.0
331
23.2
26.8
22.0
19.7
22.3
21.8
Avg. air flow
rate (L/min)
0.916
0.909
0.908
0.909
0.910
0.918
0.906
0.921
0.908
0.920
0.913
0.919
0.912
0.914
0.905
0.908
Avg. chamber
Temp. (°C)
23.3
23.3
23.8
23.7
23.6
23.9
23.3
23.8
23.5
23.8
23.3
23.4
23.8
23.3
23.8
23.6
Avg. chamber
RH (%)
44.6
41.3
47.8
53.1
52.9
46.5
41.9
43.1
41.6
33.7
44.1
43.9
45.0
41.0
47.6
52.0
Table B.2. Summary of conditions for the live tests
Ballast
ID
BL-08 a
BL-09a
BL-09c
BL-09d
BL-10
BL-lla
Duration (hrs)
19.0
25.2
24.5
26.2
24.6
25.6
Avg. air flow
rate (L/min)
0.903
0.895
0.917
0.901
0.905
0.910
Avg. chamber
Temp. (°C)
45.7
28.2
28.1
28.6
27.1
27.9
Avg. chamber
RH (%)
40.6
52.1
50.5
49.9
52.0
49.5
1 Ballast BL-08 failed during the test. See section 4.3.6 for details.
99
-------
Table B.3. Summary of test conditions for the effect of ambient temperature w
Ballast
ID
BL-02
BL-05
BL-09c
BL-12
Test Duration
(hrs)
18.5
17.5
17.1
19.6
17.0
16.7
18.0
65.8
17.4
17.4
16.7
18.0
65.6
17.4
17.3
17.7
17.6
17.1
19.6
16.9
Avg. air flow
rate (L/min)
0.904
0.893
0.909
0.910
Avg. chamber
Temp. (°C)
23.1
29.9
34.8
39.7
44.6
23.0
29.4
34.8
39.8
44.8
23.2
29.6
34.6
39.5
44.5
22.9
29.7
34.6
39.5
44.4
Avg. chamber
RH (%)
48.6
48.7
41.4
35.2
30.2
50.4
49.8
45.3
38.2
32.1
49.4
49.0
41.5
16.6
12.4
44.2
43.0
32.9
20.6
10.4
[a] Environmental data were recorded every 15 minutes.
100
-------
Table B.4. Congener emission rates for four light ballasts at different temperatures (units: ug/g)
Ballast
ID
Temperature
°C
Congener ID
#13
#18
#17
#15
#22
#52
#49
#44
#64
23.1
9.43 xlO'3
2.44X10'
29.9
2.67xlO"
8.53 xlO"
BL-02
34.8
2.88xlO'
8.97xlO'
39.7
9.10xKT
3. 00 xlO'2
7.88X10'
5.17X10'
5.53 xlO'
44.6
8.32xlO"
2.56xlO'2
7.25 xlO"
4.97xlO"
5. 47 xlO"
23.0
4.25
BL-05
29.4
3. 03 xlO'2
LOlxlO'
2.15X10'
2.75 xlO'
34.8
6.37xlO"2
1.93 xlO"
4.62xlO"
S.llxlO"
5.15xKT
2.28xlO"
. 03 xlO"
39.8
8.63 xlO'
2.55xlO'
6.89xlO'
7.17X10'
6.68xlO'
3.40xlO'
3.98xlO'
44.8
2.86xl(T3
1.14X10'1
3.80X10'
1.20X10'2
1.13X10'2
9.11X10'
4.49X10'
5.91xKT
23.2
5.53 xlO'
'3
29.6
1.17X10'
3.43xlO'
rv-3
•v4
BL-09c
34.5
2.70X10'
8.47X10'
4.33 xlO'
34.6
. 98 xlO"
1.22 xlO"2
6.92xlO"
2.65 xlO"
1 12- 10^
39.5
2.25 xlO'3
5.71xlO'
1.66xlO'
1.06X10'
3.03X10'
T4
-V4
44.5
4.93 xlO'3
8.56X10'
3.13X10'2
2.10X10'2
5.79X10'
3.62X10'
2.40X10'
22.9
1.52X10"1
5.19xlO"2
1 13- 10^
29.7
4.64X10'1
1.35X10'1
5.06X10'
rv-a
•V*
BL-12
34.6
7.48X10'1
2.30XKT1
8.45 xlO'2
39.5
1.46X10"1
3.83x10°
1.14x10°
5. 60 xlO"1
2.08X10"1
1.57X10"1
8.98xlO"2
9.47xlO"2
44.4
2.24X10'1
5.43x10°
1.72x10°
9.53
3.64X10'1
2.54X10'1
1.54X10'1
1.61X10'1
[a] Values in strikethrough font are below the practical quantification limit.
101
-------
Appendix C
Simulating the Long-term PCB Emissions from Caulk
C.I Model description
The emissions data presented in this report represent the emission status of the caulk samples as they were
tested. To estimate their emissions in the past or future, one must rely on mathematical models. Emissions
of volatile and semivolatile chemicals from solid building materials have been extensively studied in the
past two decades and, consequently, a series of mass transfer models have been developed (Little et al,
1994; Huang and Haghighat 2002; Xu and Yang, 2003; Deng and Kim, 2004; Qian et al., 2007). While
differing in complexity and applicability, they have several features in common:
• They are all derived from Pick's second law;
• They use the same set of parameters to describe the source: the content of the chemical in the source, the
solid/air partition coefficient, the diffusivity (i.e., diffusion coefficient) of the chemical in the source,
and the area and thickness of the source; and
• They, with one exception, all require solving a non-linear equation.
The model used here was developed by Little and his co-workers in 1994. The computer program that
implements this model was developed by Guo (2000).
C.2 Parameter Estimation
C.2.1 Estimation of the solid/air partition coefficient
The solid/air partition coefficient (K) is defined by Equation C.I:
C" (C.I)
where K = solid/air partition coefficient (dimensionless)
Cs = concentration of a congener in the solid phase ((ig/cm3)
Ca = concentration of the congener in the air in equilibrium with Cs ((ig/cm3)
Several empirical models are available for estimating K from the vapor pressure of the chemical. Equation
C.2 (Guo, 2002) is one of them:
102
-------
= 8.78- 0.785 InP (C2)
Where K = solid/air partition coefficient (dimensionless)
P = vapor pressure (torr)
C.2.2 Estimation of the diffusion coefficient for congener #52 in caulk
The diffusion coefficient for congener #52 in caulk was roughly estimated by using the mass transfer model
developed by Little et al. (1994), implemented in the SLAB program in IAQX (Guo, 2000). The original
code was modified to allow for calculation of residuals. The micro chamber data and the calculated partition
coefficient from Equation C.2 were used as the input of the model. The diffusion coefficient was estimated
by minimizing the residuals (i.e., least squares). The estimated diffusivity from four sets of micro chamber
data was 2.25x10-" (m2/h) or 6.25 x 10'15 (m2/s).
C.2.3 Estimation of the diffusion coefficient for other congeners in caulk
For a given class of chemicals (e.g., PCB congeners), the following correlation exists:
A l»'J (C3)
where DI and D2 = diffusivity in the solid source for chemicals 1 and 2 (m2/h)
nil and m2 = molecular weight for chemicals 1 and 2
For nonporous material, the index (n) ranges from 5.94 to 7.45 (Guo, 2002). An average of 6.63 was used
here.
C.3 Model input
The following parameters were used in the simulations. Congener specific parameters are given in Table
C.I.
• Room volume 300
• Air change rate 1
• Content of Aroclor 1254 in caulk 100000 (jig/g)
• Caulk area 0.5 (m2)
• Caulk thickness 0.0l(m)
• Caulk density 1.5 (g/cm3)
103
-------
Table C.I. Content in caulk, partition and diffusivity coefficients for four congeners in
Aroclor 1254
Parameter
Congener content in Aroclor 1254 M
Congener content in caulk
Partition coefficient (K)
Diffusion coefficient (D)
Units
%
ug/m3
—
m2/h
Congener ID
#52
5.38
8.07xl09
6.54xl07
2.25X10'11
#77
0.03
4.50xl07
4.13xl07
2.25 xlO'11
#154
0.04
6.00xl07
4.29xl07
5.52xlO'12
#187
0.25
3.75xl08
1.49xl08
3.02X10'12
[a] The percent weight data was from Frame et al. (1996).
C.4 Simulation results
Simulations were made for air concentrations and percent of congener mass emitted over a 50-year period.
The results are presented in Figures 5.1 and 5.2 in the main body of this report.
C5. Limitations
The partition and diffusion coefficients are two key parameters in this model. The values of these
parameters are rough estimates based on best available knowledge. Developing methods for accurately
determine these parameters is a key to improving the uncertainty in the model.
104
-------
Appendix D
Simulation of a Failed Light Ballast
D.I Purpose
This appendix describes how to use an existing spill model to roughly predict the concentrations of
individual congeners or total PCB concentrations following the rupture of the capacitor in a light ballast
unit.
D.2 Model Description
The spill model used for a PCB spill from a failed light ballast was originally developed for petroleum-
based solvents, which contain hundreds of hydrocarbons. The model uses the most abundant hydrocarbons
in the solvent to estimate the emissions of the total VOCs. The emissions from spilled PCB fluids are
similar. A full description of the model can be found in Guo (2000). The simulation program can be
downloaded from the EPA website http://www.epD.gov/nrmrl/appcd/mmd/iaq.html.
D.3 Assumptions
The simulation was based on the following assumptions:
• The PCB fluid in the light ballast is Aroclor 1242
• The volume of the fluid ejected during the failure is 20 mL
• After the burst, the fluid quickly either condenses or deposits on the nearby surfaces
• There is no human intervention after the failure
• The initial concentration surge due to high temperature is ignored
D.4 Model Input
The six most abundant congeners in Aroclor 1242 were used for the simulation. Their properties are
summarized in Table D.I. Other environmental parameters are as follows:
• Room volume 230 m3
• Ventilation rate 0.5 air change per hour
• Density of Aroclor 1242 1.35g/cm3
• Total spill area 0.04 m2
105
-------
Table D.I. Physical properties of the congeners used in the simulation
Congener ID
#8
#18
#28
#31
#33
#70
Chlorine
#
2
3
3
3
3
4
MW
199.1
233.6
233.6
233.6
233.6
268.0
pW
(torr)
1.19xl03
6.38xl04
2.43 xlO4
2.60xl04
2.21xl04
4.73 xlO4
Da[bl
(m2/h)
0.0199
0.0191
0.0191
0.0191
0.0191
0.0183
Content'01
(mg/g)
64.8
91.4
73.1
78.2
53.5
37.0
[a] Method B in Fischer et al. (1992).
w FSG method; calculated using program PARAMS (Quo, 2005).
w Congener content in Aroclor 1242 (Frame et al., 1996, Table 4A, column 7).
D.5 Simulation Results
The predicted time-concentration profile is shown in Figure D. 1. The peak concentration for total PCBs is in
agreement with the monitoring data reported by MacLeod (1981). The predicted concentrations remain high
for the simulation period because it was assumed that there are not clean-up measures.
o
B
%
a
o
centrat
=
o
-
0
n -
Total PCBs
#18
500 1000
Elapsed Time (h)
1500
2000
Figure D.I. Predicted concentrations of "total PCBs" and congener #18 following light ballast failure
106
-------
D.6 Limitations
This model was developed for evaporation of volatile organic compounds from a shallow pool of solvent
mixture. It has not been tested for PCBs. The simulation described above does not take into account of any
clean-up procedures.
To simulate the leakage of PCBs from a light ballast located inside of an enclosure, a two-compartment
model is needed, where the enclosure is considered a compartment because the enclosure is not air tight and
is designed to be convection cooled. The two-compartment spill model requires knowledge about the air
change rate between the enclosure and the room air.
107
------- |