EPA/600/R-11/156B
April 2012
Laboratory Study of Poly chlorinated Biphenyl (PCB)
Contamination and Mitigation in Buildings
Part 3. Evaluation of the Encapsulation Method
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
Nancy F. Roache, Rayford A. Stinson, Joshua A. Nardin, Robert H. Pope,
Corey A. Mocka, and Russell D. Logan
ARCADIS U.S., Inc.
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.
-------
Executive Summary
E.I Background
Encapsulation, one of the most commonly used abatement techniques for contamination in buildings,
involves painting the contaminated surfaces with a coating material or sealant that serves as a barrier to
prevent the release of a contaminant from the source, thereby improving the environmental quality in the
building. The practice of encapsulating polychlorinated biphenyl (PCB)-contaminated surfaces began in the
early 1970s and is still being used today. Although different levels of protective effects have been reported,
a number of questions remain regarding this mitigation method, including:
• To what extent can encapsulants provide protection from PCB contamination in buildings?
• How long does the protective effect last?
• What are the key attributes of a good encapsulant for PCBs?
• What are the key factors that affect the performance of the encapsulants?
• What are the limitations of the encapsulation method?
This study addresses some of these questions and the results should be useful to mitigation engineers,
building owners and managers, decision-makers, researchers, and the general public.
E.2 Objective
This study sought to develop a basic understanding of the encapsulation method for reducing PCB
concentrations in indoor air and contaminated surface materials and of the behavior of encapsulated sources.
The objectives of this study were to:
• Select and develop experimental methods to evaluate the abilities of selected coating materials to
encapsulate PCBs,
• Identify useful tools for studying the behavior of encapsulated sources and predicting the performances
of PCB encapsulants,
• Determine the key factors that affect the performance of the encapsulants, and
• Evaluate the effectiveness and limitations of the encapsulation method for reducing PCB concentrations
in indoor air and contaminated surface materials.
in
-------
E.3 Methods
E.3.1 Technical Approach
This study used a combination of laboratory testing and mathematical modeling to address some of the key
issues regarding the encapsulation method, and was comprised of three components: sink tests, wipe
sampling tests, and barrier modeling. The sink tests determined the sorption concentrations of PCBs. The
experimental results were used to (1) rank the encapsulants by their resistance to PCB sorption and (2)
estimate the partition and diffusion coefficients, two key parameters required by the barrier model. The wipe
sampling tests measured the PCB concentrations at the encapsulated surfaces and the results were used to
rank the encapsulants by their resistance to PCB migration from the source. The relationship between these
two experimental methods is discussed in Section 6.7. A barrier model was used to study the general
behavior of encapsulated sources and determine the key factors that may affect the performance of the
encapsulation.
E.3.2 Test Materials
Ten coating materials were selected for this study. They included coating types that had been used as PCB
encapsulants in the field, such as epoxy and polyurethane coatings, and several commonly used coating
materials, such as latex paint and petroleum-based paint.
E. 3.3 Sink Tests in Small Chambers
Sink tests were conducted in small environmental chambers, as illustrated in Figure E. 1. The source
chamber provided gas-phase PCBs emitted from building caulk. Coating materials (encapsulants) were
applied to stainless steel disks that measured 1.27 cm in diameter and the disks were cured in a fume hood.
For each encapsulant, 20 disks were placed in the test chamber (Figure E.2). The tests were conducted at 23
°C, 46% relative humidity, and one air change per hour. During the tests, four of the 20 disks were removed
from the source chamber at a given time, followed by subsequent removals of four disks at four different
times. This procedure was followed for all encapsulants tested, and the PCB concentrations associated with
the encapsulated disks were determined by extraction with hexane and analysis by gas
chromatography/mass spectrometry (GC/MS).
Source Chamber Test Chamber
1 Fan ' Fan
I vl
Sliced Caulk 71 Sink Materials
PUF PUF
Figure E.I. Schematic of the two-chamber system for sink tests
IV
-------
Figure E.2. Encapsulant disks in the test chamber
This test method provides a means of screening coating materials and determining their resistance to PCB
sorption. The test results, expressed as concentration of adsorbed PCBs, were used to rank the encapsulants.
In addition, the data were used to estimate two physical properties of the encapsulants that control the
movement of PCBs in the encapsulated sources, i.e., the material/air partition coefficient and solid-phase
diffusion coefficient, which are required for use in the barrier models (Section E.3.5).
E. 3.4 Wipe Sampling over Encapsulated Sources
Wipe sampling is the most commonly used sampling method for surface contamination. To test the
performance of coating materials as PCB encapsulants, 6 in x 3 in (15.2 cm x 7.6 cm) aluminum panels
were coated with an alkyd primer that contained 13000 ppm Aroclor 1254. The panels were then
encapsulated with ten types of coating materials. For each encapsulant, four panels were kept at room
temperature and without lighting, and four panels were placed under UV light and at 60 °C in an accelerated
weather chamber for two weeks. For each panel, wipe samples were collected three times over a three-
month period. The PCB concentrations in the wipe samples, indicators of the amounts of PCBs that had
migrated from the source through the layers of the encapsulants, were used to rank the performance of the
encapsulants.
E. 3.5 Using a Barrier Model
Barrier models are a group of mass transfer models developed in recent years for studying the behavior of
encapsulated sources. A fugacity-based, multi-layer model developed by Yuan et al. (2007) was used in this
study. The material/air partition coefficients and solid-phase diffusion coefficients estimated from the sink
tests were used as inputs to the model. The outputs from the model included the concentration profiles of
PCBs in the source and encapsulant layers as functions of time and depth and the contribution of the
encapsulated source to indoor air concentrations as a function of time. The modeling results allowed the
calculation of the average PCB concentrations in the layers of the encapsulant and the concentrations of
v
-------
PCB at the exposed surfaces of the encapsulant at different times. These concentrations were then used to
rank the performance of the encapsulants.
E.4 Findings
E. 4.1 Sink Tests
The experimentally-determined sorption concentrations for a water-borne acrylic coating material and an
epoxy coating material are shown in Figures E.3 and E.4, respectively. The sorption concentrations differed
by roughly a factor of 20 between the two coating materials, indicating that the epoxy coating is more
resistant to the sorption of PCBs than the water-borne acrylic coating.
10
E
u
4-1
oi
u
O
U
o
to
P
->
§
l->
0.001
--•
,-*
X
--•-- PCB-17
--•-- PCB-52
--&•- PCB-66
--X-- PCB-101
--*•- PCB-105
__0- PCB-110
---!-- PCB-118
--«-- PCB-154
100 200 300 400
Elapsed Time (h)Title
500
Figure E.3. Experimentally determined sorption concentrations of PCB congeners for a waterborne
acrylic coating
VI
-------
(N
U
-? 1 -
BO
'^
re
k.
•M n 1 -
s
o
(J
= n m .
1
o
in
n nm .
• •-""""""£• '""J
X- - — ~ "" ~r§ ~ - - - — d~ -- " " " . - - A
_ - - -O ~ _ _ ;fi- — -A" ^
£-::::"* Jar-- """*""
«• ^/R ^ ^
--.-
--A--
--X--
--/K--
1 —
- •»-
PCB-52
PCB-66
PCB-101
PCB-105
PCB-118
PCB-154
100 200 300
Elpased Time (h)
400
500
Figure E.4. Experimentally determined sorption concentrations of PCB congeners for an epoxy
coating
(The concentrations of congener #17 were below the practical quantification limit)
The experimentally-determined sorption concentrations can be used directly to rank the encapsulants. As
shown in Figure E.5, the three epoxy coatings performed better than the rest of the coating materials.
ST 25
u
20 -
is H
CD
o
u
= 5-1
Q.
o 0
to
^
Aroclor 1254
n n „
X *
»y
Figure E.5. Calculated sorption concentrations of Aroclor 1254 for the ten encapsulants
(Aroclor concentrations were calculated base on five predominant congeners; t = 433 h)
vn
-------
The sorption concentrations were used to obtain rough estimates of the material/air partition coefficients and
solid-phase diffusion coefficients for the ten encapsulants. Table E.I summarizes the results for congener
#52. These two types of coefficients are the key properties of the encapsulants that affect their
performances. The combined effect of the partition and diffusion coefficients can be represented by the sink
sorption index, or SSI (Equation E.I; from Guo, et al, 2012). As shown in Figure E.6, the SSIs correlate
well with the sorption concentrations.
SSI = -log(KmaDm) (E.I)
;
where Kma = material/air partition coefficient (dimensionless)
Dm = solid-phase diffusion coefficient (m2/h)
Table E.I. Estimated material/air partition coefficients and solid-phase diffusion coefficients for
congener #52 for the 10 coating materials that were tested w
Encapsulant
ID
01
02
03
04
05
06
07
08
09
10
Name
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-low VOC
Epoxy-no solvent
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyurethane
K^ (Dimensionless)
Mean
1.93xl07
2.05xl07
1.34xl07
3.05xl06
1.78xl06
2.02xl06
7.90xl06
1.62xl07
1.78xl07
5.93xl06
RSD
29.4%
19.6%
35.9%
57.2%
62.3%
67.1%
20.6%
35.7%
9.5%
8.8%
Dm(m2/h)
Mean
4.88xlO"10
1.75xlO-10
2.06xlO"10
3.76X10'11
1.89xlO"12
8.36xlO'12
9.50x10""
4.53X10'10
1.34xlO"09
1.12X10'10
RSD
58.6%
36.0%
73.8%
72.2%
92.0%
75.9%
45.3%
60.1%
18.0%
19.4%
SSI
2.0
2.4
2.6
3.9
5.5
4.8
3.1
2.1
1.6
3.2
[a] Methods for calculating the partition and diffusion coefficients for other congeners are described in Section 4.1.3.
vin
-------
0.001
1.00 2.00 3.00 4.00
SSI
5.00
6.00
Figure E.6. Correlation of sink sorption indices (SSIs) and the experimentally-determined sorption
concentrations for congener #52 at time (t) = 433 h for the ten encapsulants
E.4.2 Wipe Sampling Tests
Although the wipe sampling tests were totally different from the sink tests in terms of mass transfer
mechanisms, the two methods yielded rather similar results for the performances of the ten coating materials
that were tested. As shown in Figure E.7, the three epoxy coatings performed well. The two methods
showed very different results for the polyurea elastomer, however. This coating performed poorly in the sink
test (Figure E.5) but performed well in the wipe sampling tests (Figure E.7). One factor that may have
partially contributed to this inconsistency is the thickness of the encapsulant. The polyurea elastomer had the
thickest film in the wipe sampling tests. As discussed in Sections 4.2.6 and 5.3.6 in the main body of this
report, the PCB surface-wipe concentration at the encapsulated surface decreases as the thickness of the
encapsulant increases.
E. 4.3 Barrier Modeling
Barrier models are a group of mass transfer models that compute the concentration profiles of PCBs in the
source and the encapsulant as functions of depth and time (Figures E.8 and E.9). These models also compute
the contribution of the encapsulated source to PCB concentrations in indoor air (Figure E.10).
IX
-------
800
*
*
-------
Concrete/ Lacquer-primer
1 day
10 days
100 days
1000 days
5000 days
0.00 0.02 0.04 0.06 0.08 0.10
x(mm)
Figure E.9. Concentration profiles for congener #110 in the layer of encapsulant (lacquer primer) as
a function of depth [C2(x)j
(The interface between the source and the encapsulant is at x = 0 mm; the exposed surface is
at x = 0.1 mm; the initial concentration in the source (C0i) is 100 ug/g)
1.2
^ 0.9
£
O
S 0.6
4-1
0)
u
§ 0.3
0.0
Lacquer primer
Epoxy-waterborne
1000 2000 3000 4000 5000
Elapsed Time (days)
Figure E.10. Concentration of congener #110 in room air due to emissions from the encapsulated
sources as function of time
(The initial concentration in the source is 100 ug/g; the source area is 10 m2)
XI
-------
The modeling results showed that, for a given PCB source and encapsulant pair, a linear correlation exists
between the initial concentration in the source and the average concentration in the encapsulant (Figure
E. 11), indicating the limitations of the encapsulation method: (1) Encapsulation is not effective for reducing
surface concentrations and indoor air levels for sources that have high PCB content, and (2) The upper limit
of the PCB content in the source for which encapsulation would be effective is determined by the
performance of the encapsulant and mitigation goal. The more stringent the goal is, the lower the
concentration in the source is allowed. More details are provided in Section 6.1.
400
•55 30°
(S 200
0)
oo
S
a; 100
•Lacquer primer
•Epoxy-waterborne
200 400 600 800 1000
Initial Concentration in Source (u.g/g)
1200
Figure E.ll. Average concentration in the layer of encapsulant (average C2) as a function of initial
concentration in the source (t = 100 days)
Using the partition coefficients and diffusion coefficients obtained from the sink tests, the relative
performance of the ten encapsulants can be ranked. As an example, Figure E.12 compares the predicted
PCB concentrations at the exposed surface when a source is encapsulated with different encapsulants.
xn
-------
//•/
'////,
Figure E.12. Ranking of encapsulants by the PCB concentration at the exposed surface in the layer of
the encapsulant [C2(Surface)]
(For congener #110;t = 500 days; initial concentration in source = 100 (ig/g)
E. 4.4 Effectiveness of the Encapsulation Method
Both the experimental results and the mathematical modeling showed that selecting proper encapsulants can
effectively reduce the PCB concentrations at the exposed surfaces. However, the encapsulation method has
its limitations. To estimate the upper limit of the PCB concentration in the source for encapsulation, several
factors must be considered, including the mitigation goals, the properties of the source, the properties of the
encapsulant, and the environmental conditions.
Results from the wipe sampling tests showed that, when the source contained approximately 13000 ppm
PCBs, the PCB concentrations in the wipe samples collected from encapsulated panels ranged from 10.1 to
584 ug/100 cm2, depending on the encapsulant used. If the mitigation goal is to keep the PCB concentration
in the wipe samples below 1 ug/100 cm2, the PCB concentration in the source cannot be higher than 1287
ppm even with the best encapsulant tested. Furthermore, if a safety factor of 3 is considered, the PCB
concentration in the source must be below 430 ppm for successful encapsulation.
E. 4.5 Summary of Major Findings
The major findings of this study are as follows:
• Encapsulation can be used as an interim solution to mitigating PCB contamination in buildings.
• The encapsulation method is most effective for contaminated surfaces that contain low levels of PCBs.
xni
-------
• As demonstrated in Part 2 of this report series, the secondary sources may become emitting sources of
PCBs after the primary sources are removed. Because of their large quantities, mitigating secondary
sources is difficult and costly. The encapsulation method has the potential to substantially reduce the
cost by not having to remove the contaminated materials from the building.
• Selecting high-performance coating materials is a key to effective encapsulation. Multiple layers of
coatings enhance the performance of the encapsulation. Post-encapsulation inspection and monitoring is
essential for successful encapsulation.
• For effective encapsulation, the maximum allowable concentration of PCBs in the source is estimated to
be 430 ppm, assuming (1) the maximum allowable PCB concentration in the wipe sample is 1 ug/100
cm2, (2) the most effective encapsulant we tested is used, and (3) the safety factor is 3.
• Encapsulating primary sources, such as old caulk, that contain high concentrations of PCBs can be
beneficial, but may not be sufficient to reduce the surface and air concentrations to desirable levels.
• The experimental methods developed in this report can be used to screen more coating materials.
E.5 Study Limitations
This study was limited to laboratory testing with a limited scope. Only ten coating materials were tested.
There are many coating materials that can potentially be used as PCB encapsulants. The test results of this
study may not be applicable to the similar products that were not tested even within the same class of
coatings.
This study was narrowly focused on the effectiveness and limitations of the encapsulation method, the
performances of a limited number of encapsulants, and the factors that may affect the performance of
encapsulation. It is not a comprehensive evaluation of the encapsulation method, which involves multiple
steps.
This study investigated liquid encapsulants only. Encapsulation by using solid materials was not studied. In
practice, multiple coating materials are often used (such as using a primer before applying the encapsulant),
which were not tested in this study.
Part of the wipe samples were analyzed by a commercial analytical laboratory. The results did not meet all
the data quality criteria to qualify as quantitative data. The accuracy and precision of the data were in the
range of 25% to 50%. Thus, the data generated by the commercial laboratory should be considered semi-
quantitative.
The material/air partition coefficients and solid-phase diffusion coefficients reported are rough estimates.
For more accurate measurements, the two parameters must be determined separately.
The correlation of the PCB concentration in the surface material with the concentration in the wipe samples
is poorly understood. This data gap makes it difficult to link the wipe sampling results to the barrier models.
xiv
-------
TABLE OF CONTENTS
Executive Summary iii
E.I Background iii
E.2 Objective iii
E.3 Methods iv
E.3.1 Technical Approach iv
E.3.2 Test Materials iv
E.3.3 Sink Tests in Small Chambers iv
E.3.4 Wipe Sampling over Encapsulated Sources v
E.3.5 Using a Barrier Model v
E.4 Findings vi
E.4.1 Sink Tests vi
E.4.2 Wipe Sampling Tests ix
E.4.3 Barrier Modeling ix
E.4.4 Effectiveness of the Encapsulation Method xiii
E.4.5 Summary of Major Findings xiii
E.5 Study Limitations xiv
List of Tables xviii
List of Figures xx
Acronyms and Abbreviations xxiv
1. Introduction 1
1.1 Background 1
1.2 Goals and Objectives 2
1.3 Technical Approach 2
1.4 About This Report 3
2. Experimental Methods 4
2.1 Test Specimens 4
2.2 Sink Tests in 53-L Environmental Chambers 4
2.2.1 Test Facility 4
2.2.2 Preparation of the Coating Materials 7
2.2.3 Test Procedure 9
2.3 Wipe Sampling over Encapsulated Sources 9
2.3.1 Preparation of Source Panels 9
xv
-------
2.3.2 Application of Encapsulant 10
2.3.3 Aging at Room Temperature 11
2.3.4 Accelerated Aging 12
2.4 Sampling and Analysis 13
2.4.1 Internal Standards and Recovery Check Standards 13
2.4.2 Air Sampling 13
2.4.3 Extraction of Encapsulant-Coated Disks 13
2.4.4 Wipe Sampling 14
2.4.5 Sample Analysis 15
3. Quality Assurance and Quality Control 16
3.1 QA/QC forthe In-house Analytical Laboratory 16
3.1.1 GC/MS Instrument Calibration 16
3.1.2 Detection Limits 17
3.1.3 Environmental Parameters 18
3.1.4 Quality Control Samples 19
3.1.5 Recovery Check Standards 22
3.2 QA/QC for Using a Commercial Analytical Laboratory 22
3.2.1 QA/QC Procedure 23
3.2.2 Data Quality Indicators (DQIs) 23
3.2.3 Data Quality Evaluation 23
3.2.4 Conclusion Related to Data Quality Review 25
4. Experimental Results 26
4.1 Sink Tests 26
4.1.1 Test Conditions 26
4.1.2 Experimentally Determined Sorption Concentrations 28
4.1.3 Estimation of the Partition and Diffusion Coefficients 32
4.2 Wipe Sampling over Encapsulated Sources 35
4.2.1 PCB Concentrations in the Source 35
4.2.2 Thicknesses of the Dry Films of the Encapsulants 35
4.2.3 Wipe Samples for Not-encapsulated Sources That Had Undergone Aging at Room
Temperature 36
4.2.4 Encapsulated Sources That Had Undergone Ageing at Room Temperature 36
4.2.5 Encapsulated Sources That Had Undergone Accelerated Aging 40
4.2.6 Additional Wipe Sampling Tests 43
5. Mathematical Modeling 48
5.1 Model Description 48
xvi
-------
5.1.1 Available Barrier Models 48
5.1.2 The Concept of Fugacity 48
5.1.3 The Fugacity-Based Barrier Model 49
5.2 Input Parameters 51
5.2.1 Parameters Required by the Model 51
5.2.2 Parameter Values forthe "Base-case" Scenario 51
5.3 General Behavior of Encapsulated Sources 52
5.3.1 Concentration Profiles in the Source 52
5.3.2 Concentration Profiles in the Encapsulant Layer 54
5.3.3 Average Concentration in the Encapsulant Layer 55
5.3.4 Concentration atthe Exposed Surface 56
5.3.5 Contribution to PCB Concentrations in Room Air 56
5.3.6 Effect ofthe Thickness ofthe Encapsulant 56
5.3.7 Effect of Contaminant Concentration in the Source 61
5.4 Ranking the Encapsulants 64
5.4.1 Performance Indicators 64
5.4.2 Input Parameters forthe Barrier Model 64
5.4.3 Ranking the Encapsulants Based on Absolute Concentrations 65
5.4.4 Ranking the Encapsulants Based on Percent Reduction of Concentrations 67
5.5 Limitations of Mathematical Modeling 69
6. Discussion 70
6.1 Effectiveness and Limitations ofthe Encapsulation Method 70
6.2 Selection of Encapsulants 71
6.3 Potential Effect ofthe Weathering of Encapsulants on their Encapsulating Ability 72
6.4 Encapsulating Encapsulated Sources 73
6.5 Effectiveness of Encapsulating Sources with High PCB Content 73
6.6 Relationship between the Sink Tests and the Wipe Sampling Tests 73
6.7 Study Limitations 74
7. Conclusions 75
8. Recommendations 77
Acknowledgments 78
References 79
Appendix A. Evaluation ofthe Wipe Sampling Method 82
Appendix B. Resistance ofthe Encapsulants to Abrasion 84
XVII
-------
List of Tables
Table E.I. Estimated material/air partition coefficients and solid-phase diffusion coefficients for
congener #52 for the 10 coating materials that were tested viii
Table 2.1. Coating materials tested (product names, binder types, recommended uses, and
recommended application methods) 5
Table 2.2. Coating materials tested (principal solvents, VOC content, solid content, and
recommended application rates) 6
Table 2.3. Application methods and number of coats for the encapsulated panels 11
Table 3.1. GC/MS calibration for PCB congeners from Aroclor 1254 17
Table 3.2. IAP results for each calibration related to this study 18
Table 3.3. Instrument detection limits (IDLs) for PCB congeners on GC/MS (ng/mL) 18
Table 3.4. Background concentrations of PCBs (|ig/m3) in the chamber for the sink test 19
Table 3.5. Concentration of PCBs in the field blank samples (ng) 20
Table 3.6. Concentration of PCBs in the method blank samples (fig/cm2 wipe sample) for the
accelerated weathering process 21
Table 3.7. Concentration of PCBs in the method blank samples ((ig/cm2 wipe sample) for panels
in the storage cabinet 21
Table 3.8. Average recoveries of DCCs for the tests of the encapsulants 22
Table 3.9. Criteria for determining the usability of data reported by the commercial laboratory 23
Table 3.10. QC samples for evaluating the accuracy of the analytical results reported by the
commercial laboratory 24
Table 3.11. QC samples for evaluating the precision of the analytical results reported by the
commercial laboratory 24
Table 3.12. QC samples for evaluating the potential contamination in the laboratory and during
transportation of samples 24
Table 3.13. Recovery of the recovery check standards (RCSs) for all wipe samples analyzed by the
commercial laboratory 25
Table 4.1. Conditions of the test chamber 26
Table 4.2. Thicknesses of the dry films of the encapsulants used for the sink test 26
Table 4.3. Estimated partition coefficient (Kma), diffusion coefficient (Dm) for the reference
congener (#52) and index a in Equation 4.2 34
Table 4.4. Concentrations of target congeners in three batches of dry primer 35
Table 4.5. Thickness of dry films of the encapsulants for the wipe sampling tests 36
Table 4.6. Concentrations of Aroclor 1254 in wipe samples taken from not-encapsulated source
panels 36
Table 4.7 Percent reduction of PCB concentrations in wipe samples for encapsulated PCB
sources 40
xvm
-------
Table 4.8. PCB concentrations in cured substrates 46
Table 5.1. Input parameters for the fugacity model 51
Table 5.2. Base-case values for the simulations 52
Table 5.3. Partition coefficients (Kma) and diffusion coefficients (Dm) for congener #110 for the
source and encapsulants 52
Table 5.4. Material/air partition coefficients (Kma) and solid-phase diffusion coefficients (Dm) for
congener #110 used for ranking the encapsulants 64
Table 5.5. Ranking the encapsulants by percent reduction of the average concentration in the top
0.1 mm of the layer, i.e., the thickness of the encapsulant 68
Table 5.6. Ranking the encapsulants by percent reduction of the concentration at the exposed
surface 68
Table 5.7. Ranking the encapsulants by percent reduction of the concentration in room air 69
Table 6.1. Calculated maximum allowable concentrations in the source for effective encapsulation
with two mitigation goals based on the PCB concentration in wipe samples (Wmax) 72
xix
-------
List of Figures
Figure E. 1. Schematic of the two-chamber system for sink tests v
Figure E.2. Encapsulant disks in the test chamber v
Figure E.3. Experimentally determined sorption concentrations of PCB congeners for a waterborne
acrylic coating vi
Figure E.4. Experimentally determined sorption concentrations of PCB congeners for an epoxy
coating vii
Figure E.5. Calculated sorption concentrations of Aroclor 1254 for the ten encapsulants vii
Figure E.6. Correlation of sink sorption indices (SSIs) and the experimentally-determined sorption
concentrations for congener #52 at time (t) = 433 h for the ten encapsulants ix
Figure E.I. Concentration of Aroclor 1254 in the first round wipe samples taken over encapsulated
PCB panels that underwent aging at room temperature x
Figure E.8. Concentration profiles for congener #110 in the source encapsulated with a lacquer
primer [Ci(x)] x
Figure E.9. Concentration profiles for congener #110 in the layer of encapsulant (lacquer primer)
as a function of depth [C2(x)J xi
Figure E. 10. Concentration of congener #110 in room air due to emissions from the encapsulated
source as function of time xi
Figure E. 11. Average concentration in the layer of encapsulant (average C2) as a function of initial
concentration in the source (t = 100 days) xii
Figure E. 12. Ranking of encapsulants by the PCB concentration at the exposed surface in the layer
of the encapsulant [C2(Surface)] xiii
Figure 2.1. Schematic of the two-chamber system for sink tests, showing the air flows and
sampling locations (PUF samples) 7
Figure 2.2. Chamber system for sink tests: source chamber (top) and test chamber (bottom) 7
Figure 2.3. Painted stainless steel panel after the disks were punched out 8
Figure 2.4. Sample stage with aluminum pin mounts 8
Figure 2.5. Sample stages in the test chamber 9
Figure 2.6. Taped panel (left); panel after primer was applied (right) 10
Figure 2.7. Re-taped panel with dried primer (left); panel after application of the encapsulant
(right) 10
Figure 2.8. Plastic cabinets (left); panels inside a plastic cabinet (right) 11
Figure 2.9. Wooden cabinets that housed the plastic cabinets 12
Figure 2.10. QUV Accelerated Weathering Tester: exterior (left); UV lamps (right) 12
Figure 2.11. Wipe sampling process: wipe on panel (left); wipe covered with foil (center); roller
method (right) 14
Figure 4.1. Concentrations of the target congeners in the air of the test chamber (normal scale) 27
xx
-------
Figure 4.2. Concentrations of the target congeners in the air of the test chamber (semi-log scale) 27
Figure 4.3. Experimentally-determined sorption concentrations as a function of time for the
Acrylate-waterborne coating material (top: normal scale; bottom: semi-log scale). 29
Figure 4.4. Experimentally-determined sorption concentrations as a function of time for the
Epoxy-low VOC coating material (top: normal scale; bottom: semi-log scale). 30
Figure 4.5. Experimentally-determined sorption concentrations for congener #52 for the ten
encapsulants (t = 433 h) 31
Figure 4.6. Experimentally-determined sorption concentrations for congener #110 for the ten
encapsulants (t = 433 h) 31
Figure 4.7. Calculated sorption concentrations for Aroclor 1254 for the ten encapsulants (t = 433
h) 32
Figure 4.8. Goodness-of-fit for estimating the partition and diffusion coefficients for the Acrylic-
solvent coating material 33
Figure 4.9. Concentration of Aroclor 1254 in the first-round wipe samples taken over encapsulated
PCB panels that had undergone aging at room temperature (error bar = ±1 SD) 37
Figure 4.10. Concentrations of Aroclor 1254 in the second-round wipe samples taken over
encapsulated PCB panels that had undergone aging room temperature (error bar = ±1
SD) 38
Figure 4.11. Concentrations of Aroclor 1254 in the third-round wipe samples taken over
encapsulated PCB panels that had undergone aging at room temperature (error bar = ±1
SD) 38
Figure 4.12. Concentrations of Aroclor 1254 for the sum of three rounds of wipe samples taken over
encapsulated PCB panels that had undergone aging at room temperature 39
Figure 4.13. Concentration of Aroclor 1254 in the first-round wipe samples taken over encapsulated
PCB panels that had undergone accelerated aging (error bar = ± 1 SD) 41
Figure 4.14. Concentration of Aroclor 1254 in the second-round wipe samples taken over
encapsulated PCB panels that had undergone accelerated aging (error bar = ±1 SD) 41
Figure 4.15. Concentration of Aroclor 1254 in the third-round wipe samples taken over
encapsulated PCB panels that had undergone accelerated aging (error bar = ±1 SD) 42
Figure 4.16. Concentrations of Aroclor 1254 for the sum of three rounds of wipe samples taken over
encapsulated PCB panels that had undergone accelerated aging 42
Figure 4.17. Comparison of wipe sampling results (the sum of three wipes) for the two aging
methods 43
Figure 4.18. Concentrations of target congeners in wipe samples taken at 167 elapsed hours 44
Figure 4.19. Concentrations of target congeners in wipe samples taken at 692 elapsed hours 44
Figure 4.20. Concentrations of target congeners in wipe samples taken at 1245 elapsed hours. 45
Figure 4.21. Concentrations of Aroclor 1254 in three wipe samples taken at 167, 692, and 1245
elapsed hours 45
Figure 4.22. Effect of source substrate on PCB concentrations in wipe samples — the sources
(primer and caulk) were encapsulated with the Lacquer-primer (error bar = ±1 SD) 47
xxi
-------
Figure 4.23. Effect of source substrate on PCB concentrations in wipe samples — the sources
(primer and caulk) were encapsulated with Polyurethane (error bar = ±1 SD) 47
Figure 5.1. Schematic representation of the double-layer model (Yuan et al., 2007) 49
Figure 5.2. Concentration profiles for congener # 110 in the source encapsulated with a Lacquer-
primer [Ci(x)]. 53
Figure 5.3. Concentration profiles for congener # 110 in the source encapsulated with a waterborne
epoxy coating [Q(x)] 53
Figure 5.4. Concentration profiles for congener # 110 in the encapsulant layer (Lacquer primer) as
a function of depth 5 4
Figure 5.5. Concentration profiles for congener # 110 in the encapsulant layer (Epoxy-waterborne)
as a function of depth 55
Figure 5.6. The average concentration of congener #110 in the encapsulant layer (C2) as a function
of time 56
Figure 5.7. Concentration of congener #110 at the exposed surface of the encapsulant [C2(x=L2)]
as a function of time 57
Figure 5.8. Concentration of congener # 110 in room air due to emissions from the encapsulated
source as a function of time 57
Figure 5.9. Effect of the thickness of the encapsulant on the average concentration of congener
#110 in the encapsulant layer (average C2) — Case 1: Lacquer primer 58
Figure 5.10. Effect of the thickness of the encapsulant on the average concentration of congener
# 110 in the encapsulant layer (average C2) — Case 2: Epoxy-waterborne 5 8
Figure 5.11. Effect of the thickness of the encapsulant on the concentration of congener #110 at the
exposed surface of the encapsulant [C2(x=L2)] — Case 1: Lacquer-primer 59
Figure 5.12. Effect of the thickness of the encapsulant on the concentration of congener #110 at the
exposed surface of the encapsulant [C2(x=L2)J — Case 2: Epoxy-waterborne 59
Figure 5.13. Effect of encapsulant thickness on the concentration of congener # 110 in room air due
to emissions from the encapsulated source — Case 1: Lacquer-primer 60
Figure 5.14. Effect of encapsulant thickness on the concentration of congener #110 in room air due
to emissions from the encapsulated source — Case 2: Epoxy-waterborne 60
Figure 5.15. Average concentration of congener #110 in the encapsulant layer (average C2) as a
function of initial concentration in the source (t = 100 days) 61
Figure 5.16. Average concentration of congener #110 in the encapsulant layer (average C2) as a
function of initial concentration in the source (t = 1000 days) 61
Figure 5.17. Concentration of congener # 110 at the exposed surface of the encapsulant layer [C2(x =
L2)] as a function of initial concentration in the source (t = 100 days) 62
Figure 5.18. Concentration of congener #110 at the exposed surface of the encapsulant layer [C2(x =
L2)] as a function of initial concentration in the source (t = 1000 days) 62
Figure 5.19. Contribution of the encapsulated source to the concentration of congener #110 in room
air as a function of initial concentration in the source (t = 100 days) 63
Figure 5.20. Contribution of the encapsulated source to the concentration of congener #110 in room
air as a function of initial concentration in the source (t = 1000 days) 63
xxn
-------
Figure 5.21. Ranking of encapsulants by the average concentration in the encapsulant layer
(Average C2) 65
Figure 5.22. Ranking of encapsulants by the concentration at the exposed surface of the encapsulant
layer [C2 (x=L2)] 66
Figure 5.23. Ranking of encapsulants by the air concentration due to emissions from the
encapsulated source 66
Figure 5.24. Concentration profiles for congener #110 in not-encapsulated concrete at t = 500 days 67
xxin
-------
Acronyms and Abbreviations
DAS Data acquisition system
DCC Daily calibration check
DQI Data quality indicator
GC/ECD Gas chromatography/electron capture detector
GC/MS Gas chromatography/mass spectrometry
IAP Internal audit program
IDL Instrument detection limit
LC Laboratory control
ND Not detected
NELAP National Environmental Laboratory Approval Program
NERL National Exposure Research Laboratory
ORD Office of Research and Development
PCB Polychlorinated biphenyl
PQL Practical quantification limit
PUF Polyurethane foam
QA Quality assurance
QAPP Quality Assurance Project Plan
QC Quality control
RCS Recovery check standard
RH Relative humidity
RRF Relative response factor
RSD Relative standard deviation
SD Standard deviation
TCMX Tetrachloro-ra-xylene or tetrachlorometaxylene
UV Ultraviolet
VOC Volatile organic compound
xxiv
-------
1. Introduction
1.1 Background
Creating a barrier between the source of contaminants and the surrounding environment is one of the
common abatement techniques for contamination in structures and buildings (Esposito et al., 1987). The
encapsulating barriers may take different forms such as painting and coating, plaster, concrete casts and
walls. Painting and coating techniques are the most common forms used inside buildings. Encapsulation has
been used successfully for decontaminating asbestos (Brown, 1990; Brown and Angelopoulos, 1991;
ASTM, 2010a), lead paint (ASTM, 2004a, 2004b, and 2011), and methamphetamine (Martyny, 2008) in
buildings.
Encapsulating structures and buildings contaminated with polychlorinated biphenyls (PCBs) began in the
early 1970s (Willett, 1972, 1973, 1974, 1976) and has been used since then (Mitchell and Scadden, 2001;
Scadden and Mitchell, 2001; Pizarro et al., 2002; EH&E, 2012). Willett (1974) tested the feasibility of
twelve coating materials as PCB encapsulants for the interior of concrete silos coated with a PCB-containing
material. The author found that nine of the twelve coatings reduced the concentration of PCBs in the silage
compared to silage adjacent to control surfaces. Coatings carried by water or a solvent in which PCBs are
not readily soluble were the most effective barriers. Coating materials with a base coat to seal the surface
prior to application of the surface coating were more effective than two applications of a single formulation.
Hydraulic cement with an acrylic bonder and water-carried epoxy effectively reduced residues in silage
from contaminated dairy farm silos.
Mitchell and Scadden (2001) indicated that important properties to consider when choosing an encapsulant
include elongation (i.e., elasticity or rigidity), dry film thickness, hardness, drying or curing time, and
compatibility with existing surfaces. Epoxy-type coatings are widely used for PCB encapsulation. Epoxy
coatings generally consist of a three-part epoxy-polyamide coating applied in a primer layer, clad leveler,
and surface layer. Encapsulants applied to floors should include two coatings of contrasting color to indicate
when resurfacing is required due to wear. Such practice is also routinely applied on exterior walls.
Scadden and Mitchell (2001) reported a case study in which the PCB-contaminated floors were cleaned by
multi-step surface washing and then encapsulated with two coats of a high-solid, water-based epoxy with
contrasting colors. The authors recommended that a penetrating primer be used to help seal the concrete
surface before applying the epoxy. They also noted that compliance with the manufacturer's epoxy mixing
instructions is critical. Failure to follow the manufacturer's instructions when mixing the activator
compound in the epoxy can cause a reduction in epoxy strength and result in undesirable soft spots and
cracking. The epoxy application also needs to be performed under optimum environmental conditions (dry
with stable temperatures) to get the best results. For floors, anti-slip materials may need to be included as
part of the epoxy topcoat or placed on top of the final epoxy surface to reduce the slip hazard created by the
smooth epoxy finish.
Pizarro et al. (2002) investigated encapsulation via cleaning and epoxy-coating of PCB-contaminated
concrete samples from industrial plants. They concluded that epoxy coatings can be an appropriate
-------
encapsulation system if the surface is prepared properly and the temperature in the area is not too high. They
also concluded that metal sheet barriers could be used for high-temperature applications.
A recent literature review on PCB remediation methods (EH&E, 2012) summarized the most recent
developments in applying the encapsulation method to buildings contaminated with PCBs. There are
commercially available coating materials that are designed specifically for encapsulating PCBs (TWO
Teknik, 2011; Robnor Resins, undated; MIC, undated).
Despite the long history of encapsulating PCBs and different levels of success reported by researchers, a
number of questions regarding this mitigation method remain, including:
• To what extent can encapsulants provide protection from PCB contamination in buildings?
• How long does the protective effect last?
• What are the key attributes of a good encapsulant for PCBs?
• What are the key factors that affect the performance of the encapsulants?
• What are the limitations of the encapsulation method?
This study answers some of these questions by using a combination of laboratory testing and mathematical
modeling. The results should be useful to mitigation engineers, building owners and managers, decision-
makers, researchers, and the general public.
1.2 Goals and Objectives
In this study, we sought to develop a basic understanding of the encapsulation method for PCB-
contaminated surface materials and the behavior of encapsulated sources. The objectives were to (1) select
or develop experimental methods to evaluate the abilities of selected coating materials to encapsulate PCBs,
(2) identify useful tools for studying the behavior of encapsulated sources and predicting the performance of
PCB encapsulants, (3) determine the factors that affect the performance of the encapsulants, and (4) evaluate
the effectiveness and limitations of the encapsulation method for PCB sources in buildings.
1.3 Technical Approach
A combination of laboratory testing and mathematical modeling was used to evaluate the encapsulation
method. The approach involved the use of three interrelated components, i.e., sink tests, wipe sampling tests,
and mathematical modeling.
The sink tests compared the sorption of PCBs from air by different encapsulants. The test results, expressed
as sorption concentrations, were used to rank the encapsulants based on their resistance to PCB sorption.
More importantly, the test results were used to estimate the solid/air partition coefficients and the solid-
phase diffusion coefficients for the encapsulants, two key parameters that affect the performance of the
encapsulants. An encapsulant that has smaller partition coefficient and diffusion coefficient has greater
-------
resistance to PCB sorption from the air and resistance to PCB migration from the source into the
encapsulant layer. Thus, the sink tests provided a screening method for comparing encapsulants.
Wipe sampling is one of the most commonly used methods for measuring surface contamination. The PCB
concentration in the wipe sample collected from the encapsulated surface is an indicator for the amount of
PCBs that has migrated from the source to the encapsulant layer. Thus, for a given source, the lower the
concentration in the wipe sample is, the better the encapsulant performs. In this study, encapsulated PCB
sources were prepared, some of which underwent natural aging, while others were subjected to accelerated
aging. Wipe samples were taken over a three-month period, and the results were used to rank the coating
materials. The accelerated aging tests were conducted in order to evaluate the potential effect of
deterioration of the encapsulants in their performances as PCB barriers.
Mathematical modeling is an essential tool for evaluating the performance of encapsulation. Previous
researchers developed several mass transfer models, known as the barrier models, for this purpose. With the
partition and diffusion coefficients obtained from the sink tests, these models can be used to study the
behavior of encapsulated sources and to evaluate the relative performances of the encapsulants, at least in
semi-quantitative terms.
1.4 About This Report
This is the third report in the publication series entitled Laboratory Study ofPolychlorinatedBiphenyl
(PCB) Contamination and Mitigation in Buildings, produced by the National Risk Management Research
Laboratory in EPA's Office of Research and Development (ORD). The first report (Guo et al., 2011) was a
characterization of primary sources that was focused on PCB-containing caulking materials and light
ballasts. The second report (Guo et al., 2012) summarized the research results for PCB transport from
primary sources to PCB sinks, including interior surface materials and settled dust. This report is focused on
the evaluation of the encapsulation method for controlling the concentrations of PCBs in buildings. This
study was limited to a laboratory investigation, and it complements and supplements an ongoing field study
in school buildings conducted by the National Exposure Research Laboratory (NERL, 2010) in EPA ORD.
-------
2. Experimental Methods
2.1 Test Specimens
Ten coating materials were selected for testing (Tables 2.1 and 2.2). The selected coating materials
represented a variety of binder systems, including epoxy, acrylic, polyurethane, polyurea, alkyd, and latex
systems. The criteria for selecting the test specimens were as follows:
• Coating types, such as epoxy and polyurethane coatings, that have been used as PCB encapsulants in the
field (Mitchell and Scadden, 2001; EH&E, 2012).
• The coating materials must be commercially available "off-the-shelf products.
• The coatings must be suitable for the substrates of concern.
• Some commonly-used interior-coating materials, such as latex and alkyd paints, must be included for
comparison.
Although some of the coating products listed in Table 2.1 have been used as PCB encapsulants for a long
time, none of them was marketed as PCB encapsulants. A silicon-based coating material is currently being
sold as a PCB encapsulant (TWO Teknik, 2011). This material was not included in the study because of our
inability to obtain the product in a timely fashion.
Mention of trade names in Table 2.1 is only for product identification; it is not an endorsement of the
products, and it is not meant to discriminate against the products that were not tested.
2.2 Sink Tests in 53-L Environmental Chambers
22.7 Test Facility
The sink tests were conducted in a two-chamber system as shown in Figures 2.1 and 2.2. The system
consisted of two identical 53-L stainless steel chambers, which conformed to ASTM D-5116 (ASTM,
201 Ob). Both chambers were housed in an incubator. The source chamber contained an open Petri dish that
held 10 g of PCB-containing caulk sample to serve as a stable source of gas-phase PCBs. The caulk was an
interior window sealant the authors obtained from a pre-demolition building and contained approximately
10% Aroclor 1254 (See Appendix A in Guo et al, 2012). The test encapsulant materials, made as mini-
disks, were placed in the test chamber. During the test, the PCB concentrations in the outlet air of the test
chamber were monitored, and the encapsulant disks were removed from the test chamber at different times
to determine their content of PCB congeners. Details about this system were described by Guo et al. (2012).
-------
Table 2.1. Coating materials tested (product names, binder types, recommended uses, and recommended application methods)
ID
01
02
03
04
05
06
07
08
09
10
Product Name
Protective Coatings Series 156
Smooth Enviro-Crete
All Surface Enamel Latex Base
MOD AC Exterior Waterproof
Coating F-100OTC
Sikagard62[a]
Industrial & Marine Coatings
Macropoxy 646 M
Protective Coatings Series 151-
1051Elasto-GripFCw
Rust-O-Lastic Universal Lacquer
Resistant Primer
All-Surface Enamel Oil Base
Gloss
EPL-9 Self Leveling Polyurea
Elastomer
Fast-Drying Polyuretnane
Short Name
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-no solvent
Epoxy-low VOC
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyuretnane
Binder or Base Material
modified waterborne
acrylate
acrylic latex
solvent acrylic
solvent-free epoxy
low- VOC polyamide epoxy
waterborne modified
polyamine epoxy
talc and quartz
oil-based enamel
polyurea
polyurethane
Recommended Use
concrete and masonry
wood, metal, drywall,
interior/exterior
concrete, cinder block, brick
concrete, steel
steel, concrete
cementitious and other porous
substrates
metals, interior/exterior
wood, metal, drywall,
interior/exterior
serf-leveling base coat; deck,
crack and floor repair
wood
Recommended
Application Method
airless/conventional sprayer,
brush, roller
brush, roller, airless sprayer
brush, roller, airless sprayer
brush, roller, airless sprayer
airless/conventional sprayer,
brush, roller
airless/conventional sprayer,
brush, roller
brush, roller, airless sprayer
brush, roller, airless sprayer
spray
natural bristle brush, foam
brush, or lambswool applicator
[a] This is a two-part coating system.
-------
Table 2.2. Coating materials tested (principal solvents, VOC content, solid content, and recommended application rates)
ID
01
02
03
04
05
06
07
08
09
10
Short Name
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-no solvent
Epoxy-low VOC
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyurethane
Principle Solvent
Type
-
water, 2-(2-methoxyethoxy)ethanol
mineral spirits
-
xylene, polyamide
-
methyl isobutyl ketone, methyl propyl
ketone, xylene, ethylbenzene
mineral oil
-
mineral spirits
Content (%w/w)
-
38,5
28
0[b]
17,12
-
-
44
0
48
VOC Content
(g/L)
49
132
445
0
255
175
339
498
400
Solid Content
(% w/w)
50.9 (v/v)
56
65
100
83
17 (v/v)
76
58
100
-
Recommended
Coverage Rate
(ft2/gal) [al
100-200
350-400
690
150-250
116-232
180-400
430-570
350-400
100
500
[b]
1 To convert (ft2/gal) to (m2/L), multiply the values in the table by 0.0245.
According to the MSD S for this product, Part A contains unspecified amount of aromatic hydrocarbon blend and Part B contains benzyl alcohol.
-------
Source Chamber
Test Chamber
1 Fan
Sliced Caulk
PUF
Figure 2.1. Schematic of the two-chamber system for sink tests, showing the air flows and sampling
locations (PUF samples)
Figure 2.2. Chamber system for sink tests: source chamber (top) and test chamber (bottom)
222 Preparation of the Coating Materials
For the chamber tests, the encapsulants were applied to stainless steel disks with diameters of 0.5 inch (1.27
cm). To prepare the disks, the encapsulants were painted onto thin stainless steel panels with a paint brush,
and the panels were placed in a ventilated fume hood for curing. The disks were created by using a steel
arch punch (Figure 2.3). Twenty-four disks were prepared for each encapsulant, four of which were
designated as background samples and placed directly into 20-mL vials for extraction and analysis. The
-------
remaining 20 disks were lightly adhered to an aluminum pin mount with a diameter of 0.7 inch (1.78 cm).
To determine the weight of the dry film, the disks were weighed before painting and after curing. The
thickness of the dry film on each coated disk was measured by using a micro-caliper. Aluminum stages were
prepared with 12 pin mounts per stage (Figure 2.4). A total of 200 sample disks were placed inside the test
chamber (Figure 2.5).
Figure 2.3. Painted stainless steel panel after the disks were punched out
Figure 2.4. Sample stage with aluminum pin mounts
-------
Figure 2.5. Sample stages in the test chamber
2 2. 3 Test Procedure
Prior to the sink test, an air sample was collected from the exhaust flow from the test chamber by using a
polyurethane foam (PUF) sampler. The exhaust flow from the source chamber was then redirected to serve
as the inlet flow for the test chamber to begin the testing phase. Daily PUF samples were taken overnight on
the test chamber and occasionally from the source chamber to ensure the consistency of the emissions. At an
elapsed time of 72 hours, the test chamber was disconnected briefly from the source flow and opened inside
a fume hood. Four disks of each type of encapsulant were removed from the chamber and placed in a 20-mL
vial for extraction and analysis. The chamber was then placed back in the incubator immediately and
reconnected to the flow of the source chamber. Daily PUF sampling was resumed. This process was
repeated four more times at elapsed times of 168, 267, 360 and 433 hours.
2.3 Wipe Sampling over Encapsulated Sources
2 3.1 Preparation of Source Panels
To prepare the PCB source for encapsulation, a calculated amount of Aroclor 1254 was added to an alkyd
primer in a 60-mL amber jar. The jar was shaken in a paint shaker (Red Devil, Model #54100Fi) for 15
minutes. Aluminum panels that measured 6 in by 3 in (15.2 cm by 7.6 cm) were used to create the source
panels. Painters' tape (Frogtape, ShurTech Brands, Avon, OFT) was used to cover the edges of the panel,
leaving an area of 8.73 cm x 5.56 cm in the center for painting. The Aroclor 1254 primer was applied to the
panel using a Crescendo Airbrush (Model 175-7, Badger Air-Brush Co., Franklin, IL) with a nitrogen gas
flow. The panels were left to cure for a minimum of 48 hours in a fume hood prior to removing the tape
(Figure 2.6).
-------
Figure 2.6. Taped panel (left); panel after primer was applied (right)
For comparison, some source panels were prepared with a two-part polysulfide caulking material (Thiokol®
2235M Industrial Polysulfide Joint Sealant). An aliquot of Aroclor 1254 was added to Part A of the
polysulfide caulk system since it was less viscous and more suitable for mixing. Part B was then added to
Part A, and the two parts were mixed until they were homogenized. Then the caulk was applied to taped
panels using a 20-mil precision wet film applicator (Paul N. Gardner Company, Inc.). The tape was removed
from the panel 48 hours after application.
2. 3.2 Application of Encapsulant
The resulting panel from Section 2.3.1 was taped 4 mm from the primer area (Figure 2.7). Encapsulants
were applied over the PCB primer according to the parameters in Table 2.3. The panels were cured in the
fume hood from 24 to 48 hours prior to use. For each encapsulant, four panels underwent aging at room
temperature (described in Section 2.3.3) and the remaining four panels underwent accelerated aging
(described in Section 2.3.4).
Figure 2.7. Re-taped panel with dried primer (left); panel after application of the encapsulant (right)
10
-------
Table 2.3. Application methods and number of coats for the encapsulated panels
ID
01
02
03
04
05
06
07
08
09
10
Short Name
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-no solvent
Epoxy-low VOC
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyurethane
Application Method
Spray
Spray
Brush
Brush/Roller
Brush
Spray
Spray
Spray
Brush
Spray
Number of Coats
2
2
2
2
1
2
2
2
1
2
2 3.3 Aging at Room Temperature
The prepared panels that were allowed to undergo aging at room temperature were placed in plastic cabinets
housed in a wooden cabinet (Figures 2.8 and 2.9). Four separate plastic cabinets were placed inside the
wooden cabinet, one plastic cabinet each for the non-PCB primer panels, non-PCB encapsulated panels,
PCB primer panels, and PCB encapsulated panels. Each plastic cabinet contained a 12-VDC computer
cooling fan that circulated the air within the compartment. All of the plastic cabinets were vented into the
chemical exhaust system of the laboratory. The plastic cabinets had no lighting and were kept at room
temperature.
Figure 2.8. Plastic cabinets (left); panels inside a plastic cabinet (right)
11
-------
Figure 2.9. Wooden cabinets that housed the plastic cabinets
2. 3.4 Accelerated Aging
Accelerated aging was performed on the prepared panels using a QUV Accelerated Weathering Tester (Q-
Lab Corporation, Westlake, OH) (Figure 2.10). The tester was placed in a stainless steel tunnel to ventilate
any possible PCB emissions to the exhaust system of the laboratory. The QUV tester contained eight 4-ft
UVA-340 lamps (four on each side) and four UV monitoring sensors. The lamps emitted a peak wavelength
of 340 nm. The irradiance for the device was set to 0.89 W/m2/nm for each test. Both sides of the tester held
12 racks, each containing two test panels. The irradiance of the QUV tester was calibrated using a CR-10
Calibration Radiometer (Q- Lab Corporation, Westlake, OH) prior to testing.
Figure 2.10. QUV Accelerated Weathering Tester: exterior (left); UV lamps (right)
12
-------
It should be noted that the QUV chamber maintains a constant temperature during a test. Thus, the chamber
cannot be used to evaluate the effect of temperature changes that could occur during normal diurnal and
seasonal changes.
2.4 Sampling and Analysis
2.4.1 Internal Standards and Recovery Check Standards
Three 13C-labeled PCB congeners were used as the internal standards (ISs): 13Ci2-PCB-4,13Ci2-PCB-52, and
13Ci2-PCB-194 (Wellington Laboratories, Shawnee Mission, KS). The internal standard solution for spiking
contained 10 ug/mL of each IS.
Three chlorinated compounds were used as the recovery check standards (RCSs): 2,4,5,6-tetrachloro-w-
xylene or TMX (Ultra Scientific, North Kingstown, RI), 13C12-PCB-77, and 13C12-PCB-206 (Wellington
Laboratories, Shawnee Mission, KS). The RCS solution for spiking contained 5 ug/mL of each RCS.
2.4.2 Air Sampling
For the sink test, air samples were collected onto PUF sampling cartridges (pre-clean certified, Supelco, St.
Louis, MO) by using a mass flow controller (Model FC-269, Coastal Instruments, Burgaw, NC) and a
vacuum pump (Model 2565B-50, Welch, Skokie, IL). The sampling flow rate was set by the mass flow
controller and measured by using a Gilian Gilibrator-2 Air Flow Calibrator (Scientific Instrument Services,
Ringoes, NJ) before and after each sampling period. After the sample was collected, the glass holder with
the sample inside was wrapped in a sheet of aluminum foil, placed in a scalable plastic bag, and stored in the
refrigerator at 4 °C until extraction. Two field samples were collected during the sink tests and the results
are presented in Table 3.5.
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 (ultra grade or
equivalent, Fisher Scientific, Pittsburgh, PA). Fifty microliters of recovery check standards were spiked onto
the PUF samples inside the Soxhlet extractor. The samples were extracted for 16 to 24 h. The extract was
concentrated to about 50 to 75 mL using a Snyder column. The concentrated solution was then 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 Corporation, Kansas City, MO).
The 1-mL solution was cleaned with sulfuric acid (certified plus grade or equivalent, Fisher, 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 spiked with 10 \\L of the internal
standard solution, after which the extract was transferred to GC vials for analysis by GC/MS. The final
sample contained 50 ng/mL of each RCS and 100 ng/mL of each IS.
2. 4.3 Extraction of Encapsulant-Coated Disks
The encapsulant-coated disks that were removed from the test chamber were extracted by using a modified
sonication method (Guo et al., 2011). The disks were extracted for 30 min in a scintillation vial with 10 mL
hexane (ultra grade or equivalent, Fisher Scientific, Pittsburgh, PA) and approximately 100 mg of sodium
sulfate (anhydrous grade or equivalent, Fisher Scientific, Pittsburgh, PA) using a sonicator (Ultrasonic
13
-------
Cleaner FS30, Fisher Scientific, Pittsburgh, PA). Before extraction, 100 uL of the recovery check standard
solution were added to the extraction solution. After extraction, 990 (iL of the extract was placed in a 1-mL
volumetric flask containing 10 uL of the internal standard solution and the contents of the volumetric flask
were transferred to GC vials for analysis. The final sample contained 50 ng/mL of each RCS and 100 ng/mL
of each IS.
The sonication method was chosen for solid samples because (1) its extraction efficiency is as good as the
Soxhlet method for the particular sample types associated with this study; (2) sonication involves fewer
steps than the Soxhlet method, reducing the possibility of sample losses; (3) sonication consumes much less
solvent. A disadvantage of the sonication method is that it cannot extract large samples such as the PUF
samples.
2. 4.4 Wipe Sampling
The wipe sampling method used in this study was based on a modified California roller method (Fuller et
al, 2001). The roller, which weighed 1.04 kg, was custom made and consisted of a stainless steel cylinder (5
cm diameter and 5 cm width) and a 24-cm wooden handle. To obtain a wipe sample, the encapsulated panel
was placed on a polished granite block (30 cm x 23 cm x 5 cm). A sterile gauze pad, or "wipe," was placed
on a piece of aluminum foil and saturated with 2 mL of FIPLC Grade, submicron-filtered 2-propanol (Fisher
Scientific, Hampton, NH). The wipe was placed on top of the painted portion of the panel (Figure 2.11). To
keep the roller PCB-free, the wipe was covered with an 8 x 15-cm piece of aluminum foil. The roller was
passed over the foil-covered wipe 10 times along the longer side of the panel and then 10 times along the
shorter side of the panel. A stopwatch was used to control the speed of the roller so the 20 passes were
completed in one minute. The wipe was removed from the panel and stored in a 20-mL scintillation vial
until the sample was extracted according to the procedure described in Section 2.4.3.
Figure 2.11. Wipe sampling process: wipe on panel (left); wipe covered with foil (center); roller
method (right)
14
-------
Evaluation of this method showed that 2-propanol had a collection efficiency similar to that of hexane and
that the use of aluminum foil on top of the wipe did not affect the collection efficiency. Details are presented
in Appendix A.
The wipe sampling method described above was developed exclusively for this study to achieve better
precision and repeatability than the commonly-used hand-wipe method. This wipe sampling method is not
recommended for other uses.
2.4.5 Sample Analysis
The analytical method used in this project was a modification of EPA Method 8082A (U.S. EPA, 2007) and
EPA Method 1668B (U.S. EPA, 2008a). The procedures are detailed in Part 1 of this report series (Guo et
al.,2011).
15
-------
3. Quality Assurance and Quality Control
Quality assurance (QA) and quality control (QC) procedures were implemented in this project by following
the guidelines and procedures detailed in the approved Category II Quality Assurance Project Plan (QAPP),
Poly chlorinated Biphenyls (PCBs) in Caulk: Evaluation of coatings for encapsulating building materials
contaminated by polychlorinated biphenyls (PCBs) and a NASA method for PCS destruction. 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 that
analyses were conducted. The QA/QC activities and results that are specific to this study are described in
Section 3.1. Data that did not meet the data quality indicators (DQIs) specified in the QAPP were not
presented. Data quality indicators (DQIs) are presented in the first report of this report series (Guo et al.,
2011).
The wipe samples presented in Sections 4.2.4 and 4.2.5 were analyzed by a commercial analytical
laboratory. The QA/QC procedures and data quality evaluation for those samples are discussed in Section
3.2.
3.1 QA/QC for the In-house Analytical Laboratory
3.1.1 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 current profiles for target analytes relative to those of
the internal standard. The calibration standards (AccuStandards, New Haven, CT) were prepared at six
concentrations, ranging approximately from 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 |iL
of the prepared standards in triplicate at each concentration level. Table 3.1 summarizes all GC/MS
calibrations conducted for the project, including the practical quantification limit (PQL, i.e., the lowest
calibration concentration) and the highest calibration concentration. The percent relative standard deviation
(RSD) of the average RRF met the data quality indicator (DQI) goal of 25%.
The Internal Audit Program (IAP) was implemented to minimize systematic errors. Prepared by personnel
other than the analyst, the IAP standards contained three calibrated PCB congeners and were analyzed after
the calibration was completed. The IAP standards were purchased from a supplier (ChemService,West
Chester, PA) that was different from the supplier for the calibration standards, and the concentrations of
PCB congeners in the standards were certified.
Table 3.2 presents the results of the analysis of the IAP standards for each calibration. The recoveries of the
IAP standards ranged from 80% to 115% and the percent RSDs ranged from 0.13% to 1.40%. All the results
met the criteria for IAP analysis, i.e., within 100 ± 25% of recovery and 25% of RSDs for triplicate
analyses.
16
-------
3.1.2 Detection Limits
After each calibration, the lowest calibration standard was analyzed seven times, and the instrument
detection limit (IDL) was determined from three times the standard deviations of the measured
concentrations of the standard. The IDLs for all calibrated PCB congeners are listed in Table 3.3. The
detection limits for the sonication method were reported in the report entitled Laboratory Study of
Poly'chlorinated' Biphenyl (PCB) Contamination and Mitigation in Buildings, Part 2. Transport from
Primary Sources to Building Materials and Settled Dust (Guo et al., 2012).
Table 3.1. GC/MS calibration for PCB congeners from Aroclor 1254 w
Date
Analytes
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX (RCS)
13C12-PCB-77 (RCS)
13C12-PCB-206
(RCS)
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
7/18/2011
RRF
0.84
1.11
0.98
0.94
1.25
1.39
1.39
1.31
1.07
0.70
0.46
1.20
1.03
%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
12/21/2011
RRF
0.76
1.10
0.86
0.83
1.13
1.23
1.17
0.94
0.86
0.61
0.44
1.00
0.98
%RSD
9.35
3.10
8.43
8.02
11.8
17.7
8.65
12.3
12.3
10.4
3.04
18.6
13.5
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
Hi Cal [bl
(ng/mL)
200
200
200
199
200
200
201
202
200
199
200
200
200
The Data Quality Indicator (DQI) goal for %RSD was 25%.
High calibration concentration.
17
-------
Table 3.2. IAP results for each calibration related to this study'
Calibration
2/14/2011
7/18/2011
12/21//2011
Analyte
PCB-52
PCB-101
PCB-77
PCB-52
PCB-101
PCB-77
PCB-52
PCB-101
PCB-77
IAP Concentration
(ng/mL)
100
100
100
80.0
80.0
80.0
40.0
40.0
40.0
Avg. Recovery
%
104
94
80
116
104
94
115
103
91
%RSD
(n=3)
0.13
0.33
0.64
0.38
1.20
1.40
0.36
0.41
1.22
1 The DQI goal for IAP recovery was 75% to 125%.
Table 3.3. Instrument detection limits (IDLs) for PCB congeners on GC/MS (ng/mL)
Analytes
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX (RCS)
13C12-PCB-77 (RCS)
13C12-PCB-206 (RCS)
2/14/2011
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
7/18/2011
0.40
0.26
0.52
0.58
0.45
0.42
0.34
0.47
0.31
0.37
0.34
0.30
1.33
12/21/2011
1.28
0.15
0.47
0.59
0.51
0.67
0.54
0.58
0.52
0.69
0.35
0.62
0.54
3.1.3 Environmental Parameters
The temperature and relative humidity (RH) sensors used to measure environmental conditions for the small
chamber sink test were calibrated in EPA's Metrology Laboratory in July, 2010. Environmental data in the
small chamber, such as temperature and RH, were recorded by the OPTO 22 data acquisition system (DAS).
18
-------
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 test in the small chamber. The Gilibrator was
calibrated by EPA's Metrology Laboratory. The environmental parameters that were measured are
presented in Table 4.1 in Section 4.1.1. The data met the data quality goals.
The accelerated weathering chamber (QUV chamber) was set to be operated at a 340 nm irradiance of 0.89
W/m2/nm) and 60 °C. A CR-10 Calibration Radiometer (Q-Lab Corporation, Westlake, OH) was used to
calibrate the irradiance of the QUV chamber prior to each test. The device was connected to the appropriate
port on the QUV control panel. Lamp type UV-A was selected on the radiometer, and the sensor was placed
into the calibration port on the back panel of the chamber. Following the appropriate procedural steps, the
radiometer automatically calibrated and updated the irradiance of the sensor to 0.89 W/m /nm. The process
was repeated for all four sensors on the QUV chamber. The CR-10 manufacturer, Q-Lab, calibrated the
radiometer on 10/23/2009. The temperature and relative humidity in the QUV chamber were not monitored.
3.1.4 Quality Control Samples
The quality control samples discussed here include background, field blank, method blank, and replicate
samples.
A typical background sample showed the contribution of the contamination in the empty chamber, the
sampling device, and the clean air supply. The results of the sink tests are summarized in Table 3.4. Some of
the PCB concentrations in the small chamber tests were above the PQL, possibly due to carryover from
previous tests. These high backgrounds did not affect the test results because the PCB concentrations in
chamber air were monitored (See Figure 4.1) and because they can be considered as part of the source for
the test chamber. No background samples were collected for the QUV chambers.
Table 3.4. Background concentrations of PCBs (jig/m3) in the chamber for the sink test'"1
Analyte
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
Empty chamber
0.08
3.17
0.70
0.25
Q QQ
0.11
0.06
Q QJ
Q QQ
Q QQ
Chamber with substrate
Q QO
0.58
0.21
Q QO
0.11
Q QQ
Q Qg
0.04
Q Ql
Q QQ
Values in strikethrough are below PQL.
19
-------
Field blank samples were acquired to determine the 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. The results are presented in Table 3.5. The target PCB congener concentrations in
the field blanks were below PQL for all collected samples.
Table 3.5. Concentration of PCBs in the field blank samples (ng) w
Analyte
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
Test ID
Wipe Sample
(ng/wipe)™
Q gQ
n fr\
u ._*u
n nn
u .uu
O39
O33
O4*
Q ()g
(^
Q og
OrW
Air in Sink Test
(ng/PUF)
Q QQ
n nn
u .uu
n nn
u .uu
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Air in Sink Test
(ng/PUF)
Q QQ
n nn
u. uu
n nn
u. uu
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Values in strikethrough are below PQL.
For the second wipe; the panels had undergone accelerated aging; average of quadruple samples.
Method blank samples were collected and subjected to the complete sample preparation and analytical
procedure. The results from the method blank samples are presented in Tables 3.6 and 3.7. The values are
all below PQL. During the lengthy storage before analysis, labels were misplaced for the method blanks of
the third wipe for panels undergoing aging at room temperature, resulting in loss of those samples.
On each day of analysis, at least one standard was analyzed as a daily calibration check (DCC) to document
the performance of the instrument. DCC samples were analyzed at the beginning and during the analysis
sequence on each day. Table 3.8 summarizes the average recovery of the DCCs for the tests. The recoveries
met the laboratory criterion of 75 to 125% recovery for acceptable performance of the GC/MS instrument.
20
-------
Table 3.6. Concentration of PCBs in the method blank samples (jig/cm2 wipe sample) for the
accelerated weathering process W[bl
Analyte
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
Sample Batch /Wipe Sample ID
Batch 1
Third Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Batch 2
Third Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Batch 2
Fourth Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Batch 3
First Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Batch 3
Second Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Batch 3
Third Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Values in strikethrough are below
Average of quadruple samples.
PQL.
Table 3.7. Concentration of PCBs in the method blank samples (jig/cm2 wipe sample) for panels
in the storage cabinet W[bl
Analyte
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
Sample Batch / Wipe Sample ID
Batch 2A[C]
Third Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
0.00
Batch 2B[d]
Third Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Batch 2A[C]
Fourth Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Batch 2B[d]
Fourth Wipe
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
Q QQ
LaJ Values in strikethrough are below PQL.
M Average of duplicate samples.
[c] For panels that had undergone accelerated aging.
[d] For panels that had undergone aging at room temperature.
21
-------
Table 3.8. Average recoveries of DCCs for the tests of the encapsulants
Analyte
PCB-17
PCB-52
PCB-101
PCB-154
PCB-110
PCB-77
PCB-66
PCB-118
PCB-105
PCB-187
TMX (RCS)
13C12-PCB-77 (RCS)
13C12-PCB-206 (RCS)
Average Recovery
101%
101%
102%
103%
104%
108%
103%
98%
104%
104%
104%
108%
100%
SD
5.57%
4.23%
6.29%
7.07%
6.46%
8.24%
7.59%
11.1%
9.81%
10.0%
6.51%
6.94%
7.61%
%RSD
5.51
4.19
6.14
6.86
6.23
7.66
7.40
11.31
9.43
9.62
6.26
6.43
7.61
NW
124
124
124
124
124
124
124
124
124
124
124
124
124
1N is the number of DCCs analyzed.
3.1.5 Recovery Check Standards
Three recovery check standards (RCSs), i.e., TMX, 13Ci2-PCB-77, and 13Ci2-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 happened mostly
during bulk analysis, the extract was diluted, and the analysis of the sample was repeated. In such cases,
recoveries of RCSs were not reported. The analytical results were 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.
As indicated in Section 3.2 (below), some extracted wipe samples were sent to a commercial analytical
laboratory for analysis. No RCSs were added to those samples during extraction. RCSs were added by
mistake to three samples, and those data were not reported.
3.2 QA/QC for Using a Commercial Analytical Laboratory
This study created a large number of samples for determination of PCB concentrations. Because of the
limited capacity of the in-house analytical laboratory, 240 wipe sample extracts (i.e., all the data presented
in Sections 4.2.4 and 4.2.5) were sent to a commercial analytical laboratory for determination of PCBs as
Aroclor 1254. This section describes the QA/QC procedure and the result of the data quality review.
22
-------
3.2.1 QA/QC Procedure
The commercial analytical laboratory that we selected was certified by the National Environmental
Laboratory Approval Program (NELAP), and it specializes in the analysis of PCBs (Aroclors and
congeners) by GC/ECD and GC/MS using standard analytical methods. EPA SW-846 Method 8082 (U.S.
EPA, 2008b) was used for the wipe samples reported in Section 4.2.4 and 4.2.5.
Five QC samples were included in each batch of samples sent to the commercial analytical laboratory for
the determination of PCB content. The QC samples were not the same set for different batches. These QC
samples included:
• Two certified Aroclor 1254 standard solutions for evaluating the accuracy of the analytical results.
• Two aliquots of the same hexane extract solution for evaluating the precision of the analytical results.
• One solvent blank for determining the presence of any potential sample contamination in the laboratory
and during transportation.
12.2 Data Quality Indicators (DQIs)
The data quality indicators (DQIs) presented in Table 3.9 were defined before receiving any data from the
commercial laboratory.
Table 3.9. Criteria for determining the usability of data reported by the commercial laboratory
Purpose
Accuracy
Precision
Laboratory
contamination
DQI
Relative error (Er)[a]
RSDM
Solvent blank (C0)
Data usability
Accepted as
quantitative data
|Er|<25%
RSD<25%
c0 < PQL
Accepted as
semi-quantitative data
25% < |Er < 50%
25% 50%
RSD > 50%
C0 > 2 PQL
Based on analysis of certified Aroclor 1254 standards: Er = (reported value - certified value) / certified value.
Based on analysis of two aliquots of the same hexane extract.
3.2.3 Data Quality Evaluation
The results for the accuracy, precision, and solvent blank analyses are presented in Tables 3.10 through
3.12.
The laboratory used tetrachloro-ra-xylene (TCMX) and PCB congener #209 as the recovery check standards
(RCSs), and the recoveries of the RCSs were satisfactory (Table 3.13).
23
-------
Table 3.10. QC samples for evaluating the accuracy of the analytical results reported by the
commercial laboratory w
Sample Batch
1
2
Standard
Solution"11
Al
A2
Bl
B2
A
B
Concentration (ug/mL)
Certified
0.202
0.202
0.8072
0.8072
0.8072
0.605
Reported
0.264
0.236
1.08
0.94
1.17
0.733
Relative Error
30.7%
16.8%
33.8%
16.5%
44.9%
21.2%
Notes
[c]
[d]
[a] EPA Method 8082.
M Aroclor 1254 standard solution, certified by AccuStandard, New Haven, CT.
[c] Re-analysis of liquid standard Al.
[d] Re-analysis of liquid standard B1.
Table 3.11. QC samples for evaluating the precision of the analytical results reported by the
commercial laboratory [a]
Sample Batch
1
2
Reported Concentration (jig/mL)
Aliquot 1
4920
6780
Aliquot 2
3960
3520
RSD
15.3%
44.8%
[a] The two aliquots were the same hexane extract of a field caulk sample; EPA Method 8082.
Table 3.12. QC samples for evaluating the potential contamination in the laboratory and during
transportation of samples [a]
Sample Batch
1
2
Concentration (ug/mL)
ND
0.0405
1 Hexane solvent blank; EPA Method 8082.
24
-------
Table 3.13. Recovery of the recovery check standards (RCSs) for all wipe samples analyzed by the
commercial laboratory
Sample Group
Aging room
temperature
Accelerated aging
Total No. of
Samples
120
119W
RCS recovery between 60% and 140%
TCMX
No. of Samples
116
118
% of Total
96.7%
99.2%
PCB-209
No. of Samples
116
118
% of Total
96.7%
99.2%
1 One missing sample excluded.
3.2.4 Conclusion Related to Data Quality Review
Judging from the pre-defined DQIs in Table 3.9 and the results of the analyses of the QC samples presented
in Tables 3.10 through 3.13, the wipe sample data presented in Sections 4.2.4 and 4.2.5 had uncertainties
between 25% and 50% as estimated based on the accuracy and the precision of the QC samples and, thus,
should be considered to be semi-quantitative (See Table 3.9). The data is still useful to compare the relative
performance of the encapsulants because the PCB concentrations in wipe samples covered a range of more
than an order of magnitude for the encapsulants evaluated.
25
-------
4. Experimental Results
4.1 Sink Tests
4.1.1 Test Conditions
The conditions of the test chamber are shown in Table 4.1. The test duration was 433 hours. The thicknesses
of the dry films of the encapsulants are presented in Table 4.2. The concentrations of the target PCB
congeners in the air of the test chamber are shown in Figures 4.1 and 4.2.
Table 4.1. Conditions of the test chamber
Parameter
Temperature (°C)
Relative humidity (%)
Air change rate (If1)
Mean [al
23.2
46.0
1.05
SD
0.08
1.34
0.01
[a]n=1684.
Table 4.2. Thicknesses of the dry films of the encapsulants used for the sink test
Coating
ID
01
02
03
04
05
06
07
08
09
10
Coating
Name
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-low VOC
Epoxy-no solvent
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyurethane
Film Thickness (mm)
Mean[a]
0.095
0.067
0.305
0.399
0.501
0.168
0.191
0.127
1.63
0.056
SD
0.009
0.026
0.052
0.092
0.156
0.015
0.060
0.043
0.298
0.022
[a] n = 24 for each encapsulant.
26
-------
4.0 -,
CO
1
c
•220-
£
4-1
0)
= 10 -
o i-u |
u
n n 4
• • • B • " " ,
f
•
•
i
v XX XX XXXX
vxXXX x X*
^ o 0 0 0 0 0 0 Op 0 0 0 p
• tt!7
• #52
A #66
XttlOl
X#105
• #110
+ #118
• #154
100 200 300
Elapsed Time (h)
400
500
Figure 4.1. Concentrations of the target congeners in the air of the test chamber (normal scale)
(Several congeners are obscured in this figure; they can be seen in Figure 4.2)
10
•2 0.1
ro
4-1
0)
u
c
o
0.01
0.001
f
1
X
^
31.
?•
F
t
\
XXX x XxX
ooo o 0°°
^ A
XX X YY*
XX
• •
A A
s»
XX
xxxx
ooo0
A* A A
• t * * *
xxxx
+ #17
• #52
A #66
X#101
X#105
O#110
+ #118
• #154
100 200 300
Elapsed Time (h)
400
500
Figure 4.2. Concentrations of the target congeners in the air of the test chamber (semi-log scale)
27
-------
4.1.2 Experimentally Determined Sorption Concentrations
The experimental results were expressed in sorption concentrations as defined by Equation 4.1:
W
C.=- (4.1,
where Cm = sorption concentration (ug/cm2)
W = amount of PCB congener detected in the encapsulant sample (fig)
A = exposed area of the encapsulant (cm2)
Figures 4.3 and 4.4 show two examples of the sorption concentration profiles. The sorption concentrations
for the Epoxy-low VOC encapsulant (Figure 4.4) are more than one order of magnitude lower than those for
the Acrylate-water borne encapsulant (Figure 4.3) indicating that the epoxy coating is more resistant to the
sorption of PCBs and, thus, has a significantly greater encapsulating ability than the acrylate coating.
Figures 4.5 through 4.7 compare the sorption concentrations for congeners #52 and #110 and Aroclor 1254
for all the encapsulants tested. In Figures 4.5 and 4.6, all the sorption concentration data were above the
practical quantification limits (PQLs). The range of the sorption concentrations for the ten coating materials
was from 0.018 to 3.91 ug/cm2 for congener #52 and from 0.012 to 0.373 ug/cm2 for congener #110. The
Aroclor concentrations in Figure 4.7 were calculated based on five individual congeners (Guo, et al., 2011,
Section 4.1.10). In all cases, the three epoxy coatings showed greater resistance to PCB sorption than the
rest of the coating materials.
Note that the sorption concentrations were calculated based on the surface area of the test specimen. Thus,
the difference in dry film thickness between the coating materials (See Table 4.2) does not affect the test
results.
28
-------
Z..U -
4
;ntration
-> h
D I
U
1
O
Q.
O
to
0.0 I
X
X
X
X
X
X
X
X
X
X
X
X
X
X ^ — "" ^^
_ *--'""
' - *" "" J^i
1 ^^^"^Tw^^T1 ^^^^^^ ~ ^^^R~~ i ^^^-^ j ^^^^^
--•-- PCB-17
--»- PCB-52
--A-- PCB-66
--X-- PCB-101
--)«-- PCB-105
--0-- PCB-110
1 PCB 118
--«-- PCB-154
100 200 300
Elapsed Time (h)
400
500
J.U •
(N
£
u
"SB !
^i -1
^
o
CD
i o.i -
Ol
u
C
3
O= 0 01
\J.\J J.
's.
o
(/)
n nm .
• — — — - fl
. - - *
— *•""" — — -^
B SA. — — — —
•• •• ' 'N « fl
.• •* /*"*w ^ ^ ^ ^*-*
^^ — — ~ "" ^*
v^ ~ "* ^ .. — — *" _ .
-"•""" •ArC-"""^^ _ ^
(^ * ' ^ ^ ^ ^ * J)*' ^ — ^k" ^ ^ ^ N/
*» /\~ ^ ^ ** ** .A* **"'"" ^^
A ^ ^ " & ^^f ~ ~ Nl^. — — "" ~
• »'"" »-""""
^ _ .. X' '
X''"
--•-- PCB-17
--»- PCB-52
--A-- PCB-66
--X-- PCB-101
--*-- PCB-105
__0-- PCB-110
1 PCB-118
--«-- PCB-154
100 200 300 400
Elapsed Time (h)Title
500
Figure 4.3. Experimentally-determined sorption concentrations as a function of time for the
Acrylate-waterborne coating material (top: normal scale; bottom: semi-log scale).
29
-------
O
'^
to
1_
4-1
c
8
o
u
o
1
o
I/)
Ons -
Onfi -
OflA -
Om .
0.00 I<
(
/
/
/
/
/
/
/ „, -» "*
tf X ^
„, *^r x 4^
m- ** x ^
,••'' -'' ,j0'".-'1'
,^^i-^r|::'---'-j:::"i
) 100 200 300 400 5C
--••- PCB-52
--A-- PCB-66
--*- PCB-101
--*- PCB-105
--0- PCB-110
-H-- PCB-118
--»- PCB-154
Elpased Time (h)
0.10
5 o.oi -
u
I
o
to
0.00
,• ---- D
x-
....o...r
100 200 300
Elpased Time (h)
400
--»- PCB-52
--A-- PCB-66
--X-- PCB-101
--*•- PCB-105
--0- PCB-110
-H-- PCB-118
--»- PCB-154
500
Figure 4.4. Experimentally-determined sorption concentrations as a function of time for the Epoxy-
low VOC coating material (top: normal scale; bottom: semi-log scale).
(The concentrations of congener #17 were below the practical quantification limit)
30
-------
sr 5-°
u
"M 4.0
I 3.0
CD
k.
4-1
§ 2.0
o
u
c
_g
'43
Q.
O
to
1.0
0.0
Congener #52
Figure 4.5. Experimentally-determined sorption concentrations for congener #52 for the ten
encapsulants (t = 433 h)
u
3 0.3
o
'43
CD
£ 0.2
E
01
u
E
u 0.1
Q.
o 0.0
to
"
Congener#110
Figure 4.6. Experimentally-determined sorption concentrations for congener #110 for the ten
encapsulants (t = 433 h)
31
-------
x x
/> '/
Figure 4.7. Calculated sorption concentrations for Aroclor 1254 for the ten encapsulants (t = 433 h)
4.1.3 Estimation of the Partition and Diffusion Coefficients
The experimentally-determined sorption concentrations were used to estimate the material/air partition
coefficients and solid-phase diffusion coefficients for the ten encapsulants. The two parameters were
estimated by fitting a mass transfer model to the experimental data by non-linear regression. The details of
this parameter estimation method are described in Part 2 of this report series (Guo et al., 2012, Appendix C).
The estimated partition and diffusion coefficients and calculated sink sorption indices (SSIs) for conger #52
— the reference congener — for the ten encapsulants are presented in Table 4.3. Figure 4.8 shows how well
the model fit the experimental data. The relative standard deviations (RSDs) for the estimated partition
coefficients ranged from 8.8 to 67% based on three estimates; the RSD range for the diffusion coefficients
was from 18% to 92%. Thus, the results should be considered rough estimates.
Selection of the reference congener is arbitrary. In this study, congener #52 was selected as the reference
congener because of its abundance in the air and in the sink material (Guo et al., 2012, Section 6.2.4). The
coefficients for the other congeners can be calculated from Equations 4.2 and 4.3:
= K
i ma_0
(4.2)
^,=^oP
" 1WJ
where Kma_; = material/air partition coefficient for congener i (dimensionless)
Kma o = material/air partition coefficient for the reference congener (dimensionless)
P0 = vapor pressure for the reference congener (torr)
(4.3)
32
-------
P; = vapor pressure for congener i (torr)
a = material-dependent constant from Table 4.3
Dmj = solid-phase diffusion coefficient for congener i (m2/h)
Dm o = solid-phase diffusion coefficient for the reference congener (m2/h)
m0 = molecular weight of the reference congener (g/mol)
ni; = molecular weight of congener i (g/mol)
(3 = 6.5 for all materials.
Once the material/air partition coefficients for two materials are known, the material/material partition
coefficient of the two materials can be calculated from Equation 4.4 (Kumar and Little, 2003). The
material/material partition coefficient is a key parameter that controls the migration of PCBs from the source
to the encapsulant.
K
ma 1
K
(4.4)
where K12 = material/material coefficient between material 1 and material 2 (dimensionless)
Kma i = material/air partition coefficient for material 1 (dimensionless)
Kma 2 = material/air partition coefficient for material 2 (dimensionless)
0.0
100
200 300 400
Elapsed Time (h)
500
Figure 4.8. Goodness-of-fit for estimating the partition and diffusion coefficients for the Acrylic-
solvent coating material
(The lines are model fit; the symbols are experimental data; r2 is the overall coefficient of
determination for the non-linear regression)
33
-------
Table 4.3. Estimated partition coefficient (Kma), diffusion coefficient (Dm) for the reference congener (#52) and index a in
Equation 4.2 [a]
Encapsulant
ID
01
02
03
04
05
06
07
08
09
10
Name
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-low VOC
Epoxy-no solvent [d]
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyurethane
K^ (Dimensionless)
Mean
1.93xl07
2.05xl07
1.34xl07
3.05xl06
1.78xl06
2.02 xlO6
7.90xl06
1.62xl07
1.78xl07
5.93 xlO6
RSD
29.4%
19.6%
35.9%
57.2%
62.3%
67.1%
20.6%
35.7%
9.5%
8.8%
Dm(m2/h)
Mean
4.88xlO"10
1.75X10-10
2.06xlO"10
3.76x10-"
1.89xlO'12
8.36xlO'12
9.50x10""
4.53 xlO'10
1.34X10'09
1.12X10'10
RSD
58.6%
36.0%
73.8%
72.2%
92.0%
75.9%
45.3%
60.1%
18.0%
19.4%
a (Dimensionless)
Mean
0.418
0.582
0.646
0.912
1.007
0.981
0.829
0.478
0.232
0.896
RSD
0.0%
0.0%
0.0%
5.8%
0.0%
5.0%
0.0%
0.0%
0.0%
0.0%
,^[b]
0.9814
0.9849
0.9922
0.9544
0.9741
0.9819
0.9854
0.9981
0.9502
0.9887
SSI[cl
Mean
2.0
2.4
2.6
3.9
5.5
4.8
3.1
2.1
1.6
3.2
RSD
6.8%
3.6%
6.7%
6.6%
12.2%
7.2%
3.2%
7.5%
5.2%
3.0%
[a] The mean and RSD values were based on three estimates.
[b] Overall coefficient of determination for the non-linear regression.
[c] SSI is sink sorption index; SSI = -log (Kma Dm), from Quo et al. (2012).
[d] Most sorption concentration data were below the PQLs for this encapsulant.
34
-------
4.2 Wipe Sampling over Encapsulated Sources
4.2.1 PCB Concentrations in the Source
The PCB source used in the wipe sampling tests was prepared by mixing a calculated amount of the Aroclor
1254 standard with an alkyd primer (Section 2.3.1). Three batches of PCB primer were prepared and the
PCB concentrations in the dry primer are presented in Table 4.4.
Table 4.4. Concentrations of target congeners in three batches of dry primer w
Congener/Aroclor
ID
#17
#52
#66
#77
#101
#105
#110
#118
#154
#187
Aroclor 1254
PCB Concentration (jig/g)
Batch l[bl
4m
523
74.7
Q QQ
877
317
927
633
101
32rS4
13100
Batch 2[cl
^92-
552
84.3
4r56
917
334
1003
526
108
32^6
13300
Batch 3[cl
23$
526
66.4
Q QQ
845
300
886
599
101
32^
12600
Ial Values in strikethrough font are below the PQLs.
[bl For not-encapsulated PCB panels (Section 4.2.3).
[cl For encapsulated PCB panels (Sections 4.2.4 and 4.2.5).
4.2.2 Thicknesses of the Dry Films of the Encapsulants
As shown in Table 4.5, there were significant differences in the thickness of the dry films among the ten
encapsulants because of differences in their viscosities. This factor is difficult to control when comparing
the performance of different encapsulants. Each coating material has recommended range of application
rate. For thick products, only one coat was applied; two coats were applied for the thin products. More
discussion on the effect of encapsulant thickness is presented in Section 5.3.6.
35
-------
Table 4.5. Thickness of dry films of the encapsulants for the wipe sampling tests1'
Encapsulant
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-low VOC
Epoxy-no solvent
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyurethane
Thickness of the Dry Film (mm)
Aging at room
temperature
0.122±0.011
0.112±0.014
0.361 ±0.041
0.162 ±0.018
0.289 ± 0.046
0.092 ± 0.003
0.122 ±0.018
0.076 ± 0.009
0.368 ±0.139
0.098 ±0.005
Accelerated aging
0.129 ±0.010
0.091 ±0.015
0.438 ±0.054
0.175 ±0.031
0.274 ±0.107
0.103 ±0.008
0.116±0.019
0.077 ±0.012
0.373 ±0.044
0.108 ±0.008
1 Results are mean ± SD for n = 4.
4.2.3 Wipe Samples for Not-encapsulated Sources That Had Undergone Aging at Room Temperature
For comparison with encapsulated source panels, wipe samples were taken for not-encapsulated PCB primer
panels. The results, summarized in Table 4.6, were used to calculate the percent reduction of PCBs in wipe
samples for encapsulated sources (Section 4.2.4). Note that wipe samples were taken from each panel three
times. The later samples had lower concentrations because of the loss of PCBs due to the earlier wipe
sampling.
Table 4.6. Concentrations of Aroclor 1254 in wipe samples taken from not-encapsulated source
panels
Wipe
ID
First wipe
Second wipe
Third wipe
Elapsed
Time (h)
170.5
695.7
1248.3
Concentration of Aroclor 1254 (ng/100 cm2)
Panel 1
2294
1221
969
Panel 2
2313
1241
1221
Panel 3
2150
1227
908
Panel 4
2156
1400
1087
Statistics
Average
2228
1273
1046
SD
87.1
85.5
138
n
4
4
4
4.2.4 Encapsulated Sources That Had Undergone Ageing at Room Temperature
The wipe sample data presented in this section should be considered semi-quantitative. Details about the
data quality review are given in Section 3.2.
36
-------
The PCB concentrations, as Aroclor 1254, in the three rounds of wipe samples are presented in Figures 4.9
through 4.11. Figure 4.12 compares the sum of three rounds of wipe sampling for the encapsulants and the
relative contribution of each wipe to the total amount of PCBs collected in three wipes. The ranking of the
ten encapsulants based on wipe sampling showed some resemblance to the results of the sink tests (Figure
4.7). The three epoxy coatings performed well in both tests. Although the two types of tests were based on
completely different mass transfer mechanisms (i.e., migration from a solid source versus deposition from
the air), such consistency demonstrates that these tests are useful for evaluating the performance of different
encapsulants.
Polyurea-elastomer was an obvious exception. This encapsulant showed the worst performance in the sink
test but had very good performance in wipe sampling. The difference in the thickness of the encapsulants in
the wipe sampling tests may have contributed partially to this discrepancy. As shown in Table 4.5, the
Polyurea-elastomer was the thickest among all the encapsulants tested. Results of mathematical modeling
presented in Section 5.3.6 show that, for a given source and a given encapsulant, the PCB concentration at
the exposed surface decreases when the thickness of the encapsulant increases. Thus, increased thickness
reduces the PCB concentration at the exposed surface. In future tests, this coating material should be further
evaluated.
_. 800
(N
E
u
1 600
5 400
(N
200 -
X
' /
c/ ^
,£> «e
t = 167 hours
-------
800
Figure 4.10. Concentrations of Aroclor 1254 in the second-round wipe samples taken over
encapsulated PCB panels that had undergone aging room temperature (error bar = ±1
SD)
800
^ ^ , ^ J* J" ^
<.x .\.'° ,f> 4? v"!?
& & & -^ $r
* U XV «^ A .;
^
Figure 4.11. Concentrations of Aroclor 1254 in the third-round wipe samples taken over encapsulated
PCB panels that had undergone aging at room temperature (error bar = ±1 SD)
38
-------
1600
Figure 4.12. Concentrations of Aroclor 1254 for the sum of three rounds of wipe samples taken over
encapsulated PCB panels that had undergone aging at room temperature
The percent reduction of PCBs in the wipe samples for the encapsulated sources was calculated by using
Equation 4.5:
% Reduction = -^ x- x 100%
Cn
(4.5)
where C0 = PCB concentration in the wipe sample for the not-encapsulated source panel (ug/100 cm2)
Cx = PCB concentration in the wipe sample for the encapsulated source panel (fig/100 cm2)
As shown in Table 4.7, the percent reduction for the ten encapsulants ranged from 40.3% to 99.9%.
39
-------
Table 4.7 Percent reduction of PCB concentrations in wipe samples for encapsulated PCB
sources
Encapsulant
ID
01
02
03
04
05
06
07
08
09
10
Encapsulant
Oil enamel
Epoxy-low VOC
Epoxy-waterborne
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Lacquer primer
Polyurethane
Epoxy-no solvent
Polyurea elastomer
Percent Reduction of Aroclor 1254 in
Wipe Samples
Wipel
86.1%
98.4%
94.3%
89.8%
77.9%
89.6%
73.8%
89.4%
99.5%
97.2%
Wipe 2
77.3%
99.0%
97.3%
80.1%
67.3%
89.6%
73.8%
82.7%
99.1%
93.6%
Wipe3
74.2%
100.0%
99.8%
69.3%
40.3%
86.1%
72.0%
98.7%
99.9%
99.2%
4.2.5 Encapsulated Sources That Had Undergone Accelerated Aging
The wipe sample data presented in this section should be considered semi-quantitative. Details about the
data quality review are presented in Section 3.2.
The concentrations of Aroclor 1254 in the wipe samples taken from the panels that had undergone
accelerated aging were significantly lower than those from the panels that had undergone aging at room
temperature. Over half of the wipe samples had concentrations in the noise level of the analytical method
(Figures 4.13 through 4.17). One possible cause of this difference may have been the elevated temperature
(60 °C) in the QUV chamber. During the two-week period of testing, PCBs may have been driven off the
source surface in significant quantities. Although the test results are difficult to interpret, the following
observations can still be made:
• The epoxy coatings and the polyurea elastomer performed well relative to other encapsulants, consistent
with the results from aging at room temperature.
• The main purpose of the accelerated aging tests was to determine whether degradation of the
encapsulant may cause increased concentrations of PCBs at the encapsulated surface. The test results
did not indicate any such effect.
UV irradiation is a key factor that causes degradation of polymers in the coating materials. Given that the
UV intensity inside buildings is much lower than in the ambient environment, the degradation of coating
materials is much slower indoors than outdoors. Thus, in most cases, the degradation of coating materials is
not a primary concern for PCB encapsulants inside buildings. Weathering of the encapsulant could be a
concern, however, if the source is on the exterior side of the building.
40
-------
,-, 80°
(N
U
3 60°
400
(N
_O
o 200
t « 350 hours
Figure 4.13. Concentration of Aroclor 1254 in the first-round wipe samples taken over encapsulated
PCB panels that had undergone accelerated aging (error bar = ±1 SD)
o
o
1
in
(N
_o
o
800
600
400
t « 800 hours
Figure 4.14. Concentration of Aroclor 1254 in the second-round wipe samples taken over
encapsulated PCB panels that had undergone accelerated aging (error bar = ±1 SD)
41
-------
CM
U
O
S Ron -
1
N 400 -
_o
X
t« 1500 hours
n
I I r^n r^n
/ y y ,/ y y ,/ ,/ ^ y
Figure 4.15. Concentration of Aroclor 1254 in the third-round wipe samples taken over encapsulated
PCB panels that had undergone accelerated aging (error bar = ±1 SD)
^ 1600
"E
u
O
O
a
in
IN
800
.2 400
u
O
DWipeS
• Wipe 2
DWipel
Figure 4.16. Concentrations of Aroclor 1254 for the sum of three rounds of wipe samples taken over
encapsulated PCB panels that had undergone accelerated aging
42
-------
2000
u
O
O
m
IN
1000 -
o
73 500
O
D Natural aging
D Accelerated aging
i
&
Figure 4.17. Comparison of wipe sampling results (the sum of three wipes) for the two aging methods
4.2.6 Additional Wipe Sampling Tests
4.2.6.1 Effect of the Thickness of the Encapsulant
The effect of the thickness of the encapsulant on the PCB concentration in wipe samples was evaluated by
applying one and two coats of an encapsulant to the same types of source panels. The test conditions were as
follows:
PCB source
Substrate
Source area
Encapsulant
Thickness of encapsulant
Application method
Aging method
Wipe sampling times
Alkyd primer mixed with 0.71% of Aroclor 1254
6 in x 3 in (15.2 cm x 7.6 cm) aluminum panels
51.8±0.42cm2(n = 6)
Acrylic-solvent
0.277 ± 0.032 mm (n = 3) for one coat
0.441 ± 0.045 mm (n = 3) for two coats
roller
At room temperature without lighting
167, 692, and 1245 elapsed hours
For a given source and a given encapsulant, the test results show that the PCB concentration in the wipe
samples decreases as the thickness of the dry film of the encapsulant increases (Figures 4.18 through 4.21).
43
-------
#52 #66 #101 #105 #110 #118 #154 #187
Congener ID
Figure 4.18. Concentrations of target congeners in wipe samples taken at 167 elapsed hours
(The encapsulant was Acrylic-solvent; error bar = ±1 SD)
#52 #66 #101 #105 #110 #118 #154 #187
Congener ID
Figure 4.19. Concentrations of target congeners in wipe samples taken at 692 elapsed hours
(The encapsulant was Acrylic-solvent; error bar = ±1 SD)
44
-------
#52 #66 #101 #105 #110 #118 #154 #187
Congener ID
Figure 4.20. Concentrations of target congeners in wipe samples taken at 1245 elapsed hours.
(The encapsulant was Acrylic-solvent; error bar = ±1 SD)
800
Wipel
Wipe 2
WipeS
Figure 4.21. Concentrations of Aroclor 1254 in three wipe samples taken at 167,692, and 1245
elapsed hours
(The encapsulant was Acrylic-solvent; error bar = ±1 SD)
A similar test was conducted with Epoxy-no solvent as the encapsulant. The results were not reported
because most target congeners in the wipe samples were below the PQLs. Again, the results confirm that
this coating material performed well as a PCB encapsulant.
45
-------
4.2.6.2 Effect of Source Substrate
The properties of the source substrate may affect the performance of the encapsulant. A substrate with a
large material/air partition coefficient and a small solid-phase diffusion coefficient has a tendency to resist
the migration of PCBs from the source to the encapsulant (Guo et al, 2012). Thus, the same encapsulant
may perform differently for different sources. To evaluate the potential effect of the source substrate, PCB
source panels were prepared with an alkyd primer and a polysulfide caulk that contained approximately the
same concentrations of Aroclor 1254 (Table 4.8). Panels for each of the source substrates were encapsulated
with Lacquer-primer and Polyurethane. Four encapsulated panels were made for each substrate/encapsulant
combination. Wipe samples were taken after 188 hours and the results are presented in Figures 4.22 and
4.23. For the panels coated with the Lacquer-primer, the congener concentrations in wipe samples for the
alkyd primer panels were, on average, 34% lower than for the caulk panels, indicating that the alkyd primer
is a greater sink for PCBs than the caulk.
A more significant difference was observed between the two substrates when they were encapsulated with
Polyurethane. The thickness of the Polyurethane layer was a negative value when measured over the caulk,
suggesting that the encapsulant had penetrated into the caulk. The "mixing" of the encapsulant with the
source presents a complicated case for the encapsulation method, and its potential effects on the
performance of the encapsulant should be evaluated further.
The source/encapsulant partition coefficient, defined by Equation 4.4, is an important property of the source
substrate that affects the performance of the encapsulant. A source substrate with a large source/encapsulant
partition coefficient tends to "keep" PCBs in the source and, thus, reduce the concentration in the
encapsulant.
Table 4.8. PCB concentrations in cured substrates
Congener/Aroclor
ID
#17
#52
#66
#77
#101
#105
#110
#118
#154
#187
Aroclor 1254
Concentration (ug/g)
Primer
1.59
313
50.9
1.22
596
211
634
472
70.3
19.3
8820
Caulk
2.51
332
58.8
0.95
607
228
585
560
61.3
13.0
8164
46
-------
15
E 12
u
O
o
J
£
O
ro
4-1
0)
u
§
#52
#105 #101 #110
Congener ID
#118
Figure 4.22. Effect of source substrate on PCB concentrations in wipe samples — the sources (primer
and caulk) were encapsulated with the Lacquer-primer (error bar = ±1 SD)
#52
#105
#101 #110
Congener ID
#118
#154
Figure 4.23. Effect of source substrate on PCB concentrations in wipe samples — the sources (primer
and caulk) were encapsulated with Polyurethane (error bar = ±1 SD)
47
-------
5. Mathematical Modeling
5.1 Model Description
5.1.1 Available Barrier Models
Barrier models are mass transfer models for predicting the behavior of encapsulated sources. Several models
are available for evaluating the effect of the barrier layer on emissions of chemical substances from solid
building materials (Little et al., 2002; Kumar and Little, 2003; Hu et al, 2007; Yuan et al, 2007). The first
barrier model (Little et al., 2002) gives the exact solution to the case in which the barrier material is applied
to one side of the source panel. Unfortunately, one parameter in the model (CL2) was either mistyped or
undefined. The model developed by Kumar and Little (2003) is for predicting the rate of mass transfer
between a double-layer material and indoor air. The model is in the form of the exact solutions and the
material can be either a source or a sink. The model developed by Hu et al. (2007) provides a generalized
form of the exact solutions to the emissions from multi-layered building materials. The model developed by
Yuan et al. (2007) is a fugacity model for layered materials. As a barrier model, it works for cases in which
the barrier material is applied to either one or two sides of the source panel. The model is solved
numerically. In this study, the fugacity model was used to investigate the general behavior of encapsulated
sources, and the factors that affect the performance of the encapsulants. It was also used to rank the ten
encapsulants that were tested.
5.1.2 The Concept of Fugacity
Fugacity can be regarded as the "escaping tendency" of a chemical substance from an environmental
compartment or phase (Mackay and Paterson, 1981). The fugacity of a chemical is linked to its
concentration, as shown in Equation 5.1:
C = ZF (5.1)
where C = concentration of the chemical in the compartment (mol/m3)
Z = fugacity capacity of the compartment for the chemical [mol/(m3 Pa)]
F = fugacity of the chemical in the compartment (Pa)
Fugacity capacity (Z) is an important parameter that dictates the movement of contaminants between
environmental compartments. The contaminants tend to accumulate in the compartments with large fugacity
capacities. The direction of the flux of the contaminant between two compartments is determined by the
fugacity difference between the two compartments (i.e., from the compartment with a higher fugacity to the
compartment with a lower fugacity).
One of the advantages of the fugacity models over conventional concentration models is the continuity of
the fugacity at the interface of the two compartments because such continuity often simplifies the numerical
computations, especially when the model includes partial differential equations.
48
-------
The literature on the concept and principles of fugacity modeling is widely available. Concise but rather
comprehensive discussions on fugacity modeling can be found in the articles by Mackay and Paterson
(1981, 1982).
5.1.3 The Fugacity-Based Barrier Model
The fugacity-based barrier model developed by Yuan et al. (2007) was used to evaluate the performance of
the encapsulant. The model can be applied to a source with either one or two sides encapsulated. Figure 5.1
is the schematic representation of the double-layer model in which only one side is encapsulated.
1 2
x-L-L
x=0
Yin, Q y, Q
Room Air (y = ZairF2|x=L1+L2)
V,A
Layer 2: Encapsulant (F2, Z2)
Layer 1: PCB Source (F1; ZJ
"
/\
x
Figure 5.1. Schematic representation of the double-layer model (Yuan et al., 2007)
The definitions of the symbols are as follows:
FI, F2 = fugacity of the PCB congener in layers 1 (source) and 2 (encapsulant), (Pa)
Z1; Z2 = fugacity capacity of the PCB congener in layers 1 (source) and 2 (encapsulant), [mol/(m3 Pa)]
Zair = fugacity capacity for air; at room temperature, Zair ~ 40 [mol/(m3 atm)] or 4* 10"4 [mol/(m3 Pa)]
L], L2 = thickness of layers 1 (source) and 2 (encapsulant), (m)
V = volume of room (m3)
Q = air change flow rate (m3/s)
A = exposed area of the source that is encapsulated (m2)
y = concentration of contaminant in room air (ug/m3)
ym = concentration of contaminant in the inlet of air change flow, used to represent the source in the
room (i.e., ym = emission rate divided by the air change flow rate) (ug/m3)
x = distance from the bottom of the source layer; x=L]+L2 at the exposed surface (m)
49
-------
The model was developed with the following assumptions: (1) There is no contaminant flux through the
bottom layer (Equation 5.2); (2) The air in the room is well mixed and, thus, the mass balance for the
contaminant in room air can be established (Equation 5.3, which can be converted to Equation 5.4 and then
to Equation 5.5); (3) The contaminant at the surface of the top layer is always in equilibrium with the room
air (Equation 5.6); (4) The contact between the two layers is perfect and, thus, the fugacity at the interface is
continuous (Equations 5.7 and 5.8).
= 0 (5.2)
x=0
(5.3)
dx
VZair —2- = QZair Fm + D2AZ2-±- QZairF2 (5.4)
dt dx
(5.5)
2 dx 2 dt
where Fin=yinIZair
__Q__
2~ AK2D2
V
AK2
K2=Z2IZair
' = Z*rF^,^ (5.6)
(5.7)
(5.8)
where DI = diffusion coefficient for the contaminant in the source layer (m2/s)
D2 = diffusion coefficient for the contaminant in the barrier layer (m2/s)
KI = material/air partition coefficient of the source layer for the contaminant (dimensionless)
50
-------
K2 = material/air partition coefficient of the barrier layer for the contaminant (dimensionless)
Zi = KI Zair
Given the fugacities of the contaminants at time t = 0, Equations 5.5 and 5.8 can be solved numerically. The
MATLAB code of the model used in this study was provided by Drs. John Little and Zhe Liu of Virginia
Polytechnic Institute and State University, the co-authors of this model (Yuan et al, 2007).
5.2 Input Parameters
5.2.1 Parameters Required by the Model
This fugacity model requires 13 input parameters if one side of the source panel is encapsulated (Table 5.1).
Among these parameters, the partition coefficients, diffusion coefficients, the initial concentration, and the
film thickness of the encapsulant are the most important parameters. For air concentrations, the ventilation
rate and source area are also important.
Table 5.1. Input parameters for the fugacity model
Parameter
Exposed area of the source
Thickness of the source layer
Thickness of the barrier layer
Molecular weight of the contaminant [a]
Initial concentration of the contaminant in the source layer
Initial concentration of the contaminant in the barrier layer
Concentration of the contaminant in the inlet air M
Material/air partition coefficient for the contaminant and source layer
Material/air partition coefficient for the contaminant and the barrier layer
Diffusion coefficient for the contaminant and in the source layer
Diffusion coefficient for the contaminant and in the barrier layer
Room volume
Air change flow rate
Symbol
A
Li
L2
MW
Coi
Cfl2
Yin
KI
K2
Di
D2
V
Q
Unit
m2
m
m
Hg/mol
Hg/m3
Hg/m3
Hg/m3
dimensionless
dimensionless
m2/s
m2/s
m3
m3/s
[a] For converting fugacity (Pa) to concentration ((ig/m3). To convert the concentration in the solid material from ((ig/m3)
to (ng/g), the density of the material is required.
M Parameter ym is also used to represent the emissions from other indoor sources.
5.22 Parameter Values for the "Base-case " Scenario
Table 5.2 lists the "base-case" values of nine parameters for use in all the simulations unless indicated
otherwise. The remaining four parameters, the partition and diffusion coefficients for the source and
encapsulant, are given in Table 5.3. The source material is assumed to be concrete and the encapsulant is
51
-------
either Lacquer primer or Epoxy-waterborne. The values in Table 5.3 are for congener #110, the most
abundant congener in Aroclor 1254.
Table 5.2. Base-case values for the simulations
Parameter
Exposed area of the source (concrete)
Thickness of the source layer
Thickness of the barrier layer
Molecular weight of the contaminant (congener #110)
Initial concentration of congener #1 10 in the source layer
Initial concentration of congener #1 10 in the barrier layer
Concentration of congener #1 10 in the inlet air
Room volume
Air change flow rate
Symbol
A
Li
L2
MW
Coi
CCG
Ym
V
Q
Value
10
0.005
0.0001
3.27xl08
2.00xl08
0
0
100
2.78xlO"2
Unit
m2
m
m
Hg/tnol
Hg/m3
Hg/m3
Hg/m3
m3
m3/s
Notes
= 5mm[a]
= 0.1 mm
= 327 g/mol
= 100 ng/g
= 100m3/h
1 This is not the thickness of the concrete structure; it is the thickness of the layer that is contaminated with PCBs.
Table 5.3. Partition coefficients (Kma) and diffusion coefficients (Dm) for congener #110 for the
source and encapsulants w
Material Category
Source
Encapsulant
Material Name
Concrete [a]
Epoxy-waterborne^
Lacquer primer^
Kma
6.95 x 107
1.73 x 107
4.85 x 107
Dm(m2/s)
4.00 x 10'15
1.25 x 10'15
1.28x 10'14
fromGuoet al. (2012).
Calculated from Equations 4.2 and 4.3 using the data in Table 4.3.
To convert the concentrations in the solid phases from (ug/m3) to (ug/g), the density is assumed to be 2.0
g/cm3 for the source (concrete) and 1.2 g/cm3 for the encapsulant.
5.3 General Behavior of Encapsulated Sources
5.3.1 Concentration Profiles in the Source
Using the parameters in Tables 5.2 and 5.3 as input to the fugacity model described in Section 5.1.3, the
concentration profiles in the source and encapsulant layers can be calculated. As shown in Figures 5.2 and
5.3, a concentration gradient develops at the interface of the source and the encapsulant (i.e., x = 5 mm),
resulting in a decreased driving force for PCB migration from the source to the encapsulant. The
development of the concentration gradient is faster for the Lacquer primer (Figure 5.2) than for the Epoxy
(Figure 5.3) because the former has a greater diffusion coefficient (Table 5.3).
52
-------
120
100
-^ 60
">T
O
40
20
Concrete/ Lacquer-primer
•1 day
•10 days
•100 days
•1000 days
•5000 days
2 3
x(mm)
Figure 5.2. Concentration profiles for congener #110 in the source encapsulated with a Lacquer-
primer [Ci(x)].
(The interface of the source and the encapsulant is at x = 5 mm; the initial concentration in
the source (C0i) was 100 ug/g.)
"3)
120
100
60
•=: 40
o
20
Concrete / Epoxy-waterborne
1 day
10 days
100 days
1000 days
5000 days
23
x(mm)
Figure 5.3. Concentration profiles for congener #110 in the source encapsulated with a waterborne
epoxy coating [Ci(x)j
(The interface of the source and the encapsulant is at x = 5 mm; the initial concentration in
the source (C0i) was 100 ug/g.)
53
-------
5.3.2 Concentration Profiles in the Encapsulant Layer
The profile of the contaminant in the encapsulant layer is more complex than in the source (Figures 5.4 and
5.5). The accumulation of contaminant in the encapsulant is controlled by two factors, i.e., the gain due to
the flux of the contaminant from the source and the loss due to emissions to room air. The net effects are: (1)
the contaminant "fills up" the encapsulant layer quickly in the early days; (2) the concentration of the
contaminant at x = 0 (i.e., the interface of the source and the encapsulant) decreases over time; and (3) the
concentration of the contaminant at x = 0.1 (i.e., the surface of the encapsulant that is exposed to air)
increases at first and then decreases over time.
The concentration at the interface of the source and the encapsulant (i.e., x = 0) is greater for the Lacquer
than for the Epoxy because the former has a greater material/air partition coefficient. The concentration
gradient in the encapsulant layer is steeper for the Epoxy than for the Lacquer because the former has a
smaller diffusion coefficient.
Concrete/ Lacquer-primer
0.00 0.02 0.04 0.06
x(mm)
0.08
0.10
Figure 5.4. Concentration profiles for congener #110 in the encapsulant layer (Lacquer primer) as a
function of depth
(The interface of the source and the encapsulant is at x = 0 mm; the exposed surface is at x =
0.1 mm; the initial concentration in the source (C0i) was 100 ug/g)
54
-------
50
40
Concrete/ Epoxy-waterborne
•1 day
•10 days
•100 days
•1000 days
•10000 days
0.00 0.02 0.04 0.06
x(mm)
0.08
0.10
Figure 5.5. Concentration profiles for congener #110 in the encapsulant layer (Epoxy-waterborne)
as a function of depth
(The interface of the source and the encapsulant is at x = 0 mm; the exposed surface is at x =
0.1 mm; the initial concentration in the source (C0i) was 100 ug/g.)
5.3.3 Average Concentration in the Encapsulant Layer
With the concentration profiles shown in Figures 5.4 and 5.5, the average concentrations in the encapsulant
can be calculated from Equation 5.9:
C2= —
1
2L
(5.9)
2 i=0
where C2 = average concentration in the encapsulant layer (ug/g)
C2(x) = concentration in the encapsulant layer at depth x (ug/g)
L2 = thickness of the encapsulant layer (m)
C2l = the ith data point for C2(x), (ug/g)
C2l+1 = the (i+l)th data point for C2(x), (ug/g)
n+1 = number of data points.
Figure 5.6 shows the average concentration in the encapsulant layer as function of time. The contaminant
accumulates in the encapsulant rapidly in the early hours, followed by a slow decrease. The decrease is
55
-------
caused mainly by the concentration gradient formed at the interface of the source and the encapsulant
(Figures 5.2 and 5.3).
•Lacquer primer
•Epoxy-waterborne
1000
2000 3000 4000
Elapsed Time (days)
5000
Figure 5.6. The average concentration of congener #110 in the encapsulant layer (C2) as a function
of time
(The initial concentration in the source, C0i, was 100 ug/g.)
5.3.4 Concentration at the Exposed Surface
The PCB concentration on the exposed surface of the encapsulant is an important parameter for evaluating
the performance of an encapsulant because the concentration at the surface is linked to dermal exposure and
to the contribution of the encapsulated source to the PCB concentration in the air. As shown in Figure 5.7, a
significant difference exists between the two encapsulants due to the combined effect of the partition and
diffusion coefficients. Note that the concentration at the exposed surface (Figure 5.7) is always less than the
average concentration in the encapsulant layer (Figure 5.6).
5.3.5 Contribution to PCB Concentrations in Room Air
One of the main goals of encapsulating PCB sources is to reduce the PCB concentrations in room air. Figure
5.8 shows the contribution of the encapsulated source to the PCB concentrations in room air. For
comparison, the air concentration due to emissions from not-encapsulated concrete is also included. The
difference between the two encapsulants is more significant in the short term than in the long term.
5.3.6 Effect of the Thickness of the Encapsulant
The effect of the thickness of the encapsulant on the average concentration in the encapsulant layer is
complex. As shown in Figures 5.9 and 5.10, as the thickness increases, the average concentration increases
at first and then decreases.
56
-------
•Lacquer primer
•Epoxy-waterborne
1000 2000 3000 4000
Elapsed Time (days)
5000
Figure 5.7. Concentration of congener #110 at the exposed surface of the encapsulant [C2(x=L2)] as a
function of time
(The initial concentration in the source, C0i, was 100 ug/g.)
£
O
4J
(D
4-1
0)
U
£
O
u
3.0
2.0
1000
•Lacquer primer
•Epoxy-waterborne
•Not-encapsulated concrete
2000 3000 4000
Elapsed Time (days)
5000
Figure 5.8. Concentration of congener #110 in room air due to emissions from the encapsulated
source as a function of time
(The initial concentration in the source, C0i, was 100 ug/g.)
57
-------
0)
00
50
40
30
0
Concrete/ Lacquer-primer
•t= 100 days
•t= 1000 days
0.00 0.20 0.40
Thickness of Encapsulant (mm)
0.60
Figure 5.9. Effect of the thickness of the encapsulant on the average concentration of congener #110
in the encapsulant layer (average C2) — Case 1: Lacquer primer
0)
00
S
0)
25
Concrete/ Epoxy-waterborne
•t = 100 days
•t = 1000 days
0.0 0.1 0.2 0.3 0.4 0.5
Thickness of Encapsulant (mm)
0.6
Figure 5.10. Effect of the thickness of the encapsulant on the average concentration of congener #110
in the encapsulant layer (average C2) — Case 2: Epoxy-waterborne
The effect of the thickness of the encapsulant on the concentration at the exposed surface is much simpler,
i.e., as the thickness increases, the surface concentration decreases (Figures 5.11 and 5.12). The contribution
of the encapsulated source to the concentrations of PCBs in room air follows a similar trend (Figures 5.13
and 5.14).
58
-------
40
^ 20
10
•t = 100 days
•t = 1000 days
Concrete/ Lacquer primer
0.0 0.2 0.4
Thickness of Encapsulant (mm)
0.6
Figure 5.11. Effect of the thickness of the encapsulant on the concentration of congener #110 at the
exposed surface of the encapsulant [C2(x=L2)] — Case 1: Lacquer-primer
Concrete / Epoxy-waterborne
t = 100 days
t = 1000 days
0.0 0.1 0.2 0.3 0.4 0.5
Thickness of Encapsulant (mm)
0.6
Figure 5.12. Effect of the thickness of the encapsulant on the concentration of congener #110 at the
exposed surface of the encapsulant [C2(x=L2)] — Case 2: Epoxy-waterborne
59
-------
1.0
0.8
0 0.6
£
4-1
I
o
u
0.4
r 0.2
0.0
•t = 100 days
•t = 1000 days
Concrete/ Lacquer primer
0.0 0.1 0.2 0.3 0.4 0.5
Thickness of Encapsulant (mm)
0.6
Figure 5.13. Effect of encapsulant thickness on the concentration of congener #110 in room air due to
emissions from the encapsulated source — Case 1: Lacquer-primer
0.30
0.25
0.20
to
0.00
Concrete / Epoxy-waterborne
•t= 100 days
•t= 1000 days
0.0 0.2 0.4
Thickness of Encapsulant (mm)
0.6
Figure 5.14. Effect of encapsulant thickness on the concentration of congener #110 in room air due to
emissions from the encapsulated source — Case 2: Epoxy-waterborne
60
-------
5.3.7 Effect of Contaminant Concentration in the Source
For a given source and a given encapsulant, the initial concentration in the source affects the average
concentration in the encapsulant layer, the concentration at the exposed surface of the encapsulant, and the
concentration in room air in a similar manner, i.e., linear relationships exist in all the cases. (Figures 5.15
through 5.20). All the simulation results are for congener #110.
400
•53 SOD -H
0)
00
£
I
•Lacquer primer
•Epoxy-waterborne
200
100
200 400 600 800 1000 1200
Initial Concentration in Source (u.g/g)
Figure 5.15. Average concentration of congener #110 in the encapsulant layer (average C2) as a
function of initial concentration in the source (t = 100 days)
160
"55 120
u
0)
00
s
SJ
80
40
0 -I
•Lacquer primer
• Epoxy-waterborne
200 400 600 800 1000
Initial Concentration in Source (u.g/g)
1200
Figure 5.16. Average concentration of congener #110 in the encapsulant layer (average C2) as a
function of initial concentration in the source (t = 1000 days)
61
-------
400
"53 300
•Lacquer primer
•Epoxy-waterborne
200 400 600 800 1000 1200
Initial Concentration in Source (u.g/g)
Figure 5.17. Concentration of congener #110 at the exposed surface of the encapsulant layer [C2(x
L2)] as a function of initial concentration in the source (t = 100 days)
120
"55 90
•Lacquer primer
•Epoxy-waterborne
0 200 400 600 800 1000 1200
Initial Concentration in Source (u.g/g)
Figure 5.18. Concentration of congener #110 at the exposed surface of the encapsulant layer [C2(x
L2)] as a function of initial concentration in the source (t = 1000 days)
62
-------
8.0
6.0
.E 4.0
c
o
•Lacquer primer
•Epoxy-waterborne
re
4-1
0)
u
c
o
2.0
o.o ^
0 200 400 600 800 1000 1200
Initial Concentration in Source (u.g/g)
Figure 5.19. Contribution of the encapsulated source to the concentration of congener #110 in room
air as a function of initial concentration in the source (t = 100 days)
3.0
c
o
4J
£
4-1
I
O
•Lacquer primer
•Epoxy-waterborne
200 400 600 800 1000 1200
Initial Concentration in Source (u.g/g)
Figure 5.20. Contribution of the encapsulated source to the concentration of congener #110 in room
air as a function of initial concentration in the source (t = 1000 days)
63
-------
5.4 Ranking the Encapsulants
5.4.1 Performance Indicators
Barrier models are useful tools for ranking the relative performances of the encapsulants once their
material/air partition coefficients and solid-phase diffusion coefficients are obtained. Three indicators were
used to compare the performance of the encapsulants:
• The average concentration of the contaminant in the encapsulant (C2).
• The concentration of the contaminant at the exposed surface of the encapsulant (C2atx = L2).
• The concentration of the contaminant in room air due to emissions from the encapsulated source (Ca).
The first indicator is a measure of the level of PCBs in the encapsulant layer. The second indicator is for
surface contamination, which is linked to wipe sampling. The third indicator is an estimate of the
contribution of the encapsulated source to PCB contamination in indoor air. For selecting encapsulants,
these values should be as small as possible. Practically, the third indicator is often of primary concern.
5.4.2 Input Parameters for the Barrier Model
Parameters for PCB congener #110, the most abundant congener in Aroclor 1254, were used to rank the
encapsulants. The material/air partition coefficients and the solid-phase diffusion coefficients presented in
Table 5.4 were calculated using Equations 4.2 and 4.3 and data in Table 4.3. Note that the units for the
diffusivity have been converted from (m2/h) to (m2/s). Other model parameters are from Table 5.2.
Table 5.4. Material/air partition coefficients (Kma) and solid-phase diffusion coefficients (Dm) for
congener #110 used for ranking the encapsulants
ID
01
02
03
04
05
06
07
08
09
10
Encapsulant
Acrylate-waterborne
Acrylic-latex enamel
Acrylic-solvent
Epoxy-low VOC
Epoxy-no solvent
Epoxy-waterborne
Lacquer primer
Oil enamel
Polyurea elastomer
Polyurethane
Kma
(dimensionless)
4.82 x 107
7.33 x 107
5.50 x 107
2.24 x 107
1.61 x 107
1.73x 107
4.85 x 107
4.62 x 107
2.97 x 107
4.21 x 107
Dm
(m2/s)
6.56 x lO'14
2.36 x lO'14
2.77 x 10'14
5.06 x lO'15
2.54 x 10'16
1.12x 10'15
1.28x 10'14
6.09 x 10'14
1.81 x 10'13
1.50x 10'14
64
-------
5.4.3 Ranking the Encapsulants Based on Absolute Concentrations
Figures 5.21 through 5.23 rank the ten encapsulants based on the three performance indicators described in
Section 5.4.1. The rankings based on the second and third criteria are similar to the rankings from the
experimental results, but the ranking based on the average concentration in the encapsulant showed a
different pattern. As discussed in Section 5.3.2, the accumulation of contaminant in the encapsulant is
controlled by two factors, i.e., the gain due to the contaminant flux from the source and the loss due to
emissions to room air. Thus, a good encapsulant that significantly reduces the concentrations of PCBs at the
exposed surface and in the indoor air does not necessarily perform well in lowering the average
concentration in the encapsulant layer.
)
tM
o
a)
o>
5
a>
//////+
vy
Figure 5.21. Ranking of encapsulants by the average concentration in the encapsulant layer
(Average C2)
(For congener #110; t = 500 days; initial concentration in source = 100 (ig/g)
65
-------
Figure 5.22. Ranking of encapsulants by the concentration at the exposed surface of the encapsulant
layer [C2 (x=L2)]
(for congener #110; t = 500 days; initial concentration in source = 100 (ig/g)
Figure 5.23. Ranking of encapsulants by the air concentration due to emissions from the encapsulated
source
(For congener #110; t = 500 days; initial concentration in source = 100 (ig/g)
66
-------
5.4.4 Ranking the Encapsulants Based on Percent Reduction of Concentrations
Sometimes it is more useful to rank the encapsulants based on percent reduction of the PCB concentrations
as compared to the not-encapsulated source. Figure 5.24 shows the concentration profiles of congener #110
in the not-encapsulated source (concrete). The average concentration in the top layer (0.1 mm thick) is 19.7
ug/g; the concentration at the exposed surface is 13.2 ug/g; the air concentration is 0.38 ug/m . The percent
reductions are shown in Tables 5.5 through 5.7. Overall, the epoxy coatings perform well in keeping the
concentration low at the exposed surface and in the room. The acrylic coatings performed poorly.
As shown in Table 5.5, a good encapsulant, which effectively reduces the PCB concentration at the exposed
surface and the contribution to air pollution, does not necessarily perform well in keeping the average
concentration low in the encapsulant layer because a good encapsulant reduces the PCB loss due to
emissions and, thus, facilitates the accumulation of PCBs in the encapsulant layer.
The results shown in Tables 5.5 through 5.7 also show that selecting the wrong encapsulant may make the
contamination worse (i.e., negative percent reduction). A key factor that determines the distribution of PCBs
between the source and encapsulant is the partition coefficient between the two phases. PCB molecules tend
to migrate from the source to, and concentrate in, the encapsulant if the latter has a large material/air
partition coefficient, which leads to a large material/material partition coefficient (Equation 4.4).
120
100
80
•~ 60
u
40
20
012345
x(mm)
Figure 5.24. Concentration profiles for congener #110 in not-encapsulated concrete at t = 500 days
(The blue dotted line separates the top layer (0.1 mm thick) from the rest of the source)
67
-------
Table 5.5. Ranking the encapsulants by percent reduction of the average concentration in the top
0.1 mm of the layer, i.e., the thickness of the encapsulant
(For congener #110; initial concentration in source =100 ug/g; t = 500 days)
Encapsulant
Polyurea elastomer
Epoxy-low VOC
Oil enamel
Polyurethane
Acrylate-waterborne
Epoxy-waterborne
Epoxy-no solvent
Lacquer primer
Acrylic-solvent
Acrylic-latex enamel
% Reduction
50.7%
33.3%
21.3%
18.4%
18.2%
15.7%
7.8%
5.6%
2.7%
-29.5%
Rank
1
2
3
4
5
6
7
8
9
10
Table 5.6. Ranking the encapsulants by percent reduction of the concentration at the exposed
surface
(For congener #110; initial concentration in source =100 ug/g; t = 500 days)
Encapsulant
Epoxy-no solvent
Epoxy-waterborne
Epoxy-low VOC
Polyurea elastomer
Polyurethane
Oil enamel
Lacquer primer
Acrylate-waterborne
Acrylic-solvent
Acrylic-latex enamel
% Reduction
96.1%
84.5%
55.2%
28.2%
0.0%
-12.1%
-15.4%
-17.1%
-33.3%
-79.2%
Rank
1
2
3
4
5
6
7
8
9
10
68
-------
Table 5.7. Ranking the encapsulants by percent reduction of the concentration in room air
(For congener #110; initial concentration in source =100 ug/g; t = 500 days)
Encapsulant
Epoxy-no solvent
Epoxy-waterborne
Epoxy-low VOC
Polyurethane
Lacquer primer
Polyurea elastomer
Acrylic-solvent
Oil enamel
Acrylate-waterborne
Acrylic-latex enamel
% Reduction
89.8%
62.7%
16.7%
1.0%
0.8%
-0.9%
-1.1%
-1.2%
-1.3%
-2.0%
Rank
1
2
3
4
5
6
7
8
9
10
5.5 Limitations of Mathematical Modeling
The simulation results presented above used the three performance indicators: the average concentration in
the encapsulant layer, the concentration at the exposed surface of the encapsulant later, and the contribution
of the encapsulated source to the concentration in air. They were used to better understand the general
behavior of encapsulated sources and compare the relative performances of the encapsulants. These
indicators are difficult to measure in the real-world situations.
The partition and diffusion coefficients used as the input of the barrier model were rough estimates. The
average RSD was 35% for the partition coefficients and 55% for the diffusion coefficients.
The simulations were conducted by assuming that the PCB concentration in the source is uniform initially.
In the real world, a concentration gradient may exist in many PCB-contaminated building materials.
Although the barrier model was developed based on mass transfer theories, its long term predictions have
not been validated by experimental data. Thus, the simulation results presented above should be considered
semi-quantitative and can only compare the relative performances of the encapsulants.
69
-------
6. Discussion
6.1 Effectiveness and Limitations of the Encapsulation Method
The experimental data (Sections 4.1.2 and 4.2.4) and the results of mathematical modeling (Sections 5.4.3
and 5.4.4) showed that selecting high-performance encapsulants can effectively reduce the PCB
concentrations in the encapsulant layer, at the exposed surfaces, and in indoor air. On the other hand, the
encapsulation method has its limitations. As shown in Figure 4.9, when the source contained 13000 ug/g
PCBs, the PCB concentrations in the wipe samples for the encapsulated panels ranged from 10.1 to 584
ug/100 cm2 depending on the encapsulant used. Thus, if the goal is to keep the PCB concentration in the
wipe sample below 10 ug/100 cm2, only one encapsulant barely met the requirement.
Estimating the upper limit of the PCB concentration in the source for successful encapsulation is more
difficult than it appears because several factors must be considered. These include selection of the
performance criteria and safety factor, the properties of the encapsulant (e.g., the resistance to PCB
migration and the thickness of the coating), and the properties of the source (e.g., partition coefficient). If
the ultimate goal is to control the PCB concentration in room air, the area of the source and the
environmental conditions (e.g., ventilation rate and the presence of other sources) should also be considered.
Depending on the mitigation goals, the performance criteria can be the PCB concentration in wipe samples,
the average PCB concentration in the layer of the encapsulant, the PCB concentration at the exposed,
encapsulated surface, and the contribution to the PCB concentration in room air. Among these criteria, wipe
sampling is the easiest to implement. Use of other criteria relies heavily on mathematical modeling. It
should be noted that the concentration in the wipe sample is closely related to, but not the same as, the
concentration at the exposed surface because the solvent used for wipe sampling may penetrate into some
substrates.
According to the results of mathematical modeling, the average concentration in the encapsulant, the
concentration at the exposed surface, and the concentration in room air all showed linear relationships with
the initial concentration in the source (Section 5.3.7). Such linear relationship should also apply to wipe
samples because they are related to the concentrations at the exposed surface. Thus, the upper limit of the
encapsulating ability of a coating material can be estimated from Equation 6.1:
CW
C = — (6 1)
max sf w ^'}
where Cmax = maximum allowable concentration of PCBs in the source for effective encapsulation (ug/g)
C = measured PCB concentration in the source where the wipe sample is taken (ug/g)
Wmax = mitigation goal expressed as the maximum allowable PCB concentration in wipe samples
(ug/100 cm2)
Sf = safety factor (dimensionless)
70
-------
W = measured PCB concentration in the wipe sample for the encapsulated source (fig/100 cm2)
Among the four parameters on the right-side of the equation, C and W are either from experimental data or
mathematical modeling, whereas Wmax and Sf are determined by the decision-makers or risk assessors. For
example, for the wipe sampling tests described in Section 4.2, C ~ 13000 ug/g (Table 4.4). If the Epoxy-no
solvent is used as the encapsulant, then W = 10.1 ug/100 cm2 (Figure 4.9). For demonstration purposes,
Wmax is set to 1 ug/100 cm2 and Sf to 3. Then Cmax can be calculated:
If Wmax is relaxed to 10 ug/100 cm2, then Cmax = 4300 (ug/g).
One factor that Equation 6. 1 does not consider is the thickness of the encapsulant. In general, the thickness
of the encapsulant used in the field should be comparable with or greater than the thickness of the
encapsulant used in the laboratory testing from which parameters C and W are obtained.
As we demonstrated in Part 2 of this report series (Guo et al., 2012), the interior surfaces contaminated with
PCBs due to sorption from room air, also known as PCB sinks or "secondary sources", may become
emitting sources after the primary sources are removed. Because of their large quantities, mitigating these
"secondary sources" is difficult and costly. The encapsulation method has the potential to substantially
reduce the cost by not having to remove the contaminated materials from the building.
A disadvantage of using wipe sampling as the performance criteria is that the PCB concentration in the wipe
samples does not correlate to the concentration in room air because the latter is also dependent on the area of
the source, the ventilation rate, and the presence of other PCB sources. As a practical matter, post-
encapsulation monitoring (e.g., wipe and air sampling) is essential for successful encapsulation.
6.2 Selection of Encapsulants
Resistance to PCB migration is one of the key factors for selecting proper encapsulants for PCB sources.
The results of both the sink tests (Section 4.1) and the wipe sampling tests (Section 4.2) showed that the
performances of the ten coating materials were significantly different. Table 6.1 compares the performances
of the ten encapsulants based on the maximum allowable PCB concentrations in the source (Cmax) for
effective encapsulation. The results were calculated by using Equation 6.1 and the wipe sampling data
presented in Figure 4.9.
Overall, the epoxy coatings performed better than the other types of coating materials because they were
more effective in reducing the surface concentrations. The performance of the Polyurea elastomer should be
re-evaluated in future studies because the sink tests and wipe sampling tests yielded difference results
(Figures 4.7 versus Figure 4.9). As a general guideline, the coating materials that have smaller material/air
partition coefficients and smaller diffusion coefficients perform better in reducing the concentrations of
PCBs at the exposed surface and in the indoor air (Figures 5.22 and 5.23). An encapsulant that has a small
material/air partition coefficient also has a small fugacity capacity, resulting in more resistance to PCB
migration from the source. Similarly, an encapsulant that has a small diffusion coefficient impedes the
71
-------
mobility of the PCB molecules, thereby creating a steep concentration gradient, which, in turn, helps reduce
the PCB concentrations at the exposed surface.
Table 6.1. Calculated maximum allowable concentrations in the source for effective encapsulation
with two mitigation goals based on the PCB concentration in wipe samples (Wmax) [al
Encapsulant
Lacquer primer
Acrylic-latex enamel
Oil enamel
Polyurethane
Acrylic-solvent
Acrylate-waterborne
Epoxy-waterborne
Polyurea elastomer
Epoxy-low VOC
Epoxy-no solvent
Maximum Allowable PCB Concentration
in the Source, C^, (jig/g) [bl
For Wmax=l jig/100 cm2
7.4
8.8
14
18
19
19
34
69
120
430
For Wmax= 10 jig/100 cm2
74
88
140
180
190
190
340
690
1200
4300
See Table 3.10 for the accuracy of the wipe sample data.
Results are rounded to two significant digits.
There are many types of coating materials on the market that can potentially be used as encapsulants for
PCB sources. Although the epoxy coatings performed well among the ten coating materials we tested, they
may not be the best encapsulants available. The authors recommend that more types of coating materials be
tested in future studies, including silicon-based coating materials. Polyurea elastomer should also be
included because this study gave inconsistent results.
In practice, several more factors should be considered when selecting proper encapsulants, including
elongation (i.e., elasticity or rigidity), dry-film thickness, hardness, drying or curing time, compatibility with
existing surfaces, and ease of application (Mitchell and Scadden, 2001). Successful encapsulation also
depends on other factors, such as surface preparation and post-encapsulation monitoring (EH&E, 2012).
6.3 Potential Effect of the Weathering of Encapsulants on their Encapsulating Ability
Polymeric materials are the bases of most coating materials. Environmental conditions may cause
degradation, or weathering, of the polymeric materials. The major factors that contribute to material
degradation include ultraviolet (UV) irradiation, moisture, elevated temperature, and temperature
fluctuations. Although the accelerated aging tests described in Section 4.2.5 were not conclusive, no serious
PCB breakthroughs were observed. Given that the intensity of UV irradiation is much weaker and the
72
-------
temperature fluctuations are much smaller indoors than outdoors, the degradation of coating materials in the
indoor environment is expected to be much slower than in the outdoor environment.
Because of the harsh conditions in the ambient environment, encapsulants applied to the exterior surfaces
may deteriorate faster than those applied to the interior surfaces. Thus, post-encapsulation monitoring is
even more important for encapsulating exterior surfaces. For future studies, the effect of weathering should
be investigated by conducting the weathering tests under realistic or simulated outdoor environmental
conditions. Such tests are time-consuming and costly, however.
Another factor to be considered to judge the performance of the encapsulation is the change of the PCB
concentrations in the encapsulant layer over time. If the concentrations increase continuously over time, the
protective effect of the encapsulation may eventually fail. However, the modeling results presented in
Sections 5.3.3 and 5.3.4 suggests otherwise, i.e., the peak concentrations in the encapsulant layer occurred in
a several weeks, followed by a decrease in concentrations over time due to the formation of a concentration
gradient in the source.
6.4 Encapsulating Encapsulated Sources
The performance of an encapsulant may deteriorate over time due to environmental factors such as wearing
and aging. Adding a new layer of encapsulant may improve the protective effect because of the added
thickness (See Section 5.3.6) and coverage of damaged spots. When used in conjunction with the post-
encapsulation monitoring plan, such practice may prolong the protective effect of the encapsulation.
6.5 Effectiveness of Encapsulating Sources with High PCB Content
This study demonstrated that, although some of coating materials we tested performed much better than
others as PCB encapsulants, none of them is truly impenetrable to PCB molecules. Thus, as discussed
above, coating materials alone may not be effective in meeting mitigation goals for sources that have a high
PCB content. However, under certain circumstances, encapsulating sources with high levels of PCBs could
still be beneficial. For example, there may be a substantial volume of caulking that is scheduled for removal
at some mitigation sites. Developing and implementing a remediation plan requires proper planning and
funding. In such cases, encapsulating the caulk may help reduce potential exposures during this time period.
Using encapsulation under such circumstances must be considered as a short-term interim measure.
6.6 Relationship between the Sink Tests and the Wipe Sampling Tests
At first glance, the sink tests and wipe sampling tests are unrelated to each other because they are based on
completely different mass transfer mechanisms. In fact, the two experimental methods are closely related
because of Equation 6.3 (i.e., Equation 4.4):
V
ma-1
where K]2 = material/material coefficient between material 1 and material 2 (dimensionless)
Kma i = material/air partition coefficient for material 1 (dimensionless)
tc. ?\
(6-3)
73
-------
Kma 2 = material/air partition coefficient for material 2 (dimensionless)
The material/air partition coefficient (Kma) is not only a key factor that determines the sorption concentration
in the sink tests, it also determines the partition coefficient between the source and the encapsulant
(Equation 6.3). Thus, an encapsulant with a small material/air partition coefficient (Kma 0 also has a small
encapsulant/source partition coefficient (Ki2), which means a greater resistance to PCB migration from the
source (material 2) to the encapsulant (material 1).
6.7 Study Limitations
This study was limited to laboratory testing with a limited scope. Only ten coating materials were tested.
There are many coating materials that can potentially be used as PCB encapsulants. The test results of this
study may not be applicable to the similar products that were not tested even within the same class of
coatings. One coating material that is currently sold as a PCB encapsulant by a foreign manufacturer was
not tested because of the authors' inability to obtain the product for testing.
This study was narrowly focused on the effectiveness and limitations of the encapsulation method, the
performances of a limited number of encapsulants, and the factors that may affect the performance of
encapsulation. It is not a comprehensive evaluation of the encapsulation method.
The wipe sample data presented in Sections 4.3.4 and 4.3.5 did not meet all the data quality goals. The
uncertainty of the data was between 25% and 50% as estimated based on the accuracy and the precision of
the QC samples. Thus, the data must be considered as semi-quantitative.
The solid/air partition coefficients and solid-phase diffusion coefficients reported in Table 4.3 are rough
estimates. More accurate measurements of these properties for encapsulants are needed because they are the
key parameters that affect the performance of encapsulants. They are also the key input parameters for the
barrier models. For more accurate measurements, the two parameters must be determined separately.
Wipe sampling is a simple and useful way to determine the PCB contamination at surfaces, including
encapsulated surfaces. However, the correlation between the concentrations of PCBs in wipe samples and
the concentrations in the solid material is poorly understood. For example, the difference in porosity of the
surface materials may cause significant difference in the wipe sampling results. To use the wipe sampling
method to monitor the performance of the encapsulants, standardization of the method, such as selection of
the solvent, is needed. Hexane, the commonly used solvent for wipe sampling for PCBs, can destroy some
of the coating materials, causing difficulty in sample analysis.
In Section 5.3, the general behavior of encapsulated sources was evaluated by using a barrier model. The
simulations were made for congener #110, the most abundant congener in Aroclor 1254, with one set of
input parameters. In real world settings, multiple congeners and multiple sets of environmental parameters
should be considered.
74
-------
7. Conclusions
Ten coating materials were tested for their performances as encapsulants for PCBs by using two
experimental methods, i.e., sink tests and wipe sampling tests [Sections 2.1, 2.2 and 2.3]. In general, the
results from the two types of tests yielded similar results. The performances of the encapsulants differed
significantly. Selecting high-performance encapsulants is a key step. Overall, the three epoxy coatings
performed better than the other coatings [Sections 4.1, 4.2, and 6.2]. An encapsulant with a smaller
material/air partition coefficient and a smaller solid-phase diffusion coefficient performs better in reducing
the PCB concentrations at the exposed surface and in indoor air [Sections 4.1.3, 5.3, and 6.2]. Increasing the
thickness of encapsulant (e.g., applying multiple coats) helps reduce the PCB concentration at the exposed
surface and, thus, reduce the contribution of the encapsulated source to the PCB concentration in indoor air.
[Sections 4.2.6.1 and 5.3.6]
Encapsulation can be used as an interim solution to mitigating PCB contamination in buildings [Section
6.1]. It may not be effective in meeting mitigation goals for sources that contain high concentrations of
PCBs such as PCB-containing caulking material because, for a given source and a given encapsulant, the
concentration of PCBs in the encapsulant layer is proportional to the PCB concentration in the source and
because none of the coating materials we tested was truly impenetrable to PCBs. [Sections 5.3.7 and 6.5]
Encapsulation is most effective for contaminated interior surfaces that have large areas and that contain low
levels of PCBs. Because of their large quantities, mitigating these "secondary sources" is difficult and
costly. The encapsulation method has the potential to substantially reduce the cost by not having to remove
the contaminated materials from the building. [Section 6.1]
Barrier models are useful tools for studying the general behavior of encapsulated sources and ranking the
performances of encapsulants. These models complement and supplement the experimental results by
providing three criteria for evaluating encapsulants, i.e., (1) the average concentrations of PCBs in the
encapsulant layer; (2) the concentrations of PCBs at the exposed surface of the encapsulated source; and (3)
the contribution of the encapsulated source to the concentrations of PCBs in indoor air. In most cases, the
second and third criteria are more stringent than the first criterion. The material/air partition coefficient and
the solid-phase diffusion coefficient are two key parameters that link the experimental results to the barrier
models. [Section 5]
Determination of the upper limit of the PCB concentration in the source for successful encapsulation
depends on several factors including the mitigation goals, the properties of the source, the properties of the
encapsulant, and the environmental conditions. A combination of experimental testing and mathematical
modeling is the best approach to determining the limitations of the encapsulation method. Wipe sampling is
the most common method for measuring surface contamination and can be used as a criterion for evaluating
the performances of encapsulants [Sections 6.1 and 6.2].
Equation 6.1 can be used to estimate the maximum allowable concentration of PCBs in the source for
effective encapsulation (Cmax). For example, based on the wipe sampling results presented in Section 4.2.4,
Cmax is estimated to be 430 ug/g, assuming (1) the maximum allowable PCB concentration in the wipe
75
-------
sample is 1 ug/100 cm2, (2) Epoxy-no solvent coating — the most effective encapsulant we tested — is
used, and (3) the safety factor is 3. [Section 6.1]
Although the epoxy coatings performed well in this study, they may not be the best encapsulants available
because many coating materials, including silicon-based coatings, were not tested [Sections 2.1, 6.2, and
6.7]. The two test methods described in this study [Sections 2.2 and 2.3] can be used to screen a wide range
of coating materials.
The results from accelerated aging tests for encapsulated PCB sources were inconclusive. Because the
intensity of the UV light is much weaker and the temperature is much stable indoors than outdoors, the
deterioration of the encapsulant in the interior of the building due to weathering is expected to be much
slower and, thus, the effective encapsulation for the interior of the building is expected to last longer than for
the exterior of the building. In either case, post-encapsulation monitoring is essential. [Sections 4.2.5 and
6.3]
Building owners should be aware of the effectiveness and limitations of the encapsulation method and the
factors that may affect the effectiveness of the method. Selecting high-performance coating material is a key
to successful encapsulation of PCB-contaminated surfaces. A long-term monitoring plan is essential to
ensure the integrity of the seal. [Sections 6.1 and 6.2]
76
-------
8. Recommendations
This study is limited in scope. The authors recommend the following topics for future research.
1. Screen (e.g., by the wipe sampling method) a wide range of coating materials to determine their
encapsulating abilities for PCBs. Candidate coating materials should include all those that have been
used as the PCB encapsulants in the field and all those that are currently marketed as PCB encapsulants.
Polyurea coatings should be included because this study gave inconsistent results.
2. Evaluate the performances of the encapsulants by using realistic source substrates such as masonry. A
major difficulty in conducting such tests is to develop the sources in which the PCBs are uniformly
distributed.
3. Evaluate the encapsulation methods that use more than one type of coatings, such as the use of a primer
or atop coat over the encapsulant. The effectiveness and performance of non-liquid materials should
also be investigated. For example, using solid films (such as flexible metallic tapes) that are
impenetrable to PCBs may allow encapsulation of sources with higher PCB content.
4. Develop experimental methods that can evaluate the effects of weathering on the performance of the
encapsulants. Data from such tests will help evaluate the effectiveness of encapsulating exterior walls.
5. Develop methods to determine the material/air partition coefficients and solid-phase diffusion
coefficients for PCB congeners more accurately to reduce the uncertainty in the predictions by the
barrier models. These parameters are also essential for using the source and sink models.
6. Develop an integrated modeling framework for PCBs in buildings to allow decision-makers, school
managers, building owners, and practitioners to evaluate mitigation options, including the effect of
encapsulation on indoor air quality, and to compare the effectiveness of the mitigation methods.
Developing such a framework also sheds light on data gaps. In the past two decades, many mass
transfer models have been developed for emissions from building materials, sorption by interior
surfaces (i.e., the sink effect), and contaminant barriers. While these models are essential tools that have
helped us better understand the movements of PCBs in buildings, none of them can handle the complex
cases presented by the PCB contamination in buildings. For the framework to be useful to a broad range
of users, it should allow for multiple sources and sinks in the room, including layered sources (e.g.,
encapsulated sources) and layered sinks (e.g., painted masonry walls). Other useful simulation
capabilities include PCB migration from primary sources (e.g., caulk) to adjacent materials (e.g.,
masonry walls), removal of primary sources, use of stand-alone air cleaning devices, variable
ventilation rate, and temperature changes.
77
-------
Acknowledgments
The authors thank Drs. John Little and Zhe Liu of the Virginia Polytechnic Institute and State University for
technical consultation on barrier models and for providing the MATLAB code for the barrier model used in
this report; Jacqueline McQueen of EPA's Office of Science Policy for assistance in communications; and
Robert Wright of EPA's National Risk Management Research Laboratory and Joan Bursey of EPA's
National Homeland Security Research Center for QA support.
78
-------
References
(Website accessibilities were last verified on November 15, 2011)
ASTM (2004a). El795-04 Standard specification for non-reinforced liquid coating encapsulation products
for leaded paint in buildings. ASTM International, West Conshohocken, PA.
ASTM (2004b). El797-04 Standard specification for reinforced liquid coating encapsulation products for
leaded paint in buildings. ASTM International, West Conshohocken, PA.
ASTM (2010a). £1494-92(2010) Standard practice for encapsulants for spray- or trowel-applied friable
asbestos-containing building materials. ASTM International, West Conshohocken, PA.
ASTM (201 Ob). ASTM D5116-10 Standard guide for small-scale environmental chamber determinations of
organic emissions from indoor materials/products. ASTM International, West Conshohocken, PA.
ASTM (2010c). ASTM D4060-10 Standard test method for abrasion resistance of organic coatings by the
taber abraser. ASTM International, West Conshohocken, PA.
ASTM (2011). E1796-03(201 l)el Standard guide for selection and use of liquid coating encapsulation
products for leaded paint in buildings. ASTM International, West Conshohocken, PA.
Brown, S. K. (1990). Development of test methods for assessing encapsulants for friable asbestos insulation
products, Journal of Coatings Technology, 62(782): 35-40.
Brown, S. K., and Angelopoulos, M. (1991). Evaluation of erosion release and suppression of asbestos fiber
from asbestos building products, American Industrial Hygiene Association Journal, 52(9): 363-371.
Esposito, M. P., Mccardle, J. L., Crone, A. H., Greber, J. S., Clark, R., Brown, S., Halowell, J. B., Langham,
A., and Mclardish, C. D. (1987). Decontamination Techniques for Buildings, Structures and Equipment,
Noyes Data Corporation, Park Ridge, NJ, 252 pp.
EH&E (2012). Literature review of mitigation methods for PCBs in buildings, prepared for U.S. EPA by
Environmental Health & Engineering, Inc., Report EPA/600/R-12/034.
Fuller, R., Klonne, D., Rosenheck, L., Eberhart, D., Worgan, J., and Ross, J. (2001). Modified California
roller for measuring transferable residues on treated turfgrass, Bulletin of Environmental Contamination and
Toxicology, 67: 787-794.
Guo, Z., Liu, X., Krebs, K. A., Stinson, R. A., Nardin, J. A., Pope, R. H., Roache, N. F. (2011). Laboratory
study of polychlorinated biphenyl (PCB) contamination and mitigation in buildings — Part 1. Emissions
from selected primary sources, EPA/600/R-11/156, U.S. EPA, Office of Research and Development,
National Risk Management Research Laboratory, 127 pp.
79
-------
Guo, Z., Liu, X., Krebs, K. A., Greenwell, D. J., Roache, N. F., Stinson, R. A., Nardin, J. A., and Pope, R
H. (2012). Laboratory study of polychlorinated biphenyl (PCB) contamination and mitigation in buildings
— Part 2. Transport from primary sources to building materials and settled dust, EPA/600/R-1 l/156a, U.S.
EPA, Office of Research and Development, National Risk Management Research Laboratory, 166 pp.
Hu, H. P., Zhang, Y.P., Wang, X.K., and Little, J. C. (2007). An analytical mass transfer model for
predicting VOC emissions from multi-layered building materials with convective surfaces on both sides,
International Journal of Heat and Mass Transfer, 50: 2069-2077.
Kumar, D. and Little, J. C. (2003). Characterizing the source/sink behavior of double-layer building
materials, Atmospheric Environment, 37: 5529-5537.
Little, J. C., Kumar, D., Cox, S. S., and Hodgson, A. T. (2002). Barrier materials to reduce contaminant
emissions from structural insulated panels. Advances in Building Technology, Anson, M., Ko, J. M., and
Lam, E. S. S. (eds.), Elsevier Science Ltd., Vol. 1, pp 113-120.
Mackay, D. and Paterson, S. (1981). Calculating fugacity. Environmental Science & Engineering, 15(9):
1006-1014.
Mackay, D. and Paterson, S. (1982). Fugacity revisited. Environmental Science & Engineering, 16(12):
654A-660A.
Martyny, J. W. (2008). Encapsulation of methamphetamine-contaminated building materials, National
Jewish Medical and Research Center.
http://health.utah.gov/meth/html/Decontamination/AdditionalResources.html
MIC (undated). PCB encapsulation. Midwest Industrial Coatings-WI, LLC.
http://www.midwestindustrialcoatings.com/pcb.html
Mitchell, S. J. and Scadden, R. A. (2001). PCB decontamination methods for achieving TSCA compliance
during facility decommissioning projects. The National Defense Industrial Association Annual
Environmental Symposium, Austin, Texas, 25 April 2001, Technical Paper 0102.
http://www.westonsolutions.com/about/news_pubs/tech_papers/MitchellNDIA01.pdf
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
Pizarro, G. E. L., Dzomak, D. A., and Smith, J. R. (2002). Evaluation of cleaning and coating techniques for
PCB-contaminated concrete. Environmental Progress, 21(1): 47-56.
Robnor Resins (Undated). Robnor Polyurethane Applications.
http://www.robnor.co .uk/index.php/products/polyurethane .html
80
-------
Scadden, R. A. and Mitchell, S. J. (2001). A case study of the continued use of PCB-contaminated concrete
through implementation of the 40 CFR 76130(p) use authorization, 10th Annual RCS, National Convention,
Austin, TX, October 2001. Technical Paper #0204.
http://www.westonsolutions.com/about/news_pubs/tech_papers/ScaddenRCS2001.pdf
TWO Teknik (2011). TWO SCI - Silicon PCB Stop (SPS).
http://www.twoteknik.dk/index.php?content=page&id=47
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). 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/1668.pdf
U.S. EPA (2008b). Test methods for evaluating solid waste, physical/chemical methods, in EPA publication
SW-846, U.S. EPA, Government Printing Office, Washington, D.C.
Willett, L. B. (1972). Barrier coatings to prevent polychlorinated biphenyl (PCB) contamination of silage.
Ohio Agricultural Research and Development Center Research Summary, 59: 28-32.
Willett, L. B. (1973). The evaluation of barrier coatings in farm silos which were PCB-contaminated. Ohio
Agricultural Research and Development Center Research Summary, 69: 36-39.
Willett, L. B. (1974). Coatings as barriers to prevent polychlorinated biphenyl contamination of silage.
Journal of Dairy Science, 57(7): 816-825.
Willett, L. B. (1976). Coatings form effective barriers in PCB -contaminated silos. Ohio Report on Research
and Development in Agriculture, Home Economics, and Natural Re sources. 61(1): 5-6.
Yuan, H. Little, J. C., Marand, E., and Liu, Z. (2007). Using fugacity to predict volatile emissions from
layered materials with a clay/polymer diffusion barrier. Atmospheric Environment, 41: 9300-9308.
81
-------
Appendix A. Evaluation of the Wipe Sampling Method
A.1 Purpose
As described in Section 2.4.3, the wipe sampling method used in this study was developed based on a
modified California roller method. This method requires placing a piece of aluminum foil between the wipe
and roller. The potential loss of PCBs due to the use of the aluminum foil must be evaluated.
Hexane is the most widely used solvent for wipe sampling for PCBs. However, hexane was not suitable for
this study because this solvent may destroy some painted surfaces, so we used 2-propanol instead. To make
certain that 2-propanol had adequate collection efficiency for PCBs, side-by-side comparisons were made
for the two solvents.
A.2 Method
The PCB source was created by mixing tetrachloro-ra-xylene (TCMX, 0.00134% by weight), a commonly
used surrogate compound for PCBs, with an alkyd primer. Ten aluminum panels were painted with the PCB
primer. The area of the source was 60 cm2. After the primer was cured, wipe samples were taken by the
roller method described in Section 2.4.3 with hexane and 2-propanol as the solvents. For each sampling
event, the aluminum foil was extracted separately.
A.3 Results
A. 3.1 Effect of Using Aluminum Foil
As shown in Table A. 1, of the total TCMX collected by the sampling method, the aluminum foil contained
less than 1% TCMX for 2-propanol and less than 3% TCMX for hexane.
Table A.I Effect of using aluminum foil on wipe sampling
Solvent
2-Propanol
Hexane
TCMX in
wipe (ng)
980
1064
852
1097
852
1097
1068
732
1071
1204
TCMX in aluminum foil
Amount (ng)
6.43
8.04
3.92
5.94
1.88
27.8
23.4
18.3
28.2
17.3
Fraction of total
0.7%
0.7%
0.5%
0.5%
0.2%
2.5%
2.1%
2.4%
2.6%
1.4%
82
-------
A. 3.2 Comparison of Solvents
Using the TCMX concentrations in the wipe samples presented in Table A.I, the statistics for the two
solvents were calculated (Table A.2). The t-test yielded a two-tailed p value of 0.519, which means that the
difference between these two solvents is not statistically significant.
Table A.2. TCMX concentrations in wipe samples: comparison of two solvents'"1
Solvent
2- propanol
Hexane
Statistics
n
5
5
Mean
(jig/100 cm2)
1.65
1.76
SD
(jig/100 cm2)
0.20
0.30
RSD
12.0%
17.3%
LaJ The wiped area was 59 cm .
83
-------
Appendix B. Resistance of the Encapsulants to Abrasion
The abilities of the encapsulants to resist abrasion were tested by using the Standard Test Method for
Abrasion Resistance of Organic Coatings by the Taber Abraser (ASTM, 201 Oc) by a commercial paint-
testing laboratory. The test results, summarized in Table B.I, are reported as wear index. The lower the wear
index, the more resistant the coating is to abrasion.
Table B.I. Wear Indices for the 10 coating materials tested.
Encapsulant
ID
09
05
06
10
02
04
07
08
01
03
Encapsulant
Polyurea elastomer
Epoxy-no solvent
Epoxy-waterborne
Polyurethane
Acrylic-latex enamel
Epoxy-low VOC
Lacquer primer
Oil enamel
Acrylate-waterborne
Acrylic-solvent
Wear Index
18
70
119
120
130
136
283
298
515
893
SD[al
1.4
0.7
19.1
17.0
24.7
7.1
18.4
17.7
79.9
34.6
Ranking
1
2
3
4
5
6
7
8
9
10
1 For duplicate panels.
84
------- |