Flame Retardants in Printed Circuit Boards Appendices December 2014 Updated Draft Report DFE


vvEPA

United States
Environmental Protection
Agency



u.s. ERA

FLAME RETARDANTS IN PRINTED CIRCUIT BOARDS

& fWWWWI

M 3100

!• 9012B

APPENDICES

December 2014

UPDATED DRAFT REPORT

A-l

-------
FLAME RETARDANTS IN PRINTED CIRCUIT
BOARDS: APPENDIX A

Yamada, Takahiro; Striebich, Richard. Open-
burning, Smelting, Incineration, Off-gassing of
Printed Circuit Board Materials Phase I Flow
Reactor Experimental Results Final Report.
Environmental Engineering Group, UDRI. August
11,2008

A-2

-------
Open-burning, Smelting, incineration, off-gassing of printed circuit
board materials, Phase I Flow Reactor Experimental Results
Final Report (August 11,2008)

Takahiro Yamada and Richard Striebich

Environmental Engineering Group
University of Dayton Research Institute

300 College Park KL102
Dayton, OH45469

A-3

-------
1. Introduction and Background

In this study we investigated the controlled exposure of various printed circuit boards (PCBs)
laminates to high temperature conditions. This work, combined with more realistic combustion
studies (Cone Calorimeter) will allow us to better understand the mechanisms of PCB thermal
destruction. This information will be used to evaluate existing and candidate flame retardants
used in the manufacturing of the PCBs. The combination of better controlled experiments with
actual combustion experiments will allow researchers and manufacturers to determine whether
candidate flame retardant material is better or worse than the existing formulations.

2. Experimental Setup

Figures 1 and 2 show an overview photo and a schematic of the experimental setup designed for
the project. A straight 28.5" long quartz reactor with 9.5><7mm o.d.xi.d. (QSI, Fairport Harbor,
OH) was used for pyrolysis experiments, and same reactor with 3* 1mm i.d.xo.d. stem attached
to the straight main reactor at 5 Vi" from the reactor inlet end (QSI, Fairport Harbor, OH, custom
order) was used for the oxidation experiments. The narrow tubing was installed to introduce
oxygen for the combustion tests. Figure 3 shows detailed design of the modified reactor. New
reactor was used for each sample for pyrolysis experiments (100% N2). The same reactor was
used for the experiment with 10 and 21% O2 and N2 as bath gas. The samples were gasified
under pyrolytic condition for all experiments as seen in Figure 2. Blank experiments were
performed for each experiment, both pyrolysis and oxidation, to ensure that there was no carry
over from the previous experiments. The reactors were installed into 3-zone temperature
controlled furnace, diameter and 24" length, SST-0.75-0-24-3C-D2155-AG S-LINE
(Thermocraft, Winston-Salem, NC.).

A-4

-------
Figure 1. Overview of experimental Setup

Figure 2. Schematic of experimental setup used for this project

Figure 3. Detailed schematic of reactor inlet

Figure 4 shows the reactor temperature profiles at 300, 700, and 900°C. Based on the profiles,
effective length was determined to be 18" (from 6" to 24"). The effective length was used to set
gas flow rate to maintain 2 sec. of residence time for each temperature. The transfer line between
the reactor and GC oven was heated above 250°C.

A-5

-------
• 300C	¦ 700C	A 900C

1000
800
600
400
200

J

i







A

A





II

II

II

II

¦l

¦l

¦l

l«

|l

II

II

'l

'l











a

L

l

1

1

I

I

I1

1

1



k























































i

l















































1

1





























































¦

<

1

»

>













































~



i.

4























































l"'























































) '

L

6 12 18 24

L 30

The

beginning of
heat element

Distance from Inlet Reactor End (in)





The end of
heat element

Figure 4. Reactor temperature profiles for 300, 700, and 900°C

As shown in Figure 5, samples were gasified using a pyroprobe, CDS 120 Pyroprobe (CDS
analytical Inc., Oxford, PA). The sample (circuit board laminate) was cut into a small piece, 1.5
- 2 mm wide x 1cm long, and inserted into quartz cartridge, 3 x4mm i.d.xo.d. 1" length (CDS
analytical Inc. Oxford, PA) as shown in Figure 6. The cartridge was then inserted into pyroprobe
for the gasification. When the sample was gasified, the pyroprobe temperature was increased
from room temperature to 900°C with a 20°C/ms ramp rate and held for 20 sec. at the final
temperature. The gasification process was repeated 3 times to ensure complete gasification. The
exhaust gas was passed through an impinger containing 20mL HPLC grade ultra-pure water
(Alfa Aesar, Ward Hill, MA) in a 40mL amber vial (WHEATON Industries Inc., Millville, NJ).
A small part of gas (lmL/min. flow rate) was introduced to Gas chromatograph / Mass
Spectrometer (HP 5890/5970 GC/MSD, Hewlett Packard, Pasadena, CA). The GC column used
for the analyte separation was DB-5MS, 30m length, 0.25mm i.d., 0.25[j,m thickness (Agilent
J&W, Foster City, CA).

A-6

-------
Figure 5. Pyroprobe Pt filament

Figure 6. Pyroprobe cartridge with sample

-------
3. Experimental Conditions

Table 1 and 2 show the experimental conditions that were investigated in Phase I of the flow
reactor study. For the sample without copper laminate both pyrolysis and oxidation experiments
were performed. The samples with copper laminate were only subject to pyrolysis. Selected
experiments were repeated for pyrolysis at 700°C and 21% O2 at 900°C. The oxygen
concentrations of 10 and 21% were obtained by mixing nitrogen with 50% oxygen. The tables
describe experiments conducted on a "no Flame Retardant" sample (NFR), a conventional
"Brominated Flame Retardant" sample (BrFR), and candidate phosphorus sample (PFR).

Table 1 Experimental condition for the samples without Cu laminate (Unit: °C)

Table 2 Experimental condition for the samples with Cu laminate (Smelting) (Unit: °C),

Table 3 shows N2 and O2 (50%) flow rates for each temperature and oxygen concentration. The
flow rate was set to obtain 2 sec. residence time in the flow reactor, 18" length x 7mm i.d.

Table 3 N2, Q2, and total flow rate used for each experimental condition (Unit: mL/min),

Prior to the flow reactor incineration tests, thermogravimetric analysis (TGA) was conducted to
determine final gasification temperatures. TGA for all samples in N2 and air environments are
shown in Tables A1 to A6 of Appendix A. Table 4 shows initial and final gasification
temperatures for each sample in N2 and air environments. The gasification initial and final
gasification temperatures vary for each sample. Those temperatures were lower when air was
used for the gasification in general. No weight loss was observed over 900°C for all samples;
therefore, pyroprobe final gasification temperature was set to 900°C.

-------
Table 4 Samp

e gasification starting and final temperatures, and its

Approx. Starting

Weight Loss (%)

Temperature (°C)

Temperature (°C)

Phosphorous Flame

Phosphorous Flame

Phosphorous Flame

4.2 Major Combustion Byproduct Analysis

The major peaks of the total ion chromatograms (TIC) were identified for the each flame
retardant sample and experimental condition. Samples were introduced into the GC oven at a
flow rate of lmL/min., and cryogenically trapped at -30°C during combustion tests. After the
sample gasification and combustion, helium was introduced into the system for 3 minutes to
sweep the reactor system and pressurize GC column. The oven was, then, heated at 20°C/min
ramp rate up to 300°C and held 10 minutes. The results are shown in Figure B1 to B27 in
Appendix B. Some of the experiments were repeated to examine the consistency of the
experimental device. The repeatability experiments were conducted for the pyrolysis at 700°C,
and combustion with 21% O2 at 900°C for each of three samples. The results from these
experiments are shown in Figure 3B, 8B, 12B, 17B, 22B, and 27B in Appendix B. Most of the
compounds identified were aromatics. The most prevalent compounds from most pyrolysis and
oxidation experiments were benzene, toluene, xylene and its isomers, phenol, methylphenol and
its isomers, dimethyl phenol and its isomers, styrene, benzofuran and its derivatives,
dibenzofuran and its derivatives, xanthene, naphthofuran and its derivative, naphthalene,
biphenyl, biphenylene, fluorine, phenanthrene/anthracene. Major brominated compounds found
from the brominated flame retardant include bromo - and dibromo-phenols and hydrogen
bromide. Five largest peaks for each sample are listed in Table 5 for each temperature and
oxygen concentration. Phenol, methylphenol, toluene, xylene, and benzene were often observed
as major products. Dibromophenol was observed for brominated flame retardant at low
temperature, and HBr was major brominated compound at the high temperature. Combined with
TIC shown in Appendix B, it is observed that in the pyrolytic environment (100%N2) brominated
flame retardant reduces number of byproducts at all temperatures, especially effective at low

-------
temperature (300°C). In the oxidative environment (10 and 21% O2) the brominated flame
retardant also reduces both number of combustion byproducts and their amount at all
temperatures. Phosphorous flame retardant reduces amount of combustion byproducts.

Increased oxygen level reduces number and amount of combustion byproducts. Increased
temperature also reduces number and amount of combustion byproducts, and byproducts are
decomposed to smaller compounds at the high temperature. Number of brominated compounds
were found at the trace level, and the identification of these compounds is described in Section
4.3. No phosphorous containing combustion byproducts were identified from the major peak of
phosphorous flame retardant combustion test. Phosphorus flame retardant combustion tests at
900C with 21% oxygen were repeated after the completion of a series of combustion tests which
produced skeptical results. When experiments were conducted under this condition initially,
only water was observed with very minor combustion byproduct peaks. When experiments were
repeated later, combustion byproducts were observed. TICs shown in Figure B26 and 27 are
results from the repeated experiments. The reason why only water was observed is still
unknown; however, problems with the mass selective detector (MSD) at that time could have
caused poor sensitivity. Byproducts observed in these most recent experiments were more
consistent with similar conditions and reactant feeds. Table 6 summarizes amount of sample
gasified and its weight loss.

e 5. Major Combustion Byproducts under Different Experimental Conditions

Major Combustion Byproducts (5 largest peaks in this order, top to
bottom) and Remarks

Methylphenol
Toluene
Xylene
Xanthene

Methyl ethylphenol
Methylphenol
Dibromophenol
Toluene

(only mono-ring
aromatics as a major
peaks)

Methylphenol
Dimethylpehnol
Toluene
Benzene

Oxidation
(21%)

Methyl ethylphenol
Bromophenol
Dibromophenol
Tetramethylbenzene

Methylphenol
Dimethylphenol
Toluene
Xylene

Methylphenol
Toluene
Xylene
Benzene

Phenol
Toluene
Benzene
Methylphenol
Methylb enzofuran
(HBr observed)

Methylphenol
Toluene
Benzene
Xylene

-------
Table 5. Major Combustion Byproducts under Different Experimental Conditions (Cont'd)

Temp.
(°C)

Environment

Major Combustion Byproducts (5 largest peaks with this order, top to
bottom) and Remarks





Non-FR

Br-FR

P-FR

700

Oxidation

Phenol

Benzene

Phenol



(10%)

Benzene

Phenol

Benzene





Toluene

Toluene

Toluene





Methylphenol
Styrene

Styrene
Naphthalene
(next biggest is
bromophenol, then
HBr)

Methylphenol
Styrene

700

Oxidation

Benzene

Phenol

Benzene



(21%)

Phenol

Benzene

Phenol





Benzofuran

HBr

Toluene





Toluene

Dibenzofuran

Styrene





Styrene

Naphthalene

Methylbenzofuran

900

Pyrolysis

Benzene

Benzene

Benzene





Toluene

Toluene

Naphthalene





Naphthalene
Biphenylene
Benzofuran

Naphthalene

Styrene

Indene

Toluene

Biphenylene

Anthracene



Oxidation

Benzene

Benzene

Benzene



(21%)

Naphthalene
Benzofuran

Naphthalene
HBr

Naphthalene
Phenanthrene





Toluene

Phenanthrene

Toluene





Biphenylene
(Benzene and
naphthalene are the
major products,
others are minor)

Benzonitrile

Biphenylene

Table 6. Amount of Samples Gasified and Their Gasification Rates

Sample

O2 Cone.
(%)

Temp. (C)

Sample
Loaded(g)

Amount
Gasified (g)

Gasification
% by weight

Remarks

NFR

0

300

0.013644

0.005086

37.3



700

0.013336

0.005013

37.6



0.014391

0.005431

37.7

Duplicate

900

0.013610

0.005175

38.0



10

700

0.012586

0.004722

37.5



21

700

0.013780

0.005072

36.8



900

0.013405

0.004966

37.0



0.012944

0.004566

35.3

Duplicate

NFR w/Cu

0

900

0.022023

0.004382

19.9



A-ll

-------
Table 6. Amount of Sample

Gasified and its Gasification Rate (Cont't

Sample
Loaded(g)

Amount
Gasified (g)

Gasification
% by weight

4.3 Detailed Brominated Flame Retardant Combustion Byproducts Analysis
Product yields

The major products generated at each temperature for each material are readily identified by GC-
MS analysis. However, because the samples after pyrolysis or oxidation are so complex,
additional analysis must be performed to examine the brominated byproducts constituents for
each sample. Since analysis of the products using standards is difficult due to the fact that there is
a thermal reactor in front of the GC-MS, the concentrations of the major compounds were
estimated. At 300°C in 0% oxygen atmosphere, the monobromophenol yield was estimated to be
1.2% of the mass of the board used. This estimate was calculated from the percentage of the
laminate gasified (37% from Table 5), and the area percentage of chromatographic response from
monobromophenol compared to the entire chromatographic run response (3.3%). The yield of the
other major product (dibromophenol) was estimated to be 0.61% of the weight of the board
exposed. These yields of the major products give an idea of the probable yield of the minor
products.

The major products reported for the brominated flame retardants were the mono and
dibrominated phenols. On the trace level (estimated as less than 1% of the total gaseous product
mixture), a wide variety of compounds were formed as shown in Table 7. Various brominated
aliphatic compounds were observed in small amounts, but the majority of compounds observed
were brominated aromatics. Generally aromatic compounds are more stable, so this observation
is appropriate.

-------
Fate of brominated combustion byproducts

It is clear that some of the compounds reported for trace brominated organics were probably
formed as products of incomplete combustion. This can be deduced because bromobenzene was
not observed at 300°C reactor temperature, but was observed in high amounts (on the trace level)
at higher temperatures. We suspect that the bromophenols are relatively stable at 300°C, but do
degrade at higher temperatures to form bromobenzenes and in one case, trace amounts of
bromobenzene diol. Even at reactor temperatures of 900°C in an air atmosphere, there was some
indication of the survival of these compounds through the reactor. At 900°C, the four brominated
compounds that could be observed were bromobenzene, bromobenzene diol, monobromophenol
and dibromophenol. Blank runs (no sample) were conducted between analyses for many of the
samples, and specifically between the 700°C oxidation experiment and the 900°C oxidation
experiment. None of the major or minor compounds were observed in these blank experiments.

Even trace concentrations of brominated compounds were a surprise at these conditions.
Oxidation at 900°C should have been sufficient to completely oxidize the entire sample. It could
be explained as follows: The sample was gasified instantaneously using pyroprobe. Because the
amount of gas generated was relatively large compared to the carrier gas, it might have created
oxygen deficit environment locally, and also there might not be enough time for gasified sample
to be mixed with oxygen. Less surprising was the survival of the bromobenzene and the
bromobenzene diol which were not present at temperatures of 300°C and were present at 700 and
900°C experiments. These clearly were formed as products during their time in the reactor, and
the degradation of these compounds was not completed by the time these compounds escaped the
high temperature reactor. From all this, we have learned that even at 2 seconds residence time in
an air atmosphere, there is a small amount of bromine which will not be converted to HBr. The
great majority of the brominated compounds, at these high temperatures, do convert to HBr.
However, on the trace level, there is good evidence that compounds are surviving the exposure.
This experimental system, because of its small sample size and short sampling time are not
appropriate to observe the formation of brominated dibenzodioxins and brominated
dibenzofurans. These types of compounds will be investigated in the larger scale systems.

-------
Table 7 Identified Brominated Byproducts





Area counts (x10E-06) from the Total Ion Current for each compound





pyrolysis (N2 atmosphere)

oxidation (21% 02 atmosphere)

MW,
g/mol

compound

300

700

900

blank

300

700

900

blank





2-1-2

2-1-4

2-18-3

2-18-2

4-3-2

4-3-4

4-3-6

4-3-5





















120

Br propene

4.9

ND

ND

ND

0.2

0.1

ND

ND

122

Br propane

1.0

ND

ND

ND

ND

ND

ND

ND

136

Br butane

25.5

ND

ND

ND

6.6

ND

ND

ND

172

Br phenol

101.0

84.0

ND

ND

130.0

147.0

31.1

ND

250

Br2 phenol

55.0

27.7

ND

ND

93.0

69.6

7.5

ND

206

Br naphthalene

ND

ND

ND

ND

ND

ND

ND

ND

262

Br dibenzodioxin

ND

ND

ND

ND

ND

ND

ND

ND

246

Brdibenzofuran

ND

ND

ND

ND

ND

ND

ND

ND

156

Br benzene

0.1

4.7

ND

ND

ND

14.0

10.0

ND

234

Br2 benzene

ND

0.0

ND

ND

ND

1.1

1.4

ND

214

Br propyl phenol

3.5

3.4

ND

ND

14.0

0.1

0.2

ND

292

Br2 propyl phenol

ND

ND

ND

ND

ND

ND

ND

ND

290

Br2 propenyl
phenol

2.3

ND

ND

ND

2.1

ND

ND

ND

4.4 Phosphorous Flame Retardant Combustion Byproducts Analysis

With regard to phosphorous-containing trace organic compounds, we were not able to observe,
even on the trace level, any phosphorus containing organic compounds. Several different
phosphorous compounds were selected which were aromatic phosphorus containing compounds,
including phenylphosphine, dimethyl phenylphosphine, phenylphosphinic acid, C3 phenyl
phosphine, phenylphosphonic acid, hydroxyphenylphosphonic acid, and C4 phenylphosphine.
The major ions from these compounds were checked for the phosphorous containing laminate
materials, and none of these compounds were observed, even on the trace level.

The literature suggests that radical capture is not the mechanism of flame retardancy in
phosphorous containing materials as it is with the brominated materials. Levchik and Weil1
report some good information about these flame retardant materials. In our sample, we suspect
that a aminophenyl phosphorous compound was used in the formulation as we do observe, on a
trace level, the compound aniline as one of the compounds formed at 300°C. Since many of the
phosphorous retardants work by forming phosphate on the surface of the material they are
protecting and "crusting" up the surface, we would expect aromatic formation from phenyl
groups in the flame retardant formulation and the phenol degradation to take place. We do
observe more polycyclic aromatic hydrocarbon (PAH) formation in this retardant than in the
brominated retardant. The mechanism by which phosphorous FRs retard flame (surface
complexes and PO2 interaction with H/OH) prohibits incorporation of phosphorus with stable
organic compounds. Thus, the phosphorous compounds could not be observed downstream of
the reactor.

A-14

-------
4.5 Hydrogen Chloride Analysis

During the course of experiments we were informed by the EPA that at least some (if not all) of
the samples contained chlorine. Standard epoxies used for the laminate contain 1000 to 2500
ppm (0.1 to 0.25 wt %) chlorine. Therefore, we also examined if exhaust gas contained hydrogen
chloride. Hydrogen chloride was found from brominated flame retardant pyrolysis and
combustion tests, and phosphorus flame retardant pyrolysis tests. No hydrogen chloride was
found from non-flame retardant pyrolysis and combustion tests. We did not look for chlorinated
organics, such as polychlorinated dibenzodioxin, in these samples as there was an extremely low
possibility of forming these organics at measurable levels with a flow reactor..

4.6 Aqueous Sample Analysis

The aqueous samples collected from combustion tests of BrFRs (w/o Cu) at 900°C with 21%
oxygen, and pyrolysis of BrFRs (w/o Cu) at 900°C, were analyzed for bromine ion concentration.
Results are shown in Table 8 and Figure CI and C2 in Appendix C.

The samples were analyzed using a colorimetric method called Flow Injection Analysis (FIA)' .
In this analysis, bromine ions react with reagents to form a colored complex which absorbs at
590 nm. The absorbance measured at 590 nm is directly proportional to the bromine ion
concentration of the sample. Standards of 1, 2, 5, and 10 ppm are used for comparison to the
sample solutions (R = 0.9995). Figures CI and C2 show the results of these two analyses. The
sample labeled Blank 30 did not generate a peak as would be expected. The sample labeled
BrFRCuP -1 (bromine flame retardant with Cu laminate) produced a negative peak, which was
observed in both runs. It is believed that some other ion in the sample matrix may have reacted
with method reagents to create a colored complex with a lower absorbance than the carrier
solution. A TIC taken at the same time (Figure B9) also showed no HBr and no other
brominated compounds. It is possible that Br reacted with copper in the pyroprobe to form
CuBr2, and it could have been condensed elsewhere on the reactor wall and transfer line. The
aqueous samples from the Br flame retardant without Cu laminate showed bromine ion in it.
Based on the XRF analysis, averaged Br concentration in the flame retardant sample was 6.17%.
The expected Br ion concentration from two brominated flame retardant combustion tests were
14.0 and 13.8 ppm if all bromine converted to HBr. 63 and 51% bromine was recovered as HBr
from the aqueous samples. The TIC taken at the same time (Figure B21 and B22) also
consistently showed a large HBr peak.

Table 8 Aqueous sample analysis for Br ion concentration

Br Ion Concentration (ppm)

Br flame retardant w/o Cu 1st run (BrFR921-l)

Br flame retardant w/o Cu 2nd run (BrFR921-2)

Br flame retardant w/ Cu (BrFCuPl)

After the flow reactor combustion test, Br transport efficiency test was conducted using
tetrabromobisphenol A (TBBPA) (Aldrich, St. Louis, MO) as a Br source. TBBPA was

-------
dissolved into methylene chloride and dried in the quartz cartridge that was used for sample
gasification. TBBPA was gasified in same manner as PCB samples. Reactor temperature was
set at 700°C, and gasified TBBPA was carried by N2 through reactor at the residence time of 2
sec. Sample was purged through a 40cc vial that contained 20cc HPLC grade ultrapure water.
Results were summarized in Table 9. Br recovery rate was 33.2%. At 700°C TBBPA will most
likely decompose to HBr, or dissociated Br atom may react with the quartz reactor tube. The
surface analysis and/or extraction of the reactor and transport line between reactor and vial could
be further performed to elucidate the Br recovery rate if funding situation allows us to do so.
Also our water impinger may not be sufficient to capture all HBr.

Table 9 Br transport test using TBI

»PA as a Br source

Br Introduced
as TBBA
(mg)

Expected Br if all Br
converted to HBr
(ppm)

Br recovered
from aqueous
sample (ppm)

Recovery
Rate as Br
(%)

5. Literature Review and Comparison

Relevant literature data for Br flame retardant circuit board and TBBPA pyrolysis and
combustion experiments was reviewed after the experiment to better understand our
experimental results. Grause et al 4 conducted the pyrolysis of TBBPA containing paper
laminated printed circuit board (PCB). The major constituents and their wt% of TBBA
containing PCB are C (57.0%), H (6.3%), and Br (3.64%). The sample was pyrolised in a quartz
glass reactor. The sample was heated from 50 to 800°C with a heating rate of lOK/min. and a N2
flow of lOOmL/min. The volatile products were gathered in four gas washbottles each containing
50mL of methanol. HBr content was determined by ion-chromatography (IC), and organic
products were analyzed by GC-MS. Methylated phenols and methylated benzene derivatives
were the most prominent degradation products after phenol. Also brominated phenols were
found among the degradation products of TBBA, with main products being 2-bromophenol, 2,4-
and 2,6-dibromophenols, and 2,4,6-tribromophenol. Most of the bromine was released in the
form of HBr (87%), another 14% was bound in organic compounds, and about 1.8% of original
bromine content was left in the residue. The release of the brominated aromatics was completed
below 400°C. However, only 50% of the bromine was released as HBr at this temperature.
Another 37% of HBr was released from the resin between 400 and 700°C. Barontini et al.5'6
investigated thermal decomposition products and decomposition pathways of electronic boards
containing brominated flame retardants using thermogravimetric (TG) FTIR and laboratory-scale
fixed bed tubular batch reactor coupled with GC-MS/FID. The major constituents and their wt%
are C (22.1-27.4%), H (2.0-2.4%), and Br (6.0- 6.9%). The degradation products identified
includes non-brominated aromatics (phenol, biphenyl, anthracene/phenanthrene, dibenzofuran,
dibenzo-p-dioxin, bisphenol A), brominated benzene, phenols, and dibenzofurans and dioxins.

Chien et al. studied behavior of Br in pyrolysis of the printed circuit board waste. Pyrolysis of
the printed circuit board wastes was carried out in a fixed bed reactor at 623-1073K for 30 min.
in N2. Condensable product gases were analyzed using FTIR, and non-condensable gases were
scrubbed with NaOH solution. The main constituents and their wt% are C (52.2%), H (6.11%),

-------
Br (8.53%), and copper (9.53%). Approximately 72.3% of total Br in the printed circuit board
waste was found in product gas mainly as HBr and bromobenzene. Cu-0 and Cu-(0)-Cu species
were observed in the solid residues. No Cu-Br species was found in the solid residue. Barontini

et al. ' also conducted TBBPA decomposition product analysis. The analytical technique
applied was similar to the one they conducted for Br flame retardant containing electronic
boards. Major products formed were HBr, phenol, mono, di, and tribromophenols, bisphenol A,
and brominated bisphenol A.

Our results show small amount of HBr for brominated flame retardant pyrolysis at 700°C, and
oxidation with 21% O2 at 300°C, and large amount of HBr for the oxidation with 10 and 21% O2
at 700°C and 21% O2 at 900°C. Our HBr recovery rate could have been greater, if multiple series
of impingers and more water were used. Also if samples were captured using methanol
impingers and analyzed using GC-MS as Grause et al. performed, instead of cryogenical trap,
more brominated organic could have been identified, even though we had also identified many
brominated organic compounds at the trace level. Experimental setup and analytical procedure
will be reconsidered and redesigned for Phase II experiment for the better sample identification
and bromine mass balance.

In this work, the controlled thermal exposure of flame-retardant and non-flame retardant
laminates was examined. Results for brominated flame retardant laminates showed that
bromophenol and dibromophenol were the main brominated organic products, with estimated
yields of 1.2% for methylbromophenol and 0.67% for the dibromophenol. The responses for
methylbromophenol and Dibromophenol decreased with increasing temperature, and were below
detectable levels for oxygen free experiments. However, oxidation experiments indicated that
even at 900°C, some amounts of organic bromine containing compounds survived. In addition,
bromobenzene and substituted bromophenols were formed at high temperatures, even though
they were not formed at the 300°C exposure (in both oxidation and pyrolysis). It is possible that
these bromophenols and bromobenzenes will be sources for the formation of products in the cone
calorimeter experiments, such as dioxins and furans.

Organic phosphorus compounds were not observed in the reactor exhaust gases during
phosphorus FR experiments. When phosphorus containing flame retardants are used, the product
distribution is similar to the non-flame retardant laminate experiments, in that there is a wide
variety of polycyclic aromatic hydrocarbons (PAHs) such as benzene, toluene, xylene, and
naphthalene. The results from this study suggests that cone calorimeter experiments will
generate a large amount of PAH type compounds for all of the laminate systems but that the
brominated system is likely to yield brominated dioxins and furans because of the relatively high
yields of brominated phenols observed at high temperatures in this study. In addition, the
compounds we should expect in the cone calorimeter are higher yields of methylbromophenol,
dibromophenol, bromobenzene (mono and di) as well as brominated and nonbrominated
fragments of bisphenol A, such as C3 substituted bromophenol, bromomethylphenol and the like.
All of the laminates formed large amounts of phenol and alkyl substituted phenols.

-------
These experiments did not use enough mass of laminate to perform dioxin and furan analysis on-
line. The investigation of these compounds should be performed with larger masses of sample
and using off-line analysis as it is being performed for the cone calorimeter experiments. The lab
scale experiments indicate that even under well controlled conditions, it is difficult to completely
degrade the brominated phenols, even at 900°C. While most of the bromine is converted to HBr,
its conversion is not complete unless very well controlled mixing is available to expose all of the
gaseous products to 21% oxygen.

References:

1.	Levchik, S. V.; Weil, E. D. Fire Sci. 2006, 24, 345-364.

2.	Anagnostopoulu, P. I.; Doupparis, M. A. Anal. Chem. 1986, 58, 322-326.

3.	Clesceri, L. S.; Greenberg, A. E.; Trussell, R. R.; American Public Health Association,
1989, p 4-11.

4.	Grause, G.; Furusawa, M.; Okuwaki, A.; Yoshioka, T. Chemosphere 2008, 71, 872-878.

5.	Barontini, F.; Cozzani, V. J. Anal. Appl. Pyrolysis 2006, 77, 41-55.

6.	Barontini, F.; Marsanich, K.; Petarca, L.; Cozzani, V. Ind. Eng. Chem. Res. 2005, 44,
4186-4199.

7.	Chien, Y.-C.; Wang, Y. P.; Lin, K.-S.; Huang, Y.-J.; Yang, Y. W. Chemosphere 2000,
40, 383-387.

8.	Barontini, F.; Cozzani, V.; Marsanich, K.; Raffa, V.; Petarca, L. J. Anal. Appl. Pyrolysis
2004, 72, 41-53.

9.	Barontini, F.; Marsanich, K.; Petarca, L.; Cozzani, V. Ind. Eng. Chem. Res. 2004, 43,
1952-1961.

A-18

-------
Appendix A
Thremogravimetric Analysis (TGA)

100	;

95

90

85

80

75

70

65

IS000 Cu in nitrogen
IS000 in nitrogen

200

400

600	800

1000

TemperaturejC

Figure Al. TGA in N2 for Non-flame Retardant Sample with and without Cu Laminate

A-19

-------
ISOOO in air

100 200 300 400 500 600 700 800 900

TemperaturejC

Figure A2. TGA in Air for Non-flame Retardant Sample without Cu Laminate

A-20

-------
IS405 Cu in nitrogen
IS405 in nitrogen

1000

TemperaturejC

Figure A3. TGA in N2 for Brominated Flame Retardant Sample with and without Cu

Laminate

A-21

-------
IS405 in air

100 200 300 400 500 600 700 800 900
TemperaturejC

Figure A4. TGA in Air for Brominated Flame Retardant Sample without Cu Laminate

A-22

-------
IS499 Cu in nitrogen
IS499 in nitrogen

200	400	600	800	1000

TemperaturejC

Figure A5. TGA in N2 for Phosphorous Flame Retardant Sample with and without Cu

Laminate

A-23

-------
IS499 in air

100 200 300 400 500 600 700 800 900
Temperature°C

Figure A6. TGA in Air for Phosphorous Flame Retardant Sample without Cu Laminate

A-24

-------
Appendix B

Total Ion Chromatogram Obtained from Circuit Board Combustion Byproducts Analysis
Table B1 Chemical Name - Structure Reference Table

A-25

-------
A-26

-------
Table B1 Chemical Name - Structure Reference Table

Anthracene

Acetic Acid

Bromophenol
(one of isomers)

Methyl ethylphenol
(one of isomers)

Hydroxybiphenyl
(one of isomers)

Ethenylnaphthalene
(one of isomers)

Acenaphthylene

Methyl ethylphenol
(one of isomers)

Benzonitrile

Cont'd)

H,C-

OH

OH

Br

HO

/ \

CH(CH3)2

C,H,

\ /



A-27

-------
T1

Figure Bl. Total Ion Chromatogram (TIC) of Non-flame Retardant Sample
under Pyrolysis Condition at 300°C

A-28

-------
Xylene	Methylphenol

Time— >-

Figure B2. Total Ion Chromatogram (TIC) of Non-flame Retardant Sample
under Pyrolysis Condition at 700°C

A h u n d a n c e

TIC: 2-15-2.D

T i in e - - >

Figure B3. Overlaid TIC for Repeated Experiment (Non-flame Retardant Sample

under Pyrolysis Condition at 700°C)

A-29

-------
T i m e - - >

Figure B4. Total Ion Chromatogram (TIC) of Non-flame Retardant Sample under

Pyrolysis Condition at 900°C

A-30

-------
1 8e+07

1 .Se-i-O"?'

1 4e+07

1 .Ses-i-O"?'

SOOOOOO

6000000

4000000

2000000



2.00 4.00 6.00 S.OO 10.00 12.OO 14.OO 16.00 1S.OO 20.00

Figure B5. Total Ion Chromatogram (TIC) of Non-flame Retardant Sample with Cu
Laminate under Pyrolysis Condition at 900°C. Peak identifications are same as above

(Figure B4).

A-31

-------
Methylphenol

Time—>

Figure B6. Total Ion Chromatogram (TIC) of Brominated Flame Retardant Sample

under Pyrolysis Condition at 300°C

A-32

-------
0H Methylphenol
Isomers

Abundance

1.8e+07
1,6e+ 07
1,4e+ 07
1. 2e+ 07
1e+ 07
8000000
6000000
4000000
2000000



2.00 4.00 6.00 8.00 10.00 12. OO 14.00 16.00 18.00 20.00

i o-- >•

Figure B7. Total Ion Chromatogram (TIC) of Brominated Flame Retardant Sample

under Pyrolysis Condition at 700°C

A b u n d a n c e

1 . 8 e + 0 7

1 . 4 e + 0 7

1 . 2 e + 0 7

1 e + 0 7

0 0 0 0 0 0

6 0 0 0 0 0 0

4 0 0 0 0 0 0

2 0 0 0 0 0 0



iUU

T I C : 2 - 1 - 4 . D

T i C : 2 - 1 - 6 . D

2.00 4.00

.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00

T i m e - - >

Figure B8. Overlaid TIC for Repeated Experiment (Brominated Flame Retardant Sample

under Pyrolysis Condition at 700°C)

A-33

-------
'T1

Figure B9. Total Ion Chromatogram (TIC) of Brominated Flame Retardant Sample
with Cu Laminate under Pyrolysis Condition at 900°C

A-34

-------
Methylphen
ol Isomers

OH

Figure BIO. Total Ion Chromatogram (TIC) of Phosphorous Flame Retardant Sample

under Pyrolysis Condition at 300°C

A-35

-------
'11 nn >•

Figure Bll. Total Ion Chromatogram (TIC) of Phosphorous Flame Retardant Sample

under Pyrolysis Condition at 700°C

A b u n d a n c e

TIC : 2 - 8 - 4 . D

Figure B12. Overlaid TIC for Repeated Experiment (Phosphorous Flame Retardant
Sample under Pyrolysis Condition at 700°C)

A-36

-------
'Tim

Figure B13. Total Ion Chromatogram (TIC) of Phosphorous Flame Retardant Sample
with Cu Laminate under Pyrolysis Condition at 900°C

A-37

-------
Xylene

Time—>

Figure B14. Total Ion Chromatogram (TIC) of Non-flame Retardant Sample
under 10% 02 Condition at 700°C

A-38

-------
Xylene
Isomers

Time—>

Figure B15. Total Ion Chromatogram (TIC) of Non-flame Retardant Sample
under 21% 02 Condition at 700°C

A-39

-------
A b u si d a si c e

3 e + 0 7

2.5e+07

2 e + 0 7

1 .5 e + 0 7

1 e + 07

5000000

2.00 4.00 6.00 8.00 1 0.00 1 2.00 1 4.00 1 6.00 1 8.0020.0022.0024.0026.00

Figure B16. Total Ion Chromatogram (TIC) of Non-flame Retardant Sample
under 21% 02 Condition at 900°C

A b u si d a si c e

3 e + 0 7

2 .5 e + 0 7

2 e + 0 7

1 .5 e + 0 7

1 e + 07

5000000

TIC : 3-1 8-4 .D

T IC : 3-18-7.D

^			A	A	j	

2.00 4.00 6.00 8.00 1 0.00 1 2.00 1 4.00 1 6.00 1 8.0020.0022.0024.0026.00

f s m e - - >

Figure B17. Overlaid TIC for Repeated Experiment (Non-flame Retardant Sample under

21% 02 Condition at 900°C)

A-40

-------
OH

Figure B18. Total Ion Chromatogram (TIC) of Brominated Flame Retardant Sample

under 21% 02 Condition at 300°C

A-41

-------
Figure B19. Total Ion Chromatogram (TIC) of Brominated Flame Retardant Sample

under 10% 02 Condition at 700°C

A-42

-------
OH

T i m e - - >

Figure B20. Total Ion Chromatogram (TIC) of Brominated Flame Retardant Sample

under 21% 02 Condition at 700°C

A-43

-------
Tim e—>

Figure B21. Total Ion Chromatogram (TIC) of Brominated Flame Retardant Sample

under 21% 02 Condition at 900°C

Abundance

TIC: 4-3-8.D

Tim e—>

Figure B22. Overlaid TIC for Repeated Experiment (Brominated Flame Retardant Sample

under 21% 02 Condition at 900°C)

A-44

-------
A b u n d a n ¦

2 . 4 e +	0	7

2 . 2 e +	0	7

2 e +	0	7

1 . 8 e +	0	7

1 . 6 e +	0	7

1 . 4 e +	0	7

1 . 2 e +	0	7

1	e +	0	7

8 0 0 0 0	0	0

6 0 0 0 0	0	0

4 0 0 0 0	0	0

2 0 0 0 0	0	0

Xylene
Isomers

CH,

I C : 3 - 1 9 - 2 . D

Methylphenol
Isomers

OH

Dimethylphenol
Isomers

2.00 4.00 6.00 8.00 10.0012.0014.0016.0018.0020.0022.0024.0026.00

Figure B23. Total Ion Chromatogram (TIC) of Phosphorous Flame Retardant Sample

under 21% 02 Condition at 300°C

A-45

-------
Time—>

Figure B24. Total Ion Chromatogram (TIC) of Phosphorous Flame Retardant Sample

under 10% 02 Condition at 700°C

A-46

-------
Figure B25. Total Ion Chromatogram (TIC) of Phosphorous Flame Retardant Sample

under 21% 02 Condition at 700°C

A-47

-------
Figure B26. Total Ion Chromatogram (TIC) of Phosphorous Flame Retardant Sample

under 21% 02 Condition at 900°C

T i m e - - >

Figure B27. Overlaid TIC for Repeated Experiment (Phosphorous Flame Retardant
Sample under 21% 02 Condition at 900°C)

A-48

-------
Appendix C
Aqueous Sample Ion Chromatogram Analysis

rvi.-s 1.. I.-...,-.,, i nnnMir.r mft ¦-.Y-,Tr^iT Tra ainunA rot jrh..r...--i" An.,ivi,. r,

rr7Fa

U

Time 973.63 Second; Amp: 0 Volts

Figure CI. FIA Analysis of Aqueous Samples Run 1

Blank 30: Blank Sample

BrMB 1: Aqueous sample for TBBA standard used for Br mass balance test.

BrMB2: Bromide standard for cross check

BrFR921-l: Aqueous sample for Br flame retardant combustion test at 900°C with 21%
02.

BrFR921-2: Aqueous sample for Br flame retardant combustion test at 900°C with 21%
02, repeated.

BrFRCuPl: Aqueous sample for Br flame retardant with Cu laminate combustion test
at 900°C in pyrolysis.

A-49

-------
Figure C2. FIA Analysis of Aqueous Samples Run 2

Blank 30: Blank Sample

BrMB 1: Aqueous sample for TBB A standard used for Br mass balance test.

BrMB2: Bromide standard for cross check

BrFR921-l: Aqueous sample for Br flame retardant combustion test at 900°C with 21%
02.

BrFR921-2: Aqueous sample for Br flame retardant combustion test at 900°C with 21%
02, repeated.

BrFRCuPl: Aqueous sample for Br flame retardant with Cu laminate combustion test
at 900°C in pyrolysis.

A-50

-------
FLAME RETARDANTS IN PRINTED CIRCUIT
BOARDS: APPENDIX B

Sidhu, Sukh; Morgan, Alexander; Kahandawala,
Moshan; Chauvin, Anne; Gullett, Brian; Tabor,
Dennis. Use of Cone Calorimeter to Estimate
PCDD/Fs and PBDD/Fs Emissions From
Combustion of Circuit Board Laminates. U.S.
EPA and UDRI. March 23, 2009

A-51

-------
USE OF CONE CALORIMETER TO ESTIMATE PCDD/Fs AND PBDD/Fs EMISSIONS
FROM COMBUSTION OF CIRCUIT BOARD LAMINATES

Sukh Sidhu, Alexander Morgan, Moshan Kahandawala,

Anne Chauvin, Brian Gullett, Dennis Tabor
UDRI and EPA
March 23, 2009

The purpose of this study was to use a cone calorimeter to measure emissions from fully
ventilated combustion of printed circuit board laminates. The cone calorimeter (FTT Dual Cone
Calorimeter) was modified in order to allow for isokinetic sampling of the exhaust gas. USEPA
method 23 was used to sample and analyze Polychlorinated Dibenzo-p-Dioxins and Furans
(PCDD/Fs) and Polybrominated Dibenzo-p-Dioxins and Furans (PBDD/Fs) from combustion of
circuit board laminates. The cone calorimeter experiments were conducted at the University of
Dayton Research Institute (UDRI). The exhaust gas samples were extracted and analyzed at the
EPA Research Triangle Park laboratory. This report presents and discusses experimental and
analytical data from both institutions.

BrFR or BFR or BR FR = laminate containing brominated flame retardant
PFR = laminate containing phosphorous based flame retardant
NFR = laminate without a flame retardant

A-52

-------
MATERIAL AND METHODS

Cone Calorimeter

The cone calorimeter is a fire testing instrument that measures the inherent flammability of a
material through the use of oxygen consumption calorimetry [1], It is based on the principle that
the net heat of combustion of any organic material is directly related to the amount of oxygen
required for combustion [2], The cone calorimeter is a standard technique under ASTM E-
1354/ISO 5660 [3, 4] and is commonly used as a fire safety engineering tool. Under the ASTM
E-1354/ISO 5660 method, small samples (100 cm squares up to 50-mm thick) of combustible
materials are burned and a wide range of data can be obtained. Through oxygen consumption
calorimetry, heat release rate data can be obtained and sensors on the cone calorimeter can
measure smoke release, CO/CO2 production rates, mass loss rate and several other flammability
properties such as time to ignition and fire growth rate.

A schematic of the UDRI cone calorimeter apparatus is shown in Figure 1. At the core of the
equipment is a radiant cone heater, hence the name 'cone calorimeter'. A sample is placed at the
center of the cone heater on the sample holder with dimensions of 100 mm x 100 mm. The cone
heater provides a constant heat flux to the sample. Ignition of the sample is provided by a spark
igniter located above the sample. The exhaust gas contains smoke and products of combustion.
The constant ventilation is maintained by the blower. The cone calorimeter mimics a well-
ventilated forced combustion of an object being exposed to a constant heat source and constant
ventilation [5, 6],

-------
Several measurements can be obtained from the cone calorimeter. A load cell continuously
measures the mass loss of the sample as it burns. Gases from the fire are carried past a laser
photometer beam to measure smoke density and to a sampling ring which carries the gases to a
combined CO/CO2/O2 detector. Once the gases from the sampling ring have been analyzed, one
can obtain CO and CO2 production rates as a function of time which can give insight into the
heats of combustion for the material, as well as combustion efficiency. Oxygen consumption is
measured in the exhaust stream using an oxygen sensor (paramagnetic). The heat release rate is
determined from oxygen consumption calorimetry. Temperature and pressure measurements are
also taken at various locations in the exhaust duct.

The Cone calorimeter data collected during a test can reveal scientific information about material
flammability performance. All measured data are defined below:

loser photometer beam
including temperature measurement

Figure 1. Schematic of Cone Calorimeter used at UDRI

A-54

-------
Time to ignition (Tig): Measured in seconds, this is the time to sustained ignition of the
sample. Interpretation of this measurement assumes that shorter times to ignition mean that
samples are easier to ignite under a particular heat flux.

Heat Release Rate (HRR): The rate of heat release, in units of kW/m , as measured by
oxygen consumption calorimetry.

Peak Heat Release Rate (Peak HRR): The maximum value of the heat release rate during the
combustion of the sample. The higher the peak HRR, the more likely that flame will self-
propagate on the sample in the absence of an external flame or ignition source. Also, the
higher the peak HRR, the more likely that the burning object can cause nearby objects to
ignite.

Time to Peak HRR: The time to maximum heat release rate. This value roughly correlates
the time it takes for a material to reach its peak heat output, which would in turn sustain
flame propagation or lead to additional flame spread. Delays in time to peak HRR are
inferred to mean that flame spread will be slower in that particular sample, and earlier time to
peak HRR is inferred to mean that the flame spread will be rapid across the sample surface
once it has ignited.

Time to Peak HRR - Time to Ignition (Time to Peak HRR - Tig): This is the time in
seconds that it takes for the peak HRR to occur after ignition rather than at the start of the test
(the previous measurement). This can be meaningful in understanding how fast the sample
reaches its maximum energy release after ignition, which can suggest how fast the fire grows
if the sample itself catches fire.

-------
Average Heat Release Rate (Avg HRR): The average value of heat release rate over the
entire heat release rate curve for the material during combustion of the sample.

Starting Mass, Total Mass Lost, Weight % Lost. These measurements are taken from the
load cell of the cone calorimeter at the beginning and end of the experiment to see how much
total material from the sample was pyrolyzed/burned away during the experiment.

Total Heat Release (THR). This is measured in units of MJ/m and is basically the area
under the heat release rate curve, representing the total heat released from the sample during
burning. The higher the THR, the higher the energy content of the tested sample. THR can
be correlated roughly to the fuel load of a material in a fire, and is often affected by the
chemical structure of the material.

Total Smoke Release: This is the total amount of smoke generated by the sample during
burning in the cone calorimeter. The higher the value, the more smoke generated either due
to incomplete combustion of the sample, or due to the chemical structure of the material.
Maximum Average Heat Rate Emission (MAHRE): This is a fire safety engineering
parameter, and is the maximum value of the average heat rate emission, which is defined as
the cumulative heat release (THR) from t=0 to time t divided by time t [7], The MAHRE can
best be thought of as an ignition modified rate of heat emission parameter, which can be
useful to rank materials in terms of ability to support flame spread to other objects.

Fire Growth Rate (FIGRA): This is another fire safety engineering parameter, determined by
dividing the peak HRR by the time to peak HRR, giving units of kW/m per second. The
FIGRA represents the rate of fire growth for a material once exposed to heat, and higher
FIGRA suggest faster flame spread and possible ignition of nearby objects [1],

-------
Isokinetic Sampling

In this project, the cone calorimeter was utilized to combust the various circuit board
laminates and collect products released during their combustion. The USEPA method 23 was
used to isokinetically sample a portion of the exhaust gases flowing through the exhaust duct.
The cone calorimeter was modified to allow for the isokinetic sampling device to be inserted into
the exhaust duct.

The main characteristic of isokinetic sampling is that the extraction of the gas sample from
the main gas stream is at the same velocity as the gas travelling through the stack. This sampling
method is easily adaptable and is commonly used to test for many organic pollutants such as
polychlorinated biphenyls (PCBs), dioxins/furans and polycyclic aromatic hydrocarbons (PAHs)
[8], The compounds of interest are retained in a glass fiber filter and Amberlite XAD-2 adsorbent
resin.

Apex Instruments Model MC-500 Series Source Sampler Console and Isokinetic System
were used for this experiment and contained five main components: the source sampler console,
the external vacuum pump unit, the probe assembly, the modular sample case and the umbilical
cables. A picture of the Apex instrument isokinetic source sampling equipment is shown in
Figure 2.

A-57

-------
Modular sample case

Source sampler console

Figure 2. Isokinetic Sampling train used at UDRI

The modular sample case contained a heated box for the filter assembly and a cold box
for the impinger glassware and condenser. The sampling nozzle of the heated transfer line was
inserted into the exhaust duct, which was modified by adding holes into the side to allow for the
device to be inserted. Figure 3 shows the modifications made to the exhaust system of the cone
calorimeter. A picture of the cone calorimeter and the isokinetic sampling system assembly is
shown in Figure 4.

Figure 3. Modification of duct and sampling port of the UDRI cone calorimeter

A-58

-------
Figure 4. Cone calorimeter and isokinetic sampling system assembly

The heated probe connected the nozzle to the filter assembly where the soot was retained.
The mass of the filter before and after sampling was recorded to obtain the mass of soot formed
during the combustion of the samples (see data in the Appendix, Table 1). The filter assembly
was also connected to a condenser followed by an adsorbent trap and a series of four impingers.
The moisture formed in the condenser deposited as droplets in the first empty impinger and
therefore could not be quantified. The adsorbent trap contained about 40 g of hydrophobic resin
XAD-2, glass wool and 100 |iL of surrogate standard solution. The surrogate standard solution

13

contained C12 labeled standards of PCDD/Fs to evaluate the method. Due to lack of standards

13

for PBDD/Fs, no 1JCi2 labeled standards of PBDD/Fs were spiked into the samples prior to
sampling. XAD-2 was used to absorb the soluble organic compounds from the effluent gas. The
second impinger contained about 100 mL of water, the third one was empty and the fourth one
contained about 200 g of silica gel and was connected to a thermocouple. All three impingers
were used to collect any extra moisture in the effluent gas. The mass of silica gel was recorded

A-59

-------
before and after sampling to obtain the mass of moisture content in the effluent gas (see data in
Appendix, Table 1). The third impinger appeared to stay dry throughout the experiment (few
water droplets on the sides could not be quantified). The amount of water in the second impinger
was recorded before and after sampling (see data in Appendix, Table 1) and appeared to
decrease. This might be explained by the fact that some of the water could have been carried
away by the effluent gas and was collected in the fourth impinger with the silica gel.

After assembling the sampling train, the system had to be checked for leaks. Throughout the
runs, the temperature inside the probe and inside the filter was controlled and maintained at
120°C from the source sampler console. The cold box temperature was maintained under 20°C
by adding ice water to it. The pump flow rate was maintained at 0.1104 L/s and the exhaust flow
rate was maintained at 15 L/s throughout the experiment. The flow rate through the probe was
controlled and maintained steady by adjusting the flow rate through the stack and therefore a
pitot tube was not necessary.

After sampling, the filter and soot, as well as the soot in the probe, nozzle and front half of
the filter holder, XAD-2 resin and water from the second impinger were combined for a single
analysis. The filter was placed in container No. 1. Container No. 2 contained the soot deposited in
the nozzle, transfer probe and front half of filter holder as well as all the methylene chloride and
acetone rinses. Container No. 3 contained the same material as container No. 2 with toluene as
the rinse solvent. The water was also placed in a container for analysis and the silica gel was
discarded. After sampling, the duct and exhaust hood were dismantled and thoroughly cleaned
with hexane to avoid any risk of contamination from combustion of one type of circuit board to
the next. The sampling method and sample recovery followed the USEPA method 23 for the

-------
determination of emissions of PCDD's and PCDF's from stationary sources (9). A schematic of
the isokinetic sampling train is shown in Figure 5.

Stack wall

Figure 5. Schematic of isokinetic sampling train

For the first set of experiments (combustion of BrFR laminate), the temperature inside the
stack dropped below 100°C before it even reached the sampling probe. The temperatures below
100°C can lead to condensation inside the stack; therefore, to prevent condensation inside the
stack and ensure proper transport of gaseous organic compounds formed, a heating tape was
wrapped around the stack to maintain the temperature inside the stack between 100°C and 130°C
during combustion. In order to monitor the temperature inside the stack during combustion of the
samples, a thermocouple was placed on the inside wall of the stack right behind the nozzle. Two
other thermocouples were added to the outside wall. Please see Appendix, Table 3 for inside wall
temperature data. Note that for the first set of experiments (BrFR) the cone calorimeter did not
have the heating tape and thermocouples. However, a repeat run was made for the BrFR laminate
which included the heating tape around the stack and thermocouples.

A-61

-------
Samples tested

Three types of circuit board samples were provided: laminates containing brominated
flame retardant, non-halogen flame retardant (Phosphorous- based) and no-flame retardant. The
laminates were very thin (~0.4mm thick) and contained copper strips. They were made of a
mixture of epoxy resin and e-glass [1], The three types of circuit board are summarized in Table
1.

Table 1. Circuit Board Types

Circuit Board
types

Description

Picture

BrFR

Circuit board containing
Brominated Flame
Retardant



HI

NFR

Circuit board without
Flame Retardant

H

PFR

Circuit Board containing
Phosphorous Flame
Retardant

H

A-62

-------
Preparation of Samples

Since the laminates provided were too large to be tested as is in the cone calorimeter, the
samples were cut into roughly 100 cm square pieces for cone calorimeter testing. Samples were
not conditioned in any way prior to testing. Depending upon how the original laminates were
cut, the samples had 1 or 2 copper strips as shown in Figure 6.

Figure 6. Two-strip and one-strip circuit boards

Initially, it was estimated that 6 thin laminates had to be stacked and burned together in
order to reach a temperature inside the duct of about 120°C during combustion (120°C is the
USEPA method 23 recommended transfer line temperature); this was also the maximum number
of laminates per stack for which the exhaust gas flow rate was sufficient to remove the smoke
produced during combustion (if the number of laminates per stack was increased, smoke came
into the lab). The laminate pieces were selected and configured in six layer stacks where 2 x two-
strip laminates and 4 x one-strip laminates where stacked together. The stacking sequence
ensured that each test sample had the same amount of copper metal in similar configuration.
One single one-strip laminate as well as one single two-strip laminate were also burned
separately to determine the effect of copper on burning patterns and smoke emissions. Each

A-63

-------
sample was wrapped in aluminum foil such that only the upper side was exposed to the constant
heat flux. The aluminum foil helped to keep the samples together as they burned (preventing
them from falling from the sample holder) and directed the smoke and flames toward the exhaust
hood. Figure 7 shows a sample wrapped in aluminum foil.

Figure 7. Sample wrapped in aluminum foil

Five runs were conducted in series for each circuit board type where the first three runs
consisted of 6- layer samples and the last two runs consisted of 1 one-strip laminate and 1 two-
strip laminate sample. The combustion products for all five runs were collected for a single
analysis for a given type of circuit board. The initial mass of each sample wrapped in aluminum
foil was recorded for each run and is summarized in Table 2. Table 2 also summarizes the
sequence in which the samples were burned.

A-64

-------
Table 2. Description of Samples

Circuit
Board
Type

Date
sampled

Run

Number
of

laminates

Description (one or two-
strip laminate)

Sample ID





1

6

2 two-strip and 4 one-
strip

Br FR Epoxy Laminate, 6
plies, run 1





2

6

2 two-strip and 4 one-
strip

Br FR Epoxy Laminate, 6
plies, run 2

BrFR

06/05/08

3

6

2 two-strip and 4 one-
strip

Br FR Epoxy Laminate, 6
plies, run 3





4

1

one-strip

Br FR Epoxy Laminate, 1
ply, 1 Cu Strip, run 4





5

1

two-strip

Br FR Epoxy Laminate, 1
ply, 2 Cu Strips, run 5





1

6

2 two-strip and 4 one-
strip

No FR Epoxy Laminate,
6 plies, run 1





2

6

2 two-strip and 4 one-
strip

No FR Epoxy Laminate,
6 plies, run 2

NFR

06/16/08

3

6

2 two-strip and 4 one-
strip

No FR Epoxy Laminate,
6 plies, run 3





4

1

one-strip

No FR Epoxy Laminate,
1 ply, 1 Cu Strip, run 4





5

1

two-strip

No FR Epoxy Laminate,
1 ply, 2 Cu Strips, run 5





1

6

2 two-strip and 4 one-
strip

Non Hal FR Epoxy
Laminate, 6 plies, run 1





2

6

2 two-strip and 4 one-
strip

Non Hal FR Epoxy
Laminate, 6 plies, run 2

PFR

06/17/08

3

6

2 two-strip and 4 one-
strip

Non Hal FR Epoxy
Laminate, 6 plies, run 3

4

1

two-strip

Non Hal FR Epoxy
Laminate, 1 ply, 2 Cu
Strips, run 4





5

1

one-strip

Non Hal FR Epoxy
Laminate, 1 ply, 1 Cu
Strip, run 5





1

6

2 two-strip and 4 one-
strip

Br FR Repeat run 1

(Repeat
BrFR)

06/18/08

2

6

2 two-strip and 4 one-
strip

Br FR Repeat run 2

3

6

2 two-strip and 4 one-
strip

Br FR Repeat run 3





4

1

one-strip

Br FR Repeat run 4





5

1

two-strip

Br FR Repeat run 5

A-65

-------
Sampling

The cone calorimeter experiments were conducted on a FTT Dual Cone Calorimeter
following the ASTM E-1354-04 method at one heat flux (50 kW/m ), but some modifications
were made to the method: the isokinetic sampling system was added to sample the exhaust gas
and the heating tape was wrapped around the duct for the NFR, PFR, BrFR and BrFR (repeat)
samples. A constant heat flux of 50 kW/m was maintained by setting the cone temperature at
about 759°C. Samples were tested in triplicate without frame and grid, with the back side of each
sample wrapped in aluminum foil and an exhaust flow was maintained at 15 L/s. All samples
were tested copper side up [3], The initial and final ambient conditions during the combustion of
samples were recorded and are summarized in Table 3.

Table 3. Ambient conditions during experiment



BrFR

NFR

PFR

BrFR (repeat)

Initial

Final

Initial

Final

Initial

Final

Initial

Final

Temperature (°C)

26.5

27.5

26.5

NA

24

28

24

24

Humidity (%)

46

45

33

32

35

29

35

34

Pressure (mbar)

1088

1088

1084

1084

1091

1089

1087

1086

Each sample was ignited and allowed to burn until the flames disappeared. For the 6-
layer Non Hal FR Laminate run 2 and 3, and Br FR Laminate repeat run 3, the flame had to be
re-ignited shortly after initial ignition. The burning times for each sample as well as the initial
mass, mass burnt and volumes of gas sampled were recorded and are summarized in Table 4.

A-66

-------
Table 4. Data taken during Combustion of Samples

Sample ID

Starting
mass
(g)

Mass
lost
(g)

Total
sampling
time (s)

Volume
sampled
(ft3)

Comments

Br FR Epoxy Laminate, 6 plies, run 1

61.8

19.2

426

10.1

No heating
tape around

cone
calorimeter
duct

Br FR Epoxy Laminate, 6 plies, run 2

62.2

18.5

400

Br FR Epoxy Laminate, 6 plies, run 3

60.4

17.6

374

Br FR Epoxy Laminate, 1 ply, 2 Cu Strips, run 5

11.9

2.5

99

Br FR Epoxy Laminate, 1 ply, 1 Cu Strip, run 4

10.2

2.8

89

No FR Epoxy Laminate, 6 plies, run 1

61.5

16.6

512

12.4

Heating
tape

No FR Epoxy Laminate, 6 plies, run 2

64.5

15.9

622

No FR Epoxy Laminate, 6 plies, run 3

63.8

17.6

534

No FR Epoxy Laminate, 1 ply, 2 Cu Strips, run 5

12.6

3.4

129

No FR Epoxy Laminate, 1 ply, 1 Cu Strip, run 4

11.0

3.5

110

Non Hal FR Epoxy Laminate, 6 plies, run 1

63.3

14.3

670

13.9

Heating
tape; Run 2
and 3 were
re-ignited
after 4 min

Non Hal FR Epoxy Laminate, 6 plies, run 2

64.3

14.9

668

Non Hal FR Epoxy Laminate, 6 plies, run 3

64.5

13.8

652

Non Hal FR Epoxy Laminate, 1 ply, 2 Cu Strips,
run 4

12.6

2.2

179

Non Hal FR Epoxy Laminate, 1 ply, 1 Cu Strip,
run 5

11.0

2.8

145

Br FR Repeat run 1

61.64

19.1

360

10.5

Heating
tape; Run 3
was re-
ignited
after 1 min

Br FR Repeat run 2

60.03

18.5

300

Br FR Repeat run 3

61.25

18.7

300

Br FR Repeat run 4

10.65

1.3

60

Br FR Repeat run 5

12.15

3.4

60

All conditions during the combustion of the samples and collection of organic compounds are
summarized in Table 5.

A-67

-------
Table 5. Summary of Conditions during Combustion of Samples

Parameters

Conditions

Heat Flux (kW/m2)

50

Stack Gas Flow Rate (L/s)

15

Sampling Flow Rate (L/s)

0.1104

Pump Flow Rate (L/s)

0.1104

Probe Temperature (°C)

120

Filter Temperature (°C)

120

Cold Box Temperature (°C)

<20

Cone Temperature (°C)

759

Extraction and Analysis

After sampling, Container No. 1 (filter), Container No. 2 (soot deposited in the nozzle,
transfer probe and front half of filter holder as well as all the methylene chloride and acetone
rinses), Container No. 3 (same material as container No. 2 with toluene as the rinse solvent), and
an another container containing the XAD-2 and glass wool were sealed and recorded on a chain
of custody form. All containers were sent to the EPA Research Triangle Park laboratory for
extraction and analysis.

The EPA Research Triangle Park laboratory received the samples from UDRI and
confirmed them against the chain of custody form. The samples had been spiked at UDRI with
PCDD/F pre-sampling spikes to confirm the sampling process. The samples were spiked again
just before extraction with PBDD/F surrogates and internal standards for both the PCDD/F and

A-68

-------
PBDD/F. The samples were then extracted with methylene chloride for 3.5 hours and then with
toluene overnight. The cooler methylene chloride extraction is used in low light conditions to
extract the majority of the brominated compounds due to concerns that they could degrade due to
light exposure, the higher extraction temperature of toluene, and longer extraction times. The
toluene extraction procedure was used to ensure that the standard method of extraction (EPA
Method 23 for Dioxin Analysis) was also completed. After extraction, the extracts were
concentrated with a Snyder column and then filtered. The final volume was 1 milliliter. The
extracts were very dark so only one quarter of the extract was used for further clean-up and
analysis. Equal portions of the methylene chloride and toluene extracts were combined and
diluted with hexane for the clean-up. The extracts were then processed through acidic, neutral,
and basic silica gel, and then adsorbed onto basic alumina and washed with dilute methylene
chloride in hexane. The target compounds were then transferred to carbon/celite with 50/50
methylene chloride/hexane, washed with benzene/ethyl acetate and then eluted from the carbon
celite with toluene. The final fraction was concentrated to 100 microliter and analyzed with high
resolution gas chromatography/high resolution mass spectrometry [10],

The samples were analyzed using an isotope dilution method where isotopically labeled
internal standards and surrogate standards were incorporated prior to sampling and extraction.
The surrogate standards were spiked prior to sampling and their recoveries gave a measure of the
sampling process efficiency. The internal standards were spiked prior to extraction and allowed
quantifying the PCDD/Fs and PBDD/Fs present in the samples. According to the USEPA
method 23, recoveries of the pre-extraction standards must be between 40 and 130 percent for
tetra- through hexachlorinated compounds and 25 to 130 percent for the hepta- and

-------
octachlorinated homologues. All recoveries for PCDD/Fs pre- sampling surrogate standards must
be between 70 and 130 percent [9], Percent recovery limits for PBDD/Fs are not available at the
moment. Overall, it was found that PCDD/Fs pre-sampling and pre-extraction surrogate standard
recoveries fell within the acceptable range (see Appendix 2 for recoveries data). Standard

12

recoveries never fell below the lowest limit, but for the isotopes 13C 2,3,7,8 - TeCDF in the
BrFR run and 13 C12 1,2,3,4,7,8,9 - HpCDF in the PFR run, the percent recovery was slightly
above the highest limit, which means that there was a possibility of breakthrough in the sampling
train.

A blank run sample was also analyzed for PCDD/Fs and PBDD/Fs analysis to demonstrate that
no contamination was contributed by laboratory instruments (see Appendix 2 for data).

RESULTS AND DICUSSION

CO/CO2 production/ O2 consumption data

The gas sampled in the sampling ring was analyzed by a CO/CO2/O2 detector which
allowed measurement of CO/CO2 production rates and O2 consumption rate as a function time.
The total production rates and consumption rates per initial sample mass are presented in Table
6. Note that for the repeat run for BrFR samples, CO/CO2/O2 data is not provided because it is
not affected by the temperature of exhaust duct.

A-70

-------
Table 6. Total CO/CO2 production rate and O2 consumption rate data

Sample ID

Total CO2
produced (g)

Total CO2
produced (g)/
starting mass
(g)

Total O2

consumed

(g)

Total CO
produced (g)

Br Epoxy Laminate, 6 plies, run
1

23.7

0.4

18.3

2.7

Br Epoxy Laminate, 6 plies, run
2

23.4

0.4

17.9

2.5

Br Epoxy Laminate, 6 plies, run
3

20.3

0.3

15.1

2.6

Br Epoxy Laminate, 1 ply, 2 Cu
Strips, run 5

8.0

0.7

2.9

0.8

Br Epoxy Laminate, 1 ply, 1 Cu
Strip, run 4

6.9

0.7

2.3

0.7

No FR Epoxy Laminate, 6 plies,
run 1

35.9

0.6

26.6

1.4

No FR Epoxy Laminate, 6 plies,
run 2

39.3

0.6

28.6

2.3

No FR Epoxy Laminate, 6 plies,
run 3

37.4

0.6

28.1

1.7

No FR Epoxy Laminate, 1 ply,
2 Cu Strips, run 5

14.6

1.2

5.4

1.0

No FR Epoxy Laminate, 1 ply,
1 Cu Strip, run 4

14.2

1.3

5.3

1.2

Non Hal FR Epoxy Laminate, 6
plies, run 1

29.2

0.5

20.5

2.7

Non Hal FR Epoxy Laminate, 6
plies, run 2

31.7

0.5

22.5

2.7

Non Hal FR Epoxy Laminate, 6
plies, run 3

30.0

0.5

21.0

2.7

Non Hal FR Epoxy Laminate, 1
ply, 2 Cu Strips, run 4

13.0

1.0

3.7

1.4

Non Hal FR Epoxy Laminate, 1
ply, 1 Cu Strip, run 5

11.2

1.0

3.3

1.5

A-71

-------
PCDD/Fs and PBDD/Fs Data

For each type of circuit board laminates, combustion product samples from five runs
were combined and analyzed to determine total dioxin concentration. The emission levels of
Polychlorinated Dibenzo-p-Dioxins and DibenzoFurans (PCDD/Fs) are reported using both ng
per Kg of laminate and as ng- Toxic equivalent (TEQ) per Kg of laminate. The TEQ
concentration expresses the overall toxicity of a dioxin mixture relative to the toxicity of 2,3,7,8-
TeCDD. Each dioxin congener is assigned a toxic equivalent factor (TEF) value based on its
relative toxicity to the toxicity of 2,3,7,8- TeCDD [11], The WHO 2005 TEF values for all 7
dioxin and 10 furan chemical compounds analyzed are presented in Table 7 [12],

Table 7. Toxic Equivalent Factors of Chlorinated Congeners



2005 WHO (Mammals/Humans)

Isomer.

Toxicity Equiv.



Factor

2,3,7,8 - TeCDD

1

1,2,3,7,8 -PCDD

1

1,2,3,4,7,8-HxCDD

0.1

1,2,3,6,7,8-HxCDD

0.1

1,2,3,7,8,9-HxCDD

0.1

1,2,3,4,6,7,8 -HpCDD

0.01

1,2,3,4,6,7,8,9 - OCDD

0.0003

2,3,7,8 - TeCDF

0.1

1,2,3,7,8 -PCDF

0.03

2,3,4,7,8 - PCDF

0.3

1,2,3,4,7,8-HxCDF

0.1

1,2,3,6,7,8-HxCDF

0.1

2,3,4,6,7,8 - HxCDF

0.1

1,2,3,7,8,9-HxCDF

0.1

1,2,3,4,6,7,8 -HpCDF

0.01

1,2,3,4,7,8,9-HpCDF

0.01

1,2,3,4,6,7,8,9 - OCDF

0.0003

A-72

-------
The total TEQ was calculated by summing the multiplication of each congener
concentration in the flue gas by its corresponding TEF. The congener concentration (in ng/kg)
was calculated from the data obtained from the HRGC/HRMS analysis (in ng/train) and based on
the basis of total sampling as shown:

in a\

Concentration —

\xg'

Total flow rate in duct Total congener in extract (.Ji§/1rain)

Flow through sampling line " Initio! mass of circuit board (kg)

Congeners concentrations below the limit of detection were regarded as zero and reported as less
than limit of detection (
-------
Table 8. Results showing PCDD/Fs concentration in ng- Toxic equivalent (TEQ) per Kg of

laminate in the emission samples from combustion of circuit board samples

Isomer.

TEQ (ng/kg)

PFR Epoxy
laminate

BR FR Epoxy
laminate

BR FR Epoxy
laminate,
repeat run

NFR Epoxy
laminate

2,3,7,8 - TeCDD

-------
Table 9. Results showing PCDD/Fs concentration (in ng/Kg of laminate) in the emission samples

from combustion of circuit board samples

Isomer.

Cone, (ng/kg)

PFR Epoxy
laminate

BR FR

Epoxy
laminate

BR FR Epoxy
laminate,
repeat run

NFR Epoxy
laminate

2,3,7,8 - TeCDD

-------
The results obtained from the analysis of emissions for PBDD/Fs concentrations in the
extracts are presented in Table 10. For the PFR laminates and NFR laminates, no brominated
congener was detected. The OcBDD and OcBDF compounds were not reported for all circuit

13

boards types because OcBDD/F needed separate clean-up and the C12 labeled OcBDD
surrogate standard did not elute from the carbon column during extraction procedure. The data
for the BR FR laminates BrFR (first run and repeat run) were consistent. For the first set of
experiments, it was found that 3213.8 ng PBDD/Fs per kg of laminates was produced. For the
repeat run, it was found that 3389.7 ng PBD/Fs per kg of laminates was produced. No published
data on PBDD/Fs concentrations in ng per kg of combustible material burned where found to
compare the results.

A-76

-------
Table 10. Results showing PBDD/Fs concentration (in ng/Kg of laminate) in the emission

samples from combustion of circuit board laminates

Isomer.

Concentration (ng/kg)

PFR Epoxy
laminate

BR FR

Epoxy
laminate

BR FR Epoxy
laminate, repeat
run

NFR Epoxy
laminate

2,3,7 TrBDD*

ND

24.4

ND

ND

2,3,7 TrBDF*

ND

ND

ND

ND

2,3,7,8 TeBDD

ND

112.4

88.7

ND

2,4,6,8 TeBDF

ND

172.3

173.0

ND

2,3,7,8 TeBDF

ND

855.4

536.6

ND

1,2,3,7,8 PeBDD

ND

ND

ND

ND

1,2,3,7,8 PeBDF

ND

325.1

300.1

ND

2,3,4,7,8 PeBDF

ND

163.7

112.3

ND

1,2,3,4,7,8/1,2,3,6,7,8 HxBDD

ND

ND

ND

ND

1,2,3,7,8,9 HxBDD

ND

ND

ND

ND

1,2,3,4,7,8 HxBDF

ND

107.5

96.1

ND

1,2,3,4,6,7,9 FtpBDD* **

ND

ND

ND

ND

1,2,3,4,6,7,8 HpBDD* **

ND

ND

ND

ND

1,2,3,4,6,7,8 HpBDF

ND

1453.0

2082.9

ND

OcBDD

NR

NR

NR

NR

OcBDF

NR

NR

NR

NR

Total conc. (ng/kg)

-

3213.8

3389.7

-

*Not present in the standard; assignment based on isotope theoretical ratios and retention times of matching internal
standards and native congeners; quantified based on concentration of the congeners of the same bromination level
present in the standard

**Assignment based on the elution order of HpCDD congeners on the DB5 column.

ND= not detected

NR= not reported (OcBDD/F would need separate clean-up; 13C OcBDD did not elute from carbon column)

A-77

-------
Heat release data and fire behavior

The combined cone calorimeter heat release data are shown in Table 11. Data for the 6-
ply laminate stacks was not reproducible in all aspects of heat and smoke release due to erratic
physical effects of burning, which are described below. Data from single ply laminates with one
or two strips was also difficult to compare to each other, since the amount of copper metal had
some effects on the amount of heat released. It should be noted that for the repeat run for BrFR,
heat release data and fire behavior are not provided as they are not impacted by heating of the
exhaust duct.

A-78

-------
Table 11. Combined Heat Release Rate data

Description

Sample
Thickness
(mm)

Time
to
ignition

(s)

Peak
HRR

(kW/m2)

Time
to

Peak
HRR

Time

to
Peak
HRR
-Tig

Average
HRR

Starting

Mass

Total

Mass
Loss

Weight

%

Lost

Total
Heat

Release

Total
smoke

Release

Avg.
Effective

Heat of
Comb.

MAHRE

FIGRA





(s)

(s)

(kW/m2)

(g)

(g)

(%)

(MJ/m2)

(m2/m2)

(MJ/kg)

(kW/m2)



Br Epoxy Laminate, 6 plies, run 1

3.1

12

242

178

166

68

61.9

19.2

31.0

23.8

2394

12.35

93

1.36

Br Epoxy Laminate, 6 plies, run 2

2.9

14

204

222

208

69

62.2

18.5

29.8

23.4

2019

12.63

75

0.92

Br Epoxy Laminate, 6 plies, run 3

3.0

13

237

208

195

63

60.4

17.6

29.1

19.6

2046

11.06

68

1.14

Br Epoxy Laminate, 1 ply, 2 Cu Strips, run 5

0.4

8

171

20

12

53

11.9

2.5

21.0

3.8

449

15.12

83

8.55

Br Epoxy Laminate, 1 ply, 1 Cu Strip, run 4

0.5

10

185

25

15

43

10.2

2.8

27.4

3.2

424

10.94

76

7.39

No FR Epoxy Laminate, 6 plies, run 1

3.1

14

173

240

226

79

61.5

16.6

27.0

35.5

1401

21.40

96

0.72

No FR Epoxy Laminate, 6 plies, run 2

3.3

15

177

250

235

72

64.5

15.9

24.6

37.9

1350

23.83

85

0.71

No FR Epoxy Laminate, 6 plies, run 3

3.2

17

196

288

271

80

63.8

17.6

27.6

37.5

1310

21.37

88

0.68

No FR Epoxy Laminate, 1 ply, 1 Cu Strip, run 4

0.5

13

379

24

11

97

11.0

3.5

31.9

7.2

329

19.98

138

15.77

No FR Epoxy Laminate, 1 ply, 2 Cu Strips, run 5

0.6

15

265

50

35

81

12.6

3.4

27.0

7.4

353

21.46

111

5.29

Non Flal FR Epoxy Laminate, 6 plies, run 1

3.1

190

152

262

72

64

63.3

14.3

22.6

27.1

1310

18.90

57

0.58

Non Flal FR Epoxy Laminate, 6 plies, run 2

3.2

190

134

326

136

72

64.3

14.9

23.2

30.0

1336

20.13

59

0.41

Non Flal FR Epoxy Laminate, 6 plies, run 3

3.2

206

222

230

24

74

64.5

13.8

21.4

28.0

1209

20.33

59

0.96

Non Flal FR Epoxy Laminate, 1 ply, 2 Cu Strips, run 4

0.5

17

104

29

12

46

12.6

2.2

17.4

4.9

283

22.22

41

3.58

Non Flal FR Epoxy Laminate, 1 ply, 1 Cu Strip, run 5

0.5

15

231

29

14

62

11.0

2.8

25.5

4.5

276

15.47

63

7.96

A-79

-------
Along with the heat release data in Table 11, the heat release rate curves are plotted in
Figures 8-10. Each of the laminates had their own fire behavior which is described separately
below.

Brominated FR Epoxy Laminate Fire Behavior

For the 6-ply laminate stacks, the only reproducible part of the heat release phenomena
was the initial ignition and the detection of the 1st HRR peak, given the observed fire behavior of
these samples this correlates nicely. Each of the 6 ply laminate stacks, upon exposure to the
cone heater, began to smoke within 10 seconds of heat exposure, and then the samples quickly
foamed up as a large bubble and ignited. This rapid ignition flashed off quickly and then died
back with some edge burning on the top ply, followed by a decrease in heat release. Then the
underlying material began to ignite which led to a 2nd HRR peak. These flames continued to
grow until all of the remaining plies foamed up and flames began to come out from the sides of
the sample. This rapid flare up led to the final HRR peak between 150 and 250 seconds as
shown in Figure 8. After this rapid flare up the flames began to die down and eventually the
sample extinguished. One sample (HRR-3) actually self extinguished after the 1st HRR peak and
reignited after a brief delay (Figure 8 left), again attesting to the physical effects of burning
laminate stacks which led to irreproducibility in the HRR curves. Final chars were primarily
glass laminate with blackened metal strips. Some soot/char was present on the lower laminates,
but the top laminate was a light grey in color and had very little soot/char carbon present. Due to
the sample foaming late in the fire, the shutters of the cone calorimeter could not be closed at the
end of the test - otherwise the shutters would have crushed the sample residue which would have

A-80

-------
led to a false load cell (weight loss) result which would have affected many other cone
calorimeter measurements. So, after the last flame went out, the sample was allowed to stay
under the cone heater for another 60 seconds to collect good baseline data. This change in
procedure is noteworthy since it may have burned off the residual carbon on the top ply of the
burned laminates since for the single ply laminates, carbon char was found after the sample
extinguished. Another thing to note for these samples is that, after ignition and once the flames
had grown sufficiently, wherever the sample was burning next to copper, the flames were a
bright blue in color, typical for burning of copper salts. The flame color was yellow to orange
where there was no copper.

For the single ply laminates (Figure 9 left) the observed behavior of burning was different
than that observed with the 6 ply laminate stacks. Upon exposure to the cone heater, the sample
rapidly began to smoke, and then quickly foamed up and ignited. The flames grew quickly in
intensity and then rapidly extinguished as the epoxy in this thin sample burned away. Final chars
were black with carbon/soot noted along with blackened Cu metal strips. There does appear to
be some slight difference in HRR behavior for the single and 2 Cu metal strip laminates in that
the single Cu strip sample has two peaks of HRR while the double Cu strip sample has only 1
peak of HRR. As described above, blue flames were seen where the sample was burning next to
the Cu metal strips.

No Flame Retardant Epoxy Laminate Fire Behavior

The fire behavior of laminates with no flame retardant (control) in the cone calorimeter
was very different than that observed for the brominated flame retardant samples. First of all,

A-81

-------
none of the laminates (either 6 ply or single ply) foamed up upon exposure to the cone heater.
Instead, the laminates had a strong tendency to warp and bend up towards the cone heater with
snapping and popping heard right before ignition. This behavior was so pronounced for the 6-
ply laminates that the cone calorimeter shutters could not be closed when the sample
extinguished as the laminate plies had curled up into the space where the shutters would
normally close.

Fire behavior of the 6-ply laminates with the non-flame retardant epoxy began with
smoke being released shortly after exposure to the heat source (about 12 seconds after start of
test) followed shortly thereafter by ignition of the sample. Some blue flames (of lesser blue
color intensity than that seen with the brominated FR epoxy laminates) were observed, but for
the most part the color of the flames were orange-yellow with some smoke/soot observed at all
times. As with the brominated 6-ply stacks, the 6-ply stacks of non-FR epoxy showed
irreproducible fire behavior as the top ply would ignite, settle down in heat release/flame
intensity, and then the second ply underneath would ignite. Sometimes the top ply would
provide sufficient insulation to delay ignition of the underlying plies (see HRR-2 and HRR-3 in
Figure 8 right) and in other cases the top ply would deform so much that most of the underlying
2nd ply would be exposed to the cone heater. With all these physical effects of burning, the HRR
data for this sample showed a lot more scatter different HRR curve shape, as can be seen in
Figure 9 (right). The HRR peak occurred when the bottom 4 plies would finally all ignite at
once, leading to a slow rise in heat release followed by a slow steady decrease in HRR
whereupon the sample finally extinguished. The final chars from these 6-ply laminates showed
very little carbon char; just some soot and the blackened/oxidized copper metal strips.

A-82

-------
For the single ply no FR epoxy laminates (Figure 9, right), the samples smoked, began to
pop and deform (as seen with the 6 ply laminates) and then rapidly ignited and burned out. No
blue flames were observed for these samples when they were burning. As with the 6 ply
laminates, the shutters could not be closed at the end of the test due to laminate deformation.
The final chars were the same as those observed with the 6-ply laminate stacks, with only
fiberglass and blackened metal remaining. Unlike with the single ply brominated FR epoxy
laminate HRR data, there is a lot more difference in HRR behavior of 1 Cu metal strip and 2 Cu
metal strip HRR data for the non-halogenated FR epoxy laminates (Figure 9 right), but the
reason for this major difference is not clear since the observed fire behavior was very similar for
both samples. A likely explanation though is that the amount of Cu metal on the surface affected
the amount of surface available for burning and pyrolysis.

Non-Halogenated Flame Retardant Epoxy Laminate Fire Behavior

Fire behavior for the non-halogenated flame retardant epoxy laminates (assumed to be
phosphorus-based flame retardant) was different than the other two types of epoxy laminates.
Phosphorus-based flame retardants in epoxies tend to be condensed phase char formation
systems, so that when they burn they convert the carbon-based epoxy "fuel" into graphitic-type
protective chars which slow down the rate of mass loss and heat release. Indeed, this type of
behavior was observed for the 6-ply laminate stacks, as the samples did ignite rapidly after
exposure to the cone heater, but they then extinguished and did not re-ignite for another 150
seconds after the 1st initial ignition (see Figure 10 left). When these laminate stacks were
exposed to the cone heater, they smoked and made crackling/popping sounds (caused by

A-83

-------
delamination) within 10 seconds of exposure to the cone heater. Shortly after that, they ignited,
but then the flames died down quickly and the flame went out. The spark igniter was reinserted
and eventually the sample reignited. The sample deformed and curled up towards the cone
heater towards the end of the test such that the shutters could not be closed at the end of the test.
During the burning of the sample, no blue flames were observed, only yellow/orange flames with
smoke were seen. At the edges of the sample and towards the end of the test some white colors
could be seen at the bottom of the flame, which confirms the presence of phosphorus-based
flame retardants. The final chars were black, but the fiberglass could be seen through this black
char, which was more than just soot. The copper metal strips were completely blackened. As
with the other 6-ply laminate stack data, due to the physical effects during burning, the HRR
curve shapes were not very reproducible, but the times to ignition and flameout were
reproducible within the cone calorimeter test % error of about 10%.

For the single ply laminates, the effect of the copper strips was more pronounced than
that seen with the other samples. The sample with only one copper strip rapidly burned off while
the sample with two copper strips did not burn as intensely and took a little longer to burn.
Otherwise the fire behavior of this sample was very similar to that of the 6 ply laminate stacks,
with the sample smoking and cracking right before ignition, and the laminate curling up towards
the cone heater by the end of the test [1],

A-84

-------
250

200

Brominatad FR Epoxy Laminates
With Copper Strips - 8 Ply Stack

No FR Epo• v Laminates ply stack HRR

(-T-

150

100

Sample extinguished (HRR-3) sample relgnrted (HRR-3)

0 100 200 300 400 500 600 TOO

Time ($)

Figure 8. HRR for 6 ply Br Flame Retardant Epoxy Laminate Stacks (left) and No Flame
Retardant Epoxy Laminate Stacks (right).

Br FR Epoxy Laminate 1 ply stack HRR

Mo FR Epoxy Laminate 1 ply stack HRR

40	60

Time (s)

60 80 100 120 140
Time (s)

Figure 9. HRR for 1 ply Br Flame Retardant Epoxy Laminates (left) and 1 ply No Flame
Retardant Epoxy Laminates (right).

A-85

-------
Non-Hal FR Epo Laminate 6 ply stack HRR	Non-Hal FR Epoxy Laminate 1 ply stack HRR

Figure 10. HRR for 6 ply Phosphorous based Flame Retardant Epoxy Laminate Stacks (left) and
HRR for 1 ply Phosphorous based Flame Retardant Epoxy Laminates (right).

Conclusion

Laminates' Fire Behavior and Heat Release Data

There are four major conclusions that can be made about these samples from the observed
physical fire behavior and from the recorded heat release/smoke release measurements:

1) The 6 ply laminate samples showed erratic HRR behavior due to the physical effects of
laminates igniting and curling/foaming/charring at different rates from stack to stack, even
with the same material. This type of behavior would be normal for a non-coherent stack of

A-86

-------
laminates which would have nothing adhering them together and instead would have air gaps
between each ply to allow for additional heat release and secondary fire events to occur.
6-ply laminates showed lower peak HRR compared to single ply laminates. The likely
reason for this is that the underlying laminates pull some heat away from the top laminate
which makes the 6 ply stack act a little bit more like a thermally thick sample than a
thermally thin sample like the single ply laminates. However, it is well known that for the
cone calorimeter that sample thickness affects heat release results, and therefore it is not
surprising that the peak HRR is higher for the single ply laminates when compared to the 6-
ply laminate stacks.

The amount of Cu metal on the surface appears to have a slight effect on time to ignition.
The more Cu metal present, the more likely that time to ignition will be delayed by a few
seconds. This makes sense as the Cu metal can reflect some heat energy back, and, can
conduct some of the heat energy out and away from the epoxy laminate. However, the 2-3
second delay in time to ignition, while seen in all of the samples, isn't significant in regards
to overall fire behavior of these materials. Once the single ply laminates ignite, they rapidly
go to peak HRR and then extinguish as the fuel is rapidly burned off.

Since peak HRR and moment specific data is difficult to compare between samples due to
physical effects of burning, it is better to look at total HR and total smoke when comparing
between samples. By doing this the following trends appear: Brominated FR epoxy has
highest smoke release and lowest total heat release. The non-FR epoxy control has the
highest heat release and middle-level smoke release. The non-halogenated FR epoxy has the
lowest smoke release (although similar to the non-FR epoxy) and middle level total heat
release.

A-87

-------
Since the purpose of these experiments was to generate a total amount of material to burn for
emissions testing, the total smoke and total heat release data indicate that the experiments were
in general a success and that all experiments done did yield a controlled amount of burning
material. So while individual specimens tested may not correlate exactly in regards to specific
moments of heat release, the total amount of fuel burned/smoke released from specimen to
specimen did correlate well, indicating that the cone calorimeter did provide controlled burning
specimens over a total amount of sampling time. This is important for the emissions testing
since the sampling is done over the total amount of sample burned rather than a specific moment
in time of burning [1],

PCDD/Fs and PBDD/Fs emission data

No significant concentrations of PCDD/Fs were found after sampling and analysis of emissions
from the combustion of BrFR laminates containing brominated flame retardant, PFR laminates
containing non-halogen flame retardant (Phosphorous- based), and NFR laminates containing
no-flame retardant. Most targets pollutants were found to be below the limit of detection of the
analysis. The targets that were detected appeared to be a carry over from a standard. The results
obtained from the analysis of emissions for PBDD/Fs concentrations in the extracts confirmed
the presence of pollutants for the combustion of BrFR laminates containing brominated flame
retardant. The laminates contained copper strips which could have promoted the formation of
dioxins in the emissions. No published data on PBDD/Fs concentrations in ng per kg of
combustible material burned was found to compare the results of this study. For the PFR
laminates and NFR laminates, no PBDD/F congener was detected.

A-88

-------
REFERENCES

[1]	Morgan, Alexander B. Cone Calorimeter Analysis of Circuit Board Laminates.

(Personal communication, March 2009)

[2]	Fire Protection Engineering. http://www.wpi.edu/Academics/Depts/Fire/Lab/Cone/

[3]	ASTM E1354 "Standard Heat Method for Heat and Visible Smoke Release Rates for

Materials and Products Using an Oxygen Consumption Calorimeter"

[4]	"ISO/FDIS 5660-1 Reaction-to-fire tests - Heat release, smoke production and mass

loss rate - Part 1: Heat Release (cone calorimeter method)" and "ISO/FDIS
5660-2 Reaction-to-fire tests - Heat release, smoke production and mass loss rate
- Part 2: Smoke production rate (dynamic measurement)"

[5]	Babrauskas, V. Specimen Heat Fluxes for Bench-scale Heat Release Rate Testing.

Fire andMaterials 1995, 19, 243-252.

[6]	Babrauskas, V.; Peacock, R. D. Heat Release Rate: The Single Most Important Variable in

Fire Hazard. Fire Safety Journal 1992, 18, 255-272.

[7]	Duggan, G. J.; Grayson, S. J.; Kumar, S. "New Fire Classifications and Fire Test Methods for

the European Railway Industry" . Flame Retardants 2004 Proceedings, January 27-28,
2004, London, UK, Interscience Communications

[8]	Apex Instruments. Isokinetic Source Sampler (500-Series models), Operator's

Manual.

[9]	USEPA Method 23 "Determination of Polychlorinated Dibenzo-p-dioxins and

Polychlorinated Dibenzofurans from Municipal Waste Combustors"

[10]	Tabor, Dennis. Summary of extraction and analysis procedure (personal communication,
March 2009).

A-89

-------
[11]	WHO, Environmental Health Criteria 205, 1998

[12]	WHO. The 2005 World Health Organization Reevaluation of Human and Mammalian

Toxic Equivalency Factors For Dioxins and Dioxin-Like Compounds. Toxicological
Sciences. 93(2), 223-241 (2006;.

APPENDIX I: SAMPLING DATA

Table 1.

Note: All masses are in grams

BFR

NFR

PFR

BFR
(repeat)

Mass of cap+container

209.44

209.87

207.68

209.53

Mass of cap+container+water (pre-sampling)

309.78

311.95

308.24

282.99

Mass of cap+container+water (post-sampling)

309.11

310.3

307.36

282.06

(pre-sampling water) - (post-sampling water)

0.67

1.65

0.88

0.93

Mass of cap+container

68.15

68.17

68.15

68.15

Mass of cap+container+silica gel (pre-sampling)

269.06

268.16

268.04

268.35

Mass of cap+container+silica gel (post-sampling)

271.06

270.75

270.93

270.18

Mass of water absorbed in silica gel

2

2.59

2.89

1.83

Mass of cap+container

207.9

209.02

208.61

209.05

Mass of cap+container+XAD

247.99

249.09

248.95

249.05

Mass of XAD (pre-sampling)

40.09

40.07

40.34

40

Petri dish

68.24

68.23

68.23

68.23

Petri dish+filter (pre-sampling)

68.66

68.65

68.64

68.65

Mass of filter (pre-sampling)

0.42

0.42

0.41

0.42

Mass of container+cap

209.88

209.13

207.49

208.61

Mass of container+cap+filter (post-sampling)

210.38

209.62

207.99

NA

Mass of filter (post-sampling)

0.5

0.49

0.5

NA

Mass of soot

0.08

0.07

0.09

NA

Table 2.



BFR

NFR

PFR

BFR (repeat)

Soot formed (g)

0.08

0.07

0.09

NA

Mass burned (g)

10.1

12.4

13.9

10.5

soot formed/mass
burned (g/g)

0.00792

0.00565

0.00647

NA

A-90

-------
Table 3.



BFR REPEAT

PFR



Time
(h:m:s)

Inside Wall
temperature
(°C)

Mass
(g)

Comments

Time
(h:m:s)

Inside Wall
temperature
(°C)

Mass
(g)

Comments

Run
1

0:00:00

95

61.3



0:00:00

96

63.2



0:01:44
0:02:44
0:03:36
0:04:30
0:05:20
0:06:09

104
124
134
122
116
110

57.4

52.1
44.6

43

42.5

42.2

ignition
max temp

removed

0:03:00
0:05:00
0:06:00
0:07:15
0:08:30
0:09:15
0:10:15
0:11:45
0:13:40

95

108

128

130

121

115

110

106

105

61.9
60.3
55.3

50.5
48

47.6
47

46.6
46

max temp
removed

Run

2

0:09:34

103

59.8



0:16:35

102

63.8





0:09:44
0:10:44
0:11:44
0:12:36
0:13:45
0:14:46

107
111
127
133
121
116

58.4

56.2

49.3

43.4
41.8
41.3

ignition

max temp
removed

0:17:37
0:19:10
0:20:10
0:21:10
0:22:25
0:23:36
0:24:36
0:25:36
0:27:03
0:28:31
0:30:56

101
100
117
123
130
128
119
113
109
107
105

63.1
62.1

59.4
57.3

52.5
48.9
47.9

47.3

46.7

46.4

45.8

ignition
re-ignited
max temp

removed

Run
3

0:17:17

107

61



0:33:45

102

64.2





0:17:46
0:18:16
0:19:16
0:20:18
0:21:32
0:22:30

109
109
119
131
126
118

59.8

59.2

54.7

46.8
42.8

42.3

ignition
re-ignited

max temp

removed

0:34:57
0:36:36
0:37:25
0:39:03
0:40:23
0:41:45
0:42:50
0:44:35
0:46:32

104
102
107
130
127
118
112
109
107

63.2
62

60.6
53.1
49.4

48.3
47.9

47.4
46.8

ignition

re-ignited
max temp

removed

Run

4

0:26:40

108

10.4



0:49:12

104

10.6





0:27:01
0:27:23
0:27:57

111
114
113

9.9
7.8
9.1

ignition
max temp
removed

0:49:42
0:50:20
0:52:49

114
114
109

8.4
7.7
7.2

no flame
removed

Run

0:31:07

107

11.9



0:55:30

105

12.5











A-91







-------
5



















0:31:20

110

10.6



0:57:00

113

9.7





0:31:42

114

8.5

max temp

0:57:39

113

10.1

no flame



0:32:00

113

8.5

removed

0:58:29

110

10

removed

NFR

Time
(h:min:sec)

Inside Wall

temperature Mass (g) Comments
(°C)	

0:05:31	118	61.1

0:07:00	127	50.5

0:08:23	132	46.5 max temp

0:10:00	122	45.1

0:10:55	117	44.7

0:11:47	114	44.3

0:12:51	\_\2	43.8 removed

0:16:09	107	64.2

0:18:09	113	61.7

0:19:19	120	58.6

0:20:32	129	54.3

0:21:30	131	50.3 maxtemp

0:22:55	125	47.8

0:23:58	120	47.1

0:25:09	116	48.4

0:26:18	113	47.4 no flame

0:27:44	ILL	46.2	removed

0:30:46	107	63.6

0:31:46	111	62

0:32:45	111	61.2

0:34:06	126	58.5

0:35:06	131	54.4 maxtemp

0:36:41	134	50.6

0:37:30	128	46.1

0:38:44	121	45.7 no flame

0:40:08	116	44.8	removed

0:43:39	109	10.8

0:44:00	121	8

0:44:22	124	7.1 maxtemp

0:44:58	120	6.9

0:45:52	116	6_8	removed

0:49:16	111	12.1

0:49:32	112	11

0:50:06	123	8.5 maxtemp

0:51:00	117	8.6

0:51:48	114	8.3 removed

A-92

-------
Additional Comments

NFR: Stack conditions after experiment:

Outside Wall temperature: 167°C
Inside Wall temperature: 112°C

PFR: Stack conditions after experiment:

Outside Wall temperature: 155°C and 162°C (2 thermocouples on outside wall)
Inside Wall temperature: 74°C

BFR REPEAT : Stack conditions after experiment:

Outside Wall temperature: 158°C and 164°C (2 thermocouples on outside wall)
Inside Wall temperature: 96°C

APPENDIX 2: ANALYSIS DATA
PCDD/Fs:

Pre-extraction surrogate recovery limits:



Surrogate Recovery limits (range in %)

13C12-2 MCDF

25.0

130

13C12-2 MCDD

25.0

130

13C12-2,4 DCDF

25.0

130

13C12-2,7 DCDD

25.0

130

13C12-2,4,8 TrCDF

25.0

130

13C12-2,3,7,8 TeCDF

25.0

130

13C12-2,3,7,8 TeCDD

25.0

130

13C12-1,2,3,7,8 PCDF

40.0

130

13C12-1,2,3,7,8 PCDD

40.0

130

13C12-1,2,3,6,7,8 HxCDF

40.0

130

13C12-1,2,3,6,7,8 HxCDD

40.0

130

13C12-1,2,3,4,6,7,8 HpCDF

40.0

130

13C12-1,2,3,4,6,7,8 HpCDD

40.0

130

13C12-1,2,3,4,6,7,8,9 OCDD

25.0

130

A-93

-------
Pre- sampling surrogate recovery limits:



Pre Spike Recovery Limits (range in %)

13C12-2,8-DCDF

70.0

130

13C12-2,3-DCDD

70.0

130

13C12-2,3,7-TrCDD

70.0

130

37C14-2,3,7,8-TeCDD

70.0

130

13C12-2,3,4,7,8-PCDF

70.0

130

13C12-l,2,3,4,7,8-HxCDF

70.0

130

13C12-l,2,3,4,7,8-HxCDD

70.0

130

13C12-l,2,3,4,7,8,9-HpCDF

70.0

130

BR FR Epoxy Laminate:

Sampled: 6/05/08
Extracted: 7/15/08
Acquired: 01/27/09

Sample description/Narrative: Sample Rerun; Elevated Standard Recoveries are due to a large
interferentpeak causing reduced signal on the TeCDD Recovery Standard.

Pre Extraction
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

13C12-2,3,7,8 TeCDF

135.0

F

13C12-2,3,7,8 TeCDD

125.9

P

13C12-1,2,3,7,8 PCDF

108.6

P

13C12-1,2,3,7,8 PCDD

93.4

P

13C12-1,2,3,6,7,8 HxCDF

68.7

P

13C12-1,2,3,6,7,8 HxCDD

65.3

P

13C12-1,2,3,4,6,7,8
HpCDF

59.6

P

13C12-1,2,3,4,6,7,8
HpCDD

78.6

P

13C12-1,2,3,4,6,7,8,9
OCDD

67.3

P

A-94

-------
Pre-Sampling
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

37C14-2,3,7,8-TeCDD

91.3

P

13C12-2,3,4,7,8-PCDF

91.8

P

13C12-1,2,3,4,7,8-
HxCDF

108.1

P

13C12-1,2,3,4,7,8-
HxCDD

112.9

P

13C12-1,2,3,4,7,8,9-
HpCDF

112.7

P

Isomer.

ng/train



2005 WHO
(Mammal/Humans
) Toxicity Equiv.
Factor

TEQ
ng/train

2,3,7,8 - TeCDD

0.029

LOD

1

0.00000

1,2,3,7,8 - PCDD, co-elution

0.095

LOD

1

0.00000

1,2,3,4,7,8-HxCDD, co-
elution

0.113

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDD

0.103

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDD

0.113

LOD

0.1

0.00000

1,2,3,4,6,7,8-HpCDD

0.196

LOD

0.01

0.00000

1,2,3,4,6,7,8,9 - OCDD

0.231

LOD

0.0003

0.00000

2,3,7,8 - TeCDF

0.03

LOD

0.1

0.00000

1,2,3,7,8 -PCDF

0.064

LOD

0.03

0.00000

2,3,4,7,8 - PCDF

0.064

LOD

0.3

0.00000

1,2,3,4,7,8 -HxCDF

0.032

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDF

0.029

LOD

0.1

0.00000

2,3,4,6,7,8 - HxCDF

0.036

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDF

0.04

LOD

0.1

0.00000

1,2,3,4,6,7,8 -HpCDF

0.084



0.01

0.00084

1,2,3,4,7,8,9-HpCDF

0.064

LOD

0.01

0.00000

1,2,3,4,6,7,8,9 - OCDF

0.131

LOD

0.0003

0.00000

EMPC=Est. Max. Possible

ND = not detected	Concentration	Total TEQ

NS= not spiked	LOD=Limit of Detection	ng/train	0.0008

A-95

-------
NFR Epoxy Laminate:

Sampled: 6/16/08
Extracted: 7/15/08
Acquired: 12/15/08

Sample description/Narrative: All detected targets appear to be carry over from a Standard.

Pre Extraction
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

13C12-2,3,7,8 TeCDF

88.1

P

13C12-2,3,7,8 TeCDD

88.0

P

13C12-1,2,3,7,8 PCDF

97.4

P

13C12-1,2,3,7,8 PCDD

101.8

P

13C12-1,2,3,6,7,8 HxCDF

75.9

P

13C12-1,2,3,6,7,8 HxCDD

73.6

P

13C12-1,2,3,4,6,7,8
HpCDF

67.9

P

13C12-1,2,3,4,6,7,8
HpCDD

85.1

P

13C12-1,2,3,4,6,7,8,9
OCDD

72.4

P



Pre-Sampling
Surrogates

%

Recovery



37C14-2,3,7,8-TeCDD

90.0

P

13C12-2,3,4,7,8-PCDF

100.9

P

13C12-1,2,3,4,7,8-
HxCDF

104.2

P

13C12-1,2,3,4,7,8-
HxCDD

111.1

P

13C12-1,2,3,4,7,8,9-
HpCDF

115.5

P

A-96

-------
Isomer.

ng/train



2005 WHO
(Mammals/Humans)
Toxicity Equiv.
Factor

TEQ
ng/train

2,3,7,8 - TeCDD

0.013

LOD

1

0.00000

1,2,3,7,8 - PCDD, co-elution

0.015

LOD

1

0.00000

1,2,3,4,7,8 - HxCDD, co-elution

0.024

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDD

0.022

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDD

0.024

LOD

0.1

0.00000

1,2,3,4,6,7,8 -HpCDD

0.06



0.01

0.00060

1,2,3,4,6,7,8,9 - OCDD

0.096



0.0003

0.00003

2,3,7,8 - TeCDF

0.036



0.1

0.00360

1,2,3,7,8 -PCDF

0.014

LOD

0.03

0.00000

2,3,4,7,8 - PCDF

0.014

LOD

0.3

0.00000

1,2,3,4,7,8 -HxCDF

0.018

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDF

0.016

LOD

0.1

0.00000

2,3,4,6,7,8 - HxCDF

0.02

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDF

0.022

LOD

0.1

0.00000

1,2,3,4,6,7,8-HpCDF

0.028



0.01

0.00028

1,2,3,4,7,8,9 -HpCDF

0.025

LOD

0.01

0.00000

1,2,3,4,6,7,8,9 - OCDF

0.063

LOD

0.0003

0.00000

ND = not detected	EMPC=Est. Max. Possible Concentration	Total TEQ

NS= not spiked	LOD=Limit of Detection	ng/train 0.0045

PFR Epoxy Laminate:

Sampled: 06/17/08
Extracted: 07/15/08
Date Acquired: 12/15/08

Sampled description/ Narrative: All detected targets appear to be carry over from a Standard.

A-97

-------
Pre Extraction
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

13C12-2,3,7,8 TeCDF

90.0

P

13C12-2,3,7,8 TeCDD

89.4

P

13C12-1,2,3,7,8 PCDF

109.9

P

13C12-1,2,3,7,8 PCDD

110.9

P

13C12-1,2,3,6,7,8 HxCDF

70.4

P

13C12-1,2,3,6,7,8 HxCDD

69.2

P

13C12-1,2,3,4,6,7,8
HpCDF

64.4

P

13C12-1,2,3,4,6,7,8
HpCDD

80.2

P

13C12-1,2,3,4,6,7,8,9
OCDD

72.5

P



Pre-Sampling
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

37C14-2,3,7,8-TeCDD

105.3

P

13C12-2,3,4,7,8-PCDF

115.5

P

13C12-1,2,3,4,7,8-
HxCDF

119.9

P

13C12-1,2,3,4,7,8-
HxCDD

128.5

P

13C12-1,2,3,4,7,8,9-
HpCDF

135.2

F

A-98

-------
Isomer.

ng/train



2005 WHO
(Mammals/Humans)
Toxicity Equiv.
Factor

TEQ
ng/train

2,3,7,8 - TeCDD

0.012

LOD

1

0.00000

1,2,3,7,8 - PCDD, co-elution

0.015

LOD

1

0.00000

1,2,3,4,7,8 - HxCDD, co-elution

0.025

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDD

0.023

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDD

0.025

LOD

0.1

0.00000

1,2,3,4,6,7,8 -HpCDD

0.036

LOD

0.01

0.00000

1,2,3,4,6,7,8,9 - OCDD

0.047

LOD

0.0003

0.00000

2,3,7,8 - TeCDF

0.024

EMPC

0.1

0.00240

1,2,3,7,8 -PCDF

0.013

LOD

0.03

0.00000

2,3,4,7,8 - PCDF

0.013

LOD

0.3

0.00000

1,2,3,4,7,8 -HxCDF

0.014

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDF

0.013

LOD

0.1

0.00000

2,3,4,6,7,8 - HxCDF

0.016

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDF

0.018

LOD

0.1

0.00000

1,2,3,4,6,7,8-HpCDF

0.015

LOD

0.01

0.00000

1,2,3,4,7,8,9-HpCDF

0.02

LOD

0.01

0.00000

1,2,3,4,6,7,8,9 - OCDF

0.047

LOD

0.0003

0.00000

ND = not detected	EMPC=Est. Max. Possible Concentration	Total TEQ

NS= not spiked	LOD=Limit of Detection	ng/train 0.0024

BR FR Epoxy Laminate repeat run:

Sampled: 06/18/08
Extracted: 07/15/08
Acquired: 12/09/08

Sampled description/ Narrative: All detected targets appear to be carry over from a Standard.

A-99

-------
Pre Extraction
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

13C12-2,3,7,8 TeCDF

109.5

P

13C12-2,3,7,8 TeCDD

114.9

P

13C12-1,2,3,7,8 PCDF

112.3

P

13C12-1,2,3,7,8 PCDD

110.2

P

13C12-1,2,3,6,7,8 HxCDF

52.2

P

13C12-1,2,3,6,7,8 HxCDD

56.6

P

13C12-1,2,3,4,6,7,8
HpCDF

47.9

P

13C12-1,2,3,4,6,7,8
HpCDD

55.4

P

13C12-1,2,3,4,6,7,8,9
OCDD

49.2

P





Pass or Fail

Pre-Sampling
Surrogates

% Recovery

recovery
limits

37C14-2,3,7,8-TeCDD

96.4

P

13C12-2,3,4,7,8-PCDF

100.9

P

13C12-1,2,3,4,7,8-





HxCDF

120.5

P

13C12-1,2,3,4,7,8-





HxCDD

126.4

P

13C12-1,2,3,4,7,8,9-





HpCDF

127.2

P

A-100

-------
Isomer.

ng/train



2005 WHO
(Mammals/Humans)
Toxicity Equiv.
Factor

TEQ
ng/train

2,3,7,8 - TeCDD

0.036

LOD

1

0.00000

1,2,3,7,8 - PCDD, co-elution

0.036



1

0.03600

1,2,3,4,7,8 - HxCDD, co-elution

0.052



0.1

0.00520

1,2,3,6,7,8 -HxCDD

0.036



0.1

0.00360

1,2,3,7,8,9-HxCDD

0.056



0.1

0.00560

1,2,3,4,6,7,8-HpCDD

0.092



0.01

0.00092

1,2,3,4,6,7,8,9 - OCDD

0.172



0.0003

0.00005

2,3,7,8 - TeCDF

0.072



0.1

0.00720

1,2,3,7,8 -PCDF

0.06



0.03

0.00180

2,3,4,7,8 - PCDF

0.06



0.3

0.01800

1,2,3,4,7,8 -HxCDF

0.084



0.1

0.00840

1,2,3,6,7,8 -HxCDF

0.076



0.1

0.00760

2,3,4,6,7,8 - HxCDF

0.1



0.1

0.01000

1,2,3,7,8,9-HxCDF

0.116



0.1

0.01160

1,2,3,4,6,7,8-HpCDF

0.14



0.01

0.00140

1,2,3,4,7,8,9 -HpCDF

0.132



0.01

0.00132

1,2,3,4,6,7,8,9 - OCDF

0.22



0.0003

0.00007

ND = not detected	EMPC=Est. Max. Possible Concentration	Total TEQ

NS= not spiked	LOD=Limit of Detection	ng/train 0.1188

Blank run:

Sampled: 05/29/08
Extracted: 07/15/08
Acquired: 01/27/09

Sample Description/ Narrative: sample rerun.

A-101

-------
Pre Extraction
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

13C12-2,3,7,8 TeCDF

90.6

P

13C12-2,3,7,8 TeCDD

86.3

P

13C12-1,2,3,7,8 PCDF

78.5

P

13C12-1,2,3,7,8 PCDD

79.8

P

13C12-1,2,3,6,7,8 HxCDF

73.6

P

13C12-1,2,3,6,7,8 HxCDD

72.2

P

13C12-1,2,3,4,6,7,8
HpCDF

66.1

P

13C12-1,2,3,4,6,7,8
HpCDD

86.0

P

13C12-1,2,3,4,6,7,8,9
OCDD

77.1

P



Pre-Sampling
Surrogates

%

Recovery

Pass or

Fail
recovery
limits

37C14-2,3,7,8-TeCDD

100.9

P

13C12-2,3,4,7,8-PCDF

112.8

P

13C12-1,2,3,4,7,8-
HxCDF

118.4

P

13C12-1,2,3,4,7,8-
HxCDD

122.2

P

13C12-1,2,3,4,7,8,9-
HpCDF

109.2

P

A-102

-------
Isomer.

ng/train



2005 WHO
(Mammals/Humans)
Toxicity Equiv.
Factor

TEQ
ng/train

2,3,7,8 - TeCDD

0.026

LOD

1

0.00000

1,2,3,7,8 - PCDD, co-elution

0.043

LOD

1

0.00000

1,2,3,4,7,8 -HxCDD, co-
elution

0.061

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDD

0.056

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDD

0.061

LOD

0.1

0.00000

1,2,3,4,6,7,8 -HpCDD

0.129

LOD

0.01

0.00000

1,2,3,4,6,7,8,9 - OCDD

0.152

LOD

0.0003

0.00000

2,3,7,8 - TeCDF

0.029

LOD

0.1

0.00000

1,2,3,7,8 -PCDF

0.033

LOD

0.03

0.00000

2,3,4,7,8 - PCDF

0.033

LOD

0.3

0.00000

1,2,3,4,7,8 -HxCDF

0.033

LOD

0.1

0.00000

1,2,3,6,7,8 -HxCDF

0.03

LOD

0.1

0.00000

2,3,4,6,7,8 - HxCDF

0.036

LOD

0.1

0.00000

1,2,3,7,8,9-HxCDF

0.041

LOD

0.1

0.00000

1,2,3,4,6,7,8 -HpCDF

0.036

LOD

0.01

0.00000

1,2,3,4,7,8,9 -HpCDF

0.048

LOD

0.01

0.00000

1,2,3,4,6,7,8,9 - OCDF

0.113

LOD

0.0003

0.00000

ND = not detected	EMPC=Est. Max. Possible Concentration	Total TEQ

NS= not spiked	LOD=Limit of Detection	ng/train	ND

PBDD/Fs:

BR FR Epoxy Laminate:

Sampled: 6/05/08
Extracted: 7/16/08
Acquired: 02/17/09

A-103

-------
Pre Extraction
Surrogates

%

Recovery

13C 237 TrBDD (IS)

87.0

13C 2378 TeBDD (IS)

56.4

13C 123678 HxBDD (IS)

115.1

13C 123789 HxBDD (IS)

96.3

13C OcBDD (IS)

NR

13C 2468 TeBDF (DSSP)

123.7

13C 12378 PeBDD (DSSP)

127.9

Isomer

ng/train

237 TrBDD*

0.08

237 TrBDF*

ND

2378 TeBDD

0.37

2468 TeBDF

0.56

2378 TeBDF

2.80

12378 PeBDD

ND

12378 PeBDF

1.06

23478 PeBDF

0.54

123478/123678 HxBDD

ND

123789 HxBDD

ND

123478 HxBDF

0.35

1234679 HpBDD7**

ND

1234678 HpBDD7**

ND

1234678 HpBDF

4.76

OcBDD

NR

OcBDF

NR

* not present in the standard; assignment based on isotope theoretical ratios and retention times of matching internal
standards and native

congeners; quantified based on concentration of the congeners of the same bromination level present in the standard
** assignment based on the elution order of HpCDD congeners on the DB5 column
ND = not detected
NS= not spiked

EMPC=Est. Max. Possible Concentration
LOD=Limit of Detection (S/N=3)

NR=not reported (OcBDD/F would need separate clean-up;13C OcBDD did not elute from carbon
column)

A-104

-------
NFR Epoxy Laminate:

Sampled: 6/16/08
Extracted: 7/16/08
Acquired: 02/17/09

Pre Extraction
Surrogates

%

Recovery

13C 237 TrBDD (IS)

108.9

13C 2378 TeBDD (IS)

89.7

13C 123678 HxBDD (IS)

132.8

13C 123789 HxBDD (IS)

102.4

13C OcBDD (IS)

NR

13C 2468 TeBDF (DSSP)

103.7

13C 12378 PeBDD (DSSP)

113

Isomer

ng/train

237 TrBDD*

ND

237 TrBDF*

ND

2378 TeBDD

ND

2468 TeBDF

ND

2378 TeBDF

ND

12378 PeBDD

ND

12378 PeBDF

ND

23478 PeBDF

ND

123478/123678 HxBDD

ND

123789 HxBDD

ND

123478 HxBDF

ND

1234679 HpBDD7**

ND

1234678 HpBDD7**

ND

1234678 HpBDF

ND

OcBDD

NR

OcBDF

NR

* not present in the standard; assignment based on isotope theoretical ratios and retention times of matching internal
standards and native

congeners; quantified based on concentration of the congeners of the same bromination level present in the standard
** assignment based on the elution order of HpCDD congeners on the DB5 column
ND = not detected
NS= not spiked

EMPC=Est. Max. Possible Concentration
LOD=Limit of Detection (S/N=3)

NR=not reported (OcBDD/F would need separate clean-up;13C OcBDD did not elute from carbon
column)

A-105

-------
PFR Epoxy Laminate:

Sampled: 06/17/08
Extracted: 07/15/08
Date Acquired: 12/15/08

Pre Extraction
Surrogates

%

Recovery

13C 237 TrBDD (IS)

79.6

13C 2378 TeBDD (IS)

61.1

13C 123678 HxBDD (IS)

122.6

13C 123789 HxBDD (IS)

116.1

13C OcBDD (IS)

NR

13C 2468 TeBDF (DSSP)

117.6

13C 12378 PeBDD (DSSP)

139.1

Isomer

ng/train

237 TrBDD*

ND

237 TrBDF*

ND

2378 TeBDD

ND

2468 TeBDF

ND

2378 TeBDF

ND

12378 PeBDD

ND

12378 PeBDF

ND

23478 PeBDF

ND

123478/123678 HxBDD

ND

123789 HxBDD

ND

123478 HxBDF

ND

1234679 HpBDD7**

ND

1234678 HpBDD7**

ND

1234678 HpBDF

ND

OcBDD

NR

OcBDF

NR

* not present in the standard; assignment based on isotope theoretical ratios and retention times of matching internal
standards and native

congeners; quantified based on concentration of the congeners of the same bromination level present in the standard
** assignment based on the elution order of HpCDD congeners on the DB5 column
ND = not detected
NS= not spiked

EMPC=Est. Max. Possible Concentration
LOD=Limit of Detection (S/N=3)

NR=not reported (OcBDD/F would need separate clean-up;13C OcBDD did not elute from carbon
column)

A-106

-------
BR FR Epoxy Laminate repeat run:

Sampled: 06/18/08
Extracted: 07/16/08
Acquired: 02/17/09

Pre Extraction
Surrogates

%

Recovery

13C 237 TrBDD (IS)

77.2

13C 2378 TeBDD (IS)

57.1

13C 123678 HxBDD (IS)

112.5

13C 123789 HxBDD (IS)

120.9

13C OcBDD (IS)

NR

13C 2468 TeBDF (DSSP)

110.5

13C 12378 PeBDD (DSSP)

139.6

Isomer

ng/train

237 TrBDD*

ND

237 TrBDF*

ND

2378 TeBDD

0.24

2468 TeBDF

0.47

2378 TeBDF

1.45

12378 PeBDD

ND

12378 PeBDF

0.81

23478 PeBDF

0.30

123478/123678 HxBDD

ND

123789 HxBDD

ND

123478 HxBDF

0.26

1234679 HpBDD7**

ND

1234678 HpBDD7**

ND

1234678 HpBDF

5.64

OcBDD

NR

OcBDF

NR

* not present in the standard; assignment based on isotope theoretical ratios and retention times of matching internal
standards and native

congeners; quantified based on concentration of the congeners of the same bromination level present in the standard
** assignment based on the elution order of HpCDD congeners on the DB5 column
ND = not detected
NS= not spiked

EMPC=Est. Max. Possible Concentration
LOD=Limit of Detection (S/N=3)

NR=not reported (OcBDD/F would need separate clean-up;13C OcBDD did not elute from carbon
column)

A-107

-------
Blank run:

Sampled: 07/15/08
Extracted: 07/16/08
Acquired: 02/17/09

Pre Extraction
Surrogates

%

Recovery

13C 237 TrBDD (IS)

117.3

13C 2378 TeBDD (IS)

93.5

13C 123678 HxBDD (IS)

118.1

13C 123789 HxBDD (IS)

106.0

13C OcBDD (IS)

NR

13C 2468 TeBDF (DSSP)

105.3

13C 12378 PeBDD (DSSP)

112.1

Isomer

ng/train

237 TrBDD*

ND

237 TrBDF*

ND

2378 TeBDD

ND

2468 TeBDF

ND

2378 TeBDF

ND

12378 PeBDD

ND

12378 PeBDF

ND

23478 PeBDF

ND

123478/123678 HxBDD

ND

123789 HxBDD

ND

123478 HxBDF

ND

1234679 HpBDD7**

ND

1234678 HpBDD7**

ND

1234678 HpBDF

ND

OcBDD

NR

OcBDF

NR

* not present in the standard; assignment based on isotope theoretical ratios and retention times of matching internal
standards and native

congeners; quantified based on concentration of the congeners of the same bromination level present in the standard
** assignment based on the elution order of HpCDD congeners on the DB5 column
ND = not detected
NS= not spiked

EMPC=Est. Max. Possible Concentration
LOD=Limit of Detection (S/N=3)

NR=not reported (OcBDD/F would need separate clean-up;13C OcBDD did not elute from carbon
column)

A-108

-------
FLAME RETARDANTS IN PRINTED CIRCUIT
BOARDS: APPENDIX C

U.S. EPA. Analysis of Circuit Board Samples by
XRF. Original Report - July 28, 2008. Revised
Report - March 23, 2009. Prepared by Arcadis.

A-109

-------
Imagine the

SEFW

Analysis of Circuit Board Samples
by XRF

Report

Original Report - July 28, 2008
Revised Report - March 23, 2009

A-110

-------
DISCLAIMER: The USEPA Design for the Environment Program has provided
additional information in Appendix B and Appendix C to further explain methods
and results. This information is critical for interpreting the main report,
especially in regards to chorine measurements. Results found in the main
report are not complete without the information in the appendices, and cannot
be correctly understood or interpreted without their aid.

A-lll

-------
Analysis of Circuit Board
Samples by XRF

Report

Prepared for:

U.S. Environmental Protection
Agency

Air Pollution Prevention and Control
Division

Research Triangle Park, NC 27711

Prepared by:

ARCADIS

4915 Prospectus Drive

Suite F

Durham

North Carolina 27713
Tel 919.544.4535
Fax 919.544.5690

Our Ref.:

RN990234.0037

This document is intended only for the use
of the individual or entity for which it was
prepared and may contain information that
is privileged, confidential and exempt from
disclosure under applicable law. Any
dissemination, distribution or copying of
this document is strictly prohibited.

A-112

-------
Table of Contents

1.	Statement of Work

2.	Introduction

3.	Experimental

3.1	Sample preparation

3.1.1	Phase 1

3.1.2	Phase 2

3.1.2.1	Sub-sampling

3.1.2.2	Milling

3.1.2.3	Homogenization and sub-sampling

3.1.2.4	Pellet Preparation

3.1.2.5	Preparation of Spiked Sample

3.2	Analysis

3.3	Quantification

4.	Data

4.1	Phase 1

4.2	Phase 2

115

115

116

116
116
116
116
116
118

118

119

120

120

121

121

122

5. Conclusions

128

Tables

Table 1: Samples Received	115

Table 2: Milling parameters	117

Table 3. Coarse fraction Milling Parameters	118

Table 4. Pellet Press Parameters	119

Table 5. Composition of Spiked Sample 7	120

Table 6. Results for Phase 1 Samples	121

A-113

-------
Table of Contents

Table 7. Elemental Concentrations, weight %	122

Table 8. Sample 7, Short Term Reproducibility, weight %	124

Table 9. Sample Preparation Reproducibility, Sample 7	125

Table 10. Recovery of Spikes, Sample 7, weight %	126

Figures

Figure 1 . Sieved Circuit Board	117

Appendices

Appendix A: Responses to Questions

Appendix B: Laminate Etching and Chlorine Measurements

Appendix C: ISOLA Experiment Demonstrating the Impact of the Etching Process
on Chlorine Measurements

A-114

-------
Analysis of Circuit
Board Samples by
XRF

Report

1.	Statement of Work

The following report is in response to a task under Work Assignment (WA) No. 3-37,
that consisted of an elemental analysis of two sets of circuit boards samples by X-ray
Fluorescence (XRF) Spectrometry. This report describes the results of those analyses
and provides discussions of several questions that have arisen from these analyses.

2.	Introduction

Under two separate events, described as "Phase 1" and "Phase 2," circuit board
samples were received for analyses. Table 1 presents this information.

Table 1: Samples Received

Laminate
#

Phase

Laminate
type

1

1

NFR

2

1

BFR

3

1

PFR

4

2

HF

5

2

HF

6

2

HF

7

2

HF

NFR : Non-flame Retardant; BFR: Bromine Flame Retardant; PFR: Phosphorous
Flame Retardant; HF: Halogen-free

Each board was received "mostly" free of copper plating. Phase 2 samples were
accompanied by a letter that indicated 12" by 12" samples of "halogen-free laminates."
Inspection of each showed a rectangular area of plated copper in one corner of each
sample that was used to identify each sample. Further inspection showed that some
samples had additional small, random areas of elemental copper. This was also true of
the phase 1 samples.

A-115

-------
Analysis of Circuit
Board Samples by
XRF

Report

3. Experimental
3.1 Sample preparation

3.1.1	Phase 1

As directed, phase 1 samples were cored in the shop at random locations. These
circuit board disks were sized to be a slip fit to our standard sample cups. Separate
disks were cut for each individual analysis.

3.1.2	Phase 2

As agreed prior to sample receipt, samples were homogenized, powdered, pelletized,
and analyzed by XRF. One sample was prepared and analyzed in duplicate. One
spiked sample was prepared and analyzed.

3.1.2.1	Sub-sampling

To minimize the errors of heterogeneity, each board was sub-sampled from several
locations. One board was weighed at ~ 79 g. per square foot. To ensure that any one
sample was of sufficient size to provide sufficient material for sample, replicate, and
spike, it was decided to sample 21-1" diameter locations in a representative manner.
Boards were delivered to the shop, which laid out a 9 by 7 grid. With directions to avoid
potential elemental copper, all edge areas were not sampled. 21 of the remaining 35
positions were sampled by coring.

3.1.2.2	Milling

The 21 disks from each sample were homogenized by milling. A Spex Certiprep model
6850 Freezer/Mill was used for this step. This instrument is basically a hammer mill
operating at liquid nitrogen temperatures. All 21 disks were added to a sample tube
along with the stainless steel, SS, hammer. This instrument has the capacity to handle
a single sample of this size. Table 2 provides the operating parameters for the first
milling operation.

A-116

-------
ARCADIS

Analysis of Circuit
Board Samples by
XRF

Report

Table 2: Milling parameters

Operation

Value

Pre-cool time

15 min.

# of cycles

4

Milling time

3 min.

Re-cool time

10 min.

After samples had warmed back to room temperature, they were opened and
examined. The milling was considered generally acceptable, with a large fraction of the
sample present as powder. A fraction of each sample, however, was present as large
flakes. Figure 1 shows one sample after size classification.

Figure 1 . Sieved Circuit Board

It was unclear whether this coarse flake fraction (left) represented a surface treatment
coating or was merely incomplete milling of a homogeneous sample. After discussions
it was decided to sieve, re-mill the coarse fraction, and combine. A W.S. Tyler Number
18 sieve, Tyler Equivalent 16 mesh, was used for the fractionation. The fine fraction
was transferred to a pre-cleaned 40 mL sample vial while the coarse fraction was
returned to the cryo-mill for further milling. Table 3 provides the operating parameters
for this second milling operation.

A-117

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Table 3. Coarse fraction Milling Parameters

Operation

Value

Pre-cool time

10 min.

# of cycles

4

Milling time

2 min.

Re-cool time

5 min.

Less stringent conditions were used since the coarse fraction represented a smaller
sample. Coarse fractions were found to range between ~ 1 g and 3 g. This second
milling operation was successful and the sample fractions were combined.

3.1.2.3	Homogenization and sub-sampling

Sample homogenization began with the coring of multiple discs spanning the area of
each sample. It continued with the cryo-milling operation described in the previous
section. It was finalized just prior to sample weighing by sample riffling. A Humboldt
Mfg. Co. Model H-3971C archeological grade rifflerwas used for this purpose. This
model was designed for samples in the several gram range. A riffler has the purpose of
sub-sampling a larger powdered sample in a statistically equivalent manner that is
particle-size and density independent. It achieves this by fractionating the total sample
through multiple, equivalents sized paths leading to two or more sample buckets. No
assumptions, however, can be made that the sub-samples will remain equivalent if
time is allowed to pass. Riffling must be done immediately prior to sample use.

This riffler is manufactured of SS (stainless steel). It consists of a hopper, a gate,
multiple equivalent alternating vertically angled slots, and two buckets. It may be used
for both homogenization and sub-sampling and was used for both purposes in this
project. The entire sample was passed through the riffler twice. After the second pass,
sample material in one bucket was returned to the sample vial. The sub-sample in the
second bucket represented ~ 4 g at this point. This fraction was passed through the
riffler one more time. Each bucket contained about 2 g, which was the correct size for
preparing a single XRF pellet.

3.1.2.4	Pellet Preparation

Pellets were prepared by pressing a mixture of powdered sample with a polymeric
binder. 2 grams of sample were weighed and transferred to a boron carbide mortar and

A-118

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

pestle. The sample was ground for a period, though little grinding took place at this
stage for these samples. 2 mL of Spex Liquid Binder, equivalent to 200 mg of binder in
a dichloromethane carrier, were added using a Gilson Microman positive displacement
pipettor. Sample was mixed until the sample returned to a free-flowing state. Sample
was transferred to 32 mm dies with vacuum port. Pellet was pressed under vacuum in
a Spex 3630 X-press programmable hydraulic press. Table 4 presents the pelletizing
parameters.

Table 4. Pellet Press Parameters

Operation

Value

Applied pressure

20 tons

Hold time

1.1 min.

Release time

1.0 min.

Formed pellets were transferred to Millipore 47 mm Petrislides for identification and
stored in a silica gel controlled desiccator until ready for analysis.

As agreed, one sample was prepared in duplicate. As agreed, one sample was spiked
with known masses during the pellet preparation stage. After discussions with the
work assignment manager and the industry committee, spiking materials and elements
were selected as described in the next section. Based upon data from the first set of
circuit boards; spikes were prepared for aluminum, calcium, and copper.

3.1.2.5 Preparation of Spiked Sample

As directed, one sample was prepared by spiking with known masses of certain
analytes to provide data on recovery. Sample 7 was chosen since that sample
represents the most complete data set. In other words, sample 7 was prepared in
duplicate and analyzed in replicate. This sample had the most data available for
comparison to the spiked sample.

Based upon data from the Phase 1 circuit boards; spikes were prepared for aluminum,
calcium, and copper using reagent grades of Al203, CaC03, and CuS04, respectively.
This gave us data on a fourth element; S. Table 5 provides data on the preparation of
the spiked sample.

A-119

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Table 5. Composition of Spiked Sample 7

Material

Mass, g

Sample 7

1.761

Al203

0.0383

CaCC>3

0.1504

C11SO4

0.0505

Total

2.0002

The four materials listed in Table 5 were weighed in the amounts described in Table 5
and mixed manually using mortar and pestle. A pellet was prepared from this mix as
described in the previous section.

3.2	Analysis

Pressed sample pellets were analyzed on a Panalytical model PW2404 wavelength
dispersive X-Ray Fluorescence Spectrometer equipped with the PW2540 sample
changer. The instrument is equipped with both flow and scintillation detectors plus five
crystals. The instrument is controlled and acquires data using the manufacturer's
software, SuperQ. The entire spectrum is acquired as 10 sub-scans using variations in
applied power, crystal, detector, filter material, and goniometer setting.

Data were acquired using the application, IQ+Metalloids. IQ+Metalloids is a variation
of the manufacturer supplied application, ZIQ+. IQ+Metalloids adds 4 channels to
provide increased sensitivity for the elements: arsenic, selenium, mercury, and lead.
The increased sensitivity comes from increased counting times while the goniometer
sits at the peak maxima. ZIQ+ is a full scan application, which optimizes sample
throughput.

3.3	Quantification

Data acquired as above are quantified using the manufacturer supplied software, IQ+.
IQ+ is a matrix independent, fundamental parameters based quantification program.

A-120

-------
Analysis of Circuit
Board Samples by
XRF

Report

4. Data
4.1 Phase 1

Table 6 presents the data for the three Phase 1 samples. Each was analyzed in
duplicate; where each analysis also represents a replicate sample preparation (cores
from different locations on the board). To be explicit, due to sample decomposition
within the instrument, each sample core was analyzed once. During analysis, the
whole-board cores charred. Replicate analysis on charred samples seemed neither
good chemistry nor good for the instrument.

Table 6. Results for Phase 1 Samples

Sample

1-NFR

2-BFR

3-PFR

Element

Mean, %

% RSD

Mean, %

% RSD

Mean, %

% RSD

Na

0.109

1.76

0.01



0.114

67.47

Mg

0.008

5.38





0.0070



Al

0.083

31.94

1.042



0.773

5.50

Si

0.398

37.02

0.145

2.34

0.201

8.84

P

0.0016

16.26

0.0017

23.03

4.19

1.75

S

0.010

14.89

0.0081

60.67

0.013

8.03

CI

0.878

9.91

0.591

42.27

0.517

11.30

K

0.0078

27.70

0.0043



0.0070

49.55

Ca

2.62

10.04

1.29

33.60

2.49

4.67

Ti

0.061

9.09

0.038

25.42

0.060

4.20

Cr

0.0039







0.0044



Fe

0.036

9.69

0.033

28.74

0.038

2.30

Cu

0.054

1.03

1.81

137.65

3.59

13.93

As

0.0008

17.32

0.056

27.16

0.0011



Br





6.13

22.53

0.0047

12.49

Sr

0.064

4.72

0.064

28.89

0.083

1.08

Pb

0.0007

30.44





0.0007



Zr





0.0088







NFR : Non-flame Retardant; BFR: Bromine Flame Retardant; PFR: Phosphorous
Flame Retardant; HF: Halogen-free

ARCADE

A-121

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Results above are the average of duplicate samples; reproducibility is also presented
as % relative standard deviation, % RSD. In Table 6, an empty cell under a Mean
column heading indicates that this element was not detected in either replicate of this
sample. An empty cell under % RSD indicates that the element was only observed in
one of the replicates of that sample.

In examining Table 6, the most striking feature is the very large % RSDs found for
several results. This is true for all three samples. This is attributed to circuit board
heterogeneity.

4.2 Phase 2

Table 7 presents the data acquired under this task. Colored cells represent not
detected elements for the respective samples.

The first pellet (sample 7) was analyzed three times within a 1 hour period to provide
data on short term reproducibility. These data are provided in Table 8.

As directed, one sample was selected for replicate sample preparation and analyses.
These data may be found in Table 9. Here, both "Replicate 1" and "Replicate 2"
represent the mean determinations of triplicate data collections on a single pellet.

The results for sample 7 spiked as described in Table 5 are provided in Table 10. For
comparison the results from replicate preparations of sample 7 are repeated from
Table 9.

Table 7. Elemental Concentrations, weight %

Element

4

5

6

7

F







0.054

Na

0.135

0.143

0.121

0.151

Mg

0.663

0.085

0.410

0.375

Al

2.76

5.65

6.35

5.30

Si

15.65

9.23

7.77

10.07

P

1.42

0.84

0.74

0.68

S

0.0104

0.0050

0.0049

0.0098

CI

0.449

0.427

0.488

1.044

K

0.0161

0.0126

0.0087

0.0123

A-122

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Element

4

5

6

7

Ca

5.39

4.58

4.47

5.64

Ti

0.107

0.096

0.093

0.117

Cr

0.0184

0.0045

0.0058

0.0065

Fe

0.135

0.067

0.064

0.088

Ni

0.0044







Cu

0.051

0.041

0.047

0.056

Zn

0.0050

0.0031

0.0044

0.0043

Br



0.0012



0.0012

As





0.00071

0.00116

Sr

0.0616

0.0627

0.0581

0.0722

Zr

0.0038







Ba





0.0168



Pb

0.00084





0.00087

A-123

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Table 8. Sample 7, Short Term Reproducibility, weight %

Element

Rep 1

Rep 2

Rep 3

Mean

% RSD

F





0.05028

0.05028



Na

0.148

0.1447

0.1473

0.146667

1.19

Mg

0.3678

0.3776

0.3834

0.376267

2.10

Al

5.305

5.253

5.325

5.294333

0.70

Si

9.97

9.972

10.04

9.994

0.40

P

0.6837

0.6793

0.6879

0.683633

0.63

S

0.0122

0.008915

0.00974

0.010285

16.62

CI

0.9215

0.8356

0.813

0.8567

6.68

K

0.01335

0.01237

0.01404

0.013253

6.33

Ca

5.659

5.674

5.614

5.649

0.55

Ti

0.1199

0.1182

0.114

0.117367

2.59

Cr

0.006383

0.007127

0.006177

0.006562

7.62

Fe

0.09025

0.09096

0.09163

0.090947

0.76

Ni











Cu

0.059

0.05484

0.05479

0.05621

4.30

Zn

0.00449

0.003899

0.00459

0.004326

8.63

Br

0.001292

0.001128

0.001084

0.001168

9.39

As









Sr

0.072

0.07354

0.07197

0.072503

1.24

Zr







Ba





Pb

0.000619 0.000709 0.001066 0.000798

29.65

A-124

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Table 9. Sample Preparation Reproducibility, Sample 7

Element

Replicate 1

Replicate 2

Mean

% RSD

F

0.0503

0.0570

0.0537

8.91

Na

0.1467

0.1558

0.1513

4.29

Mg

0.3763

0.3731

0.3747

0.60

Al

5.294

5.302

5.298

0.10

Si

9.994

10.143

10.069

1.05

P

0.6836

0.6713

0.6774

1.29

S

0.01029

0.00934

0.00981

6.84

CI

0.86

1.23

1.04

25.36

K

0.0133

0.0113

0.0123

11.40

Ca

5.649

5.625

5.637

0.30

Ti

0.11737

0.11597

0.11667

0.85

Cr

0.00656

0.00653

0.00655

0.32

Fe

0.09095

0.08504

0.08799

4.75

Ni









Cu

0.05621

0.05573

0.05597

0.61

Zn

0.00433

0.00428

0.00430

0.74

Br

0.0012

0.0012

0.0012

0.00

As



0.0012

0.0012



Sr

0.07250

0.07199

0.07225

0.50

Zr









Ba









Pb

0.00080 0.00095 0.00087

12.38

A-125

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Table 10. Recovery of Spikes, Sample 7, weight %

Element

Sample 7 Mean
(Table 9)

Sample 7 Spike

Mean % Recovery

Recovery % RSD

Al

5.298

5.193333

91

0.5

Ca

5.637

8.201

103

0.9

Cu

0.05597

1.019333

97

1.3

S

0.00981

0.614233

119

2

F

0.0537







Na

0.1513

0.147767

111

4

Mg

0.3747

0.293467

89

0.4

Si

10.069

8.333

94

0.5

P

0.6774

0.5176

87

0.5

CI

1.04

0.846133

92

8

K

0.0123

0.010305

95

2

Ti

0.11667

0.100767

98

2

Cr

0.00655

0.006422

111

15

Fe

0.08799

0.072413

93

2

Ni









Zn

0.00430

0.004176

110

5

Br

0.0012

0.001184

115

0

As

0.0012

0.00118

115

23

Sr

0.07225

0.066293

104

1.2

Zr









Ba









Pb

0.00087

0.000601

78

15

The spiking of a non-blank material provides results that are slightly difficult to interpret.
The spiked material acts as a diluent for all elemental results that are not added as part
of the spiking process. Iron and magnesium in Table 10 are an example of this.

A-126

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

The proper calculation is described by equations 1 and 2.

%analytei * Samplel

+ HGRAVu *SPike

%Theoretical Re cov eryi = 100 *

100

Equation 1

Samplel + ^ Spike.

%SpikeRe cov ery =100*

%SpikedSamplei

Equation 2

VoTheoretical Re cov eryt

Where sample 7 and Spike; refer to the values found in Table 5, %analyte values are
found in the first column of Table 9. GRAVy refers to the gravimetric factor for analyte i
in spike material j.

To be more explicit, one example of Spikej from Table 5 would be Al203. The only
analytej in alumina would be aluminum. Therefore, GRAVy in this case would be the
gravimetric factor for aluminum in alumina. The gravimetric factor is a well established
concept in quantitative chemistry and is defined as the molecular weight of the analyte,
Al, divided by the molecular weight of the form it is in, alumina.

Table 10 presents these spike recovery data. Spike recovery data are presented in the
final two columns to represent the mean spike recovery and the % variance (based
upon 1 a of triplicate analyses performed on the spiked sample pellet) about that
mean. Fluorine was not observed in the spiked sample despite having been reported in
Tables 7, 8, and 9. As Table 8 demonstrates, fluorine is not dependably quantified at
this level. The values in blue represent those analytes for which spikes were introduced
into the sample. Black values are strictly based upon the dilution effect mentioned
above.

2*MW_of_Al

0.529527

(2* MW _of _AJ + 3* MW _of _0)

A-127

-------
Analysis of Circuit
Board Samples by
XRF

Report

5. Conclusions

Several conclusions may be observed from the data presented here.

•	The Phase 1 sample preparation of cored boards did not provide quality data. This
likely had to do with two aspects. First, these boards are heterogeneous. This can
be seen in the data variability associated with "replicate" samples cored from
different locations on the boards. The second is that the cored boards charred
during analysis. Due to this, we were unwilling to perform replicate analyses on
any of these Phase 1 samples.

•	The Phase 2 efforts to achieve homogeneous samples were successful. Sampling
of several aliquots across the circuit boards followed by milling and riffling has
achieved reproducible results. This is observed, in particular, in Table 7 where
replicate samples were prepared.

•	From this it may be inferred that the circuit boards are heterogeneous. The
analysis of cored single disks, while the cheaper approach, does not provide
dependable data. This was seen in the phase 1 analyses.

•	Pellets prepared from these powdered samples are robust and may be used for
multiple analyses without significant deterioration.

•	The cryo-mill is an appropriate approach to powdering this type of sample. Other
mills, hammer and ball mills may also work.

•	It is unclear whether the flaked material found after the first milling represents the
effect of surface coating or not. It is also possible that it is the result of samples
larger than desirable for that size sample container on the cryo-mill

•	The pellets prepared by the methods described in this memo were of good quality.
However, separation by sieving could have been carried out more extensively and
would have ultimately resulted in pellets that were stronger and more
homogeneous than those achieved during this work.

•	Table 8 describes the short term reproducibility achieved for multiple analyses of a
single pellet. The standard deviations described in this table provide one approach
to detection limits by this method.

A-128

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

•	Table 10 described the recovery of spiked materials. Four elements were
deliberately spiked during these experiments. Recovery for these spikes is very
good. Copper and calcium, in particular, are excellent at 97 and 103 % recovery.
Aluminum and sulfur at 91 and 121 % are also very good recovery. The low
recoveries for lead are not considered significant since this element was not spiked
and because this element is very close to detection limits. This is seen in Table 8
where %RSD for lead is 30% and the individual analyses are only 6-10ppm.

•	The results for chlorine are somewhat unclear. Data for this element shows
somewhat more variance than is seen for most other elements. It must be
considered possible that some or all of the chlorine represents contamination from
the Liquid Binder carrier material, dichloromethane. Two steps, mixing the sample
plus binder till it returns to a free flowing state, and operation of the pellet dies
under vacuum, were specifically included as quality assurance steps to minimize
dichloromethane retention. No proof is available either way. This could be
investigated in future work by preparing pellets with both liquid binder and binder
pellets. The latter are solvent free.

However, the Phase 1 chlorine results are also high and variable. No
dichloromethane was used in the preparation of these Phase 1 samples.

•	When certified standard reference materials are not available for the sample
matrix, spiked samples become the best alternative available. This approach is
highly dependent upon operator experience and attention to detail. Additional
replicates, spiking with other elements would be appropriate for the future.

•	The submittal letter described these samples as "halogen free laminates". This
data found one or more halogen in each sample. Chlorine was found in all
samples, though the source of that chlorine remains an open question. Separate
from chlorine, however, fluorine was found in 1of 7 samples and bromine in 4 of 7
samples. Laboratory contamination does not appear to be a source for either of
these elements.

•	During the quantification process, matrix of these boards was described as an
organic polymeric material. This was used as a "balance compound" during
quantification. This was an assumption in the absence of better information. The
data can be re-calculated should this be an invalid assumption.

•	We have investigated interactions between bromine and arsenic as a result of
questions from the committee. As described in a separate section, it is likely that

A-129

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

the majority of the arsenic response in the high bromine Phase 1 sample is due to
a bromine interference. As described, two corrective approaches are available that
could be investigated and implemented in future work.

A-130

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report
Appendix A

Appendix A: Responses to Questions
A1. Comments from Draft Version

SS = stainless steel

Yes. The appropriate section has been edited.

Homogenization and Sub-sampling section. Does "several gram range" refer to 2 to 10
grams?

Yes, though it is not that specific. The actual capacity is restricted by the mass
that can be held in the 2 buckets. That varies with the density of the material.

What is the composition of this binder? Would it have any influence of the results?

As described in that section, this binder is composed of a polymer dissolved in
dichloromethane at a concentration of 100 mg of polymer per 1 mL of solution.

The exact composition of the polymer is not provided by the manufacturer, of
course; its elemental composition is based upon carbon, hydrogen, oxygen,
and nitrogen (per the retailer's literature).

As an organic structure, the polymer does not have any specific response by
XRF; though it may contribute in some small fashion to the baseline. We have
found no evidence of elemental contamination from this liquid binder material
and it has been used in this laboratory for many years. As described in
previous communications, the solvent, dichloromethane, could contribute to
the chlorine response...if it remained in the pellet until analysis. Our pellet
preparation procedures are designed to prevent residual dichloromethane in
the prepared pellets.

Are there quality controls associated with this (ZIQ+) analysis? Can you briefly mention
what they are?

On a monthly basis, drift is measured and a correction factor is calculated and
stored. This is based upon the analysis of a manufacturer-supplied drift
standard.

A-131

-------
Analysis of Circuit
Board Samples by
XRF

Report

Appendix A

On a monthly basis, control charts are maintained based upon the analyses of
4 historical standards. These control charts are used to alert personnel to
instrumental problems.

For each analysis by this program identification is based upon a
manufacturer's supplied library of peaks.

Additional quality control is based upon what the customer specifies. This can
include replicate analyses of each pellet or other sample form, analyses of
replicate pellets, homogenization procedures, analyses of standard reference
materials, when available, and preparation and analysis of spiked samples.
For the Phase 2 samples, all of these except standard reference materials
were implemented.

Could you express variability as percent coefficient of variation?

This has been done in the pertinent tables.

Could you provide all the raw data for the replicates in an appendix? Printouts of raw
data from the computer would be fine. Since the final mean value is a mean of two
means, would you agree that expressing the standard deviation or standard error with
the means for replicates 1 and 2 would be appropriate?

This raw data will follow separately.

How was the spiking done? Can you add that to the methods section?

A separate experimental section was implemented for this version of the
report. The description of the spiking process may be found there.

Why did the wt% ofAl not increase with spiking? Ca, S and Cu all increased markedly.

Each additional spiking compound acts as a diluent on the others. As such it is
quite possible for a spiked element to be lower on a concentration basis and
yet be correct.

A-132

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report
Appendix A

Could you provide the gravimetric factors for the analytes so that myself and the
partners can understand the calculations? Refers to equation 1

An expanded description of gravimetric factors has been added above. They
may also be found in reference books, such as Lange's Handbook of
Chemistry.

Should the %analytej be expressed as a percent or as a decimal in this equation?

Refers to equation 1

%analyte should be used in the percent form. This is why there is a factor of
100 in the equation.

Why is spikej in the denominator, preceded by a sum sign? I see only one value in
Table 8 (Now table 5). Refers to equation 1

The equation includes a I because there are 3 spiking compounds added to
the sample. J is the counting integer for the multiple spiking compounds and
varies from 1-3. The summation is correct. Sample 7+1 = 2.002, as the final
row of Table 5 indicates.

Why is this so high? (Refers to sulfur) I understand variability around 100% but does
119% suggest a measurement problem? Similar for Br and As - 115%

While sulfur is an element we are "watching," we are not prepared at this time
to declare that there is a problem needing resolution with this element.

Consider equation 2, where the numerator is based upon experimentally
acquired data from the XRF. Similarly, the denominator of 2 comes from
equation 1 and also includes experimentally acquired data; both XRF and
balance. There is variability in both the numerator and denominator of equation
2 and we would need additional data to be certain biases existed here.

Bromine and arsenic are present at 12 ppm in the unspiked sample. For
arsenic, in particular, this must be considered at the detection limit since it was
observed in only 1 of 2 replicate samples. At this level for these elements,
noise becomes more important and the difference between 100 % and 115 %
cannot be considered significant for a single sample.

A-133

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report
Appendix A

Conclusions: Could you explain this sentence? The standard deviation describes the
detection limits? Doesn't it describe the variability around the mean?

One definition of detection limit is na; where n is an integer selected based
upon the desired confidence level. To be done properly, detection limits are
measured using dilute samples. In many cases that is shortcut by using the na
calculation.

Conclusions: Where appropriate, could you provide the detection limits, e.g. for lead?

As described in the previous response, this depends upon the confidence level
desired. N = 3 is generally considered a reasonably conservative approach.

Referring to Table 6, short term reproducibility, of the draft report, we can use
a = 0.000237 weight %. 3a then becomes 0.0007 weight % for lead. This is
strictly an estimate that would need to be confirmed experimentally.

Conclusions: Brian etal, could you elaborate your conclusions here ... e.g. Brian
commented that based on the phase I XRF data, these high chlorine levels may be
accurate. Dennis commented that he saw decreasing CI concentrations as he made
replicate measurements

Simply put, both the range of concentrations and variability are similar between
phase 1 and phase 2 samples. Chlorine in phase 1 samples ranged from 0.5
to 0.9 % and had % RSDs ranging from 10 to 40. Similarly, phase 2 samples
ranged from 0.4 to 1 % while the % RSD of replicate sample preparations was
25 % for sample 7. And, since no binder was used for the phase 1 samples,
there is every indication that the chlorine concentrations observed during
phase 1 are real.

Dennis may be referring to the chlorine data where the replicates could be
exhibiting a decreasing trend with time. This is, however, a small trend, from
0.92 to 0.81 % across triplicate analysis.

All phase 2 samples exceed the "halogen free" definition for chlorine. Sample 7
is simply consistently high across several sample preparations and analyses.

Conclusions: Yes this is correct- can you explain what a "balance compound"is and
how it is used?

A-134

-------
Analysis of Circuit
Board Samples by
XRF

Report

Appendix A

In the absence of information about the organic mass present, the material that
is not observed by XRF, the quantification program will assign the full sample
mass to the analytes observed. This will usually result in unacceptably high,
and wrong, results. Informing the program that there is a balance compound
present avoids this.

Bromine-Arsenic Question

In an e-maildatedJuly1^2008^^^^^^^Htransrr|ittec' a communication
from	regarding a potential interference

between bromine and arsenic by XRF. The following figure was prepared by
from the phase 1 analytical results and was attached to these messages.

Br Vs As in Epoxy Resins

2	3	4	5	6	7	8

Br (%)

Figure A-l. Bromine vs. Arsenic in Phase 1 samples

This graph clearly shows a direct relationship between the Phase 1 bromine and
arsenic results. While there are more than one possible explanation for such a causal

A-135

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report

Appendix A

relationship,

warns of a spectral interference leading to arsenic false

positives. After investigating the data, there is every indication that he is correct.

The instrument is currently not operational while it awaits the arrival and installation of
a new chiller. If the instrument were up, running several known standards would have
been the most appropriate approach to investigating this potential interference. Since
we do not have that option at the moment, the following several paragraphs consider
the question.

Tables A-1 and A-2 provide information on instrumental operational parameters for the
several sub-scans and channels that were used for these analyses. "LOCorr" is the
acronym for line overlap correction; it is marked yes for the all sub-scans and channels.
While the several acronyms used in these tables are not important; what is important is
that:

•	Channel 2 defines the conditions under which the arsenic data was collected

•	Sub-scan 3 defines the conditions under which bromine data was collected

•	Channel 2 instrumental conditions match those used under sub-scan 3

A-136

-------
Analysis of Circuit
Board Samples by
XRF

Report

Appendix A

Table A-l. Arsenic and Bromine Scans

Analyte

Line

Scan or
channel

Use
LOCorr

Measured
(kcps)

LO Corrected
(kcps)

Used
(kcps)

Calculated
(kcps)

Difference
(kcps)

As

KB

Ch 2

Yes

5.539

5.539

5.532

5.532

0

Br

KB1,3

Sc 3

Yes

542.153

542.153

541.46

541.475

-0.016

Table A-2. Line Selection Parameters

Scan or
channel

X-tal

Detector

Collimator
(pm)

Tube
Filter

kV

mA

Start
(°)

End
(°)

Step
(°)

material / (Jtn

Sc 1

LiF220

Scint

150

Brass /100

60

66

14.02

18.58

0.04

Sc 2

LiF200

Scint

150

Brass / 300

60

66

12.02

20.99

0.03

Sc 3

LiF220

Scint

150

None

60

66

26.63

44.98

0.05

Sc 4

LiF220

Scint

150

Al /200

60

66

42.03

61.98

0.05

Sc 5

LiF220

Duplex

150

None

50

80

61.03

126

0.05

Sc 6

LiF200

Flow

150

None

32

125

76.04

146

0.08

Sc 7

Ge

Flow

300

None

32

125

91.05

146

0.1

Sc 8

PE

Flow

300

None

32

125

100.1

114.9

0.12

Sc 9

PE

Flow

300

None

32

125

130.1

147

0.12

Sc 10

PX1

Flow

300

None

32

125

20.08

59.98

0.15

Ch 1

LiF220

Scint

150

None

60

66

40.35

40.35

0

Ch 2

LiF220

Scint

150

None

60

66

43.58

43.58

0

Ch 3

LiF220

Scint

150

Al /200

60

66

45.64

45.64

0

Ch 4

LiF220

Scint

150

Al /200

60

66

51.65

51.65

0

ARCADE

A-137

-------
ARCADE	Analysis of Circuit

Board Samples by
XRF

Report

Appendix A

It is, therefore reasonable to examine the sub-scan 3 data for evidence of spectral
interference. Figure A-2 provides an expanded view of sub-scan 3 in the vicinity of the
arsenic Kp lines. In Figure A-2, we can observe that the bromine Ka1,2 doublet is in
the vicinity of the arsenic Kp lines. The horizontal colored line below the doublet
represents the calculated baseline. The green vertical hashmarks to the right of the
doublet represent predicted arsenic peak locations. As can be seen from the cells at
lower left, the graphic crosshairs are at the arsenic Kp3 line and it can be seen that the
tail of the bromine doublet contributes a non-zero response at this 20 angle. Figure A-3
expands the bromine tail region of this spectrum.

1 I 2 3 I 4 I 5 I 6 I 2 i B ) 3 I 1fl ]

[CIRCUITBOARD IS405 FRONT]

41.0	42.0

PW2404 XRF spectrometer

Line:jfos fKB3~" " 0rder.fi j 71% Diff-dftoOl ' X: [43.616*
Recipe: |EPA Sample par.: fSoldCompounds f<: 43.616 * Y: 12.194 kcps

' Y;p2.194 kcps X]jfi A I

Figure A-2. Sub-scan 3, Bromine doublet

A-138

-------
Analysis of Circuit
Board Samples by
XRF

Report

Appendix A

36-

4-

0 —i—i—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—j—i—i—i—i—|—i—i—i——j—i—i—i—i—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—|—

43.40	43.50	43.60	43.70	43.81

Lj^:p_. 0td9(:||— |||gj x xp|^r— xlfjAjfj

Figure A-3. Bromine Tail in Arsenic Region

Having said that there is spectral overlap of bromine on arsenic, just as

noted, we must also note that Table A-2 says that line overlap correction is
used. Having said that, we must also note that the arsenic response in the LO Corr cell
is identical to the measured value, which would seem to contradict that.

Examining Figure A-3 it looks a lot as if the 5.539 kcps measured value in Table A-2
comes from the difference between the calculated background at the crosshair and the
bromine tail response. The question remains as to whether or not corrective
procedures have been implemented. The Panalyticalsoftware provides 2 approaches
to corrective action that are applicable to interferences. One is the already mentioned
line overlap correction. The other is a line specific, as opposed to sub-scan specific,
background correction procedure. Details on these procedures are not available to the
operator within the IQ+ quantification program.

While the details of such applications as IQ+Metalloids are not available through the
IQ+ program, they can be found via the Setup program. Here we can find that channel
2, arsenic, was set up without any background points. Four are available to provide
from 0th to 4th order regressions of curved backgrounds in the vicinity of an analytical

A-139

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report
Appendix A

channel. By using the channel set button on the bottom of the application specific
page, one arrives at a graphic representation of the appropriate standard. On this
page, there is a box for defining line overlap interferences. For arsenic in the
IQ+Metalloids application no line overlaps are defined.

In summary, the above suggests there is a strong probability that an uncorrected
bromine interference on arsenic exists in this application. Once the instrument is back
up, the new chiller is installed, running of standards while modifying the application;
followed by re-running certain samples would be appropriate.

There are two comments to be made on this subject

•	The applications that are currently on this instrument were set up by the
manufacturer's representative during installation of the software

•	As noted in the last few paragraphs, the operator does not have easy access to
such details as background correction and line overlap correction.

A-140

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report
Appendix B

Appendix B: Laminate Etching and Chlorine Measurements

Both phase 1 and phase 2 samples were sent directly from each manufacturer to
David Bedner at ISOLA. Mr. Bedner prepared the laminates for the experiments by
etching a portion of the copper from the laminate using standard methods and
procedures.

To prepare the copper clad laminates for etching, 33% of the copper was masked with
an acrylic tape and 66% of the copper was left exposed. Standard Cupric Chloride
solution (2.5% Normal, 130°F) was then applied to the laminate using a Chemcut
Etcher model GSK-168 with a line speed of 1.5 feet per minute. Thirty-three percent of
each sample's copper surface remained intact after etching. Once etching was
complete, the samples were sent to the appropriate laboratory for combustion testing
and XRF analysis.

Laminate suppliers certified that the supplied pre-preg samples met the IPC's halogen
free definition of less than 900 ppm chlorine (Table B-1). However, the etching
process described above caused residual chlorine to be left on the laminates, as
demonstrated by a subsequent experiment conducted by ISOLA (Appendix C). As a
result, the measured chlorine levels noted in Tables 6 and 7 of the report should be
considered in the context of the procedures used to etch the laminates. Furthermore,
elemental composition was measured using XRF analysis, which some partners view
as less quantitative than other methods. In addition, phase 1 samples were not
homogenized prior to analysis, whereas phase 2 samples were homogenized.

Dichloromethane was used during homogenization, but specific steps were taken to
prevent the samples from retaining any dichloromethane.

A-141

-------
ARCADE

Analysis of Circuit
Board Samples by
XRF

Report
Appendix B

Table B-1. Laminate suppliers' independent chlorine analyses

Sample Number

Chlorine concentration in the laminate
based upon suppliers analysis by an
independent third party

4

Not provided

5

317 ppm

Method : IC

6

290 ppm

Method: IC

7

265 ppm

Method: IC

Due to this information, which was discovered after original preparation of the report,
DfE would like to alter the tenth conclusion bullet in the report as following (page 15,
second bullet):

"The results for chlorine are higher than predicted based on halogen free definitions
(<900 ppm chlorine) and are likely due to contamination with chlorine during the
etching process when the laminates were prepared. Data for this element also shows
somewhat more variance than is seen for most other elements. A second possibility of
chlorine contamination was the Liquid Binder carrier material, dichloromethane used
for phase 2 sample preparation. Two steps, mixing the sample plus binder till it returns
to a free flowing state, and operation of the pellet dies under vacuum, were specifically
included as quality assurance steps to minimize dichloromethane retention. Chlorine
results for Phase 1 laminates, where no homogenization was done and therefore no
dichloromethane was used, are also high and variable. Therefore, chlorine
contamination likely came from the etching process. To demonstrate this Mr. Bedner
did an experiment comparing chlorine levels of laminates prepared in three different
ways. Results are shown in Appendix C."

A-142

-------
Analysis of Circuit
Board Samples by
XRF

Report

Appendix C

Appendix C: ISOLA Experiment Demonstrating the Impact of the Etching
Process on Chlorine Measurements

Samples of two laminates, one with a brominated flame retardant and one with a flame
retardant that was not brominated, were each prepared one of three ways: 1) copper
was peeled from the laminate, i.e. no etching, 2) copper was etched from the laminate
using the standard method described in Appendix B or 3) copper was etched from the
laminate using the standard method described in Appendix B, followed by an additional
de-ionized water rinse before analysis. Chlorine content was analyzed using XRF and
results were reported as relative chlorine content compared to known quantity of
bromine or another element (proprietary). The results are shown in the Tables and
Figures below. Standard etching resulted in 7-9 times more chlorine compared to un-
etched laminate whereas additional water rinsing yielded only 2-3 times more chlorine
than the un-etched laminate.

Laminate manufacturers typically measure elemental concentrations by IC and believe
this is the most accurate method for determining element levels. XRF was chosen for
this experiment for the objective of determining general differences in composition
between laminate samples, to aid in choosing a diverse set of laminates for Phase II
experiments.

XRF measurement



Br

CI

X

16533-1

96.85

3.15



BrFR No Etch

95.98

4.02





94.69

5.31



Average

95.84

4.16



16533-2

75.20

24.80



BrFR Normal Etch

71.05

28.95





69.30

30.70



Average

71.85

28.15



16533-3

95.47

4.53



BfFR Extra Rinse

89.25

10.75





90.31

9.69



Average

91.68

8.32



16533-4



2.27

72.57

PFR No Etch



4.63

68.57





2.13

72.41

Average



3.01

71.18

ARCADE

A-143

-------
Analysis of Circuit
Board Samples by
XRF

Report

Appendix C



Br

CI

X

16533-5



21.41

56.73

PFR Normal Etch



16.55

61.49





13.07

62.12

Average



17.01

60.11

16533-6



7.54

59.80

PFR Extra Rinse



7.23

58.63





8.81

58.51





7.86

58.98

Chlorine Pick-up from Etcher

c 0.45

<1)

f 0.4
o 0.35


0)

i

I Bromine
I non-Bromine

No etch Normal Extra No etch Normal
etch rinse	etch

Extra
rinse

Sample Conditions

CI pick "normal"	CI pick up X-Rinse

Bromine Samples 9x	2x
non-Bromine

Samples 7x	3x

A-144

-------
FLAME RETARDANTS IN PRINTED CIRCUIT
BOARDS: APPENDIX D

U.S. EPA. Flame Retardant in Printed Circuit
Boards Partnership: Short Summary of
Elemental Analyses. DRAFT. December 9, 2009.

*This Short Summary is based on the work
presented in the following three documents,
which are also included in Appendix D:

ICL Industrial. JR 22 - Br and CI Analysis in
Copper Clad Laminates - part II. February 12,
2009. (See page A-150)

ICL-IP Analysis of Laminate Boards. Memo
from Stephen Salmon. November 16, 2009.
(See page A-152)

Dow. Analysis of Chlorine and Bromine.
November 2, 2009. (See page A-156)

A-145

-------
Flame Retardant in Printed Circuit Board Partnership
Short Summary of Elemental Analyses

December 9, 2009

Dow and ICL-IP tested the seven laminate samples for elemental composition. Dow tested for
bromine and chlorine using neutron activation (NA). ICL-IP tested for aluminum, calcium,
magnesium, and phosphorus using ICP, bromine using titration, and chlorine using ion
chromatography. Results from Dow and ICL-IP are shown alongside prior XRF results.

Aluminum, Calcium, and Magnesium

The partnership had previously decided to analyze levels of aluminum, calcium, and magnesium
to determine whether any of these elements were present as a flame retardant filler, such as
Al(OH)3, Mg(OH)2 or CaCC>3. As is shown in ICL's report, results for Al, Ca, and Mg were not
repeatable. In addition, results were low and further testing showed that Al, Ca, and Mg were
not completely digested in the initial procedure. This led ICL to conclude that the Al, Ca, and
Mg were most likely from glass fiber or glass treatment, and not from a flame retardant filler
(personal communication with ICL, Dec 2009). For these reasons, we do not summarize results
for Al, Ca, and Mg here, but instead focus on phosphorus, bromine, and chlorine.

Phosphorus

As is shown in Table 1 and Figure 1, phosphorus levels are highest in laminate 3. There is some
discrepancy between XRF and ICP results, but both test methods agree that laminate 3 has the
highest level of phosphorus.

Table 1. Phosphorus



Test Method



ICP

XRF

Laminate

wt%

±

wt%

±

1

0.011

0.0068

0.0016

0.00036

2

0.012

0.0013

0.0017

0.00054

3

1.7

0.020

4.2

0.10

4

1.1

0.054

1.4

n/a

5

0.80

0.0065

0.84

n/a

6

0.69

0.0065

0.74

n/a

7

0.52

0

0.68

0.0049

1: Confidence intervals are based on variance among reported
values. It is not possible to determine the extent to which these
intervals account for measurement uncertainty.

n/a: not applicable (not enough data to determine confidence
bounds)

A-146

-------
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0

Phosphorus

4 5
Laminate ID

¦ ICP
~ XRF

Figure 1. Phosphorus levels measured by ICP and XRF

A-147

-------
Bromine

As is shown in Table 2 and Figure 2, bromine levels are highest in laminate 2. There is some
discrepancy in results for laminate 1 (titration results are an order of magnitude higher than
neutron activation results), but keep in mind that prior testing did not show noticeable levels of
brominated dioxins or furans for laminate 1. Laminates 3 through 7 appear to have negligible
amounts of bromine (two to three orders of magnitude lower than for laminate 2).

Table 2. Bromine



Test Method



Titration

Neutron Activation

XRF

Laminate

wt%

±

wt%

±

wt%

±

1

0.7

n/a

0.0017

0.00093

n.d.

n/a

2

8.1

n/a

7.2

0.30

6.1

1.9

3

<0.04

n/a

0.0038

0.000063

0.0047

0.00015

4

<0.04

n/a

0.00054

0.00012

n.d.

n/a

5

<0.04

n/a

0.0026

0.0011

0.0012

n/a

6

<0.04

n/a

0.00011

0.0000098

n.d.

n/a

7

<0.04

n/a

0.0014

0.000079

0.0012

0.00012

1: Confidence intervals are based on variance among reported values. It is not possible to
determine the extent to which these intervals account for measurement uncertainty.

n/a: not applicable (not enough data to determine confidence bounds)





n.d.: not detected











9.0
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0

Bromine

[

1	1







J,





























--~-r

-L|	1	1	1	1	

¦	Titration

¦	NA
~ XRF

4 5
Laminate ID

Figure 2. Bromine levels measured by titration, neutron activation (NA), and XRF

A-148

-------
Chlorine

Table 3 and Figure 3 show noticeably lower chlorine results with neutron activation and ion
chromatography than with XRF (order of magnitude difference), which is as expected under the
revised washing protocols. Despite potential discrepancies between test methods, the results
show that chlorine levels are similar between laminates, and along the order of l/100th to 1/10th
of a percent by weight.

Table 3. Chlorine



Test Method



Ion Chromatography

Neutron Activation

XRF

Laminate

wt%

±

wt%

±

wt%

±

1

0.06

n/a

0.075

0.0013

0.88

0.12

2

0.02

n/a

0.073

0.018

0.59

0.35

3

0.02

n/a

0.062

0.0013

0.52

0.081

4

<0.02

n/a

0.063

0.00065

0.45

n/a

5

0.02

n/a

0.060

0.0023

0.43

n/a

6

0.04

n/a

0.046

0.0033

0.49

n/a

7

<0.02

n/a

0.030

0.0020

1.0

0.065

1: Confidence intervals are based on variance among reported values. It is not possible to
determine the extent to which these intervals account for measurement uncertainty.

n/a: not applicable (not enough data to determine confidence bounds)





1.2
1.0
0.8

sO

| 0.6
0.4
0.2
0.0

Chlorine

I

¦	Ion
chromatography

¦	NA

~ XRF

3 4 5
Laminate ID

Figure 3. Chlorine levels measured by ion chromatography, neutron activation (NA), and XRF
Note: Ion chromatography results for laminate 4 and 7 were below detection limits, and are
shown in Figure 3 as one-half the detection limit.

A-149

-------
A

.ICS. Industrial.

PRODUCTS

BROMINE COMPOUNDS LTD.

P.O.Box 180 Beer-Sheva 84101 Israel

Tel:+972-8-6297001 Fax:+972-8-6297412

Iris Ben-David, Ph.D.

RD Division
www. i cl-industrial .com

bendavidi@icl-ip.com

02/12/2009

To: Pierre Georlette
From: Dr. Iris Ben David

Re: JR 2293 - Br and CI Analysis in Conner Clad Laminates - part II

Following our previous report on the analysis of bromine and chlorine in Copper Clad
laminates (see Appendix-1) we received a request for analyzing the halides in these samples at
levels under 0.5 %. We analyzed the samples using ion chromatography, with detection limit of
0.02 % for chlorine and 0.04 % for bromine.

The results are summarized in the table.

Sample ID

Br Content (%)

CI Content (%)

EPA-1

0.7 1

0.06

EPA-2

8.1 1

0.02

EPA-3

< LOD

0.02

EPA-4

< LOD

< LOD

EPA-5

< I.OD

0.02

EPA-6

< I.OD

0.04

EPA-7

< LOD

< LOD

Notes;

1) Determined by titration - see Appendix-1.

Please let us know if you need any additional analyses for these samples.

With Best Regards,

/^/
-------
Appendix-1: Our report from November 11, 2009 - JR 2283.

11/11/2009

To:	Pierre Georlette

From: Dr. Iris Ben David

Re: JR 2283 -Br & CI Analysis in Conner Clad Laminates

We received seven samples of Copper Clad laminates (marked EPA-1 to EPA-7). We analyzed the samples for
their bromine and chlorine contents. Two of the samples had metal strips on them; we examined only the metal free
section, in comparison with the other samples.

The Br/CI contents are given below:

Sample II)

Br Content

("I Content

EPA-1

0.7% (iO.4%)1

n.d.2

EPA-2

8.1 %3 (± 0.2 %)4

n.d.

EPA-3

n.d.

< 0.5 %4

EPA-4

n.d.

< 0.5 %

EPA-5

n.d.

< 0.5 %

EPA-6

n.d.

< 0.5 %

EPA-7

n.d.

< 0.5 %

Notes:

2)	The uncertainty at 1 % level is 5 %.

3)	n.d. = Not detected.

4)	Average of 5 specimens (including the second set of samples EPA 2).

5)	The uncertainty at 10 % level is 2 %.

The analytical method used has a limit of quantification of 0.5 %. At levels under 0.5 % the uncertainty is >50%.
If the accuracy at lower levels of halides is important and should be determined, we can use a different analytical
method. Upon request, the analytical results will be available within a month.

With Best Regards,

JR 2293 - Br & CI in copper clad laminates - part II.doc

2/2

-------
Date:

November 16, 2009

Subject: Analysis of Laminate Boards.

From:	Stephen Salmon, ICL-IP

Determination of P, Al, Ca, Mg

Analyses were completed on seven laminate boards. The results show repeatability was
very good for P, but very poor for Al, Ca, and to a lesser extent Mg. The nature of the
sample matrix appears to be the problem. Details are given below.

The laminate boards were sampled by taking very thin slices across areas that did not
contain any of the copper cladding. The slivers were cross cut to produce very small
pieces. This material was mixed and sub-sampled for acid digestion to get a
representative sample across the board. It was noted that this cutting procedure produced
some very fine glass dust from the edges of the pieces. Some of this dust was included in
the sub-samples.

The samples were digested with sulfuric acid using nitric acid and 30% hydrogen
peroxide as needed to destroy the organic matrix. The resulting solution contained the
insoluble fiberglass. The digested samples were filtered through 0.45 um polypropylene
syringe filters into 100-mL volumetric flasks and made to volume at 4% sulfuric acid.
The samples prepared in triplicate were analyzed by inductively coupled plasma-optical
emission spectroscopy (ICP-OES) using calibration standards matched to the 4% sulfuric
acid of the samples.

A-152

-------
Results for triplicate analyses of the seven laminate boards are shown in Table 1.

Table 1

ICP Analysis of slivered laminate boards

Sample ID

wt% Al

wt% Ca

wt % P

wt% Mg

EPA-1 A

0.21

0.54

0.017

<0.01

EPA-1 B

0.26

0.62

<0.01

0.010

EPA-1 C

0.19

0.45

0.010

<0.01

EPA-2 A

0.31

0.78

0.011

0.013

EPA-2 B

0.32

0.79

0.011

0.013

EPA-2 C

0.39

0.93

0.013

0.016

EPA-3 A

0.21

0.50

1.71

<0.01

EPA-3 B

0.40

0.32

1.71

<0.01

EPA-3 C

0.48

0.78

1.74

<0.01

EPA-4 A

0.35

0.68

1.14

0.080

EPA-4 B

1.60

3.34

1.07

0.14

EPA-4 C

0.27

0.74

1.16

0.070

EPA-5 A

1.09

0.69

0.80

0.014

EPA-5 B

2.34

0.51

0.81

0.013

EPA-5 C

0.34

0.26

0.80

<0.01

EPA-6 A

2.67

1.63

0.68

0.056

EPA-6 B

2.96

1.37

0.69

0.046

EPA-6 C

2.21

0.72

0.69

0.040

EPA-7 A

2.86

1.74

0.52

0.085

EPA-7 B

3.09

2.14

0.52

0.10

EPA-7 C

1.81

0.96

0.52

0.059

The results show that only P determination was repeatable. To check if the fine glass
dust that was included at various levels in the acid digested samples skewed the results
four of the laminate boards were prepared again in triplicate. This time a single chip of
sample of the desired weight was cut out of three sections of the laminate board. The
acid digestion and ICP-OES analyses were repeated.

A-153

-------
The results of this evaluation are shown in Table 2.

Table 2

Repeat Digestions on single laminate board chips.

Sample ID	wt% Al wt% Ca wt % P wt% Mg

EPA-4 A chip

0.31

0.66

1.18

0.070

EPA-4 B chip

0.22

0.72

1.23

0.068

EPA-4 C chip

0.23

0.73

1.23

0.073

EPA-5 A chip

0.38

0.25

0.81

0.004

EPA-5 B chip

0.80

0.67

0.83

0.010

EPA-5 C chip

0.85

0.57

0.83

0.011

EPA-6 A chip

2.91

1.35

0.63

0.043

EPA-6 B chip

0.77

0.85

0.70

0.018

EPA-6 C chip

1.87

1.29

0.69

0.024

EPA-7 A chip

0.49

0.24

0.50

0.017

EPA-7 B chip

0.39

0.34

0.51

0.016

EPA-7 C chip

0.43

0.35

0.51

0.012

The results show that P again was very repeatable and matched the values from digestion
of the small pieces. Al and Ca, and to a lesser extent Mg, again showed very poor
repeatability.

The acid digestion of the single chip samples resulted in four small sheets of fiberglass
from each sample. These were recovered from the filtration step and the washed
fiberglass was dried and weighed. The fiberglass was subjected to the acid digestion
procedure again and an ICP-OES analysis showed significant and variable amounts of Al
and Ca had not been recovered by the first digestion. Mg showed the same to a lesser
extent, but P was not detected indicating quantitative recovery in the original digestion.

Table 3 shows the results of this evaluation.

Table 3

Redigestion of fiberglass recovered from digestion of single chips.
Sample ID	wt% Al wt% Ca wt % P wt% Mg

EPA-6 A chip 2nd

0.45

0.50

nd

0.016

EPA-6 B chip 2nd

0.64

1.30

nd

0.030

EPA 6 C chip 2nd

0.41

0.086

nd

0.011

The conclusion is that Al and Ca are in the fiberglass or can not be separated from the
sample matrix quantitatively. This is also the case for Mg, but to a lesser extent. P,

A-154

-------
however, is quantitatively recovered from the laminate board matrix by the procedure
used.

Determination of Br and CI

An analysis of slivered laminate board for halogens was attempted by metallic sodium
reflux in isopropanol with silver nitrate titration for Br and CI. Unfortunately, the
laminate board matrix proved to be impervious to extraction by the reagent and this
approach had to be abandoned.

Samples of the seven laminate boards were sent to ICL in Israel for sample preparation
by sodium peroxide bomb. Preliminary results are shown below. Other results are
pending and will be sent when available.

Date:	11/11/2009

To:	Pierre Georlette

From:	Dr. Iris Ben David

Re:	JR 2283 - Br & CI Analysis in Copper Clad Laminates

We received seven samples of Copper Clad laminates (marked EPA-1 to EPA-7). We
analyzed the samples for their bromine and chlorine contents. Two of the samples had
metal strips on them; we examined only the metal free section, in comparison with the
other samples.

The Br/CI contents are given below:

Sample ID

Br Content

CI Content

EPA-1

0.7% (± 0.4 %)1

n.d2

EPA-2

8.1 %3 (± 0.2 %)4

n.d.

EPA-3

n.d.

< 0.5 %4

EPA-4

n.d.

<0.5%

EPA-5

n.d.

<0.5%

EPA-6

n.d.

<0.5%

EPA-7

n.d.

<0.5%

Notes:

1)	The uncertainty at 1 % level is 5 %.

2)	n.d. = Not detected.

3)	Average of 5 specimens (including the second set of samples EPA 2)

4)	The uncertainty at 10 % level is 2 %.

The analytical method used has a limit of quantification of 0.5 %. At levels under 0.5 %
the uncertainty is >50%. A different analytical method will be used to get more precise CI
results. The analytical results will be available within two weeks.

A-155

-------
Triplicate samples were prepared by transferring 0.3 grams respectively into pre-cleaned
0.25-dram polyethylene vials. Samples were measured for thickness and cleaned with
isopropanol prior to placing into the vials. Areas with copper were not sampled.
Standards of chlorine, bromine were prepared from standard solutions and placed into
pre-cleaned 0.25 dram vials. The standards were diluted to the same volume as the
samples and the vials heat-sealed. The samples, standards and blanks were irradiated and
counted in four batches. Triplicate samples of EPA -2 were irradiated separately using
O.Olgrams. The higher concentration of bromine identified interferes with the detection
of chlorine. Thickness was measured in triplicate using a micrometer.

Sample ID

20 min @ 250 kW

10 min @250 kW

10 min @30 kW
10 min decay

CI (ppm)
td =1 h
= 1 h

Br (ppm)
td =1 h

tc= 1 h

CI (ppm)
td =1 h

tc= 1 h

Br (ppm)
td =1 h

tc = 1 h

CI (ppm)
td =1 h

tc= 1 h

Br (ppm)

EPA 1

760±40

15.5±0.8

740±40

9.7±0.5

740±40

25.9±1.3

EPA 3

630±30

38.2±1.9

630±30

37.8±1.9

610±30

37.1±1.9

EPA 4

640±30

4.5±0.2

630±30

5.2±0.3

630±30

6.5±0.3

EPA 5

600±30

20.6±1.0

580±30

37.8±1.9

620±30

20.1±1.0

EPA 6

440±20

1.0±0.1

440±20

1.1±0.1

490±20

ND@2ppm

EPA 7

290±10

13.3±0.7

320±20

14.7±0.7

290±10

14.0±0.7



10 min@5kw: CI td =10 min, tc = 7 min;

3r td = 5 hour, tc = 1.5 hour

Sample ID

CI (ppm)

Br (wt%)

CI (ppm)

Br (wt%)

CI (ppm)

Br (wt%)

EPA 2

650±130

6.9±0.3

920±180

7.4 ±0.4

630±130

7.3±0.4

Thickness

Inch

Inch

Inch

Average± Stdev

EPA 1

0.018

0.021

0.019

0.019±0.002

EPA 2

0.016

0.018

0.018

0.018±0.001

EPA 3

0.019

0.019

0.02

0.020±0.001

EPA 4

0.018

0.017

0.02

0.019±0.001

EPA 5

0.018

0.018

0.018

0.018±0.001

EPA 6

0.017

0.017

0.017

0.017±0.001

EPA 7

0.018

0.018

0.018

0.018±0.001

A-156

-------
FLAME RETARDANTS IN PRINTED CIRCUIT
BOARDS: APPENDIX E

University of Dayton Research Institute. Use of
Cone Calorimeter to Identify Selected
Polyhalogenated Dibenzo-P-Dioxins/Furans and
Polyaromatic Hydrocarbon Emissions from the
Combustion of Circuit Board Laminates. October
22, 2013.

A-157

-------
USE OF CONE CALORIMETER TO IDENTIFY SELECTED POLYHALOGENATED
DIBENZO-P-DIOXIN S/FURAN S AND POLY AROMATIC HYDROCARBON
EMISSIONS FROM THE COMBUSTION OF CIRCUIT BOARD LAMINATES

Final Report

Prepared for the U.S. Environmental Protection Agency

Sukh Sidhu, Alexander Morgan, Moshan Kahandawala, Kavya Muddasani
University of Dayton Research Institute
300 College Park, Dayton, OH 45469

Brian Gullett, Dennis Tabor
U.S. Environmental Protection Agency
Office of Research and Development
Research Triangle Park, NC 27711

October 22, 2013

A-158

-------
TABLE OF CONTENTS

1	Executive Summary	167

2	Introduction	169

2.1	Electronic Waste	169

2.2	Performance Requirements for Printed Circuit Boards	170

2.3	Project Goal	171

3	Experimental Methods	171

3.1	Laminate Preparation	173

3.2	Component Mixture Preparation and Component Mixture Grinding	175

3.3	Combustion Testing	176

3.3.1	Cone Calorimeter Apparatus Description	176

3.3.2	Cone Calorimeter Testing Methods	179

3.3.3	Sampling Train	179

3.3.4	Samples Tested	182

3.4	Sample Handling and Custody	183

3.4.1	Shipping Custody	183

3.4.2	Sample Identification and Log	183

3.5	By-product Extraction	183

3.5.1	Organic Compound Target List	184

3.5.2	EPA-RTP Experimental Strategy	184

3.5.3	Same-Sample Extraction of PCDD/Fs and PBDD/Fs	186

3.5.4	Cleanup and Fractionation of PCDD/Fs and PBDD/Fs	186

3.6	Dioxin/Furan Analysis	187

3.6.1	HRGC/HRMS Calibration and Maintenance	187

3.6.2	HRGC/HRMS Analysis	187

3.6.3	Data Processing and Reporting	188

3.6.4	Quality Assurance/Quality Control	188

3.6.5	Pre-Sampling Spikes Quality Criteria and Performance	189

3.6.6	Pre-Extraction Spikes Quality Criteria	190

3.7	Polyaromatic Hydrocarbon Analysis	192

3.8	Organophosphorus and Chlorinated Benzene/Phenol Analysis	193

4	Results and Discussion	193

A-159

-------
4.1	Total Mass Burned	193

4.2	Smoke	194

4.3	CO/CO2 Emissions	196

4.4	Particulate Matter Emissions	198

4.5	PBDD/Fs and PCDD/Fs Emission Factors	199

4.6	PAH Emissions	202

4.7	Heat Release (Flammability) Results	209

5	Conclusions	212

6	Acknowledgments	213

7	Appendix A: Circuit Board Flammability Data	215

8	Appendix B: Experimental Conditions	234

9	Appendix C: Elemental Analyses of Component Mixtures	235

10	References	236

A-160

-------
LIST OF TABLES

Table 2-1. E-Waste by Category in 2010	170

Table 3-1. Overview of Phase II Testing Methodology	172

Table 3-2. Emission/Combustion Tests for Phase II DfE Work	173

Table 3-3. Copper Area of Circuit Board Laminates	174

Table 3-4. Blend of Components to Mimic Circuit Board Components	176

Table 3-5 Laboratory ID Coding System	183

Table 3-6. PCDD/Fs and PBDD/Fs Target Analytes	184

Table 3-7. Composition of the PCDD/Fs Sample Spiking Solution	189

Table 3-8. Composition of the PBDD/Fs Sample Spiking Solution	189

Table 3-9. Pre-Sampling Spike Recovery Limits [%]	190

Table 3-10. Pre-Extraction Spike Recovery Limits [%]	191

Table 4-1. Total Mass Burned Per Sample	194

Table 4-2. Smoke Release Data	195

Table 4-3. CO/CO2 Emission Factors	197

Table 4-4. PM Emission Factors	198

Table 4-5. PBDD/Fs Emission Factors	201

Table 4-6. PAH Emission Factors from EPA List of 16* Priority PAHs for BFR and NFR at 50
and 100 kW/m2	206

Table 4-7. PAH Emission Factors from EPA List of 16* Priority PAHs for HFR and 1556 HFR
at 50 and 100 kW/m2	207

Table 4-8. Toxic Equivalent Emission Factors of Carcinogenic PAHs from EPA List of 16*
Priority PAHs	207

Table 4-9. Organophosphorous Compounds Detected	209

Table 4-10. Heat Release Summary for Laminates and Laminates + Component Powders Tested
at 50 kW/m2	211

Table 4-11. Heat Release Summary for Laminates and Laminates + Component Powders Tested
at 100 kW/m2	212

Table 7-1. Heat Release Rate Data (50 kW/m2)	216

A-161

-------
Table 7-2. Heat Release Data (100 kW/m2)	228

Table 8-1. Ambient Conditions during Cone Testing	234

Table 9-1. Elemental Analyses of Component Mixtures	235

A-162

-------
LIST OF FIGURES

Figure 3-1. Overview of Workflow for Combustion Testing and Analysis	172

Figure 3-2. NFR Sample	175

Figure 3-3. BFR Sample	175

Figure 3-4. HFR Sample	175

Figure 3-5. 1556-1II R Sample	175

Figure 3-6. Cone Calorimeter Schematic	177

Figure 3-7. Total Sampling Train Coupled with UDRI Cone Calorimeter	181

Figure 3-8. Schematic of Total Sampling Train	182

Figure 3-9. Original RTP Experimental Strategy	185

Figure 4-1. Smoke Release Plot	196

Figure 4-2. CO/CO2 Emission Factors Plot	198

Figure 4-3. Particular Matter (PM) Emission Factors	199

Figure 4-4. PBDD/Fs Emission Factors Plot for ND=0 and EMPC=EMPC	202

Figure 4-5. PAH Emission Factors Plotted for Naphthalene and Higher Molecular Weight PAHs
Detected from the EPA List of 16* Priority PAHs	203

Figure 4-6. PAH Emission Factors for Fluorene and Higher Molecular Weight PAHs Detected
from the EPA List of 16* Priority PAHs	204

Figure 4-7. Emission Factors of Carcinogenic PAHs from the EPA List of 16* Priority PAHs 205

Figure 4-8. Toxic Equivalent Emission Factors of Carcinogenic PAHs from EPA List of 16*
Priority PAHs Compared at 50 kW/m2 Conditions	206

Figure 7-1. HRR for BFR Sample	218

Figure 7-2. Final Chars for BFR Sample	218

Figure 7-3. HRR for BFR + P Sample	219

Figure 7-4. Final Chars for BFR + P Sample	219

Figure 7-5. HRR for BFR + PHF Sample	220

Figure 7-6. Final Chars for BFR + PHF Sample	220

Figure 7-7. HRR for NFR Sample	221

A-163

-------
Figure 7-8. Final Chars forNFR Sample	221

Figure 7-9. HRR for HFR Sample	222

Figure 7-10. Final Chars for HFR Sample	222

Figure 7-11. HRR for HFR + P Sample	223

Figure 7-12. Final Chars for HFR + P Sample	223

Figure 7-13. HRR for HFR + PHF Sample	224

Figure 7-14. Final Chars for HFR + PHF Sample	224

Figure 7-15. HRR for 1556 HFR Sample	225

Figure 7-16. Final Chars for 1556 HFR Sample	225

Figure 7-17. HRR for 1556 HFR + P Sample	226

Figure 7-18. Final Char for 1556 HFR + P Sample	226

Figure 7-19. HRR for 1556 HFR + PHF Sample	227

Figure 7-20. Final Chars for 1556 HFR + PHF Sample	227

Figure 7-21. HRR for BFR Sample	229

Figure 7-22. Final Chars for BFR Sample	229

Figure 7-23. HRR forNFR Sample	230

Figure 7-24. Final Chars forNFR Sample	230

Figure 7-25. HRR for HFR Sample	231

Figure 7-26. Final Chars for HFR Sample	231

Figure 7-27. Heat Release Rate Plot	232

A-164

-------
LIST OF ACRONYMS

1556 HFR

1556 halogen-free flame retardant

ASTM

American Society for Testing and Materials

Avg HRR

Average heat release rate

BFR

Brominated flame retardant

CIL

Cambridge Isotope Laboratories

CO/CO2

Carbon monoxide/carbon dioxide

Dffi

Design for the Environment Program

DQI

Data quantity indicator

DQO

Data quantity objective

EM PC

Estimated maximum possible concentration

EMT

Environmental Monitoring Technologies Inc

EPA

U.S. Environmental Protection Agency

E-waste

Electronic waste

FIGRA

Fire growth rate

FMS

Fluid Management Systems Inc

FIT

Fire testing technology

GC/MS

Gas chromatography/mass spectrometry

HFR

Halogen-free flame retardant

HRGC

High resolution gas chromatography

HRMS

High resolution mass spectrometry

HRR

Heat release rate

ISO

International Organization for Standardization

KOH

Potassium hydroxide

LRMS

Low resolution mass spectrometry

MARHE

Maximum average rate of heat emission

NFR

No flame retardant

NGO

Non-governmental organization

NRMRL

National Risk Management Research Laboratory

OSL

EPA Organic Support Laboratory

P

Populated by halogen components

PAHs

Polyaromatic hydrocarbons

PBDD/Fs

Polybrominated dibenzo-p-dioxins/furans

PCB

Printed circuit board

PCDD/Fs

Polychlorinated dibenzo-p-dioxins/furans

Peak HRR

Peak heat release rate

PFK

Perfluorokerosene

PHF

Populated by low-halogen components

PM

Particulate matter

PUF

Polyurethane foam

R&D

Research and development

RoHS

Restriction of Hazardous Substances

RTP

EPA Research Triangle Park

TBBPA

Tetrabromobisphenol A

TEF

Toxic equivalent factor

TEQ

Toxic equivalent quantity

A-165

-------
THR

Total heat release

Tig

Time to ignition

UDRI

University of Dayton Research Institute

UL

Underwriters Laboratories

UV

Ultraviolet

WEEE

Waste Electrical and Electronic Equipment

A-166

-------
1 Executive Summary

The U.S. Environmental Protection Agency (EPA) Design for the Environment (DfE) program
convened a partnership to conduct an alternatives assessment for TBBPA in printed circuit
boards. The partnership determined that combustion testing of sample laminates using the
alternatives would strengthen the assessment and industry decision-making on use of
alternatives. This report explains the outcome of that testing.

The purpose of this study was to understand the potential emissions of halogenated dioxins or
furans and polyaromatic hydrocarbons (PAHs) from burning circuit board laminates. The
methods of this study mimic two types of fire events: open burn and incineration of electronic
waste (e-waste), both of which are used for precious metal recovery. While difficult to model
these two complex fire scenarios exactly, the University of Dayton Research Institute (UDRI)
utilized a cone calorimeter, a fire safety engineering instrument capable of simulating these
scenarios and measuring combustion efficiency.

Combustion conditions, as well as model samples for burning, were selected with input from a
group of stakeholders "Partnership" assembled by DfE. These stakeholders included circuit
board laminate manufacturers, flame retardant producers, government regulators, and non-
governmental organizations (NGOs) with vested interests in the potential emissions from these
burning items. Some stakeholders funded the UDRI experiments while EPA funded the sample
extractions and dioxin/furan analyses.

The results of this study show that when these materials are burned, even at high heat flux that
would attempt to mimic an incinerator, various pollutants are released. Further, flame retarded
materials release more PAHs and other pollutants when burning compared to materials that are
not flame retarded, but this is expected and indicates that the flame retardants are working as
designed. Specifically, the retardation of flame and combustion will result in more incomplete
combustion products.

The combined dioxin/furan and PAH emission studies suggest that circuit board polymers cannot
be analyzed in isolation when determining emissions; the entire populated board must be
considered. While certain pollutants were found in both flame retardant and non-flame retardant
circuit boards, toxicity studies were not conducted. Therefore the relative toxicity of the
combustion by-products from the different laminate formulations can only be partially
calculated.

While the exact flame retardants used in this study were not identified to the Partnership, the
flame retardant chemistry of these materials behaved as expected. Brominated flame retardants
inhibited combustion and produced brominated phenols (detected, but not quantified), dioxins,
furans, and other aromatics during burning. Non-halogenated flame retardants (presumed to be
phosphorus-based) slowed down burning through char formation. This generated more PAHs
than the non-flame retardant circuit boards in certain circumstances (lower heat flux) but less
PAHs when compared to BFRs.

In general, these emissions fit the known combustion chemistry of these flame retardants classes.
Therefore, this study contributes data supporting the approach that, to achieve both fire safety

A-167

-------
and lower emissions, disposal must be done properly with full incineration and appropriate air
pollution control devices in place.

Despite this confirmation of open burning pollution, the study does also leave some questions
unanswered. The results from this study are not definitive regarding which specific pollutants
were released since chemical identification was limited. Further, the results do not show which
chemistries and circuit board components may lead to lower emissions, even under simulated
incineration conditions. A cone calorimeter may not achieve temperatures as high as those of
real-world incinerators. The high heat flux results may not be fully indicative of real-world
emissions should printed circuit boards be put into an incinerator. Because some flame retardants
(including those in this report) inhibit combustion even at very high heat fluxes, additional
research is needed to identify circuit board flame retardant chemistry with lower environmental
and human health impact emissions. Incinerator conditions are likely to reduce the emissions, but
additional emission controls (baghouses, filters) may be needed to prevent all emissions of
concerns as the efficiency of an incinerator is a function of its design and actual operation
temperatures.

Finally, this study demonstrated that the technique of using the cone calorimeter (ASTM E1354)
for emission studies in combination with a custom-built emissions capture sampling train was
successful with small samples. Specifically, the cone calorimeter can be used to collect
emissions from circuit board materials without having to conduct actual open burns. However
this proved to be a labor intensive analytical technique needing refinement of procedures. To
summarize the findings of this study:

50 kW/m2 heat flux:

•	BFR: PBDD/Fs emitted. PAHs emitted at higher levels compared to other samples.

•	HFR: PAHs emitted at higher levels than NFR sample.

•	NFR: PAHs emitted at lowest levels compared to other samples.

100 kW/m2 heat flux:

•	BFR: PBDD/Fs emitted. PAHs emitted at higher levels compared to other samples.

•	HFR: PAHs emitted at lowest levels compared to other samples.

•	NFR: PAHs emitted at a level slightly lower than the BFR sample.

Effect of components on emissions:

•	PBDD/Fs: PBDD/Fs were similar or lower than sample without components.

•	PAHs: In general, presence of components reduced PAH emissions for BFR, were similar or
slightly higher for HFR and were lower for 1556 HFR. The size of these differences varied
depending on how PAHs were defined (see section 4.6).

Smoke, PM, CO and CO2 release:

•	Smoke release was higher for BFR than HFR laminates. Smoke release was higher with
components due to greater amount of material. PM generally had small differences between
samples. There were negligible differences in CO release between samples. CO2 release was
lowest for BFR but with small differences between samples. Results are complex and
smoke/PM results do not always correlate.

A-168

-------
2 Introduction

2.1 Electronic Waste

According to statistics gathered by the Electronics TakeBack Coalition, which were derived from
EPA statistics, 2.4 million tons of e-waste were generated in 2010, only 27% of which was
recycled (see Table 2-1).1 However, with the price of precious metals and rare earths increasing
due to demand and geopolitical issues, there is increased demand to recycle electronics in order
to recover the metals and rare earths. One of the more popular and cost-effective techniques for
this type of metal/rare earth recovery is incineration, which burns off the polymeric components
of the e-waste and leaves behind inorganic ash. This ash can be further smelted down and refined
to isolate the precious metals and rare earths. When incineration is not conducted properly, the
combustion of polymeric components creates toxic by-products that can be released into the
environment. Improper incineration of electronics in developing countries, as seen in popular
magazines like National Geographic2, has led to concerns about the improper disposal of these
products and has influenced the research in this report. Improper disposal of waste that leads to
widespread environmental damage and under-ventilated toxic by-product release is highly
undesirable and illegal in many countries. This issue may be attributable to companies sending e-
waste to countries with looser regulations for improper incineration instead of following
incineration regulatory standards in place in many developed countries. The drivers for improper
waste disposal are numerous, but ultimately financial, and the drive to recover precious metals is
causing more developed countries to keep the wastes inside borders to recycle materials via
internal infrastructure. However, even for operations that will utilize clean burning incinerators
and afterburner/scrubber technology, there still needs to be some knowledge of what is being
released from burning this waste so incinerators can be designed and engineered correctly.

A-169

-------
Table 2-1. E-Waste by Category in 2010

I.-W asle l)\ Ton in 2010

Products

Tolal disposed"" (Ions)

Trashed (Ions)

Kcoclcd (Ions)

Kcocling Kale (%)

Computers

423,000

255,000

168,000

40%

Monitors

595,000

401,000

194,000

33%

Hard copy devices

290,000

193,000

97,000

33%

Keyboards and Mice

67,800

61,400

6,460

10%

Televisions

1,040

864,000

181,000

17%

Mobile devices

19,500

17,200

2,240

11%

TV peripherals*

Not included

Not included

Not included

Not included

Total (tons)

2.440.000

1.790.000

649.000

2 7%

K-\\asle l>\ I nil in 2010

Products

lolal disposed"" (unils)

Trashed (unils)

Kcoclcd (unils)

Recycling Kale ("")

Conipuiei'b

')<)<) ()()()

U 'DO 000

'Ml (iOO OOO

40" u

Monitors

35,800,000

24,100,000

11,700,000

33%

Hard copy devices

33,600,000

22,400,000

11,200,000

33%

Keyboards and Mice

82,200,000

74,400,000

7,830,000

10%

Televisions

28,500,000

23,600,000

4,940,000

17%

Mobile devices

152,000,000

135,000,000

17,400,000

11%

TV peripherals*

Not included

Not included

Not included

Not included

Total (units)

384,000,000

310,000,000

73,700,000

19%

Computer products include CPUs, desktops, and portables.

Flard copy devices are printers, digital copiers, multi-functions and faxes.

Mobile devices are cell phones, personal digital assistants (PDAs), smartphones, and pagers.

*Study did not include a large category or e-waste: TV peripherals, such as VCRs, DVD players, DVRs, cable/satellite receivers, converter boxes,
game consoles.

""Disposed" means going into trash or recycling. There totals don't include products that are no longer used, but which are still stored in homes
and offices.

1 Table adapted from "Facts and Figure on E-Waste and Recycling", Electronics TakeBack Coalition, 2012. Statistics from "Electronics Waste
Management in the United States Through 2009", U.S. EPA, 2011.

2.2 Performance Requirements for Printed Circuit Boards

The materials in printed circuit boards are influenced by performance and regulatory
requirements that must be met by manufacturers. These selections ultimately influence the
emissions from these components when they burn. For electronic products produced today,
numerous environmental requirements must be met. Environmental regulations in the European

"3

Union, namely the Restriction of Hazardous Substances (RoHS) and Waste Electrical and
Electronic Equipment (WEEE)4 directives have been driving the elimination of specific metals
and organic compounds of environmental concern so that incineration and recycling are easier,
and in the event of improper disposal, environmental damage is limited. Regulations from one
nation automatically affect other nations as most electronics manufacturers prefer to produce for
a global market rather than tailor specific products for specific markets that would result in
higher manufacturing and research and development (R&D) costs.

Flame retardants are added to consumer products, including printed circuit boards, to protect
highly flammable polymers against potential fire/ignition risks. The primary fire risk that flame
retardants are protecting against in circuit boards is that of an electrical fault or short circuit
ignition source that can cause the polymer (typically an epoxy) to thermally decompose and
ignite. This ignition site can lead to flame spread across the board and can cause the electronic
casing (also typically made out of flammable polymer) to also ignite, which may lead to flame
spread out of the electronic device into a larger compartment such as a home, a vehicle, or a

A-170

-------
mass transport structure (e.g., subway, train, bus), which may contain other flammable products
that can cause the initial fire to further propagate. If a fire gets out of control, one might
hypothesize that because flame retardants may prevent a product from being fully consumed in
an accidental fire event, there is less total emissions when compared to a non-flame retardant
product that fully ignites. This is especially true if the non-flame retardant product is composed
of a high heat release material which in turn causes other nearby objects to burn and lead to a
large fire event (flashover). It should be pointed out though that this toxic emission reduction
enabled by flame retardant products in the event of accidental fires is only realized in life cycle
models if that product is disposed of properly at the end of its lifetime.5'6' If products are not
disposed of properly then flame retardants have some potential to leach into the environment and
lead to measureable levels of pollution. The flame retardant technology in use today for most
circuit boards typically consists of brominated bisphenol A epoxies that are co-polymerized into
the circuit board, or are reactive phosphorus-based flame retardants that are also co-polymerized
into the circuit board.8'9'10 These technologies have been in use for decades because they are cost-
effective and reliable while not compromising other essential epoxy circuit board properties
(e.g., electrical insulation properties, mechanical). These systems in place today served as the
baseline for the DfE project initially conducted in 2008-09 to study the emissions of circuit
boards using brominated and phosphorus-based flame retardants.11

2.3 Project Goal

The goal of this project was to understand the potential emissions of halogenated dioxins,
halogenated furans, and PAHs and fire characteristics of a standard tetrabromobisphenol A
(TBBPA) laminate compared to different halogen-free laminates in various scenarios with and
without typical circuit board components. The methods of this study mimic two types of fire
events used for precious metal recovery: open burning and proper incineration. Definitions of
open burning and proper incineration are needed here:

•	Open burning means that combustion is done in a crude vessel, open to the environment,
where there are no good engineering measures in place to capture emissions or drive the
combustion process to completion.

•	Proper incineration means that combustion is carried out in a system designed and
engineered to fully combust a material can capture its emissions through the use of
afterburner and baghouse-type emissions capture systems.

The results will provide scientific information to aid electronics and electrical manufacturers in
their decision-making processes to design and choose sustainable and environmentally-friendly
materials for their products.

3 Experimental Methods

A series of circuit boards were selected based on Phase I of this project to be tested under various
conditions mimicking open burning and incineration operations. The components used on circuit
boards were ground up and combusted along with the copper-clad circuit board laminate to
simulate the potential emissions from printed circuit board e-waste. An overview of the testing
methodology for Phase II of this project is provided in Table 3-1.

A-171

-------
Table 3-1. Overview of Phase II Testing Methodology

Laminates Burned (Acronym)

TBBPA laminate (BFR)

Non-flame retardant laminate (NFR)

Halogen-free flame retardant laminate (HFR)
Halogen-free flame retardant laminate (1556-HFR)

Components Burned

Standard halogen components (P)
Low-halogen components (PHF)

Laminate/ Component
Combinations Burned

BFR + standard halogen components (BFR +P)
BFR + low-halogen components (BFR + PHF)
HFR + standard halogen components (HFR + P)
HFR + low-halogen components (HFR + PHF)

1556-HFR + standard halogen components (1556HFR + P)
1556-HFR + low-halogen components (1556HFR + PHF)

Scenarios (Heat Flux)

Open Burn (50 kW/m2) (Laminate Name -50)
Incineration (100 kW/m2) (Laminate Name - 100)

Analytes Tested

Polybrominated dibenzo-p-dioxins/furans (PBDD/Fs)
Polyaromatic hydrocarbons (PAHs)

Multiple entities were responsible for conducting different parts of Phase II's combustion testing
experiment. Figure 3-1 depicts the workflow throughout the project. DfE facilitated and oversaw
the workflow by communicating directly with Isola, Seagate, UDRI, and EPA Research Triangle
Park (RTP).		

Seagate

Component mixture
preparation

Isola

Laminate
preparation

EMT

Component mixture
grinding

~

UDRI

Combustion
testing

~

UDRI

Phosphorus and
PAH analysis

«<

RTP

•	Byproduct
extraction

•	Dioxin/furan
analysis

Figure 3-1. Overview of Workflow for Combustion Testing and Analysis

The circuit board laminates selected and the conditions used to burn the components and circuit
board combinations are shown in Table 3-2. This experimental plan was created with input from
the DfE stakeholders participating in this project including government officials, NGOs, circuit
board laminate manufacturers, electronics producers, and flame retardant producers. The
instrument and method selected to mimic open burning and incineration was the cone
calorimeter, which is a standard fire science measurement tool (ASTM El354, ISO 5660) used to
quantify heat release, smoke release, and CO/CO2 emissions from burning objects in a variety of
fire scenarios. This tool was chosen based on UDRI hypothesis that it could mimic burning
conditions of interest to the program while providing quantitative emissions on complex

A-172

-------
heterogeneous circuit board samples. More specifically, the cone calorimeter provided a dynamic
model in that it could burn a realistic amount of material (an actual circuit board laminate with
components or component mimics) and be instrumented in such a way to capture all of the
emissions from that burning event.

UDRI and EPA conducted the experiments in Table 3-2 in 2011. The original experiment plan
included a third combustion scenario for low-oxygen combustion. These low-oxygen
experiments were not carried out because the low-oxygen attachment for the cone calorimeter
was unable to yield dependable results for simulated smelting conditions at 100 kW/m2 heat flux
at 10% O2. The investigators discovered that when a sample was initially pyrolyzed/burned
under these conditions, combustion gases escaped from the top of the unit where they could
potentially be exposed to more oxygen. This event could lead to a more complete combustion
and thus generate inaccurate results. For reasons of integrity and efficiency, UDRI and the
partnership collectively decided to exclude the 100 kW/m heat flux at 10% O2 test condition
from the study.

Table 3-2. Emission/Combustion Tests for Phase II DfE Work

Ileal
Mux

( omhiislion
atmosphere

Sample
description

# of
blank
I'll its'

# of
laminale
burns

PliDD/l-s

Tesl
lilanks
for
PliDD/l-s

PA lis

Phosphorus

50
kW/m2

Air
(Open-burn)

BFR

2

2

X

X

X

X

BFR + P

2

2





X

X

BFR + PHF

2

2

X

X

X

X

HFR

1

2





X

X

HFR + P

1

2



X

X

X

HFR + PHF

1

2





X

X

1556 HFR

1

2





X

X

1556 HFR + P

1

2



X

X

X

1556 HFR +
PHF

1

2





X

X

NFR

1

2





X

X

100
kW/m2

Air

(Incineration)

NFR

1

2



X

X

X

BFR

1

2

X

X

X

X

HFR

1

2





X

X

Subtotal

16

26









Total (blanks + laminates)

42









for PBDD/Fs carry-over. The blanks were clean; therefore the number of blanks in subsequent sets of samples was
reduced.

3.1 Laminate Preparation

The laminate manufacturer Isola was responsible for laminate preparation. Each laminate was
61cm x 46cm (2,806cm2) and had a 4-ply 2116 Taiwan glass S409 finish. These samples were
prepared by pressing each side of the laminates with loz of shiny copper from Nan Ya and

A-173

-------
etching a portion of the copper from the laminate using standard methods and procedures, just as

12

was done during Phase I testing (see Phase 1 Report) , followed by a rinse with dilute KOH. To
prepare the copper clad laminates for etching, a portion of the copper was masked with an acrylic
tape and the rest of the copper was left exposed. Standard cupric chloride solution (2.5% normal,
266°C) was then applied to the laminate using a chemical etching machine. Etched laminates
were then washed with KOH (2.5% normal) to remove residual chlorine. During preliminary
testing, laminates were washed only with water and not with KOH. However, it is standard
practice in industry to wash laminates with dilute KOH after etching, so the partnership decided
to replicate this approach to reflect real-world conditions.

Due to a miscommunication, Isola initially etched off 25% of the copper, leaving 75% of the
surface area covered by copper. However, the partnership agreed that a copper surface area of
approximately 33% would be more representative of real-world conditions. The copper was
distributed evenly over the surface in a way that allowed UDRI to cut the laminate into 100mm x
100mm squares for combustion testing, each containing an equal amount of copper. In order to
achieve a surface area as close as possible to 33% and also obtain an even distribution of copper,
Isola etched the copper so that 25% remained on one side, and 37.5% on the other side. This
resulted in total surface area coverage of 31%. The total amount of copper present in the actual
samples is shown in Table 3-3. Pictures of representative samples of the four different copper
clad sample types are provided in Figure 3-2 through Figure 3-5.

Table 3-3. Copper Area of Circuit Board Laminates

Sample Deseriplion-lleal l-'lu\ (k\\/iir)

Copper area conlenl l%l

BFR - 50

32.01

BFR - 50

32.56

BFR - 100

32.95

BFR - 100

32.85

BFR + P - 50

33.86

BFR + P - 50

33.50

BFR + PHF - 50

32.85

BFR + PHF - 50

32.76

HFR - 50

32.66

HFR - 50

32.78

HFR - 100

32.72

HFR - 100

32.68

HFR + P - 50

32.98

HFR + P - 50

32.65

HFR + PHF - 50

32.96

HFR + PHF - 50

31.90

1556 HFR - 50

32.92

1556 HFR - 50

32.86

1556 HFR + P - 50

33.12

1556 HFR + P - 50

33.10

1556 HFR + PHF-50

32.87

1556 HFR + PHF-50

32.68

NFR-50

32.75

A-174

-------
Sample Description-Meal l lu\ (k\\/iir)

Copper area conienl ('"¦)

NFR-50

32.80

NFR - 100

32.22

NFR - 100

32.25

Figure 3-2. NFR Sample

Figure 3-3. BFR Sample

Figure 3-4. HFR Sample

Figure 3-5.1556-HFR Sample

3.2 Component Mixture Preparation and Component Mixture Grinding

Seagate prepared a standard mixture of components, which Environmental Monitoring
Technologies, Inc. (EMT) ground up and sent to UDRI for combustion testing. The mixture was
combusted with selected laminate samples to simulate populated circuit boards. Both a low-
halogen mixture and a standard halogen mixture were prepared and were added to the laminates.
To the extent possible, the types of components in the low-halogen and standard halogen
mixtures were made identical. Seagate formulated and supplied the mixtures based on the
electronic components found on standard disk drive boards. Seagate provided as much detail as
possible about the composition of the ground-up mixtures and calculated the amount to add to
each laminate sample. The mixtures included integrated circuits, resistors, capacitors, connectors
(main source of plastic housing), shock sensors, and accelerometers. The partnership decided to
grind up components into a mixture prior to combustion testing. The blend of components that
was ground up to mimic circuit board components is shown in Table 3-4. Since the chemical

A-175

-------
composition of the component mixtures will determine emissions, Seagate provided information
on the chemicals present in the component mixtures, which is shown in Appendix C: Elemental
Analyses of Component Mixtures.

There are a few advantages to using ground-up components instead of whole components:

•	More reliable results: Combustion results are consistent for ground-up components, but
are not consistent for whole components. This is because small changes in the placement
of whole components on the boards can affect the amount and type of materials that come
into contact with each other during combustion, which affects the formation of
combustion by-products.

•	Better estimate of worst-case-scenario: Using ground-up components ensures maximum
contact between component materials and would give a higher probability of producing
combustion by-products.

•	More inclusive sample: Capacitors can be included in the mixture of ground-up
components, as they are not an explosion hazard when ground-up.

•	Less variability in sample preparation: Components do not have to be attached to the
laminate, which removes potential sources of variability (e.g., human error that might
occur while fixing components to the laminate and increased probability of introducing
contaminants).

Table 3-4. Blend of Components to Mimic Circuit Board Components

(nmpnni-nl

Amount (ii)

T\|)ic;il l»( «'

( oinpoiiciil Mix

Resistor (fixed)

0.07

30.77

Capacitor

1.59

694.51

Shock Sensor

0.03

10.94

Xstr (thermistor, bipolar transistor, FET)

0.08

33.19

Frequency Drive

0.06

25.38

EMIRFI Filter

0.02

6.57

Inductor

0.53

229.82

Integrated Circuit (custom drive specific, linear, memory)

1.64

718.82

Connector

3.05

1335.17

Total

7.05

3085.17

Typical circuit board component mass/surface area of board is 0.128 g/cm . The component mixture
loading used for experiments was 0.1 g/cm2 (10±0.05 g/100 cm2 of laminate burned).

3.3 Combustion Testing

3.3.1 Cone Calorimeter Apparatus Description

A cone calorimeter (FTT, United Kingdom) housed at UDRI was modified and used to
characterize emissions from combustion of various printed circuit board laminate samples. The
cone calorimeter is a fire testing instrument which quantitatively measures the inherent
flammability of material through the use of oxygen consumption calorimetry, and is a standard
technique14 under ASTM E-1354/ISO 5660. This instrument was designed primarily as a fire
safety engineering tool, but has found great utility as a scientific tool for understanding fire

A-176

-------
performance in relation to regulatory pass/fail tests as will be referred to in the next paragraph. In
effect, it mimics a well-ventilated forced combustion scenario of an object being exposed to a
constant heat source and constant ventilation (Figure 3-6). This scenario represents many real
world fires where an object or material is aflame and radiates heat to other objects that also catch
fire as a result. The cone calorimeter serves as a very useful fire safety engineering tool by
looking at the heat release rates of a material under these forced conditions.

By studying the various parameters measured by the cone calorimeter, one can correlate the cone
calorimeter measurements to other tests, or, bring understanding of how a material behaves when
a flame is exposed to various fire scenarios. Work on comparing cone calorimeter to other tests
has included full scale flammability tests,15 bench scale tests like UL-94 or limiting oxygen

16 20	21	22

index, " automotive material flame spread tests, wire and cable flame spread tests, and other
types of fire tests/scenarios23"26. A schematic of the cone calorimeter basic setup is shown in
Figure 3-6.

Loser photometer beam
including te-mpe-roture measurement

Te-mperature and differential
pressure measurements token here

Soot sample tube

Exhaust
blouier

Soot collection filter —

Sxhoust
hood

Cone heater
-Spark igniter
Specimen

Load ob 11

Vertical orientation

Figure 3-6. Cone Calorimeter Schematic

Several measurements can be obtained from the cone calorimeter. The cone calorimeter at UDRI
is equipped with a laser for smoke measurements (laser photometer beam in Figure 3-6), oxygen
sensor (paramagnetic) for measuring oxygen consumption, and load cell for measuring mass loss
as the sample pyrolyzes during heat exposure. The instrument at UDRI also has a CO/CO2
(infrared-based) detection system, allowing for the measurement of CO/CO2 production as a
function of time during sample combustion. From these parts of the instrument, various
measurements are collected during each test which can reveal scientific information about
material flammability performance. These include:

A-177

-------
Time to ignition (Tig): Measured in seconds, this is the time to sustained ignition of the
sample. Interpretation of this measurement assumes that earlier times to ignition mean that
the sample is easier to ignite under a particular heat flux.

Heat Release Rate (HRR): The rate of heat release, in units of kW/m , as measured by
oxygen consumption calorimetry.

Peak Heat Release Rate (Peak HRR): The maximum value of the heat release rate during the
combustion of the sample. The higher the peak HRR, the more likely that flame will self-
propagate on the sample in the absence of an external flame or ignition source. Also, the
higher the peak HRR, the more likely that the burning object can cause nearby objects to
ignite.

Time to Peak HRR: The time to maximum heat release rate. This value roughly correlates
the time it takes for a material to reach its peak heat output, which would in turn sustain
flame propagation or lead to additional flame spread. Delays in time to peak HRR are
inferred to mean that flame spread will be slower in that particular sample, and earlier time to
peak HRR is inferred to mean that the flame spread will be rapid across the sample surface
once it has ignited.

Time to Peak HRR - Time to Ignition (Time to Peak HRR - Tig): This is the time in
seconds that it takes for the peak HRR to occur after ignition rather than at the start of the test
(the previous measurement). This can be meaningful in understanding how fast the sample
reaches its maximum energy release after ignition, which can suggest how fast the fire grows
if the sample itself catches fire.

Average Heat Release Rate (Avg HRR): The average value of heat release rate over the
entire heat release rate curve for the material during combustion of the sample.

Starting Mass, Total Mass Lost, Weight % Lost: These measurements are taken from the
load cell of the cone calorimeter at the beginning and end of the experiment to see how much
total material from the sample was pyrolyzed/burned away during the experiment.

Total Heat Release (THR): This is measured in units of MJ/m and is the area under the heat
release rate curve, from time to ignition to time to flameout, representing the total heat
released from the sample during burning. The higher the THR, the higher the energy content
of the tested sample. THR can be correlated roughly to the fuel load of a material in a fire,
and is often affected by polymer chemical structure.

Total Smoke Release: This is the total amount of smoke generated by the sample during
burning in the cone calorimeter from time to ignition to time to flameout. The higher the
value, the more smoke generated either due to incomplete combustion of the sample, or due
to polymer chemical structure. Note that this is a light obscuration measurement, and the
smoke measurement does not discriminate between particulate matter (PM) which obscures
light and organic vapors/pyrolyzed molecules which also may obscure light.

Maximum Average Rate of Heat Emission (MARHE): This is a fire safety engineering
parameter,27 and is the maximum value of the average rate of heat emission, which is defined
as the cumulative heat release (THR) from time t=0 to t divided by time t. The MARHE can
best be thought of as an ignition modified rate of heat emission parameter, which can be
useful to rank materials in terms of ability to support flame spread to other objects.

Fire Growth Rate (FIGRA): This is another fire safety engineering parameter, determined by
dividing the peak HRR by the time to peak HRR, giving units of kW/m2 per second. The
FIGRA represents the rate of fire growth for a material once exposed to heat, and higher
FIGRA suggest faster flame spread and possible ignition of nearby objects.

A-178

-------
• CO/CO2 Yields: This is the total measured amounts of CO/CO2 measured during testing,
pre-ignition and post-ignition. The yields are in units of kg gas (CO, CO2) per kg sample.

3.3.2	Cone Calorimeter Testing Methods

Circuit board samples were provided as very thin (0.4mm to 0.6mm thick) epoxy + e-glass
laminates. These laminates contained copper plating in squares on both sides of the laminates
and were cut in such a way that each sample had the same amount of copper metal present in the
same configuration. Since the laminates provided were too large to be tested as is in the cone
calorimeter, the samples were cut into 100 cm2 square (±0.1cm2) pieces for cone calorimeter
testing. Samples were not conditioned in any way prior to testing. All of the samples were tested
as single ply laminates, with some of the laminates also having ground component powder put
upon them in lOg batches prior to testing in the cone. Any powder used was weighed out right
before the cone experiment and spread evenly across the sample surface. The powder was not
conditioned before use but was always kept in a sealed jar and was weighed out with a typical
benchtop digital scale (accurate to +/- lOmg).

Samples tested included epoxy with brominated flame retardant (BFR), epoxy with non-flame
retardant (NFR), and two epoxies each with different halogen-free flame retardant additives
(HFR). Powders put on the board samples include standard halogen-containing component
powder (P) and low halogen-containing component powder (PHF).

Cone calorimeter experiments were conducted on a FTT Dual Cone Calorimeter as per the
ASTM E-1354-07 method at two heat fluxes (50 kW/m2 and 100 kW/m2). Samples were tested
in triplicate without frame and grid, with the back side of each sample wrapped in aluminum foil.
The only deviation from the ASTM method was that an exhaust flow of 15 L/s was used instead
of the standard 24 L/s exhaust flow rate. The lower flow rate was used to better mimic the "open
burning" fire scenario as the normal 24 L/s flow rate would give more oxygen to the fire than is
typically seen in a "open burning" flaming combustion scenario. Heat release rate data from cone
calorimeter can be found in Appendix A: Circuit Board Flammability Data.

3.3.3	Sampling Train

The total sampling train was designed and constructed specifically for these experiments to
collect the total exhaust gas emitted from the combustion of samples in a standard cone
calorimeter (Figure 3-7 and Figure 3-8). Sampling the total exhaust reduces the amount of
sample that has to be burned to characterize and quantify emissions. The exhaust duct on the
FTT Dual Cone Calorimeter from Fire Testing Technology Limited, UK, was modified to enable
connecting of the total sampling train. The exhaust hood above the combustion zone was
connected to the sampling exhaust duct (110mm in diameter) with a cooling jacket (not used for
these experiments). The sampling exhaust duct was connected to a stainless steel filter holder
61cm x 25.5cm x 2.5cm. The filter holder holds three 20.5cm x 25.5cm filters. The filter holder
was connected to an amber-glass coiled-condenser to cool the hot gas flowing before it entered
an amber-glass cartridge containing four polyurethane foam (PUF) cartridges of 10cm x 5cm
meant to capture semi-volatile organic compounds. Amber glass is important to note here since
many of the chemical species of interest in this study can be UV light sensitive. The PUFs were
retained by a fritted Teflon disk inside the cartridge. The gas exiting the PUFs was passed

A-179

-------
through an impinger which was connected to a vacuum pump and the gas exiting the pump was
directed to the cone calorimeter exhaust system through a wire reinforced vacuum tube.

At the beginning of each sampling period after assembling the sampling train, the system was
checked for leaks. Once any leaks were fixed, the air flow was set to 15 L/s by turning the
vacuum pump on and using a gate valve to control the air flow. All the circuit board laminate
samples tested were exposed to a heat flux of 50 kW/m2 or 100 kW/m2 For additional details on
the cone heater temperature (which is not the temperature that the samples encountered during
burning), see Appendix B: Experimental Conditions. Once the cone reached its set temperature,
the cone calorimeter ignition was turned on and samples were placed in the sample holder at the
center of the cone heater and ignited. Once the samples ignited, they were allowed to burn until
no flame and smoke were detectable. During sampling, the gas temperature inside the sampling
train was constantly monitored at eight different positions. The first two thermocouples (T1 and
T2) were placed inside the stainless steel duct at 5cm and 25.5cm from the exhaust hood above
the cone to monitor the gas temperature entering the duct (Tl) and entering the filter holder (T2).
The third thermocouple (T3) was placed at the outlet of the filter holder (or entrance of
condenser). The fourth thermocouple (T4) was positioned at the inlet of the PUF cartridge and
the fifth thermocouple (T5) was placed to monitor the gas temperature exiting the PUF cartridge.
The cold bath temperatures are adjusted to maintain the PUF cartridge exit gas temperatures (T5)
to ~20-25°C. However, the average gas temperatures exiting the PUFs were ~30°C for all
experiments. The other thermocouples were used to monitor the water bath temperatures for the
stainless steel duct water jacket, the condenser, and the glass cartridge water jacket. All
thermocouples used were 3mm sheath diameter, grounded, type K thermocouple probes from
Omega Engineering, Stamford, Connecticut. During sampling, the pressure dropped inside the
sampling train and the flow through the sampling train was constantly monitored by a digital
gauge manometer placed at the pump inlet and by a differential flow meter on the cone
calorimeter exhaust system, respectively. When the soot particles started to build up on the glass
filter and decreased the gas flowing through it, the flow was adjusted by opening the gate valve
situated at the inlet of the pump.

Post-sampling, the sampling train was disassembled; the condensate from the condenser was
recovered to a pre-cleaned container for analysis, the various components of the train were
covered with hexane-rinsed aluminum foil and transported to the recovery lab. In the recovery
lab, the filters and PUFs were removed, the filters were weighed to determine their PM loading
and the entire sampling train (from the hood and duct work above the cone/combustion zone) up
to the inlet of the impinger was rinsed with three solvents (methanol, methylene chloride and
toluene, respectively) to recover condensed material for analysis. All solvent rinses, condensate,
PUFs and filters were stored in pre-cleaned amber glass containers with Teflon lined caps; the
solvent levels were marked with the appropriate labels; and were refrigerated till they were either
shipped to the analytical lab or were analyzed at UDRI using GC/MS. The glass fiber filter and
PUF adsorbents were shipped to the Organic Support Laboratory (OSL) of EPA at RTP where
they were combined together, extracted, and analyzed for PxDD/Fs. After extraction, the OSL of
EPA at RTP shipped back a part of the PUF and Filter extract to UDRI to analyze for PAHs and
phosphorous-containing compounds. The analytical methods used to quantify involved isotope
dilution and internal standard procedures that are described later in Sections 3.6 through 3.8.
After the final solvent rinse (i.e., toluene), the metal duct and filter holder were rinsed with

A-180

-------
methylene chloride and covered with hexane-rinsed aluminum foil until the next experiment; the
glassware was rinsed with Sparkleen soap solution/deionized water and baked at 475°C for 8
hours in a Barnstead Thermolyne Pyro-clean Trace oven for baking glassware. After baking, the
glassware was rinsed with methylene chloride and covered with hexane-rinsed aluminum foil. A
field blank was performed to check for carry over and memory effects.

All fluorescent lights in the laboratory, as well as in the fume hood, were covered with clear UV-
absorbing filters supplied by UV Process Supply, Chicago, Illinois. This was done to
minimize/eliminate decomposition of UV light sensitive compounds from the pre-sampling
surrogates and samples recovered from the experiments. The three solvents used were toluene
(Envisolv, 34413) and Methanol (Pestanal, 34485) purchased from Sigma-Aldrich, Milwaukee,
Wisconsin and Methylene Chloride (Pestisolv, PS 724) purchased from Spectrum Chemicals,
New Brunswick, New Jersey at purity levels required as per EPA method 23 for analysis of
dioxins and furans. The 150 mm glass-microfiber filters (TE-EPM2000) without binder were
purchased from Whatman, USA. The PUFs were purchased from Tisch Environmental. The
PlJFs and the filters were cleaned by the OSL at EPA, RTP by Soxhlet extraction with
methylene chloride for 16 hours and wrapped in aluminum foil, labeled, and shipped to UDRI in
airtight cans to use for sampling.

Impinger

out et ine connected to exhaust duct

Pump

Filter Holder

Condenser

Cone Heater

Cooling Water system

Chiller

Figure 3-7. Total Sampling Train Coupled with UDRI Cone Calorimeter

A-181

-------
sanpling line
for Cjj/CCi
nalysis

Note: the
sanpling lines
loci ions (f or
CyCOj aid
ions aialysis)
3 e iubj ai to
changes
bsed on
shi edown
eKperiments

-------
2

had to be tested; they were cut into 100cm square pieces. Four types of laminates were tested for
Phase II: laminate without flame retardant (NFR), laminate containing brominated flame
retardant (BFR), laminate containing halogen-free flame retardant (HFR), and laminate
containing halogen-free flame retardant (1556-HFR). The printed circuit board laminate samples
were tested at two different heat fluxes to mimic different combustion scenarios. The lower heat
flux (50 kW/m ) was used to mimic an "open burn" type of event and the higher heat flux (100
kW/m2) was used to mimic an incinerator furnace condition that would be encountered during
incineration of the boards.

3.4 Sample Handling and Custody

3.4.1 Shipping Custody

Samples were collected at UDRI, packaged, and shipped by UPS to RTP. In RTP, the samples
were received and brought to the laboratory and then opened by the laboratory custodian. The
samples were stored in laboratory refrigerators until extraction. The sample custody form was
included in the shipping cooler, and the UPS records are the custody records for the transfer from
UDRI to RTP. The boxes and coolers were sealed with tape and the tape was removed in the
laboratory.

3.4.2 Sample Identification and Log

Each sample was given an identifying laboratory code number and name (laboratory ID). The
laboratory ID was assigned to the samples upon receiving and samples were logged in the
sample ID log book along with the sample name and project description. The code sequence was
explained to the laboratory personnel to prevent sample mislabeling. Proper application of the
code simplified sample tracking throughout the handling, analysis, and reporting processes.
Table 3-5 shows the laboratory ID coding that was used in this study. PUF and Filters were not
given separate numbers.

Table 3-5 Laboratory ID Coding System

YYMMXX

Laboratory
ID Code

Sample Type

YYMM Year and month of the sample logging in the laboratory system
XX	Consecutive sample number of the given year (YY) and month (MM)

3.5 By-product Extraction

After the samples were collected and shipped back to RTP, the EPA OSL performed extraction,
cleanup, and fractionation of samples provided by UDRI. The extracts were analyzed using High
Resolution Gas Chromatography/High Resolution Mass Spectrometry (HRGC/HRMS) for target
PCDD/Fs and PBDD/Fs (Table 3-6). The results were reported in a spreadsheet to UDRI for
inclusion in the final report (results were reported as amounts per sampling train). In very early
samples, less than ten percent of the dioxins and furans were found in the sampler rinses and the
rinses would cause very high shipping costs, so only the PUF and filters from each sample were
sent to RTP for extraction and analysis.

A-183

-------
3.5.1 Organic Compound Target List

Chlorinated and brominated dioxins and furans (PCDD/Fs and PBDD/Fs, respectively) were
targeted in this project. Analysis concerned 2,3,7,8-substituted congeners of PCDD/Fs (17
congeners) and their brominated counterparts (only 13 2,3,7,8 PBDD/Fs congeners were reported
due to limited availability of commercial standards). Table 3-6 presents the congener-specific list
of PCDD/Fs and PBDD/Fs target analytes.

Table 3-6. PCDD/Fs and PBDD/Fs Target Analytes

PCDD/Fs targets	PBDD/Fs targets

TeCDD	TeBDD

PCDD*	PBDD

HxCDD	HxBDD

HxCDD	HxBDD

HxCDD	HxBDD

HpCDD	HpBDD

OCDD	OBDD

Congener
Pattern

2,3,7,8

1,2,3,7,8
1,2,3,4,7,8

1.2.3.6.7.8

1.2.3.7.8.9
1,2,3,4,6,7,8

2,3,7,8	TeCDF	TeBDF

2,4,6,8	***	TeBDF**

1,2,3,7,8	PCDF	PBDF

2,3,4,7,8	PCDF	PBDF

1,2,3,4,7,8	HxCDF	HxBDF

1.2.3.6.7.8	HxCDF	***

1.2.3.7.8.9	HxCDF*	***
2,3,4,6,7,8	HxCDF*	***

1.2.3.4.6.7.8	HpCDF	HpBDF

1.2.3.4.7.8.9	HpCDF

1,2,3,4,6,7,8,9	OCDF	OBDF
* Were reported as co-elution.

** FromTeBDF homolog group 2,4,6,8 -TeBDF can be reported because it was present in the calibration solution
and therefore has an accurate retention time.

*** In the various calibration solutions, 18 different congener patterns were included, e.g. 2,3,7,8. Of the 18
individual congener patterns that were looked for, five were only in one of the solutions (either bromo or chloro).

3.5.2 EPA-RTP Experimental Strategy

Figure 3-9 presents the original experimental strategy for RTP's part of the project. The first
phase of this project was extraction, cleanup and fractionation (described in detail in Section
3.5.3 and Section 3.5.4 of this report) of samples provided by UDRI for HRGC/HRMS
instrumental analysis of PCDD/Fs and PBDD/Fs. The second phase described in detail in Section
3.6.2 was the instrumental analysis. The third phase of the analysis was data processing and
reporting (see Section 3.6.3 for details).

A-184

-------
Figure 3-9. Original RTP Experimental Strategy.

The actual work added a step to the PCDD/Fs cleanup and dropped the PBDD/Fs cleanup.

A-185

-------
3.5.3 Same-Sample Extraction of PCDD/Fs and PBDD/Fs

Extraction of sampling trains for PBDD/Fs and PCDD/Fs measurements was performed by
sequential Soxhlet extraction: overnight (16 hours) with methylene chloride, followed by
overnight (16 hours) extraction with toluene. This project had such a large sample volume that
the regular 3.5 hours methylene chloride extraction did not give enough cycles for the extraction.
Before extraction, samples were spiked with the internal standard mixtures. Pre-extraction spikes
were purchased from Cambridge Isotope Laboratories Inc., Andover, Massachusetts (EDF-5408,
EDF-4137A). The composition of 13C-labeled PCDD/Fs and PBDD/Fs pre-extraction internal
standard mixes is given in Table 3-7 and Table 3-8. All solvents were
HPLC/GC/spectrophotometry grade ACS/HPLC certified (Burdick and Jackson, Honeywell,
Muskegon, Michigan).

3.5.4 Cleanup and Fractionation of PCDD/Fs and PBDD/Fs

For determination of PBDD/Fs and PCDD/Fs, one-quarter of the extract was cleaned and
fractionated using an automated liquid chromatography multicolumn Power Prep/Dioxin System
(FMS Fluid Management Systems, Inc., Watertown, Massachusetts). One-twentieth of the
extract was sent to UDRI for further analysis of other target compounds. The remainder of the
extract was archived. Prior to the automated cleanup process, extracts were concentrated and
then diluted in hexane, causing precipitation of non-dioxin-like compounds that could have
caused interferences in the analysis. This step was repeated until no more precipitate formed and
the extract was less than ten percent toluene. The extracts were then loaded and pumped
sequentially through individual sets of FMS proprietary columns. Acidic and multilayer silica,
carbon, and alumina columns were pre-packed, disposable cartridges available from FMS Fluid
Management Systems, Inc., U.S.A. The previous experiments on HRGC/HRMS analysis of
some combustion-related matrices showed interferences from other compounds that interfere
with quantitative determination of the target compounds (PCDD/Fs and PCBs)1. This
interference necessitates the introduction of an additional cleanup step, prior to the usual
automated PowerPrep liquid chromatography cleanup used in the OSL for same-sample
determination of PBDD/Fs and PCDD/Fs from combustion flue gas. The additional step
involved passing the extract through a large acidic silica gel column for the cleanup of the raw
extract and concentration of the eluate to 0.5ml. This additional cleanup step was repeatedly
performed until the extract was clear at 0.5ml volume. If the extract was not clear the eluate was
diluted to 12ml with hexane and processed again. This clear 0.5ml of extract was then diluted to
12ml in hexane and processed through multilayer silica (4g acid, 2g base, and 1.5g neutral)
column, followed by a basic alumina (1 lg) column and also a carbon column (0.34g).
Composition of elution solutions and elution volumes are presented in Figure 3-9 of this report.
To quantitate the PBDD from a single aliquot of extract, an additional step was added after the
toluene elution of the carbon column, in which the alumina column was washed with 100ml of
methylene chloride and that eluate was concentrated and exchanged into decane. In the later
samples this portion was analyzed separately. It has been determined since the 2009 publication2
that a separate FMS cleanup for the PBDD/Fs was not necessary, just this additional alumina

1	Data not published, information archived and available from OSL.

2	Tabor D., Gullett B.K., Same-Sample Determination of Ultratrace Levels of Polybromodiphenylethers,
Polybromodibenzo-p-dioxins/Furans, and Polychlorodibenzo-p-dioxins/Furans from Combustion Flue Gas. Anal.
Chem. 2009, 81, 4334-4342

A-186

-------
column wash. Also, the removal of the carbon column step completely (as was done previously)
was considered insufficient cleanup for most samples. The final eluates were then spiked with
pre-analysis compounds, and then decane was concentrated to a final volume of about 25|il.

3.6 Dioxin/Furan Analysis

3.6.1	HRGC/HRMS Calibration and Maintenance

EPA methods require that a laboratory record be maintained of all calibrations, including daily
calibration checks. These daily checks ensure continued reliable operation and provide the
operator warnings of abnormal operation.

The following calibration activities were conducted:

•	Daily optimization of the HRMS instrument was carried out using a perfluorokerosene
(PFK) calibration standard; static resolving power checks were performed before and
after data acquisition to demonstrate the required resolution of 10 000 (5% valley).

•	Bromodioxin/furan and chlorodioxin/furan calibration standard solutions (please see
Section 3.5.1. for details) were used for the initial calibration of the HRGC/HRMS. The
medium concentration standard was used for calibration verification according to
requirements of U.S. EPA M-23.3

•	The daily calibration was acceptable if the concentration of each labeled and unlabeled
compound is within the calibration verification limit of 25-30%. If all compounds met the
acceptance criteria, calibration was verified and analysis of standards and sample extracts
proceeded. When any compound failed its respective limit, recalibration for all congeners
was performed. In addition, the ion abundance ratios were within the allowable control
limits of 15%.

Instrument maintenance was conducted as recommended by the manufacturer and on an as-
needed basis. Replacement parts, including columns and filaments, were maintained in the
laboratory to minimize downtime. Service engineers' visits were utilized in major failure
situations and for annual preventive maintenance.

3.6.2	HRGC/HRMS Analysis

For analysis of tetra- through octa-BDD/Fs, the GC was equipped with 15m DB-5 (0.25[j,m film
thickness x 0.25mm i.d.) column (J&W Scientific, Folsom, California). For analysis of tetra-
through octa-CDD/Fs, a 60m RTX-Dioxin-2 (Restek, Bellefonte, Pennsylvania) column was
used (0.25[j,m film thickness x 0.25 mm i.d.).

The GC oven temperature for PBDD/Fs analysis was programmed from 130°C to 320°C at
10°C/min (21 minute hold). The temperature program for PCDD/Fs went from an initial
temperature of 150°C to 260°C at 10°C/min with a final hold time of 55 minutes. The carrier gas
(helium) flow rates were 1 and 1.2ml/min for PBDD/Fs and PCDD/Fs, respectively. The
PCDD/Fs flow was ramped to 1.5ml/min after 15 minutes. Two microliters (2|iL) of the extract

3 U.S. EPA Test Method 23. Method 23 - Determination of Polychlorinated Dibenzo-p-dioxins and Polychlorinated
Dibenzofurans from Municipal Waste Combustors; Office of Solid Waste and Emergency Response, Environmental
Protection Agency: Washington, DC, 1996.

A-187

-------
was injected under splitless mode (injection port temperature set as 300°C and 270°C for
brominated and chlorinated targets, respectively).

The HRMS was operated in an electron ionization (35 eV and 650 [iA current) selective ion
recording (SIR) mode at resolution R > 10 000 (5% valley). The temperature of the ion source
was 280°C for the PBDD/Fs analyses, whereas for PCDD/Fs, the ion source was kept at 250°C.
The two strongest ions in the molecular cluster were monitored in every retention time window
for each native and labeled PBDD/Fs and PCDD/Fs based on mass spectroscopy libraries and
literature data, unless interferences are present. Peak responses for each of the two selected
molecular ion clusters must be at least 2.5 times the noise level (S/N > 2.5), otherwise the
compound was considered below the limit of detection. The bromine/chlorine isotope ratio for
the two molecular ion clusters was within ±15% of the correct isotope ratio, if not they were
flagged EMPC (Estimated Maximum Possible Concentration).

The standards used for PBDD/Fs identification and quantification were a commercially available
set of calibration standards that contained native target tetra- through octabromodioxins and/or
furans at concentrations from 0.4 to 4.0 (CS-2) through 50-500 (CS-5) ng/ml depending on the
degree of bromination (EDF-5407, CIL Cambridge Isotope Laboratories Inc., U.S.A.). The
standards used for chlorinated dioxin/furan identification and quantification were a mixture of

13

standards containing tetra- to octa-PCDD/Fs native and C-labeled congeners designed for
modified U.S. EPA Method 23 (ED-2521, EDF-4137A, EDF-4136A, EF-4134, ED-4135, CIL
Cambridge Isotope Laboratories Inc., U.S.A.). The PCDD/Fs calibration solutions were prepared
in house and contain native PCDD/Fs congeners at concentration from 1 (ICAL-2)-20 (ICAL-6)
ng/ml.

3.6.3	Data Processing and Reporting

For the data collection, Mass Lynx software (Waters, Milford, Massachusetts), version 4.1 was
used (including Target Lynx 4.1. for processing and quantitation). Data processing included not
only the determination of PCDD/Fs and PBDD/Fs concentrations, but also the determination of
the method detection and quantitation limits (LOD and LOQ, respectively). Every set of data was
reported as ng per train. For PCDD/Fs analysis, data would have been reported as ng-TEQ per
train, if the analyses were accepted (pre-sampling surrogate problems will be detailed later).

3.6.4	Quality Assurance/Quality Control

The data quality objectives (DQOs) define the critical measurements needed to address the
objectives of the test program, and specify tolerable levels of potential errors associated with
data collection as well as the limitations of the use of the data. The data quality indicators (DQIs)
are specific criteria used to quantify how well the collected data meet the DQOs. The DQI goals
for the critical measurements correspond to and are consistent with the standards set forth in each
respective referenced EPA Method. DQI goals will correspond to recovery criteria of the labeled
standards in the respective reference methods. The DQI goals specified for the respective
sampling method used by UDRI sampling team, such as pre-sampling surrogates recoveries are
not included in the DQOs, but were reported to UDRI, along with quality criteria guidelines.

A-188

-------
Composition of labeled pre-sampling (surrogate standards), pre-extraction (internal standards)
and pre-injection (recovery standards) spiking solutions are given in Table 3-7 and Table 3-8.

Table 3-7. Composition of the PCDD/Fs Sample Spiking Solution

Spiking Solution

Analytes

Concentration (jig/ml)

Special Notes

Surrogate standards

3 Cl4-2,3,7,8-TCDD

1.25

Added to the sample prior to

(Field spikes)

13C12-1,2,3,4,7,8-HxCDD

2.5

sampling

EDF-4136A*

13C12-2,3,4,7,8-PeCDD

2.5





13C12-1,2,3,4,7,8-HxCDF

2.5





13C12-l,2,3,4,7,8,9-HpCDF

2.5



Internal standards

13C12-2,3,7,8-TCDD

1.25

Added to the sample prior to

EDF-4137A*

13C12-l,2,3,7,8-PeCDD

2.5

extraction



13C12-1,2,3,6,7,8-HxCDD

2.5





13Cn-l,2,3,4,6,7,8-HpCDD

2.5





13C12-OCDD

5





13C12-2,3,7,8-TCDF

1.25





13C12-l,2,3,7,8-PeCDF

2.5





13C12-1,2,3,6,7,8-HxCDF

2.5





13C12-l,2,3,4,6,7,8-HpCDF

2.5



Recovery Standards

13C12-1,2,3,4-TCDD

5

Added to extracts prior to

ED-252i*

13C12-1,2,3,7,8,9-HxCDD

5

analysis

Commercially available from CIL Cambridge Isotope Laboratories Inc., U.S.A.
Table 3-8. Composition of the PBDD/Fs Sample Spiking Solution

Spiking Solution

Analytes

Concentration (ng/ml) Special Notes

Surrogate standard
(Field spikes)
EF-5410*

Internal standards
EDF-5408*

Ci2-l,2,3,4,7,8-TeBDF

3C12-2,3,7,8-TBDD

3Ci2-l,2,3,7,8-PeBDD

3C12-l,2,3,4,7,8-HxBDD

3C12-1,2,3,6,7,8-HxBDD

3C12-l,2,3,4,6,7,8-HpBDD

3C12-0BDD

3C12-2,3,7,8-TBDF

3C12-2,3,4,7,8-PeBDF

3C12-l,2,3,4,7,8-HxBDF

3C12-l,2,3,4,6,7,8-HpBDF

3cv-obdf

Recovery Standards
EDF-5409*

Ci2-l,2,3,7,8-PeBDF
1,2,3,7,8,9-HxBDD

C

100

100
100
250
250
500
750
100
100
250
500
750

100
250

Added to the sample prior
to sampling

Added to the sample prior
to extraction

Added to extracts prior to
analysis

Commercially available from CIL Cambridge Isotope Laboratories Inc., U.S.A.

3.6.5 Pre-Sampling Spikes Quality Criteria and Performance

A group of carbon-labeled PBDD/Fs and PCDD/Fs congeners (Table 3-7. and Table 3-8) were
added to the PUF sorbent before the sample was collected in UDRI. The surrogate recoveries
were measured as relative to the internal standards and were a measure of the sampling train
collection efficiency.

A-189

-------
OSL provided results of pre-sampling spikes recovery to UDRI, using the acceptance criteria
outlined in Table 3-9.

Table 3-9. Pre-Sampling Spike Recovery Limits [%]

Pre-sampling spike

Minimum

Maximum

PCDD/Fs

%

%

37Cl4-2,3,7,8-TeCDD

70.0

130

13C12-2,3,4,7,8-PCDF

70.0

130

13C12-1,2,3,4,7,8-HxCDF

70.0

130

13C12-1,2,3,4,7,8-HxCDD

70.0

130

13C12-l,2,3,4,7,8,9-HpCDF

70.0

130

PBDD/Fs

%

%

13C12-l,2,3,4,7,8-TeBDF

70.0

130

The pre-sampling surrogates recovery acceptance criteria were as recommended by U.S. EPA
Method 23 for chlorinated dioxins 4 There is no standard method guidance for PBDD/Fs pre-
sampling surrogates recovery; hence Method 23 acceptance criteria were used for brominated
targets.

Upon analysis of the PCDD/Fs samples, the pre-sampling surrogates were found to be absent
from seven of the ten samples requested for PCDD/Fs analysis. Because this constituted a large
majority of the PCDD/Fs samples and that there were no PCDD/Fs detected in the first phase of
this project, the investigators decided not to report PCDD/Fs data. In the samples that were
analyzed, there were virtually no PCDD/Fs detected consistent with the first phase of the project
but it would be consistent with complete loss of target compounds which is highly unlikely given
the PBDD/Fs data. Given both of these possibilities, not reporting the data was of the most
objective action.

There was significant brominated interference in 6 of 18 tests. The six tests with bromine
interference were all the samples that had standard halogen-containing ground components
added. This reduced the number of measured experimental samples to 12. In the PBDD/Fs
samples there was also a brominated pre-sampling surrogate. The recoveries for the 12 samples
ranged from 0.8% recovery to 234% recovery. Four samples appear to have been double-spiked
with recoveries near 200% and the sample near 0% recovery was probably not spiked. Five of
the remaining samples were between 90 and 110% recovery. The other two samples had low
recovery which was not likely due to spiking problems.

3.6.6 Pre-Extraction Spikes Quality Criteria

A group of 11 PBDD/Fs and 9 PCDD/Fs 13C-labeled internal standards (see Table 3-7. and Table
3-8), representing the tetra- through octa-halogenated homologs, were added to every sample

4 U.S. EPA Test Method 23. Method 23 - Determination of Polychlorinated Dibenzo-p-dioxins and Polychlorinated
Dibenzofurans from Municipal Waste Combustors; Office of Solid Waste and Emergency Response, Enviromnental
Protection Agency: Washington. DC, 1996.

A-190

-------
prior to extraction. The role of the internal standards is to allow quantification (via the isotope
dilution internal standard methodology) of the native targets in the sample as well as to
determine the overall method efficiency.

Recovery criteria for the internal standards of PBDD/Fs and PCDD/Fs are given in Table 3-10.

Table 3-10. Pre-Extraction Spike Recovery Limits [%]

Pre-extraction spike

Minimum

Maximum

PCDD/Fs

%

%

13C12-2,3,7,8 TeCDF

40.0

130

13Ci2-2,3,7,8 TeCDD

40.0

130

13C12-1,2,3,7,8 PCDF

40.0

130

13C12-1,2,3,7,8 PCDD

40.0

130

13Ci2-l,2,3,6,7,8 HxCDF

40.0

130

13C12-1,2,3,6,7,8 HxCDD

40.0

130

13Ci2-l,2,3,4,6,7,8 HpCDF

25.0

130

13C12-1,2,3,4,6,7,8 HpCDD

25.0

130

13Ci2-l,2,3,4,6,7,8,9 OCDD

25.0

130

PBDD/Fs

%

%

13C12-2,3,7,8-TBDF

40.0

130

13C12-2,3,7,8-TBDD

40.0

130

13C12-2,3,4,7,8-PeBDF

40.0

130

13C12-l,2,3,7,8-PeBDD

40.0

130

13C12-1,2,3,4,7,8-HxBDF

40.0

130

13C12-1,2,3,4,7,8-HxBDD

40.0

130

13C12-1,2,3,6,7,8-HxBDD

40.0

130

13C12-l,2,3,4,6,7,8-HpBDF

25.0

130

13C12-l,2,3,4,6,7,8-HpBDD

25.0

130

13C12-OBDD

25.0

130

13c12-obdf

25.0

130

The pre-extraction internal standard recovery acceptance criteria were as recommended by U.S.
EPA Method 23 for chlorinated dioxins.5 There is no standard method guidance for PBDD/Fs
pre-extraction internal standards recovery; U.S. EPA Method 23 criteria were therefore used for
brominated targets.

5 U.S. EPA Test Method 23. Method 23 - Determination of Polychlorinated Dibenzo-p-dioxins and Polychlorinated
Dibenzofurans from Municipal Waste Combustors; Office of Solid Waste and Emergency Response, Enviromnental
Protection Agency: Washington DC, 1996.

A-191

-------
As was mentioned before, the PCDD/Fs results were considered not reportable and the pre-
extraction results are not reported as well.

The brominated pre-extraction spikes mostly passed the PCDD/Fs criteria up to the hexa
congeners but the hepta and octa congeners were frequently below the PCDD/Fs criteria
although detectable. In the original QAPP, the table for the PBDD/Fs pre-extraction spike
criteria was not the table of criteria specified in the Method 23 for PCDD/Fs pre-extraction
spikes (Table 3-10).

3.7 Polyaromatic Hydrocarbon Analysis

Combustion by-products were collected into PUF and filter and Soxhlet extracted using both
methylene chloride and toluene, yielding two separate samples for analysis. The sampling train
was also rinsed sequentially with methanol, methylene chloride, and toluene following each
experiment to collect any by-products that may not have been collected by the PUF or filter. The
methanol rinse was solvent extracted with the methylene chloride rinse (liquid-liquid extraction)
and separated, yielding two separate samples from the three rinses. Therefore, UDRI tested four
different sample media for the presence of PAHs: (1) methylene chloride from methanol and
methylene chloride rinses, (2) toluene rinse, (3) methylene chloride Soxhlet extraction of PUF
and filter, and (4) toluene Soxhlet extraction of PUF and filter. Using samples from brominated
laminate tests, the PAH content of the rinses were compared to the PAH content of the PUF/filter
extracts. Methylene chloride and toluene rinses from experiments with BFR + P - 50 (E6), BFR -
100 (E15), and BFR + PHF - 50 (E30) were analyzed (for Experiment # see Appendix B:
Experimental Conditions). Experiment BFR - 100 (El5) was used to analyze the toluene rinse
and was compared to the extract. For methylene chloride, most of the PAHs (EPA list of priority
PAHs) in the rinse were estimated to be <10% of the magnitude of the PAHs from the extract.
This excludes naphthalene and compounds lighter than fluorine where breakthrough was likely.
The naphthalene and lighter compounds were less than 1% in the rinses when compared to the
PUF/filter extracts. Even in the extract, the naphthalene signal was significantly smaller than the
other PAHs detected probably due to breakthrough through the PUF. UDRI found -90% of the
PAHs to be in the methylene chloride extracts compared to <10% in the methylene chloride
rinses. The level of PAHs detected in the toluene extract was <1% and in the toluene rinse was
<0.1%. These findings and budgetary constraints led the researchers to decide to only analyze the
methylene chloride extracts. PAHs were thus only measured for the methylene chloride
extraction samples for the remainder of the project.

A-192

-------
3.8 Organophosphorus and Chlorinated Benzene/Phenol Analysis

The chromatograms from PAH analysis were used to generate library search reports to determine
the presence of organophosphorous compounds. In addition, since no attempt was made to
analyze for chlorinated dioxins and furans due to reasons explained in Section 3.6.5, an attempt
was made to determine the presence of chlorinated benzenes and phenols known to be precursors
for the formation of halogenated dioxins and furans. The following integration events were used
when generating the library search reports: initial area reject at 1%; initial peak width of 0.02;
shoulder detection off; initial threshold of 16. The compound with the highest match quality is
reported for the compounds detected.

4 Results and Discussion

The purpose of this study as part of the U.S Environmental Protection Agency (EPA) Design for
the Environment (DfE) program was to understand the potential emissions of halogenated
dioxins or furans, and polyaromatic hydrocarbons (PAHs) from burning circuit board laminates.
This objective was achieved by using the cone calorimeter to expose circuit board laminates to
simulated combustion scenarios under ventilated fire conditions (15 L/s) at two heat fluxes (50

2	2	2

kW/m and 100 kW/m ). The 50 kW/m heat flux was chosen to mimic open burn conditions
when circuit boards are improperly burned for precious metal recovery. The higher heat flux, 100
kW/m , was chosen to mimic incineration conditions that would be used to recover/smelt away
precious metals and properly dispose of e-waste. Since the sampling train for this study
prevented the normal collection of oxygen consumption calorimetry data (Sections 3.3.1 to
3.3.3), experiments were done using the normal cone calorimeter exhaust system to collect data
for heat release (see Appendix A: Circuit Board Flammability Data), smoke yield, fire safety
information, oxygen consumption rates, CO/CO2 production rates, and effective heats of
combustion needed to attempt to correlate back to observed emission products. The emphasis of
this section of the report is on the emissions observed from the cone calorimeter (smoke,
CO/CO2) which will then be later compared to the emissions data collected from the sampling
train.

4.1 Total Mass Burned

The total mass of each type of printed circuit board laminate sample burned for the cone
calorimeter total sampling train experiments is given in Table 4-1. Total mass is important for
determining emissions factors; the amount of flammable mass burned will determine how much
total emissions are obtained.

A-193

-------
Table 4-1. Total Mass Burned Per Sample

Sample Description-Heat Flux (kW/m2)

Total Mass Burned per Sample (g)

BFR - 50

11.8

BFR - 50

13.6

BFR - 100

14.3

BFR - 100

15

BFR + P - 50

20

BFR + P - 50

20.4

BFR + PHF - 50

18.2

BFR + PHF - 50

17.3

HFR - 50

8.9

HFR - 50

8.1

HFR - 100

13.3

HFR - 100

13.3

HFR + P - 50

18.1

HFR + P - 50

19.8

HFR + PHF - 50

19.6

HFR + PHF - 50

18.6

1556 HFR - 50

9.3

1556 HFR - 50

9.7

1556 HFR + P - 50

17.9

1556 HFR + P - 50

17.8

1556 HFR + PHF-50

16.4

1556 HFR + PHF-50

15.9

NFR-50

16.5

NFR-50

15.6

NFR - 100

7.9

NFR - 100

OO
00

4.2 Smoke

Smoke data obtained using the standard cone calorimeter (without the total sampling train) for all
of the printed circuit board samples are shown in Table 4-2. Total smoke release was affected by
both component blend and flame retardant chemistry, with flame retardant chemistries always
having higher smoke release than the non-flame retardant samples. It should be noted that smoke
release in the cone calorimeter is a simple light obscuration measurement and may be composed
of many different components. While smoke is a good indication of incomplete combustion, its
presence cannot be directly correlated to emissions of concern (PM, PAH, dioxins, etc.). Instead,
smoke provides some insight into likely emissions trends from the different flame retardant
chemistries.

A-194

-------
Table 4-2. Smoke Release Data

Sample Description-Heat Flux (kW/m2)

Average smoke release.
N=3 per sample*

(m2/m2)

NFR-50

222.03

BFR - 50

479.10

HFR - 50

250.80

1556 HFR - 50

246.33

NFR - 100

214.73

BFR - 100

439.77

HFR - 100

264.83

BFR + P - 50

691.80

HFR + P - 50

438.53

1556 HFR + P - 50

397.43

BFR + PHF - 50

468.13

HFR + PHF - 50

353.43

1556 HFR + PHF-50

309.23

* Raw data listed in appendix

The smoke release information is also presented in Figure 4-1 and the following conclusions can
be made.

Brominated Flame retardant (BFR) - When compared to the other chemistries, BFR smoke
release was more than 50 to 90% greater than HFR samples. This is expected due to the flame
retardant mechanism of BFR which inhibits vapor phase combustion and in turn creates more
smoke. As heat in the flame increases due to higher heat flux, more of the smoke should burn
away and total smoke should decrease; this is observed in Figure 4-1.

Halogen-Free Flame retardant (HFR) and 1556 Halogen-Free Flame retardant (1556 HFR)

- Due to the mechanism of flame retardancy, which should be condensed phase char formation
assuming that the halogen-free flame retardants are phosphorus-based, lower smoke release is
observed compared to the BFR laminates. Unlike the BFR laminates, as heat flux is increased for
HFR, a slight increase (5.6 %) in total smoke was observed compared to NFR(-4.6%). This may
be due to the fact that the higher heat flux of burning is causing more of the PAHs in the char of
the samples to become pyrolyzed and form soot and condensed phase soot precursors. However,
this difference between NFR and HFR samples is within the percentage error of the cone
calorimeter smoke measurement device (± 10%). The difference should be considered with
caution even though the trend was reproducible with the triplicate cone calorimeter experiments
conducted.

No Flame retardant (NFR) - These materials show the lowest smoke release as expected since
they have no flame retardants present. However, the difference compared to HFR is within the
margin of error of the measurement device as described above.

A-195

-------
Halogenated and Low-Halogen Components - The addition of powdered components produced
variable smoke release results (-2.2 to 74.6 %) compared to the laminates alone. For example,
the addition of halogen containing components to BFR increased smoke by 44.2%, but when
low-halogen component powders were present, total smoke was reduced by 2.2%. The addition
of halogen containing components to halogen-free laminates provided the highest increases in
smoke release 74.6% and 61.3% for HFR and 1556 HFR laminates respectively. Halogen-free
component powders yielded a smaller increase in smoke compared to the halogen-containing
component powders, with a reduction in total smoke (2.2%) seen with BFR laminates, and only a
40.9% and 25.6% increase for HFR and 1556 HFR laminates respectively. The extra flammable
mass in both powders contributes to some smoke from burning, but the presence of halogen
increased smoke release even more.

800.00

Total Smoke Release

700.00

600.00

*£ 500.00
oJ"

l/»

S 400.00

a;
cc

a> 300.00
o

200.00

100.00
0.00

<&\ <&\ ^ 4

. .	\ % x % \

-% \ X '% \ % \ \ \ % \

° ° %

Sample Description

Ao



Figure 4-1. Smoke Release Plot
4.3 CO/CO2 Emissions

The brominated FR laminates, with or without components, show lower emissions of CO2 than
the other sample types (1.05 to 1.28 kg/kg compared to 1.3 to 1.62 kg/kg for HFR and 1.85 and
1.67 kg/kg for NFR) (Table 4-3 and Figure 4-2). Less total CO2 is observed because bromine
inhibits full combustion of carbon to CO2. However, a significant increase in CO is not always
observed with the samples tested in this study when CO2 emissions decrease. Therefore, the data
only support the idea that the brominated FR compounds reduce total CO2 emissions when

A-196

-------
2	9

combusted under open burn (50 kW/m heat flux) or incinerator (100 kW/m heat flux)
conditions. The mass balance of emissions must lie in other gases and compounds if the CO2
emissions are lower. The non-halogenated FR laminates have similar CO yields when compared
to the BFR compounds, but higher CO2 yields. This makes sense in that the flame retardants are
causing more char formation, which would lower the total amount of carbon that is combusted.
Since the non-halogenated laminates do not contain halogens that can affect combustion
chemistry, CO2 yields should be higher. The non-flame retardant samples burn with the highest
CO2 yields but have CO emissions roughly equal to or higher than the other flame retardant
systems when burned at low heat flux (50 kW/m2). This is because in the flame retardant
systems, potential carbon is present as PAHs and soot rather than being partly oxidized. Total
mass burned (total potential carbon that could convert to CO or CO2; see Table 4-1) does not seem
to correlate well to average CO and CO2 emissions, allowing combustion chemistry of the
boards, flame retardants, and components to explain to CO/CO2 emissions factors.

Table 4-3. CO/CQ2 Emission Factors

Sample Description-Heat Flux (kW/m2)

Av Post Ignition

CO Yield

C02 Yield

(kg/kg)

BFR - 50

0.15

1.05

BFR - 100

0.14

1.06

BFR + P -50

0.13

1.12

BFR + PHF - 50

0.14

1.28

HFR - 50

0.18

1.59

HFR - 100

0.11

1.44

HFR + P - 50

0.16

1.50

HFR + PHF - 50

0.12

1.52

1556 HFR - 50

0.12

1.42

1556 HFR + P - 50

0.10

1.30

1556 HFR + PHF-50

0.10

1.62

NFR-50

0.20

1.85

NFR - 100

0.07

1.67

A-197

-------
Post Ignition C0/C02 Emission Factors

Figure 4-2. C0/C02 Emission Factors Plot
4.4 Particulate Matter Emissions

The cone calorimeter data (Table 4-4 and Figure 4-3) demonstrates that most of the samples have
similar PM emissions when components are present, but can vary depending on base resins. The
halogen-free flame retardant (HFR) at 50 kW/m2 has the highest level (40% higher than BFR 50
kW) of PM emitted during burning. This relates to the condensed phase mechanism of action,
where the phosphorous flame retardant reacts with the polymer and is involved in its charring.
These charred and cross-linked polymer components will have chemical structures similar to
soot precursors, and as those molecules pyrolyze off the surface of the burning circuit board,
higher amounts of PM may be seen. The BFR compounds do show some higher PM emissions
when compared to the NFR and HFR + component blends. While smoke yields were higher for
BFR compounds compared to other sample types (Table 4-2 and Figure 4-1), PM was not always
higher for BFR. This may simply indicate that the smoke produced by burning BFR materials is
not captured by the PM filters in our experiments or that the smoke measured by the cone
calorimeter system was not a particulate but was instead organic vapors which obscured light.

Table 4-4. PM Emission Factors

Sample Description-Heat Flux (kW/m2)

PM, g/kg fuel in

BFR - 50

24.05

BFR - 100

23.11

BFR + P - 50

22.66

BFR + PHF - 50

20.85

HFR - 50

33.48

A-198

-------
Sample Description-Heat Flux (kW/m2)

PM, g/kg fuel in

HFR - 100

21.02

HFR + P - 50

18.59

HFR + PHF - 50

19.32

1556 HFR - 50

23.54

1556 HFR + P - 50

17.93

1556 HFR + PHF-50

13.42

NFR-50

17.28

NFR - 100

17.70

40.00

PM Emission Factors

^ \ \ \ % % \ \ % % \ % %
^ * A ^ \ ^ \ \ % % s° %

*o «a

-------
samples selected for PBDD/Fs analysis at EPA. Due to problems with the pre-sampling spike,
the PCDD/Fs analysis was not quantitated. In the PBDD/Fs analysis, four blanks were added to
the fourteen samples selected, yielding 18 samples. Of the 18 total samples, 12 were able to be
quantitated. The six samples that could not be quantitated were of brominated flame retardant
with halogenated components. The quantitation could not be done due to significant interference
that caused the internal standards to not be useable for quantitation. Analysis of one sample on a
LRMS in full scan resulted in insufficient sensitivity to identify the compound emissions.

PBDD/Fs compounds were quantitated in 12 samples. Six of these samples were BFR laminates
and six were combustion blanks. Five of the six blanks had significantly lower levels of
PBDD/Fs compared to the laminate samples. For the higher concentrated PBDD/Fs detected, the
difference in detection level between the combustion blanks and the BFR laminates was as large
as a factor of 100. For example, the detection of 1,2,3,4,6,7,8 - HpBDF in all but the first blank
ranged from not detected to 0.3 ng/train compared to 4 to 9 ng/train for the six BFR laminate
samples. In a system that is as complex as the calorimeter and has as many reused parts very low
levels in the actual heated calorimeter blanks are not surprising.

The chromatographic peaks for the 2,3,7,8 congeners were small compared to the non-2,3,7,8
congeners based on visual confirmation. This finding was confirmed by quantification of a single
non-2,3,7,8 congener. 2,4,6,8-TeBDF congener was a factor of four higher than the highest of
the 2,3,7,8-Br-substituted toxic congeners in the samples. Other visible brominated compounds
in the chromatograms were of similar concentrations.

The total PBDD/Fs emission from the cone calorimeter experiments shown in Table 4-5 and
Figure 4-4 indicate that brominated flame retardant (BFR) laminates have higher total PBDD/Fs
emission factors than brominated flame retardant laminates with halogen-free components. For
all six brominated samples, PBDD/Fs were released in the range of 1.89 to 4.14 ng/g (Table 4-5)
with variability that suggests there is no large difference between each sample based on only
N=2. Figure 4-4 is based on the average emission factors and suggest differences in the samples
that cannot be conclusive without larger sample sizes.

Brominated dioxins and furans were not analyzed in the NFR and HFR systems since these
systems were free of brominated FR structures (TBBPA) that could have formed PBDD/Fs
compounds.

Interestingly, the addition of components did not appear to increase PBDD/Fs emissions. This
may due to (1) a chemical interaction between the halogen-free component powder and
PBDD/Fs, (2) a dilution effect from the additional non-halogenated mass burned contributing to
the total mass lost used in the emission factor calculation, or (3) a combination of both. At this
time, it is not be possible to clearly discern given the data scatter between the replicates shown in
Table 4-5.

Based on the available data, the conclusion is that PBDD/Fs are detected in the emissions of
these brominated samples.

A-200

-------
Table 4-5. PBDD/Fs Emission Factors

Analyte

Sample Description - Heat flux (kW/m2)

BFR-

50

BFR-

50

BFR-
100

BFR-
100

BFR +
PHF-50

BFR +
PHF-50

ND=0,EMPC=EMPC

ng/g

2,3,7,8 - TBDD

0.00E+00

0.00E+00

0.00E+00

0.00E+00

0.00E+00

0.00E+00

1,2,3,7,8 -PeBDD

3.72E-01

1.79E-01

1.85E-01

3.25E-01

1.20E-01

1.42E-01

1,2,3,4,7,8 + 1,2,3,6,7,8 - HxBDD

1.38E-01

9.57E-02

1.25E-01

1.49E-01

8.79E-02

6.94E-02

1,2,3,7,8,9 -HxBDD

6.97E-02

4.68E-02

5.45E-02

7.65E-02

4.49E-02

3.16E-02

1,2,3,4,6,7,8 -HpBDD

8.76E-02

7.73E-02

1.42E-01

1.18E-01

7.36E-02

7.18E-02

1,2,3,4,6,7,8,9 - OBDD

0.00E+00

0.00E+00

0.00E+00

0.00E+00

0.00E+00

0.00E+00















2,3,7,8 - TBDF

0.00E+00

0.00E+00

0.00E+00

0.00E+00

0.00E+00

0.00E+00

1,2,3,7,8 -PeBDF

5.81E-01

0.00E+00

1.59E-01

2.24E-01

2.42E-01

2.79E-01

2,3,4,7,8 -PeBDF

8.90E-01

5.14E-01

2.47E-01

4.06E-01

3.60E-01

6.11E-01

1,2,3,4,7,8 -HxBDF

1.32E+00

6.60E-01

2.29E-01

9.04E-01

4.86E-01

5.72E-01

1,2,3,4,6,7,8 -HpBDF

5.68E-01

3.45E-01

4.21E-01

6.25E-01

2.48E-01

3.11E-01

1,2,3,4,6,7,8,9 - OBDF

7.35E-02

5.57E-02

-

0.00E+00

0.00E+00

0.00E+00















Total PBDD/Fs
(ND=0; EMPC= 0)

3.21E+00

1.97E+00

1.56E+00

2.83E+00

1.66E+00

2.06E+00

Total PBDD/Fs
(ND=0; EMPC= EMPC)

4.10E+00

1.97E+00

1.56E+00

2.83E+00

1.66E+00

2.09E+00

Total PBDD/Fs
(ND=DL; EMPC= EMPC)

4.14E+00

2.05E+00

1.89E+00

3.07E+00

2.09E+00

2.63E+00

The laminate samples with halogenated components (BFR-P) could not be quantitated due to significant halogenated interference.

"EMPC" indicates that the bromine isotope ratio for the two molecular ion clusters was not within ±15% of the correct isotope ratio. When the
two molecular ions are not within the correct isotope ratio, the two molecular ions are quantitated separately and the smaller quantitation is
denoted EMPC. The EMPC notation identifies that the presence of an additional molecule may be influencing the detection level of the
compounds of interest.

A-201

-------
PBDD/Fs Emission Factors

1,2,3,4,6,7,8,9-OBDF
1,2,3,4,6,7,8- HpBDF
1,2,3,4,7,8- HxBDF
I 2,3,4,7,8-PeBDF
1,2,3,7,8- PeBDF
I 2,3,7,8-TBDF
1,2,3,4,6,7,8,9-OBDD
I 1,2,3,4,6,7,8- HpBDD
I 1,2,3,7,8,9- HxBDD
1,2,3,4,7,8 + 1,2,3,6,7,8 - HxBDD
I 1,2,3,7,8- PeBDD
I 2,3,7,8-TBDD

Sample Description

Figure 4-4. PBDD/Fs Emission Factors Plot for ND=0 and EMPC=EMPC

The laminate samples with halogenated components (BFR + P) could not be quantitated due to significant
interference.

4.6 PAH Emissions

Table 4-6, Table 4-7 and Figure 4-5 show the total PAH emission factors for the 16 EPA priority
PAHs quantified for the different printed circuit board laminates tested using the cone
calorimeter. Brominated flame retardant (BFR) laminates burned at 50 kW/m2 heat flux had the
highest total PAH emissions and no flame retardant (NFR) laminates burned at 50 kW/m heat
flux had the least. At a higher heat flux (100 kW/m2), the NFR sample showed 29% higher PAH
emissions than the halogen-free (HFR) sample at the same heat flux. Emissions for the BFR were
similar at both heat flux levels.

The observed trends of PAH emissions make sense in light of both the known and assumed
flame retardant mechanisms for the two types of flame retardant systems. Since the BFR is a
vapor phase flame retardant, any combustion of that flame retardant with decomposing epoxy
structures should generate more incomplete combustion products. In the case of the HFR system,
it is assumed a phosphorus-based flame retardant is present, which has more of a condensed
phase (char formation) mechanism and binds up most of the possible PAH structures on the
burned sample residue rather than created in the flame front as seen with BFRs. The results
presented in Figure 4-5 support this general trend with a wide range of PAH products detected.
The presence of component powders affected PAH emissions for both BFR and HFR systems.
PAH emissions were reduced for the 1556 HFR samples that had components compared to the

A-202

-------
other HFR samples. In some cases, a slight increase in PAH emissions was noted for the other
HFR laminates when components were present. For the BFR systems, the presence of
components slightly lowered total PAH emissions.

Since PAHs are known to be the nascent precursors of soot, a higher presence of PAHs should
lead to higher PM yields from combustion. In this study, the PM yields (Table 4-4 and Figure
4-3) and the PAH emissions (Table 4-6 and Figure 4-5) did not always have this positive
correlation. Typically, naphthalene yields should have been higher than the other PAHs detected.
Analysis of our methods to determine breakthrough of PAHs during sampling at these high
velocities has shown that fluorene and heavier compounds are captured using 4 PUFs in the glass
cartridge that holds the PUFs and that acenaphthylene breakthrough was almost 50%. However,
since the carcinogenic PAHs are of interest and the extraction of eight PUFs is complex, no
attempt was made to prevent breakthrough of compounds lighter than fluorene by increasing the
number of PUFs. Figure 4-6 displays the PAH emissions data excluding compounds with a lower
molecular weight than fluorene likely to have had breakthrough. The same emission trends were
observed when naphthalene, acenapthylene, and acenapthene were excluded, suggesting that no
crucial information was lost by not sampling compounds requiring eight sampling PUFs.

PAH Emission Factors

6.0E+00

5.0E+00

4.0E+00

5

DO



3.0E+00

g 2.0E+00

1.0E+00

0.0E+00

-------
2.5E+00

2.0E+00

5

6J>

t/T
X

2

1.5E+00

l.OE+OO

5.0E-01

O.OE+OO

PAH Emission Factors (Flourene +)

Benzo[ghi]perylene
Dibenz[ah]anthracene
lndeno[123cd]pyrene
Benzo[a]pyrene

-	¦ Benzo[b+k]fluoranthene

-	¦ Chrysene

-	¦ Benz[a]anthracene
" ¦ Pyrene

~ ¦ Fluoranthene

¦	Anthracene

_ ¦ Phenanthrene

¦	Fluorene

% % % % % \ \ \
XO	X> /O. O/y V/") /O /O.	/v. /<~. MO V>0

^^ V %> % \> % ^ % >

^	^ % */> >So

-vC"

so

V





Sample Description

Figure 4-6. PAH Emission Factors for Fluorene and Higher Molecular Weight PAHs Detected from the EPA
List of 16* Priority PAHs

+Bcn/o|b |fluoranthene andbenzo[k]fluoranthene are reported together

When looking solely at the release of known carcinogenic PAHs (Figure 4-7), trends similar to
those in Figure 4-5 and Figure 4-6 are observed. BFR systems produce more of the carcinogenic
PAHs than the HFR or NFR systems. The addition of components does not appear to drastically
affect the yields of carcinogenic PAHs. The presence of components decreases the yields in
some cases probably due to a dilution effect from the added mass when calculating emission
factors. The high heat flux can cause the NFR system to give off just as much carcinogenic
PAHs as a flame retardant + component system from a lower heat flux. When looking at only the
toxic equivalent emission factors of carcinogenic PAH values (Figure 4-8), it is again observed
that BFR has the highest value followed by the HFR systems and then the NFR system.

A-204

-------


8.0E-01



7.0E-01

2

6.0E-01





BJ)



i/T

5.0E-01

X



<



0.



u

4.0E-01

'E



OJ



SP



0
c

3.0E-01

'o







fC

u

2.0E-01



1.0E-01



0.0E+00

Carcinogenic PAH Emission Factors

Benzo[ghi]perylene
Dibenz[ah]anthracene
lndeno[123cd]pyrene
Benzo[a]pyrene
Benzo[b+k]fluoranthene
I Chrysene
Benz[a]anthracene

° x x°* °° % \xx%°* °° %

so ^ % % X
so so ^

so ^
so

X.

Sample Description



Figure 4-7. Emission Factors of Carcinogenic PAHs from the EPA List of 16 Priority PAHs

*Benzo[b]fluoranthene andbenzo[k]fluoranthene are reported together

A-205

-------
Toxic Equivalent Emission Factors of Carcinogenic PAHs

5

M

£

OJ

o5

1.6E-01

1.4E-01

1.2E-01

1.0E-01

8.0E-02

6.0E-02

re

4.0E-02

2.0E-02

0.0E+00

Sample Description

lndeno[l,2,3-cd]pyrene
Dibenzo[a,h]anthracene
I Chrysene

I Benzo[g,h,i]perYlene
I Benzo[b+k]fluoranthene
Benzo[a]anthracene
I Benzo[a]pyrene

Figure 4-8. Toxic Equivalent Emission Factors of Carcinogenic PAHs from EPA List of 16 Priority PAHs
Compared at 50 kW/m2 Conditions

*Benzo[b]fluoranthene andbenzo[k]fluoranthene are reported together

Table 4-6. PAH Emission Factors from EPA List of 16 Priority PAHs for BFR and NFR at 50 and 100

kW/m2

Analyte

Sample Description - Heat flux (kW/m2)

BFR - 50

BFR - 100

BFR + P -

50

BFR +
PHF - 50

NFR-
50*

NFR - 100

Emission Factors, g/kg

Naphthalene

4.3E-01

2.1E-02

3.1E-02

2.5E-02

4.1E-03

7.7E-03

Acenaphthylene

2.6E+00

2.9E+00

1.8E+00

1.7E+00

1.9E-01

2.9E-01

Acenaphthene

1.1E-02

5.4E-03

6.3E-03

5.4E-03

0.0E+00

0.0E+00

Fluorene

2.3E-01

2.1E-01

1.8E-01

1.8E-01

7.2E-02

2.7E-01

Phenantlirene

8.0E-01

9.6E-01

8.3E-01

8.2E-01

1.2E-01

6.1E-01

Anthracene

8.7E-02

1.0E-01

9.3E-02

9.4E-02

4.9E-02

2.2E-01

Fluoranthene

1.6E-01

1.5E-01

1.7E-01

1.5E-01

2.7E-02

1.1E-01

Pyrene

1.4E-01

1.7E-01

1.5E-01

1.1E-01

4.3E-02

1.7E-01

B enz [o] antliracene

1.3E-01

7.3E-02

1.2E-01

9.9E-02

1.6E-02

3.4E-02

Chrysene

2.6E-01

2.2E-01

2.7E-01

2.5E-01

3.3E-02

8.2E-02

Benzo [b +A]fluoranthene

1.2E-01

9.4E-02

1.1E-01

1.1E-01

3.3E-02

5.5E-02

Benzo[o] pyrene

1.0E-01

9.2E-02

8.6E-02

7.2E-02

2.3E-02

4.0E-02

Indeno | /,2,j-«/|pyrene

5.6E-02

4.5E-02

5.3E-02

4.0E-02

1.4E-02

3.6E-02

A-206

-------
Analyte

Sample Description - Heat flux (kW/m2)

BFR - 50

BFR - 100

BFR + P -

50

BFR +
PHF - 50

NFR-
50*

NFR - 100

Emission Factors, g/kg

Dibenz[o, /?] anthracene

2.6E-02

2.7E-02

2.5E-02

2.1E-02

0.0E+00

0.0E+00

Benzo [g, h, ;']perylene

4.8E-02

3.7E-02

4.7E-02

2.7E-02

8.2E-03

2.7E-02

Total 16 EPA PAHs

5.22E+00

5.08E+00

3.93E+00

3.69E+00

6.24E-01

1.95E+00

Benzo [b]fluoranthene andbenzo[k]fluoranthene are reported together
*From a single run

Table 4-7. PAH Emission Factors from EPA List of 16* Priority PAHs for HFR and 1556 HFR at 50 and 100

kW/m2



Sample Description - Heat flux (kW/m2)



HFR -

HFR-

HFR + P

HFR +

1556

1556

1556 HFR



50*

100

-50

PHF - 50

HFR - 50

HFR + P
-50

+ PHF-

50

Analyte





Emission Factors

> g/kg





Naphthalene

7.9E-03

8.4E-03

6.7E-03

6.7E-03

1.9E-02

6.3E-03

1.6E-02

Acenaphthylene

5.1E-01

5.5E-01

7.7E-01

6.2E-01

9.6E-01

7.4E-01

7.1E-01

Acenaphthene

7.9E-03

3.6E-03

1.8E-03

6.7E-03

6.7E-03

0.0E+00

6.9E-03

Fluorene

1.9E-01

1.6E-01

1.7E-01

2.4E-01

4.8E-01

1.9E-01

3.1E-01

Phenantlirene

4.5E-01

3.6E-01

4.7E-01

3.4E-01

6.0E-01

5.4E-01

4.2E-01

Antliracene

1.1E-01

9.3E-02

9.8E-02

8.6E-02

1.3E-01

9.7E-02

8.7E-02

Fluoranthene

8.7E-02

7.5E-02

1.0E-01

1.0E-01

1.4E-01

1.3E-01

1.3E-01

Pyrene

1.2E-01

1.1E-01

1.0E-01

7.8E-02

1.2E-01

8.3E-02

7.9E-02

B enz [o] antliracene

3.6E-02

1.9E-02

4.0E-02

3.6E-02

5.0E-02

5.5E-02

4.2E-02

Clirysene

7.9E-02

4.1E-02

1.3E-01

9.6E-02

2.0E-01

1.9E-01

1.4E-01

Benzo [b +A]fluoranthene

7.9E-02

3.1E-02

6.7E-02

6.4E-02

1.1E-01

8.6E-02

8.1E-02

Benzo [o] pyrene

4.0E-02

2.0E-02

4.2E-02

3.4E-02

4.5E-02

5.3E-02

4.1E-02

Indeno | /,2,j-«/|pyrene

2.4E-02

2.6E-02

2.4E-02

1.8E-02

3.7E-02

3.1E-02

2.4E-02

Dibenz[o, /?] antliracene

0.0E+00

0.0E+00

1.2E-02

1.0E-02

2.2E-02

1.8E-02

1.3E-02

Benzo [g, h, ;']perylene

2.4E-02

1.6E-02

1.5E-02

1.2E-02

1.9E-02

1.6E-02

1.2E-02

Total 16 EPA PAHs

1.74E+00

1.51E+00

2.04E+00

1.75E+00

2.93E+00

2.24E+00

2.11E+00

+Ben/o | b | flnorantlienc andbenzo[k]fluoranthene are reported together
*From a single run

Table 4-8. Toxic Equivalent Emission Factors of Carcinogenic PAHs from EPA List of 16* Priority PAHs

Carcinogenic -PAHs

Toxic

Equivalency
Factor (TEF)

Toxic Equivalent Emission Factors of
Carcinogenic PAHs (g/kg)

BFR

HFR

1556 HFR

NFR

Benzo [o] pyrene

1

1.0E-01

4.0E-02

4.5E-02

2.3E-02

B enzo \a\ antliracene

0.1

1.3E-02

4.0E-03

4.5E-03

1.6E-03

Benzo [b +A]fluoranthene

0.1

1.2E-02

7.9E-03

1.1E-02

3.3E-03

Benzo [g, h, ;']perylene

0.01

4.8E-04

2.4E-04

1.9E-04

8.2E-05

Clirysene

0.01

2.6E-03

7.9E-04

2.0E-03

3.3E-04

Dibenzo [a, h] antliracene

0.1

2.6E-03

0.0E+00

2.2E-03

0.0E+00

Indeno [ 1,2,3-cd] pyrene

0.1

5.6E-03

2.4E-03

3.7E-03

1.4E-03

*Benzo[b]fluoranthene andbenzo[k]fluoranthene are reported together

A-207

-------
Although attempts were also made to determine presence of other chlorinated benzenes/phenols
known to be PCDD/Fs precursors, none were detected at the sample concentrations analyzed for
PAHs. No significant presence of chlorobenzenes and phenols detected in the laminate burns is a
likely indicator of a negligible presence of chlorinated dioxins under the conditions explored in
this study. However, the absence of PCDD/Fs cannot be conclusively stated without further
analysis of more concentrated samples or attempts to analyze extracts for PCDD/Fs disregarding
the previously discussed issues related to the absence of the chlorinated pre-sampling surrogates.

Scanning for organophosphorus was also done because it was believed that the non-halogenated
flame retardants present in the samples were phosphorus-based. The detection of
organophosphorus emissions would indicate the presence of a vapor phase flame retardant while
the detection of no organophosphorus emissions would indicate the presence of a condensed
phase flame retardant. The organophosphorous compounds detected in this study are given in
Table 4-9. As Table 4-9 shows, different compounds were detected from the repeat burn of the
same laminate. The environmental and health effects of the compounds detected are not
evaluated in this report to explain their impact. From a flame retardant perspective, some of the
compounds fit with known flame retardant chemistry while others are likely post-combustion
reaction products or reactions between the phosphorus flame retardant and parts of the circuit
board. For example, the phosphorous compounds with silicon in their chemical structure are
likely present due to reactions between organophosphorus and e-glass during burning. The
presence of any halogen-phosphorus compounds is likely due to reaction between halogen and
organophosphorus during burning. Other organophosphorus compounds present that contain
phosphonic or phosphinic acids are decomposition products of known phosphorus flame
retardants, especially compounds containing phenyl groups. However, it should be recognized
that the exact phosphorus flame retardant used in these systems was not reported to UDRI,
leaving the interpretation of the data based upon information in open literature for phosphorus
flame retardants. Combustion chemistry is complex, especially when many components are
present, and the list of compounds detected is not surprising.

A-208

-------
Table 4-9. Organophosphorous Compounds Detected

Laminate
Description

Organophosphorous Compounds Detected

Area

%

BFR -50

1 -Ethyl-1 -hydridotetrachlorocyclotriphosphazene

0.04

BFR -50

Silanol, trimethyl-, pyrophosphate

0.51

BFR + P -50

Phosphonic acid, methylenebis-, tetrakis(trimethylsilyl) ester

0.17

0,0'-(2,2'-Biphenylylene)thiophosphoric acid

0.38

BFR + P -50

Bis(4-methoxyphenyl)phosphinic acid

0.1

BFR + PHF-50

Silanol, trimethyl-, pyrophosphate^: 1)

0.08

1 -Phosphacyclopent-2-ene, 1 -methyl -5-methylene-2,3 -diphenyl-

0.61

4-Phosphaspiro[2.41hept-5-ene, 4-methyl-5,6-diphenyl-

0.15

Bis(4-methoxyphenyl)phosphinic acid

0.15

BFR + PHF-50

1 -Phosphacyclopent-2-ene, 1 -methyl -5-methylene-2,3 -diphenyl-

0.23

BFR -100

Ethylphosphonic acid, bis(tert-butyldimethylsilyl) ester

8.33

BFR -100

Methylenebis(phosphonic acid), tetrakis(3-hexenyl) ester

0.29

HFR +P-50

Phosphonic acid, phenyl-, diethyl ester

0.25

HFR + PHF-50

(2-Bromo-3 -methylphenyl) diphenylphosphine

0.34

HFR + PHF-50

Phosphine imide, P,P,P-triphenyl-

0.3

1556 +P-50

Phosphorane, 1 lH-benzo[a]fluoren-l-ylidenetriphenyl-

0.43

1 -Phosphacyclopent-2-ene, 1 -methyl -5-methylene-2,3 -diphenyl-

0.53

1556 +PHF-50

Phosphine imide, P,P,P-triphenyl-

0.21

4.7 Heat Release (Flammability) Results

The flammability data for the laminate samples and laminates + component powders are shown
in Appendix A. Since material flammability/fire safety was not the primary focus of this study, it
is not a primary focus of the Results and Discussion section. Instead, suggestions are provided on
how the heat release results should and should not be interpreted and used.

The circuit board samples in this report are likely formulated to pass a small flame test, such as
UL-94 V-0/-1/-2 (ASTM D3801), or a glow wire test (ASTM D6194) that mimics a short circuit
ignition scenario. The cone calorimeter used in this report represents a well-ventilated fire
scenario when it is run at a flow of 24 L/s as per the ASTM E1354 method. It better represents a
larger fire source and not the small ignition source typically seen in electronic circuit boards. In
this report, the cone calorimeter experiments were run at a lower flow rate of 15 L/s, which
would roughly simulate open burn type conditions, not an intense well ventilated fire. Further,
where ASTM D3801 uses a small flame source, the cone calorimeter uses a radiant heater, which
in this case was set to heat fluxes of 50 and 100 kW/m and represent a medium sized and a very
large scale fire, respectively. The measurement of heat release from materials that were not
designed to protect against robust heat sources like that of the cone calorimeter is a limitation of
this study. It should not be used to infer the fire safety of the products in their respective
scenarios. Each fire test used for regulating flame retardant materials is tailored for a specific fire
risk scenario; the standards are not interchangeable. Therefore, the cone calorimeter data in this

A-209

-------
study is best used to understand how much heat an object gives off when burned in a situation
where it is well ventilated and a robust heat source is present. With this in mind, heat release rate
and smoke data from the cone calorimeter testing of circuit boards can be used to better
understand:

•	Heat output from the burning material when properly disposed of (100 kW/m heat flux
conditions) to know if the laminate gives off enough heat to run the incinerator cleanly.

•	Heat output if e-waste was to be used for waste-to-energy processes (how much energy
would be generated by the burning of e-waste).

•	Relative rankings on flame retardant performance outside the regulatory test scenario for
which it was designed. Specifically, cone calorimeter measures can inform how the materials
would contribute to a larger fire event (server room fire, house fire) when set afire by another
object in the same room. The lower the heat release of the material, the less likely it will
contribute negatively to a large fire event, or, spread fire should it be exposed to heat and
flame.

While the cone calorimeter data can be useful, care should be taken when using it for the
selection of fire safe materials, or in the case of this report, figuring out which flame retardant
chemistry (brominated or non-halogenated) is appropriate for a particular need. Cone calorimeter
data can guide selections, but each material scientist and engineer will need to look closely at the
fire standards to decide what aspect of fire performance certain materials must meet.

Although cone calorimeter measurements can give insight into heat output and comparative
flame retardant performance, there are conclusions that cannot be made with the
flammability/heat release data in this report:

•	The measured heat release of each of the system does not infer that any one material is safer
than another from a fire safety perspective. Since the cone calorimeter measures flammability
in a different way than other regulatory tests, a low heat release in the cone calorimeter does
not ensure a "pass" result in a regulatory test. A lower peak HRR would mean that the
burning laminate would be less likely to ignite other nearby objects though. A lower total
HR would indicate that if the burning laminate was fully burned, it would contribute less
total heat (fuel) to the overall fire.

•	Smoke release in the cone calorimeter is very much a function of the combustion conditions
used in the test. Smoke release may be more intense or less intense under different ventilation
conditions and the results cannot be used to infer that a particular material will be better or
worse than another in a different flaming combustion configuration/scenario. Smoke release
in the cone calorimeter is very different than smoke release from a full high heat flux fire and
is also very different than smoke release from a small flame ignition source.

•	Cone calorimeter data has a known % error of ±10%.

With the above caveats in mind, the following trends are observed in Table 4-10 and Table 4-11:

•	At a heat flux of 50 kW/m , the flame retardant systems show lower peak heat release when
compared to the non-flame retardant systems. The non-halogenated "1556 HFR" sample
shows the lowest flammability overall but also has a lower amount of total mass lost,
suggesting that it either has more non-combustible mass present or is a more robust char
forming flame retardant system.

A-210

-------
•	The addition of component powders generally increased total heat release and had mixed
effects on peak HRR.

•	At a heat flux of 100 kW/m , only the brominated flame retardant continues to lower heat
release (peak HRR and total HR) versus the non-flame retardant control. The non-
halogenated system gives heat release roughly equal to, or slightly higher, than the non-flame
retardant system.

Table 4-10. Heat Release Summary for Laminates and Laminates + Component Powders Tested at 50 kW/m2

Sample
Description -
Heat Flux (50
kW/m2)

Sample

Thickness

(mm)

Time to
ignition

(s)

Peak
HRR

(kW/m2)

Average
HRR

(kW/m2)

Weight
% Lost

(%)

Total
Heat
Release
(MJ/m2)

Total
smoke
Release
(m2/m2)

MARHE

(kW/m2)

BFR-1

0.49

11

279.0

65.31

37.2

4.4

485.2

115.6

BFR-2

0.49

10

272.4

64.23

39.8

4.8

496.9

114.2

BFR-3

0.50

10

296.5

91.31

37.5

4.8

455.2

146.8

BFR + P-1

0.49

9

280.2

81.29

29.3

6.9

719.9

127.7

BFR + P -2

0.48

8

265.0

79.41

28.8

6.9

698.5

116.3

BFR + P -3

0.49

14

255.7

79.94

27.9

6.6

657.0

105.9

BFR + PHF-1

0.48

12

279.3

83.44

25.2

6.8

467.1

111.7

BFR + PHF -2

0.48

18

331.4

88.70

25.1

6.9

446.5

107.5

BFR + PHF -3

0.48

14

266.8

81.37

24.9

6.9

490.8

108.4

NFR-1

0.43

11

406.1

77.77

32.3

5.8

228.3

130.0

NFR-2

0.41

11

391.6

87.52

28.4

6.1

199.0

139.4

NFR-3

0.44

12

445.9

88.69

34.9

6.5

238.8

140.8

HFR-1

0.57

12

406.7

98.15

35.8

7.8

240.2

141.4

HFR-2

0.56

15

292.1

84.51

32.3

6.7

237.5

106.9

HFR-3

0.58

17

368.5

94.59

34.2

7.3

274.7

124.7

HFR + P-1

0.56

10

267.4

88.64

25.0

8.2

451.2

116.1

HFR + P -2

0.58

8

278.9

102.55

25.9

9.6

461.4

139.8

HFR + P- 3

0.58

14

303.5

102.61

25.6

9.2

403.0

128.4

HFR+ PHF -1

0.58

21

343.0

111.98

25.1

9.8

330.9

128.4

HFR + PHF -2

0.57

31

294.0

96.43

21.5

7.8

372.5

92.4

HFR + PHF -3

0.56

26

271.1

86.55

22.5

8.0

356.9

98.5

1556 HFR -1

0.46

14

181.2

55.56

27.2

4.2

270.5

76.0

1556 HFR -2

0.45

24

205.9

50.88

23.0

3.6

232.1

60.7

1556 HFR -3

0.46

16

230.9

63.06

25.3

4.6

236.4

84.1

1556 HFR + P-1

0.46

12

165.7

73.22

23.3

6.6

400.4

93.1

1556 HFR+ P-2

0.46

9

185.9

68.54

20.9

6.1

382.6

92.3

1556 HFR+ P-3

0.45

9

165.8

71.18

22.8

6.6

409.3

92.2

1556 HFR+PHF-1

0.45

18

196.7

76.26

20.0

6.4

293.6

88.3

1556 HFR+ PHF-2

0.46

22

209.4

83.15

20.4

7.1

324.0

88.6

1556 HFR +PHF -3

0.46

22

220.6

81.50

20.5

6.5

310.1

84.4

A-211

-------
Table 4-11. Heat Release Summary for Laminates and Laminates + Component Powders Tested at 100
kW/m2

Sample
Description -
Heat Flux
(100 kW/m2)

Sample

Thickness

(mm)

Time to
ignition

(s)

Peak
I IKK

(kW/m2)

Average
I IKK

(kW/m2)

Weight
% Lost

(%)

Total
Heat
Release
(MJ/m2)

Total
smoke
Release
(m2/m2)

MARHE

(kW/m2)

BFR -1

0.41

3

226.7

55.5

41.1

4.5

475.6

128.5

BFR-2

0.42

5

390.6

80.4

45.8

5.7

451.0

180.2

BFR-3

0.40

3

356.8

77.0

45.3

5.4

392.7

189.4

NFR -1

0.32

3

356.4

79.7

36.5

5.3

194.6

188.4

NFR-2

0.35

4

490.5

94.5

38.9

6.6

230.1

201.3

NFR-3

0.34

4

387.5

70.8

37.5

5.0

219.5

152.5

HFR -1

0.49

6

494.7

104.0

38.6

7.4

231.4

205.4

HFR-2

0.48

6

495.2

104.9

35.8

7.5

237.5

215.9

HFR-3

0.49

5

367.1

120.0

40.5

10.2

325.6

200.5

5 Conclusions

While the cone calorimeter is a useful instrument for measuring flammability from a fire safety
perspective, the use of the cone calorimeter in this study was as a combustion science tool. Heat
fluxes plus a lower flow rate were chosen to represent potential open burn (50 kW/m2) and
incineration for metal recovery (100 kW/m ). The following general trends were observed:

50 kW/m2 heat flux:

•	BFR: PBDD/Fs emitted. PAHs emitted at higher levels compared to other samples.

•	HFR: PAHs emitted at higher levels than NFR sample.

•	NFR: PAHs emitted at lowest levels compared to other samples.

100 kW/m2 heat flux:

•	BFR: PBDD/Fs emitted. PAHs emitted at higher levels compared to other samples.

•	HFR: PAHs emitted at lowest levels compared to other samples.

•	NFR: PAHs emitted at a level slightly lower than the BFR sample.

Effect of components on emissions:

•	PBDD/Fs: PBDD/Fs were similar or lower than sample without components.

•	PAHs: In general, presence of components reduced PAH emissions for BFR, were similar or
slightly highly for HFR and were lower for 1556 HFR. The size of these differences varied
depending on which PAHs were summarized (see section 4.6).

•	PAH emissions and smoke release of laminates with low halogen components were slightly
lower than standard components across all three difference laminates.

Smoke, PM, CO and CO2 release:

•	Smoke release was higher for BFR than HFR laminates. Smoke release was higher with
components due to greater amount of material. PM generally had small differences between
samples. There were negligible differences in CO release between samples. CO2 release was

A-212

-------
lowest for BFR but with small differences between samples. Results are complex and
smoke/PM results do not always correlate.

The results of this report do not suggest that any one material is safer than another in regards to
fire safety. The results do show that the flame retardants lower heat release under flaming
combustion even at high heat fluxes.

Overall, the results clearly show that all of the samples generated combustion by-products other
than CO2 and water. The flame retardant samples in some cases generated more pollutants than
the NFR samples, as one would expect since the flame retardants are inhibiting combustion. Any
system that slows down flaming combustion will generate higher levels of smoke, CO, PM, and
other incomplete combustion products. A flame retardant with a vapor phase mechanism (such as
BFR) will generate more species than a flame retardant that uses a condensed phase mechanism
(assumed to be the case of the phosphorus-based HFR system). It is important to look at flame
retardant chemistry, flame retardant mechanism, polymer decomposition chemistry, and fire
scenario (heat, ventilation) to determine what sorts of species may be formed during accidental
fires (where flame retardants serve as passive protection) or intentional ones (proper and
improper incineration).

The other major finding of this report is that the cone calorimeter was able to obtain a diverse
amount of information about emissions from circuit boards. For the brominated laminate with
halogenated components, the complexity of the emissions made them difficult to separate and
identify but the results show that pollutants exist. Further work and separation science would be
needed to achieve that higher level of data resolution with these particular samples.

Based upon the results in this report, users of flame retardants for circuit boards should realize
that if PCBs or other e-waste is to be incinerated for precious metal recovery, it should be done
properly with good incinerator control to address the pollutant emissions that will occur. Even
non-flame retardant boards when incinerated improperly will release pollutants of concern, as
was seen from the data in this report. Emissions may have been lower, but they were still present.
The use of flame retardants is a technology compromise: it provides fire safety performance
(thus lowering risk of short circuit ignitions in daily use) but will generate higher pollutants
when incinerated improperly. Other environmental concerns may drive the selection of different
flame retardant chemistry, but from emissions alone, such a decision cannot be made. With
careful attention to polymer thermal decomposition chemistry and combustion science, it may be
possible to generate a flame retardant in the future which provides fire protection and minimizes
emissions/pollutants of concern during burning. If there is a desire to develop clean burning
flame retardant materials, entirely different flame retardant chemistries must be developed.
Otherwise, the safest solution to this problem is to recover precious metals via well controlled
incineration with regulatory emissions controls in place as well as cost-effective methods of e-
waste collection and disposal.

6 Acknowledgments

The authors wish to thank Kathleen Beljan, Mary Galaska, and Kathy Schenck of UDRI for their
assistance with the cone calorimeter tests and Anne Chauvian and Saikumar Chalivendra for
their initial support for the modified experimental design work. Barbara Wyrzvkowska-Ceradini

A-213

-------
and Craig Williams assisted with sample extraction, clean-up and analysis at EPA labs. Funding
and materials for the project were provided by Albemarle, Boliden, BSEF, Chemtura, Clariant,
Ciba Specialty Chemicals, Dell, Fujitsu-Siemens, Hewlett-Packard, IBM, ICL-IP America Inc.,
Intel, Isola, ITEQ, Matsushita Electric Industrial and Matsushita Electric Works, Nabeltec,
Panasonic, Seagate, Sony, Supresta, & U.S. EPA.

A-214

-------
7 Appendix A: Circuit Board Flammability Data

Along with emissions data, heat release information as per ASTM E1354 was also collected.
This data is reported in below as a function of heat flux and samples tested. Observed fire
behavior, final chars, and heat release rate curves are given. The data is presented for the
purposes of completeness in this report. It does not infer any particular level of fire safety about
the samples tested. Merely it shows what the measured heat release information was from these
samples when tested at 15 L/sec exhaust flow in triplicate as per the ASTM methodology.

In the section below, BFR indicates a brominated flame retardant system being tested, while HF
indicates halogen-free flame retardant and NFR indicates that the sample had no flame retardant
present. Component blends are identified as "Comp", meaning a component blend where
halogen was present in the component blend powder, and as "HF Comp" meaning the mostly
halogen-free component blend was used.

A-215

-------
Heat Release Rate-50 kW/m2

Table 7-1. Heat Release Rate Data (50 kW/m2)

Sample
Description -
Heat Flux
(50 kW/m2)

Sample

rhickness

(mm)

rime to
ignition

(s)

Peak
HRR

(kW/m2)

Time to

Peak

HRR

(s)

Average
HRR

(kW/m2)

Starting

Mass

(g)

Total

Mass
Loss

GO

Weight %
Lost

(%)

Total
Heat
Release
(MJ/m2)

Total
smoke
Release
(m2/m2)

Avg. Effective
Heat of Comb.
(MJ/kg)

MARHE

(kW/m2)

FIGRA

BFR-1

0.5

11

279

20

65

10.5

3.9

37.2

4.4

485

15.14

116

13.95

BFR-2

0.5

10

272

20

64

10.8

4.3

39.8

4.8

497

11.21

114

13.62

BFR -3

0.5

10

296

25

91

10.4

3.9

37.5

4.8

455

17.58

147

11.86

BFR + P -1

0.5

9

280

30

81

20.5

6.0

29.3

6.9

720

11.92

128

9.34

BFR + P -2

0.5

8

265

35

79

20.5

5.9

28.8

6.9

699

11.71

116

7.57

BFR + P -3

0.5

14

256

34

80

20.4

5.7

27.9

6.6

657

11.50

106

7.52

BFR + PHF -1

0.5

12

279

33

83

20.3

5.1

25.2

6.8

467

13.09

112

8.46

BFR + PHF -2

0.5

18

331

37

89

20.3

5.1

25.1

6.9

447

13.39

108

8.96

BFR + PHF -3

0.5

14

267

32

81

20.5

5.1

24.9

6.9

491

13.14

108

8.34

NFR-1

0.4

11

406

28

78

9.3

3.0

32.3

5.8

228

18.66

130

14.50

NFR -2

0.4

11

392

26

88

9.1

2.6

28.4

6.1

199

22.87

139

15.06

NFR -3

0.4

12

446

29

89

9.5

3.3

34.9

6.5

239

19.36

141

15.37

HFR-1

0.6

12

407

31

98

11.4

4.1

35.8

7.8

240

19.00

141

13.12

HFR -2

0.6

15

292

39

85

11.5

3.7

32.3

6.7

238

17.75

107

7.49

HFR -3

0.6

17

368

36

95

11.4

3.9

34.2

7.3

275

18.44

125

10.24

HFR + P -1

0.6

10

267

45

89

21.2

5.3

25.0

8.2

451

15.36

116

5.94

HFR + P -2

0.6

8

279

39

103

21.6

5.6

25.9

9.6

461

17.01

140

7.15

HFR + P- 3

0.6

14

304

41

103

21.5

5.5

25.6

9.2

403

16.50

128

7.40

HFR+PHF-1

0.6

21

343

49

112

21.5

5.4

25.1

9.8

331

17.90

128

7.00

HFR + PHF -2

0.6

31

294

47

96

21.4

4.6

21.5

7.8

373

16.67

92

6.26

HFR + PHF -3

0.6

26

271

43

87

21.3

4.8

22.5

8.0

357

16.38

99

6.30

1556 HFR -1

0.5

14

181

32

56

10.7

2.9

27.2

4.2

271

14.16

76

5.66

1556 HFR -2

0.5

24

206

38

51

10.5

2.4

23.0

3.6

232

14.61

61

5.42

1556 HFR -3

0.5

16

231

30

63

10.7

2.7

25.3

4.6

236

16.38

84

7.70

A-216

-------
Sample
Description -
Heat Flux
(50 kW/m2)

Sample

rhickness

(mm)

rime to
ignition

(s)

Peak
I IKK

(kW/m2)

Time to
Peak
I IKK

(s)

Average
I IKK

(kW/m2)

Starting

Mass

(g)

Total

Mass
Loss

GO

Weight %
Lost

(%)

Total
Heat
Release
(MJ/m2)

Total
smoke
Release
(m2/m2)

Avg. Effective
Heat of Comb.
(MJ/kg)

MARHE

(kW/m2)

FIGRA

1556 HFR + P -1

0.5

12

166

49

73

20.6

4.8

23.3

6.6

400

13.56

93

3.38

1556 HFR + P-2

0.5

9

186

34

69

20.6

4.3

20.9

6.1

383

13.99

92

5.47

1556 HFR + P-3

0.5

9

166

45

71

20.6

4.7

22.8

6.6

409

13.86

92

3.69

1556 HFR +PHF -1

0.5

18

197

34

76

20.0

4.0

20.0

6.4

294

15.73

88

5.79

1556 HFR + PHF-2

0.5

22

209

39

83

20.6

4.2

20.4

7.1

324

16.49

89

5.37

1556 HFR +PHF -3

0.5

22

221

44

82

20.5

4.2

20.5

6.5

310

15.31

84

5.01

A-217

-------
BFR Fire Behavior

Upon exposure to the cone heater, the sample began to smoke and make crackling sounds
very quickly. It then burst into flame with orange, blue, and purple colors noted. The sample was
noted to curl up some during burning with the 21K' sample curling and delaminating to a severe
degree such that the cone heater shutters could not close at the end of the experiments. ITeat
release was reproducible (Figure 7-1) and the final chars (Figure 7-2) were blackened with
copper plates noted. The sample where the shutters could not be closed is shown on the far left of
Figure 7-2 where the surface char has be slowly burned away leaving behind just copper and
fiberglass. So with sufficient heat and oxygen, eventually most of the carbon can be burned
away/ consumed.

BFR HRR



v.





I

	HRR-1

	HRR-2

	HRR-3

—



















Heat Flux
50 KW/m2









Flow Rate
15 Us















\









V





v—.—





0	20	40	60	80	100

Time (s)

Figure 7-1. HRR for BFR Sample

Figure 7-2. Final Chars for BFR Sample

BFR + P (populated halogen components)Fire Behavior

Fire behavior of this sample was the same as the BFR sample, but the flame colors were
more muted. The component powder was also noted to spit and pop a bit, with occasional pieces
of the powder leaving the aluminum foil holder. Heat release rates (Figure 7-3) were
reproducible indicating that the powder did not inhibit burning behavior. Final chars (Figure 7-4)
were black with yellowish-black powder on top.

A-218

-------
300

250

200

<4
£

£ 150
cr
tr
x

100

50

0

0	20	40	60	80	100 120

Time (s)

Figure 7-3, HRR for BFR + P Sample

Figure 7-4. Final Chars for BFR + P Sample

BFR + PHF(Populated halogen-free t omponents) h ire Behavior

Upon exposure to the heater, the sample smoked and crackled, and then ignited on one
side of the sample with the flames sweeping across the surface quickly. Flames were noted to be
blue and purple in color, and the component powder had a tendency to crackle and bubble,
suggesting the presence of thermoplastic material in the HF powder. HRR was fairly
reproducible (Figure 7-5) although the 2nd sample (HRR-2) has a higher peak HRR and delayed
time to ignition when compared to the other two samples. Final chars (Figure 7-5) were black
with copper squares noted. From this observation the halogen-containing component powder
does not flow (Figure 7-4) and may contain less thermoplastic material as opposed to the
halogen-free component powder which appears to burn up more completely and leave less of a
powdery residue.

BFR+P HRR











I

	HRR-1

	HRR-2

	HRR-3

.....























Heat Flux
50 kW/m2











Flow Rate
15 Us



		,













i









i i i

A-219

-------
350

300

250

200

<;

i;

en

§= 150

100

50

0

0	20	40	60	80	100 120

Time (s)

Figure 7-5. HRR for BFR + PHF Sample

Figure 7-6. Final Chars for BFR + PHF Sample

NFR Fire Behavior

Upon exposure the cone heater, the sample made a lot of crackling noises, and then began
to smoke before quickly igniting. The sample curled quite a bit during burning such that the
shutters could not be closed at the end of the experiment. Heat release (Figure 7-7) was very
reproducible and the final chars (Figure 7-8) show just the copper and fiberglass as most of the
residual carbon was burned away since the shutters would not close. Therefore any char which
had self-extinguished during the test was slowly pyrolyzed away until the sample could be
removed from the cone calorimeter.

BFR + PHF HRR





I







HRR-1
HRR-2
HRR-3































Heat Flux
50 kW/m2











Flow Rate
15 Us































I







A-220

-------
NFR HRR

Figure 7-7. HRR for NFR Sample

Figure 7-8. Final Chars for NFR Sample

lll R Fire Behavior

Upon exposure to the cone heater, the sample began to crackle and then smoke, followed
by ignition. The sample burned with some white colors, suggesting the presence of a
phosphorus-based flame retardant. The first sample curled during the test and the shutters could
not be closed. Some scatter in the HRR was noted (Figure 7-9), especially in the peak ITRR
values. Final chars (Figure 7-10) in general show black-grey chars on the surface of the
fiberglass, but some char is noted on the copper squares as well.

A-221

-------
HFR HRR









	HRR-1

	HRR-2

	HRR-3









Heat Flux
50 kW/m2







Flow Rate
15 Us













V .









0	50	100	150

Time (s)

Figure 7-9. HRR for HFR Sample

Figure 7-10. Final Chars for HFR Sample

HI R P (Comp) Fire Behavior

Upon exposure to the cone heater, the sample began to smoke right away, followed an
ignition and some loud crackling noises. Some parts of the powder also spat out of sample
surface during this burning behavior with some flames going out sideways from under the
powder. Some blue flames were noted at the beginning and end of the test. The third sample
tested had some curling and the shutters could not be closed at the end of the test. Heat release
(Figure 7-11) showed some scatter in the peak HRR values, but the scatter was not severe. Final
chars (Figure 7-12) were completely black and the powder is of a similar color, unlike the BFR
sample above which had the same component powder but the powder char was of a different
color at the end of the test (Figure 7-4). The curling observed for the 3rd sample can be seen in
the middle of Figure 7-12.

A-222

-------
HFR + PHRR

Figure 7-11. HRR for HFR + P Sample

Figure 7-12. Final Chars for HFR + P Sample

HFR + PHF Fire Behavior

Fire behavior for this sample was similar to that of the sample above, except no blue
colors were noted. All of the samples had a tendency to curl such that it was difficult to close the
shutters at the end of the test. Loud crackling and popping was heard, but no bubbling seen this
time as was observed for the BFR + PHF sample. HRR showed some scatter in the time to
ignition and peak HRR values (Figure 7-13). Final chars (Figure 7-14) showed intact charred
powder, but with more residual color noted. Some of the copper squares can be seen under the
charred component powder.

A-223

-------
HFR + PHF HRR

Figure 7-13. HRR for HFR + PHF Sample

Figure 7-14. Final Chars for HFR + PHF Sample

1556 HFR Fire Behavior

Upon exposure to the cone heater, the sample was heard to crackle and pop, then smoke,
then ignite. The sample had small flames which were not as sooty as those seen in previous
samples. The sample also curled during burning, but flaked apart as it burned, suggesting the
presence of a phenolic resin, or some sort of charring polymer. HRR (Figure 7-15) was not very
reproducible for this sample, with notable variability in the peak HRR and time to peak HRR
behavior. Final chars (Figure 7-16) are black and grey with regions of soot on the surface. Some
of the copper squares have moved suggested they debonded from the surface during burning.

A-224

-------
1556 HFR HRR











I

	HRFf-1

	HRR-2

	HRR-3





















Heat Flux
50 kW/m:















Flow Rate
15 Us





















!

I











0	20	40	60	80	100 120

Time (s)

Figure 7-15. HRR for 1556 HFR Sample





1 t'^l p £¦¦ fa



(

|

/

s>

Vj|

M

HI 1

I

1 V

lj

J





Figure 7-16. Final Chars for 1556 HFR Sample

1556 HFRU P Fire Behavior

Fire behavior for this sample was similar to that of sample 1556 HFR, but some blue
flames were noted as well. No real curling of the sample occurred when the powder was present,
but some spitting of the component powder out of the sample holder was noted. HRR (Figure
7-16) was fairly reproducible, with only the 2nd sample (HRR-2) showing variability in the peak
FIRR and time to peak HRR. Final chars (Figure 7-17) were black underneath with copper
squares and the powder was a dark yellow-green in color.

A-225

-------
1556 HFR + PHRR







	HRR-1

	HRR-2

	HRR-3







Heat Flux
50 kW/m2

Flow Rate
15 Us









1







0	50	100	150

Time (s)

Figure 7-17. HRR for 1556 HFR + P Sample

Figure 7-18. Final Char for 1556 HFR + P Sample

1556 HFR+ PHF Fire Behavior

Fire behavior for this sample was also similar to that of sample 1556 HFR, that some
colors were seen in the flames toward the end of the test with some blue and blue/green colors
noted. HRR (Figure 7-19) was reproducible and the final chars (Figure 7-20) were black and
grey with the powder being mostly intact.

A-226

-------
1556 HFR + PHF HRR







	HRR-1

	HRR-2

	HRR-3









Heat Flux
50 kW/m2









Flow Rate
15 Us









...







0	50	100	150

Time (s)

Figure 7-19. HRR for 1556 HFR + PHF Sample

Figure 7-20. Final Chars for 1556 HFR + PHF Sample

A-227

-------
Heat Flux-100 k\V m

Table 7-2. Heat Release Data (100 kW/m2)

Sample
Description
- Heat Flux
(50 kW/m2)

Sample
Thickness
(mm)

Time
to

ignition

(s)

Peak
I IKK

(kW/m2)

Time
to

Peak
I IKK

(s)

Average
I IKK

(kW/m2)

Starting
Mass

(r)

Total

Mass
Loss
(R)

Weight

%

Lost

(%)

Total
Heat
Release
(MJ/m2)

Total
smoke
Release
(m2/m2)

Avg.
Effective
Heat of
Comb.
(MJ/kg)

MARHE

(kW/m2)

FIGRA

BFR-1

0.4

3

227

15

56

10.2

4.2

41.1

4.5

476

11.05

129

15.11

BFR-2

0.4

5

391

15

80

10.7

4.9

45.8

5.7

451

11.58

180

26.04

BFR -3

0.4

3

357

15

77

10.4

4.7

45.3

5.4

393

11.72

189

23.79

NFR-1

0.3

3

356

15

80

8.8

3.2

36.5

5.3

195

17.75

188

23.76

NFR -2

0.4

4

490

15

94

9.5

3.7

38.9

6.6

230

18.37

201

32.70

NFR -3

0.3

4

387

15

71

8.8

3.3

37.5

5.0

220

15.91

153

25.83

HFR-1

0.5

6

495

20

104

10.9

4.2

38.6

7.4

231

18.49

205

24.74

HFR -2

0.5

6

495

20

105

11.2

4.0

35.8

7.5

238

20.75

216

24.76

HFR -3

0.5

5

367

25

120

14.1

5.7

40.5

10.2

326

17.95

201

14.68

A-228

-------
BFR Fire Behavior

Upon exposure to the cone heater, the sample quickly began to smoke and crackle, and
then ignited quickly. The flames were noted to be orange and blue in color. With some of the
samples, smoke would shoot out the sides of the sample and escape the cone calorimeter exhaust
ducting. Some of the samples also curled/deformed during testing. Heat release (Figure 7-21)
showed some notable scatter in the peak ITRR value for the 1st sample (HRR-1). The reasons for
this scatter with the 1st sample are not clear at this time, but perhaps this sample had slightly less
flammable epoxy mass than the other two samples tested. Final chars (Figure 7-22) were dark
grey with exposed glass fiber and burned/damaged copper metal squares.

400

350

300

250

§ 200

-------
NFR 100 kWHRR

Time (s)

Figure 7-23. 11R R for NFR Sample

Figure 7-24. Final Chars for NFR Sample

HFR Fire Behavior

Upon exposure to the heater, the sample began to smoke and crackle, with more of a
whiter smoke noted prior to ignition. Some deformation during burning was noted, and the
sample was noted to have a distinct smell to it when removed from the cone heater. HRR was
reproducible for the 1st two samples (HRR-1, HRR-2), but the third sample (HRR-3) shows a
lower peak HRR and a bit of delay in time to peak HRR (Figure 7-25). Again, reasons for this
difference are unclear at this time. Since some of the samples deformed greatly during testing, it
was not possible to close the cone heater shutters at the end of the test and so the samples were
exposed to additional heat at the end of the test after extinguishment which burned off additional
surface char, yielding light grey specimens of bare glass fiber (Figure 7-26). One of the samples

A-230

-------
did not deform as much and the shutters could be closed, giving a specimen with more surface
char (middle of Figure 7-26).

HF 100 kW HRR











—	HRR-1

—	HRR-2

—	HRR-3







Heat Flux:
100 kW/mJ









































	 _l

	

0	20	40	60	80	100

Time (s)

Figure 7-25. HRR for HFR Sample

Figure 7-26. Final Chars for HFR Sample

A-231

-------
Heat Release Rate

¦ Time to ignition (s) I Time to Peak HRR (s) aPeak HRR (kW/m2)

Figure 7-27. Heat Release Rate Plot

Overall Remarks on 50 kW/m Heat Flux Sample Burning Behavior:

There are notable interactions between the component powder and the polymer
decomposition chemistry going on as these samples burn. Brominated FR epoxy reacts
differently with halogen-containing and halogen-free component powder, as does the halogen-
free epoxy. The 1556 HFR sample also shows some differences when exposed to the two
different powders, but not to as great a degree seen with the BFR and HF epoxy samples. The
behavior of the HF comp powder is worth noting on here since in one case it showed bubbling
but not in others. This may be due to a unique flame retardant reaction in the presence of
brominated epoxy, but no obvious reason for this behavior can be given at this time.

The BFR samples, as expected, gave off lots of smoke and pyrolyzed some of the copper
away in the form of copper halides, which were seen in the flames as blue colors. The HF
samples showed some white colors indicating phosphorus release, but no blues until halogen-
containing component powder was added, suggesting that less copper was pyrolyzed during
burning. The 1556 HFR samples showed color in the presence of the halogenated powder, and
surprisingly in the presence of the HF component powder as well, indicating the components
again have an effect on metal pyrolysis/thermal reaction behavior.

Overall Remarks on Burning Behavior 100 kW m Heat Flux:

2

At 100 kW/m heat flux, the differences in fire behavior between the samples tested were
minimal, but there were some differences noted in physical burning behavior which correlate to

A-232

-------
2

the fire behavior noted at 50 kW/m heat flux. The brominated FR epoxy does give off more
smoke and does inhibit combustion as expected, and the blue colors noted during burning are
visual evidence of bromine reacting with copper under burning/pyrolysis conditions. The non-FR
sample burns quickly and rapidly (as a sample with no flame retardant should), and the non-
halogenated FR sample also shows physical fire behavior similar to that of the non-FR sample.
The non-halogenated FR has an equally high effective heat of combustion to that of the non-FR
sample which may just suggest that the flame retardant mechanism for this material has little
effect at very high heat fluxes, or at least does not inhibit combustion as much at very high heat
fluxes. Smoke release is slightly higher though, and so the non-halogenated FR sample is having
some effect on combustion products even if no change in measured heat of combustion is
observed.

A-233

-------
8 Appendix B: Experimental Conditions

Table 8-1. Ambient Conditions during Cone Testing

Experiment

#

Laminate
Description-Heat
Flux-kW/m2

Ambient Conditions

Temperature

°C

Relative
Humidity

%

Pressure
mbar

Cone Set
Temperature

°C

E2

BFR - 50

24

22

998

731

E4

BFR - 50

22.5

46

974

721

E6

BFR + P - 50

22.5

32

969

721

E8

BFR + P - 50

23

36

980

721

E10

BFR + PHF - 50

23

43

980

721

E30

BFR + PHF- 50

22.5

37

978

725

E12

NFR -100

22.5

45

981

978

E13

NFR -100

24

47

982

978

E15

BFR -100

23

43

975

937

E16

BFR -100

22.5

38

987

927

E18

HFR -100

22.5

44

986

924

E19

HFR -100

22.5

42

986

922

E21

NFR-50

22.5

38

987

740

E22

NFR-50

22.5

41

982

736

E24

HFR - 50

23

37

985

736

E25

HFR - 50

23

27

996

736

E27

1556 HFR - 50

22

37

986

727

E28

1556 HFR - 50

22

40

980

725

E32

HFR + P - 50

22

35

995

722

E33

HFR + P - 50

21.5

28

991

722

E35

HFR + PHF - 50

21.5

26

981

721

E36

HFR + PHF - 50

21.5

32

992

721

E38

1556 HFR + P - 50

22

32

981

721

E39

1556 HFR + P - 50

21.5

33

981

721

E41

1556 HFR + PHF-50

21.5

24

998

719

E42

1556 HFR + PHF-50

20.5

35

990

719

A-234

-------
9 Appendix C: Elemental Analyses of Component Mixtures

Table 9-1. Elemental Analyses of Component Mixtures

Substance

Low Halogen: Total
Mass(g) per 3052.25
g of mixture

Non-Low Halogen:
Total Mass(g) per
3052.25 g of mixture

1,4-BENZENEDICARBOXYLIC ACID, POLYMER WITH [1,1'-BIPHENYL]-4,4'-DIOL,

845.140

0.000

4-HYDROXYBENZOIC ACID, 6-HYDROXV-2-NAPHTHALENECARBOXVLIC ACID AND N-(4-HYDROXVPHENYL)ACETAMIDE (9CI)

845.140

0.000

1,4-B IS (2,3-E POXYP RO POXY) B UTA N E

0.002

0.002

ACRYLIC RESIN

0.135

0.135

AG (Silver)

8.208

8.208

AL (Aluminum)

0.004

0.004

AL203 (Aluminum oxide)

41.150

41.150

ANTIMONY TRIOXIDE

0.000

0.000

ARALDITE GY 250

1.721

1.721

AU (Gold)

7.065

7.065

B (Boron)

0.000

0.000

BARIUM TITANATE(IV)

453.479

453.479

BASIC DUROMER: POLYURETHANE RESIN (COMPOUND OF A POLYMERIC NETWORK)

1.082

1.082

BERYLLIUM

0.000

0.000

BROMINE

0.086

0.085

C.I. PIGMENT BLACK 28

0.281

0.281

CALCIUM

0.000

0.000

CALCIUM MONOXIDE

0.157

0.157

CALCIUM-CARBONATE

1.866

1.866

CARBON BLACK

12.662

1.318

CHLORINE

0.086

5.757

CHROMIUM

0.001

0.001

CHROMIUM(lll)OXIDE

0.355

0.355

COBALT, ELEMENTAL

0.615

0.615

COPPER (METALLIC)

425.069

425.069

COPPER OXIDE (CUO)

9.852

9.852

CRISTOBALITE

1.174

1.174

DIIRON-TRIOXIDE

121.742

121.742

DODECANE

0.014

0.014

DUMMY SUBSTANCE

0.002

0.002

Epoxy Resin

33.936

33.936

FE (Iron)

8.160

8.160

FIBROUS-GLASS-WOOL

277.933

453.768

FLOWERS OF ZINC (Zinc Oxide)

29.989

29.989

FORMALDEHYDE, OLIGOMERIC REACTION PRODUCTS WITH 1-CHLORO-2,3-EPOXTPROPANE AND PHENOL

1.906

1.906

FRITS, CHEMICALS

0.280

0.280

FUSED SILICA

374.758

374.758

IN (Indium)

0.000

0.000

LEAD

0.170

0.170

LEAD (II) OXIDE

0.062

0.062

LEAD (II) TITANATE

0.767

0.767

MAGNESIUM TITANIUM OXIDE (MG7103)

9.767

9.767

MAGNESIUM-OXIDE

0.131

0.131

MANGANESE

0.031

0.031

MO (Molybdenum)

0.355

0.355

NICKEL

101.263

101.263

NICKEL OXIDE

26.977

26.977

P (Phosphorous)

0.036

0.036

PALLADIUM

0.451

0.451

P-F-R-2

25.913

25.913

Polyphenylene Sulfide



674.980

SI (Silica)

14.265

14.265

SILICA

0.761

0.761

SILICONE

2.555

2.555

SN (Stannum/Tin)

7.623

7.623

SOLVENT NAPHTHA (PETROLEUM), HEAVY AROM.

0.018

0.018

STABILIZATION UV, LIGHT, HEAT

2.094

2.094

TUNGSTEN (W)

0.780

0.780

ZINC POWDER - ZINC DUST (NOT STABILIZED)

199.323

199.323

A-235

-------
10 References

(1)http://www.	electronicstakeback.com/wp-

content/uploads/Facts and Figures on EWaste and Recvcling.pdf (accessed 03/10/12)

(2)http://ngm.nationalgeographic.com/2008/01/high-tech-trash/carroll-text	(accessed 03/10/12)

(3)http://www.rohs.eu/english/index.html	(accessed 03/10/12)

(4)http://en.wikipedia.org/wiki/Waste	Electrical and Electronic Equipment Directive
(accessed 03/10/12)

(5)	"Fire-LCAModel: TV Case Study" Simonson, M.; Blomqvist, P.; Boldizar, A.; Moller, K.;
Rosell, L.; Tullin, C.; Stripple, H.; Sundqvist, J. O. SP Report 2000:13 ISBN 91-7848-811-7
Printed in 2000.

(6)	"Emissions from Fires Part I: Fire Retarded and Non-Fire Retarded TV-Sets" Blomqvist, P.;
Rosell, L.; Simonson, M. Fire Tech.2004, 40, 39-58.

(7)	"Emissions from Fires Part II: Fire Retarded and Non-Fire Retarded TV-Sets" Blomqvist, P.;
Rosell, L.; Simonson, M. Fire Tech.2004, 40, 59-73.

(8)	"A review of current flame retardant systems for epoxy resins" Weil, E. D.; Levchik, S. J.
Fire. Sc/,2004, 22, 25-40.

(9)	"Thermal decomposition, combustion and flame-retardancy of epoxy resins - a review of the
recent literature" Levchik, S. V.; Weil, E. D. Polym. Int.2004, 53, 1901-1929.

(10)	"A Review of Recent Progress in Phosphorus-based Flame Retardants" Levchik, S. V.;

Weil, E. D. J. Fire Sc/,2006, 24, 345-364.

(1 l)http://www.epa.gov/dfe/pubs/proiects/pcb/index.htm (accessed 03/10/12)
(12)http://www.epa.gov/dfe/pubs/proiects/pcb/full report pcb flame retardants report draft 11
10 08 to e.pdf (accessed 03/10/12)

(14)	Cone Calorimeter: ASTM E1354-07 "Standard Heat Method for Heat and Visible Smoke
Release Rates for Materials and Products Using an Oxygen Consumption Calorimeter"

(15)	"Full Scale Flammability Measures for Electronic Equipment" Bundy, M.; Ohlemiller, T.
NIST Technical Note 1461 August 2004, US Dept. of Commerce.

(16)"Bench-Scale	Flammability Measures for Electronic Equipment" Bundy, M.; Ohlemiller, T.

NISTIR (National Institute of Standards and Technology Internal Report) 7031 July 2003, U.S.
Dept. of Commerce.

(17)	"Mechanistic Study of the Combustion Behavior of Polymeric Materials in Bench-scale
Tests. I. Comparison between Cone Calorimeter and Traditional Tests" Costa, L.; Camino, G.;
Bertelli, G.; Borsini, G. Fire and Materials 1995, 19, 133-142.

(18)	"Oxygen Index: Correlations to Other Fire Tests" Weil, E. D.; Hirschler, M. M.; Patel, N.
G.; Said, M. M.; Shakir, S. Fire and Materials 1992, 16, 159-167.

(19)	"Comprehensive fire behaviour assessment of polymeric materials based on cone
calorimeter investigations" Schartel, B.; Braun, U. e-Polymers 2003, No. 13. http://www.e-
polymers.org/papers/schartel_010403.pdf.

(20)	"Cone calorimeter analysis of UL-94 V-rated plastics" Morgan, A. B.; Bundy, M. Fire and
Materials 2007, 31, 257-283.

(21)	"Ignition performance of new and used motor vehicle upholstery fabrics" Spearpoint, M.;
Olenick, S. M.; Torero, J. L.; Steinhaus, T. Fire and Materials 2005, 29, 265-282.

(22)	"Assessment of Flame Retardancy in Polyolefin-Based Non-Halogen FR Compounds"
Cogen, J. M.; Whaley, P. D.; Lin, T. S.; Bolz, K. Proceedings of the International Wire and
Cable Symposium (2004).

A-236

-------
(23)	"The Assessment of Full-Scale Fire Hazards from Cone Calorimeter Data" Petrella, R. V. J.
Fire Sci. 1994, 12, 14-43.

(24)"Some	comments on the use of cone calorimeter data" Schartel, B.; Bartholmai, M.; Knoll,
U. Polym. Degrad. Stab. 2005, 88, 540-547.

(25)	"Heat release rate measurements of thin samples in the OSU apparatus and the cone
calorimeter" Filipczak, R.; Crowley, S.; Lyon, R. E. Fire Safety Journal 2005, 40, 628-645.

(26)	"Heat Release Rate: The Single Most Important Variable in Fire Hazard" Babrauskas, V.;
Peacock, R. D. Fire Safety Journal 1992, 18, 255-272

(27)"New	Fire Classifications and Fire Test Methods for the European Railway Industry"
Duggan, G. J.; Grayson, S. J.; Kumar, S. Flame Retardants 2004 Proceedings, January 27-28,
2004, London, UK, Interscience Communications

A-237

-------