United States
Environmental Protection
Agency
Office of
Toxic Substances
Washington DC 20460
EPA-560/5-85-022
May, 1985
Toxic Substances
&EPA
Products of
Thermal Degradation
of Dielectric Fluids
Tetra CDFs
Tetra CDDs
-------
PRODUCTS OF THERMAL DEGRADATION OF DIELECTRIC FLUIDS
By
Stephen E. Swanson
Mitchell D. Erickson
Leslie Moody
WORK ASSIGNMENT NO. 23
INTERIM REPORT NO. 2
EPA Contract No. 68-02-3938
MRI Project No. 8201-A(23)
May 23, 1985,
For
U.S. Environmental Protection Agency
Office of Toxic Substances
Field Studies Branch, TS 798
401 M Street, S.W.
Washington, DC 20460
Attn: Joseph J. Breen, Project Officer -
Daniel T. Heggem, Work Assignment Manager
-------
DISCLAIMER
This document has been reviewed and approved for publication by
the Office of Toxic Substances, Office of Pesticides and Toxic Substances,
U.S. Environmental Protection Agency. The use of trade names or commercial
products does not constitute Agency endorsement or recommendation for use.
-------
PREFACE
This report presents the results of a portion of Work Assignment
No. 23 on U.S. Environmental Protection Agency Contract No. 68-02-3938, "In-
cineration Testing of PCBs." The work was done at Midwest Research Institute
(MRI) during the period September 28, 1984, to January 28, 1985. This work
was performed as a follow-up to work presented previously in "Thermal Degrada-
tion Products From Dielectric Fluids," by Mitchell D. Erickson, Christopher J.
Cole, Jairus D. Flora, Jr., Paul G. Gorman, Clarence L. Haile, Gary 0. Hinshaw,
Fred C. Hopkins, and Stephen E. Swanson, Interim Report No. 1 on this work
assignment, November 19, 1984 (Report No. EPA-560/5-84-009). Mitchell D.
Erickson was the MRI Work Assignment Leader. This report was prepared by
Dr. Erickson, Stephen E. Swanson and Leslie Moody. The thermal combustion
system was operated by Gary D. Hinshaw, Christopher J. Cole, Paul G. Gorman,
and Fred C. Hopkins. Laboratory work was done by Mr. Swanson and Ms. Moody.
The GC/MS data were acquired by John Gamble, Jon Onstot, Gil Radolovich, and
Margaret Wickham. Mass spectral data were interpreted by Dr. Erickson,
Mr. Swanson and Ms. Moody.
The EPA Work Assignment Manager, Daniel T. Heggem of Field Studies
Branch, provided helpful guidance and advice.
RESEARCH INSTTUTE
Clarence L. Haile
Deuty Program Manager
rohn E. Going
Program Manager
Approved:
.^A^tUJU6^
James L. Spigarelli, Director
Chemical and Biological Sciences
Department
-------
TABLE OF CONTENTS
Page
I. Introduction . 1
II. Summary 1
III. Experimental Methods 2
A. Thermal Combustion System 2
B. Chemical Analysis 4
C. Selection of Compounds for Evaluation. ...... 4
IV. Results and Discussion 6
A. Continuous Monitoring Results 6
B. PCB-Containing Matrices 6
C. Non-PCB Dielectric Fluids 12
D. Quality Assurance and Quality Control 20
V. References 24
-------
LIST OF FIGURES
Number Title Page
1 Summary of conversion efficiencies (%) for total PCDFs
from Phase III and Phase IV runs 3
2 PCDF formation in PCB-spiked mineral oil by homolog
(Phase III and Phase IV results) 11
3 PCDF and PCDD formation from trichlorobenzene trans-
former fluid (Phase IV results) 15
4 PCDF and PCDD formation from trichlorobenzene trans-
former fluid (Phase III and Phase IV results) 16
5 PCDF formation from Wecosol® 17
6 Tetra- and pentaCDFs in sample 11-19-94-WEC ... 18
7 Hexa-, hepta-, and octaCDFs in sample 11-19-94-WEC 19
8 PCDF formation from OCDF-spiked feed, expressed as
percent of OCDF fed into system 21
9 Tetra- and pentaCDFs in sample 11-30-100-OCDF 22
10 Hexa-, hepta-, and octaCDFs in sample 11-39-100-OCDF. ... 23
LIST OF TABLES
Number Title Page
1 Operating conditions for Phase IV tests 5
2 PCDF formation from PCB-spiked mineral oil 7
3 Percent conversion PCB to PCDF 8
4 PCB destruction efficiencies for PCB-spiked mineral oil . . 10
5 PCDF formation from non-PCB dielectric fluids and OCDF. . . 13
6 PCDD formation from non-PCB dielectric fluids and OCDF. . . 14
IV
-------
I. INTRODUCTION
The Environmental Protection Agency (EPA) has issued an Advance
Notice of Proposed Rulemaking (ANPR) (USEPA 1984a) and proposed rule (USEPA
1984b) to gather data on the specific risks posed by fires involving electri-
cal transformers that contain polychlorinated biphenyls (PCBs) and also on
mechanisms for mitigating or eliminating.these risks. A work assignment was
issued to Midwest Research Institute (MRI) to support EPA's data-gathering
activities under the ANPR. Specifically, MRI was asked to study the combus-
tion of transformer dielectric fluids to investigate the potential for forma-
tion of toxic products such as polychlorinated dibenzofurans (PCDFs) and poly-
chlorinated dibenzo-p-dioxins (PCDDs). A report (Erickson et al. 1984) pre-
sents the results of previous work completed on this work assignment.
This report describes the results of a series of experiments which
were conducted to augment the data presented in the previous report (Erickson
et al. 1984). The work presented here includes both repeat runs of those pre-
viously reported and also investigations of materials not previously tested.
The next section presents a summary of the results. Section III presents the
experimental methods. Section IV presents the results of the tests performed
with the dielectric fluids.
II. SUMMARY
Two types of experiments were performed during Phase IV of this
project. First, a series of repeat runs were made under identical conditions
to those previously reported (Erickson et al. 1984). In these runs, the feed
'oil was mineral oil spiked with individual PCB congeners as in Phase II, and
mineral oil spiked with Aroclor 1254 as in Phase III. Also during Phase IV,
repeat runs were made using a trichlorobenzene dielectric fluid. The second
group of Phase IV runs were performed to investigate PCDF and PCDD formation
from the combustion of other non-PCB containing dielectric fluids which had
not previously been tested. The three dielectric fluids which were tested
were tetrachloroethylene (Wecosol®), a paraffinic high temperature hydrocarbon
(RTEmp®), and an esterified high temperature hydrocarbon (Dielectric fluid
HT-310-S). Finally two runs were made in which the feed oil consisted of
mineral oil spiked with octachlorodibenzofuran (OCDF). These runs were made
to better characterize the chemistry of PCDF formation.
The results of the repeat runs correlate well with the previously
reported results. The formation of PCDFs and PCDDs from the PCB-spiked min-
eral oil, as well as the calculated destruction efficiencies for PCBs, closely
match the previously reported results. The repeat chlorobenzene runs show
higher levels of PCDF formation than in Phase III. However, several concen-
trations in the previous report are "greater than" values; hence, comparison
is difficult.
-------
Significant amounts of PCDFs and PCDDs were formed from the tetra-
chloroethylene fluid. The two high temperature hydrocarbon fluids did not
produce PCDFs or PCDDs. Figure 1 summarizes the conversion efficiency of
dielectric fluid to PCDF for the different dielectric fluid types tested in
Phases III and IV. It appears that, under these thermal destruction condi-
tions, the extent of PCDF formation varies among dielectric fluids by several
orders of magnitude. The highest PCDF formation was noted for PCB-containing
dielectric fluids, and the least formation was observed in tetrachloroethylene.
It should be noted that during Phase II the thermal combustion system was op-
timized for PCDF formation from PCBs. The results for the OCDF-spiked feed
oil runs showed formation of a number of lower chlorinated PCDF homologs with
a variety of isomers within each homolog.
III. EXPERIMENTAL METHODS
A. Thermal Combustion System
1. Description
MRI has developed a bench-scale thermal combustion test system which
can be used to examine various combustion processes. It has also been used
to provide data on the incinerability (i.e., destruction efficiency) of haz-
ardous compounds in solid or liquid waste material, and to provide data on
products of incomplete combustion that may be formed. This system can pro-
vide combustion data on gram-quantity samples of materials in either solid
or liquid form or even semisolid materials, such as tars. In contrast, other
related systems often can handle only very small quantities of pure compounds.
The system is described in detail in a previous report which pre-
sents the results of previous work on this work assignment (Erickson et al.
1984). Briefly, the system consists of a volatilizing/pyrolysis heater for
the sample, an air preheater furnace, and the main combustion furnace (all
electrically heated). Separate volatilization/pyrolysis furnaces are used
for solid/semisolid feed and for liquid feed. Gas flow through the combus-
tion furnace is laminar and can be varied to provide different gaseous resi-
dence times. The combustion temperature can also be varied, up to a maximum
of 1200°C (2200°F). The liquid feed rate, which is controlled by a syringe
pump, can be varied over a nominal range of 1 uL/min to 1 mL/min.
Primary operating conditions that can be varied and controlled in
this system are temperature, oxygen concentration, and residence time. For
this program, the system has been operated using continuous injection of
liquid feeds.
The 02, C02, and CO concentrations in the effluent gases were con-
tinuously monitored during each run. Semivolatile organic compounds were also
collected in the effluent stream by adsorption and concentration in an XAD-2
sampling trap.
-------
PCDFs
PCDDs
HP'r-
10'
,-2
-o
V
01
U.
I'
CL.
w
o>
w
O
10-3
10-5
c
o
U
icr6
10-7
PCBs Feed (16 Runs)
Trichlorobenzene
Feed (2 Runs)
Tefrrochloroefhylene
Feed (2 Runs)
PCB Feed
(2 Runs)
PCDFs
PCDDs
Figure 1. Summary of conversion efficiencies for total
PCDFs from Phase III and Phase IV runs.
-------
2. Operating Conditions
The pertinent operating conditions are listed in Table 1. The ther-
mal combustion system was operated under the conditions optimized for PCDF
formation from PCBs and used in all of the Phase III experiments described
previously (Erickson et al. 1984). However, it is recognized that the optimal
conditions for PCDF formation from other compounds are probably different.
The key operational parameters were a combustion temperature of 675°C, with a
residence time of 0.8 s, with 8% excess oxygen. All samples were fed at a
rate of 14 uL/min.
B. Chemical Analysis
Samples were analyzed as described in the previous report (Erickson
et al. 1984). Each effluent sample consisted of two parts, the XAD-2 resin
cartridge and a solvent rinse of the sample collection apparatus. The XAD-2
samples were spiked with 13C-surrogates and Soxhlet-extracted. This extract
was combined with the associated solvent rinse to make a combined effluent
extract. The combined extract was evaporatively concentrated to 2 mL. One-
half of each extract was cleaned by chromatography on acidified silica and
acidified alumina. The cleaned extracts were analyzed for PCDFs and PCDDs.
The remaining extract aliquot was analyzed for PCBs. Some of these extracts
were also screened for other chlorinated organics.
The high resolution gas chromatography/electron impact mass spec-
trometry (HRGC/MS) analysis of these effluent extracts was performed using
selected ion monitoring. The analytical conditions used were:
Column: 30 m x 0.25 mm fused silica column, wall-coated with DB-5
Column Temperature: 100°C (1 min hold) to 325°C at 10°C/min
Injector: Grob-type, 45 s splitless, 280°C
Electron Energy: 70 eV
C. Selection of Compounds for Evaluation
1. PCBs
Five repeat experiments were performed in which the feed oil was
PCB-spiked mineral oil. In three of these experiments (runs 90-92), the min-
eral oil (Exxon HPLX 355077) was spiked with 50 ug/g each of 2,3,5,6-tetra-,
3,3',4,4',5,5'-hexa-, and 2,2',4,4',6,6'-hexachlorobiphenyl. These compounds
were purchased from Ultra Scientific. Runs 90-92 were performed under the
same conditions as run 26 reported from the previous work. In two experiments
(runs 97 and 98), the mineral oil (Exxon HPLX 355077) was spiked with 50 ug/g
of Aroclor 1254 (Monsanto). These two runs are duplicates of runs 45 and 46
from the previous work.
2. Chlorobenzene
A Chlorobenzene dielectric fluid was fed during two runs (runs 106
and 107). This Chlorobenzene fluid (Electro-Chem FR-15, Standard Chlorine
Chemical Company in Kearny, NJ) was determined to contain mostly trichloro-
benzene isomers with some tetrachlorobenzene. This is the same Chlorobenzene
fluid which was previously tested as runs 57 and 58.
-------
Table 1. Operating Conditions for Phase IV Tests
Feed material
Run no.
Combustion Run
temperature time
(°C) (min)
Excess Residence
oxygen time C02
(s)
CO
Combustion
efficiency
Pyrolysis
temperature
PCB congeners
PCB congeners
PCB congeners
None
Tetrachloroethy1ene
Tetrachloroethy1ene
None
Aroclor 1254
Aroclor 1254
Mineral oil
OCDF
OCDF
OCDF
HTH (para)
HTH (ester)
None
Trichlorobenzene
Trichlorobenzene
None
11-14-90-M+M+M 678
11-14-91-M+M+M 678
11-16-92-M+M+M 678
ll-16-93-MO(B)a 677
11-19-94-WEC 676
11-20-95-WEC 676
ll-20-96-WEC(B)a 676
11-28-97-M50 677
11-28-98-M50 677
11-29-99-MO 678
11-30-100-OCDF 678
11-30-101-OCDF 679
11-30-102-OCDF(B) 677
12-3-103-HTH(P) 676
12-4-104-HTH(E) 677
12-4-105-HTH(B) 676
12-5-106-CLBZ 677
12-5-107-CLBZ 677
12-5-108-CLBZ(B) 677
58
60
61
55
60
57
56
61
57
61
37
35
55
44
45
45
16
16
61
7.9
7.9
7.9
7.9
7.8
8.0
8.0
7.9
8.0
7.9
7.9
7.9
7.9
7.9
8.0
8.0
7.8
8.1
7.9
0.80
0.81
0.79
0.79
0.80
0.79
0.80
0.82
0.81
0.79
0.81
0.78
0.77
0.79
0.81
0.81
0.79
0.78
0.78
0.22
0.21
0.27
0.07
0.08
0.07
0.06
0.12
0.12
0.13
0.14
0.15
0.06
0.09
0.12
0.07
0.08
0.07
0.07
0.37
0.40
0.29
0.0005
0.03
0.01
0.0005
0.32
0.33
0.52
0.33
0.33
0.0003
0.09
0.21
0.0002
0.16
0.16
0.0009
37
34
48
99
70
85
99
27
26
19
29
30
99
47
35
99
32
29
98
397
397
397
397
397
397
397
397
397
397
397
398
398
444
444
445
397
397
397
.Runs with "(B)" suffixes are blank runs with no feed.
Mineral oil with no PCBs added.
-------
3. Wecosol®
This dielectric fluid, containing tetrachloroethylene (> 99%), was
also tested on runs 94 and 95. It was obtained from Westinghouse Corporation,
Pittsburgh, PA.
4. HTH
Two high temperature hydrocarbon dielectric fluids were tested on
runs 103 and 104. The first type of fluid is a paraffinic hydrocarbon [desig-
nated HTH(P) in run 103] labeled "RTEmp®." It was obtained from RTE Corpora-
tion, Waukesha, WI. The other fluid is an esterified hydrocarbon [designated
HTH(E) in run 104]. It was labeled "Dielectric fluid HT-310-S" and was ob-
tained from Standard Chlorine Chemical Company, Kearny, NJ.
5. Octachlorodi benzofuran
Two runs were completed (runs 100 and 101) using a feed which con-
sisted of 500 ug/g of octachlorodibenzofuran (OCDF) from Ultra Scientific in
mineral oil (Exxon HPLX 355077). It was apparent from the previous work that
a variety of PCDF congeners are formed from many precursors in the thermal
combustion system. It was not apparent from this work, the available litera-
ture, or communication with other researchers, whether these;PCDFs were formed
directly from precursors or whether a highly chlorinated dibenzofuran (e.g.,
heptaCDF or octaCDF) was first formed and then dechlorinated. It was decided
that using a feed oil spiked with OCDF could provide insight about the chem-
istry of PCDF formation.
IV. RESULTS AND DISCUSSION
A. Continuous Monitoring Results
Table 1 lists the thermal combustion system operating conditions
for the tests completed during this phase of the program. This table shows
the combustion temperatures, the percent excess oxygen measured in the combus-
tion effluent, the calculated residence time, the percent C02 and percent CO
measured in the combustion effluent, the calculated combustion efficiency,
and the pyrolysis temperature.
B. PCB-Containing Matrices
1. PCB Isomers ,
In three runs (Nos. 94, 95, 96), the feed oil consisted of mineral
oil spiked with 500 ug/g each of the three PCB isomers 2,3,5,6-tetra-,
3,3'4,4',5,5'-hexa-, and 2,2',4,4',6,6'-hexachlorobiphenyl (replicates of
run 26 in the previous work). The results of analysis of PCDFs in these runs
are listed in Table 2. Table 3 lists the conversions of PCBs to PCDFs. The
literature (Buser et al. 1978; Buser and Rappe 1979) suggests that each of
the three PCB isomers produces specific PCDF isomers as shown below:
-------
Table 2. PCDF Formation from PCB-Spiked Mineral Oil
Run no.
11-14-90-M+M+M
11-14-91-M+M+M
11-16-92-M+M+M
ll-16-93-MO(B)C
11-28-97-M50
11-28-98-M50
11-29-99-MO
MonoCDF
(ng)
770
810
810
< 5d
-
DiCDF
(ng)
550
520
< 5
< 5
-
-
TriCDF
(ng)
2,100
4,500
3,200
< 5
-
-
TetraCDF
(ng)
1,400
2,600
2,000
82
250
190
< 10
PentaCDF
(ng)
240
320
830
73
30
20
< 10
HexaCDF
(ng)
73
< 30
280
85
< 30
31
< 30
HeptaCDF
(ng)
< 20b
< 20
< 20
110
< 20
16
< 20
OctaCDF
(ng)
< 10
< 10
< 10
12
< 10
< 10
< 10
PCDFsa
(ng)
5,100
8,800
7,100
360
280
240
0
b~The not detected values (< X) were considered to be zero for calculating total PCDF concentration.
Not detected; estimated method quantitation limit is given.
.Runs with "(B)" suffixes are blank runs with no feed.
- = Not analyzed.
-------
Table 3. Percent Conversion PCB to PCDF
MonoCDF DiCDF TriCDF TetraCDF PentaCDF HexaCDF HeptaCDF OctaCDF
Conversion Conversion Conversion Conversion Conversion Conversion Conversion Conversion Total
efficiency efficiency efficiency efficiency efficiency efficiency efficiency efficiency PCDF
Run no. (%) (%) (%) (%) (%) (%) (%) (%) (%)
11-14-90-M+M+M
11-14-91-M+M+M
11-16-92-M+M+M
11-28-97-M50
11-28-98-M50
NCa
NC
NC
_c
NC
NC
NC
-
0.61
1.14
0.88
-
0.20
0.37
0.29
0.69
0.55
0.035
0.046
0.12
0.082
0.058
0.021
Ob
0.077
0
0.091
NC
NC
NC
0
0.047
NC
NC
NC
0
0
0.37
0.85
0.39
0.77
0.70
bNC = Not calculated; conversion efficiencies only calculated for proposed reaction mechanisms.
0 = not detected.
- = not analyzed
-------
2,3,5,6-Tetrachlorobiphenyl —^ 1,2,4-Trichlorodibenzofuran
2
3,3',4,4',5,5'-Hexachlorobiph'enyl —^ 2,3,4,6,7,8-Hexachlorodibenzofuran
U2
r» 1,3,7,9-Tetrachlordibenzofuran
2,2',4,4',6,6'-Hexachlorobiphenyl -jj-J
2U 1,3,4,7,9-Pentachlorodibenzofuran
The conversion efficiencies listed in Table 3 assume the above reactions apply.
As noted in the previous report, only one isomer each of tn'CDF and tetraCDF
was present in the samples. However, several pentaCDF and hexaCDF isomers
were identified, indicating that either the chemistry of the PCB-to-PCDF con-
version is more complex than originally thought or the PCDFs are degrading
and rearranging after the initial formation. A total PCB to total PCDF con-
version is also listed for each run in Table 3. No PCDDs were detected in
these samples.
The PCDF formations in the three M+M+M runs (90, 91, and 92) were
similar. The results are also in agreement with those previously obtained
under these experimental conditions (Table 14 in Erickson et al. 1984). Al-
though the current results are not as high as those in the previous report,
they are higher than the amounts observed under any other experimental condi-
tions. Thus, the current results corroborate the conclusion reached in the
previous report that combustion at 675°C for 0.8 s with 8% excess oxygen is
optimum for formation of PCDFs from PCBs.
The combustion effluents from these PCB-spiked mineral oil runs
(Nos. 90, 91, and 92) were also analyzed for PCBs. From these results, the
PCB destruction efficiencies were calculated for each run using the formula
presented in the previous report. The destruction efficiencies are shown in
Table 4. These results (91, 93, and 85%) compare well with the previously
reported destruction efficiencies (Table 30 in Erickson et al. 1984).
2. Aroclor 1254
In two runs (97 and 98) the feed oil consisted of mineral oil spiked
with 50 ug/g of Aroclor 1254 (Monsanto Corporation). These runs duplicate
runs 45 and 46 from the previous work. The results of the analysis of PCDFs
in these runs are also shown in Table 2. These results are generally compar-
able to those previously reported. The calculated conversion efficiencies
for PCB to PCDF are shown in Table 3. No PCDDs were found in the samples.
A conversion efficiency is given for each PCDF homolog. Unlike the conversion
efficiencies calculated in individual PCB isomers, these conversion efficien-
cies were calculated using the total concentration of Aroclor 1254 feed.
Figure 2 shows a plot of the nanograms of each PCDF homolog detected in these
runs per milliliter of spiked mineral oil feed versus homolog. This figure
combines the data from this phase of work with a similar graph from the pre-
vious report (Figure 22 in Erickson et al. 1984). The overall formation
efficiency is similar to that previously reported. However, the PCDF distri-
bution in the current work tends toward higher chlorination.
-------
Table 4. PCB Destruction Efficiencies for
RGB-Spiked Mineral Oil
Destruction efficiency
Run no. (%)
11-14-90 M+M+M 81
11-14-91 M+M+M 93
11-16-92 M+M+M 85
11-28-97 M50 73
11-28-98 M50 72
10
-------
10,000
1,000
100
u_
Q
U
a.
O)
10
Not
Detected
Mono Di Tri Tetra Penta
PCDF
Hexa Hepta Octa
Figure 2. PCDF formation in PCB-spiked mineral oil by homolog (Phase III
and Phase IV results). Closed symbols are averages of two
values; open symbols are single determinations; missing points
are no data. Phase IV data are superimposed on Phase III data,,
previously published as Figure 22 in Erickson et al. (1984).
-------
The combustion effluent from the Aroclor 1254-spiked feed runs (Nos.
97 and 98) was also analyzed for PCBs. The PCB destruction efficiency was
calculated as described in the previous report. These results are shown in
Table 4. These two calculated destruction efficiencies (73 and 72%) compare
well with the calculated destruction efficiencies reported previously (Table
30 in Erickson et al. 1984).
C. Non-PCB Dielectric Fluids
1. Chlorobenzene Fluid
In runs 106 and 107 the feed was a non-PCB containing chlorobenzene
dielectric fluid (RF-15 from Standard Chlorine Chemical Company). These runs
duplicate runs 57 and 58 from the previous work. The major compounds identi-
fied in the analysis of the combustion effluents include dichloro-, trichloro-,
tetrachloro-, pentachloro-, and hexachlorobenzene. No PCBs were detected in
chlorobenzene combustion effluents.
The results of the analysis of PCDFs in these runs are shown in
Table 5. The PCDF levels detected for runs 106 and 107 are higher than pre-
viously reported (runs 57 and 58 in Erickson et al. 1984). However, several
concentrations in the previous report are "greater than" values; hence, com-
parison is difficult.
Table 6 lists the amounts of PCDDs formed from these chlorobenzene
dielectric fluid runs. The total amount and overall distribution of PCDDs
detected are comparable to the levels previously reported (runs 57 and 58 in
Erickson et al. 1984). Figure 3 shows a plot of the nanograms of each PCDF
and PCDD homolog detected in these runs per milliliter of chlorobenzene fluid
fed versus homolog. Figure 4 combines these data with a similar plot from
the previous report (Figure 25 in Erickson et al. 1984). The isomeric dis-
tributions of the PCDFs and PCDDs in these runs were similar to those from
the previous runs (see Figure 20 in Erickson et al. 1984).
2. Wecosol®
As shown in Table 5 and Figure 5, fairly consistent amounts of
tetraCDF through octaCDF were formed from thermal reaction of Wecosol®. While
the conversion efficiency (Figure 1) is low relative to pure PCBs, it does
correspond roughly to the PCDF formation from 500 ppm Aroclor 1254 (Erickson
et al. 1984). Low amounts of hexa- through octaCDD were observed (Table 6)
in the Wecosol® runs. However, only slightly lower amounts of the PCDDs were
found in the run immediately following the two Wecosol® runs. Therefore, the
significance of the PCDD values cannot be determined from these data.
Figures 6 and 7 present extracted ion current profiles of the PCDFs.
The isomeric distribution from the tetrachloroethylene precursor is slightly
different from those previously observed from reaction of PCBs or trichloro-
benzene (Figures 10-20 in Erickson et al. 1984).
12
-------
Table 5. PCDF Formation from Non-PCB Dielectric Fluids and OCDF
Run no.
MonoCDF DiCDF TriCDF TetraCDF PentaCDF HexaCDF HeptaCDF OctaCDF PCDFs
(ng) (ng) (ng) (ng) (ng) (ng) (ng) (ng) (ng)
11-19-94-WEC
11-20-95-WEC
ll-20-96-WEC(B)D
11-30-100-OCDF
11-30-101-OCDF
11-30-102-OCDF(B)
12-3-103-HTH(P)
12-4-104-HTH(E)
12-4-105-HTH(B)C
12-5-106-CLBZ
12-5-107-CLBZ
12-5-108-CLBZ(B)c
-a - - 630
2,400
< 10
3,600
3,700
170
33
< 10
1,100
1,100 5,900 32,000 79,000
570 1,700 3,700 32,000
84 140 240 610
600
2,400
91
6,200
4,300
120
< 10
< 10
1,500
93,000
48,000
960
920
1,800
140
29,000
42,000
190
25
< 30
1,000
9,800
22,000
1,300
1,100
1,500
150
41,000
50,000
260
35
40
340
19
690
1,300
270
320
42
14,000
20,000
150
62
< 10
80
< 10
70
750
3,500
8,400
420
94,000
120,000
890
160
40
4,020
220,000
109,000
5,400
u~ = Not analyzed.
Runs with suffixes "(B)" are blank runs with no feed.
Extracted after the OCDF and CLBZ samples (Run Nos. 11-30-100-OCDF, 11-30-101-OCDF, 12-5-106-CLBZ, and
12-5-101-CLBZ). Observed PCDFs may be carryover in glassware or carryover in thermal combustion system.
-------
Table 6. PCDD Formation from NorrPCB Dielectric Fluids and OCDF
Run no.
11-19-94-WEC
11-20-95-WEC .
ll-20-96-WEC(B)D
11-30-100-OCDF
11-30-101-OCDF
11-30-102-OCDF(B)
12-3-103-HTH(P)
12-4-104-HTH(E)
12-4-105-HTH(B)
12-5-106-CLBZ
12-5-107-CLBZ
12-5-108-CLBZ(B)
MonoCDD
(ng)
_a
-
-
-
-
-
-
-
-
24
81
< 5
DiCDD
(ng)
-
-
-
-
-
-
-
-
40
150
< 5
TriCDD TetraCDD
(ng) (ng)
< 10
< 10
< 10
< 10
< 10
< 10
< 10
< 10
< 10
1,100 600
350 680
< 5 < 10
PentaCDD
(ng)
< 10
< 10
< 10
< 10
< 10
< 10
< 10
< 10
< 10
< 10
63
< 10
HexaCDD
(ng)
57
51
< 30
< 30
< 30
< 30
< 30
< 30
< 30
< 30
94
< 30
HeptaCDD
(ng)
130
72
72
< 20
< 20
< 20
< 20
< 20
< 20
< 20
< 20
< 20
OctaCDD
(ng)
120
49
46
< 10
< 10
< 10
< 10
< 10
< 10
< 10
< 10
< 10
PCDDs
(ng)
310
170
120
< 80
< 80
< 80
< 80
< 80
< 80
1,800
1,400
< 100
.- = Not analyzed.
Runs with suffixes "(B)" are blank runs with no feed.
-------
o
D
"D
I
o>
c
100.000,000
10.000,000
1,000,000
100,000
10,000
1,000
100
10
«
Not'
Detected
Mono
Di
Tri
Tetra Penta
Homolog
Hexa
Hepta
Octa
Figure 3. PCDF and PCDD formation from trichlorobenzene transformer fluid (Phase IV results).
Closed symbols are averages of two values; open symbols are single determinations.
-------
u
D
-a
O)
100,000.000
10,000.000
1,000,000
100.000
10,000
1,000
100
10
Not'
Detected
_ o
Mono Di Tri Tetra Penta
Homo log
Hexa
Hepta Octa
Figure 4. PCDF and PCDD formation from trichlorobenzene transformer fluid
(Phase III and Phase IV results). Solid lines are Phase IV re-
sults; broken lines are Phase III results, originally published
as Figure 25 in Erickson et al. (1984). Closed symbols are
averages of two values; open symbols are single determinations.
-------
10,000
1,000
15
*5
Q
U
Q.
O)
c
100
10
ND
Mono
Di
Tri "Tetra Penta
PCDF
Hexa Hepta Ocha
Figure 5. PCDF formation from Wecosol®. Closed symbols are averages of two values;
missing points are no data.
-------
MID MASS CHROMATOGRANS DATAi 8281A14R1 «1
81/14/85 12:35:88 CALIj MID380A09R1 §2
SAMPLE: RUN 11-19-94-HEC,SONIFIED,2UL INJ
CONOS.i -2208EMU 78EU IMA 38M-DB5 188-1H-325-18/-45SS GROB
RANGE: G 1, 1 LABEL: N 8, 4.8 QUAN: A 8* 1.8 J 8 BASEi U 28,
SCANS 1885 TO 1258
TetraCDFs
676864.
386
385.988
i 8.588
CO
.8-1
348
1486848.
PentaCDFs
i
1050
17:32
339.980
i 8.560
1108
18:22
1158
19:13
1288
20:83
1250 SCAN
28:53 TIME
Figure 6. Tetra- and PentaCDFs in sample 11-19-94-WEC.
-------
96.8-1
374
100.0-1
408
30.1-.
444
HID MASS CHROMATOGRAMS DATAi 8201A16R1 11
81/16/85 10:22:60 CALI: MID355A16R1 #3
SAMPLES RUN 11-19-94-WEC FOR CL6-8,SONIFIED,2UL INJ
CONDS.: -2290EMU 70EU IMA 30M-DB5 100-1H-325-10/-45SS GROB
RANGE: G I, 1 LABEL: N 0, 4.0 QUANi A 9, 1.8 J 0 BASEt U 20, 3
SCANS 1850 TO 1350
HexaCDFs
OctaCDFs
Figure 7. Hexa-, Hepta-, and OctaCDFs in sample 11-19-94-WEC.
473600.
373.800
± 8.508
489472.
407.888
i 0.580
147456.
443.808
t 8.588
1059
28:34
1180
21:33
1150
22:32
1208
23:31
1258
24:29
1 * i * *
1308
25:28
1358 SCAN
26i27 TIME
-------
These samples were also analyzed by full scan HRGC/MS. A large
amount of tetrachloroethylene was observed. Chlorinated benzenes were also
observed. A variety of nonchlorinated compounds were also identified, includ-
ing smaller polynuclear aromatics, hydrocarbons, phthalate esters, and alde-
hydes. A more detailed identification of these products will be reported at
a later date.
3. HTH
No appreciable amounts of PCDFs or PCDDs were found in the reaction
products of HTH (ester) or HTH (para) runs, listed in Tables 5 and 6. The
small amounts of PCDFs found in run 103 were apparently carryover from the
previous OCDF runs. This conclusion is supported by the fact that the isomer
patterns in the HTH and OCDF runs were similar.
These samples were also analyzed by full scan HRGC/MS. A variety
of C8-C20 hydrocarbons, smaller polynuclear aromatics and their alkyl adducts,
aliphatic alcohols, aldehydes and ketones were observed. A more detailed
identification of these products will be reported at a later date.
4. Octachlorodibenzofuran
Two runs, followed by a blank run, were conducted with OCDF in min-
eral oil at 500 ug/mL. As discussed in Section III.C.5, the objective of
these experiments was to assess the potential for formation of lower chlori-
nated homologs from OCDF. As noted in Table 5 and Figure 8, substantial
amounts of the lower PCDF homologs were formed. Only about 8% of the OCDF
feed passed through the reaction system unchanged. No PCDDs were detected
(Table 6). Preliminary, nonquantitative analyses indicated that small amounts
of mono- through triCDFs were also formed.
Figures 9 and 10 present extracted ion current profiles of the PCDFs
formed from the OCDF feed. The isomeric distribution of the PCDFs has some
similarities to those observed from combustion of tetrachloroethylene (Figures
6 and 7), PCBs, or trichlorobenzene (Figures 10-20 in Erickson et al. 1984).
Further work is needed to identify specific similarities and differences.
D. Quality Assurance and Quality Control
A variety of QA and QC activities were conducted throughout the
course of this work to assure that the results were acceptable. The project
records were inspected by MRI's Quality Assurance Unit and any deficiencies
were corrected. The data were spot-checked by the work assignment leader and
reviewed by the Quality Assurance Unit.
Various routine QC activities were conducted to provide data on the
data quality. All thermal combustion runs were conducted in duplicate (see
Tables 2, 5, and 6) to provide a measure of precision. Isotopically labeled
compounds were added to assess the overall accuracy. The recovery of 13C-TCDD
averaged 66 ± 23% for 16 samples, based on external standards. Similarly,
recoveries for 13C-TCDF averaged 69 ± 17% and 13C-OCDD averaged 45 ± 15% (14
samples). The differences in number of samples arose from interferences in
certain samples which prevented quantisation of one or more of the surrogates.
20
-------
30
20
10
c
o
c
o
U
Mono
Di
Tri Telra Penta
PCDF
Hexa
Hepfa
Octa
Figure 8. PCDF formatior. from OCDF-spiked feed, expressed as percent of OCDF fed
into system. Closed symbols are averages of two values; missing points
are no data.
-------
SCANS 1885 TO 1258
MID MASS CHROMATOGRAMS DATAi 8281A15R3 «1
01/15/85 11:16:88 CALI: MID388A89R1 tt2
SAMPLE: RUN 11-30-188-OCOF PCOO/F/CL4-5 SONICATED 2UL INJ
CONDS.: -2200EMU 78EU IMA 38M-DB5 188-1H-325-18/-45SS GROB
RANGE: G 1, 1 LABEL: N 8, 4.8 QUAN: A 8/ 1.8 J 8 BASEi U 29, 3
18.4n
386
ro
100.01
348
2363390.
385.988
i 8.589
12877888.
339.988
i 8.588
1288
20:83
1258 SCAN
28:53 TIME
Figure 9. Tetra- and PentaCDFs in sample 11-30-100-OCDF,
-------
MID MASS CHROMATOGRAMS DATA: 8281A17R2 ttl
91/17/85 8:54:68 CM.Il MID353A16R1 t3
SAMPLE: RUN 11-38-100-OCDF FOR CL6-8,SONIFIED,2UL INJ
CONOS.: -2209EMU 70EU IMA 30M-OB5 108-1H-325-18/-45SS GROB
RANGE: G 1, 1 LABEL: N 0, 4.9 QUAN: A 8, 1.8 J 8 BASE: U 28, 3
SCANS 1858 TO 1358
84.
374
160.0-
408
62.2n
444
1050
20:34
1100
21:33
1150
22:32
HeptaCDFs
OctaCDFs
1288
23:31
1258
24:29
A
1388
25:28
Figure 10. Hexa-, Hepta-, and OctaCDFs in sample 11-39-100-OCDF.
18371886.
373.880
± 8.588
12384388.
487.880
i 0.588
7651328.
443.888
i 8.588
1358 SCAN
26:27 TIME
-------
The values fbr the native PCDFs and PCDDs presented in this report have been
corrected for the surrogate recovery, as described in the Chemical Analysis
Section in the previous report (Erickson et al. 1984).
V. REFERENCES
Erickson et al. 1984. Thermal degradation products from dielectric fluids.
Interim report. Report EPA-560/5-84-009, NTIS PB85/38535. Washington, DC:
Office of Toxic Substances, U.S. Environmental Protection Agency. Contract
68-02-3938.
USEPA. 1984a. Polychlorinated biphenyls (PCBs); manufacture, processing,
distribution in commerce and use prohibitions; use in electrical transformers.
Advanced notice of proposed rulemaking. (49 FR 11070-11083).
USEPA. 1984b. Polychlorinated biphenyls (PCBs); manufacture, processing,
distribution in commerce and use prohibitions; use in electrical transformers.
Proposed rule. (49 FR 39966-39989).
Buser HR, Bosshardt H-P, Rappe C. 1978. Formation of polychlorinated di-
benzofurans (PCDFs) from the pyrolysis of PCBs. Chemosphere 7(1):109-119.
Buser HR, Rappe C. 1979. Formation of polychlorinated dibenzofurans (PCDFs)
from the pyrolysis of individual PCB isomers. Chemosphere 8(3):157-174.
24
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-560/5-85-022
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Products of Thermal Degradation of Dielectric Fluids
5. REPORT DATE
May 1985
6. PERFORMING ORGANIZATION CODE
8201A23
7. AUTHOR(S)
Stephen E. Swanson, Mitchell D. Erickson, Leslie Moody
8. PERFORMING ORGANIZATION REPORT NO.
Interim Report No. 2
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
10. PROGRAM ELEMENT NO.
Work Assignment No. 23
11. CONTRACT/GRANT NO.
EPA Contract No. 68-02-3938
12. SPONSORING AGENCY NAME AND ADDRESS
Field Studies Branch, TS 798
Office of Toxic Substances, U.S. EPA
401 M Street, S.W.
Washington, DC 20460
13. TYPE OF REPORT AND PERIOD COVERED
Interim (11/84 - 5/85)
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
The EPA Work Assignment Manager is Daniel T. Heggem, (202) 382-3990.
The EPA Project Officer is Joseph J. Breen, (202) 382-3569.
. ABSTRACT
, s report describes the results of a series of experiments which were con-
ducted to augment the data presented in a previous report (EPA 560-5-84-009). The work
presented here includes both repeat runs of those previously reported and also investi-
gations of materials not previously tested. The results of the repeat runs correlate
well with the previously reported results. The formation of PCDFs and PCDDs from the
PCB-spiked mineral oil, as well as the calculated destruction efficiencies for PCBs,
closely match the previously reported results. The repeat chlorobenzene runs show
higher levels of PCDF and PCDD formation than previously. However, several concentra-
tions in the previous report are "greater than" values; hence, comparison is difficult.
Significant amounts of PCDFs and PCDDs were formed from the tetrachloroethylene fluid.
The two high temperature hydrocarbon fluids did not produce PCDFs or PCDDs. It appears
that, under these thermal destruction conditions, the extent of PCDF formation varies
among dielectric fluids by several orders of magnitude. The highest PCDF formation was
noted for PCB-containing_dielectric fluids (about 0.01% of the amount fed), trichloro-
benzene had about 5 x 10~4 conversion efficiency, and the least formation was observed
in tetrachloroethylene (about 4 x 10 6). It should be noted that the thermal combus-
tion system was optimized for PCDF formation from PCBs and the optimal conversion con-
ditions for the other fluids may produce higher PCDF yields. The results for the OCDF-
spiked feed oil runs showed formation of a number of lower chlorinated PCDF homologs
with a variety of isomers within each homoloa. _
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
PCBs; PCDF; PCDD; OCDF; Polychlorinated
biphenyl; Polychlorinated dibenzofuran;
Polychlorinated dibenzo-jD-dioxin;
Tetrachloroethylene; Trichlorobenzene;
Dielectric fluid; Combustion; Pyrolysis;
Transformer fires; PCB fires
o. Dial ni BUTlOiN STATcMcNl
19. SECURITY CLASS (TntS Report)
UNCLASSIFIED
21. NO. OF PAGES
Unlimited
30
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
------- |