31...... .r.
.21. t. .9

.. vfil;
:3 .x

 

 

:
f ..I

55.31.

 

Ein:....fi.¢._%£. iii....... ‘ ,_ . ‘ _ . . . . _ . . 3.2;

 

RSITY Ll ISBRARIE

11111111111111111111111111111111 11 1011

3 1293 014103

111

 

 

 

 

 

 

 

This is to certify that the

dissertation entitled

THE QUANTITATIVE ANALYSIS OF POLYCHLORINATED BIPHENYIS AND
THEIR METABOLITFS BY GAS CHROMATOGRAPHY AND ION-MOLECULE
REACTION DETECTION IN A TANDEM MASS SPECTROMETER

presented by

Ronald Frankfin Lopshii'e'i

has been accepted towards fulfillment
of the requirements for

Ph. D. . Chemistry

degree in

 

 

flan/A? $4,.

Major professor

Date W [(9]. [i?(

MSU is an Affirmative Action/Equal Opportunity Institution O~12771

 

 

 

 

LIBRARY
Michigan State
Unlverslty

 

 

 

PLACE IN RETURN BOXto remove thIe checkout {mm your record.
To AVOID FINES return on or before due due.

__—______fi
DAWIDUE DATE DUE DATE DUE

 

 

 

 

  

M1192300

 

 

 

 

 

 

 

-— 11 -—1

 

 

If
poi

MSU Ie An Affirmative Action/E“ Opportunlty Inotltulon
Wm

 

 

 

THE QUANTITATIVE ANALYSIS OF POLYCHLORINATED
BIPHENYLS AND THEIR METABOLITES BY GAS
CHROMATOGRAPHY AND ION—MOLECULE REACTION
DETECTION IN A TANDEM MASS SPECTROMETER

By

Ronald Franklin Lopshire

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCPOR OF PHILOSOPHY

Department of Chemistry

1994

ABSTRACT

THE QUANTITATIVE ANALYSIS OF POLYCHLORINATED
BIPHENYLS AND THEIR METABOLITES BY GAS
CHROMATOGRAPHY AND ION-MOLECULE REACTION
DETECTION IN A TANDEM MASS SPECTROMETER

By

Ronald Franklin Lopshire

An analytical method for the quantitative analysis of polychlorinated biphenyls
(PCBs) using gas chromatography and mass spectrometry was developed. The
method uses the exchange reaction of oxygen for chlorine with the molecular anions
of PCBs in an MS/MS experiment with a triple stage quadrupole mass spectrometer.
The instrumental parameters explored for instrumental optimization for this
reaction include: type and pressure of moderating gas for the production of
molecular anions; type and pressure of collision (reactant) gas; and collision energy
and translational kinetic energy correction factor. Instrument scan modes are
optimized for application of the method to sediment samples containing PCBs. The
method focusses on analysis of coplanar PCB congeners that have been shown to
exhibit dioxin-like toxicity. The specificity of the method is explored with respect to
congeners that oo-elute with the coplanar congeners in a GC experiment. The
method is expanded to identify other chlorinated aromatic compounds in sediment
samples containing significant amounts of PCBs.

ACKNOWLEDGMENTS

I am deeply indebted to my guidance committee — Professors Chris Enke, Jack
Watson, Jack Holland, Matt Zabik and Kim Dunbar. The knowledge gained during
my years in graduate school goes far beyond academic pursuits. In spite of my bull-
headed attitudes at times, their support has allowed me to grow as a person both
professionally and personally. The skills that I developed as an analytical chemist
and mass spectroscopist have been and will be invaluable professionally. I
particularly want to express my gratitude to Chris for his patience, understanding
and guidance.

I would like to thank my family for their support in this work. My parents, Mary,
Wayne and Frank, have always been there for me and my family. We love you and
will never be able to repay you for your love and support. I would hope that my
daughters, Michelle and Chris, know how much I appreciate the sacrifices they have
made during this time. It is tough enough growing up without having your father
”spazzed out" half the time due to the pressures of graduate school. Of course, none
of this work would have even been attempted, let alone accomplished, without the
love and encouragement of Lorna. She should enjoy the fruits of this labor as much

as 1, due to the extreme sacrifices she has made during this endeavor.

Iwould like to thank the associates in all of my research groups for their continuous
help and support— Brian Eckenrode, Norm Penix, Sherri Adams, Mark Cole, Jon
Wahl, Tim Heath, Steve Chan, Kate Noon, Kevin M‘Nitt, Mel Micke, Mike

Davenport, Doug Gage and many others. I will always cherish their fiiendship and

help.

I would like to thank the administration of Michigan State University and the
College of Natural Science for making me really appreciate my undergraduate
education. The most disappointing aspect of my graduate career was the way I was
treated as a graduate student by the University. Kate Noon said it best, ”As a
graduate student, you lose your status as an adult.” This attitude filters down to all
departments, and as such, these most adversely affected are the undergraduate
students. It is absolutely appalling the way these young adults are treated by the
University, faculty and teaching assistants. I feel fortunate that I received my
undergraduate education at a school with no graduate program where higher

learning is given more than just lip service.

Lastly, I would like to thank the faculty of the Chemistry Department at Purdue
University in Fort Wayne (IPFW). Their support and guidance provided me with the
skills and incentive to pursue my goals in this field. Iwould like to dedicate this work
to Professor Ken Stevenson at IPFW, Chairman of the Department of Chemistry.
Without Ken’s guidance, support and understanding over the years, I would still be

nothing more than a laid-off autoworker.

TABLE OF CONTENTS

LIST OF TABLES ............................................... viii
LIST OF FIGURES ............................................... x
Chapter 1 ....................................................... 1
Introduction ............................................... 1
Chapter 2 ....................................................... 5
Introduction to Polychlorinated Biphenyls ....................... 5
Synthesis and Use ..................................... 5
Environmental Ramifications ........................... 6

Biological Ramifications ............................... 7

Analytical Significance ................................. 9
Nomenclature and Composition ......................... 9

Chapter 3 ....................................................... 16
Analytical Methodology for Polychlorinated Biphenyls ............. 16
Introduction ......................................... 16

Sample Preparation ................................... l7

Extraction ..................................... 17

Clean-up ...................................... 20
Non-chromatographic Methods .......................... 22
Chromatographic Methods ............................. 23

Gas Chromatography ............................ 23

High Performance Liquid Chromatography .......... 25

Hyphenated Methods ............................ 25

Other Chromatographic Techniques ................ 26

Mass Spectrometric Detection ........................... 26

Electron Impact Ionization ....................... 28

Positive Chemical Ionization ...................... 29

Electron Capture Negative Ionization ............... 29

Tandem Mass Spectrometry ....................... 30

Data Reduction ...................................... 32

Summary ............................................ 35

Chapter 4 ....................................................... 36
Exchange of Oxygen for Chlorine in Molecular Anions of

Chlorinated Aromatics in a Tande- Mass Spectrometer ..... 36
Experimental ........................................ 36
Gas Chromatography ............................ 36

Mass Spectrometry .............................. 37
Generation of the M' Ion .............................. 37
Positive Chemical Ionization ...................... 37
Electron-Capture Negative Ionization .............. 40

Ion Source Optimization for ECN I ................. 45
Oxygen-Chlorine Exchange ............................. 54
Odd- vs. Even-Electron Anions .................... 54
Reaction Scheme ............................... 56
MS/MS Reaction Parameters ........................... 58
Voltage Programming ............................ 58
Translational Kinetic Energy Correction ............. 60
Collision Reagents and Pressures .................. 64
Collision Energy ................................ 67
Collision RF Linkage ............................ 71
Summary of Experimental Details ....................... 73
Chapter 5 ....................................................... 76

Development of an Analytical Method for the Analysis of
Polychlorinated Biphenyls using the Exchange Reaction of
Oxygen for Chlorine with Molecular Anions of Chlorinated

Aromatics in a Tandem Mass Spectrometer ................ 76
Introduction ......................................... 76
Experimental ........................................ 76
Gas Chromatography ............................ 77

Mass Spectrometry .............................. 77

Gas Chromatography .................................. 77
Mass Spectrometry .................................... 79
Instrumental Tune .............................. 79

Signal Optimization ............................. 81

Full Scan Acquisition ............................ 83
Selected Reaction Monitoring ..................... 94
Quantitation ......................................... 106
Relative Response Factors ........................ 106
Internal Standards .............................. 106
Chlorinated Analogs ....................... 108

Isotope Dilution .......................... 109

Summary ............................................ 109

Chapter 6 ....................................................... l 12
Application of the MS/MS Method using the Oxygen-Chlorine
Exchange Reaction for the Analysis of Polychlorinated

Biphenyls in River Sediment Samples ..................... 112
Introduction ......................................... 112
Experimental ........................................ 113
Gas Chromatography ............................ 113

Mass Spectrometry .............................. 114
Quantitation ......................................... 115
Precision ...................................... 118
Interferences ................................... 119
Accuracy ...................................... 120
Results ............................................. 120
Chapter 7 ....................................................... 124

Application of the MS/MS Method using the Oxygen-Chlorine
Exchange Reaction to the Analysis for Polychlorinated
Dibenzodioxins and Dibenzofurans in Samples containing

Polychlorinated Biphenyls .............................. 124

Introduction ......................................... 124
Experimental ........................................ 126

Gas Chromatography ............................ 126

Mass Spectrometry .............................. 126

Isotopic Distribution .................................. 127

Molecular Anions ..................................... 127

Dioxins and Furans ................................... 140

Conclusions .......................................... 142

Chapter 8 ....................................................... 146
Conclusions ............................................... 146
Instrumental Parameters ............................... 146

Other Chlorinated Aromatics ........................... 148

Future Studies ....................................... 148

LIST OF REFERENCES .......................................... 150

LIST OF TABLES

Table 2.1. The number of possible congeners for each homolog. .......... 12
Table 2.2. Average Aroclor composition by homolog (percent by weight). . . 13

Table 2.3. Twelve PCB congeners that have been shown to exhibit dioxin-
like toxicity. ............................................... 15

Table 5.1. The structure of the twelve PCB congeners that have been
shown to exhibit dioxin-like toxicity and those congeners that co-
elute. Relative retention times (RRT) are with respect to
octachloronaphthalene (77). .................................. 90

Table 5.2. The response factors of several PCB congeners with respect to
octachloronaphthalene for a GC/MS/MS experiment where the
products of the reaction of molecular ions with oxygen are
monitored. ................................................ 107

Table 5.3. The responses of the twelve coplanar PCB congeners with
respect to the three 13C-Iabelled standards. ...................... 110

Table 6.1. The results of the analysis of an Aroclor 1242 standard for the
twelve coplanar congeners compared to values reported in
literature. ................................................. 121

Table 6.2. The results of the quantitation by GC/MS/MS of two Aroclor
1242 samples, one of which was autoclaved and the other
inoculated and incubated under anaerobic conditions. ............. 123

Table 7.1. The molecular anions [M+n]’ observed in the primary mass
spectra of polychlorinated aromatics designated as n,x where 1:

indicates the chlorine number. ................................ 133
Table 7.1 (continued). ............................................. 134
Table 7. 1 (continued). ............................................. 135
Table 7.1 (continued). ............................................. 136

Table 7.1 (continued). ............................................. 137
Table 7.1 (continued). ............................................. 138

Table 7.1 (continued). ............................................. 139

LIST OF FIGURES

Figure 2.1. The biphenyl structure indicating numbering convention. ...... 10

Figure 3.1. A flow chart (method of Webb and M‘Call) for the
quantitation of samples with mixed Aroclor patterns. .............. 33

Figure 4.1. The ECNI mass spectrum of a tetrachlorobiphenyl. The three
regions of interest are the molecular ion region, the oxygen-for-
chlorine exchange reaction region and the loss of chlorine from the
molecular ion region. ........................................ 44

Figure 4.2. The ECNI mass spectrum of a tetrachlorobiphenyl indicating
the [M]' ' radical ions. ....................................... 46

Figure 4.3. The ECNI mass spectrum of a pentachlorobiphenyl indicating
the [M]' ' radical ions ....................................... 47

Figure 4.4. The ECNI mass spectrum of a hexachlorobiphenyl indicating
the [M]’ ° radical ions. ....................................... 48

Figure 4.5. The ECN I mass spectrum of a heptachlorobiphenyl indicating
the [M]’ ’ radical ions. ....................................... 49

Figure 4.6. The ion current of the molecular anion of a
pentachlorobiphenyl as the pressure of the CI moderating gas is
varied from high to low ....................................... 51

Figure 4.7. A plot of electron energy versus ion current of the molecular
anion of a pentachlorobiphenyl. ............................... 53

Figure 4.8. The offset voltage settings for 01, OZ, and Q3 for three
settings of MSMSC. ......................................... 62

Figure 4.9. The plots of the DOFF voltage as a function of the ratio of the
daughter mass to the parent mass. ............................. 63

Figure 4.10. The ion current of the product of the oxygen-for-chlorine
exchange reaction as a function of the translational kinetic energy

correction factor (MSMSC). .................................

Figure 4.11. Ion current of the product of the oxygen-for-chlorine

exchange reaction as a function of the pressure of the collision gas. . . .

Figure 4.12. The ion current of the oxygen-for-chlorine exchange reaction

product of a tetrachlorobiphenyl as a function of collision energy. . . .

Figure 4.13. The ion current of the oxygen-for-chlorine exchange reaction

product of a pentachlorobiphenyl as a function of collision energy. . . .

Figure 4.14. The ion current of the oxygen-for-chlorine exchange reaction

product of a hexachlorobiphenyl as a function of collision energy. . . .

Figure 4.15. The transmission of the molecular anion of a
pentachlorobiphenyl as a function of the collision energy (02

voltage offset) with no collision gas being introduced ..............

Figure 5.1. The relationship of an ion and its isotope indicating the
significance of using the exact mass for a set mass in an MS/MS
experiment and the effect on selectivity of under-resolving the

instrument. ...............................................

Figure 5.2. The ECNI mass spectrum of a tetrachlorobiphenyl indicating

the loss of 35u in the ion source. ..............................

Figure 5.3. The ECNI mass spectrum of a pentachlorobiphenyl indicating

the loss of 3411 in the ion source. ..............................

Figure 5.4. The ECNI mass spectrum of a hexachlorobiphenyl indicating

thelossof34uintheionsource. ..............................

Figure 5.5. The ECNI mass spectrum of a heptachlorobiphenyl indicating

the loss of34u in the ion source. ..............................

Figure 5.6. The ECNI mass spectrum of a pentachlorobiphenyl which co-
elutes with a coplanar congener in a GC analysis. This congener
exhibits the loss of 35u in the source as opposed to the loss of 34u

observed with the coplanar congeners. .........................

.65

66

. 68

69

70

. 72

82

86

87

88

89

.91

Figure 5.7. The ECNI mass spectrum of a hexachlorobiphenyl which co-
elutes with a coplanar congener in a GC analysis. This congener
exhibits the loss of 35u in the source as opposed to the loss of 3411
observed with the coplanar congeners. .......................... 92

Figure 5.8. The ECNI mass spectrum of a heptachlorobiphenyl which co-
elutes with a coplanar congener in a GC analysis. This congener
exhibits the loss of 34u in the source, and as such, interferes in a
GC/MS/MS analysis of coplanar congeners ....................... 93

Figure 5.9. Mass chromatograms at m/z 273, 307, 341 and 375 and a
reconstructed total ion chromatogram (RIC) that represent the
analysis by GC/MS/MS of four PCB congeners that co—elute with
coplanar congeners. ......................................... 95

Figure 5.10. The mass and reconstructed total ion current (RIC)
chromatograms from a GC/MS/MS experiment where the exchange
reaction of oxygen for chlorine was monitored for
trichlorobiphenyls in an Aroclor 1242 standard. .................. 96

Figure 5.11. The mass and reconstructed total ion current (RIC)
chromatograms from a GC/MS/MS experiment where the exchange
reaction of oxygen for chlorine was monitored for
tetrachlorobiphenyls in an Aroclor 1242 standard. ................ 97

Figure 5.12. The mass and reconstructed total ion current (RIC)
chromatograms from a GC/MS/MS experiment where the exchange
reaction of oxygen for chlorine was monitored for
pentachlorobiphenyls in an Aroclor 1242 standard. ................ 98

Figure 5.13. The mass and reconstructed total ion current (RIC)
chromatograms from a GC/MS/MS experiment where the exchange
reaction of oxygen for chlorine was monitored for
hexachlorobiphenyls in an Aroclor 1242 standard. ................ 99

Figure 5.14. The mass and reconstructed total ion current (RIC)
chromatograms from a GC/MS/MS experiment where the exchange
reaction of oxygen for chlorine was monitored for
heptachlorobiphenyls in an Aroclor 1242 standard. ............... 100

Figure 5.15. The mass and reconstructed total ion current (RIC)
chromatograms from a GC/MS/MS experiment where the exchange
reaction of oxygen for chlorine was monitored for
octachlorobiphenyls in an Aroclor 1242 standard. ................. 101

xii

Figure 5.16. The mass and reconstructed total ion current (RIC)
chromatograms from a GC/MS/MS experiment where the exchange
reaction of oxygen for chlorine was monitored for
nanochlorobiphenyls in an Aroclor 1242 standard. ................ 102

Figure 5.17. The reconstructed total ion current (RIC) chromatogram of
a GC/MS/MS experiment using product and precursor scanning of
the instrument for the same reaction indicating the product scan
mode to be the more sensitive. ................................ 104

Figure 6.1. The plot of the ion current, standard deviation and relative
standard deviation as the conversion dynode voltage was varied
from 3 to 20 kV. ............................................ 116

Figure 6.2. The plot of the ion current, standard deviation and relative
standard deviation as the electron multiplier voltage was varied
from 500 to 2500 volts. ....................................... 117

Figure 7. 1. The molecular anion clusters of polychlorinated biphenyls
calculated using the natural abundances of carbon, chlorine and
hydrogen. ................................................. 128

Figure 7.2. The structures and numbering conventions for six compounds
whose chlorinated analogs are of environmental significance. ....... 129

Figure 7.3. Generated spectra from isotope calculations of the molecular
ion region of six tetrachlorinated aromatics. ..................... 130

Figure 7.4. The molecular anion clusters of six tetrachloro- aromaties
calculated using the natural abundances of carbon, chlorine and
hydrogen. ................................................. 131

Figure 7.5. The reconstructed total ion current (RIC) chromatogram
generated by monitoring the oxygen-for-chlorine exchange reaction
of the molecular anions [M+2]- of PCDDs in a GC/MS/MS
experiment. ............................................... 141

Figure 7.6. The reconstructed total ion current (RIC) chromatogram
generated by monitoring the oxygen-for-chlorine exchange reaction
of the molecular anions [M+2]- of tetrachlorodibenzofurans in a
GC/MS/MS experiment. ..................................... 143

Figure 7.7. The reconstructed total ion current (RIC) chromatogram
generated by monitoring the oxygen-for-chlorine exchange reaction
of the molecular anions [M+2]- of tetrachlorodibenzofurans in a

GC/MS/MS experiment. ..................................... 144

Chapter 1

Introduction

The detection and quantitation of polychlorinated biphenyls (PCBs) in a wide range
of samples and matrices poses a unique and challenging analytical problem. The
purpose of the present work has been to use tandem mass spectrometry (MS/MS)
to develop a unique but robust and routine method for the detection and
quantitation of individual PCB congeners in PCB-contaminated samples. In
addition, the methods developed have been extended to include the detection of

PCB metabolites in samples with a significant PCB contribution.

As a preface to the discussion of the analytical methodology involved with this work,
a brief discussion of the h'story of PCBs as an environmental and clinical problem
is included. This discussion also contains the terminology associated with the
chemistry and analysis of PCBs.

The current work involves the development of an MS/MS method for the detection
and quantitation of individual PCB congeners. It is, therefore, desirable to include
a brief synopsis of the limitations of the analytical methods currently used for

l

2
routinePCBanalyses. These methods include gaschromatography, electron capture

detection and mass spectrometry, and have generally been used not for congener-
specific analyses, but for the determination of total PCBs or for the determination

of a specific commercial mixture, such as an Aroclor.

The method of analysis for individual PCB congeners presented in thiswork involves
the detection of products of an ion-molecule reaction carried out in the collision
chamber of a triple-quadrupole mass spectrometer. The molecular anions of PCB
congeners are allowed to react with molecular oxygen to produce an exchange of an
oxygen atom for chlorine, and the products of the reactions are monitored. The
parameters involved in these reactions include the kinetic energy and collision

pressure.

The exchange reaction of oxygen for chlorine with the molecular anions of PCB
congenersismonitoredinconjunction with chromatographicseparation asamethod
of analysis for individual PCB congeners. The development of the method includes
the optimization of several instrumental parameters involved in the chromatography
and mass spectrometry. The precision and accuracy of the method are also
compared to methods commonly used. The precision obtained is less than ten
percent which is comparable for most congeners to MS techniques and of an order
acceptable for methods such as GC/ECD. The accuracy of the method was
determined by comparison with published results of analysis of Aroclor standards.

Even though the composition of Aroclor mixtures varies widely, the results were of

3

the same order of magnitude for the congeners compared.

While the development of the MS/MS method for PCB analysis has merit on its own,
the method is applied to the analysis of real samples. Specific applications include
using the method for the analysis of PCB-contaminated soil and sediment samples.
The method is applied to the analysis of samples used in the study of microbial

action on individual PCB congeners.

An analytical method was also developed using the exchange reaction of oxygen for
chlorine to include the analysis of PCB metabolites and other chlorinated aromatics
in samples with significant amounts of PCBs. Specific metabolites include
polychlorinated dibenzodioxins (PCDDs) and polychlorinated dibenzofurans
(PCDFs).

A summary of the limitations and general observations of the method of analysis
presented is offered. The primary advantage of the method is due to the high
selectivity provided by the MS/MS technique. Other methods such as GC/ECD and
GC/MS do not provide for the quantitation of lower mass congeners in the presence
of higher mass congeners that chromatographically co-elute. The observations
presented include a discussion of the critical parameters involved in the
development of the method, the application of the method to specific samples and
possible interferences observed with the method. The primary interference with the

method occurs in some cases when congeners co-eluting with those of interest

4

exhibit rearrangement in the ion source to produce precursor ions of the same
chemical composition. Future considerations offered include the further exploration

of all possible interferences with the method.

Chapter 2

Introduction to Polychlorinated Biphenyls

Synthesis and Use. The synthesis of polychlorinated biphenyls (PCBs) was first
reported in 1881 by Schmidt and Schultz (1), but industrial production of the
unnatural complex mixtures did not begin until 1929. Due to their unique physical
and chemical properties, PCBs were soon in widespread use as nonflammable
lubricants, plasticizers, flame retardants, insulators, pesticide extenders, dust
suppressors, vapor suppressors and high-pressure sealants. These applications
found use in transformers, capacitors, pesticides, printing inks, paints, de-dusting
agents and several other products. PCBs were manufactured and marketed under
several trade names including Aroclor, Clophen, Phenoclor, Fenclor, Kaneclor and

Phyralene.

Commercial production of PCBs is accomplished by chlorination of biphenyl with
chlorine gas. By controlling reaction conditions and using catalysts such as ferric
chloride, mixtures of PCB congeners are produced with the desired overall chlorine
content (2). In theory, 209 chlorinated biphenyl isomers are possible, but in
practice only a little over one hundred are probable. The properties of these

5

6
mixtures, as expected, are dependent on the degree of chlorination. Boiling points
range from 278 C for the 21% chlorinated product to over 420 C for the 60%
chlorinated product. These substances can be distilled at atmospheric pressure
without decomposition and are soluble in hydrocarbons but only very slightly soluble
in aqueous media. In addition, PCBs are also produced during incineration of

chlorinated aromatics and as by-products in several commercial chemical processes.

Monsanto began production of Aroclors in 1930, and before ceasing production in
1977, became the world’s largest producer of PCBs. It has been estimated that total
U. S. production of PCBs through 1976 was 1.3 billion pounds, 93% of which was
produced by Monsanto. Worldwide commercial production through 1980 is
estimated to be 2.4 billion pounds. Annual production of PCBs as by-products is
estimated to be 100,000 pounds (3).

Environmental Illuminations. A unique aspect of the PCB problem is that unlike
other environmental contaminants such as DDT, the commercial applications of
PCBs did not generally include distribution in the environment (4). And yet
today PCBs are ubiquitous contaminants and have been found all over the globe.
Three significant events occurred which led to worldwide concern. In 1966, the
Swedish scientist Jensen drew attention to the fact that PCBs were found in fish and
birds (5). In 1968, an accident in western Japan poisoned 1300 people. The
so-called "Yusho" poisoning occurred when a heat exchanger leaked PCBs into a

batch of rice oil (6). This accident was the precursor of several similar but

7
mailer events in this country. Then in 1975, Renate Kimbrough of the Centers for

Disease Control found that high dosage feedings of Aroclor 1260 (a commercial
mixture of highly chlorinated PCBs) caused liver cancer in rats (7). These events
spurred a tremendous amount of PCB research which led to their eventual
regulation. Since these events, however, studies have shown that the clinical
ramifications of the Yusho poisoning were actually due to the polychlorinated
dibenzofuran (PCDF) content of the PCB-containing oil.

Surprisingly, or not, as early as 1936, occupational exposure was reported to cause
toxic effects and limits were set. But itwas not until the Yusho incident that the US.
FDA conducted surveys to determine the extent and levels of PCB contamination
which might find ways into the food chain. In the early 1970s, it was determined that
67% of food packaging and 19% of the food therein contained PCBs —- 75% of
infant food cereals were found to be contaminated (8). In 1973, temporary
tolerances of PCB levels were established for several classes of food. Under the
Toxic Subdance Control Act of 1976, the EPA was granted authority to enforce
their continued use. As of November 1, 1979, PCBs could no longer be used in new

systems, and all equipment containing PCBs was to be phased out by July 1, 1984
(9).

Biological Ramificetlons. As a biological hazard, chronic exposure tends to be more
serious than acute exposure. The toxicity of individual congeners decreases with

increasing chlorine substitution at the ortho position (10),(11),(12). The

8
toxicity of PCB congeners is measured as dioxin-like toxicity, and only those
congeners with no or one ortho chlorine and at least tetrachloro- substitution are
particularly toxic. PCBs are soluble in human adipose tissue, leach into most
biological fluids, particularly blood and milk, and tend to bio-accumulate.
Symptoms of exposure include chloracne, fatigue, nausea, vomiting, mild jaundice,
abdominal pain, loss of hair, loss of libido, numbness of extremities, headaches and
darkening of the skin. Long range effects include neural disturbances such as
memory loss, impaired circulation, impaired digestion, bone and joint deformities,
morphological changes in adult teeth and poor teeth development in children. Of
particular importance is the fact that many symptoms take years to manifest

themselves after exposure.

Since the 1975 Kimbrough study showed that high levels of Aroclor 1260 caused
liver cancer in rats, several other investigations have supported these findings. It
does appear, however, that only highly chlorinated mixtures produce these efiects,
whereas mixtures such as Aroclor 1242 do not. Three significant conclusions have
resulted from these studies. One, the PCB-treated rats, including those with liver
tumors, lived significantly longer than the untreated rats. TWO, the PCB-treated rats
had significantly fewer cancers of all types. And three, the liver tumors, although

officially classified as cancerous at that time, did not metastasize to other organs.

Several studies have been made on humans, specifically General Electric Company

workers in New York State, exposed to PCBs and PCB-containing substrates. These

9
epidemiological studies included those done by the National Institute for

Occupational Safety and Health (NIOSH) and David Wegrnan of the University of
Lowell. In all cases, no association of PCB exposure and human cancer was found.
Dr. Kimbrough, now with the Environmental Protection Agency, has reviewed these
studies and has stated "No conclusive evidence thus far reported shows that

occupational exposure to PCBs causes increased incidence of cancer.” (7)

Analytical Significance. From an analytical standpoint, the analysis of PCBs has
undergone several changes. In the US, most analyses in the past have classified
PCBs as total PCBs or as individual or mixtures of Aroclors since Monsanto had
been the major producer. Even so, procedures including extraction, clean-up,
separation, identification and quantitation have been difficult at best. Now with the
interest in individual PCB congeners, the analytical problem is formidable indeed.
The future trend will no doubt be in the development of effective and efficient
methods for the identification and quantitation of PCB congeners in a wide variety

of environmental and biological matrices.

Nomenclature and Composition. Polychlorinated biphenyl isomers (congeners) are
named with the numbers of the chlorine substitution (2-6 and 2’-6’) about the
biphenyl structure with the two carbons at the linkage being number 1 and 1 ’ —for
example, 3,3 ’,4,4 ’-tetrachlorobiphenyl (Figure 2. 1). Since free rotation is possible
about the biphenyl linkage, the 2-position is identical to the 6-position for a mono-

substituted ring. Numbering is always done to achieve the lowest possible numbers

10

 

Figure 2.1. The biphenyl structure indicating numbering convention.

11

with the primed numbers reported second. In addition, each congener has been
assigned a number from 1 to 209 (13). This numbering convention has been
adopted by IUPAC such that IUPAC #77 is a valid designation for the above-
mentioned tetrachlorobiphenyl. The logic for the numbering of all 209 congeners
was originally followed for all but two of the 209 congeners (14). This has since
been corrected for classification by IUPAC. Nonetheless, early work in the

literature often refers to IUPAC #200 as #201 and #201 as #200.

All PCB isomers with the same number of chlorines are classified as belonging to the
same homologous series (Table 2.1). This is not strictly a correct use of the term
homologous, since a homologous series generally refers to a class of compounds
differing only by some structural unit such as -CH2 in CH3(CH2),,CH3 for a series of
alkanes. Thus, each PCB isomer (or congener) belongs to one of eleven (including
biphenyl itself) homologous series. A homolog, therefore, of a pentachlorobiphenyl

is any other pentachlorobiphenyl.

The vast majority of PCBs found in the environment in the United States are from
commercial mixtures produced by Monsanto and marketed under the trade name
Aroclor. These mixtures were designated as Aroclor 12pp where the 12 is indicative
of the two phenyl rings (12 carbons) and the pp is indicative of the percent chlorine
by weight in the total mixture such as Aroclor 1242 being 42% by weight chlorine
(Table 2.2) (15). Aroclor 1016 is composed of a mixture of PCB congeners with

16% by weight chlorine, but it is a distillation product of another Aroclor and not

12
Table 2.1. The number of possible congeners for each homolog.

Composition of PCB Homologs

 

Empirical Molecular Percent Number of
Formula Weight Chlorine Congeners

 

can“, 154.1 0 1

CuII,Cl 188.0 19 3
(3,211.0, 222.0 32

CnH7Cl, 256.0 41

camel, 289.9 49

curacy 323.9 54
Cal-1,0, 357.8 59

CuH3CI, 391.8 63
camel, 425.8 66
CuHCl, 459.7 69

c,,c1lo 493.7

 

13
'Tabk325L AnnnageAwnxflorconqxxdfionlnflhonuflog(juneentbvafighfiy

Anuxior(knnposnhnrbyIkneenthehflu:

 

Aucekn'

 

1221 1232 1016 1242 1248 1254 1260

 

 

10

50

 

14

from chlorination of biphenyl. Thus, the 10 in Aroclor 1016 is not indicative of ten

carbons.

A major focus of the current work is with the toxicity of sediment samples containing
PCBs. The twelve (12) PCB congeners that have been classified as exhibiting dioxin-
like toxicity (14) are listed in Table 2.3. All of these congeners contain either one or
no chlorine ortho to the biphenyl linkage. This allows for these congeners to form
a conformation where the two aromatic rings are coplanar. Also, all of these

congeners contain at least four (4) chlorines.

15

Table 2.3. Twelve PCB congeners that have been shown to exhibit dioxin-like
toxicity.

IUPAC NUMBER ORTHO CHLORINES CONGENER STRUCTURE

345,4-CB

 

34,34-CB

 

345,24-CB

 

245,34-CB

 

2345,4-CB

 

234,34-CB

 

345,34-CB

 

245,345-CB

 

2345,34-CB

 

234,345-CB

 

345,345-CB

 

2345,345-CB

 

 

 

Chapter 3

Analytical Methodology for Polychlorinated Biphenyls

Introduction. Most of the currently employed methods of analysis for PCBs (and
other chlorinated aromatics) involve extensive sample preparation followed by
chromatography with one of a wide variety of detectors. The amount and type of
sample clean-up is generally dependent on the specificity of the detection scheme
used and on the need to protect the cleanliness of the chromatographic apparatus.
The various types of detectors used offer a wide range of detection limits and

dynamic working ranges, and no one method can be justified for all applications.

This chapter offers a brief literature review of many of the more commonly used
methods of analysis for PCBs. PCB analyses may be categorized as one of three
types —— total PCBs, Aroclor composition or individual PCB congeners. With the
current emphasis on classifying PCB samples with respect to toxicity, the need for
congener specific analytical methods is becoming more important. It is this need to
which this research is focussed. As such, this discussion of the analytical
methodology currently in use for PCBs places emphasis on the inadequacies of these
methods with respect to congener-specific analysis. In addition, the specificity of

16

17

each method is examined for the required sample preparation.

Sample Preparation. The extent of the preparation of PCB-containing samples is
defined by the specificity of the method(s) being used for analysis and by the type of
sample. Typically, samples are subjected to extraction with non-polar solvents and
extensive clean-up involving separation with adsorbent and/or conventional
chromatographic columns. Often, further contaminant removal is done with

chemical degradation and precipitation.

Extraction. The extraction of PCBs from solid samples such as soils, sediments and
sludges is usually done with a non-polar solvent such as hexane. The most commonly
employed extraction method involves the use of a Soxhlet apparatus with continual
extraction and distillation for several hours up to overnight. For sludge samples, due
to their high organic content, several extraction procedures have been evaluated
(16) with respect to sample recovery. For Soxhlet extraction, the sludge sample
is often mixed with sodium sulfate and extracted with a mixture of dichloromethane
and hexane. The efficiency of this method varies greatly with respect to the particle
size of the sample due to solvent-sample contact. The best recoveries (> 80%) have
been reported using centrifuging of an emulsified mixture of the sample with
acetone/dichloromethane/hexane (3 ). In all cases, repetition of the extraction

process improves recovery.

With sediment and soil samples, the lower organic content generally allows for

18

‘plified extraction and clean-up procedures. A most important step involves
pulverization and thorough mixing of the sample. Soxhlet extraction works
exceptionally well for these samples using a variety of solvent mixtures including
acetone, dichloromethane and hexane (3, 17-20). In almost all cases, recoveries are
improved if the sample is first wetted with an agent such as water, acetone or
acetonitrile or mixtures of same. Paper samples are similarly extracted, but with
KOH saponification as adopted by the Association of Official Analytical Chemists
(AOAC) (21 ),(22).

In recent years, the use of supercritical fluid chromatography (SFC) and
supercritical fluid extraction (SFE) have been applied to analysis and sample clean-
up for many environmental methods (2.7),(24). The main advantage of SFE
for sample clean-up is that the method is very amenable to automation and typically
does not require the use of large volumes of organic solvents. Much of the early
work with SFE for PCB samples was devoted to those samples containing relatively
high concentrations of total PCBs (25-30). More recently, results have been
published for the use of SFE for low nanogram per gram levels of PCBs in sediment
samples (31 -33). An important consideration with SFE is the choice of trapping
systems. These include liquid, cryogenic and solid-phase systems. Much of the
advantage of mag SFE for PCB samples is often lost with liquid or cryogenic
systems, however, since extensive clean-up is still required. With solid-phase
trapping, extraction, clean-up and concentration for PCB samples maybe performed

simultaneously (31). An SFE method for PCB-containing sediment samples has also

19
been developed that includes sulfur removal using elemental copper (34). This

allows for the elimination of an additional clean-up step in sample preparation, and

also eliminates the need for using elemental mercury which has other environmental

ramifications.

With aqueous samples, classical liquid-liquid extraction with a non-polar solvent
(hexane, dichloromethane) is commonly employed (35). For many samples,
prior to extraction, adjusting the pH of the samples offers no advantages. For paper
mill effluents, extraction followed by KOH saponification provided for best
recoveries (36),(37). Other extraction techniques for aqueous samples
include continuous liquid-liquid extraction (CLE) (38),(39), sorbent column
extraction (40-43) and purge-and-trap techniques (44). In addition, the sorbent
column extraction technique is commonly employed in the field as a sampling
technique.

In the interest of completeness, extraction techniques for air and biological samples
are also presented. Air samples are usually acquired with a sorbent material such as
polyurethane foam, Florisil or XAD-2 (3) . The sorbent matrix is then extracted with
a non-polar solvent system. Blood samples are typically extracted with liquid-liquid
partitioning according to the protocols of the US. EPA (45) or the US. Centers
for Disease Control (46),(47). Adipose tissue and food substances may be
extracted with petroleum ether and then the PCBs extracted from the petroleum

ether with acetonitrile (48). This method may also be used for fish samples.

20
Milk, both bovine and human, is similar in fat content to adipose tissue and may be

extracted similarly as outlined by AOAC (49). Other biological samples
routinely extracted for analysis for PCBs include eggs (50), bird tissue (51),

earthworms (52) and plant tissue (53).

The determination of PCB content in oils has been of analytical significance because
the major commercial use of PCBs was as transformer and related oils. Also, the
Yusho incident (see Chapter 2) prompted interest in oils used in animal and human
food sources. For most oil samples, simple dilution with hexane or other non-polar
solvent systems is the most common technique as outlined by EPA methods

(54).

Regardless of the sample under consideration or the analytical methods of detection
being employed, extraction or dilution with or into a non-polar solvent system is
necessary. Even with the MS/MS method involved in this research, placing a raw
sample directly into the source could conceivably interfere with maintaining a

constant and reliable flux of thermal electrons in the ECNI plasma.

Clean-up. After extraction or dilution of the sample, the amount and extent of
sample clean-up is dependent on the specificity of the analytical method, the
detection scheme employed and on the type of analysis — total PCBs, Aroclor
composition or identification and quantitation of individual congeners. These clean-

up steps often include back-partitioning with solvents, column elution and chemical

21

treatment. Of course, the extent of this clean-up is also dependent on the matrix

effects inherent in the type of sample being analyzed.

Back-partitioning of many samples is employed as a clean-up and extraction
procedure. The most common procedure is with acetonitrile partitioning. In this
classic procedure (45), the PCBs are extracted from the non-polar solvent system
with acetonitrile. After dilution with water, the PCBs are partitioned back into the
non-polar solvent by aqueous dilution of the acetonitrile extract. This method allows
for the removal of many of the polar components from the extracted sample. Other

partitioning techniques have also been employed (55).

Perhaps the most common clean-up technique is the use of adsorbent column
chromatography. Adsorbents include Florisil (22),(45),(56), silica gel (5 7-60),
alumina (61) and carbon (62),(63). Carbon column techniques are also
used to separate PCB congeners with no ortho chlorines from those with ortho
chlorines (64). Other column clean-up techniques include HPLC (65), thin

layer chromatography (TLC) (3) and gel permeation chromatography (GPC)
(66).

Chemical methods of clean-up include degradation and sulfur removal. Commonly
used chemical methods include acid degradation (65, 67- 70), chromic trioxide
oxidation (71) and saponification (72). Sulfur removal from sample extracts

is typically done by shaking with elemental mercury (73), treatment with

22
tetrabutylammonium sulfite (74) or treatment with cyanide (75).

For GC/ECD analysis of PCBs, most samples are subjected to extraction (or
dilution), back-partitioning, adsorbent column elution and sulfur removal. All of this
clean-up is required due to the lack of specificity of the detector. For GC/EI-MS
methods, most of these procedures are still followed in an effort to keep the
chromatographic apparatus clean. For direct insertion EI-MS methods for some
targeted compounds, some of these steps may be reduced or eliminated. This still
requires much method development for each sample matrix encountered, though
since high background ion current may be detrimental. For other modes of MS
ionization, background levels are again dependent on the sample matrix. For direct
insertion MS/MS methods, simple extraction (or dilution) may be the only
requirement as a screening method. For the current work, however, all samples used
were subjected to the same clean-up procedures for purposes of comparison to the

methods used by collaborators.

Non-chromatomphic Methods. Many non-chromatographic techniques have been
used for the analysis of PCBs. Generally, though, none of these techniques possess
the specificity and/or sensitivity to be of use quantitatively for trace PCB
determinations. Nuclear magnetic resonance (NMR) has been used to characterize
the chemistry of PCB congeners (76-78), but has not been used for the quantitative
analysis of real samples. Infrared (IR) (79-81) and Fourier transform infrared

(FTIR) (82) spectrometry have been employed as screening techniques for total

23
PCB content. FI‘IR has also been used in the field for determining the PCB content

of transformer oils (83). While some of these techniques have been used as

chromatographic detectors (FTIR), most are used for confirmation of other
techniques.

Chromatographic Methods. The majority of chromatographic methods for the
analysis of PCBs has been with gas chromatography. Other techniques, though, have
included high pressure liquid chromatography (HPLC), thin layer chromatography
(TLC) and hyphenated methods such as GC/GC. These methods may be used for

sample clean-up and/or determination.

Gas Chromatography. Both packed column (glass or stainless steel) and wall-coated
open tubular (WCOT - typically fused silica) gas chromatographic with a wide
variety of detectors have been employed for the trace analysis of PCBs. Packed
column GC has been the most widely used technique, and several reviews have
appeared in the literature (84-89). One work investigated 13 different GC stationary
phases for observed theoretical plates and overall PCB resolution (90). It was
calculated that of the 21,945 pom‘ble pairs of PCB congeners, 465 would be
indistinguishable using the best column available. It would take multiple analyses

using at least five columns to resolve all possible pairs.

Increasingly, high resolution (WCOT) GC has become the method of choice for
PCB analysis due to the extremely high number of theoretical plates (up to 100,000)

24
and hence, resolution available (76, 77, 90-93). For congener specific analysis, high
resolution GC is required. The majority of methods have used methyl silicone, 95%
methyl/5% phenyl silicone or 94% methyl/5% phenyl/1% vinyl silicone as stationary
phases. A comparison of packed column GC to capillary column GC has been
presented in the literature (94).

The most common detectors used with GC for PCB analysis are ECD and MS. Mass
spectrometric detection is discussed later in this chapter. The electron capture
detector (ECD) is selective for analytes such as halogenated compounds and is
generally the most sensitive of all GC detectors for PCBs (95). Other
halogenated compounds will interfere in a PCB analysis, as will some esters,
nitrogen-containing compounds and sulfonated compounds. With the use of an
ECD, extensive sample clean-up is necessary to eliminate many of these
interferences. Even so, for some samples an ECD is not suitable for PCB detection
(96). Mullin has determined the relative responses to an ECD of all 209 PCB

congeners ( 77).

Other GC detectors used for PCB analysis include the Hall electrolytic conductivity
detector (HECD), the flame ionization detector (FID), the photoionization detector
(PID) and the thermal conductivity detector (TCD). While the I-[ECD is selective
for chlorinated and brominated compounds, the other detectors are generally
regarded as universal detectors, and as such, are not as selective (or sensitive) as an

ECD. The literature contains many reviews of the relative merits of the various GC

25
detection methods. Pellizzari has reviewed the GC detectors used for PCB analysis

(95).

HighPerformanceLiqur‘d Chromatography. Highperformanceliquid chromatography
(HPLC) may be normal phase or reverse phase. Normal phase HPLC consists of
uug a non-polar solvent system such as hexane and a polar stationary phase such
as alumina or silica. With reverse phase HPLC, a polar solvent system (water,
methanol, acetonitrile, etc.) is used with a non-polar stationary phase such as
chemically bonded C1,3 - silica gel. Reverse phase HPLC is more commonly used for
PCBs. Commonly used HPLC detectors in increasing order of selectivity are
ultraviolet, fluorescence and electrochemical. HPLC is primarily used for PCB
determination when combined as part of the clean-up procedure (97),(98).
The major limitation of HPLC as a routine method for PCBs is that none of the
commonly employed HPLC detectors are as sensitive or selective as those used with

GC.

Hyphenated Methods. Several investigators have used GC/GC or ”heart cut" methods
where the effluent of one column is redirected onto a second column using a
different stationary phase (14),(99),(100). With this method, Schulz, et. al,
were able to resolve all detectable congeners of several commercial mixtures of
PCBs. Since this technique is not commercially available, though, this could not be

termed a routine method.

26
Other Chromatographic Techniques. Other chromatographic techniques used for

PCB analysis include thin layer chromatography (TLC) (101),(102). TLC
is most commonly used as a method for analyzing food samples for PCBs and as a
qualitative confirmation technique for other methods. Perchlorination techniques
are actually modified GC methods where all congeners are exhaustively
perchlorinated in-line and the resultant decachlorobiphenyl content detected
(103). The major problem with this technique is that biphenyl itself is detected,
and generally high blanks result. A‘technique just the opposite of perchlorination is
termed carbon skeleton chromatography where all congeners are dechlorinated in
line and the resultant biphenyl content is detected (104-106). The same problems

with perchlorination techniques are inherent in this method.

Mass Spectrometric Detection. The use of mass spectrometry for PCB analysis has
generally been limited to gas chromatographic detection. The modes of MS usually
employed are electron impact (EI) and electron capture negative ionization (ECNI).
EI methods provide structural information for purposes of identification, but ECNI
methods are generally two to three orders of magnitude more sensitive. Positive
chemical ionization (PCI) may also be used with some improvement in selectivity
over El, but is generally employed only in situations where ECNI is unavailable. As
tandem techniques (MS/MS) become more routine, methods are developed forPCB
analysis by MS/MS due to enhancements in selectivity. MS/MS methods may be
applied to any of the ionization modes, and these methods are categorized as one

of two types— collisionally-induced (or activated) dissociation (CID or CAD) or

27
ion-molecule reactions. Any of these methods may be used as gas chromatographic

detection schemes or used with direct sample introduction such as with a solids
probe. Additionally, with the advent of ionization modes such as electrospray (E81)
and atmospheric pressure chemical ionization (APCI), HPLC/MS methods may

become more routine.

The choice of mass spectrometer is usually made with regard to analytical needs and
to cost. A simple GC/EIMS system may cost as little as $50,000 USD, and a high
resolution double-focussing sector instrument with many modes of sample
introduction and ionization may cost close to 8 1,000,000 USD. GC/MS systems with
chemical ionization and both positive and negative ion polarity modes are readily
available but at higher costs than that of EIMS systems. If available, high resolution
mass spectrometry can provide much more information than low resolution
instruments particularly with regard to possible interferences. For example, high
resolution MS would be needed to distinguish between the hexachlorobiphenyl ion
CnCIOH“ m/z 357.84443, and the pentachlorodibenzo dioxin ion C1202”CI,”CI2H3,
m/z 357.851662. A low resolution MS instrument is typically tuned such that at m/z
358, the resolution (< 5% valley) calculated as m/Am where m and Am are the m/z
values of two adjacent peaks in the mass spectrum is 358. In order to resolve the two
aforementioned ions then, the resolving power of the instrument must be at least
49,480. Using this method of calculating resolution, with a quadrupole instrument,
the peak shape remains constant across the mass range, but the resolution increases

with m/z. With a high resolution sector instrument, the resolution is fixed by slit

28

widths and focussing and remains relatively constant across the mass range, but the

peak widths broaden with increasing m/z.

Electron Impact Ionization. GC/EIMS has become a routine method for the analysis
of PCBs (78,83, 107-109). While analytical response varies from instrument to
instrument, the reproducibility from day to day is actually often better than that of
an ECD. Using full scan modes of operation, EIMS spectra provide much more
information than other modes of detection but at limits of quantitation (LOO) up
to four orders of magnitude higher than that of an ECD. As such, this technique is
often used for confirmation for ECD methods. Using single ion monitoring (SIM)
or multiple ion detection (MID) modes, much information is sacrificed but at an
improvement in LOQ of up to two orders of magnitude. Many other limited scan

range techniques have been proposed (91 ),(110),(111 ).

Differences in response are generally observed between sector, time-of-flight (T‘OF)
and quadrupole mass spectrometers. Since there are many h'braries of El spectra
available, it is important that data are generated under standard conditions. Most
BI data are acquired using electron energies of 70 eV. Also, instrument tunes are
generally adjusted according to the spectral characteristies of a cah’bration and
tuning compound such as decafluorotriphenylphosphine (DFI‘PP). An alternative
to this approach, is for the analyst to generate his or her own library of MS data with

standards for comparison with sample data.

29

Positive Chemical Ionization. By introducing a moderating gas such as methane into
ion source of a mass spectrometer, a plasma is generated that allows for the
protonation of analyte molecules (see Chapter 4). These processes, positive
chemical ionization (PCI), are much "softer" in that little fragmentation is observed.
For many analytes, this is a complementary technique in that PCI provides
molecular weight information and BI provides structural information. For PCBs,
however, this is not much of an advantage since the molecular ion is of significant
abundance for all congeners. Also, PCI is not as reproducible as El. Nonetheless,
several researchers have used the PCI technique for the screening and/or

determination of PCBs (107,112-118).

Electron Capture Negative Ionization. Using the same plasma of methane ions as in
PCI, a good source of thermalized electrons is produced. Under these conditions,
good electron-capturing analytes such as PCBs may be analyzed by electron capture
negative ionization (ECNI - see Chapter 4) (62119-125). Basically, the same
ionization processes occur in ECNI as in ECD modes of detection. Also, aswith PCI
techniques, instrumental parameters are more critical in maintaining reproducible
results. Operated in a SIM or MID mode (m/z 35, m/z 37 and/or the molecular
anion), ECNI analysis of PCBs approaches the detection limits of an ECD
(126). And, as with an ECD, responses of the 209 PCB congeners varies over
several orders of magnitude. Since this method offers an enhancement in selectivity

over that of an ECD, some of the sample clean-up may be eliminated (127).

30
Tandem Mass Spectrometry. By selecting a specific ion (cation or anion) produced
in the source, reacting this ion in a collision cell and mass selecting the resultant
product or products, an MS/MS spectrum is produced. This technique is necessarily
one of the most selective analytical techniques available (128),(129). In
addition, the reaction spectra allow one to regain the structural information lost
when more sensitive modes of ionization such as chemical ionization are used
(124),(130). The majority of early MS/MS work done with PCBs has involved
collisionally induced dissociation (CID) (131). Even with the enhanced

selectivity, CID MS/MS is not very useful for distinguishing positional isomers.

Neither the ECD nor the MS detectors are totally specific for halogenated species;
the ECD responds to many electron-capturing species besides chlorine, and MS
responds to virtually all hydrocarbons often present in real samples. MS/MS, on the
other hand, generally provides extremely high chemical selectivity. A GC/MS/MS
method could easily be restricted to chlorinated biphenyls; at worst these
interferences can be identified and compensated for (132). It has been shown
with GC/MS/MS that the analyte fragments can be identified with little interference
(133). This technique as implemented, though, was not used for quantitation.
Attempts have been made to use direct-probe MS/MS for PCB analysis, but again

quantitation was diffith and yielded erroneously high results (134).

None of the MS techniques have been very helpful in resolving positional isomers

of molecules (135),(136). MS/MS techniques, however, hold some

31

promise for alleviating this problem (137). For example, PCB isomers have
been distinguished by GC-MS/MS using oxygen-enhanced negative chemical
ionization (ONCI/MS/MS) by Guevremont and co-workers (138). In this work,
congeners of the same molecular weight and similar retention times were shown to
produce significantly different precursor-to-product fragmentation patterns
resulting in the relative abundances of each being distinguishable for the co-elutants.
Several rules for the identification of congeners were then postulated based on the
stability of precursor ions [M-Cl+ O]‘ with respect to chlorine position (and/or
hydrogen position) about the biphenyl structure. The development and application
of such rules would allow the possibility for the identification of PCB congeners and
their relative concentrations even if two isomers co-elute. Further development of
this relative intensities method using other modes of ionization and other MS/MS

techniques has yet to be explored.

In most cases, positional isomers do not yield unique precursor-product ion
combinations but rather different intensities of the same combinations. At a given
collision energy, then, it may be possible to distinguish one positional isomer from
another based on the ratio of two product ions. The same difference in product ion
intensities, however, may also be induced by varying instrumental parameters such
as ion source voltages and temperature. Thus, the use of such ratios is rarely reliable
and reproducible as an analytical method. In these cases, isomer-specific ion
molecule reactions with MS/MS shows some promise with regard to the

characterization of positional isomers even without GC separation. Some reactions

32
have been shown to be highly selective with regard to positional substitution.

Preliminary results obtained from reactions of the brominated analogs (PBBs) with
nitrogen, ammonia and deuterated water clearly distinguish the isomers with respect
to bromine position (139). For example, resultant spectra for
2,2’-dibromobiphenyl and 4,4’-dibromobiphenyl using the reaction of the negative
molecular ions with DzO are clearly distinguishable. This allows the possibility for
developing rules (and hence software) for these reactions based on chlorine position

on the biphenyl structure.

Data Reduction. Many methods have been used to process the MS data produced
in the analysis of PCBs. Most of these, however, have been used to characterize
samples as to total PCB content or Aroclor content. For the sake of completeness,

a few of these are presented.

One of the most prominent quantitation techniques for PCBs was originated by
Webb and M‘Call for packed column GC/ECD methods (140). In this
technique, samples are quantitated as total PCB content or as the content of specific
Aroclors for the case of mixtures of Aroclors. All PCB peaks are identified by
retention times relative to that of p, p ’-DDE, and a flow chart (Figure 3. 1) was

developed for quantitating samples with mixed Aroclor patterns.

Most GC peaks that contain more than one component are not precisely co-eluting

but rather ”overlapping" with maximum concentrations at slightly different times. In

33

Method of Webb and M°Ccll

 

 

RRTofFItsfPeck <47?

 

 

ml

 

 

 

 

lNo

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

lslhereoDisfinct
pedtwflhRRUU? ”WW
val 1N0 val lNo
Use l242for Use 1242 for Use 125m RRT>=70?
Peds<=RR184 Peds<=RRT70 Peds<=RRT104
IslheteoDisfhct UselZéOfor
PeokwiflrRRTW? AlPeoks

 

 

 

Figure 3.1. A flow chart (method of Webb and M‘Call) for the quantitation of

YES

1N0

 

 

Use 1254f0tAl
Feds <=RRT174

 

 

 

v i

 

 

UselZéOfor
AlOlherPede

 

 

samples with mixed Aroclor patterns.

 

 

 

 

34

these cases it is possible to produce clean MS spectra for each component effectively
improving peak recognition over the observed GC resolution by an order of
magnitude. Most MS spectra obtained in the course of a GC/MS run are a
superposition of several individual component spectra. By examining each spectrum
it is possible to identify the individual component spectra by testing the intensity of
each peak for synchronicity with the others. In this way, components can be
identified which do not even have sufficient concentration to appear on the total ion
current mass chromatogram. The identification of structural isomers which have
similar spectral masses at different relative intensities is sometimes possible by
comparing ratios of the intensities of fragment ions. This method for extracting the
necessary information from the GC./MS database has been shown to be feasible and

effective and has only recently been implemented in modern data systems.

Biller and Biemann generated reconstructed mass spectra and mass-resolved gas
chromatograms from total ion current (TTC) mass chromatograms (141). By
plotting m/z versus relative intensity for each scan number for only those ions which
exhibit their maximum intensities in that scan, a clean spectrum is produced which
is free of interference from the co-elutant. Also, by then regenerating the mass
chromatogram from the reconstructed mass spectra, much higher resolution is
achieved than from a TTC plot since the leading and tailing edges have been
discarded. The reconstructed mass spectra for samples spiked with known
compounds have been subjected to automated identification with h’brary data-bases

with excellent results. Also, Blaisdell and Sweeley have developed the algorithms

35
used in clean-up methods for MS data fields (142), and Gates and workers have

developed a computer system for quantitative GC/MS (143). It would be of use
to develop the software compatible with existing data systems to analyze the data
from standard GC/MS experiments for synchronicity of MS spectra. This would
include an improvement over current methodology with new smoothing and
correlation techniques. The result would be the production of a modified file of mass
spectra versus chromatographic time in which the mass spectra are essentially free
of background noise and random mass peaks and in which the chromatographic time

resolution is improved by approximately ten-fold.

Several workers have reported using regression techniques for both GC/BCD
(144) and GC/MS (145) analyses of Aroclors and their mixtures. Results
of such techniques have ranged from fair to good. By identifying up to 40 GC peaks
from several representative Aroclors, a matrix is generated for each sample, peak
by peak, as linear combinations of known Aroclors. In addition to this multiple
regression analysis (MRA), other more sophisticated classification schemes have
been reported (146),(147).

Summary. In this chapter, many of the commonly employed methods of analysis for
PCBs have been presented. All of these methods have limitations with respect to
cost and ease of implementation. In addition, as alluded to earlier, most have
moderate to severe limitations with respect to congener specific analysis. It was

toward these limitations that the research presented in this work was directed.

Chapter 4

Exchange of Oxygen for Chlorine in Molecular Anions of Chlorinated

Aromatics in a Tandem Mass Spectrometer

The products of the reaction of PCB molecular anions with molecular oxygen were
monitored as a detection scheme for chromatographically-separated PCB con geners
in an MS/MS experiment. In order to develop this method, the effects of several
instrumental parameters on this reaction were studied. These parameters included
the choice of reagents and pressures for both the ion source, and the collision cell
and the energy of the reaction as controlled by the collision cell offset voltage. Ako
studied were the effects of translational kinetic energy correction, electron energy,

and source and manifold temperatures on ionization and product formation.

Experimental. The Aroclor standards were provided by John Quensen, and all other
individual congeners and standards were provided by George Frame. All standards

were prepared by dilution with Organic Residue Analysis grade hexane (J .T. Baker).

Gas Chromatography. All gas chromatographic methods were performed with a
Varian 3400 gas chromatograph equipped with a 30 meter x 0.250 mm column of
36

37
0.25 mm DB-l phase (methyl silicone, J & W Scientific). Helium was used as the GC

carrier gas. The Varian 3400 was equipped with an injector body that accommodates
both split and splitless modes of injection. All experiments were performed using the

splitless mode of injection.

Mass Spectrometry. Early MS and MS/MS experiments were done with a Finnigan
TSQ-70B mass spectrometer operated in a negative ion mode. Later, the T80
instrument was upgraded to a T‘SQ-700. The pertinent changes included the
installation of an octapole as a collision cell (the TSQ-70 instruments use a
quadrupole) and replacing the PDP-l 1/73 data system with a DEC Station.
Ammonia, methane, argon or deuterated ammonia were used as chemical ionization

reagent gases.

Generation of the M' Torr. As a prelude to a discussion of negative ion formation
in MS, a discussion of positive chemical ionization is presented. It is important to
note that those instrumental parameters such as ion source pressure and electron
energy must be set differently for all three modes of classical ionization— electron
impact, positive chemical ionization and negative ion production under high

pressure ionization conditions.

Positive Chemical Ionization. Ions formed by true chemical ionization are those ions
that result from secondary and tertiary ionization processes. Generally, for positive

chemical ionization (PCI), the analyte molecules react with an abundance of reagent

38
gas ions produced by electron impact ionization of the reagent gas (148). When

methane is used as the reagent gas, the primary reagent gas ions produced are:

CH, + e‘ —» CH,*' + 2e" m/z 16
CH,+' -v CHf' + H2 m/z 14
CH,+' -’ CHf' + H’ m/z 15
CH," + CH, - CH,*' + ”CH3 m/z 17
CH,"‘ + CH, -. C,H,*' + H2 m/z 29

The major analyte ions produced are then:

M + CH,+ _. [NH-H]+ + CH, [Ms-1]+

M + CH,+ -. [M+CH,]+ [M+ 171*

M + 02H: ~ [M+C2H5r [M+29r

39

While there may be the formation of free protons with these processes, their
abundance is probably limited due to the high abundance of species with significant
proton affinities. As such, the formation of protonated analytes from association
with free protons is generally ignored as insignificant compared to other processes.
Fragmentation observed with PCI generally includes elimination of a functional
group X together with a hydrogen atom from the protonated molecule. This process
typically involves transfer of a proton from a reagent ion to a specific site (such as
a heteroatom) on the sample molecule (149),(150). This process may be
summarized as:

AX+RH+-.R+AXH"+HX+

where AX is the sample molecule, X is the functional group or heteroatom and RH”
is the reagent ion. Other fragmentation reactions observed with PCI include hydride
abstraction (150-157) and electrophilic attack of a carbon/carbon single bond by a

proton with the resultant cleavage of the sigma bond (154),(158).

Optimum PCI conditions generally exist when the reagent gas pressure in the CI ion
volume is approximately 1 Torr. Th's roughly corresponds to 5-10 Torr on the ion
source Convectron gauge or 5-10 x 103 mTorr on the ion gauge. The optimum
pressure for analyte ion production is not the pressure that results in the highest
abundance of reagent gas ions. At higher reagent gas pressures, reagent gas ions
become more abundant, but the abundance of analyte ions decreases as their

formation and the ability to be extracted from the ion source is suppressed.

40
While El tuning compounds (such as PFI'BA - FC43) are used to tune the

instrument for positive chemical ionization, the tuning compound ions are not the
result of chemical ionization. Although the higher mass PFTBA ions are generally
more abundant with PCI, the same ions are observed with PCI as with E1. If one
wishes to optimize the CI reagent gas pressure by monitoring the abundance of an
analyte ion that undergoes PCI, the analyte must be introduced in some steady state

fashion such as via a micro valve in plumbing connected to the solids probe.

Positive chemical ionization is generally a sofier ionization technique than E1 in that
much less fragmentation is observed. The choice of reagent gas is analyte dependent.
Those analytes that have proton affinities less than that of the reagent gas will not
become protonated, but may form positive adduct ions. For analytes having a proton
affinity greater than that of the reagent gas, the analyte becomes protonated, and
quite often, the base peak represents the protonated molecule [M+H]+. As such,
PCI is a complementary technique to El. Molecular weight information is provided

by PCI, and structural information by El.

Electron-Capture Negative Ionization. Negative ions are produced in all modes of
ionization including El. Generally, though, molecular anions are produced in a
significant abundance only in the presence of thermal electrons, which can be
provided by a significant amount of a moderating or reagent gas. While positive ions
observed under CI conditions are usually adducts of the neutral analyte molecule

with protons or protonated species, almost all negative ions observed with high

41

moderating gas pressures are the result of the capture of a thermalized electron or
electron capture negative ionization (ECNI). There are, however, five prominent

types of negative ion production (149),(159),(160):

M + e' (< 1 eV) -» M‘ Resonance Electron Capture
MX + e' (< 15 eV) -e M“ + X0 Dissociative Electron Capture
MX + e’ (> 10 eV) -* M' + X” + e' Ion-pair Formation
M + X‘ -» MX‘ Negative Ion-molecule Reaction
M + X" - M" + X Charge-exchange Reaction

Notice that only the negative ion-molecule reaction is analogous to PCI, and thus,
could properly be called NCI. The vast majority of negative ion processes observed
are a result of one of the first two processes which produce M" or [M-H]’.
Nonetheless, all negative ion processes are quite often categorized (incorrectly) as
negative chemical ionization or NCI. Notice, abo, that the electrons involved in
these processes are not those that result directly from the higher energy filament
emission. The thermalized electrons that are involved in electron-capturing
processes are 1) those electrons released during the ionization of the moderating gas

and 2) those electrons emitted from the filament that lose most of their energy

42

during collisions with molecules in the relatively high pressure of the ion source. In
addition, unlike in PCI, since the ions observed during tuning are the result of the
same ionization processes by which analyte ions are produced, these ions may quite
often be used for the optimization of all instrumental parameters. The ions observed
( M' and [M-X]‘ ) with both analyte and tuning compounds are the result of
resonance or dissociative electron capture, and as such, optimizing instrumental
parameters such as moderating gas pressure with the tuning compound should also

allow for optimum conditions for analyte anion production.

Electron capture negative ionization is generally one of the softest ionization
techniques available in mass spectrometry. The molecular anion is almost always
represented by the base peak, and very little fragmentation is generally observed. As
such, ECNI is usually the most sensitive technique available for those analytes
amenable to these processes. Generally, any species that can be detected by electron
capture detection (ECD) is amenable to ECNI-MS. These include halogenated
compounds, sulfonated compounds, many nitrogen containing compounds
(particularly those with basic nitrogens such as amines), many oxygen containing
compounds, and some aromatic and polyaromatic compounds. While this work is
prirnarilyconcernedwith chlorinated aromaticcompounds, otherelectron-capturing

compounds must not be overlooked as possible interferences.

Amajor difference between ECN I and other ionization techniques, is that these low-

energy processes are much more susceptible tovariations in source conditions. Small

43

voltages or charging phenomena on source elements can drastically compromise the
instrument tune. Also, it is quite possible to saturate the ionization capacity of the
source (depletion of thermalized electrons) without saturating the analyzer or
detector. With ECNI, very little fragmentation is observed in the primary mass
spectra of PCB congeners (161),(162). Generally, disregarding chloride
ions, the most abundant ion produced is in the molecular ion region of the spectrum.
This ion maybe the [M]’ or [M-H]' ion or the [M+2]' or [M+4]‘ ions in the case
of the more chlorinated species due to the presence of 3" Cl isotopes (taking M as the
PCBwith only 3’Cl). Also observed in the primary mass spectrum of a PCB congener
is the isotopic cluster at -19 u (”Cl-“O or 35-16) due to the oxygen-chlorine
exchange reaction products formed in the source from the molecular anions. These
ions are usually ten to thirty percent as abundant as the molecular ions. The
molecular ions can also lose a chlorine to form the isotopic cluster at -35 u, but this
region is generally only one to twenty percent as abundant as the molecular ion
region. This does not imply that the loss of chlorine from a molecular ion is not
favorable, only that the charge may stay with the chlorine as opposed to the biphenyl
structure. Rarely are any other ions observed with relative abundances more than
a few tenths of a percent. Thus, only ions from these three regions are logical choices
as precursors in an MS/MS experiment. This is shown in Figure 4.1, which is the
ECNI mass spectrum of a tetrachlorobiphenyl.

Negative ion formation, particularly in the case of chlorinated aromatic compounds,

is generally due to the capture of a thermal electron produced during the ionization

 

 

 

 

 

 

 

  

 

 

 

   

 

 

m“ I as: j s as
1°. 1 ' 3.04
”‘1
‘ C1
-— .. © © .1
U)
at, c. E
E “T g
— :3
(D 1 3" O
4“): C
U 40 . 2
a) ‘ ”is
a .
j ”I, If, ' P: m] '30 :Ill"?'m

 

 

m/z

Figure 4.1. The ECNI mass spectrum of a tetrachlorobiphenyl. The three regions
of interest are the molecular ion region, the oxygen-for-chlorine exchange reaction
region and the loss of chlorine from the molecular ion region.

45

of the moderating gas in the source and may proceed by one of the two electron-
capturing processes. For this work, the two ionization products of interest are the
[M]' and [M-H]' ions. Whether one or the other or both are produced is a function
of the electron affinities of the species involved. The major difference in these two
species other than mass is that [M]' ' is a radical anion and [M-H]' is not. It has
been previously reported that for PCBs, those species with five or more chlorines
generally yield [M] ' ' ions as a result of resonance electron capture and those with
four or fewer chlorines yield [M-H]' ions as a result of dissociative electron capture
(163),(164). In the present work, resonance electron capture has also
been observed with a few trichlorobiphenyls and all the tetrachlorobiphenyls,
producing a significant amount of the [M]" ' ion. In Figures 4.2 through 4.5 are
shown the mass spectra of selected tetra- through heptachlorobiphenyls. As is
shown, the predominant ions are molecular anions [M]' ' with nominal masses of
m/z 290, m/z 324, m/z 358 and m/z 392. The anion [M-H] ' is significant only with the
tetrachlorobiphenyl. For each of these, though, the ratio [M]' ’/[M-H]‘ could be

estimated from the excess abundance of [M+l]' over 13.1%.

Ion Source Optimization for E CNI. The major parameters to be considered for
generating negative ions are moderating gas pressure, type of moderating gas and
filament electron energy. Since the primary ionization processes involve the
interaction of analyte molecules with thermalized electrons, the generation of a large
abundance of thermal electrons is desired. This usually requires that the electron

energy be lower than that used for E1 or PCI and the moderating gas pressure to be

 

 

 

 

 

 

 

 

 

 

 

 

 

We 1 ”2 “l r. as
so C1 C1 290
C E
m 30- a)
'E t:
— 3
(1) 29,4 0
“>- C
.1—
9 9
a) 25.5 351
a
:19 271 295
3‘1.
I II?“ as 2”rs s 2 I ll»?
:3 1- .L T -. g I lfil,f,',.n.» .4
no no 210 m g» , soo

M‘.
m/z

Flame 4.2. The ECNI mass spectrum of a tetrachlorobiphenyl indicating the [M] " ‘
radical ions.

47

 

Fire I 7
w C1 C1
.1 e e a

Cl 324, 33s

3“

40"

Relotlve Intensity
Ion Current

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

214
330
l
W 301
F5
395,
1“ 6 0!
. pp, r
33}
u. 14 3 9 333 339
r.-:r:|!l., -..l.1r:t--lll.li:h&‘1f:.1l.l|!--|.'
380 390 300 310 320 ‘ no , 340

M
m/z

Figure 4.3. The ECNI massspectrum of apentachlorobiphenylindicating the [M]’ °
radical ions

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[1:5 ’1 ‘—1
rev m '35:,
3s:

,0. 325
8 E
.8? .. w

h

C 3%. ISI ‘—
3 m a
> C
a:
U ‘0‘ 3f. 2
(D
(Z

30‘ 310 err

33 343
T I m
I I 31: 3'45
1 o 1 31: m
- -I.1.?’.’.-..n -gl’3:_1’.-’.‘.’1- :1: t
355 - no ' sic ' 3J0 use ”'3 , sic

 

 

m/2

Figure 4.4. The ECNI mass spectrum of a hexachlorobiphenyl indicating the [M]' '
radical ions.

49

 

 

 

 

 

 

 

F l .
190 W 3,2" P It. ‘0:

7; C1 C1 4—
C c:
0) w Q)
+"' L—
C h
— m 3
(I) 3., U
4?.
+- C
(U 40‘ O
(D
(z 1“

3."

30‘ :51 4'00
m
l 3‘“ 315 | I (on
3 m m 3 3 4 o 4
”7 I. LL1- -._--n,_- riff: -ulf ..... H:
350 no sis 530 ' we do do

 

 

 

 

 

 

 

 

 

M.
m/z

Figure 4.5. The ECNI massspectrum of a heptachlorobiphenyl indicating the [MT ’
radical ions.

50
somewhat higher than that used for PCI.

The use of highly basic reagent gases can sometimes result in the production of
higher abundances of thermalized electrons. While many CI/MS techniques employ
methane as a moderating gas, it was observed that ammonia offers advantages for
negative ion formation. This difference was not studied in a quantitative fashion, but
the abundance of negative ions of PCBs produced in the source and extracted into
the precursor analyzer appeared to be up to two to three times more abundant when
using ammonia rather than methane as the moderating gas. In addition, maximum
[M]‘ ‘ ion formation was observed when the ammonia pressure was approximately
ten percent higher than the pressure normally used with methane. Not only did this
choice result in somewhat greater negative ion formation in the source, but greater
MS/MS product formation was observed with the use of ammonia as opposed to
methane. This could be due to either somewhat reduced fragment ion formation
with ammonia or to a greater abundance of thermal electrons. These phenomena
were observed early in the development of this method, and no attempt at
quantifying these observations was made. The optimum pressure of ammonia was
determined to be approximately 9500 mTorr as measured by the ion source
Convectron gauge by monitoring the formation of radical anions in an MS
experiment with a steady state of Aroclor 1242 being leaked into the source (Figure
4.6).

On a TSQ-700 there are two voltages associated with the filament. The voltage

51

Moderating 6303 Pressure vs TIC

4110.59 CI Gas Pressure vs TIC

443930. 0 WK
Total Ion

Current

V

 

 

 

 

 

   
  

 

Source Pressur

L-I‘
52.00 0.5-] .5 mTorr

1127.81

 

 

 

Soon Number

figure 4.6. The ion current of the molecular anion of a pentachlorobiphenyl as the
pressure of the CI moderating gas is varied from high to low.

52

across the filament is set such that resistive heating of the filament occurs with
electrons being emitted. In order to form a stable environment for ionization, the
electron emission is set (0- 1000 uA). The voltage across the filament is varied to
maintain that emission as source conditions change with analyte introduction (such
as chromatographic elution). In addition to the voltage applied across the filament,
the entire filament is biased (negatively) with respect to ground. Since the ion source
(ion volume) is at ground potential, this filament bias is the energy of the electrons
being emitted. With El experiments, the filament electron energy is usually set to 70-
eV. This is due to the ion production with respect to fragmentation of most MS ion
sources being relatively stable at this value. Also, most EI analyses are performed
for the purposes of searching generated El spectra against a database (library) which
was also generated with 70 eV electrons. With PCI, most analyses are performed
with higher electron energies (100-200 eV) again for reasons of stability and for

purposes of enhancing the production of [M+H]+ molecular ions.

For the current work in ECN I modes of ionization, the filament emission was
maintained at 300 u-amperes. Higher emission currents generally provide for more
molecular anions being produced, but at a compromise in filament lifetime. At an
emission of 300 u—amperes, the production of M" optimized at an electron energy
of approximately 50 eV (Figure 4. 7). The energywas varied from 35-200 eV and the
ion current monitored. Lower energies were not studied since the instrument
automatically lowers the filament emission at low energies. This is a feature that

prevents the user from destroying the filament by "boiling off' significant amounts

53

Electron Energy vs TIC

ElectronEnergyvs‘lIC

2940063 JUL
103407€t6.00

 

Totol lon Current

 

6261555‘
rrrrnsn nn

 

 

 

35 T95
Electron Energy (volts)

Figure 4.7. A plot of electron energy versus ion current of the molecular anion of
a pentachlorobiphenyl.

54

of electrons, but not removing them from the region with an adequate electron

energy.

The temperature of the ion source has been shown to have an effect on the
production of molecular anions (165). For the present work, MS
chromatographic analyses were conducted at 100 C, 150 C ("ISO-700 default) and
250 C (See Chapter 5). At the low temperature, a more significant amount of
background was observed than at the higher temperatures. At the high temperature,
more fragmentation in the source was observed than with the lower temperatures.
As such, all experiments were conducted with the source at 150 C and the manifold
(analyzer) at 70 C. In an effort to eliminate chemical background, the system was
generally "baked out" during the evening with a source temperature of 250 C and a
manifold temperature of 150 C, and then returned to 150 C and 70 C, respectively,

prior to analysis.

Oxygen-Chlorine Exchange. By forming the precursor ion [M]' under
elecfion-capture negative ionization conditions and selecting it in the first
quadrupole of a triple-stage quadrupole mass spectrometer and introducing oxygen
into the second quadrupole (collision cell), the exchange reaction of oxygen for
chlorine may be observed. The same reaction is not observed with the [M-HT
species. This phenomenon may be attributed to the fact that the former is an odd-

electron moiety and the latter even-electron.

55
Odd- vs. Even-Electron Anions. It has been observed that only radical anions of PCBs

will undergo the oxygen-chlorine exchange reaction. Other precursor ions such as
the [M-H]’ ions will not undergo oxygen exchange. These ions are even-electron
aromatic species forwhich the oxygen-chlorine exchange reaction isnot energetically
favorable. The radical anions, however, are odd-electron and non-aromatic, and as
such, are amenable to reactions that produce more stable products. The m/z 290 ion,
which is the 35C1 molecular anion of tetrachlorobiphenyls, is also present in the
primary mass spectrum of many pentachloro- congeners as a result of the loss of
chlorine from the 13C molecular anion. The ion formed by loss of chlorine in the
source has an even number of electrons. The oxygen-chlorine exchange reaction,
however, is specific for the odd-electron molecular anion. Therefore, the MS/MS
detection of m/z 271 ions formed from m/z 290 will be completely free of
interference from the even-electron m/z 290 ions formed by the loss of chlorine from
pentachlorobiphenyls in the source. This also explains the fact that the oxygen-for-
chlorine exchange reaction in the source is only observed in the mass spectrum in the

neutral loss of 19 u from the molecular ion and its isotopes.

Several researchers have investigated the reaction of molecular oxygen in an MS
source with chlorinated hydrocarbon radicals in both MS (166) and MS/MS
(138) experiments. Indeed, it has long been known that oxygen will quench radical-
initiated chain reaction mechanisms (167). Recently, the use of this reaction
in an MS/MS experiment has been explored as a detection scheme for

tetrachlorodibenzo-p-dioxins and tetrachlorodibenzofurans (168),(169).

56
It has been proposed that the oxygen atom(s) in these species are instrumental in the
reactions of their precursor ions with molecular oxygen. While this may be true, it
is obviously not a prerequisite for reaction because the same reaction is observed
with PCBs, and this reaction has been reported for other chlorinated aromatics

(170).

Reaction Scheme. The reaction of a hexachlorobiphenyl molecular anion with
molecular oxygen is shown in Scheme I. The molecular ion shown is only an
example, and presumably, due to resonance, any of the carbon atoms of the biphenyl
skeleton could exhibit the higher radical electron density. In addition, in the case of
asymmetric congeners, the charge may exist on either of the two rings or on both in
resonance equilibrium. Perchlorinated aromatic compounds, such as
decachlorobiphenyl and octachloronaphthalene, exhibit responses to this reaction
that are considerably less than that of the congeners with lower degrees of
chlorination. This would indicate that, while hydrogen atoms attached to the ring are
not necessary for this reaction to occur, those positions of chlorination which are

ortho to hydrogens may be more favorable due to steric considerations.

While the exact mechanism of this reaction is not certain, the product ion is most
likely that represented in Scheme 1. This product, unlike the precursor anion, is a
highly stable moiety with total aromatic character. In addition, the charge residing
on the oxygen atom is the only configuration possible in which each atom of the

product is surrounded by a full complement of eight electrons. The stability of this

57

SCHEME I

c.

.0, mm
01

.. o o
C 'r

58
product may also explain the fact that the exchange reaction products do not

undergo further exchange of chlorine for oxygen.

MS/MS Reaction Parameters. When acquiring data in an MS/MS mode (product,
precursor or neutral loss), the voltages applied to all of the analyzer elements afier
the collision cell need to be modified from those set during tuning to compensate for
the collision energies of the MS/MS reactions. On a TSQ-700 this is accomplished
automatically using a voltage programming algorithm that takes into account the
original tune voltages, the energy of the collisions and the masses and energies of the
product ions. The elements affected are Lens Stack 3 or the Einsel lens assembly
between the collision cell (02) and the product analyzer (Q3) —— L31, L32 and L33
— and the product analyzer offset (DOFF). These analyzer elements are referred
to aslinkparameters since their values are linked tothe mass currently being filtered.

As such, link parameters are used to tune the instrument.

Voltage Programming The goal of voltage programming is to set the voltages on the
analyzer elements such that the Translational Kinetic Energy of the product ions
(T‘KE,) is the same as that experienced by isobaric ions in the Q3MS mode of
scanning. As an example, the Daughter (product) OFFset (DOFF) parameter for an
MS/MS reaction will be explored. For the moment, Lens Stack 3 will be ignored.
Specifically, the reaction of m/z 500 (positive) at 100 eV to produce a product at m/z
250 will be used. Also, it is assumed that the Q3MS tune resulted in DOFF = -5.2

volts at m/z 250.

59

As a first apprar‘ation, it is assumed that no loss in total energy of the products
occurred as a result of the dissociation reaction. This assumption will be dealt with
later. This means that the product ion would have a TKE, = 50 eV —- half the
mass, half the energy. Since the product ion must have only 5.2 eV in Q3, it must
experience a decelerating potential (DP) of 44.8 volts. Now, since COFF = -100 V,

DOFF must be set at -55.2 V; i.e.,

(1) DP = DOFFM, - TKE, and
(2) DOFFW = COFF + DP or
(3) DOFFm = DOFFM, + COFF - TKE, .

Now, since
(4) TKE, = COFF (MJM) , where

M, is the mass of the product and M, is the mass of the precursor, substituting for

TKE, leaves

(5) norrm = nose,“ + core - comm or

(6) norrm = none” + coma-MM .

60
Now, substituting the values for the above experiment into Equation (6) does indeed

result in DOFF = -55.2 V. This equation would work well, however, only for those

ions not losing energy in the collision process.

Translational Kinetic Energy Correction. Since most reactions result in some energy
loss, TKE, will be some fraction of Equation (4). This is accounted for by
incorporating into the equation a translational kinetic energy correction factor,

MSMSC, or

(7) TKE, = COFF (MJMp) (MSMSC/100) where

MSMSC is expressed in percent and is a number between 0 and 100. This allows the

instrument to account for an MS/MS product ion losing all, none or part of its

theoretical energy. Now, substituting into Equations (5) and (6) leaves

(8) norrm = nose,” + COFF - commrmsmscrroo) or

(9) norrm = norrm + COFF[l-(M,/M,)(MSMSC/100)] .

As an example, the two extreme cases are considered. For the case of metastable

decomposition (i.e., no collision gas), no energy is lost during the reaction, and

MSMSC should be set at 100. The would be identical to the above example of m/z

61
500 producing m/z 250 at 100 eV — DOFFm would be set at -55.2 V. For the

other extreme, all translational kinetic energy is lost during CID (i.e., TKE, =0).

This is common for many reactions at high CID gas pressures. In this case, Equation

(9) reduces to
(10) DOFFm = DOFFM, + COFF (MSMSC = 0).
Thus, the product ion must be accelerated — DP is negative for positive ions. For

the ion at m/z 250 with COFF = -100 and TKE, = 0, DOFFm should be -105.2V.

In the above examples, only the DOFF parameter was considered. These are shown
graphically in Figures 4.8 and 4.9. As originally stated, the voltages of all of the
analyzer elements afier the collision cell (L31, I32, 133, and DOFF) must be
programmed to reflect the energies of the collisions and products. The more

generalized form of Equation (9), therefore, is

(11) LTABW,” = LTAB,,, + COFF[l-(MJM,)(MSMSCIloo)] where

LTABW", and LTABm are L31, L32, L33, and DOFF. Thus, the voltages of these
elements in MS/MS modes of acquisition are calculated from Equation (1 1) and not
derived from an MS/MS link table. MSMSC becomes more significant for
experiments conducted at low pressures where product ions do not lose all

theoretical energy. From Equation (11), it is also apparent that the use of MSMSC

62

 

  

 

 

o «-
DOE-651V
V
O
In
2
3
G
I
M

 

 

lon Flight Path —>

Figure 4.8. The offset voltage settings for Q], Q2, and Q3 for three settings of
MSMSC.

63

 

 

 

 

arrow
sarcoma.

" Dore-52v

D

0

I

r

V

0

I

I “d.”

‘ mrwcrcrcc-tm

‘ ear-imam

' mmomcoo

a j

0 r

"a.

Figure 4.9. The plots of the DOFF voltage as a function of the ratio of the daughter
mass to the parent mass.

64
(greater than 0%) becomes more critical for experiments conducted at high energies

and with products from mall neutral losses.

By introducing Aroclor 1242 into the source in a steady state, the kinetic energy
correction factor (MSMSC) was varied from 0 to 100% and the ion current from the
reaction of m/z 326 with molecular oxygen to produce m/z 307 monitored (Figure
4.10). As was expected with these low collision pressure, low energy experiments,
MSMSC was shown to have minimal effect. This was observed early in thiswork but
not tabulated until later. As such, all experiments were conducted with MSMSC set

to the default value of 70%.

Collision Reagents and Pressures. The maximum product formation from a given
abundance of precursor ion is dependent upon both collision pressure and collision
energy. Optimization of product ion production for this ion-molecule reaction
occurs when the pressure of oxygen in the collision cell is such that the collision
cross-section allows for a high probability of interaction with a limited amount of
scattering and fragmentation. Thus, higher pressures of oxygen require higher
collision offset energies for maximum product formation. As the amount of oxygen
is increased, however, the energy required to maintain optimum conditions becomes
so great that either the likelihood of reaction is diminished or the products cannot
be focussed into the entrance of the product analyzer. This relationship is shown in
Figure 4. 11 where the reaction of a pentachlorobiphenyl was monitored as the

collision gas pressure was varied from 10 mTorr to non-existent. The optimum

65

Tronslofionol Kinetic Energy Correction vs 11C

 

 

 

 

 

 

 

 

 

 

 

i
4.—
C
9).
5
U
C
.9
6 V
'5
j...
0 100

MSMSC

Figure 4.10. The ion current of the product of the oxygen-for-chlorine exchange
reaction as a function of the translational kinetic energy correction factor
(MSMSC).

120'

reassessment “W
Hill!

A

g

an
o
A 4

L

RELATIVE INTENSITY
1.5 e

A

 

PRESSURE (mTORR)

figure 4.11. Ion current of the product of the oxygen-for-chlorine exchange reaction
as a function of the pressure of the collision gas.

67

reaction parameters were determined to be about 1 mTorr of oxygen in the collision
cell with 1.5 -eV (laboratory) precursor ions. This corresponded to collisions of about
1.0 eV in energy (See below).

Early MS/MS experiments with PCBs were done by monitoring the oxygen-chlorine
exchange reaction products formed by reaction with ambient air. While this proved
to be a viable alternative to pure oxygen, a reduction in sensitivity was observed.
Dilution of the oxygen with unreactive gases results in optimization at higher
pressures where the rates of scattering and fragmentation are also increased (Figure

4.11).

Collision Energy. With the collision gas (oxygen) pressure set to approximately 1
mTorr, the voltage offset of Q2 was varied from -2 volts to 8 volts and the ion
current of oxygen-for-chlorine exchange reaction products of tetra-, penta- and
hexachlorobiphenyls monitored (Figures 4. 12-4. 14). Although the optimum energy
of an ion-molecule reaction in an MS/MS experiment is dependent on all parameters
such as collision pressure and type of collision gas, the energy (Q2 offset) is clearly

the most critical.

The optimum energy for an MS/MS reaction with a triple-stage quadrupole
instrument is determined and set by varying the voltage offset of the collision cell.
This offset is the energy of the MS/MS reaction in the laboratory frame of reference.

In order to determine the true energy of the reaction(s), the precursor ion

Collision Energy vs TIC

 

m/z 292 to m/z 273

 

Totol Ion Current

 

 

 

-2.oor 8.00*
T Q2 Offset (volts)

Figure 4.12. The ion current of the oxygen-for-chlorine exchange reaction product
of a tetrachlorobiphenyl as a function of collision energy.

69

Collision Energy vs TIC

 

 

 

 

m/z 326 to m/z 307

1%
z:
:3
0
C
9
2'9
,9

l I

-2.00* 8-00‘

Q2 Offset (volts)

Figure 4.13. The ion current of the oxygen-for-chlorine exchange reaction product
of a pentachlorobiphenyl as a function of collision energy.

70

Collision Energy vs TIC

 

m/z 360 to m/z 341

Total Ion Current

 

 

 

-1001: 8.00’
Q2 Offset (volts)

Figure 4.14. The ion current of the oxygen-for-chlorine exchange reaction product
of a hexachlorobiphenyl as a function of collision energy.

71

transmission is monitored as a function of the Q2 voltage offset with no collision gas
being introduced (Figure 4. 15). The difference of the optimum Q2 voltage offset for
product ion formation and the inflection point of the energy plot for the precursor
ion transmission is the true optimum energy for the given reaction. As shown, the
inflection point occurred at approximately 0.5 volts and the optimum offset at
approximately 1.5 volts. Figure 4.15 shows the energy range of the ions that are
extracted from Q1. All ions to the left of the offset used get in with the excess energy

given by their displacement to the left

Having performed the energy vs. ion current experiments on several occasions over
many months, it was observed that the optimum value is dependent on ion source
conditions. Specifically, as the ion source becomes more and more contaminated,
the optimum Q2 voltage offset occurs at higher values. This is also true of the
precursor ion transmission. Thus, even when the optimum reaction energy
(laboratory) occurs at higher voltages, the true energy of the reaction(s) remains

relatively constant at approximately 1 eV.

Collision RF Linkage. The default setting for the RF power on the collision cell (Q2)
on a TSQ-700 is linked to the precursor analyzer (Q 1). This peak-to-peak voltage
will be set from 2-8 kV depending on the mass being filtered by Q1. This is a
problem for MS/MS reactions produced at high collision pressu res (> 1 mTorr), low
energies (< 40 eV) and where the ratio of the product-to-precursor mass is low (<

0.5). For reactions under these conditions, product transmissions may be extremely

Collision Energy vs TIC

 

m/z 326 1

All ions
extracted

 

 

 

 

 

4...
C
2 lnflection Point
8 < 50% of ions
C <— extracted
0 from source
: due to asymmetry
.9.
8
No ions
xtracted
. l
I
-6.00* 4.00*
Q2 Offset (volts)

Figure 4.15. The transmission of the molecular anion of a pentachlorobiphenyl as
a function of the collision energy (Q2 voltage offset) with no collision gas being
introduced.

73

low or even non-existent due to the product ions being produced near the entrance
of the collision cell. At the relatively high RF powers that are optimum for the
precursor ion, quite often the lower mass product ion is unstable. The provision for
changing the 02 linkage to Q3 is available on the TSQ-700, and this capability offers

a significant improvement in product transmission for many experiments.

For the present work, changing the linkage of the 02 RF power to 01 offered little
or no improvement in product ion production. This was not surprising since the
collision gas pressures of the reactions being studied were so low and the product
ions were only 19 u less than that of the precursor ions. This phenomenon was
observed on several occasions and was not tabulated. As such, all experiments were

conducted under default conditions — 02 RF power linked to 01.

Summary of Experimental Details. In conclusion, the following results were

observed with the reactions of the molecular anions of PCBs:

O The odd-electron molecular anions M" of PCBs will undergo an exchange
of chlorine for oxygen in an MS/MS experiment in a triple-stage quadrupole
mass spectrometer. It will be shown later that the even-electron [M-HT

species will not.

6 Under CI conditions, mono- and di-substituted PCBs produce negative

anions only by dissociative EC ([M-H]‘). Some tri- and all tetra-substituted

74
PCBs produce a significant amount of M" through resonance EC. All penta-

through decal-substituted PCBs exclusively produce molecular anions (M")

through resonance EC.

Molecular anion production maximizes with moderating gas pressures of
approximately 9500 mTorr as measured by the TSQ-700 ion source gauge.
This pressure is slightly higher than the optimum (8500 mTorr) observed with
PCI.

Using a filament emission of 300 u—amperes, molecular anion production

optimizes with an filament electron energy of 50 eV.

Ammonia was observed to be a better CI moderating gas for negative ion

production than methane.

Product ion formation for the oxygen-for-chlorine exchange reaction
optimized with a collision offset of approximately 1 eV. This represents the
optimum compromise between transission and fragmentation and/or

scattering losses.

Translational kinetic energy correction through QB is not a critical factor for

these small neutral-loss, low-pressure reactions.

75
Collision cell (02) RF linkage is not a critical factor for these mall neutral-

loss experiments.

The use of pure oxygen as a collision gas provides an enhancement of

product ions over the use of ambient air.

Chapter 5

Development of an Analytical Method for the Analysis of
Polychlorinated Biphenyls using the Exchange Reaction of Oxygen
for Chlorine with Molecular Anions of Chlorinated Aromatics in a

Tandem Mass Spectrometer

Introduction. Using the exchange reaction of oxygen for chlorine with
polychlorinated biphenyl anions, an analytical method for monitoring reaction
products in a GC/MS/MS experiment was developed. The method development
included optimization of chromatographic and mass spectrometric instrumental
parameters such as GC temperature programming, the MS tune and MS scan
functions for both full scan and selected reaction monitoring (SRM) modes of
acquisition. For quantitation, the selection of an internal standard or standards was

determined based on both GC retention time and relative instrumental response.

Experimental. The Aroclor standardswere provided by John Quensen, and all other
individual congeners and standards were provided by George Frame. All standards

were prepared by dilution with Organic Residue Analysis grade hexane (J .T. Baker).

76

77

Gas Chromatography. All gas chromatographic methods were performed with a
Varian 3400 gas chromatograph equipped with a 30 meter x 0.250 mm column of
0.25 mm DB-l phase (methyl silicone, J & W Scientific). Helium was used as the GC
carrier gas. The Varian 3400 was equipped with an injector body that accommodates
both split and splitless modes of injection. All experiments were performed using the

splitless mode of injection.

Mass Spectrometry. Early MS and MS/MS experiments were done with a Finnigan
TSQ-7OB mass spectrometer operated in a negative ion mode. Later, the T80
instrument was upgraded to a T‘SQ-700. The pertinent changes included the
installation of an octapole as a collision cell (the TSQ-70 instruments use a
quadrupole) and replacing the PDP-l 1/73 data system with a DEC Station.
Ammonia, methane, argon or deuterated ammonia were used as chemical ionization
reagent gases. The CI reagent gas pressure was set to approximately 9500 mTorr as
measured by the source Convectron gauge and was optimized daily using
perfluorotributylamine. Pure oxygen or ambient airwas introduced into the collision
cell. The source was held at 150 C and the manifold at 70 C. The instrument utilizes
a 20—kV dynode which was set at 10 kV. All reactions were generated using electron
energies of 50 eV, collision offset energies of 1-2 eV (laboratory), and collision

pressures from 0.5 mTorr to 1.5 mTorr.

Gas Chromatography. The three most common stationary phases used for the GC

78
analysisof chlorinated aromatics are methyl silicone (DB-1), 95 % methyl/5 % phenyl

silicone (DB-5) and 94% methyl/5% phenyl/1% vinyl silicone (SE-54). The latter
phase, SE-54, provides resolution of more congeners (188 of 209) than the other
two. For many modes of detection, such as ECD and FID, chromatographic
resolution provides ahnost all of the selectivity of the method. With MS detection,
quantitation of higher mass congeners that co-elute with other lower mass congeners
is possible due to differences in mass spectral ion abundances. The purpose of the
current research was to develop an MS/MS method that provided increased
selectivity such that ultimate chromatographic resolution was not necessary. All
chromatographic experiments were therefore performed using DB-l even though

other phases provide better chromatographic resolution of PCB congeners.

The gas chromatograph was equipped with a 30 meter x 0.250 mm column of 0.25
mm DB-1 phase (methyl silicone, J & W Scientific). The GC was temperature
programmed—40 Cfor twominutes, 40 C to 160 C at20 Cper minute, 160 Cto
270 C at 5 C per minute and held at 270 C for seven minutes. Helium was used as
the GC carrier gas with a column head pressure of 22.5 psi which corresponds to a
linear flow of approximately 45 cm/second. Splitless injection was used with the split
valve closed for the initial two minutes and open for thirty minutes. The injectorwas
held at 270 C and the transfer line held at 275 C. The effluent end of the column was
placed directly into the ion source. These conditions corresponded to those used by
collaborators in other laboratories with other modes of detection (MS and ECD).

Under these conditions, the PCB congeners of interest eluted from 10 to 25 minutes

79
after injection.

Generally, the split valve is held closed as long as possible in order to place more
analyte on column. The trade-off is that with longer times, excessive tailing of the
solvent peak may interfere with the elution of the analyte. Using negative ion MS,
the solvents generally used (methanol or hexane) do not exhibit a response, and
therefore, solvent tailing is not an issue. However, the ECNI plasma used for
ionization is affected by the solvent front. When monitored, the solvent typically
suppresses the background and is observed as a ”negative" peak. As such, it is still
important for this method to open the split valve and sweep the solvent out of the

injector but with a minimum of analyte loss.

Mass Spectrometry. In addition tothe instrumental parameters discussed in Chapter
4, collision energy, collision gas pressure and translational kinetic energy correction,
all other instrumental parameters were studied for the purpose of improving
precursor and product ion trannission and detection. These involved those
instrumental parameters that control both MS and MS/MS tunes and those that
control the scan functions of the instrument in both full scan and single or multiple

ion detection modes of acquisition.

Instrumental Time. The voltages (RF and DC) of each element of the analyzer of the
T80 instrument are controlled by link tables. Each link table is specific for a given

analyzer element for the given scan mode (Q 1, Q3 or MS/MS) and ion polarity and

80

may consist from one to several tune points across the scan range. As the instrument
is scanned from low to high mass, the voltage of each analyzer element is varied
according to its specific link table. Also, with the T80, the instrument must be
cah’brated prior to acquisition. This is accomplished by setting the 01 or Q3
cah’bration link table with the exact masses of the tuning compound
(perfluorotributylamine -— PFI'BA or FC43). For the current work, at least two
instrument tunes were generated and used. One was generated at unit (baseline)
resolution for MS acquisitions and the other an "open-resolution" tune for MS/MS

experiments.

The resolution of a quadrupole mass filter is determined from the DC-to-AC ratio
of the voltage placed on the quadrupole assembly. Thus, there is a compromise
between resolution and tranmission. This ratio was set to achieve baseline
resolution (<5 % valley between a peak and its isotope). The DC offset of the
filtering quadrupole was set to achieve good Gaussian peak shapes. The DC offsets
of the filtering quadrupoles will affect the resolution, and as such, both parameters
must be set iteratively. Th'n also involves a compromise with sensitivity, but good
peak shape is of paramount importance with respect to centroiding and tune and
cah'bration stability. All other parameters, lenses and RF powers of the non-filtering
quadrupole(s) and collision cell, were set for maximum transmission. In some cases,
there was a slight compromise between sensitivity and stability (such as the middle
lenses of the Einsel lens assemblies). In these cases, stability was always chosen. By

using these criteria, instrumental tunes were generated that were stable with respect

81

to cah'bration, resolution and sensitivity over several days and even weeks.

Signal Optimization. When using the T80 in an MS/MS mode, two mass filters are
used in line. As such, ahnost 99% of all ions generated in the source are filtered out
in order to achieve unit mass resolution in both (01 and QB) mass filters. Thus, it
is advantageous to enhance the number of ions striking the detector without
sacrificing selectivity. In an MS/MS experiment, one mass filter, typically Q 1, is used
for a set mass — the quadrupole is maintained at a specific DC-to-AC ratio. The
resolution may be ”opened up" to a significant degree, however, but such that still
only ion current at one m/z value is transmitted (Figure 5. 1). PRES and DRES are
the instrumental parameters linked to mass that are used to set the resolution of the
instrument. Also, typically with a baseline-resolved MS tune, ions are slightly over-
resolved in an MS/MS mode due to over-filtering. As such, in MS/MS the resolution
may be opened up slightly thus improving sensitivity but still achieving baseline

resolution.

For full scan MS/MS acquisitions, the resolution of the precursor quadrupole (01)
was adjusted to approximately 20% valley and the product analyzer (QB) from 1%
to %5 valley. This offered a factor of two improvement in sensitivity as measured
with the PFTBA tuning compound. For selected reaction monitoring experiments,
the resolution of both filters was further adjusted such that three to five times the
ion current was observed over that with the MS tune. In both cases, selectivity for

ion current at each m/z value was preserved. Also, the quadrupole offsets were

Set Mass Window

‘- Set Mass Window
Fixed at l DAC Step

9 M E M+l

R E Basellneresolutlon

E """""""" . l """"""""""""""

s 53 /\

/ ii /

D 5; / \

R :: \

E 55/ \ o

S , .............. £31 .................... R gets".
: l

l "‘ \

/
/
/

 

--------------T§1

1--------------

O /

 

mass

Figure 5.1. The relationship of an ion and its isotope indicating the significance of
using the exact mass for a set mass in an MS/MS experiment and the effect on
selectivity of under-resolving the instrument.

83
adjusted to maintain proper peak shapes. In all cases, after adjusting quadrupole

resolution, the instrument was recalibrated.

Full Scan Acquisition In many MS/MS experiments, the products are the result of
collisionally-induced dissociation (CID), and as such, the product analyzer is
scanned from low mass to a mass near (above or below) the precursor mass. Since
this mass range can be up to several hundred u, scan times must be at least several
hundred mini-seconds in order to sample the mass spectral data adequately.
Typically, in order to achieve precision at the 95% confidence level with respect to
peak height and area, at least 10 samples per peak are required. The TSQ-700 has
a limit of approximately 12,000 samples per second. With chromatographic peaks of
2 to 3 seconds, the compromise is therefore between mass spectral and
chromatographic sampling. Since the current work involved monitoring only a single
reaction product at a time, all quantitative work was done in selected reaction mode
(SRM) at relatively short scan times, and the compromise in sampling was not an

issue.

Full scan acquisition was used to generate MS precursor spectra but not for
quantitative purposes. As such, mass spectral sampling was always maintained at a
minimum of 10 samples per ion. This resulted in only a few samples per
chromatographicpeak which were averaged for qualitative purposes. These spectra
were used to determine the exact mass of the molecular anions to be used as

percursors for the MS/MS experiments. While the exact precursor masses could be

84
and were calculated, us-g the precursor masses as observed allowed for daily drifts

in cah‘bration. Also, full-scan MS data provided information on precursor ions that

could interfere in the selectivity of the ion-molecule reactions being studied.

As previously stated, only odd-electron molecular anions will undergo the exchange
of oxygen for chlorine in an MS/MS experiment. Since all of the tetra, penta, hexa
and heptachlorobiphenyls produce this anion in the MS source, all of the coplanar
congeners of interest for this work are amenable to this reaction. Most congeners
exhibit ions in the source that are the result of the loss of chlorine (loss of 35u).
These ions are not amenable to the exchange reaction, however. The specificity of
this reaction for PCB congeners with a given amount of chlorine substitution is
challenged only when the formation of two species occurs in the source. As
mentioned previously, when a radical molecular anion loses a chlorine atom in the
source, the resultant ion is an even-electron species that will not undergo chlorine
exchange with oxygen. If this ion, however, gains or loses a hydrogen atom, whether
concerted with the loss of the chlorine atom or not, before being extracted from the
source, the product is once again an odd-electron species that will undergo
oxygen-chlorine exchange for a net loss of 19 mass units. This is illustrated in Figures
5.2 through 5.5, where the primary mass spectra of tetra- through
heptachlorobiphenyls show that there is a net loss of 35 mass units (11) from the
molecular anion for the tetrachlorobiphenyl congener and a net loss of 34 u for the
homologs of pentachloro- and greater substitution. This was found to be true for all

twelve of the coplanar congeners and was not affected by varying instrumental

85

parameters such as temperature and CI moderating gas pressure. Due to these
regions of the spectra not exhibiting typical isotopic patterns for polychlorinated

species, there is apparently also some loss of 35 and 36 mass units.

While this reaction is specific for a given amount of chlorine substitution, it may not
be totally specific for a given PCB isomer class. It was thought that the problem of
a net loss of 34u in the source from a molecular anion might be overcome with the
appropriate choice of moderating gas, since the source of the attached hydrogen
atom might be due to the moderating gas. By choosing a gas that contains no
hydrogen (such as argon), this interference might be eliminated. However, when
either argon or deuterated ammonia was used as a moderating gas, the net loss of
3411 was still observed. This demonstrated that the source of the H was not the
moderating gas. The loss of HCl from a molecular anion would still be an
interference, but would only be a problem if the combination of abundance and
response to this reaction exceeded a few percent of that of a co-eluting congener of

interest.

Even though all twelve coplanar congeners exhibit a net loss of 34u in the source, the
specificity of this method is compromised only if a congener that co-elutes with one
of these twelve also loses 34u in the source. When four of these co-elutants (Table
5. 1) were mass analyzed, three exhibited losses of 35u and one showed a loss of 34u
as shown for three of the four congeners in Figures 56-58. This is consistent with

results reported by others in that some congeners lose 34u in the source and others

 

Flirts l 1

 

 

 

 

 

 

 

 

 

 

 

 

100a ”3 [:3
eo- 29?
C1 C1
>3
g so~ 4a
3 ' ‘1’
E: l:
a, 5
>
‘33 S:
a 0
'—g 40" H
Q
m 25.5 131
zo-
2 e 211
i ' 29.5
3”.
I ass 21;
2 5 7s 9 2 397
311 . ”Lethe! ILL PCP ,!:+ l!.
25's 25's r 2% sic are zero

 

 

 

mlz

Figure 5.2. The ECNI mm spectrum of a tetrachlorobiphenyl indicating the loss of
35u in the ion source.

87

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FF“ 1 3“ 1. 3. ,
recs s.1
sod
C C1

5? , 321, ape

on +3

a so- Cl 5
+5 292 E
H 290 I :3

g 0
e“ a
g o
r-i ro~ H

Q)
a:

2;;
also
zo-l ' so?
I”
391
2“ s 309
' P231 313
i 3231 332
220 14 s 9 333 339
-* 1| "'1 is: l l-‘ 133L311"! -LL ..... [r
are ''''' 290 {Jo 31'; 22's no 340

 

 

m/z

figure 5.3. The ECNI mass spectrum of a pentachlorobiphenyl indicating the loss
of 34u in the ion source.

 

 

 

 

 

 

 

[he I j 2..
100‘ see Fins
C1 2s:

.0. 33‘
is
u-r
m +3
3 ‘°‘ 5
+5 33‘. 15. a
H ’34 :3
g 0
a“ a
t; o
.—o “l" I—t
q) Jfl

2nd

:10 3'41
13 343
i ' ass
:12 3'45 '
3 41 o 3 7 372 315
-o- - “fan-.1 ltr'ifl'm"l Li}: J. .......

 

 

 

 

 

 

 

 

 

 

 

 

T I ' r r
320 no .140 3‘0 360 170 no

m/z

figure 5.4. The ECNI mass spectrum of a hexachlorobiphenyl indicating the loss of
34u in the ion source.

89

 

 

 

 

 

 

 

 

 

 

 

 

 

I fl lot
100' ' 394 l'
ass 1.0
C1 C1 C1
o o .
C1 C1
5‘
ea 8
8 se- :1
OJ
H
s g
g 0
our 5
*5; o
l—i 40" H
Q
m
360
I
re:
30‘ 351 clue
ass
I are I 102
1311 39 3 o 4
’“wl I 1:- .1 ifi 11.12.1111.
“a vvvvvv mi ' .1? r aria ' 330 we h no ‘ do

 

 

m/z

Figure 5.5. The ECNI mass spectrum of a heptachlorobiphenyl indicating the loss
of 34u in the ion source.

90

Table 5.1. The structure of the twelve PCB congeners that have been shown to
exhibit dioxin-like toxicity and those congeners that co-elute. Relative retention

times (RRT) are with respect to octachloronaphthalene ( 77).

TWELVE'I‘OXIC CO-ELUT
CONGENERS

345,4—CB

2356, 4-CB
345,26-CB
7346,26-CB

.6149

 

236,34-CB

 

236,245-CB

 

 

 

 

 

 

 

 

 

 

 

 

 

91

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100T 32‘ ":9?
C1 C1 C1
C1
32? ape
'5 +3
c: ‘° :I
3 i3
*3 s
Q)
.2 c:
H 2n 0
'2 «Pl I—r
Q)
m 307
2.; p2 '
3
313 05.] 3'30
20«
P4 so;
as
as
i as 31.1 333
2 - e
31.1.1 l1: l l 1111:1311 -13:
200 ' m :30 31's ‘ are ' 2:10 ' 24's

 

m/z

Flare 5.6. TheECNTmassspectrum of a pentachlorobiphenylwhich co-elutes with
a coplanar congener in a GC analysis. This congener exhibits the loss of 35u in the
source as opposed to the loss of 34a observed with the coplanar congeners.

 

 

 

 

J." -0 C.
10°] F ‘ . 92
C1 C1 C1 C1
. © ©
'°‘ or
e a a
‘0‘!
=3 r:
3 3511 Q)
:l l:
H 325 S
g ' O
mi :3
t; o
«r l—t
F" ass
0
m 221
an
33.1 I
as
zo- 33? l as;
32:
as
ass
Po at;
l 131 an I I lug e 51 31s
2 - L'Jttg- LJ'JJ --f ..... 1.1. ...... L"
a2? aaro . as'o ago see an arc

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

figure 5.7. The ECNI mass spectrum of a hexachlorobiphenyl which co-elutes with
a coplanar congener in a GC analysis. This congener exhibits the loss of 35u in the
source as opposed to the loss of 34u observed with the coplanar congeners.

93

394 to us

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

too- '1.as
ass
Cl CIC Cl
.. o o ..

C1
is
.8 a
s “‘ s
4.2
c: E
H are 5
g as} O
efi :
t" o
g 402 l—l
d)
n:

aso
an
and ars
l 490
3&3. 3:17
364 373.
35.7.. are l
s sqz
ass a 4 o r
’i’| 11..-. [13:34 llifmiiw
«'8 """ um ""fifi?" 736' ' «in «us do aria

 

m/z

figure 5.8. The ECNI mass spectrum of a heptachlorobiphenyl which co-elutes with
a coplanar congener in a GC analysis. This congener exhibits the loss of 3411 in the
source, and as such, interferes in a GC/MS/MS analysis of coplanar congeners.

94
lose 35u (23). When these congeners are analyzed in a GC/MS/MS experiment by

monitoring the products of reactions with oxygen, the ions that result from a loss of
chlorine in the source do not react while those ions that result from a net loss of 3411
in the source do react. These phenomena are represented in the results shown in
Figure 5.9. As is shown, the heptachlorobiphenyl, which loses 34u in the source,
exhibits the products of several reactions. Each ion from a subsequent loss of
chlorine and gain of hydrogen readily undergoes reaction with molecular oxygen.
The penta- and hexachlorobiphenyl con geners, however, produce only odd-electron
molecular anions and even-electron fragmentation products in the source, and as

such, only undergo one reaction.

Selected Reaction Monitoring. In MS/MS selected reaction monitoring (SRM) is
analogous to selected ion monitoring (SIM) in MS. These modes of detection are
generally used for quantitation with chromatographic experiments. For MS analyses,
SIM provides a significant improvement in signal-to-background with a loss in
selectivity. For MS/MS, either full-scan or SRM, a reduction in selectivity may not

always be the case since the MS/MS reaction may be specific for a given analyte.

Using an Aroclor 1242 standard, the oxygen-for-chlorine exchange reaction was
monitored for all 10 homologous series in separate experiments (Figures 5. 10-5. 16).
No reactions were observed for the mono-, dichloro- and decachlorobiphenyls and
very little for the trichlorobiphenyls. This also supports the premise that these

congeners form ahnost exclusively [M-HI in the ion source, and being even-electron

95

 

 

 

 

 

 

 

 

- “dim m 'm
“1 use
H. 1 t. I h J i i l i ' i. in. ' l LIFI :11" I
,ma
1.!!!
29
ea . a
d) 1.187 Q)
a E
Ci
H a
d)
> m m C;
s ~ A - - - .. .
fl ”‘8’” _ _ Pm H
é: A1 “a.
‘3 .ms
1.357
5
utfireVfi'Vfi'[rvvvrvvvv'vvvvlrvrv'fv—fi'vevrrrtvv'evvavrve
I“ 300 300 “I ”I m

Scan Number

fiyrre 5.9. Mass chromatograms at m/z 273, 307, 341 and 375 and a reconstructed
total ion chromatogram (RIC) that represent the analysis by GC/MS/MS of four
PCB congeners that co-elute with coplanar congeners.

 

 

 
 
 
 
 

 

 

 

 

 

 

 

me
till
19m
ii:
iii
.33
E W 2222 E
3 "f7 2119 19:45 5
g i'l" 1111 ' o
'5‘ «1 "99 ms g
.53 1mg ma 14" '-'
g ' ' 21:24
l
2m 2533
2% mg 2476 .
S 1 “we.“
""1 It trTrTrrltrr

ll" 21" 1m 1m Nil 2m 2600

Scan Number

figure 5.10. The mass and reconstructed total ion current (RIC) chromatograms
from a GC/MS/MS experiment where the exchange reaction of oxygen for chlorine
was monitored for trichlorobiphenyls in an Aroclor 1242 standard.

 

 

 

m -
us:
me .
till I illii
ism

ll '
.1?
3 s
3 it. 2
3 - i s
3 l o
"i l ‘ s
.99. it Hi ' 2519 H
3° 21:36 . mu

m ' ~ '
ii 21: 1 159! ill! .
1 ms. 1 497 zmmzzr 2‘92 2751
21:11 :lilllms ' 23:55 24m
3" ' 233121139. .1
.. L - ” ‘

 

 

23" 23", till llil 2310 iii! liii 163i Nil Nil lil

Scan Numb er

figure 5.11. The mass and reconstructed total ion current (RIC) chromatograms
from a GC/MS/MS experiment where the exchange reaction of oxygen for chlorine
was monitored for tetrachlorobiphenyls in an Aroclor 1242 standard.

98

 

  
       
     
   

 

 

 

lit
zm
ms:
till 3"“
1.197
ll
.3:
2 2'5
3 ll é’
*5 s
d)
.3 . ' - c:
3 w , 2759 ,2.
g .26" mi 24:31
"was ugly ' .
1" 21:3. , 267i (:11 2862
m 12:52: ' n: . 25:26
2r 15“ 2m 1'“ use
A um 23:)
' Ill lrli T

 

 

r1]
llil llil 222i 2552 262i 2622 llll 275i 2222 222i 2922

Sean Number

figure 5.12. The mass and reconstructed total ion current (RIC) chromatograms
from a GC/MS/MS experiment where the exchange reaction of oxygen for chlorine
was monitored for pentachlorobiphenyls in an Aroclor 1242 standard. .

 

 

 

 

 

  

ill
2766 .
ll .
. 1 5‘ lElll
ill . . 2
2m . .. it.”
thli
Ii'
é’
E " *8
3 ll- ' $3
15. s
E 3
til it 2729 , -' .2
'33 266 21:36 .
m 22.61 i "92
2t
2925
'63:!
..

 

 

 

26" 262i .27" 2162' 2m 2262 2!" 2m 2222

Scan Number

fiyrre 5.13. The mass and reconstructed total ion current (RIC) chromatograms
from a GC/MS/MS experiment where the exchange reaction of oxygen for chlorine
was monitored for hexachlorobiphenyls in an Aroclor 1242 standard.

100

 

 

 

 

 

RIC
m:
25:20 it '5
[00- . *
1966 2.50:
16:26

8t
.1?
U) +1

:1
g at o
c: I:1
I-t , 5
g 0
.... s:
E W - . .2
§ 102 ‘
23:3:
204
2773 -
24m 317
' . | 20:19!
0. 1 r I

 

 

 

Nil 275! mu 2m 2m 2m am am am am am

Scan Number

figure 5.14. The mass and reconstructed total ion current (RIC) chromatograms
from a GC/MS/MS experiment where the exchange reaction of oxygen for chlorine
was monitored for heptachlorobiphenyls in an Aroclor 1242 standard.

101

 

 

 

 

 

 

 

 

 

 

 

 

m ,
am
or - 3M5
' 3931 1.927
at
.1?
U) +3
5 w 5
*a‘ t:
H =3
a a:
05
cs 40‘ +3
'5 29." am
(:3 I
w 937 2995 3133 fl
1 I -
A A I A u 32'”
'TllrlT—UTIII [I‘Tml—ITIUII IIIUTI1II IUITITT

2m 285' 2m 29“ am 3050 am am 3m 32“ ll"

Scan Number

Figure 5.15. The mass and reconstructed total ion current (RIC) chromatograms
from a GC/MS/MS experiment where the exchange reaction of oxygen for chlorine
was monitored for octachlorobiphenyls in an Aroclor 1242 standard.

102

 

 

am
am
m- 29.94 3““
3.316
W
a; am a
a 27:5; a
3 at ' a»
c: i:
"" {3'
d)
>
2:3 5
.53 ll+ r-4
a?
ma
2r 9..“
. l
TTl—rlml‘l—lrl l TI—Tl—rljlilrrll'llr

 

 

 

 

 

I
am 3059 31" 315i am 3259 .3390 3350

Scan Number

figure 5.16. The mass and reconstructed total ion current (RIC) chromatograms
from a GC/MS/MS experiment where the exchange reaction of oxygen for chlorine
was monitored for nanochlorobiphenyls in an Aroclor 1242 standard.

103

anions, do not undergo this reaction. In addition, the presence of the three
nanochlorobiphenyls is observed with a substantial amount of reaction product.
These congeners are typically not able to be detected, much less quantitated, using

ECD or MS modes of detection with this Aroclor.

Since the two choices in MS/MS are using each analyzer to either select a specific
ion (set mass) or to scan, there are four modes of MS/MS — product scans,
precursor scans, constant neutral loss (or gain) scans and SRM. The TSQ-700 is
under total digital control, and as a result, unlike many analog-controlled
instruments, the instrument must be scanned in order to function. Thus, for SRM
experiments, the T‘SQ-700 is operated in one of the other three modes but scanned
over a very narrow mass range. This narrow mass range scanning may or may not
actually involve analyzer voltages changing depending on the width of the scan

window and the resolution of the DAC control of the voltages.

In order to determine which mode of MS/MS scanning -— product, precursor or
neutral loss — offered best sensitivity and precision, several scans of each mode
were acquired for the given oxygen-chlorine exchange reactions. In all cases,
regardless of the instrument tune used, the product scan mode was shown to exhibit
maximum sensitivity (Figure 5. 1 7) . There is no inherent reason that one mode should
necessarily be more sensitive than the other two for SRM experiments. However, in
ahnost all cases, the product analyzer (Q3) on a TSQ-700 is more sensitive than the

precursor quadrupole (Q 1). This is generally due to the better quadrupole rod

104

 

 

 

 

RIC iitll
1 ill '1 51,
Product Scan

llJ
£9
'8 59+ 4.:
‘3 c:
a) a:
*a‘ s
H :3
a: (L)
>
£3 49* 8
—~ W ""
a:
CI:

I?reu:1111scxr'53cn311

29‘

fl

' "IllillIIUIIIIIIIIIIIIIITIIIIIIIIIIIIUITWTI'IITTIIIITIIUlliilltliIIIIIIIIIIWU

 

500 1800 158% 2935 2589 399% 3569 4086

Scan Number

figure 5.17. The reconstructed total ion current (RIC) chromatogram of a
GC/MS/MS experiment using product and precursor scanning of the instrument for
the same reaction indicating the product scan mode to be the more sensitive.

105
assembly being used as Q3. This is desired since most MS/MS experiments are
product scans, and a better analyzer is required for scanning than for a set mass

filter. Also, Q3 is always less contaminated than 01.

Since product scans were used for SRM, it was necessary to select the size of the Q3
scan window. Generally, to a limit, as the size of the window is narrowed, an
enhancement in signal to background is observed. The limit is a function of the
stability of the instrument cah'bration. The specifications of this stability for the
T’SQ-700 is :t 10 mmu over 8 hours. This is also supported from experiments that
showed best signal-to-background when the scan window was maintained from 50
to 100 mu. Smaller windows resulted in a significant increase in signal noise due
to scan-to-scan variation in the observed mass of the peak. Due to the method that
is used by the T‘SQ-700 to sample and sum to registers the data, the apparent signal
actually decreases as the scan window is narrowed. Also, the stability of the
instrument cah'bration required that the precursor set mass be defined to the nearest

10 mmu.

The other variable studied was the SRM scan time. On a TSQ-700, the scan time for
SRM is the actual time spent sampling ion current. The inter-scan time is then the
sum of the scan time plus the voltage settling time plus the time of the electronic
overhead. As stated previously, with SRM very rapid scan cycling is possible such
that the compromise between chromatographic sampling and MS sampling is not an

issue. Using scan times of 10 to 50 ms and voltage settling times of 5 to 10 ms

106
provided optimum signal to background with adequate sampling of both even when

cycling rapidly between reaction products. Faster scan times require the voltage
settling to be increased to eliminate memory or carry-over in the signal such that the
overall duty cycle suffers. Also, with very rapid scanning, ion flight time corrections
become an issue. The capability exists to correct for this, but this is only necessary
when chromatographic sampling is an issue in SRM such that extremely rapid

scanning of the instrument is necessary.

Quantitation. Many PCB analyses are performed chromatographically using
octachloronaphthalene as an internal standard. For ECD or MS routines,
octachloronaphthalene is a reasonable choice since it is chromatographically
resolvable from PCB congeners and exhibits a response factor of the same order of
magnitude as that of the PCB congeners. This or any single internal standard may
not be the best choice, however, for the oxygen-chlorine exchange reaction. While
octachloronaphthalene will undergo chlorine exchange with oxygen, response
factors for PCB congeners vary more than two orders of magnitude with respect to

this internal standard.

Relative Response Factors. The relative response factors with respect to
octachloronaphthalene for several congeners that undergo chlorine exchange with
oxygen have been determined, and these vary from approximately one to five

hundred (Table 5. 2).

107
Table 5.2. The response factors of several PCB congeners with respect to

octachloronaphthalene for a GC/MS/MS experiment where the products of the
reaction of molecular ions with oxygen are monitored.

“ com “
7'7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

34,34—CB 1.6
85 234,24-CB 290
105 234,34-CB 18
109 236,34-CB 88
118 245,34—CB 11
126 345,34-CB 14
132 W 94
149 236,245-CB 190
153 245,245-CB 130
156 2345,34—CB 14
169 345,345-CB 6.0
176 2346,236-CB 48
182 2346,2450 500
197 2346,2346-CB 130
199 2345,356-CB 580
205 23456,345-CB 160
m6 23456.2345-CB 230
207 23456.346-CB 220
208 23456.2356—CB 120

 

 

108
Internal Standards. Since the response factors vary dramatically with the degree and

position of chlorine substitution, it was of interest to explore using other or multiple
internal standards for these analyses. It was desired to find at least one chlorinated
aromatic compound that undergoes this reaction, does not co-elute with any of the
PCB congeners of interest, and exhibits a response to this reaction, two to three

orders of magnitude greater than that of octachloronaphthalene.

Chlom' ated Angggs. The first species chosen as an additional internal standard was
pentachlorobenzene. This standard elutes in a PCB chromatographic analysis
between the solvent (hexane) and the monochlorobiphenyls, and undergoes the
oxygen-chlorine reaction with a response approximately 500 times that of
octachloronaphthalene. Thus, this standard meets all of the desired criteria and

would seem to be an ideal choice. However, on both of the columns generally used

for PCB analyses

 

5% phenyl and 5% phenyl/1% vinyl methyl silicones ——
pentachlorobenzene exhibits excessive tailing for many GC temperature programs.
This was thought to be the result of some condensation phenomenon occurring in
the injection port or transfer line; and some improvement in peak shape was
observed when the temperatures of the injection port and transfer line were both
increased to 300C. The tailing was not decreased sufficiently, though, to allow for
reproducible responses to be observed. Since this is the upper temperature limit of
these stationary phases, increases in temperature were not feasible. By using lower
initial temperatures, tailing was reduced significantly, but extended analysis times

were necessary. Thiswould indicate that instead of condensation occurring, thermal

109

breakdown of the molecule was occurring in the injection port.

The next molecule chosen as a candidate for use as an additional internal standard
for a multiple internal standard scheme was aldrin. This is a hexachloro-, tetracyclic
compound that meets all of the aforementioned desirable chromatographic and
response properties. In addition, under the conditions favorable for PCB analyses,
chromatographic peak shapes are acceptable. Unfortunately, this molecule for the
most part undergoes dissociative electron capture ionization, and as such, not
enough [M]’ ' is produced to allow for its use as an internal standard. Other
chlorinated species examined were dieldrin, heptachlor, and heptachlor epoxide.

These species also were unusable due to the same problems associated with aldrin.

@tgm Dilution. Many MS methods of analyses for PCBs utilize isotopic dilution
as a means of providing for an internal standard for quantitation (24). It was decided
that this would also be the best choice for this method. Since the response of most
congeners to the oxygen/chlorine exchange reaction generally increases with the
amount of chlorine substitution, three 13C-labelled congeners (34,34-CB, 345,34-CB
and 345,345-CB) were chosen to be used as internal standards. The responses of the
twelve coplanar congeners with respect to these labelled standards are listed in

Table 5.3.

Summary. The aforementioned experiments resulted in the following observations

for setting instrumental parameters to observe the oxygen-chlorine exchange

110

Table 5.3. The responses of the twelve coplanar PCB congeners with respect to the
three ”C-labelled standards.

Relative Response Factors
IUPAC # Congener ”C477 1’C-#126 1’C—#169

 

 

 

 

 

 

 

 

 

 

 

 

81 345,4-CB 1.7 0.15 0.097
77 34,34—CB 0.82 0.087 0.055
123 345,24-CB 7.7 0.77 0.48
118 245,34~CB 7.1 0.71 0.45
114 2345,4vCB 8.1 0.77 0.49
105 234,34-CB 7.7 0.76 0.48
126 345,34-CB 11 1.1 0.72
167 245,345oCB 55 5.5 3.5
156 2345,34-CB 56 55 3.5
157 234,345-CB 62 6.1 3.8
169 345,345-CB 15 1.5 0.95
189 2345,345-CB 56 5.5 3.5

 

 

 

 

111

reaction as a chromatographic detection scheme for quantitative purposes:

0 The mass resolution may be decreased with the instrument tune offering an

improvement in signal but still allowing for mass specificity.

0 Product scanning is the more sensitive scan mode for SRM.

O Scanwindowsin SRM of 50to 100 mu offer optimum signaltobackground.

0 Scan times of 10 to 50 ms allow for adequate chromatographic and mass

spectral sampling.

0 Voltage settling times of 5 to 10 ms allow for the elimination of memory

effects with a minimum compromise in duty cycle.

0 Using isotopically-labelled PCBs as internal standards met all the desirable
characteristics of internal standards. The only disadvantage as a general

method may be in some cases price and availability.

Chapter 6

Application of the MS/MS Method using the Oxygen-Chlorine
Exchange Reaction for the Analysis of Polychlorinated Biphenyls in

River Sediment Samples

The analytical method using the oxygen-chlorine exchange of PCBs was applied to
the quantitative analysis of river sediment samples. These samples were studied for
the purpose of determining the amount of dechlorination that had occurred under
anaerobic experimental conditions. The dioxin-like toxicity of these samples was
determined for those samples that had been innoculated with dechlorinating
microbes and those that had been autoclaved. The toxicity levels were calculated
based on the concentrations of the twelve PCB congeners that had been shown to
exhibit dioxin-like toxicity. This method was used since the dioxin-like toxicity of
PCB-containing samples had been shown to derive ahnost solely from these

congeners (100).

Introduction. Those parameters that were studied for the application of this method

to the analysis of PCB-containing sediment samples included the preparation and

112

113
analysis of PCB standards using 13C-labeled internal standards, the precision and

accuracy of the method and the presence of possible analytical interferences. The
quantitative results for several sediment samples of varying sources are also
presented. As a preface to this discussion, the experimental parameters used for

sample preparation and instrumental analysis are provided.

Experimental. The Aroclor standards were provided by John Quensen, and all other
individual congeners and standards were provided by George Frame. All standards
were prepared by dilution with Organic Residue Analysis grade hexane (J .T. Baker).
River sediment samples spiked with Aroclor 1242 (ppm levels) were extracted with
50/50 acetone/hexane. Each extract was subjected to Florisil clean-up and then
shaken with mercury for sulfur removal and finally concentrated to ten milliliters.
While these clean-up procedures are not necessary for sample analysis by
GC/MS/MS, this was done such that these samples could also be analyzed by
W.

Gas Chromatography. All gas chromatographic methods were performed with a
Varian 3400 gas chromatograph equipped with a 30 meter x 0.250 mm column of
0.25 mm DB-l phase (methyl silicone, J 8: W Scientific). The GC was temperature
programmed—40 Cfortwominutes, 40C to 160 C at20Cperminute, 160 Cto
270 C at 5 C per minute and held at 270 C for seven minutes. Helium was used as
the GC carrier gas. The Varian 3400 was equipped with an injector body that

accommodates both split and splitless modes of injection. All experiments were

114
performed using the splitless mode of injection with the split valve closed for the

initial two minutes and open for thirty minutes. The injector was held at 270 C and

the transfer line held at 275 C.

Mass Spectrometry. Early MS and MS/MS experiments were done with a Finnigan
"ISO-70B mass spectrometer operated in a negative ion mode. Later, the T80
instrument was upgraded to a T‘SQ-700. The pertinent changes included the
installation of an octapole as a collision cell (the TSQ-70 instruments use a
quadrupole) and replacing the PDP-l U73 data system with a DEC Station.
Ammonia was used as the chemical ionization reagent gas. The CI reagent gas
pressure was set to approximately 9500 mTorr as measured by the source
Convectron gauge and was optimized daily using perfluorotributylamine. Pure
oxygen was introduced into the collision cell. The source was held at 150 C and the
manifold at 70 C. All reactions were generated using electron energies of 50 eV,
collision offset energies of 1-2 eV (laboratory), and collision pressures from 0.5

mTorr to 1.5 mTorr.

Usingperfluorotributylamine (PFIBA),two instrumentaltuneswere generated, one
at unit resolution and one with open resolution. The open resolution tune was
generated with 01 being approximately twice as under-resolved (percent valley) as
Q3. All MS experiments were performed in the QlMS mode using the baseline-

resolved tune, and all MS/MS experiments performed with the open-resolution tune.

115
The instrument utilizes a 20-kV conversion dynode that may be set from 3-20 kV,

and an electron multiplier that may be set from 0-3000 volts. For ions less than
approximately 200 u, the gain in the observed signal as the dynode is increased
becomes non-linear. For the ions of interest for the current work (> 250 11), though,
increases in dynode and/or multiplier voltage causes a significant increase in the
number of ADC counts observed as the ion current. In order to provide optimum
precision for the analytical method, these voltages are set for maximum signal for
the most concentrated standard but such as not to saturate the data register
(268,000,000 counts). Using PFTBA anions, in separate experiments the dynode and
multiplier voltages were varied and the relative standard deviation of the observed
signal plotted (Figures 6. 1 and 6.2). It was determined that best precision occurred
when the conversion dynode was operated from 8 to 18 kV and the electron
multiplier from 1000 to 2000 volts. As such, for the current work, the dynode was set

to 10 kV and the multiplier to 1500 volts.

Quantitation. Using the oxygen-chlorine exchange reaction as an analytical method
involved setting all parameters affecting the scan of the instrument. Scan times and
scan ranges were set to accommodate the simultaneous monitoring of up to four
specific reactions at one time but still allow for adequate chromatographic sampling.
The voltage settling time was set to 6 milliseconds to prevent unwanted memory
effects. For this method, lowest detection limits, best precision and good
chromatographic sampling occurred when the instrumentwas operated in a product

scan mode of operation with a 2n scan range (see below) about the chosen product

116

Conversion Dynode Voltage vs TIC

 

 

 

 

 

2396.62
621819.
R D
S TIC
Std Dev
20525.41
531.89
0.36 / '\
an 3.00" 20.00'
Conversion Dynode (kV)

Figure 6.1. The plot of the ion current, standard deviation and relative standard
deviation as the conversion dynode voltage was varied from 3 to 20 kV.

117

Electron Multiplier Voltage vs TIC

 

 

 

34.15 ”Rage
40755 .00 /\
00334 .00
RSD Std Dev
\
\
TIC

0.00
0.00
0.00 /

 

 

500 Electron Multiplier (volts) 2500

figure 6.2. The plot of the ion current, standard deviation and relative standard
deviation as the electron multiplier voltage was varied from 500 to 2500 volts.

118

and a scan rate of 20 scans per second. This allowed for one PCB reaction product
to be monitored at any given time, and quantitation was achieved by using peak
areas of mass chromatograms. The product chosen could be changed over the course
of a chromatographic run allowing for the analysis of all the congeners of interest
in a single GC run. Since there are never more than three PCB congeners of
differing degrees of chlorination eluting in given region of the chromatograms
produced with the stationary phase used, all possible PCB exchange reactions could
be monitored in a single GC run by monitoring three reactions in each scan. For
quantitation, four reactions were monitored to include the 13C-labelled internal
standards. This raised detection limits by less than an order of magnitude, which may
or may notbe detrimental in the analysis of the coplanar congeners in some samples.

For the coplanar congeners, detection limits ranged from low to mid pg/ul levek.

For the analysis of the sediment samples, only the coplanar congeners and the ”C-
labeled analogs of three of these were of interest. Therefore, for this specific
method, only two reactions were monitored at one time. The reactions monitored
were changed according to chromatographic retention time to accommodate all four

coplanar homologs and 13C-labelled analogs of three of these homologs.

Precision. The precision of such a selective reaction method is influenced by several
factors. It was apparent early in the development of this method that gas flows
and/or pressures must be maintained in a relatively steady state over fairly long

periods of times. Thus, great care was exerted in setting and monitoring these

119
parameters. The valves controlling both the moderating and the CID gases were

replaced with fine high precision metering valves. Even so, in order to achieve the

best precision, the use of gas flow controllers may be required.

In addition to constant gas flows, the instrument needs to be tuned in such a fashion
that the tune does not drift significantly over a period of several hours or even days.
Th'm requires that the instrument (source, quadrupoles and lenses) be very clean.
Also, the voltages and parameters used to tune the instrument were set not for
maximum transmission, but for best stability over long periods of time. This was
particularly important since this was a chromatographic experiment, and injections
were made no more frequently than every 45 minutes. Without the availability of an
autosampler, a typical analysis took up to three days for all injections including
samples, standards and blanks. This was the reason that, even with a good stable
instrument tune, scan windows for SRM were maintained at 2n. Better signal-to-
noise could be achieved with naller scan windows but only if the instrument was

recahhrated daily which was counterproductive.

Interferences. Many of the inherent problems with ECD and MS modes of
chromatographic detection for the analysis of PCB-containing samples are
eliminated by lung this MS/MS method. Much of the sample cleanup such as
Florisil column elution and sulfur removal with mercury may not be necessary for
MS/MS detection. This is often done, however, to keep the GC injector liner and

column and the MS ion source clean. With MS detection it is difficult to reliably

120

quantitate an individual PCB congener that co-elutes with a congener containing
one more chlorine, since that congener generally loses a chlorine in the source. That
ion’s l"'C isotope [M-34} has the same mass as the molecular anion of the co-eluting
congener. As such, the precision and accuracy of monitoring that anion

quantitatively in a SIM experiment suffers drastically.

As discussed in Chapter 5, the primary interference with this MS/MS method is in
the case where a higher homolog co-eluting PCB congener loses a chlorine and gains
a hydrogen in the source for a net loss of 3411. Fortunately, many of the congeners
that co-elute with the coplanar PCB congeners do not exhibit this loss. It would be
advantageous, therefore, to explore other GC stationary phases that may offer worse
chromatographic resolution overall, but allow all twelve coplanar PCB congeners to

be resolved from higher homolog congeners that lose 3411 in the ion source.

Accuracy. The accuracy of the quantitative results of this method were compared to
those recently reported by Schulz (14) for the analysis of several commercial
mixtures of PCBs including Aroclor 1242 by means of multiple GC with electron
capture detection. The results are listed in Table 6.1. The discrepancies between the
two methods may be due to differences in the two Aroclor 1242 mixtures used. A
major problem with determining accuracy, then, is that there is no generally agreed

upon standard Aroclor mixture.

Results. As mentioned previously, the focus of this research was to develop a

121

Table 6.1. The results of the analysis of an Aroclor 1242 standard for the twelve
coplanar congeners compared to values reported in literature.

 

 

 

 

 

.

 

 

 

 

 

‘ 4

 

 

 

 

 

 

:2
§

' The amounts from literature were estimated for an identical sample
based on the amounts given for all Aroclor 1242 congeners.
" The literature percents are from Reference 14.

122

method of analysis for PCB congeners to aid in toxicity assessment and in
dechlorination studies. Recent work with PCB-contaminated river sediment samples
has shown evidence for anaerobic dechlorination of individual PCB congeners. It
was therefore important to develop a method of analysis for those congeners that
may be involved in these processes. In particular, the congeners of interest are those
twelve congeners that have been shown to exhibit dioxin-like toxicity. The results of
the analysis of two river sediment samples spiked with Aroclor 1242, one of which
was autoclaved and the other inoculated under anaerobic conditions, are shown in
Table 6.2. Two of the more important and more abundant congeners are 34,34 CB
and 234,34 CB. Both of these co-elute with other congeners in a GC method. Due
to the high specificity of the oxygen-chlorine exchange reaction, however, these can
be quantitated using this method. It is apparent that these results support the

premise that dechlorination is occurring under anaerobic conditions.

123

Table 6.2. The results of the quantitation by GC/MS/MS of two Aroclor 1242
samples, one of which was autoclaved and the other inoculated and incubated under
anaerobic conditions.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mmmnmnma

'Allamountsarein ug/LAllpercentsarebasedonthetotalamountoftheautoclavedsamples.

Chapter 7

Application of the MS/MS Method using the Oxygen-Chlorine
Exchange Reaction to the Analysis for Polychlorinated
Dibenzodioxins and Dibenzofurans in Samples containing

Polychlorinated Biphenyls

The analytical method using the oxygen-chlorine exchange reaction was used for
detection of polychlorinated dibenzodioxins (PCDDs) and polychlorinated
dibenzofurans (PCDFs) in PCB-containing sediment samples. Early experiments
had indicated the presence of furans but not dioxins, and as such, PCDF standards

were prepared for purposes of comparison but not quantitation.

Introduction. The main focus of this research was to develop MS/MS and
GC/MS/MS methods for the quantitation of individual PCB congeners, and in
particular those congeners that have been shown to exhibit dioxin-like toxicity, in
sediment samples. It has been shown, though, that the overall toxicity of PCB-
containing samples is quite often due not to the PCB content of these samples but

to the presence of PCDFs (7). A PCDF mixed standard was prepared and analyzed

124

125

by monitoring the exchange of oxygen for chlorine with the molecular anions of
tetra- through octachloro dibenzofurans in a GC/MS/MS experiment. This
experimentwas also done with an autoclaved and inoculated Aroclor 1254 standard

and with two PCB-containing sediment samples.

The toxicity of chlorinated aromatics is usually determined relative to that of 2,3,7,8-
TCDD (tetrachlorodibenzodioxin) (10-12). In the case of PCB congeners, it has
been shown that 34,34 CB (chlorobiphenyl) is the most toxic of all 209 congeners.
It is also speculated that this is as expected from the fact that this congener in the
coplanar (the two aromatic rings in the same plane) conformation has the four
chlorine nuclei in ahnost the exact geometric configuration as that of the 2,3,7,8
TCDD molecule. As stated in the previous chapter, those PCB congeners that have
been shown to exhibit dioxin-like toxicity are only those congeners of at least tetra
substitution thatcan exist in the coplanar conformation. Thisconformation is stericly

possible only with congeners with one or no chlorines ortho to the biphenyl linkage.

In the environment, the fate of individual PCB congeners includes chlorine uptake,
photo-induced dechlorination and dechlorination due to microbial activity. The
oxidation of PCB congeners to dioxins, furans and ethers has been demonstrated in
the laboratory. There is evidence that these processes may also be photo-induced in
the environment. It was therefore expedient to not only develop a method to screen
for these possible metabolites, but to also investigate the possibility of these

compounds interfering in the analysis of individual PCB congeners.

126
Experimental The Aroclor and sediment samples were provided by John Quensen,
and the PCDF standards were purchased from AccuStandard, New Haven, CT. All
standards were prepared by dilution with Organic Residue Analysis grade hexane
(J.T. Baker). River sediment samples spiked with Aroclor 1242 were extracted with
50/50 acetone/hexane. Each extract was subjected to Florisil clean-up and then
shaken with mercury for sulfur removal and finally concentrated to ten milliliters.
While this extensive clean-up procedure is not necessary for sample analysis by

GC/MS/MS, this was done such these samples could also be analyzed by GC/ECD.

Gas Chromatography. All gas chromatographic methods were performed with a
Varian 3400 gas chromatograph equipped with a 30-meter x 0.250-mm column of
0.25-mm DB-l phase (methyl silicone, J & W Scientific). The GC was temperature
programmed—40 Cfor two minutes, 40 Cto 160 C at20Cper minute, 160 Cto
270 C at 5 C per minute and held at 270 C for seven minutes. Helium was used as
the GC carrier gas. The Varian 3400 was equipped with an injector body that
accommodates both split and splitless modes of injection. All experiments were
performed umg the splitless mode of injection with the split valve closed for the
initial two minutes and open for thirty minutes. The injector was held at 270 C and

the transfer line held at 275 C.

Mass Spectrometry. All experiments were done with a Finnigan TSQ-700 mass
spectrometer operated in a negative ion mode. Ammonia was used as the chemical

ionization reagent gas. The CI reagent gas pressure was set to approximately 9500

127

mTorr as measured by the source Convectron gauge. Pure oxygen was introduced
into the collision cell. The source was held at 150 C and the manifold at 70 C. All
reactions were generated using electron energies of 50 eV, collision offset energies

of 1-2 eV (laboratory), and collision pressures from 0.5 mTorr to 1.5 mTorr.

Isotopic Distribution. The isotopic distribution of molecular anions of
polychlorinated aromatics provides for several anions as candidates as precursor
ions in an MS/MS experiment. In the case of any polychlorinated aromatic, the most
abundant anion in the molecular cluster is the [M]" (or [M-H]') anion for congeners
comprised of less than four chlorines, the [M+2]' anion for congeners comprised
of four to seven chlorines and the [M+4]‘ anion for congeners comprised of over

seven chlorines. This is indicated for PCBs in Figure 7.1.

The isotopic distribution of six tetrachlorinated aromatics (Figure 7.2) of particular
environmental importance are shown in Figures 7.3 (spectra) and 7.4 (text listing of
exact masses). These distributions are theoretical and were calculated using the
natural abundances of 12C, 13C, 1H, 2H, 1“O, “O, 35Cl and 3“’Cl. With a perfectly
cah'brated instrument, the molecular anion region of an ECN I spectrum of these

analytes should appear as shown for those compounds not exhibiting any hydrogen
stripping ( [M-Hr ).

Molecular Anions. The molecular anions (chlorine isotopes only) of six chlorinated

compounds are indicated in Table 7.1. The six compounds chosen include PCBs,

128

 

 

 

 

 

 

 

 

 

 

 

 

 

 

n=0 n=1
.. M'=154 .1 'i' M‘=133
- .. 0
T, _. 01.. 4
' 11:3 - n=4
- Nfiéfli .. "Theamz
—* I!“ =———..-l-———-al 17"."

 

 

 

 

n=6

 

 

 

 

 

 

 

 

 

 

 

 

1' l1; .. _ _ 0 l l Ila...“
-5" I. ' fl we
. -. n=9 . w n=10

 

 

 

 

 

 

 

 

 

 

L°M+4'=464 " 044-42493

. 7
””1135." __.__J 1 1|; .... -

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“'lr—v "' V:
n=2
M'=222
.. .. 0
.. -50". _ E
.0 "” .... 11:5
.. M+2°=3213
-: il ILL,” .. -
"" ‘0'" [:3
. .- n==8
-. M+4'=430
.. i ‘"
.. : Jilli- : --

 

 

 

Figire 7.1. The molecular anion clusters of polychlorinated biphenyls calculated
ung the natural abundances of carbon, chlorine and hydrogen.

129

Biphenyl

0 a
2 O 7
3 I I o
4

Dibenzofuron

 

Benzene Phenol

mare 7.2. The structures and numbering conventions for six compounds whose
chlorinated analogs are of environmental significance.

130

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.. .... 0301,11, ‘ ..... c,,c1,H,o
.. M+2°=292 -. M+2‘=308
m ..,.. - w
I ' ||TJ in ... a. see .a. I lusl Lain”... s. .5.
mfi - "" Pr: .... "" I”
-. cucuip -. cuc1,H,o,
.._. M+2°=306 | ..... M+2‘=322
-4 n, " W
W a. ll I’L'e' 0:.- “ in .. ... ”I |_ “"5” .. __
....-~" (a. ,
,, ..... _, .....
0,013, 1 c,CI,H,o
_, M+2°=2 16 - M+2'=232
- "0
..,.. ..,..
f - Ill-w. _, .. Jill”... .. ..

 

 

 

 

 

 

 

 

 

 

 

 

Figure 7.3. Generated spectra from isotope calculations of the molecular ion region
of six tetrachlorinated aromaties.

131

 

 

 

 

 

 

Tmtobigheuyl Tetrachlomphenyi ether
CuHoCL CuI'LCLO
n M+n 96 Abundance n M+n 96 Abundance
0 289.922 77.67 0 305.917 7754
1 290.926 10.53 1 306.921 1054
2 291.919 100.00 2 307.914 100.00
3 292.923 13.50 3 308.918 1353
4 293.917 48.49 4 309.912 48.62
5 294.920 6.49 5 310.915 6.53
6 295.914 10.56 6 311.909 10.64
7 296.917 139 7 312.912 1.41
8 297.911 0.90 8 313.907 0.92
9 298.914 0.11 9 314.909 0.12
10 299.917 0.01 10 315.912 0.01
11 316.915 ( 0.01
Tetrachiorodibenzofuran Tetrachiorodibenzodioxin
CEILCLO CithCLOz
n M+n 96W n M+n “Abundant:
0 303.901 7754 0 319.896 77.42
1 304.905 1052 1 320.900 1053
2 305.899 100.00 2 321.894 100.00
3 306.902 1350 3 322.897 1354
4 307.896 48.62 4 323.89) 48.75
5 308.899 651 5 324.894 655
6 309.893 10.64 6 325.888 10.73
7 310.896 1.40 7 326.891 1.42
8 311.891 0.92 8 327.886 0.94
9 312.894 0.12 9 328.889 0.12
10 313.896 0.01 10 329.891 0.01
11 314.899 ( 0.01 11 330.893 ( 0.01
Tetrachlorobenzene Tetrachlorophenol
CstCl. CstCloO

n M+n 96 Abundance n M +n 95 Abundance
0 213.891 78.06 0 229.886 77.94
1 214.894 5.28 1 230.889 5.30
2 215.888 100.00 2 231.88 100.00
3 216.891 6.76 3 232.886 6.79
4 217.885 48.09 4 233.880 48.31
5 218.889 3.24 5 234.88 3.27
6 219.882 1030 6 235.877 1038
7 220.886 0.69 7 236.881 0.70
8 221.879 0.84 8 37.875 0.86
9 222.885 0.06 9 38.878 0.06
10 223.886 ( 0.01 10 239.880 ( 0.01

Figure 7.4. The molecular anion clusters of six tetrachloro- aromatics calculated
using the natural abundances of carbon, chlorine and hydrogen.

132
chlorobenzenes, PCDFs, PCDDs, chlorophenols and chlorodiphenyl ethers. These

compounds are of significance environmentally and have been shown to be the
products of microbial and photo-induced reactions of PCBs, and as such, are often
present in PCB-containing samples. The molecular anions of these compounds are
indicated in Table 7.1 as n,x where n is indicative of the isotope [M+n]" andx is the
chlorine number. The Table includes the molecular anions of isomers that are of at
least trichloro- substitution. Many of these anions, such as [M+n]' where n is
greater than 10 for x less than 8, are of such low natural abundance that their
presence is never observed. This is the reason that a 13C isotopically-labeled standard

may be used for a hexachlorobiphenyl.

Some of the anions listed in Table 7.1 do not exist as radical anions, and as such, are
not amenable to the oxygen-for-chlorine exchange reaction. For example, the anion
cluster at m/z 222 for PCBs is associated with the dichloro congeners which exist
with ECNI only as [M-H]‘. For all six compounds, M' and the associated isotopes
are observed with at least tetrachloro substitution. By reading across the Table for
a particular anion, it is apparent which compounds may interfere with the analysis
of any chlorinated aromatic compound. It is also of use in choosing a precursor

anion for the MS/MS exchange reaction experiment.

The appropriate choice of a precursor anion must be made with regard to
abundance and the possibility of interferences from chlorinated metabolites. For

example, monitoring the reaction of m/z 304 to m/z 285 for TCDFs would be free of

133

Table 7.1. The molecular anions [M+n]‘ observed in the primary mass spectra of
polychlorinated aromatics designated as n,x where x indicates the chlorine number.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[Mfll' €60.11... C.CI.H.. (31.00.11... (160.031.. c,ocur,, 00.11.
180 0,3
182 2,3
184 4.3 0.0
186 6,3
188 0,1
190 2,1
192
194
196 0,3
198 2,3
200 4.3
202 0,1 6,3
204 2,1 0,1
206 2,1
208
210
212
214 0,4
216 2.4
218 4,4 0,1
220 6,4 2,1
222 0,2 8,4
224 2.2
226 4.2
228

 

 

 

 

 

 

134

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7.1 (confirmed).
[mar cucm... 0.0.15... 0600135.. 0.0.0.21... 0.003.. 0cm.
230 0,4
232 2,4
4,4
0,2 6,6
2.2 3,. 02
4.2 2.2
242 4,2
246
248 0,5
250 2.5
252 4,5 0,3
6.5 2.3
266 03 8.5 4.5
258 2,3 10,5 6,3
260 4,3
262 6.3
264 0,5 0,3
266 2.5 2.3
268 4,5 4,3
270 6.5 6.5 6.3
272 2.5 8.5
274 4,3 10,5
276 6,3
278

 

 

 

 

 

 

135

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7.1 (continued).
[M+n]" Ciao-“10a C‘CIJ-I“ Cum-Hen Cacao-Hex (2.00.112. 00:“:
2m
282 0,6
284 2,6
286 4,6 0,3
288 8,6 2,3
2!) 0,4 10,6 4,3
292 2,4 12,6 6,3
294 4,4
296 6,4
298 8,4
300
302
304 0,4
306 2,4 0,4
308 4,4 2,4
310 6,4 4,4
312 8,4 6,4
314 3.4
316
318
320 0,4
322 2,4
324 0,5 4,4
326 2,5 6,4
328 4,5 8.4

 

 

 

 

 

 

136

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7.1 (continued).
[M341]- Ciro-Hie. CsCLHu CuOCllHu €11qu C,OCl.H,,, 00-”:

330 6,5

332 8,5

334 10,5

336

338 0,5

340 2,5 0,5

342 4,5 2,5

344 5.5 4,5

346 3.5 5,5
10.5 3.5

350 10,5

352

354 0,5

356 2,5 0,5

358 0,6 4,5 2,5

360 2,6 6,5 4,5

362 4,6 8,5 6,5

364 6,6 10,5 8,5

366 8,6 10,5

368 10,6

370 12,6

372 0,6

374 2,6 0,6

376 4,6 2,6

378 8,6 4,6

 

 

 

 

 

 

137

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7.1 (continued).

[M‘Hll' (3110.316. (3.200.114. Cacao-Hex (3600.115. 00.31
380 10,6 6,6
382 12,6 8,6
384 10,6
386 12,6
388 0,6
390 2,6
392 0,7 4,6
394 2,7 6,6
396 4,7 8,6
398 6,7 10,6
400 8,7 12,6
402 10,7
404 12,7
406 14,7 0,7
408 2,7 0,7
410 4,7 2,7
412 6,7 4,7
414 8,7 6,7
416 10,7 8,7
418 12,7 10,7
420 14,7 12,7
422 0,7 14,7
424 2,7 16,7
426 0,8 4.7
428 2,8 6,7

 

 

 

 

 

 

138

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7.1 (continued).
1M+nr cuasi... €600.12. 0.0.0.21... w. 0cm.

430 4,8 8,7

432 6,8 10,7

434 8,8 12,7

436 10,8 14,7

438 12,8

440 14,8 0,8

442 16,8 2,8 0,8

444 4,8 2,8

446 6,8 4,8
8.8 6.8

450 10,8 8,8

452 12,8 10,8

454 14,8 12,8

456 16,8 0,8 14,8

458 2,8 16,8

460 0,9 4,8

462 2,9 6,8

464 4,9 8,8

466 6,9 10,8

468 8,9 12,8

470 10,9 14,8

472 12,9 16,8

474 14,9

476 16,9 0,9

478 18,9 2,9

 

 

 

 

 

 

139

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7.1 (continued).

[M+nr €60.21... 060035.. €60.03!" @0031" 0cm.
4&3 4,9
482 6,9
484 8,9
486 10,9
488 12,9
4% 14,9
492 16,9
494 0,10 18,9
496 2,10
498 4,10
500 6,10
502 8,10
504 10,10
506 12,10
508 14,10
510 16,10 0,10
512 18,10 2,10
514 20,10 4,10
516 6,10
518 8,10
520 10,10
522 12,10
524 14,10
526 16,10
528 18,10

 

 

 

 

 

 

140

interference from the M' of tetrachlorodiphenyl ethers even though choosing m/z
306 as a precursor would be warranted from its abundance. Also, monitoring the
reaction of m/z 458 to m/z 439 would be specific for octachlorodibenzo dioxins (the
[M+ 16]' is not abundant enough to be observed in a routine analysis) even though
m/z 460 is more abundant but is also observed with nonachlorobiphenyls. While
using the exchange reaction method for targeted analyses with direct introduction
into the ion source (such as solids probe), the choice of precursor is of paramount
importance in preserving the specificity of the method. With the increased specificity

of chromatographic detection, this may not be necessary for some analytes.

Dioxins and Eurans. As stated previously, much of the toxicity of PCB-containing
samples can often be attributed to the PCDF content of the sample. This may be
true even when the amount of total PCDF concentration is much less than that of
total PCBs. It is obvious that with these concentrations, GC/ECD or GC./MS will not
suffice for identification and quantitation for these samples. The oxygen-for-chlorine
exchange reaction has been used in the past to determine the presence of PCDDs
in sediment samples, and none have been observed at a quantifiable level. The
GC/MS/MS analysis of a standard mixture of 5 tetrachlorodibenzo dioxins and 1
pentachlorodibenzo dioxin is shown in Figure 7 .5 . Indicated in the chromatograms
are the reactions for tetra-, penta- and hexachlorodibenzo dioxins. Notice that in this
"pure’ standard of only Cl, and C15 dibenzo dioxins, there is a significant response at

approximately scan number 875 for a Cl, dibenzo dioxin.

141

1.100

Tetra

50“

 

1L

 

 

 

 

 

:5 Pento

C 50 E
'9 ll 03
g .613
->- c
'8 o
— ‘°‘ Hex —
11> 0' J

a L

sol

 

 

 

rvvij’fV—V' vvf VV fr VVV ‘7 VVV r vv wifTW—v v

Scan Number

Figure 7.5. The reconstructed total ion current (RIC) chromatogram generated by
monitoring the oxygen-for-chlorine exchange reaction of the molecular anions
[M+2]- of PCDDs in a GC/MS/MS experiment.

142

The tetra- through octachloro-substituted PCDFs were monitored for an autoclaved
and an inoculated sample spiked with Aroclor 1254 standard (congeners at the ppm
level) (Figure 7.6) and two sediment samples from natural sites contaminated with
PCBs (Figure 7. 7). The chromatograms for the Aroclor 1254 samples indicate that
microbial action may have resulted in a slight decrease in the overall PCDF content
(39,000 ADC counts versus 77,000 counts), but there is virtually no difference in the
overall pattern. This is not generally the case with sediment samples showing
apparent dechlorination (see Chapter 6). These results indicate that even with a
pure Aroclor standard, there is a significant amount of other chlorinated aromaties
present. This is not surprising, since, as stated previously (Chapter 2), the toxicity of
many PCB-containing samples is quite often due to the PCDF-content of these

samples.

For the two sediment samples, a significant amount of PCDFs is indicated from the
chromatographic data. These samples are from two different environmental sites
that have been found to be PCB-contaminated. As the dechlorination of PCB-
containing samples is monitored, it may be worthwhile to also monitor other
chlorinated aromatics. This data also demonstrates the inadequacies of other
detection methods such as ECD or MS for low levels of analytes in the presence of
high levels of similar compounds.

Conclusions. The data from the analysis of PCB-containing samples for PCDDs and

PCDFs indicates that using the oxygen-for-chlorine exchange reaction in a

143

“a euuaee na-enav-ua “0.... none. . o

Inoculated

 

 

 

 

i
'1'
5
Ion Current

 

Relative Intensity

"‘ Autoclaved

 

 

 

 

Scan Number

Ilium 7.6. The reconstructed total ion current (RIC) chromatogram generated by
monitoring the onygen-for-chlorine exchange reaction of the molecular anions
[M+2]- of tetrachlorodibenzofurans in a GC/MS/MS experiment.

144

 

 

 

 

 

 

“' m 79:.“

H7 Sediment
g z .
.93 .. aha-3:11- 9
C ..... .2... 2...... ...... a”... .... h
6 ““ "“ 8
>
:1: C
£2 £2
1]) F3 Sediment
c2

 

 

 

 

 

 

 

 

Scan Number

Figure 7.7. The reconstructed total ion current (RIC) chromatogram generated by
monitoring the axygen-for-chlorine exchange reaction of the molecular anions
[M+2]- of tetrachlorodibenzofurans in a GC/MS/MS experiment.

145
GC/MS/MS experiment may be of utility for the analysis of many chlorinated

aromatics. Several issues remain to be investigated. One, each of the chlorinated
compounds studied needs to be further analyzed to determine which if any exhibit
the loss of 34 u in the source as discussed for PCBs in previous chapters. Also, for
some samples, other chlorinated compounds such as naphthalenes, anthracenes,
phenanthrenes and biphenyienes should be studied analytically. Furthermore, for
each of the chlorinated aromatics studied including the PCDDs and PCDFs, these

methods need to be developed with individual quantitative standards.

Chapter 8

Conclusions

Due to the problems inherent with ECD and MS methods of detection for PCB
samples, an MS/MS method was developed for the chromatographic detection of
samples containing PCBs and other chlorinated aromatics. This method employs
monitoring the products of ion-molecule reactions generated in a triple-stage
quadrupole using oxygen as a reagent. The products observed were the result of an
exchange of an oxygen atom for a chlorine atom for a net loss of 19 u. The method
development included instrument optimization for production of the precursor ions
(molecular anions) and for the MS/MS reaction products. The method was also used
to detect the presence of other chlorinated aromatics (specifically polychlorinated
dibenzofurans and polychlorinated dibenzodioxins) insamples containing significant

amounts of PCBs. Also studied were possible interferences with this method.

Instrumental Parameters. The instrumental parameters studied that affect the
formation of molecular anions in the MS ion source included source temperature,
type and pressure of moderating gas and filament electron energy. Itwas determined

that molecular anion production was optimum with ammonia as a moderating gas

146

147

and at a pressure of 8 x 1045 Torr as measured by the manifold ion gauge. The ion
source was operated at 150 C, and the filament emission was generated with an
electron energy of 50 eV. Instrumental tunes were generated with baseline
resolution for MS experiments and with slightly less resolution (<25% valley) for

MS/MS experiments.

The significant instrumental parameters controlling the generation of ion-molecule
reaction products in an MS/MS experiment include type and pressure of collision
gas, reaction energy and translational kinetic energy correction. Both ambient air
and pure oxygen were used as the collision gas, but the use of oxygen provides for
greater yields in reaction products. Unlike higher-energy CID experiments, the use
and/or amount of translational kinetic energy correction is not critical for product
ion trannission into the detector with this ion-molecule reaction. Clearly the most
critical parameter controlling the amount of reaction product observed in this
MS/MS experiment is the collision energy. The optimum energy for this reaction was

determined to be appru-ately 1 eV.

In applying this reaction as an analytical method for the determination of PCBs in
sediment samples, up to two orders of magnitude of improvement in signal is
realized using the selected-reaction monitoring (SRM) mode of detection instead
of full-scan acquisition. Also, with the instrumental tunes and set masses used, the
product scan mode is more sensitive than precursor or neutral loss scan modes for

SRM. Chromatographic conditions were used that allowed for comparison with

148
GC/ECD methods for the same samples. Due to the selectivity of the SRM method,

GC run times could be shortened considerably for routine analysis of PCB samples.

Other Chlorinated Aromatics. The oxygen-for-chlorine reaction was also applied as
a detection method forpoiychlorinated dibenzofurans (PCDFs) and polychlorinated
dibenzodioxins (PCDDs). This was particularly useful for detecting these
compounds in samples containing significant amounts of PCBs. In ahnost all
sediment samples analyzed, the presence of PCDDs was not observed. In ahnost all
sediment samples analyzed, however, the presence of significant amounts of PCDFs

was observed.

Future Studies. With chromatographic detection for some coplanar PCB congeners,
congeners that co-elute and lose 34 u in the ion source interfere with this MS/MS
method. Each congener that may interfere, then, must be analyzed as a pure
standard in order to determine which of these congeners lose 34 u and which lose
35 11. Also, it may be possible to use other GC stationary phases that offer overall
less chromatographic resolution for PCB congeners, but might provide for fewer

interferences with the GC/MS/MS method.

Since the MS/MS method is applicable for many chlorinated compounds even in the
presence of significant amounts of PCBs, it may be possible to determine the fate of
chlorinated compoundsin samples subjected tomicrobial action. These results could

also be compared to other types of analysis such as total organic chlorine content

149

and toxicity assessment. These methods could also be used to determine the
inadequacies of methods such as GC/ECD where PCDFs and PCDDs are possible

interferences.

Due to the high selectivity offered by MS/MS detection, particularly with the use of
ion-molecule reactions, other reagent gases should be explored. These would include
other radical molecules that may undergo the same type of reaction or reactions as
the oxygen-for-chlorine exchange. In addition, other reagents may prove to be even
more selective than oxygen and other free radicals. Ideally, these reagents could also
be selective for the differences in the steric conformations exhibited by various PCB
congeners. Specifically, an optimum reagent may be one that reacts only with those
congeners that form the coplanar conformation, particularly since these are the

congeners that have been shown to exhibit dioxin-like toxicity.

(1)
(2)
(3)

(4)
(5)
(6)

(7?

(8)

(99

(109

(11)

(129

(137

(14)

ILISVI'()I7IKIEFTEJKIEPI(IEMS

Schmidt, H.; Schultz, G. Ann. China, 1881, 207, 338.
Holden, A. V. Nature, 1970, 228, 1220.

Erickson, M. D. Analytical Chemistry of PCB ’s, Buttcrworth: Stoneham,
MA, 1986.

DiNardi, S. R.; Desmarais, A. M. Chemistry, 1976, 49(4), 14.
Jensen, S. New Sci, 1966, 32, 612.

Saeki, S.; Tsutsui, A; Oguri, K.; Yoshimura, H.; Hamana, M. Fukuoka Acta
Med, 1971, 62, 20.

Hamilton, S. B. PCBs in the Hudson River, General Electric Company:
Fairfield, CT: 1990.

Cairns, T.; Siegrnund, E. G. Anal. Chem, 1981, 53, 1183A.

Fed. Reg., May 9, 1980, 45, 30980.

Safe, 8.; Bandiera, S.; Sawyer, T.; Robertson, L; Safe, L.; Parkinson, A.;
Thomas, P.E,.; Ryan, D.E.; Peik, L.M.; Levin, W.; Denomme, M.A.; Fuhita,
Y. EHP, Environ. Health Perspect, 1985, 60, 47-56.

Leece, B.; Denomme, M.A.; Towner, R.; Angela Li, S.M.; Safe, S. J.
Toxicol. Environ. Health, 1985, 16, 379-388.

Kannan, N.; Tanabe, S.; Tatsukawa, R. Bull. Environ. Contarn. Toxicol,
1988, 41, 267-276.

Ballschmiter, K.; Zell, M. Fresenius Z. Anal. Chem, 1980, 302, 20.
Schulz, D.E.; Petrick, G.; Duinker, J .C. Environ. Sci. Technol, 1989, 23,
852.

150

(15)

(16)

(17)

(18)

(19)

(20)

(21)

(22)

(23)

(24)

(25)

(26)

(27)

(28)

151

Brinkrnan, U. A. T.; De Kok, A. In Halogenated Biphenyls, Terphenyls,
Naphthalenes, Dibenzodiaxins and Related Products; Kimbrough, R. D., Ed.;
Elsevier/North-Holland: New York, NY, 1980.

Rodriguez, C. F.; M‘Mahon, W. A.; Thomas, R. E. Method Development for
Determination of Polychlorinated Hydrocarbons in Municipal Sludge, Final
Report, Contract No. 68-03-2606, Environmental Protection Agency, EPA-
600/2-80-029; NT'IS No. PB 82-234 071: March, 1980.

Goerlitz, D. F.; Law, L. M. J. Assoc. Off. Anal. Chem, 1974, 57(1), 176.
Eder, G. Chemosphere, 1976, 5(2), 101.

Chan, A. S.; Carton, J.; Lee, H. B. J. Assoc. Off Anal. Chem, 1979, 62(6),
1312.

Seidl, G.; Ballschmiter, K. Chemosphere, 1976, 5, 373.

Young, S. J .; Finsterwalder, C.; Burke, J. A. J. Assoc. 0))“. Anal. Chem 1973,

56, 957.

Association of Official Analytical Chemists, Official Methods of Analysis of
theAssociation of Official Analytical Chemists, 13m ed, Horwitz, W., Ed.,
1980, 475.

King, J. W. J. Chromatogr. Sci, 1989, 17, 355.

Hawthorne, S. B. Anal. Chem , 1990, 62, 633A.

Hawthorne, S. B.; Krieger, M. S.; Miller, D. J. Anal. Chem, 1989, 61, 736.
Onuska, F. 1.; Terry, K. A. I. High Resolut. Chromatogr., 1989, 12, 527.

Hawthorne, S. B.; Miller, D. J. J. Chromatogr., 1987, 403, 63.

Miller, D. J .; Schantz, M.; Chesler, S. N. J. Chromatogr., 1986, 363, 397.

(29)

(30)

(31)

(32)

(33)

(34)

(35)

(36)

(37)

(38)

(39)

(40)

(41)

(42)

(43)

152

Hawthorne, S. B.; Langenfeld, J. J .; Miller, D. J .; Burford, M. D. Anal.
Chem, 1992, 64, 1614.

Langenfeld, J. J .; Hawthorne, S. B.; Miller, D. J .; Pawliszyn, J. Anal. Chem,
1993, 65, 338.

David, F.; Vershuere, M.; Sandra, P. FreseniusJ. Anal. Chem, 1992, 344,
479.

Nam, K. S.; Kapila, S.; Viswanath, D. S.; Clevenger, T. E.; Johansson, J .;
Yanders, A. F. Chemosphere, 1989, 19, 33.

Nam, K. S.; Kapila, S.; Yanders, A. F.; Puri, R. K. Chemosphere, 1990, 20,
833.

Bowadt, S.; Johansson, B. Anal. Chem, 1994, 66, 667.

Millar, J. D.; Thomas, R. E.; Schattenberg, H. J. Anal. Chem, 1981, 53, 214.
Easty, D. B.; Wabers, B. A. Tech. Assoc. Pulp and Paper Ind, 1978, 61 , 71.
Delfino, J. J.; Easty, D. B.Anal. Chem, 1979, 51, 2235.

Godefroot, M.; Stechele, M.; Sandra, P.; Verzele, M. J. Higt Resol Chrom.
and Chrom Comm, 1982, 5, 75.

Nickerson, G. B.; Likens, S. T. J. Chromatogr., 1966, 21, 1.

Coburn, J. A.; Valdmanis, I. A.; Chau, S. Y. J. Assoc. 0175 Anal. Chem,
1977, 60(1), 224.

Gesser, H. D.; Chow, A.; Davis, F. C. Anal. Letters, 1971, 4(12), 883.

Picer, M.; Picer, M. J. Chromatogr., 1980, 193, 357.

Leoni, V.; Puccetti, G.; Colombo, R. J .; d’Ovidio, A. M. J. Chromatogr.,
1976, 125, 399.

(44)

(45)

(46)

(47)

(48)

(49)

(50)

(51)

(52)

( 53)

(54)

(55)

(56)

153
Colenutt, B. A; Thorburn, S. Intern. J. Environ. Anal. Chem, 1980, 7, 231.

Analysis of Pesticide Residues in Human and Environmental Samples, A
Compilation of Methods Selected for Use in Pesticide Monitoring Programs,
EPA-600/8-80-038, Watts, R. R., Ed., US. EPA: Research Triangle Park,
NC, June, 1980.

Needham, L. L.; Burse, V. W.; Price, H. A.J. Assoc. Off Anal. Chem, 1981,
64(5), 1131.

Burse, V. W.; Needham, L. L.; Korver, M. P.; Lapeza Jr., C. R.; Liddle, J.
A.; Bayse, D. D]. Assoc. 0172 Anal. Chem, 1983, 66(10), 32.

Mills, P. A.; Onley, J. H.; Gaither, R. A. J. Ass. Oflic. Anal. Chem, 1963,
46(2), 186.

Association of Official Analytical Chemists, Official Methods of Analysis of
theAssociation of Ofi‘icial Analytical Chemists, 13m ed, Horwitz, W., Ed.,
1980, 475.

Zitko, V. Bull Environ. Contam. Toxicol, 1976, 16(4), 399.
Bagley, G. E.; Reichel, W. L.; Cromartie, E. J. Assoc. 0177c. Anal. Chem,
1970, 53(2), 251.

Tarradellas, J .; Diercxsens, P.; Bouche, M. B. Intern. J. Environ. Anal.
Chem, 1982, 13, 55.

Ernst, W.; Schaefer, R. G.; Goerke, H.; Eder, G. Z. Anal. Chem, 1974, 272,
358.

Environmental Protection Agency, The Analysis of Polychlorinatd Biphenyls
in Dansformer Fluid and Waste Oils, Office of Research and Development,
Environmental Monitoring and Support laboratory, Cincinnati, OH,
February 1981.

Seidl, G.; Ballschmiter, K. Chemosphere, 1976, 5, 363.

Food and Drug Administration, Pesticide Analytical Manual, Volume I,
August 1, 1977.

(57)

(58)

(59)

(60)

(61)

(62)

(63)

(64)

(65)

(66)

(67)

(68)

(69)

(70)

154
Bergman, A.; Reutegardh, L.; Ahlman, M. J. Chromatogr., 1984, 291, 392.
Huckins, J. N.; Stalling, D. L.; Johnson, J. L]. Assoc. Off. Anal. Chem,
1976, 59(5), 975.
Zitko, V. J. Chromatogr., 1971, 59, 444.
Griffin, D. A.; Marin, A. B.; Deinzer, M. L. J. Assoc. 0172 Anal. Chem, 1980,

63(5), 959.

Thin-Layer Chromatography: A Laboratory Handbook, Stahl, E., Ed.,
Springer-Verlag: New York, NY, 1969.

Dougherty, R. C. in Biochemical Applications of Mass Spectrometry, First
Supplemental Volume, Waller, G. R.; Dermer, O. C. Eds, Wiley-
Interscience Publications: New York, NY, 1980.

Smith, L. M.; Stalling, D. L.; Johnson, J. L.Anal. Chem, 1984, 56, 1830.

Jensen, 8.; Sundstrom, G. Ambio, 1974, 3(2), 70.

Bellar, T. A.; Lichtenberg, J. J. The Determination of Polychlorinated
Biphenyls in Transformer Fluid and Waste Oils, US. EPA Publication EPA-
600/4-81-045, 1981.

Stalling, D. L.; Tindle, R. C.; Johnson, J. L. J. Assoc. Ojfic. Anal. Chem,
1972, 55(1), 32.

Haile, C. L.; Baladi, E. Methods for Determining the Total Polychlorinated
Biphenyl Emissions for Incineration and Capacitor and Dansformer Filling
Plants, U. S. EPA Publication, EPA-600/4-77-048, NTIS No. PB-276
745/7G1, November 1977.

Lincer, J. L. Progr. Anal. Chem, 1973, 5, 109.

Veierov, D.; Aharonson, N. J. Assoc. Oflic. Anal. Chem, 1980, 63(2), 202.

Lamparski, L. L.;Nest1ick, T. J.Anal. Chem, 1982, 54, 402.

( 71)

(72)

(73)

(74)

( 75)

(76)

(77)

(78)

(79)

(80)

(81)

(82)

(83)

(84)

155
Underwood, J. C. Bull Environ. Contam Toxicol, 1979, 21 , 787.

Goerlitz, D. F.; Law, L. M. Bull. Environ. Contam Toxicol, 1971, 6, 9.

Jensen, S.; Sundstrom, G. Anal. Chem, 1977, 49(2), 316.

Mattson, P. E. J. Chromatogr., 1976, 124, 265.

Wilson, N. K.; Anderson, M. in Mass Spectrometry and NMR Spectroscopy in
Pesticide Chemistry, Haque, R.; Biros, F. J ., Eds., Pelnum Press: New York,
NY, 1973.

Mullin, M. D.; Sawka, G.; Safe, L. M.; M‘Crindle, 8.; Safe, S. H. J. Anal.
Toxicol, 1981, 5, 138.

Mullin, M. D.; Pochini, C. M.; M°Crindle, S.; Romkes, M.; Safe, S. H.; Safe,
L. M. Environ. Sci. Technol, 1984, 18, 468.

Hutzinger, 0.; Safe, 8.; Zitko, V. The Chemistry of PCB ’8, CRC Press:
Cleveland, OH, 1974.

Webb, R. G.; M‘Call, A. C]. Assoc. 0177c. Anal. Chem, 1972, 55(4), 746.
Nyquist. R. A.; Putzig, C. L.; Peterson, D. P. Appl. Spectrsoc., 1983, 37(2),
140.

Chen, J -Y. T.; Gardner, A. M. Amer. Lab., 1983, 26.

Nordstrom, R. J .; M°Intosh, B. in Proceedings: 1981 PCB Seminar, Report
No. EPRI-EL-25 73, Addis, G.; Marks, J ., Eds., Electric Power Research
Institute: Palo Alto, CA, 1982.

Fishbein, L. J. Chromatogr., 1972, 68, 345.

Fuller, B; Gordon, J .; Kornreich, M. Environmental Assessment of PCBs in

the Atmosphere, EPA-450/3- 77-045, U. S. EPA: Washington, DC, April,
1976.

(85)

(86)

(87)

( 88)

(89)

(90)

(91)

(92)

(93)

( 94)

(95)

(96)

156

Krull, I. S. in Residue Reviews, Gunther, F. A.; Gunther, J. D., Eds., New
York, NY, 1977.

Margeson, J. H. Methodology for Measurement of Polychlorinated Biphenyls
in Ambient Air and Stationary Sources - A Review, Report No. EPA-600/4-77-
021, U. S. EPA, April, 1977.

The Analysis of Chlorinated Biphenyls, EPA Reference OPTS-62017,
Chemical Manufacturer’s Association, August 21, 1981.

Sherma, J. inAdvances in Chromatography, Giddings, J. C.; Grushka, E.;
Keller, R. A.; Cazes, J ., Eds., Marcel Dekker, Inc.: New York, NY, 1975.

Albro, P. W.; Haseman, J. K; Clemmer, T. A.; Corbett, B. J. J.
Chromatogr., 1977, 136, 147. Errata in]. Chromatogr., 1977, 139, 404.

Safe, S.; Mullin, M.; Safe, 1...; Pochini, C.; M‘Crindle, S.; Romkes, M. in
Physical Behavior of PCBs in The Great Lakes, Mackay, D; Paterson, S.;
Eisenreich, S. J .; Simmons, M. S., Eds., Ann Arbor Science Publishers, Inc.:
Ann Arbor, MI, 1983.

Erickson, M. D.; Stanley, J. 8. Methods of Analysis for By-Product PCBs -
Literature Review and Preliminary Recommendations, Interim Report No. 1
EPA-560/5-82-005, NTIS No. PB83 126 573, Office of Toxic Substances,
U.S. EPA: Washington, DC, 1982.

Mullin, M. D.; Filkins, J. C. Advances in the Identification and Analysis of
Organic Pollutants in Water, Volume 1, Keith, L. H., Ed., Ann Arbor Science
Publishers, Inc.: Ann Arbor, MI, 1981.

Mullin, M. D.; Pochini, C. M.; Safe, S. H.; Safe, L. M. PCBs: Human and
Environmental Hazards, D’Itri, F. M.; Kamrin, M. A, Eds., Butterworth
Publishers: Boston, MA, 1983.

Jennings, W. Gas Chromatography with Glass Capillary Columns, 2‘Id Ed. ,
Academic Press: New York, NY, 1980.

Pellizmri, E. D. Environmean Health Chemistry, M‘Kinney, J. D., Ed., Ann
Arbor Science Publishers, Inc.: Ann Arbor, MI, 1981.

Hanneman, L. F. The New PCB Issue - Analysis of Specific Congeners
Produced by Unintentional By-Product Chemistry, Capillary Chromatography
’82 - An International Symposium, Tarrytown, NY, 1982.

(97)

(98)

(99)

(100)

(101)

(102)

(103)

(104)

(105)

(106)

(107)

(103)

(109)

(110)

157
Hanai, T.; Walton, H. F. Anal. Chem 1977, 49(6), 764.

Van Vliet, H. P. M.’ Bootsman, T. C. J. Chromatogr., 1979, 185, 483.

Aue, W. A.; Kapila, SJ. Chromatogr. Sci, 1973, 11, 255.

Duinker, J. C.; Schulz, D. E.; Petrick, G. Anal. Chem, 1988, 60, 478.
Mulhern, B. M. J. Chromatogr., 1968, 34, 556.

Mulhern, B. M.; Cromartie, E.; Reichel, W. L.; Belisle, A. J. Assoc. Offic.
Anal. Chem, 1971, 54(3), 548.

Trotter, W. J .; Young, S. J. V. J. Assoc. 0177c. Anal. Chem, 1975, 58(3), 466.
Berg, O. W.; Diosady, P. L.; Rees, G. A. V. Bull. Environ. Contam Toxicol,
1972, 7(6), 338.

Zimmerli, B. J. Chromatogr., 1974, 88, 65.

Cooke, M.; Nickless, G.; Prescott, A. M.; Roberts, D. J. J. Chromatogr.,
1978, 156, 293.

Oswald, E. O.; Albro, P. W.; M‘Kinney, J. D. J. Chromatogr., 1974, 98, 363.
Safe, S. in National Conference on Polychlorinated Biphenyls, November 19-
21, 1975, Ayer, F. A., Ed., U. S. EPA: EPA-560l6-EflO4, NTIS No. PB 253-

248, March, 1976.

Stan, H. J. inPesticideAnalysis, Das, K. G., Ed., Marcel Dekker, Inc.: New
York, NY, 1981.

Tindall, G. W.; Wininger, P. E. J. Chromatogr., 1980, 196, 109.

(111)

(1129

(113)

(1149

(115)

(116)

(117)

(118)

(1199

(120)

(121)

(122)

(123)

(124)

158

Erickson, M. D.; Stanley, J. S.; Turman, J. K.; Going, J. E.; Redford, D. P.
in Polynuclear Aromatic Hydrocarbons: Formation, Metabolism and
Measurement, Cooke, M; Dennis, A. J ., Eds., Battelle Press: Columbus,
OH, 1983.

Oswald, E. O.; Levy, 1...; Corbett, B. J .; Walker, M. P. J. Chromatogr., 1974,
93, 63.

Iida, Y.; Daishima, S.; Kashiwagi, T. Seikei Daigaku Kogakubu Kogaku
Hokolat, 1975, 19, 1461.

Stalling, D. L. The Institute of Electrical and Electronics Engineers, Inc., 1976,
Annals No. 75CH1004-I 7-5.

Cairns, T.; Jacobson, R. A. J. Chem Infor. Comp. Sci, 1977, 17(2), 105.
Sawyer, L. D. J. Assoc. Ofi‘ic. Anal. Chem, 1978, 61(2), 272.

Cairns, T.; Siegmund, E. G.Anal. Chem, 1981, 53(11), 1599.

Iida, Y.; Daishima, S.; Kashiwagi, T. Bunseki Kagaku, 1983, 32(2), 80;

Chem Abstr. 98:15473 7f.

Dougherty, R. C. Biomed. Mass Spectrom, 1981, 8(7), 283.

Dougherty, R. C. Origins of Life, 1981, 11, 71.

Dougherty, R. C.; Roberts, J. D.; Tannenbaum, H. P.; Biros, F. J. in Mass
Spectrometry and NMR Spectroscopy in Pesticide Chemistry, Haque, R.;
Biros, F. J ., Eds., New York, NY, 1973.

Dougherty, R. C.; Whitaker, M. J .; Smith, L. M.; Stalling, D. L.; Kuehl, D.
W. Environ. Health Persp. , 1980, 36, 103.

Kuehl, D. W.; Whitaker, M. J .; Dougherty, R. C. Anal. Chem, 1980, 52,
935.

Pellimri, E. D.; Tomer, K. B.; Mosely, M. A. inAdvances in the
Identification and Analysis of Organic Pollutants in Water, Volume 1, Keith,
L. H., Ed., Ann Arbor Science Publishers, Inc.: Ann Arbor, MI, 1981.

(125)

(126)

(127)

(128)

(129)

(130)

(131)

(132)

(133)

(134)

(135)

(136)

(137)

159
Pellizzari, E. D.; Cooper, S. D.; Hartwell, T. D.; Whitehurst, D. A.Analysis

of Adipose and Blood Sera Samples for Individual PCB Isomers, Final Report,
U. S. EPA: Kansas City, MO, Contract No. 68-01-5915, Task 46, 1983.

Pellizmri, E. D.; Moseley, M. A.; Cooper, S. D.; Harry, J. V.; Demian, B.
A.; Mullin, M. D. inAdvances in Exposure, Health and Environmental
Effects Studies of PCBs: Symposium Proceedings, Davenport, R. J .; Bernard,
B. K., Eds., U. S. EPA: Washington, DC, Report No. LSI-TR-507-l37B,
NTIS No. PB84-l35771, December, 1983.

Dougherty, R. C.; Whitaker, M. J .; Tang, S. Y.; Bottcher, R.; Keller, M.;
Kuehl, D. W. in Environmental Health Chemistry, The Chemistry of
Environmental/{gents as Potential Human Hazards, McKinney, J. D., Ed.,
Ann Arbor Science Publishers, Inc.: Ann Arbor, MI, 1981.

Yost, R. A.; Enke, C. G. Anal. Chem, 1979, 51, 1251A.
Stults, J. T.; Holland, J. F; Enke, C. G.Anal. Chem, 1983, 55, 1323.

Webster, G. R. B.; Birkholz, D. A. Mass Spectrometry in Environmental
Sciences, Karasek, F. W.; Hutzinger, 0.; Safe, S., Eds., Plenum Press: New
York, NY, 1985.

Hunt, D. F.; Shabanowitz, J .; Harvey, T. M.; Coates, M. Anal. Chem, 1985,
57, 525.

Greathead, R. J .; Ashcroft, A. E. Chem. Anal, 1987, 91, 211.

Bonner, R. F. Int. J. Mass Spectrom and Ion Phys., 1983, 48, 311.
Voyksner, R. D.; Hass, J. R.; Sovocool, G. W.; Bursey, M. M.Anal. Chem,
1983, 55, 744.

Safe, S.; Hutzinger, O.; Jamieson, W. D. Org. Mass Spectrom, 1973, 7, 169.

Levy, L. A; Oswald, E. O. Biomed. Mass Spectrom, 1976, 3, 88.

Roach, J. A. G.; Andrujewski, D. Chem Anal, 1987, 91 , 187.

(138)

(139)

(140)

(141)

(142)

(143)

(144)

(145)

(146)

(147)

(148)

(149)

(150)

(151)

(152)

160

Guevremont, R.; Yost, R. A.; J amieson, W. D. Biomed. Mass Spectrom. ,
1987, 14, 435.

Chakel, J. D. Ph. D. Diss., Michigan State University. East Lansing, MI,
1982.

Webb, R. G.; M‘Call, A. C.J. Chromatogr. Sci, 1973, 11, 366.

Biller, J. E.; Biemann, K.Anal. Lett., 1974, 7, 515.

Blaisdell, B. B.; Sweeley, C. C.Anal Chim Acta, 1980, 117, 1.

Gates, S. C.; Smisko, M. J .; Ashendel, C. L.; Young, N. D.; Holland, J. F.;

Sweeley, C. C. Anal. Chem, 1978, 50, 433.

Dunn III, W. J.; Stalling, D. L.; Schwartz T. R.; Hogan, J. W.; Petty, J. D.;
Johansson, E.; Wold, S.Anal. Chem, 1984, 56, 1308.

Liu, R. H.; Ramesh, S.; Liu, J. Y.; Kim, S.Anal. Chem, 1984, 56, 1808.
Kowalski, B. R.; Bender, C. F.Anal. Chem, 1972, 44, 1405.

Wold, S. Pattern Recognition, 1976, 8, 127

Harrison, A. G. Chemical Ionization Mass Spectrometry, CRC Press: Boca
Raton, FL, 1982.

Watson, J. T. Introduction to Mass Spectrometry, 2"" ed., Raven Press: New
York, NY, 1985, p 192.
Milne, G. W. A.; Lacey, M. J. Crit. Rev. Anal. Chem, 1974, 4, 45.

Field, F. H.; Munson, M. S. B. J. Am Chem Soc., 1967, 89, 4272.

Munson, M. S. B.; Field, F. H. J. Am Chem. Soc., 1967, 89, 1047.

(153)

(154)

(155)

(156)

(157)

(158)

(159)

(160)

(161)

(162)

(163)

(164)

(165)

(166)

(167)

161
Field, F. H. J. Am Chem Soc., 1968, 90, 5649.

Hunt, D. F.; M‘Ewen, C. N. Org. Mass Spectrom, 1973, 7, 441.

Munson, B.Anal Chem, 1971, 44, 3781.

Field, F. H. J. Am Chem Soc., 1970, 92, 2672.

Dzidic, 1.; M‘Closkey, J. A. J. Am Chem Soc., 1971, 93, 4955.

Field, F. H. Accounts Chem Res, 1968, 1, 42.

Gregor, I. K.; Guilhaus, M., Int. J. Mass Spectrom Ion Proc., 1984, 56, 167.
Bowie, J. H., Mass Spectrom Rev., 1984, 3, 161-208.

Dougherty, R. C.; Dalton, J.; Biros, F. J ., Org. Mass Spectrom, 1972, 6,
1171-1181.

Stemmler, E. A.; Hites, R. A., Chem Anal, 1987, 91 , 42-59.

Daishima, S.; Iida, Y.; Kanda, F., J. Trace Microprobe Tech, 1989, 7, 87-102.
Field, F. H., Proc. Amer. Soc. Mass Spectrom , 1980, 28, 2- 13.

Stemmler, E. A.; Hites, R. A., Negative Chemical Ionization Mass Spectra,

Harcourt: New York, NY, 1988.

Hunt, D. F.; Harvey, T. M.; Russell, J. W.,]. C. S. Chem Comm, 1975, 151-
152.

Sykes, P.,A Guidebookto Mechanism in Organic Chemistry, Longman
Group, Ltd.: New York, NY, 1986.

162
(168) Kostiainen, R.; Auriola, S., Rap. Comm Mass Spectrom, 1988, 2, 135-137.

(169) Kostiainen, R.; Auriola, S., Org. Mass Spectrom, 1990, 25, 255-259.

(170) Heath, T. Ph. D. Diss. , Michigan State University: East Lansing, MI, 1991.

“I91919911“