Citation
Applications of pyrolysis-gas chromatography to pharmaceutical and clinical analysis

Material Information

Title:
Applications of pyrolysis-gas chromatography to pharmaceutical and clinical analysis
Creator:
Roy, Timothy Armand, 1947-
Publication Date:
Language:
English
Physical Description:
ix, 143 leaves : illustrations. ;

Subjects

Subjects / Keywords:
Chromatography, Gas ( mesh )
Pharmaceutical Chemistry thesis Ph. D
Dissertations, Academic -- Pharmaceutical Chemistry -- UF
Genre:
Academic theses ( lcgft )
Academic theses ( fast )

Notes

Thesis:
Thesis (Ph. D.) - University of Florida.
Bibliography:
Includes bibliographical references (leaves 140-142).
General Note:
Manuscript copy.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Timothy Armand Roy. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
024231090 ( ALEPH )
22838980 ( OCLC )

Downloads

This item has the following downloads:


Full Text
APPLICATIONS OF PYROLYSIS-GAS CHROMATOGRAPHY TO
PHARMACEUTICAL AND CLINICAL ANALYSIS
By
TIMOTHY ARMAND ROY
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA


DEDICATION
To the late Dr. Stephen Szinai, scientist, teacher,
and friend whose openness and encouragement made this
work possible.


ACKNOWLEDGEMENTS
I am indebted to my supervisory committee, Dr. Stephen
G. Schulman, Chairman, Dr. K.F. Finger, Dr. J.A. Zoltewicz
and Dr. B.S. Andresen, for their time and guidance in the
preparation of this manuscript.
I would also like to acknowledge Dr. Donald Chichester
for his help in editing the text and buffering my assaults on
the King's English.
Special thanks and appreciation go to Dr. Schulman for
accepting the responsibilities of committee chairman following
the loss of Dr. Szinai.
Thanks also to Carolyn Grantham for typing and to Gail
Clifford for graphics.
Ill


TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS . . . . iii
ABSTRACT iv
SECTION I - INTRODUCTION 1
Reaction Gas Chromatography 1
Nature of the Chemical Reaction in Pyrolysis-
Gas Chromatography 2
Pyrolysis Classification Based on Extent of
Degradation 4
Utilization of the Pyrolysis-Gas Chromatogram:
The Pyrogram 7
Recent Applications of Pyrolysis-Gas Chroma¬
tography 9
Polymer Analysis 9
Hydrocarbons - Pyrolysis Mechanisms and
Kinetics 12
Identification of Microorganisms and Fungi... 14
Toxicological and Pharmaceutical
Analysis 17
Purpose of Present Research 20
SECTION II - QUALITATIVE IDENTIFICATION OF FOOD AND
DRUG MATERIALS USING PYROLYSIS-GAS
CHROMATOGRAPHY 2 4
Advantages of Pyrolysis-Gas Chromatography 24
Increased Peak Identification Ability 24
Simplification or Elimination of Derivati-
zation Procedures 25
iv


TABLE OF CONTENTS (Continued)
Page
Characterization of Saccharin (o-benzo-
sulfimide ) by Pyrolysis-Gas Chromatography 26
Discussion and Results 27
Saccharin and sodium saccharin 27
Saccharin in soft drinks 28
Saccharin in a multivitamin product 29
Experimental 31
Materials 31
Apparatus 31
Procedure 32
Characterization of Penicillins and Cephalosporins
by Pyrolysis-Gas Chromatography 33
Discussion and Results 33
Benzyl penicillins 33
Isoxazolyl penicillins 35
Methicillin, nafcillin, penicillin V.... 35
Cephalosporins 36
Experimental 37
Apparatus 37
Antibiotics 38
Procedure 3 8
SECTION III - QUANTITATIVE ANALYSIS OF FOOD AND DRUGS
USING PYROLYSIS-GAS CHROMATOGRAPHY 39
Prerequisites for Quantitation 39
Cracking Severity Measurements 40
Peak Identification 42
Quantitative methods in GC and PGC 42
v


TABLE OF CONTENTS (Continued)
Page
Origin of fragmentation products 43
Comparison of classical and non-
classical thermolysis mechanisms 44
Quantitation of Penicillins and Cephalosporins.... 47
Discussion and Results 48
Cracking severity measurements 48
Peak identification - instrumental
methods 4 9
Peak identification - benzyl penicil¬
lins 49
Peak identification - isoxazolyl
penicillins 53
Peak identification - methicillin and
penicillin V 54
Peak identification - cephalosporins.... 55
Preparation of standard curves 56
Experimental 57
Apparatus 57
Antibiotics 58
Procedure 58
Quantitation of Saccharin 59
Discussion and Results 59
Cracking severity measurements 59
Peak identification - saccharin (sodium
salt) 59
Preparation of standard curves 60
Determination of sodium saccharin in
diet beverages 60
vi


TABLE OF CONTENTS (Continued)
Page
Experimental 64
Materials 64
Apparatus 65
Procedure 65
SECTION IV - ADDITIONAL APPLICATIONS OF PYROLYSIS-GAS
CHROMATOGRAPHY AND CONCLUDING REMARKS 67
The Use of Pyrolysis-Gas Chromatography for the
Diagnosis and Study of Metabolic Disorders 67
The Urine Pyrogram 67
Pyrolysis of serum and urine 67
PGC/MS analysis of urine 68
Utilization of the Urine Pyrogram 7 0
Present use of GC/MS for diagnosis of
metabolic disorders 70
Application of PGC and PGC/MS to the
diagnosis of metabolic disorders 72
Concluding Remarks 7 5
Qualitative Applications of Pyrolysis-Gas
Chromatography 7 5
Quantitative Applications of Pyrolysis-Gas
Chromatography 77
APPENDIX I - TABLES 83
APPENDIX II - FIGURES 87
REFERENCES 14 0
BIOGRAPHICAL SKETCH 143
vii


Abstract of Dissertation Presented to the
Graduate Council of the University of Florida in
Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy
APPLICATIONS OF PYROLYSIS-GAS CHROMATOGRAPHY TO
PHARMACEUTICAL AND CLINICAL ANALYSIS
By
Timothy Armand Roy
December, 1976
Chairman: Stephen G. Schulman
Major Department: Pharmaceutical Chemistry
Pyrolysis-gas chromatography (PGC) has established
itself as an effective technique for the identification of
non-volatile materials such as polymers, paints, resins and
microorganisms. PGC has the ability to provide a reproducible
succession of peaks from a single parent material which, in
effect, represents a fingerprint of that compound, similar to
that provided by an infrared spectrum. The pyrogram can con¬
siderably increase the certainty of identification and provide
a means for classification. Thus far, applications of PGC
to the pharmaceutical and related areas have not appeared in
the literature. The quantitative potential of the technique
has remained untapped even in those areas where it has been
Vlll


used most extensively over the past 15 years. The advent
of more sophisticated pyrolysis instrumentation now enables
the entire heating profile to be defined, thus assuring a
high degree of reproducibility and increasing the potential
for quantitative applications.
This study investigates the possible qualitative and
quantitative applications of PGC to pharmaceutical analysis,
in two series of experiments, with a number of penicillins
and cephalosporins and with several food and drug items
containing sodium saccharin. In addition, preliminary
investigations are discussed concerning use of the technique
for the diagnosis and study of a variety of metabolic dis¬
orders. These studies are based on abnormalities observed
in the pyrograms of urine samples.
IX


SECTION I
INTRODUCTION
Reaction Gas Chromatography
Pyrolysis-gas chromatography (PGC) is a specific
application of reaction gas chromatography coupling
gas chromatography with chemical reactions. The
combination of physical methods with chemical reactions
widens the scope of analysis and provides an increased
potential for problem solving. In chromatographic
terms, this added potential is reflected in an ability
to alter partition ratios or detector responses through
the chemical transformation of one or more components.
The subtractive method of reaction gas chromatography
and PGC represent the two extreme examples of partition
ratio alteration. In the former the ratio is made so
large that the compound is not eluted at all. In the
latter the ratio is changed such that an otherwise highly
retained material may be eluted under less rigorous
operating conditions.
1


2
Nature of the Chemical Reaction
in Pyrolysis-Gas Chromatography
Pyrolysis may be defined as decomposition at
elevated temperatures in the absence of oxygen. When
a material is heated to high enough temperatures,
thermal excitation of the individual molecules of that
material becomes sufficient to cleave certain bonds in
the molecules and generate free radical fragments.
Pyrolysis conditions can usually be varied such that
fragmentation results in the formation of comparatively
volatile components which can then be analyzed by con¬
ventional gas chromatography. The nature and the quantity
of these fragments will reflect the elemental and structural
character of the parent material as well as the conditions
of pyrolysis.
The free radicals initially formed during pyrolysis
arise from the splitting of the molecule at its weakest
links. The average carbon-carbon bond energy, or bond
dissociation energy, is about 90 Kcal/mole.^ Thermal
excitation of molecules becomes sufficient to break such
bonds at temperatures in the range of 450° to 650°C.^ The
mechanism which best accounts for the kinetics of organic
pyrolysis reactions was first proposed by Rice.^ Subsequent
work has strengthened his basic premises which are summarized
below.


J
To illustrate, consider the dissociation of a
paraffin:
1) R* + P -> R* + RH
An initially formed free radical (R*) abstracts
a hydrogen atom from a neutral paraffin molecule (P).
The structure of the newly formed free radical (R*) is
dependent on the ease of abstraction of a hydrogen from
a primary, secondary, or tertiary position on the neutral
paraffin molecule. In studies of hydrogen abstraction
from aliphatic hydrocarbons, the rates have been found
to be in the order, primary order is independent of the nature of the attacking
free radical species and is attributed to the strengths
of the C-H bonds being broken. Consequently, the weakest
of the C-H bonds, the tertiary, are broken at the fastest
rate. Looked at another way, the most reactive hydrogens,
the tertiary, are abstracted at the fastest rate.
2) Ri R'i 0L1 + R*
Although free radical rearrangements are not common,
they do occasionally occur in hydrocarbon free radicals,
following the propagation sequence, if the newly formed
radical is a C^-fragment or larger. The rearrangement can
take the form of a hydrogen shift from a primary to a
secondary site which requires an activation energy of 4
Kcal/mole. Either the R* or R'* species will subsequently
fragment a,8 to the carbon atom bearing the unpaired electron,


4
with the dissociation favoring formation of the more
stable free radical (R*) if more than one 3-bond occurs
on the same molecule. This reaction is accompanied by
the formation of the olefin (OL).
3) R* + 0Ln * RnR*
Another type of propagation reaction is now possible
wherein a free radical species can add to an olefin.
Addition occurs in such a way as to form the more stable
free radical, i.e., primary and secondary radicals would
react to form secondary and tertiary radicals, respectively.
4) a. 2R* -> RR
b. 2R* -* R + OL
These two radical reactions terminate the chain
process by a) combination of two radicals or b) hydrogen
abstraction by one radical from another to generate, in
this hydrocarbon example, an alkane and an alkene.
Pyrolysis Classification Based on Extent
of Degradation
As mentioned earlier, the volatile fragments formed
as a result of pyrolysis will reflect the elemental and
structural character of the parent material. In practice,
the pyrolysis conditions can be varied such that either
one or the other of these features will predominate. The
parameter which most affects the nature of the fragments


5
formed on pyrolysis (the pyrolysate) is the final pyrolysis
4
temperature. Beroza and Coad's review of reaction gas
chromatography grouped pyrolyses according to the extent
of degradation suffered by the parent material. According
to this classification, pyrolysis carried out in the 100°-
300°C range is designated as thermal degradation. This
is the mildest of pyrolytic procedures and often can occur
in a conventional injection port of a gas chromatograph,
requiring no special apparatus. For example, when this
5
procedure is applied to tert-amine oxides they undergo
Cope elimination:
I I
H
V
NR
+
100°-150°C
0 -
A
>
C = C + H-O-NRo
x- \ z
The subsequent analysis can then be performed on the olefin
or hydroxylamine generated, or both; the method selected
depends on chromatographic conditions.
Mild pyrolysis describes thermal reactions carried
out in the 300°-500°C range where many types of carbon-
carbon bond cleavage can occur. It has been reported that
amino acids subjected to pyrolysis at 300°C for three
minutes decarboxylate and give characteristic "amine
6
profiles." Lower aliphatic branched and straight-chain
alcohols have been dehydrated by passing the compounds
through a quartz column (at 470°-560°C) containing Chromasorb
7
P. Identification is based on the corresponding olefins
that are generated.


6
Normal pyrolysis covers the 500°-800°C range, and
it is in this zone that the majority of PGC analysis
has been done. At these temperatures most materials
are fragmented into small molecules. However, a sub¬
stantial amount of information concerning the structure
characteristics of the parent molecule can be derived
from their analysis. Polymer chemists have had a great
8
deal of success in deducing the structures of polymers,
9 10
plastics, elastomers, and related materials from
analysis of pyrolysis fragments generated at these
temperatures. Typically, compounds are assayed according
to the amounts of low-boiling material produced upon
fragmentation which are directly correlated with the type
11
and number of functional groups present in the parent.
Vigorous pyrolysis (800°-1100°C) causes extensive
fragmentation of molecules and will usually result in a
pyrolysate dominated by low-molecular-weight compounds.
The extreme temperatures used approach conditions of
elemental analysis and often much of the information
obtained under normal pyrolyses (500°-800°C) concerning
molecular structure is lost in the higher temperature
range due to the extensive dismantling of the parent
compound. Nevertheless, a large number of successful
analyses have been conducted under these vigorous condi¬
tions including the present work to be described in later
sections.


7
Utilization of the Pyrolysis-Gas
Chromatogram: The Pyrogram
The chromatogram resulting from PGC is commonly
referred to as the pyrogram and this term will be used
in the remainder of this report. It may be inferred
from the preceding discussion of free radical mechanisms
and the nature of the pyrolysate that the chief concern
of PGC is the identification of the individual peaks
which make up the pyrogram. Unquestionably, this type
of analysis provides an abundance of information which
facilitates parent structure elucidation and reaction
mechanisms. Peak identification, however, is not essential
for qualitative work within a class of compounds and this
point is demonstrated in many of the more recent applica¬
tions of PGC discussed in a later part of this manuscript
(see Section I - Recent Applications of Pyrolysis-Gas
Chromatography).
The widespread use of PGC as an analytical tool can
be accounted for primarily on the basis of the excellent
reproducibility of pyrograms that can be achieved. Within
certain operational limits (which are still not precisely
defined) the pyrolysis of a wide variety of materials from
simple hydrocarbons to polymers to microorgamisms has
resulted in fragmentation patterns that approach infrared
spectra in their consistency. Unfortunately, reproducibility
on an interlaboratory basis has been severely hampered by
the variety of different pyrolysis units constructed by


8
individual laboratories. This lack of standardization
has made it virtually impossible to compile a common
set of pyrograms analogous to the universal collections
of infrared spectra, mass spectra, etc. The situation
is improving with the emergence of more sophisticated
commercial instrumentation which offers a high degree of
control over pyrolysis conditions thus assuring more
reproducible characterization of materials.
Where conditions have been standardized, pyrograms
have been cataloged for qualitative identification. The
12
Federal Bureau of Investigation has set up a PGC
laboratory in Washington, D.C. and has obtained pyrograms
of all of the different paints that have been used by
car manufacturers for the past 15 years. This constitutes
a reference library of well over 1,000 pyrograms. Paint
chips recovered following a hit-and-run automobile accident
can be sent to this laboratory for PGC analysis and the
identity of the vehicle type confirmed by comparison of
the pyrogram with those in the F.B.I. library. This
technique is called fingerprinting and experimental evidence
would seem to justify use of the term.
Some variation of pyrolysis and column conditions can
usually be found such that the fragmented material will
yield a reproducible succession of peaks that will be
unique within its class of compound. Thus, a data bank such
as that compiled by the F.B.I. allows for the differentiation


9
of over 1,000 paints, even those from the same manu¬
facturer, strictly on the basis of the "fingerprint"
provided by PGC.
The fingerprinting technique enjoys widespread
13
popularity in the areas of fiber and textile analysis,
14 15
toxicology, and microbiology. In those situations
where analysis is limited to a particular type of non¬
volatile material or compound, fingerprinting is often
a sufficient method. Moreover, analysis which can be
carried out without the need for more sophisticated peak
analyzing apparatus offers considerable cost benefits in
instrumentation, servicing, and operation. A better
appreciation of the analytical capabilities of the various
PGC methods is essential for determining the type of
instrumentation necessary for a particular analytical
problem. A review of the other current applications of
PGC is offered to provide such insight.
Recent Applications of Pyrolysis-Gas Chromatography
Polymer Analysis
Previous to the introduction of PGC techniques,
characterization of organic polymers was done primarily
with nuclear magnetic resonance (NMR) and infrared
16 17
spectroscopy. Perry and Martinez and Guiochon have
discussed the advantages of PGC over these methods in detail


iü
However, the single greatest advantage of PGC is the
ability to fragment these complex materials into simpler
pieces which can then be separately characterized and
quantitated or collectively interpreted as the polymer
fingerprint.
17
Martinez and Guiochon were able to correctly
identify the various types of phenol-formaldehyde
polycondensates using PGC. A number of resins which
they prepared from various mixtures of pure phenol,
3-methyl phenol, and 3,5-dimethyl phenol, all provided
reproducible pyrograms easily distinguishable from one
another on the basis of qualitative or quantitative
differences. Under the PGC conditions they used, the
majority of the pyrolysis fragments consisted of a variety
of methyl-substituted phenols which were identified by
a comparison of the retention times of the pyrolyses
peaks with the retention times of solutions of known
phenols. In this way, the mole percent of the phenolic
compounds formed by pyrolysis were determined and equated
with the relative concentration of the phenols, 3-methyl
phenol and 3,5-dimethyl phenol, in the prepared resins.
Semiquantitative analysis of methacrylate and
styrene polymers in organic coatings was achieved by
18
Esposito using the internal standardization technique
with PGC. Pyrolysis conditions were adjusted so that a
maximum yield of monomer units were produced from the


11
internal standard and the polymers to be assayed. The
polyethylmethacrylate internal standard yielded 98%
ethylmethacrylate on pyrolysis. The polymers assayed
also produced essentially one peak on heating, greatly
reducing the possibility of interference from plasticizers
and modifying resins in the coatings.
19
Iglauer and Bentley developed an elaborate analytic
system for organic polymer identification which consisted
Q
of a pyrolysis unit coupled to an 1100 Pyrochrom Pyrolyzer'w'
manufactured by Chemical Data Systems. This unit was
modified by those investigators in order to allow separation
of the pyrolysate into a dual flow system equipped with
thermal conductivity and flame ionization detectors. The
smaller fragments arising from pyrolysis are diverted to
a self-contained gas chromatograph within the Pyrochrom
unit. Here the compounds pass through the thermal con¬
ductivity detector and their identity and quantity are
determined by comparison with a reference model for the
internal gas chromatograph which indicates the positions
of all major small molecules, e.g., CO, CH4, CO2, C2H4,
C2Hg, C^Hg, H20, S02, HCN, CHICHO and C^Hg. Larger
fragments are diverted to a second gas chromatograph
equipped with the flame ionization detector. The column
oven is operated with linear temperature programming which
helps to resolve the higher boiling components in a
reasonable time period. The identity of these peaks is
also based on comparison with a reference model.


12
This apparatus generates a substantial amount of
information concerning the nature of the polymer. Both
of the gas chromatographs produce pyrograms which can
be used as low-and high-molecular-weight fingerprints
of the material. The identification of the individual
peaks in the low-molecular-weight pyrogram provides
information concerning the functionality of the polymer,
whereas similar analysis of the high-molecular-weight
fragments can yield useful information as to the nature
of monomeric units and molecular components of the
parent material.
Hydrocarbons - Pyrolysis Mechanisms and Kinetics
Much of the work done with hydrocarbons has been
concerned with the mechanisms of pyrolysis and the effect
of varying pyrolysis conditions on thermolysis patterns
20
and product distribution. Fanter et al. studied the
pyrograms of 83 Cg-Cpg compounds pyrolyzed over a 50°C
temperature range (575°—625°C). They found that the PGC
patterns changed very little over this range and that all
of the 83 hydrocarbons could be differentiated on the bases
of their fragmentation patterns with the exception of cis-
trans isomers. Prior to the pyrolysis experiment, the
retention times of all the C-^-C^g normal alkanes were
determined, as well as those for a number of alkenes and
alkynes. In this way, the authors were able to ascertain


1J
the relative amounts of these materials formed upon
fragmentation of the parent molecule. Their data for
n-hexane, in particular, was in good agreement with
Rice's proposed free radical mechanism discussed earlier
(See Section I - Nature of the Chemical Reaction in
Pyrolysis-Gas Chromatography).
21
Levy and Paul used hexadecane as a test compound
to study changes in pyrolysis fragmentation products as
a function of pyrolysis conditions and sample size. In
order to better quantify their results, the authors
normalized the distribution of pyrolysis fragmentation
products so that the total area of all the peaks in the
pyrogram (excluding residual unpyrolyzed parent compound)
n
would equal 100% = 100 E (X /EX ) , where X is the area
i=l i i i
of any peak in a pyrogram of n peaks. Fragmentation data
for hexadecane covering a pyrolysis range of 580°-650°C
was collected and analyzed. values for peaks corres¬
ponding to C^-Cg olefins were plotted against the percent
relative cracking ratio which is essentially the fraction
of original compound which has been decomposed and is a
reflection of increasing pyrolysis temperatures. As
expected, the trend was toward an increasing percentage
of smaller fragments and a decreasing percentage of larger
fragments as the percent relative cracking ratio (cracking
severity) increased. For this particular compound, the
pattern seemed to center around 1-pentene which maintained


14
a slope very close to zero over the temperature range
studied. The change in the fragmentation pattern as a
function of sample size was insignificant even with a 100-
fold increase in the amount of hexadecane pyrolyzed.
22
In a subsequent paper, Groenendyk et al. were able
to determine the thermolytic- dissociation rates of both
hexadecane and a fatty acid methyl ester by varying
reaction temperature and flow rate through the pyrolysis
chamber. The first order pyrolysis rate was then
calculated from In C/Co = Ktr where tr is the temperature
corrected residence time of the reactant in the pyrolysis
chamber, K is the rate constant, C is the remaining
concentration of parent compound after reaction and Co,
the original concentration of parent compound. These
rates were then fitted to the Arrhenius equation by linear
regression techniques to determine the activation energy
and the pre-exponential term.
Identification of Microorganisms and Fungi
23
In 1965, Reiner introduced PGC as a technique to
detect, characterize and classify bacteria by visual
examination of their pyrograms. Since that time, PGC has
been used for the differentiation of a wide variety of
microorganisms and fungi.
24
Sekhon and Carmichael utilized PGC techniques in
lieu of the traditional classification methods of fungi


(i.e., those based on gross and microscopic morphology)
to characterize a number of dermatophytes belonging
to the genera Nannizzia, Arthroderma and Microsporum.
Pyrograms of replicate samples prepared from the same
colony or from two separate colonies of the same strain
appeared nearly identical. However, the authors did
find that sample size, colony age, and especially the
nature of the culture medium all exerted noticeable
effects on the quantitative and qualitative appearances
of the pyrograms.
2 5
Meuzelaar and in't Veld, using a modified PGC
system with a high-frequency induction heating filament
pyrolyzer (Curie point pyrolyzer), were able to attain a
greater degree of qualitative and quantitative repro¬
ducibility with bacterial test samples. In order to
assure good reproducibility of sample sizes, the authors
developed a technique whereby a 5-15 V1ICS2 suspension of
the freeze-dried bacteria (Neisseria meningitidis and
Neisseria sicca) was applied to the filament as a thin and
uniform coating. This method allows for the most efficient
and reproducible heat transfer to the sample, which,
according to Farre"-Ruis and Guiochon,^ is one of the
controlling factors in the decomposition of polymers and
probably of many other organic materials. The use of the
Curie point pyrolyzer offers additional control over another
factor in decomposition which is considered crucial by


16
Farrea-Ruis and Guiochon, the rate of heating. The ferro¬
magnetic filaments are well known to be highly reproducible
in attaining the Curie temperature and the high-freguency
induction heating of these filaments assures a uniform
heating rate (milliseconds). Use of these innovative
methods allowed Meuzelaar and in't Veld a level of
qualitative and relatively quantitative reproducibility
such that the pyrograms obtained in the series from the
same samples were nearly superimposable.
Continuing work in this area has led to increasingly
sophisticated instrumentation and technique. Using 500-ft
capillary columns and disposable pyrolysis chambers,
27
Quinn was able to resolve approximately 200 peaks from
the pyrolysates of representative bacteria, yeasts, fungi,
and mycoplasma. Sensitivity of the system was such that
peaks equivalent to 1 nanogram were detectable. Other
28
researchers have been able to show differentiation of
the fungi belonging to the Aspergillus flavus group at
the species and strain level using apparatus such as
Quinn's. Presently, materials being examined include
Salmonella, cell walls of bacteria, whole cells,
2 9
Leptospira, and Clostridia, among others.
In many respects, PGC compares favorably with more
conventional "wet" biological and serological methods
for classification and identification of biological
materials, especially because it is simple, rapid, and


well suited to automation and electronic data processing.
However, lack of interlaboratory uniformity of pyrolysis
systems again precludes a wide-spread use of PGC in
these biological areas as it does in those areas
discussed previously. There is evidence, however, that
the ribbon-type probe/pyrolysis unit used in the present
study is becoming standard apparatus in the field and
should lead to uniform PGC instrumentation in the coming
years.
Toxicological and Pharmaceutical Analysis
The combination of gas chromatography and pyrolysis
as an analytical tool in drug and poison identification
was first popularized by Dr. Paul Kirk of the School of
Criminology at the University of California, Berkeley
30
campus. As early as 1962, Kirk and Nelson were able
to characterize 27 barbiturates solely on the basis of
the "fingerprints." A later experiment carried out with
these same compounds attempted to identify the pyrolysis
fragments. Although many similar fragments were found in
the series of 27 drugs, a number of unique nitrile
derivatives was identified which represented the pre¬
dominant peaks in the pyrograms and, in fact, were found
to be the very peaks which had allowed Kirk and Nelson
previously to differentiate the compounds. The nitrile
derivatives contained the methylene carbon atom with the


18
double substitution characteristic of the individual
barbiturates, e.g., the major peak in the pyrogram of
probarbital (R^ = ethyl, R£ = isopropyl) was found to
be 3-methyl-2-ethyl butanenitrile:
Me
I
H_C-C-C-C=N
I
Et
0
Kirk and his associate were able to refine chroma¬
tographic and pyrolytic conditions so that the various
barbiturates could be identified (within the group) by
the one characteris.tic nitrile peak in the pyrogram.
These examples of imides breaking down to nitriles must
be considered an energetically favorable thermolysis
process as this same phenomenon has been observed
during the present PGC studies of saccharin (o-sulfo-
benzoic acid imide), the penicillins and cephalosporins.
30
In addxtion to the barbiturates,- Kirk and Nelson also
studied the phenothiazines and the morphine alkaloids.
Neither of these classes of compounds produced a
characteristic peak such as was seen in the barbiturates;
however, their identification could be based on the
variations in the amounts of low-molecular-weight pyrolysis
products generated. The phenothiazines were characterized
principally by the relative amounts of methane, ethene and
propene produced, whereas the morphine alkaloids were
characterized by the amounts of methane and ethene.


Computer analysis of the relative peak heights of these
prominent peaks permitted the identification and
differentiation of the individual members of these
drug groups.
A final example of a rather sophisticated and
expensive analytical technique involving pyrolysis in
31
combination with GC/MS is described by Merritt et al.
which combines PGC/MS with a computer data system. In
this experiment, the phenylthiohydantoin (PTH) derivatives
formed during Edman degradation of proteins are charac¬
terized by PGC which provides unique fingerprints of each
PTH derivative. These pyrograms are then analyzed peak by
peak with the mass spectrometer. This data is fed to the
computer where the pyrogram is transformed from a histo-
graphic to a Cartesian form, i.e., the abscissa is divided
into a number of positions corresponding to the retention
times for the various common constituents of the pyrograms
and the ordinate is divided into arbitrary units for
assignment of intensity levels. On this basis, a
diagnostic code for the individual PTH derivatives is
developed. For example, proline PTH was represented in
this manner as CIF2GIJ2L5. Such information can then be
stored and ultimately recalled to yield the protein amino
acid sequence.


Purpose of Present Research
The present work was concerned with developing new
analytical methods for food and drug analysis using
pyrolysis techniques. PGC complements conventional gas
chromatography in that it lends itself to the analysis
of non-volatile materials. As one traverses a series of
materials arranged in order of decreasing volatility, a
point is reached at which assays, using conventional gas
chromatography techniques (involving derivatization and/or
abnormal chromatographic parameters), become too time
consuming and often fail to maintain minimum limits of
precision and accuracy. It is at this point that the
analyst must decide on ancillary methods. In this labora¬
tory the alternatives investigated have been high-pressure
liquid chromatography and pyrolysis-gas chromatography.
Studies of the latter method are described in the following
sections.
The major concerns of these studies were (1) standardi¬
zation of PGC to foster a uniform pyrolysis system and aid
in the interlaboratory reproducibility of experiments; (2)
extraction of the maximum amount of information from as
simple and economical a pyrolysis system as possible; (3)
development of a quantitative method with a degree of
precision and accuracy comparable to conventional gas
chromatography.


21
On the first point, one can only attempt to abide
by the trends within the field. An exhaustive literature
review of this area was carried out before deciding on
the filament-type platinum ribbon probe and versatile
pyrolyzer unit manufactured by Chemical Data Systems (Figure
1). This instrument offers a high degree of control over
the various pyrolysis parameters discussed previously
and makes possible that degree of reproducibility essential
for the standardization of the technique.
With regard to the second point, expense and
versatility, the pyrolysis unit used in the present
study can be purchased for approximately $1300 (1976), and
is readily adapted to integrate with a number of commercially
available gas chromatographs. Assuming most analytical
laboratories are already in the possession of a gas chroma¬
tograph, this outlay can be considered nominal when viewed
against the total cost of outfitting such a laboratory for
general instrumental analysis.
Needless to say, pyrolysis systems can be highly
sophisticated and so correspondingly costly, for example,
systems such as those involving integrated PGC/MS/computer
analysis. A wealth of information can be provided by such
systems which include a peak analyzing component, yet
pyrograms alone are usually quite sufficient for qualitative
work. Individual peak analysis does, however, become
essential where the concern is for the mechanism of


22
thermolysis or the determination of mole percent of
monomers as in the polymer studies described. It will
also be pointed out later that peak analysis may well
be a prerequisite for absolute quantitation (in contrast
to relative quantitation, discussed in the polymer
studies) as an assurance of good linearity.
Finally, it has been noted that the pyrolysis
techniques reviewed in the literature, to date, have
been limited to qualitative and relative quantitative
analyses. Despite the increased refinements in pyrolysis
systems, no studies have been reported on the absolute
quantitative analysis of materials using pyrolysis
techniques. As used here the term, absolute quantitation,
means reproducibly pyrolyzing the entire amount of sample
applied to the heating element. In a majority of the
studies reviewed, no attention was paid to the percentage
of the parent material that is actually pyrolyzed. Those
studies which consider this parameter (cracking severity-
percent relative cracking ratio) report figurés of 20% or
below, a range where the formation of bimolecular reaction
products higher in molecular weight than the starting
material is supposedly minimized.
The results obtained in the present investigation
suggest that in pharmaceutical applications of pyrolysis
techniques to qualitative and quantitative analysis, there
is need for concern over cracking severity and secondary


23
reaction products only insofar as they interfere with the
reproducibility that is essential for such analysis.
These topics are discussed more fully in the
following sections on the characterization of a
number of food and drug materials using PGC methods.


SECTION II
QUALITATIVE IDENTIFICATION OF FOOD AND DRUG
MATERIALS USING PYROLYSIS-GAS CHROMATOGRAPHY
Advantages of Pyrolysis-Gas Chromatography
In Section I it was mentioned that the major use of
PGC up to the present has been for qualitative analysis
of non-volatile materials. The special advantages of
this technique for chemical characterizations will be
reviewed before discussing its specific applications in
the present study.
Increased Peak Identification Ability
In practice, gas chromatography is not frequently
used as a technique for peak identification. However,
there are a number of methods which improve peak identity
one of which consists of adding a quantity of the material
sought to the sample containing the unknown and then to
chromatograph this mixture. Should the compound sought
be present in the unknown, one of the peaks in the chroma¬
togram will show a relative increase in intensity as
compared with the peak produced by the original sample.


25
This is evidence that the known substance added has the
same retention time as the material corresponding to the
specific peak. A difficulty with this addition procedure
is the possibility that the test material may contain
some other substance having the same retention time as
the added known compound under the set of chromatographic
conditions employed. As discussed earlier, in Section I,
studies of the application of PGC have demonstrated the
ability of this process to yield a reproducible succession
of peaks from a single compound. The pyrogram, in effect,
provides a fingerprint of the substance, and substantially
increases the certainty of identification.
With PGC, in contrast to conventional gas chroma¬
tography, each one of a series of peaks in the pyrogram
which relates to the compound sought is increased when a
quantity of the suspected substance is added. Thus the
need to repeat chromatographic analysis with addition of
known substances is greatly diminished or eliminated.
Simplification or Elimination of Derivatization Procedures
Conventional gas chromatographic analysis of non¬
volatile materials usually necessitates prior derivatization.
This step often requires isolation of the compound sought
from an impure sample and this introduces two potential
sources of error. These are (a) inefficient isolation and
(b) irreproducible derivatization and/or formation of
multiple products. As opposed to the necessity for volatile


26
samples in conventional gas chromatography, those for
PGC must be as non-volatile as possible, otherwise they
may be lost from the pyrolysis probe before thermal
decomposition takes place. In most cases, salt formation
is the only step needed to keep the sample on the pyrolysis
probe should the parent material prove too volatile under
the chromatographic conditions employed.
Characterization of Saccharin (o-benzosulfimide)
by Pyrolysis-Gas Chromatography
Interest in the safety of saccharin (Figure 2) and
sodium saccharin for human consumption has stimulated
a number of investigations into methods of assay for
these sweetening agents. Conacher and 0'BrienJ reported
a gas chromatographic (GC) method for the determination
of saccharin in soft drinks, using diazomethane as a
methylating agent. According to their report, the
methylation of saccharin consistently gave two peaks in
a 17:3 ratio which they postulated to be the N-methyl deriva¬
tive of saccharin and the O-methyl derivative of pseudo¬
saccharin, respectively. Before derivatizing the saccharin,
which was present in the soft drinks as the sodium salt,
a number of acid-base extractions was required in order to
effect isolation. Ratchik and Viswanathan^ later reported
the determination of saccharin in a number of pharmaceutical
products utilizing silylation with N,O-bis(trimethyIsilyl)


27
acetamide.. This procedure offered some improvement
upon Conacher and O'Brien's derivatization method which
utilized diazomethane, a potentially explosive and rather
difficult reagent to prepare and use. Here, also, lengthy
extraction procedures were necessary prior to the deriva¬
tization step which, although comparatively safe and
nearly quantitative, still required over an hour's time
for sample preparation.
In view of the cumbersomeness of procedures thus
far reported, it was decided to investigate the application
of pyrolysis methods to the determination of saccharin,
in the hope of simplifying analytical work.
Discussion and Results
Saccharin and sodium saccharin. Before attempting
assay of these sweetners in food or drug materials, it
was necessary to develop an appropriate set of PGC
operation conditions for the characterization of saccharin
and its sodium salt. The pyrograms of saccharin (Figure 3)
and of sodium saccharin (Figure 4), under the conditions
used, were nearly identical. Several coinciding peaks
in the two pyrograms are seen which differ in their
relative intensities within each pyrogram. In addition,
the relative overall intensities of the two pyrograms,
derived from nearly equal quantities of material, differ
considerably. The comparatively lower intensity of the
saccharin pyrogram may have been due, at least partially,


28
to evaporation of the free imide (mp 229-230°C); pyrolysis
inlet-chamber temperature 130°C) following insertion of
the probe and prior to pyrolysis.
The pyrograms were highly reproducible, providing a
succession of peaks for which retention times and relative
sizes formed an easily recognizable pattern. During studies
on the effects of the final temperature of pyrolysis it was
found that the patterns in Figures 3 and 4 were reproducible
at temperatures as low as 700°C. However, it was further
observed that at increased temperatures (up to 900°C) the
salt and the free imide gave pyrograms of greater intensity
with no resulting change in the pattern of the pyrogram
(based on peak heights and retention times of four pre¬
dominant peaks). Pyrolysis at temperatures from 900° to
1000°C caused no discernible change in the pattern or
intensity of either pyrogram. No attempt was made at this
stage to quantitate the sweeteners; however, considerable
effort was later expended on that objective. This is
discussed in Section III.
Saccharin in soft drinks. Once a set of standard
conditions had been developed for PGC characterization of
saccharin and its sodium salt, identification of the sweetener
in diet soft drinks was investigated. The essential features
of the sodium saccharin pyrogram are clearly apparent in the
pyrogram of a typical artificially sweetened beverage (Figure
5). In experiments on the pure imide salt it was found that
amounts as small as 1.0 yg produced recognizable pyrograms.


29
With smaller quantities, a majority of the distinguishing
peaks in the pyrogram was lost. When a known portion of
pure sodium saccharin was added to a sample of beverage
of similar size, the resulting pyrogram (Figure 6) showed
an increase in intensity of those peaks which characterized
the pyrogram of the pure sodium saccharin standard (Figure
5). There could be no reservations concerning the identity
of the material since characterization was based on the
presence of a multitude of peaks representing a finger¬
print of the compound, thereby eliminating the need for
further qualitative analysis. In addition, the procedure
required no isolation or derivatization of free saccharin.'
Since it was present as its sodium salt, a form most
suited for PGC analysis, the saccharin could be directly
characterized by placing a few microliters of the soft
drink on the platinum ribbon of the probe without additional
preparation.
Saccharin in a multivitamin product. A number of
multivitamin products for children which are on the
market contain sodium saccharin as a sweetening agent.
The relatively complex composition of the multivitamin
products provided a rigorous test of PGC. Each tablet
of the selected product contained substantial quantities
of fat-soluble vitamins (vitamin A, 1.7 mg, vitamin D, 0.01
mg); and water-soluble vitamins (three B-vitamins, 19 mg.
total, vitamin C, 75 mg); as well as large quantities of
excipients (400 mg). These were in addition to the sodium


JO
saccharin (1.35 mg by weight, equivalent to 1.20 mg
of saccharin). The extraction (described below in the
Experimental Section) removed the greater part of these
fat-soluble and water-soluble materials. Interference
with the analysis of the free imide from residual
ether-soluble materials could be minimized by pre-firing
the sample in a sealed, nitrogen-swept apparatus to
reduce the possibility of secondary reactions. Repeated
firings at temperatures of 100-110°C were carried out until
no vapors could be detected rising from the ribbon surface
during heating. The sample could then be analyzed under
the usual pyrolysis conditions. The resulting pyrogram
(Figure 7) clearly showed the presence of a number of the
characteristic peaks of the pure saccharin pyrogram. Assuming
complete extraction of saccharin from the tablet, this 5.0
y 1 sample contained a maximum of 6.0 yg of the imide in
addition to residual impurities. This quantity was at least
twice as much as that required to produce a recognizable
pyrogram of the imide; that is, one which displays, in a re¬
producible fashion, a majority of the distinguishing peaks
observed in the pure saccharin pyrogram. Again, addition
of a known quantity of pure saccharin (Figure 8) reaffirmed
the identity of the extract, distinguishing between those
peaks due to residual impurities and those due to saccharin.
As with the soft drink analysis, no attempt was made, at
this stage, to carry out quantitation; therefore the free
imide was not converted into the less volatile and more
easily quantitated sodium salt.


31
Experimental
Materials. Solvents - All solvents were reagent grade
and were not further purified.
Saccharin and sodium saccharin - The sodium
saccharin was reagent grade (Penick). The saccharin,
34
furnished by another source, was prepared by dissolving
sodium saccharin in water and adding excess 6N HC1. The
precipitate was extracted with ethyl acetate and crystal¬
lized by evaporating the solvent (m.p. 229-230°C, un¬
corrected; lit. 228°C).
Soft drinks - These were purchased commercially:
Diet-Rite Cola, Diet Rite Orange, Diet Rite Ginger Ale,
Diet-Rite Grape, Tab, Fresca, and Canada Dry Ginger Ale.
Multivitamin produce - Purchased commercially:
Elusivol Multivitamin Chewable Tablets, Ayerst Laboratories,
Inc.
Apparatus. Gas Chromatograph - Carle Model 311 with
modified inlet for pyroprobe interface which was mounted
externally. The chromatograph was equipped with a flame
ionization detector (FID) and a thermal conductivity detector
(TCD) and dual 6' x 1/8" stainless steel columns, packed with
8% OV-101 on 100/120 mesh Gas Chrom Q support. Column
temperatures were variable from 125-150°C; carrier gas,
helium; ambient flow rate 50 ml/min.


Pyrolysis Unit - Chemical Data Systems Pyroprobe
150 equipped with platinum ribbon probe. Operating
conditions: final pyrolysis temperature 900°C; rate of
temperature rise 20°C/Millisecond; pyrolysis interval 1.0
second.
Procedure. (a) Soft Drinks - 1 or 2 ml of the beverage
were placed in a small test tube and gently shaken to drive
off most of the dissolved CC>2• Depending upon the
indicated concentration of sodium saccharin in the beverage
(generally in the range of 300-400 yg/ml), a 5-10 yl quantity
was directly applied to the ribbon probe. The sample was
evaporated by setting the final pyrolysis temperature at
100°C and firing several times prior to inserting the probe
in the pyrolysis interface.
(b) Multivitamin Product - One tablet (average weight
500 mg) was placed in a 120-mm test tube and pulverized
with a stirring rod. A small portion (5 ml) of diethyl
ether was added and the mixture could be either centrifuged
or simply allowed to settle for a minute, after which most
of the ether was removed with a pipet. This procedure
was repeated five times, or until evaporation of the ether
portion revealed no detectable residues in the flask. The
remaining solid in the test tube containing the sodium
saccharin was mixed with 2 ml of 6N IIC1 to convert the
saccharin to the imide. This mixture was extracted with


33
three 5-ml portions of ether which were then passed through
anhydrous Na2S0^ and collécted. The solvent was evaporated
under vacuum and the residue dissolved in 1.0 ml of
absolute ethanol. A 5-y1 aliquot of this solution was
usually sufficient for direct analysis by PGC.
Characterization of Penicillins and
Cephalosporins by Pyrolysis-Gas Chromatography
These antibiotics are B-lactams. They are amino acid
derivatives in which the amino acids are simultaneously
substituted bases and substituted acids. This confers
on them the property of being amphoteric. Because of
their pronounced polar nature, chromatographic determinations
of these compounds has been limited to paper chromatography
and more recently thin-layer chromatography.
Qualitative analysis of a number of penicillins (Figure
9) and cephalosporins (Figure 10) was investigated, using
PGC, in the hope of developing a detection technique which
would complement these separation methods.
Discussion and Results
Benzyl penicillins. The pyrograms of the penicillins
studied are shown in Figures 11-20. In this series, the
basic penicillanic acid nucleus remains unchanged. Dif¬
ferentiation among this group was thus dependent on the
fragmentation patterns of the different side chains and


34
their effect on the overall fragmentation pattern of
the molecule.
All the penicillins studied provided unique
pyrograms under the given conditions (see Experimental
section, below), with the exception of carbenicillin
(Figure 11) and penicillin G (Figures 12 and 13) whose
pyrograms could not be distinguished from one another.
Structurally, these two compounds differ only at the
benzyl carbon of the side chain (R-^) where carbenicillin
has a carboxyl group in place of one of the two hydrogens
in penicillin G. Groenendyl et. al. explored the use
of pyrolysis for the identification of functional groups
in parent molecules. Under the conditions employed by
those workers (pyrolysis temperature 600°C; pyrolysis
interval 4 seconds) it was found that carboxylic acids
and esters could be identified by the high yield of CO2
and H2O. The decarboxylation of carbenicillin seems a
likely route of decomposition, especially because it is
a (3-keto acid which should readily lose CO2 at elevated
temperatures. The CO2 given off was virtually unretained
and remained undetected amidst the other low-boiling
fragments that make up the first 30 seconds or so of the
pyrogram. This pyrolytic pathway would result in essentially
identical pyrograms for carbenicillin and penicillin G.
The pyrogram of ampicillin (Figure 14) was easily
distinguished from those of carbenicillin and penicillin G,
despite the fact that the only structural difference is,


35
again, at the benzyl carbon of the side chain where
ampicillin has a primary amine function. No analogous
mechanism exists here for elimination of the substituent
as in the case of carbenicillin. A different degradation
pathway is followed, as is evident from the pyrogram.
A close study of the pyrograms of the several benzyl
penicillins showed that the differences really were
quantitative, i.e., the retention times of the major peaks
were the same, but they differed in relative intensity.
The predominant peak in the penicillin G and carbenicillin
pyrograms was at ~6-1/2 min., whereas the predominant peak
for ampicillin occurred at '-1-3/4 min. All three pyrograms
had common peaks at these and other retention times; only
the relative intensities differed.
Isoxazolyl penicillins. These three penicillins were
readily differentiated from the other antibiotics and from
each other. Oxacillin, (Figure 15) the parent compound,
had a unique peak at -8-2/3 min., and lacked completely
the intense, more highly retained peaks that characterized
the mono- and dichloro- derivatives. Cloxacillin (Figure
16) and dicloxacillin (Figure 17) both had a rather intense
peak at -5-3/4 min.; dicloxacillin displayed an additional
large peak at almost 16 min. Interestingly, all three of
these penicillins exhibited an intense peak at -1-3/4 min.,
as was seen in the benzyl penicillins and in the benzyl
cephalosporins.
Methicillin, nafcillin, penicillin V. These three
compounds are substantially different in structure from each


36
other and from the rest of the penicillins .investigated.
This difference is seen in their pyrograms (Figures 18-
20) where there were no major peaks at those retention
times that characterized the benzyl penicillins.
Unquestionably, fragments derived from the side
chains of these and the other antibiotics represent the
major peaks in the pyrograms obtained in this study.
This could explain the noticeable lack of peaks in the
nafcillin pyrogram (Figure 19) especially when the physical
characteristics of the most likely fragments formed upon
pyrolysis are considered. In this case, the pyrolysis
temperature used (875°C) would be expected to generate
either (3-ethoxynaphthalene (bp 280°C) or B-naphthol (bp
295°C) in large quantities. Either one of these would be
indefinitely retained on the column (oven temperature 100°C)
and so would be absent from the pyrogram.
Cephalosporins. The pyrograms of the cephalosporins
studied are shown in Figures 21-24. Two of these anti¬
biotics gave virtually identical pyrograms. The peak seen
at ~3 min. in the pyrogram of cephalexin (Figure 21) and
cephaloglycin (Figure 22) consistently appeared at a slightly
greater intensity in the pyrogram of the latter. However,
because the peak was of very low intensity, it could not be
used as the criterion for differentiation of the two com¬
pounds. These two benzyl cephalosporins both displayed a
very intense peak at ~1-3/4 min. which was previously noted


37
for the benzyl and isoxazolyl penicillins. The common
structural unit to be found in all of these compounds is
the side chain, - C^, and where there is a nitrogen
function attached to C^, the ~1-3/4 min. peak is the most
intense peak in the pyrogram. In addition, the pyrograms
of these two cephalosporins also contained peaks at -6-1/2
and -14 min., similar to benzyl penicillins.
The remaining two cephalosporins are structurally
unique, as were their pyrograms. The simplicity of the
cefazolin pyrogram (Figure 23) probably relates to a facile
fragmentation of the R^-tetrazol derivative and the R^~
thiadazol ring, into low-boiling fragments.
Experimental
Apparatus. Gas Chromatograph - Varian Model 2740 with
modified inlet for pyroprobe interface which was mounted
externally. The chromatograph was equipped with an F.I.D.
detector. Column temperature was 100°C; carrier gas, helium
at 60 ml/min. ambient flow rate; chromatographic column: 6' x
1/8" i.d. stainless steel column packed with 3% XE-60 on 80-100
mesh Gas Chrom Q solid support.
Pyrolysis Unit - Chemical Data Systems Pyroprobe 150
equipped with platinum ribbon probe. Operating conditions:
final pyrolysis temperature 875°C; rate of temperature rise
20°C/millisecond; pyrolysis interval 1.0 second.


38
Antibiotics. All of the drugs studied were obtained
from the pharmacy at the University Teaching Hospital. Most
of these were in the injectable form, as their sodium salts.
Those in capsule form were present as the free acid and
were converted to their sodium salts by dissolution in
an equimolar aqueous solution of sodium hydroxide. The
possibility of contamination from sodium citrate, citric
acid, or dihydrogen sodium phosphate in the buffered
preparations was examined and found to be negligible. As
an example, a 4.67% by weight additive of sodium citrate
in one buffered penicillin G preparation had to be increased
60-fold before its contribution to the pyrogram was de¬
tectable .
Procedure. Qualitative analysis of the antibiotics
was carried out with amounts ranging from 10 yg to 40 yg of
the sodium salts depending on the complexity and intensity
of the pyrogram.


SECTION III
QUANTITATIVE ANALYSIS OF FOOD AND DRUGS
USING PYROLYSIS-GAS CHROMATOGRAPHY
Prerequisites for Quantitation
It was mentioned previously that the quantitative
applications of PGC have been limited to the areas of
polymer and hydrocarbon analysis. For the former, relative
quantitation of monomer units has been achieved with PGC,
aiding in the classification of the parent polymer. For
the latter, the relative percent distribution of pyrolysis
fragments has been determined for a number of hydrocarbons.
This data has greatly facilitated the deduction and pre¬
diction of pyrolysis mechanisms.
Both these areas of application represent examples of
relative quantitation, i.e., a measure of the relative amounts
of fragments, normalized and compared to one another. In
neither instance is there any attempt made to volatilize
the entirety of the sample, and only in the case of the
hydrocarbon analysis is the amount of parent material placed
on the heating surface or the amount of parent material
actually pyrolyzed taken into consideration.
39


40
One reason why absolute quantitation (an absolute
measure of the entire sample placed on the heating element)
has not been more closely examined has been the lack of
suitable instrumentation to provide the type of reproduci¬
bility essential for such quantitation. In addition to
this, one observes that pyrolysis of single components often
results in the formation of a multitude of peaks which would
seemingly complicate any attempts at quantitation. Also,
further difficulties may arise where cracking ratios or
cracking severities (see Section I: Hydrocarbons-Pyrolysis
Mechanisms and Kinetics) are in excess of 20%. This is an
area where formation of bimolecular reaction products higher
in molecular weight than the starting material have been
. , 7,35,36
reported.
The pyrolysis unit used by this group offered a high
degree of control over the various pyrolysis parameters.
This capacity for reproducibility was reflected in the
reproducibility of the pyrograms obtained in the qualitative
work described. It was felt that the previous performance
of the system justified an investigation of quantitative
applications.
Cracking Severity Measurements
Regardless of the complications reported to arise when
the cracking severity exceeded 20% of the parent material,
it was obvious that absolute quantitation necessitated a


41
burn or pyrolysis efficiency approaching 100%. A simple
test was devised to determine if and under what conditions
the pyrolysis unit employed could meet this requirement.
The material to be tested was applied to the ribbon
surface, usually as an aqueous solution. This method of
application assured a relatively uniform distribution of
sample over the "hot" middle 20 mm of the 2.0 X 40 mm
ribbon. Once applied, the water could be evaporated by
one of three methods, 1) pulsing the probe externally at
100°C, 2) pulsing the probe internally at 100°C or, 3)
drying externally with the aid of a gentle stream of warm
air. Chromatographic conditions employed for a particular
analysis as well as the nature of the material under study,
determined which method was used. Prior to pyrolysis,
attenuation was adjusted to give the maximum recorder
response (based on the most intense peak in the pyrogram)
for the sample size and pyrolysis temperature being tested.
Following an initial pyrolysis and generation of the pyrogram,
the sensitivity was increased ten-fold and a second pyrolysis
was performed. After dividing the peak response in the second
pyrolysis by 10, the cracking severity or burn efficiency
could be determined using the formula, P^/(P-^+P2), where P^ =
peak response for the initial pyrolysis and P2 = peak response
for the second pyrolysis. It was generally found that the
cracking severity was more a function of the final pyrolysis
temperature than the duration of pyrolysis or sample size (at


42
least up to 100 yg). Near 100% pyrolysis efficiency was
usually achieved with pyrolysis intervals of from 1 to 5
seconds at temperatures above 750°C. More often than
not, the appearance of the pyrograms was unaltered through
the range 750-1000°C and the tendency was to use final
pyrolysis temperatures in excess of 800°C for quantitative
studies. These efficiency tests were carried out only with
2-10 yg sample sizes since it was felt that any deviation
which might arise with larger sample sizes (due to in¬
complete pyrolysis) would be reflected in a deviation from
linearity in the preparation of standard curves.
Peak Identification
Quantitative methods in GC and PGC. One of the diffi¬
culties which was mentioned, that could hamper absolute
quantitation in PGC, was the multiplicity of peaks which
normally result from fragmentation of a single component.
This multiplicity would only be accentuated under the
conditions necessary to achieve near 100% pyrolysis efficiency.
In theory, the entire area under the curve of a pyrogram
could be integrated to achieve quantitation. In practice,
this would be an extremely cumbersome method of assay and
no doubt, one lacking in precision and accuracy. A more
conventional and expedient method which is used in gas
chromatography, would be individual peak area analysis, or
preferably, peak area estimation by peak height measurements.
The question that arises is whether measurement of the peak


43
height of one or two peaks in a pyrogram made up of many
peaks is sufficient for quantitation of the parent material.
Experimentally, it has been found, by the present authors,
that most pyrograms do contain at least one relatively
intense peak. By manipulating pyrolytic, or more often
chromatographic conditions, this peak can often be resolved
into a shape which lends itself to peak height measurement.
This certainly is a necessary condition for quantitation
of this peak; however, in PGC it would be insufficient
insofar as quantitation of the parent material is concerned.
Here, the identity and origin of the peak must also be inves¬
tigated. The reason for this stems from the fact that the
parent material undergoes radical fragmentation and product
formation prior to chromatography and detection.
Origin of fragmentation products. The nature of these
products will be a function of the parent material, the amount
of the parent material pyrolyzed, the pyrolysis temperature,
the cracking severity, and the pyrolysis interval. All of
these factors can influence the mechanism by which the ini¬
tially formed free radicals will propagate, i.e., via uni-
molecular decomposition reactions or bimolecular combination,
disproportionation and other reactions. Although unimolecular
decomposition has been found to be the primary process under
controlled conditions with simple molecules,^'^'^ pyrolysis
of more complicated species, under the vigorous conditions
necessary for near 100% pyrolysis efficiency, would be
expected to generate a greater variety of free radicals (of


44
varying reactivity) as well as an overall, greater number
of free radicals. Such conditions may well lead to the
formation of an abnormally high number of bimolecular
reaction products and possibly products of secondary
reactions. In qualitative analysis, all the parameters
mentioned above which control the nature of the fragmen¬
tation pattern, can be held constant and thus affect the
type of reproducibility reported here and in other studies
on pyrolysis. However, in quantitative analysis and the
construction of standard curves, the concentration or mass
is necessarily a variable. With changes in sample size,
the probability of bimolecular interaction also changes.
Consequently, quantitative schemes based on peaks originating
from unusual bimolecular reactions would be expected to re¬
flect deviations in linearity due to the non-classical
mechanisms of their formation. Peak identification then
allows for the selection of the peak or peaks which will
provide the best linearity and the best representation of
the quantity of the parent sample.
Comparison of classical and non-classical thermolysis
mechanisms. The mechanisms advanced by Rice^ are detailed
in Section I of this manuscript. These mechanisms are based
on peak characterizations and kinetic data from the pyrolysis
of light alkanes at temperatures which expose these molecules
to little more energy than that needed for decomposition
(400-650°C). In addition, the degree of pyrolysis (cracking


4 b
severity) is purposely kept very small (0.2-20%) to aid
*k
in determining the identity of primary products. The
conditions found necessary to achieve near 100% cracking
severity are quite stringent by comparison and therefore
some deviations from Rice's mechanisms should be considered
(Refer to Section I for an explanation of the symbols - "Nature
of the Chemical Reaction in Pyrolysis-Gas Chromatography"):
1) R* + P -> R* + RH
1
This step becomes less probable when pyrolysis conditions
approach 100% cracking severity.
2) R# + R* + OL + R*
11 12
3 7
Doue and Guxochon have reported that some Cr to C0 alkanes
6 9
undergo very fast 1-5 and 1-4 isomerizations. Others'^
have shown that isomerizations by 1-5 and more distant
hydrogen-atom transfers are probably much faster than
radical decomposition by bond rupture 6 to the radical
site. The conditions necessary for near 100% cracking
severity would seem to have two opposing affects on radical
rearrangements. On one hand, the large energy input, during
•k
The identity of the primary products formed after the
initial fragmentation can be facilitated by plotting the
product composition against percentage composition of the
original substrate; extrapolation of the curves to zero
decomposition indicates the primary products of the decom¬
position and their relative amounts.


46
pyrolysis at elevated temperatures, should generate
relatively energetic radical ions capable of bond re¬
arrangements. At the same time, however, this large
energy input would also tend to produce smaller radical
ions, in which case, isomerization would be a much slower
process than fragmentation. Another point to consider,
with molecules more complex than hydrocarbons, would be
the possibility of rearrangements involving aryl shifts,
where the transition state is stabilized by delocalization
of the unpaired electron over the orbital system of a
benzene nucleus, e.g.,
PhMeC-CH
I 2
Ph
PhMe£-CH2
PhMeC-CH0
I 2
Ph
The second part of step 2, propagation via decomposi¬
tion, may well remain the predominant propagation reaction
even at the high radical concentrations produced during
quantitative pyrolysis. However, high concentrations
of energetic species may result in an increase of bi-
molecular propagation reactions such as step 3 in Section I,
3) R* + OL R R*
n n n n
This may be the source of the so-called secondary reaction
products which are higher in molecular weight than the parent


compound. If two such species react early in the chain
process there is a possibility that the final termination
product will be larger than the parent.
4) 2R* -> RR
2R* -> R + OL
Under conventional pyrolysis conditions used in the study
of thermal radical kinetics and mechanisms, termination
of the reaction by radical/radical interaction is unlikely
to occur to any significant extent until the concentration
of large fragments has dropped to a very low level. The
lower the initial free radical concentration and the lower
the reactivity of the radical toward combination (either
as a result of resonance stabilization or steric inacces-
sability of the free electron), the more likely that these
bimolecular termination steps will occur with simple
fragments, i.e., H*, CH*, and H*. The conditions employed
here for quantitative analysis might very well lead to
"premature" combination or disproportionation reactions
resulting in the formation of non-classical fragmentation
products.
Quantitation of Penicillins and Cephalosporins
In Section II the characterization of a number of
antibiotics using PGC was discussed in detail. The


48
pyrograms of these materials proved to be highly re¬
producible and a re-examination showed that many of
them contained peaks which could potentially serve as
the basis for quantitation of the parent drug. Cracking
severity and peak identification studies were first
carried out, adhering to the prerequisites for quantita¬
tion. The peak identification method employed consisted
of a pyrolysis unit interfaced with a GC/MS system which
is described in full below.
Discussion and Results
Cracking severity measurements. For this experiment
it was decided to chose one antibiotic from each of the
four groups into which they had been divided during the
qualitative studies (see Section II: Characterization
of Penicillins and Cephalosporins by PGC; Discussion and
Results). Once selected, penicillin G (benzyl penicillin,
Figure 12), oxacillin (isoxazolyl penicillin, Figure 15),
methicillin (Figure 18) and cephalexin (cephalosporin,
Figure 21) were then tested according to the procedure
described previously. Table I shows the results of these
cracking severity measurements and it can be seen that,
within the range of quantities studied, pyrolysis was
essentially complete. Again, it was assumed that any
substantial change in this degree of efficiency with
greater amounts of material would be reflected by deviations
in standard curves.


49
Peak identification - Instrumental methods. A
variety of methods has been established for the identi¬
fication of gas chromatographic peaks. However, in
terms of the certainty of results and the expediency
with which they are obtained, the integrated GC/MS
system is superior to any available. Fortunately, GC/MS
facilities were available to the authors and the pyrolysis
unit used in these studies was readily coupled to the
system. Normally, experimental conditions were determined
in advance on a separate PGC apparatus before peak analysis
was carried out on the PGC/MS instrument. The same
pyrolysis unit and GC columns were used in both systems.
It was decided at the outset that those peaks most likely
to lend themselves to peak height measurement should be
of primary concern. Consequently, small or unresolved
peaks were usually ignored, certainly not from a lack of
interest in their identity, but because of the limits of
the PGC/MS unit which had no computer data system.
Peak identification - benzyl penicillins. The
pyrograms of these penicillins and the identity of their
major peaks are shown in Figures 25-27. Benzyl nitrile
{6h minute peak) was present in all of these compounds in
varying amounts. Its mass spectrum (Figure 36) is
characterized by a large M-l ion which is common for
nitriles having hydrogens a to the CN group. This product


50
★
may originate by a unimolecular process since the basic
structural unit (Ph-C-C-N) is contained in the acyclic
portion of all these penicillins. In the benzyl-
unsubstituted penicillin G, dehydration and cleavage
of the acyclic portion of the parent molecule leads
directly to the product:
-> Ph-CH2-CEN
For ampicillin and carbenicillin, generation of the
product by a unimolecular process would involve loss of the
benzyl substituent, intramolecular hydrogen abstraction via
a 1-5 radical isomerization, subsequent loss of water and
fragmentation 8 to the odd electron on the developing
nitrile carbon:
★
Pyrolysis conditions here were the same as those used in
the qualitative studies described in Section II and repre¬
sent a substantial departure from the classical conditions
employed by Rice^ and other workers21'22 qn their mechan¬
istic studies. Little information is available concerning
mechanisms under these extreme conditions and consequently,
the reactions depicted here are intended to represent only
the possibility for product formation based on an extra¬
polation of known classical mechanisms.


51
O
Ph-CH -C-N-
2
'N'
7^
O II H
CO -
2
CH
CO
•y
Ph-CH -C=N
-N-
Aj
2
CH
CH
CH
Ph-CH -CEN
2
A comparison of the size of the benzylnitrile peak
in these three pyrograms shows that the amount formed
from pyrolysis of ampicillin is measurably less than for


52
the other two penicillins. This could point to cleavage
of the amine function as a rate-limiting step (AH° = +40
Real) as opposed to the generation of CO2(AH^ = -94 Real)
in carbenicillin, and the straightforward dehydration and
cleavage reaction possible with the unsubstituted penicillin
G. Also, retention of the amine moiety and generation of
benzonitrile appears to be a preferential fragmentation
pathway with ampicillin, as can be seen from the intensity
of the 1 3/4-minute peak. Loss of water and hydrogen,
followed by cleavage of the bond between the benzyl
carbon and the carbonyl group of the side chain, could
lead to the formation of benzonitrile (mass spectrum,
Figure 37) with ampicillin as well as with the other a-
amino benzyl antibiotics studied:
Ph
Two other interesting compounds identified in this
series were bibenzyl (mass spectrum, Figure 38) and 1-
phenyl-2-propanone (mass spectrum, Figure 39). Bibenzyl


was found in all three pyrograms and is clearly the product
of a bimolecular radical/radical termination reaction. The
benzyl radical is quite stable and less reactive than, for
example, simple alkyl radicals, because of delocalization
of the unpaired electron over the tt orbital system. Again,
it is noted that the amount of this radical formed from
pyrolysis of ampicillin is measurably less than for
penicillin G and carbenicillin. l-phenyl-2-propanone appears
to be formed from the combination of a methyl radical and
the benzyl carbonyl radical. Although this is a bimolecular
process, the reaction may be intramolecular, because of the
proximity of the two methyl groups on the C-3 carbon of the
thiazolidine ring. The combination of a benzyl and an acetyl
radical may seem a more realistic route to the product; how¬
ever, no reaction is immediately obvious which would generate
the latter species.
Peak identification - isoxazolyl penicillins. The
pyrograms of these penicillins and the identity of their
major peaks are shown in Figures 28-30. Again, benzonitrile
was a predominant peak occurring in all three pyrograms.
A comparison of these peaks in the three pyrograms shows a
decrease in intensity with increasing substitution of the
benzene ring. Presumably cleavage of the isoxazolyl ring
is equally probable for all three species and the relative


proportion of benzonitrile formed reflects a decreasing
probability of consecutive scissions of the two carbon-
chlorine bonds.
The mass spectra of o-chlorobenzonitrile (Figure 40)
and 2,6-dichlorobenzonitrile (Figure 41) are characterized
by the isotopic clusters around the molecular ion resulting
from the presence of one and two chlorine atoms. Clearly,
these two compounds are products of a unimolecular frag¬
mentation process and would be likely candidates for
quantitation of the parent material.
Peak identification - methicillin and penicillin V.
As was decided earlier, no extraordinary attempts would be
made to characterize minor or unresolved peaks. Hence,
nafcillin (Figure 19) was not investigated in this quanti¬
tative study. The pyrogram of methicillin (Figure 31)
is dominated by one peak which was found to be 1,3-dimethoxy-
benzene (mass spectrum, Figure 42). No evidence was found
to suggest the formation of a substituted species, i.e.,
a combination of the 1,3-dimethoxybenzene radical with
some radical other than HI This may be indicative of rapid
hydrogen abstraction from the acid side of the molecule
following cleavage of the phenyl-carbonyl bond.
The major peak in the penicillin V pyrogram (Figure 32)
was found to be phenol (mass spectrum, Figure 43). The
phenoxy radical is quite stable due to the delocalization
of the unpaired electron to the ortho- and para-positions


of the benzene ring. It might, therefore, be expected
that dimeric products would form here as occurs with
those compounds which generate the analogous benzyl
radical (on thermolysis). Five distinct dimeric products
are possible (this excludes peroxide formation); however,
none of these was identified in the pyrogram.
The phenylacetate (mass spectrum, Figure 44) formation
is probably analogous to the formation of l-phenyl-2-
propanone seen earlier in penicillin G. Because the
origin of the compound is questionable and the amount
produced is rather small, it was not considered a good
choice for quantitative study.
Peak identification - cephalosporins. Due to a lack
of prominent peaks in its pyrogram, cefazolin (Figure 23)
was not investigated in the quantitative study. The pyro-
grams of the other three cephalosporins and the identity
of their major peaks are shown in Figures 33-35.
As was the case for ampicillin, the two a-amino benzyl
cephalosporins revealed a large benzonitrile peak in addition
to a comparatively small benzylnitrile and bibenzyl peak.
Apparently the fragmentation processes for these compounds,
with regard to the acyclic side chains, are quite similar,
if not identical, and seem to be unaffected by the structural
changes in the acid portion of the molecules.
The acyclic side chains of cephalothin (Figure 35) and
penicillin G (Figure 25) are identical, except for the sub¬
stitution of thiophene for benzene in the former. The identity


of the three major peaks in the cephalothin pyrogram are,
in fact, the sulfur analogs of those peaks seen in the
penicillin G pyrogram. The mass spectra of these compounds
(Figures 45-47) are all characterized by the strong thenyl
ion at m/e 97, and the M+2 peak of the sulfur isotope.
Interestingly, the R2 substituent on the C-3 carbon of
the cephalosporanic acid nucleus would appear to be a
source of acetyl radicals. Despite this, there appeared
to be no disproportionate amount of 2-thienyl-2-propanone
formed upon pyrolysis of cephalothin. This may lend support
to the suggestion of a radical/radical combination reaction
between benzylcarbonyl (thenylcarbony1) and methyl, to
form the substituted propanone.
Preparation of standard curves. After completion of
the cracking severity and peak identification studies it
was decided to attempt the construction of standard curves
using the same four antibiotics employed in the cracking
severity experiment.
The major peak in both the oxacillin and cephalexin
pyrograms had been shown to be benzonitrile. This compound
appeared to be the product of a unimolecular fragmentation
process and therefore, a species which should display good
linearity, at least over a limited concentration range.
The major peak in the methicillin pyrogram (1,3-dimethyl-
benzene) and in the penicillin G pyrogram (benzylnitrile)
also appeared to be derived from unimolecular processes


which should display a minimum of concentration dependence
and therefore, good linearity.
The sample sizes used in preparation of the curves
ranged from 10 nanograms to 100 micrograms. Each curve
consisted of a minimum of 27 data points from 9 different
sample sizes across the indicated range. Because of the
large range, the data points were plotted in logarithmic
form (Figure 48). Table II provides an explanation of the
symbols used on the graph along with the log-log slopes and
the regression coefficients. The adherence to linearity
was quite good throughout the entire,range with no percep¬
tible deviations at large sample sizes for any of the four
antibiotics tested.
No attempt was made to quantitate the other antibiotics
used in the study, as it was felt that the four that were
chosen fairly represented the group and, indeed, were
representative of a larger group of non-volatile materials
which do not lend themselves to conventional gas chroma¬
tography methods. The experimental results clearly show
the potential for quantitative application of the PGC
system employed in the assay of these non-volatile materials.
Experimental
Apparatus. Gas Chromatograph - Varian Model 2740 with
modified inlet for pyroprobe interface which was mounted


58
externally. The chromatograph was equipped with an F.I.D.
detector. Column temperature was 110°C; carrier gas,
helium at 60 ml/min. ambient flow rate; chromatographic
column; 6 ft. x 1/8 inch i.d. stainless steel column
packed with 3% XE-60 on 80-100 mesh Gas Chrom Q solid
support.
Pyrolysis Unit - Chemical Data Systems Pyroprobe
150 equipped with platinum ribbon probe. Operating
conditions: final pyrolysis temperature 900°C; rate of
temperature rise 20°C/millisecond; pyrolysis interval
1.0 second.
Mass Spectrometer - Dupont 490-F single focusing,
magnetic sector instrument.
Antibiotics. (See "Experimental," Section II).
Procedure. Cracking severity measurements are
described in the discussion and results portion of this
section. Samples ranging from 10 yg to 40 yg of the sodium
salts of the antibiotics were pyrolyzed and subjected to
combined gas chromatographic-mass spectrometric (70 eV
electron impact) analysis utilizing a jet separator.
Identification of peak components in the pyrograms was
based on comparison with authentic samples or the matching
41
of mass spectra obtained m reference files.


Quantitation of Saccharin
Discussion and Results
Cracking severity measurements. The use of
chromatographic and pyrolytic conditions very similar
to those employed for the qualitative and quantitative
analysis of the antibiotics produced a sodium saccharin
pyrogram consisting of just two major peaks. Cracking
severity measurements were carried out (monitoring both
peaks) following the procedure described earlier in this
section. The results were very similar to those obtained
in the antibiotic series and it was decided to continue
on to the next prerequisite phase, peak identification.
Peak identification - saccharin (sodium, salt). PGC/MS
analysis revealed the two major peaks in the sodium saccharin
pyrogram to be benzonitrile and biphenyl, one of which had
been observed previously in a number of the antibiotic
pyrograms. It may be recalled from Section I that Kirk
and NelsonJU had observed the formation of nitriles upon
pyrolysis of the barbiturates. The conditions they used
were quite similar to those employed in the present quan¬
titative studies. In light of these results, it can be
concluded that nitriles will be predominant pyrolysis products
for compounds containing the amide, imide, sulfonamide,
sulfimide and related functional groups.
It was decided to base the Quantitative studies on the
benzonitrile peak, since its formation was probably the


60
result of a concentration independent intramolecular
process involving interaction of the saccharin nucleus
with the two water molecules of the hydrated salt. Either
of these intimate waters of hydration could supply the
hydrogen atom for the aromatic ring following scission
of the C-S bond, ortho to the evolving nitrile function.
Preparation of standard curves. During the course of
these experiments, many standard curves of sodium saccharin
were prepared prior to assaying the various diet beverages.
All curves displayed excellent linearity (regression co¬
efficients >0.997) throughout the range studied (1.0-10.0
Mg) •
Determination of sodium saccharin in diet beverages.
Several commercial diet beverages (sodium saccharin content:
50-150 mg/12 fluid oz.) were randomly selected and quan¬
titation was attempted according to the procedure used
earlier in the qualitative studies (see Section II). This
consisted of evaporating a few microliters of the beverage
on the ribbon surface and directly pyrolyzing the residue.
This process was carried out on several different samples
and the results showed a high degree of precision; however,
the experimental values were always higher than the levels
indicated for the particular beverage analyzed. It was
suspected that the high values were a result of interference
from some common ingredient in the beverages which, upon
pyrolysis, generated benzonitrile or another species having


61
a peak coincident with that of benzonitrile. A check
of the various beverages revealed that sodium benzoate
was the only such common ingredient indicated on the
labels. Although the amount of sodium saccharin varied
from product to product, the benzoate was always present
to the extent of 1/40 of 1% (or, 0.25 mg/ml as opposed to
saccharin levels ranging from 0.14-0.44 mg/ml). Simon
and Giacobbo^ had earlier studied the pyrolysis of sodium
benzoate and, under conditions approximating those used
in this study, they identified the major pyrolysis product
as benzene, accompanied by smaller quantities of benzyl
alcohol, phenol, biphenyl and benzaldehyde. Sodium benzoate
was then studied by the present authors under conditions
established for saccharin quantitation. Results similar
to Simon and Giacobbo's were obtained. Two major and two
minor peaks were identified by a comparison with retention
times of those products described above. Samples of sodium
saccharin and sodium benzoate were pyrolyzed separately and
as mixtures and it .was determined that the benzaldehyde
fragment arising from the pyrolysis of sodium benzoate was
coincident with the benzonitrile peak in the sodium saccharin
pyrogram.
Attempts were made to alter pyrolytic and chromato¬
graphic conditions to resolve these two peaks; however, little
success was achieved and it was decided to further purify


the samples prior to pyrolysis using extraction methods.
The difference between the pKa values of saccharin (1.60)
and benzoic acid (4.19) was sufficient to allow selective
extraction of the two compounds by appropriate acidification
of the decarbonated beverage. Application of this technique,
to the determination of sodium saccharin in two name brand
diet beverages, gave the results summarized in Table III.
Although this method of analysis proved to be substantially
more time consuming than direct assay, it still offered
the advantage of elimination of the need for derivatization
3 2 3 3
described in previous papers. ' Reproducibility appeared
to be limited, not by the instrument itself, but rather by
other random errors associated with separation, purification
and chromatographic analysis.
The series of experiments described in this section
indicate that quantitation in PGC is certainly as practical
as it is in conventional gas chromatography. As in GC, the
degree of accuracy and precision attained depends upon the
care taken with the experimental work. Ultimately a reliable
internal standard method will have to evolve for PGC in
order to increase the limits of its accuracy and precision
and to increase the scope of its application. The develop¬
ment of internal standards will follow much the same criteria
as that used in conventional GC. The internal standard must
give the same, or nearly the same, response as the species


63
being assayed. In additionr the internal standard should
elute close to the peak of interest, yet, be well resolved..
Consequently in GC, hydrocarbons are selected as internal
standards for hydrocarbon analysis, fatty acids for fatty
acid analysis, etc. In PGC, the internal standard should
undergo a similar transformation on pyrolysis, which is
all the more reason to use a homolog of the compound of
interest. As an example, benzonitrile is seen to arise from
the pyrolysis of a number of compounds containing amide or
imide type functional groups, with a variety of substituents
at the site of functionality and in the aromatic ring.
Logically, one would investigate those compounds which also
contain these functional groups with some slight variation:
0
II


These experiments have shown that there are a
number of core functional groups which either survive or
evolve from the vigorous conditions necessary to achieve
absolute quantitation. These core species can often be
found, in some form, incorporated into the structure of a
large variety of high-molecular weight compounds, polar
compounds, and generally non-volatile materials. Under
the proper pyrolysis conditions, these core species can
be generated from the parent compound and thus serve as
the basis for quantitating a large number of materials.
Accordingly, quantitation based on benzonitrile could be
internally standardized with the addition of, e.g., an
appropriately substituted amide which then would serve as
the PGC internal standard for several antibiotics, a sweeten¬
ing agent, and a number of sedative/hypnotics and PTH amino
acid derivatives.
Experimental
Materials. Solvents - All solvents were reagent
grade and were not further purified.
Sodium saccharin - (C^H^NO^SNa*2H2O) Mallinckrodt
U.S.P. powder.
Diet beverages - These were purchased commercially
(see "Experimental," Section II).


65
Apparatus. Gas Chromatograph - Varían model 2740
with modified inlet for pyroprobe interface which was
mounted externally. The chromatograph was equipped
with an F.I.D. detector. Column temperature 150°C;
^chromatographic column 12 ft. x 1/8 inch i.d. stainless
steel column packed with 3% XE-60 on 80-100 mesh Gas Chrom
Q solid support.
Pyrolysis Unit - Chemical Data Systems pyroprobe 150
equipped with platinum ribbon probe. Operating conditions:
final temperature 900°C; rate of temperature rise 20°C/
millisecond; pyrolysis interval 1.0 second.
Mass Spectrometer - Dupont 490-F single focusing,
magnetic sector instrument equipped with a jet separator
for PGC and GC/MS.
Procedure. Quantitation of diet beverages - 20 ml
of the soft drink was pipetted into a 100-ml separatory
funnel, made basic with 2 ml of 10.N NaOH, and extracted
with two 25-ml portions of diethyl ether. The combined
ether extracts were washed with 20 ml of water, the ether
discarded, and the washings were combined with the original
aqueous layer. The aqueous layer was acidified to a pH of
3.5-3.6 and extracted with three 25-ml portions of diethyl
ether. The combined ether extracts were washed with 20 ml
of water, dried over anhydrous sodium sulfate and evaporated-
to dryness (benzoic acid fraction). The original aqueous
layer was further acidified to a pH near 0 and extracted


with three 25-ml portions of ethyl acetate. The combined
extracts were washed with 20 ml of water, dried over an¬
hydrous sodium sulfate and evaporated to dryness on a
rotary evaporator. The residue was dissolved in 1.0 ml
of 0.025N NaOH and was used immediately for PGC analysis.


SECTION IV
ADDITIONAL APPLICATIONS OF PYROLYSIS-GAS
CHROMATOGRAPHY AND CONCLUDING REMARKS
The Use of Pyrolysis-Gas Chromatography for the
Diagnosis and Study of Metabolic Disorders
The Urine Pyrogram
Pyrolysis of serum and urine,. During the course of
the present work, preliminary studies of drug assay in
biological fluids was undertaken. Much of this work was
concerned with removal of endogenous materials from the
fluids (via extraction and precipitation methods) which
would interfere with the pyrolysis patterns of drugs. Use
of standard separation methods made it possible to reduce
this interference, in serum, to just one peak in the
pyrogram of a treated blank sample.
The urine pyrogram, however, was seemingly unaffected
by changes in pH, extraction with solvents, absorption
methods, etc,. If PGC analysis were to be pursued with
urine, it would probably have to be carried out on the
residues remaining after workup of the fluid for conventional
GC urinalysis.
The reproducibility of the urine pyrogram is quite
67


amazing, considering the heterogeneous nature of the
solution. The physical and chemical properties of urine
are exceedingly variable and change substantially with the
nature of the diet. Nevertheless, the pyrograms of urine
obtained from a number of adult males are qualitatively
superimposable, almost peak for peak (see Figure 49 for
an example of a typical urine pyrogram). Differentiation,
thus far, has been based solely on differences in the
relative peak heights in the pyrograms. As might be ex¬
pected, these quantitative differences can be detected in
the urine pyrogram of an individual when monitored over
several days. Such changes are presumably dietary-in nature
No study has yet been carried out to determine if acute
ailments have an effect on the appearance of a normal urine
pyrogram.
PGC/MS analysis of urine. These studies had not been
completed at the time of writing; however, those peaks which
have been identified are indicated in Figure 49. Pyrolyses
were carried out at high temperatures (900°C for 1.0 sec.)
and this is reflected in the nature of the compounds noted.
These species represent some of the core compounds mentioned
in Section III which survive or evolve in these extreme
pyrolysis conditions. These compounds probably originate
from several different materials normally found in the urine
Amino acids are excreted in the urine to the extent of 2-3
gm/day and would be expected to contribute significantly to


31
the urine pyrogram. Merritt et al. have shown that
several of the amino acids will generate compounds
identified in Figure 49 under conditions very similar
to those used in the present study:
Phenylalanine
Tyrosine
Proline
Hydroxyproline
Pyrrole
Phenylalanine and tyrosine might also be expected to
generate benzylnitrile, benzene and phenol. Benzonitrile
may be formed as a result of a bimolecular reaction of the
benzene and nitrile radicals; the source of the nitrile
radical presumably could be urea which is excreted in the
urine in amounts up to 20 gm/day by adults. Reaction of
this radical with fragments from the alkyl amino acids, or
direct unimolecular generation from aliphatic peptides may
explain the presence of acrylonitrile.
Chromatographic conditions presently employed can be
improved to maximize the resolution and peak shapes in the
pyrograms. The reproducibility of the urine pyrogram was
first observed using a 6-foot, SE-30 column and operating
isothermally at 110°C. These conditions produced a pyrogram
which consisted of only five predominant peaks. The present
system utilizes a 12-foot, XE-60 column and is programmed


70
from 35-175°C at 10°C/min. These conditions produce a
pyrogram'such as the one seen in Figure 49. The present
system can be further improved upon by extending the pro¬
grammed temperature range; however, as the upper limit is
increased, complications may arise with PGC/MS analysis due
to contamination o'f the GC detector and mass spectrometer
from column bleed.
Utilization of the Urine Pyrogram
Present use of GC/MS for diagnosis of metabolic dis¬
orders . An increasing number of human diseases is now
known to result from defects in some biochemical process.
Clinically, these disorders often manifest themselves as
severe mental deficiencies, epilepsy, muscular dystrophy,
failure to thrive, irritability and acidosis among other
symptoms. Biochemically, these symptoms are usually related
to enzyme deficiencies leading to accumulation of certain .
toxic metabolities in the blood and tissues. Early diagnosis
and treatment of these disorders often depends on the de¬
tection of these abnormal metabolites in the body fluids.
Thus, a rapid method of analysis and characterization
of these metabolites becomes a critical factor in preventing
progression of the disease state.
A substantial amount of work has been done in this
area since the early 70's using GC/MS methods. Williams
and Halpern^ identified a number of common urinary and


71
serum amino acids by GC/MS of their neopentylidene alkyl
ester derivatives. By comparison with the amino acid
profiles from urine samples of mentally retarded children
and other selected patients, they found that they were
able to detect 15 of the known inborn errors of amino
acid metabolism. Nyhan^ was able to successfully develop
simple screening methods for a number of amino acid and
purine metabolic abnormalities using GC/MS methods to
analyze urine extracts.
The advantages of GC/MS over older methods are
numerous. The latter were often based on color reactions,
e.g., ninhydrin with amino acids, and consequently were
often dependent on the presence of a functional group like
the amino or keto group. On the other hand, volatilization
via derivatization, followed by GC/MS, allows for the
identification of over 75 different compounds” in the
normal urine sample. Not only are the amino acids
characterized but also the a-keto carboxylic acids and
the carbonyl compounds which are formed as a result of
amino acid metabolism. Even with derivatization, GC/MS is
still a much quicker method than the older schemes which
required paper chromatography or two-dimensional thin layer
chromatography. It is worth noting that about 1970 the
discovery of new inborn errors of amino acid metabolism
began to level off from the sharp incline of the previous
20 years. According to Nyhan^ this was a function of the


fact that most of those things that could have been
discovered by the limited conventional color tests had
already been discovered. However, with the advent of
GC/MS methods, the discovery of inborn errors began to
increase again, at a greater rate than noted in the 1950-
1970 period.
The drawbacks to the GC/MS method applied here arise
from the same difficulties encountered in all assay methods
of non-volatiles. Derivatization, in particular, silylation,
often results in the formation of numerous products if there
is more than one reactive group in a molecule. Multiple
derivatization is also a problem when diazomethane is used
and, of course, these difficulties will carry over and
complicate the mass spectral as well as the chromatographic
analysis.
App1ication of PGC and PGC/MS to the diagnosis of
metabolic disorders. If the pyrogram, seen in Figure 49,
accurately depicts a normal urine sample, then it is highly
possible that a pyrogram may also be representative of
urine containing abnormal levels of metabolites. In a
preliminary investigation, several urine samples were
obtained from pediatric patients suffering from a variety
of inborn metabolic defects, as well as other illnesses
caused by metabolic dysfunctions. In addition to these,
several samples from normal adults and children were also
acquired to serve as standards. The experimental conditions


7 3
employed were the same as those described earlier. Pyrolysis
of ~5 pi of pooled adult urine produced the pyrogram seen
in Figure 49. The samples used in this experiment were
subjected to PGC on three consecutive days so that the
reproducibility of the method could be established. Samples
were kept at 4°C to retard any bacterial growth. Previous
4 5
studies indicate that if urine is allowed to stand for
several hours at room temperature, microorganisms introduced
into the urine from the skin during voiding can produce
appreciable amounts of unusual metabolites. No changes
were noted in the pyrolysis patterns of these samples over
the time span of the experiment. Pyrograms obtained from
the first series of pyrolyses were essentially superimposible
on those from the last series of pyrolyses, after correction
for sample size.
Since only a few of the peaks in the urine pyrogram
have been identified, differentiation of the pyrograms
presently has to be based on differences in the overall
fingerprint of the sample. Even then, there are discernable
differences between the urine pyrograms from the diseased
patients and the standard, and between the urine pyrograms
of the diseased patients themselves. It was mentioned
earlier that many of these diseases are characterized by
the abnormal accumulation of some endogenous material due
to enzyme defects; consequently, what appears to be a "new"
peak in a diseased urine sample may also be present in a


7 4
normal urine pyrogram, but originating from the analyte
in much smaller amounts. Figure 50 represents an example
of a pyrogram of a urine sample from a patient suffering
from methylmalonic aciduria. This disease is presumably
caused by a defect of the methylmalonyl CoA mutase enzyme
and is characterized, biochemically, by an abnormal accum¬
ulation of methylmalonic acid, propionate, a-methylaceto-
acetic acid, butanone, hexanone, and glycine in the blood,
tissues and urine. In normal urines methyl malonic acid
appears in extremely small quantities. This urine sample
was quite dilute and some of the characteristic urine peaks
may be lost because of this. This fact accentuates the
intensity of the peaks seen at ~3, 12 and 16 minutes. The
~12 minute peak may be benzylnitrile and the small peak at
~5h minutes is probably pyrrole. The ~16 minute peak is
unidentified, and based on the difference in peak shape it
is probably a material different from that seen at approxi¬
mately the same retention time in the standard urine pyrogram
(Figure 49). The peak at ~3 minutes in Figure 50 has not
been observed previously in normal or abnormal urines and,
although its origin is unknown, it may nevertheless serve
as a basis for characterization of this disorder.
Many hundreds of urine samples will have to be screened
before the utility of this procedure can be determined. A
patient with a particular disease should be monitored over
an extended period of time and results compared with those
from other patients with the same disorder so that any common


features of the urine pyrogram can be established. PGC
and/or PGC/MS would be a faster method than GC/MS for
screening of large numbers of urine samples since the
need for isolation and derivatization would be eliminated.
However, the identification of new metabolites which may
occur in previously undiagnosed diseases would be a very
difficult task with PGC/MS, since the materials in the
urine sample undergo substantial chemical transformation
prior to detection. Such cases would best be dealt with
using GC/MS where the new species could be directly
identified.
Concluding Remarks
Qualitative Applications of Pyrolysis-Gas Chromatography
The ability to characterize materials based on their
pyrolysis fragmentation patterns is presently the most
desirable feature of PGC. Its fingerprinting techniques
have proven invaluble in identification and classification
of non-volatile substances. The present study has inves¬
tigated the qualitative applications of PGC in the pharma¬
ceutical and related areas. The results obtained are
promising and indicate that PGC can be utilized as a
research tool and potentially as a routine analytical tool
in these areas.
The characterization and differentiation of the


76
penicillins and cephalosporins demonstrates the utility
of PGC as an ancillary analytic method which can comple¬
ment GC and HPLC. In addition, the specificity provided
by the pyrogram measurably increases the certainty of
identification and decreases the time required for analysis.
This increased specificity is especially useful when one
is attempting to differentiate between two members of the
same group of compounds. Whereas two homologs may have
similar peak shape and retention time in GC and HPLC, the
difference between the pyrograms of the compounds will
usually be definitive. If a sample of the suspected unknown
is available, then it can be added to verify its presence.
The advantage of the pyrogram, of course, is that every
peak in the pyrogram of the compound will increase as the
quantity added increases, thereby providing a multiple
reassurance of its identity. It has been noted in the
literature that PGC can frequently differentiate between
positional isomers whose mass spectra are identical. PGC
data which has been classified by set theory^® has made it
possible to distinguish between geometrical isomers with
seemingly identical pyrograms. The fingerprinting technique
is limited in its ability to distinguish between compounds
in a mixture, except for, perhaps, simple molecules whose
pyrograms are correspondingly simple.
In Section I, pyrolysis was classified according to
the extent of degradation of the parent material. An example


of "thermal degradation" wag given where tertiary amine
oxides were pyrolyzed at 150°C and converted to the
corresponding olefins. Under these controlled conditions
it may be possible to examine mixtures of homologs. As an
example, a series of 3-substituted carboxylic acids could
be decarboxylated in the pyrolysis chamber and then
successfully resolved as the corresponding alkenes. This
type of selectivity would not be feasible with compounds
containing multiple functional groups, like the antibiotics,
where it was found necessary to extensively fragment the
molecule in order to increase volatility. The experiments
performed with urine samples demonstrate a different
situation. Pyrolysis of such a complex mixture virtually
precludes the identification of any one particular metabo¬
lite which may represent the biochemical manifestation of
a disease. Nevertheless, once the disease is characterized,
the overall urine fingerprint of subsequent patients
afflicted by that disease may provide the basis for the
most expeditious diagnosis.
Quantitative Applications of Pyrolysis-Gas Chromatography
This area of application has received very little
attention over the years. This stems partially from the
lack of available instrumentation which is capable of the
required precision and accuracy. It has been pointed out
that the apparatus used in this study offers as high a


78
degree of control over the .heating profile as is commer¬
cially available. The cracking severity measurements
carried out, show that the entirety of a sample can be
pyrolyzed; however, the heating profile should be defined
for each particular compound or group of compounds so that
the best compromise can be reached between pyrolysis
conditions and burn efficiency. The possible decarboxy¬
lation of the 3-substituted carboxylic acids, noted above,
may serve as an example. Before attempting to construct
a standard curve, two questions should be answered: 1) What
is the minimum temperature allowable which will effect
decarboxylation and volatilization? 2) What is the minimum
temperature and burn time allowable that will effect a 100%
burn efficiency at the high end of the range to be studied,
e.g., 1-20 yg? If 20 yg cannot be totally volatilized
without increasing the severity of pyrolysis conditions
to a point where side reactions complicate analysis, then
the assay of mixtures (but not single components) becomes
improbable. These optimum conditions will vary from one
type of compound to another depending upon the nature and
number of the functional groups present, and there is always
a possibility that they may not exist at all. The tertiary
amine oxides are easily volatilized and chromatographed at
150°C, whereas the present workers were seemingly unable to
affect the prohibitively large partition ratios of the
tetracyclines, even under the most vigorous pyrolysis
conditions.


Peak identification was described as a prerequisite
to quantitative analysis. This procedure at least allows
the analyst some inhight as to the origin of a species and
the mechanism of its formation. In addition, such infor¬
mation would be essential for the development of appropriate
internal standards. Although no attempt was made to
quantitate the changes in the amount of bimolecular reaction
products formed in these experiments, it was obvious from
the pyrograms that such species as bibenzyl increased in a
non-linear fashion with increased sample size. Peak iden¬
tification eliminates the possibility of erroneously basing
quantitation on this sort of bimolecular product whose
formation is concentration-dependent.
One of the major concerns of this study, stated in
Section I, was the attainment of the maximum amount of
information from as simple and economic a pyrolysis system
as possible. The authors were fortunate enough to have
access to a mass spectrometer for peak analysis, and,
considering the complexity of most pyrograms, PGC/MS is
not only the most expeditious method of peak analysis but
the most practical as well. This does not, however, neces¬
sitate that the PGC unit be permanently integrated with a
mass spectrometer. On the contrary, even when used as a
research tool, the pyrolysis unit need only be interfaced
with the GC/MS for the initial characterization of the
pyrogram. For example, PGC/MS analysis of the 12 antibiotics


80
studied, required two workers about two hours to complete.
This, of course, excludes the job of interfacing the
pyrolysis unit and the time involved in the interpretation
of the mass spectral data. Once these interpretations
are completed however, these antibiotics can be characterized
and quantitated routinely, without further need to employ
the mass spectrometer. Certainly on an intralaboratory
basis this PGC/MS data can be stored and referred to for
ensuing routine PGC analysis. As PGC methods become more
standardized, the analyst will hopefully be able to rely
on reference tables for peak analysis. Reproducibility
could then be checked by a comparison of the shape, size
and retention time of one or two peaks in the pyrogram
which may represent readily available chemical entities.
The future of PGC as a qualitative and quantitative
tool depends on the willingness of individual investigators
to collaborate in the development of a standard reproducible
methodology. This is especially critical when identification
is based on fingerprinting, as is the case with micro¬
organisms paints and resins, and the urine samples discussed
above. In these instances, all variables will have to be
rigidly defined to assure interlaboratory reproducibility;
this includes chromatographic as well as pyrolytic parameters.
The authors have attempted to maintain some consistency in
experimental conditions in order to minimize the number of
variables involved in an analysis. The rate of temperature


81
increase should be as fast as possible. This parameter
remained constant at 20°C/milliseconds (the maximum con¬
trollable rate of the instrument) throughout these studies.
Pyrolysis temperatures were necessarily variable; however,
a 1.0-second burn time was found sufficient for all
experiments carried out. Pyrolysis intervals up to 30
secs, can be found in the literature but, these are for
very large sample sizes where no possibility exists for a
uniform hea't exchange between the source and the material.
Chromatographic conditions are usually more varied as is
characteristic of GC. Nevertheless, high flow rates are
recommended to minimize the time spent in the reaction zone
by the pyrolysis fragments. This helps to avoid high
concentrations of free radicals and prevent bimolecular
and secondary reactions.
In these studies a helium flow rate of 60 ml/min. was
used in all experiments. Calculations indicated that this
rate was really not much faster than the optimum rate
required for a minimum plate height (HETP). It was felt
that helium, rather than nitrogen, was the best choice of
carrier gas since it has been reported that nitrogen pyrolysis
products have been identified in the pyrograms of aromatic
carbocyclics, presumably as a result of reaction with the
carrier gas. Finally, silicone XE-60 evolved as the liquid
phase of choice for all analyses performed. Although both
OV-225 and XE-60 have the same recommended maximum


ü ¿
temperature (275°C), it was found that there was sub¬
stantially less column bleed from the XE-60, in addition
to better overall resolving power with this stationary
phase.


APPENDIX I


Table I. Cracking Severity Measurements of Four Representative Antibiotics.
Antibiotic
Cephalexin
Oxacillin
Methicillin
Penicillin G
Severity5'^
99.4
98.4
98.1
100.0
Cracking
Relative S.D. (%)c
0.2
0.9
0.7
Sample Weight (yg)
4
2
10
5
alst burn efficiency - 1/p where P^ = peak response for the initial pyrolysis
and, P2 = peak response for the second pyrolysis.
^These figures represent an average of at least four determinations.
cRelative standard deviation = S.D./X-


85
Table II. Symbol Key and Statistics for Standard Curves
in Figure 48.
Antibiotic â–  Log-Log Slope. Regression Coefficient
| Oxacillin
0.97
0.996
Cephalexin
1.08
0.999
0 Penicillin G
0.89
0.993
/\ Methicillin
0.88
01 9 98


86
Table III. Determination of Saccharin in Two Name Brand
Diet Beverages.
Amount
Determined3
(mg/12 fluid oz.)
Beverage
Sample 1
Sample 2
Sample 3
Sample 4
Cola A
6 6.3
76.5
63.6
Cola B
38.4
36.2
41.7
40.5
aEach result is the average of at least three determinations
on the same bottle.


APPENDIX II


rp-i
'1CJU1
1. Pyroorobe 150 solids pyrolyzei


o
SACCHARIN (o-BENZOSULFIMIDE)
Figure 2. Saccharin structure.
cc


RESPONSE
90
Figure 3. Pyrogram of saccharin.


RESPONSE
j i
Figure 4.
Pyrogram of sodium saccharin.


Full Text
APPLICATIONS OF PYROLYSIS-GAS CHROMATOGRAPHY TO
PHARMACEUTICAL AND CLINICAL ANALYSIS
By
TIMOTHY ARMAND ROY
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA

DEDICATION
To the late Dr. Stephen Szinai, scientist, teacher,
and friend whose openness and encouragement made this
work possible.

ACKNOWLEDGEMENTS
I am indebted to my supervisory committee, Dr. Stephen
G. Schulman, Chairman, Dr. K.F. Finger, Dr. J.A. Zoltewicz
and Dr. B.S. Andresen, for their time and guidance in the
preparation of this manuscript.
I would also like to acknowledge Dr. Donald Chichester
for his help in editing the text and buffering my assaults on
the King's English.
Special thanks and appreciation go to Dr. Schulman for
accepting the responsibilities of committee chairman following
the loss of Dr. Szinai.
Thanks also to Carolyn Grantham for typing and to Gail
Clifford for graphics.
Ill

TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS . . . . iii
ABSTRACT iv
SECTION I - INTRODUCTION 1
Reaction Gas Chromatography 1
Nature of the Chemical Reaction in Pyrolysis-
Gas Chromatography 2
Pyrolysis Classification Based on Extent of
Degradation 4
Utilization of the Pyrolysis-Gas Chromatogram:
The Pyrogram 7
Recent Applications of Pyrolysis-Gas Chroma¬
tography 9
Polymer Analysis 9
Hydrocarbons - Pyrolysis Mechanisms and
Kinetics 12
Identification of Microorganisms and Fungi... 14
Toxicological and Pharmaceutical
Analysis 17
Purpose of Present Research 20
SECTION II - QUALITATIVE IDENTIFICATION OF FOOD AND
DRUG MATERIALS USING PYROLYSIS-GAS
CHROMATOGRAPHY 2 4
Advantages of Pyrolysis-Gas Chromatography 24
Increased Peak Identification Ability 24
Simplification or Elimination of Derivati-
zation Procedures 25
iv

TABLE OF CONTENTS (Continued)
Page
Characterization of Saccharin (o-benzo-
sulfimide ) by Pyrolysis-Gas Chromatography 26
Discussion and Results 27
Saccharin and sodium saccharin 27
Saccharin in soft drinks 28
Saccharin in a multivitamin product 29
Experimental 31
Materials 31
Apparatus 31
Procedure 32
Characterization of Penicillins and Cephalosporins
by Pyrolysis-Gas Chromatography 33
Discussion and Results 33
Benzyl penicillins 33
Isoxazolyl penicillins 35
Methicillin, nafcillin, penicillin V.... 35
Cephalosporins 36
Experimental 37
Apparatus 37
Antibiotics 38
Procedure 3 8
SECTION III - QUANTITATIVE ANALYSIS OF FOOD AND DRUGS
USING PYROLYSIS-GAS CHROMATOGRAPHY 39
Prerequisites for Quantitation 39
Cracking Severity Measurements 40
Peak Identification 42
Quantitative methods in GC and PGC 42
v

TABLE OF CONTENTS (Continued)
Page
Origin of fragmentation products 43
Comparison of classical and non-
classical thermolysis mechanisms 44
Quantitation of Penicillins and Cephalosporins.... 47
Discussion and Results 48
Cracking severity measurements 48
Peak identification - instrumental
methods 4 9
Peak identification - benzyl penicil¬
lins 49
Peak identification - isoxazolyl
penicillins 53
Peak identification - methicillin and
penicillin V 54
Peak identification - cephalosporins.... 55
Preparation of standard curves 56
Experimental 57
Apparatus 57
Antibiotics 58
Procedure 58
Quantitation of Saccharin 59
Discussion and Results 59
Cracking severity measurements 59
Peak identification - saccharin (sodium
salt) 59
Preparation of standard curves 60
Determination of sodium saccharin in
diet beverages 60
vi

TABLE OF CONTENTS (Continued)
Page
Experimental 64
Materials 64
Apparatus 65
Procedure 65
SECTION IV - ADDITIONAL APPLICATIONS OF PYROLYSIS-GAS
CHROMATOGRAPHY AND CONCLUDING REMARKS 67
The Use of Pyrolysis-Gas Chromatography for the
Diagnosis and Study of Metabolic Disorders 67
The Urine Pyrogram 67
Pyrolysis of serum and urine 67
PGC/MS analysis of urine 68
Utilization of the Urine Pyrogram 7 0
Present use of GC/MS for diagnosis of
metabolic disorders 70
Application of PGC and PGC/MS to the
diagnosis of metabolic disorders 72
Concluding Remarks 7 5
Qualitative Applications of Pyrolysis-Gas
Chromatography 7 5
Quantitative Applications of Pyrolysis-Gas
Chromatography 77
APPENDIX I - TABLES 83
APPENDIX II - FIGURES 87
REFERENCES 14 0
BIOGRAPHICAL SKETCH 143
vii

Abstract of Dissertation Presented to the
Graduate Council of the University of Florida in
Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy
APPLICATIONS OF PYROLYSIS-GAS CHROMATOGRAPHY TO
PHARMACEUTICAL AND CLINICAL ANALYSIS
By
Timothy Armand Roy
December, 1976
Chairman: Stephen G. Schulman
Major Department: Pharmaceutical Chemistry
Pyrolysis-gas chromatography (PGC) has established
itself as an effective technique for the identification of
non-volatile materials such as polymers, paints, resins and
microorganisms. PGC has the ability to provide a reproducible
succession of peaks from a single parent material which, in
effect, represents a fingerprint of that compound, similar to
that provided by an infrared spectrum. The pyrogram can con¬
siderably increase the certainty of identification and provide
a means for classification. Thus far, applications of PGC
to the pharmaceutical and related areas have not appeared in
the literature. The quantitative potential of the technique
has remained untapped even in those areas where it has been
Vlll

used most extensively over the past 15 years. The advent
of more sophisticated pyrolysis instrumentation now enables
the entire heating profile to be defined, thus assuring a
high degree of reproducibility and increasing the potential
for quantitative applications.
This study investigates the possible qualitative and
quantitative applications of PGC to pharmaceutical analysis,
in two series of experiments, with a number of penicillins
and cephalosporins and with several food and drug items
containing sodium saccharin. In addition, preliminary
investigations are discussed concerning use of the technique
for the diagnosis and study of a variety of metabolic dis¬
orders. These studies are based on abnormalities observed
in the pyrograms of urine samples.
IX

SECTION I
INTRODUCTION
Reaction Gas Chromatography
Pyrolysis-gas chromatography (PGC) is a specific
application of reaction gas chromatography coupling
gas chromatography with chemical reactions. The
combination of physical methods with chemical reactions
widens the scope of analysis and provides an increased
potential for problem solving. In chromatographic
terms, this added potential is reflected in an ability
to alter partition ratios or detector responses through
the chemical transformation of one or more components.
The subtractive method of reaction gas chromatography
and PGC represent the two extreme examples of partition
ratio alteration. In the former the ratio is made so
large that the compound is not eluted at all. In the
latter the ratio is changed such that an otherwise highly
retained material may be eluted under less rigorous
operating conditions.
1

2
Nature of the Chemical Reaction
in Pyrolysis-Gas Chromatography
Pyrolysis may be defined as decomposition at
elevated temperatures in the absence of oxygen. When
a material is heated to high enough temperatures,
thermal excitation of the individual molecules of that
material becomes sufficient to cleave certain bonds in
the molecules and generate free radical fragments.
Pyrolysis conditions can usually be varied such that
fragmentation results in the formation of comparatively
volatile components which can then be analyzed by con¬
ventional gas chromatography. The nature and the quantity
of these fragments will reflect the elemental and structural
character of the parent material as well as the conditions
of pyrolysis.
The free radicals initially formed during pyrolysis
arise from the splitting of the molecule at its weakest
links. The average carbon-carbon bond energy, or bond
dissociation energy, is about 90 Kcal/mole.^ Thermal
excitation of molecules becomes sufficient to break such
bonds at temperatures in the range of 450° to 650°C.^ The
mechanism which best accounts for the kinetics of organic
pyrolysis reactions was first proposed by Rice.^ Subsequent
work has strengthened his basic premises which are summarized
below.

J
To illustrate, consider the dissociation of a
paraffin:
1) R* + P -> R* + RH
An initially formed free radical (R*) abstracts
a hydrogen atom from a neutral paraffin molecule (P).
The structure of the newly formed free radical (R*) is
dependent on the ease of abstraction of a hydrogen from
a primary, secondary, or tertiary position on the neutral
paraffin molecule. In studies of hydrogen abstraction
from aliphatic hydrocarbons, the rates have been found
to be in the order, primary order is independent of the nature of the attacking
free radical species and is attributed to the strengths
of the C-H bonds being broken. Consequently, the weakest
of the C-H bonds, the tertiary, are broken at the fastest
rate. Looked at another way, the most reactive hydrogens,
the tertiary, are abstracted at the fastest rate.
2) Ri R'i 0L1 + R*
Although free radical rearrangements are not common,
they do occasionally occur in hydrocarbon free radicals,
following the propagation sequence, if the newly formed
radical is a C^-fragment or larger. The rearrangement can
take the form of a hydrogen shift from a primary to a
secondary site which requires an activation energy of 4
Kcal/mole. Either the R* or R'* species will subsequently
fragment a,8 to the carbon atom bearing the unpaired electron,

4
with the dissociation favoring formation of the more
stable free radical (R*) if more than one 3-bond occurs
on the same molecule. This reaction is accompanied by
the formation of the olefin (OL).
3) R* + 0Ln * RnR*
Another type of propagation reaction is now possible
wherein a free radical species can add to an olefin.
Addition occurs in such a way as to form the more stable
free radical, i.e., primary and secondary radicals would
react to form secondary and tertiary radicals, respectively.
4) a. 2R* -> RR
b. 2R* -* R + OL
These two radical reactions terminate the chain
process by a) combination of two radicals or b) hydrogen
abstraction by one radical from another to generate, in
this hydrocarbon example, an alkane and an alkene.
Pyrolysis Classification Based on Extent
of Degradation
As mentioned earlier, the volatile fragments formed
as a result of pyrolysis will reflect the elemental and
structural character of the parent material. In practice,
the pyrolysis conditions can be varied such that either
one or the other of these features will predominate. The
parameter which most affects the nature of the fragments

5
formed on pyrolysis (the pyrolysate) is the final pyrolysis
4
temperature. Beroza and Coad's review of reaction gas
chromatography grouped pyrolyses according to the extent
of degradation suffered by the parent material. According
to this classification, pyrolysis carried out in the 100°-
300°C range is designated as thermal degradation. This
is the mildest of pyrolytic procedures and often can occur
in a conventional injection port of a gas chromatograph,
requiring no special apparatus. For example, when this
5
procedure is applied to tert-amine oxides they undergo
Cope elimination:
I I
H
V
NR
+
100°-150°C
0 -
A
>
C = C + H-O-NRo
x- \ z
The subsequent analysis can then be performed on the olefin
or hydroxylamine generated, or both; the method selected
depends on chromatographic conditions.
Mild pyrolysis describes thermal reactions carried
out in the 300°-500°C range where many types of carbon-
carbon bond cleavage can occur. It has been reported that
amino acids subjected to pyrolysis at 300°C for three
minutes decarboxylate and give characteristic "amine
6
profiles." Lower aliphatic branched and straight-chain
alcohols have been dehydrated by passing the compounds
through a quartz column (at 470°-560°C) containing Chromasorb
7
P. Identification is based on the corresponding olefins
that are generated.

6
Normal pyrolysis covers the 500°-800°C range, and
it is in this zone that the majority of PGC analysis
has been done. At these temperatures most materials
are fragmented into small molecules. However, a sub¬
stantial amount of information concerning the structure
characteristics of the parent molecule can be derived
from their analysis. Polymer chemists have had a great
8
deal of success in deducing the structures of polymers,
9 10
plastics, elastomers, and related materials from
analysis of pyrolysis fragments generated at these
temperatures. Typically, compounds are assayed according
to the amounts of low-boiling material produced upon
fragmentation which are directly correlated with the type
11
and number of functional groups present in the parent.
Vigorous pyrolysis (800°-1100°C) causes extensive
fragmentation of molecules and will usually result in a
pyrolysate dominated by low-molecular-weight compounds.
The extreme temperatures used approach conditions of
elemental analysis and often much of the information
obtained under normal pyrolyses (500°-800°C) concerning
molecular structure is lost in the higher temperature
range due to the extensive dismantling of the parent
compound. Nevertheless, a large number of successful
analyses have been conducted under these vigorous condi¬
tions including the present work to be described in later
sections.

7
Utilization of the Pyrolysis-Gas
Chromatogram: The Pyrogram
The chromatogram resulting from PGC is commonly
referred to as the pyrogram and this term will be used
in the remainder of this report. It may be inferred
from the preceding discussion of free radical mechanisms
and the nature of the pyrolysate that the chief concern
of PGC is the identification of the individual peaks
which make up the pyrogram. Unquestionably, this type
of analysis provides an abundance of information which
facilitates parent structure elucidation and reaction
mechanisms. Peak identification, however, is not essential
for qualitative work within a class of compounds and this
point is demonstrated in many of the more recent applica¬
tions of PGC discussed in a later part of this manuscript
(see Section I - Recent Applications of Pyrolysis-Gas
Chromatography).
The widespread use of PGC as an analytical tool can
be accounted for primarily on the basis of the excellent
reproducibility of pyrograms that can be achieved. Within
certain operational limits (which are still not precisely
defined) the pyrolysis of a wide variety of materials from
simple hydrocarbons to polymers to microorgamisms has
resulted in fragmentation patterns that approach infrared
spectra in their consistency. Unfortunately, reproducibility
on an interlaboratory basis has been severely hampered by
the variety of different pyrolysis units constructed by

8
individual laboratories. This lack of standardization
has made it virtually impossible to compile a common
set of pyrograms analogous to the universal collections
of infrared spectra, mass spectra, etc. The situation
is improving with the emergence of more sophisticated
commercial instrumentation which offers a high degree of
control over pyrolysis conditions thus assuring more
reproducible characterization of materials.
Where conditions have been standardized, pyrograms
have been cataloged for qualitative identification. The
12
Federal Bureau of Investigation has set up a PGC
laboratory in Washington, D.C. and has obtained pyrograms
of all of the different paints that have been used by
car manufacturers for the past 15 years. This constitutes
a reference library of well over 1,000 pyrograms. Paint
chips recovered following a hit-and-run automobile accident
can be sent to this laboratory for PGC analysis and the
identity of the vehicle type confirmed by comparison of
the pyrogram with those in the F.B.I. library. This
technique is called fingerprinting and experimental evidence
would seem to justify use of the term.
Some variation of pyrolysis and column conditions can
usually be found such that the fragmented material will
yield a reproducible succession of peaks that will be
unique within its class of compound. Thus, a data bank such
as that compiled by the F.B.I. allows for the differentiation

9
of over 1,000 paints, even those from the same manu¬
facturer, strictly on the basis of the "fingerprint"
provided by PGC.
The fingerprinting technique enjoys widespread
13
popularity in the areas of fiber and textile analysis,
14 15
toxicology, and microbiology. In those situations
where analysis is limited to a particular type of non¬
volatile material or compound, fingerprinting is often
a sufficient method. Moreover, analysis which can be
carried out without the need for more sophisticated peak
analyzing apparatus offers considerable cost benefits in
instrumentation, servicing, and operation. A better
appreciation of the analytical capabilities of the various
PGC methods is essential for determining the type of
instrumentation necessary for a particular analytical
problem. A review of the other current applications of
PGC is offered to provide such insight.
Recent Applications of Pyrolysis-Gas Chromatography
Polymer Analysis
Previous to the introduction of PGC techniques,
characterization of organic polymers was done primarily
with nuclear magnetic resonance (NMR) and infrared
16 17
spectroscopy. Perry and Martinez and Guiochon have
discussed the advantages of PGC over these methods in detail

iü
However, the single greatest advantage of PGC is the
ability to fragment these complex materials into simpler
pieces which can then be separately characterized and
quantitated or collectively interpreted as the polymer
fingerprint.
17
Martinez and Guiochon were able to correctly
identify the various types of phenol-formaldehyde
polycondensates using PGC. A number of resins which
they prepared from various mixtures of pure phenol,
3-methyl phenol, and 3,5-dimethyl phenol, all provided
reproducible pyrograms easily distinguishable from one
another on the basis of qualitative or quantitative
differences. Under the PGC conditions they used, the
majority of the pyrolysis fragments consisted of a variety
of methyl-substituted phenols which were identified by
a comparison of the retention times of the pyrolyses
peaks with the retention times of solutions of known
phenols. In this way, the mole percent of the phenolic
compounds formed by pyrolysis were determined and equated
with the relative concentration of the phenols, 3-methyl
phenol and 3,5-dimethyl phenol, in the prepared resins.
Semiquantitative analysis of methacrylate and
styrene polymers in organic coatings was achieved by
18
Esposito using the internal standardization technique
with PGC. Pyrolysis conditions were adjusted so that a
maximum yield of monomer units were produced from the

11
internal standard and the polymers to be assayed. The
polyethylmethacrylate internal standard yielded 98%
ethylmethacrylate on pyrolysis. The polymers assayed
also produced essentially one peak on heating, greatly
reducing the possibility of interference from plasticizers
and modifying resins in the coatings.
19
Iglauer and Bentley developed an elaborate analytic
system for organic polymer identification which consisted
Q
of a pyrolysis unit coupled to an 1100 Pyrochrom Pyrolyzer'w'
manufactured by Chemical Data Systems. This unit was
modified by those investigators in order to allow separation
of the pyrolysate into a dual flow system equipped with
thermal conductivity and flame ionization detectors. The
smaller fragments arising from pyrolysis are diverted to
a self-contained gas chromatograph within the Pyrochrom
unit. Here the compounds pass through the thermal con¬
ductivity detector and their identity and quantity are
determined by comparison with a reference model for the
internal gas chromatograph which indicates the positions
of all major small molecules, e.g., CO, CH4, CO2, C2H4,
C2Hg, C^Hg, H20, S02, HCN, CHICHO and C^Hg. Larger
fragments are diverted to a second gas chromatograph
equipped with the flame ionization detector. The column
oven is operated with linear temperature programming which
helps to resolve the higher boiling components in a
reasonable time period. The identity of these peaks is
also based on comparison with a reference model.

12
This apparatus generates a substantial amount of
information concerning the nature of the polymer. Both
of the gas chromatographs produce pyrograms which can
be used as low-and high-molecular-weight fingerprints
of the material. The identification of the individual
peaks in the low-molecular-weight pyrogram provides
information concerning the functionality of the polymer,
whereas similar analysis of the high-molecular-weight
fragments can yield useful information as to the nature
of monomeric units and molecular components of the
parent material.
Hydrocarbons - Pyrolysis Mechanisms and Kinetics
Much of the work done with hydrocarbons has been
concerned with the mechanisms of pyrolysis and the effect
of varying pyrolysis conditions on thermolysis patterns
20
and product distribution. Fanter et al. studied the
pyrograms of 83 Cg-Cpg compounds pyrolyzed over a 50°C
temperature range (575°—625°C). They found that the PGC
patterns changed very little over this range and that all
of the 83 hydrocarbons could be differentiated on the bases
of their fragmentation patterns with the exception of cis-
trans isomers. Prior to the pyrolysis experiment, the
retention times of all the C-^-C^g normal alkanes were
determined, as well as those for a number of alkenes and
alkynes. In this way, the authors were able to ascertain

1J
the relative amounts of these materials formed upon
fragmentation of the parent molecule. Their data for
n-hexane, in particular, was in good agreement with
Rice's proposed free radical mechanism discussed earlier
(See Section I - Nature of the Chemical Reaction in
Pyrolysis-Gas Chromatography).
21
Levy and Paul used hexadecane as a test compound
to study changes in pyrolysis fragmentation products as
a function of pyrolysis conditions and sample size. In
order to better quantify their results, the authors
normalized the distribution of pyrolysis fragmentation
products so that the total area of all the peaks in the
pyrogram (excluding residual unpyrolyzed parent compound)
n
would equal 100% = 100 E (X /EX ) , where X is the area
i=l i i i
of any peak in a pyrogram of n peaks. Fragmentation data
for hexadecane covering a pyrolysis range of 580°-650°C
was collected and analyzed. values for peaks corres¬
ponding to C^-Cg olefins were plotted against the percent
relative cracking ratio which is essentially the fraction
of original compound which has been decomposed and is a
reflection of increasing pyrolysis temperatures. As
expected, the trend was toward an increasing percentage
of smaller fragments and a decreasing percentage of larger
fragments as the percent relative cracking ratio (cracking
severity) increased. For this particular compound, the
pattern seemed to center around 1-pentene which maintained

14
a slope very close to zero over the temperature range
studied. The change in the fragmentation pattern as a
function of sample size was insignificant even with a 100-
fold increase in the amount of hexadecane pyrolyzed.
22
In a subsequent paper, Groenendyk et al. were able
to determine the thermolytic- dissociation rates of both
hexadecane and a fatty acid methyl ester by varying
reaction temperature and flow rate through the pyrolysis
chamber. The first order pyrolysis rate was then
calculated from In C/Co = Ktr where tr is the temperature
corrected residence time of the reactant in the pyrolysis
chamber, K is the rate constant, C is the remaining
concentration of parent compound after reaction and Co,
the original concentration of parent compound. These
rates were then fitted to the Arrhenius equation by linear
regression techniques to determine the activation energy
and the pre-exponential term.
Identification of Microorganisms and Fungi
23
In 1965, Reiner introduced PGC as a technique to
detect, characterize and classify bacteria by visual
examination of their pyrograms. Since that time, PGC has
been used for the differentiation of a wide variety of
microorganisms and fungi.
24
Sekhon and Carmichael utilized PGC techniques in
lieu of the traditional classification methods of fungi

(i.e., those based on gross and microscopic morphology)
to characterize a number of dermatophytes belonging
to the genera Nannizzia, Arthroderma and Microsporum.
Pyrograms of replicate samples prepared from the same
colony or from two separate colonies of the same strain
appeared nearly identical. However, the authors did
find that sample size, colony age, and especially the
nature of the culture medium all exerted noticeable
effects on the quantitative and qualitative appearances
of the pyrograms.
2 5
Meuzelaar and in't Veld, using a modified PGC
system with a high-frequency induction heating filament
pyrolyzer (Curie point pyrolyzer), were able to attain a
greater degree of qualitative and quantitative repro¬
ducibility with bacterial test samples. In order to
assure good reproducibility of sample sizes, the authors
developed a technique whereby a 5-15 V1ICS2 suspension of
the freeze-dried bacteria (Neisseria meningitidis and
Neisseria sicca) was applied to the filament as a thin and
uniform coating. This method allows for the most efficient
and reproducible heat transfer to the sample, which,
according to Farre"-Ruis and Guiochon,^ is one of the
controlling factors in the decomposition of polymers and
probably of many other organic materials. The use of the
Curie point pyrolyzer offers additional control over another
factor in decomposition which is considered crucial by

16
Farrea-Ruis and Guiochon, the rate of heating. The ferro¬
magnetic filaments are well known to be highly reproducible
in attaining the Curie temperature and the high-freguency
induction heating of these filaments assures a uniform
heating rate (milliseconds). Use of these innovative
methods allowed Meuzelaar and in't Veld a level of
qualitative and relatively quantitative reproducibility
such that the pyrograms obtained in the series from the
same samples were nearly superimposable.
Continuing work in this area has led to increasingly
sophisticated instrumentation and technique. Using 500-ft
capillary columns and disposable pyrolysis chambers,
27
Quinn was able to resolve approximately 200 peaks from
the pyrolysates of representative bacteria, yeasts, fungi,
and mycoplasma. Sensitivity of the system was such that
peaks equivalent to 1 nanogram were detectable. Other
28
researchers have been able to show differentiation of
the fungi belonging to the Aspergillus flavus group at
the species and strain level using apparatus such as
Quinn's. Presently, materials being examined include
Salmonella, cell walls of bacteria, whole cells,
2 9
Leptospira, and Clostridia, among others.
In many respects, PGC compares favorably with more
conventional "wet" biological and serological methods
for classification and identification of biological
materials, especially because it is simple, rapid, and

well suited to automation and electronic data processing.
However, lack of interlaboratory uniformity of pyrolysis
systems again precludes a wide-spread use of PGC in
these biological areas as it does in those areas
discussed previously. There is evidence, however, that
the ribbon-type probe/pyrolysis unit used in the present
study is becoming standard apparatus in the field and
should lead to uniform PGC instrumentation in the coming
years.
Toxicological and Pharmaceutical Analysis
The combination of gas chromatography and pyrolysis
as an analytical tool in drug and poison identification
was first popularized by Dr. Paul Kirk of the School of
Criminology at the University of California, Berkeley
30
campus. As early as 1962, Kirk and Nelson were able
to characterize 27 barbiturates solely on the basis of
the "fingerprints." A later experiment carried out with
these same compounds attempted to identify the pyrolysis
fragments. Although many similar fragments were found in
the series of 27 drugs, a number of unique nitrile
derivatives was identified which represented the pre¬
dominant peaks in the pyrograms and, in fact, were found
to be the very peaks which had allowed Kirk and Nelson
previously to differentiate the compounds. The nitrile
derivatives contained the methylene carbon atom with the

18
double substitution characteristic of the individual
barbiturates, e.g., the major peak in the pyrogram of
probarbital (R^ = ethyl, R£ = isopropyl) was found to
be 3-methyl-2-ethyl butanenitrile:
Me
I
H_C-C-C-C=N
I
Et
0
Kirk and his associate were able to refine chroma¬
tographic and pyrolytic conditions so that the various
barbiturates could be identified (within the group) by
the one characteris.tic nitrile peak in the pyrogram.
These examples of imides breaking down to nitriles must
be considered an energetically favorable thermolysis
process as this same phenomenon has been observed
during the present PGC studies of saccharin (o-sulfo-
benzoic acid imide), the penicillins and cephalosporins.
30
In addxtion to the barbiturates,- Kirk and Nelson also
studied the phenothiazines and the morphine alkaloids.
Neither of these classes of compounds produced a
characteristic peak such as was seen in the barbiturates;
however, their identification could be based on the
variations in the amounts of low-molecular-weight pyrolysis
products generated. The phenothiazines were characterized
principally by the relative amounts of methane, ethene and
propene produced, whereas the morphine alkaloids were
characterized by the amounts of methane and ethene.

Computer analysis of the relative peak heights of these
prominent peaks permitted the identification and
differentiation of the individual members of these
drug groups.
A final example of a rather sophisticated and
expensive analytical technique involving pyrolysis in
31
combination with GC/MS is described by Merritt et al.
which combines PGC/MS with a computer data system. In
this experiment, the phenylthiohydantoin (PTH) derivatives
formed during Edman degradation of proteins are charac¬
terized by PGC which provides unique fingerprints of each
PTH derivative. These pyrograms are then analyzed peak by
peak with the mass spectrometer. This data is fed to the
computer where the pyrogram is transformed from a histo-
graphic to a Cartesian form, i.e., the abscissa is divided
into a number of positions corresponding to the retention
times for the various common constituents of the pyrograms
and the ordinate is divided into arbitrary units for
assignment of intensity levels. On this basis, a
diagnostic code for the individual PTH derivatives is
developed. For example, proline PTH was represented in
this manner as CIF2GIJ2L5. Such information can then be
stored and ultimately recalled to yield the protein amino
acid sequence.

Purpose of Present Research
The present work was concerned with developing new
analytical methods for food and drug analysis using
pyrolysis techniques. PGC complements conventional gas
chromatography in that it lends itself to the analysis
of non-volatile materials. As one traverses a series of
materials arranged in order of decreasing volatility, a
point is reached at which assays, using conventional gas
chromatography techniques (involving derivatization and/or
abnormal chromatographic parameters), become too time
consuming and often fail to maintain minimum limits of
precision and accuracy. It is at this point that the
analyst must decide on ancillary methods. In this labora¬
tory the alternatives investigated have been high-pressure
liquid chromatography and pyrolysis-gas chromatography.
Studies of the latter method are described in the following
sections.
The major concerns of these studies were (1) standardi¬
zation of PGC to foster a uniform pyrolysis system and aid
in the interlaboratory reproducibility of experiments; (2)
extraction of the maximum amount of information from as
simple and economical a pyrolysis system as possible; (3)
development of a quantitative method with a degree of
precision and accuracy comparable to conventional gas
chromatography.

21
On the first point, one can only attempt to abide
by the trends within the field. An exhaustive literature
review of this area was carried out before deciding on
the filament-type platinum ribbon probe and versatile
pyrolyzer unit manufactured by Chemical Data Systems (Figure
1). This instrument offers a high degree of control over
the various pyrolysis parameters discussed previously
and makes possible that degree of reproducibility essential
for the standardization of the technique.
With regard to the second point, expense and
versatility, the pyrolysis unit used in the present
study can be purchased for approximately $1300 (1976), and
is readily adapted to integrate with a number of commercially
available gas chromatographs. Assuming most analytical
laboratories are already in the possession of a gas chroma¬
tograph, this outlay can be considered nominal when viewed
against the total cost of outfitting such a laboratory for
general instrumental analysis.
Needless to say, pyrolysis systems can be highly
sophisticated and so correspondingly costly, for example,
systems such as those involving integrated PGC/MS/computer
analysis. A wealth of information can be provided by such
systems which include a peak analyzing component, yet
pyrograms alone are usually quite sufficient for qualitative
work. Individual peak analysis does, however, become
essential where the concern is for the mechanism of

22
thermolysis or the determination of mole percent of
monomers as in the polymer studies described. It will
also be pointed out later that peak analysis may well
be a prerequisite for absolute quantitation (in contrast
to relative quantitation, discussed in the polymer
studies) as an assurance of good linearity.
Finally, it has been noted that the pyrolysis
techniques reviewed in the literature, to date, have
been limited to qualitative and relative quantitative
analyses. Despite the increased refinements in pyrolysis
systems, no studies have been reported on the absolute
quantitative analysis of materials using pyrolysis
techniques. As used here the term, absolute quantitation,
means reproducibly pyrolyzing the entire amount of sample
applied to the heating element. In a majority of the
studies reviewed, no attention was paid to the percentage
of the parent material that is actually pyrolyzed. Those
studies which consider this parameter (cracking severity-
percent relative cracking ratio) report figurés of 20% or
below, a range where the formation of bimolecular reaction
products higher in molecular weight than the starting
material is supposedly minimized.
The results obtained in the present investigation
suggest that in pharmaceutical applications of pyrolysis
techniques to qualitative and quantitative analysis, there
is need for concern over cracking severity and secondary

23
reaction products only insofar as they interfere with the
reproducibility that is essential for such analysis.
These topics are discussed more fully in the
following sections on the characterization of a
number of food and drug materials using PGC methods.

SECTION II
QUALITATIVE IDENTIFICATION OF FOOD AND DRUG
MATERIALS USING PYROLYSIS-GAS CHROMATOGRAPHY
Advantages of Pyrolysis-Gas Chromatography
In Section I it was mentioned that the major use of
PGC up to the present has been for qualitative analysis
of non-volatile materials. The special advantages of
this technique for chemical characterizations will be
reviewed before discussing its specific applications in
the present study.
Increased Peak Identification Ability
In practice, gas chromatography is not frequently
used as a technique for peak identification. However,
there are a number of methods which improve peak identity
one of which consists of adding a quantity of the material
sought to the sample containing the unknown and then to
chromatograph this mixture. Should the compound sought
be present in the unknown, one of the peaks in the chroma¬
togram will show a relative increase in intensity as
compared with the peak produced by the original sample.

25
This is evidence that the known substance added has the
same retention time as the material corresponding to the
specific peak. A difficulty with this addition procedure
is the possibility that the test material may contain
some other substance having the same retention time as
the added known compound under the set of chromatographic
conditions employed. As discussed earlier, in Section I,
studies of the application of PGC have demonstrated the
ability of this process to yield a reproducible succession
of peaks from a single compound. The pyrogram, in effect,
provides a fingerprint of the substance, and substantially
increases the certainty of identification.
With PGC, in contrast to conventional gas chroma¬
tography, each one of a series of peaks in the pyrogram
which relates to the compound sought is increased when a
quantity of the suspected substance is added. Thus the
need to repeat chromatographic analysis with addition of
known substances is greatly diminished or eliminated.
Simplification or Elimination of Derivatization Procedures
Conventional gas chromatographic analysis of non¬
volatile materials usually necessitates prior derivatization.
This step often requires isolation of the compound sought
from an impure sample and this introduces two potential
sources of error. These are (a) inefficient isolation and
(b) irreproducible derivatization and/or formation of
multiple products. As opposed to the necessity for volatile

26
samples in conventional gas chromatography, those for
PGC must be as non-volatile as possible, otherwise they
may be lost from the pyrolysis probe before thermal
decomposition takes place. In most cases, salt formation
is the only step needed to keep the sample on the pyrolysis
probe should the parent material prove too volatile under
the chromatographic conditions employed.
Characterization of Saccharin (o-benzosulfimide)
by Pyrolysis-Gas Chromatography
Interest in the safety of saccharin (Figure 2) and
sodium saccharin for human consumption has stimulated
a number of investigations into methods of assay for
these sweetening agents. Conacher and 0'BrienJ reported
a gas chromatographic (GC) method for the determination
of saccharin in soft drinks, using diazomethane as a
methylating agent. According to their report, the
methylation of saccharin consistently gave two peaks in
a 17:3 ratio which they postulated to be the N-methyl deriva¬
tive of saccharin and the O-methyl derivative of pseudo¬
saccharin, respectively. Before derivatizing the saccharin,
which was present in the soft drinks as the sodium salt,
a number of acid-base extractions was required in order to
effect isolation. Ratchik and Viswanathan^ later reported
the determination of saccharin in a number of pharmaceutical
products utilizing silylation with N,O-bis(trimethyIsilyl)

27
acetamide.. This procedure offered some improvement
upon Conacher and O'Brien's derivatization method which
utilized diazomethane, a potentially explosive and rather
difficult reagent to prepare and use. Here, also, lengthy
extraction procedures were necessary prior to the deriva¬
tization step which, although comparatively safe and
nearly quantitative, still required over an hour's time
for sample preparation.
In view of the cumbersomeness of procedures thus
far reported, it was decided to investigate the application
of pyrolysis methods to the determination of saccharin,
in the hope of simplifying analytical work.
Discussion and Results
Saccharin and sodium saccharin. Before attempting
assay of these sweetners in food or drug materials, it
was necessary to develop an appropriate set of PGC
operation conditions for the characterization of saccharin
and its sodium salt. The pyrograms of saccharin (Figure 3)
and of sodium saccharin (Figure 4), under the conditions
used, were nearly identical. Several coinciding peaks
in the two pyrograms are seen which differ in their
relative intensities within each pyrogram. In addition,
the relative overall intensities of the two pyrograms,
derived from nearly equal quantities of material, differ
considerably. The comparatively lower intensity of the
saccharin pyrogram may have been due, at least partially,

28
to evaporation of the free imide (mp 229-230°C); pyrolysis
inlet-chamber temperature 130°C) following insertion of
the probe and prior to pyrolysis.
The pyrograms were highly reproducible, providing a
succession of peaks for which retention times and relative
sizes formed an easily recognizable pattern. During studies
on the effects of the final temperature of pyrolysis it was
found that the patterns in Figures 3 and 4 were reproducible
at temperatures as low as 700°C. However, it was further
observed that at increased temperatures (up to 900°C) the
salt and the free imide gave pyrograms of greater intensity
with no resulting change in the pattern of the pyrogram
(based on peak heights and retention times of four pre¬
dominant peaks). Pyrolysis at temperatures from 900° to
1000°C caused no discernible change in the pattern or
intensity of either pyrogram. No attempt was made at this
stage to quantitate the sweeteners; however, considerable
effort was later expended on that objective. This is
discussed in Section III.
Saccharin in soft drinks. Once a set of standard
conditions had been developed for PGC characterization of
saccharin and its sodium salt, identification of the sweetener
in diet soft drinks was investigated. The essential features
of the sodium saccharin pyrogram are clearly apparent in the
pyrogram of a typical artificially sweetened beverage (Figure
5). In experiments on the pure imide salt it was found that
amounts as small as 1.0 yg produced recognizable pyrograms.

29
With smaller quantities, a majority of the distinguishing
peaks in the pyrogram was lost. When a known portion of
pure sodium saccharin was added to a sample of beverage
of similar size, the resulting pyrogram (Figure 6) showed
an increase in intensity of those peaks which characterized
the pyrogram of the pure sodium saccharin standard (Figure
5). There could be no reservations concerning the identity
of the material since characterization was based on the
presence of a multitude of peaks representing a finger¬
print of the compound, thereby eliminating the need for
further qualitative analysis. In addition, the procedure
required no isolation or derivatization of free saccharin.'
Since it was present as its sodium salt, a form most
suited for PGC analysis, the saccharin could be directly
characterized by placing a few microliters of the soft
drink on the platinum ribbon of the probe without additional
preparation.
Saccharin in a multivitamin product. A number of
multivitamin products for children which are on the
market contain sodium saccharin as a sweetening agent.
The relatively complex composition of the multivitamin
products provided a rigorous test of PGC. Each tablet
of the selected product contained substantial quantities
of fat-soluble vitamins (vitamin A, 1.7 mg, vitamin D, 0.01
mg); and water-soluble vitamins (three B-vitamins, 19 mg.
total, vitamin C, 75 mg); as well as large quantities of
excipients (400 mg). These were in addition to the sodium

JO
saccharin (1.35 mg by weight, equivalent to 1.20 mg
of saccharin). The extraction (described below in the
Experimental Section) removed the greater part of these
fat-soluble and water-soluble materials. Interference
with the analysis of the free imide from residual
ether-soluble materials could be minimized by pre-firing
the sample in a sealed, nitrogen-swept apparatus to
reduce the possibility of secondary reactions. Repeated
firings at temperatures of 100-110°C were carried out until
no vapors could be detected rising from the ribbon surface
during heating. The sample could then be analyzed under
the usual pyrolysis conditions. The resulting pyrogram
(Figure 7) clearly showed the presence of a number of the
characteristic peaks of the pure saccharin pyrogram. Assuming
complete extraction of saccharin from the tablet, this 5.0
y 1 sample contained a maximum of 6.0 yg of the imide in
addition to residual impurities. This quantity was at least
twice as much as that required to produce a recognizable
pyrogram of the imide; that is, one which displays, in a re¬
producible fashion, a majority of the distinguishing peaks
observed in the pure saccharin pyrogram. Again, addition
of a known quantity of pure saccharin (Figure 8) reaffirmed
the identity of the extract, distinguishing between those
peaks due to residual impurities and those due to saccharin.
As with the soft drink analysis, no attempt was made, at
this stage, to carry out quantitation; therefore the free
imide was not converted into the less volatile and more
easily quantitated sodium salt.

31
Experimental
Materials. Solvents - All solvents were reagent grade
and were not further purified.
Saccharin and sodium saccharin - The sodium
saccharin was reagent grade (Penick). The saccharin,
34
furnished by another source, was prepared by dissolving
sodium saccharin in water and adding excess 6N HC1. The
precipitate was extracted with ethyl acetate and crystal¬
lized by evaporating the solvent (m.p. 229-230°C, un¬
corrected; lit. 228°C).
Soft drinks - These were purchased commercially:
Diet-Rite Cola, Diet Rite Orange, Diet Rite Ginger Ale,
Diet-Rite Grape, Tab, Fresca, and Canada Dry Ginger Ale.
Multivitamin produce - Purchased commercially:
Elusivol Multivitamin Chewable Tablets, Ayerst Laboratories,
Inc.
Apparatus. Gas Chromatograph - Carle Model 311 with
modified inlet for pyroprobe interface which was mounted
externally. The chromatograph was equipped with a flame
ionization detector (FID) and a thermal conductivity detector
(TCD) and dual 6' x 1/8" stainless steel columns, packed with
8% OV-101 on 100/120 mesh Gas Chrom Q support. Column
temperatures were variable from 125-150°C; carrier gas,
helium; ambient flow rate 50 ml/min.

Pyrolysis Unit - Chemical Data Systems Pyroprobe
150 equipped with platinum ribbon probe. Operating
conditions: final pyrolysis temperature 900°C; rate of
temperature rise 20°C/Millisecond; pyrolysis interval 1.0
second.
Procedure. (a) Soft Drinks - 1 or 2 ml of the beverage
were placed in a small test tube and gently shaken to drive
off most of the dissolved CC>2• Depending upon the
indicated concentration of sodium saccharin in the beverage
(generally in the range of 300-400 yg/ml), a 5-10 yl quantity
was directly applied to the ribbon probe. The sample was
evaporated by setting the final pyrolysis temperature at
100°C and firing several times prior to inserting the probe
in the pyrolysis interface.
(b) Multivitamin Product - One tablet (average weight
500 mg) was placed in a 120-mm test tube and pulverized
with a stirring rod. A small portion (5 ml) of diethyl
ether was added and the mixture could be either centrifuged
or simply allowed to settle for a minute, after which most
of the ether was removed with a pipet. This procedure
was repeated five times, or until evaporation of the ether
portion revealed no detectable residues in the flask. The
remaining solid in the test tube containing the sodium
saccharin was mixed with 2 ml of 6N IIC1 to convert the
saccharin to the imide. This mixture was extracted with

33
three 5-ml portions of ether which were then passed through
anhydrous Na2S0^ and collécted. The solvent was evaporated
under vacuum and the residue dissolved in 1.0 ml of
absolute ethanol. A 5-y1 aliquot of this solution was
usually sufficient for direct analysis by PGC.
Characterization of Penicillins and
Cephalosporins by Pyrolysis-Gas Chromatography
These antibiotics are B-lactams. They are amino acid
derivatives in which the amino acids are simultaneously
substituted bases and substituted acids. This confers
on them the property of being amphoteric. Because of
their pronounced polar nature, chromatographic determinations
of these compounds has been limited to paper chromatography
and more recently thin-layer chromatography.
Qualitative analysis of a number of penicillins (Figure
9) and cephalosporins (Figure 10) was investigated, using
PGC, in the hope of developing a detection technique which
would complement these separation methods.
Discussion and Results
Benzyl penicillins. The pyrograms of the penicillins
studied are shown in Figures 11-20. In this series, the
basic penicillanic acid nucleus remains unchanged. Dif¬
ferentiation among this group was thus dependent on the
fragmentation patterns of the different side chains and

34
their effect on the overall fragmentation pattern of
the molecule.
All the penicillins studied provided unique
pyrograms under the given conditions (see Experimental
section, below), with the exception of carbenicillin
(Figure 11) and penicillin G (Figures 12 and 13) whose
pyrograms could not be distinguished from one another.
Structurally, these two compounds differ only at the
benzyl carbon of the side chain (R-^) where carbenicillin
has a carboxyl group in place of one of the two hydrogens
in penicillin G. Groenendyl et. al. explored the use
of pyrolysis for the identification of functional groups
in parent molecules. Under the conditions employed by
those workers (pyrolysis temperature 600°C; pyrolysis
interval 4 seconds) it was found that carboxylic acids
and esters could be identified by the high yield of CO2
and H2O. The decarboxylation of carbenicillin seems a
likely route of decomposition, especially because it is
a (3-keto acid which should readily lose CO2 at elevated
temperatures. The CO2 given off was virtually unretained
and remained undetected amidst the other low-boiling
fragments that make up the first 30 seconds or so of the
pyrogram. This pyrolytic pathway would result in essentially
identical pyrograms for carbenicillin and penicillin G.
The pyrogram of ampicillin (Figure 14) was easily
distinguished from those of carbenicillin and penicillin G,
despite the fact that the only structural difference is,

35
again, at the benzyl carbon of the side chain where
ampicillin has a primary amine function. No analogous
mechanism exists here for elimination of the substituent
as in the case of carbenicillin. A different degradation
pathway is followed, as is evident from the pyrogram.
A close study of the pyrograms of the several benzyl
penicillins showed that the differences really were
quantitative, i.e., the retention times of the major peaks
were the same, but they differed in relative intensity.
The predominant peak in the penicillin G and carbenicillin
pyrograms was at ~6-1/2 min., whereas the predominant peak
for ampicillin occurred at '-1-3/4 min. All three pyrograms
had common peaks at these and other retention times; only
the relative intensities differed.
Isoxazolyl penicillins. These three penicillins were
readily differentiated from the other antibiotics and from
each other. Oxacillin, (Figure 15) the parent compound,
had a unique peak at -8-2/3 min., and lacked completely
the intense, more highly retained peaks that characterized
the mono- and dichloro- derivatives. Cloxacillin (Figure
16) and dicloxacillin (Figure 17) both had a rather intense
peak at -5-3/4 min.; dicloxacillin displayed an additional
large peak at almost 16 min. Interestingly, all three of
these penicillins exhibited an intense peak at -1-3/4 min.,
as was seen in the benzyl penicillins and in the benzyl
cephalosporins.
Methicillin, nafcillin, penicillin V. These three
compounds are substantially different in structure from each

36
other and from the rest of the penicillins .investigated.
This difference is seen in their pyrograms (Figures 18-
20) where there were no major peaks at those retention
times that characterized the benzyl penicillins.
Unquestionably, fragments derived from the side
chains of these and the other antibiotics represent the
major peaks in the pyrograms obtained in this study.
This could explain the noticeable lack of peaks in the
nafcillin pyrogram (Figure 19) especially when the physical
characteristics of the most likely fragments formed upon
pyrolysis are considered. In this case, the pyrolysis
temperature used (875°C) would be expected to generate
either (3-ethoxynaphthalene (bp 280°C) or B-naphthol (bp
295°C) in large quantities. Either one of these would be
indefinitely retained on the column (oven temperature 100°C)
and so would be absent from the pyrogram.
Cephalosporins. The pyrograms of the cephalosporins
studied are shown in Figures 21-24. Two of these anti¬
biotics gave virtually identical pyrograms. The peak seen
at ~3 min. in the pyrogram of cephalexin (Figure 21) and
cephaloglycin (Figure 22) consistently appeared at a slightly
greater intensity in the pyrogram of the latter. However,
because the peak was of very low intensity, it could not be
used as the criterion for differentiation of the two com¬
pounds. These two benzyl cephalosporins both displayed a
very intense peak at ~1-3/4 min. which was previously noted

37
for the benzyl and isoxazolyl penicillins. The common
structural unit to be found in all of these compounds is
the side chain, - C^, and where there is a nitrogen
function attached to C^, the ~1-3/4 min. peak is the most
intense peak in the pyrogram. In addition, the pyrograms
of these two cephalosporins also contained peaks at -6-1/2
and -14 min., similar to benzyl penicillins.
The remaining two cephalosporins are structurally
unique, as were their pyrograms. The simplicity of the
cefazolin pyrogram (Figure 23) probably relates to a facile
fragmentation of the R^-tetrazol derivative and the R^~
thiadazol ring, into low-boiling fragments.
Experimental
Apparatus. Gas Chromatograph - Varian Model 2740 with
modified inlet for pyroprobe interface which was mounted
externally. The chromatograph was equipped with an F.I.D.
detector. Column temperature was 100°C; carrier gas, helium
at 60 ml/min. ambient flow rate; chromatographic column: 6' x
1/8" i.d. stainless steel column packed with 3% XE-60 on 80-100
mesh Gas Chrom Q solid support.
Pyrolysis Unit - Chemical Data Systems Pyroprobe 150
equipped with platinum ribbon probe. Operating conditions:
final pyrolysis temperature 875°C; rate of temperature rise
20°C/millisecond; pyrolysis interval 1.0 second.

38
Antibiotics. All of the drugs studied were obtained
from the pharmacy at the University Teaching Hospital. Most
of these were in the injectable form, as their sodium salts.
Those in capsule form were present as the free acid and
were converted to their sodium salts by dissolution in
an equimolar aqueous solution of sodium hydroxide. The
possibility of contamination from sodium citrate, citric
acid, or dihydrogen sodium phosphate in the buffered
preparations was examined and found to be negligible. As
an example, a 4.67% by weight additive of sodium citrate
in one buffered penicillin G preparation had to be increased
60-fold before its contribution to the pyrogram was de¬
tectable .
Procedure. Qualitative analysis of the antibiotics
was carried out with amounts ranging from 10 yg to 40 yg of
the sodium salts depending on the complexity and intensity
of the pyrogram.

SECTION III
QUANTITATIVE ANALYSIS OF FOOD AND DRUGS
USING PYROLYSIS-GAS CHROMATOGRAPHY
Prerequisites for Quantitation
It was mentioned previously that the quantitative
applications of PGC have been limited to the areas of
polymer and hydrocarbon analysis. For the former, relative
quantitation of monomer units has been achieved with PGC,
aiding in the classification of the parent polymer. For
the latter, the relative percent distribution of pyrolysis
fragments has been determined for a number of hydrocarbons.
This data has greatly facilitated the deduction and pre¬
diction of pyrolysis mechanisms.
Both these areas of application represent examples of
relative quantitation, i.e., a measure of the relative amounts
of fragments, normalized and compared to one another. In
neither instance is there any attempt made to volatilize
the entirety of the sample, and only in the case of the
hydrocarbon analysis is the amount of parent material placed
on the heating surface or the amount of parent material
actually pyrolyzed taken into consideration.
39

40
One reason why absolute quantitation (an absolute
measure of the entire sample placed on the heating element)
has not been more closely examined has been the lack of
suitable instrumentation to provide the type of reproduci¬
bility essential for such quantitation. In addition to
this, one observes that pyrolysis of single components often
results in the formation of a multitude of peaks which would
seemingly complicate any attempts at quantitation. Also,
further difficulties may arise where cracking ratios or
cracking severities (see Section I: Hydrocarbons-Pyrolysis
Mechanisms and Kinetics) are in excess of 20%. This is an
area where formation of bimolecular reaction products higher
in molecular weight than the starting material have been
. , 7,35,36
reported.
The pyrolysis unit used by this group offered a high
degree of control over the various pyrolysis parameters.
This capacity for reproducibility was reflected in the
reproducibility of the pyrograms obtained in the qualitative
work described. It was felt that the previous performance
of the system justified an investigation of quantitative
applications.
Cracking Severity Measurements
Regardless of the complications reported to arise when
the cracking severity exceeded 20% of the parent material,
it was obvious that absolute quantitation necessitated a

41
burn or pyrolysis efficiency approaching 100%. A simple
test was devised to determine if and under what conditions
the pyrolysis unit employed could meet this requirement.
The material to be tested was applied to the ribbon
surface, usually as an aqueous solution. This method of
application assured a relatively uniform distribution of
sample over the "hot" middle 20 mm of the 2.0 X 40 mm
ribbon. Once applied, the water could be evaporated by
one of three methods, 1) pulsing the probe externally at
100°C, 2) pulsing the probe internally at 100°C or, 3)
drying externally with the aid of a gentle stream of warm
air. Chromatographic conditions employed for a particular
analysis as well as the nature of the material under study,
determined which method was used. Prior to pyrolysis,
attenuation was adjusted to give the maximum recorder
response (based on the most intense peak in the pyrogram)
for the sample size and pyrolysis temperature being tested.
Following an initial pyrolysis and generation of the pyrogram,
the sensitivity was increased ten-fold and a second pyrolysis
was performed. After dividing the peak response in the second
pyrolysis by 10, the cracking severity or burn efficiency
could be determined using the formula, P^/(P-^+P2), where P^ =
peak response for the initial pyrolysis and P2 = peak response
for the second pyrolysis. It was generally found that the
cracking severity was more a function of the final pyrolysis
temperature than the duration of pyrolysis or sample size (at

42
least up to 100 yg). Near 100% pyrolysis efficiency was
usually achieved with pyrolysis intervals of from 1 to 5
seconds at temperatures above 750°C. More often than
not, the appearance of the pyrograms was unaltered through
the range 750-1000°C and the tendency was to use final
pyrolysis temperatures in excess of 800°C for quantitative
studies. These efficiency tests were carried out only with
2-10 yg sample sizes since it was felt that any deviation
which might arise with larger sample sizes (due to in¬
complete pyrolysis) would be reflected in a deviation from
linearity in the preparation of standard curves.
Peak Identification
Quantitative methods in GC and PGC. One of the diffi¬
culties which was mentioned, that could hamper absolute
quantitation in PGC, was the multiplicity of peaks which
normally result from fragmentation of a single component.
This multiplicity would only be accentuated under the
conditions necessary to achieve near 100% pyrolysis efficiency.
In theory, the entire area under the curve of a pyrogram
could be integrated to achieve quantitation. In practice,
this would be an extremely cumbersome method of assay and
no doubt, one lacking in precision and accuracy. A more
conventional and expedient method which is used in gas
chromatography, would be individual peak area analysis, or
preferably, peak area estimation by peak height measurements.
The question that arises is whether measurement of the peak

43
height of one or two peaks in a pyrogram made up of many
peaks is sufficient for quantitation of the parent material.
Experimentally, it has been found, by the present authors,
that most pyrograms do contain at least one relatively
intense peak. By manipulating pyrolytic, or more often
chromatographic conditions, this peak can often be resolved
into a shape which lends itself to peak height measurement.
This certainly is a necessary condition for quantitation
of this peak; however, in PGC it would be insufficient
insofar as quantitation of the parent material is concerned.
Here, the identity and origin of the peak must also be inves¬
tigated. The reason for this stems from the fact that the
parent material undergoes radical fragmentation and product
formation prior to chromatography and detection.
Origin of fragmentation products. The nature of these
products will be a function of the parent material, the amount
of the parent material pyrolyzed, the pyrolysis temperature,
the cracking severity, and the pyrolysis interval. All of
these factors can influence the mechanism by which the ini¬
tially formed free radicals will propagate, i.e., via uni-
molecular decomposition reactions or bimolecular combination,
disproportionation and other reactions. Although unimolecular
decomposition has been found to be the primary process under
controlled conditions with simple molecules,^'^'^ pyrolysis
of more complicated species, under the vigorous conditions
necessary for near 100% pyrolysis efficiency, would be
expected to generate a greater variety of free radicals (of

44
varying reactivity) as well as an overall, greater number
of free radicals. Such conditions may well lead to the
formation of an abnormally high number of bimolecular
reaction products and possibly products of secondary
reactions. In qualitative analysis, all the parameters
mentioned above which control the nature of the fragmen¬
tation pattern, can be held constant and thus affect the
type of reproducibility reported here and in other studies
on pyrolysis. However, in quantitative analysis and the
construction of standard curves, the concentration or mass
is necessarily a variable. With changes in sample size,
the probability of bimolecular interaction also changes.
Consequently, quantitative schemes based on peaks originating
from unusual bimolecular reactions would be expected to re¬
flect deviations in linearity due to the non-classical
mechanisms of their formation. Peak identification then
allows for the selection of the peak or peaks which will
provide the best linearity and the best representation of
the quantity of the parent sample.
Comparison of classical and non-classical thermolysis
mechanisms. The mechanisms advanced by Rice^ are detailed
in Section I of this manuscript. These mechanisms are based
on peak characterizations and kinetic data from the pyrolysis
of light alkanes at temperatures which expose these molecules
to little more energy than that needed for decomposition
(400-650°C). In addition, the degree of pyrolysis (cracking

4 b
severity) is purposely kept very small (0.2-20%) to aid
*k
in determining the identity of primary products. The
conditions found necessary to achieve near 100% cracking
severity are quite stringent by comparison and therefore
some deviations from Rice's mechanisms should be considered
(Refer to Section I for an explanation of the symbols - "Nature
of the Chemical Reaction in Pyrolysis-Gas Chromatography"):
1) R* + P -> R* + RH
1
This step becomes less probable when pyrolysis conditions
approach 100% cracking severity.
2) R# + R* + OL + R*
11 12
3 7
Doue and Guxochon have reported that some Cr to C0 alkanes
6 9
undergo very fast 1-5 and 1-4 isomerizations. Others'^
have shown that isomerizations by 1-5 and more distant
hydrogen-atom transfers are probably much faster than
radical decomposition by bond rupture 6 to the radical
site. The conditions necessary for near 100% cracking
severity would seem to have two opposing affects on radical
rearrangements. On one hand, the large energy input, during
•k
The identity of the primary products formed after the
initial fragmentation can be facilitated by plotting the
product composition against percentage composition of the
original substrate; extrapolation of the curves to zero
decomposition indicates the primary products of the decom¬
position and their relative amounts.

46
pyrolysis at elevated temperatures, should generate
relatively energetic radical ions capable of bond re¬
arrangements. At the same time, however, this large
energy input would also tend to produce smaller radical
ions, in which case, isomerization would be a much slower
process than fragmentation. Another point to consider,
with molecules more complex than hydrocarbons, would be
the possibility of rearrangements involving aryl shifts,
where the transition state is stabilized by delocalization
of the unpaired electron over the orbital system of a
benzene nucleus, e.g.,
PhMeC-CH
I 2
Ph
PhMe£-CH2
PhMeC-CH0
I 2
Ph
The second part of step 2, propagation via decomposi¬
tion, may well remain the predominant propagation reaction
even at the high radical concentrations produced during
quantitative pyrolysis. However, high concentrations
of energetic species may result in an increase of bi-
molecular propagation reactions such as step 3 in Section I,
3) R* + OL R R*
n n n n
This may be the source of the so-called secondary reaction
products which are higher in molecular weight than the parent

compound. If two such species react early in the chain
process there is a possibility that the final termination
product will be larger than the parent.
4) 2R* -> RR
2R* -> R + OL
Under conventional pyrolysis conditions used in the study
of thermal radical kinetics and mechanisms, termination
of the reaction by radical/radical interaction is unlikely
to occur to any significant extent until the concentration
of large fragments has dropped to a very low level. The
lower the initial free radical concentration and the lower
the reactivity of the radical toward combination (either
as a result of resonance stabilization or steric inacces-
sability of the free electron), the more likely that these
bimolecular termination steps will occur with simple
fragments, i.e., H*, CH*, and H*. The conditions employed
here for quantitative analysis might very well lead to
"premature" combination or disproportionation reactions
resulting in the formation of non-classical fragmentation
products.
Quantitation of Penicillins and Cephalosporins
In Section II the characterization of a number of
antibiotics using PGC was discussed in detail. The

48
pyrograms of these materials proved to be highly re¬
producible and a re-examination showed that many of
them contained peaks which could potentially serve as
the basis for quantitation of the parent drug. Cracking
severity and peak identification studies were first
carried out, adhering to the prerequisites for quantita¬
tion. The peak identification method employed consisted
of a pyrolysis unit interfaced with a GC/MS system which
is described in full below.
Discussion and Results
Cracking severity measurements. For this experiment
it was decided to chose one antibiotic from each of the
four groups into which they had been divided during the
qualitative studies (see Section II: Characterization
of Penicillins and Cephalosporins by PGC; Discussion and
Results). Once selected, penicillin G (benzyl penicillin,
Figure 12), oxacillin (isoxazolyl penicillin, Figure 15),
methicillin (Figure 18) and cephalexin (cephalosporin,
Figure 21) were then tested according to the procedure
described previously. Table I shows the results of these
cracking severity measurements and it can be seen that,
within the range of quantities studied, pyrolysis was
essentially complete. Again, it was assumed that any
substantial change in this degree of efficiency with
greater amounts of material would be reflected by deviations
in standard curves.

49
Peak identification - Instrumental methods. A
variety of methods has been established for the identi¬
fication of gas chromatographic peaks. However, in
terms of the certainty of results and the expediency
with which they are obtained, the integrated GC/MS
system is superior to any available. Fortunately, GC/MS
facilities were available to the authors and the pyrolysis
unit used in these studies was readily coupled to the
system. Normally, experimental conditions were determined
in advance on a separate PGC apparatus before peak analysis
was carried out on the PGC/MS instrument. The same
pyrolysis unit and GC columns were used in both systems.
It was decided at the outset that those peaks most likely
to lend themselves to peak height measurement should be
of primary concern. Consequently, small or unresolved
peaks were usually ignored, certainly not from a lack of
interest in their identity, but because of the limits of
the PGC/MS unit which had no computer data system.
Peak identification - benzyl penicillins. The
pyrograms of these penicillins and the identity of their
major peaks are shown in Figures 25-27. Benzyl nitrile
{6h minute peak) was present in all of these compounds in
varying amounts. Its mass spectrum (Figure 36) is
characterized by a large M-l ion which is common for
nitriles having hydrogens a to the CN group. This product

50
★
may originate by a unimolecular process since the basic
structural unit (Ph-C-C-N) is contained in the acyclic
portion of all these penicillins. In the benzyl-
unsubstituted penicillin G, dehydration and cleavage
of the acyclic portion of the parent molecule leads
directly to the product:
-> Ph-CH2-CEN
For ampicillin and carbenicillin, generation of the
product by a unimolecular process would involve loss of the
benzyl substituent, intramolecular hydrogen abstraction via
a 1-5 radical isomerization, subsequent loss of water and
fragmentation 8 to the odd electron on the developing
nitrile carbon:
★
Pyrolysis conditions here were the same as those used in
the qualitative studies described in Section II and repre¬
sent a substantial departure from the classical conditions
employed by Rice^ and other workers21'22 qn their mechan¬
istic studies. Little information is available concerning
mechanisms under these extreme conditions and consequently,
the reactions depicted here are intended to represent only
the possibility for product formation based on an extra¬
polation of known classical mechanisms.

51
O
Ph-CH -C-N-
2
'N'
7^
O II H
CO -
2
CH
CO
•y
Ph-CH -C=N
-N-
Aj
2
CH
CH
CH
Ph-CH -CEN
2
A comparison of the size of the benzylnitrile peak
in these three pyrograms shows that the amount formed
from pyrolysis of ampicillin is measurably less than for

52
the other two penicillins. This could point to cleavage
of the amine function as a rate-limiting step (AH° = +40
Real) as opposed to the generation of CO2(AH^ = -94 Real)
in carbenicillin, and the straightforward dehydration and
cleavage reaction possible with the unsubstituted penicillin
G. Also, retention of the amine moiety and generation of
benzonitrile appears to be a preferential fragmentation
pathway with ampicillin, as can be seen from the intensity
of the 1 3/4-minute peak. Loss of water and hydrogen,
followed by cleavage of the bond between the benzyl
carbon and the carbonyl group of the side chain, could
lead to the formation of benzonitrile (mass spectrum,
Figure 37) with ampicillin as well as with the other a-
amino benzyl antibiotics studied:
Ph
Two other interesting compounds identified in this
series were bibenzyl (mass spectrum, Figure 38) and 1-
phenyl-2-propanone (mass spectrum, Figure 39). Bibenzyl

was found in all three pyrograms and is clearly the product
of a bimolecular radical/radical termination reaction. The
benzyl radical is quite stable and less reactive than, for
example, simple alkyl radicals, because of delocalization
of the unpaired electron over the tt orbital system. Again,
it is noted that the amount of this radical formed from
pyrolysis of ampicillin is measurably less than for
penicillin G and carbenicillin. l-phenyl-2-propanone appears
to be formed from the combination of a methyl radical and
the benzyl carbonyl radical. Although this is a bimolecular
process, the reaction may be intramolecular, because of the
proximity of the two methyl groups on the C-3 carbon of the
thiazolidine ring. The combination of a benzyl and an acetyl
radical may seem a more realistic route to the product; how¬
ever, no reaction is immediately obvious which would generate
the latter species.
Peak identification - isoxazolyl penicillins. The
pyrograms of these penicillins and the identity of their
major peaks are shown in Figures 28-30. Again, benzonitrile
was a predominant peak occurring in all three pyrograms.
A comparison of these peaks in the three pyrograms shows a
decrease in intensity with increasing substitution of the
benzene ring. Presumably cleavage of the isoxazolyl ring
is equally probable for all three species and the relative

proportion of benzonitrile formed reflects a decreasing
probability of consecutive scissions of the two carbon-
chlorine bonds.
The mass spectra of o-chlorobenzonitrile (Figure 40)
and 2,6-dichlorobenzonitrile (Figure 41) are characterized
by the isotopic clusters around the molecular ion resulting
from the presence of one and two chlorine atoms. Clearly,
these two compounds are products of a unimolecular frag¬
mentation process and would be likely candidates for
quantitation of the parent material.
Peak identification - methicillin and penicillin V.
As was decided earlier, no extraordinary attempts would be
made to characterize minor or unresolved peaks. Hence,
nafcillin (Figure 19) was not investigated in this quanti¬
tative study. The pyrogram of methicillin (Figure 31)
is dominated by one peak which was found to be 1,3-dimethoxy-
benzene (mass spectrum, Figure 42). No evidence was found
to suggest the formation of a substituted species, i.e.,
a combination of the 1,3-dimethoxybenzene radical with
some radical other than HI This may be indicative of rapid
hydrogen abstraction from the acid side of the molecule
following cleavage of the phenyl-carbonyl bond.
The major peak in the penicillin V pyrogram (Figure 32)
was found to be phenol (mass spectrum, Figure 43). The
phenoxy radical is quite stable due to the delocalization
of the unpaired electron to the ortho- and para-positions

of the benzene ring. It might, therefore, be expected
that dimeric products would form here as occurs with
those compounds which generate the analogous benzyl
radical (on thermolysis). Five distinct dimeric products
are possible (this excludes peroxide formation); however,
none of these was identified in the pyrogram.
The phenylacetate (mass spectrum, Figure 44) formation
is probably analogous to the formation of l-phenyl-2-
propanone seen earlier in penicillin G. Because the
origin of the compound is questionable and the amount
produced is rather small, it was not considered a good
choice for quantitative study.
Peak identification - cephalosporins. Due to a lack
of prominent peaks in its pyrogram, cefazolin (Figure 23)
was not investigated in the quantitative study. The pyro-
grams of the other three cephalosporins and the identity
of their major peaks are shown in Figures 33-35.
As was the case for ampicillin, the two a-amino benzyl
cephalosporins revealed a large benzonitrile peak in addition
to a comparatively small benzylnitrile and bibenzyl peak.
Apparently the fragmentation processes for these compounds,
with regard to the acyclic side chains, are quite similar,
if not identical, and seem to be unaffected by the structural
changes in the acid portion of the molecules.
The acyclic side chains of cephalothin (Figure 35) and
penicillin G (Figure 25) are identical, except for the sub¬
stitution of thiophene for benzene in the former. The identity

of the three major peaks in the cephalothin pyrogram are,
in fact, the sulfur analogs of those peaks seen in the
penicillin G pyrogram. The mass spectra of these compounds
(Figures 45-47) are all characterized by the strong thenyl
ion at m/e 97, and the M+2 peak of the sulfur isotope.
Interestingly, the R2 substituent on the C-3 carbon of
the cephalosporanic acid nucleus would appear to be a
source of acetyl radicals. Despite this, there appeared
to be no disproportionate amount of 2-thienyl-2-propanone
formed upon pyrolysis of cephalothin. This may lend support
to the suggestion of a radical/radical combination reaction
between benzylcarbonyl (thenylcarbony1) and methyl, to
form the substituted propanone.
Preparation of standard curves. After completion of
the cracking severity and peak identification studies it
was decided to attempt the construction of standard curves
using the same four antibiotics employed in the cracking
severity experiment.
The major peak in both the oxacillin and cephalexin
pyrograms had been shown to be benzonitrile. This compound
appeared to be the product of a unimolecular fragmentation
process and therefore, a species which should display good
linearity, at least over a limited concentration range.
The major peak in the methicillin pyrogram (1,3-dimethyl-
benzene) and in the penicillin G pyrogram (benzylnitrile)
also appeared to be derived from unimolecular processes

which should display a minimum of concentration dependence
and therefore, good linearity.
The sample sizes used in preparation of the curves
ranged from 10 nanograms to 100 micrograms. Each curve
consisted of a minimum of 27 data points from 9 different
sample sizes across the indicated range. Because of the
large range, the data points were plotted in logarithmic
form (Figure 48). Table II provides an explanation of the
symbols used on the graph along with the log-log slopes and
the regression coefficients. The adherence to linearity
was quite good throughout the entire,range with no percep¬
tible deviations at large sample sizes for any of the four
antibiotics tested.
No attempt was made to quantitate the other antibiotics
used in the study, as it was felt that the four that were
chosen fairly represented the group and, indeed, were
representative of a larger group of non-volatile materials
which do not lend themselves to conventional gas chroma¬
tography methods. The experimental results clearly show
the potential for quantitative application of the PGC
system employed in the assay of these non-volatile materials.
Experimental
Apparatus. Gas Chromatograph - Varian Model 2740 with
modified inlet for pyroprobe interface which was mounted

58
externally. The chromatograph was equipped with an F.I.D.
detector. Column temperature was 110°C; carrier gas,
helium at 60 ml/min. ambient flow rate; chromatographic
column; 6 ft. x 1/8 inch i.d. stainless steel column
packed with 3% XE-60 on 80-100 mesh Gas Chrom Q solid
support.
Pyrolysis Unit - Chemical Data Systems Pyroprobe
150 equipped with platinum ribbon probe. Operating
conditions: final pyrolysis temperature 900°C; rate of
temperature rise 20°C/millisecond; pyrolysis interval
1.0 second.
Mass Spectrometer - Dupont 490-F single focusing,
magnetic sector instrument.
Antibiotics. (See "Experimental," Section II).
Procedure. Cracking severity measurements are
described in the discussion and results portion of this
section. Samples ranging from 10 yg to 40 yg of the sodium
salts of the antibiotics were pyrolyzed and subjected to
combined gas chromatographic-mass spectrometric (70 eV
electron impact) analysis utilizing a jet separator.
Identification of peak components in the pyrograms was
based on comparison with authentic samples or the matching
41
of mass spectra obtained m reference files.

Quantitation of Saccharin
Discussion and Results
Cracking severity measurements. The use of
chromatographic and pyrolytic conditions very similar
to those employed for the qualitative and quantitative
analysis of the antibiotics produced a sodium saccharin
pyrogram consisting of just two major peaks. Cracking
severity measurements were carried out (monitoring both
peaks) following the procedure described earlier in this
section. The results were very similar to those obtained
in the antibiotic series and it was decided to continue
on to the next prerequisite phase, peak identification.
Peak identification - saccharin (sodium, salt). PGC/MS
analysis revealed the two major peaks in the sodium saccharin
pyrogram to be benzonitrile and biphenyl, one of which had
been observed previously in a number of the antibiotic
pyrograms. It may be recalled from Section I that Kirk
and NelsonJU had observed the formation of nitriles upon
pyrolysis of the barbiturates. The conditions they used
were quite similar to those employed in the present quan¬
titative studies. In light of these results, it can be
concluded that nitriles will be predominant pyrolysis products
for compounds containing the amide, imide, sulfonamide,
sulfimide and related functional groups.
It was decided to base the Quantitative studies on the
benzonitrile peak, since its formation was probably the

60
result of a concentration independent intramolecular
process involving interaction of the saccharin nucleus
with the two water molecules of the hydrated salt. Either
of these intimate waters of hydration could supply the
hydrogen atom for the aromatic ring following scission
of the C-S bond, ortho to the evolving nitrile function.
Preparation of standard curves. During the course of
these experiments, many standard curves of sodium saccharin
were prepared prior to assaying the various diet beverages.
All curves displayed excellent linearity (regression co¬
efficients >0.997) throughout the range studied (1.0-10.0
Mg) •
Determination of sodium saccharin in diet beverages.
Several commercial diet beverages (sodium saccharin content:
50-150 mg/12 fluid oz.) were randomly selected and quan¬
titation was attempted according to the procedure used
earlier in the qualitative studies (see Section II). This
consisted of evaporating a few microliters of the beverage
on the ribbon surface and directly pyrolyzing the residue.
This process was carried out on several different samples
and the results showed a high degree of precision; however,
the experimental values were always higher than the levels
indicated for the particular beverage analyzed. It was
suspected that the high values were a result of interference
from some common ingredient in the beverages which, upon
pyrolysis, generated benzonitrile or another species having

61
a peak coincident with that of benzonitrile. A check
of the various beverages revealed that sodium benzoate
was the only such common ingredient indicated on the
labels. Although the amount of sodium saccharin varied
from product to product, the benzoate was always present
to the extent of 1/40 of 1% (or, 0.25 mg/ml as opposed to
saccharin levels ranging from 0.14-0.44 mg/ml). Simon
and Giacobbo^ had earlier studied the pyrolysis of sodium
benzoate and, under conditions approximating those used
in this study, they identified the major pyrolysis product
as benzene, accompanied by smaller quantities of benzyl
alcohol, phenol, biphenyl and benzaldehyde. Sodium benzoate
was then studied by the present authors under conditions
established for saccharin quantitation. Results similar
to Simon and Giacobbo's were obtained. Two major and two
minor peaks were identified by a comparison with retention
times of those products described above. Samples of sodium
saccharin and sodium benzoate were pyrolyzed separately and
as mixtures and it .was determined that the benzaldehyde
fragment arising from the pyrolysis of sodium benzoate was
coincident with the benzonitrile peak in the sodium saccharin
pyrogram.
Attempts were made to alter pyrolytic and chromato¬
graphic conditions to resolve these two peaks; however, little
success was achieved and it was decided to further purify

the samples prior to pyrolysis using extraction methods.
The difference between the pKa values of saccharin (1.60)
and benzoic acid (4.19) was sufficient to allow selective
extraction of the two compounds by appropriate acidification
of the decarbonated beverage. Application of this technique,
to the determination of sodium saccharin in two name brand
diet beverages, gave the results summarized in Table III.
Although this method of analysis proved to be substantially
more time consuming than direct assay, it still offered
the advantage of elimination of the need for derivatization
3 2 3 3
described in previous papers. ' Reproducibility appeared
to be limited, not by the instrument itself, but rather by
other random errors associated with separation, purification
and chromatographic analysis.
The series of experiments described in this section
indicate that quantitation in PGC is certainly as practical
as it is in conventional gas chromatography. As in GC, the
degree of accuracy and precision attained depends upon the
care taken with the experimental work. Ultimately a reliable
internal standard method will have to evolve for PGC in
order to increase the limits of its accuracy and precision
and to increase the scope of its application. The develop¬
ment of internal standards will follow much the same criteria
as that used in conventional GC. The internal standard must
give the same, or nearly the same, response as the species

63
being assayed. In additionr the internal standard should
elute close to the peak of interest, yet, be well resolved..
Consequently in GC, hydrocarbons are selected as internal
standards for hydrocarbon analysis, fatty acids for fatty
acid analysis, etc. In PGC, the internal standard should
undergo a similar transformation on pyrolysis, which is
all the more reason to use a homolog of the compound of
interest. As an example, benzonitrile is seen to arise from
the pyrolysis of a number of compounds containing amide or
imide type functional groups, with a variety of substituents
at the site of functionality and in the aromatic ring.
Logically, one would investigate those compounds which also
contain these functional groups with some slight variation:
0
II

These experiments have shown that there are a
number of core functional groups which either survive or
evolve from the vigorous conditions necessary to achieve
absolute quantitation. These core species can often be
found, in some form, incorporated into the structure of a
large variety of high-molecular weight compounds, polar
compounds, and generally non-volatile materials. Under
the proper pyrolysis conditions, these core species can
be generated from the parent compound and thus serve as
the basis for quantitating a large number of materials.
Accordingly, quantitation based on benzonitrile could be
internally standardized with the addition of, e.g., an
appropriately substituted amide which then would serve as
the PGC internal standard for several antibiotics, a sweeten¬
ing agent, and a number of sedative/hypnotics and PTH amino
acid derivatives.
Experimental
Materials. Solvents - All solvents were reagent
grade and were not further purified.
Sodium saccharin - (C^H^NO^SNa*2H2O) Mallinckrodt
U.S.P. powder.
Diet beverages - These were purchased commercially
(see "Experimental," Section II).

65
Apparatus. Gas Chromatograph - Varían model 2740
with modified inlet for pyroprobe interface which was
mounted externally. The chromatograph was equipped
with an F.I.D. detector. Column temperature 150°C;
^chromatographic column 12 ft. x 1/8 inch i.d. stainless
steel column packed with 3% XE-60 on 80-100 mesh Gas Chrom
Q solid support.
Pyrolysis Unit - Chemical Data Systems pyroprobe 150
equipped with platinum ribbon probe. Operating conditions:
final temperature 900°C; rate of temperature rise 20°C/
millisecond; pyrolysis interval 1.0 second.
Mass Spectrometer - Dupont 490-F single focusing,
magnetic sector instrument equipped with a jet separator
for PGC and GC/MS.
Procedure. Quantitation of diet beverages - 20 ml
of the soft drink was pipetted into a 100-ml separatory
funnel, made basic with 2 ml of 10.N NaOH, and extracted
with two 25-ml portions of diethyl ether. The combined
ether extracts were washed with 20 ml of water, the ether
discarded, and the washings were combined with the original
aqueous layer. The aqueous layer was acidified to a pH of
3.5-3.6 and extracted with three 25-ml portions of diethyl
ether. The combined ether extracts were washed with 20 ml
of water, dried over anhydrous sodium sulfate and evaporated-
to dryness (benzoic acid fraction). The original aqueous
layer was further acidified to a pH near 0 and extracted

with three 25-ml portions of ethyl acetate. The combined
extracts were washed with 20 ml of water, dried over an¬
hydrous sodium sulfate and evaporated to dryness on a
rotary evaporator. The residue was dissolved in 1.0 ml
of 0.025N NaOH and was used immediately for PGC analysis.

SECTION IV
ADDITIONAL APPLICATIONS OF PYROLYSIS-GAS
CHROMATOGRAPHY AND CONCLUDING REMARKS
The Use of Pyrolysis-Gas Chromatography for the
Diagnosis and Study of Metabolic Disorders
The Urine Pyrogram
Pyrolysis of serum and urine,. During the course of
the present work, preliminary studies of drug assay in
biological fluids was undertaken. Much of this work was
concerned with removal of endogenous materials from the
fluids (via extraction and precipitation methods) which
would interfere with the pyrolysis patterns of drugs. Use
of standard separation methods made it possible to reduce
this interference, in serum, to just one peak in the
pyrogram of a treated blank sample.
The urine pyrogram, however, was seemingly unaffected
by changes in pH, extraction with solvents, absorption
methods, etc,. If PGC analysis were to be pursued with
urine, it would probably have to be carried out on the
residues remaining after workup of the fluid for conventional
GC urinalysis.
The reproducibility of the urine pyrogram is quite
67

amazing, considering the heterogeneous nature of the
solution. The physical and chemical properties of urine
are exceedingly variable and change substantially with the
nature of the diet. Nevertheless, the pyrograms of urine
obtained from a number of adult males are qualitatively
superimposable, almost peak for peak (see Figure 49 for
an example of a typical urine pyrogram). Differentiation,
thus far, has been based solely on differences in the
relative peak heights in the pyrograms. As might be ex¬
pected, these quantitative differences can be detected in
the urine pyrogram of an individual when monitored over
several days. Such changes are presumably dietary-in nature
No study has yet been carried out to determine if acute
ailments have an effect on the appearance of a normal urine
pyrogram.
PGC/MS analysis of urine. These studies had not been
completed at the time of writing; however, those peaks which
have been identified are indicated in Figure 49. Pyrolyses
were carried out at high temperatures (900°C for 1.0 sec.)
and this is reflected in the nature of the compounds noted.
These species represent some of the core compounds mentioned
in Section III which survive or evolve in these extreme
pyrolysis conditions. These compounds probably originate
from several different materials normally found in the urine
Amino acids are excreted in the urine to the extent of 2-3
gm/day and would be expected to contribute significantly to

31
the urine pyrogram. Merritt et al. have shown that
several of the amino acids will generate compounds
identified in Figure 49 under conditions very similar
to those used in the present study:
Phenylalanine
Tyrosine
Proline
Hydroxyproline
Pyrrole
Phenylalanine and tyrosine might also be expected to
generate benzylnitrile, benzene and phenol. Benzonitrile
may be formed as a result of a bimolecular reaction of the
benzene and nitrile radicals; the source of the nitrile
radical presumably could be urea which is excreted in the
urine in amounts up to 20 gm/day by adults. Reaction of
this radical with fragments from the alkyl amino acids, or
direct unimolecular generation from aliphatic peptides may
explain the presence of acrylonitrile.
Chromatographic conditions presently employed can be
improved to maximize the resolution and peak shapes in the
pyrograms. The reproducibility of the urine pyrogram was
first observed using a 6-foot, SE-30 column and operating
isothermally at 110°C. These conditions produced a pyrogram
which consisted of only five predominant peaks. The present
system utilizes a 12-foot, XE-60 column and is programmed

70
from 35-175°C at 10°C/min. These conditions produce a
pyrogram'such as the one seen in Figure 49. The present
system can be further improved upon by extending the pro¬
grammed temperature range; however, as the upper limit is
increased, complications may arise with PGC/MS analysis due
to contamination o'f the GC detector and mass spectrometer
from column bleed.
Utilization of the Urine Pyrogram
Present use of GC/MS for diagnosis of metabolic dis¬
orders . An increasing number of human diseases is now
known to result from defects in some biochemical process.
Clinically, these disorders often manifest themselves as
severe mental deficiencies, epilepsy, muscular dystrophy,
failure to thrive, irritability and acidosis among other
symptoms. Biochemically, these symptoms are usually related
to enzyme deficiencies leading to accumulation of certain .
toxic metabolities in the blood and tissues. Early diagnosis
and treatment of these disorders often depends on the de¬
tection of these abnormal metabolites in the body fluids.
Thus, a rapid method of analysis and characterization
of these metabolites becomes a critical factor in preventing
progression of the disease state.
A substantial amount of work has been done in this
area since the early 70's using GC/MS methods. Williams
and Halpern^ identified a number of common urinary and

71
serum amino acids by GC/MS of their neopentylidene alkyl
ester derivatives. By comparison with the amino acid
profiles from urine samples of mentally retarded children
and other selected patients, they found that they were
able to detect 15 of the known inborn errors of amino
acid metabolism. Nyhan^ was able to successfully develop
simple screening methods for a number of amino acid and
purine metabolic abnormalities using GC/MS methods to
analyze urine extracts.
The advantages of GC/MS over older methods are
numerous. The latter were often based on color reactions,
e.g., ninhydrin with amino acids, and consequently were
often dependent on the presence of a functional group like
the amino or keto group. On the other hand, volatilization
via derivatization, followed by GC/MS, allows for the
identification of over 75 different compounds” in the
normal urine sample. Not only are the amino acids
characterized but also the a-keto carboxylic acids and
the carbonyl compounds which are formed as a result of
amino acid metabolism. Even with derivatization, GC/MS is
still a much quicker method than the older schemes which
required paper chromatography or two-dimensional thin layer
chromatography. It is worth noting that about 1970 the
discovery of new inborn errors of amino acid metabolism
began to level off from the sharp incline of the previous
20 years. According to Nyhan^ this was a function of the

fact that most of those things that could have been
discovered by the limited conventional color tests had
already been discovered. However, with the advent of
GC/MS methods, the discovery of inborn errors began to
increase again, at a greater rate than noted in the 1950-
1970 period.
The drawbacks to the GC/MS method applied here arise
from the same difficulties encountered in all assay methods
of non-volatiles. Derivatization, in particular, silylation,
often results in the formation of numerous products if there
is more than one reactive group in a molecule. Multiple
derivatization is also a problem when diazomethane is used
and, of course, these difficulties will carry over and
complicate the mass spectral as well as the chromatographic
analysis.
App1ication of PGC and PGC/MS to the diagnosis of
metabolic disorders. If the pyrogram, seen in Figure 49,
accurately depicts a normal urine sample, then it is highly
possible that a pyrogram may also be representative of
urine containing abnormal levels of metabolites. In a
preliminary investigation, several urine samples were
obtained from pediatric patients suffering from a variety
of inborn metabolic defects, as well as other illnesses
caused by metabolic dysfunctions. In addition to these,
several samples from normal adults and children were also
acquired to serve as standards. The experimental conditions

7 3
employed were the same as those described earlier. Pyrolysis
of ~5 pi of pooled adult urine produced the pyrogram seen
in Figure 49. The samples used in this experiment were
subjected to PGC on three consecutive days so that the
reproducibility of the method could be established. Samples
were kept at 4°C to retard any bacterial growth. Previous
4 5
studies indicate that if urine is allowed to stand for
several hours at room temperature, microorganisms introduced
into the urine from the skin during voiding can produce
appreciable amounts of unusual metabolites. No changes
were noted in the pyrolysis patterns of these samples over
the time span of the experiment. Pyrograms obtained from
the first series of pyrolyses were essentially superimposible
on those from the last series of pyrolyses, after correction
for sample size.
Since only a few of the peaks in the urine pyrogram
have been identified, differentiation of the pyrograms
presently has to be based on differences in the overall
fingerprint of the sample. Even then, there are discernable
differences between the urine pyrograms from the diseased
patients and the standard, and between the urine pyrograms
of the diseased patients themselves. It was mentioned
earlier that many of these diseases are characterized by
the abnormal accumulation of some endogenous material due
to enzyme defects; consequently, what appears to be a "new"
peak in a diseased urine sample may also be present in a

7 4
normal urine pyrogram, but originating from the analyte
in much smaller amounts. Figure 50 represents an example
of a pyrogram of a urine sample from a patient suffering
from methylmalonic aciduria. This disease is presumably
caused by a defect of the methylmalonyl CoA mutase enzyme
and is characterized, biochemically, by an abnormal accum¬
ulation of methylmalonic acid, propionate, a-methylaceto-
acetic acid, butanone, hexanone, and glycine in the blood,
tissues and urine. In normal urines methyl malonic acid
appears in extremely small quantities. This urine sample
was quite dilute and some of the characteristic urine peaks
may be lost because of this. This fact accentuates the
intensity of the peaks seen at ~3, 12 and 16 minutes. The
~12 minute peak may be benzylnitrile and the small peak at
~5h minutes is probably pyrrole. The ~16 minute peak is
unidentified, and based on the difference in peak shape it
is probably a material different from that seen at approxi¬
mately the same retention time in the standard urine pyrogram
(Figure 49). The peak at ~3 minutes in Figure 50 has not
been observed previously in normal or abnormal urines and,
although its origin is unknown, it may nevertheless serve
as a basis for characterization of this disorder.
Many hundreds of urine samples will have to be screened
before the utility of this procedure can be determined. A
patient with a particular disease should be monitored over
an extended period of time and results compared with those
from other patients with the same disorder so that any common

features of the urine pyrogram can be established. PGC
and/or PGC/MS would be a faster method than GC/MS for
screening of large numbers of urine samples since the
need for isolation and derivatization would be eliminated.
However, the identification of new metabolites which may
occur in previously undiagnosed diseases would be a very
difficult task with PGC/MS, since the materials in the
urine sample undergo substantial chemical transformation
prior to detection. Such cases would best be dealt with
using GC/MS where the new species could be directly
identified.
Concluding Remarks
Qualitative Applications of Pyrolysis-Gas Chromatography
The ability to characterize materials based on their
pyrolysis fragmentation patterns is presently the most
desirable feature of PGC. Its fingerprinting techniques
have proven invaluble in identification and classification
of non-volatile substances. The present study has inves¬
tigated the qualitative applications of PGC in the pharma¬
ceutical and related areas. The results obtained are
promising and indicate that PGC can be utilized as a
research tool and potentially as a routine analytical tool
in these areas.
The characterization and differentiation of the

76
penicillins and cephalosporins demonstrates the utility
of PGC as an ancillary analytic method which can comple¬
ment GC and HPLC. In addition, the specificity provided
by the pyrogram measurably increases the certainty of
identification and decreases the time required for analysis.
This increased specificity is especially useful when one
is attempting to differentiate between two members of the
same group of compounds. Whereas two homologs may have
similar peak shape and retention time in GC and HPLC, the
difference between the pyrograms of the compounds will
usually be definitive. If a sample of the suspected unknown
is available, then it can be added to verify its presence.
The advantage of the pyrogram, of course, is that every
peak in the pyrogram of the compound will increase as the
quantity added increases, thereby providing a multiple
reassurance of its identity. It has been noted in the
literature that PGC can frequently differentiate between
positional isomers whose mass spectra are identical. PGC
data which has been classified by set theory^® has made it
possible to distinguish between geometrical isomers with
seemingly identical pyrograms. The fingerprinting technique
is limited in its ability to distinguish between compounds
in a mixture, except for, perhaps, simple molecules whose
pyrograms are correspondingly simple.
In Section I, pyrolysis was classified according to
the extent of degradation of the parent material. An example

of "thermal degradation" wag given where tertiary amine
oxides were pyrolyzed at 150°C and converted to the
corresponding olefins. Under these controlled conditions
it may be possible to examine mixtures of homologs. As an
example, a series of 3-substituted carboxylic acids could
be decarboxylated in the pyrolysis chamber and then
successfully resolved as the corresponding alkenes. This
type of selectivity would not be feasible with compounds
containing multiple functional groups, like the antibiotics,
where it was found necessary to extensively fragment the
molecule in order to increase volatility. The experiments
performed with urine samples demonstrate a different
situation. Pyrolysis of such a complex mixture virtually
precludes the identification of any one particular metabo¬
lite which may represent the biochemical manifestation of
a disease. Nevertheless, once the disease is characterized,
the overall urine fingerprint of subsequent patients
afflicted by that disease may provide the basis for the
most expeditious diagnosis.
Quantitative Applications of Pyrolysis-Gas Chromatography
This area of application has received very little
attention over the years. This stems partially from the
lack of available instrumentation which is capable of the
required precision and accuracy. It has been pointed out
that the apparatus used in this study offers as high a

78
degree of control over the .heating profile as is commer¬
cially available. The cracking severity measurements
carried out, show that the entirety of a sample can be
pyrolyzed; however, the heating profile should be defined
for each particular compound or group of compounds so that
the best compromise can be reached between pyrolysis
conditions and burn efficiency. The possible decarboxy¬
lation of the 3-substituted carboxylic acids, noted above,
may serve as an example. Before attempting to construct
a standard curve, two questions should be answered: 1) What
is the minimum temperature allowable which will effect
decarboxylation and volatilization? 2) What is the minimum
temperature and burn time allowable that will effect a 100%
burn efficiency at the high end of the range to be studied,
e.g., 1-20 yg? If 20 yg cannot be totally volatilized
without increasing the severity of pyrolysis conditions
to a point where side reactions complicate analysis, then
the assay of mixtures (but not single components) becomes
improbable. These optimum conditions will vary from one
type of compound to another depending upon the nature and
number of the functional groups present, and there is always
a possibility that they may not exist at all. The tertiary
amine oxides are easily volatilized and chromatographed at
150°C, whereas the present workers were seemingly unable to
affect the prohibitively large partition ratios of the
tetracyclines, even under the most vigorous pyrolysis
conditions.

Peak identification was described as a prerequisite
to quantitative analysis. This procedure at least allows
the analyst some inhight as to the origin of a species and
the mechanism of its formation. In addition, such infor¬
mation would be essential for the development of appropriate
internal standards. Although no attempt was made to
quantitate the changes in the amount of bimolecular reaction
products formed in these experiments, it was obvious from
the pyrograms that such species as bibenzyl increased in a
non-linear fashion with increased sample size. Peak iden¬
tification eliminates the possibility of erroneously basing
quantitation on this sort of bimolecular product whose
formation is concentration-dependent.
One of the major concerns of this study, stated in
Section I, was the attainment of the maximum amount of
information from as simple and economic a pyrolysis system
as possible. The authors were fortunate enough to have
access to a mass spectrometer for peak analysis, and,
considering the complexity of most pyrograms, PGC/MS is
not only the most expeditious method of peak analysis but
the most practical as well. This does not, however, neces¬
sitate that the PGC unit be permanently integrated with a
mass spectrometer. On the contrary, even when used as a
research tool, the pyrolysis unit need only be interfaced
with the GC/MS for the initial characterization of the
pyrogram. For example, PGC/MS analysis of the 12 antibiotics

80
studied, required two workers about two hours to complete.
This, of course, excludes the job of interfacing the
pyrolysis unit and the time involved in the interpretation
of the mass spectral data. Once these interpretations
are completed however, these antibiotics can be characterized
and quantitated routinely, without further need to employ
the mass spectrometer. Certainly on an intralaboratory
basis this PGC/MS data can be stored and referred to for
ensuing routine PGC analysis. As PGC methods become more
standardized, the analyst will hopefully be able to rely
on reference tables for peak analysis. Reproducibility
could then be checked by a comparison of the shape, size
and retention time of one or two peaks in the pyrogram
which may represent readily available chemical entities.
The future of PGC as a qualitative and quantitative
tool depends on the willingness of individual investigators
to collaborate in the development of a standard reproducible
methodology. This is especially critical when identification
is based on fingerprinting, as is the case with micro¬
organisms paints and resins, and the urine samples discussed
above. In these instances, all variables will have to be
rigidly defined to assure interlaboratory reproducibility;
this includes chromatographic as well as pyrolytic parameters.
The authors have attempted to maintain some consistency in
experimental conditions in order to minimize the number of
variables involved in an analysis. The rate of temperature

81
increase should be as fast as possible. This parameter
remained constant at 20°C/milliseconds (the maximum con¬
trollable rate of the instrument) throughout these studies.
Pyrolysis temperatures were necessarily variable; however,
a 1.0-second burn time was found sufficient for all
experiments carried out. Pyrolysis intervals up to 30
secs, can be found in the literature but, these are for
very large sample sizes where no possibility exists for a
uniform hea't exchange between the source and the material.
Chromatographic conditions are usually more varied as is
characteristic of GC. Nevertheless, high flow rates are
recommended to minimize the time spent in the reaction zone
by the pyrolysis fragments. This helps to avoid high
concentrations of free radicals and prevent bimolecular
and secondary reactions.
In these studies a helium flow rate of 60 ml/min. was
used in all experiments. Calculations indicated that this
rate was really not much faster than the optimum rate
required for a minimum plate height (HETP). It was felt
that helium, rather than nitrogen, was the best choice of
carrier gas since it has been reported that nitrogen pyrolysis
products have been identified in the pyrograms of aromatic
carbocyclics, presumably as a result of reaction with the
carrier gas. Finally, silicone XE-60 evolved as the liquid
phase of choice for all analyses performed. Although both
OV-225 and XE-60 have the same recommended maximum

ü ¿
temperature (275°C), it was found that there was sub¬
stantially less column bleed from the XE-60, in addition
to better overall resolving power with this stationary
phase.

APPENDIX I

Table I. Cracking Severity Measurements of Four Representative Antibiotics.
Antibiotic
Cephalexin
Oxacillin
Methicillin
Penicillin G
Severity5'^
99.4
98.4
98.1
100.0
Cracking
Relative S.D. (%)c
0.2
0.9
0.7
Sample Weight (yg)
4
2
10
5
alst burn efficiency - 1/p where P^ = peak response for the initial pyrolysis
and, P2 = peak response for the second pyrolysis.
^These figures represent an average of at least four determinations.
cRelative standard deviation = S.D./X-

85
Table II. Symbol Key and Statistics for Standard Curves
in Figure 48.
Antibiotic â–  Log-Log Slope. Regression Coefficient
| Oxacillin
0.97
0.996
Cephalexin
1.08
0.999
0 Penicillin G
0.89
0.993
/\ Methicillin
0.88
01 9 98

86
Table III. Determination of Saccharin in Two Name Brand
Diet Beverages.
Amount
Determined3
(mg/12 fluid oz.)
Beverage
Sample 1
Sample 2
Sample 3
Sample 4
Cola A
6 6.3
76.5
63.6
Cola B
38.4
36.2
41.7
40.5
aEach result is the average of at least three determinations
on the same bottle.

APPENDIX II

rp-i
'1CJU1
1. Pyroorobe 150 solids pyrolyzei

o
SACCHARIN (o-BENZOSULFIMIDE)
Figure 2. Saccharin structure.
cc

RESPONSE
90
Figure 3. Pyrogram of saccharin.

RESPONSE
j i
Figure 4.
Pyrogram of sodium saccharin.

RESPONSE
©
Figure 5.
Pyrogram of a 5 yl (~2 yg) sample of Diet Rite
ginger ale.

RESPONSE
93
Pyrogram of a 5-ul sample of beverage in Figure 5
with the addition of 1.0 yl of a 2.0% sodium
saccharin solution.
Figure 6.

RESPONSE
TIME (MINUTES)
Figure 7.
Pyrogram of extract (5 jil)
vitamin chew tablet.
from
Clusivol
©
multi

RESPONSE
TIME (MINUTES)
Figure 8. Pyrogram resulting from same sample size as in
Figure 7 but with 1.0 yl of a 1.0% solution of
saccharin added to the probe prior to analysis.

o
II
R-C-NH-CH
0^
BASIC PENICILLIN STRUCTURE
Figure 9. Basic penicillin structure.
cr

o
R-C-NH-CH-r"
COO"
BASIC CEPHALOSPORIN
STRUCTURE
Figure 10. Basic cephalosporin structure.
CO

RESPONSE
TIME (MINUTES)
Figure 11.
Pyrogram of carbenicillin.

RESPONSE
Figure 12. Pyrogram of penicillin G (sodium salt)

RESPONSE
Figure 13.
Pyrogram of penicillin G (potassium salt)

UJ
CO
z
o
CL
CO
LU
CL
Figure 14. Pyrogram of ampicillin.

TIME ( MINUTES)
Figure 15. Pyrogram of oxacillin.

RESPONSE
TIME (MINUTES)
Figure 16.
Pyrogram of cloxacillin.

RESPONSE
i i i 1 1 1 r
TIME (MINUTES)
Figure 17
Pyrogram of dicloxacillin

l±J
in
2:
o
a
m
UJ
cc
T T 1 1-
TIME (MINUTES)
Figure 18.
Pyrogram of methicillin.

tu
CO
2
o
íl¬
eo
UJ
oc
TIME (MINUTES)
Figure 19. Pyrogram of nafcillin.

RESPONSE
TIME (MINUTES)
Figure 20. Pyrogram of penicillin V.

TIME (MINUTES)
Figure 21. Pyrogram of cephalexin.

RESPONSE
H
NH2
TIME (MINUTES)
Figure 22. Pyrogram of cephaloglycin.

TIME (MINUTES)
Figure 23. Pyrogram of cefazolin.

RESPONSE
Figure 24.
Pyrogram of cephalothin.

RESPONSE
i i 2
Figure 25.
Peak identification in penicillin G pyrogram.

RESPONSE
113
TIME (MINUTES)
Peak identification in carbenicillin pyrogram.
Figure 26.

BIBENZYL
Figure 27. Peak identification in ampicillin pyrogram.

RESPONSE
T • I 1 T
TIME ( MIN UTES)
Figure 28.
Peak identification in oxacillin pyrogram.

RESPONSE
TIME (MINUTES)
Figure 29. Peak identification in cloxacillin pyrogram.

RESPONSE
TIME (MINUTES)
Figure 30. Peak identification in dicloxacillin pyrogram

RESPONSE
i 1 r
TIME (MINUTES)
Figure 31. Peak identification in methicillin pyrogram.

RESPONSE
TIME (MINUTES)
Figure 32.
Peak identification in penicillin V pyrogram.

RESPONSE
TIME (MINUTES)
Figure 33. Peak identification in cephalexin pyrogram

RESPONSE
Figure 34. Peak identification in cephaloglycin pyrogram.

RESPONSE
Figure 35.
Peak identification in cephalothin pyrogram.

RELATIVE ABUNDANCE
50 100 150
M/e
Figure 36. Mass spectrum of benzylnitrile.

RELATIVE ABUNDANCE (7J
O
CO
o _•
r-
o _
CD
ID
O _{_
07 T
CD
PJ
50
M/e
100
150
Figure 37. Mass spectrum of benzonitrile.
124

RELATIVE ABUNDANCE
o
CT’
Q J_
! i)|
50
100 150
M/'e
!0 0
Figure 38. Mass spectrum of bibenzyl
h
tv.
LT

RELATIVE ABUNDANCE
Figure 39. Mass spectrum of 1-pheny1-2-propanone.
126

RELI1TIVE ABUNDANCE (7.1
O !
LT; T
o
¡ ! ; : ' ¡ 1 ‘ i
50 100 150
M/e
Figure 40.
Mass' spectrum of o-chlorobenzonitrile.

RELATIVE ABUNDANCE
Figure 41. Mass spectrum of 2,6-dichlorobenzonitrile
tv-
CC

rjHbÃœMflGU
Figure 42
Mass spectrum of 1,3-dimethoxybenzene

RELATIVE ABUNDANCE
Figure 43. Mass spectrum of phenol.
Of T

RELOT I VE ABUNDANCE
M/e
Figure 44. Mass spectrum of phenylacetate
TfT

RELHT I VE REUNIDUNTE
Figure 45. Mass spectrum of 1-(2-thienyl)-2-propanone.
zft

RELOT I VE PBUNOPNCE
i : 1 i r ' ! ! 1 T i â–  T J ' T T i I
50 100 150
M/e
Figure 46. Mass spectrum of 2-thenylnitrile.
1J J

REDUNDANCE
l—1
C-
I
lu
i
!
o
;
50 1Ó0 150 200
M/e
Figure 47. Mass spectrum of bithenyl

LOG RESPONSE
i J b
LOG MICROGRAMS
Figure 48. Quantitation of four antibiotics (see Table II)
by peak height. The lines represent the
calculated linear regression lines through the
data.

Figure 49.
Pyrogram of pooled urine sample (5 yl) from healthy adult males.


Figure 50. Pyrogram of urine sample (10 yl) from pediatric patient suffering
from methylmalonic aciduria.

RESPONSE
139

REFERENCES
1.
2.
3.
4 .
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Pryor, W.A., "Introduction to Free Radical Chemistry,"
Prentice-Hall, Englewood Cliffs, N.J., 1966.
Pryor, W.A., "Radical Reactions," McGraw-Hill Book
Company, New York, N.Y., 1966.
Rice, F.O., "Free Radicals," (Collected Papers of F.O.
Rice), The Catholic University Press, Washington, D.C.,
1958 .
Beroza, M. and Coad, R.A., J. Gas Chromatog., 4, 199
(1966).
Craig, J.C., Mary, N.Y. and Roy, S.K., Anal. Chem., 36,
1142 (1964) .
Winter, L.N. and Albro, P.W., J. Gas Chromatog., 2, 1
(1964).
Dhout, J.H., Analyst, 89', 71 (1964).
Sellier, N., Jones, C.E.R. and Guiochon, G., J. Chromatog.
Sci., 13, 383 (1975).
Haslam, J., Chem. Age (London), 82, 169 (1959).
Feuerberg, H. and Weigel, H., Z. Anal. Chem., 199, 121
(1964).
Groenendyl, H., Levy, E.J. and Sarner, S.F., J. Chromatog.
Sci., 8, 599 (1970).
Wolf, C.J., Walker, J.Q. and Levy, R.L., Industrial
Research, 13, 40 (1971).
Derminot, J. and Rabourdin-Berlin, C., Bull. Inst. Text.
Fr., 25, 721 (1971).
Kirk, P.L., 17th Pittsburgh Conf. Anal. Chem. and Appl.
Spectroscopy, Abst. 71, 73 (1966).
140

141
15. Myers, A. and Smith, R.N.L., Chromatographia, 5, 521
(1972) .
16. Perry, S.G., J. Gas Chromatog., 5, 77 (1967).
17. Martinez, J. and Guiochon, G., J. Gas Chromatog., 5,
146 (1967).
18. Esposito, G.G., Anal. Chem., 36, 2183 (1964).
19. Iglauer, N. and Bentley, F.F., J. Chromatog. Sci., 12,
23 (1974).
20. Fanter, D., Walker, J.Q. and Wolf, C.J., Anal. Chem.,
40, 2168 (1968).
21. Levy, E.J. and Paul, D.G., J. Gas Chromatog., 5, 136
(1967) .
22. Groenendyk, H. , Levy, E.J. and Sarner, S.F., J. Chromatog.
Sci., 8, 117 (1970).
23.Reiner, E. , Nature, 206 , 1272 (1965).
24. Sekhon, A.S. and Carmichael, J.W., Can. J. Microbio.,
18, 1593 (1972).
25. Muezelaar, H.L.C. and in't Veld, R.A., J. Chromatog.
Sci., 10, 213 (1972) .
26. Farre''-Ruis, F. and Guiochon, G. , Anal. Chem., 40,
998 (1968).
27. Quinn, P.A., J. Chromatog. Sci.., 1_2, 796 (1974).
28. Vincent, P.G. and Kulek, M.M., Appl. Microbiol., 20,
957 (1970).
29. Wolf, C.J., Walker, J.Q. and Levy, R.L., Industrial
Res., 13, 40 (1971).
30. Kirk, P.L. and Nelson, D.F., Anal. Chem., 34 , 899 (1962).
31. Merritt, C., DiPietro, C., Robertson, D.H. and Levy,
E.J., J. Chromatog. Sci., 12, 668 (1974).
32. Conacher, H.B.S. and O'Brien, R.C., J. Ass. Offic.
Anal. Chem., 53, 1117 (1970).
33. Ratchik, E.M. and Viswanathan, V., J. Pharm. Sci., 64,
133 (1975).

i 34. Couch, M.W., Das, N.P., Scott, K.N., Williams, C.M.
and Foltz, R.L., Biochem. Med. , _8, 362 (1973).
35. Steacie, E.W.R., "Atomic and Free Radical Reactions,"
Vol. 1, Second Edition, Reinhold Publishing Corp., New
York, 1954.
36. Keulemans, A.I.M. and Perry, S.G., Fourth International
Gas Chromatography Symposium, June, 1962, Hamburg,
West Germany.
37. Doue, F. and Guiochon, G., J. Chem. Phys., 73, 2804
(1969) . ~
38. Semenov, N.N., "Some Problems in Chemical Kinetics and
Reactivity," Princeton University Press, Princeton, N.J.,
1958 .
39. Voge, H.H. and Good, G.M., Ind. Eng. Chem., 73, 593
(1949) .
40. Gordon, A.S. and McNesby, J.R., J. Chem. Phys., 31, 853
(1959).
41.Eight Peak Index of Mass Spectra, 2nd Ed., Vols. I-III,
Mass Spectrometry Data Centre, AWRE Aldermaston, Reading
RGI, 4PR, UK, 1974.
42.
Simon, W.
and Giacobbo, H.,
Chem. Ingr.
Tech.,
I UJ
709
(1965).
43.
Williams,
831 (1973
K.M. and Halpern,
) •
B., Aust. J
. Biol.
Sci .
, 26,
44.
Nyhan, W.
L., presented at a
workshop on
application
of
gas chromatography-mass spectrometry to the investigation
of human disease in Montreal, Canada, 1973.
45. Perry, T.L. and Hansen, S., unpublished results.
46. Merritt, C., Jr. and Robertson, D.H., Anal. Chem., 44,
60 (1972).

BIOGRAPHICAL SKETCH
Timothy Armand Roy was born on 16 April, 1947, in
Rochester, New York. He graduated from East High School in
June 1965. In 1969 he received his B.A. degree in Chemistry
from the State University College of New York at Fredonia
and immediately began graduate work at that institute. After
one year of study he was drafted into the U.S. Army where he
served as a military policeman in Vietnam during most of his
16 month tour. Upon discharge from the service, he re-entered
graduate school and completed the requirements for the M.S.
degree in Chemistry in September 1973. In January 1974 he
began graduate studies in the Department of Pharmaceutical
Chemistry, College of Pharmacy, University of Florida. During
his graduate studies he authored or co-authored four scientific
publications in the areas of organic synthesis and pharma¬
ceutical analysis. He is a member of the American Chemical
Society, the Florida and New York chapters of the American
Chemical Society, and the Student American Pharmaceutical
Association. Upon receiving his doctorate he expects to seek
employment in the pharmaceutical industry as an analytical
chemist.
143

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Stephe#i G. ScHulman, Chairman
Assoc/Cate Professor of
Pharmaceutical Chemistry
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
/
—
y
ft 'ZerC'JU
J.A. Zoltewicz
Professor of Chemistry
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
'jULfTlA^
B.S. Andresen
Assistant Professor of
Pharmaceutical Chemistry
This dissertation was submitted to the Graduate Faculty of
the College of Pharmacy and to the Graduate Council, and was
accepted as partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
December, 1976
Dean, Graduate School