 

V
m :r.
n. , ‘ ' "ET-i...
‘K‘J'A 51': , .2
“—30“. ‘ .x._1o~
waxw‘; _
'xkxj.‘7§ E‘PLI‘sz‘.‘. -L

4 .33.
.551“ $3
E15231)

3-3733“
‘ ‘ .

E331} * .3
32%“.

. f, a
‘Lrsi;£§,;w--'
' ' I

V n

4 .3
'4 5“" . '
mum. vdm‘
:2'N::;;ra~"
; «410 Alf-1,-
1

.- .
gﬁ’a-sr',
r3315 1'
{.72.

. .u
a ‘7”...
.41". {.2’. .
.... 1 ,.

-u
1.?
"mu”
MW.
07
”I
. .

- 7...;1. .n-zJ
Jr: 1:7}2‘3.
mam/'6,“

7r

 

IIIIIIIIIHIIIIZIILHIIIIHIJOIHIIHHUHIII IIIIIIIIIIIIIIIII

914 6022

This is to certify that the

dissertation entitled

Further Development and Characterization of the
Desorption Ionization Technique, K IDS,
and Investigation of the Mechanisms of Fragmentation
and Ionization of Desorption Ionization Techniques

presented by

Karen Jean Light

has been accepted towards fulﬁllment
of the requirements for

Ph.D. Chemistry

 

 

degree in

///z;

Major professor

Date 11/55/70

MSUL: an Afﬁrmau'w' Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY
‘Mlchtgan State
, University

 

 

U-MQUW vvvvvvv

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

\I
r“

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution
cMma-DJ

 

w—

 

FURTHER DEVELOPMENT AND CHARACTERIZATION OF THE
DESORPTION IONIZATION TECHNIQUE, K+IDS,
AND INVESTIGATION OF THE MECHANISMS OF FRAGMENTATION
AND IONIZATION OF DESORPTION IONIZATION TECHNIQUES

By

Karen Jean Light

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemistry

1990

 

ABSTRACT

FURTHER DEVELOPMENT AND CHARACTERIZATION OF THE
DESORPTION IONIZATION TECHNIQUE, K+IDS, AND INVESTIGATION OF
THE MECHANISMS OF FRAGMENTATION AND IONIZATION OF
DESORPTION IONIZATION TECHNIQUES

By

Karen Jean Light

Further development and characterization of K+ ionization of desorbed
species, K+IDS, is performed to gain a better understanding of this desorption
ionization technique. K+IDS utilizes a thermionic emitter material as a source
for gas phase K+ ions and rapid heating of thermally labile compounds to
promote desorption of intact analyte molecules and neutral thermal
degradation products. K"~ adduct ions are formed when K+ emission occurs
simultaneously with sample desorption. A two-filament probe tip design,
which improves the overlap of K+ emission with sample desorption, is
evaluated. The K+ emitter surface is resistively heated to temperatures in
excess of 900°C and the sample filament wire is radiatively heated to
temperatures between loo-200°C. Temperature studies of the sample ﬁlament
wire show the effects of the heating rate and the ﬁnal temperatures reached
by this sample ﬁlament wire on the K+IDS mass spectra. The applicability of
the K+IDS technique is extended to the analysis of cardiac glycosides with both
molecular weight and structural information obtainable from the K+IDS mass
spectra. A new nomenclature scheme is proposed that allows for the accurate
identification of neutral thermal degradation products and fragment ions
produced by desorption ionization techniques, such as K+IDS. This naming
scheme allows the designation of hydrogen shifts, multiple cleavages. and

adduct formation observed in the analysis of these cardiac glycosides.

 

 

Adaptation of the 10108 technique to a double-focusing mass
spectrometer is explored to investigate the mass limit of the K+IDS technique.
Modifications to the pre-existing direct chemical ionization probe and power
supply of the JEOL HX-llO double-focusing mass spectrometer are made to
perform the K+IDS experiment. Difﬁculties arise in achieving a fast scan rate
that can detect the rapid changes in the K+IDS mass spectra with the double-
focusing instrument.

The FAB mass spectrometric analysis of the cardiac glycosides is
performed to gain a better understanding of this desorption ionization
technique compared to the K+IDS technique. A detailed mechanistic study of
the FAB mass spectrum of digoxin is described. All information obtained by a
double-focusing mass spectrometer, including tandem mass spectrometry and
high resolution peak matching. is used to deduce the structural formulas of the
fragment ions observed in the FAB mass spectrum and the possible

fragmentation pathways.

This thesis is dedicated to my entire family and especially my
mother and father whose never-ending love, support and
encouragement inspired me to achieve.

iv

 

ACKNOWLEDGMENTS

I owe much to my mentor John Allison whose support and guidance was
always strongest when I needed it most. I really admire and appreciate his
dedication to his graduate students and to the development of science. I would

also like to thank Dan Bombick and Dan Kassel for introducing me to “K+IDS”
and their leadership and support in teaching me the “art” of graduate
research.

Then there were the “three stooges” who always seemed to be there
with a geod laugh or reassuring smile to keep me going through good and bad.
Thanks “big brother” for all of our good conversations, hospitality trips, and
the good advice (although the last one you were wrong on....I just had to be
patient!). To Gary, Kurt and Jason, who brought new life - and partying - to
our group, thanks for all the fun times. Someday, Jason, maybe I’ll give you
another chance to spike one in my face!

The grad school experience would not have been complete without the
“flamingo crew”. I’ll never forget all the good times we shared - the
wallpaper party, late night swims, and of course the waterbed to name a few.
May our blenders always churn pink. The annual winter ski trips were such a
great getaway from all of the horrors of grad school such as the seminars and
2nd year orals. Someday Sabo, maybe the rest of us will figure out just what
“we touched on” that evening in wintry north! A special thanks to Kris for
being such a good friend and to Judy for being such a good roommate and
friend throughout the five-year graduate school experience.

Iowe many thanks to all those hard-working “specialists“ of the Mass
Spectrometry Facility. I will really miss all of you! Thanks Doug for all your
conﬁdence in me, Mel for all your assistance in keeping the computer world
running, Melinda for all the extras you do, and Bev for all the hard work you
do to make us look good! And then there’s Mike...who had an uncanny way of
brightening up the day. I have no doubt that if it weren’t for
your...patience?...the HP’s would have been history long ago! Thanks for all
the extra help in keeping them running. I’m really going to miss being able
to bug you when something needs fixing. But I guess Jason can fill in that
role! Thanks for all the electronics lessons you gave while fixing the
instruments.

And last, but most importantly, I would like to thank Jon for all the
confidence, self-assurance, and joy he has brought to my life through all of
his love and support. Your ability to make me laugh in the face of despair
helped me through many stressful times. You are the best friend I could ever
have and I love you for it.

TABLE OF CONTENTS

List of tables

List of

Chapter

1.
2.

Chapter

1.
2.

ﬁgures

1. Introduction
Desorption ionization techniques
K+IDS Theory
A. Production of gas-phase K+ ions
B. Desorption of thermally labile compounds
C Experimental considerations for the K+IDS technique

2. Characterization of the K+IDS technique
Introduction
Characterization and modiﬁcation of the K+IDS probe
A. Original single filament probe design
B. Dual filament probe design
Temperature studies of the K+IDS two-ﬁlament probe design
A. Determining surface temperatures in a vacuum
chamber
1. Experimental design
2. Results
B. Correlation of K+ IDS mass spectra with sample
surface temperature
C Real time determination of sample desorption
temperatures
1. Experimental design
2. Results
D. Activation energy determinations based on K+IDS
analyses

. Summary

3. Applications of the K+IDS technique
Introduction

K+IDS mass spectrometric analysis of cardiac glycosides
Utility of the K+IDS technique for mixture analysis
Final comments

4. Li+IDS

. Introduction

Experimental considerations for producing a Li+ thermionic
emitter

Li+IDS mass spectra
Final comments

vi

viii

46
47

56

67

69
69

69
74
8O

81
81

82

84
93

 

Ci

Chapter 5. Implementation of the K+IDS technique on a
double-focusing mass spectrometer '
1. Introduction
2. Experimental design
A. Modiﬁcation of the DCI power supply
B. Conversion of the K+IDS power supply to control the
DCI/K+IDS probe
3. Results
4. Final Comments

Chapter 6. FAB mass spectrometric analysis of cardiac glycosides
1. Introduction
2. Experimental section
3. Results
A. Positive FAB mass spectra of some cardiac glycosides
B. Negative FAB mass spectra
4. Summary

Appendix A.

Figures A1-A25. Plots of K+ adduct ion abundance vs. sample
ﬁlament wire temperature for a variety of compounds.

Figures A26-A31. Activation energy determinations based on plots
of K+ adduct ion abundance vs. III.

Figures A32-37. Activation energy determinations based on plots
of ln(ln(Ao/At)) vs. 1/T.

Figures A38-A42. Sample Li+1DS mass spectra.

Appendix B.

“Mass Spectrometric Analysis of Cardiac Glycosides by the
Desorption/Ionization Technique Potassium Ion Ionization of
Desorbed Species.”

Appendix C.

“Mechanistic Considerations of the Protonation and Fragmentation
of Highly Functionalized Molecules in FAB: High Resolution MS and
MS/MS Analysis of the Ions Famed by Fast Atom Bombardment of
Digoxin and Related Cardiac Glycosides.”

References

94
94

95
97

98

99
105

106
106
106
107
107
115
121

123

126

151

157

163

168

177

196

 

Table 4.1

Table 5.1

Table 6.1

Table 6.2

LIST OF TABLES

The m/z values of the Na+ and Li+ adduct ions of the

oligomers of PPG 725.

Values of m/z for the K+ adduct ions of the oligomers of
polyethylene glycol 1000.

Fragment ion assignments for digoxin, digitoxin, gitoxin,
and acetyldigitoxin.

Fragment ions observed in the positive and negative FAB

mass spectra of ouabain and possible assignments based
on the LKA nomenclature (see Appendix B).

viii

88

103

110

117

 

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

1.1

1.2

1.3

2.1

2.2

2.3

2.4

2.5

2.6

2.7
2.8

2.9

2.10

2.11

2.12

LIST OF FIGURES

Schematic of the ion source of the quadrupole mass
spectrometer and the experimental design of the K+IDS

~ experiment.

Original K+IDS probe tip design.

Schematic diagram of the electronics of the K+IDS power
supply designed and constructed by Marty Rabb.

Example of the 1.2-elimination mechanism for the thermal
degradation of sucrose.

Arrhenius plot of the log k vs. 1/T for desorption and thermal
degradation processes. The slope of the line is a function of
the activation energy.

Structure of tristearin and the thermal degradations observed
as K+ adduct ions in the K+IDS mass spectra.

The abundances of the K+ adduct ions of the intact molecule
(m/z 929) and of the thermal degradation products of
tristearin versus scan number.

K+IDS mass spectra of tristearin averaged over scans early
and late in the same analysis.

TIC chromatograms of the sample desorption of a fatty acid

mix obtained by El overlapped with the K+ emission for the
original probe tip design and the new two-ﬁlament probe tip
design.

Two-filament K+IDS probe tip design.

Experimental setup for the temperature measurements of
the K+IDS probe performed inside a glass vacuum chamber.

K+ Emitter bead and sample ﬁlament wire temperature versus
time when 3.0A is applied to the emitter.

Temperature of the emitter surface and sample filament wire
vs. current applied to the emitter.

Heating rate of the sample ﬁlament wire vs. emitter current.

Temperature of the emitter surface, emitter ceramic post,
and the sample filament wire vs. emitter current.

ix

 

10

14

17

19

20

21

23

25

28

31

32

33

35

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

2.13

2.14

2.15

2.16

2.17

2.18

2.19

2.20

2.21

2.22

2.23

2.24

2.25

2.26

X

Plot of the K+IDS mass spectra of tristearin when 2.0A

and 2.5A are applied to the K1‘ emitter and the sample is
on the sample ﬁlament wire only.

‘Plot of the K+IDS mass spectra of tristearin when 3.0A

and 3.5A are applied to the K+ emitter and the sample is
on the sample filament wire only.

Plot of the K+IDS mass spectra of tristearin when 2.0A
and 3.0A are applied to the K+ emitter. In this case, the

thermal mass of the K+ emitter was reduced by decreasing
the length of the ceramic support rod.

K+IDS mass spectrum of sucrose with sample on the support
wire only. 1 = 2.5A Sample desorption was ﬁrst observed at

scan 30. T(sample wire) = 150°C.

K+IDS mass spectrum of hexaglycine with the sample on the
support wire only. 1 = 3.0A Sample desorption was ﬁrst

observed at scan 22. T(sample wire) = 150°C.

Structure of hexaglycine and the thermal degradations
observed as K+ adduct ions.

Abundance of the [M]K+ ion of sucrose vs. Temperature (K)
of the sample ﬁlament wire.

Abundance of the [M]K+ ion and K+ adduct ions of thermal
degradation products of melezitose vs. temperature (K).

Structure of melezitose and the thermal degradation
products that are observed as K+ adducts.

Abundance of the K+ adduct ions of digoxin vs. sample
filament wire temperature (K).

Abundance of the [M]K+ ions of sucrose and melezitose vs.
temperature of the sample ﬁlament wire.

Abundance of the [M]K+ ions of palmitic acid, sucrose, and
melezitose vs. temperature (K) of the sample filament wire.

Abundance of the [M]K+ ion of sucrose vs. 1/T (K).
Ea calculated from the slope of the line of the first half of
the experiment = 13 kcal/mol.

Abundance of the [M]K+ ion of palmitic acid vs. l/T (K).
Ea calculated from the slow for NT > 2.7 = 23 kcal/mol.

37

38

40

43

44

45

48

50

51

53

54

55

57

59

Figure

Figure

Figure

Figure

Figure

Figure
Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure
Figure
Figure

Figure

Figure

2.27

2.28

2.29

2.30

3.1
3.2

3.3

3.4

3.5

3.6

4.1

4.2

4.3

4.4

4.5

4.6

5.1

5.2

SUCTOSC.

xi

Arrhenius plot of ln(ln(Ao/Am vs. VT for palmitic acid.
Ea calculated from the slope = 12 kcal/mol

Arrhenius plot of ln(ln(Ao/At» vs. 1/T for the [M]K+ ion of
Ba calculated = 28 kcal/mol.

Structure of methionine-enkephalin and the thermal

degradations observed as K+ adduct ions.

Arrhenius plot of ln(ln(Ao/At)) vs. III for three fragment
K+ adduct ions of met-enkephalin. Ea(259) = 27 kcal/mol,
Ea(392) = 37 kcal/mol, Ea(449) = 52 kcal/mol.

Published fast atom bombardment mass spectrum of digitonin.

Structure of digitonin.
K+IDS mass spectrum of digoxin.

Mass chromatograms of the [M]Na+ ions of the eight
individual components present in the mixture.

Plot of the Na+IDS mass spectra vs. scan number for the
analysis of the eight component mixture.

The averaged Na+IDS mass spectrum of the eight component
mixture containing organic acids, fatty acids, and steroids.

Na+IDS mass spectrum of PPG 725 averaged over the entire
experiment.

Li+1DS mass spectrum of PPG 725 averaged over the entire
experiment.

Li+IDS
single

mass spectra of PPG 725 vs. scan number for a
experiment.

Li+1DS mass spectrum of palmitic acid.

Li+IDS mass spectrum of cholesterol.

Li+IDS mass spectrum of digitoxigenin.

Schematic of the modiﬁed JEOL HX-llO DCI probe tip for
use with the K+IDS experiment.

Parameter ﬁle for data collection on the JEOL PIX-110 mass
spectrometer with the modiﬁed DCI/K+1DS probe and
K+IDS power supply.

62

64

65

66

71

72

73

76

78

79

85

86

87

89

91

92

95

96

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

5.3

5.4

5.5

5.6

6.1

6.2

6.3

6.4

6.5

6.6

6.7
6.8

6.9

xii

Temporal dependence of the K+IDS mass spectra of PEG 1000
obtained with the JEOL HX-llO double-focusing mass
spectrometer.

.Mass chromatograms of the TIC and selected K+ adduct ions

produced by the K+IDS analysis of PEG 1000.

The averaged K+1DS mass spectrum of PEG 1000 from scans
15-40 obtained on the JEOL HX-llO double—focusing mass
spectrometer.

K+IDS mass spectrum of digoxin obtained on the JEOL
HX-llO double-focusing mass spectrometer.

Structure of digoxin and other similar cardiac glycosides
studied. The fragmentations observed in the FAB mass
spectrum of digoxin are labeled.

Positive FAB mass spectrum of digoxin. The glycerol cluster
ions are denoted by *.

Positive FAB mass spectrum of gitoxin. The glycerol cluster
ions are denoted by *.

Positive FAB mass spectrum of digitoxin. The glycerol
cluster ions are denoted by *.

Positive FAB mass spectrum of acetyldigitoxin. The glycerol
cluster ions are denoted by *.

Negative FAB mass spectrum of digoxin. The glycerol cluster
ions are denoted by *.

Structure of ouabain.

Positive FAB mass spectrum of ouabain. The glycerol cluster

ions are denoted by *.

Negative FAB mass spectrum of ouabain. The glycerol cluster

ions are denoted by *.

100

101

102

104

108

109

111

113

114

116

118

119

120

CHAPTER 1. INTRODUCTION

1. DESORPT ION IONIZATION TECHNIQUES

Mass spectrometry, as an analytical tool, has been traditionally used for
the analysis of volatile analytes or compounds that are easily derivatized to
some volatile species. The common ionization technique for such volatile
samples is electron ionization (El). This ionization technique is well
characterized and the mechanisms of ion formation and fragmentation are
well documented.1 Another ionization technique that is commonly used for
volatile species is chemical ionization (CI)2, which is considered to be a softer
ionization technique than EI, and provides information about the sample that
complements E1 results. CI utilizes gas phase reagent ions to ionize the sample
in the gas phase via ion/molecule reactions. Adduct ions are often produced
with this ionization method that contain the intact molecular species. Little
fragmentation is observed with C1 relative to El, hence CI is considered as a
soft ionization process. Both EI and CI provide a means of ionizing gas phase
species, but do not provide a means of transferring the sample from the
condensed phase to the gas phase. This greatly limits the applicability of these
techniques to volatile species.

Desorption ionization (DI) techniques circumvent this sample volatility
requirement by providing a means of transporting thermally labile
compounds from the condensed phase (either solid or liquid) into the gas
phase and by providing a means of ionizing the sample. The development of
these techniques has greatly expanded the realm of mass spectrometry,
especially in the field of biochemistry where most of the compounds of
interest are large multifunctional molecules which are thermally labile.
There are many desorption ionization techniques currently in use with
continual development of new techniques. Field desorption (FD) was one of
the first ionization techniques developed for nonvolatile compounds.3 More
recently, plasma desorption (PD)4, laser desorption (LD)5, fast atom
bombardment (FAB)°, and liquid secondary ionization mass spectrometry
(LSIMS)7 have been developed as desorption ionization techniques for
thermally labile compounds. These DI techniques utilize particle
bombardment of the sample to transfer energy to the sample to form gas phase

ionic species. For example, LD utilizes photons to bombard the sample, FAB

1

2

utilizes fast heavy atoms (Xe or Ar) and LSIMS utilizes ions (Cs+) for
bombardment. The overall idea is the same, but the techniques vary in the
amount of energy imparted and the rate at which the energy is deposited into
the sample. This difference in energy deposition, coupled to chemical
differences, leads to differences in mechanisms of ion formation and
fragmentation, and differences in the mass spectra obtained. Also direct
insertion techniques such as DEI8 and DC]9 have been developed which
combine the traditional EI and CI techniques with methods of transferring
thermally labile compounds into the gas phase. Ionization techniques also
have been developed to combine liquid chromatography with mass
spectrometry (LC/MS). Thermospraylo and continuous flow FAB11 are among
these LC/MS techniques that are gaining much attention. The difﬁculty in
this endeavour is dealing with the large volume of solvent while maintaining
the vacuum system for mass spectrometric analysis.

There are many different DI techniques currently in use, all of which
have their strengths and weaknesses. None of these techniques is all
inclusive or solves all of the mass spectrometric needs for the analysis of
thermally labile compounds. FAB and LSIMS, perhaps the most widely used DI
techniques, require the use of a liquid matrix, such as glycerol or nitrobenzyl
alcohol, to dissolve the sample, control the sampling rate, and aid in ion
formation. Due to the presence of this matrix, many ions originate from the
matrix which complicate the mass Spectra and often interfere with the
observation of sample-related ions. This is especially true in the low-mass
range where ions from the matrix often dominate the mass spectra and mask
analyte ions present. Therefore, mass spectrometry/mass spectrometry
(MS/MS) is often necessary to obtain low mass information without
interference from the matrix.12 This technique allows for the parent ion to be
selected and only those fragments produced from the selected parent ion are
observed in the MS/MS spectrum. LD mass spectrometry has proven to be a
very useful technique especially for large biomolecules where much energy
is needed to ionize and desorb the intact species with little fragmentation.13
This technique, however, requires very SOphisticated, expensive
instrumentation, and much care and experience to keep both the laser and the
mass spectrometer operational simultaneously. Because there is no ideal DI

technique currently available, this field of mass spectrometry is constantly

developing.

 

3

The main analytical goal of these DI techniques is to generate gas-phase
ions that provide molecular weight and structural information pertaining to
the compound of interest. Cationization is a processithat may assist in the mass
spectrometric determination of molecular weight. Cationization typically
produces [M]A“' ions where A is most often an alkali metal cation such as Na+
or K+. This adduct ion formation has been utilized with LD, FAB, FD, and LSIMS
to aid in molecular weight determination. More recently, cationization has
been utilized as a means of obtaining additional structural information. Mallis
and Russel reported differences in fragmentation of [M]H+ and [M]Na+ ions of
small peptides such as hippuryl-L-histidyl-L-leucine (HHL) when collisionally
activated.14 The Na+ adduct ions were produced by adding 1 ug of NaCl to the
sample. They suggest that the highly polar functional groups commonly
present in the thermally labile compounds studied, such as peptides and
saccharides, have different H+ and N“ ion affinites (where A is an alkali metal
such as Li, Na, or K). Thus, the binding sites of H+ and A+ may be different and
induce different fragmentations with the aid of collisionally activated
dissociation. Many other research groups have reported similar advantages of
performing collisionally induced dissociation on alkali adduct ions of
oligosaccharides” and peptides16 for increased structural information over
protonated species. The addition of alkali salts to the sample or matrix is the
most common way of producing adduct ions with these DI techniques. The
other way of obtaining alkali ions is by their presence as impurities in the
sample.

Cationization is not a new concept. Rollgen and coworkers were among
the ﬁrst to recognize the utility of forming adduct ions with alkali metal
cations.” Their early studies involved formation of Li+ adducts with volatile
and thermally labile compounds. They applied electric fields to solid Lil to
produce a large flux of Li+ ions in the gas phase. They also studied the
mechanisms of Na+ and K+ adduct ion formation formed by laser desorption.18
They determined that cationization can be a surface reaction, when alkali salts
are in contact with the sample, or adduct ion formation can be a gas phase
addition process. To prove this gas phase adduct ion process, Rollgen et al.
designed a two-ﬁlament probe to spatially separate the alkali ion source (alkali
salt) from the sample holder.19 This design still produced adduct ion
formation, suggesting a gas-phase mechanism of alkali ion attachment. Van

der Peyl, Haverkamp, and Kistemaker supported this gas phase mechanism of

4

adduct ion formation with a series of studies performed by LD mass
spectrometry.20 The experimental design insured that the K+ emitter
(obtained commercially) was spatially separated from the sample housing with
no line-of-sight between the two components. They observed K1” attachment to
an intact gas-phase sucrose molecule when there was no possibility of the
adduct being formed by surface reactions.

Most of the developmental work with DI techniques is in creating
alternative methods and trying to increase the mass range rather than in
gaining a better understanding of the mechanisms of fragmentation and ion
formation involved. The ionization methods which are gaining the most
attention and which may revolutionize the role of mass spectrometry in
biological research are matrix-assisted laser desorption and electrospray
ionization (ESI).21 It has been demonstrated that both LD22 and E8123 are
capable of extending the ionization and molecular weight determinations to
compounds with molecular weights greater than 100,000 daltons. Electrospray
ionization was ﬁrst developed by Dole and coworkers to produce gas-phase
macroions that were detected in a Faraday cage with the molecular weight
determined by st0pping potentials.24 Yamashita and Fenn were among the
ﬁrst researchers to combine 1381 with mass spectrometry (ESI-MS) in 1984.25
By 1988, Fenn and coworkers had gained much attention and interest in their
new ionization method with the report of intact multiply charged ions of
proteins up to a molecular weight of 40,000 daltons.2° Since then, much
interest and research has been devoted to the development of this technique
and increasing the mass limit of ESI-MS. Electrospray ion formation is a two-
step process where highly charged droplets are dispersed at pressures
approaching atmospheric pressure, followed by droplet evaporation. Briefly,
a high electric ﬁeld is applied to a small ﬂow of liquid from a capillary tube.
This electric ﬁeld results in the formation of highly charged liquid droplets
due to the disruption of the liquid surface. The solvent evaporates leaving
behind multiply charged analyte molecules. A more detailed explanation of
this technique can be found elsewhere.21

The desorption ionization technique examined in this dissertation
differs from many of the DI techniques discussed above in that the sample is
not bombarded by particles to promote vaporization. but instead is heated
radiatively (or resistively) to induce desorption of the sample. This D1

technique is called K+IDS, potassium ion ionization of desorbed species, and

5

was developed at Michigan State University as an economical and simple way
of analyzing thermally labile analytes with a quadrupole mass spectrometer.27
This technique utilizes a thermionic emitter, a material that produces a large
ﬂux of ions when heated, and rapid heating of a thermally labile compound to
promote desorption of both the intact molecule and neutral products of
thermal degradation. The thermionic emitter used produces a large flux of K+
ions in the gas phase. These K+ ions attach to the neutral gas-phase species
desorbed from the sample to produce K+ adducts. No matrix is required with
this technique. Therefore, all of the ions observed are K+ adducts
representative of the sample. The mass spectra produced are easily
interpreted based on one fragmentation mechanism (1,2-elimination).
Implementation of this technique to a quadrupole mass spectrometer that is
equipped with a direct insertion probe inlet is inexpensive and simple. The
further development and characterization of this technique is one of the main

goals of. this research.

2. K+IDS THEORY

There are two concepts that are important to the K+IDS technique. One
is the use of thermionic emitters to form gas-phase K+ ions that are used as
reagent ions in a way similar to those used in chemical ionization. The other
concept is the use of rapid heating of a thermally labile compound to promote
vaporization of the sample with little thermal degradation. When these two
processes are combined and occur simultaneously, a K+IDS mass spectrum can
be obtained. The K“ ions attach to neutral gas-phase species, possibly in a
three-body collision process, and are observed as K" adducts. This K“
attachment induces little or no fragmentation and merely samples the neutrals

present in the gas phase which are produced by the heating of the sample.
A. PRODUCTION OF GAS-PHASE K+ IONS

The most important aspect of the K+IDS technique is the production of a
large ﬂux of K+ ions in the gas phase. This is accomplished with the use of
thermionic emitter materials that are patterned after the aluminosilicate
mixtures described by Blewett and Jones in 1936.28 They determined that the
mixture A1203:Li20:28102 was the best mixture for producing a large ﬂux of Li+

6

ions. The same mixture doped with K20 instead of LizO has proven to be a good
source of K+ ions when the mixture is heated to temperatures between 800 and
1200°C. Extensive discussions of how these materials are capable of emitting
metal ions can been found elsewhere.29 The emission of K+ ions is dependent
primarily on the work function of the surface and the ionization energy of
the potassium atom. The work function of the surface is essentially a measure
of the electron afﬁnity of the surface. This term is explained and discussed in
detail by Bombick.29

B. DESORPT ION OF THERMALLY LABILE COMPOUNDS

The other concept important to the K+IDS technique is the transfer of
the thermally labile compounds into the gas phase with minimal thermal
degradation of the sample. The use of rapid heating to promote desorption of
the intact molecule over competitive decomposition reactions of thermally
labile compounds was described by Bcuhler and coworkers for the mass
spectrometric analysis of underivatized peptides.30 The enhancement of the
volatility of a complex thermally labile species due to rapid heating of the
sample is based on a kinetic analysis of the competitive decomposition and
vaporization processes. This will be explored in more detail later. Low
temperatures and slow heating rates will promote more decomposition than
high temperatures and fast heating rates, which favor desorption of the intact
molecule. If this relationship holds, then the temperature variables and
heating rates may provide a means of adjusting and controlling the types of
information obtained. The ideal case would be to use heating rates and
experimental conditions that allowed for the detection of degradation products
early in the analysis followed by desorption of the intact species all within one
K+IDS mass spectrometric analysis. Therefore, both structural and molecular
weight information would be obtained and the relationship between the
temperature of the surface and the appearance of different adduct ions could
be used to deduce information about the structure of the analyte and the
relative strengths of the bonds broken to form the various degradation

products. This goal is pursued in Chapter 2.

7
C. EXPERIMENTAL CONSIDERATIONS FOR THE K+IDS TECHNIQUE.

The K+IDS technique is performed at Michigan State University on an
HP 5985 GC/MS/DS quadrupole mass spectrometer that has a mass range of 10-
1000 daltons. A schematic of the ion source and the K+IDS apparatus are shown
in Figure 1.1. A modiﬁed direct insertion probe is used to insert the K+
aluminosilicate bead into the center of the ion source. The tip of this modified
probe is shown in Figure 1.2. The K+ bead ﬁlament is made by threading 0.007
inch rhenium wire through a two-holed ceramic (3/64" outer diameter, 0.010"
inner diameter) leaving a 1mm 100p at the top. This tip is spot-welded to nickel
wire leads that are fed through the K+IDS probe. The aluminosilicate mixture
is then applied to the tip of this ﬁlament loop to make the K+ bead in one of two
ways. The dry aluminosilicate powder can be melted on a piece of platinum
with an oxygen/acetylene torch. The filament loop is dipped into the molten
bead material to capture some of the glass onto the wire. Another way to make
a K" head is by ﬁrst making a slurry of the aluminosilicate powder in acetone.
A Pasteur pipette is used to transfer this slurry to the wire 100p. Then the tip is
inserted into a bunsen burner flame to evaporate the solvent and slightly
harden the bead surface. The ideal bead, made with either technique, has a
1.1111 coating of the aluminosilicate mixture completely covering the bare wire.
A bead that is too large takes longer to produce K+ ions once current is applied
to the bead. On the other hand, if the wire is not completely covered, the
spectra seem to contain much more noise, possibly due to ionization of gas-
phase molecules on the bare, hot wire surface. Once the bead is made, it is
conditioned inside the ion source of the mass spectrometer by slowly ramping
the current applied to the bead. A current ramp of approximately lA/min in
0.25A steps from GA to 3.8A is a good rate for conditioning. A current source is
used to provide 0-4A to the K+ bead in order to heat the thermionic emitter
material. Also, a low voltage power supply is used to provide a bias voltage to
the tip of the K4“ bead of 0 to +10 V with respect to the Cl volume (repeller). A
power supply was designed that satisfied both of these requirements at
Michigan State University by M. Rabb. A schematic of the electronic design of
this power supply is provided in Figure 1.3. The K+IDS experiment is generally
performed with a current of 2.2-3.5 A and a bias voltage of +2 to +4 V. More

details of the technique are provided elsewhere.29

aluminosilicate
bead covering
Rhenium wire

I / [Rhenium wire (0.007“)

  
 
   

T HP -
0° odmsglas two holed ceramic probe "9 spot

u D e (0050.0 0) welded to probe
Mass Filter ° ' ' assembly

 

 

 

Solids
Probe Power Supply to

—
l —

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—'-1
0
23 g Inlet Heat Bead
E E L / 1| (0:04:11
'0 E _ I 'll 7
o r I
c— U .-
8 o L_:]
I 'o 1 11
C
O - IL+
'r
HP5985 Bias an Bead Relative to
Ian Source CI Volume

(CI Volume)

Figure 1.1 Schematic of the ion source of the quadrupole mass spectrometer
and the experimental design of the K+IDS experiment.

game a: Sea moi .265 m; 9:5

 

 

 

 

 

 

 

 

 

 

 

. ......................... i ZQQDm i >_QQDw
” . .................... Egon 0.585
ml zoom mecca
\. iiiii I ll....--/.
n/ 11111 I 111-- \.

 

 

 

 

  

”.4/ 2292

+
9
.ﬁgktL‘DPM

 

 

 

 

 

 

 

1.).”

 

Q---

 

81

A

9,

3
nan-o..."

$34“
<14

 

 

 

 

 

 

“ 'b—z
(Unit. t) E” m .
2 6 1

   

(unit 2)

 

 

 

 

 

+ 14v

N638!)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C.°.E
W -t 10:» 5%
t s“ -t low 5%
n),
vmeasure 5
T ﬁrm Gilt
:Fﬂu
E 1%
rs tz—zoo , .
a!) Power 6
3H1? <——— S l ,
1“" 3%v i opt; ﬂ:
1 7 - _T
Ref
V019
In

 

1:3; £521

 

 

 

 

 

l'rzss-u leDGE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 +11”!

 

 

 

:HEEJ

 

 

Figure 1.3 Schematic diagram of the electronics for the K+IDS
power supply designed and constructed by Marty Rabb.

 

11

Calibration of the instrument, particularly the mass axis, is performed
in the positive EI mode in the traditional way for GC/MS and/or DIP analyses.
PFI‘BA is introduced into the ion source at a pressure of 2-5 x 10'6 torr with the
ion source temperature at ISO-200°C. The voltages of the focusing lenses and
the other variables are adjusted for Optimum E1 results. The K+IDS signals are
too short-lived to allow for good calibration of the instrument with K+ adduct
ions. Once the instrument has been calibrated (especially the mass axis) in the
El mode, the source is cooled down to room temperature for the K+IDS
experiment. The El ﬁlament is turned off for the K+IDS experiments by setting
the emission current and the electron energy to their lowest possible values.
The K+ emission is monitored with the override tuning program and the
focusing lenses and other source variables affecting the ion signal are
adjusted for optimum K+ transmission, introducing only minimal changes in
the parameters from the original El tuning ﬁle.

The original K+IDS experiment or "Classic K+IDS", which used a single
K1” bead ﬁlament, is performed in the following way. (Details of how the
technique has changed and how it is currently done will be presented in
Chapter 2.) The sample is deposited onto the K+ bead as either a solution or a
slurry in some organic solvent. Acetone is the preferred solvent as it
evaporates quickly, however, methanol also is used frequently as many of the
samples will not dissolve in acetone. Water as a solvent is avoided whenever
possible as it is difﬁcult to evaporate completely and often produces spectra
which contain much more noise than those from samples with the other
solvents. This phenomenon is not completely understood, but serves as an
observation and guideline. Once the sample is deposited onto the K" bead and
all of the solvent has evaporated, the probe is inserted into the ion source of
the mass spectrometer. Data collection is started immediately and then a preset
current of approximately 3.0A and a preset bias voltage of approximately +3V
is selected. The preset current allows for rapid heating of the emitter. This is
in contrast to the temperature (current) ramping procedure commonly used
with many other techniques including DCI. This K+IDS procedure is the
original mode of operation with a single ﬁlament probe tip. Details on the
modiﬁcations made to the procedure and probe tip design that are currently

employed will be discussed in Chapter 2.

CHAPTER 2. CHARACTERIZATION OF THE K+IDS TECHNIQUE

1. INTRODUCTION

Initial development and characterization of the K+IDS technique by
Bombick29 showed the utility of this technique and the wide applicability to a
variety of thermally labile compounds including saccharides, peptides,
polymers and steroids. Bombick explored the mechanisms of adduct formation
and fragmentation characteristics of the K+IDS technique. The addition of a
neutral collision gas produced an increase in adduct formation for small
analyte molecules.29 Under normal K+IDS experimental conditions, Bombick
determined that the lower limit of detection for polyphenylether was
approximately 500 ng at a signal-to-noisc ratio of ﬁve. When N2 was
introduced into the ion source as a neutral collision gas to a source pressure of
approximately one torr, the detection limit for polyphenylether was lowered to
approximately 250 ng at a signal-to-noise ratio of ﬁve.29 This increase in
adduct formation with the presence of a collision gas conﬁrms that K+ adduct
formation is a three-body collision process in the gas phase, at least for small
analyte molecules. The three-body collision process may not be the only way
of stabilizing the excited K" adduct ion. Woodin and Beauchamp proved that
Li+ attachment to an analyte molecule produced an excited complex that could
be stabilized by the emission of an infrared photon, instead of by a three-body
collision.31 These bimolecular infrared radiative association reactions are
assumed to be the dominant stabilization mechanism at very low pressures
where the time between collisions exceeds 100 milliseconds. This photon
emission process has gained p0pularity as a possible mechanism for
interstellar chemistry, but is not likely to be responsible for the stabilization
of the K" adduct ions formed in K+IDS. Suppose a thermally labile analyte, A, is

desorbed intact into the gas phase as shown in reaction 2.1. A K“ ion may

As —’ Ag 2.1
Ag +K+ —9 AK” 2.2
AK‘H‘ -) Ag +K+ 2.3

12

13

attach to this gas-phase neutral with the binding energy of the K+ ion to the
analyte of approximately 20 kcal/mol.32 This leads to an activated complex,
AK‘”, as shown in reaction 2.2. This excess energy present in AK“ can be
dissipated in several ways. The energy can be used to dissociate the complex
back to A and K1“, reaction 2.3. A collision with a third body or emission of a
photon can eliminate some of this excess energy so that dissociation (reaction
2.3) is no longer possible and the adduct is stabilized. If ions only spend
approximately 10'6 sec in the ion source, the pressure in the source, or in
some localized region, needs to be 10‘2 torr in order to have an average of one
collision/ion in 10'°sec.33 This localized high pressure necessary for the
stabilizing third body collision is likely to occur near the K" emitter where a
large flux of K+ ions and neutral analyte species are present in the gas phase.
Therefore, it is reasonable to assume that the majority of the K" adduct
formations are due to a three-body collision process.

Bombick also suggested that degradation of the compounds analyzed by
K+IDS appear to follow one simple fragmentation mechanism. This mechanism
is a 1.2-elimination mechanism for cleaving a skeletal bond, which produces
two neutral products. An example of this mechanism is shown in Figure 2.1
for the thermal degradation of sucrose. The ions observed in the K+IDS mass
spectrum of sucrose are labeled in Figure 2.1. It is very common for
saccharides to fragment at the glycosidic hands when thermally stressed. This
characteristic fragmentation of saccharides will be further explored and
utilized in Chapter 3. All thermal degradation products observed to date with
the K+IDS technique can be explained by means of this 1.2-elimination
mechanism. The most common fragmentations observed with the K+IDS
technique are cleavage of the glycosidic bonds for compounds containing
saccharides and dehydration for compounds containing hydroxyl groups.

The concept of “rapid heating” of a thermally labile sample to promote
desorption has been studied by Beuhler and coworkers.30 They described two
ways of enhancing the volatility of thermally labile compounds such as
underivatized peptides. One way of achieving this volatility enhancement is to
use relatively inert surfaces for deposition of the samples. One study has been
performed where volatility enhancement has been achieved by depositing
peptides, such as thyrotropin releasing hormone (TRH), onto teflon.34 This
inert surface reduces the energy of bonding of the molecules to the surface

and thus makes it easier to preferentially break these surface-to-molecule

14

 

.3203 .o cosmonaut 35.05 as. .5.
EmEazooE :o:~:.E__o-~.p on. .o 2953 _..u 0.59...

 

 

.: 8F .: «or
:o :0
:oazo o: o: \ :o 0:
m3 £8 + SN nE
+V— L. +¥ .o.
o 0
10:6 10:8

c2555.? . N. _.

:o I o:
o:
zofo . 0)) :o o:
O O .

IONIC IONIC

 

 

bat
dil

cnl
pri

111C

15

bonds with little degradation of the analyte. Other researchers have utilized
different materials for sample supports to achieve the same results of
enhancing thesample volatility. Materials used as sample support surfaces
primarily for DCI techniques include polyimide-coated wires,” various
metals,3° Vespel,37 nitrocellulose,38 and SE-30-coated glass (SE-30 is a dimethyl
siloxane polymer).39 This variable of sample support surface and its effect on
the volatilization of thermally labile compounds has not been pursued to date
with the K+IDS technique. It is an area of future study that may prove
necessary when trying to extend the mass range of this technique and to work
with more thermally fragile compounds.

Beuhler and coworkers also suggest the use of “rapid heating” of a
sample for volatility enhancement.30 This is based on the kinetics of
competitive desorption versus thermal degradation processes. They showed
how the rate at which energy is deposited in a solid sample can enhance
desorption of the intact molecule over thermal degradation of the sample on
the surface. Arrhenius plots can be used to illustrate how kinetic
considerations of vaporization versus degradation can be used to control the
amount of decomposition that occurs upon rapid heating of thermally labile
samples. The rate of vaporization of a neutral fragment is related to the
activation energy for decomposition.30 This activation energy is lower than
the activation energy for evaporation of the intact parent molecule. The
Arrhenius equation (Equation 2.4) provides a relationship between the rate of

product formation, It, and the activation energy, E a» and the inverse

temperature, 1/T.

k = Acxp[-.E_a_] 2.4
RT
or lnk = lnA - [E] 2.5
RT

Therefore, Arrhenius plots of the relative ion abundances of the intact
molecule and thermal degradation products (which are a function of In k)
versus 1/T must intersect at some crossover temperature (Tc) since the
activation energies and, thus, slapes of the plots are different for these two

processes of desorption of the intact molecule and thermal degradation. A

5P
161
dc
he
du
hi;

16

generic Arrhenius plot for desorption and degradation is given in Figure 2.2.
At low temperatures (1/1‘ > l/Tc), the rate of degradation is greater than the
rate of desorption and, therefore, mostly K" adducts of thermal degradation
products would be observed in the K+IDS experiment. At high temperatures
(1/I‘ < life), the rate of desorption is greater than the rate of decomposition
and K" adducts of the intact molecule also would be observed in the K+IDS mass
spectrum. Therefore, rapidly heating the sample allows for the crossover
temperature to be exceeded while there is still enough sample to allow
desorption of the thermally labile compound to dominate. The selection of the
heating rate of the sample may be used to control the information obtained
during the K+IDS experiment. Molecular weight information is obtainable at
high rates of heating. Structural information, from thermal degradation
products, is available at lower rates of heating and lower ﬁnal temperatures.
Monitoring the surface temperature from which sample desorption
occurs also provides other interesting information. Williams and coworkers
investigated the kinetics of the volatilization process for thermally labile
compounds.40 Both slow (1-5K/min) and rapid (1000 K/sec) heating
techniques were utilized in these experiments. The N-acetyl(L-alanine)n
methyl esters were studied with the slow heating rates while rapid heating
chemical ionization was applied to polyalcohols and saccharides. The two
techniques gave similar results regarding the temperature dependence of the
rate of ion production of the molecular adduct ions. The activation energies
for the volatility of the intact sample molecules were calculated from the
slopes of the Arrhenius plots of the abundance of the molecular adduct ion
versus the reciprocal of the surface temperature. The activation energies
calculated increased with molecular weight within a class of compounds such
as the methyl esters. These activation energies also quantitatively resembled
the heats of vaporization obtained experimentally. Since intermolecular
hydrogen bonding is a dominant interaction that is reflected in the heat of
vaporization, the activation energies obtained for volatilization of the sample
may be quantitatively related to the degree of hydrogen bonding. Ideally
these temperature studies can be applied to the K+IDS technique to allow for
the determination of the relative activation energies for the volatilization of
different compounds and of different fragmentations from a given compound.
The ﬁrst indication that temperature could be used to alter the niass

spectra obtained with the K+IDS technique was from early studies

17

Desorption

log k

Degradation

 

 

Desorption favored (— —> Decomposition favored
1/T

Figure 2.2. Arrhenius plot of the log k vs. V1" for desorption and
thermal degradation processes. The slope of the line is a function
of the activation energy.

18

with tristearin. Tristearin is a fatty acid derivative of glycerol; the structure is
shown in Figure 2.3. The arrows on the ﬁgure show the fragmentations that
are observed during the K+IDS analysis and the resulting m/z values of the K+
adduct ions produced. Varying the current applied to the K+ emitter for each
analysis, produces different K+IDS mass spectra. At higher currents and thus
faster heating rates and higher ﬁnal temperatures, only the K1” adduct ions of
the intact molecule are observed. At lower currents and thus lower heating
rates, more thermal degradation products are observed as K" adduct ions. In
addition, variations in the K+IDS mass spectra can be observed at intermediate
heating rates within one experiment, when there is sufficient sample present.
This is illustrated in Figure 2.4 which is a plot of the relative abundance versus
scan number (time) for a variety of K“ adduct ions of thermal degradation
products and of the intact molecule of tristearin. Early in the experiment, the
thermal degradation products are abundant as the temperature is close to the
crossover point and, thus, the rate of vaporization and degradation are
competitive. Later in the experiment as the temperature increases, the rate of
vaporization is greater than the rate of thermal degradation and thus the mass
spectrum mainly contains the [M]K+ ion at m/z 929. Figure 2.5 shows two
K+IDS mass spectra of tristearin obtained by averaging the spectra in these
two different regions of the experiment. Mass spectrum A is from the early
part of the experiment and mass spectrum B is from the latter part of the same
experiment. These within-run variations prompted interest in pursuing this
temperature variable as a means of controlling the information obtainable
with the K+IDS technique.

It is the goal of this part of the research to exploit these variations in
the K+IDS mass spectra and alter the K+IDS mass spectra obtained with the use
of the temperature/heating rate variable. The ideal mass spectrum of any
compound contains both molecular weight and structural information.
Alteration of this temperature variable may allow for the production of K+IDS
mass spectra that contain both molecular weight and structural information.
The main difﬁculty with this endeavor is the dynamic nature of the K+IDS
experiment and the short-lived signals which typically last less than one
minute. Modifications to the probe design and experimental conditions are

made to allow for better control of this heating rate variable and to increase

the lifetime of the adduct ion signal.

19

PLW. - 890
mlz 4' (mm’ = «112 929

HC~

  

 

112C>‘

 

0
0 — é"(CH2 )‘6CH3

0
54042116013

0 .
"' O &‘(CH2)‘6CH3
0H+KI ONOK.
lﬂ/Z lﬂlz
663 307

 

+H+KI

«V:
647

./W

"V2
323

Figure 2.3 Structure of tristearin and the thermal degradations
observed as K+ adduct ions in the K+IDS mass spectra.

20

.8962: $8 399, 5.69th no weaned cosmomame .952: m5 8 new
3% NE: 2822: 695 m5 6 82 88cm t. on. 6 805223 m5 ed 239“.

mum—232 23%..
NN mm 0N VN MN NN _N

P . - . .

.6 ........... 0 .......... 9.

  

         
 
      
    

  
  

.l.b.1.l.@!\.\.

mmm. ~\E / x

on m. 9 ti

. p. p

l ........ 0 ................
...t.a.wan..umhminwou.. o .

\\/.

\
\\ anIN\E

  

nNm . ~\E

mmw - ~\E

2

tom.

uONm

tomv

uovw

 

BONVONDBV NOI

RELATIVE INTENSITY

RELATIVE INTENSITY

21

 

 

 

 

 

 

 

 

 

 

 

 

B 929
' 663
.. 913
‘ 323
307 647
379
I l A t l l I
I 1 1 Ii I I I I I I I I I I 1 I I ‘l I I ‘l I I I I I I I I j I
200 300 400 500 600 700 800 900 1000
ml:
A 9”
I ‘1 I‘ll I ‘I I I I I I W liuI ‘1 I I 1 1 I T 1.1T! I ‘1! I I l I I I I ‘I I U I
200 300 400 500 600 700 000 900 1000
mlz

Figure 2.5 K+IDS mass spectra of tristearin averaged over scans early (A)

and late (B) in the same analysis.

22

2. CHARACTERIZATION AND MODIFICATION OF THE K+IDS PROBE
A. ORIGINAL SINGLE F'ILAMENT PROBE DESIGN

As mentioned in Chapter 1. the most important aspect of the K+IDS
technique is the temporal and spatial overlap of the emission of K+ ions in the
gas phase and the desorption of neutral molecules and thermal degradation
products of the sample. Characterization of the K+IDS probe was performed to
determine the overlap of these two processes and to formulate ways of
improving the sensitivity of the technique. Experiments were designed to
monitor the two processes of K" emission and sample desorption individually
and correlate the results to determine the temporal overlap. EI was used to
monitor the desorption of the sample from the probe tip. The sample,
suspended or dissolved in solvent, was applied in the normal manner for a
traditional K+IDS experiment on the tip of the K+ bead. After the solvent
evaporated, the probe was inserted into the instrument with the probe tip
positioned just at the edge of the ion source. This placement was chosen so that
the K+ bead did not interfere with the path of the electrons from the BI
ﬁlament and to insure that desorption of the sample was from the current
applied to the probe and not from electron bombardment of the surface. There
were no ions detected from the sample with only the El filament on. However,
sample desorption was detected almost immediately (by El) after the current
was applied to the K4" head. The K+ emission was monitored in a separate
experiment with the normal K+IDS setup with the exceptions that the mass
range was adjusted to detect m/z 39 (K+) and the multiplier voltage was lowered
so that the strong m/z 39 signal did not saturate the detector. In this case there
was a lag time of approximately eight seconds before K4“ emission began. (This
time lag for K+ emission varies with the thickness of the K4“ glass covering the
ﬁlament wire.) The TIC chromatograms of these two experiments are overlaid
and shown in Part A of Figure 2.6. For both experiments, the current was
applied to the K+ bead at the same time (scan 7) and the scan rate of the
quadrupole was 1 second/scan over the mass range of 20-420 daltons. From
this overlap, it can be seen that most of the gas-phase sample molecules were

gone by the time 10‘ ions were produced in a large abundance. This poor

23

A. ORIGINAL PROBE TIP DESIGN

 

 

 

I applied to emitter

E} K* emission

sample desorption

B. TWO-FILAMENT PROBE TIP DESIGN

 

 

 

 

 

- “ \;( z x) ’X \\:\\\\\\\:
- . ‘3:«;( (\K/v sgmdﬁé w< \‘ \‘Q.
: [{AKV ﬁg) ”wwﬁxym’z“

- // 3;;3’1‘AI‘A)’ /¥x)8\/\:1\}AAXA6\)\)$ (A
14 2'4 34 6‘4 74

Sc:n

Figure 2.6 TIC chromatograms of the sample desorption of a fatty acid mix

obtained by El overlapped with the K+ emission for the original probe tip
design and the new two-filament probe tip design.

24

overlap resulted in poor sensitivity of the K+IDS technique as most of the

sample was desorbed too rapidly and, thus, was not detected.
B. DUAL FILAMENT PROBE DESIGN

To improve on the poor overlap of the two processes involved in the
K+IDS technique, new probe designs were pr0posed and tested for
improvement in this overlap and for ease of construction. The goal was to
substantially improve upon this overlap by either speeding up the onset of K+
emission or by slowing down the onset of sample desorption. Many different
ideas were addressed to solve this overlap problem including physically
separating the two processes and providing two different power supplies for
independent control of the K" bead and sample heating. Logistically, the
independent temperature control of the two processes was not feasible. Only
one probe inlet was available for a 1/4 inch outer diameter probe. This small
diameter greatly limited the design possibilities. The final design for the new
probe tip chosen after many tests with different wires. ribbons and other
support materials is shown in Figure 2.7. This design resembles the two-
ﬁlament “push-rod” probe tip design reported by Rollgen and co-workers for
the production of alkali metal adduct ions of thermally labile compounds.19
The probe tip shown in Figure 2.7 has a separate sample post very similar in
design to the original K4” bead ﬁlament. A 0.007” diameter wire (usually
tungsten/rhenium instead of the pure rhenium used for K+ bead ﬁlament for
economical reasons) is threaded through a two-holed ceramic support post
with a loop of wire left at the top. This sample wire is spot welded to two
support posts that are cemented into a four-holed ceramic in the tip of the
probe. This sample support is not directly heated, but is radiatively heated by
the K" bead when in close proximity to the tip of the bead (the distance
between the K+ emitter and the sample wire is less than or equal to one
millimeter).

The same overlap test of the K+ emission and sample desorption was
performed with this two-ﬁlament probe design. The sample is now placed only
on the wire loop at the tip of the sample support post. (A twist in the wire loop
just above the ceramic post helps prevent the sample from creeping down the
wire and into the ceramic post). The desorption of the sample is observed by El

in the same manner as in the previous test. The results of this overlap test are

25

 

 

233m
@585

 

 

.cgmmn a: Boa work EmEmEézF 5m 9:9”.

 

232m
526d

 

 

 

 

 

 

 

 

 

 

 

 

 

EEEerv. I.

 

 

 

 

 

26

shown in Figure 2.6 B. A much improved overlap of the two processes is now
obtained. The K“ emission began sooner due to the absence of sample covering
the K" bead as in the previous design. But more importantly, the onset of
sample desorption occurred later with respect to application of current to the
emitter. Thus, the temporal overlap of K" emission and sample desorption is
greatly improved. The sensitivity of the K+IDS technique improved 30- to 40-
fold with the development of this two-filament probe design. Kassel
documented the increase in sensitivity of this two-ﬁlament probe design for
the analysis of methyl stearate,41 which is a compound typically used for
determining relative sensitivities of ionization techniques such as El and CI.
Monitoring the [M]K"‘ ion of methyl stearate with selected ion monitoring
(SIM), the absolute peak height obtained with the original single ﬁlament
probe design was 140 arbitrary units and the peak area was 2.300 arbitrary
units squared. For the two-ﬁlament K+IDS probe, the absolute peak height was
5,100 arbitrary units and the peak area was 26,000 arbitrary units squared for
the [M]K+ ion of methyl stearate of equal sample volume. These results show
the large increase in sensitivity obtained with the two-ﬁlament probe design
for the K+IDS technique. This increased sensitivity is obtained for most other
thermally labile compounds. Kassel determined the detection limit of the
K+IDS technique with the two-ﬁlament probe tip for the analysis of methyl
stearate to be 3 ng in the SIM mode and 10 ng in the full scanning mode (50-
450 daltons and 0.5 sec/scan) at a signal-to-noise ratio of five.41 This two
ﬁlament probe design is used to obtain the remainder of the K+IDS mass
spectral data presented, unless otherwise noted.

The success of this two-filament probe design introduces some
uncertainty as to why the K+IDS technique works well for the analysis of
thermally labile compounds. The new design produces relatively the same
K+IDS mass spectra as were previously obtained with the original single
ﬁlament probe design. However, the heating rates of the K+ bead surface
(where sample was originally placed) and the separate sample post are not the
same. This prompted more detailed studies of the temperatures of these
different surfaces and the actual heating rates needed to desorb the thermally
labile compounds studied. The term “rapid heating” is generic in nature, as is
the Arrhenius plot in Figure 2.2 which provides no idea of the actual
temperature needed to reach the crossover point and promote vaporization of

the intact thermally labile molecule. Beuhler reported that the volatilization

27

of sucrose reaches half of the maximum rate by 111°C and that rapid heating
of the peptide TRH to 215°C produces twice as much protonated parent ion as
fragment ion.34 These temperatures are much lower than previously expected
to promote vaporization of thermally labile compounds based on the
understanding of the K+IDS technique. In the original probe design, the
samme was deposited directly onto the K" bead surface which reached very
high temperatures (approximately 1000°C) at the onset of the K+IDS
experiment. Thus, it was assumed that these high temperatures were actually
needed to desorb the thermally labile molecules intact. The success of the two
ﬁlament probe design, however, questions this original assumption since the
sample is no longer achieving the same high temperatures and yet the same
K+IDS mass spectra are obtained. Several experiments were designed to gain a
better understanding of these heating rates and temperatures needed for the

desorption of thermally labile compounds.

3. TEMPERATURE STUDIES OF THE K+IDS TWO-FILAMENT PROBE
DESIGN.

A. DETERMINING SURFACE TEMPERATURES IN A VACUUM CHAMBER

The first experiments to study the temperatures of the different
surfaces of the K+IDS probe were performed in a glass vacuum chamber
separate from the mass spectrometer for easier access to the probe. The
experimental design is shown in Figure 2.8. An evacuated three-neck round
bottom ﬂask was used to mimic the ion source of the mass spectrometer. An
optical pyrometer was initially used to measure the temperature of the K4" bead
surface. It was determined that the emitter reached temperatures between 800
to 1250°C when K+ emission was observed (heating currents > 2.0A). Heating
the K" emitter inside the clear glass vacuum chamber allowed for visual
inspection of the heating process and the differences of the two ﬁlaments.
The K+ bead tip glowed bright red at low currents (1.0 to 2.0A) and became
bright white when the heating currents were greater than 2.0A. The ceramic
post of the K+ bead remained an orange-tinted glow throughout most of the
experiments. In contrast, the sample ﬁlament did not show any physical signs
of heating, which was the first evidence of the drastic temperature
differences between the two surfaces. This visual observation prompted

further studies of the temperatures of the two surfaces and of the heating

28

ll

 

 

Glass vacuum manifold

 

Three-neck glass
round bottom flask

Iii

 

 

 

 

 

 

 

 

K+ IDS probe

Thermocouple

D

 

 

 

 

Current and biasing
voltage power supply

 

 

Figure 2.8 Experimental setup for the temperature measurements

 

Glass 'U" tube

temperature meter

 

 

 

 

 

Rough
pump

 

of the K* IDS probe performed inside a glass vacuum chamber.

 

 

 

29

rates of the surfaces when different currents are applied to the K+ emitter

filament.
1. EXPERIMENTAL DESIGN

To determine the heating rates, a thermocouple that had a fast response
time and could respond to a wide temperature range (25-1400°C) was needed.
Two different kinds of thermocouples were obtained from OMEGA Engineering,
Inc. (Stamford, CT) as well as a monitoring device. Type K and Type C thin,
bare wire thermocouples were obtained for their quick response to
temperature changes. The Type K thermocouple is chromel/alumel with an
upper temperature limit of 1250°C. The type C thermocouple is composed of two
wires which are tungsten 5% rhenium/tungsten 26% rhenium and it has an
upper temperature limit of 2320°C. Both of these thermocouples were
converted into vacuum tight probes with glass tubing and epoxy. Small glass
tubing was used to insulate the two thermocouple leads from one another. The
wires and small tubing was threaded through a larger glass tube with an outer
diameter of 0.25 inch and approximately six inches long. Five-minute epoxy
was used to seal up both ends of the larger glass tube to make a vacuum tight
probe. This probe and the K+IDS probe were inserted into the" round bottom
flask, sealed with O-ring connectors and evacuated with the vacuum chamber.
The tip of the thermocouple could be positioned on any surface of the K+IDS
probe, while under vacuum.

The Type K thermocouples obtained were 0.001 inch in diameter and,
thus, could respond to temperature changes more quickly than the Type C
thermocouples which were 0.003 inch in diameter. However, the Type K
thermocouples were ﬁne hair-like wires that broke easily and proved to be too
difficult to work with. Thus, the Type C thermocouples were used for the
initial heating rate studies. The Type C thermocouple leads were connected to
an OMEGA DP-80 series thermocouple monitor that was equipped with an
analog output option card so that a signal representing the thermocouple
temperature could be sent to a chart recorder. The analog output was
connected to a Houston Omnigraphic 2000 X-Y recorder and a temperature-
versus-time plot was obtained. The X-axis was set at 25 sec/cm and the Y-axis
was set at 1V/in. The OMEGA temperature monitor was programmed so the

analog output of zero V corresponded to +25°C and 10V corresponded to +1525°C.

3 0
2. RESULTS

The heating rates were approximated from plots of temperature versus
time at the different current values applied to the emitter. An example of the
temperature vs. time plots for the two different probe tip surfaces (K+ bead
and sample wire support) is shown in Figure 2.9. The two measurements were
made consecutively with the same thermocouple. The thermocouple was
repositioned onto the different surface and the bead was cooled between
experiments. For the data shown in Figure 2.9, 3.0A were applied to the K”
emitter at the start of each experiment. As can be seen in these plots, not only
is there a discrepancy in the ﬁnal temperatures of these two surfaces, but also
a difference in the heating rates of the two surfaces. This difference in
heating rates is indicated by the slopes of the heating curves. This process was
repeated for six different currents between 2.25A and 3.50A. Figure 2.10 shows
a plot of the ﬁnal temperatures of the K“' emitter surface and the sample
support surface as a function of current applied to the emitter measured with
the Type C thermocouple probe. The heating rates of the two surfaces were
approximated by dividing the temperature obtained at half the time it took to
reach the maximum temperature by the half time, or T1/2/t1/2. The heating
rate of the emitter surface is hard to determine with this method as the
temperature is changing so rapidly that the monitoring device is not keeping
up with the change. The heating rate (Tug/tug) is approximately 200°C/sec
during this initial heating period of the K“ bead. The sample surface has a
much lower heating rate and therefore changes in the heating rate as a
function of the emitter current can be observed. Figure 2.11 shows a plot of
the approximate heating rates (Tl/zltl/z) of the sample support surface as a
function of current applied to the emitter. The data in Figures 2.10 and 2.11
demonstrate that the current applied to the emitter affects both the final
temperatures of the surfaces of interest and the heating rates. As the current
increases so does the final temperature and the heating rate (especially for
the sample support surface). These figures also emphasize the major
differences in the final temperatures and the heating rates of the two
different surfaces. '

This proof of the differences in heating rate and final temperature of
the sample in the two different K+IDS experimental designs prompts the need

to explain why the K+IDS mass spectra remain unchanged. One possible

31

.5580 m5 2 83% m_ o. cm a 9:.
82? ESQSES 9? E952: macaw 95 been .mz_Ew<+x.m mazesgh

 

 

 

 

 

 

 

 

 

 

N2“
.
_ a
E 08:. u _
_ _
: : : : 3 8 8 .2 no
at" n n I I ._ u x J...
.
. I ma
" iﬁ ..... Wm:
55.55% H l E e
iiiiiiiiii lliitiiurv meH
II on.
II who I.
m m:
3:: 2 P... vulntiwiw ._.
I one
SE .2 025.82 a m
a
[I one I“
b
II an?
amt—2m KC 11 mm:
iiiiiiiiiiii riiiiluiw meH
LI 3?

1300-
12001
1100:
1000':
goal
800-
700-:

q

600-
J

500‘
-l

400-

Temperature (°C)

300 -
200 -

 

100-
a

32

a me)
e T(Bead)

El

 

oiﬁﬁ '

I ' I ‘ I ' I f I ' I ‘ I

1.001.251.501.75 2.00 2.25 2.50 2.75 3.00

Current (A)

Figure 2.10 Temperature of the emitter surface and
sample filament wire vs. current applied to the emitter.

33

 

 

25-
a

204

E)?
El
Q
9 A 15-
37%
a: $3 '5'
a: ET
0)\
.E 5 10- a
a; l-
0)
I
El
5 - E
0 ' 1 1 I ' I ' l
2.0 2.5 3.0 3.5 4.0

Current (A)

Figure 2.11 Heating rate of the sample filament wire
vs. emitter current.

 

34

explanation for this may have to do with where the sample resides when
applied to the K+ bead versus the second ﬁlament support wire. It is possible.
for instance, when applying sample that is suspended in a solvent such as
methanol to the K+ bead in the original probe design, that some of the sample
may creep down the ceramic post of the emitter tip. The sample observed with
K+IDS utilizing the old probe design could be the result of desorbing sample
from the ceramic and not from the tip of the K+ bead which heats up so
rapidly. The sample that resides directly on the tip of the K+ bead may desorb
too quickly, before the production of K+ ions, and therefore not be detected.
The temperature of the ceramic post of the K+ bead was monitored in a manner
previously described to test the heating rates. The thermocouple was
consecutively positioned on the three different surfaces: the K+ bead tip, the
ceramic post of the K+ bead, and the sample support wire. The temperature of
each surface was monitored as a function of time with the same current
applied to the K+ bead for each experiment. These measurements were
repeated for a range of currents between 2.25A and 3.00A applied to the K+
bead. Just as before, the probe tip was allowed to cool down to approximately
room temperature before each analysis. The results of this test are shown in
Figure 2.12. The heating of the ceramic post closely resembles the heating of
the sample support surface. which adds some experimental support to the
above hypothesis. However, it is hard to prove this notion that the observed
sample in the original design is desorbing from the ceramic below the bead as
it is difﬁcult to observe where the sample actually resides on the K+ bead and
post when applied to the tip. It is much easier to observe the deposition of the
sample on the bare wire typically used for the second ﬁlament sample support.

The two-filament design fosters the notion that surface bombardment
could play a role in the K+IDS technique since the two filaments are in such
close proximity. Another temperature-related experiment was designed to
dispel this premise. If many K+ ions were bombarding the sample surface, a
temperature rise of the sample surface would be observed. Once the
temperature of the sample surface equilibrated with a constant current being
applied to the emitter, variations were made in the bias voltage applied to the
tip of the K+ emitter. Varying the bias voltage applied to the tip of the K+
emitter from 0V to +150V at a constant heating current did not change the
temperature of the sample surface signiﬁcantly. This supports the notion that

the sample support is being heated radiatively and not by bombardment. This

35

1200 -
a
A
1000 4 ‘
A
A d
33, 800 A T(bead)
g 0 T(ceramic)
g 600 - El T(sample)
o.
E
CD
1.—
400 ~ o
o
o a
El
200 - 0 la
a
o l I I ‘ I I ﬂ

 

 

Current (A)

Figure 2.12 Temperature of the emitter surface, emitter
ceramic post, and the sample filament wire vs. emitter current.

36

provides additional support to the mechanism of adduct formation that K+
attachment is occurring in the gas phase and not on the surface.

Changes in heating rate, by changing the current applied to the emitter
and the design of the probe tip, affect the K+IDS mass spectra obtained. A
series of K+IDS experiments of tristearin with 2.0A. 2.5A, 3.0A, and 3.5A applied
to the K+ bead were performed with very interesting results. The onset of K+
adduct ion formation varied with current and the relative appearance of the
parent adduct ion of the intact molecule versus the K"' adduct ions of thermal
degradation products varied with current. The 3-D plots of these four
experiments are shown in Figures 2.13 and 2.14. The current was ﬁrst applied
at scan number 10 for each analysis and the scan time was 0.5 sec/scan. From
these plots it can be seen that increasing the current increases the heating
rate of the surface and thus the K+IDS mass spectra are observed earlier in the
run. Also. there is some difference in the relative appearance time of the
fragment ions versus the molecular adduct ion at m/z 929. For example, the
experiment obtained at 2.5A shows that thermal degradation products are
observed by scan 20 and the [M]K+ ion is not observed until scan 27. This may
be an example of a slower heating rate where the temperature is below the
crossover point at the beginning of the experiment and thus thermal
degradation products dominate. In agreement with the Arrhenius plot (Figure
2.2), the temperature has reached the crossover temperature at scan 27 and
therefore vaporization of the intact molecule dominates. For the experiment at
the higher current of 3.0A, the heating rate may be great enough that this
crossover point is reached at the beginning of ion formation. It is also
possible that in this case, the limiting factor in ion formation is the production
of K+ ions. Any thermal degradation occurring before scan 16 is not detected
as K+ emission has not begun. This is supported by the fact that the next
experiment at 3.5A has the same lag time (16 scans, or 3 seconds after current
is ﬁrst applied) before any K+ adduct ions are observed. The poor sensitivity
of this last experiment of the series suggests that most of the sample has
desorbed prior to the onset of K+ emission. These changes in appearance times
of different fragment ions and the intact molecule with changes in the
surface temperature and heating rate support the idea that the thermal
decomposition of the sample is occurring on the surface. The products are
desorbed into the gas phase followed by K+ attachment. K+ attachment does

not induce fragmentation, but merely samples the neutral gas-phase species.

37

Tristearin on sample wire 1:2.0A

 

Tristearin on sample wire I=2.5A

 

 

I .
R81. . 50
Int. :M

 

 

Scan '

 

 

 

 

 

 

 

l‘
‘ITTI IIII [III ‘leI’ III! IIII1TIIIIIIII[\

140 240 340 440 540, 640 740 E40 946

O

m/z

Figure 2.13 Plot of the K+IDS mass spectra of tristearin when 2.0A
and 2.5A are applied to the K+ emitter and the sample is on the
sample filament wire only.

38

Tristearin on sample wire I=3.0A

 

Tristearin on sample wire I=3.SA

 

Figure 2.14 Plot of the K+IDS mass spectra of tristearin when 3.0A
and 3.5A are applied to the K+ emitter and the sample is on the
sample filament wire only.

39

This analysis was repeated with a different K+ bead and sample support
designed to decrease the heating rates of both surfaces. The thermal mass of
the K+ bead was reduced (thinner coating of aluminosilicate on the emitter
wire) and the ceramic post on the sample support was reduced by half. The
results of the K+IDS experiments performed with this probe design when 2.0A
and 3.0A were applied to the K“ bead are shown in Figure 2.15. Again the
current was applied at scan 10 for both of these analyses. The heating rate of
the sample surface is greatly increased with this new bead design. The intact
neutral molecule of tristearin ﬁrst desorbs at scan 15 when 2.0A is applied to
the emitter compared to desorption ﬁrst occurring at scan 45 when the same
current was applied to the slower heating design (Figure 2.13 A). Also, it
appears that K+ emission is starting sooner with the thinner emitter as adduct
ions are observed 3 to 4 scans earlier than for the previous design. The
thickness of the K+ emitter and the amount of ceramic support post on the K”
emitter and the sample ﬁlament affect the rate of heating of the sample and
the onset of K+ emission. These variables in the probe tip design may be
adjusted to obtain the desired K+IDS mass spectra, either promoting thermal
degradation or desorption of the intact analyte molecule. For most thermally
labile compounds studied to date with the two ﬁlament K+IDS probe tip, the
first design with more ceramic covering both the K+ bead and the sample
wires produces better results than the design with reduced thermal mass (less

ceramic post).

B. CORRELATION OF K+ IDS MASS SPECTRA WITH SAMPLE SURFACE
TEMPERATURE

Preliminary K+IDS studies, especially of mixtures, indicate that
different compounds are observed at different times during a K+IDS analysis
and, therefore, require different surface temperatures to desorb and be
observed. If these temperatures at which different compounds vaporize can
be determined, then it is possible, based on the Arrhenius equation, that
activation energies, or heats of vaporization, may be determined for
desorption of these different compounds. K+IDS provides a means of
monitoring this sample desorption by the formation of [M]K+ adduct ions for
thermally labile compounds, M. It is assumed that the desorption of the

analyte is the rate limiting step in the formation of the adduct ions, which is

4o

Tristearin on sample wire with bare wire emitter I:2.0A

4O

Scan 0

 

140 240 340 440 540 640 740 840 946°

Tristearin on sample wire with bare wire emitter I=3.0A

901.

16
Int. mm
- 1m 13 Scan 0
x 11L 1 ll
ﬁ.
; 11

IIII]IIIIleTrlIIIIIIIIIIIIIIIIIIIIIIIII\

140 240 340 440 540 640 740 840 946

 

 

 

m/z

Figure 2.15 Plot of the K+IDS mass spectra of tristearin when 2.0A and
3.0A are applied to the K+ emitter. In this case, the thermal mass of the
K+ emitter was reduced by decreasing the length of the ceramic support rod.

41

explained in more detail later. Thus, by monitoring the surface temperature of
the sample ﬁlament and the formation of [M]K+ adduct ions, the temperature at
which thermally labile compounds desorb can be determined. Beuhler et al.
reported experimentally determined activation energies for the production of
both parent and fragment ions of some small saccharides and erythritols.39
These activation energies were calculated based on the Arrhenius plots of the
rate of production of these sample ions as a function of the reciprocal of the
absolute temperature of the sample support wire. The basic trend observed
was an increase in activation energy with increasing molecular weight. The
main discrepancy observed with the calculated activation energies was for the
formation of the fragment ions of sucrose. Two of the fragment ions had very
similar activation energies that were much lower than for the formation of
the other fragment ion and the parent adduct ion. This suggests that there
may be more than one mechanism involved and that there may be processes
involving the monolayer molecules on the bare metal surface versus
multilayer molecules on other molecules. Thus. the determination of
activation energies for the formation of the different fragment ions and
parent adduct ion observed may aid in the understanding of their origins.

The main goal of these K+IDS temperature studies is to determine the
temperatures required to desorb intact thermally labile compounds and
produce thermal degradation products which are detected as K+ adducts in a
K+IDS mass spectrum. Once this goal is achieved, the next step is to determine
the activation energies of these desorption and degradation processes. The
ﬁrst experiments designed to correlate the temperature measurements of the
sample support surface to the appearance of the K+IDS mass spectra required
two steps and two experimental designs. The temperature measurements of the
sample support surface were performed in the vacuum chamber as previously
described. Plots of temperature versus time for various currents applied to the
K+ emitter were obtained. Once the temperature measurements were
completed, the same probe system was used to obtain K+IDS mass spectra of
compounds at the various currents used for the temperature measurements.
Current was first applied to the K+ bead at scan 10 and the scan rate was 0.5
sec/scan. From this information, the time of sample desorption could be
calculated based on the formation of K+ adduct ions. The previously acquired

temperature versus time plot, obtained at the appropriate heating current, is

42

used to determine the temperature of the sample filament at the time of
desorption.

The results of such experiments are shown for a variety of compounds.
From the 3-D plot of the K+IDS analysis of tristearin at 3.0A, Figure 2.14A, it was
determined that desorption (observed as a peak at m/z 929 representative of
the [M]K+ ion) was ﬁrst observed at scan 16. The temperature of the surface at
scan 16 was extrapolated from the temperature vs. time plot of 3.0A to be 165°C.
When only 2.5A was applied to the emitter (Figure 2.13B), the desorption of
tristearin did not appear until scan number 27. In this case, the temperature
was determined to be approximately 170°C at the onset of desorption. These two
temperatures are within the experimental errors for the measurements. The
lower heating current produces a slower heating rate for the sample wire and
therefore it takes longer to reach the same temperature. The same types of 3-D
plots are shown in Figures 2.16 and 2.17 for the K+IDS analysis of sucrose and
hexaglycine respectively. It was calculated from the temperature vs. time
plots that sample desorption of the intact molecule for both of these
compounds occurred when the sample wire temperature was 150°C. The
structure of hexaglycine is shown in Figure 2.18. The thermal degradation of
hexaglycine observed in the K+IDS mass Spectrum of this compound is labeled
in Figure 2.18 along with the m/z values for the corresponding K+ adduct ions.

This experimental design allows for the correlation of temperature
measurements performed in the vacuum chamber to be made with the K+IDS
mass spectral data of several types of compounds to determine the
temperatures of sample desorption. This experimental design, however, lends
itself to much systematic error. One source of error is in the conﬁguration of
the K+IDS probe itself. In order for these experiments to give accurate
temperature measurements. the distance between the sample wire and the K+
bead must remain constant while obtaining the temperature measurement and
the K+IDS mass spectrum. This distance greatly affects the heating rate of the
sample wire. It is difficult to ensure that this distance remains constant
during transportation between the two experimental setups for temperature
measurement and K+IDS mass spectral data collection. Another source of error
in this design is with the mass spectrometer data system. It was discovered
midway through these temperature studies that the data system does not always
continuously collect data as previously assumed. When the data collection

program is operated in the real time mode of monitoring the mass spectra. a

43

.0009 u A95, maEmmz. .8 $8 8 8288 55 mm; 53.8% maEmm

<3 n _ .28 9;,
toaaam m5 .5 29:8 53> 898m B 828QO $9: mo_+x .mg 929;

EE
ovv can own omm ovm omw ovw
OF brb b — n n n n — n h - bit— r- n n — n n n u — n n h -
mp l1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a zoom

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.~c_
._mm

 

 

 

 

 

 

 

 

 

 

 

 

5N N}:

 

44

Doom? u 8:2, maEmmv... .Nm cmom am 82820 3.: mm; 83.8% maEmm
<o.mn_ Eco we? togaam
m5 :0 29:8 9: 53> 862988: B :58QO gas mm: v. K To. 2:9”.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NE.
one cow com com omm com
or
/—f\ n n h n — n P p n - p p h — b - p — n p n n
m? k x
d.
ON {INN NN .-
/M N
I? I N
mN [MN - NN
Mix m . . II\ I
u cmom on M ._ L~
4 “a .
IN .l 1
mm MN "a .u
/T a“. I
0? UP“ .m u“ ”I _ .U\
4 “u n f
me nu .
4 a
om MN \ ._mm
4 \

 

 

 

 

\ x N, x

+ 9+5; NE NE mm as mum NE

45

-----------------

ﬂ---
I

§

4. \
K m/z 342

1""

..................

1""
7i
+':

\
m/z 285

‘---

1""
X
+

\
m/z 228

I
c
I MW = 360

m [M]K+= 399

Figure 2.18 Structure of hexaglycine and the thermal
degradations observed as K+adduct ions.

46

delay time of four seconds occurs every time the computer redraws the mass
axis at the bottom of the screen. During this four second delay, data are not
being collected. This is a very serious problem for this project as the -K+IDS
spectra are so short-lived and variable in nature that four seconds is a very
large gap in data collection. Also. this lag time creates a problem when trying
to calculate the time interval of sample desorption based on the scan number
and, thus. time. This lag time was taken into consideration when calculating
the temperatures of sample desorption discussed, but it still adds some

uncertainty to the extrapolation.
C. REAL-TIME DETERMINATION OF SAMPLE DESORPTION TEMPERATURES.

1. EXPERIMENTAL DESIGN

The uncertainties in the temperature measurements described with the
previous design led to the development of a simultaneous temperature and
mass spectral data collection process. The ideal experiment would provide real
time temperature measurement of the sample support surface while desorption
via the K+IDS technique is being monitored. In this case. collection of the
K+IDS mass spectrum is made simultaneously with collection of the
temperature vs. time plots of the sample wire ﬁlament. The thermocouple is
designed to be the sample support ﬁlament. Chromel and alumel wires 0.01
inch in diameter are threaded through the K+IDS probe and spot-welded
together at the tip to form the thermocouple junction for a Type K
thermocouple. A small piece of two-holed ceramic is used at the tip to mimic
the previous sample filament and provide the same heating characteristics.
The thermocouple wires are obtained from OMEGA Engineering, Inc. with a
teflon coating that provides electrical isolation of the wires through the
length of the probe. The insulation has been removed at the tip where the
wires are exposed so as not to interfere with the K+IDS analysis. The sample is
placed directly on the tip of the thermocouple and, therefore, the temperature
of the surface can be measured directly as the sample desorbs from the
surface. Apiezon Q sealing compound (Kurt J. Lesker, Inc., Clairton, PA) is used
to create a temporary vacuum seal around the thermocouple leads at the back
end of the probe. The thermocouple is connected to the same OMEGA DP-80

series monitor used in the previous studies. The output of the temperature

47

monitor is connected to a SOLTEC 1242 chart recorder to obtain a temperature
versus time plot of the sample wire surface. There still exists some uncertainty
with this design as the temperature is still extrapolated from the plot.
However, the variable distance between the sample wire (thermocouple) and
the K" bead is no longer a factor as the analyte surface temperature is
monitored simultaneously with the collection of the K+IDS mass spectra. Also.
the mass spectral data are collected in the chromatographic display mode
which is a constant data collection mode and do not suffer from the time lags
previously mentioned for the real time spectral monitoring mode. Therefore,
this experimental design should provide more accurate measurements of the

surface temperature during sample desorption.
2. RESULTS

The goal of this experiment is to obtain a sample desorption proﬁle in
terms of the sample filament wire temperature in order to gain a better
understanding of the temperatures needed to observe different thermal
degradation products relative to the desorption of the intact molecule. K+IDS
provides a way of monitoring the production of these gas-phase degradation
products and intact molecules by the production of K+ adducts. The raw K+IDS
mass spectral data were correlated with the temperature measurements
obtained from the chart recorder plots, in the same manner as the previous
experimental design, to create adduct ion abundance versus temperature plots.
The plot in Figure 2.19 shows the abundance of the molecular adduct ion (the
[M]K+ ion is represented by a peak at m/z 381) as a function of surface
temperature for the K+IDS analysis of sucrose. Each data point represents one
mass spectral scan collected during the experiment. These data were obtained
with a scan speed of 0.5 sec/scan over a mass range of 134-634 daltons. The
erratic signal is characteristic of the results of these experiments. Some
possible explanations for the origin of this adduct ion fluctuation are that the
K+ ion signal is ﬂuctuating, the desorption of the neutral species is
fluctuating, or there is some instrumental problem. A thorough cleaning of
the entire instrument and repairs of weak electrical connections resulted in a
decrease in these fluctuations, but not their complete elimination. The overall
proﬁle of Figure 2.19 resembles a Gaussian peak with observance of the [M]K+

ion of sucrose being observed when the sample ﬁlament is between 225-365°C.

 

48

1200 -

1000 "

800 '-

 

600 -‘

Abundance

400 -

200-

 

 

I I I
400 500 600 700

Temp K

Figure 2.19 Abundance of the [M]K"'ion of sucrose vs.
Temperature (K) of the sample filament wire.

49

Another example of the abundance of the sample K+ adduct ions versus
temperature of the sample support wire is shown in Figure 2.20 for the K+IDS
analysis of melezitose. The relative abundance of the [M]K+ ion (m/z 543) and
the K+ adduct ions of the most abundant thermal degradation products (m/z
363, m/z 201, and m/z 183) are plotted versus the temperature of the sample
wire ﬁlament. The structure of melezitose and the origin of the fragment ions
observed in the K+IDS mass spectrum are shown in Figure 2.21. There is a
slight variation in the temperatures at which the maximum abundance of
these adduct ions are observed. A better comparison may be the differences in
the overall adduct ion proﬁles as opposed to comparisons of the temperatures
of maximum abundance. The adduct ion proﬁle of the [M]K"’ ion of m/z 543 is
slightly shifted to the right on the temperature axis with respect to the other
K+ adduct ion proﬁles of the degradation products. This shift is not substantial
enough to suggest any major differences in the formation of these different
K4“ adduct ions. Repetitive experiments with melezitose were performed and
the plots of adduct ion abundance vs. sample. filament temperature are
provided in Appendix A. It has proven more difﬁcult than expected to adjust
the heating conditions to produce a K+IDS experiment where the mass spectra
are changing with time and analyte temperature. The main adjustments to the
probe tip made in an attempt to change the heating rate of the analyte
included changing the spacing between the two ﬁlaments and the amount of
ceramic post present on both the emitter and sample filament. Reducing the
amount of ceramic post on the sample wire ﬁlament seems to be a better way of
increasing the heating rate than decreasing the amount of ceramic post on
the K+ emitter. Both of these changes would increase the heating rate of the
sample, but the latter would result in bare wire on the K+ emitter that typically
results in mass spectra that contain more noise due to more chances of surface
ionization and electron ionization.

Compounds with multiple functional groups are likely candidates for
observing variations in the K+IDS mass spectra within a single experiment as
these compounds have many possible fragmentation sites. Methionine-
enkephalin and leucine-enkephalin are pentapeptides that were studied with
the hope of observing some variations in the K+IDS mass spectra with time
(sample wire temperature). Plots of the abundance versus temperature for the
most abundant fragment ions, observed as K+ adduct ions, of these peptides are

provided in Appendix A. The desorption ion proﬁles of these fragment ions as

50

800 -

 

 

600-
m/z 543
m/z 363
Q)
8 mazot
(U
'8 400 - mlz 183
3
.0
<<
200-

 

 

 

0 lﬁj

I I ‘ U I i 1 V I
520 540 560 580 600 620 640
T (K)

Figure 2.20 Abundance of the [M]K+ion and K+adduct ions of
thermal degradation products of melezitose vs. temperature (K).

51

MW=504

  

 

 

[M]K+= 543
,, 381—'—H.2°—. 363
.v"" +
CHZOH K
H o H
m 5 G H m cuzou
H m —1"—-> H
H

dc----

 

 

 

Figure 2.21 Structure of melezitose and the thermal
degradation products that are observed as K adducts.

52

a function of temperature overlapped substantially with no trend observed as
to the order of desorption of the different thermal degradation product ions.
The same is true for digoxin, which is a cardiac glycoside that is analyzed and
discussed in detail later. There is little difference in the sample wire
temperature when the fragment adduct ions of digoxin and the K+ adduct ion
of the intact molecule are observed. Shown in Figure 2.22 is the plot of the
abundance of the K+ adduct ions of the intact molecule (represented by a peak
at m/z 819) and the thermal degradation products of digoxin (represented by
peaks at m/z 169. m/z 299, m/z 429, m/z 559. and m/z 689) versus the
temperature of the sample ﬁlament. The overlap of all these K+ adduct ion
profiles suggests that this experiment was performed near the crossover
temperature where the rates of degradation are similar to the rate of
desorption of the intact molecule.

Even though these within-run variations have proven difﬁcult to
reproduce, there does seem to be some temperature differences between the
desorption of the intact molecule of different compounds. For example. Figure
2.23 contains the abundance of the molecular adduct ions, [M]K+, versus
surface temperature plots for sucrose (m/z 381) and melezitose (m/z 543)
obtained when the analytes were codeposited onto the sample wire ﬁlament.
As expected from the previous plots, observance of the K+ adduct ions of intact
sucrose molecules begins prior to observance of the K1” adduct ions of intact
melezitose molecules. The peak shapes of these K+ adduct ion proﬁles of the
intact analyte molecules resemble Gaussian shaped peaks that are shifted with
analyte ﬁlament temperature. This shift can be used to indicate the relative
thermal lability of the analytes. Instead of focusing on variations in the
appearance of different K" adduct ions from one compound as a function of
sample filament temperature, it may prove more beneficial to consider the
appearance of [M]K+ ions of different compounds as a function of temperature.
A nonequimolar three-component mixture of palmitic acid, sucrose, and
melezitose was analyzed by K+IDS while monitoring the sample wire
temperature. Three distinct [MjK+ adduct ion proﬁles are obtained with this
mixture for the three different components. The abundance vs. temperature
plot for the [M]K+ ion of each component is shown in Figure 2.24. The
molecular adduct ion profiles shift to the right along the temperature axis
with an increase in molecular weight of the analyte. This same trend was

observed by Williams, et. al. when they calculated activation energies for the

600-

500-

400-

Abundance
w
8

2001

100-

i . .thbrﬁﬁusp ..‘

e F n

.0
‘\

 

53

 

Q.‘.

0
c--. '

I V I V I

600
T (K)

_a_
...............
---...--

-o-o*o-o

--*-c

I ‘I 1 1

700

Figure 2.22 Abundance of the K+adduct ions of

digoxin vs. sample filament wire temperature (K).

I'll/Z169
"V2299
I'll/Z429
"V2559
M689
I'll/2819

Abundance

54

600-

500'

400-

 

30° " o ““9““ Sucrose
—ﬁ— Melezitose

 

200- !

 

100‘

 

 

300 0460 ' 500 660 ' 760
Temp (K)
Figure 2.23 Abundance of the [M]K+ ions of

sucrose and melezitose vs. temperature (K)
of the sample filament wire.

55

4003-

3000q 9
----o--- Palmitic Acid

~
.
.0

”‘0'” Sucrose x 5
—9— Melezitose x 20

Abundance
'3
8

 

10a)-

 

...Q...
.....ou-ooonoouoo-o-OIOIOO-0

 

 

o "‘1ﬁ"'1'1“I'ﬁ‘
250 350 450 550 650

Temp (K)

Figure 2.24 Abundance of the [M]K+ ions of palmitic
acid, sucrose, and melezitose vs. temperature (K)

of the sample filament wire.

56

volatilization of a series of N-acetyI-L-alanyl methyl esters and noted a regular
increase in activation energy with molecular weight.“0 The appearance of
[M]K+ ions as a function of temperature did follow this trend that an increase
in molecular weight leads to higher temperatures required for desorption of
the intact analyte. Additional plots of the abundance of the [MlK+ and K+
adduct ions of neutral thermal degradation products versus sample ﬁlament
temperature for other compounds studied are provided in Appendix A for

future reference.

D. ACTIVATION ENERGY DETERMINATIONS BASED ON K+IDS ANALYSES

Activation energies for the processes leading to production of K“ adduct
ions of the intact molecule and neutral thermal degradation products can be
determined from the log of the rate constant for producing these adduct ions
as a function of the reciprocal of the absolute temperature of the sample
ﬁlament. The rate of production of these K+ adduct ions, or abundance, may be
a good approximation for the rate constant of the volatilization of thermally
labile compounds. Williams et al. determined activation energies for
vaporization of thermally labile compounds by rapid heating and DCI using
this approximation.40 They assumed that the absolute abundance of the adduct
ions was a function of the rate constant. Thus, by plotting abundance vs. 1/1‘.
the activation energy was determined from the slope of the line of this
Arrhenius p10t. Activation energies, or heats of vaporization, for the
desorption of the intact analyte molecule of palmitic acid and sucrose were
determined by this method with the K+IDS technique. Detection of the peaks
representative of the [M]K+ ions was used to monitor sample desorption. The
plot of intensity of the peak representative of the [M]K+ ion of sucrose versus
III is shown in Figure 2.25. It is assumed that at the maximum intensity, the
sample is over half depleted. Therefore. the activation energy is calculated
with the ﬁrst half of the plot only, 1/1‘ > 1.75 x 10'3K'1,when the amount of
sample is not the limiting factor. From the slope of the best fit line. the
activation energy, or heat of vaporization. of sucrose is calculated to be 13
kcal/mol. This is significantly lower than the activation energy for the
formation of the [M]NH4+ ion of sucrose reported as 38.9 kcal/mol by Williams
et al. obtained by rapid heating DCl.40 This method of calculating the

activation energy of vaporization does not take into account the competitive

 

57.

1200 -

1000-l

800‘

 

600-

Abundance

400-

200‘

 

 

 

1/T x1000

Figure 2.25 Abundance of the [M]K+ ion of sucrose vs. 1/T (K).
Ea calculated from the slope of the line of the first half of the

experiment = 13 kcal/mol.

58

processes of thermal degradation of the sucrose molecule or the continual
depletion of the sample with time. The more thermal degradation occurring,
the less vaporization of the intact molecule can occur, both of which are also
dependent on the amount of sample present on the sample wire. This could be
the reason for the discrepancy in the heat of vaporization of sucrose
calculated based on the K+IDS technique compared to that calculated based on
the DCI rapid heating experiment. The calculation of the activation energy for
the desorption of palmitic acid, determined by the formation of [MlK+ adduct
ions, was performed with slightly better results. The plot of the intensity of
the peak representing the [M]K"‘ ion versus 1]? is shown in Figure 2.26. The
slope of the ﬁrst half of the experiment, when it is assumed that the sample
supply is not the limiting factor, is used to calculate an activation energy of 23
kcal/mol. The literature value for the heat of vaporization of palmitic acid is
17.6 kcal/mol.“2 In this case, the compound does not undergo any thermal
degradation and, therefore, essentially all of the sample can be desorbed intact
and monitored by the production of [M]I(+ ions. No competing degradation
reactions are present. This leads to a more accurate calculation of activation
energy for the desorption process, based on the assumption that the formation
of the [M]K+ ion is a function of the rate of vaporization of palmitic acid. Other
plots of abundance vs. UT, and the activation energy determinations for
additional compounds are provided in Appendix A for future reference.

The formation of K+ adduct ions depends on several factors that may not
allow for this simpliﬁed analysis. The availability of K+ ions, the amount of
sample present, the competitive thermal degradation reactions, and the
desorption processes all affect the production of K" adducts. A few of these
possible reactions are shown below in equations 2.6-2.10 where M is the intact
molecule in the solid phase (s) or the gas phase (g), and D is a neutral product
of thermal degradation in the condensed phase (s) or the gas phase (g).
Reaction 2.6 is the desorption of the intact analyte molecule into the gas phase.
This reaction has a rate constant associated with it and the activation energy
for this process is the heat of vaporization of the analyte, M. There is also a
rate constant for the formation of the K+ adduct ion of this gas-phase neutral
through reaction 2.7. This reaction is assumed to be rapid once the analyte
molecule is present in the gas phase and is not considered the rate-limiting
step. The other option for the analyte species present on the surface is

thermal degradation, represented by reaction 2.8. There is a rate constant

59

5000 '-

4000 "

3000 -

 

Abundance

N

O

8
l

 

1000-

 

 

1/Tx1000

Figure 2.26 Abundance of the [M]K’ ion of palmitic acid vs. 1rr (K).
Ea calculated from the slope for 1/T > 2.7 = 23 kcal/mol.

60

Ms -> M8 2.6
Mg+ K‘? -—> MK+ 2.7
Ms —) D5 2.8
D5 -9 D8 2.9
Dg+K+ —) DK+ 2.10
associated with each degradation process. The vaporization of these

degradation products each have a corresponding rate constant. Again, it is
assumed that the K+ adduct formation of these gas phase species, reaction 2.10.
is not the rate-limiting reaction in the formation of [D]K+. It is unclear
whether the degradation process or the vaporization of the degradation
products is the rate limiting step in observing the K+ adduct ions of these
products. Therefore, the activation energies for the formation of the gas
phase degradation products could be dependent on either the degradation
process or the vaporization process.

The most important factor in the production of K+ adduct ions of the
intact analyte molecule and thermal degradation products may be the finite
sample supply on the sample ﬁlament. If ﬁrst order kinetics is assumed for
the depletion of sample with time, the following derivation may apply. A0 is
the total amount of sample initially applied to the ﬁlament (t=0) and At is the
remaining amount of sample at time t. This analysis is for neutral species and
the processes involved in the thermally induced conversion from the
condensed phase to the gas phase. K+IDS is a very useful technique in these
analyses as the abundance of K+ adducts, [M]K+ and [D]K+, reflect the
abundance of the species M and D in the gas phase and thus provides a means
of monitoring the vaporization of the analyte. From this derivation, the
activation energy can be calculated from the slope of the plot of ln(ln(Ao/At)
versus 1/T. This assumes that ln(t) is negligible as a variable due to the short
reaction time of the K+IDS experiment (less than one minute). The difﬁculty
with this derivation is in determining the amount of sample remaining as the
analysis proceeds. This is especially true when the sample undergoes much

thermal degradation and, thus, many pathways of sample depletion are

61

a ... mun 2-11
do

which is equivalent to:

k = 1.1 n "—0 2.12
I A;
Ea . .
k = A ex 727 Arrhenius equation 2.4

llni‘i = Acxp[-.Ei] 2.13

r A, RT

“{1} + In 1 12 = lnA - [3] 2.14
I A; RT

lnl {‘3 =-[E—al+(lnA +lnt) 2.15
A, R

present. A simple compound that does not decompose during the K+IDS
analysis is a good starting point for these activation energy determinations.
One such compound is palmitic acid which has a K+IDS mass spectrum with
only one ion, the [M]K+ ion at m/z 295, indicating that the sample does not
decompose when rapidly heated. This simpliﬁes the determination of the ratio
of Ao/At. Assuming complete vaporization of the sample, A0 is calculated as the
sum of the abundance of the [M]K‘*’ ion for the entire experiment. A;, the
amount of sample left at time t, is calculated by the total [M]K+ ion observed
during the experiment, A0, minus the sum of the [MlK‘l' ion observed until time
t. The plot of ln(ln(Ao/At» versus III‘ for palmitic acid is shown in Figure
2.27. This data are from the same experiment as was used for the plot in Figure
2.26. The slope of the plot from the best ﬁt line of the entire data set in Figure
2.27 is used to calculate an activation energy of 15 kcal/mol. The plot seems to
have two regions, or two slopes. If the activation energy is calculated based
only on the ﬁrst half of the experiment (1/1‘ > 0.0027). it would have a value of
25 kcal/mol. The activation energy calculated from the second half of the
experiment (1/T < 0.0027) is 5 kcal/mol. The overall activation energy
calculated as 15 kcal/mol is closest to the literature value of 17.6 heal/mm.“2

Since this method of calculation takes into account the depletion of the sample

62

, ln(ln(Ao/At))

 

 

 

1/Tx1000

Figure 2.27 Arrhenius plot of ln(ln(Ao/At)) vs. 1/T for palmitic
acid. Ea calculated from the slope = 12 kcal/mol.

63

with time, the entire plot should be representative of the activation energy
and should be used in the calculation. Duplicate activation energy
determinations for the vaporization of palmitic acid based on other K+IDS
experimental results are presented in Appendix A.

This method of activation energy determination may be applied to
sucrose. In this case, the analyte does undergo thermal degradation which
complicates the calculations. Initially, these thermal degradation products are
ignored and A0 is calculated as the sum of the abundance of only the [M]K+ ion
of sucrose observed over the entire experiment. The AI is calculated the same
way as for palmitic acid, as if no analyte decomposition is occurring. The plot
of ln(ln(Ao/Al» vs. VT for sucrose is shown in Figure 2.28. The data points
are more linear than for the plots of the palmitic acid data. It appears that
these assumptions for the calculation of A0 are valid based on the appearance
of the plot. The activation energy is calculated as 28 kcal/mol which is close to
the reported experimental activation energy of 38.9 kcal/mol for the
vaporization of sucrose and formation of the NH4+ adduct ion.“o These
preliminary results of activation energy determinations lend some validity to
the proposed method of analysis. The calculated activation energies are the
same order of magnitude as reported values.“"»"’2 More replicate
determinations of activation energies need to be performed in order to
determine the reproducibility and validity of these results.

Methionine-enkephalin (met-enkephalin) is a pentapeptide that
experiences much thermal degradation during a K+IDS analysis. The structure
of this peptide is shown in Figure 2.29. The fragmentation pathways that
produce the most abundant thermal degradation products as observed by K+
adduct formation are labeled in Figure 2.29. The [M]I(+ ion of this peptide is
not observed as an abundant adduct ion in the K+IDS mass spectrum.
Therefore, only the activation energies for the production of the gas phase
thermal degradation products that are observed as the most intense peaks
representative of K+ adduct ions can be calculated. Again, these activation
energies could either be for the thermal degradation of the analyte or the
vaporization of these thermal degradation products, whichever is the limiting
step in being observed as K+ adduct ions. The simplest approach would be to
assume there is no interdependence of the thermal degradation processes and
that each adduct ion can be considered independently. If this is pursued. then
A0 and A; can be determined for each adduct ion. The results of such

 

 

ln(ln(Ao/At)

64

 

 

 

21

0'

-2-

_4..

-6 a , . , . , . . a . 4'—
1.5 1.6 1.7 1.8 1.9 2.0 2.1

1/T x1000
Figure 2.28 Arrhenius plot of ln(ln(Ao/At)) vs. 1/T for the [M]K+
ion of sucrose. Ea calculated = 28 kcal/mol.

65

T”?
mO-cw-rw
0‘9
.----1 ......
i“
m/z 259
x.‘ K"' (In-‘2
: 0‘0
‘x--.: ......... '. .....
g I
y M-I
H I
f“
O=T
l"
Q—CHF‘E”
O=(|2
M-l

--‘

'unn I‘--‘--
i
I o

1""

I

I
I
I
I

\

\
K+ m/z 449

 

\
m/z 392

MW = 573
[M]K+= 612

Figure 2.29 Structure of methionine-enkephalin and the

thermal degradations observed as

K+adduct ions.

66

2 -
-{.\
E3 \
O ‘\
o‘-
o - . +
o a + ln(ln(Ao/At)) 392
°. ‘ o ln(ln(Ao/At» 449
+ + ln(ln(Ao/At)) 259
a ’2‘ . +
g ta
0
S,
E O a
E 47 T
.6-
'8 ‘ I V I ' I ‘ I ' I ‘ I

 

 

1.6 1.7 1.8 1.9 2.0 2.1 2.2

1/T x1000

Figure 2.30 Arrhenius plot of In(ln(Ao/At)) vs. VT for three
fragment K+ adduct ions of met-enkephalin.

Ea (259) = 27 kcal/mol, Ea (392) = 37 kcal/mol,

Ea (449) = 52 kcal/mol.

67

calculations are plotted in Figure 2.30 for K+ adduct ions of three thermal
degradation products. As expected, the activation energy for the production of
gas phase thermal degradation products of met-enkephalin does increase with
molecular weight of the decomposition product. The values of the aciivation
energies calculated for these processes do seem to be reasonable based on the
previous activation energy determinations for vaporization of thermally
labile compounds. However, repetitive experiments of met-enkephalin
indicate that these values are not reproducible within experimental errors.
Additional Arrhenius plots of K+IDS experiments are provided in Appendix A.
The data from one experiment were used to calculate the activation energy for
the formation of one of the fragment K+ adduct ions (observed as a peak at m/z
259) as being 4 kcal/mol. This value is much below the experimentally
acceptable deviation and points to a possible problem in the method of
calculation. The assumptions used for these calculations, mainly that the
thermal degradation processes are independent of each other, are probably
over simpliﬁcations of the processes involved in the K+IDS technique. This
does, however, provide a starting point for the determination of activation
energies based on K+IDS experimental data. These preliminary results indicate
that these determinations are possible with the K+IDS technique, and that

further studies and modiﬁcations are needed.

4. SUMMARY

Continuation of the temperature studies would benefit from further
alterations in the experimental design. First of all, the computerization of the
temperature measurement would be a signiﬁcant improvement. The output of
the OMEGA temperature monitor could either be connected to a separate
computer or directly to the HP computer of the instrument for a digital
readout of the temperature versus time. Ideally, the temperature measurement
would be synchronized with the scanning of the mass spectrometer which
would eliminate many of the sources of error currently present. This would
allow for more accurate temperature determinations at specific times instead
of the current manual measurement of the chart recorder temperature plots.

Another aspect of this study which needs further testing is the variable
adduct ion signal that plagued many of the preliminary results obtained. A

more detailed examination of this problem may lead to ways to reduce some of

68

this signal ion ﬂuctuation. Possible sources of this varying signal are the K”
emitter and the production of K+ ions. In the past, this K+ signal was
relatively constant and very strong. However, when the sample adduct ion
ﬂuctuation was troublesome, the K+ signal also appeared to ﬂuctuate more
than usual. If the source of K+ ions ﬂuctuated it would follow that the K“
adduct ion production and signal would also ﬂuctuate. Whether this K+
variation was the result of the probe construction or the performance of the
instrument is unclear.

One way to avoid this ﬂuctuating signal may be to use selected ion
monitoring (SIM) to monitor the production of just the [M]K+ ion and one or
two thermal degradation products. This data collection mode allows for more
signal averaging and faster scanning capabilities than the full mass range
scanning procedure typically used. However, one disadvantage to this mode. is
the limited number of adduct ions that can be observed simultaneously. This
would inhibit the ability to observe within run variations in the production of
different adduct ions at different temperatures and times. The main advantage
to using SIM is the increased data collection that could be obtained over the
desorption proﬁle of the sample. Since the K+IDS experiment is typically very
fast with analyte ion signals lasting less than one minute, this higher
sampling rate could greatly improve the results of determining the
temperatures at which desorption and thermal degradation processes of

thermally labile compounds occur.

CHAPTER 3. APPLICATIONS OF THE K+IDS TECHNIQUE

1. INTRODUCTION

Previous researchers have applied the K+IDS technique to a wide
variety of compounds. Bombick showed the utility of K+IDS for the analysis of
saccharides, peptides, steroids, antibiotics, and some salts. The K+IDS
technique also has proven to be very effective for the analysis of polymers in
determining the average molecular weight, oligomer distribution, and
impurities present.43 Kassel extended the applicability of Na+IDS to the
determination of organic acids in complex mixtures and metabolic profiling.44
(Na"’IDS is the same as K+IDS except NazO is used in the aluminosilicate mixture
instead of K20 and, therefore, thermally produces Na+ ions. Na"’IDS seems to be
more sensitive for some compounds including the organic acids and induces
more fragmentation than K+ attachment does in other types of compounds). A
procedure was developed based on the Na+le technique for the rapid
screening of urine samples for the detection of organic acidemias. These are
physiological disease states caused by errors in certain metabolic pathways
that can be characterized by elevated concentrations of one or more organic
acids in the body's plasma or urine. Na+IDS provided a simple. quick method
for analyzing urine samples for these diseases.44 Even in these early stages of
development, the K+IDSINa+lDS technique has proven to be a very useful DI

technique for a wide variety of compounds.
2. K+IDS MASS SPECTROMETRIC ANALYSIS OF CARDIAC GLYCOSIDES

DI techniques, in general, have been applied to a wide range of
thermally labile compounds especially in the ﬁeld of biochemistry where most
compounds of interest contain a large number of functional groups and are
considered thermally labile. One group of compounds that has received much
attention in the DI ﬁeld is the cardiac glycosides. These compounds contain a
sugar portion and a steroid moiety. The cardiac glycosides are used as
therapeutic agents for the treatment of congestive heart failure and certain
heart arrythmias. These compounds have been studied by almost every mass
spectrometric ionization technique.” FAB is the most widely used DI

technique used to date for the analysis of thermally labile compounds.

69

70

Therefore. this technique will serve as a standard for comparison of the K+IDS
technique.

A reported FAB mass spectrum of digitonin is shown in Figure 3.1 and
serves as a good example of the mass spectra obtained with FAB.“6 The
structure of digitonin. which is a cardiac glycoside with a molecular weight of
1228 daltons, is shown in Figure 3.2 This mass spectrum is characteristic of the
FAB technique with glycerol used as the matrix. In the molecular weight
range, there are at least three peaks representative of the intact molecule; the
[M]H"'ion, the [M]Na+ ion, and the [M+glycerol]H+ ion. This multiple
representation often complicates the molecular weight determination. Other
protonated ions representative of the analyte from cleavages of the glycosidic
bonds are observed in the middle mass range. The low-mass range is cluttered
with many ions representative of the liquid matrix, glycerol. This makes it
very difficult, if not impossible, to obtain information about the sample from
this low-mass range of the mass spectrum. In summary, FAB mass spectra are
often characterized by decreasing abundance of sample ions with increasing
molecular weight (or m/z value), many fragment ions representative of the
sample, protonated and glycerol adduct ions representative of the intact
molecule, and a complicated low-mass region.

In contrast, the K+IDS mass spectrum of digoxin is shown in Figure 3.3.
Digoxin is a cardiac glycoside similar to digitonin with fewer sugar residues
and, therefore, has a molecular weight (780 daltons) within the mass range of
the quadrupole mass spectrometer used for K+IDS. This mass spectrum is
characteristic of K+IDS mass spectra in that the peak representing the [M]K+
ion is the most intense peak in the mass spectrum and the majority of the
peaks observed represent the K+ adduct ions of the intact molecule and neutral
thermal degradation products of the analyte. The fragmentations are simple
1,2-eliminations about the glycosidic bonds. The mass spectrum is
representative of the sample with no interference from addition of a solvent
or liquid matrix. This comparison of FAB and K+IDS for the analysis of cardiac
glycosides shows the utility of the K+IDS technique for the analysis of
thermally labile compounds and some of its advantages over the FAB
technique. This does not mean that it is expected for K+IDS to replace FAB, but
rather that the K+IDS technique does have some advantages to offer for the
analysis of thermally labile compounds. K+IDS is a simple technique that

produces relatively simple mass spectra that are representative of the sample.

71

 

me.c_c2_o_n Co 526QO mme EmEEmnEon 52m 89 85:33. mm 9:9...

 

 

cA—Osgo O IV I X
LIZ + 3: Cap cg + .< a. 4 LI” + .( l < l y 68h l/ O—O

.AIn+ut.(I2_

 

 

 

 

«9
pp: 03‘

/

. AI + w200>40(|OI.

'v'v

Vvvvv

 

% ‘aouepunqe menu

72

n”IO

0

IO

fo

.cEQEU Co 93025 Nd 9:9”.

mmmw u >22

OI

IO

IO

xoaxo

 

 

 

 

OI

Iofo

IO
IO OI

Iofo

IO
IO OI

73

 

 

.Exooﬁ Co :55QO $2. wary. md 9:9".

N\E

com oov

 

 

 

 

arm

 

 

mam

 

 

mmm

 

mac

 

 

 

man I

 

 

AJJSNBLNI BALLV'IBU

74

The K+IDS mass spectra are easy to interpret based on one mechanism of
fragmentation, the 1.2-elimination mechanism.

Due to these encouraging preliminary results, K+IDS was used to analyze
several cardiac glycosides to allow for the evaluation of K+IDS in the context of
other DI techniques. The results of the K+IDS mass spectrometric analysis of
the cardiac glycosides have been published and this article is provided in
Appendix B. The results of this study show the strengths of K+IDS for the
analysis of the cardiac glycosides and related compounds. The mass spectra are
clean and only contain K+ adduct ions representative of the sample. The base
peak in the mass spectrum represents the [M]K+ ion which provides molecular
weight information. Also. thermal degradation products, primarily from
fragmentation of the glycosidic bonds between the sugars. are observed as K"
adduct ions. These ions provide structural information. All of the
fragmentations observed can be explained by the 1.2-elimination mechanism
(see Figure 2.2). Therefore, another useful application of the K+IDS technique

to these cardiac glycosides has been established.

3. UTILITY OF THE K+IDS TECHNIQUE FOR MIXTURE ANALYSIS

One interesting aspect of the K+IDS technique is the utility for mixture
analysis. Bombick was able to accurately determine the average molecular
weight of polymers that are composed of mixtures of oligomers.‘43 Kassel
successfully performed numerous studies utilizing Na‘l’IDS for analyzing the
organic acids in urine samples with minimal sample preparation.“ However,
quantitative determinations of individual components within mixtures has
proven difﬁcult with the K+IDS technique, as with other Dl techniques. There
seems to exist a mass dependence on the extent of desorption of free fatty acids.
This is discussed in more detail elsewhere.41 There are several characteristics
of the K+IDS technique that contribute to its applicability for mixture analysis.
K" ions attach to neutral gas-phase species during the K+IDS technique and do
not induce any fragmentation. Therefore, K+ adduct formation will not
complicate the mass spectrum, but merely provide a good representation of the
gas-phase neutral species representative of the sample. K+IDS produces
primarily [M]K+ ions of most analyte species with some K+ adduct ion
formation of neutral thermal degradation products. Thus, the K+IDS mass

spectra of mixtures provide molecular weight information of the components.

75

Since the K+IDS technique requires no matrix, the K+IDS mass spectra of
mixtures are less complex than the mass spectra obtained with other DI
techniques, such as FAB, which do require liquid matrices. This simpliﬁes the
analysis of the K+IDS mass spectra of mixtures.

The K+IDS analysis of a mixture that contains different types of
compounds also provides very encouraging results as to the utility of K+IDS for
mixture analysis. A mixture of fatty acids, low molecular weight organic acids,
and steroids provides an interesting challenge. In order to analyze this
mixture by more traditional E1 or GC/MS analysis, several different sample
preparation techniques and derivatizations would be necesary. An equimolar
mixture of eight compounds representing these three groups was made and
analyzed by Na+IDS. This mixture contained three organic acids: salicylic acid
(MW = 138), cinnamic acid (MW = 148) and mandelic acid (MW = 152); three
saturated fatty acids: myristic acid (MW = 228), palmitic acid (MW = 256) and
stearic acid (MW = 284); and two steroids: 58-pregnen-3B-ol-20-one (MW = 316)
and cholesterol (MW = 386). The Na+IDS mass spectra of each of these eight
compounds was obtained individually for identification of the components in
the Na+IDS mass spectrum of the mixture. All of the Na+IDS mass spectra of the
individual components, except for mandelic acid, represent only one abundant
ion, the [M]Na"' ion. The Na+IDS mass spectrum of mandelic acid has a base
peak (at m/z 157) corresponding to the [M-H20]Na+ ion in addition to a strong
peak at m/z 175 ([M]Na“'). One uliter of the equimolar mixture dissolved in
methanol was applied to the sample support wire of the Na+IDS probe and the
solvent was evaporated. Approximately five umoles of each component were
present on the probe tip at the beginning of each analysis. The TIC mass
chromatogram and the [M]Na"‘ mass chromatograms of the eight components
from the Na+IDS mass spectrometric analysis of the mixture are shown in
Figure 3.4. There is a considerable difference in the maximum abundance of
each component as denoted on the chromatograms in Figure 3.4. For instance,
the maximum abundance of the [M]Na+ ion of myristic acid is labeled as 27,136,
whereas the maximum abundance of the [M]Na+ ion of cinnamic acid is 6,112.
These differences could be the result of differences in the efficiency of
desorption of the intact species and adduct ion formation. Another reason for
these discrepancies in abundance is probably due to differences in the onset
of desorption and the overlap of this process with the emission of Na+ ions.

This can be seen in the plot of the Na+IDS mass spectra versus scan number of

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 100%: 16736. 409
E 100%: 11336. 339
f 100%: 23168. 307
q

5 100%: 26640. 279
1

f 100%: 27136. 251
q

f 100%: 286}[“\\\‘\\‘—F 175

+v -, -~Jr- -, if

E 100%: 6112 171
E 100%: 2684 161
7v 1 r-~ -, “T ~
E 100%: 11796:///7\\\\\“-~‘¥ TlC
0 ' ' {0 V ' 7' 30 ' V 40

Cholesterol

SB-Pregnen-

38-ol-20-0ne

Stearic Acid

Palmitic Acid

Myristic Acid

Mandelic Acid

Cinnamic Acid .

Salicylic Acid

Figure 3.4 Mass chromatograms of the [MJNa+ ions of the eight individual

components present in the mixture.

77

this eight-component mixture shown in Figure 3.5. The x-axis is the m/z scale,
the y-axis is scan number, and the z-axis is abundance. The low-mass ions are
from the organic acids. Notice that these ions do not increase in abundance
with scan number for the ﬁrst few scans as is observed for all the other ions at
higher m/z values. This suggests that the formation of these ions is limited by
the production of Na+ ions. The neutral molecules of these organic acids
(salicylic, cinnamic, and mandelic acids) are probably desorbing before Na+
ions are being produced from the Na+ emitter. Therefore, when they are ﬁrst
observed, they are already at, or beyond, their maximum desorption. There is a
slight delay in the observance of the [M]Na+ adduct ions of the steroids of m/z
339 and 409 with respect to the observance of the fatty acids of lower m/z
values. This is in agreement with the discussions presented in Chapter 2 that
there is an increase in heat of vaporization with increasing molecular weight.
Therefore, the lower molecular weight fatty acids desorb prior to the higher
molecular weight steroids and are observed by their corresponding Na+
adducts in the Na+le mass spectrum before the [M]Na"’ adducts of the steroids.
The Na+IDS mass spectrum obtained from averaging scans 12-18 of this
analysis of the mixture is shown in Figure 3.6. The [M]Na"’ adduct ions of the
eight components are all observed. The m/z values of these [M]Na"’ ions are
provided with their corresponding analyte names in Figure 3.6. The least
abundant molecular adduct ions in this mass spectrum are from the three
lower molecular weight organic acids. These components may be desorbing
before the onset of K+ emission and, thus, part of the sample is undetected. It
is very interesting to note that all of the compounds. though differing in
structure. are desorbed intact with the Na+IDS technique and detected. Little
thermal degradation is observed for any of these analytes, except mandelic
acid which does exhibit dehydration. No evidence exists for any reaction
between these eight components in the mixture, either on the surface or in
the gas phase. As previously mentioned. K"' and Na+ ions primarily add to the
neutral molecule without inducing any fragmentation. These results further
support the utility of the Na+IDS/K"'IDS technique for mixture analysis. Any
reaction products observed in mixtures would probably be the result of

interactions in the sample prior to analysis and are not the result of rapid

heating.

+9N[|OJGIS€|OL|Q] 1

+eN[euo-oz-|o-qe-ueuﬁaid-gg] .

.2111be 0112619] 1

+9NID!0V outwled] 1

+enlptov 011911le 1

+9Nlpl0v ouepuew]
+9Nlpl0‘v' atweuurol

 

Z+BNIPIOV 011401198]

+eulo H- 010v ouepuew]

 

\\.

 

78

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

41:
<0 8
N01
NP (3
NO
NO)
"Q
PF
FLO
F?
1 < F09
FN
4 < 1.1—
..1 v-O
-1. q-o
——_ '0
t ._..d V
.1
1 ‘ l'
.4
.1 L
.4
d
and .-
11 1 ___L‘
a 1 __“ L
d 1 l—l
.1 l
.41 —l 1 ‘ 11°)
1‘—_—‘
J —-‘ ‘ d—d 7
d i—d l l d
_‘ -—1—‘ i _
1
q“ I d -
.——.1 1
1
P
1 i—I—I
t o
1 J L._“’
, N
‘ 1
1 1
\ .' r
1 1
\V 1‘ 1
\‘ 1 d 1 P
1 ‘d .1
\\‘ 1L":
a: II at 1-
Eat
E K o
_I l ‘0
-a-) 1-
[I

Figure 3.5 Plot of the Na+ IDS mass spectra vs. scan number for the analysis of

the eight component mixture.

79

mo... .298.on
8m ace-8-6- 2-5899 mm
Ba 22 262m
and 22 2222
+2.2. .2. 65858

NE

EN
mm:
C...
5..

+625.

22 6:222
22 26222
22 252.20

22 222:3

:2. .95an00

o: ow». oov own com ovm own com 0mm com ovw om_w com om:

 

 

mov mmm

 

 

mom

 

mum

22966 new 522 Em. .mgom 2590 9.59:8
2258 E23800 £96 on. .o Egon-am mmmE mo. +mz 63896 9: md 659”.

cap 0:.

F1 1- — n11p1 C1

 

 

:mm

PEEEFPIFFE. ________L

I 1.! 14 111“—

m:

:L

 

.

For

E: r

 

 

Misueim BARBIGH

80

4. FINAL COMMENTS

K+IDS and Na+IDS have been shown to be useful Dl techniques for the
analysis of a wide range of compounds including the cardiac glycosides. All of
the ions observed in the K+IDS mass spectra are K+ adduct ions of the intact
molecule or neutral thermal degradation products representative of the
analyte, produced by the 1.2-elimination mechanism. Since K+IDS requires no
matrix, all of the ions observed are representative of the sample. Thus, K+IDS
mass spectra are typically cleaner with less chemical noise than the mass
spectra produced by the more traditional DI techniques such as FAB.

In addition, K+IDS provides an easy way to analyze mixtures as there is
no matrix interference to complicate the already complex mass spectrum from
the sample. Also, the mass spectrum is representative of the mixture with no
mixing or interactions occurring as a result of the ionization process.
Molecular weight information of the components in the mixture is usually
obtained due to the intense peaks representing the [M]K"’ ions that are

characteristic of the K+IDS technique.

CHAPTER 4. Li+IDS

1. INTRODUCTION

Often the K+IDS analysis of thermally labile compounds produces only
[M]K'*' ions and, thus, only molecular weight information is obtained. Ideally,
mass spectrometric analyses provide both molecular weight and structural
information concerning the analyte of interest. One way of adjusting the
K+IDS technique to obtain structural information is by altering the
temperature and heating rate variables to change the rate of desorption
versus the rate of thermal degradation. This was discussed in detail in chapter
2. Another variable that may be used to increase fragmentation of some
compounds is the choice of the metal ion used with this DI technique. K+
merely attaches to neutrals in the gas phase with all observed fragmentation
occurring prior to 10" attachment.

There are many possible metal ions that can be used for the
cationization of thermally labile compounds that may induce fragmentation
and thus provide more structural information. Blewett and Jones reported
thermionic emitters that could produce Mg", Ca+, Ga", In", Rb", and Al“ in
addition to the alkali ion emitters.28 More recently. sources of ions such as Cu+
and Ag+ have been reported.47 Extensive research has been performed on the
gas-phase reactivity of transition metal ions such as Fe“, Co+, and Ni+ with
molecules containing functional groups and for alkanes.48 Many of these
metal ion reactions are speciﬁc enough to be useful for structural
determination. Transition metal ions have been reportedly used for chemical
ionization analyses.49 The main difﬁculty with transition metal ions is that
they have the higher ionization energies than alkali metals and, therefore,
the transition metal ions are more difficult to ionize thermally. It may be
difﬁcult to create an emitter that produces a high enough flux of gas-phase
transition metal ions for the “M+IDS” technique. Some encouragement.
however, has come with the report of thermally generated Cr"' ions.so
Volkening and Heumann used borosilicate matrices containing the transition
metals as a source of the transition metal ions.50 If this technique could be
used to produce transition metal ions such as Co"‘ or Fe", it would greatly
expand the realm of the K+lDS approach. Normal K+IDS could be used to

determine molecular weight information and Co+IDS could be used to obtain

81

82

information concerning the structure of the analyte and functional groups
present.

Li+ has been shown to be a reactive ion and induce fragmentation in
some organic molecules.51 Allison and Ridge suggested a mechanism by
which alkali ions may react with organic molecules such as alkyl halides and
alcohols.52 For example, Li+ reacts with small alcohols such as t—butanol to
form I.i(C4I-lg)+ and Li(H20)+. The Li+ ion induces fragmentation of this
alcohol to produce the corresponding oleﬁn and H20. Na+ and K+ only form
adduct ions with t-butanol with no fragmentation observed. Li+IDS may
provide a way of obtaining structural information by inducing fragmentation
of compounds that only desorb intact and. therefore. are only observed as
[M]K"’ ions in the traditional K+IDS technique.

2. EXPERIMENTAL CONSIDERATIONS FOR
PRODUCING A Li+ THERMIONIC EMITTER

The construction of a Li+ thermionic emitter follows the design of the
K+IDS emitter. The aluminosilicate mixture is composed of Li20:A1203:28i02.
This mixture is more difﬁcult to formulate into a bead than is the K“ mixture.
Two methods of making a 16" bead on a rhenium wire loop were discussed in
Chapter 2. One method is to use an oxygen/acetylene torch to melt the alkali
aluminosilicate on a platinum crucible. The rhenium wire 100p is then dipped
into the molten mixture to make a head. The melting temperature of the Li"
aluminosilicate is higher than the K+ aluminosilicate mixture. Therefore, the
lithium mixture is harder to melt with the oxygen/acetylene torch without
also melting the platinum crucible used as the support. This proves to be a
very tedious way of forming a Li+ emitter. The other procedure for making an
emitter bead is with an acetone slurry of the aluminosilicate mixture. Some of
the slurry is deposited onto the rhenium wire loop and the solvent is allowed to
evaporate. Then the mixture is hardened in the flame of a bunsen burner.
This slurry method of bead making is also more difﬁcult with the Li" mixture.
When the dry powder is inserted into the bunsen burner flame, much popping
occurs probably due to the formation of N02 gas (from LiNO3). This popping
causes most of the aluminosilicate to fall off the wire. The best way to deal
with this problem is to carefully melt the aluminosilicate mixture several

times with the oxygen/acetylene torch, re-grinding between each melt. This

83

procedure will reduce most, if not all, of the LiNO3 to Li20. For this procedure,
the aluminosilicate does not have to be molten enough to submerge the wire
in, but just melted enough for LiNO3 to decompose. Therefore, it is easy to
accomplish without melting the platinum. This acetone slurry method with
premelted aluminosilicate powder proved to be the easiest way to make Li+
beads.

Once the Li+ emitter is made, the conditioning process is performed
inside the ion source of the mass spectrometer as usual. It takes much longer
to condition the Li"’ bead to achieve a strong Li+ signal than it does for the K+
bead to achieve a K+ signal. This is primarily due to impurities of the other
alkali metals present in the LiNO3 and their relative ionization energies. The
ionization energies for the alkali metals are as follows: Li is 5.392 eV. Na is
5.139 eV. K is 4.341 eV. Rb is 4.4177 eV. and Cs is 3.894 W.” Therefore, any
alkali metal impurities with lower ionization energies in the Li glass will be
preferentially ionized, before Li.

The ﬁrst Li+ bead was conditioned for a long period of time in order to
reduce the signals from other alkali ions present as impurities in the Li+ bead.
This extensive conditioning to produce a Li+ emitter is similar to the process
reported gy Weber and Cordes for the conditioning of a Na source to eliminate
K+ impurities by heating the source at 1000°C for over six hours-54 After ﬁve
hours of conditioning, the Li+ emitter still produced 100 times more Na+ ions
than Li+ ions. Varying the bias voltage applied to the tip of the bead did affect
the ratio of the alkali ions observed. When the bias voltage was less than 1.0V,
more Li+ ions were observed than Na'l‘ ions. The bias voltage also affected the
Li+lK+ ratio in a similar manner. However, the absolute abundance of these
impurity alkali ions was still greater than desired. Another disturbing
observation was the strong appearance of Na" and 10' ions when current was
ﬁrst applied to a cold Li+ bead. Only after heating for approximately 30-60
seconds did the Li+ signal become the dominant metal ion observed. A second
Li+ head was initially conditioned at a low-current (<2.0A) for approximately
three hours with the goal of preferentially ionizing Na and K, which have
lower ionization energies than Li. After this period of low current
conditioning, the current and bias voltage were adjusted to obtain an optimum
Li+ signal. In this case. the Li+ signal was much larger than the impurity K4“
and Na+ signals. This low-current conditioning appears to be a good way of

conditioning the Li+ bead to decrease the presence of the impurities.

84

3. Li+IDS MASS SPECTRA

This offshoot of the K+IDS project was performed during the early
stages of the development of the two-filament probe tip prior to the
temperature studies and characterization of the probe tip design. Therefore,
the data presented here should only be considered as preliminary results to
show that Li+ attachment to intact thermally labile molecules does occur and to
suggest that this Li+ attachment may lead to increased fragmentation. These
results cannot be considered conclusive of the reactive nature of Li"’ ions (or
nonreactive) upon attachment. In Chapter 2, evidence was presented that
variations in the probe tip conﬁguration and design can result in differences
in the mass spectra obtained. During these Li+IDS studies, little attention was
given to these design details. Therefore. some of the differences in the Li+lDS
mass spectra compared to the Na+IDS and K+IDS mass spectra could be the
result of the probe tip design instead of the reactive nature of Li+ compared to
N a+ and K+, however, the latter may still be a factor.

The polymer propylene glycol 725 (PPG 725) is one compound that does
show some differences in the Li+IDS mass spectra compared to the Na+IDS mass
spectra. The Na+IDS mass spectrum of PPG 725 is shown in Figure 4.1 and the
Li+IDS mass spectrum of PPG 725 is shown in Figure 4.2. Both of these mass
spectra are obtained by averaging over the entire experiment. The two mass
spectra are very similar in the high—mass range where the ions are the
corresponding Na+ and Li" adduct ions, respectively, of the oligomers
represented by H[CH2CH2CH20]nOH where n=8 to 16. The m/z values of the
oligomer Na" and Li" adduct ions as a function of n are given in Table 4.1. The
low-mass range of the two mass spectra in Figure 4.1 and Figure 4.2 show
considerable differences in the abundance of dehydration products observed.
The metal adduct ions of the low molecular weight oligomers (n=1 to 7) are of
very low abundance in the Na+IDS mass spectrum in Figure 4.1 compared to
the Li+IDS mass spectrum in Figure 4.2. Also, dehydration products of the low
molecular weight oligomers are observed as Li+ adduct ions with a large
relative abundance. It is possible that the Li+ attachment is inducing some of
this observed dehydration. Figure 4.3 shows a plot of the Li+ mass spectra of
PPG 725 as a function of scan number. This is a good example of the changes

in the Li+IDS mass Spectra with time during a single analysis. As the

2:68.833 2.2.3 35 .35 892623 man can. .o 62.6QO 89: mo. +m2 I: 359...

 

85

 

 

 

 

 

 

as
am: am: ado mom am: can mam mm: mm

__L1P.1_._1_1. +1L_-L1_ .. . p1m_._..-F....._1L_-.1_L_1LFF

11 1. 4 u a -. -. - q a. .41.:4 241.44-1J1

mom mom
::a .
mm: new
:N: 1
men
emu mew

 

 

 

 

Misuelm 84191813

86

2:68.898 2:5 35 85 89262... mun odd :0 2.25QO 22:: mo. +3 me 359“.

 

 

 

 

 

 

 

 

 

 

 

N\E
mFoF arm arm mph are arm Gr? mwm ¢FN mz.
__-__:_ .:- F: :.- .....:-:_:1.1.P_1:-b+:_mPL._ 1
GP?

05” ﬁrm 1.

5: new 313 am:
:8 2 Ba: -

m th mam Fﬁw
mmv NNN
hvm I
hmwmhh moo
j
FNh moo

 

 

 

 

 

MSUGIUI aAﬂelatl

87

2:68.598 29% m .9 .883: 58 «.29? mwm and .o 9.8% mmmE met... 34 2:9”.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NE
92:. 0mm 0mm own 000 00m 09: 0mm cam 00—.
/_D th—yb .hP—Dlh I—D .- I—Ith—b .- h—P Pb .— h-F-D—bbbP
o\
1N 1 1. i N
0.. rVN 1 11 11 -1 -- 1-1.11.a.111u_....-1. Teal-1M1!
/\ t . 111‘ 1‘ 44¢
K I 1 I 1 - ‘1 1.4 11‘ I‘dl“ ‘11“ I
IN 1 H 4 u u1 1‘ 1d 1.-1 u N
ww r\ 1 _ u a _ n- -1 1. -._-11L
L. 1 - _ J 1 a A . 11-1.111l414411ﬂ
/11 .1 .. . x . n. .1 -1 m. .12.“:
a :mom I‘ —4 1hd - ——:d m ‘1 I1 111111 11111 144N114] .1
MN pN _ : _ A J...— u . . 1 1 . u. .i .
1N HF nu an J 1.. a 1 4 .1 41.114 «C-
l? a a. . _ - : _ .1 .1 1.. 411 .1. 111
K a . . u .1 A : 4 1 1 - 1 - 11 .1111 .mm
1?

 

 

 

 

 

 

 

 

 

 

 

88

experiment progresses, the temperature of the sample wire ﬁlament increases.
and the molecular weight of the observed oligomer adduct ion increases. This

is consistent with the findings of the temperature studies in Chapter 2.

Table 4.1 The m/z values of the Na+ and Li+ adduct ions of the oligomers of PPG
725, H[CH2CH2CH20]DOH.

 

 

m/z value of m/z value of
n [H[CHzCHzCH20]n01-1]Na+ [H[CH2_C_H_2CH20]nOH]Li+
1 138 122
2 197 181
3 255 239
4 313 297
5 331 315
6 389 373
7 447 431
8 505 489
9 563 547
10 621 605
11 679 663
12 737 721
13 795 779
14 853 837
15 911 895
16 969 953
17 1027 1011

Several other compounds were analyzed by the Li" emitter. The Li+IDS
mass spectrum of palmitic acid is shown in Figure 4.4. The base peak is the
[M]Li+ ion of m/z 263. If the lower-mass region is magnified. there are some
other ions at very low abundance that may be the result of the fragmentation
of palmitic acid. There is a signiﬁcant peak at m/z 245 corresponding to the
[M-H20]Li+ ion. This dehydration is uncommon for the K+IDS analysis of

89

.28 058.3 8 Eazomam mmmE wo_+_._ 3. 9:9“.

 

 

 

 

N\E
CmN CCN CNN CCN CmN CvN CCN CNN CwN CCN CC? Cmr our 00—. me va Cmr CNF
_P_F...—b:._.:LPCL._..—r:._L..L_...._..:_..:—c..._::
_. 1 .. 1. 1 :. I: 1..
NFN
_..
.._ f
PX CN XL
mVN
mCN 1.
+25;

 

 

 

 

Ausuatul aAuelau

90

palmitic acid and, therefore. may be the result of Li" attachment. The ion of
mass 217 is at a relatively low abundance, but could be explained as resulting
form the loss of formic acid (HCOOH) from the [M]Li+ ion of palmitic acid. Co+.
another reactive metal ion, has been shown to eliminate CH202 and H20 in
addition to other products from organic acids.49b This provides support for
the possibility of Li+ reacting with fatty acids to eliminate similar products.
Li+IDS was used to analyze two steroid molecules to see if any induced
fragmentation resulted from the ionization process. The Li+1DS mass spectrum
of cholesterol is shown in Figure 4.5. The base peak represents the [M]Li+ ion
of mlz 393. The Na+ and K+ molecular adduct ions are also observed with mlz
409 and mlz 425, respectively. as these metal ions are present as impurities in
the Li+ emitter. It was hoped that this steroid. which contains several
functional groups, would exhibit more fragmentation due to the reactive Li"
ion attachment. The low-mass ions in the mlz range 135-175 do not resemble
typical adduct ions as they appear at odd and even mlz values unlike K" or Li+
adduct ions that should only appear at odd mlz values for cholesterol.
Compounds that contain only carbon. hydrogen and oxygen. as does
cholesterol, should be represented by even-mass neutrals from both the intact
molecule and from products of 1.2-elimination. Therefore all adduct ions
formed by the attachment of Li" (mlz 7) or K+ (mlz 39) to these even-mass
neutrals should have odd mlz values. This discrepancy suggests that the low-
mass ions observed in Figure 4.5 are surface ionization products. Figure 4.6
contains the Li+IDS mass spectrum of digitoxigenin which is the steroid
portion of digitoxin (a member of the cardiac glycoside family). The [M]Li"’
ion is observed as the base peak at mlz 381. Again, some molecular adduct ions
of digitoxigenin from the alkali metal impurities are present to a small extent
at mlz 397 ([M]Na+) and m/z 413 ([M]K+). No major fragmentation is observed
in the Li+IDS mass spectrum of digitoxigenin. There are some low-mass ions
present in this mass spectrum that are from residual PPG 725 present either on
the probe tip or in the source. This polymer was analyzed just prior to the
analysis of digitoxigenin. It is not uncommon for residual polymer to be
present for the next few experiments. Other Li+IDS mass spectra of thermally

labile compounds are provided in Appendix A.

91

one omv or... 0mm

—-—-bL_—_~P-
a 1 1

cum can can own omm chm omN cam era car 0:. omw 00..

._o§mo_oco .o 826QO $9: mot: m4. 9:9...

NE.

—-_-_-p_-—L-—-—--—--—-——~_phqu~PbL—b—-—_bnn

 

xi

+

 

«2.2

+

+3.2.

 

a

 

 

AusuezuI ennmeu

92

.559x256 _o 83.8QO 39: mo. 1.: 3. 9:9".

#8
o: omv oov can own ovo 0N0 oom omm cow ovm CNN com of. 00—. 3: our

l—rrtur—LcTan—aF—nrv-r..._.1.:1_.Wm ._: _...VIWFL tL—E1w—.Cm1_.1_WL_1CJ_1L1PLJ_L l
111%.; 1.1.1 134 111.1111 4 111.1 1 =1 J .1

mom

 

_.mm

 

 

 

 

Aigsuetul aAnBIaa

93

4. FINAL COMMENTS

These preliminary results show that it is possible to obtain Li"
attachment to the intact molecule of several thermally labile compounds
including cholesterol, digitoxigenin and palmitic acid. and to the oligomers of
the polymer PPG 725. The Li+IDS mass spectra obtained did show some
possibility of induced fragmentation of the analyte from Li+ attachment. This
is primarily in the form of dehydration of the intact analyte molecule. More
carefully controlled studies need. to be performed to accurately compare the
results of Li+IDS with K+IDS and/or Na+IDS and ascertain whether the Li+1DS
technique will provide a means of obtaining more structural information from
the different thermally labile compounds. None of the emitters used for this
preliminary study were used for more than one day as the direct insertion
probe was converted back to K+IDS for use with other on-going research
projects. This makes it very difﬁcult to control all of the variables that affect
the mass spectra with this technique, including the size of the emitter bead,
the amount of ceramic support post present. and the distance between the
sample wire surface and the emitter. Recently, more modiﬁed direct insertion
probes have been constructed for the K+IDS project and. thus, it would be
feasible to dedicate one probe to the Li+IDS technique for an extended period of
time. This would allow for more extensive studies with Li+IDS to determine
whether more structural information is obtainable than with the traditional
K+IDS. Also, consecutive Li+IDS and K+IDS analyses of the same samples could
be made with all instrumental parameters constant. This would provide ideal
conditions for comparing the results obtained with Li+IDS and K+IDS to
determine if Li+ attachment does induce any fragmentation for the thermally
labile compounds studied. The preliminary results are encouraging, but more
experimentation is necessary to determine the utility of using a Li" ion emitter
for the “K+IDS” technique.

CHAPTER 5. IMPLEMENTATION OF THE K+IDS TECHNIQUE ON A
DOUBLE-FOCU SIN G MASS SPECTROMETER.

1. INTRODUCTION

Another offshoot of the original research project was to determine the
mass limit of the K+IDS technique. The current trend in the development of D1
techniques is towards extending the mass range of these mass spectrometric
methods. All previous work with K+IDS was performed on a quadrupole mass
spectrometer with a mass limit of 1.000 daltons. Therefore, to determine the
mass limit of this technique, it was apparent that another mass spectrometer
was needed. The instrument readily available for such an endeavor was the
JEOL PIX-110 double-focusing mass spectrometer which has a mass limit of
14,000 daltons. Modiﬁcations were made to adapt the K+IDS technique for use
on this magnetic sector instrument. ' The JEOL HX-llO mass spectrometer is
more automated than the HP 5985 quadrupole mass spectrometer previously
used in terms of the direct inlet probe and electronic controls, and is not
designed for easy modiﬁcation. Also, the inherent differences between
magnetic sector and quadrupole mass spectrometers, such as the ion source
voltage, add to the complexity of the modiﬁcations needed.

Initial efforts to implement the K+IDS technique on the double-focusing
mass spectrometer dealt with modifying the existing direct chemical ionization
(DCI) probe and DCI electronics of the JEOL HX-llO mass spectrometer, since
the DCI ﬁlament closely resembles that used for the K+IDS technique. DCI
involves placing a nonvolatile sample directly on the surface of an extended
probe and inserting the probe tip into the ion source.55 Often the tip of the
probe is heated with a programmable current control as is provided on the
JEOL instrument. This heating promotes desorption of thermally labile
compounds into the gas phase. followed by traditional C1 or E1 (DEI). The main
differences between this technique and K+IDS are the rate of heating of the
sample (slow ramping for DCI and rapid heating for K+IDS) and the CI reagent
ions (typically ions of ammonia or methane for DCI versus K+ ions for K+IDS).
The JEOL HX-llO DCI probe consists of a platinum wire ﬁlament loop connected
to the probe tip through which a heating current is applied. A second method
for implementing K+IDS on the JEOL HX-llO mass spectrometer dealt with

using the modified DCI probe of the mass spectrometer combined with the

94

95

original K+IDS power supply. These two experimental designs are explained in

detail and the results obtained with each setup are presented and evaluated.

2. EXPERIMENTAL DESIGN

The automated nature of the JEOL HX-llO vacuum system makes it
difﬁcult to construct a direct insertion probe suitable for the K+IDS technique.
The JEOL DCI probe. however, can be modiﬁed to make a K+IDS probe very
similar in appearance to the original probe design used with the quadrupole
instrument. The normal DCI ﬁlament is converted to a K" emitter by threading
the ﬁlament wire (0.007" Re) through a two-holed ceramic support post
(identical to that used with the standard 'K+IDS probe) and then making the K“
glass bead on the wire loop protruding from the ceramic. This is shown in
Figure 5.1. A sample support loop also can be spot-welded to the outer cone of
the DCI probe. The instrument controls (parameter ﬁles and control panel)
must be set in the DCI mode, as the DEI mode was never completed in the
software package of the instrument. The source may still remain in the El
mode, however. A sample parameter ﬁle used to perform the K+IDS
experiment on the JEOL HX-llO mass spectrometer with the modiﬁed DCI probe

is shown in Figure 5.2. Calibration of the mass axis was performed by positive

CTRL Type: SCAN Group: lofl MS Unit: 0 File: HSPAR.PAR Date:
22-NOV-88
DA5000 Acquisition Parameters
A1: DY HSCSKPOS 2: ST HSSTATIC
Bl: Acell.V.: 6.0kV/DAC 2: Ion Mode: DCI 3: Polarity: P08. 4: Temp.
C1: Resolution: 1000 2: Filter: AUTO 3: Inlet: DIR 4: Store: ALL

< Value Manual >
D1: Scan Slope: 1.0 2: Cycle Time: 1.0[0.7 + 0.3] 3: M. Field:HS
131' Max Run Time 2: Scan Type: MF, LI, UP, DA 3: DataType:

Mi,HE
Fl: Thres.: 500-500 2: Valley: 50% 3: Accum.: Range 4: Amp.Sw:
2/60%/100

<E/D Range AMP
G]: 2: 100.000-1500.000

Q1: Note: K+IDS with modiﬁed DCI probe and K+IDS power supply.

Figure 5.2. Parameter file for data collection on the JEOL HX-llO
with the modified DCI/K+IDS probe and K+IDS power supply.

96

K+ bead
T ceramic post
1 1 cm ‘
filament wire
contact to source block

metal ribbon

 

 

 

contact to DCI
power supply

 

 

 

DCI probe body

 

 

 

.0
h-

 

contact pin

Figure 5.1. Schematic of the modified JEOL HX-110
DCI probe tip for use with the K" IDS experiment.

97

FAB with CsI/KI with the same scan slope as used for the K+IDS/DCI
experiments. The variables that need further study are the scan slope and
cycle time. The K+IDS experiment is very fast and most of the sample is
depleted within 1 minute. Therefore, fast scan rates are necessary to obtain all
the information possible from this DI technique. Typical scan rates used for
FAB are 1 scan/minute, or slower. Fast scanning results in poor ion statistics
and possible problems with the mass axis calibration. A detailed analysis of the
limitations of the JEOL PIX-110 instrument regarding fast scanning needs to be
performed to determine the optimum scan slope and cycle time for the K+IDS
technique at the desired mass ranges (e.g., 100-1000 daltons and 100-1500
daltons). Preliminary studies show that a one-second scan slope is often too
great, depending on the mass range, and often leads to erroneous results as the
magnet and data system can not keep up with the fast scanning. The result is
the appearance of peaks at erroneous values and a distorted or poorly

calibrated mass axis.
A. MODIFICATION OF THE DCI POWER SUPPLY.

The existing DCI power supply of the JEOL HX-llO mass spectrometer was
modiﬁed so that it could be used as the power supply for the K+IDS technique.
The DCI power supply was originally designed to apply current through a
ﬁlament similar to that used for the K+IDS emitter. However, instead of a
sudden input of a constant preset current, as is used for K+IDS, the JEOL DCI
power supply was designed to produce a current ramp that varied from 1/32 to
4A per minute with continuous repetition of this current cycle. There was no
way to hold the current at the maximum value and no way to control the
maximum current obtainable. Also the ramping rates were all too low to heat
up the K+ emitter as rapidly as needed for optimum results with the K+IDS
technique. The maximum current obtainable with the original electronic
design and with a K+ head on the ﬁlament was approximately 1A. This is below
the optimum current for a traditional K+IDS probe which usually requires 2-
3A for the best results.

The ﬁrst modiﬁcation to the DCI controls was the addition of two “TEST”
buttons placed on the front panel of the DCI power supply that allowed for by-

passing the current ramping controls. In the test mode, the current can

98

immediately be placed at the maximum value and be held at that current until
the “MIN" switch is activated. With this addition, the current supplied to the
DCI/K+IDS ﬁlament can be immediately set at the maximum current with the
“MAX" button and remain constant. These “TEST" buttons are used to provide
the maximum rate of heating, but do nothing to increase the maximum current
available. Originally there were two 40 resistors (R20 and R21 on the
schematic diagram) in the current system that were replaced by two to
resistors. Thus, the overall resistance was changed from 20 to 0.50. With this
modiﬁcation the maximum current obtainable is approximately 2A with the K"
bead connected (3A with no load). The DCI current meter on the main
instrument panel no longer reads the correct current that is passing through
the DCI ﬁlament due to the modifications with the resistors. The meter value
times four gives the approximate current ﬂow. Therefore, the maximum
current of 2A is read as 0.5A on the meter. This was still not enough current to
perform the K+IDS experiment optimally. Also, the inability to control the
value of the constant current applied to the K+ bead was another major

drawback of this experimental design.
B. CONVERSION OF THE K+IDs POWER SUPPLY TO OON'IROL THE DCI/K+IDS PROBE.

The second attempt at applying the K+IDS technique to the JEOL HX-llO
instrument centered around ﬂoating the K+1DS power supply so that it could
be connected to the DCI probe and used as the controllable current source for
the DCI/K+IDS probe. An isolation transformer was purchased from Heyboyer
Transformer. Grand Haven, M1 to ﬂoat the K+IDS power supply at the high
voltage of the source in the magnetic sector instrument. The transformer and
K+IDS power supply were placed inside a wooden crate to protect the user from
the power supply which is ﬂoating at the acceleration voltage (typically 6-10
kV). Plexiglass was placed on the front of the crate to allow for monitoring
and controlling the current supplied to the bead. The REF IN for the bias
voltage was left unconnected. At this point, no bias voltage has been applied to
the K+ bead as it is difﬁcult to ﬁnd a true reference voltage that would not
interfere with the current loop. Therefore, the bias voltage is set to zero. The
two current connections are made with pins inside the surge ﬁlter located
behind the panel under the vacuum control diagram. One lead is connected to

pin #6 which is the contact to the source block and the other lead is connected

99

to the top lead (from the source) that was connected to pin #10 (from the DCI
control box). The DCI control box is now disconnected from the system and the
K+IDS power supply is connected to the DCI probe contacts through the surge
ﬁlter. Many precautions need to be taken when using this system. All the
connections should be clear and not be touching anything. The panel safety
rollers must be taped in to override the safety switch and allow for the
acceleration voltage to be turned on. The maximum accelerating voltage used
with this setup should be 6 kV. Accelerating voltages higher than this result
in overloading and destroying the K+IDS power supply bias voltage meter (and
possibly other components). There are holes in the Plexiglass cover of the
crate to change the current and turn the power supply off and on with a

rubber hose or wooden dowel.

3. RESULTS

The ﬁrst modiﬁcations to the DCI control unit, the addition of the test
switches and the replacement of the resistors, did allow for K+IDS experiments
to be performed on the JEOL PIX-110 mass spectrometer, but with only marginal
results. The only compounds for which acceptable K+IDS mass spectra were
obtained with this system was the polyethylene glycols (PEG). The K+IDS mass
spectra of PEG 1000 obtained with these modiﬁed DCI controls on the JEOL HX-
110 are shown in Figure 5.3. The m/z values for the K" adduct ions of the
oligomer series ions are given in Table 5.1. These spectra closely resemble
those obtained on the HP 5985 for similar compounds of lower mass (that are
within the mass limit of the HP quadrupole instrument). There is a time
dependence for these mass spectra where the lower-mass fragments are
observed as K+ adduct ions earlier during the run and higher-mass adduct ions
are observed later during the run. This can be seen in Figure 5.4 in the mass
chromatogram of several K"' adduct ions of different oligomers. The averaged
mass spectrum over this entire desorption proﬁle from scan 15 through 50 is
shown in Figure 5.5. This shows a very interesting distribution of adduct ions
throughout the mass range. These are the same results as are obtained with
the K+IDS performed on the quadrupole instrument. A detailed analysis of
polymers by K+IDS has been made elsewhere.43 The main problem with this

design for K+IDS is the limited current that can be applied and the lack of

RELATIVE ABUNDANCE RELATIVE ABUNDANCE RELATIVE ABUNDANCE

RELATIVE ABUNDANCE

100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100
1 Scan 20
u « L
" >
‘ r-
7 b
C. "' i-
‘ r
1. i L
p
a. ‘ I ll L '-
‘ D
1 .
-T 1 llllll‘J l d 7 1 ll lilllAAJAAL-ALAAAAAA LA.._AAA; A A A A p
v r r v v . t ' ' ' I ' ' V T ﬁ ﬁ ‘ ﬁ ﬁ ' ' ‘ ‘ '
1..
1 can as t
4
s >
G. " r-
1 .
4 ,
C. '1 .-
4 >
C. '1 r-
1 l-
4 b
i .
2° ‘ l .—
1 >
i l l ;
r M1 '1 1“ l v 4" ‘4‘ If lﬁA ‘ t A v V
1..
1 can a :
i
.o 1 I
1 :
‘0 " .-
1 b
‘ I’
1 >
I :
2° " p-
4 b
p
- .4.I.ll_-IL111.I-1L41l41|411-llrll41IJIIIIIILIJI I l l I L
' V V V ‘V V V V V W T V V V V V V
10.
I Icon as
4 b
C. J p-
4 r
4 b
so 4 _
1 b
1 b
O. " p-
1 >
i :
2° " .—
1 b
1 l >
-+ A .Lll‘AAA L--A-_.Lt—.AA‘LIARAAILA‘LLAllAuAlLAEAl‘L‘ 1‘ L‘ jl ‘ﬂl l l l L] 1‘7.‘ ['LlA 4... P
U V ﬂ V f ' V V V V V V V V T T v f f w

200 «no see one 1000 1200 1400
I12

Figure 5.3. Temporal dependence of the K+ IDS mass spectra of PEG 1000
obtained with the JEOL HX-110 double-focusing mass spectrometer.

101

.89 Own. a magmas 872
m5 .3 8883 «:2 888 N. 3823 new 0:. m5 .0 3996895 392 in 2:9“.

mmmEDz 2<Om
om om ov cm

> b b h b I h b I h h

 

mm? 1

 

_..Nx

mom I

 

 

1 1111 /

 

 

 

 

 

 

Q; 1< i
a. l 6
_.CN_. NI._.x I ll J<
o:

 

 

 

 

 

 

NS

AJJSNSLNI BALLV'IBH

102

 

oovw

.85608QO 89: 93326538 of 7x: 1.0m...
65 :0 8598 3-3 9.58 60: co? Own. .0 €28QO 89: mo. Iv. 80.296 9:. .m.m 9:9“.

N>2
oouw ooow com com 00? com

3.33; 2...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

om

ov

om

Mgsuetul OAIIBIOH

om

oow

103

Table 5.1 Values of m/z for the K+ adduct ions of the oligomers of polyethylene
glycol 1000.

 

n H[OCH;CH2]nOH]K+ n H[OCH2CH2]nOH]K+
1 101 18 849
2 145 19 893
3 189 20 937
4 233 21 981
5 277 22 1025
6 321 23 1069
7 365 24 1113
8 409 25 1157
9 453 26 1201
10 497 27 1245
11 541 28 1289
12 585 29 1333
13 629 30 1377
14 673 31 1421
15 717 32 1465
16 761 33 1509
17 805 34 1553
control over this current. Different types of compounds require different

heating currents and energy to desorb with little or no fragmentation and be
detected as K+ adduct ions. As the search for the upper mass limit of K+IDS
continues. higher ﬁnal currents and heating rates will probably be needed to
desorb larger molecules intact. Therefore, this initial design proved incapable
of replicating the K+IDS technique satisfactorily and providing control over
the current.

The second design, which utilizes the K+IDS power supply to apply
current to the DCI/K+IDS probe, does allow for the maximum current to be
controlled and optimized for different types of compounds. The K+IDS mass
spectrum of digoxin obtained with the K+IDS power supply providing current
to the DCI probe is shown in Figure 5.6. As can be seen, this mass spectrum
closely resembles the K+IDS mass spectrum of digoxin obtained on the

quadrupole mass spectrometer (see Appendix B). Not all of the fragment

38896QO $88
9580212238 or 7x1 40w... 9: :0 8598 sxomﬁ .o 828QO mmmE mo. .2 .o.m 9:9“.

 

104

 

 

 

 

Neg
com com ooe com
«1 _ _ .C
1
Few .
mmm .om
moo .
a. H
+ x5. 8“ 1 9.
amp 1cm
.om
+8 .
mar

 

 

00—.

 

Aigsuetul eAgtelea

105

adduct ions and [M]K+ ion are detected at the correct masses. This is most likely
due to the fast scanning rate, pushing the limits of the magnet and the data
system, and affecting the mass axis calibration. The major problem with the
application of the K+IDS technique to the JEOL PIX-110 magnetic sector mass

spectrometer is the scan rate limitation.

4. FINAL COMNIENTS

The dynamic nature of the K+IDS technique and the short-lived signals
(less than one minute) makes it highly unlikely that implementation of this
technique to the JEOL HX-110 double-focusing mass spectrometer will be a
successful endeavor. The scan rates needed to observe the desorption of the
analyte and the changes in the K+IDS mass spectra with time are too great for
the magnetic sector instrument to handle them with good results. The K+IDS
technique is more suited to quadrupole mass spectrometers or time-of-ﬂight
instruments that can handle scan rates of 1 scan/second or greater.

This application of K+IDS to the HX-110 mass spectrometer may be useful
in the study of mechanisms of ion formation and fragmentation of the FAB
technique. Since K+ ions only attach to neutral species and do not induce
fragmentation, the formation of K" adduct ions provide information about the
origin of fragmentation. Any thermal degradation products that are formed
on the surface and desorb into the gas phase as neutrals should be observed as
K+ adduct ions. Whereas, any fragmentations induced by the initial
protonation of the molecule will not be observed as K+ adduct ions. This
combination of K+IDS with FAB may provide a way of determining whether
fragmentations are occurring in the gas phase following protonation or prior
to ionization during the desorption process. The K" emitter provides a good
means of injecting a large ﬂux of K" ions into the gas phase, a process which
is required for the formation of K" adduct ions. Therefore. it may be
beneﬁcial to combine these two techniques to gain a better understanding of
the mechanisms of fragmentation and ionization of FAB. The combination of
the traditional FAB technique for desorption and K+ adduct formation in the
gas phase for ionization (K+ chemical ionization) is currently being pursued

in our lab.56

CHAPTER 6. FAB MASS SPECTROMETRIC ANALYSIS
OF CARDIAC GLYCOSIDES.

1. INTRODUCTION

Interest in desorption ionization (DI) techniques and the understanding
of the mechanisms of ionization and fragmentation led to a side project that
utilizes FAB. One goal of this study was to gain a better understanding of a
more traditional DI technique, FAB, compared to K+IDS in terms of sample
preparation, experimental procedures required, and the mass spectra obtained.
Ultimately, a direct comparison of these two DI techniques in terms of
mechanisms of ion formation and fragmentation would be ideal, but is beyond
the scope of this minor research project. Since much work was done with
cardiac glycosides utilizing the K+IDS technique, these compounds were
chosen for this comparative FAB study. The previous K+IDS study of these
biomolecules showed very interesting results as molecular weight and
sequence information was easily obtainable from the K+IDS mass spectra
(Appendix B). However, K+ IDS is a technique that does not induce
fragmentation upon ionization. All fragmentation occurs prior to ionization
in the condensed phase and K'I' ions merely attach to the neutral species
present in the gas phase. FAB, on the other hand, is a D1 technique where
ionization (protonation) often induces fragmentation which may lead to
further structural information. Therefore, it was of interest to study the FAB
mass spectra of the cardiac glycosides to investigate the types of
fragmentation of these multifunctional compounds that occur with this DI
technique and whether the mechanisms of fragmentation can be determined.

2. EXPERIMENTAL SECTION

The cardiac glycosides were obtained from Sigma Chemical Co., St. Louis,
Mo. and were used without further purification. Acetyldigitoxin was
purchased from ICN K&K Laboratories, Cleveland, OH. The samples were
dissolved in methanol to concentrations of approximately lug/pl. One to two-pl
aliquots were transferred to the FAB probe sample target and mixed with the
glycerol matrix. The FAB analyses were performed on a JEOL HX-llO double-

106

107

focusing mass spectrometer of forward geometry with an accelerating voltage
of lOkV and a FAB gun voltage of 6kV with Xe as the FAB gas.

3. RESULTS
A. POSITIVE FAB MASS SPECTRA OF SOME CARDIAC GLYCOSIDES

The structures of four of the cardiac glycosides studied are shown in
Figure 6.1. Digoxin was chosen for a detailed mechanistic study due to its
relative ease of ionization by FAB and the abundance of fragment ions
observed with FAB. The goal was to determine if the information obtainable
from a double-focusing mass spectrometer (i.e., peak matching, B/E linked
scanning, and FAB) was sufﬁcient for use in deducing the structural formulas
of the ions and the fragmentation pathways. The results of this study are
published and this article is presented in Appendix C. Many fragment ions are
observed in the FAB mass spectrum of digoxin as seen in Figure 6.1. There are
fragment ions representative of both ends of the molecule. However, the
majority of the types of fragment ions of digoxin contain the aglycone portion
of the molecule which suggests that this steroid structure may be directly
involved in the ionization of this molecule and carries the charge in the
fragment ions. Based on the results obtained from the linked scanning and
peak matching experiments, it is determined that the majority of the fragment
ions observed can be formed by remote site fragmentation of the protonated
molecule with the charge localized on the aglycone portion of the molecule.
Even though this seems to be the dominant mechanism of fragment ion
formation, it is possible that some of the fragment ions which do not contain
the aglycone could be the result of charge-initiated fragmentation from
protonation of one of the glycosidic bonds. These mechanisms are explained
in more detail in the article in Appendix C.

During the preliminary FAB study, FAB mass spectra of all the cardiac
glycosides shown in Figure 6.1 were obtained. An analysis of these mass
spectra, with comparison to the FAB mass spectrum of digoxin provide some
interesting differences due to only minor differences in the structures of
these compounds. Figure 6.2 contains the published FAB mass spectrum of
digoxin to provide a basis for these comparisons. The proposed fragment ion

assignments for digoxin, gitoxin. digitoxin, and acetyldigitoxin are given in

108

-H + _
[H083 ]H [11083082081 Hm”

 

 

 

 

 

 

mlz 131 m/z 391
+l\ +I'\
——> H V H
CH3 CH3 C":5
o o o 0“
-0 -0 —0
H0
R3 an OR
——> H ——> H ——> H
\1+ \2+ \E-t-
mlz 651 m/z 521 mlz 391

[AOS1OS2 OH]H"' [AOS1OH]H+ [AOH]H"'

D=Digoxin F=Acetyldigitoxin
R 1: OH R1: H
R2: H R2: H
R 3: OH R3: OAC
E=Digitoxin G=Gitoxin
R1: H R1: H

Figure 6.1 Structure of digoxin and other similar cardiac glycosides
studied. The fragmentations observed in the FAB mass spectrum of
digoxin are labeled.

109

.. B 88ch 9m «:2 8626 .983 mi .c_xom_n Co 826me mmmE m<.... 958d No 92mm

 

 

 

 

NE.
co m o om
o.m x ._.
08 m9.
mom
5N Em
+ £2 :5

00¢

an
«no

man

 

mvm

 

_.mw

 

 

10¢

MISUGIUI 81119188

I
O
(D

 

00..

 

110

Table 6.1. The assignments are designated by the LKA nomenclature scheme,

which is explained in detail in Appendix B.

TABLE 6.1 Fragment ion assignments for digom'n, digitoxin, gitordn, and
acetyldigitoxin.

W3 mlzlalue
digoxin and digitoxin acetyldigitoxin
gitoxin

[M]H+ 781 765 807
[A0810820H]H+ 651 635 635
[A081 OSzOH-HzOH-P' 633 617 617
[A081 OHJH+ 531 505 505
[AOSloH-H20]H+ 513 487 487
[AOHJH+ 391 375 375
[HOSaOSzOSl°H]H+ 391 391 433
[AOH-H20]H+ 373 357 357
[HOS3082081 'H-HzO]H+ 373 373 415
[AOH-2H20]H+ 355 339 339
[HOS3OSzOS1'H-2H20]H+ 355 355 397/355b
[AOH-3H20]H+ 337 N Ac N Ac
[143083032081 ‘H-3H20]H+ 337 337 379/337
[HOS3OSz'H'H20]H+ 243 243 285/243
[HOSg'H]H+ 131 131 173/131
[HOS3'H-HzO]H+ 113 113 155/113

a. These assignments are based on the Light, Kassel, Allison (LKA)
nomenclature scheme. For an explanation, see Appendix B.

b. There are two possible m/z values for this fragment ion with and without
the acetyl group on the terminal sugar.

c. This fragment is not applicable to the compound. Digitoxin can not lose
3H20 from its aglycone since there are only 20H groups present.

The FAB mass spectrum of gitoxin is shown in Figure 6.3. This
compound has the same molecular weight as digoxin, but varies in the position
of one -OH group on the aglycone, as shown in Figure 6.1. The [M]H+ ion of
gitoxin is present in the FAB mass spectrum at m/z 781. There is less
fragmentation (lower abundance of fragment ions) of gitoxin, where the

retention of the charge is on the aglycone, than was observed in Figure 6.2 for

111

 

com

+ 1.2.

 

.. .3 88:8 2m .82 .288 .283 mg... .526 _o 825QO «was m<u 928d Qo 9:9...

con com com oov com

awn.

 

 

 

 

 

 

mvm

 

«u

 

 

«Pr

 

Fur

 

 

 

 

 

om

ov

cm

on

cor

MisuewI eniteleu

112

digoxin. The low-mass fragment ions, with charge retention on the sugar
portion of the molecule, are observed in the gitoxin spectrum. It should be
noted that while concentrations of the cardiac glycosides in solution that are
used in this preliminary study are approximately equal, the ratio of sample
solution to glycerol matrix is not accurately controlled. Therefore. among
these cardiac glycoside FAB mass spectra, the ratio of peaks representing
matrix ions to sample ions (both molecular and fragment) may vary between
different spectra. The fragment ion abundance vs. molecular ion abundance
of one mass spectrum can be compared to the same abundance ratio for
another compound.

The FAB mass spectrum of another similar cardiac glycoside, digitoxin,
is shown in Figure 6.4. This compound differs from digoxin in that it has one
less -OH group on the aglycone and, hence, the molecular weight is 16 daltons
lower. The mass spectrum of digitoxin does contain a peak at m/z 765
representing the [M]H+ ion. The “mass symmetry" that is present in the
digoxin molecule between the aglycone and the sugars does not exist for
digitoxin. Therefore, the series of fragment ions in the 300-400 dalton mass
range, that each had two structural assignments for digoxin, are now split by
16 daltons due to the difference in the aglycone. Each fragment ion in the
300-400 dalton mass range in the digitoxin mass spectrum has one structural
assignment and is less abundant than the corresponding doublet fragment ion
in the mass spectrum of digoxin which has two structural assignments. For
example, the peak at mlz 373 in the FAB mass spectrum of digoxin has two
structural assignments. In the digitoxin FAB mass spectrum, one of these
stuctures is respresented by a peak at m/z 373, while the other is represented
by a peak 16 daltons lower at m/z 357. The low-mass sugar ions of m/z 113, mlz
131, and m/z 243 are present in the mass spectrum of digitoxin as they are for
digoxin.

Acetyldigitoxin is another cardiac glycoside that is similar to digitoxin
except for the presence of an acetyl group on the terminal sugar (digitoxose).
The structure is shown in Figure 6.1 and the FAB mass spectrum of
acetyldigitoxin is given in Figure 6.5. The peak at mlz 807 represents the
[M]H'*' ion and the peaks at m/z 635 and m/z 505 are representative of the loss
of one and two sugars from the compound, with charge retention on the
aglycone. The ion of m/z 415 represents the structure [HOS3082081'H-H20]H+

where 83 contains the acetyl group. There are no peaks observed in the mass

113

com

.. .3 88:8 96 «:2 .28.... .983 9:. .c_x2_m_o .o 628QO «was 9E 9:68 «.0 9:9”.

 

 

+ as: .

 

#5

mac

com

mom

 

oov

 

com

 

 

mvm

cow

 

cor

 

 

m:

 

5..

Q i: : _

 

 

 

 

 

0N

ov

om

om

cow

Mgsuetul eAuelea

.. .3 86:8 9m 82 8.8.0 .983 9:. .:_x2_a_u_b8m Co 52:88 89: BR 828: md 98E

 

114

 

 

Néz
com com com com oov oom CON ooF
blip 1L p p p h p m ndlpl b n - ~114P1dup 4:11.111:
96*.
:2
8m 8 .
moo .
wow
wow

«FF

 

 

1 car

 

hisuetm emeleu

115

spectrum that represent H20 losses from this peak at m/z 415. The aglycone
fragment [AOH]H+ is observed as a peak at m/z 375 and does exhibit water loss
that is represented by peaks at m/z 357 and m/z 339. The peak at m/z 285
represents a sugar ion that contains the acetyl group on the terminal sugar,
with a structure [HOS3OS2'H-H20]H+. This fragment is also represented by the
peak at mlz 243 without the acetyl group in which case the “H20" loss is the
acetylated hydroxy group on the terminal sugar. The peaks representative of
a single sugar are observed at mlz 113 and 131 when no acetyl group is present
and at mlz 155 and 173 when the acetyl group is present on the sugar moiety.
The minor differences in the structures of these cardiac glycosides
created observable differences in the FAB mass spectra in comparison to
digoxin. For all of these compounds, peaks representing an [M]H+ ion and
basically the same fragment ions were observed in the FAB mass spectra.
However, the structural differences did seem to affect the overall desorption
ionization efﬁciency of the sample, as the relative intensities of the peaks
representing the [M]H+ ions differed. Many of the differences observed were
due to mass shifts caused by the structural changes. Other noticeable
differences were in the relative intensities of peaks representing the
fragment ions from dehydration products. More detailed mechanistic studies
of these other cardiac glycosides need to be performed before many

conclusions can be drawn based on these preliminary observations.

B. NEGATIVE FAB MASS SPECTRA

The negative FAB mass spectra of digoxin and ouabain were obtained for
Preliminary comparison of the behavior of cardiac glycoside compounds
under conditions of negative vs. positive ion formation. Similar studies of both
Positive and negative FAB mass spectra of cardiac glycosides utilizing different
FAB matrices have been performed by Pare et al.57 Figure 6.6 shows the
negative FAB mass spectrum of digoxin. The peak at m/z 779 corresponds to
the [M'HJ‘ ion, which is more intense in this mass spectrum than the
COrresponding peak at mlz 781 representing the [M]H+ ion in the positive FAB
mass sPectrum (Figure 6.2). This suggests that ionization occurs more readily
in the negative ion mode and/or subsequent fragmentation is less prevalent.
The fragmcntations at the glycosidic bonds, with charge retention on the

aglycone-cont.aining portion, are observed at m/z 649 and mlz 519 in the

116

000

. =12.

 

 

.. 3 88:8 88 82 88:8 8.820 9:. .8898 8 628QO 89: m<n 82882 88 9:9”.

005

mvo

com com .

8 8m 3..

00¢

com

 

oom

Far

 

wmw

cor

 

 

Alswalw GADBIGH

117

negative FAB mass spectrum. The low-mass sugar ions, which are very
abundant in the positive mode, are practically non-existent in the negative
mode. This suggests there may be interesting differences in ion formation and
fragmentation between the negative FAB and positive FAB mass spectrometric
analysis of digoxin, and suggests the need for a similar mechanistic study of
digoxin to be carried out in the negative mode as was presented in Appendix C
for digoxin in the positive mode.

Another cardiac glycoside for which both positive and negative FAB
mass spectra were obtained is ouabain. The structure of this compound is
shown in Figure 6.7. This cardiac glycoside contains only one sugar and the
aglycone contains many more functional groups than the aglycone of digoxin.
The positive ion mass spectrum of ouabain is shown in Figure 6.8 and the
negative ion mass spectrum is shown in Figure 6.9. The m/z values and
possible assignments of the analyte ions observed in both the positive and

negative FAB mass spectra of ouabain are given in Table 6.2.

Table 6.2. Fragment ions observed in the positive and negative FAB mass
spectra of ouabain and possible assignments based on the LKA
nomenclature (see Appendix B).

E . | . I E 9 E M S I
Assignment mlz M W
['M]H+ or [AOS10H]H+ 585 583 [M-H]' or [AOS10]‘
[AOH]H"' 439 437 [A0]'
[A‘H]H+ 421 419 [A'2H]'
[A‘H-H20]H+ 403 401 [A'ZH-H201'
[A'H-2H20]H+ 385 383 [A’ZH-2H20]’
[A'H-Hzo-OCHzJH+ 373 371 [A'ZH-Hzo-OCH2}
[A'H-3H20]H+ 367 369* [A’2H-H20-CH3OH]'
[A'H-2H2()-OCH2]H+ 355 353 [A’2H-2H20-OCH2J'
[A'H-4H20]H+ 349 347 [A'zH-4H20}
[A'H-3H20-OCH2]H+ 337 331 [A'ZH-3H20-OCH2T

[A'H-4H20-OCH2]H+ 319 319 [A'2H-4H20-OCH2]'
163 [HOSOI'

* There are several ions near this m/z value in the negative FAB spectrum
shown in Figure 6.9 that may provide structural information. The
assignment provided for the ion of m/z 369 is only proposed as one possible
assignment.

OH OH

118

 

OH

MW: 584

Figure 6.7 Structure of ouabain.

119

com

 

.. 3 88:8 2m 22 .288 .9096 9:. .5330 .o 828QO $9: m<u 928a 9o 9:9“.

com

.nga

 

oom

 

 

 

 

 

 

Neg
cow com
7 _;
Fmv men men
woe
5mm
mum 5mm
mmm mmm
mov

cow

 

00.
,,b
-ou
. H
.oe mm
m.
m.
. 8
. WM
.oo MW
e.
um
.om
cor

 

 

120

com

now
new

use
. $-95:

 

.. B 88ch 2m 20. .230 689% 9:. .5330 .o 88.8% mmmE m<u mémomz m6 9ng

- 5-2.

 

com

 

hmv

 

 

2:
.
a: i,
_ w . f
1 a W
has mom
a:
not
_ . H
,mwwrmr Tove:
we.
. m.
. 3
. M
.8w
5.
. M
.8
ﬂ
. oS

 

 

 

121

In the positive ion mass spectrum, there is a peak representing the [M]H+ ion,
and much fragmentation of the aglycone is observed in the m/z 300-400 dalton
mass range. These fragment ions are probably from multiple dehydrations
and loss of the functional groups of the aglycone. Very few low-mass sugar
ions are observable to a significant extent. In the negative FAB mass
spectrum, there is an intense peak at mlz 581 representing the [M-H]’ ion.
The same fragment ions from dehydration of the aglycone are present. but
they have a lower relative abundance than those in the positive ion mode. The
negative mass spectrum does, however, contain low-mass fragment ions from
the sugar portion of the compound. The peak at mlz 369 in the negative FAB
mass spectrum does not ﬁt the pattern with the rest of the fragment peaks that
are 2 daltons lower than corresponding positive ion fragments. It appears that
there is some different mechanism in operation for the formation of this
fragment ion. There are several peaks in this mass range of 365-371 daltons
that may provide structural information. However, before “over-analysis” of
these results is attempted these experiments should be repeated for

verification.

IV. SUMMARY

The positive FAB mass spectra of the cardiac glycosides contained some
very interesting differences due to slight structural variances.
Understanding how these structural differences are responsible for the
changes in the FAB mass spectra requires more detailed studies similar to the
one completed with digoxin, which is presented in Appendix C. One
interesting observation made from these FAB experiments is the static nature
of the FAB mass spectra over the long-lived signals (> 15 minutes). This is in
contrast to the K+IDS mass spectra where the spectra often change with time
and the signals are very short-lived (< one minute). There appear to be more
variables to control in the K+IDS technique, such as current and bias voltage
applied to the emitter and the distance of the sample from the emitter, that
may affect the mass spectra obtained and possibly increase the information
obtainable. For instance, as described in Chapter 2, the order in which
fragmentations and desorption of the intact molecule occur may provide some
insight into the nature of those fragmentations or the corresponding products

formed. FAB has few variables that can be changed to alter the mass spectra

122

obtained. The main variable is the matrix used which primarily affects the
ionization efficiency and the longevity of the signal. Changing the matrix
used for FAB. does not change the fragment ions observed or the overall
appearance of the FAB mass spectrum. Therefore, K+IDS may have an
advantage over the FAB technique in that several variables can be used to
modify the mass spectra obtained for the desired results. This is contigent,
however, on the ability to control all of these variables of the K+IDS
technique.

From these preliminary negative FAB mass spectra of digoxin and
ouabain, it appears that there may be some interesting differences between
positive and negative ion formation of the cardiac glycosides by FAB
ionization. It may prove interesting to try some other matrices, that may be
more amenable to formation of the [M-H]’ ions of these molecules, such as
triethanolamine (TEA) which is often used as a matrix for negative FAB. Also,
the ratios of sample to matrix can be adjusted to obtain the optimum sample ion

formation and fragmentation.

APPENDIX A

Plots of K+ adduct ion abundance vs. sample ﬁlament wire temperature (K).
Figure A1. Abundance of fragment adduct ions of met-enkephalin vs. sample
filament wire temperature (K).

Figure A2. Abundance of fragment adduct ions of met-enkephalin vs. sample
ﬁlament wire temperature (K).

Figure A3. Abundance of fragment adduct ions of met-enkephalin vs. sample
filament wire temperature (K).

Figure A4. Abundance of fragment adduct ions of leu-enkephalin vs. sample
filament wire temperature (K).

Figure AS. Abundance of fragment adduct ions of leu-enkephalin vs. sample
filament wire temperature (K).

Figure A6. Abundance of the [M]K+ ion of sucrose vs. sample ﬁlament wire
temperature (K).

Figure A7. Abundance of the [M]K+ ion of sucrose vs. sample ﬁlament wire
temperature (K).

Figure A8. Abundance of the K" adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

Figure A9. Abundance of the K+ adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

Figure A10. Abundance of the K4” adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

Figure All. Abundance of the K+ adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

Figure A12. Abundance of the K4” adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

Figure A13. Abundance of the K” adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

Figure A14. Abundance of the K+ adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

Figure A15. Abundance of the K+ adduct ions of melezitose vs. sample ﬁlament
wire temperature (K).

123

Figure A16. Abundance of the

124

[M]K+ ions of sucrose and melezitose vs. sample

filament wire temperature (K).

Figure A17. Abundance of the

[M]K+ ions of sucrose and melezitose vs. sample

filament wire temperature (K).

Figure A18. Abundance of the
wire temperature (K).

Figure A19. Abundance of the
wire temperature (K).

Figure A20. Abundance of the
melezitose vs. sample ﬁlament

Figure A21. Abundance of the
melezitose vs. sample ﬁlament

Figure A22. Abundance of the
temperature (K).

Figure A23. Abundance of the
temperature (K).

Figure A24. Abundance of the

[M]K+ ion of palmitic acid vs. sample ﬁlament
[M]K+ ion of palmitic acid vs. sample ﬁlament
[M]K+ ions of palmitic acid, sucrose, and

wire temperature (K).

[M]K"' ions of palmitic acid, sucrose, and

wire temperature (K).

[M]K"’ ion of glycerol vs. sample ﬁlament wire

[M]K+ ion of glycerol vs. sample ﬁlament wire

[M]K+ ion of triethylene glycol vs. sample

ﬁlament wire temperature (K).

Figure A25. Abundance of the

filament wire temperature (K).

[M]K+ ion of triethylene glycol vs. sample

Activation energy determinations based on plots of
K+ adduct ion abundance vs. III

Figure A26. Ea determinations
mlz 392 of met-enkephalin vs.

Figure A27. Abundance of the
from the slope = 23 kcal/mol.

Figure A28. Abundance of the
= 4 kcal/mol.

Figure A29. Abundance of the
Figure A30. Abundance of the

Figure A31. Abundance of the

from the Arrhenius plot of the abundance of
UT.

[M]K“' ion of palmitic acid vs. 1/1‘. Ba calculatedd

[M]K+ ion of palmitic acid vs. 1/1‘. Ba calculated

[M]K+ ion of glycerol vs. l/T (K).
[MJK+ ion of glycerol vs. 1/T (K).

[M]K"‘ ion of glycerol vs. In (K).

125

Activation energy determinations based on plots of ln(ln(Ao/At)) vs. 1/T.

Figure A32. Arrhenius plot of ln(ln(Ao/At)) vs. 1/1‘ for the [M]K+ ion of
palmitic acid. Ea calculated = 14 kcal/mol.

Figure A33. Arrhenius plot of ln(ln(Ao/At)) vs. 1/T for the [M]K+ ion and a K+
adduct ion of a fragment of melezitose. Ea(543) = 22 kcal/mol and Ea(363) = 47
kcal/mol.

Figure A34. Arrhenius plot of ln(ln(Ao/At)) vs. l/T for the [M]K+ ion and a K+
adduct ion of a fragment of melezitose. Ea(543) = 35 kcal/mol and Ea(363) = 40
kcal/mol.

Figure A35. Arrhenius plot of ln(ln(Ao/At)) vs. 1/1‘ for the [M]K+ ion and a K+
adduct ion of a fragment of melezitose. Ea(543) = 41 kcal/mol and Ea(363) = 47
kcal/mol.

Figure A36. Arrhenius plot of ln(ln(Ao/At)) vs. III for the [M]K+ ions of

melezitose (mlz 543) and sucrose (mlz 381). Ea(543) = 20 kcal/mol and Ea(381) =
15 kcal/mol.

Figure A37. Arrhenius plot of ln(ln(Ao/At)) vs. 1/1‘ for the m/z 392 fragment
adduct ion of met-enkephalin. Ea calculated = 37kcal/mol.

Sample Li+IDS mass spectra.

Figure A38. Li+IDS mass spectrum of polyethylene glycol 600 averaged over
scans 30-55.

Figure A39. Plot of the Li+IDS mass spectra of polyethylene glycol 600 vs. scan
number.

Figure A40. Li+IDS mass spectrum of methionine-enkephalin.
Figure A41. Li+IDS mass spectrum of triphenylamine.

Figure A42. Li+IDS mass spectrum of violuric acid.

126

    

 

 

 

 

 

10000 -
O
8000 "
l ’4 —D— mlz 259 x 10
W (Tl/2392
8 6000 " ---"0'"' 01/2449 X 10
C
(U
'0
C
:3
.O
<
4000 "
J
2000 -'
0 ' l
450 A 600

Figure A1 . Abundance of fragment adduct ions of met-
enkephalin vs. sample filament wire temperature (K).

127

60000 a

_,./

50000 -

40000“ —a— m/z259x1o
----v-- m/2392

----- o- --- mlz 449 x10
30000 -

Abundance

 

20000 ‘

 

10000 -

 

 

 

300 350 400 450 500 550 600 650 700
NW

Figure A2. Abundance of fragment adduct ions of met-
enkephalin vs. sample filament wire temperature (K).

128

 

soooo -
O
‘ i
5
20000 - g
i
a) ‘ —a— mlz 259 x 10
0
g --+- mlz 392
'0
g -------o m/z 449
.0
<

 

 

 

Figure A3. Abundance of fragment adduct ions of met-
enkephalin vs. sample filament wire temperature (K).

129

20000 '-

15000 . /3

 
     

 

  

8 —-9— mlz 259 x10
C
58 '0000 " W mlz 374
C
3 ------o-- mlz 431 x10
< :

sooo - ,9

 

‘0

   
 
 

 

0?.

  

 

Figure A4. Abundance of fragment adduct ions of leu-
enkephalin vs. sample filament wire temperature (K).

130

—9— 01/2259 x10

m/2431x10

W mlz 374
.-.--°-...

 

 

 

12000 -

10000 -

8000 '-

q
0
m
6

8:20:34.

4000 -

2000 '-

 

650

600

550

500

450

Temp (K)

Figure A5. Abundance of fragment adduct ions of leu-
enkephalln vs sample filament wure temperature (K)

Abundance

131

3m)—

2m)-

 

 

100‘

 

 

1 I "V I f V

v . . l . .
450 500 550 600

Temp (K)

Figure A6. Abundance of the [M]K* ion of sucrose
vs. sample filament wire temperature (K).

Abundance

132

 

 

 

 

 

 

1500-
1000‘
——9_ 111/Z381
500‘
OJ—w—I—I-w—l-w—I—I-‘lvuwﬁ.#VAVjvv.tl
400 450 500 550 600 650

Temp (K)

Figure A7. Abundance of the [M]K+ ion of sucrose
vs. sample filament wire temperature (K).

133

 

 

 

 

 

 

5m)-
0
400-
9
i
a
if
a) 300‘ ’t: I —a— m/2543+-1
O i‘ c
g is.” (I ----o--— macaw
‘D g:
c
:3
.o
a o,
9
. W:
o 00:: ° 0
. \
100. A
.6”. ‘ ° 1
o o ".
0 V"lj"l"'l"'l“'lf"l“‘1

 

520 540 560 580 600 620 640 660

Temp (K)

Figure A8. Abundance of the K" adduct ions of melezitose
vs. sample filament wire temperature (K).

134

 

 

 

 

 

 

 

700-

600-

500-
8 400-
g —G— mle43+-1
"o .
g "-9“- W2363+-1
.o
< 300- ----- 0"" m/2201+-1

-----o---- m/2183+-1
200.
100‘
0 r“1ﬁ" "'|“‘T"'l“ﬁ

l
520 540 560 580 600 620 640

Temp (K)

Figure A9. Abundance of K’ adduct ions of melezitose
vs. sample filament wire temperature (K).

135

 

 

 

 

 

 

 

400-
300‘ p
‘ l
—a— m/2543+-1
(D
2 ——o— mlzae3+-1
(U
‘2 20°“ —-— rn/2201+-1
3
‘2 —'°— m/zl83
100-
o Y ﬁi V i

 

 

' I ' I ‘ I I I
520 540 560 580 600 620 640

Temp (K)

Figure A10. Abundance of the K" adduct ions of melezitose
vs. sample filament wire temperature (K).

136

600 -

500 '-

400 - —9— mlz 543 +-1

—9— mlz 363 +-1

300 '-

Abundance

200 -

 

100‘ n a

O

 

‘Y UV‘Ij‘j1

' I
600 620 640

o ‘I V V I t T I I I

. . ,
520 540 560 580

Temp (K)

Figure A11. Abundance of the K1 adduct ions of melezitose
vs. sample filament wire temperature (K).

137

 

 

 

 

300 '1

amo-
a)
g o
58 ,° —9— mlz 543+~1
C O
3 —0— mlz 363 +-1
< O

1001 I
. i
. In.)
0 T I V V I 1 1 Vﬁ' if V V V T I I I V I I
450 500 550 600 650

Temp (K)

Figure A12. Abundance of the K+ adduct ions of melezitose
vs. sample filament wire temperature (K).

138

1200 n

1000 4

°“-”°.'.‘.'uum-.

8

S

8

.9

800 -‘ 5
I -—G-— mlz543+-1

‘ ““9"” 01/2363 +4

Abundance
O)
O
O
l

 

400 "

 

 

oj'ﬁ'Y‘YIVUVIU‘V'UYVIU'1'

540 560 580 600 620 640 660

Temp (K)

Figure A13. Abundance of K‘ adduct ions of melezitose
vs. sample filament wire temperature (K).

139

 

 

 

 

 

 

 

300-
200-
a, —a— mle43+-1
0
g ~-~~v~- mlz 363 +-1
'U
C
:3
.0
<
100-
q
0"'I' 'I"‘fﬁ"l"l
540 560 580 600 620 640
Temp (K)

Figure A14. Abundance of the K1 adduct ions of melezitose
vs. sample filament wire temperature (K).

140

1200 a

1000 -'

 

 

 

 

 

    

 

 

Q)
0
c —G— mle43+-1
“3 eoo~
‘O
S ---+-- m/2363+o1
.D
<
400-
200‘
.0
‘1‘; £5,200.
0 ~o ‘3‘.
o ..... .,....,-..-.....,
450 500 550 600 650 700
Temp(K)

Figure A15. Abundance of the K+ adduct ions of melezitose
vs. sample filament wire temperature (K).

141

300 -

““0““ Sucrose
—f=|'— Melezitose

3
S
200 - 3'
i

f

8

 

 

 

 

Abundance

 

 

100'

 

 

 

Figure A16. Abundance of the [M]K" ions of sucrose and
melezitose vs. sample filament wire temperature (K).

142

 

 

 

      
 

 

 

1500 -
g
6 b
§
1000- f g
\ 5 .
5
8 ‘ l 2...... s
c g e a! ucrose
(U i i
'2 5 3 —ﬁ— Melezitosexz
3 9.
< "<2
3 i
500-
3
6
.3
“ 3
.5
f
. g ..
ﬁ/ 0 OX
0 I U i j I r I U U I U I' ‘I ‘j’ Tﬁ' I V 'm
500 550 600 650

400 450

Temp (K)

Figure A17. Abundance of the [M]K+ ions of sucrose and
melezitose vs. sample filament wire temperature (K).

Abundance

143

5000'“

4000"

3000"

2000"

 

1000"

 

 

 

 

300 350 400 450 500

Temp (K)

Figure A18. Abundance of the [M]K+ ion of palmitic acid
vs. sample filament wire temperature (K).

Abundance

143

5000'“

 

 

 

Temp (K)

Figure A18. Abundance of the [M]K+ ion of palmitic acid
vs. sample filament wire temperature (K).

144

2000 -

1500 "

Abundance
3
O
O
l

 

500 ‘

 

 

Figure A19. Abundance of [M]K+ ion of palmitic acid
vs. sample filament wire temperature (K).

145

 

 

 

 

2000-
1500- 5 3.?
a, -------o PalmiticAcid
w = ,0?
g ' v; , g W SucrosexS
o . 0 .
g 1000 - 5 ' -—9— Melezitose
C o
3 l :
.o :
< a
500-
.5
:' «o
.-' 04"
0 ﬂ 3. . . 9 .
250 350

Temp (K)

Figure A20. Abundance of the [M]K+ ions of palmitic acid,
sucrose, and melezitose vs. temperature (K).

146

 

 

3000 -

9

2000 -

-------o- m 295 +-1
cu °.
2 ........o...... mlz 381 +-1
m
'g —a— mlz 543 X5
3
.0
<

1000 J
o 1
300

 

Figure A21. Abundance of the [M]K+ ions of palmitic acid,
sucrose, and melezitose Vs. temperature (K).

Abundance

147

mxmo-

uxmo-

 

 

 

 

 

I I I I

I U U U I I I I I I I U
300 350 400 450 500 550

T (K)

Figure A22. Abundance of the [M]K* ion of glycerol
vs. the sample filament wire temperature (K).

Abundance

148

40000 -

30000 ‘

20000 -'

 

10000 "

 

 

 

 

Figure A23. Abundance of the [M]K“ ion of glycerol
vs. the sample support surface temperature (K).

Abundance

149

30000 -

20000 '-

10000 -

 

 

 

 

 

 

Temp (K)

Figure A24. Abundance of the [M]K+ ion of triethylene
glycol vs. sample filament wire temperature (K).

Abundance

150

 

 

   

 

      

 

 

20000 "
15000 '1
10000 "'
5000 "
1
0" " Itlt1ﬁv'tivtlt
300 350 400 450 500 550 600
Temp (K)

Figure A25. Abundance of the [M]K* ion of triethylene
glycol vs. sample filament wire temperature (K).

151

     

 

 

1oooo -
Ea calculated from slope of abundance vs. 1fl' (K) for
the mlz 392 fragment of met-enkephalin = 67 kcal/mol

aooo -
y = 7.0317e+4 - 3.3937e+4x me = 0.887
El
6000 ~

(D

O

C

(U

‘C

C

:3

.0

< 4000 -

2000 -
o I 1 I F I E l
1 a 1 9 2.0 2 1 2 2

1/Tx1000

Figure A26. Ea determination from the Arrhenius plot
of the abundance of m/z 392 of met-enkephalin vs. 1/T

Abundance

152

5000‘

y = 3.70656+4 - 1.1881e+4x R"2 = 0.890

4000'

3000‘

2000‘

1000‘

 

 

 

2.7

1H x1000

Figure A27. Abundance of the [M]K+ ion of palmitic acid
vs. 1/1' Ea calculated from slope = 23 kcal/mol.

153

1500 -

1ooo -
y = 4891.5 - 1924.7x R‘Z = 0.599

Abundance

500 -

 

 

 

1ffo1000

Figure A28. Abundance fo the [M]K" ion of palmitic acid
vs. 1fl'. Ea calculated = 4 kcal/mol.

15000 ‘

10000 "

Abundance

5000 "‘

 

154

y = 7.784le+4 - 2.4219e+4x R"2 = 0.856

Ea calculated from the slope for the
[M]K+ ion of glycerol = 48 kcal/mol.

 

 

2.8

I ' I ' I I I

2.9 3.0 3.1 3.2 3.3

1/T x1000

Figure A29. Abundance of the [M]K+ ion of glycerol vs. m (K).

155

    

 

 

20000 -
y = 1.9983e+5 - 6.32926+4x R‘2 .-= 0.816
J
Ea calculated from the slope for the

8 [M]K+ ion of glycerol = 126 kcal/mol
C
:0
.0 0000 -
C
3
.0
<

0 v I I I j I I

2.9 3.0 3.1 3.2 3.3
1H x1000 K

Figure A30. Abundance of the [M]K+ ion of glycerol vs. 1/T K.

156

 

 

 

40000 -
T y = 1.1428e+5 - 3.5870e+4x R‘Z = 0.614

30000 -
8 Ea for the [M]K+ ion of glycerol calculated
§ 20000 .. from the slope of the plot = 71 kcal/mol
C
23
.0
«<

10000 -
0
a]

 

1/T x1000

Figure A31. Abundance of the [M]K+ ion of glycerol vs. 1/T (K).

n(ln(Ao/At))

157

  

y = 14.062 - 6794.8x R42 = 0.98'.

-2.000 “

-4.000 "'

-6.000 "‘

 

 

-8.000ﬁﬁv111'1'1v1ﬁ1

0.0018 0.0020 0.0022 0.0024 0.0026 0.0028 0.0030 0.0032

1/T

Figure A32. Arrhenius plot of ln(ln(Ao/At))
vs. 1/T for the [M]K* ion of palmitic acid.
Ea calculated = 14 kcal/mol.

158

 

 

2 _
a
a o
o
1 - 90. y = 17.948 - 10.975x R"2 = 0.815
a
o
o o y = 38.991 - 23.503): R"2 = 0.993
e
O .1
g a 0 El 1010(AO/A1)543
B
s a 0 lnln(Ao/At) 363
c a
-- .. e
E. 1 a
a
an
- “13
2 o a
e a ‘3
o
-3- O
'4 I l V I j l ‘ I ‘ l
1.5 1.6 1.7 1.8 1.9 2.0
1/Tx1000

Figure A33. Arrhenius plot of ln(ln(Ao/At)) vs. 1/T for the
[M]K+ ion and a K" adduct ion of a fragment of melezitose.
Ea(543) = 22 kcal/mol and Ea(363) = 47 kcal/mol.

159

   
 
 
  
  

 

 

2 .—
. y = 29.070 - 17.846x R42 = 0.978
O
o .. y = 33.549 - 20.176x R‘Z = 0.904
la lnln(Ao/At) 543
:5 O lnln(Ao/At) 363
s -2 -
E
.E
.4 -I
o
'6 ' I ' I ' ﬁ ﬁ' I
1.5 1.6 1.7 1.8 1.9

1/Tx1000

Figure A34. Arrhenius plot of ln(ln(Ao/At)) vs. 1fT of the
[M]K+ ion and a K+ adduct ion of a fragment of melezitose.
Ea(543) = 35 kcal/mol and Ea(363) = 40 kcal/mol.

160

y = 34.555 - 20.732x R‘2 = 0.981

0 - ° y = 40.488 - 23.7251: R‘2 = 0.980

5' a lnln(Ao/At) 543

S a O

2 a o lnln(Ao/At) 363
E’

:5

 

 

1.6 ' 1.7 I 1.8 v 1.9 i 2.0
1fl’x1000
Figure A35. Arrhenius plot of ln(ln(Ao/At)) vs. 1/T for the

[M]K+ ion and K' adduct ion of a fragment of melezitose.
Ea(543) = 41 kcal/mol and Ea(363) = 47 kcal/mol.

161

    
  

 

 

2 .—
y = 13.529 - 7.4660x R‘2 = 0.984
0 y = 17.298 ~ 10.207x RAZ = 0.968
a lnln(Ao/At) 381
o
:3 o 0 lnln(Ao/At) 543
it
2 2- °
1.5
E
-4-
-6 ' I f I i I I f ﬁ I
1.4 1.6 1.8 2.0 2.2 2.4

1/Tx1000

Figure A36. Arrhenius plot of ln(ln(Ao/At)) vs. VT for the
[M]K ions of melezitose (m/z 543) and sucrose (m/z 381).

Ea(543)= 20 kcal/mol and Ea(381) = 15 kcal/mol.

162

   

y = 33.881 - 18.854): R‘2 a 0.961

ln(ln(Ao/At))

 

 

.8 "V'I‘V‘V'V'I‘"Yj—‘II'V‘ITYUj1

1.6 1.7 1.8 1.9 2.0 2.1 2.2

1/T x1000

Figure A37. Arrhenius plot of In(ln(A0/At)) vs. 1/T
for the m/z 392 fragment adduct ion of met-enkephalin.
Ea calculated = 37kca|lmo|

163

.926 mcmom 86 89296 com .83 82.3828 8 E2868 89: mo. +3 .mm< 9:9“.

#5
one owo onm cam 09. omv o; own own 0mm 0mm oPN our cow

—-h-——P-—--Ll—PP-.—-b-—n-Ln—nn-n—ppnp 71+? .1
1 —1 A— — d
I

mmm

 

mom 1
00¢ FNV hhm PNN th

0mm .-
mmN

 

 

 

 

mm. 1
3&8

 

_.ow hmw

ovw

 

 

 

 

Misueim 84119168

164

mm

/.\

ov 7%

* cmom mv L ll-l-II—Il:.i_

om. RF .
mm .F .
/L~H-_.
00 ﬂ .

.cmnEac cmom .m> com 6020 0535028 .0 9.8% 0.8:. mo_+_._ 9: _o 8.6 .mm< 9:9".

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

165

.c__m.._amxcm-mc_co_£mE 8 8:8QO 0%.: mo. +3 .ov< 239“.

NE.
ova one own 0; 000 can own cum com 0mm ovm own cum own oom omv omv onc

g

 

 

 

on”.
b

h’
4‘:

00%. on». 03 omv omv owv oov can own cum own can own can own own com

b‘Lth'bpr’lF-t>L—bF”_PP'b_hFDD—’DP__bbe-"Pb-’I.F__Lbb
l-1““ ‘ .1 1 ‘1 ‘1-1 .1“1 1‘ 1‘1‘111111 ‘ 1 li-l1d1‘ﬂ- -‘-1‘

 

_ au ‘1'. ‘l—-I“.‘{‘ 1 ‘

\um com own

I

l

vmm

I

mx

'l'l

 

 

 

com omN omm onm com 0mm ovw com CNN 05 com oar om? our cow of. 3... of.

mvm vm— ow? th I
mam IN on. I

‘1‘]

NNN

 

 

 

 

Ausuaxu' annelaa

166

65.53559: .0 828on 89: mo. +3 .52 93mm

NE

0mm 0mm Ohm owm omm ovm omm CNN OHN oom oma omH on.“ oma omH ova omH

_E__::_ E.:_

bun—pup-—--——-ppp—b-h—FPbb—u-.-——-b—-——b——pup—hhhn

 

 

mmm

mmm

 

 

h:_rp
__‘

 

 

167

.23 25.03 .o 828QO 89: mo. t... .33. 9:9...

0mm 0mm Ohm 0mm 0mm ovN 0mm ONN OHN oom om.“ omH ova om." omH ova oma

__.:—_:LL:F—p.:—:bp_:_L_E._bbp.__:.—L:p—~brb_._: :”____L.+LLL|_LLLL‘

 

 

mva

 

 

 

 

APPENDIX B

168

169

BIOMEDICAL AND ENVIRONMENTAL MASS SPECTROMETRY. VOL. l8. I77—l84 (I989)

 

Mass Spectrometric Analysis of Cardiac Glycosides
by the Desorption/Ionization Technique Potassium
Ion Ionization of Desorbed Species

 

Karen J. Light, Daniel B. Kassel and John Allison‘
Department of Chemistry. Michigan State University. East Lansing. Michigan 48824, USA

 

The analysis of cardiac glycosides by the desorption/ionization (D/l) spectrometric technique potasinm ion
ionization ofdesorbed species (K‘IDS) is presented. K‘IDS mass spectra ofdigitonin, digoxin, digoxigenin. digi-
toxin and ouabain are deemed to demonstrate the capabilities of this Dll method. The K‘IDS analysis consists of
two steps: thermal desorption of neutral molecules representative of the analyte. followed by gu-phaae addition of
K‘ ions to these species. Structural and molecular weight information of the cardiac glycosides is obtained with the
K’IDS technique. The most intense peak in the K‘IDS mass spectrum of an analyte, M, is frequently the [MIK’
ion. Interpretation of the K‘IDS mass spectra is simple. since one thermal degradation mechanism dominates.
This mechanism is a l.2-elimination process. A variation of the original K’IDS technique, performed by changing
the ionizirg metal from K‘ to Na" (i.e. Na‘lDS), is presented for the analysis of digoxin. The Na‘lDS mass
spectrum of digoxin contain more structural information than the K’lDS mass spectrum of that compound. “is
may lead to a means of controlling the types of information obtainable with this D]! techiqne by varying the
cation that '5 thermionically generated. K‘lDS analyses can be performed rapidly, no sample derivatintion is

 

necessary. no matrix is required and little instrument modiﬁcation is necessary.

 

INTRODUCTION

 

The cardiac glycosides are an important class of
cardenolide-type compounds. some of which are used as
therapeutic agents for the treatment of congestive heart
failure and certain heart arrythmias.l Their primary
effect is to increase the efficiency and contractility of the
heart muscle. One concern in using these compounds as
therapeutic agents is the very narrow range of concen-
trations over which their effects are therapeutic (e.g., for
digoxin, this range is 0.5—2.5 ng/ml in serum). Concen-
trations above 3.0 ng/ml are considered as toxic.2 As a
result, methods for identifying and accurately measuring
the relative concentrations of these drugs are needed.
Mass spectrometry offers one such method for accu-
rately detecting low levels of these compounds.

The analyses of cardiac glycosides, using a wide
variety of mass spectrometric approaches, have been
reported. These include electron ionization (El),3 chemi-
cal ionization (Cl),‘ desorption chemical ionization
(DCI),s ﬁeld ionization (F 1),“ ﬁeld desorption (FD),7 fast
atom bombardment (FAB),' liquid chromatography/
mass spectrometry (LC/MS)° and laser desorption
(LD).'° Extensive comparisons of these mass spectro-
metric ionization techniques for cardiac glycoside evalu-
ation have been presented in the literature.'°' Recently
the desorption/ionization (D/l) technique, K" ioniza-
tion of desorbed species (K*lDS), was reported for the
analysis of thermally labile compounds by mass spec-
trometry.” Presented here are the results of K‘IDS
analyses of several cardiac glycosides, to allow for
evaluation of the method in the context of other D/l

0887—6l34/89/030177-08 $05.00
© I989 by John Wiley & Sons. Ltd.

techniques. The results show the strength of K‘lDS for
the analysis of cardiac glycosides. Extensive compari—
sons to other techniques are limited here, to avoid
redundancy with recently published review articles.‘°'

The D/l technique K‘lDS utilizes a thermionic
emitter which produces a high flux of alkali ions in the
gas phase when heated to temperatures in excess of
800 °C.‘2 When a thermally labile compound is depos-
ited onto the surface of such an emitter, which is then
rapidly heated, K" adduct ions of desorbing neutrals
are formed in the gas phase.“ K‘lDS mass spectra of
most thermally labile compounds frequently contain
K+ adducts of the intact molecule (M), [M]K‘, and of
various thermal decomposition products (DI), [DJK*.
The extent to which desorption and/or decomposition
of M occurs is dependent on the heating rate and the
ﬁnal temperature achieved, and, of course. on the mol-
ecule M. Beuhler and coworkers have proposed that the
rapid heating of a sample to a high temperature can
access conditions at which the rate of vaporization of a
thermally labile compound may exceed the rate of
decomposition. ' 3

In reviewing the literature for mass spectrometric
analyses of cardiac glycosides, we found a number of
approaches used to identify fragment ions produced by
the various D/l mass spectrometric techniques. Many
designation schemes were sufficiently vague that the
reader must calculate the exact composition of the ion.
Others were more complex than necessary for determin-
ing the sequence of carbohydrates. Therefore we devel-
oped a scheme for our needs that will be used here and
in subsequent publications. Details are provided in the
Appendix. We have found it to be useful for discussion

Received 25 May I988
Revised 6 September I988

 

 

170

I78 K. J. LIGHT. D. B. KASSEL AND J. ALLISON

of ionic and neutral species derived from saccharides
and saccharidc-containing compounds.

 

EXPERIMENTAL

 

The cardiac glycosides were obtained commercially and
used without further puriﬁcation or derivatization. The
digitonin was obtained from Merck & Co. Inc.,
Rahway. New Jersey. The digoxin. digitoxin, ouabain
and digoxigenin samples were obtained from the Sigma
Chemical Co, St Louis, Missouri. The samples were
suspended or dissolved in acetone to allow transfer of
l—2 pg of sample onto the tip of the probe. The solvent
was then evaporated before inserting the probe into the
mass spectrometer.

All mass spectrometric analyses were performed on
an unmodiﬁed HPS985 GC/MS/DS quadrupole mass
spectrometer with a mass range of lO—lOOO u, equipped
with a direct insertion probe (DIP) inlet. The mass spec-
trometer ion source was operated in the El ‘open'
source conﬁguration. The construction of the K‘lDS
probe has been described previously." The thermionic
emitter is an alkali aluminosilicate mixture
(lK,O:lAI,O,:ZSi0,).”

In the early stages of the K‘lDS probe design,
analyte was deposited directly onto the K‘ bead."
Desorption of analyte from the bead surface occurred
rapidly when the K+ emitter was resistively heated to
high temperatures (>800°C) while the onset of K+
emission occurred after 5—6 s. Formation of adduct ions
requires a sufficient temporal overlap of the alkali emis-
sion and analyte desorption." In order to improve the
temporal overlap between the emission of K‘ ions from
the thermionic emitter and the desorption of analyte
neutrals, a modiﬁcation of the original probe design was
made. This new design incorporates a separate ‘ﬁla-
ment‘, on which the sample is deposited." The second
ﬁlament is not directly heated, but is in close proximity
to the bead and is therefore radiatively heated. Initial
results using this two-ﬁlament design have shown an
approximate «lo-fold enhancement in sensitivity. This
new probe design is currently under detailed study for
further optimization and characterization.

CHZOH

0

OH
HO 0

HOH

04on
0

oK‘ 0H
m/z 201(— HO

OH

162u

 

RESULTS AND DISCUSSION

 

The K‘IDS mass spectra of the simplest sugars glucose
and sucrose have been reported.” Abundant molecular
adduct ions as well as K+ adducts of thermal degrada—
tion products can be formed. For sucrose, a disaccha-
ride, ions are observed at m/z 201 and 219 which are
formed by a 1.2-elimination about the glycosidic linkage
(Scheme I).

Cleavage of one bond (3 glycosidic bond) followed by
an H-shift (to the glycosidic oxygen) occurs to produce
two stable neutral species in a low-energy process. This
mechanism is dominant for most types of molecules
because little energy is required. For example, consider
a simple molecule such as methyl ethyl ether. While
cleavage of the C—0 bond to yield two radicals. reac-
tion (1), requires 318 U/mol, only 67 kJ/mol is required
for a 1.2-elimination, i.e. C-O cleavage with an accom-
panying H-shift, reaction (2).

CH2 _ CH2 0C2H5 ‘ 'WJ (I)
J. 1......
CH2 - CH2 . “OCH; (2)

This mechanism is dominant in the thermal degradation
of saccharide-containing molecules analyzed by K‘lDS.
Degradation occurs in the condensed phase in a K‘IDS
analysis; both neutrals from this process can desorb
into the gas phase and undergo K+ attachment to
produce [DJK‘ adducts Such 1.2-elimination reac-
tions dominate thermal degradation processes at high
temperatures and short times, and can be used to
explain most of the [DJK+ ions formed in K‘IDS
analyses. Dehydration is another low-energy thermal
degradation pathway, again a 1.2-elimination. Both the
parent molecule, M, and decomposition products, D.,
can undergo dehydration reactions (when hydroxy
groups are present) and desorb, leading to ions such as
[M — H,O]K" and [DI — H,O]I(‘.

The ﬁrst cardiac glycoside that will be discussed is
digitonin (C56H92029. mol.wt = 1228, structure I; Fig
Al). The complete K’IDS mass spectrum of digitonin
cannot be presented here since the instrument used has

CHZOH

0

HO
CHZOH

0H

1,2-eliminetion

04on
0

“0 CH CH .K. /
2 a m 22:9
0H
IBOu

Scheme 1

171

MASS SPECTROMETRIC ANALYSIS OF CARDIAC GLYCOSIDES 179

a limited mass range (lo-1000 u). The molecular adduct
ion, [M]K‘, would be at m/z 1267. Despite this draw-
back we begin our discussion with digitonin since it is
the most complex of the cardiac glycosides studied. The
ions observed will serve as a framework for discussing
the nomenclature scheme that we use and the relation-
ship between the ions observed and the structure of the
analyte. It is assumed that the 1.2-elimination mecha-
nism, which dominates the thermal degradation of
many other types of molecules, is Operative for cardiac
glycosides This will be the basis for the assignments of
the observed ions, and is reasonable as the following
discussion will show. Unfortunately K‘lDS mass
spectra are short-lived, and the quadrupole instrument
currently being used cannot perform exact mass mea-
surements, but the consistent observation of the 1.2-
elimination degradation products allows these assign-
ments to be made with some conﬁdence.

Using the nomenclature presented in the Appendix,
the digitonin molecule is designated as
AOS,OS,(OT,)OS,OS4 (or as M, see structure I in
Appendix). The ions observed below in]: 1000 are listed
in Table 1. Based on the mechanism observed for other
compounds. especially sucrose. it is postulated that the
glycosidic bonds will be the site of thermal degradation.
For example, if the 0-8‘ bond is cleaved, the fragment
designated as S. has a mass of 163 u. A K‘ adduct is
observed at m/z 201, which is the alkali ion attached to
a neutral of mass 162 u, indicating that an H has shifted
off of the sugar (via a 1,2-elimination) The other neutral
product of this degradation reaction, [M - (S; ")] has a
mass of 934 u. The vaporization of this latter molecule
followed by K’ attachment leads to an ion at m/z 973.
The observation of both ions in the K‘IDS mass spec-
trum further supports the proposed elimination of S; ".
In this way, the remaining ions in the mass spectrum
are easily related to the structure of the molecule, as
indicated in Table 1. It should be stressed again that the
large number of molecules studied to date strongly
support this mechanism, making interpretation simple.

 

Table l. Assignment of the ion in the K‘IDS
maasspectrnm ofdigitonin
ann' Aauonnnnt
1267' [M]K' = [AOS,OS,(OT,)OS,OS.]K‘

973 [Aos,os,(0°")os,o'"]x°
943 [M - (s.os;“nx'

927 [M - (s.os,or"nxr

811 [AOS,OS,(O°")O'”]K’

795 impurity“ = [A'OS.O$,(O°")O’"]K’
633 [AOS,‘"]K’

649 [AOS,O’”]K‘

495 (M — (AOS,OS,(O’")0‘")]K’
487 [AO’"]K°

471 impurity” = [A'O’”]K’

363 [s.os;"]xr

201 [s;"]x°

171 U5“]K‘

' Ion not observed. since it exceeds the mass limit
of the mass spectrometer used in this study.

“The desrgnallon N indicates that the Impurity
contains an aglycone structure that differs from
that in digitonin. as suggested in Ref. ‘6.

 

The thermal conversion of one molecule into two stable
molecules, by this 1.2-elimination mechanism, is ther-
modynamically the most favored process.

Even though the mass limit of the instrument used in
this study prohibited the detection of the [M]K+ ion at
m/z 1267, it has been observed by another researcher on
a higher-mass instrument.“ In this case the [M]K‘ ion
was not the base peak in the mass spectrum, but it was
the only important ion above m/z 1000.

There were a few peaks that did not easily correlate
with the structure of digitonin but did correlate with a
common impurity of digitonin that has been identiﬁed
by others” and observed by an FAB analysis per-
formed in this laboratory. The impurity corresponded
to a dcoxy version of the genin of digitonin, and the
impurity ions were 16 u less than major ions of digito-
nin. Therefore the assignment of these ions as impurities
logically followed.

Another reason for beginning this discussion with
digitonin is to show the time dependence of the spectra
obtained with the K‘lDS technique, unlike other D/l
techniques such as FAB. A three-dimensional plot
depicting the mass spectra collected v. scan number
(time) for the entire K‘IDS analysis of digitonin is
shown in Fig. l. The scan rate was I s/scan and the
analyte signal typically lasted for approximately 30 s.
Early in the run, low—mass ion formation was favored,
while at intermediate times the higher-mass ions
became prevalent. Late in the run, both intermediate
and low-mass ions were observed. Activation energies
for the formation of speciﬁc decomposition products
may differ and would explain the observed temperature/
time dependence.” Thus it may be possible to extract
information about speciﬁc decomposition pathways (i.e.
the relative activation energies for the formation of the
products) from the temperature or time at which they
are formed during a K ‘103 analysis. and relate these to
structural features.

The emphasis of this paper, however, is to demon-
strate the wealth of information concerning the struc-
ture and carbohydrate sequence of cardiac glycosides
that can be obtained with the K’lDS technique, and
the simplicity of K‘lDS mass spectra. The remaining
K’lDS mass spectra that will be shown are generated
by averaging spectra from all of the scans over the
entire desorption proﬁle (approximately 30 s in
duration).

DIGOXIN 0
/
Hocn

CM3
CH3 CH3 CH3
0 o o 0“
o 0 0
HO
OH 0H 0H

Digoxin (CullaOu, mol.wt = 780, structure II). a
commonly used therapeutic agent, is a three-sugar
cardiac glycoside. Molecular weight as well as abundant
structural information can be obtained for digoxin
using the K * IDS technique. The averaged K ' IDS mass
spectrum of digoxin is shown in Fig 2 and assignments

(II)

Rn].
Int.

100 200 300 400 500 600 700

mlz

17

 

800

2

K. J. LIGHT. D. B. KASSEL AND J. ALLISON

Scan 0

 

soo‘b

900

Figure 1. Time dependence of the 10108 mass spectra of digitonin (structural. mol.wt -1228).

for the observed ions are summariud in Table 7. The
base peak for digoxin is the [M]K‘ ion at m/z 819. All
of the major cleavages at the glycosidic linkages
produce two neutral species that undergo K+ addition.
These bond cleavages follow the 1.2-elimination mecha-
nism as described for digitonin and sucrose. The low-
mass ion series will be addressed shortly.

DIGITOXIN

Digitoxin (C.,H“Ou. mol.wt a: 764, structure III),
is a cardiac glycoside that differs from digoxin by the
absence of the hydroxy group on the Q12) position of
the genin portion of the molecule and thus has a molec-
ular weight 16 u below digoxin. The K‘IDS mass spec-
trum of digitoxin is very similar to that of digoxin. The
K+ adduct ions observed and their assignments are
given in Table 2. All of the ions containing the genin
portion are shifted 16 u lower in the digitoxin mass

(III)

 

 

TahlelAmig-nentofionshtheK‘IDSmspectraof

 

 

an

RELATIVE INTENSITY

 

 

 

 

'itltlltk%jrvvt ‘I’YYY‘IYY
0° 20°

I ’00 400

Iisoxh. chm xin. ouabain and disotise-n‘
mp Autumn
Digoxin 819 [M]K‘ - [AOS,OS,OS,]K’
801 [M - H,O]K’
689 [AOS,OS,O’°']K‘
659 [AOS.O”‘]K’
429 [AO”']K‘
[M - (A0’")IK‘
299 [$.085"]K‘
Digitoxin 803 [M]K’ - [AOS,OS,OS,]K’
785 [M - H,O]K°
673 [AOS,OS,O“']K’
643 [AOS,O”‘]K‘
41 3 [AO°")K‘
429 [M - (AO”')]K°
299 [S,OS;"]K°
Ouabain 623 [M]K’ - [AOSJK’
605 [M - H,O]K°
477 [AO°”]K’
459 ' [A"‘]l(’
Digoxiganin 429 [M]K’
411 [M - H,O]K’
393 [M - 2H,O]K’
I“
an
ﬂ”
Y'IVYI‘IU'I‘YTTer] I]
500 ‘00 700 .00 .00
mlz

Figure 2. K‘IDS mass spectrum of digoxin (structure ll. mol.wt = 780).

MASS SPECI' ROMET RIC ANALYSIS OF CARDIAC GLYCOSIDES

spectrum relative to those formed from digoxin. For
digoxin, two structures are assigned for m/z 429, but
only one contains the aglycone portion of the molecule.
For digitoxin, there is still an m/z 429, and the product
containing the aglycone is now 16 u lower at m/z 413.
This suggests that, for digoxin, the ion current at m/z
429 represents both structures as indicated in Table 2.

OUABAIN

H0

 

0H 0“

Another cardiac glycoside, ouabain (c,,H..o,,.
mol.wt - 584, structure IV), is well characterized by its
K‘lDS mass spectrum. The averaged mass spectrum of
a K’IDS analysis of ouabain is shown in Fig 3. This
glycoside contains only one sugar and a slightly differ-
ent steroid moiety. The ion of highest ml: in the spec-
trum, which is also the base peak, provides the
molecular weight information, since it is the [M]K+ ion
at m/z 623. Structural information is also available. The
ion at m/z 477 is the K * adduct of the aglycone portion
and is formed in a manner identical to that described
for structural ions observed for digitonin. Table 2 gives
the assignments of the major ions in this mass spectrum.
Interestingly, the same low-mass ion series between m/z
105 and 215 u that was observed for digoxin is present
in the K ’IDS mass spectrum of ouabain. These ions are
separated by 12, 13 and 14 u, which suggests an aro-
matic system. Either the sugars or the aglycone portion
could be thermally degrading to an aromatic species.
Ouabain and digoxin have different numbers of sugar
groups in the molecule. Therefore it seems that this low-
mass ion series might be from the aglycone, upon con-
version to a highly reduced form. If this low-mass ion
series proves useful for indicating the presence of a
steroid group, this would offer an advantage of K’IDS
over FAB in which this low-mass ion series would be
lost in the chemical noise associated with that tech-
nique.

173

181

To investigate the origin of this low-mass ion series
digoxigenin was studied, which is not a cardiac glyco-
side but rather the genin, or steroid portion, of digoxin.
The K‘lDS mass spectrum of digoxigenin (C,,H,‘O,.
mol.wt = 390, structure V) is shown in Fig. 4. Once

DIGOXIGENIN
0
/
“O C 0
CH
3 (V)
on

“0

again, the base peak is the [M]K‘ ion at m/z 429. The
same low-mass ion series, between all: 105 and 215,
does appear which shows that they arise from the
steroid portion of the molecule. The prominent low-
mass ions are tabulated in Table 3 along with proposed
assignments. The justiﬁcation for the proposed assign.
ments is as follows. These ions are not typical of those
observed in K’IDS analyses of compounds containing
only C, H and 0 since they appear at both odd and
even nt/z values. The mechanism that is consistently
operative in K’IDS should only form even mass
neutral products for these molecules which then appear
as odd mass ions following K+ (mlz 39) attachment.
The mass differences between the ions might suggest
that they are derived from aromatic species. It is sug-
gested that these low-mass ions are products of surface
ionization of the reduced aglycone moiety (CHI-In,
mol.wt a 216, structure VI) (or its precursors) for two
reasons. First, surface ionization of molecules with low

REDUCED AGLYCONE

(VII

ionization energies has been observed on these emit-
ters." Second, the ionization energy. LE. for multi-
cyelic hydrocarbons decreases as the size of the
molecule increases and the extent of unsaturation

 

 

L

RELATIVE INTENSITY
l a

1

Jill It

I

   

 

623

 

477

JI A-

 

 

 

rYT

 

l
I
no ‘I .0

Yﬁfj‘l’Yﬁ'j I

1
11171
m

M

I T I l 1 l I I

e90 000

Figure 3. K‘IDS mass spectrum of ouabain (stnrcture IV. mol.wt - 5&4).

174

182 K. J. LIGHT. D. B. KASSEL AND J. ALLISON

 

,—— X5
.1
128

< 115

RELANVEINTENSNV

429

 

 

 

141 165
105 153
, 179
202
_J f LI 1
j
90 190 290

"V2

 

Figure 4. K’lDS mass spectrum of digoxigenin (structure V. mol.wt - 390).

 

Table3. Low-mass ion series observed in the K’IDS
spectra of onahain, digoxin anddigoxigenln

am W ark m
105 [c.H.]° 165 [camr
115 MW 178 re..H..1°
123 [C,.H.] ° res [c,,H,] r
141 [C"H,] ’ 202 [C,.H,°]°
153 IciaHaI. 2‘5 [Canl‘

 

increases (e.g l.E.(benzene) a 9.25 eV. I.F.(nap-
thalene)= 8.12 eV. I.E.(anthracene)= 7.5 eV)." This
would also explain why the same low-mass ions are
observed for cardiac glycosides with aglycones that
contain different functional groups. Work is still in
progress to determine whether this proposed origin of
these low-mass ions is valid. However, in the ﬁnal
analysis, these ions are very characteristic of a steroid
structure, both in terms of the mass range at which they
appear, and when they appear during the K‘IDS
analysis.

In addition to K‘lDS, analyses have been performed
using an Na’ emitter. The Na’IDS mass spectrum of
digoxin is shown in Fig 5. There are several noticeable
differences in the Na‘IDS mass spectrum of digoxin to
that obtained utilizing K‘IDS (Fig 2). The [ijNa+
ion is not the base peak of the spectrum in Fig. 5. The
lower-mass ion peaks from cleavage of each glycosidic
bond (observed as Na‘ adducts) are relatively more

intense than in the K ’IDS mass spectrum. Also the loss
of H30 from each of the major fragments is observed,
providing additional structural information. This sug-
gests that Na+ may promote fragmentation (when K’
does not) upon adduct formation in the gas phase, such
as reaction (3).

Na‘ + DI -o (D. + Na*}‘ _. (D, — H,O]K‘ + H10
(3)

Thus, the metal ion used for ionization in this process is
a variable that can be selected to obtain different infor-
mation about the analyte. It has been shown that Li‘ is
more reactive in the gas phase20 than Na‘ or K‘, and
would be more likely to undergo reactions such as (3).

 

CONCLUSIONS

 

K‘IDS has been demonstrated as a useful ionization
technique for the mass spectrometric analysis of cardiac
glycosides, and the mass spectra obtained are compar-
able to those obtained from other desorption/ionization
techniques that are currently used in mass spectrometry.
The K‘IDS mass spectra are obtained very rapidly,
within 1 min, and are relatively clean. Thermal pro-
cesses occurring at these short times can produce intact
desorbed molecules, even though these compounds are
considered to be ‘thermally labile‘. Also. the operative

 

283

413

- 153

HELATWEINIENSHV
4

 

I‘—
p

I 1
trt

543
673

903

 

 

 

r fir 1 L :A r t I t ‘7 1r r I
12° 23° ’30 43°
M

Ll sisal 1 Al
'l‘YY-ijr‘lﬁ

1‘7

I
910 sec 720 nae

Figure 6. Na’lOS mass spectrum of digoxin (structure If. mol.wt = 780).

175

MASS SPECTROMETRIC ANALYSIS OF CARDIAC GLYCOSIDES 183

mechanisms under these conditions appear to be simple
1.2-eliminations, thus thermal degradation products can
be easily related to the parent compound. K‘ ions
appear to have a sufficient afﬁnity for organic molecules
such that adduct formation can occur. However, the
energy released on complexation is insufficient to induce
fragmentation, so the K+ ions simply ‘sample' the
desorbed species. For molecules which only desorb
intact, with no decomposition at temperatures used in
K’IDS, only the [M]K‘ ion is observed. However. the
same type of analysis with another thermionically gen-
erated cation, such as Na+ or Li‘, may provide some
structural information via gas-phase reactions. Thus by
changing the metal ion used in the analysis, it may be

possible to obtain different types of information about
the analyte. We are currently adapting this technique to
a JEOL HX-llO mass Spectrometer, for studies of
higher molecular weight compounds, and to perform
B/E linked-scanning analyses of speciﬁc ions.

Acknowledgements

The authors wish to thank T. Brody for providing the cardiac glyco-
side samples. This work was made possible through the ﬁnancial
support of: the Analytical Labs of the Dow Chemical Company; the
United States Department of Agriculture (grant no. USDA-ARS-59-

’ 32U4-7-107, administered by the MSU Center for Environmental

Toxicology); and the National Institutes of Health (NIH grant no.
880048046).

REFERENCES

 

1. T. W. Smith. Digitalis Glycosides. Grune 8t Stranon. Orlando
(1986).

2. A. Goth. Medical My. 11th sdn. p. 431. Mosby. St
Louis (1984).

3. (a) F. C. Falkner. J. Frolich and J. T. Watson. Org. Mass
Spectrom. 7. 141 (1973); (b) P. Brown. F. Bruschvveiler and
G. R. Pattit. Halv. Chim.Acta 65. 531 (1972).

4. J. Vine. L Brown. J. Boutagy. R. Thomas and 0. Nelson.
Biomed. Mass Spectrom. 6. 415 (1979).

5. (a) A. Kappaler and U. Richli, Advances in Mass Spectrom-
etry. Part 8. 10th Int. Mass Spectrom. Cont. p. 1497. Mley.
New York (1985); (b) A. P. Bruins. Int. J. Mass Spectrom.
Ion Proc. 48. 185 (1983); (c) A. P. Bruins. Anal. Chem. 52.
605 (1980).

6. (a) P. Brown. F. Bruschvvailor and G. R. Pattit. Org. Mass
Spectrom. 6. 573 (1971); (b) P. Brown. F. Brusehweilor. G.
R. Pottit and T. Reichstoin. J. Am. Chem. Soc. 92. 4470
(1970).

7. (a) T. Komori. 1’. Kawasaki and H. R. Schulten. Mass Spoc-
trom. Rev. 4. 255 (1985); (b) H. R. Schulton. T. Komori. T.
Nohara. R. Hiquchi and T. Kawasaki. Tetrahedron 34. 1003
(1978); (c) H. R. Schulten and D. E. Games. Biomed. Mass
Spectrom.1. 120 (1974).

8. (a) J. R. J. Pare. P. Lafontaina. J. Bolanger. W. W. 8y, N.
Jordan and J. C. K. Loo. J. Phann. Biomed. Anal. 5. 131
(1987); (b) R. lsobe. T. Kornori. F. Abe and T. Yamauchi.
Bloated. Environ. Mass Spectrom. 1 3. 585 (1986).

9. O. E. Games. M. A McDowall. K. Lavsen. K. H. Scholar. P.

Dobberstain and D. L Govver. Biomed. Mass Spectrum. 11.
87 (1984).

10. (a) R. E. Shomo II. A Chandrasekaran. A. G. Marshall, R. H.
Reuning and L W. Robertson. Blamed. Environ. Mass Spac-
rrom. 15. 295 (1988); (b) J. C. Tabst and R. J. Cotter. Anal.
Chem. 56. 1662 (1984); (c) M. A. Posthumus, P. G. Kistema-
ker. H. L. C. Meuzelaar and M. C. Ten Noevor do Brauw. Anal.
Chem. 50. 985 (1978).

11. D. D. Bombick and J. Allison. Anal. Chem. 59. 458 (1987).

12. J. P. Blewett and E. J. Jones. J. Phys. Rev. 60. 464 (1936).

13. R. J. Beuhler. E. Flanigan. L J. Greene and L Friedman. J.
Am. Chem. Soc. 96. 3990 (1974).

14. D. B. Bombick and J. Allison. Anal. Chim. Acta 208. 99
(1988).

15. D. B. Kassel and J. Allison. Biomod. Env. Mass Spectrum. 17.
221 (1988).

16. W. J. Simonsick. E. l. du Pont de Nemours It Co., Philadelp-
hia, personal communication (1988).

17. Y. M. Yang, H. A Lloyd. L K. Pannoll. H. M. Fates. R. O.
Macfarlane. C. J. McNeal and Y. lto. Blamed. Env. Mass
Spectrom. 13. 439 (1986).

18. D. Bombick. J. D. Pinkston and J. Allison. Anal. Chem. 66.
396 (1984).

19. J. L Franklin. J. G. Dillard. H. M. Rosanstock. J. T. Horton. K.
Draxl and F. H. Field. Nat. Stand. Rel. Data 5a.. Nat. Bur.
Stand. (US) 26. 289 (1969).

20. J. Allison and D. P. Ridge. J. Am. Chem. Soc. 101. 4998
(1979).

APPENDIX

 

Analyzing the K’IDS mass spectra of the cardiac gly—
cosides required the aid of a labeling scheme to accu-
rately identify the fragments observed as K” adduct
ions. Such a scheme was not found in the literature that
would show the exact designation, including the H-
shifts observed, and would simply relate the fragments
to the intact molecule. Therefore a scheme was designed
for use in this paper as well as subsequent works uti-
lizing K‘IDS and FAB from our laboratory on
saccharide-containing compounds. This scheme was
pattegned after one proposed by Pettit and Brown in
1971.

Digitonin (structure I; Fig. AI) will be used as an
example for explaining this naming scheme since it is
the most complex compound discussed in this paper.
The non-sugar or steroid portion of the molecule is
labeled A for aglycone. The numbering of the sugars in
this molecule is similar to that used for designating the

position of C atoms in alkanes. The sugars in the main
(longest) chain are labeled as S, , where n is the index for
the position of that sugar along the chain, in this case
the distance from the aglycone. The sugars in the
branch are labelled T... The glycosidic oxygens between
sugars are explicitly designated in the naming scheme.
Therefore, we designate the molecule in a linear short-
hand form as AOS,OS,(OT3)OS3OS.. The branched
sugars have an index number that identiﬁes their dis-
tance from the aglycone.

With the designation of the parent molecule com-
pleted, the identiﬁcation of cleavage products can be
addressed. As discussed in the paper, the 0—54 bond is
broken with a concurrent H-shift to the genin portion.
This H-shift is designated by “—H‘ and ‘+H' super-
scripts; thus the two neutrals would be labeled as [SI "]
and [AOS,OS,(OT,)OS§"]. The later fragment could
also be labeled as [M — (S; ")1 with the selection

176

[84 K. J. LIGHT. D. B. KASSEL AND J. ALLISON

DIGITONIN

 

r3-o- T2—o—s.—o—a E aos,os,(or,)os,os4

53-0
l

s4§o

 

ts.’“lx'

laos .os,tor,)os,0°" I V

Figure A1. The structure at digitonin is shown. and its relationship to the shorthand designation scheme. The use at the designation
scheme for labeling ’lragment ions‘ is shown here lor the ions observed at m/z 201 and 973.

between the two possible names depending on the
emphasis of the discussion and/or convenience. The
ions observed with the K‘IDS technique are K‘
adducts of these neutral fragments, so they are rep-
resented as [ ]K*. This naming scheme allows for
designation of multiple cleavages, such as those
observed in the analysis of digitonin. For example, two

1.2-eliminations of the digitonin molecule lead to the
formation of the ion appearing at m/z 811 which is
designated [AOS,OS,(O*")O* JK‘. The 0-8, and
O-‘I‘3 bonds are broken and the S, portion retained
both glycosidic oxygens and gained the H's from the
H-shifts.

APPENDIX C

177

178

 

 

Mechanistic Considerations of the Protonation
and Fragmentation of Highly Functionalized
Molecules in Fast Atom Bombardment: High
Resolution Mass Spectrometry and Tandem
Mass Spectrometry Analysis of the Ions
Famed by Fast Atom Bombardment of
Digoxin and Related Cardiac Glycosides

Karen J. Light and John Allison

Department of Chemistry. Michigan State University. East Lansing, Michigan, USA

 

High resolution mass spectrometry and tandem mass spectrometry analyses of the major
ions of digoxin formed by fast atom bombardment are presented and discussed to investi-
gate the mechanisms through which fragment ions are formed. Similar cardiac glycosides
are also analyzed to provide support for the proposed fragment assignments. Remote site
fragmentation with the charge localized on the aglycone portion of the molecule may pro-
vide an explanation for the fragment ions observed in these studies because the majority of
these ions contain the aglycone portion of the molecule. The results obtained parallel previ-
ously reported results from an ammonia chemical ionization mass spectral study of cardiac

glycosides. (I Am Soc Mass Spectrum 1990, 1, 455—472)

 

 

its power for structure elucidation is built, is the

understanding of the mechanisms by which
molecular ions fragment in the gas phase. For the
ionization technique used most often in mass spec—
trometry. electron ionization (El), fragmentation
mechanisms have been well characterized and utilized
to understand the relationship between the m/z val-
ues and relative abundances of the ions formed and
the structure of the compound under study [1]. When
BI is utilized, the relative abundance of a fragment ion
in a mass spectrum, as well as its m/z value, gives
some information on the structure of the ion, the
environment in the molecule from which it is formed,
and the type of mechanism involved in its formation
[1]. However, mass spectrometry has moved away
from El—based techniques toward a variety of desorp-
tion ionization techniques that can be applied to the
analysis of larger, more highly functionalized
molecules. While the number of ionization methods
available for mass spectrometry is now substantial.
the mechanisms through which fragment ions are
formed via these methods are not well understood.

THE cornerstone of mass spectrometry, on which

 

Address reprint requests to John Allison, Department of Chemistry.
Michigan State University, East Lansmg, MI 48824.

© 1990 American Society for Mass Spectrometry
1044-0305/90/3350

and their establishment has not been extensively pur-
sued to date. Frequently there is little discussion of
fragmentation mechanisms in the mass spectrometry
analysis of large molecules, and those mechanisms
that are proposed and utilized usually have not been
substantiated. Often the interpretation relies only on
the presence/absence of ion current at a particular
m/z value. with abundance information being of rela-
tively little utility.

The understanding of fragmentation mechanisms
not only facilitates the interpretation of mass spectra
of unknown compounds, but is vital when known
compounds are under study in which an isotopic label
has been incorporated. and the position and extent of
the label incorporation must be ascertained. This has
become important for larger molecules in the study of
metabolic pathways [2] in which a labeled compound
is introduced into a system and its fate is followed
with mass spectrometry, by monitoring label incorpo-
ration into metabolites. Thus. mechanistic aspects of
fragmentation for larger molecules should be consid-
ered when ionization methods other than El are used.
As it is somewhat impractical to expect that extensive
labeling studies be performed on large molecules. we
evaluate here the use of the tools that are available on
a conventional double-focusing sector instrument—
high resolution mass spectrometry (peak matching)

Received February 27, 1990
Accepted May 10, 1990

456 LIGHT AND ALLISON

and collisionally activated dissociation (CAD) method-
ology—for providing insights into the ions produced
by fast atom bombardment (FAB). In particular, we
focus on the FAB mass spectra of cardiac glycosides.
and further focus on the molecule digoxin. We have
chosen a case where the protonated molecule and
many fragment ions are observed in the FAB mass
spectrum. The CAD spectra obtained from linked
scanning at constant B/E for the ions observed, and
the results of accurate mass measurements, will be
presented and discussed. We will assume that for this
type of analyte molecule the dominant mode of ion-
ization is essentially glycerol chemical ionization in
which protonated glycerol (or some fragment ion de-
rived from glycerol) protonates the desorbed neutral
molecule in the gas phase. and fragmentation follows
protonation [3].

The data will be evaluated in the context of basic
questions concerning the site of protonation and frag-
mentation mechanisms for these highly functionalized
molecules, typical of those studied by FAB and liquid
secondary ion mass spectrometry (LSIMS).

Experimental

The cardiac glycosides were obtained from Sigma
Chemical Co., St. Louis, MO. and were used with-
out further purification. Acetyldigitoxin was purchased
from ICN K&K Laboratories, Cleveland. OH. The sam-
ples were dissolved in methanol to concentrations of
approximately lag uL". Two microliters were trans-
ferred to the FAB probe tip and mixed with the gly-
cerol matrix. All FAB analyses were performed on
a JEOL FIX-110 double-focusing mass spectrometer
(JEOL. Ltd.. Tokyo, Japan) of forward geometry with
an accelerating voltage of 10 kV and a FAB gun volt-
age of 6 kV with xenon FAB gas. Peak matching was
performed with a resolving power of 7.000 or more.
using glycerol cluster ions as reference ions.

All CAD experiments were performed by linked
scanning (at constant B/E) controlled with the JEOL
IMA-DASOOO software and using helium as the col-
lision gas. The ability to compare CAD spectra in
terms of the daughter ions observed was important
for their utility in the mechanistic considerations ad-
dressed in this article. This suggested the need for
performing these CAD experiments under single colli-

sion conditions, as opposed to multiple collision condi- '

tions where CAD of CAD products could be observed.
Therefore all CAD experiments were performed by
introducing helium into the collision cell so that the
signal for the parent ion was attenuated by 10%.
which produced single collision conditions [4]. In our
initial studies we were intrigued that low mass ions at
m/z 113 and m/z 131 were dominant ions in the FAB
spectrum of digoxin, but were not observed as CAD
daughter ions of the [Mjl-l+ ion of digoxin. This was
particularly distressing in light of a previously re-

179

I Am Soc Mass Spectrom 1990, I. 455-472

ported CAD mass spectrum of digoxin in which these
two low mass ions were observed [5]. Our prelimi-
nary CAD data were collected with no instrumental
changes from that used in the FAB mode and with a
tuning file that contained ions from m/z 39 to m/z 984
(from a mixture of K1 and Csl). However, the B/E
linked scanning software program used with the IEOL
HX—llO double-focusing mass spectrometer to per-
form these CAD experiments requires a good mag-
netic ﬁeld calibration table. A relation between the
calibration table lowest mass number (m'), the linked
scan parent ion mass number (m,), and the lowest
observable daughter ion (m2) is suggested in the
)EOL instruction manual by the following formula [6]:
m‘ = (m2)2/(m,). Therefore a different calibration
compound. Ultramark 1621. from I’CR lnc..
Gainesville, FL. was selected to allow for the con-
struction of a tuning file that contained many low
mass ions. with m/z 1 being the lowest mass ion
included. In addition to this change of calibration
compound, performing CAD experiments on the JEOL
HX-llO requires repositioning the conversion dynode
and opening the slit between the electric sector and
the magnetic sector to enhance detection of low mass
fragment ions that have lower kinetic energies than
their parent ions [7]. These modifications allowed us
to observe the expected low mass daughter ions at
m/z 113 and m/z 131 from the protonated digoxin
molecule.

In some cases, where the parent ion was within
approximately 5 u of a mass spectral peak derived
from glycerol alone, interference from the glycerol
matrix appeared in the CAD spectra and made it
necessary to use an alternative matrix. thioglycerol.
Another form of glycerol interference occurred when
the parent ion selected for CAD analysis had a low
relative abundance compared to the glycerol adduct
ions. In this case, a cluster of ions 90, 92. 94, and 96 u
lower than the nominal mass of the parent ion, with
successive losses of 90. 92. 94, and 96 u. appeared in
the linked scan spectrum. These "92" losses were not
observed when the matrix was thioglycerol. but were
observed with glycerol alone. Therefore, it was deter-
mined that these ions originated from the glycerol
matrix. The artifact peaks clustered 92 u below the
parent ion are denoted by T in the B/E mass spectra
that are presented.

Results

The fast atom bombardment mass spectrum of digoxin — ex-
act mass measurements and fragment ion assignments: The
fragmentation pathways operative in the FAB analysis
of digoxin are the subject of this article and thus
discussion will begin by presenting the data obtained
for digoxin. Initial assignments for the fragment ions
are proposed to provide a basis for discussing the
possible mechanism(s) of fragmentation. This creates
a difficult situation because the identities of the frag.

I Am Soc Mass Spectrom 1990, I, 435-472

. H e . H 0
(H05 3 )H [NOS3 052 ()51 ]H

     

ml: 131 ml: 391

  

ml: 651
(1105‘ cs, onw‘

ml! 521
“05‘ OH]! °

mix 1191
[AOHIH ’

D=Digoxin F=Acetyldigitoxin

R' s OH ‘1 u H

II:- II Ila II

R3 I 0H '3 I OAC
E=Digitoxin G=Gitoxin

“I I H ‘l a R

R1 e H I:- OH

“3 s OH I}: 0“

Figure 1. Structures of the cardiac glycosides. The m/z values
of the fragment ions labeled on the figure are for digoxin.

ments are necessary for determining the fragmenta-
tion mechanism(s) and likewise the fragmentation
mechanism is required to know the exact structure of
the fragments that we wish to describe in the ionic
assignments. Whereas mechanistic possibilities are be
ing evaluated throughout this article, the nomencla-
ture used to identify different fragments is the
Light/Kassel/Allison scheme, which has previously
been described [8]. The Light/Kassel/Allison scheme
is designed as a precise shorthand for discussing both
ionic and neutral variants of the types of molecules
addressed in this article. This nomenclature allows for
more mechanistic detail than the more common
Domon/Costello nomenclature [9]. which is mecha-
nistically neutral. Where appropriate, reference is
made to the Domon/Costello scheme to provide addi-
tional clarity for those readers more familiar with this
latter system.

The structure of digoxin is shown in Figure 1 (case
D). Also shown in Figure 1 are the proposed fragment
ions for digoxin. The FAB mass spectrum of digoxin is
shown in Figure 2a. and the linked scan spectrum of
the [M]H* ion of digoxin (where M is the intact
molecule), m/z 781, is shown in Figure 2b. The FAB
mass spectrum contains peaks representative of the
protonated digoxin molecule [M]H+ at m/z 781, frag-
ment ions derived from the analyte digoxin, and the
glycerol matrix adduct ions, [(glycerol),,]H+ (denoted
by ‘). All of the fragment ions appear at odd m/z
values and are thus even electron ions. There are a
few types of ions that suggest patterns of fragmenta-
tion. The first of these is represented by the ion

180

HIGHLY FUNCTIONALIZED MOLECULES IN FAB 457

current at m/z 651. The terminal glycosidic bond is
cleaved such that the glycosidic oxygen remains on
the fragment containing the aglycone. Although the
actual fragmentation follows protonation, to demon-
strate how the nomenclature scheme is used, consider
the neutral digoxin molecule, designated as
[AOSIOSZOS3OHL If the C—0 bond of the terminal
glycosidic linkage were cleaved as shown in Figure 1,
two radical fragments would be formed with masses
of 131 and 649 u. This bond cleavage appears to be
accompanied by an H-shift toward the glycosidic oxy-
gen. producing two neutral species with masses of
130 and 650 u that are designated by the Light/Kas-
sel/Allison shorthand notation as [H055 HI and
[AOS,OSZOH]. respectively. Both of these neutral
species are observed in protonated form in the FAB
mass spectrum at m/z values of 131 and 651, respec-
tively (see Figure 2a). These ions are labeled as
[H055 H[H+ and [A0510520H|H* and are con-
firmed by exact mass measurements as shown in
Table 1. In the designation scheme of Domon and
Costello [9]. these ions correspond to the BI and Y2
fragments, respectively.

The tentative structural assignments of the frag-
ment ions observed in the FAB mass spectrum of
digoxin using the Light/Kassel/Allison scheme are
presented in Table 1. in addition to the designations
based on the Damon/Costello nomenclature. Peak
matching results. and the relative errors between the
proposed structure exact mass calculations and the
experimentally determined exact mass, are also given
in Table 1. All uncertainties are within 2 mmu and
thus adequately support the elemental compositions
of these fragment ions as listed. There is more than
one way that the fragment ions can be labeled using
the Light/Kassel/Allison designation scheme. For ex-
ample. the ion labeled (HOS; "]H* could also be
labeled [H053+ ]; both designations have the same
chemical formula but carry different implications in
terms of the process by which the ion is actually
formed (as will be discussed below). The structural
assignments given in Table 1 for the fragment ions
observed provide a starting point for discussing frag-
mentation pathways and may, of course, be altered as
mechanistic information is obtained. Similar to the
terminal glycosidic bond cleavage that leads to the ion
at m /z 651, fragmentation occurs about the other two
glycosidic bonds of digoxin followed by H-shifts to
produce Y ions at m/z 521, [AOS,OH]H*, and m/z
391, [AOH)H*. The ion at m/z 391 may also be
[HOS3OSZOSf Hll-I *2 These assignments for the frag-
ment ions based on the low resolution FAB mass
spectral data are confirmed by peak matching results.
as presented in Table 1. The two assignments for the
ion current at m/z 391 will be discussed below.

Another type of ion formed in this experiment is
represented by the ion at m/z 633. This ion could be
formed by cleavage of the 2 bond instead of the y bond
in the terminal glycosidic bond, as designated in (I) to

181

458 LIGHT AND ALLISON 1 Am Soc Mass Spectrom 1990. 1. 455-472

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 
 

 

 

100‘ 4 ‘3‘ l
‘11: ‘
80“
ui
‘z’
3 +
g 60" 651 [MW
in ' 521 781
( 243
g 503
E ‘0. 485 533
i i
I: i 33255 A I a A ‘5 L 1L 1
20" 373 7‘ x5.0
. 391
I .. - A 'JJ Y 1 41 r v 1 1‘ .4
200 400 600 800
(a) mlz
100‘ 781
[mm+
243
80‘
w .
o i
z 1
3
z 60
D ‘ 391 521
tn 113
( 373
. 97 131 261 533
l“ 40‘ 355
a 1 [ j ...[
_.i
.. , l l l I l l
g ‘ f I ' ' ‘ ' rﬁ ﬂ ' ' I ' ‘ V ‘ I ' ‘ ' ' I ' '
20‘
. 1 x10.0
0
100 200 300 400 500 600 700 800
(b) mlz

Figure 2. (a) The FAB mass spectrum of digoxin in glycerol. The glycerol adduct ions are labeled
as ‘. (b) The linked scan CAD spectrum of the digoxin parent ion [MIH’ at mlz 781.

produce a Z, ion [9]. In this case the glycosidic oxy- 147 and 633 u. An H-shift toward the glycosidic oxygen
gen remains with the nonreducing terminal sugar. This would produce two neutral species of masses 148 and

cleavage would produce two fragments with masses of 632 u. respectively. Only one of these species is ob-
served, in a protonated form. at m/z 633, which may

suggest that this ion is formed by loss of H20 from
the ion at m/z 651 instead of involving an uncommon
cleavage about the z glycosidic bond [10]. (The two
C—O bonds in glycosidic linkages are chemically dis-
tinct; they tend to fragment more readily at the nonre-
ducing end of the bond for typical glycosides, such as
is seen for the cardiac glycosides upon acid catalyzed

 

J Am Soc Mass Spectrom 1990, 1. 455-472

182

HIGHLY FUNCTIONALIZED MOLECULES IN FAB

 

 

 

Table 1. Digoxin fragment ions: possible assignments and peak matching results
m/z Assignment LKA' designation DC” designation Ammu‘
781 c..H..o;, iAos.os,os,0H1H°; [MIH' (M+Hl‘ -o.2
651 C35H550 ;, lAOS.0$,0H|H° Y2 + 1.0
633 C35H530 io lAOS.OS,OH-H,OIH' 2, —0.7
521 CnH.50.° IAO$.OHIH‘ v. - 0.2
503 CnHuO; lAOS.OH-H,OIH‘ Z. +1.6
391 C,,H,,O; (811‘ IAOHIH' v, -1.6
C..H,.O; (19) lHOS,0$,OS;"IH' B3 0.0
373 Cal-1330“ 1591 lAOH-H201H' Zo — 1.0
C..H,,O; (41) lHOS;O$,OS;"-H,OIH° 8,-H20 —0.6
355 Cut-13.0; (821 lAOH-2H201H' Zo-H20 — 1.5
C..H,,O; (18) IHOS,OS,OS ;“-2H,OIH ‘ 83- 2H,0 e
337 CanO; (76) lAOl-l- 3H,OIH‘ Zo— 2H,O - 1.0
C..H,,O; (24) IHO$,OS,OS ;“-3H,OIH‘ 83— 3H,O + 0.9
243 C.,H.,O; lHOS,O$;"—H,O)H' 8,-H,0 + 2.2
131 c.H.,o; lHOS;“lH‘ a. -0.6
113 c.H,o; lHOS;“-H,OIH' 8. — H10 -o.2
97 C.H,O‘ lS —H,OIH ' 7 .+ 0.3

 

4S9

' Light/KassellAllison scheme presented in ref 5.
' Demon/Costello scheme presented in ref 9.
‘ Ammu - measured mass-calculated mass.

‘ Percent contributions of each component based on normalized peak heights averaged for three high resolution
scans. These indicate that major contributions to ion Currents in the mlz 33 7- 391 region are aglycone-containing

fragments.

° ton intensity is too low for accurate peak match; see text.

hydrolysis. See, for example, ref 10.) This presents a
question as to how to designate the ion at m/z 633 un-
til the mechanism is determined. It could be labeled
as [AOSIOSz‘h' [H+ or [AOS,OSIOH—HZO]H*. The
first designation suggests its formation via a one step
process; by cleavage of the 2 bond, whereas the sec-
ond notation carries no implication concerning the site
of H20 loss, but does connote a two-step process. Ob-
viously, information on the fragmentation pathway is
required to determine which of these two designations
is most appropriate. However, most fragment ions ob-
served in the spectra of glycosides appear to be formed
by cleavage of the y glycosidic bond with retention of
the glycosidic oxygen on the reducing portion of the
molecule (containing the aglycone), accompanied by an
H-shift toward this glycosidic oxygen [9]. If this cleav-
age (to produce Y ions) is most prevalent for digoxin,
then the ion at m/z 633 is most probably formed by
the loss of H20 from either the aglycone or one of the
sugars in the ion observed at m/z 651.

The final type of ions is in the 300 u range that
begins with m/z 391, with a series of ions 18 u lower,
at m /z 373. 355, and 337, which do not simply corre-
late with primary fragments of the molecule (as in the
discussion above), and must be the result of multiple
dehydration steps. The ion at m/z 391 has two possi-
ble identities due to the nearly symmetric nature of
this molecule (in terms of mass). Cleavage of the
AO—S1 glycosidic bond in the neutral digoxin molecule
followed by an H-shift toward the aglycone would

produce two neutral species, each with a mass of 390,
both of which could appear in the mass spectrum in
the protonated form at m/z 391. Peak matching re-
sults are vital for this group of ions that can each have
two possible origins. Without knowing the exact com-
position of this mass spectral peak with a nominal
mass of 391, it would be impossible to discuss the
fragmentation mechanism(s) involved. The exact
masses of the aglycone structure and sugar portion
differ by approximately 50 mmu, and thus peak
matching and /or high resolution scanning can be used
to determine the contribution to each peak from the
aglycone portion and the contribution from the sugar
portion of the molecule. Peak matching results are
presented in Table 1 and the deviations of experimen-
tally obtained exact masses from calculated exact mass
determinations are all within 2 mmu. These results
indicate that the ions in the 300 u range each have
two components. one from each end of the molecule.
The contribution to the peak at m/z 355 from the
sugar portion could be detected, but that signal is too
weak to allow an accurate mass measurement. The
relative contributions, from the aglycone portion and
the sugar portion of the molecule, within these dou-
blet peaks, do vary throughout the series of 300 u-
range ions. High resolution scanning is used to deter-
mine the relative contributions of each component to
the ion current of each nominal mass. The results of
three experiments of high resolution scanning for the
ions at m/z 391, m/z 373, m/z 355, and m/z 337 were

460 LIGHT AND ALLISON

183

I Am Soc Mass Spectrom 1990. I, «155-472

Table 2. Relative intensity data from the CAD mass spectra of digoxin fragment ions

 

Parent ions (nominal mass)‘

 

 

 

781 651 633 521 503 391 373 355 337 243 131 113

Daughter ions

(nominal mass)

763 3.9

651 28.9

633 1.9 2.8

615 0.6 4.3

521 2 6.8 23.7

503 1.8 1.4 12.2 5.3

485 0.8 1.6 2.8 2.4 7.8

467 0.4 0.7 2.6

391 2.2 3.3 5.1 4.4 31.2

373 1.9 2.3 2.9 5.4 6.9 8.0

355 0.9 1.3 2.0 3.5 6.0 5.9 19.3

337 0.7 0.5 1.0 1 7 0.8 5.8 23.6

319 0.6 2.2

279 0.6

261 1.0 0.6 0.4 0.5

243 5.4 3 1 9.4 0.6 1.3 0.6 1.0

225 1.3 0.2 0.4 0.5 1.0

149 0.2 0.9

131 1.5 5 2.0 1.8 1.2 0.3 0.4 0.7

113 1.9 1.4 2.4 1.2 2.3 0.3 0.4 0.3 4.9
97 1.6 0.8 1 2 0.3 10.5 0.3 9.3 0.1
95 0.7 0.2 2.0
85 0.4
83 0.1 0.2
69 0.6 0.5
2 59 28 69 26 60 18 39 26 3 12 6 3

 

' Parent ions are the [NIH ' and the fragment ions observed in the FAB mass spectrum of digoxin.

averaged and the percent contribution of each compo-
nent is presented in Table 1. These ratios vary some-
what with experimental conditions, such as ion source
pressure, which may suggest that some fragmentation
occurs as CAD within the ion source. In most cases,
the contribution to each doublet peak is greater from
the aglycone portion of the molecule than from the
sugar portion. The peak at m/z 373 contains the
largest contribution to the ion current from the sugar
portion of the molecule than any of the other 300
u-range ions.

One case where low resolution mass spectrometry
and peak matching capabilities are insufficient to sug-
gest, unambiguously, a relationship between the m/z
value of a fragment ion and some substructural feature
of the original molecule is the ion at m/z 243. Based
on the nominal mass assignments and peak matching,
it is determined that this ion is from the sugar por-
tion of the molecule but could have one of two assign-
ments: [H053052 'H —l‘le]I-l+ or [H ‘52051’H]H* .
This ion also appears as a daughter ion from CAD
analyses of all of the fragment ions that contain at
least two sugars. In an attempt to differentiate between
these two assignments, the FAB spectrum of a similar
compound was obtained. Acetyldigitoxin (structure F,
Figure 1) was chosen for study because of its similar-

ity to digoxin with two exceptions: the aglycone con-
tains one less OH group (like digitoxin) and, more im-
portant, the terminal sugar contains an acetyl group.
The ion observed at m/z 243 in the FAB mass spec-
trum of digoxin does shift to m/z 285 for acetyldig-
itoxin and suggests that this ion contains the termi-
nal sugar where the acetyl group is located. There-
fore, we will assume that the assignment for m/z 243,
[H053052‘H—H201H1, is correct and is included in
Table 1.

Linked scan mass spectral data on ions formed from digoxin
in the fast atom bombardment mass Spectrum. In addi-
tion to the peak matching information, the linked
scan CAD mass spectra of the parent ion [MIH+ and
the major fragment ions of digoxin were obtained to
provide additional information that may aid in the
elucidation of fragmentation mechanisms. The results
from these CAD experiments are presented in Table
2.

One difficulty with the B/E linked scan technique
performed on a forward geometry double-focusing
mass spectrometer is that the resolution of the parent
ion selection is low [11]. In practice, the acceptance
window for the parent ion selection is approximately
5 u wide. Therefore, the CAD spectra of the doublet

] Am Soc Mass Spectrom 1990, I, 455-472

184

HIGHLY FUNCTIONALIZED MOLECULES IN FAB 461

 

 

 

 

 

 

 

 

 

100: 521 651
80‘
m .
o I
i 1
o 60‘ 243 39‘
g ‘ 373
m I
< . 131 485
m ‘0. 113 355 503
> , 97
I: l 337 I
< l
_. f, _ ‘le l, I
g 20‘ I ‘ ' ' ‘ 1
. Tx10.0
‘ T
U v v - 1'1? ' l 1 JJ 1 A 1 L
, ...4. f...j....,.1.....

100 200 300

400 500 600

W2

mlz

Figure 3. The linked scan CAD spectrum of the digoxin fragment ion at mlz 651. The peaks
labeled T are glycerol interferents that are present due to the glycerol cluster ion 6 u from the

parent ion at mlz 651 (see text).

ions, such as those at m /z 373, will contain a mixture
of daughter ions from both of the isomass species.
Also, this wide parent ion selection window can lead
to artifact peaks from glycerol cluster ions (if glycerol
is the matrix) when the m/z value of the desired
parent ion is within approximately 5 u of a glycerol
cluster ion.

The CAD mass spectrum of the [MIH+ ion is
presented in Figure 2b and that for the fragment ion
at m/z 651 is shown in Figure 3 as an example of the
linked scan spectra obtained for the major fragment
ions of digoxin. The two spectra in Figures 2b and 3
are very similar in appearance and content even
though the two parent ions differ by one sugar. To
obtain these linked scan data, the helium collision gas
pressure was set to produce a 10% attenuation of the
[Mll-l+ ion and was maintained at this same pressure
to obtain the CAD linked scan spectra of the major
fragment ions of digoxin. The results of these CAD
experiments, the daughter ion m/z values and rela-
tive abundances, for all of the major fragment ions of
digoxin are listed in Table 2. These relative abun-
dances are from the normalization of the fragment ion
currents to that of the parent ion. Therefore, the
absence or presence of the daughter ions and their
relative abundances can be compared within each B/E
linked scan, but the abundances between linked scans
for different parent ions should not be compared
directly. The last row in Table 2 contains the sum of
the relative abundances of all fragment ions produced
from each selected parent ion (the sum of the columns)
and gives some measure of the relative extent of
fragmentation that occurs upon CAD of each species.

One ion not listed in this table is that at m/z 175. This
fragment ion is not observed as a daughter ion in the
majority of the linked scan spectra collected. This ion
is believed to originate from a ring fragmentation of
the middle sugar, referred to as an A 3 cleavage, using
the Damon/Costello nomenclature. According to
Damon and Costello [9], this ring cleavage is not
common in the FAB analysis of saccharide-containing
compounds in the positive ion mode. This ion at m/z
175 is the only ring cleavage product observed in
these studies.

Discussion

Before evaluating the data in search of clues to the
fragmentation mechanisms that may be operative for
protonated digoxin, a discussion of some of the likely
mechanistic possibilities will be presented. When con-
sidering mechanisms, one can certainly benefit from
the approach used by McLafferty [1] in the context of
fragment ions formed by El. One should first consider
the description of the ionized molecule (where are the
charge/radical sites?), then propose possible fragmen-
tations based on established mechanisms and chem-
ical intuition. In the case of a protonated molecule,
an even electron ion, the literature on chemical ion-
ization (CI) mass spectrometry [12] certainly provides
a useful framework. Thus, we first ask the question
concerning the nature of the protonated molecule pro-
duced by FAB. Where is it protonated [13]? (We note
that much work has been done to investigate the site
of protonation. in molecules that contain more than one
basic site. See, for example, ref 13.) It is difficult to an-

462 LIGHT AND ALLISON

H.
e“ b "a k

0
2
o\
cu3 5 H0
1 ‘Il

H H'

OH d cu3 ;’ c
o’—‘V\
I
o

185

J Am Soc Mass Spectrom 1990, 1, 455-472

   
 
 

H0

51'

Figure 4. Extended structure of digoxin based on crystallographic data with possible sites of

protonation labeled a through I.

swer this question for many reasons. We do not know
how the molecule is protonated. lt presumably occurs
by gas phase proton transfer [3] from protonated glyc-
erol and/or some fragment ion derived from glycerol
such as m/z 45. It is probable that protonation of any
of the heteroatoms in the molecule could occur be-
cause the proton transfer is probably from an oxygen-
containing Lewis base to one of the oxygen atoms in
the cardiac glycoside. There are many different types
of sites on the digoxin molecule with different proton
affinities. An extended structure of digoxin, based on
the crystallographic analysis of the molecule [14], is
presented in Figure 4 with possible protonation sites
indicated. We have estimated the proton affinities of
these sites, labeled a through I in Figure 4, and these
values are presented in Figure 5. The proton affini-
ties of the different sites on digoxin are approximated
based on smaller compounds, resembling these sites,
whose proton affinities are known [15]. In those cases
where there are multiple interactions, estimates for the
increase in proton affinity (PA) due to secondary in-
teractions have been made and are discussed in Ap-
pendix 1. Based on the estimates in Figure 5, it ap-
pears that many of the possible sites of protonation
of the digoxin molecule lie in the PA range of 190-200
kcallmol. Also shown in Figure 5 are possible protonat-
ing species derived from glycerol. Candidates as pro-
tonating reagent ions are selected based on the FAB
mass spectrum of glycerol reported by Sunner et al.
[3], which includes the protonated glycerol molecule
and lower mass glycerol fragments. Note that all of the
glycerol fragments have proton affinities lower than
that of glycerol, 209 kcallmol [3]. If the PA values in
Figure 5 are correct, proton transfer from protonated
glycerol to most of the labeled sites of digoxin would
be endothermic, however, proton transfer from the

fragment ions of glycerol is possible. The most basic
sites in the molecule appear to be the glycosidic link-
ages, which are further enhanced by additional inter-
actions. It has been shown that protonated molecules
containing two functional groups can show intramolec-
ular hydrogen bonding, even when the two groups
are separated by many methylene groups. Frequently,
10-20 kcallmol can be introduced by such secondary
interactions [16]. We propose that the site of highest

220-

Q
m— G" M“

O
°3“5° ’ 4%
_L_d_
_L

 

 

c3“7°2

’°°‘

cznso’
m

PA (ical/mot)
s
l
[-

170- ‘

4

$9.9.

 

 

Possible protonation
sites in digoxin

Ions formed by
FAB of glycerol

Figure 5. Estimates of the proton affinities of the basic sites
labeled on the digoxin structure in the figure and estimates of
the proton affinities of possible protonating agents from glycerol
and glycerol fragments.

] Am Soc Mass Spectrom 1990, I, 455—472

PA is the terminal sugar unit. The basic site formed by
the two —OH groups plus the —OR group, all three
being on the same side of the ring, resembles the inter-
actions described by Winkler and McLafferty [17] for
a protonated cyclohexanetriol. ln digoxin (solid), the
ring oxygen of sugar 2 is in close proximity to the OH
group in the 3 position on sugar 1, making site It a
possible site for a multiple interaction. Our conclusion
is that any part of the molecule may be protonated in
the FAB experiment with a glycerol matrix, although
not necessarily by the most abundant of the possible
candidates from glycerol, [glycerolll-l+ .

How, then, should we consider the protonated
molecule in the context of the fragmentation that will
follow protonation? In this regard, two extremes
have been discussed in the literature. ln the simplest
approach, the molecule is protonated and fragmenta-
tion follows directly, occurring at that protonation site
[18]. For example, it has been proposed that proto-
nated bradykinin, formed by field desorption, frag-
ments at the site of protonation and the proton does
not migrate freely about the molecule [19]. The other
extreme is a dynamic model that suggests that the
initial site of protonation would be, in this case, irrele-
vant—the proton rapidly moves from heteroatom to
heteroatom, with the possibility of fragmentation oc-
curring at every site while the proton resides at that
site [20]. We believe that the latter is more likely in
this case, and will assume that our starting point will
be a protonated molecule that has a mobile proton.

What possible fragmentation mechanisms may be
operative for this even electron ion? Mechanisms that
have been proposed include inductive cleavage pro—
cesses, fragmentation involving multiple bond cleav-
ages (1,2-elimination reactions and ring cleavage pro-
cesses), and remote site fragmentations.

Inductive Cleavage

Domon and Costello [9] have proposed a scheme for
naming fragment ions from carbohydrates, and the
process that they have proposed leading to what they
call 8, ions is shown in reaction 1 for two digitoxose

sugars.

 

186

HIGHLY FUNCTIONALIZED MOLECULES IN FAB 463

Inductive effects lead to cleavage of the glycosidic
bond at the site of protonation in (11), forming the
charge migration product (111). It has certainly been
documented that inductive processes involving oxy-
gen do readily occur [1], and such a mechanism is
reasonable for this even-electron ion (11).

cu3 cu3

lo,
0

H0 0 012

on H UN on It

(2)

 

(IV)

Multiple Bond Cleavage Processes

We will not discuss ring cleavage reactions because
only one has, possibly, been observed in this study
(at m /z 175). However, it is apparent that H—shifts do
occur, and accompany glycosidic C —O cleavages. For
example, reaction 2 has been proposed by Domon
and Costello [9] to yield Y, ions, of the type (IV). The
mechanism suggests a 1,2-elimination to leave a double
bond in the neutral sugar fragment, although H shifts
from other sites within the molecule cannot be ruled
out [21]. (H-shifts in peptides as 1,2-eliminations have
been proposed in refs 19 and 21a. H-shifts in peptides
as 1,3-eliminations have been proposed in ref 21b. H-
shifts via 7 and 8—membered rings have been discussed
in ref 21c.) Thus, the fragment ions (III) and (IV) are
proposed to come from a common intermediate, (11),
via two different mechanisms. It is also possible that
the fragment ions (In) and (IV) are formed through an-
other common intermediate and a single mechanistic
step. This is shown in reaction 3. The starting point is
an ion of the type (II), protonated on a glycosidic oxy-
gen. The charge site stimulates a 1,2-elimination reac-
tion to form the proton-bound adduct shown as (V),
which can then dissociate to form either (VI) and/or
(V11), depending on which fragment retains the pro-
ton. (This mechanism is similar to that suggested by
Stevenson's rule [1] in E1, in which separating frag-
ments compete for the charge; here the competition is
for the proton. This protic analogy to Stevenson’s rule
has not been established to date, although it has been
alluded to by Bowen et al [22].) Thus, the ions of that
type labeled (III) and (VI) are the same, except that
(V1) is formed via an H-shift. It is difficult to decide
on a designation for such ions. In our nomenclature
scheme, the two different mechanisms would suggest
that an ion such as that shown as (V I) could be labeled
as 11053052 * or [H053052'”]H*, and we have ten-

 

464 LIGHT AND ALLISON
CH3 \
\o .
«053-0 g-sr O-A
“053-0 (3)
/V-H°--- ads.— O-A
O
or 0'S|’ O-A
"053-0 "2
(VIII
0H
(VI)

tatively chosen the latter. It is also interesting to note
that while Domon and Costello [9] have proposed reac-
tion 2, they write the product not as (TV) but as (VIII),
in which the proton in this fragment ion has moved
from its initial site on the terminal hydroxy group in
(IV) to another glycosidic linkage; presumably this was
done to suggest that the proton is mobile following
protonation.

cu3
\
\o O
no on
H
OH H
(vm)

Remote Site Fragmentation

Jensen et al. [23], Adams [24], and Wysocki et al. [25]
discussed the remote site mechanism in ions for which
the charge and /or radical site are apparently far from
the site of fragmentation. Such mechanisms certainly
seem reasonable when a C—C bond is broken in a
long alkyl chain of an ion containing a single func-
tional group, and difficult to prove for multifunctional
molecules. This mechanism would suggest that proto-
nation can occur at any part of the molecule, and
fragmentation need not be local to the site of protona-
tion. Vine et al. [26] discussed the possibility of re-
mote site fragmentation occurring in the ammonia CI
mass spectra of cardiac glycosides. They suggest that
the ammonium ion may complex with the aglycone
portion of the molecule and 1,2-eliminations about the
glycosidic linkage may occur far from this site, as

187

] Am Soc Mass Spectrom 1990, 1, 455-472

C":

H053-0-52-0 o-a-rm;

(IX)
H
l (4)

. H-O-A'NH‘.

0H

NOS 3 -o-s 2 —o

(X)

 

shown in reaction 4. It is interesting to note that in a
recent article on the mass spectrometry of peptides by
johnson et al. [21a], it was proposed that much of the
fragmentation observed for peptides may occur via
remote site processes.

Adams and Cross [27] have discussed the analogy
of remote site fragmentations to thermolytic pro-
cesses. Energy is imparted into the molecular ion,
which fragments as it would if energy were added to
the corresponding neutral molecule. In this context,
we have reported the K*IDS mass spectrometric anal-
ysis of cardiac glycosides [8]. K+IDS, K‘ ionization of
desorbed species, is a technique that combines rapid
thermal processes, thermal degradation, and vapor-
ization, with gas phase K+ attachment. We note that
when digoxin is rapidly heated, 1,2-eliminations ap-
pear to readily occur about the glycosidic bonds,
yielding K+ adducts that are very similar to those
seen here in protonated form. In K+IDS, decomposi-
tion occurs before I<+ attachment (K+ attachment
does not induce much fragmentation), while in FAB
protonation presumably precedes decomposition and
often induces fragmentation. Even with these differ-
ences in fragmentation processes, preceding ioniza—
tion in K‘IDS and following ionization in FAB, most
of the same fragmentation processes are observed in
both mass spectra. The correlation between the FAB
and 10105 spectra could be fortuitous, but may sup-
port remote site processes that are not initiated by the
charge site but rather by energy deposition in general.

Mechanistic discussions of FAB presented here are
based on the assumption that protonation of the intact
molecule precedes fragmentation. However, it is pos-
sible that direct fragment ion formation from the ma-
trix upon bombardment can occur. Many of the frag-
ment ions of digoxin contain the aglycone. However,
there are some ions in the low mass range that are
from the sugar portion of the molecule. Two such ions
are at m/z 113 and 131 from the terminal sugar (Ta-
ble 1) and are the most abundant fragment ions in the
FAB mass spectrum (Figure 2a) of the digoxin sam-
ple. Additional experiments were performed to inves-
tigate the origin of these two abundant low mass ions.
The FAB mass spectra of digoxin in glycerol were col-
lected and monitored for a period of thirty minutes.

I Am Soc Mass Spectrom 1990, 1, 455-472

During this time the ions at m/z 113 and m/z 131 re-
mained dominant, even when only a trace of glycerol
remained and the [M]H* ion of digoxin was no longer
observed. These two low mass ions seem to be formed,
at least to some extent, in a different manner than the
[M]H* ion. It is possible, based on the observations
of this extended FAB experiment, that these two low
mass sugar ions may be formed directly from the liq-
uid target upon bombardment and are not only the
result of fragmentation of the [MjH+ ion. If this is
the case, then the decomposition of the [MjH+ ion
of digoxin produces predominantly daughter ions that
contain the aglycone. The ions from the sugar portion
that do not contain the aglycone may be formed by
other ionization/fragmentation processes.

The mechanisms discussed here must be evaluated
with experimental data in order to determine the most
probable fragmentation pathways. We now turn to
the peak matching and linked scan data to determine
whether these results support any of the mechanisms
discussed, or suggest others.

Discussion of Collisionally Activated Dissociation
Data

Analysis of the CAD data should provide some in-
sights into the fragment ion structures and, from
these, insights into the mechanisms through which
they are formed. The mechanisms proposed above
that are based on a localized site of protonation in the
molecule, leading to fragmentation at the site of proto-
nation, would be substantiated by a CAD mass spec-
trum of a fragment ion with very few daughter ions
produced. If the site of protonation is also the site of
fragmentation and the charge site does not migrate,
then the primary fragment [with a structure such as
(IV)] might be expected to only undergo H20 losses
with no extensive fragmentation upon CAD. Other
types of fragmentation mechanisms, where the charge
is mobile throughout the molecule or is localized fol-
lowed by remote fragmentation, may produce more ex-
tensive fragmentation of the parent species upon CAD.
Substructure-specific daughter ions may be formed,
and this concept will be used here. Suppose an ion
is observed 18 it below another fragment ion due to a
water loss, for example, the ion at m/z 633, 18 it below
the major fragment ion at rn/z 651. From where was
the water lost? The ion at m/z 651, [AOS1OS-,.OH]H+ ,
could eliminate water from the terminal sugar (52), the
interior sugar (5,), or from the aglycone (A). If there
is a daughter ion present in the CAD mass spectrum
of m/z 633 that represents the loss of an intact sugar,
a neutral loss of 130, [HOSz’”], then the water elim-
ination cannot be from the terminal sugar. However,
if a neutral loss of 112, [HOSz‘H-HzO], is observed
in the absence of a 130 loss, this would suggest that
the terminal sugar in m /z 651 was the site of H20 loss.
This neutral loss prediction is shown in Figure 6 and is

188

HIGHLY FUNCTIONALIZED MOLECULES IN FAB 465

mil ‘5!

Iaos,os,omlt‘
.1. or
II; 6).!
|A'“'° os.os,ou III ' "
“05.05; III‘
Iaosf'” os,ou|II'
' can cw

[A'“'°0$,Oll|ll' . |"‘s,oIII
IAos,oum' . I'"S.‘"l
Iaos;"'°ortm' 0 (“spin

ah $0.1 antral less I). III 511 aeerral loss In

Figure 6. Schematic diagram of the H20 losses from all: 651
that can occur to produce the ion at m/z 633. Neutral losses
observed from the CAD of m/z 633 can help determine the
structure of the ions at m/z 633.

explained in more detail in the discussion of the CAD
results for the ion at m/z 633. In a similar fashion, one
could search for ions indicative of the intact aglycone,
which would indicate that the H20 loss is not from
that portion of the selected parent ion. These types
of observations and conclusions are presented here for
the CAD data obtained for the fragment ions formed
in the FAB analysis of digoxin.

Collisionally activated dissociation of m/z 781. The dis-
cussion of the data begins with the [M]H* ion of
digoxin at m /z 781. The daughter ions of the proto-
nated molecule are very similar to those observed in
the FAB mass spectrum of digoxin except for some rel-
ative abundance variations of the fragment ions. The
same fragment ions appear in both the FAB mass spec-
trum of digoxin (Figure 2a) and the CAD mass spec-
trum of the protonated molecule (Figure 2b). This ob-
servation is consistent with the desorption ionization
mechanism that is assumed to be operative here, that
is, the fragment ions observed in the FAB spectrum of
this neutral analyte are from unimolecular decompo-
sition of the protonated, intact, molecule as opposed
to being fragments that are emerging directly from the
target upon FAB. There are some differences between
these two spectra, notably that the FAB mass spectrum
shows a more abundant [M-H201H * ion than the BIE
mass spectrum. Also, the dominant ions from the FAB
analysis of digoxin are at m /z 113 and m /z 131 and are
less abundant in the linked scan mass spectrum of the
[M]H+ ion. An explanation is suggested from the data
in Figure 5. There exists in the selvedge region of the
FAB experiment a collection of reagent ions available
for proton transfer in the m/z range 19-93. Of these,
the most abundant is protonated glycerol at m/z 93,
which may only be able to protonate the most basic
site (I), on the terminal sugar, leading to prompt for-
mation of m /z 131 and m /z 113, and a large abundance

466 LIGHT AND ALLISON

of these ions. The lower mass reagent ions protonate
other parts of the digoxin molecule, and many of these
[MlH+ ions are sufficiently long lived to be observed
in the FAB mass spectrum at m /z 781. These latter ions
are the ones chosen for CAD analysis.

Collisionally activated dissociation of m/z 651. A major
fragment ion of digoxin is observed at m/z 651 in
the FAB mass spectrum. It is also the most abundant
daughter ion of [MIH+ , as shown in Figure 2b. From
the peak matching results it is proposed that the struc-
ture of this ion corresponds to a protonated cardiac
glycoside containing only two sugars (see Table 1). The
daughter spectrum of m/z 651, shown in Figure 3,
closely resembles that of the [M]H* species of digoxin,
shown in Figure 2b. If the mechanism of fragmenta-
tion to produce the ion at m/z 651 were charge initi-
ated as shown in reaction 2, then the CAD mass spec-
trum of m/z 651 should be very simple with daugh-
ter ions at m/z [651-18]. Instead, the CAD mass spec-
trum closely resembles that of the [M]H+ ion and is
what would be expected for a CAD mass Spectrum of
a protonated cardiac glycoside containing two sugars,
[A0510520H]H* . If the mechanism of reaction 2 is
operative, the proton must be mobile and not local-
ized in the product. This requirement of proton mo-
bility to explain the extensive subsequent fragmenta-
tion of primary fragments undergoing CAD was pre-
sumably recognized by Domon and Costello [9] when
they chose the specific designation shown as structure
(VIII), with the proton moving within the ion following
fragmentation. The neutral loss of 130 from the parent
ion at m/z 651 yielding m/z 521 further supports the
structural assignment of m/z 651, as this requires that
the terminal sugar must be intact, without any H20
losses. One fragment ion of m/z 651, at m/z 467, is
not observed in the linked scan mass spectrum of the
[MjH+ ion. A possible explanation for the presence of
this daughter ion in the CAD mass Spectrum of m/z
651 is as follows. Both parent species at m/z 781 and
m/z 651 can form the daughter ion at m/z 521, which is
[AOSIOH]H+. In the CAD of m/z 781, the fragment
at m/z 521 loses one and two H205 to yield ions at
m/z 503 and m/z 485. In the B/E spectrum of m/z 651,
the fragment at m/z 521 is formed to a greater extent,
and therefore may allow for the observation of up to
three H20 losses, at m /z 503, m/z 485, and m/z 467.
The sugar component of m/z 521 can only lose two
of these water molecules as there are only two —OH
groups on the sugar, and likewise for the aglycone
portion, which has only two —OH groups. Therefore,
these multiple H20 eliminations from m/z 521 must
occur throughout the [A051 OH]H" ion and not from
one isolated part. The similarities of the CAD results
for m/z 651 and m/z 781 would be consistent with a
remote site mechanism in which the site of protona-
tion is either the aglycone or is mobile and does not
directly induce the fragmentations observed.

189

] Am Soc Mass Spectrom 1990, I, 455-472

Collisionally activated dissociation of m/z 633. Another
fragment ion of interest is observed at m/z 633 and
could have several origins. First, this ion could be a
primary fragment following protonation produced by
cleavage of the z glycosidic C—O bond on the termi-
nal sugar to produce what Domon and Costello [9]
refer to as the Z2 ion. This ion at m/z 633 could also
be a secondary fragment, as suggested by the abun-
dance ratio of m/z 651 to m/z 633 in the FAB mass
spectrum and in the CAD daughter ion spectra of m/z
651. If m/z 633 is produced by a loss of H20 from m/z
651, the H20 could be from either of the two sugars
retained in the ion, or from the aglycone. If the ion
at m/z 651, [AOSIOSOHjH‘ , loses a water molecule
from the terminal 5, sugar, then a daughter ion corre-
sponding to the neutral loss of 130, due to loss of an
intact sugar molecule, should not be observed in the
BIE mass spectrum of m/z 633. Instead, a daughter
ion corresponding to the neutral loss of 112 should be
observed, as discussed in Figure 6. Both of these neu-
tral losses are observed yielding daughter ions at m/z
503 and m/z 521. Therefore, at least some of the water
loss is from the terminal sugar, as evidenced by the
large abundance at m/z 521 (which requires an intact
aglycone and internal sugar). It is also possible that
some of the [A05] ‘“*°O$0H]H* species is present
as this would fragment upon CAD to produce daugh-
ter ions at m/z 243 (from the two sugars), m/z 391
(from the intact aglycone), and m/z 503 (from the loss
of the terminal 5; sugar), all of which are observed in
the CAD spectrum of m /z 633. The third possible iden-
tity of m/z 633 is [A'H100510520H1H+ . This species
should (and does) fragment upon CAD to produce m /z
373 (from the dehydrated aglycone) and m /z 503 (from
the y glycosidic cleavage). This species could also form
m/z 243 in the same manner that m/z 781 does, as the
two sugars from this species resemble the sugar por-
tion of the ion at m/z 781. Therefore, H20 losses are
not specific as they do not come from any one sub-
structural group. There are no daughter ions, present
or absent, that can be used to make definitive struc-
tural assignments for the species at m/z 633.

Collisionally activated dissociation of m/z 521. The frag-
ment ion observed at m/z 521 is also a prominent
daughter ion of protonated digoxin and is identified
as being structurally equivalent to a protonated cardiac
glycoside containing only one sugar (see Table 1). The
daughter ion mass spectrum of m/z 521 shows that
substantial further fragmentation occurs. Again, the
CAD results are consistent with a remote-site mech-
anism and/or a mobile proton. The parent ion at m /z
521 does not yield a daughter ion at m/z 243, which
is present in the CAD spectra of the ions at m/z 781,
m/z 651, and m/z 633. This ion at m/z 243 is indicative
of the presence of two sugars, as noted in Table 1, and
therefore is not expected as a daughter ion of m /z 521,
which contains only one sugar. This is consistent with,

] Am Soc Mass Spectrom 1990. 1, 455—472

and further supports, the assignments suggested by
the peak matching results. The CAD of m /z 521 shows
cleavage of the AO—S, bond to produce the daugh-
ter ion at m/z 391, which is also observed in the FAB
mass spectrum, and is given the assignment [AOHlH+
based on peak matching results. Also observed are the
subsequent eliminations of three H20 molecules from
the species at m /z 391. These water losses must be
coming from the two —OH groups on the aglycone and
the —OH group formed upon fragmentation (from the
glycosidic oxygen with the H-shift). Though the rela-
tive abundances are slightly different, this same series
of ions, m/z 391, 373, 355, and 337, is present in the
CAD spectra of m/z 781, 655, and 633. These 300-400
dalton-series fragment ions will be discussed in more
detail shortly.

Collisionally activated dissociation of mlz 503. The frag-
ment ion at m/z 503 is structurally similar to the ion
at m /z 633 in that these are both secondary fragments
formed by a glycosidic bond cleavage and a water loss.
The relative abundance ratio of m/z 521/503 is similar
to that for the m/z 651/633 pair. This ion fragments
upon CAD to give daughter ions at rn/z 391, m/z 373,
m/z 355, and m/z 337. The m/z 503 ion also loses one
and two waters to form daughter ions at m/z 485 and
m /z 467. The fragment ions at m /z 391 and m /z 373 are
from neutral losses of 112 and 130, as was observed
for the CAD of m/z 633. Once again the ion current
at m/z 503 is probably a mixture of species with wa-
ter loss from the aglycone portion of In /2 521 (which
would show a neutral loss of 130 upon CAD), water
loss from the sugar (which would show a neutral loss
of 112), and possibly some primary fragment from m/z
781 (cleavage at the z glycosidic bond). The abundant
daughter ion at m/z 391, which corresponds to a neu-
tral loss of 112, suggests that a large percentage of m /z
503 contains a sugar that is monodehydrated and that
this species readily fragments at the glycosidic bond to
produce the protonated aglycone species [AOHll-l+ at
m/z 391.

Collisionally activated dissociation of m/z 391, mlz 373, m/z
355, and mlz 337. There are four ions in the 300-400
u range that have two structural assignments for each
nominal mass, as shown in Table 1. Peak matching
with a resolution of 117000 or better is used to dis-
tinguish between the two peaks with the same nom-
inal mass but of different origin. The ion current ob-
served at m/z 391 has two possible origins, the agly-
cone end and the saccharide end of digoxin. Both of
these species are capable of losing up to three H20
molecules upon CAD. Thus the other three ions ob-
served in this mass range, at m/z 373, m/z 355, and
m /z 337, can be, and are, formed by H20 losses from
both species contributing to the ion current at In /2 391.
The peak matching and high resolution scanning re-

190

HIGHLY FUNCTIONALIZED MOLECULES IN FAB 467

sults show that the largest contributor to the ion cur-
rent for each of these four m /2 values is from the agly-
cone portion of the molecule. This follows with the
previous observation that the majority of the types of
fragment ions observed do contain the aglycone. The
ion at m /z 373 has the strongest contribution from the
sugar but this contribution still accounts for less than
half of the total ion current at m /z 373 (see relative con-
tributions in parentheses in Table 1). Unlike the other
300 series ions, the CAD analysis of tn /2 373 does show
some fragments indicative of the sugars. The daugh-
ter ions from sugars are observed at m/z 243, which
represents a Species containing two sugars, and at m /z
131 and m /z 113, which are derived from one sugar of
the molecule, as identified in Table 1. The CAD results
for the other three ions at m/z 391, m/z 355, and rn/z
337 support the conclusion that they primarily contain
the aglycone, since only H20 losses upon CAD are ob-
served (no fragmentation of the steroid ring structure
is observed). The relative abundances of the ions rep-
resentative of these water losses increase as the parent
ion becomes more unsaturated. For example, the loss
of H20 from m/z 391 to form m/z 373 is not as promi-
nent as the loss of H20 from the parent at m/z 373
upon CAD to form m/z 355. This is supported by the
data presented in Table 2. This suggests that once one
OH group is lost in the form of H20, subsequent de-
hydrations are more facile. It is expected, as observed,
that dehydration of the aglycone would occur before
demethylation or dehydrogenation. Consideration of
the decomposition of substituted cyclohexanes can be
used as a model to understand the energetics of H20
elimination versus CH4 elimination versus loss of H;
from the aglycone structure. The dehydration of cyclo—
hexanol to form cyclohexene requires + 10.4 kcallmole
[28]. (Thermochemical estimates are based on data
contained in ref 28.) In contrast, +18.1 kcallmole is
required to eliminate methane from methylcyclohex-
anal, and +28.4 kcallmole is required to induce the
dehydrogenation of cyclohexane. Therefore, energetic
considerations suggest that water losses are expected
before loss of CPL or H; from a cyclic system such as
the aglycone of digoxin.

The recognition that the second H20 loss occurs
more readily than the first is also observed for other
ions, such as the pairs m/z 651/63 and m/z 521/503.
The monodehydrated species in each pair (lower mass)
loses another H20 more readily (upon CAD) than the
nondehydrated species, as seen in Table 2. It is pos-
sible that, once a double bond is formed (upon H20
loss), the second loss is more facile.

Collisionally activated dissociation of m/z 243. The ion at
m/z 243 is a prominent fragment ion in both the FAB
mass spectrum of digoxin and the linked scan mass
spectrum of protonated digoxin. This ion is unique in
that it is isolated from any other peaks and is not as-
sociated with other ions 18 it higher or lower in mass.

468 LIGHT AND ALLISON

Most other major fragment ions observed in the FAB
mass spectrum have an ion 18 it below them due to
H20 loss. This ion at m /z 243 is observed as a daugh-
ter ion from m/z 781, m/z 651, m/z 633, m/z 391, and
In /2 373. These are the only species that should be able
to produce the fragment ion at m/z 243 by CAD if the
assignment for this ion is correct as given in Table 1.
The species at m/z 355 and m/z 337 have already lost
too many H205 to form the ion at m/z 243. The as-
signment proposed in Table 1 for this ion at m /z 243
is the same, minus one sugar, as that given to the ion
at m/z 373 for the sugar portion of that doublet. These
two ions may be formed by the same fragmentation
mechanism that may differ from the mechanism that
forms the ions containing the aglycone. The fact that
the ion at m/z 243 does not lose H20 suggests that the
sugars do not as readily lose H20 as does the aglycone.
Further support for the proposed assignment of m/z
243 is from the linked scan mass spectrum of this ion.
The appearance of a daughter ion at m/z 149 proves
that In /z 243 contains an intact terminal sugar. The ion
at m/z 149 is from the terminal sugar with retention of
the glycosidic oxygen, [HOS;.OH]H+ .

The most abundant daughter ion observed by the
CAD of m/z 243 is m/z 97. It is also observed in
the FAB mass spectrum of digoxin. The chemical for-
mula of this ion is determined to be QIIgO" by
peak matching (Table 1). From Table 2, it can be seen
that this daughter ion is present to a minor extent in
many of the B/E spectra collected, but is only a sig-
nificant daughter ion of the parent ions at m/z 373
and m/z 243. These two species contain a partially de-
hydrated sugar that leads to the possibility that the
structure of m/z 97 is a completely dehydrated digi-
toxose (sugar) unit in the protonated form, structure
(XI). This ion is designated as [S—H201H+ in Table 1.
It is unclear whether this ion comes from the terminal
sugar, in which case it would be labeled as [HOSfH —
H202]H*, or from an internal sugar, [*“Sz‘H
—H~,O]H+ or [* ”S, ‘” ~HZO]H* . However, structure
(X1) is supported by the results of the CAD of m/z 97.
This BIE mass spectrum of m/z 97 contained fragment
ions corresponding to the loss of CH4 and CO, which
would be expected from this proposed structure, con-
sistent with the presence of a methyl group and the
ring oxygen atom.

 

(X1)

191

I Am Soc Mass Spectrum I990. I, 455-472

 

CH) CH) C“)
mlz iri 0; o o
(—) {—9
on on /Ii' on /
on on 'o‘ii2
\L 1.1-diminution CH3 . \L helm
0' '10 o t-uze)
CM / \ Cit
3 a . 3
in]: 113 °\ / n" ,5 \ 0
o“ "2.2;...“ rad-62;“ o“ /
CH; C“; CH;
0 o O u a
k
\ _> _"'_> .
“Q— it “Q—0 “CC—f ——CI\0
04 N I
“III
J, ' rut/a I I! .9“)

C“:

O c, W
E‘\J «<1\. "”é

”1‘." rrrla 69 t

Figure 7. Possible mechanisms for the fragmentation of the
parent ion at m/z 131 upon CAD to produce the ions observed
in the linked scan (see Table 2).

Collisionally activated dissociation of m/z 131 and mlz 113.
The two low mass fragment ions at rn/z 131 and m/z
113 are very abundant in the FAB mass spectrum of
digoxin and appear to be terminal sugar fragments.
This is supported by the results of peak matching data
presented in Table 1. The m/z 131 ion fragments upon
CAD to give some interesting low mass ions. At the
low collision gas pressure used for most of this study,
the main fragment ion observed from m/z 131 is at
m/z 113 due to loss of H 20. If the collision gas
pressure is increased, to attenuate m/z 131 by more
than 10%, more fragmentation results to give all the
low mass fragments listed in Table 2 (between m/z 69
and m/z 113). Possible mechanisms for the CAD of
the primary ions at m/z 131 and m/z 113 to produce
the low mass daughter ions observed in a B/E linked
scan are outlined in Figure 7. Inductive cleavages of
rearrangement species and 1,2-eliminations are used
to explain the fragmentation of m /z 131 and m/z 113.
The daughter ion at m/z 97 is of low relative abun-
dance compared to the other daughter ions from the
CAD of m/z 131 and m/z 113. It is possible that this
fragment is formed by the loss of H 202 from the
species at m/z 131 to give the structure proposed.
However, due to the low relative abundance of this
daughter ion and the uncertainty in the fragmentation
pathway, m/z 97 is not included in the fragmentation
scheme of m/z 131 and m/z 113 shown in Figure 7.

] Am Soc Mass Spectrom 1990, I. 455-472

Relevant results from the fast atom bombardment mass
spectra of other cardiac glycosides. The effects of small
structural changes in the analyte molecule on the FAB
mass spectra and the B /E linked scan mass spectra of
the [MIH+ ions were investigated. Two cardiac gly-
cosides, digitoxin and gitoxin, were chosen for study
because of their differences in the aglycone portion of
the molecule from digoxin. The structures of these two
compounds are shown in Figure 1 where digitoxin is
structure E and gitoxin is structure G. Digitoxin has
one less OH group on the aglycone than digoxin, and
gitoxin has the same molecular formula but with differ-
ent positioning of one of the OH groups on the agly-
cone ring structure (C-12 on digoxin and C-16 on gi—
toxin). Another compound, acetyldigitoxin (structure
F) was chosen for its modification to the sugar por-
tion of the compound. The aglycone of acetyldigitoxin
is identical to that of digitoxin, but the terminal sugar
contains an acetyl group. The FAB mass spectra and
the BIE spectra of the [M]H+ ions of digitoxin, gitoxin,
and acetyldigitoxin were obtained and compared with
the corresponding spectra of digoxin in order to de-
termine whether changes in the aglycone portion or
the sugar portion of digoxin have more effect on the
types and relative abundances of the fragment ions ob-
served. The mass spectra of these related compounds
all contained relatively the same protonated fragments
(and parent ions) as observed for digoxin, with some of
the m /z values shifted due to the introduced modifica-
tions from digoxin. For example, with the compound
digitoxin, the ions in the m /z 337 to m /z 391 range
that were given two assignments for digoxin (from
the aglycone end and from the sugar portion of the
molecule for the same m/z value) were split into two
series, separated by 16 u, due to the change in the mass
of the aglycone portion of digitoxin. This further sup-
ports the double assignments given to these fragment
ions of digoxin in Table 1. The relative abundances of
the daughter ions in the BIE linked scan mass spectra
of the [MlH+ ions of the two compounds with agly-
cone modifications (digitoxin and gitoxin) are different
from digoxin. The most notable changes are that the
[M—HzolH+ ions are more abundant in the BIE mass
spectra of these two modified aglycone compounds
than for digoxin. This observation is probably a result
of the differences in locations, and hence reactivity, of
the —OH groups on the aglycone portions of these
compounds. The modifications to the sugar portion of
the compound digitoxin do not produce any notice-
able differences in the FAB mass spectrum or the BIE
spectrum of the [MjH+ ion of acetyldigitoxin from the
corresponding spectra of digoxin with respect to the
protonated species observed and their relative abun-
dances. One exception is the shift in mass of the ions
containing the terminal sugar with the added acetyl
group. Therefore, small changes in the aglycone por-
tion of these cardiac glycosides result in more dramatic
mass spectral changes than do modifications to the

192

HIGHLY FUNCTIONALIZED MOLECULES IN FAB 469

sugar portion of the molecules. The differences in the
FAB and BIE linked scan mass spectra produced by
slight alterations in the aglycone suggest that the agly-
cone does play a major role in controlling the frag-
mentation mechanism(s) and may possibly be the site
of charge localization. A possible explanation for the
importance of this aglycone on the mechanisms of frag-
mentation could be related to energetics and the pro-
ton affinities of specific sites on the aglycone. How-
ever, there is no obvious site where the charge resides
on the aglycone of digoxin. The modifications in the
aglycone of these cardiac glycosides may affect the en-
ergy that the [MlH+ ion contains as it fragments via
remote site processes and therefore may affect the rel-
ative abundances of the fragment ions observed.

Conclusions

The data presented from the FAB mass spectrum and
the MS/MS analyses of all the major fragment ions of
digoxin show that over half of the different types of
fragment ions formed contain the aglycone portion of
the molecule and that many different bonds are
cleaved. The results we obtained from the FAB mass
spectrum of digoxin can be closely compared with the
reported NH 3 CI spectrum of digoxin [26]. All of the
fragments, including those due to multiple H20
losses, that are observed in protonated form in the
FAB mass spectrum are observed in the NH, Cl
spectrum as NH: adducts. This suggests that very
similar fragmentation mechanisms are occurring with
these two techniques and that the difference between
protonation by glycerol and NH: adduct formation
does not affect the fragmentation mechanism(s) sig-
nifrcantly. If the fragment ions formed in the NH, Cl
experiment evolve from [M]NH*, as they appear to,
the charge resides on the ammonium ion and not on
some part of the digoxin molecule itself. This rules
out reaction 1 as a possible pathway of fragmentation
because it requires that the charge reside on the oxy-
gen of the glycosidic bond in order to induce frag-
mentation of this bond. Reactions 2 and 3 are possible
fragmentation pathways for NH3 Cl, as well as proto-
nation by glycerol fragment ions because the charge
site is not required to be on the oxygen atom, as in
reaction 1. However, further fragmentation of the
primary fragments would be unlikely following reac-
tion 2 because charge migration through the molecule
is required for the formation of the secondary frag-
mentations observed. If the charge site in the adduct
ion formed in NH 3 Cl is localized on the NH: ion,
then charge migration throughout the molecule would
not be possible. Therefore, reaction 3 would not pre-
dict the production of the same fragment ions in the
decomposition of protonated molecules and NH:
adducts. A remote site mechanism, reaction 4, pro-
vides the best explanation for the similarities in the
fragmentations of digoxin observed with FAB /MS and

470 LIGHT AND ALLISON

NH3 CI. In this case the charge site can be localized
on the aglycone and induce fragmentation at remote
sites throughout the molecule. As the charge is not
directly involved in the fragmentation and no charge
migration is required, it is expected, by reaction 4,
that the protonated ion and NH: adduct ion would
produce the same fragmentation products, as ob-
served, as long as the charge can be localized on the
aglycone in both cases.

The data presented here from the mass spectrome-
try and MS/MS analysis of digoxin lend further sup-
port to the role of the remote site mechanism in the
formation of the fragment ions observed. Reaction 1
can explain only a limited number of the types of
fragmentations of digoxin observed. For instance, the
ion at m/z 131 can be easily explained via reaction 1
by protonation of the oxygen atom of the terminal
glycosidic bond, inducing fragmentation to produce
the species at m/z 131. By this same mechanism,
other fragment ions at m/z 261 and m/z 391 should
be formed from protonation and fragmentation of the
other two glycosidic bonds. However, the ion current
at m/z 391 is primarily from the aglycone portion of
the molecule (see Tables 1 and 2) and no ion is
observed at m/z 261, in contradiction to what would
be expected by reaction 1. Therefore, reaction 1 fails
to provide a consistent explanation of the fragmenta-
tion pathways for the decomposition of digoxin and is
only a possible mechanism for the formation of a few
fragment ions observed.

Reactions 2 and 3, based on charge-induced frag-
mentation, can explain the majority of the primary
fragments observed in the FAB analysis of digoxin. It
is possible that these mechanisms can provide path-
ways for the extensive fragmentation of the primary
fragments as reported in Table 2. The charge site must
be able to migrate throughout the molecule following
primary fragmentation in order to explain all of the
ions observed in the MS/MS spectra, which may be
possible with the protonated species. Based on reac-
tion 3, competition for the charge should be deter-
mined by the relative basicity of the two competing
sites. This holds true for the formation of m /z 651 and

m/z 521 where fragmentation about the glycosidic
bond, following reaction 3, results in the charge site
residing on the nonreducing end of the sugar that
contains the aglycone portion of the molecule. This
nonreducing end of the sugar contains a more basic
site, due to the close proximity of two —OH groups,
than does the reducing end of the terminal sugar,
which has only a ring oxygen and double bond in
close proximity. However, this mechanism does not
hold true for the formation of the fragment ion at m/z
391. This species is primarily composed of the agly-
cone portion of the molecule, even though the agly-
cone has no -OH group or other functional group
near the site of protonation/fragmentation to increase
the basicity, whereas the sugar moiety has a ring

193

I Am Soc Mass Spectrom I990, I. 455-472 "

oxygen and double bond near the competing site. In
this case the aglycone end is not more basic than the
sugar end and therefore should not effectively com-
pete for the H’. Therefore, some other fragmentation
mechanism is required to explain the production of
m/z 391.

The remote site fragmentation mechanism can ex-
plain a high proportion of the fragment ions contain-
ing the aglycone and the occurrence of the majority of
the ions observed if the charge site is localized on the
aglycone. This requires that a basic site or stable
cation be formed for the charge to remain localized on
the aglycone throughout the extensive fragmentation
observed. There is no obvious basic site in the digoxin
molecule on the aglycone that has a higher proton
affinity than parts of the sugars, as can be seen in
Figure 5. However, the ﬁve-membered ring of the
aglycone provides a very plausible site for charge
localization. Protonation of the oxygen atom in this
strained frve-membered ring could induce ring cleav-
age followed by H migration to produce a tertiary
cation. This is shown in reaction 5. The formation of
the tertiary cation provides a charge site that hinders
free migration of the charge back onto the sugars.
This is one possible way that charge localization can
occur on the aglycone to allow for remote site frag-
mentation to produce the majority of the ions ob-
served. The modifrcations in the aglycone that were
investigated with digitoxin and gitoxin would not
prevent this charge localization and therefore would
still allow for the remote site fragmentation mecha-
nism to occur.

0 O
—" l

\i”

1 (s)

0

| +o-H O—H

\~’

O—H

I
\+ CH3

Thermodynamic considerations of these fragmenta-
tion processes support remote site fragmentation
mechanisms. Consider the case of the [MjH+ ion of a
multifunctional compound containing at least two
—OH groups, where one is protonated. If the ion is
formed with excess energy, fragmentation can occur.
In such a situation, ionic pathways are not necessarily

I Am Soc Mass Spectrom 1990, 1, 455—472

more energetically favored than neutral (remote site)
processes. These pathways are shown in reaction 6.
For example, pathway B, elimination of water due to
an inductive cleavage, yielding a carbocation, requires
approximately 23 kcal/mol [28]. In contrast, pathway
A, a remote site fragmentation to form a double bond
and H20, only requires 12 kcal/mol [28]. Thus, there
is no thermodynamic basis for excluding remote site
eliminations because they should be competitive with
ionic fragmentation mechanisms.

no —cn/ cn—onz’
\tzn2 cn2
-H -H (6 l
20 A B 20

cu ——cn
cn// 2\cu‘
\cn2 C“2/

This remote site mechanism, however, does not
explain the large ion currents at m/z 131 and m/z 113
that are present in the FAB analysis of digoxin. It is
possible that reaction 1 or 3 is the primary pathway for
producing these two ions. According to Figure 5 the
site with the highest proton affinity is on the terminal
sugar. This could allow for charge-induced fragmenta-
tion via reaction 1 or 3 to produce these two abundant
low mass fragment ions at m/z 113 and m/z 131.

It is highly probable that more than one mecha-
nism is involved in forming all of the fragment ions
observed in this FAB mass spectrometric study of
digoxin. A combination of these mechanisms can ex-
plain the results obtained satisfactorily. More exten-
sive studies on similar compounds need to be per-
formed to determine the exact fragmentation path-
ways involved in the formation of each fragment ion
observed by both primary and secondary fragmenta-
tion.

We have attempted to take a fundamental ap-
proach to the interpretation of the mass spectra ob-
tained in the FAB and CAD experiments of digoxin
and other related cardiac glycosides. The extensive
fragmentation observed for these compounds is cer-
tainly not uncommon in FAB analyses of such highly
functionalized compounds. It is possible that the ex-
tensive fragmentation observed for these types of
complex compounds may occur for different reasons
and that fragmentation patterns may not be transfer-
able to different types of compounds. However, the
information presented here may provide a starting
point for similar detailed analyses of the mechanisms

194

lllCllLY FUNCTIONALIZED MOLECULES IN FAB 471

of fragmentation of protonated molecules from other
highly functionalized compounds by FAB.

Acknowledgments

This work was supported by a grant from the Biotechnology
Research Program of the Divisrun of Research Resources of NIH
(RR-00480-20). We thank Gary A. Schultz for early discussions
and CAD experiments that led to the development of this
project, and John T. Stults tor conversations concerning the B/E
linked scanning technique.

References

l. McLafferty, F. W. Interpretation of Mass Spectra, 3rd ed.;
University Science: Mill Valley, CA, 1980.
2. Kassel, D. 8.; Glerum, M.; Robinson, B. H.; Sweeley, C. C.
Anal. Biochem. 1989, I 76, 382.
3. Sunner, I. A.; Kulatunga, R.; Kebarle, P.; Anal. Chem. 1986,
58. 1312.

. Holmes, I. L.; Org. Mass Spectrum. 1985, 20, 169.

5. Crow, F. W.; Tomer, K. 8.; Looker, I. H.; Gross, M. L.
Anal. Biochem. 1986, 155, 286.

6. Instruction Manual for MS-LNO9/08/11, Linked Neutral Loss
Scan Program, IEOL, Ltd., Tokyo, japan, No. IMS-LNS, p.
10. Sato, K.; Asada, T.; lshihara, M.; Kunihiro, F.; Kammei,
Y.; Kubota, E.; Costello, C. E.; Martin, S. A.; Scoble, H. A.;
Biemann, K. Anal. Chem. 1987, 59, 1652.

7. Holmes, 1. L.; Szulejko, I. E. Org. Mass Spectrum. 1983, I8,
273.

8. Light, K. I.; Kassel, D. 8.; Allison, I.; Biomed. Environ. Mass
Spectrom. 1989, I8, 177.

9. Domon, 8.; Costello, C. E. Cchoconjugate I. 1988, 5, 397.

10. Bochkov, A. F.; Zaikov, G. E. Chemistry of the O-Glycosidic
Bond: Formation and Cleavage; Pergamon: Oxford, 1979; pp
179-181.

11. Busch, K. L.; Glish, C. L.; McLuckey, S. A. Mass Spectrome-
try / Mass Spectrometry: Techniques and Applications of Tandem
Mass Spectrometry; VCH: New York, 1988.

12. (a) Field, F. H. In MTP International Review of Science Mass
Spectrometry Series One; Maccoll, A., Ed.; Butterworths, 1972.
(b) McLafferty, F. W. Org. Mass Spectrum. 1980, 15, 114.

13. (a) lchikawa, H.; Harrison, A. G. Org. Mass Spectrom. 1978,
l3, 389. (b) Martinsen, D. P.; Buttrill, S. E., Jr. Org. Mass
Spectrom. 1976, II, 762. (c) Pesheck, C. V., Buttrill, S. E., Ir.
I. Amer. Chem. Soc. 1974, 96, 6027. (d) Pollack, S. K.; Devlin,
I. L. III; Summerhays, K. D.; Taft, R. W.; Hehre, W. I. I.
Amer. Chem. Soc. 1977. 99, 4583.

14. Co, K.; Kartha, (3.; Chen, I. P. Acta Cryst. 1980, 836, 1811.

15. Lias, S. 0.; Liebman, I. F., Levin, R. D. I. Phys. Chem. Ref.
Data 1984, 13, 695.

16. (a) Aue, D. H.; Webb, H. M.; Bowers, M. T.; I. Amer.
Chem. Soc. 1973, 95, 2699. (b) Yamdagni, R.; Kebarle, P. I.
Amer. Chem. Soc. 1973, 95, 3504. (c) Longevialle, P.; Milne,
G. W. A.; Fales, H. M. I. Amer. Chem. Soc. 1973, 95, 6666.
(d) Houriet, R.; Rufenacht, H.; Carrupt, P. A.; Vogel, P.;
Tichy, M. I. Amer. Chem. Soc. 1983, 105, 3417. (e) Davis, D.
V.; Cooks, R. G. Org. Mass Spectrom. 1981, 16, 176. (f)
Houriet, R.; Rufenacht, H.; Stahl, D.; Tichy, M.;
Longevialle, P. Org. Mass Spectrom. 1985, 20, 300. (g) Nac-
son, 5.; Harrison, A. (3.; Davidson, W. R. Org. Mass Spec-
trum. 1986, 21, 317. (h) Longevialle, P.; Girard, ].-P.; Rossi,
I.-C.; Tichy, M. Org. Mass Spectrom. 1979, 14, 414. (i) McMa-
hon, A. W.; Chadikun, F.; Harrison, A. 6.; March, R. E.
Int. I. Mass Spectrom. Ion Process. 1989, 87, 275.

I7. Winkler, I.; McLafferty, F. W. Tetrahedron 1974, 30, 2971.

h

472 LIGHT AND ALLISON

18. Michnowicz, I.; Munson, 8. Org. Mass Spectrom. 1972, 6,
283.

19. Neumann, G. M.; Derrick, P. I. Aust. I. Chem. 1984, 37,
2261.

20. (a) Tsang, C. W.; Harrison, A. C. I. Amer. Chem. Soc. 1976,
98, 1301. (b) Mandelbaum, A.; Biemann, K. I. Amer. Chem.
Soc. 1968, 90. 2975.

21. (a) Iohnson, R. S.; Martin, S. A.; Biemann, K. Int. I. Mass
Spectrom. Ion Process. 1988, 86, 137. (b) Kiryushkin, A. A.;
Fales, H. M.; Axenrod, T.; Gilbert, E. I.; Milne. C. W. A.
Org. Mass Spectrom. 1971, 5, 19. (c) Winnik, M. A. Org. Mass
Spectrom. 1974, 9, 920.

2. Bowen, R. D.; Stapelton, B. I.; Williams, D. H. I. Chem. Soc.
Chem. Commun. 1978, 24.

23. Iensen, N. I.; Tomer, K. 8.; Gross, M. L. I. Amer. Chem.
Soc. 1985, 107. 1863.

24. Adams, I. Mass Spectrum. Rev. 1990, 9, 141.

25. Wysocki, V. H.; Bier, M. E.; Cooks, R. G. Org. Mass
Spectrum. 1988, 23, 627.

26. Vine, I.; Brown, L.; Boutagy, I.; Thomas, R.; Nelson, D.

Biomed. Mass Spectrum. 1979, 6, 415.
. Adams, I.; Gross, M. L. I. Amer. Chem. Soc. 1989, III, 435.
. Lias, S. (3.; Bartmess, I. E.; Liebman, I. F.; Holmes, I. L;
Levin, R. D.; Mallard, W. C. I. Phys. Chem. Re]. Data 1988.
17. Suppl. 1.

88

Appendix

The proton affinities listed in Figure 5 for the various
sites in digoxin, as indicated in Figure 4, are estimated
as follows. Three types of protonation sites are pre-
sent. First, we consider simple functional groups such
as double bonds (site a), —OH groups (site b), keto—
oxygens (site d), ring (ether) oxygens (site I), and
glycosidic (ether) oxygens (site i). In these cases,
small, simple molecules containing such functional
groups are used as models. Thus, the proton affinities
of these sites are based on the PA of the following
molecules: site a: 2-butene; site b: isopropanol; site d:
acetone; site I: tetrahydropyran; and site i: isopropyl
ether. The second type of protonation site is repre-
sented by compounds containing more complex func-
tional groups or multiple functional groups. For ex-
ample, site e is represented by an ester, which has a
PA different from that of an ether or a ketone. The PA
of site e is estimated based on the PA of methyl
acetate. In a similar way, site c is considered to be
similar to that found in the propenal molecule.

195

I Am Soc Mass Spectrom I990, 1, 455-472

The third type of protonation site involves multiple
interactions from heteroatoms/functional groups that
are in close proximity. Many studies reported an in-
crease in the stability of the protonated species with
the addition of functional groups, such as —OH, to
the molecule [16]. This is believed to be due to in.
tramolecular H-bonding with these functional groups
upon protonation. For example, butanol has a PA of
191 kcallmol [15] compared to 1,2,4-butanetriol, which
has a PA of 216 kcallmol [3]. This increase in PA of
25 kcallmol can be attributed to the two extra —-OH
groups, which allow for multiple H-bonding inter-
actions that increase the stability of the protonated
species. Following this pattern, the PA of propanol,
190 kcallmol [15] can be compared to the PA of glyc-
erol, which is 209 kcallmol [3]. In this case, the addition
of two -OH groups leads to an increase in the PA of
the molecule of approximately 19 kcallmol. Therefore,
we propose that similar H-bonding interactions with a
corresponding increase in PA occur with the molecule
digoxin when functional groups such as -OH groups,
ring oxygens, and glycoside oxygens are in close prox-
imity to each other, due to the stereochemistry of the
molecule. We have conservatively proposed, based on
the above PA values for linear mono and trials, that
the interaction of a single —OH group with an —OR
group (such as the glycosidic oxygen site on digoxin)
results in an increase in PA of the more basic site of
approximately 5 kcallmol. For instance, site I has two
—OH groups near the —OR (glycosidic 0), a1] of which
are on the same side of the ring and have the potential
for H-bonding. If the presence of each —OH group in.
creases the PA of this site by at least 5 kcallmol, then
site I would have a PA of at least 10 kcallmol above the
PA of the glycosidic oxygen (site i) alone. This is repre-
sented in Figure 5 as a shaded region above the value
for site i, due to the uncertainty of these estimates and
interactions. In a similar way, it has been estimated
that the presence of a ring oxygen near a glycosidic
oxygen (site i) may increase the PA of that site by at
least 3 kcallmol. These estimates for the proton affini-
ties for the possible sites of protonation on digoxin,
based on similar organic molecules and the possible
multifunctional interactions that have been shown to
occur for other molecules, are presented in Figure 5.

10.

11.

12.

13.
14.

15.

16.

17.

18.

REFERENCES

F. W. McLafferty, Interpretation of Mass Spectra, 3rd. Ed., University
Science Books, Mill Valley, CA, 1980.

M. S. B. Munson and F. H. Field, J. Amer. Chem. Soc., 88, 2621 (1966).
H. D. Beckey, Int. J. Mass Spectrom. Ion Phys., 2, 500 (1969).

D. F. Torgerson, R. P. Showronski, and R. D. Macfarlane, Biochem. Biophys.
Res. Commun., 60, 616 (1974).

M. A. Posthumas, P. G. Kistemaker, H. L. C. Meuzelaar, and M. C. Ten Nocvcr
dc Brauw, Anal. Chem, 50, 985 (1978).

M. Barber, R. S. Bordoli, R. D. Sedgwick, and A. N. Tyler, J. Chem. Soc. Chem.
Commun., 325 (1981).

A. Benninghoven, Int. J. Mass Spectrom. Ion Phys., 46, 459 (1983).

A. Dell, D. H. Williams, H. R. Morris, G. A. Smith, J. Feeney, and G. C. K.
Roberts, J. Amer. Chem. Soc., 97, 2497 (1975).

M. A. Baldwin, and F. W. McLafferty, Org. Mass Spectrum, 7, 1353 (1973).
M. L. Vestel, Mass Spectrum. Rev., 2, 447 (1983).
R. M. Caprioli, T. Fan, and J. S. Cottrell, Anal. Chem, 58, 2949 (1986); L. J.

Deterding, M. A. Moseley, K. B. Tomer, and J. W. Jorgenson, Anal. Chem., 61,
2504 (1989).

F. W. McLafferty, Ed., W, Wiley, New York (1983).
R. J. Cotter, Biomed. Environ. Mass Spectrum, 18, 513 (1989).
L. M. Mallis and D. H. Russell, Anal. Chem, 58, 1076 (1986).

R. L. Cemy, K. B. Tomer, and M. L. Gross, Org. Mass Spectrum. 21, 655
(1986); Z. Zhou, S. Ogden, and J. A. Leary, Proc. 38th ASMS Conf. Mass
Spectrom. Allied Tupics, 1990, 974.

L. M. Teesch and J. Adams, Proc. 38th ASMS Conf. Mass Spectrom. Allied
Topics, 1990, 457; R. P. Reese, R. Cemy, and M. L. Gross, J. Amer. Chem. Soc.,
111, 2835 (1989); D. Renner and G. Spiteller, Biomed. Environ. Mass
Spectrum., 15, 75 (1988); X. Tang, W. Ens, K. G. Standing, and J. B. Westmorc,
Anal. Chem, 60, 179 (1988).

F. W. Rollgen and H. R. Schulten, Z. Naturforsch., 303, 1685 (1975); F.
Borchers, U. Giessmann and F. W. Rollgen, Org. Mass Spectrum, 12, 539
(1977).

R. Stoll and F. W. Rollgen, Z. Naturfursch., 37a, 9 (1982).

196

19

20.

21.

22.

23.

24.

25.

26.

27.

28.
29.
30.

31.

32.

33.

34.

35.

36.

37.

197

G. Schmelzeisen-Redeker, U. Giessmann and F. W. Rollgen, Org. Mass
Spectrum., 20, 305 (1985).

G. J. Q. van der Peyl, K. Isa, J. Haverkamp, and P. G. Kistemaker, Org. Mass
Spectrum., 16, 416 (1981).

J. B. Fenn, M. Mann, C. K. Meng, S. F. Wong, and C. M. Whitehouse, Science.
246, 64 (1989).

M. Karas and F. Hillenkamp, Anal. Chem. 60, 2299 (1988); M. Karas, U. Bahr,
A. Ingendoh, and F. Hillenkamp, Angew. Chem. 101, 805 (1989).

M. Mann, C. K. Meng, and J. B. Fenn, Anal. Chem, 61, 1702 (1989); J. A. Loo,
H. R. Udseth, and R. D. Smith, Anal. Biochem., 179. 404 (1989).

M. Dole, L. L. Mack, R. L. Hines, R. C. Mobley, L. D. Ferguson, and M. 8. Alice,
J. Chem. Phys., 49, 2240 (1968); L. L. Mack, P. Kralik, A. Rheude, and M.
Dole, J. Chem. Phys, 52, 4977 (1970).

M. Yamashita and J. B. Fenn, J. Phys. Chem, 88, 4451 (1984).

C. K. Meng, M. Mann, and J. B. Fenn. Z. Phys. D., 10, 361 (1988).

D. Bombick, J. D. Pinkston, and J. Allison, Anal. Chem, 56, 396 (1984); D.
Bombick and J. Allison, Anal. Chem, 59, 458 (1987).

J. P. Blewett and E. J. Jones, J. Phys. Rev., 50, 464 (1936).
D. Bombick, Doctoral Dissertation, Michigan State University, 1986.

R. J. Beuhler, E. Flanigan, L. J. Greene and L. Friedman, J. Amer. Chem. Soc.,
96, 3990 (1974).

R. L. Woodin and J. L. Beauchamp, Chem. Phys., 41, 1 (1979).

See, for example B. C. Guo, B. J. Conklin, and A. W. Castleman, Jr., J. Amer.
Chem. Soc., 111, 6506 (1989).

T. A. Lehman and M. M. Bursey, [on Cyclotron Resonance Spectrometry, J.
Wiley & Sons, New York (1976).

R. J. Beuhler, E. Flanigan, L. J. Greene, and L. Friedman, Biochem. Biophys.
Res. Commun., 46, 1082 (1972).

V. N. Reinhold and S. A. Carr, Anal. Chem. 54, 499 (1982).

D. 1. Carroll, J. G. Nowlin, R. N. Stillwell, adn E. C. Homing, Anal. Chem, 53,
2007 (1981); W. Lange, M. Jirikowsky, and A. Benninghoven, Surf. Sci., 136,
419 (1984).

R. J. Cotter, Anal. Chem, 51, 317 (1979).

38.

39.

40.

41.

42.

43.
44.

45.

46.
47.

48.

49.

50.

51.

52.

53.

54.
55.

56.

198

G. P. Johnson, A. B. Hedin, P. L. Hakansson, B. U. R. Sundqvist, B. G. S. Save,
P. F. Nielson, P. Ruepstorff, K. E. Johansson, 1. Kamensky, and M. S. L.
Lindberg, Anal. Chem.,58, 1084 (1986).

J. P. Thenot, J. Nowlin, D. 1. Carroll, F. E. Montgomery, and E. C. Homing,
Anal. Chem.,51, 1101 (1979).

B. W. Williams, A. P. Irsa, H. Zmora, and R. J. Beuhler, J. Phys. Chem, 87,
2185 (1983).

D. B., Kassel], Doctoral Dissertation, Michigan State University, 1988.

CRC Handbook of Chemistry and Physics, 63rd Ed., CRC Press, Inc., Boca
Ratun, FL, 1982.

D. Bombick and J. Allison, Anal. Chim. Acta., 208, 99 (1988).

D. B. Kassel and J. Allison, Biomed. Environ. Mass Spectrum., 17, 221 (1988).
See, for example: R. E. Shumo, II, A. Chandrasekaran, A. G. Marshall, R. H.
Reuning, and L. W. Robertson, Biomed. Environ. Mass Spectrum., 15, 295
(1988).

V. N. Reinhold and S. A. Carr, Mass Spectrom. Rev., 2, 216 (1983).

P. M. Holland and A. W. Castleman, Jr., J. Amer. Chem. Soc., 102, 6174 (1980).

J. Allison, “The Gas Phase Chemistry of Transition- Metal Ions with Organic

Molecules", in Erugr gas in Inimgang; Chemistg, vol. 34, S. J. Lippard, Ed”
John Wiley and Suns,Inc.,1986.

a) M. Lombarski and J. Allison, Int. J. Mass Spectrom. Ion Phys., 49, 281
(1981); b) M. Lombarski and J. Allison, Int. J. Mass Spectrom. Ion Phys., 65,
31 (1985).

J. Vulkening and K. G. Huemann, n M r m A v n e
W, 1985, Part B, John Wiley & Sons, 1059.

See, for example: J. Allison and D. P. Ridge, J. Amer. Chem. Soc., 101, 4998
(1979); M. L. Larrivee and J. Allison ?????

J. Allison and D. P. Ridge, J. Amer. Chem. Soc., 101, 4998 (1979).

S. G. Lias, J. E. Bartmess, J. F. Liebman, J. L. Holmes, R. D. Levin, and W. G.
Mallard, I, Phys Chem. Re], Data, 17 (1988).

R. E. Weber and L. I. Cordes, Rev. Sci. Instrum, 37, 112 (1966).
R. J. Cutter, Anal. Chem, 52, 1589A (1980).

J. C. Rouse, K. J. Light, and J. Allison, Prue. 38th ASMS Conf. Mass Spectrum.
Allied Topics, 1990.455.

199

57. J. R. Pare, P. Lafontaine, J. Belanger, W. W. Sy, N. Jordan, and J. C. K. Loo, J.
Pharm. Biomed. Anal.,S, 131 (1987).