MIC CHI GANSTATE

ll lllllllllzllll

 

 

 

 

 

 

 

 

 

 

llllllllllll‘llllllllllllllll

93 01026 2073

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

Investigation of the Ion Formation and Unimoleoular
Decomposition Mechanisms in the Desorption/Ionization
Technique Fast Atom Bombardment

presented by

Jason C. Rouse

has been accepted towards fulfillment
of the requirements for

PhD. Chemistry

degree in

 

 

/%fl%

Major professor

 

Date 4 Jeff 73

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

mouthi- checkmnflom your rooofd.

PLACE iN RETURN 80X to
data duo.

1’0 AVOID FINES roturn on or baton

   
       
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU is An Affirmative Action/Equal Opportunity institution
Want

____————’

 

 

 

INVESTIGATION OF THE ION FORMATION AND UNIMOLECULAR
DECOMPOSITION MECHANISMS FOR THE
DESORPTION/IONIZATION TECHNIQUE FAST-ATOM
BOMBARDMENT (FAB) MASS SPECTROMETRY

By

Jason Craig Rouse

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Chemistry

1993

ABSTRACT

INVESTIGATION OF THE ION FORMATION AND UNIMOLECULAR
DECOMPOSITION MECHANISMS FOR THE
DESORPTION/IONIZATION TECHNIQUE FAST-ATOM
BOMBARDMENT (FAB) MASS SPECTROMETRY

By

Jason Craig Rouse

A divided probe that incorporates a potassium aluminosilicate glass target and an
analyte/glycerol matrix target, spatially-separated, was used to inject potassium ions (K+)
into the high pressure "selvedge" region formed above the analyte/glycerol matrix target
during fast-atom bombardment (FAB). [M+K]+ adduct ions that represent the types of
gas-phase neutral molecules (M) present in the selvedge region are observed. Computer
modeling assisted in designing the divided target and an additional ion optical element for
the FAB ion source to optimize interactions between K+ ions and the desorbed neutral
molecules. The capability of injecting K+ ions into the FAB experiment has utility in
both mechanistic studies and analysis. Experimental results here are consistent with a'
model for the desorption/ionization processes in FAB in which some types of neutral
analyte molecules, M, are desorbed intact and are subsequently protonated by glycerol-

chemical ionization. Unstable protonated molecules undergo unimolecular

decompositions to yield observed fragment ions. The use of K+ cationization of analytes
for molecular weight confirmation is demonstrated, as well as its utility in FAB
experiments in which mixtures are encountered.

Peptide sequence ion fragmentation mechanisms that define the relationship
between the relative intensity (RI) information in liquid secondary-ion mass spectrometry
(LSIMS), metastable ion (MI) decomposition, and single-collision, high-energy
collisional activation dissociation (CAD) mass spectra, and the protonated structure of
pentaglycine (no side chains) and methionine enkephalin (side chains) were evaluated by
using the RI distributions in the respective spectra and estimated reaction enthalpies for
the formation of each sequence fragment ion, an, bn, on, x", yn and z". The most likely
protonation sites in these pentapeptides are the amide nitrogens based on their high
estimated proton affinities and by the straightforward fragmentation mechanisms that can
be written from this site to explain the formation of each sequence ion. The RI
distributions and reaction enthalpies suggest that both simple cleavages and secondary
interactions (i.e., when a remote site of the molecule interacts with the protonation site to
induce chemistry) are competitive processes from this site in LSIMS and CAD whereas
only secondary interactions are prominent for M1. The ratios of fragment ions from each
protonated amide nitrogen were determined not to be equal and it appears secondary
interactions either redistribute the initial protonation sites and/or induce specific bn, y"
and on fragmentation. Although the most favorable pathways are not always followed,
the most stable fragment ions and neutrals are formed in every peptide fragmentation
process, depending on the amount of the internal energy in the [M+H]+, whether simple

cleavages or secondary interactions are required to propagate the chemistry.

iv

This dissertation is dedicated to my Father and in memory
of my loving Mother, Lee and Phyllis, for their endless
love, guidance, encouragement, and support which was
essential in my pursuit of this advanced degree. I want
to also thank my late Grandfather, Andrew T. F enrich,
who referred to me as "his grandson, the doctor", for
initially introducing me to science and providing me
with love, encouragement, and support needed to
pursue this profession.

ACKNOWLEDGMENTS

This advanced degree would have been impossible for me to complete alone, and
I am eternally grateful to everyone who provided me with assistance, guidance, and
moral, technical and financial support in my five years of graduate school at MSU.

I am very grateful to Jeff Gilbert because he made my move from The College of
Wooster, a small liberal arts college, to MSU a smooth transition both socially and
academically. Jeff and I were both members of the Beta Kappa Phi Fraternity at
Wooster, and, here in graduate school at MSU, we became best friends, roommates, and
members of the Allison research group. Karen Light-Wahl was another College of
Wooster graduate and a former Allison research group member who initially taught me
how to operate a mass spectrometer. I appreciate all her efforts to help me begin my
research project and I hope we can "spike" a beer (and break glass!) when I graduate.
The first year of graduate school is very intellectually challenging and I am indebted to
my fellow colleagues Kurt Kneen, Ed Townsend, Mike Waldo, and Julie Homer for their
genuine friendships and their helpful contributions in our first-year study groups. I was
glad to become good friends with fellow MSU students Kris Kurtz, Kimberly Gilbert,
Gary Schultz, Tracy Kneen, Man'a Townsend, Laura Pence, Dave Gale, Jim Ridge, Jon
Wahl, Dave Wagner, Art Harms, and Mary Puzycki. Amazingly, everyone mentioned
above would get together quite often during the week to celebrate events at the local bars,
go out to dinner, see a movie, play volleyball, etc., which was a needed break from the
stresses of graduate school.

In the NIH supported Mass Spectrometry Facility (MSF) in Biochemistry at

MSU, I am appreciative of all the technical assistance from J. Throck Watson, Doug

vi

Gage, Mike Davenport, Mel Micke, Bev Chamberlin, and Melina Beming. The MSF has
provided me a great working atmosphere, state-of-the-art instrumentation, and the
opportunity to attend the numerous mass spectrometry meetings, both local and national.
A most rewarding experience for me was performing maintenance with Mike Davenport
on the JEOL HX-l 10 because I learned much about the electronics, nuts and bolts, and
workings of a double-focusing mass spectrometer. I am obliged to Kermit Johnson and
Evy Jackson in the Max T. Rodgers NMR Facility in Chemistry for their run-time
assistance and technical support for the "bradykinin" project. Although the project is not
finished, I still learned much about the cutting-edge techniques in NMR spectroscopy. I
want to acknowledge Joe Leykam, the manager of the Structure Facility in Biochemistry,
for teaching me the principles of peptide synthesis and purification. Last, I want to
acknowledge NIH, the MSU Research Excellence Fund, and the Analytical Sciences
Group at Dow Chemical for moneys supporting my research assistantships and supplies.

I met my loving wife, Sue, at the College of Wooster my sophomore year. After I
started at MSU, we endured a difficult, long-distance relationship until my third year of
graduate school when we were finally married on August 17, 1991. We have been very
happy ever since. Sue has provided me with love, encouragement, feedback, and moral
support during my entire graduate school endeavor, and with her by my side, the difficult
times and tasks became much easier knowing she was there rooting for me.

Last, but by far the most important person during my graduate career at MSU was
my advisor John Allison. I am very grateful to receive from JA a wealth of graduate
career guidance, encouragement, and technical, moral and financial support over the
years. The most rewarding times in graduate school were one-on-one meetings with JA
either interpreting data, discussing possible research approaches, learning the correct way
to give a seminar, or writing (or downsizing) a paper together. I learned the most about
real science during those times. JA, I thank you for being a challenging but fair advisor,

and I wish the best for you in the future.

TABLE OF CONTENTS

LIST OF TABLE xi
LIST OF FIGURES x
LIST OF ABBREVIATIONS xvii
CHAPTER ONE. INTRODUCTION 1

1. Mass Spectrometry l

2. Overview of Ionization Techniques in Mass Spectrometry 3

3. The Focus, Goals, and Organization of the Dissertation 8
CHAPTER TWO. THEORY AND METHODS 13

1. The Ion Formation and Dissociation Processes in the Traditional
Ionization Techniques Electron Ionization (E1) and Chemical
Ionization (CI) - 13
2. Potassium Ion Ionization of Desorbed Species (K+IDS) 29

3. Fast-Atom Bombardment (FAB) Mass Spectrometry and Related
Particle Bombardment D/I Techniques 34

4. Principles and Utility of Collisionally-Activated Dissociation (CAD) 58
5. The Mass Analyzers Used for D/I Experiments 62

CHAPTER THREE. K+IDS-BY-FAB TARGET DEVELOPMENT AND
APPLICATION TO MECHANISTIC STUDIES OF ION FORMATION

IN FAST-ATOM BOMBARDMENT MASS SPECTROMETRY 77
1. Introduction 77
2. Development and Characterization of the K+IDS-by-FAB Target 81

3. Mechanistic Studies of Ion Formation in FAB using KflDS-by-FAB 126

4. Application of NMR for Determining the Charge State of Peptides in
Solution 142

5. Conclusions 145

vii

viii

CHAPTER FOUR. THE EVOLUTION OF FRAGMENT IONS FROM
PROTON ATED PEPTIDES IN LIQUID SECONDARY-ION MASS
SPECTROMETRY: MECHANISTIC AND THERMOCHEMICAL
CONSIDERATIONS
1. Introduction
2. Experimental Details
3. Results and Discussion
4. Conclusions

CHAPTER FIVE. FUNDAMENTAL AND ANALYTICAL
APPLICATIONS OF GAS-PHASE METAL ION/BIOMOLECULE
REACTIONS

1. Introduction
2. Reactivities between Metal Ions and Analyte Molecules in M+IDS

3. Gas-Phase Metal Cationization in FAB Using M+IDS-by-FAB

APPENDD( ONE. INJECTION OF REAGENT IONS INTO THE
SELVEDGE REGION IN FAST—ATOM BOMBARDMENT
MASS SPECTROMETRY

APPENDIX TWO. DATA FOR ESTIMATION OF REACTION
ENTHALPIES FOR PEPTIDE SKELETAL BOND CLEAVAGES

LIST OF REFERENCES

147
147
150
152
204

207
207

210

221

244

255
261

LIST OF TABLES

Table 1. PAs and IE5 for small molecules with unique functional groups. All values
were obtained from Lias et a1. [40] except when noted.

ix

Figure 1.
Figure 2.

Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure 1 1.

Figure 12.

Figure 13.

Figure 14.
Figure 15.

Figure 16.

Figure 17.

LIST OF FIGURES

Schematic diagram of a mass spectrometer.
Schematic diagram of an E1 ion source.

An EI clastogram for methanol (Adapted from [39]).

Thermochemical considerations for unimolecular decomposition of M“.
Estimated Wahrhaftig diagram for ionized CH3CH2CH2CH20CH3.
Schematic diagram of the K+IDS experiment.

An Arrhenius plot that demonstrates the competition between desorption
and degradation processes in thermal heating.

Schematic diagram of the fast atom gun (actually a SIMION model drawn to
scale) and the FAB experimental configuration on the JEOL HX—l 10.

Schematic diagram of the fast ion gun (actually a SIMION model drawn to
scale) and the LSIMS experimental configuration on the JEOL HX-l 10.

LSIMS mass spectrum of lug/11L of triglycine (MW 189) dissolved in the
glycerol matrix.

Energy deposition by a keV-Xe beam induces desorption of atoms, fragment
ions, preionized molecules, and intact molecules (not drawn to scale).

The neutral and ionic species desorbed into the selvedge region above the
FAB target during Xe bombardment.

The extraction and acceleration of M” a) from an ion source and b) the
representative potential energy surface for this ion source.

Schematic diagram of a double-focusing mass spectrometer.

Schematic diagram of a quadrupole mass analyzer.

A plot of the B, E plane for linked scans at constant B/E and B2/E on
double-focusing instruments.

A KHDS-by-FAB divided target under Xe bombardment demonstrating the
injection of desorbed K+ ions into the selvedge region for the

Figure 18.

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Figure 23.

Figure 24.

Figure 25.

Figure 26.

xi

identification of desorbed neutral molecules.

Figure 18. a) FAB mass spectra of K+ glass emitter and of b) KCl solid film
on a divided target half without matrix or analyte present on the opposite

half. c) KHDS-by-FAB mass spectrum of a KCl emitter and glycerol on
either half of a dual-surface space-divided target. The latter spectrum is

presented to show K0 is just as effective as K+ glass emitter in a K+IDS-
by-FAB experiment, but the KCl emitter is more difficult to apply cleanly
to only one target surface plus it needs to be regenerated for each run.

a) The first K+IDS-by-FAB type target employing a K+ glass emitter
reported by Akerrnann et al. [164], b) the wall and slit targets reported by
Munster et al. [166, 168], c) the dual-surface target reported by Miller et

al. [169], d) the KHDS-by-FAB "set screw" target reported by Schultz et
al. [172], e) the K+IDS-by-FAB "wall-divided" target reported by Kassel
[171], f) the K+IDS-by-FAB "K+ bead pop-up" target, g) the K+IDS-by-
FAB "vertical arrangement" FD/FAB type divided target, and h) the
vertical, dual—surface space-divided target, all which provided

unsatisfactory gas-phase K+ adduct ion formation for mechanistic studies.

a) FAB mass spectrum of glycerol (MW 92) matrix, b) typical K+IDS-by-
FAB mass spectrum when the K+ glass emitter on any divided target is
contaminated with glycerol/matrix, and c) K+IDS-by-FAB mass spectrum

of the glycerol matrix obtained with the KflDS-by-FAB space-divided
target and LO lens element installed in the ion source.

Different views of the a) K+IDS-by-FAB space-divided target with the K+
glass emitter bead and analyte/matrix sample holder, b) its orientation with
the impinging FAB beam and exiting secondary ions, and its typical c)
side and d) top dimensions.

a) FAB mass spectrum and b) early K+IDS-by-FAB mass spectrum of H-

LGG-OH (MW 245) in glycerol on the standard JEOL target and K+IDS—
by-FAB space-divided target, respectively.

Early K+IDS-by-FAB mass spectrum of sucrose (MW 342) in glycerol
using the K+IDS-by-FAB space-divided target.

Schematic diagram for the implementation of thermally-assisted K+IDS-by-
FAB on the JEOL HX-l 10.

Figure 25. Thermally-assisted K+IDS-by-FAB mass spectra of a) thermally-
heated K+ glass emitter, FAB off, and no analyte on FD/FAB sample

target; b) thermally-heated K+ glass emitter, FAB off, and glycerol on
FD/FAB target; and c) same as b except FAB is on.

SIMION models of the FD/FAB ion source coupled with the a) standard
FAB target, b) wall-divided target, c) dual-surface space-divided target,
and K+IDS-by-FAB space-divided target showing (I) analyte/matrix ion
and e) K+ ion trajectories.

xii

Figure 27. SIMION models of the FD/FAB ion source and L0 ion optical element

Figure 28.

coupled with the a) standard FAB target, b) wall-divided target, c) dual-
surface space-divided target, and KflDS-by-FAB space-divided target
showing (I) analyte/matrix ion and e) K+ ion trajectories.

Estimated extraction times for typical ions from the a) FAB target, b) dual-

surface space-divided target, and c) KflDS-by-FAB space-divided target
SIMION models, with and without L0 present, shown in Figures 26 and
27a, c, d, and e, respectively. The KflDS-by-FAB space-divided target
combined with L0 is best technique for injecting low energy K+ ions, with
longer residence times, into the selvedge region.

Figure 29. SIMION model of dual-surface space-divided target and full FD/FAB ion

Figure 30.

Figure 31.

Figure 32.

Figure 33.

source a) shows K“ and glycerol ions, extracted from each respective
target half, leave the ion source in two separate beams, b) mass spectrum
of K+ beam from K+ glass emitter target half, c) mass spectrum of the
glycerol beam from glycerol target half, (I) "rare" mass spectrum of both
K+ and glycerol ion beams, e) typical mass spectrum of both K+ and
glycerol ion beams showing glycerol contamination of K+ glass emitter,
and 1') same SIMION model as in 3, except with L0 installed, shows a
single well-defined beam leaving the ion source which eliminates ion
beam focusing problems.

Sequential K+IDS-by-FAB mass spectra of glycerol using the K+IDS-by-

FAB space-divided target a) with L0 installed initially and b) then with L0
removed which demonstrates the effectiveness of L0 for enhanced

[M+K]+ adduct formation.

The KfiDS-by-FAB mass spectra of glycerol from sequential experiments

in which the KflDS-by-FAB space-divided target was positioned a)
distant from, and b) then close to the L0 lens element resulting in an
increase in the intensity of the peak representing [G+K]+ in the latter
experiment owing to the possible creation of a higher pressure region
between the target and L0.

The K+ cationization efficiency of desorbed neutral molecules and the effect

of L0 slit width on K+ adduct formation is demonstrated in sequential
experiments using thevetin A (MW 872) and B (MW 858). A comparison

between the a) FAB mass spectrum, and the K+IDS-by-FAB mass spectra
b) with a narrow L0 slit width (out of ion optical tune) and c) with a wide

L0 slit width reveals 30% of total desorbed neutral molecules are K+
cationized, if [M+H]+ and [M+K]+ are summed from b and c, averaged,

and compared to [M+H]+ in a. Slightly more K+ adducts are obtained.
from a narrow slit width.

a) A FAB mass spectrum, b) KflDS-by-FAB mass spectrum, using with the

K+IDS-by-FAB space-divided target with L0 installed, and c) same as b,
except L0 was not installed, were obtained for Kassinin (MW 1334) at a
resolution of 1000. A comparison between b and c demonstrates that L0
enhances resolution of the peaks.

Figure 34.

Figure 35.

Figure 36.

Figure 37.

Figure 38.

Figure 39.

Figure 40.

Figure 41.

xiii

The utility of the LO lens element in front of the FAB target for delaying ion
extraction and possibly providing a higher pressure region above the FAB
target is demonstrated, in sequential scans, for the analysis of methionine
enkephalin (MW 573) in glycerol. With L0 present, a) an intense peak

representing [M+H]+ was observed in the initial FAB mass spectrum, b) a
successful KflDS-by-FAB mass spectrum validated a, and last, c) L0 was
removed resulting in a decrease in the intensity of [M+H]+.

Sequential experiments to prove K+ forms adduct ions in the gas phase and
K+ is not readily deposited into the upper layers of glycerol in K+IDS-by-
FAB. First, a) K+IDS-by-FAB mass spectrum of glycerol on K+IDS-by-
FAB space-divided target which yields 6 counts for [G+K]+, then b) the
K+ glass emitter is detached from this target and the residual glycerol is
reanalyzed by FAB to probe K+ contamination which yields 0.6 counts,

and last c) the residual K+ in new glycerol placed on the same target in b
is probed.

Cholic acid (MW 408), in glycerol, a) FAB mass spectrum, b) K+IDS-by—
FAB mass spectrum, c) FAB mass spectrum with 0.01 M KCl salts, d)
FAB mass spectrum with 0.1 M KCl salts, e) FAB mass spectrum of

analyte/matrix contaminated K+ glass emitter.

Two possible mechanisms for the formation of the [M+H-3H20]+ fragment
ion in FAB of cholic acid (MW 408).

A comparison between the molecular weight region peak intensities for the
analytes: 3) glycerol, b) cholic acid, c) digoxin, d) stachyose, e) kassinin,
and f) bradykinin dissolved in glycerol and analyzed by FAB, by K+IDS-
by-FAB, by FAB in 0.01 M KCl also, by FAB in 0.1 M KCl also, and on
the surface of the K+ glass emitter by FAB.

By applying an extremely small amount of glycerol on the target to reduce
the glycerol-related reagent ions, K+ became the dominant reagent ion as
shown in this K+IDS-by-FAB mass spectrum of digoxin (MW 780) since
the peak representing the [M+K]+ adduct ion is more intense in

comparison to [M+H]+ and there is no fragmentation observed from either
species.

a) FAB mass spectrum and b) KHDS-by-FAB mass spectrum of
benzyltriethylammonium chloride (MW 227) in glycerol, with L0 present,

obtained from a regular FD/FAB target and KflDS-by-FAB space-divided
target, respectively.

a) FAB mass spectrum and b) K+IDS-by-FAB mass spectrum, obtained

with a regular FD/FAB target and the KflDS-by-FAB space-divided
target, respectively, with L0 present, of a glycerophospholipid mixture
containing l-stearoyl-2-palmitoleoyl-3-phosphatidylcholine and/or 1-
palmitoyl-2-oleoyl-3-phosphatidylcholine as M] (MW 759), and l-
stearoyl-2-oleoyl-3-phosphatidylcholine and/or 1-palmitoyl-2-
arachidonoyl-3-phosphatidylcholine as M2 (MW 787), in glycerol.

Figure 42.

Figure 43.

Figure 44.

Figure 45.

Figure 46.

Figure 47.

Figure 48.

Figure 49.

Figure 50.

Figure 51.

Figure 52.

Figure 53.

Figure 54.

xiv

Fractional composition diagrams for polyprotic peptides a) H-VGVAPG-
0H, b) kassinin, c) bradykinin, and d) sleep-inducing peptide in which the
prominent forms at pH=7 and other peptides exhibiting similar behavior
are listed on left hand column.

a) FAB mass spectrum of H-VGVAPG-OH (MW 498), b) KflDS-by-FAB
mass spectrum of H-VGVAPG-OH, and c) KHDS-by-FAB mass
spectrum of H-VGVAUDPG-OH (MW 499) in glycerol using either the

FD/FAB or K+IDS-by-FAB space-divided target, respectively, with L0
installed.

a) FAB mass spectrum and b) KflDS-by-FAB mass spectrum of the sleep-
inducing peptide, H-WAGGDASGE-OH (MW 848), in glycerol, with L0
installed, using the FD/FAB and K+IDS-by—FAB space—divided targets,
respectively.

Representative fragment ion structures for the peptide methionine
enkephalin (MW 573).

a) LSIMS, b) MI, c) MI/CAD, and (1) "true" CAD mass spectra for
methionine enkephalin (MW 573). See text for description of spectra and
peptide sequence ion nomenclature.

a) LSIMS, b) MI, c) MI/CAD, and d) "true" CAD mass spectra for
pentaglycine (MW 303). See text for description of spectra and peptide
sequence ion nomenclature.

Estimated PAs for various basic sites on methionine enkephalin (H-
YGGFM-OH) are compared to PA of glycerol.

Bar graphs showing the percent of skeletal bond cleavage products versus
skeletal bond positions from a) LSIMS, b) MI, c) "true", single collision
CAD mass spectra of the peptide pentaglycine (H -GGGGG-0H ).
Structure and skeletal bonds are shown at the top.

Bar graphs showing the percent of skeletal bond cleavage products versus
skeletal bond positions from a) LSIMS, b) MI, c) "true", single collision
CAD mass spectra of the peptide methionine enkephalin (H-YGGFM-
0H). Structure and skeletal bonds are shown at the top.

Estimated reaction enthalpies for peptide skeletal bond cleavages in
kcal/mol.

Unimolecular decomposition pathways and their estimated reaction
enthalpies (in kcal/mol) for excited peptide [M+H]+ ions.

Bar graphs comparing normalized RIs of bu and y" peaks in the LSIMS and
MI spectra of a) triglycine, b) tetraglycine, c) pentaglycine, and d)
hexaglycine.

Fragmentation barrier profiles based on estimated reaction enthalpies for an
and b" ions from unimolecular decompositions of a) methionine

Figure 55.

Figure 56.

Figure 57.

Figure 58.

Figure 59.

Figure 60.

Figure 61.

Figure 62.

Figure 63.

Figure 64.

XV

enkephalin and b) pentaglycine protonated molecules.

Bar graphs comparing normalized RIs of 1),, and yn peaks in the MI spectra
of underivatized and acetylated forms of a) triglycine, b) tetraglycine, and
c) pentaglycine.

Fragmentation map showing the metastable decomposition pathways for
[M+H]+ of methionine enkephalin obtained from linked scans at constant

B/E and B2/E. The dotted arrows represent missing data, but the pathways
are presumed to occur based on sequence ions in same series.

a) Schematic diagram of the vacuum chamber, K+IDS power supply unit,

and picoammeter used for testing M+ thermionic emission ion currents
from aluminosilicate emitters doped with various metal substances, and b)

a plot of In+ emission ion current and emitter temperature versus heating
current for the In2032Ale3z2Si02 emitter.

a) K+IDS and b) In+IDS mass spectra of acetone (MW 58), and c) Al+IDS

mass spectrum of CCl4 (MW 152) leaked into HP ion source to 1x10'5 torr
base pressure.

a) K+IDS, b) Ga+IDS, and c) Al+IDS mass spectra of palmitic acid (MW
256), and d) Al+IDS mass spectrum of tridecanoic acid (MW 214) on the
HP mass spectrometer.

K+IDS mass spectra of a) glycerol (MW 92), b) triethanolamine (MW 149),
c) thioglycerol (MW 108), d) nitrobenzyl alcohol (MW 153), and e)
cholic acid (MW 408) mixed in glycerol.

FAB mass spectra of a) Co(NO3)3, b) Cu(NO3)3, c) Ni(NO3)3, d) K+ glass
emitter bead on wall-divided target in Figure 19c, e) Na+ glass emitter
bead on "pop-up" divided target in Figure 19f, D NaflDS-by-FAB mass
spectrum of glycerol, g) FAB mass spectrum of Li+ glass emitter bead on
K+IDS-by-FAB space-divided target, h) LifiDS-by-FAB mass spectrum
of glycerol.

3) CoflDS-by-FAB mass spectrum of glycerol and Co(NO3)2 solid

separated on a wall-divided target in Figure 1%, b) CoflDS-by-FAB mass
spectrum of H-LGG-OH (MW 245) in glycerol and Co(NO3)2 separated
on a wall—divided target, c) FAB mass spectrum of digoxin (MW 780) and

Co(NO3)2 mixed in glycerol, and d) [digoxin-H+Co]+ metastable
decomposition mass spectrum obtained by a linked scan at constant B/E.

The [M+K]+ adduct ions of H—VGVAPG-OH (MW 498) in glycerol formed

by a) KfiDS-by-FAB and c) FAB of 0.1 M KI salts in glycerol were
selected as the precursor ion to undergo metastable decomposition. In b),
the structure of H -VGVAPG-0H is shown with the prominent
fragmentation sites. The * represents a glycerol contamination peak.

The [M+K]+ adduct ions of H-VGVAPG-OH (MW 498) in glycerol formed

xvi

by a) K+IDS-by-FAB and c) FAB of 0.1 M KI salts in glycerol were
selected as the precursor ion for CAD at 50% beam attenuation. In b), the
structure of H—VGVAPG-OH is shown with the prominent fragmentation
sites.

Figure 65. Metastable decomposition mass spectra of the a) [M+H]+, b) [M+Li]+, and

c) [M+Na]+ precursor ions of the peptide H-VGVAPG-OH (MW 498) in
glycerol, 0.1 M Lil in glycerol, and 0.1 M NaI in glycerol, respectively. A
* is a glycerol contamination peak.

Figure 66. CAD mass spectra of the a) [M+H]+, b) [M+Li]+, and c) [M+Na]+ precursor
ions of the peptide H-VGVAPG-OH (MW 498) in glycerol, 0.1 M LiI in
glycerol, and 0.1 M NaI in glycerol, respectively, at 50% beam
attenuation.

Figure 67. a) Metastable decomposition mass spectrum and c) single-collision, high-

energy CAD mass spectrum of the [M+K]+ of methionine enkephalin
(MW 573) formed by FAB of 0.1 M KI salts in glycerol. The methionine
enkephalin structure and prominent fragmentation sites are shown in b).

Both [M-i-K]+ precursor peaks are on scale.

LIST OF ABBREVIATIONS

! l l v' . I l .
3N-6 ........................... Degrees of freedom in M (N is # of atoms)
AID ........................... Analog-to-digital conversion
ac ........................... Alternating current
AE ........................... Appearance energy
A ........................... Alanine
APCI ........................... Atmospheric pressure CI
API ........................... Atmospheric pressure ionization
Art ........................... Argon ion
B ........................... Magnetic sector, Magnetic field strength
BAT ........................... Best anode temperature
BET ........................... Best emitter temperature
CAD ........................... Collisionally-activated dissociation
CE ........................... Charge exchange / Capillary electrophoresis
C ........................... Cysteine
CF-FAB ........................... Continuous-flow FAB
CI ........................... Chemical ionization
Cs+ ........................... Cesium ion
D ........................... Distance, Aspartic acid
D/I ........................... Desorption/ionization
Da ........................... Dalton
dc ........................... Direct current
DCI ........................... Direct CI / Desorption CI
DEI ........................... Desorption EI
DIP ........................... Direct insertion probe
E ........................... Electric sector, Electric field strength
E ........................... Internal energy, Glutamic acid
e' ........................... Electron
E0 ........................... Activation energy
E0r ........................... Reverse activation energy
Ecom ........................... Center-of—mass energy in CAD
EE ........................... Even-electron species
EI ........................... Electron ionization
Eiab ........................... Translational energy of projectile ion
ESI ........................... Electrospray ionization
eV ........................... Electron volt
E2t ........................... Energy of the kinetic shift
F ........................... Phenylalanine
F+ ........................... Fragment ion
FAB ........................... Fast-atom bombardment
FD ........................... Field desorption

xvii

xviii

FI ........................... Field ionization

G ........................... Glycerol molecule, Glycine

GB ........................... Gas phase basicity

GC/MS ........................... Gas chromatography/mass spectrometry

Gp+ ........................... Glycerol fragment ion

H ........................... Histidine

H+ ........................... Proton

He ........................... Helium

I ........................... Isoleucine

ICR ........................... Ion Cyclotron Resonance

IE ........................... Ionization energy

k(E) ........................... Unimolecular decomposition rate function
........................... Lysine

K+ ........................... Potassium ion

K+IDS ........................... ‘ Potassium ion ionization of desorbed species

KE ........................... Kinetic energy

keV ........................... Kiloelectron volts

LC ........................... Liquid chromatography

L ........................... Leucine

LD ........................... Laser desorption

Li+ ........................... Lithium ion

LSIMS ........................... Liquid secondary-ion mass spectrometry

M ........................... Analyte molecule, Methionine

m ........................... Mass or Minute

M+ ........................... Metal cation or Molecular cation

M+‘ ........................... Molecular ion or Radical cation

M+'* ........................... Excited molecular ion

M' ........................... Radical anion

m/z ........................... Mass-to-charge ratio

MALDI ........................... Matrix-assisted laser desorption ionization

MeV ........................... Megaelectron volts

MS ........................... Mass spectrometry

MS/MS ........................... Tandem mass spectrometry

mt ........................... Mass of target gas

MW ........................... Molecular weight or Megawatt

MI ........................... M” critical configuration in transition state

N ........................... Neutral molecule in fragmentation

Na+ ........................... Sodium ion

OE ........................... Old-electron species

0 ........................... Omithine

Q ........................... Glutarnine

P(E) ........................... Probability distribution function

P ........................... Proline

P+ ........................... Product ion formed in CAD

PA ........................... Proton affinity

PD ........................... Plasma desorption

PE ........................... Potential energy

PID ........................... Photo-induced dissociation

QET ........................... Quasi-equilibrium theory

R, r ........................... Radius of E and B, respectively; Arginine

RA ........................... Relative abundance

xix

Rb+ ........................... Rubidium ion

Re ........................... Rhenium

RF ........................... Radio frequency

RI ........................... Relative intensity

RRKM ........................... Rice, Ramsperger, Kassel, Marcus theory
8 ........................... Second or degrees of freedom, 3N-6
S ........................... Serine

SID ........................... Surface—induced dissociation

SIMS ........................... Secondary-ion mass spectrometry

t ........................... Time

TIC ........................... Total ion current

T ........................... Threonine

TOF ........................... Time-of-flight

UV ........................... Ultraviolet radiation

V ........................... Volt, Accelerating voltage, Valine

W ........................... Tryptophan

v ........................... Velocity

v ........................... Vibrational frequency

MI ........................... Metastable ion

MI/CAD ........................... CAD mass spectrum containing MI
Xe+ ........................... Xenon ion

Y ........................... Tyrosine

ze ........................... Charge

[G+H]+ ........................... Protonated glycerol

[M+H]+ ........................... Protonated molecule (M)

[M+H]+* ........................... Activated/excited protonated molecule
[M+H]+* ........................... Transition state, Critical configuration
[M+K]+ .................. _. ........ Potassium adduct ion of M

[M-H]' ........................... Deprotonated molecule (M)

AGrxn ........................... Gas-phase basicity

AHf ........................... Heat of formation

Dn ........................... Thermal degradation neutral fragment

AH,“ ........................... Reaction enthalpy

CHAPTER ONE. INTRODUCTION

1. Mass Spectrometry

Mass spectrometry is an extremely powerful analytical technique that provides
molecular weight (MW) and/or structural information for nanomole quantities or less of
analyte [1]. A mass spectrometer is an analytical instrument, comprised of a sample inlet
system, ion source, mass analyzer, ion detection system, and computer, connected as
illustrated in Figure 1. The ion source, mass analyzer and ion detector regions are held
under high vacuum (<10"6 torr) in order to prevent ion scattering and additional
chemistry from ion/molecule collisions. An elaborate sample-inlet system uses a direct
insertion probe (DIP) to transfer the analyte from 760 torr into the evacuated ion source.

In the ion source, a user-selected ionization technique converts gas-phase analyte
molecules (M) into ionized intact analyte molecules, representative of the MW.

Positively-charged intact analyte molecules, M4”, [M+H]+, and [M~i-K]+ are formed by

 

 

 

 

 

 

 

 

 

 

 

 

 

Computer
Scan Signal
Sample ‘ Ion Mass Ion
Inlet ' Source ' Analyzer Detector

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Vacuum System l———

 

 

 

Figure 1. Schematic diagram of a mass spectrometer.

the energetic removal of an electron (e') from, proton (H+) attachment to, or K+
attachment to an analyte molecule (M), respectively, and negatively-charged intact
analyte ions, M' and [M-H]', are formed from the attachment of an e' or removal of a H+,
respectively. The ionization process leaves some intact analyte ions with excess
vibrational internal energy. For ions containing sufficient internal energy, subsequent
unimolecular decomposition reactions will occur to form unique fragment ion and neutral
molecule pairs, representative of analyte structure. The various types of fragment ions
formed following ionization have unimolecular decomposition mechanisms with
favorable thermodynamical properties such as activation energy, reaction enthalpy,
reaction entropy, etc. However, the abundances of these fragment ions depend on the
unimolecular decomposition kinetics for each mechanism, which depends on the internal
energy content and structural features of the ionized intact analyte molecule such as
charge-site location, gas-phase secondary structure, etc. All analyte ions formed in the
ion source volume are extracted by the ion source optics system, accelerated into a well-
defined beam, and directed towards the mass analyzer such that each ion has the same
final kinetic energy (KB).

The mass analyzer is designed to separate the ion beam that emerges from the ion
source into groups of ions according to mass-to—charge (m/z) ratios. Under computer
control, the mass analyzer is scanned so each group of ions with identical m/z values is
sequentially separated from the ion beam and focused into the ion detector (e.g., electron
multiplier, etc.) for generation of a signal proportional to the ion current. The computer
acquires the ion abundance signal for each m/z value following A/D conversion of a
voltage proportional to ion current and then generates a mass spectrum for the range of
m/z values scanned. The mass spectrum is a plot of relative abundances for each ion,
normalized to the most abundant ion, versus the respective m/z values. A certain peak at
high mass represents the stable ionized intact analyte molecules which provides the

analyte MW information. The unique "fingerprint" region in the mass spectrum, or the

peaks representing the fragment ions from the ionized analyte molecule, provide the

structural information.

2. Overview of Ionization Techniques in Mass Spectrometry

Compatibility between the analyte of interest and the available ionization
techniques on a mass spectrometer centers around how well the ionization technique
interfaces with the physical properties of the analyte and whether the ionization technique
can provide the desired MW and/or structural information. Today, a wide array of
ionization techniques is available in response to the wide variety of organic, inorganic,
biological or polymeric analytes that benefit from analysis by mass spectrometry.
Analytes can be successfully analyzed whether they are nonpolar or highly polar, low or
high MW, and gaseous, liquid or solid. A brief survey of the popular ionization
techniques is presented below.

Historically, W (El) [2] was the first successful routine ionization
technique in mass spectrometry. Analytes analyzed by El usually show extensive
fragmentation which is desirable for obtaining structural information. However, often
times a peak representing the intact ionized analyte molecules, which provides the analyte
MW information in the mass spectrum, is missing or has low abundance. This led
researchers to develop "softer" (i.e., less fragmentation) alternative ionization methods,
such as finlsLinnizntinn (F1) [3. 4]. nhntninnLinminn (CI) [5], and W
(CE) [6] to complement EI and provide the missing MW information. All of these
traditional ionization techniques require gaseous analytes or condensed-phase analyte
molecules that are highly volatile at 10'6 torr so an ample number of analyte molecules
reach the gas phase for ionization because the ionization technique does not assist in the

phase transition. In these traditional ionization techniques, ionization of the analyte to

form positive ions is accomplished by removal of an e' by a collision with a 70-eV e' in
El, exothermic transfer of a H+ from a reagent ion in CI, exothermic transfer of an e'
from a reagent ion in CB, and removal of an e' in a high electric field by tunneling in F1.
MW (APCI) [7] is essentially CI of volatile compounds at 760 torr.
Thus, volatility limits these traditional ionization techniques to the analysis of the less
polar, lower MW analytes.

The advent of W (D/I) [8] techniques, beginning with field
desorption (FD) [9] in 1969, has advanced mass spectrometry from a routine, small
molecule, organic analysis tool to a cutting-edge analytical technique for protein
sequencing in biological research. D/I techniques provide MW and/or structural
information for biomolecules and polymers that are fragile, high MW, highly
functionalized, highly polar and thermally labile. In contrast to the traditional ionization
techniques, D/I techniques provide the means for producing analyte-related ions directly
from a condensed phase sample, which eliminates the need for derivatization methods.
Historically, derivatization methods were used in E1 and CI to reduce the analyte polarity
and increase volatility, but the intact derivatized analyte ions usually fragmented
differently from the underivatized analyte ions making it difficult to study the
underivatized analyte, specifically. Although the exact ion formation mechanisms are
unknown, evidence and theory predict condensed-phase analyte molecules are first
desorbed intact into the gas-phase and, subsequently, ionized in a manner specific to the
DI] technique used. Alternatively, preionized analyte molecules may be directly
desorbed into the gas phase eliminating the need for further ionization.

mm (FD) [10, 11] was the first D/I technique amenable for the
analysis of polar, high MW, thermally labile analytes. From the end of a DIP, intact
ionized analyte molecules and a few fragment ions are generated from thermally heating
a microneedle-containing pyrolytic carbon emitter, on which a solid or liquid analyte

resides, in the presence of a very high electric field. FD is based on F1 principles,

specifically the field image theory [12]. Nonpolar analyte molecules are thermally
desorbed into the gas phase and, subsequently, near the microneedle tips, the high electric
field removes an electron by tunneling from these thermally activated molecules to form
a M“ ion. For polar molecules, the combination of thermal energy and high field
strengths induce the desorption of preionized intact analyte molecules such as [M+H]+ or
[M+Na]+ from the condensed phase [13, 14, 15]. Although FD works well to provide
MW information exclusively, the technique is difficult and non-routine because the
carbon emitter is fragile and consumable after a few experiments, exact sample placement
is critical, the "best anode temperature" (BAT) needs to be optimized for maximum intact
analyte desorption for each analyte, and the types of ions generated are transient [16].
Another early D/I technique, first introduced in 1973 by Baldwin and McLafferty
[17], was 51mg (DCI). DCI involved placing a condensed-phase analyte on a target,
attached to the end of a DIP, and inserting it into a heated EI/CI ion source volume near
the electron beam. By increasing the ion source temperature at some rate, the analyte on
the target, initially at room temperature, was rapidly heated which promoted thermal
desorption of analyte molecules into the gas phase for ionization by the traditional
techniques E1 or C1 [18]. Similar techniques, W (also DCI) [19] and
W (DEI) [20], introduced by Hunt et al. and Holland et al. in 1977,
respectively, involved rapidly heating an analyte on a pyrolytic carbon emitter to induce
thermal desorption for gas-phase ionization by C1 or E1. Daves [21] later performed DCI
by heating the analyte "dried" on a bare rhenium wire attached to a DIP, which is the
preferred DCI method used today. DCI and DEI using either an emitter or a wire have
problems similar to those in FD because the best emitter temperature (BET), similar to
the BAT, needs to be optimized for each analyte, and the analyte ions generated are
transient and last for only several seconds making accurate spectral acquisition difficult.
In addition, polar, highly functionalized analytes are thermally labile which means the

analyte is highly unstable during heating, and, as a result, neutral thermal degradation

products are desorbed into the gas phase instead of intact analyte molecules.

In the late 19708, particle bombardment D/I techniques, such as W
(PD) [22], lnnerdnsntntinn (LD) [23], and "513119" 0r"mnlnnnl_ar"§9£9m1a§tifln_mm
W (SIMS) [24], were actively pursued as alternatives to FD and the thermal
desorption techniques since the initial analyses of fragile, highly polar, high MW,
thermally labile biomolecules and polymers were successful and held much promise.
Instead of thermally heating the solid analyte as required by FD and DCI for desorbing
ions and/or molecules, the solid analyte was "dried" on a target, meaning solvents were
evaporated, and the target, attached to a DIP, was bombarded with an energetic particle
beam to induce desorption and ionization of the analyte. The respective bombardment
particles for PD, LD and "static" SIMS are MeV-fission products of radioactive
califomium—252, photons generated by a UV laser with power densities of l MW/cm2 or
greater, and low beam densities (1 nA/cmz) of keV-ions such as Ar+ or Xe+.
Unfortunately, the analyte ions were usually short-lived and transient similar to those
from FD and DCI. Bombardment of the solid analyte sample is simply another method to
energize the analyte molecules to effect ionization.

In early 1981, the advent of the DI] technique, W (FAB) [25,
26], changed mass spectrometry forever because, for the first time, these fragile, high
MW, highly polar, thermally labile biomolecule analytes could be analyzed routinely. In
FAB, on the end of a DIP, a low density beam of keV-xenon atoms impinges on a target
on which the analyte resides similar to that in "static" SIMS. However, FAB differed
from SIMS, PD, and LD because the analyte is dissolved in a polar, viscous matrix,
typically glycerol (MW 92), instead of being presented as a "dried" layer of the analyte
on the target. The matrix was the key to success in FAB because it had a relatively low
vapor pressure, and thus persisted as a solvent for a solution of analyte under vacuum.
This provided strong, long-lived, stable ion currents without degradation as opposed to

the transient signals previously obtained. However, there are drawbacks to using a matrix

because it may interfere with the analyte spectrally if a matrix peak overlaps with an
analyte peak and/or chemically if the matrix undergoes condensed-phase reactions with
the analyte during bombardment.

Two years later, W (LSIMS) [27] was introduced which combined the
"static" SIMS technique with the liquid matrix identical to that in FAB. LSIMS is
generally more sensitive than FAB, having detection limits extending down into the low
picomole region, since the keV-Cs+ beam can be highly focused at high energies which
results in higher ion yields. PD has also enjoyed renewed success since recently the
analyte is dissolved in a solution of reduced glutathion (1:1 mole ratio) and
electrosprayed on a nitrocellulose substrate (i.e., the matrix) target instead of onto a bare
thin metal target as before [28]. WW (MALDI)
combines LD with a solid matrix [29]. Here, the solid analyte/matrix is routinely
prepared by dissolving picomole amounts of analyte in a solution containing excess low
MW, highly UV-absorbing organic matrix molecules, and subsequently drying a portion
on the DIP target. With the addition of the matrix, LSIMS, PD and MALDI are capable
of desorbing and ionizing analytes having MW 5 of 10, 35, and 300 kDa, respectively.

Most recently, the ionization techniques, continuous-flow FAB (CF-FAB) [30],
and W (ESI) [31], have enabled the direct introduction of trace
amounts of analytes in their native aqueous media by flow injection, or from a liquid
chromatography (LC) or capillary electrophoresis (CE) experiment for mass spectral
analysis. In CF-FAB, the aqueous solution containing the analyte is pumped through a
capillary in the DIP to the FAB target surface which is simultaneously bombarded with
Xe atoms. This analyte-aqueous solution contains only 5% glycerol and is continuously
regenerated on the target surface, resulting in both higher signal-to-background noise and
lower detection limits. In ESI, small highly charged droplets of the analyte solution are
sprayed into a high electric field and a high pressure N2 bath gas similar to those used in

FD and APCI, respectively. As the droplets evaporate, their surface charge density

increases until the Rayleigh limit is reached, and, at this point, a coulomb explosion
occurs to produce an array of smaller charged dr0plets. Eventually, when droplets are
fully evaporated, a distribution of multiply protonated analyte molecules emerges. Mass
analysis of the array of differently charged analytes provides a database from which an
average analyte MW is calculated. ESI allows detection of analytes with MW 5 up to 100
kDa.

3. The Focus, Goals, and Organization of the Dissertation

The power of mass spectrometry for analytical applications is the ability to
identify the molecular weight and structure of the analyte under consideration through
interpretation of the respective mass spectrum. The ability of the analyst to interpret a
mass spectrum with confidence requires complete knowledge of the ion formation and
fragmentation mechanisms specific to the analyte under consideration. General
interpretation rules and fragmentation mechanisms for mass spectrometry were
successfully developed by analyzing small volatile molecules, containing no more than
two functional groups, with E1 and CI. This feat was possible because the ion formation
mechanisms for both EI and CI were well understood and these small molecules gave
very straightforward results. However, using the DH techniques, FAB and LSIMS, to
elucidate structures of polar, highly functionalized, higher MW analytes are not always
straightforward because the fragmentation mechanisms are not known with the same
degree of certainty as in E1 and CI. Unfortunately, in FAB and LSIMS, the ion formation
mechanisms are highly complex and not well understood, and the analytes commonly
analyzed, usually biomolecules, are structurally much more complex than the small.
molecules analyzed in El and CI because they have many ionizable functional groups and
three dimensional structures that are capable of folding which makes the fragmentation

mechanisms more complex to elucidate. FAB provides an abundance of very useful

structural information for biomolecules which is not readily obtainable by other analytical
or mass spectrometric techniques. Often times, the m/z and the relative intensity
information of most peaks in a typical mass spectrum cannot be fully utilized because the
fragmentation chemistry is not well understood. This dissertation will focus on the
fundamental aspects of the DH techniques, FAB and LSIMS, with the goal of trying to
better understand the ion formation and dissociation mechanisms for typical biomolecules
in an attempt to further aid the interpretation process.

In the traditional ionization techniques, EI and C1, the mechanisms by which gas-
phase analyte molecules are ionized are based on well-known gas-phase chemistry
principles such as ionization energies (IE) and proton affinities (PA), respectively. The
formation of a particular fragment ion from the ionized analyte molecule has a high
probability of occurring if the overall reaction enthalpy (Aern) for the reaction
mechanism requires little energy. More importantly, whether this particular fragment ion
is formed to a great extent depends on whether the average amount of internal energy in
the ionized analyte molecules is sufficiently high to initiate the unimolecular
decomposition processes. However, if the rate of another fragmentation reaction
mechanism is more competitive in the same range of internal energies and/or if there is
steric hindrance or interactions between two distant groups in the rate determining step
for rearrangements, then a thermodynamically favorable reaction mechanism may not
occur to a great extent. Most EI and CI fragmentation mechanisms have been elucidated
previously because the e' beam in E1 or the CI reagent ion deposits a controllable,
average amount of internal energy into the gas-phase analyte so the appearance and
disappearance of fragment ions can be controlled by adjusting the internal energies. As a
result, the changes in relative abundances of the fragment ions show the kinetic
relationships between the ionized analyte molecule, the first generation of fragment ions,
the second generation, and so on. Thermochemical data such as IE3 and appearance

energies of fragment ions can be obtained from these studies for estimating the AHrxn of

10

fragmentation mechanisms. Fragmentation mechanisms may involve rearrangements or
simple bond cleavages, and isotopic labeling can be used to differentiate between these
two processes. Isotopic labels placed in strategic locations on the analyte molecules may
or may not be shifted depending on the exact mechanism as indicated by unique shifts in
fragment ion m/z values. The standard set of fragmentation mechanisms for El and C1
are summarized in references 2 and 5, respectively. Thus, in El and CI, the fragmentation
mechanisms are well understood for most small analytes, and, as a result, unknown
analyte structures may be elucidated because these standard mechanisms represent the
links between the fragment ions in the unknown mass spectrum and the unknown analyte
structure.

In contrast to the development of El and C1, the ability to perform the particle
bombardment experiments, FAB and LSIMS, has preceded our understanding of the
underlying mechanism(s) by which the condensed-phase analyte molecules are
transformed into ionized, gas-phase analyte molecules and fragment ions. Currently,
there are many excellent theoretical studies and/or indirect studies which rely on the
condensed- and gas-phase chemistry of a select group of analytes to reveal the
mechanism(s) responsible for the production of ionized analyte molecules in FAB,
LSIMS and other D/I techniques. While much of what is known to date about the ion
formation mechanisms in FAB and LSIMS was obtained from these indirect studies, the
fact is, different analytes have different condensed- and gas-phase properties depending
on their structural features which ultimately determines the manner by which they are
ionized in FAB and LSIMS. As a result, there is an over-abundance of mechanisms and
multiple interpretations which makes the fundamental mechanism(s) that much more
difficult to identify. A better approach would be to conduct direct experimental studies to
probe for the ion formation mechanism(s). However, the particle beam energy is not
directly related to the ionization of the condensed-phase analyte molecules as in E1 or C1,

and these direct experiments need to be performed in or near the micro-volume (i.e.,

11

where gas-phase chemistry is believed to occur) above the sample probe tip, under high
vacuum, and in a high-voltage ion source. Even though direct experimentation is
extremely difficult, it has the potential to yield a plethora of new mechanistic
information.

The elucidation of fragmentation mechanisms is significantly more complicated in
FAB than in E1 because the analytes are higher MW and more highly functionalized, and
the ion formation and dissociation mechanisms are not well understood. For
biomolecules, there are more polar sites on the molecule that have the potential to be
ionized and each of these ionized sites may unimolecularly decompose with competitive
rates to give rise to an array of fragment ions. In contrast, in E1, only one or two sites are
ionized. In many cases, the exact sites of protonation in FAB are difficult to locate in
biopolymers which makes exact mechanisms more difficult to determine. Another
inability to elucidate fragmentation mechanisms with certainty, similar to that for E1, is
not knowing the average internal energy imparted into the ionized intact analyte
molecules by the FAB ion formation mechanism. This internal energy is not directly
related to the energy of the keV-particle beam as in El. In addition, some fragment ions
in the mass spectrum may result from the specific conformations of the ionized analyte
molecules in the gas phase. Determining the ionized analyte molecular
structure/confirmation in the gas phase is beyond our current analytical technologies and
is a question mass spectrometry will answer someday. While FAB expertise does exist in
some laboratories, a complete understanding of all the underlying mechanisms is still
some years away even with the research effort currently underway. Without
understanding the actual ion formation and dissociation mechanisms in a FAB D/I
experiment which link the mass spectrum to the condensed-phase analyte, it is unlikely
that ( 1) FAB and other D/I techniques will be advanced to assume new levels of
performance, (2) applications to new chemical systems or studies will occur and, (3) new

ionization techniques will evolve from these existing techniques in a timely manner.

12

Unfortunately, many D/I parameters are interlinked and conclusions are usually limited.
Chapter two will discuss FAB and LSIMS; other research related ionization
techniques such as EI, CI, and K+IDS; unimolecular decomposition theory; fundamental
mass spectrometry principles and MS/MS instrumental scanning techniques and CAD in
an attempt to bring the reader up-to-speed before discussing specific, novel research.
Both Chapter three and the published paper in Appendix one identify and evaluate the D/I
mechanisms operative in FAB using a novel, direct experimental method referred to as
KflDS-by-FAB in an attempt to better understand the ion formation mechanism in FAB.
Chapter four evaluates or elucidates the mechanisms for the formation of peptide
fragment ions under FAB and LSIMS using either available literature information,
estimated reaction enthalpies and other novel experimental methods in an attempt to
better understand the fragmentation mechanisms in FAB and LSIMS. Chapter five
describes some preliminary work that explores the metal ion/biomolecule reactions in the

gas phase with the ionization techniques, M+IDS and M+IDS-by-FAB.

CHAPTER TWO. THEORY AND METHODS

I. The Ion Formation and Dissociation Processes in the Traditional
Ionization Techniques Electron Ionization (E1) and Chemical Ionization (CI)

The theories and mechanisms surrounding ion formation and ion dissociation are
well-characterized experimentally for the ionization techniques, EI and CI, as will be
described below. This knowledge base for E1 and CI may serve as a starting point for
understanding the theories and mechanisms for ion formation and ion dissociation in

FAB and the other related D/I techniques.

ELECTRON IONIZA TION (El)

Historically, E1 was first introduced by Dempster in 1921 [32] and it evolved into
the first routine ionization technique in mass spectrometry. Improvements were made to
Dempster's design in 1929 by Bleakly [33] who collimated the e' beam with a parallel
external magnetic field, and in 1940 by Nier [34, 35] who placed a field—free region
between the El ion source and the mass analyzer to eliminate any influences from the
magnetic and electric fields on production and acceleration of ions. Before the 19505, EI
mass spectrometry (El-MS) was primarily used for accurately determining the elemental
isotopic masses and abundances [36]. Then, beginning in the 19505, the power of EI-MS
for the analysis of small organic molecules was first realized [37] because, in addition to.
molecular weight information, researchers noted that EI-MS also produced an unique
"fingerprint" of abundant fragment ions which provided specific structural information
for each analyte structure. Over the years, most fragmentation mechanisms have been

elucidated for EI-MS so unknown ionized analyte structures can be derived from the

13

14

various types of fragment ions present in the unknown mass spectrum. Presently, E1 is a
routine, well characterized ionization technique that gives reproducible, well understood
spectra, and is primarily utilized as the ionization technique in the mass spectrometer

detector of a gas chromatograph (GC/MS).

EI Instrumentation.

A schematic for an El ion source is shown in Figure 2. In El, a monoenergetic
beam of electrons (e') is the ionizing reagent for analyte molecules. Gas-phase electrons
are generated by thermionic emission from an incandescent rhenium filament. The gas-
phase electrons are accelerated into a well-defined beam with a final energy of 70 eV
since the potential difference between the rhenium filament held at 9930 V, for example,
and ion source volume held at 10000 V is 70 V. The beam of electrons traveling across
the ion source is well collimated owing to an external magnetic field (not shown in Figure
2) parallel to the electron beam which forces the electrons to travel in small helices about
the field. The electron trap is held at a higher potential than the ion source volume in
order to collect the remaining electron beam so electronic feedback can be provided to the

filament power supply to ensure that the electron emission current is kept constant. The

’
\

2'!
£5.
E’,

Repeller

 

1

G'OOOGOOOOOO'GO
+
I
H
O
D
3
fl.
0
VJ

- 'N1.
+

ZZ
3
lo

.31

 

ZZz
Z ZZZ

  

Z

 

e'
Ion Source Y
Volume Electron Trap

Figure 2. Schematic diagram of an El ion source.

15

repeller in the rear of the ion source "pushes" any newly formed ions in the ion volume
towards the front of the ion source for extraction and acceleration by the ion optics.

These ions are subsequently analyzed by the mass analyzer.

Analyte Volatility in El.

In El, either a gaseous analyte is leaked into the ion source volume or condensed-
phase analyte molecules are thermally desorbed by heating a sample holder on the end of
a direct insertion probe. Thus, volatility limits the analytes amenable for analysis by E1
to low molecular weight and low polarity (i.e., frequently less than two polar functional
groups). For a successful analysis by El, the analyte pressure needs to be at least lxlO‘5
torr. If it is greater, then ion/molecule chemistry readily occurs. Unfortunately, it is well
known that less than 1% of the total analyte is ionized while the remaining 99%+ is
pumped away by the vacuum system. The current used to heat the rhenium filament and
ion source heaters (not shown in Figure 2) raise the ion source temperature to
approximately 250° C which prevents the analyte molecules from condensing on the ion

source walls.

Formation of Ions in E1.

Ionization of a gas-phase analyte molecule (M) can occur if a 70-eV 6 passes
sufficiently close to M and upsets its electronic equilibrium. The activated M* may
either eject an e‘ to rid energy equal to its ionization energy (IE) or remain in an excited
neutral state. As a result, a positively-charged analyte radical cation, M+', also known as
the molecular ion, is formed initially by the ionization process as illustrated in Equation 1

M (g) + e' (70 eV) —-) M'” -i- 2 e' (1)
and Figure 2. For electron ionization to occur, the energy of the colliding electron must
be greater than the IE of the gas-phase M. The amount of energy imparted into M is

dependent on the closeness of the interaction between 6 and M and the structural features

16

of M. If the M” has low internal energy after ionization, it will probably be detected
intact and molecular weight information will be obtained. If M“ contains excess internal
energy after ionization, it has a high probability of undergoing unimolecular ion
decomposition processes [38] to form a fragment ion and neutral radical pair as shown in
Equation 2 and Figure 2. The fragment ion, Fn+, can undergo further unimolecular
M+‘* —> E,“ + Nno (2)

decomposition to form a secondary fragment ion if ample internal energy remains. Since
these fragmentation mechanisms in El are well characterized, many Fn+ reveal structural

features of M‘”.

Utility of Clastograrns in E1.
The clastogram (or breakdown curve) in Figure 3 is a plot of the relative

abundances of M‘” and the Fn+ ions for methanol versus energy of the e' beam. At 70

 

 

 

 

 

 

 

1007
r: 90f
[:1 30%
'3 70..
U
5
.5 60.,
fl
3 50..
<1
.5. 40“
E 30--
% 20L
a:

lO--

0 j

0

 

Electmn Beam Energy (eV)

Figure 3. An EI clastogram for methanol (Adapted from [39]).

17

eV, the rates of unimolecular decomposition are independent of e' energy because only a
small fraction of the 70 eV can be deposited into M. Since the relative abundances (RA)
of ions formed in E1 are highly reproducible and uniform at 70 eV, EI mass spectra are
obtained at this e‘ energy. The efficiency of the e' collision and the structural features of
M determine how much energy is absorbed, which is summarized by the probability
distribution function, P(E). For a 10-kV ion source, if the voltage potential on the
filament is increased from 9930 V to 9980 V, then the e' beam energy will be decreased
from 70 eV to 20 eV. At 20 eV, the rates of ion formation are highly dependent on the
energy of e‘ beam, which is not the case at 70 eV. The IE of methanol is approximately
11 eV so the CH3OH+', or M‘”, curve starts at 11 eV and 100% RA in the clastogram.
Therefore, at low e' energies near the IE, M“ will be the most intense ion in the mass
spectrum which is useful for molecular ion identification purposes. As the e' energy is
increased, the RA of M‘” decreases while the RA of the CH20H+ fragment ion increases
because, at higher e' energies, more M“ ions have more excess internal energy after
ionization which results in a higher probability of unimolecular decomposition occurring.
The shapes of the curves in Figure 3 indicate that the formation of CO+ and CHO+
fragment ions are secondary processes which originate from CH20H+ precursor. Thus, a
clastogram is useful for fragmentation studies in B] because it "maps" out the kinetics of
the fragmentation events, contained in the RAs, over an entire range of average internal
energies imparted into M+' by the e' beam energy. This aids in identifying the precursor
ions of each fragment ion. ii"

A clastogram is also useful for identifying the thermodynamics and possibly the
mechanism of each fragmentation process. In Figure 3, the appearance energy (AE) of
CH2=OH+ fragment ion is approximately 12 eV. The AB is equal to the sum of the IE of
CH3OH and the activation energy, E0, of the fragmentation pathway. At the e“ beam
energy of 12 eV in Figure 3, a certain population of M” ions have approximately 1 eV of

excess internal energy since ionization of CH30H requires 11 eV. As a plausible

18

mechanism, CH3OH+' may decomposes into CH2=OH+ if a hydrogen loss occurs by a

radical-initiated alpha-cleavage as shown in Equation 3. The estimated reaction enthalpy

I[I/\ [1+ 18 kcal/mol

+
CH2—OH -——> CH2=OH + H-

3
202 kcal/mol 168 kcal/mol 52 kcal/mol ( )

(AHrm) for the unimolecular decomposition mechanism of CH3OH+' shown in Equation
3 is 18 kcal/mol or 0.8 eV, which corresponds well to the experimental AB in Figure 3.
The heats of formation (AHf) in Equation 3 for small ions and neutral radicals are

available and listed in reference 40.

Ionization Theory for M in E].

A physical description of the ionization and fragmentation processes in E1 is
provided by the quasi-equilibrium theory (QET), developed by Rosenstock et al. [41] in
1952. In El, if M* chooses, an e' is removed in approximately 10'16 s from the excited M
in its ground-state configuration to form M+'. If internal energy remains in M'” after
losing e‘, M""* subsequently undergoes a vertical Franck-Condon transition to an excited
state, perhaps with electronic internal energy exceeding 10 eV. The electronic transition
is vertical because no change in the ground state molecular geometry occurs during the
excitation process. In an attempt to restore electronic equilibrium in M‘“, this electronic
internal energy in M+'* undergoes fast radiationless transitions between closely situated
electronic potential energy surfaces or vibrational levels and eventually becomes
randomized intramolecularly into vibrational energy among the active 3N-6 degrees of
freedom (N is the number of atoms in M, and the active degrees of freedom are 1/2 to 1/3'
of the 3N-6 modes which contain randomized energy). This randomization of internal
energy in M“ occurs for a short period of time before unimolecular ion decomposition

processes begin because energy redistribution is faster than fragmentation. In the ion

19

source volume, the fastest unimolecular decomposition processes, the simple bond
cleavages, begin after 10'13 s and the slowest processes, the cleavages involving a
rearrangement process, end around 10'6 s. The low internal energy M+' ions not
decomposed in 10‘6 s are extracted and accelerated. After acceleration, some low energy
M‘” ions may undergo metastable unimolecular decomposition [42] during their flight to

the detector, but the lowest internal energy M+° ions will be detected intact.

Unimolecular Ion Decomposition Theory.

Fragment ions are formed by unimolecular decomposition kinetics because it is
highly improbable that M“ will undergo a bimolecular collision at 1x10'6 torr. The QET
and RRKM [43, 44] theories are utilized to describe the rate of unimolecular ion
decomposition of the excited M+‘* into fragment ions. In Figure 4, if the internal energy,

E, remaining after ionization of M, exceeds the activation energy, E0, for a certain

if“ ‘

 

 

 

 

Kinetic Shift, Bi
_ .1211. _
AE E l r
130
F,{*‘
E V 7 -" — "
g 150
id
3 Aern
‘5
3
e W“ __y
1 IE (M) *
. M —> M+'
—— M (Ground state molecule)

 

 

Reaction Coordinate

Figure 4. Thermochemical considerations for unimolecular decomposition of M+'*.

20

reaction pathway, then there is a high probability that M+°* will move through the
transition state or critical configuration, Mai, to form the fragment ion, Fn+, and neutral
radical, NO, as shown in Equation 4. The critical configuration, Mni, represents the point

M‘m‘ —-) Mni —) Fn+ + My (4)
in time when enough vibrational energy accumulates in a specific bond or group of bonds
to induce either simple bond cleavage(s) or intramolecular secondary interactions and
subsequent bond cleavages in M+'. The rate of unimolecular decomposition for M+'*
increases as the internal energy, E, exceeds the activation energy, E0, and this difference
is defined as the kinetic shift, E1. Thus, the decomposition rate is a function of E and it
can be approximated by the expression in Equation 5

k(E) = I) ((E - 150)/E)~°"l (5)
where u is the vibrational frequency factor and s is the number of oscillators or active
vibrational degrees of freedom. From Equation 5, as E approaches E0, the rate slows to
zero while as E approaches infinity, the rate advances to a maximum rate of u. The
vibrational frequency factor of the bond is the maximum rate for bond cleavage. Since ‘0
for a simple cleavage, 10'14 s", is greater than t) for a rearrangement, 10-10 5'1, simple
cleavages will always dominate over a rearrangement if both pathways have identical
activation energies. If the number of oscillators, s, is increased, the internal energy will
likely be distributed over more bonds, which will result in slower decomposition rates
since the probability of ample vibrational energy localizing at a particular bond(s) is
reduced.

It is important to realize that the fragmentation pathway with the lowest reaction
enthalpy (Aern) does not necessarily produce the most abundant ions. The kinetics
often dictate which processes dominate. McLafferty [2] describes the fragment ions that.
are formed in El as being either simple cleavage mechanisms as shown in Equation 6 or
rearrangement mechanisms in Equation 7. The simple cleavage mechanisms have a

"loose" transition state in Equation 6, because the ionized sigma bond undergoes a sigma-

21

47 kcal/mol
(26V)
CH3CH2‘1'CH2CH20CH3 —> CH3CH2+ + °CH2CH20CH3 (6)
158 kcal/mol 215 kcal/mol -10 kcal/mol

H
/ (Lt/CH3 15 kcal/mol

Hcha I)” (0.7 eV)
|\,« —> 0CH2CH2CH2CH2+ + CH3OH (7)
H2O CH2
C 2 158 kcal/mol 221kcal/mol -48 kcal/mol

bond dissociation without regard to critical steric configuration when enough vibrational
energy critically displaces the bond length and angle. However, rearrangement reactions
in Equation 7 have a "tight" transition state because a critical steric configuration in the
transition state is required to bring H and +'OCH3 together intramolecularly to form the
H-OCH3 bond in step (a). The second step (b) of the reaction involves a simple inductive
cleavage to free CH3OH, and possibly a third step (c) would involve formation of a ionic
four-membered ring (not shown). The overall reaction enthalpy is low for a
rearrangement reaction because, for example, in Equation 7, the energy required for the
cleavage of the C-0 bond is recovered when a new H-O bond(s) is formed. Ironically, in
Equation 7, the reaction enthalpy for forming the ionic four-membered fragment is 15
kcal/mol higher than for the linear fragment probably because the cyclic fragment is
highly strained and not as stable. For simple cleavage reactions, the reaction enthalpy is
usually high because one bond is cleaved and none are formed. The rates of
rearrangement reactions are slower than those for simple cleavage reactions because the
rate-dependent step is the through space interaction of H and +"OCH3 where simple
cleavages just require energy deposition into the bond. Therefore, if a simple cleavage
and a rearrangement reaction had identical activation energies, the rate of the simple
cleavage would be faster because the vibrational frequency of a single bond is higher than

an intramolecular interaction. Thus, simple cleavages are dependent on the average

22

internal energy in M“ while rearrangement reactions are very dependent on the
secondary interaction rate and the steric structure of M” in the transition state, provided
the thermodynamics are reasonable for both.

In Figure 5, an estimated Wahrhaftig diagram for ionized CH3CH2CH2CH20CH3
shows the relationship between the probability distribution function, P(E), in the top plot,
and the kinetic rate constants, k(E), in the bottom plot. The P(E) describes the
probabilities of M” ions having particular internal energies, B, after ionization and/or
activation and the k(E) describes the decomposition rates of M“ as a function of E and
structure for a particular fragmentation pathway. Both the P(E) and k(E) are plotted
against E, the internal energy in the analyte after ionization with 70 eV electrons. The
P(E) was approximated from distribution of abundances for the ionic species M'”,

metastable fragment ions, and the fragment ions formed from unimolecular

 

 

 

Metastable
M4" ions
P(E)
M-P
0 C2) @ G) @
13~ I I I I 97
12~ l I | l
11- l I I O
€13: : j @l / | M“ = CH3CH2CH2CH26CH3
“ a ll 1 .. +
7" I I I i®®M —> °CH2CH2CH2CH2 + CH3OH
+
6""'—'9 I : : @M‘” —> CH2=OCH2 + CH3CH2CH2°
:jb- : : l : @M'” —>CH3CH2+ + °CH2CH20CH3
3 I I II I I I I I 7 I
o 1 2 3 4

Internal Energy, E, after Ionization (eV)

Figure 5. Estimated Wahrhaftig diagram for ionized CH3CH2CH2CH20CH3.

23

decomposition of M“ in the ion source, and it is usually based on the initial abundances
of analyte species in case secondary processes would be involved. All ions formed in the
ion source are extracted and accelerated to the mass analyzer within 10‘6 s, regardless.
Thus, for fragmentation to occur in the ion source, M” must have enough excess internal
energy to initiate a fragmentation process having a rate greater than 106 8'1. For this
specific Wahrhaftig diagram in Figure 5, the M‘" ions with an internal energy, E, less
than 0.7 eV are extracted from the ion source and detected intact without fragmenting
because all the fragmentation reactions require more internal energy to occur at an
appreciable rate. In region Q), the P(E) indicates that not many M‘” ions acquire an
energy less than 0.7 eV because the M” ion abundance is low. In region @, an M+' that
acquires between 0.7 eV and 0.9 eV will undergo metastable decomposition after
acceleration because the k(E) values for forming metastable fragment ions are less than
106 8’1 but greater than 104-5 8'1, resulting in fragmentation rates too slow to occur in the
ion source. The rearrangement fragmentation pathway, M+’—-> ~CH2CH2CH2CH2+, in
Equation 7, has an low E0 of 0.7 eV (the E05 are the E values at the start of each
decomposition rate curve on the x axis in Figure 5) in the metastable region, but it has an
estimated AE of 0.9 eV [45] for ion source fragmentation (the AEs are represented by the
dots on each decomposition rate curve in Figure 5). If M” ions have between 0.9 and 1.7
eV of B, then the M“—-) °CH2CH2CH2CH2+ reaction will kinetically dominate in region
Q) and undergo ion source decomposition. If M“ has an E between 0.7 and 0.9 eV, M“
will undergo the same fragmentation reaction, but as a metastable ion at a slower rate in
region ® outside the ion source after acceleration. The simple cleavage process,
M+‘-)CH3CH2+, in Equation 6, has an E0 of 2 eV and an estimated AE of 3 eV [45] for
competitive ion source decomposition. If M“ ions have 3 eV of E and up, M” will
unimolecularly decompose at a high rate in the ion source to form CH3CH2+ fragment
ions which is dominant fragmentation process in region 6). If M“ had an E of 2 eV, it

would likely unimolecularly decompose to form CH2=OCH3+ fragment ions as shown in

24

region @ since it is the most kinetically favored pathway in this region. A fragmentation
pathway with a low E0 will cause the M” abundance to be low or missing as will a Fn+
ion with secondary fragmentation process having a low E0; both of which are not the
cases here in Figure 5. Thus, all fragmentation pathways here originate from an ionized
site on CH3CH2CH2CH20CH 3 and are kinetically competitive as shown in the
Wahrhaftig diagram depending on E. If multiple sites on M“ are ionized, fragmentation
becomes kinetically competitive between ionization sites in addition to the competition
between local decomposition reactions at each ionized site. The M“ ion lifetimes are
dependent on E and equivalent to the inverse of k(E) values for a particular value of E.
For ion source decomposition, the simple cleavage reaction dominates from 10'13 s to 10'
3-5 3 while the rearrangement reaction are slower reactions that dominate from 103-5 s to
10'6 5. Outside the ion source, metastable M“ ions decompose between 10‘6 and 104-5 5
while intact M4” ions are stable for at least 104-5 s or longer. Thus, M+‘ ions that
undergo metastable transitions usually have a narrow distribution of low internal energies
and follow slow rearrangement reactions having low Eos. However, if both a simple
cleavage and rearrangement reaction have identical Eos in the metastable regime, then
simple cleavage will dominate over the rearrangement processes in the metastable
transition, as explained above.

The above explanations give a simple, brief overview of the physical processes
occurring during ionization and unimolecular decomposition in E1, but the events that do
occur are more complicated than portrayed above and important questions still remain as
explained by Lorquet [46]. The unimolecular ion decomposition theory explained above
can be utilized also to account for the competitive fragmentation processes following

ionization of analytes in other ionization techniques.

25

CHEMICAL IONIZA TION (CI)

Ion/molecule reaction products were often observed and considered a nuisance in
the early days of mass spectrometry because the vacuum systems were primitive and the
pressures were usually above 1x10“4 torr. Chemical ionization (CI) is an outgrowth of
fundamental gas-phase ion/molecule studies of protonated methane, CH5+, during the late
19505 and early 19605. In 1966, Munson and Field [47] introduced CI as a "soft"
ionization technique for mass spectrometry which means CI provided abundant molecular
weight and limited structural information for most analytes in contrast to El; that is, there
is little fragmentation in CI as opposed to E1. CI has become popular because it is
routinely used to complement the analyses in which EI fails to provide molecular weight
information for certain analytes. CI is presented here because the ion formation
mechanisms for D/I techniques indicate that chemical ionization processes produce the

majority of the observed protonated analytes.

CI Instrumentation.

CI is performed in a heated, EI ion source identical to Figure 2 except the ion
volume is filled with a reagent gas at medium pressure (i.e., between 0.2 and 2 torr) and
the large slits for the electron beam and the ion exit slit are greatly reduced in size to
prevent the reagent gas from leaking into the surrounding ion source housing chamber
held at low pressure. Differential pumping is needed in CI to keep the mass analyzer
pressure low at all times and to keep the ion source housing chamber as low as possible to
prevent additional ion/molecule chemistry outside of the ion source volume. Also, in CI,
the electron beam is accelerated to energies greater than 200 eV and directed into the
reagent gas, held at a medium pressure, in the ion volume as in E1. However, even at
higher 6 energies, the medium pressure still prevents the electrons from reaching the

electron trap so electronic feedback to regulate the filament current is obtained from the

26

electrons that collide with the ion source volume wall upon entrance. Another reason the
openings in the ion source volume are reduced in size is to give the newly formed reagent
ions longer residence times in the ion source volume to ensure more complete analyte

protonation before their extraction.

[Mi-HP Formation in CI.
If the reagent gas, methane, is present in excess at medium pressures, when a 200-

eV electron collides with a methane gas molecule, as shown in Equation 8, there is a high

CH4 (3) + e- (200 eV) —) CH4“ + 2 6' (8)
CH4” + CH4 (g) —> CH5+ + CH30 (9)
CH5+ + M (g) —> [M+H]+ + CH4 (g) (10)

probability that the radical cation, CH4“, will be formed by the E1 process. The CI
methane reagent ion, CH5+, is formed by an ion/molecule proton transfer reaction from a

bimolecular collision between CH4“ and another gas-phase methane molecule, present in

excess as shown in Equation 9 above. The analyte, M, either needs to be volatile at
medium pressures or thermally desorbed by DCI techniques because CI, like El, requires

M to be in the gas-phase for protonation. If possible, the analyte pressure in the ion

volume should be at least lxlO‘4 torr. If the proton affinity of M is greater than CH4,

then an exothermic proton transfer from CH5+ to M will occur to form the protonated

analyte molecule, [M+H]+, when the two species undergo a bimolecular collision as in
Equation 10 above. The proton affinity (PA) is defined as the exothermicity (i.e, PA = -A
Hun) of the proton attachment to a particular analyte, M, as shown in Equation 11. In

M (g) + H+ —> [M+H]+ PA = -A ern (11)
Equation 10, for the collision of protonated methane and M, the reaction enthalpy, A
Hm], of the proton transfer is governed by the difference in PAs of the two species (i.e., A

ern = PA(CH4)-PA(M)). It is important to realize that under the medium pressure

conditions described above, the reactions in Equations 8-10 are occurring simultaneously

27

and they represent the primary pathway for the formation of [M+H]+, although secondary
pathways exist. Finally, good yields of [M+H]+ ions are dependent on fast reaction rates
which are directly proportional to high collision rates between the reagent ions and M
brought about by the medium pressure CI environment. A table of PAs and IE5 for small

molecules with various functional groups is presented in Table 1.

Fragmentation of [M+H 1+ in CI.

The extent of fragmentation in CI depends most on the amount of internal energy
in [M+H]+, induced by protonation, and also the types of fragmentation pathways
available surrounding the protonation site. Protonation of a gas-phase M usually occurs
at sites which have the highest PAs, and the highest PAs indicate the most basic sites in
M. For biomolecules, a basic site may either be a single functional group containing a
heteroatom(s) with non-bonding electrons or a collection of two or more heteroatoms
from different functional groups. Protonation may occur at a single heteroatom basic site
or the proton may coordinate with two or more heteroatom sites which increases the
stability of the [M+H]+. However, a proton that dicoordinates with two heteroatoms sites
essentially undergoes cyclization which reduces the basicity of the site (i.e., Aern)
because a negative change in entropy occurs. The exothermicity of the proton transfer
reaction determines the exact amount of internal energy imparted in the [M+H]+. If the
internal energy is localized in M-H+ bond and/or surrounding bonds, the fragmentation
may be localized at the protonation site. If the PAs of the two reactants are similar, then
the internal energy may be less than 1 eV (23 kcal/mol) which results in abundant
[M+H]+ ions for molecular weight information and some low-energy fragmentation such
as neutral losses. In CI, most often, stable even-electron (EE) fragment ions are formed
from the low-energy elimination of stable EE molecules from the EB protonated molecule
although exceptions do exist since odd-electron (OE) species are sometimes formed

depending on the analyte. If structural and molecular weight information is desired, then

28

 

Table 1. PAs and IE5 for small molecules with unique functional groups. All values
were obtained from Lias et al. [40] except when noted.

 

Functional Group Compound PA (kcal/mol) IE (eV)
fiydrogen H2 151.2 15.4259
alkane CH4 131.6 10.85
alkane C3Hg 150.0 10.95
branched alkane CH3CH(CH3)CH3 163.3 10.57
water H20 166.5 12.612
alkyl chloride CH3CH2C1 169.0 10.97
alkyl bromide CH3CH2Br ~ 171.0 10.28
alkyl iodide CH3CH21 176.0 9.346
alkene CH3CH=CH2 179.5 9.73
benzene C6H6 181.3 9.2459
aldehyde CH3CH2C(O)H 189.6 9.953
toluene C6H5CH3 189.8 8.82
l°-alcohol C3H7OH 190.8 10.22
thiol CH3CH2CHZSH 191.6 9.195
carboxylic acid CH3CH2C(O)OH 191.8 10.525
nitrile CH3CH2CN 192.6 1 1.84
NBA N02C6H4CHZOH 194.0a
phenol C5H50H 196.3 8.47
ether CH3CH20CH3 196.4 9.72
ketone CH3C(O)CH3 196.7 9.705
ester CH3CH2C(O)OCH3 200.2 10.15
thioether CH3SCH2CH3 203.5 8.54
ammonia NH3 204.0 10.16
l°—amide CH3C(O)NH2 206.2 9.65
glycerol HOCH2CH(OH)CH20H 209.0a
aniline C6H5NH2 209.5 7.720
glycine NH2CH2C(O)OH 21 1.6 8.80
2°-amide CH3C(O)NHCH3 212.7 9.3
alanine NHzCH(CH3)C(O)OH 214.8 8.88
phenylalanine NHzCH(CH2C6H5)C(O)OH 216.5 8.4
l°-amine CH3CH2NH2 217.0 8.86
valine NH2CH(CH(CH3)2)C(O)OH 217.0 8.53
dipeptide CH3C(O)NHCH2C(O)OCH3 217.7
leucine NH2CH(CH2CH(CH3)2)C(O)OH 218. 1 8.5 l
proline NHC4H7(C(O)OH) 220.2 8.3
2°-amine CH3NH2CH3 220.6 8.23
pyridine C5H5N 220.8 9.25
methionine NH2CH(CH2CH2SCH3)C(O)OH 22 l .4 8.3
tyrosine NH2CH(CH2C6H40H)C(O)OH 222.3 8.0
dipeptide CH3C(O)NHCH(CH3)C(O)OCH3 224.5 9.2
piperidine C6H10NH 226.4 8.05
lycine NH2CH(CH2CH2CH2CH2NH2)C(O)OH 230.3 8.6
histidine NH2CH(CH2C3H3N2)C(O)OH 231.9
triethanol amine N(CH2CH20H)3 233.0a
gimme NHZCH(CH2CH2CH2NHC(=NH)NH2)C(O)OH 245.2b

 

3‘ Obtained from reference [48].
'3 Obtained from reference [49].

29

the internal energy imparted into the [M+H]+ needs to be greater than 1 eV. Extensive
fragmentation equivalent to or greater than EI can be achieved in CI if the reagent gases
such as hydrogen or methane, which have PAs of 101.2 and 131.6 kcal/mol, respectively,
are utilized. For example, protonated peptides, having PAs equal to 220 kcal/mol, would
have as much as 5.2 and 3.8 eV of internal energy if hydrogen and methane were used,
respectively, which is necessary for abundant fragment ion formation. Thus, in C1, the
extent of fragmentation from the protonated molecule can be controlled by the PA of the
reagent gas. For protonated analytes with high internal energy and two or more basic
functional groups having the potential for protonation, the fragmentation processes
observed are not well rationalized because the two fragmentation pathways are in direct
competition as are the local fragmentation pathways. Thus, the QET theory, developed to
rationalize the appearance of El spectra, may be employed in CI to explain the observed
abundances of fragment ions assuming protonation activates M and the vibrational
internal energy in [M+H]+ ions eventually concentrates in certain bond(s) causing
intramolecular rearrangements and/or bond dissociation(s) [50]. In medium pressure C1,
the rates of ion source unimolecular decomposition are faster than the rate for collisional
de-excitation of the activated [M+H]+ by the excess reagent gas, and the fragment ions

observed are formed in 10'7 s or less.

2. Potassium Ion Ionization of Desorbed Species (K+IDS)

Potassium ion ionization of desorbed species (K+IDS) [51] is a novel D/I
technique developed in the Allison research group 10 years ago that complements the
current D/I techniques. K+IDS is essentially a chemical ionization experiment using K+
as the reagent ions because the K+ ions (cat)ionize both intact analyte molecules (M) and
neutral thermal degradation (Dn) products that are thermally desorbed into the gas phase,

forming [M+K]+ and [Dn-i-K]+ adduct ions, respectively. The [M+K]+ and [Dn+K]+

30

adduct ions conveniently yield molecular weight and structural information, respectively.
K+IDS illustrates the basic, well-understood principles of desorption and ionization.
K+IDS is similar to DEI, DCI and FD in that analyte molecules are thermally desorbed
and subsequently ionized by K+ in the gas phase. The K+ adduct ion signals are still
transient in K+IDS as in other thermal D/I techniques and very thermally labile analytes
may still undergo facile degradation preventing intact desorption. In contrast to current
D/I techniques, FAB, LSIMS, MALDI, and E81, K+IDS does not require a matrix, so
matrix effects and interferences are avoided. K+IDS is a novel mechanistic tool for
studying the thermal decomposition of analytes because K‘t simply adds to the desorbed
intact neutral M and neutral degradation products, and does not induce any new
fragmentation. Thus, K+IDS allows the condensed-phase, thermal decomposition
mechanisms of thermally-activated analyte molecules to be elucidated which
complements the gas-phase, charge—induced unimolecular decomposition of intact,
activated analyte ions in the other thermal and particle beam D/I techniques. K+IDS is
introduced here because it was used in the FAB experiment to investigate the ion

formation mechanisms and was a starting point for performing M+IDS experiments.

K+IDS Instrumentation.

The K+IDS experiment is performed on the end of a modified direct insertion
probe (DIP) in the ion volume of an El ion source as shown in Figure 6. The K+ glass
emitter bead is formed by melting a small amount of a refractory mixture that contains
high purity potassium nitrate, alumina and silica in a 2:1:2 mole ratio, onto a rhenium
wire loop at very high temperatures. It has a final molar composition of
leO: lAle3z28102 after firing since N02 gas is released [52]. In a K+IDS experiment;
the K+ glass emitter bead is rapidly heated to 1000° C or more by passing 2 to 3 amperes
of current through the underlying Re wire. High voltage wire feed-throughs in the DIP

connect the K+ glass emitter to an external power supply outside the instrument. Once

31

 

 

 

 

  
    

 

  
  

 

 

 

 

    

 
  
 
 
 

 

 

      
 
    

Repel] er Ion Source Volume
+
K+ 16:4- MD2 M M +
K+ D D2131 [Mm] —>M Anal
’ ' " + ass yzer
K+ Glass / [Dn+K]
Emitter
(1000° C) Sample holder
(0—> 300° C)
.... ..I.'..-'.‘ . - . . . . . - . - - ' Ion Optics
Figure 6. Schematic diagram of the K+IDS experiment.
Desorption Favored l Degradation Favored
|
3 I
r:
a
F |
i:
:1
'° |
<
.5
O
> I
'5'
.9.
9 |
a:
l
I

 

 

High T l/T (K) Low T

Figure 7. An Arrhenius plot that demonstrates the competition between desorption and
degradation processes in thermal heating.

32

the K+ glass emitter is at high temperatures, low-energy K+ ions are thermionically
emitted into the gas phase [53]. The heat generated from the production of K+ radiatively
heats the analyte "dried" on the adjacent Re wire loop sample holder to approximately
300° C which induces thermal desorption of intact M and/or thermal degradation and
desorption of Du neutral products. A rapid heating rate will generate abundant [M+K]+
ions, while a slow heating rate will provide more abundant [Dn+K]+ ions. The two-
filament DIP design [54] in Figure 6 is based on a similar DIP developed by Rollgen et
al. [55]. This two-filament DIP performs better than the original method of placing the
analyte directly on the K+ glass emitter [56] because it delays desorption of the analyte
until abundant K+ ions are present in the gas phase. The K+ glass emitter potential can be
biased equal to or above the E1 repeller potential so the energy of the K+ ions can be
varied. K+IDS is best interfaced with a quadrupole mass analyzer because all the analyte
is desorbed and ionized within 10 s after rapid heating is initiated. This means an
extremely fast scan rate (i.e., 2 scans per second, full mass range) is required to capture as
many mass spectra of the transient ion current as possible for an accurate reconstruction
of the analyte desorption profile. Also, K+IDS has the potential to form K+ adduct ions
with a range of KEs; fortunately, the quadrupole mass analyzer will mass filter these ions

despite the wide KE range.

Thermal Desorption and Degradath of M.

Desorption of an analyte molecule from a surface occurs when all the non-
covalent, intermolecular bonds between the molecule and the surface and/or neighboring
molecules are broken. Because a highly polar biomolecule such as a peptide has many
functional groups, there is a high probability that more non-covalent, intermolecular
bonds will have to be broken for desorption of these intact species as opposed to the case
for nonpolar molecules. Thus, for highly polar peptides, substantial thermal energy,

equal to the heat of vaporization, needs to be deposited into the 3N-6 degrees of freedom

33

to induce non-covalent, intermolecular bond breaking. In K+IDS, the rapid heating
concept, first introduced by Beuhler et al. [57], is utilized to make the rate of thermal
desorption for intact neutral analyte molecules competitive with the rate of decomposition
so both molecular and structural information can be obtained. In Figure 7, desorption is
kinetically favored at higher temperatures while degradation is favored at lower
temperatures assuming the relative ion abundances are proportional to the k(E)s.
Therefore, if the analyte is heated at a high rate, less time is spent at lower temperatures,
which induce degradation, so more analyte molecules will survive and can desorb intact
at higher temperatures. If the heating rate is low, then more neutral degradation products
will be formed because the temperature will be below the crossover point for a longer
period of time. Quite often, thermal decomposition pathways for a biomolecule will lie at
a lower activation energy than thermal desorption so neutral decomposition products (Dn)
will be formed by a intramolecular bond cleavage(s) instead, owing to the thermal
liability of biomolecules. Thermal decomposition of a biomolecule results in the
formation of two stable molecular fragments. For peptides, usually one molecular
fragment is a six-membered cyclic product that contains the two residues from the N-
terrninal side and the other is a linear peptide product that contains the remaining residues
from the C-terminal side, assuming a central amide bond was cleaved. Studies of peptide
deuterium-labeled peptides indicated that 1, 2-elimination mechanisms do not occur and
that cyclization mechanisms dominate. For K+IDS analyses of peptides containing only
neutral amino acid residues, the cyclization mechanism with the lowest reaction barrier

dominates the thermal decomposition processes [58].

K4“ Ionization of Gas-Phase M and D".
The thermal desorption of many neutral molecules in the presence of K+ ions
creates a medium pressure region similar to that under CI conditions. In this region,

ion/molecule reactions from bimolecular collisions between K+ and M or D“ in the gas

34

phase result in the formation of either [M+K]+ or [Dn+K]+ adduct ions as shown in
Equation 12. The higher the medium pressure region and the better the temporal and

K+ + M (9 (DH (9) —) [M+K]+ ([Dn+K]+) (12)

[M+K]+* ([Dn+K]+*) + M (Dn) -> IM+K1+ ([Dn+K]+) + M* (Dn*) (13)
spatial overlap of K+ ions with M and D“, the higher the collision rate between K” and M
or D“, which is essential for adduct ion formation. In Equation 13, the medium pressure
region also aids in providing de-activating collisions to remove the excess energy from
the excited adduct ions so no fragmentation occurs [59]. The order in which alkali metals
induce gas-phase fragmentation is Li+ > N a+ > K+ > Rb+ > Cs+ [60]. This order is based
on the affinity of M or Dn for the particular alkali metal, in a manner similar to PAs. Of
the alkali metals, K+ is a compromise because the electrostatic interaction between M or
Dn and K+ is stronger than with C54“ or Rb+, yet it is not as strong as the "near" covalent
interaction between M and Li+. For Li+ and Na+, interaction is more "covalent" than
with K+ meaning more of the charge is passed over to the neutral molecule which results
in more charge-induced fragmentation, similar to the covalent bond between M and H+
[61]. Thus, the process of cationization of neutral molecules with alkali metal ions
produces stable adduct ions that do not readily fragment as do certain protonated
molecules, so more intact analyte ions are obtained for molecular weight information

[62].

3. F ast-Atom Bombardment Mass Spectrometry and Related D/I Particle
Bombardment Techniques

In 1981, Barber and co-workers [63, 64] reported the first high quality D/I mass
spectrum of a peptide, Met-Lys-Bradykinin (MW 1318), using fast-atom bombardment
(FAB) as a novel ionization technique on a double-focusing instrument. However, it was

not until later in 1981 when they revealed their unique sample preparation secret, which

35

was the real significance of their work, since the differences between keV-atom and ion
guns were well understood in the surface sciences. Instead of "drying" the analyte on the
target as in "static" SIMS, Barber and co-workers [65] dissolved the analyte into a polar,
viscous matrix, glycerol, which enhanced analyte sensitivity and signal stability,
eliminated analyte degradation, and extended the lifetime of the analyte ion currents from
seconds to minutes. Today, the D/I techniques FAB and LSIMS (i.e., SIMS
incorporating the liquid matrix) either utilize a fast atom or an ion gun to bombard an
analyte dissolved in a liquid matrix with keV-Xe or Cs+ particles, respectively, to induce
sputtering (i.e., emission) of ions, electrons and neutral molecules from the condensed-
phase into the gas-phase. Thus, they are essentially similar techniques that yield identical
results. Both FAB and LSIMS are commercially available and have essentially opened
the door to routine mass spectrometric analyses of biomolecules and polymers allowing
for major advances in the biological, biomedical, and polymer industries. However, the
full potential of FAB and LSIMS may be unrealized because the D/I mechanisms are still
unclear, which results in the inability to understand the mass spectrum/condensed-phase
analyte relationship and resolve matrix effects that still preclude classes of analytes from

this analysis.

FAB and LSIMS Instrumentation.

"Static" SIMS and FAB are essentially an outgrowth of the surface science
technique of "dynamic" secondary-ion mass spectrometry (SIMS) that is utilized for
depth profiling of metal, inorganic and semiconductor materials [66]. "Dynamic SIMS"
involves using a fast ion gun to bombard the sample with keV-particles at high ion beam
current densities on the order of 1.0 [IA/cm2 in comparison to 1.0 nA/cm2 from the fast.
atom and ion guns used in "static" SIMS and FAB. The fast ion gun for "static" SIMS
was introduced by Benninghoven in 1970 to obtain mass spectral information of

inorganic surfaces [67] and of organic molecules adsorbed on metal surfaces [68]. The

36

fast atom gun technology is as well known as that of ion guns [69], and during the 19605
and 19705, it was used to analyze inorganic surfaces, and organic and biological
molecules adsorbed on metal surfaces [70]. Bombardment of the these surfaces with
keV-atoms instead of ions supposedly eliminated sample charging effects [71]. In FAB
and LSIMS, the highest secondary ion yields are obtained when Xe and Cs+ are the
bombardment species, respectively, because they deliver the greatest momentum transfer
to the sample [72]. However, LSIMS is generally more sensitive than FAB, because the
Cs+ beam can be more tightly focused on the target at much higher bombardment
energies [73]. This deposits more energy into the analyte/matrix which enhances the
desorption of analyte species, increases the analyte ion currents, and essentially lowers
the detection limits for most analytes. In FAB, the Xe beam defocuses between the gun
and target which results in a large area being irradiated on the target and ultimately
reduces analyte sensitivity. The angle of incidence (i.e., the angle between the axis of the
keV-beam and the normal to the target surface) of the fast atom/ion gun that results in the
highest secondary ion yield is approximately 70° [72, 73] and instrument dependent.
Stainless steel targets are most common and least expensive, but gold- and silver-plated
targets have been found to give greater sensitivity [74]. Both FAB and LSIMS

instrumentation can be implemented on sector, quadrupole and TOF mass analyzers.

EnsLAtnnLQuns.

Barber and co-workers developed the FAB ion source for analyzing biomolecules
on a double-focusing instrument, which took advantage of its high mass and resolution
capabilities. Barber and co-workers overcame the high accelerating voltages by utilizing
a neutral keV-beam of atoms for their experiments because it was not influenced by the
high accelerating potentials in the ion source, meaning fewer precautions (or
modifications) needed to be taken in contrast to the application of ion beams in LSIMS.

The fast atom gun for the JEOL HX-l 10, illustrated in Figure 8 to scale, produces a 6-

37

   
   
      
    

Ion Source Housing & Filament (3050 V, Xe 10'5 torr)
Xe (g) + e' (200eV) A5 Xe+' + 2e'

\ Beam Steering (0.50 V & 0 V)

Collision Chamber (0 v, Xe 10-4 torr)
. CE
6+ (3 keV) + Xe (g, thermal) ' “'5

+0
CE conversion Xe (~3 keV) + X6 (thermal)
to Xe atoms

Exit Nozzle & Xe Gas Inlet (-3000 V, Xe 10'3 torr)

I
Fast Xe Atoms (6 keV)

 

Secondary Ions
to Mass A

FD/F AB
Target DIP

\_
:l l I FD/F AB Ion Source

Figure 8. Schematic diagram 0 the fast atom gun (actually a SIMION model drawn to
scale) and the F-AB experimental configuration on the JEOL HX- 110.

J

 

38

keV-beam of Xe atoms at 1.0 nA/cm2 beam densities in a series of discrete steps. In the
ion source of this fast atom gun, Xe gas is ionized by El using a highly efficient ion
source design identical to that of a residual gas analyzer [75]. The electrons are
thermionically emitted from a large Re filament (3050 V, not shown) positioned between
the "anode" (3250 V) and the ion source housing (3050 V), and are accelerated to the
anode with 200 eV of energy from the 200-V potential difference between the anode and
the ion source housing/filament. The electrons circulate in and out of the cylindrical wire
mesh anode held at 3250 V because the electrons are repelled by the opposing ion source
housing circular walls (3050 V), which enhances the ionization efficiency of this ion
source . The Xe“ ions are formed with the highest probability near the top of the anode,
and they are subsequently extracted and accelerated to a final KB of approximately 3 keV
from the potential difference between the ion source housing (3050 V) and the extraction
lens element held at zero V. The ion focus (2600 V) and x- and y-beam steering (~0 V)
lens elements comprise an einzel lens that shape Xe“ ions into a well-defined beam and
steer it towards the exit nozzle, which is biased at -3000 V in the collision chamber.
After the einzel lens, the well-defined 3-keV Xe“ beam undergoes further acceleration to
a final KB of 6 keV because the potential difference between the einzel lens (0 V) and the
exit nozzle (-3000 V) is 3000 V. In the collision chamber or exit nozzle, at least one-half
of the 6—keV Xe+' ions acquire an electron from thermal, gaseous Xe atoms, present at
medium pressures, by resonant charge exchange to produce neutral ~6 keV-Xe atoms
with minimal reduction in KE. Because the recombination energy of monatomic ions
equals the ionization energy of monatomic atoms, charge exchange between Xe't‘ atomic
ions and Xe atoms requires zero energy. After charge exchange, the 6-keV Xe neutral
beam that emerges from the exit nozzle of the gun is unaffected by electric fields in the
ion source, and it impinges on the FAB target, attached to the end of a DIP, with a
diverging beam shape. The secondary ions formed as a result of bombardment are

extracted and accelerated by the ion source optics system. The Xe” ion (black) and

39

neutral Xe trajectories (gray), shown in Figure 8, were estimated using the electrostatic
lens design and analysis computer program, SIMION 4.0, which will be described in
more detail in Chapter three and Appendix one. There are other commercially available
fast atom guns that are based on the same principles as above, but incorporate slightly

different designs [76].

W

A fast ion gun for LSIMS is somewhat more difficult to implement on a double-
focusing instrument and more precautions need to be taken. The fast atom gun should be
mounted as close to the target as possible plus the beam needs to be highly shielded from
the accelerating potentials in the ion source because they can change the Cs+ ion beam
shape and path. In addition, to generate a 10-keV beam of Cs+ ions, the gun potential
needs to be 10 kV higher than the accelerating potential, which may be as high as 10 kV
on a sector instrument, so Cs+ reaches the 10-kV target. These two problems are
alleviated in FAB because the keV neutral beam is not influenced by electric fields which
allows the fast atom gun to operate at much lower potentials. A LSIMS ion source, based
on the design by Aberth et al. [77], is available for the JEOL HX-l 10 and a schematic
diagram of it is illustrated in Figure 9 to scale. The principles of operation for a fast ion
gun in LSIMS are conceptually more simplistic than those for the fast atom guns above
because the Cs+ ions generated in the ion source of the gun are extracted, focused and
accelerated to bombard the analyte/matrix target directly, attached to the end of a DIP, in
the ion source of the mass spectrometer. Specifically, the Cs+ ions are generated by
thermionic emission from a heated Cs+ glass emitter (15 kV). They are then are extracted
and shaped into a well-defined beam by the ion extractor (14995 V), ion focus 1 (14250
V) and ion focus 2 (12750 V) lens elements which comprise an einzel lens. Finally, they
are accelerated to the final, desired KE for bombardment on the sample target (10 kV)

between ion focus 2, the C54“ shield lens element (10600 V) and the target itself.

I

I u
".1:-
u .-

Cs+ Pellet (Cs‘I' Glass Emitter) & Housing (15 kV)

 

 

 

 

 

 

 

   

Ion Extractor (14995 V)

 

 

 

 

‘/ Ion Focus 1 (14250 V)

‘ [on Focus 2 (12750 V)
/ Cs+ Gun Shield (10600 V)
I l/ Cs+ Beam (5 keV)

Seconda Ions r

to Mass Emlyw< Target (10 kV) mpg

 

 

__l

I LSIMS Ion Source

Figure 9. Schematic diagram of the fast ion gun (actually a SIMION model drawn to scale)
and the LSIMS experimental configuration on the JEOL HX—110.

41

Sample Preparation and Matrix Selection in FAB and LSIMS.

The successful and accurate analysis of an analyte in FAB and LSIMS depends on
the correct choice of matrix and additives for that particular analyte so analyte response is
maximized and sample effects, matrix effects and interferences are reduced or eliminated
[78, 79]. In most cases, the sample preparation for FAB and LSIMS involves initially
applying a small, consistent amount of the matrix to the target surface and then mixing a
known amount of analyte, usually nanomolar amounts for FAB and picomolar amounts
for SIMS, either in solid form or dissolved in a co-solvent such as water or methanol, into
the matrix. The sample on the target is then inserted into the mass spectrometer for
analysis by keV particle bombardment. The following criteria should be considered
when selecting a matrix for a particular analyte [80]. The matrix should dissolve the
analyte, possess low volatility under vacuum to preserve the analyte signal, and resist
reaction with the analyte. The standard matrix in FAB and LSIMS is glycerol (MW 92)
because it readily dissolves most polar analytes, it is viscous and inert, and it provides
strong, long-lived secondary ion yields under keV bombardment preventing analyte
degradation. However, glycerol is not always the best—suited matrix for each analyte and
application. Choosing a matrix for a particular analyte is not an exact science and trial
and error is required. The differences in the physical and chemical properties of matrices
such as PAs, solubilities, vapor pressures, heat of vaporization, viscosities, reduction
potentials, dipole moments, etc. WM. Several reviews and papers have
described the advantages, disadvantages, physicochemical properties, and applications of
alternative matrices such as thioglycerol, triethanolamine, and 3-nitrobenzyl alcohol [81,
82, 83]. If the analyte does not dissolve in the matrix, no analyte signal will be observed
by bombardment. To aid in the dissolution of an analyte, additives, such as acids/bases
and/or polar/nonpolar co-solvents, can be dissolved into the analyte/matrix to either

preionize and/or solvate the analyte, respectively.

42

Typical FAB and LSIMS Spectra.

The typical and reproducible LSIMS mass spectrum for 1 ug/uL of triglycine (M)
dissolved in the matrix glycerol (G) is shown in Figure 10. The two sets of peaks
observed for M and G are overlaid in one mass spectrum which may potentially
complicate the interpretation of analyte ions if any glycerol peaks interfere. The peak at
m/z 93 represents protonated glycerol while the peaks at rn/z 29, 45, 57, and 75 represent .
fragment ions of glycerol. The peak at m/z 185 represents the proton-bound dimer of
glycerol. Peaks representing these glycerol "cluster" ions extend out to approximately
m/z 921, which represents 10 glycerol molecules bound to a proton (i.e., m/z=92n+1
where n=10), with a decreasing RI distribution. The peak at m/z 190 represents
protonated triglycine and yields the MW information for this analyte. The structurally
significant fragment ions for triglycine are labeled and identified on the triglycine
structure in which the charged fragment is "identified" to by the arrow. The mass
spectrum essentially resembles a CI mass spectrum owing to the protonated species.
Most fragmentation involves loss of even-electron (EE) neutral fragments from EE
protonated molecules to form EE fragment ions by simple cleavages reactions and
rearrangements reactions that involve hydrogen shifts. However, some analytes such as
triglycine display a mix of EB and odd-electron (OE) fragment ions. The peaks at every
mass or so-called chemical noise is attributed to glycerol radiation-damage products from
Xe beam-induced high energy processes [84]. This mass spectrum demonstrates the
disadvantage of assessing the RI of a peak representing an analyte fragment ion in the
presence of a matrix. The R1 of the peak at m/z 76 represents the ion current from both a
13C-isotope fragment ion of glycerol and the y 1 fragment ion of triglycine. Thus, for
most analytes, FAB and LSIMS can provide molecular weight and significant structural
information, but for some analytes, these techniques are ineffective owing to the unique

chemical properties of the analyte in the presence of a matrix. In the analyses of certain

43

.555: Shook—M 05 E volcano 83 32v 050wa we 43w:— uo 8.58% an: meg—mi— .3 Emmi

as

 

 

 

 

 

 

 

 

 

 

 

 

 

es en ...s_.._...em.r.-oe
_.__.._.__s.“_________$.1.; __ ____._._______.r.________n___..__________._.33.______.._:=_____._.._._______._.__.__:___._._:j ___.__ _______ :____ i __ _
m_m mom ____ a
m: a 2 a . a
0+H=+Ma§~ Nfi VON—
m: . .
mm~ NQ n» m
9. :
. D
I ma .ov a
I e w a
n__u W2, 30/ o \zmm hm rm H mm .8 e
:o\~ U ,0 :z N 0 ..~__ . >_
n__u -EU 12.39.20) 2
i __ m. .9:
e+HI+m£ Nn Na 3» mm 0 . . o—
+E+§ +5.6". +5.é fl um um . m
o2 Lem: mm: o2

 

44

classes of analytes, understanding the current mechanistic details of desorption, ion
formation and fragmentation in FAB and LSIMS are important in resolving chemical

incompatibilities between an analyte and its associated matrix and additive(s) .

Mechanistic Aspects of FAB and LSIMS.

In FAB and LSIMS, analyte ions may arise at various points in time and space by
different mechanisms during the keV-bombardment process as determined by their
chemical nature and/or position relative to the keV-beam impact site [85]. For example,
the analyte or matrix molecules may be (I) initially preionized in solution and then
desorbed intact, (2) immediately ionized and fragmented by uptake of too much energy in
the bombardment process if too close to the impact site, or (3) desorbed intact and
ionized by gas-phase ion/molecule chemistry after desorption. Traditionally, in FAB and
LSIMS, desorption, ionization, and fragmentation are considered as sequential, separate
processes in time, and they will be described that way here [86]. Although this section
describes events that are highly controversial, debated, and not well understood, the
following discussion is based on the available information in the literature and is a good
attempt to provide insight into the mechanisms responsible for formation of the ions
typically observed in the mass spectra of biomolecules. Understanding these mechanisms

is required for successful sample preparation and accurate spectral interpretation.

t' im .

Historically, the first theoretical desorption (or sputtering) models were developed
to gain a better understanding of the sputtering mechanisms responsible for production of
atoms and ions from solid surfaces (i.e., metal, amorphous, or crystal lattice) under SIMS
keV-bombardment so that secondary ion yields could be improved [87]. The now widely
accepted collision cascade model formulated by Sigmund [88] in 1969 was first to

quantitatively describe the sputtering of atoms (not ions) from metal surfaces which is

45

based on elastic (not inelastic) momentum transfer between the keV-projectile particle
and each atom in its random path. Qualitatively, in this model, sputtering is initiated
when a keV-projectile ion or atom enters the solid and is gradually slowed by binary
inelastic collisions with various atoms until all of its translational KE has been transferred
to the surrounding atoms in the interfacial region along its track. Then, these initially
excited atoms in the track undergo second generation binary inelastic collisions with
neighboring atoms and so on, forming a "collision cascade". The initially-deposited
energy is essentially dissipated radially away from the projectile track by momenta
transfers. Energetic binary collisions near the surface may sputter atoms into the gas
phase if the surface binding energies are overcome. The atomic ions, the more important
species in a SIMS experiment, are probably formed by electronic excitation and a
subsequent electron transfer process near the initial keV-particle impact site, but
unfortunately, the sputtering of atomic ions is not as well understood as that for atoms
[89, 90]. The collision cascade model is believed to principally energize the solid, and
sputter atoms away from the target with up to 20 eV of KE . The thermal spike
desorption model [91, 92] evolvesifrom the collisional cascade later in time when the
high translational energies therrnalize the lattice structure and sputter atoms with KEs less
than 1 eV [93]. The thermal spike model describes the collisionally activated region
radially surrounding the track of the keV-projectile as being a high temperature region
that behaves like a dense gas because this lattice structure contains excessive vibrational
(thermal) and translational energies. In this hot, high pressure gaseous region, the
thermal spike model predicts that a large number of atoms/molecules in the upper surface
layers will undergo an explosive expansion into the cool vacuum similar to that in a
hydrodynamic process [94]. The collision cascade and thermal spike phenomenon for-
atoms are widely accepted, although the subject is still debated because much remains to
be learned.

The two desorption models for atoms described above have been successfully

46

extended to provide insight into the sputtering processes for keV bombardment of organic
molecules dissolved in liquid matrix. Fast Xe atoms bombard and penetrate the glycerol
matrix, which contains a dissolved analyte, from a large angle and deposit keV energy
into the upper layers of the matrix as illustrated in Figure 11. Initially, the impinging
keV—Xe atom induces a collisional cascade in glycerol which is shown by the "black"
translational energy flow paths. The desorption events are considered isolated since the
beam density and flux are such that one keV-Xe atom impinges the same area every 1-5
seconds. Because the ratio of matrix to analyte is greater than 1000: 1, glycerol molecules
uptake most of the energy deposited by the keV-Xe projectile, which essentially protects
the analyte from damage by collisional excitation. Hoogerbrugge et. al. [95] have
observed this phenomenon because sputtered N 3*, polyatomic ions, and molecules from a
glycerol matrix were energy-analyzed and had KEs of only 5 eV or less, and no high
energy tail. This indicates a temporary, less violent collision cascade occurs in liquids in

comparison to that in solids. In Figure 11, the top-layer of glycerol molecules at or near

@ Mass Analyzer

\\

 

Figure 11. Energy deposition by a keV-Xe beam induces desorption of atoms, fragment
ions, preionized molecules, and intact molecules (not drawn to scale).

47

the site of Xe impact may be so electronically and vibrationally excited that they are
sputtered intact and fragmented completely into atonrs, atomic ions, ion pairs, or small,
high energy ions according to the QET theory [96]. At a larger radius from the Xe impact
in a less energetically excited region, the desorption of less excited glycerol molecules
occurs which would fragment into one of the normally observed glycerol fragment ions
(G1:+). In the least excited outer radial region, intact cations, preionized intact analyte
ions, and intact glycerol and analyte molecules are desorbed. During the thermal spike
regime, the translationally excited molecules in the collision cascade become
vibrationally excited, so weak intermolecular bonds (i.e., H bonds and van der Waals
forces) between glycerol and analyte molecules are cleaved, ready for low energy
desorption. The highly vibrationally excited glycerol molecules (G) in the interfacial
region of the surrounding projectile track are vaporized and ionized to form G+‘ in

Equation 14. Since a peak representing G+' is not observed in FAB and LSIMS mass

XC(5 keV) + G(g) —) G” + e‘ + Xe(..6 keV) (l4)

6*” -> GF+ + N120 (15)
Xe(6 key) + 6(2) —9 612+ + A‘ + Xe(..6 key) (16)
Gp+ + G(g) —> G-GF+ + H2 (17)

spectra, the average excitation energy must be high enough that G+’* fragments
completely, according to the QET theory, to form the glycerol-related fragment ions,
GF+, by unimolecular decomposition in Equation 15. Gp+ may also be formed by the
separation of an ion pair from collisionally activated glycerol as shown in Equation 16.
Also, in Equation 17, GF+ may react with gas-phase G to form, G-Gp+, which is
essentially a fragment ion of a proton-bound dimer of glycerol [97]. In addition, in this
interfacial region, some analyte and matrix combinations undergo undesirable, keV-
particle beam induced recombination reactions such as oxidation and reduction [98, 99],
and/or additions and substitutions [100, 101]. These reactions may occur in the gas-

and/or condensed-phase, may involve ions and/or molecules of glycerol, analyte, matrix

48

additives, etc., and are probably catalyzed by the highly excited matrix molecules in the
interfacial region. Undoubtedly, these reactions are most serious when they affect the
analyte ions, so the possible particle beam-induced products should be recognized for
spectral interpretation [102, 103]. High-energy, particle beam-induced free-radical
glycerol chemistry also occurs in the interfacial region surrounding the Xe impact site
yielding an abundance of radiation-damage products [84]. At some point in time, the
entire hot, activated region surrounding the impact site will "explode" into the vacuum,
removing and vibrationally cooling several exterior layers of excited intact analyte/matrix
molecules and preionized analyte molecules, and the representative interior ionic species
and radiation products in the interfacial region. Sunner et al. [104] calls this a "phase
explosion" and discusses at great length how and why the superheated matrix irreversibly
"explodes" into the gas-phase. Similarly, in MALDI, a microvolume containing analyte
molecules entrained in a principally solid matrix absorbs laser energy and undergoes a
"collective" phase change forming 104 more intact gas—phase neutral molecules than
ionized intact analyte molecules [105]. Recent results indicate that the top most layers of
matrix and analyte are sputtered away completely exposing a new analyte/matrix surface
[106, 107] because the rate of analyte diffusion from the matrix bulk to replenish the
matrix surface is slow on the experimental time scale [25]. Thus, a matrix such as
glycerol will provide interpretable, long-lived analyte ion signals as long as volatile
radiation-damage products are produced and sputtered away, and beam-induced reactions
between the analyte and matrix are minimized. Otherwise the matrix would likely
polymerize and stop functioning. For MALDI, Vertes et al. [108] gives insight into the
vibrational energy flow in the crystalline lattice, and also between the analyte and the
lattice according to the bottleneck model which is applicable to desorption in other D/I
techniques.

The desorption of salt cations, preionized analytes and intact neutral analyte

molecules from a glycerol matrix dominate in the FAB and LSIMS experiments because

49

glycerol solvates these species and separates the ionic charges which significantly
reduces the surface binding energies required for desorption. Benninghoven [109]
introduced the precursor model to account for the desorption of preionized analytes and
cations directly from the matrix in Equation 18. Here, if a collision from the collision
M“ A‘ (aq) -> We) + A'(g) (18)
cascade provides sufficient energy, the preionized species on the surface are desorbed
into the gas phase usually with little vibrational internal energy content. Here, a large
dose of energy that is deposited locally from the collision cascade results in limited
vibrational energy gain by the analyte in comparison to global energy deposition in
thermal desorption because energy is concentrated in one part of the matrix and its
dissipation from this site is very fast (<10'12 5). For alkali and quaternary salts
traditionally thought to be desorbed as intact cations and anions, Kidwell et al. [110]
suggests that activated ion pairs are desorbed into the gas-phase instead and followed by
dissociation in the gas phase. In contrast, Sunner [111] recently suggested that desorbed

intact cations and anions pair up during desorption or the disintegration of the matrix.

As mentioned above, the molar ratio of glycerol matrix to analyte is 1000:1 so
most neutral molecules and ions desorbed by the collisional cascade and thermal phase
explosion regimes are related to glycerol. Wong et al. [107] estimated 1000 glycerol
molecules are desorbed for each Xe impact. In addition, the glycerol matrix has a high
vapor pressure under vacuum. Thus, between desorption and evaporation of glycerol
molecules, the liquid-vacuum interface is estimated to be a high pressure region. The
observation of protonated analyte molecules and EE fragment ions in typical FAB or
LSIMS mass spectra indicates CI processes may be operative in this high pressure area.
The coupling of CI with FAB for analysis of postdesorbed neutral species has resulted in

[M-I-H]+ sensitivity increases by three orders of magnitude, which indicates a large

50

number of neutral species are present in this region [112]. A similar experiment by
Schroder et al. [113] concluded gas-phase glycerol molecules, introduced into the gas-
phase from a heated inlet, were initially ionized from collisional activation by the keV—Xe
beam to produce the glycerol reagent ions, G‘”, which then protonate other gas-phase
glycerol molecules in a process very similar to that for CI. Historically, Rabalais and co—
workers [114] noted the formation of cesium iodide clusters by Ar+ bombardment of C51
using SIMS and attributed this phenomenon to a high pressure region above the target
surface they termed the "selvedge". Here, the desorbed cations and anions with similar
energies moving through this "selvedge" region would aggregate into gas-phase cluster
ions. Using SIMS, Cooks and co—workers [115, 116] used the selvedge region concept to
explain the cationization of organic molecules with protons and desorbed metal ions from
the target in Equation 19. They described the selvedge at the liquid—vacuum interface as a

M(s) —) M(g) + H+(g) (Ag+, K+) -) [M+H]+(g) ([M-I-Ag, K]+) (19)
high pressure, high temperature, and highly charged region where gas-phase,
recombination, CI-like, associativeion/molecule cherrristry occurs [117]. The height of
the selvedge region was estimated to be on the micrometer-nanometer level, but its height
should be defined as where the gas-phase chemistry ceases. In Figure 12, the neutral and
ionic species desorbed into the selvedge region by FAB are shown in addition to the
[G+H]+ and [M+H]+ ions formed.

Kebarle and co-workers [118, 119] noted if two matrices with different gas-phase
basicities (GB) are mixed on the target, with one having the typical concentration of an
analyte, and analyzed by FAB, the matrix with the lower GB will be suppressed by an
amount relative to its concentration in comparison to the matrix with the higher GB. The
GB (-AGm,) is approximately equal to the PA (-AHm.) if the Aern'T term in the Gibbs-.
Helmholtz equation Aern=Aern-A8m,°T is negligible, but if a proton dicoordinates or
cyclizes, ASrxn cannot be neglected and GB¢PA. Acid-base chemistry in the condensed

phase was ruled out because these matrices do not readily ionize or exhibit dependence

51

Mass Analyzer
[M +H1+ [G+H]"'
Salt '9' [(M+H)-H20]+
Z ’ " ~ ~ ‘ p Z
Free Vacuum [(G+H)'H20I+ / ’
..... 3:- / ®®
Selvedge Region +
[M+H]+ [Gan] ® ®
Salt G // 5
G M + Fast Atoms
G M + (j / F"

Gas Phase

  

 
 

"
.’ ~ ’1’ .\“ \“

 
  

      

Figure 12. The neutral and ionic species desorbed into the selvedge region above the
FAB target during Xe bombardment.

on bombardment time. From the above observation, they proposed the "gas collision
model", which is similar to the selvedge region model. The majority of the species
desorbed into the gas phase are glycerol (G) molecules. Also, "trickles" of analyte
molecules (M) are desorbed as are trickles of the ionic fragment species such as Gp+ that
are related to glycerol and formed near keV-Xe impact site as presented in Equations 14-
17 above. An additional, fraction of ionic species may be formed in the gas phase when
intact gas-phase glycerol molecules are collisionally activated by a colliding keV-Xe
atom and ionized by E1 or charge transfer/CE processes (see below) [120]. Primarily, the
G1:+ ionic species, the fragment ions related to glycerol (and also to a lesser degree G-
GF+ and 6"“), undergo many collisions with G and initially protonate the intact, gas-
phase glycerol molecules in the high pressure selvedge region by a process similar to that

0f CI as shown in Equation 20. A protonated glycerol molecule may undergo a

52

GF+ (G+') + G(g) —9 [G+H]+ + [Gp+-H+] ([G+’-H+]) (20)
bimolecular collision with a gas-phase analyte molecule which will result in a proton
transfer if the GB of the analyte is higher than that of glycerol as shown in Equation 21.

[G+H]+ + M(g) —> [M+H]+ + G(g) (21)
Thus, this model portrays FAB essentially as a glycerol chemical ionization experiment
using protonated glycerol molecules as the reagent ions for protonating the analyte
molecules. In addition, the analyte may be protonated by G“ or the ionic fragments of
glycerol if the analyte GB is less than that for glycerol as shown in Equation 22. Once

01:4” (64") + M(g)( —> [M+H]+ + [Gp+-H+] ([G+‘-H+]) (22)
protonated analyte molecules are formed in the selvedge region, they may undergo a
nonreactive collision to remove excess energy, a reactive collision with another G to
produce a "cluster" ion, [M+G+H]+, or an immediate extraction towards the mass
analyzer which will probably result in fragmentation if they were not vibrationally
"cooled" in the selvedge [121]. In Equation 23, if two analytes M1 and M2 are dissolved

M1(g) + [M2+H]+ —> [M1+H]+ + M2(g) (23)
in glycerol, and M1 has a higher GB than M2, but M2 has a higher solution pr,
[M1+H]+ will have a higher RA than [M2+H]+ supporting CI gas-phase chemistry in the
selvedge since proton transfer from [M2+H]+ to Mug) will occur assuming both analytes
are initially protonated equally by glycerol ionic species [122].

The formation of "cluster" ions in FAB and LSIMS represents another selvedge
region reaction type in which protonated glycerol molecules collide with G to form a
proton-bound dimer of glycerol as shown in Equation 24. The decreasing RI distribution

[G+H]+ + G(g) -> [2G+H]+ (24)
0f peaks representing the increasing size of glycerol cluster ions according to the formula,
m/z=92n+1, in a typical FAB mass spectrum indicates cluster ions are most likely formed
from a sequential clustering process in the selvedge region. Because the larger clusters

require more collisions in the selvedge, they are not formed as often as smaller cluster

lllI

f0

53

ions that require fewer collisions as indicated by the peak RIs. This distribution may also
indicate that larger clusters are not as thermodynamically stable and are not readily
formed. The other school of thought considers the possibility that both small and large
neutral and ionic cluster species are desorbed into the gas-phase instead of individual
molecules from the thermal expansion (phase explosion) of the matrix. These small and
large ionic cluster species then desolvate in the free vacuum to give the distribution of
peaks observed in the mass spectrum [94, 123]. The formation of a bare protonated
analyte or glycerol molecule from a large cluster ion seems unlikely as the dominant
process because experimentally, [G+H]+ and [2G-i-H]+ are the dominant glycerol ions
observed and a decreasing RI distribution for peaks representing the larger glycerol
cluster ions is observed. If declustering was a major process, a gaussian RI distribution
of peaks representing the "surviving" glycerol cluster ions would be expected since
complete desolvation seems unlikely under FAB and LSIMS experimental conditions
[122]. However, dissociation of cluster ions probably occurs to some extent depending on
the analyte and matrix.

For preionized analytes or cationic salts, R4N+X', in positive mode, the respective
cations, [M+H]+ and R4N+, would be desorbed into the selvedge region. The high RIs of
intact cationic substances indicates either they are desorbed in staggering numbers,
desorbed with low internal energy that prevents fragmentation and/or are initially excited
and undergo nonreactive collisions for de-excitation. If the R4N+X‘ ion pair is desorbed
intact or undergoes aggregation in the selvedge, the reaction in Equation 25 will yield

R4N+X' + [G+H]+ —> R4N+ + G(g) + HX (25)
R4N+ in the selvedge, and R4N+ may subsequently undergo an ion/molecule collision in
the selvedge with glycerol to form [R4N+G]+. In addition, preionized [M+H]+ ion may
be neutralized to M if its GB is less than that of another desorbed molecule as in Equation
23. Also, if the matrix is d0ped with KCl salts, either K+ or KCl will be desorbed into the

selvedge region to form K+ adduct ions with glycerol and analyte molecules as shown in

54

Equations 26 and 27, respectively [97]. The exchange of K+ between G and M will occur

K+ + G(g) (M(g)) —> [G+K]+ ([M+K]+) (26)

KCl(g) + [G+H]+ ([M+H]+) —> [G+K]+ ([M+K]+) + HCl (27)
as in Equation 21.

For polyaromatic hydrocarbons (PAH), the M“ ion is observed instead of a
protonated molecule which is expected because the PAs of these compounds are very low
(~18O kcal/mol) in comparison to that of glycerol (209 kcal/mol). Because protonation is
not a readily available option and PAHs are highly volatile, ionization may occur by El,
CE or collisional activation in the selvedge region from sputtered secondary electrons
during Xe bombardment as in Equation 28 [117], from a charge transfer/charge exchange

Mrs) '7 Mrg) + e'(>12 eV) -> M” (28)

process where keV-Xe ionizes residual Xe (Xer) in the ion source which then ionizes M

Xeawv) + Xer -—> XC(~keV) + Xe” + e' (29)
Xe(kev) + M(g) —> XC(~keV) + M” + e' (31)

by CE as in Equations 29 and 30 [120], and from collisional activation of M by a keV-Xe
particle as in Equation 31, respectively [120].

Thus, the high collision rate between glycerol and analyte molecules, and the
dependence of the chemistry on GBs in the high-pressure selvedge region allow CI-type
ion formation mechanisms to explain the presence of various protonated molecules,
cations, clusters ions, and molecular ions in the FAB and LSIMS mass spectra. Recently,
in the work of Rouse and Allison, desorbed analyte molecules were detected in the so-
called selvedge region using the K+IDS-by-FAB technique, which supports the
mechanisms set forth by the Kebarle and Cooks research groups (see Chapter three and
Appendix two for details) [124]. Cooks and co-workers [86] introduced a unified D/I
model based on the similarity of mass spectra from FAB, LSIMS, PD and LD analyses

for a wide range of analytes. Basically, the energy deposited by particle bombardment is

55

isomerized (i.e., redistributed) in the analyte-matrix system resulting in the desorption of
analyte species, subsequent ionization and stabilization in the high pressure "selvedge"
region by C1 processes and/or unimolecular dissociation in the free vacuum. Recently, in
MALDI, an ion formation model was proposed from fundamental experiments in which
the matrix is first photoionized and subsequent ion/molecule reactions between these
desorbed, photoionized matrix species and gas-phase analyte molecules occur in the gas-

phase [125].

Il'lll! "8'

Both the neutral and preionized species contain an undetermined, but sufficient,

amount of initial internal energy from the desorption event to induce unimolecular
decomposition (after ionization, if required). Species desorbed from the collision cascade
and thermal desorption regimes likely have high and low internal and kinetic energies,
respectively. In the high pressure selvedge region, the initial internal energies of these
species will ultimately be determined by reactive (i.e., protonation) and nonreactive
collisions which increase and decrease internal energies, respectively. The longer an ion
or molecule spends in the selvedge region, the greater the probability for more stabilizing
collisions that ultimately decrease the initial internal energies. The protonated molecules,
either preionized or formed in the selvedge region, are gradually extracted into the free
vacuum towards the mass analyzer by the ion source ion optics. In FAB and LSIMS, the
observation of metastable transitions that yield EE fragment ions following protonation of
the analyte indicates that gas-phase unimolecular decomposition occurs in the free
vacuum above the selvedge region [86]. The [M+H]+ internal energies are sufficient to
induce either fast simple cleavages or slower rearrangement fragmentation since fragment
ions are observed on the "normal" and metastable timescales. The [M+H]+ fragmentation
appears to be kinetically competitive, and is expected to be in accordance with the QET

and RRKM theories discussed above for El (see Chapter 4 for peptide fragmentation

56

mechanisms in LSIMS).

Attempts to estimate the internal energy content of desorbed intact neutral
molecules and preformed ions emerging from the selvedge region have been made
previously as described below. Hoogerbrugge et al. [126] determined the internal
temperature of glycerol molecules sputtered from a target of 267 K to be 190 K, which
indicates the desorbed glycerol molecules are cooled by expansion into the vacuum
and/or by nonreactive collisions in the selvedge. This internal temperature of sputtered
glycerol molecules corresponds to an internal energy of 0.044 eV assuming only 15% of
the total degrees of freedom are active. Derwa et al. [127] and Williams et al. [128] have
estimated, using benzylpyridinium salts in glycerol, that the internal energy content of
preionized secondary ions in FAB and LSIMS is between 0 and 4 eV; this energy
distribution is most populated between 0 and 2 eV, with a high energy tail that extends
out to 4 eV. Derwa et al. [127] used Hoogerbrugge's internal temperature value of 190 K
to calculate an internal energy of 1.13 eV for desorbed benzylpyridinium ions assuming
all 69 degrees of freedom were active. Kebarle and co—workers [118] and Hamdan [129]
found the distribution of peaks representing fragment ions in FAB and FAB—CAD (see
next section for CAD discussion) mass spectra for typical matrices (PAz200 kcal/mol)
corresponds almost identically with the RI distribution of peaks representing fragment
ions in methane CI (PA=132) and CI-CAD spectra. This indicates the protonated
molecules formed in FAB and in methane CI have very similar internal energy

distributions that extend out to 3 eV owing to the 68 kcal/mol difference in PAs.

Sample and Matrix Effects in FAB and LSIMS.

The inability to identify important analyte peaks from matrix interferences, and/or
the partial or complete analyte signal suppression often observed in FAB and LSIMS
mass spectra are referred to as sample and matrix effects [72, 78, 79, 80, 130]. A peak

representing a matrix ion that unfortunately overlaps with an important analyte peak in

57

the mass spectrum is a common occurrence. The m/z values of the analyte or matrix
peaks may be "shifted" by derivatizing the analyte or employing a different matrix,
respectively. An MS/MS product ion scan (discussed below in Sections 4 and 5) will
provide fragment/product ions free of matrix interferences. Cationization of the analyte
molecules with NaCl or KCl salts, placed in the matrix, results in a new peak 23 or 39
mass units higher than the analyte MW, respectively, that represents the highly stable
adduct ions, [M+Cat]+. These stable adduct ions do not fragment and they provide
confirming MW information, especially if the exact location or form of the MW peak
(i.e., [M+H]+, M+', a fragment ion, or [M+K]+ adduct ion) is unknown [131, 132]. If the
concentration of salts is high, as in biological samples, additional adduct ion peaks such
as [M-I-2K-H]+ will appear in the mass spectrum which is less (interpretable at higher
mass, and quite possibly the analyte peaks may become partially suppressed in the mass
spectrum . HPLC is usually used to remove salt impurities from biological samples.
Analyte suppression is a serious matrix effect, especially in analyte mixtures, because the
R15 of each component in the mass spectrum may be a poor representation of their actual
concentrations. Suppression can be caused by poor analyte solubility (see "sample
preparation" above), poor surface activity, or an unfavorable difference between the
analyte and matrix PAs (see "mechanisms" above). A "surface active" analyte
preferentially concentrates in the matrix surface which results in preferential desorption
and thus strong ion signals. Preionized analytes and hydrophobic analytes are usually
surface active while hydrophilic analytes readily dissolve and disperse in the matrix
which results in poor ion yields. The addition of acids/bases and surfactants to the matrix
may bring the hydrophilic analytes to the surface as will preionization of the analyte by
fixing a charge using known derivatization techniques to the analyte [133, 134]. Some
analytes respond better in negative ion detection mode owing to a highly acidic group,

while other analytes respond better in positive ion mode owing to a highly basic group.

58

4. Principles and Utility of Collisionally-Activated Dissociation (CAD)

For accurate determination of analyte structure, abundant and informative
fragment ions in the mass spectrum are required, as are routinely obtained by El.
However, the mass spectrometric analyses of thermally labile, high molecular weight
biomolecules require a "soft" ionization technique, either FD, FAB, LSIMS, MALDI,
PD, or ESI, which on average only impart a small amount of internal energy into the
protonated molecules in relation to their large size. In contrast, a large amount of internal
energy is imparted into small molecules by El. Thus, structurally informative
fragmentation for these high molecular weight biomolecules is limited because the small
amount of internal energy imparted into the protonated molecules is distributed over
more vibrational degrees of freedom than is the case in E1. As a result, the probability of
sufficient internal energy concentrating in a bond or group of bonds to induce fast
fragmentation in the ion source is lowered. Collisionally-activated dissociation (CAD)
[135] can be employed to impart a range of sufficient internal energies into low energy
[M-i-H]+ ions through energetic bimolecular collisions with an inert gaseous atoms to
stimulate both high- and low-energy fragmentation processes so that a wide range of
structural information can be obtained. The amount of energy imparted depends on the
number of collisions, KB of the [M+H]+ projectile ion, and the atomic size of the target
gas. In fact, CAD product ions may be obtained from any user-selected precursor ion,
which is any intact ionized analyte molecule or fragment ion formed in the ion source
before acceleration that appears in the "normal" mass spectrum. Often times, protonated
biomolecules with low internal energies also undergo metastable transitions which may
produce all or some of the complementary structural information required. However,
metastable transitions usually represent the low-energy fragmentation options for a poorly
activated protonated molecule meaning that the metastable fragment ions observed

underwent primarily rearrangement reactions. If the identical fragment ions are formed

59

in CAD and metastable decomposition, it is important to realize that the metastable
mechanisms induced by lower average internal energies may be different from the CAD
mechanisms induced by higher average internal energies. Since CAD fragmentation
occurs after the acceleration of the user-selected precursor ions, which are formed in the
ion source, tandem mass spectrometry is required for detection; this technology is
discussed in the next section as will the applications of CAD. High-energy CAD is
performed on sector instruments because the ion source accelerating voltage and ion KEs
are in the keV range while low-energy CAD is performed on instruments containing
quadrupoles that ultimately mass analyze the CAD product ions since the accelerating
voltages and ion KEs are restricted to the eV range. Low-energy CAD was not

performed in this work and is discussed elsewhere [136].

High-Energy CAD Instrumentation and Parameters.

In high-energy CAD [137] on a double-focusing instrument, a lO-keV-ion beam
traverses a collision cell, positioned between the ion source and the mass analyzer
entrance (see Figure 14), containing an inert gas such as He, Ar, or Xe at thermal
energies. The ion beam undergoes one or more energetic collision(s) with the gaseous
atoms, depending on the magnitude of the cell pressure, which in turn activates the ions
as shown for [M+H]+ in Equation 32. If the [M+H]+ is user-selected, meaning

[M+H]+1o keV + He —-) [M+H]+*(..1o keV) + He (32)

[M+H]+* -> [M+H]+1' —> Pn+ + Nn (33)
fragmentation is desired from this [M+H]+, then the mass analyzer will detect the product
ions, Pn+, formed from the unimolecular decomposition of the [M+H]+-’F in the critical
configuration following collisional activation of [M+H]+* in Equation 33. In high-
energy CAD, each successive collision imparts more energy into the colliding [M+H]+
ion which in effect opens up fragmentation pathways requiring high energy such as

simple bond cleavages and the successive decomposition of primary product ions into

60

secondary or higher product ions [138]. The number of collisions is proportional to the
target gas pressure in the collision cell. Since the pressure in the collision cell cannot be
measured directly, the number of collisions is adjusted by increasing or decreasing the
target gas pressure in the cell so the user-selected ion signal is attenuated by a certain
percentage . For example, the attenuation of the ion signal by: (1) 10% results in 100%
single collisions; (2) 50% results in 70% single, 20% double, and 10% triple collisions;
and (3) 70% results in 50% single, 30% double, 16% triple, and 4% quadruple collisions,
respectively [138]. In addition, the energy uptake by user-selected precursor ions can be
controlled by the size of collision target gas [139]. Helium is best for high-energy
collisions because it minimizes scattering and neutralization of [M+H]+ ions while still
yielding abundant CAD fragmentation from moderately efficient [M+H]+ activation.
However, the increased atomic size of Xe and Ar allows more efficient activation of
[M-I-H]+ ions, but the competing scattering and neutralization processes are increased

resulting in sensitivity losses.

High-Energy CAD Energy Deposition and Fragmentation.

The first step in high—energy CAD involves the conversion of a small portion of
the [M+H]+ ion's translational energy into additional internal energy as the result of a
bimolecular collision with the target gas. The [M+H]+ ion is excited electronically from
the electron repulsion generated as the fast moving ion approaches and separates from the
short-lived (i.e., approximately 10‘14 s), inelastic ion-target molecule collision complex.
In CAD, the maximum KE available for conversion into [M+H]+ ion internal energy,
Ecom, is calculated using center-of-mass (COM) considerations as shown in Equation 34.

Beam = Elab (mt/(mt+mion)) (34)
For example, a user-selected precursor ion, [M+H]+, with a m/z value of 500, set equal to
man, has the potential to gain 79 eV of internal energy, Ecom, assuming He is the target

gas (m = 4) and Eiab is the 10 keV accelerating voltage. However, in high-energy CAD,

61

assuming single-collision conditions exist in a He-filled collision cell, the average energy
uptake by the [M+H]"’ from the bimolecular collision in Equation 32 is only between 1
and 3 eV with a high-energy tail extending out to 15 eV [140]. The average internal
energies imparted into [M+H]+ are far less than Beam and have been shown to be
proportional to the mass of the analyte ion [14]]. The second step after [M+H]+
activation is its unimolecular decomposition into a product ion(s), Pn+, as shown in
Equation 33, which depends on its internal energy content for determining the rate of
decomposition, k(E). According to the QET and RRKM theories discussed above, the
internal energy is randomly distributed among the active 3N-6 vibrational degrees of
freedom in [M+H]+* until it concentrates in certain vibrational mode(s) to cause bond
cleavage(s) in [M+H]+*. Since the energy imparted into protonated molecules by CAD is
on average high, both simple bond cleavages and rearrangement product ions are
observed on fast and slow timescales. However, as the degrees of freedom increase with
the number of atoms in [M+H]+, the internal energy imparted into [M+H]+ is
increasingly spread over more vibrational modes and eventually it becomes ineffective in
inducing competitive CAD fragmentation rates [142]. For analytes with MW 5 greater
than 1500 Da, CAD is essentially ineffective for inducing abundant, useful
fragmentation. In addition, the CAD ion decomposition mechanisms may be different
from ion source mechanisms for the same product/fragment ion because CAD may impart
a different distribution of internal energies into the [M+H]+ ions, or low energy [M+H]+
ions for CAD activation may have different protonation sites than higher energy [M+H]+
ions that fragment in the ion source. CAD is only one method of activation, but its
simplicity makes it most popular. Surface-induced dissociation (SID) [143, 144] and
photo-induced dissociation (PID) [145, 146] techniques have demonstrated that more
internal energy than that of CAD can be deposited in large biomolecule [M+H]+ ions, but

their set up and operation are more complicated.

8f.

3B
0p§
11161

and

Will]

i002.

62

5. The Mass Analyzers Used for D/I Experiments

D/I techniques are most often performed on commercially available mass
spectrometers that employ either a double-focusing, quadrupole, or time-of-flight (TOF)
mass analyzer. As is true for ionization techniques, no one mass analyzer has the ideal
combination of mass range, mass accuracy, resolving power, sensitivity, scanning speed,
etc. for all mass spectrometric experiments, and compromises are common [147]. The
double—focusing sector mass analyzer provides high mass range, high resolution (at the
expense of sensitivity), and accurate mass determinations for ions generated by the
ionization techniques EI, CI, FI, FD, DCI, FAB, LSIMS, and CF-FAB. Unfortunately,
sector instruments are expensive, difficult to maintain, non-portable, possess slow scan
speeds (in comparison to quadrupoles and TOF over the identical mass range) and utilize
high acceleration and detector voltages. The quadrupole mass analyzer provides unit
resolution over the entire mass range, fast scan speeds, strong focusing properties and no
slits which enhances sensitivity, usability at higher pressures, easy maintenance,
portability and affordability. However, the quadrupole mass analyzer has a low upper
mass range limit in comparison to those of sector and TOF instruments, and it
discriminates against ions of high mass. Quadrupoles are most often used as mass
analyzers for the ionization techniques EI, CI, APCI, DCI, FAB, or ESI. TOF mass
analyzers almost represent the ideal mass analyzer because they boast the fastest scan
speeds, unlimited mass range, high sensitivity (no slits), easy maintenance and simple
operation. Unfortunately, the resolution is often less than 500, which severely limits the
identification and mass accuracy of high-mass ions. The addition of a reflectron [148]
and utilization of high accelerating voltages greater than 10 kV improves the resolution to
an upper limit of 1500 which is still too low for accurate mass assignments of analytes
with molecular weights above 10 kDa. TOF is the ideal mass analyzer for the pulsed

ionization techniques PD, LD, and MALDI. These ionization techniques are all capable

63

of producing ions with m/z values greater than 10 kDa which is well suited for TOF. In
addition, EI and LSIMS can be set to pulsed mode for TOF analyses as well. The
fundamentals of operation for these three analyzers will be described below, but first

extraction and acceleration of ions will be discussed which is universal to any mass

analyzer.

Ion Extraction from Ion Source.

The ions (e.g., M“) formed in the ion volume of the ion source are extracted and
accelerated by the ion optical lens system on the ion source as shown in Figure 13a. The
potential energy surface in Figure 13b is generated by placing the specified electrostatic
potentials on the lens elements of the ion optics in Figure 13a. The ion of mass, In“, and
charge, 2, in Figure 13a will move through a potential difference, V, of 10000 V between
the exit slit and L2, and, therefore, gain 10000 eV of kinetic energy (KE) if z is one.
Thus, the KB gained by the ion will be equal to the potential energy (PE) lost in Equation
35 as shown in Figure 13b. The repeller in Figure 13a simply "pushes" the newly formed

Repeller Exit Slit Re 11 er . .
(10910 V) (10008 V) Ion Optics PC EXlt Sllt

 

 

 

 

 

 

 

 

 

5 \l I
Q A
‘3 >
, Mass :73 I;
) M"""> <( . Analyzer g g)
é’ 3’
L2 m
(0 V) 0
Ion Source L1 Distance (cm)

Volume (10000 V) (8000 V)

a) b)

Figure 13. The extraction and acceleration of M“ a) from an ion source and b) the
representative potential energy surface for this ion source.

KE = 1/2mnvn2 = zeV = PE (35)
ions towards the exit slit for extraction by the ion optics. All ions are formed in the same
general area in the ion source volume, so each ion will have nearly identical final KEs
when they emanate from the ion source, but their velocities, V“, will be different and
inversely proportional to their m/z ratio. By fine tuning the potentials on the ion optical
elements, extraction of the maximum number of ions to enhance sensitivity and the

generation of a well-defined ion beam to enhance the peak shape can be achieved.

Sector Mass Analyzer.

A Nier-Johnson, double-focusing mass spectrometer is shown in Figure 14. The
forward-geometry mass analyzer contains an electric (E) and magnetic (B) sector in
tandem, often expressed as EB. The ion beam shown in Figure 14 is isobaric meaning all
ions in the beam have identical m/z values. In the E sector, a fixed electric field, E), is
created by the potential difference between the two electrostatic plates, which is linked to
the accelerating voltage. For ions to traverse the E sector of a fixed radius, R, and fixed
electric field strength, E0, the ions need to have a KE equal to the accelerating voltage, V,
as shown in Equation 36. However, isobaric ions that emanate from the main slit with

R = mnan/Eo = 2 KE/Eo = 2 WED (36)
angular divergence and/or ions with greater or lower KE in comparison to the
accelerating potential are spatially refocused, if possible, which enhances sensitivity.
Thus, there is a certain tolerance of ion kinetic energies that are passed through B because
the beta slit deterrrrines the finite range of radii to be passed onto B, so R varies with ion
KE in practice. The alpha slit width determines the maximum angular divergence of the
ion beam that enters E, and the beta slit width determines the velocity spread of the ions;
resolved by E, allowed to enter the magnetic sector, B. In B, the magnetic field extends
out of the page, denoted by "x"s in Figure 14, and is perpendicular to the velocity vector,

V“, of an ion. The force exerted by the magnetic field on an ion changes the direction of

65

Magnetic
Sector (B)

Collector Slit

   
    

I-Collision Cell

Direct Insertion Probe

 

Volume

 

Ion Source

Figure 14. Schematic diagram of a double-focusing mass spectrometer.

 

 

 

w-RFV

Figure 15. Schematic diagram of a quadrupole mass analyzer.

66

its velocity, but not its magnitude. Thus, the basis of mass separation for the magnetic
sector analyzer, as shown in Equation 37, is Newton's second law which states that ions

rn =mnvn/zeBo (37)
of different momenta, mnvn, but same 2, will have different radii, rn, in a uniform
magnetic field, B0. In the early days of mass spectrometry, ions with different radii,
having essentially different momenta, were detected simultaneously using a photographic
plate before sophisticated electronic control systems and sensitive detection systems were
developed [149]. Today, as illustrated in Figure 14, ion flux detectors are positioned at a
fixed radius, r, such that ions of different momenta are detected individually by scanning
(i.e., increasing or decreasing) the magnetic field. The ion flux detectors are more
sensitive and convenient for routine use than photographic plates. Since the ions entering
B have the same velocity, V“, as obtained from acceleration in the ion source, the velocity
term in Equation 35 can be substituted for the velocity term in Equation 37 and this
expression can be solved for any mn/z value as shown in Equation 38, assuming the
magnetic field, B“, is scanned, and the radius of ion detection, r, and the accelerating

mn/z = Bn2r2/2V (38)
voltage, V, are fixed. Thus, by scanning the magnetic field, B", to higher fields at a fixed
accelerating potential, V, each group of ions from the total ion beam with unique and
increasing mn/z values will be sequentially focused onto the collector slit before the
detector in Figure 14 when the radius, rn, of the group of ions equals the fixed radius, r, of
the collector slit. If E and B work in tandem, the double-focusing properties of the mass
spectrometer are achieved when the isobaric ions with diverging trajectories and/or
velocity differences from inhomogeneous energies are corrected spatially in both sectors
as shown in Figure 14. As a result, the beam of isobaric ions with various velocities and
directions initially are (doubly) focused to a single point on the collector slit.

In 1953, Nier and Johnson [150] designed the most sensitive and accurate, high

resolution EB double-focusing magnetic sector instrument at that time by refining the

67

various instrumental geometries to achieve true double-focusing of the isobaric ion beams
with inhomogeneous energies and/or angular deviations. The double-focusing mass
spectrometers commercially available today, such as the JEOL HX-l 10, are based on the
Nier-Johnson geometry and they include such refinements such as electrostatic
quadrupole lenses to reduce ion beam dispersion in the horizontal and vertical planes (2-
direction) before both E and B sectors. The FAB experiments described in this
dissertation were conducted on the JEOL HX-110 because of its high mass range, mass

accuracy, and sensitivity.

Quadrupole Mass Analyzer.

The quadrupole mass analyzer [151] consists of a set of 4 parallel, precision-
machined rods arranged radially, equally distant from each other and a center point as
illustrated in Figure 15. A quadrupole mass analyzer electronically "filters" low KE
energy (i.e., 100 eV or less) ions directly by their m/z values through a combination of
time-independent (dc) and time-dependent (ac, i.e., radio frequencies (RF)) potentials
applied to the four rods. Even if the ions possess a wide range of energies or angular
directions upon entering the central space between the four rods, the unit resolution
remains unaffected unlike that in sector and TOF instruments because ions are filtered by
their m/z value.

The stability and paths of ions can be calculated from the Mathieu equations, but
here, only a qualitative description will be presented. In Figure 15, the RF potentials
applied to the yz plane are 180° out of phase from the RF potentials in the x2 plane while
the dc polarities remain fixed. The quadrupole mass analyzer is scanned incrementally by
increasing both the RF and dc potentials at a fixed ratio. When the RF is positive in a
particular xz or yz plane, a positive ion will be focused to the center axis of the space

between the rods, while a negative RF potential will defocus the ion towards the rods. As

the RF voltage in the yz plane is increased during scanning, the positive ion will

68

eventually strike the rod when the RF negative potential becomes too large with respect
to the m/z value and the position of the ion in the quadrupole assembly such that the
positive RF fails to restore the ion's straying trajectories. Light ions are primarily
affected by the RF voltage while heavy ions tend to feel only the average RF force. The
dc potential in the yz plane is positive which contributes to focusing both small and large
ions in the quadrupole assembly. Heavy ions are gradually defocused by the negative dc
potential in the xz plane. As the negative dc voltage is increased, heavy ions are
eliminated when the respective RF potentials fail to restore or stabilize their trajectories
owing to their increased momentum. Thus, the yz rods pass high-mass ions, but filter
low-mass ions because the RF has negligible filtering effect on high-mass ions. The rods
in the xz plane pass low-mass ions and filter high-mass ions because the negative dc
potentials have little filtering effect on low-mass ions. By scanning or increasing the dc
and RF potentials at a constant ratio, a group of isobaric ions at a particular m/z value,
depending on the potentials, will remain stable in the xz and yz planes while the rest of
the ions are lost to the four rods. The quadrupole mass analyzer was developed by Paul
et al. [152] and recently reviewed by Dawson [153]. The K+IDS experiments described
later in this dissertation required the fast scanning capabilities of a quadrupole mass

analyzer and were performed on a Hewlett Packard quadrupole mass spectrometer.

Time-o -Flight (TOF) Mass Analyzer.
In TOF, ion packets are generated by pulsing the ion source accelerating voltage,
V, on and off rapidly. The ions in each ion packet after acceleration gain nearly identical
final KEs as explained above. As demonstrated in Equation 39, each group of ions of a
KE = l/2mnvn2 (39)
unique mass, In“, travel at unique velocities, v... TOF takes advantage of this
phenomenon by measuring the time, t, it takes for an ion packet composed of ions of

mass, In“, to translate the a hollow flight tube of distance, D, and reach the detector. The

69

mn/z values of ion packets can be determined as shown in Equation 40 by knowing V, tn,
mn/z = 2eV(t,,/D)2 (40)

and D. The TOF instrument was originally described by Wiley and McLaren [154] in

1955 and Cotter recently reviewed TOF instrumentation for applications in the biological

sciences [155].

Tandem Mass Spectrometry (MS/MS).

Until now, the mass analyzer discussion above has only considered scanning a
single mass analyzer to separate, as a function of mass, the ions formed in the ion source
volume before acceleration. Instrumentally, true tandem mass spectrometry (MS/MS)
involves interfacing two mass analyzers on either side of a field-free region that contains
a collision cell, stainless steel surface, laser excitation cell for the activation techniques
CAD, SID and PID, respectively, although MS/MS on sector instruments is more
obscure. Commercially available instrumentation capable of MS/MS are the triple
quadrupole (QqQ), double-focusing (EB or BE), hybrid instrument (BEqQ) and four-
sector (EBEB, two double-focusing mass spectrometers linked to either side of a collision
cell) mass spectrometers where B is a magnetic sector, E is an electric sector, q is an RF-
only quadrupole, and Q is a RF/dc quadrupole. MS/MS on each instrument will be
discussed in further detail below.

I! 1' EMS“:

Tandem mass spectrometry instrumentation is necessary for observing the
fragmentation of intact ionized analyte molecules and fragment ions formed in the ion
source after acceleration, specifically, low-energy metastable transitions or product ions'
from higher-energy processes formed by CAD, SID, and PID. Both types of ions arise
from a single, user—selected precursor ion, which is formed in the ion source before

acceleration and would appear in the "normal" mass spectrum. The popular MS/MS scan

IN

70

type is the product-ion scan. Here, the tandem mass analyzer detects all metastable
fragment or CAD product ions that arise from the user-selected precursor ion. Most
often, the product ion scan is used in conjunction with CAD to obtain additional,
complementary structural information for a highly stable ionized analyte ion that yields
little informative fragmentation in the ion source or for obtaining primary fragmentation
from a poorly activated [M+H]+ formed by a "soft" ionization technique such as FD.
Additionally, if a mixture of analytes was present in a sample, a product-ion scan of each
component coupled with CAD will yield only the structural information for each
individual component. Another important MS/MS scan is the precursor ion scan which
involves user-selecting the mass of a metastable fragment or CAD product ion and
scanning the MS/MS instrument to detect all the metastable or CAD product ions of that
mass that arise from all the respective decomposing precursor ions. Thus, the identity of
each precursor ion is reconstructed by the respective metastable fragment or CAD
product ion transitions because the precursor ions are impossible to detect directly.
Product and precursor ion scans can be utilized to obtain structural and mechanistic
information for the low-energy metastable and higher-energy CAD fragmentation
processes. In fact, a "fragmentation map" can be constructed and possible mechanisms
can be identified once all product ions from each precursor ion and all precursor ions
generating each product ion are deterrrrined. However, this "map" only suggests the
possibilities and should not be interpreted literally. A fragment ion produced by a low-
energy metastable transition may not be the identical mechanism by which the same
product ion is formed in high—energy CAD or in the "normal" mass spectrum because the
precursor/[M+H]+ ions in each case have different average internal energies which

ultimately determines which kinetically favorable mechanisms dominate.

WWW

Tandem mass spectrometry on the triple quadrupole instrument (QqQ) is straight

71

forward [156], so it will be introduced first. The RF-only quadrupole, q2, located
between the two mass analyzer quadrupoles, Q1 and Q3, is the collision region for CAD
or unimolecular decomposition region for metastable transitions. An RF-only quadrupole
simply transmits ions of all masses from end to end with high efficiency owing to its
excellent focusing properties. The "norm " mass spectral scan on the triple quadrupole
is achieved by setting qz and q3 to RF-only mode and scanning Q]. If collision gas is
present in qz, then low-energy CAD can be performed. If no collision gas is present, then
metastable transitions will be observed. Only low-energy CAD is possible on this
instrument because the accelerating voltage is less than 200 eV. However, it is very
efficient, meaning the conversion of precursors to product ions is high, since the scattered
ions are transmitted owing to the high focusing properties of the qg. For a product ion
scan, the precursor ion is mass-selected by Q1 and the metastable/CAD fragment/product
ions formed in q2 are mass-filtered by Q3. For the precursor ion scan, the
fragment/product ion of interest is mass-selected by Q3 and Q1 is scanned for each
precursor ion that yields that user-selected fragment/product ion in the collision region,
q2. One advantage of the triple quadrupole for MS/MS work is that unit mass can be
achieved for both the precursor/product ion selection and precursor/product ion scanning.
This is important for accurate structural and mechanistic work because Q1 only transmits
the analyte 12C isotopes of the analyte ion into q2 so only monoisotopic product ions are

detected. The double-focusing and TOF instruments do not have this built in luxury.

W

MS/MS on an EB double-focusing instrument is not as straightforward as on the
triple quadrupole. The plot of E versus B, and the example fragmentation
map/mechanism in Figure 16 should aid in understanding the "normal", product and
precursor scans on an EB instrument. For each of these scans, the accelerating voltage,

V, is held constant. Along the normal scan line, the m0+ ion represents the [M-I-H]+ ion,

72

while my’ and m2+ represent the fragment ions from m0+. A portion of the m2+ ions are
formed by a secondary fragmentation process from m1+. Since m2+, m1+, and mo+ ions
along the normal scan line in Figure 16 are produced in the ion source and have full
accelerating potential, they are transmitted through the electric sector, held constant at a
potential of E0. Next, these ions are mass-dispersed in B and individually focused onto
the detector at the appropriate magnetic field strength, Bn, during scanning. Thus, the
ions along this scan line appear in the normal mass spectrum. However, if during normal
operation an ion, mo+ undergoes a metastable transition after acceleration to form m2+ in
the first field-free region between the ion source and E, the fragment ion, m2+, will have
the same velocity, v0, as the precursor ion, mo+, but m2+ will most likely be filtered out in

E because its KE is less than that of moi: because the mass of m2+ is less than mo+. To

Metastable _(m2‘*‘)2 m2+ m1+ m0+
Transition "

  
   

 

E0

 

B/E
Scan Line

 

Electric Field (V)

 

 

 

 

Magnetic Field (T)

Figure 16. A plot of the B, E plane for linked scans at constant B/E and BZIE on double-
focusing instruments.

73

sample these metastable transitions in the first field-free region of an EB instrument, a
product ion scan can be performed by a linked scan at constant B/E [157, 158] in which
both E and B are scanned simultaneously, under computer control, holding the ratio
between the 8,, and En fields constant. In this scan type, shown in Figure 16, the
precursor ion, mot is user-selected from the "normal" mass spectrum, so the slope of the
linked-scan line at constant B/E can be calculated from the Boon ratio for moi”, where E0
is the potential of E at full accelerating voltage at which mo+ is transmitted through E,
and B0 is the magnetic field strength required to pass mo" onto the detector. Since m2+
has a lower KE than mo+, the E potential is lowered by scanning from E0 down to E2 so,
for example, m2+, with a lower KE than mo+ is allowed to pass through E as illustrated
by Equation 41. Likewise, m2+ has a lower momentum than mo+

50 = movoz/R and 132 = m2V02/R, SO M2 = mo/mz => 132 = szo/mo (41)

B0 = movolr and 8;; = mzvolr, so Bo/Bz = mo/m2 =9 B2 = szo/mo (42)

B/E ratio = Bo/Eo = constant (43)
but the same velocity, v0, so Bo needs to be lowered to B2 so m2+ can be passed onto the
detector as shown in Equation 42. Thus, for a linked scan at constant B/E, the slope of
the linear scan line in Figure 16 coordinates E, and Bn values for transmitting each
metastable ion of mo+ onto the detector. The scan slope is defined by the BolEo ratio in
Equation 43 or more accurately, Bo, since the user-selected mass is transmitted by this
field strength and E0 is fixed. The magnetic field strength, B2, in Figure 16, required for
transmitting m2+ to the detector in the linked scan at constant B/E is the identical field
strength in a "normal" scan required to transmit the mo+ —> my“ metastable transition to
the detector if it occurs in the second field-free region between E and B. Thus, for
metastable transitions after the electric sector, a classic broad peak(s) at an apparent m/z
value equal to the (m2+)2/m0+ ratio will appear in the "normal" mass spectrum.
Additionally, if the metastable transition mo+ -—) m1+ occurs in the first field-free region,

then by reducing E and B to the E1 and B1 fields, respectively, along the B/E scan line,

74

the m1+ fragment ion can be detected. Linked scans at constant B/E can be done for
fragment ions as well, such as m1+ in Figure 16, and, in this case, a peak corresponding to
m2+ would be detected along the m1+ B/E scan line according to the mechanism.

For the precursor ion scan on an EB instrument, a fragment ion, m2+, is user-
selected from the "normal" mass spectrum and a linked scan at constant B2/E, which has
a parabolic scan line extending from zero to m2+ in Figure 16, is performed to identify the
various precursor ions of mzf. The precursor ion scan is confusing because the actual
precursor ions of a metastable transition are not directly detected. Instead, the different
possible precursor ions of m2+ are represented by the metastable fragment ions from each
precursor ion that has the same mass as m2+. In Figure 16, the B2/E scan essentially
intersects the respective isobaric metastable ions in the B, E plane precisely at the B/E
scan slopes of each precursor ion. For example, in Figure 16, along the B2/E scan line of
m2+, the two broad peaks represent two metastable fragment ions equivalent in mass to
m2+ formed from the metastable decomposition of the two precursor ions, m1+ and mi:
In addition, the fragment ion m1+ has only one precursor ion, mof, in Figure 16, shown
by the single broad peak on the m1+ B2/E scan curve. In Equation 44, the Bn2/En ratio is

Bn = mnvo/r and E1] = mnv02/R =9 an/En = mnR/r2 (44)
proportional to the mass of the user-selected fragment ion mn+ if both Bn and En are
solved for v0 and set equal. The BZ/E constant for the parabolic scan line is calculated
using the user-selected fragment/product ion mass to determine the appropriate magnetic
field strength, B", from a "normal" magnetic field scan, assuming E0 is held constant.
Linked scans at constant B/E and B2/E can be also utilized for detecting CAD product
ions formed in the first field-free region assuming the collision cell is positioned between
the rear of the ion source and in front of the main slit as shown in Figure 14.

Poor resolution is obtained with linked scans at constant Bz/E, as indicated by the
broad metastable ion peaks on the B2/E scan lines in Figure 16. The wide peak widths

are proportional to the translational energy released (i.e., energy equal to the kinetic shift,

75

Eli) and the peak tailing originates from metastable transitions that occur during the late
stages of acceleration or in the early stages of the E sector. The resolution for linked
scans at constant B/E is at least unit mass because the NE scan line "cuts" these broad
peaks on the Bz/E scan line perpendicularly. Unfortunately, the resolution of precursor
ion selection for linked scans at constant B/E is low and becomes worse as the m/z value
increases since B is nonlinear and the m/z values converge. Depending on the m/z value,
a number of the analyte isotopic peaks and possibly a set of peaks from an interfering
analyte may unavoidably be user-selected. This will yield a more complex
fragment/product ion mass spectrum because the product ions will have isotopic peaks
and, if an interfering analyte exists, its mass spectrum will overlay the desired analyte
spectrum. A hybrid [159] and a four-sector mass spectrometer [160] are able to select the
precursor ion with unit mass so that only 12C isotopic product ion peaks appear in the
product ion mass spectrum. Until now, only first-generation CAD product ion spectra
have been discussed. With hybrid or four-sector instruments, second-generation product

ion mass spectra (i.e., MS/MS/MS) can be obtained [161].

W

The detection of metastable transitions and CAD (i.e., assuming the cell is located
between the ion source and flight tube entrance) fragment/product ions in linear TOF is
not possible because the fragment/product ions have the same velocity as the precursors
so both will be detected simultaneously. However, these fragment/product ions can be
detected if the TOF instrument is equipped with a reflectron. By scanning the reflectron
incrementally over small mass ranges, the fragment/product ions are sequentially focused
on the detector since they have different KEs than their precursors. The fragment/product
ions detected in each mass range increment can be assembled to form a composite
MS/MS mass spectrum. However, for large biomolecules such as proteins, the highly

efficient PID and SID techniques have the potential to activate the proteins more than

76

CAD and yield either more or entirely new products ions with better resolution, but
unfortunately, these techniques are much more difficult to implement. Recently, two
reflectron-TOF mass analyzers were linked to a PID laser excitation cell for high-speed

MS/MS applications [162].

If

[h.

CHAPTER THREE. K+IDS-BY-FAB TARGET DEVELOPMENT AND
APPLICATION TO MECHANISTIC STUDIES OF ION FORMATION IN
FAST-ATOM BOMBARDMENT (FAB)

I . Introduction

Fast-atom bombardment (FAB) has served the biological community well by
providing routine molecular weight and structural information for high molecular weight,
highly polar, thermally labile biomolecules. However, the availability of FAB preceded
an understanding of the chemical processes that yield gas-phase ions from condensed
phase molecules. Insight into the FAB desorption/ionization (D/I) mechanism would
allow for improvements in the technique for the analysis of compounds that are currently
difficult to analyze, and the evolution of new applications and new ionization techniques.
As described in Chapter two, a typical analysis by FAB involves dissolving analyte
molecules (M) in a polar, viscous matrix, usually glycerol (G), with a GM molar ratio of
100021, respectively, and subjecting the G and M mixture on a stainless steel target to 6-
keV Xe atoms for sequential desorption and ionization. Briefly, the desorption of intact
M and G molecules from either the collision cascade, thermal spike, or final phase
explosion events, in addition to the high rate of G molecule evaporation under vacuum,
all contribute to a high pressure region that exists above the FAB target called the
selvedge region. Here, in this high-pressure selvedge region, extensive gas-phase
ion/molecule chemistry similar to that in chemical ionization (CI) is substantiated by a
high rate of multiple collisions. Essentially, glycerol-CI conditions are created owing to
the large number of G molecules in the experiment. A small fraction of ionic species
related to glycerol, GF+, are responsible for initially protonating the glycerol molecules in

the selvedge region to form [G+H]+. These GF+ species are likely formed when a keV-

77

78

Xe atom collisionally activates glycerol molecules in the condensed or gas phases (see
Chapter two for specific reactions). In the selvedge, the CI reagent ions, [G+H]+,
subsequently either (1) protonate the desorbed intact neutral analyte molecules having
PAs greater than glycerol to form [M+H]+, (2) undergo subsequent collisions with G to
form the [nG-i-H]+ (n22) cluster ions, or (3) undergo extraction towards the mass
analyzer. Preionized analytes in the matrix may be desorbed directly into the gas phase
without the need for further chemistry in the selvedge region. Last, the protonated
molecules formed in the selvedge or the preionized analytes may unimolecularly
decompose in the free vacuum above the selvedge to form respective fragment ions.

In this chapter, the FAB ion formation mechanism described above will be
investigated by injecting low—energy K+ ions into the selvedge region of the FAB
experiment to probe the types of neutral species present there following desorption. This
work is an outgrowth of potassium ion ionization of desorbed species (K+IDS), explained
fully in Chapter two, which takes advantage of the fact that K+ ions only form adducts
with gas-phase molecules without inducing fragmentation as in El and CI. Thus, this
capability will allow the identification of desorbed neutral molecules by their respective
[M+K]+ adduct ions. Since we use K+ here to ionize neutral molecules desorbed by keV
Xe in the FAB experiment, we refer to this technique as K+IDS—by-FAB [163]. This
chapter was written to supplement the previously published K+IDS-by-FAB paper [124]
in Appendix one with additional studies and data not in that publication.

In K+IDS-by-FAB, the K+ ions are injected into the selvedge region using a
spatially-separated, two-stage FAB divided target as illustrated in Figure 17. One half of
the divided target is the K+ alurrrinosilicate glass emitter, identical to the emitter used in
the K+IDS experiment, and the other target half is the sample holder for the analyte and
matrix as in normal FAB. In K+IDS, when the K4“ alurrrinosilicate glass emitter is heated
to high temperatures, a large flux of K+ ions are therruionically emitted into the gas

phase, but this same K+ glass emitter will also emit copious amounts of K+ ions when

79

Mass Analyzer
/
I [G+K]+ “(MD-H 291+ ®/
+
Free Vacuum [M+K]+ [M+H]"’ [6:HL_ _ _ 7 ® /,
, —-— , ’ [M+H]+ ®
‘:_____/ [G+K]+ I; (36%
A [G+H]++G M Fast Atoms
Selvedge Region K G
K-I-

  
   

Matrix / Analyte

 

K+Glass Emitter

Figure 17. A K+IDS-by-FAB divided target under Xe bombardment demonstrating the

injection of desorbed 1K" ions into the selvedge region for the identification of desorbed
neutral molecules.

bombarded by keV-Xe atoms in FAB W [164]. The FAB beam
simultaneously impinges both target halves desorbing K+ ions from the K” glass emitter
target half, and desorbing intact analyte and glycerol molecules, preionized analyte
molecules, and ionic glycerol species (G1:+) from the analyte/matrix target half as in the
normal FAB experiment. The desorbed K+ ions may be manipulated by using both
electrostatic extraction fields in the ion source and divided target geometries, swept
across the divided target barrier in the gas phase and injected into the selvedge region
above the analyte/matrix target half. These injected K+ ions will ionize the various types-
of desorbed M and G neutral molecules present in the selvedge to form respective
[M+K]+ adduct ions which are then used for the identification of these gas-phase

molecules. A divided FAB target is used to separate the analyte/matrix from the K+

80

emitter to prevent condensed-phase interactions between the two targets and ensure only
gas-phase chemistry occurs in these FAB mechanistic studies.

Since K+ ions are injected into the selvedge region above the analyte/matrix
target, some further comments should be made on the characteristics of this region. To
date, no direct experimental evidence has definitely proved the existence, the actual high
pressure and the dimensions of the selvedge region above the FAB target. The existence
of a selvedge region is validated by: (1) the independence of Xe projectile energy on
types of ions formed in contrast to those from El and CI, (2) the observation of
protonated intact molecules in the mass spectra, and (3) their correlation with established
proton affinities (PA). The high local pressure in the selvedge region is expected based
on high vapor pressure of glycerol, 0.13 millitorr, under vacuum, and the high sputtering
yield estimated at greater than 1000 glycerol molecules per incident keV Xe atom from
the liquid target [107]. A high density of gas-phase neutral molecules exists near the
target-vacuum interface; these molecules probably undergo expansion into the free
vacuum resulting in an exponential pressure gradient that extends away from the target.
Calculations by Pachuta and Cooks [117] have indicated that within a volume of 1010 A3
the pressure exerted by 10 particles with 2 eV of energy is about 1 torr, a pressure able to
sustain CI type collisions and chemistry. Thus, if an area of 106 A2 receives one keV
particle impact per second and 10 or 1000 particles are desorbed to maintain a pressure of
l torr, then the height of this selvedge region is 1 um and 100 um, respectively. The
most useful dimensions of the selvedge region extend out beyond the target in a
decreasing exponential pressure gradient to the point where ion/molecule chemistry
ceases. Therefore, if K+ is injected above the FAB target and an [M-i-K]+ adduct ion is
formed, then this can be interpreted as evidence that K+ ions sampled a region of space in
which the pressure of the desorbed neutral molecules were sufficiently high that a
collision occurred in the time period of a few microseconds, corresponding to the lifetime

of K+ ions in the ion source.

81

2. Development and Characterization of the K +IDS-by-FAB Target

The function of a divided target in a K+IDS-by-FAB experiment is to inject K+
ions into the selvedge region to sample desorbed neutral molecules. Various divided
targets with minor geometry differences were designed and assembled with intuition
which resulted in mostly unsuccessful results. In a more rigorous attempt, SIMION, an
ion optics design and analysis computational tool, was used to assist in the development
of a successful K'I’IDS-by-FAB target. These computational experiments revealed that
the success of a K+IDS-by-FAB experiment depends essentially on the divided target
design and geometry, and how well the divided target interfaces with the extraction
potentials in the ion source of the mass spectrometer. Only after gaining insight to the
K+IDS-by-FAB experiment with SIMION computer modeling was a functional divided
target able to be designed and assembled for routine and reproducible use.

All K+IDS-by-FAB experiments were conducted on the JEOL HX-l 10 mass
spectrometer using the field desorption/fast-atom bombardment (FD/FAB) ion source and
operating at 10 kV. The K+IDS-by-FAB target was placed on the end of a direct
insertion probe and entered into the instrument in an identical manner as a regular JEOL
FD/FAB target. The ion source temperature was 25° C and the ion source operating
pressure with FAB Xe gas on was 5x10’6 torr. In this section, my experiences and

insights for designing a successful K+IDS-by-FAB divided target are discussed.

Generation of KI” Ions.

The use of a divided target for the K+IDS-by-FAB experiment requires the keV-
Xe beam to simultaneously impinge both the analyte/matrix on one target half to create
the FAB experiment, and a solid surface on opposite target half that yields gas-phase K+

ions. These K+ ions are then injected in the selvedge region above the analyte/matrix

82

target half. The method of choice for generating K+ ions in a K+IDS-by—FAB experiment
is bombarding a K+ alurrrinosilicate glass emitter surface held at room temperature in
close vicinity of the analyte/matrix target. The FAB mass spectrum of such a K+ glass
emitter surface, comprising one target half on a divided target, is shown in Figure 18a
without any analyte/matrix present on the opposite target half. Bombardment of the glass
emitter surface desorbs one dominant ion into the gas phase, K+, with an ion current
proportional to ~66 counts out of 100 possible. Thus, K“ is the only ion injected into the
selvedge region which provides for a straightforward K+ CI experiment identical to
K+IDS. The K+ glass emitter is permanently attached to the divided target so it can be
used for many experiments. However, it will eventually crack from careless handling
before K+ is ever depleted. Depending on the divided target design the K+
aluminosilicate glass emitter is prepared either by melting the tertiary rrrixture of finely
ground 2KNO3:1A1203:28102 directly on a divided target surface half with a Bunsen
burner, or by dipping and coating the end of 0.007 in. diameter rhenium wire with a
"head" of the molten 2KNO3:1A1203:28102 liquid prepared in a platinum dish by heating
with an oxygen-acetylene torch [52]. In the latter case, the K+ emitter bead is the target
surface half from which K“ is desorbed and the Re wire holding the K+ emitter bead is
spot welded to the appropriate site on the divided target. In both cases, the final molar
composition of the K+ glass emitter is 1K20:1A1203:2S102 after heating since N02 gas
is released. Often times, for direct application to a divided target surface, the spattering
of the Kf glass from N02 release during melting can potentially contaminate the
analyte/matrix divided target surface. This nuisance is elinrinated by first melting the K+
glass in a platinum dish to release all N02, cooling, re-crushing the solid glass into a fine
power with a mortar and pestle, and then melting a portion on the divided target surface.
The more common way to generate gas-phase K+ ions is to bombard a KCl film
dried on one target half from a saturated solution of KCl in water. The FAB mass

spectrum in Figure 18b shows bombardment of a divided target with one target half

83

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100‘ 39
If! . a) K + K+ glass emitter (66 counts K“)
a 80‘.
E 1
IV I
e 601
1 I
l‘ 40:
e
g l
201 +
i j Na+ (5:;
y 0‘ 213; I .__L_ I
0 50 100 150 200
m/z
100‘ 39
If? . b) K+ KC] solid emitter (74 counts 10*) I
a 80‘ l
i I I
{I . r
I ‘ '
{I 40; [(KC1)+K]+ Z
c ‘ 113
l : 1
l 201 2K, [2(KC1)+K1+C
Y ()1 l 713 1?: '
(3 50' T 150' ‘T 150 T r260
m/z
100
R j C) 93 Dual-surface space-divided target :
f . [G+H]+ with KCI emitter on one half and -
f1 30: glycerol on the other. f
i : ;
e 60: [(ZG+H)-H20]+ T
l i 16712ci+m+ :
g 40: 185 -
n .
S . 10 ’
l , [G+K]+ X .
ty 201 31%;; ] 131 [20+K]+ [33;5‘1“;
1.331% ...... we ...23_3.....J...
100 150 200 250 300
m/z

Figure 18. a) FAB mass spectra of K+ glass emitter and of b) KCl solid film on a divided

target half without matrix or analyte present on the opposite half. c) K+IDS-by-FAB mass
Spectrum of a KCl emitter and glycerol on either half of a dual-surface space-divided target.

The latter spectrum is presented to show KCl is just as effective as K"' glass emitter in a
R‘I'IDS-by-FAB experiment, but the KC] emitter is more difficult to apply cleanly to only

one target surface plus it needs to be regenerated for each run.

84

coated with KCl without analyte or matrix on the opposite half. A strong K+ ion current
proportional to 74 counts is obtained but the presence of [(KCl)n+K]+ cluster ions may
undergo undesirable gas-phase chemistry with analytes because they will also be injected
into the selvedge region and/or they may spectrally interfere with analyte peaks. In
addition, the KC] solid surface needs to be regenerated for each experiment to insure
integrity of the results which is quite tedious considering the small size of the target
halves and high probability of contaminating the analyte/matrix target half from
spattering.

The sensitivity obtained from the reactants mixed on single stage target are much
greater than the reactants separated on a divided target. The most popular method used to
generate K+ adduct ions of analyte molecules with much higher sensitivity than K+IDS-
by-FAB is to place KCl or KI salts directly in the analyte/glycerol matrix and perform
FAB. The high sensitivity achieved with this method is great for molecular weight
confirmation studies, but not for mechanistic studies because it is not certain whether the
[M+K]+ ions are desorbed intact from the matrix or formed by gas-phase ion/molecule
chemistry similar to KflDS-by-FAB. However, KflDS-by-FAB has some analytical

advantages over FAB when KCl salts are used which will be discussed in Chapter five.

Previous Experiments with Divided Targets.

The first type of K+IDS-by-FAB experiment involved bombarding a single-stage
target shown in Figure 19a in which the K+ glass emitter was melted on a stainless steel
surface and the analyte/glycerol was placed on top of the glass [164]. As a result,
abundant [M+K]+ adduct ions were detected for analytes, such as fructose, which do not
respond well in standard FAB, with detection limits 2-3 orders of magnitude lower than
the [M+H]+ ions formed from FAB of fructose in glycerol. Apparently, glycerol leached
Ki" ions from the K+ glass emitter, and the K+ adducts of the analyte were formed either

by desorption of intact [M+K]+ ions from glycerol, or in the gas phase by Kf CI of

85

3)
mm.
V/l/l/l/l/l/l/l/A

c) Top Top

      

 

 
   

Ke d) ‘\
y ‘\\\\ \ \\
K+ Glass Emitter ..‘\\\\\\\\\\\\\\\\\“\

- Analyte/Matrix

   

 

 

 

 

Figure 19. a) The first K+IDS-by-FAB type target employing a K+ glass emitter reported
by Akerrnann et al. [164], b) the wall and slit targets reported by Munster et al. [166, 168],

c) the dual-surface target reported by Miller et al. [169], d) the K+IDS-by-FAB "set screw"
target reported by Schultz et al. [172], e) the K+IDS-by-FAB "wall-divided" target reported
by Kassel [171], f) the K+IDS-by-FAB "K+ bead pop-up" target, g) the K+IDS-by-FAB
"vertical arrangement" FD/FAB type divided target, and h) the vertical, dual-surface space-
divided target, all which provided unsatisfactory gas-phase K+ adduct ion formation for
mechanistic studies.

86

desorbed intact analyte neutral molecules. Since the K+ and analyte were not separated in
the condensed phase, no mechanistic conclusions could be made. Analytically, the
analysis of fructose was straightforward and yielded only [M+K]+ adduct ions, but less
desirably cholic acid (MW 408) yielded both [M-i-K]+ and [M-H+2K]+, which is the
potassium adduct ion of the potassium salt of cholic acid. This study was an outgrowth
of an archaeological study involving 19th century ceramic pottery in which elemental
compositions and lead isotope ratios were obtained by direct keV-Xe bombardment of the
ceramics attached to a standard FAB target. A comparison between the ceramic pottery
and the National Bureau of Standards "common" lead glass led to the identification of
their geographic origin [165].

The use of wall- and slit (space)-divided FAB targets shown in Figure 19b are
utilized by FAB practitioners for exact mass measurements in which the analyte/matrix is
placed on one target half and a reference compound such as C51 is placed on the opposite
target half to reduce the number of prepared samples and to prevent unwanted
sample/matrix effects from mixing both the analyte and the reference compound together
in the same matrix [166]. As a result, exact analyte masses of the reference and analyte
can be measured with high precision and accuracy because the reference target half can
be bombarded individually before and after analyte bombardment, on the opposite target
half, by simply rotating the respective divided target half in the Xe beam. Divided targets
were also employed previously in conjunction with D/I techniques to explore gas-phase
ion formation in the high pressure selvedge region. Chiarelli and Gross [167] studied
gas-phase Na+ cationization of sucrose in the selvedge during laser desorption using a
divided target with NaCl and sucrose spatially-separated by 30 pm to ensure the laser
irradiated both halves. At low laser powers of 106 W cm'1 , they concluded that Na+ ions
and neutral species were desorbed from both target surfaces in the thermal spike regime
and a long-lived selvedge region existed above both targets since [M+Na]+ was observed,

but at higher laser powers, 1010 W cm'z, no [M+Na]+ ions were observed owing to a

87

short-lived selvedge region in which the desorbed Na+ and sucrose molecules had high
translational energies and little opportunity to overlap spatially and/or undergo three-
body collisions for stabilization. Munster et al. [168] used the wall- and slit- (space—)
divided targets in Figure 19b to show that two different matrices A and B placed on
opposite target halves undergo ion/molecule reactions in the gas phase during Xe
bombardment of both target halves, similar to a CI experiment, to form [A+B+H]+ ions.
Miller et al. [169] placed a volatile crown ether in 3-nitrobenzyl alcohol (NBA) on the top
half of the divided target, shown in Figure 19c, and placed either KCl salts in NBA or
solid KCl deposited from solution on the bottom target half, which was spatially
separated by 0.6 mm from the top target half. The bottom potassium salt target was
selectively bombarded bottom with FAB to demonstrate that gas-phase [crown+K]+
adduct ions could be formed analogous to [M+H]+ in CI processes such as protonation of
gas-phase analyte molecules or proton exchange between a protonated molecule and gas-
phase neutral analyte molecule with a higher gas-phase basicity (GB). Michaud et al.
[170] placed two different analytes, A and B, in glycerol on opposite divided target
halves separated by a 200 um gap and did not observe the expected [A+B+H]+ adduct
ions unless A was an alkali metal ion. This experiment indicates only single-site,
electrostatic interactions are formed in the gas-phase. The [A+B+H]+ adduct ions were
only observed when both A and B analytes were mixed in glycerol on the same target
which indicates the multiple ligand-ligand interactions either occur in the condensed-
phase or possibly in the selvedge region. Since they claim three-body collisions in the
gas-phase are not likely, [A+B+H]+ may be desorbed preionized from the matrix, [A+H]+
may be desorbed preionized and react with desorbed B in the gas-phase selvedge, or the
[AB] cluster may be desorbed intact and become protonated in the selvedge.

In each mechanistic study above which employs a divided target for the study of
gas-phase ion formation in FAB, there were many experimental difficulties and

precautions that needed to be taken into account to ensure the observed product/adduct

88

ions were formed from gas-phase ion/molecule reactions and not from those in
condensed-phase chemistry. Condensed-phase cross-contamination between the two
target halves as a result of condensation, splattering, sputtering and/or creeping of
reagents to the opposite target halves under keV-Xe bombardment had to be checked
after each run. Usually, each reactant on each divided target half were individually
analyzed by FAB for traces of the opposite reactant. If the condensed-phase reactants
originated from separate divided target halves, then the ion currents of the product/adduct
ions formed in gas-phase are much lower than the ion currents obtained when both
condensed-phase reactants are mixed together on the same target stage. Depending on
the divided target geometry, ions from one of the target halves may be preferentially
formed/detected if keV-Xe bombardment is skewed and/or if the potentials on the ion
source ion optics favor extraction from the "right" or "left" target. Moreover, just the
right ion optical tuning parameters are needed to draw out the ions formed in the gas-
phase above both target halves.

Kassel in 1988 [171] and Schultz in 1989 [172] performed preliminary K+IDS-by-
FAB experiments on the JEOL HX-l 10 mass spectrometer using the wall-divided target
in Figure 19c and "set-screw" target in Figure 19d, respectively, for the initial
investigation of the FAB gas-phase ion formation mechanisms. The wall in the wall-
divided target prevents cross-contamination of the K+ glass emitter by glycerol during
bombardment to ensure only gas-phase [M-I-K]+ adduct ion formation. The "set-screw"
target allowed the height of the K+ glass emitter, located on top of the set-screw, to be
adjusted in an attempt to probe the height of the selvedge region while keeping the
glycerol at the base of the set-screw separate from the emitter. Irreproducible K+ adduct
formation and unpredictable condensed-phase glycerol contamination of the K+ emitter
were major problems for both of these early K+IDS-by-FAB targets. As discovered later
in SIMION modeling, the wall- and set-screw target designs geometries did not interact

well with the characteristics of the ion source preventing formation of [M+K]+ ions.

89

KHDS-by-FAB Target Development-The Next Generation.

The current K+IDS-by—FAB work began by completely characterizing and
understanding the simple wall-divided target beyond what was done previously. The wall
of the wall-divided target was constructed using stainless steel or platinum ribbon
(~0.007 in. thick) formed as shown in the center - w — and spot welded to the
center of a standard JEOL FAB target. With the analyte/glycerol and the K+ glass emitter
placed on either half of the wall-divided target, [M+K]+ adduct formation with desorbed
neutral molecules was usually irreproducible and unpredictable, and often times, there
were spectral and visible signs of analyte/glycerol contamination of the K“ glass emitter.
In addition, optimization of the potentials of the ion optics for the extraction of the K+
adduct ions formed in the gas-phase above the wall region was not straightforward, and
usually ions were extracted from either target half instead resulting in either an
analyte/glycerol or K+ glass emitter FAB mass spectrum. The wall height was
incrementally lowered from 1.5 mm to 0.1 mm (the lowest practical limit with simple
magnification), similar to the "set-screw" target, without enhanced performance which
indicated at the time that perhaps the selvedge region was on a smaller scale, it did not
exist, or K+ ions were not successfully being injected into the FAB selvedge region above
the analyte/matrix with this target. In Figure 19f, using the same concept as the "set-
screw" target, the Re support on a K+ "bead" emitter was spot-welded to the bottom of
the target in such a way that it emerged from a hole drilled through a standard JEOL FAB
target. The analyte/matrix was placed on the JEOL target surface surrounding the K+
emitter bead. While this K+ emitter was flexible for adjustment purposes, this target also

exhibited the same drawbacks as the wall-divided and set-screw targets.

llM': "CII+EIE'-

Following these initial studies, the first problem area in K+IDS-by—FAB to receive

9O

attention was the condensed-phase cross-contamination of the K+ glass emitter by
analyte/glycerol. In Figure 20a, the normal FAB mass spectrum of glycerol is shown in
which the peak at m/z 93 represents protonated glycerol, the peaks at lower mass
represent the fragment ions of glycerol and the peaks at higher mass represent the proton-
bound cluster ions of glycerol (i.e., m/z=92n+1). In Figure 20b, a typical K+IDS-by-FAB
mass spectrum of glycerol using the wall-divided target is shown in Figure 1%. The
appearance of a peak representing K+ suggested that K+ was successfully injected into
the selvedge region, and the intense peaks representing the K+ adduct ions of glycerol
indicated there was a high conversion of desorbed glycerol molecules into K+ products.
However, the peak at m/z 169 represents [G-H+2K]+, the potassium adduct ion of the
potassium salt of glycerol, and its appearance is indicative of a condensed-phase reaction
between glycerol and the K+ glass emitter surface under bombardment in which glycerol
is sputtered or creeps from the opposite target half onto the K+ emitter. If the 10' glass
emitter is purposely contaminated with glycerol, an identical mass spectrum is obtained
as in Figure 20b. Most likely, the condensed-phase chemistry also contributed to the
observed intense [G+K]+ peak since these ions may be desorbed preionized from the
contaminated K+ glass emitter. Thus, if [G-H+2K]+ or [M-H+2K]+ peak appears in the
KflDS-by-FAB mass spectrum, then K+ adduct ion formation cannot be guaranteed to be
solely result of gas-phase ion/molecule reactions. In addition, after each KflDS-by-FAB
experiment, the divided target was examined under a microscope for visible signs of K+
emitter contamination by glycerol. This method is less reliable for identifying small
amounts of contamination but it does allow some glycerol flow paths to be identified.
Therefore, to solve this contamination problem, the K+ glass emitter cannot be physically
attached (i.e., spot welded) to the same surface as the liquid analyte/matrix because under
bombardment, glycerol "creeps" around, up and/or over the barriers separating physically
connected surfaces. In the "set-screw", wall-divided, and "pop-up" K+ emitter bead

targets KflDS-by-FAB targets in Figures 19d, e, and f, respectively, the set-screw, the

Ill
REla‘t.I\ .f\ I‘ll“.

Rela\rlv e Intens.11n\vc

91

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

    

1
R . 93
f 1 a) [G+H]+ 126+HI+ 9“
i1 801 185 HO-CHz-CH-Cflz-OH
i, 60-: Glycerol on FDIFAB target
it 4o-j I<G+H)-H201+
E ; [(G+H)-2H20]+ 75
3 20; 45 57 [3G+H]+
i . 277
i o - e r J A 1
o 50 160 150 260 ' '250' ' - '300
m/z
100
R . .
1° 1 b) [IGLKP‘ K+ emitter contaminated I
p 80‘ by glycerol -
260‘ i
1 . IG+H1+ [2G+H]+ '
{I 404 93 185 [2G+K]+ .
C ,
n K+ I
f 20 39 75 r
I I l 57 J I
Y O 6 L ‘ A "100
m/z
1
R 00; C) 93 KflDS-by-FAB of glycerol on
f 80; [G+H}+ KHDS-by-FAB space-divided target
f : with Loinstarred
1 .
V - +
c 601 [(G+H)-H20]+ [233?“
l, 40; [(G+H)-2H7_O]+ 75
g ‘ 45 57
f 20‘. ‘3‘; [ZG+K]+ [36+H}+
ty 0: 223 277
0 50 100 150 200 250 300

m/z

Figure 20. a) FAB mass spectrum of glycerol (MW 92) matrix, b) typical K+IDS-by-
FAB mass spectrum when the 10‘ glass emitter on any divided target is contaminated
with glycerol/matrix, and c) KflDS-by—FAB mass spectrum of the glycerol matrix

obtained with the K+IDS-by-FAB space-divided target and L0 lens element installed in
the ion source.

92

wall, or the Re wire holding the K+ emitter bead are all physically attached and/or spot
welded to the same surface that glycerol resides allowing the K+ emitter to be easily
contaminated despite the barrier. For the less obvious glycerol contamination of the K+
emitter on the "pop-up" target in Figure 19f, glycerol placed on the target surface creeped
up the Re wire to the K+ glass emitter and contaminated the base of the K+ glass emitter
bead where the FAB beam was likely centered and/or gas-phase glycerol from desorption
or evaporation may have condensed on the K+ glass emitter since it was positioned

surrounding a pool of glycerol on the target surface.

K" _ - _ . . n

To alleviate the contamination problem, the K+ glass emitter was physically
isolated from the analyte/matrix target surface which solved the contamination problem
in the next generation of KflDS-by-FAB targets. The divided target in Figure 19g was
constructed using Miller's target design in Figure 19c in which an "open space" or slit
separated the K+ glass emitter bead and the analyte/matrix target. However, this
particular divided target had the same undesirable characteristics as Miller's target
because ions could only be extracted from either the upper or lower target halves but not
from a point in between. Either the Xe beam could only effectively irradiate one target
half at a time owing to the target geometry, or the absence of horizontal deflection plates
in the source ion optics early in the ion extraction process, while the ions have low
translational kinetic energies (KE) resulted in the inability to control, shape and focus the
ions into a well defined beam. Since the initial ion optics in the JEOL HX-110 FD/FAB
ion source are vertical deflection plates, Miller's dual-surface divided target was rotated
90° as shown in Figure 19h and initial success was obtained as shown by the K+IDS-by-
FAB mass spectrum in Figure 18c where the KC] solid film, dried from solution, and
glycerol were placed on separate target halves. Here, the [G+K]+ adduct ions are

produced without the contarrrination-indicating [G-H+2K]+ peak at m/z 169. However,

93

as mentioned above, the KC] surface requires regeneration for each experiment so
instead, the K+ glass emitter was applied to the same dual-surface target and similar
success was achieved. Better results were obtained when a K+ glass emitter head was
positioned next to the analyte/matrix stainless steel surface of a modified FD/FAB target
as shown in Figure 21a because K+ now could be injected into the selvedge region above
the analyte/matrix target at a multitude of angled trajectories creating better spatial
overlap over a wider region than K+ injection from a flat, tilted potassium surface in
Figure 19h. This new target design is referred to as the KflDS-by-FAB space-divided
target, and its orientation in the Xe beam and dimensions are shown in Figures 21b and
21c and d, respectively. Most importantly, the K+ emitter contamination by glycerol was
eliminated because the Re wire holding the K+ bead was spot-welded to the target post
underneath the sample holder where glycerol was never found.

The next set of experiments optimized the sample holder and K+ glass emitter
geometry, the 10‘ emitter composition, and the types of matrices used with this K+IDS-
by-FAB space-divided target in order to improve the K+ adduct ion formation, and the
reproducibility of the KfiDS-by-FAB experiment. For the K+IDS-by-FAB target in
Figure 21c, the optimum K+ emitter bead position was ~0.3 rrrrn and ~1.3 mm from the
bottom and top of the angled sample holder, respectively. In the top view of the K+IDS-
by-FAB target in Figure 21d, the K’r emitter was best positioned next to the top half of
the sample holder. If the K4“ glass emitter becomes too close to the matrix/analyte surface
or the angle of the sample holder becomes too steep, then either condensation, as
explained above, or splattering of condensed-phase glycerol onto the K+ glass emitter
under Xe-keV bombardment may occur resulting in K+ emitter contamination. The
optimum angle for the sample holder is ~35° or lower because angles of 40° or more
resulted in K“ emitter glycerol contamination more frequently. Increasing the yield of K+
ions in the K+IDS-by-FAB experiment will undoubtedly generate more K+ adduct ions of

desorbed neutral molecules. Generally, more K+ and [M+K]+ adduct ions were

94

   

. Analyte/Matrix
Sample Holder

\-
--:.-...-.-:.-..-:.-.-:-.

Figure 21. Different views of the a) KflDS-by-FAB spacedivided target with the
K+ glass emitter bead and analyte/matix sample holder, b) its orientation with the

impinging FAB beam and exiting secondary ions, and its typical c) side and d) top
dimensions.

gle

~§
v

su
1(

DC

95

generated when the K+ emitter bead was larger in diameter and the composition of the K+
glass emitter was a soft, granular white glass rather than a hard, smooth clear glass. The
soft, granular white or hard, smooth clear glasses were obtained by either minimum or
maximum additional heating, respectively, during final "shaping" of the initial "chunk" of
K+ glass, which was captured from the molten glass liquid in the platinum dish, into a
"bead". For example, on average, at least 10-20 counts of K+ were detected with the K+
glass emitter bead on the K+IDS-by-FAB space—divided target in Figure 21a in contrast to
~5 counts average for the other targets in Figure 19 because the former target has a
superior geometry. K+ emitters with the molar compositions, 5 or
10KNO3:A1203:2S102, were made, but increased K+ from these K+ emitter beads was
not observed. In addition, 50% more K+ ions were detected from keV-Xe bombardment
of the K+ alurrrinosilicate glass emitter than from the K+ borosilicate glass emitter, which
was prepared with the same molar ratios as the aluminosilicate emitter. All K+IDS-by-
FAB experiments to date utilized glycerol as the matrix applied to the sample holder
because it is compatible with a wide variety of analytes and it kept the number of
additional experimental variables to a minimum. However, other matrices in
combination with K+IDS-by-FAB were tested for enhanced K+ adduct formation with
desorbed neutrals. Thioglycerol was contaminated with sodium and yielded more Na+
analyte adducts ions than K't. The high GB of triethanolarrrine resulted in only K+ adduct
ions of this matrix and no K+ adduct ions with analytes. Nitrobenzyl alcohol was not
compatible with all the analytes and usually resulted in little if any K+ adduct ion
formation. Thus, glycerol was considered to be the optimal matrix for KHDS-by-FAB
since it is salt free and yields the expected K+IDS-by-FAB mass spectrum with a variety
of analytes with minimal spectral interferences.

The early KflDS-by-FAB experiments with the K+IDS-by-FAB space-divided
target were still somewhat unpredictable and notorious for yielding low levels of K+

adduct ions similar to the previous target designs, but at least gas-phase K+ ionization of

deso
were
the n
K+ a
K‘II
Show
respe
(MW
analy

3116111.

96

desorbed molecules was observed without contamination products. Successful results
were more often obtained when the analyte to glycerol ratio was increased which lowered
the number of ionizing glycerol species in the CI environment of the selvedge favoring
K+ adduct formation over protonation. In Figure 22a and b, the FAB and a successful
K+IDS-by-FAB mass spectra of the tripeptide, H—LGG-OH (MW 245), in glycerol are
shown using a standard JEOL FAB target and the K+IDS-by-FAB space—divided target,
respectively. Additionally, in Figure 23, the K+IDS-by-FAB mass spectrum of sucrose
(MW 342) is shown using the K+IDS-by-FAB space-divided target. Unfortunately, both
analytes are contaminated with sodium which indicated sample preparation needed
attention.

Milligram amounts of the compounds used in K+IDS-by-FAB were dissolved
either in milliliter amounts of pure Milli-Q® water, methanol, or a 1:1 mixture of both to
make solutions with final concentrations ranging from 2 ug/uL to 10 ug/uL. Glass vials
with cap liners were used initially but after it was deduced that the analytes complex to
N a+ present in the glass, plastic vials were used. To ensure run-to-run consistency, a
small drop of spectroscopically pure glycerol was captured on the end of a Re wire (0.007
in.) and applied/coated to either the K+IDS-by-FAB or FD/FAB target over a large
surface area. One microliter of the analyte solution was transferred to the analyte/matrix
target and mixed with the glycerol. A microscope (x2 power) assisted in the accurate
placement of the matrix and analyte on the K+IDS-by-FAB target to prevent
contamination of the K+ glass emitter. After the analysis, the K+IDS-by-FAB target was
heated briefly in a Bunsen burner flame to remove the sample from the target surface and
the K+ glass, if contaminated. A gentler approach to renewing the K+IDS-by-FAB target
was later adopted to save the life of the K+ emitters and to ensure the removal of sodium
and potassium salts from the target surfaces. It simply involved rapidly dipping the target
in methanol, acetone, and finally hexanes. Despite the early successes with this new

target design in eliminating glycerol contamination of the K+ glass emitter and providing

97

 

 

 

 

 

 

 

 

 

 

 

 

100
e 1 a1 [M+H]+ '
L 801 H-LGG-OH .'
t b
i , FAB [M+Na]+ .
g 60; 318 f
I 1 i
n 40. .
t ‘ .
C
3 ‘ iIG+H1+ b2 "10 '
i 20: o W [a3+H]+‘ i
t 44 93 133 171 201 i
y I 30' I57 7'5“ 112 I I | 23o| .

OJ “rLL ' AA I “L' JV V r 1
o 100 200 300 1:1/00
2

100
R . 39 86 '
f I b) K+ (11 i
a 89, H-LGG-OH T
i 3 KflDS-by-FAB IM+H1+ .
y . 246 -
I j '
n .
a 1 T
2 j G b2 [M+Na]+ I
l 20: Na+30 [G+1~In]+ly323 171 [a 3+H]+’ 268IM+K1+ :
y I 23 11 201 230 284 .

I e
o« '
o 100 200 300 :30
2

Figure 22. a) FAB mass spectrum and b) early KflDS-by-FAB mass spectrum of H—LGG-

0H (MW 245) in glycerol on the standard JEOL target and KflDS-by-FAB space—divided
target, respectively.

98

as

. mx

  

. 372. c mNm
.+ Lazmsm me +~ONIII+SE

- +5.45

 

com

 

3:
+5.63

m3

CE

m2

me_ 52 L912

:59
mm

 

gowns eoE>€oommm
minced—b. 2: mean .ccccbm 5 Ag 35 secs .6 538% as: $139.0. ram .9 came

mm

mm

-o
. an
. u
. m
. ON m
. a
o
. a
. ow :
. _
. o
.8 >
A _
. u
. a
tow _
n o
. ~—
noo—

 

 

99

definite K+ adduct ion formation with glycerol and analyte molecules, a more
reproducible and predictable K+IDS-by-FAB experiment was still desired in addition to

stronger K+ adduct ions of desorbed neutrals.

W+W

A set of experiments was conducted to investigate the possibility of increasing the
yield of K+ ions from the K+ glass emitter in the KflDS-by-FAB experiment by heating
the K+ emitter similar to K+IDS in addition to Xe bombardment. However, only low
heating currents could be used to heat the K+ emitter because if the K+ emitter became
too hot too quickly, it would radiatively heat the analyte/matrix target and induce
unwanted thermal desorption and/or degradation of the analyte. These experiments were
challenging to implement on the JEOL HX-l 10 with kV-accelerating voltages because
the K+IDS power supply unit is designed to operate in conjunction with quadrupole
instruments at low accelerating voltages [173]. However, the K+IDS power supply is
best for these experiments because it allows fine control of the heating current directed
through the K+ glass emitter to prevent possible analyte thermal desorption and/or
degradation while also providing circuitry for biasing the K” bead relative to the potential
on the FAB target. The accelerating voltage on the HX-110 was reduced to 4 kV in order
to prevent damage to the K+IDS power supply. In Figure 24, the input for the reference
voltage power supply in the K+IDS power supply unit was connected to the 4 kV:t50 V
repeller reference site #5 in the surge filter of the HX-110 in order to bias the K+ emitter
and the heater circuitry equivalent to the accelerating voltage. In FAB on the HX-l 10,
the repeller voltage, i50V of accelerating voltage, is normally applied to the FAB target.
This potential is measured at reference site #10 in the surge filter, but the EI/CI repeller
voltage at reference site #5 tracks the FAB repeller at #10 exactly and was used instead
because it was remote from the FAB target. The heating current outputs from the K+IDS

power supply unit were connected to the EI/CI filament heating current reference sites #3

Figure
FABC

Ion Source Housing of JEOL HX-l 10

 

  

#3 EI/CI Filament
#4 EI/CI Filament
#5 EI/CI Repeller
#6 Source Chamber
#7 EI/CI Trap

#8 Thermocouple

#9 Thermocouple

#10 FD/FAB Repeller
#11 Accelerating V

     
   

FD/FAB Ion Source (4 kV) : (7)

7

 

 

 

 

 

Surge Filter (4 kV)
3 4 5 6 7 8 91011

f‘?OOOOOO

 
 

 

 

Reference
I Power Supply
0-20 V

 

Heater Current
Power Supply I Y3]
0-4 A 4—-—- A

  

K+IDS Power Supply Unit

 

 

120VAC

 

   

10 kV Isolation
Transformer

 

 

 

 

A wooden crate and plexi-glass shield protect
the operator since the K+IDS power supply is floated at 4 kV.

Figure 24. Schematic diagram for the implementation of thermally-assisted K+IDS-by-
FAB on the JEOL HX-l 10.

101

and #4 in the surge filter following removal of the existing connections to sites #3 and #4,
so a current loop was isolated between the K+ emitter and the K+IDS power supply.
Thus, when the accelerating voltage of mass spectrometer is turned on, the entire K+IDS
power supply will operate at 4 kV. To protect the operator and the mass spectrometer
from exposed high voltages, the K+IDS power supply was placed in a wooden crate,
fitted with a plexi-glass shield. The K+IDS power supply unit was also plugged into a
10-kV isolation transformer to prevent the 4 kV from leaking to ground or the wall AC
voltage. The K+IDS power supply was operated by a wooden rod and rubber hose
though port holes in the plexi-glass. The K+IDS-by-FAB space-divided target without
the K+ glass emitter attached was placed on the end of the FD/FAB direct insertion probe
(DIP) and inserted into the FD/FAB ion source. A K‘t emitter bead, routinely prepared as
in K+IDS (see Chapter five), was spot-weld mounted to the E1 filament leads in the E1
filament housing block and adjusted so the K+ emitter bead was centrally positioned in
the front, right of the cathode lens element slit in the FD/FAB ion source. The two leads
on the El filament block were connected to the #3 and #4 feed-throughs as is routinely
done in El with Re filaments to complete the K+ emitter heating current circuit. In Figure
25a, with a heating current of 1.8 A, just above the on-set of K+ thermionic emission, 62
counts of K“ were detected. In Figure 25b, glycerol was applied to the K+IDS-by-FAB
space-divided target and inserted into the ion source so the angled glycerol target was left
of the K4" emitter 1 m away. At a lower heating current of 1.7 A, stronger K+ signal
was detected because the analyte/matrix target surface helped focus K+ towards the mass
analyzer. In Figure 25b, 0.1 counts of [G+K]+ were also detected which can arise if K+
adds to glycerol that was thermally desorbed or evaporated from the sample holder. In
Figure 25c, the K+ emitter was held at 1.7 A while the FAB beam was activated which
resulted in a thermally-assisted KflDS-by-FAB mass spectrum of glycerol. Here, the
[G+K]+ was 20 times more abundant than in Figure 25b owing to more desorbed glycerol

molecules and K+ ions in the selvedge. Also with this experimental set-up, a normal

§

102

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I 39 -
. FAB Off .
1% 80: a) K+ Heating Current: 1.8 A L
a I Bias Voltage: 0 V .
t 1 Accelerating V: 4 kV i
{, 60‘ K+ Counts: 62 -
e ‘ .
l. 40‘ :
I ' t
C : i
3 201 .
i . 41 ,
yc'. ......--.' ...-.'
0 20 40 60 80 i9;
100
R 1 b) 39 FAB Off I
I 80: K+ Heating Current: 1.70 A i
a . Bias Voltage: 0.5 V ,
t : Accelerating V: 4 kV :
it 60‘ K+ Counts: 98 -
c ‘ b
I : [G+K]+ I
n 40‘ 131 -
E. I . - - n. n n ,:
g 20.: x 50.0 L
{ ‘ P
y 0‘ ' I 1 1 1 . I i . . V I fl 1 v ' r
0 50 100 150 200
m/z
R100: c) 185 FAB On I
I + .
'f 80- 93 ”6"“) Same as b -
[l 1 [G+H]+ .
1,. 60< ;
e l :
i401 e
e ‘ >
n ' [3G+H]+ .
is 20'. 55 15 [5131?” 167 [20+K]+ 277 E
I I
50 100 150 200 250 300
m/z

Figure 25. Thermally-assisted K+IDS-by-FAB mass spectra of a) thermally-heated K+ glass

emitter, FAB off, and no analyte on FD/FAB sample target; b) thermally-heated K+ glass
emitter, FAB off, and glycerol on FD/FAB target; and c) same as b except FAB is on.

0P
065
Jen:
TM:
gene
“'Orf‘
W1mg

into h

103

K+IDS-by-FAB mass spectrum of glycerol could be obtained without heating the K+
emitter. When both FAB was activated and the K+ emitter was heated, the K+ ion current
was approximately twice that obtained in keV-Xe bombardment of K+ glass alone.
However, biological analytes were very difficult to analyze using this method because the
analyte ion signals were transient similar to K+IDS because the heated K+ emitter
radiatively heated the sample holder on the K+IDS by-FAB target to some degree and
unfortunately induced rapid analyte thermal desorption from glycerol. In conclusion, by
thermally heating the K+ emitter, more undesirable variables were introduced into the
K+IDS—by-FAB experiment which reduced the effectiveness of KHDS-by-FAB for

informative mechanistic studies for FAB.

I n ti 1 ter 11' rim nt 'h I N

SIMION computer modeling was pursued in an attempt to optimize the injection
of all desorbed K+ ions by FAB into the selvedge region above the analyte/matrix target
half because the "brute force" methods for increasing K+ adduct formation, such as
thermally heating the 1K+ emitter or changing its composition, did not give the desired
results. The K+ ion trajectories were apparently affected by divided target design and
geometry, and the influence of the extraction fields generated by the electrostatic
potentials applied to the ion source ion optics. For years, SIMION [174], a respected ion
optics computer modeling program, has provided mass spectrometn'sts with the insight
needed to understand how ions traverse specific ion optical transport systems, which have
lens elements with applied potentials, for designing novel or modifying existing systems.
This software package is well known in the mass spectrometry community and it is
generally accepted that the theoretical models generated correspond quite well with real-
world systems. Up until now, K+IDS-by-FAB targets were designed using basic intuition
without stunning success, but the SIMION 4.0 software package [175] provided insight
into how the various K+IDS-by-FAB targets interfaced with the FD/FAB ion source

f0
101
p01

refi

C3101

each

104

mounted in the JEOL HX-l 10. As a result, this additional insight assisted in the design
of a functional K+IDS-by-FAB divided target design and realizing a required minor ion
source modification which enabled K+ to be efficiently injected into the selvedge region
and FAB mechanistic studies to proceed.

The first step in SIMION modeling of KflDS-by-FAB targets was to input the
scaled dimensions and electrostatic potentials of a two-dimensional "slice" of the target
and ion source optical elements into a small array size using the "editor" of SIMION. In
SIMION, the lens elements created in the editor are referred to as "electrodes". After
data input, the small array (e.g., 50x75 or 3750 points), containing the target ion source
electrodes and potentials, was doubled in size (i.e., 100x150 or 15,000 points), using the
"double" command, to provide the model with sufficient resolution to insure "real world"
accuracy and integrity. The upper limit for array sizes in SIMION are 16,000 points.
The same set of ion source electrode dimensions were used in each SIMION model in
Figures 26 and 27 and only the target electrode geometries changed. As a point of
reference, the slits between the cathode (L1) and lst focus (L2) electrodes are 1 and 2
mm, respectively. Additionally, the same set of electrode voltages were used for each
SIMION model in Figures 26 and 27 and these values were obtained by measuring the
actual potentials on the ion source elements on the HX-l 10 mass spectrometer during a
typical KflDS-by-FAB experiment. The JEOL FD/FAB ion source elements and
measured potentials are: FAB target (repeller), 10050 V; cathode (Ll), 10010 V; lst
focus (L2), 9000 V; 2nd focus (L3), 600 V; deflector (L4, not shown), 30 V; TIM (total
ion monitor) plate (L5), 0 V. Following data input and array size "doubling", this 15,000
point two-dimensional array of ion source "electrodes" having fixed potential values is
refined meaning a potential energy surface is calculated in which all non-fixed potential
array points in the open spaces between the electrodes are assigned a respective
calculated potential. In Figures 26 and 27, the equipotential lines C1-C6, identical in

each model, trace a specific potential value across the potential energy surface revealing

105

dew—28?: =2 +M A0 93 :2 “Egg—5832a 8 9.365 .038 823—060an mica—shaman. was .835 Bv_>_?oommm
88.8-33 3 .695 83:62—55 3 .8qu m<m Eavesm Au 05 55 3.9.8 858 :2 m<n<e 2: mo $0.36 2925. .cm Eswfi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

> 88 > 28—
> Sea u 90
S 3 >§an8
8 6 >22: "5
«U > 32: u no
> eng— u «U
1199 > 6'2: 11.. #9
Nu > em:— 954 3:32—35
6 .... AQ > 88 > 38—
3
«out; no «D
335.8an 6
méééEE > 3...:
/ I.
> 88 > 22: 11 .
«A 1— Sneak.
m0 $9635.00an
8 82.535
/ /
e0

_/

'4'
.'I‘

 

 

   

3135-....3

/

>88
m.—

m0

>§a

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

> 22:
3
no no 5
my > 82:
3.55-8an
5.23:5.

/

> :33

 

 

 

 

 

 

 

 

 

 

 

.motoauoqmb =2 +v. 8 ES :2 xEmEBEmca Q. 9:305 898 coEz—Vooaam m<nT£-mn=+M Ea Jowhfi uoEzuéuaam coursing
3 Jews. BEEF—EB 3 4038 BE Eat—Em A: 05 5:5 33:8 32:20 .5230 :2 c..— ccm 08:8 :2 man—RE of we 239: 20:45 SN 2&5

> 83 > 3.5— > 83 > 32:

NJ 1— > Sea u 00 NA —‘—

> cows .1. m0

> 32: u 9U

> 3.x: u no

> one: u «U

> :33 .1. —D
3:: _a=:82__=_.fl

 

 

 

 

 

    

823:3an
31.—adhere.

 

V///A

. A Q
Fatwa—53%

 

107

the contour of this surface. The great utility of SIMION for modeling ion optical systems
is ions with initial KEs and directions can be placed on this calculated potential energy
surface and their responses to it, based on their initial trajectories, can be observed which
reveals the subtleties of the potential energy surface. Thus, ion trajectories can be
manipulated by altering the potential energy surface which is derived from divided target
and the ion source potentials, geometries and dimensions. For simplicity, all ion
trajectories in the SIMION models in Figure 26 and 27 were calculated for ions
originating on the target surfaces and having initial kinetic energies of 0 eV. Under real
world conditions, all ions have an initial KEs and directions, however ions with zero
initial KB are most sensitive to potential energy surface and will yield more information
about it. Additionally, ion trajectories in or near the selvedge region are not accurate
because SIMION does not take into account ion/molecule reactions and non-reactive
collisions. The results extrapolated from these SIMION models were used to assist the
KflDS-by-FAB target development process by providing insight into the general ion
extraction trends from a variety of divided targets, and they should not be interpreted
literally. Last, the fast atom beam is not shown in these SIMION models, but the beam
would originate from above the Figures and strike the divided targets with an average
angle of incidence of 70°.

The first course of action with SIMION was to model each KflDS-by-FAB
divided target interfaced to the FD/FAB ion source. For reference, the "normal" FAB
experiment on the JEOL HX-l 10 is shown in Figure 26a in which the JEOL FD/FAB
target is interfaced with the FD/FAB ion source under normal operating conditions. In
Figure 26b, the KHDS-by-FAB wall-divided target is interfaced with the FD/FAB ion
source unsuccessfully because ions extracted from either target half crash into L1 with
the applied "center-extracting" ion optics potentials. Ion extraction from each target half
in the KflDS-by-FAB dual-surface space-divided target was more successful as shown in

Figure 26c. The purpose of a divided target is to inject [G into the selvedge region above

108

the analyte/matrix target. According to the SIMION models in Figures 26b and c and
assuming the "top" divided target half is the K+ glass emitter surface, K+ ions would be
extracted by the ion optics without ever passing though the "selvedge" region above the
opposite target half. Experimentally, this observation probably accounts for the lack of
K+ adduct ions formed in the gas phase observed with this target. Since the wall-divided
target is notorious for condensed-phase contamination of the K+ glass emitter and the
wall is the primary location of this contamination, the contamination-indicating ions
[G+K]+ and [G-H-1-2K]+ may be successfully extracted from this region if they were
formed near or on wall region and desorbed preionized. In Figures 26d and e, the
analyte/matrix and 10‘ ions are extracted from their respective KflDS-by-FAB space-
divided target halves, respectively, when the target was interfaced to the FD/FAB ion
source. The KflDS-by-FAB space-divided target in Figure 26c indicates that 10‘ ions are
injected into the presumed selvedge region above the analyte/matrix target which is
consistent experimentally, because the definite gas-phase K+ adduct ion formation
without condensed-phase contamination is observed with this target in comparison to the
other divided targets. However, the weak K+ adduct ion formation typically observed
using this divided target may be attributed in part to K+ ions traversing the presumed
selvedge region with KEs greater than 2 eV, which are too high for K” adduct ion
formation. K+ ions traversing the high pressure selvedge region should ideally have low
KEs and long residence times which increases the probability K+ forming adducts with
desorbed molecules through ion/molecule collisions.

Additional SIMION modeling was performed in an attempt to lower the KB of the
K+ ions in the presumed area of the selvedge while maintaining their favorable
trajectories. Success was obtained by placing a thin lens element electrode, referred to as
L0 between the K+IDS-by-FAB space-divided target and L1 electrodes as shown in
Figure 27e. Since L0 has a center slit of less than 1 m, it further limits the extent of

penetration by the extraction field, which is represented by the equipotential lines Cl-C4,

109

because these equipotential lines are now positioned between L0 and L1. Thus, the
presence of L0 essentially creates a field—free region above the target area. Without the
influence of the extraction field in the field-free region above the target area, the potential
energy surface in this region has a much shallower depression in comparison to the deep
depression formed without LO. Thus, K+ ions do not gain as much kinetic energy as
rapidly during extraction with LG present. In Figure 28c, for the KHDS-by-FAB space-
divided target with L0 present (corresponding to Figure 27e), the typical K+ ion desorbed
from the K+ emitter and injected into the selvedge will gain 2 eV of KB within this field-
free region, the upper limit for K+ adduct formation, in 4.5 yrs and will remain in the
selvedge region for approximately 1.9 us as estimated by SIMION. Without LO present,
for the same target (corresponding to Figure 26e), the typical 10 ion will gain 2 eV in
0.45 us and reside in the selvedge region for only 0.3 us. Thus, with the addition of L0,
K+ ions are still injected into the presumed area of the selvedge, gain minimal KE in the
field-free region, and essentially have much longer residence times, 6 to 10 times longer,
in the selvedge, all which in theory should substantially enhance the efficiency of
ion/molecule collisions for K+ adduct formation. In Figure 27a, the electrode L0 was
successfully placed in the "normal" FAB experiment. It was important for L0 to be
compatible with this configuration since all analytes are analyzed by FAB before K+IDS-
by-FAB to determine the natural [M+Na]+ and [M+K]+ levels, to ensure [M+K]+ adduct
ions in KHDS-by-FAB are real. In Figures 27b and c, L0 dramatically improved the
extraction efficiencies from these targets in comparison to Figures 26b and c in theory, '
but these models still did not indicate that K+ ions would interact favorably with the
selvedge region above the analyte/matrix target. Additionally, the wall-divided target in
Figure 27b would still exhibit glycerol contamination of the K+ glass emitter regardless
of L0 presence. In Figure 28a-c, the K+IDS-by-FAB space-divided target with or without
L0, corresponding to Figures 26c and 27e, respectively, allows longer K+ residence times

below 2 eV in the presumed area of the selvedge comparison to the FAB and dual-surface

110

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 20 la) . ' III/Z 39 W/O LO

3 15 ' - —-D—m/21000w/oLO

h I

E 10 / - , . mlzB9w/LO

.3 5 _ _" / ——<>—m/leOOw/LO

0 I 0 o

E ' 0' ...o°

.- O00:00.oocooooooo00909000000009.0000...0000‘.‘°°°'.°--°. I : I I

:4 0 1 2 3

Extraction Time (microseconds)

E 20 b) u ' m/z 39 w/o LO

‘5 15 .' —D—m/21000w/0L0

h I

5 10 . _ . . 1:1/239me

.g 5 l I —o—m/z1000w/Lo

E O 0.35:: ...... 0.. ooooooooooooooooooooooooooooooooooooooooooo ‘ vo°‘°.r.. 1 I :
O 1 2 3 4

Extraction Time (microseconds)

 

 

 

m/z 39 w/o LO

     
  

N

O

6
V

+ m/z 1000 w/o LO

  

u—a
LI!

 
 

 

O

 

   

m/z 39 w/LO

—a
O

    
 

——O—-m/21000w/LO

 

 

 

Kinetic Energy (eV)

012 3 45 67 89101112131415
Extraction Time (microseconds)

Figure 28. Estimated extraction times for typical ions from the a) FAB target, b) dual-

surface space-divided target, and c) KflDS-by-FAB space-divided target SIMION models,
with and without L0 present, shown in Figures 26 and 27a, c, d, and e, respectively. The

K+IDS-by-FAB space-divided target combined with L0 is best technique for injecting low
energy K+ ions, with longer residence times, into the selvedge region.

111

space-divided targets with and without L0 present, corresponding to Figures 26a and c
and 27a and c, respectively. Thus, the geometry of the KflDS-by-FAB space-divided
target without L0 is better optimized for injecting K+ ions into the selvedge.

The L0 focusing lens element was constructed from two stainless steel ribbon
strips (0.125 x 0.005 in) precisely arranged side to side in the same plane less than 1 mm
apart. Tungsten wire (.007 in) held the two lens halves together and was used in the
connection between the L0 lens and the FAB repeller voltage source, which is positioned
on the FD/FAB ion source near the DIP. The focusing lens was firmly mounted to the
FD/FAB ion source using existing screws and ceramics. The installed focusing lens
could accept any type of FD/FAB target, both stock and modified.

In practice, the experimental results obtained with the K+IDS-by-FAB space-
divided target in conjunction with L0, as modeled in Figure 27e, were significantly
improved in comparison to without L0 present, which allowed the KHDS-by—FAB
experiment to proceed as originally envisioned, that is, by efficient injection of 10‘ ions
into the selvedge region. In Figure 20c, a KflDS-by-FAB mass spectrum of glycerol
obtained with the K+IDS-by-FAB space-divided target and L0 installed is shown as proof
that this new target/ion source modification combination does achieve the desired results.
In this spectrum, K4" is injected into the selvedge region as shown by the peak
representing K+ at m/z 39, to yield peaks representing the [G+K]+ and [26+K]+ adduct
ions that have parallel abundances to their respective protonated species. No peak
representing [G-H+2K]+ appears at m/z 169 which indicates the K4” emitter is not
contaminated by glycerol and the K+ adduct ions observed are truly gas phase products.
The similar peak intensity ratios of K+ to the K+ adduct ions of glycerol indicates there is

extensive conversion of K+ into product ions in the high pressure selvedge region.

I ' ' 'inrii rt

Divided targets are notorious for providing ion source ion optical tuning problems

112

because ions are extracted from both right and left target halves instead of from a single
central target surface. In Figure 29, the SIMION model of the dual-surface space-divided
target and entire FD/FAB ion source is shown to illustrate this point. The potentials in
this particular SIMION model have an ion optical "center tune" which means essentially,
extraction of ions from either target half is equivalent and an inverted mirror-like image
of the divided target surface is reproduced on L5, although some ions do traverse L5. By
adjusting the potentials on L1, L2 and L3, so one lens element half is higher than the
other, a "right" or "left" ion optical tune can be achieved and the ions generated from the
right or the left target halves of this divided target. can be extracted, accelerated and
focused out of the ion source, through L5, preferentially over the opposite target. While
it is possible to extract ions from both target halves using a "center" ion optical tune,
SIMION and experimental results suggest the ion source geometry and range of applied
potentials will not readily allow the "dual ion beam" in Figure 29a to unify and exit L5
together, although sometimes ion source conditions will exist, such as cleanliness, which
allow unification. Experimentally, even with a center tune, tuning problems from having
a dual beam as in Figure 29a, were experienced for the wall-divided target, dual-surface
space-divided target, and to a lesser degree with the KHDS-by-FAB space-divided
targets. These tuning problems were predicted to occur by SIMION in Figures 26b-e
because the ion beams were poorly defined. Experimental evidence of preferential
extraction is shown in Figures 29b-e from a K+IDS-by-FAB study with a wall-divided
target. In sequential order, ions were extracted from: (1) the K+ emitter target half in b,
(2) the glycerol matrix target half in c, (3) both target halves in (1 without glycerol
contamination of the K+ glass emitter which was a rare occurrence, and (4) both target
halves in as (1 except the K+ emitter was contaminated which most often was the case'
with this target. The addition of L0 to the K+IDS-by-FAB dual-surface divided target
SIMION model in Figure 29f, having a "center tune", produces a well-defined ion beam,

composed of ions extracted from both target halves, that exits L5. Moreover, the

113

 
 
 
  

\ \ \
Dual-Surface Space- I
DividedTarget

 

.1
-
_.
-
-

 

 

 

      
   

 

 

 

 

 

 

 

10050 V F

10010 v 9000 v 600 v 0 v

loo—39 100 93 [ZG+H] E

80‘ b) K+ 801C) [G+H]+ 185 E

so .
4o
20
0o

 

 

 

 

 

 

 

 

 

 

 

      
 
  

93 [261-HF:
801d) [G+H]+ 185 :
so :
4o ;
20 IO“ 57 75 tG+K1+ i
0.9241511 113111, “E

OWTWS'O' ' '160"""15'O'm '2

m/
Dual-Surface Space- 1
DividedTarget ,1 1

l

\.

l .

 

 

  

 

n

u- j
*—
l

at: .

 

 
     

  

10050 V
C]

  
 

 

1 I—
L3

L1 C5 L2 L5
10010V 9000V 600V 0V

Figure 29. SIMION model of dual-surface space-divided target and full FD/FAB ion source
a) shows K" and glycerol ions, extracted from each respective target half, leave the ion

source in two separate beams, b) mass spectrum of K+ beam from K+ glass emitter target
half, c) mass spectrum of the glycerol beam from glycerol target half, (I) "rare" mass
spectrum of both K+ and glycerol ion beams, e) typical mass spectrum of both K"' and
glycerol ion beams showing glycerol contamination of K+ glass emitter, and 0 same

SIMION model as in a, except with LO installed, shows a single well-defined beam leaving
the ion source which eliminates ion beam focusing problems.

114

combination of L0 with all K+IDS-by-FAB targets in the SIMION models shown in
Figures 27 result in well-defined ion beams being accelerated. Experimentally, for a
typical run using K+IDS-by-FAB space-divided target with LO installed, the K+ ions from
the right target half, K+ adduct ions from the gas-phase, and regular FAB ions from the
left target half all appear in a KHDS-by-FAB mass spectrum with a normal "center tune".
In theory, the well-defined ion beams should increase the sensitivity for a typical analyses
but unfortunately, the FAB beam is partially blocked by L0 resulting in a narrower

portion of the divided target being irradiated.

V ' i ' l t.

Experiments were conducted to substantiate and optimize the use of L0 in the
K+IDS-by-FAB experiment after it proved successful in (1) increasing K+ adduct ion
formation of desorbed neutral molecules for a variety of analytes, and (2) correcting
tuning problems for divided targets. In Figure 30, sequential KHDS-by-FAB
experiments with glycerol show the [G+K]+ ion relative abundance was 30 counts when
LO was installed and 11 counts when LO was removed in the next experiment which
suggests that L0 actually does increase K+ adduct formation in the K+IDS-by-FAB
experiment. In Figure 31, increased [G+K]+ adduct formation using glycerol was
demonstrated when the K+IDS-by-FAB space-divided target was close to L0 rather than
distant as indicated by the 4 and 0.7 count difference, respectively. Perhaps this
experiment indicates a local high pressure region exists in small volume between target
and lens which enhances adduct ion-forming collisions and/or termolecular stabilization.
However, there were never any signs of condensation of vapor-phase glycerol on the LO
lens element. In Figure 32, the K+ conversion efficiency of desorbed neutral molecules
and the effect of L0 slit width on K+ adduct formation is demonstrated for sequential
experiments with the analyte mixture thevetin A (MW 872) and B (MW 858). Consistent

and identical amounts of sample and matrix were mixed on the K+IDS-by—FAB target for

115

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100
‘ ) 93 185 .
R : a [G+H]+ I
f 30. L01nstalled [26+H]+ _
a ; mlz 131,30Counts ;
l ‘ .
v 60- .
e ‘ .
1 [G+Kl+ :
p 40- 131 -
e 75 ~
3 20' K.» [ZG+K]+ Z
i 1 39 45 223 [3gzm+ ;
y 0* 29 167 _
.l .
o 50 100 150 200 250 300
m/z
- I
1004 b) 75 93 135
R 1 [G+H]+ [ZG+H]+ _
f 80- LORemoved ~
f 2 W2 131,11 Counts ;
{I 604 57 '
e l [3G+H]+ .
I . .
n 40- 45 277
t .
S ‘ .
‘ G+K] '
S 20.. [ .
i . 1291;; 131 1.7 .1 [ZG+K]+ ~
y I , ; 223 2 9 I
0- ‘ * -
o 50 100 150 200 250 300
m/z

Figure 30. Sequential KfIDS-by-FAB mass spectra of glycerol using the K+IDS-by-FAB
space-divided target a) with L0 installed initially and b) then with L0 removed which
demonstrates the effectiveness of L0 for enhanced [M+K]+ adduct formation.

116

 

 

 

 

 

 

 

 

 

 

 

 

100
g j a) Divided target 185
1 80: away from L0 93 [ZG+H]+
é‘ . m/z 131, 0.7 counts [G+H]+
1 I
X 60:
I ‘ [G W
n ‘ +
t8 40: 13 1 167
g . 75 AW
1 20* X5
g, 3 K+ 57 l
O: 39 j ‘ F Lfi * .‘ v ‘v v ‘: :lw 1 v . v f
0 50 100 150 200 250
m/z
100
R 1 b) Divided target 185
]e 80‘ close to L0 + [2G+H]+
ii : m/z 131, 4 counts [694?]
‘v i
e 60‘ [G+K]+
}, 1 131
t 40‘
f1 3 167
'S 20" 75 X5
i 1 Id
y ‘ 39 57 J
0+ ' " A: {if-l" ' ‘ ' 1 ' v '1 v ‘1 v "v A-‘ v u v # r r
0 50 100 150 200 250
m/z

Figure 31. The KHDS-by-FAB mass spectra of glycerol from sequential experiments in

which the KflDS-by-FAB space—divided target was positioned a) distant from, and b) then
close to the LO lens element resulting in an increase in the intensity of the peak

representing [G+K]+ in the latter experiment owing to the possible creation of a higher
pressure region between the target and LO.

117

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.
R .
‘f ‘ a) FAB
a , 859
t . [Thevetin B+HJ+
1 24
C 1
I t
It! ‘ [Thevetin A+H]+
e 1: 873
n
s 1
1
y
850 860 870 880 890 900 910 19112/0
2
1
1; j b) Narrow L0 slit width 897
a]! 1 859 K‘HDS-by-F AB [Thevetin B+K]+
E ' [Thevetin B+H1+
g [Thevetin A+K]"’
911
[I] ‘ [Thevetin A+H]+
g 873
II
S j 1.
i 1 1
y 01 A 14‘ ‘LLA -1 A- a A
850fi ' 8'60 H 870' ' ' 880' ' - 8'90 ' ' 900 ' 'fi9‘16 ' ' 9'20
m/z
1
R j ) Wide L0 slit width
6 ‘ c 859 +
,1, 1 [Thevetin B+H]+ K ms'bY'FAB
t I 897
. , +
2 1 [Thevetin A+Hl+ [Thevetin B+K]
I I 873 [Thevetin A+K]+
n j ‘ 911
t .
e d
n 1
5 1
1 111 11 l 1 111
y0.1.11l111 1111111111 11....111111111111 11..ii..111 1111
850 ' ' 860' ”T8116 " 880' ' 8'90 ” 900 h 910' ' ' 230
2

Figure 32. The K+ cationization efficiency of desorbed neutral molecules and the effect of

L0 slit width on K+ adduct formation is demonstrated in sequential experiments using
thevetin A (MW 872) and B (MW 858). A comparison between the a) FAB mass spectrum,

and the KflDS-by-FAB mass spectra b) with a narrow LO slit width (out of ion optical
tune) and c) with a wide LO slit width reveals 30% of total desorbed neutral molecules are

K+ cationized, if [M+H]+ and [M+K]+ are summed from b and c, averaged, and compared
to [M+H]+ in 8. Slightly more K+ adducts are obtained from a narrow slit width.

118

these sequential experiments, and the spectra were normalized to the same level. In
Figure 32b, at a narrow L0 slit width, the peaks representing [M+K]+ adduct ion at m/z
897 and 911 are more intense in comparison to the peaks representing the protonated
species. However, in Figure 32¢, a wide LO slit width resulted in a reversal of the peak
intensities from Figure 32b which suggests the K+ gas-phase ions have longer residence
times, resulting in enhanced K+ adduct ion formation, when the LO slit width is very
narrow. By summing the normalized peaks representing the [M+H]+ and the [M+K]+
ions at m/z 859 and 897, respectively, for thevetin in Figures 32b and c, taking an
average, and comparing this average to the normalized peak representing [M+H]+ at m/z
859 in Figure 32a, ~30% of the desorbed neutral molecules or [M+H]+ ions are lost in the
KflDS—by-FAB experiment. Of these remaining desorbed neutral molecules, only half
undergo ion/molecule chemistry with K+ to form [M+K]+ adduct ions, as indicated in
Figures 32b and c, which suggests H” and 10‘ may be in competition for analytes in the
selvedge region under certain situations. Thevetin is a cardiac glycoside and the PA of
cardiac glycosides were estimated to be 1ess than the PA of glycerol which means
glycerol cannot readily protonated thevetin [176]. Instead, the less abundant fragment
ions of glycerol, GP“, which are estimated to be 1% of the 1000 desorbed glycerol
molecules, are believed to protonate thevetin A or B in this case. Thus, because the
abundant [G-i-H]+ present in the selvedge is not an effective protonating agent and the
612+ species are as populated as K+, K+ has a better chance of forming adduct ions with
thevetin as shown by roughly equal [M+K]+/[M+H]+ ratios. The K+IDS-by-FAB data
presented in this thesis support this phenomenon because [M+K]+/[M+H]+ ratios are
larger for steroids and oligosaccharides than for peptides and other nitrogen containing
compounds. Next, the well-defined ion beams observed in the KflDS-by-FAB SIMION
models with L0 present in Figures 27a-e are believed occur in practice because ion tuning
is enhanced as explained above. Additionally, the resolution of the peaks representing

[M+H]+ and [M+K]+ for a high molecular weight peptide such as kassinin (MW 1334) is

119

H-Asp-Val-Pro-Lys-Ser-Asp-Gln-Phe-Val-Gly-Leu-Met-NHZ

 

 

 

 

 

 

 

 

R j ) [M+H]+ FAB

f 1 a 1335 FD/FAB target

i‘ ‘.

1 .

V a

e .

I

n

t

e

n

.S

1

t

y “A. . . . LLLA - ..... - .
1320 1330 1340 1350 1360 1370 1380

m/z

g j b) [M+H]+ K+Il)S-by-FAB

1 1 3 3 5 space-divided target

a j with L0 present

i i

v u

e .

I - [M+K]+

n . 1373

t 1 1

e i

{J 1

l I

§ I‘llalnllll‘h ‘lnnhhllll‘h‘llh511‘klt111111311“ 1111

 

1320 1330 1340 1350 1360 1370 113/80
2

 

 

 

 

 

R 1

e 1 C KHDS-by-FAB

l j ) [NI-1H]+ space-divided target

f‘ 3 1335 without L0 present

1.

C

I

n

g . [M+K]+

2 3 1373

i «1

t 4

y W

1320 ' T1330 1 ' 1340 I ' 150 ' ' 1360 FT '13'70 ‘ f ‘1380
m/z

Figure 33. a) A FAB mass spectrum, b) K+IDS-by-FAB mass spectrum, using with the

KflDS-by-FAB space-divided target with L0 installed, and c) same as b, except L0 was not
installed, were obtained for Kassinin (MW 1334) at a resolution of 1000. A comparison
between b and c demonstrates that L0 enhances resolution of the peaks.

120

increased. In Figure 33b, for L0 installed in the ion source, baseline resolution is
obtained for K+IDS-by-FAB of kassinin. In comparison, in Figure 33c, L0 was removed
from the ion source which resulted in poorer resolution. Both of these spectra were
obtained with the KflDS-by-FAB spaced divided target at a resolution of 1000 which is
too low to adequately resolve the high mass kassinin peaks at m/z 1335 under normal
circumstances. In Figure 34, as a final test, the L0 lens element was used in conjunction
with a regular FD/FAB target in three successive experiments with the peptide
methionine enkephalin (MW 573) in glycerol, as shown, which further suggested that L0
delays the extraction of ions responsible for protonation and a high pressure region exists
between the target and L0. In Figure 34a, the presence of L0 allows the intensity of the
protonated molecule at m/z 574 to exceed the intensity of the peak representing the F
immonium ion at m/z 120, which always has a high intensity independent of [M+H]+. In
a successive experiment shown in Figure 34c, the L0 lens element was removed and the
peak representing the [M+H]+ ion dropped to "normal" intensity which resulted in a
normal FAB mass spectrum of methionine enkephalin with the expected ion distribution.
A successful K+IDS-by-FAB experiment with methionine enkephalin was performed
midway between these two experiments, with L0 installed, to ensure the L0 lens element
was installed and working correctly. In this KfiDS-by-FAB mass spectrum of
methionine enkephalin, the intensity of the peak representing [M-i-K]+ ions is one-third of
[M+H]+ which further supports the fact analytes with higher PAs than [G+H]+ are
protonated more competitively than K+ cationization. The K+IDS-by-FAB experiment
works best if the ion source is clean, L0 is clean, and the edges of L0 are finely sanded
clean. These prerequisites enhance sensitivity, ion extraction and ion optical tuning. In
addition to the simple planer L0 lens element modeled (i.e., — —) in SIMION, other
L0 designs and geometries were attempted such as angled lens element with a center slit
(i.e., / \), a lens element "wrapped" around the target (i.e., 1— __l), etc. In

general, these more complex geometries usually resulted in greater chances of

121

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  
  
    

 

 

 

 

 

 

 

    
 
    

 

 

1 IF FAB with L0 installed 574
g I a) 120 [Mi-HP
; 80: [G+H]+ a]
t 1 93 136
ii 60- [26+H]+
:3 1 185
1‘ 40: 612
g1 1 75 b2 b3 1’2
1 201 (357 147 221278297 )3 a4 c 05 XS
y 01 3O . : 354 397 442 529
0 100 200 300 400 500 600
mlz
R1 1b) 120 KflDS-by-FAB with Loinstalled [M1H]+
‘f 803 F 574
a . 136
1 I 01
v 60-
e .
I I
1‘ 40.
e
n : K+ [G+H]+ GF—28 b b3
1 20. 39 3 11471772221278 ’2 a4
1 1 75 297 )3 397
y 0 1.1 1 3§4 1 1 1 1
100 200 300 400 500 """
R1 1C) 93T 11,201? FAB withoutLOimtalled
f 801 lG+H1+ [26+H]+
a '.
i l 75
g 60:
I j 57 136
1‘ 49. 45 “I
e .
n , all-rib
‘ +
i 20'. 14761548152 b3 1’2 [M+H] X5
t . 177 278297 1’3 04 574
Y 0+ 221 354 397
0 100 200 300 400 500 600

m/z

Figure 34. The utility of the LO lens element in front of the FAB target for delaying ion
extraction and possibly providing a higher pressure region above the FAB target is
demonstrated, in sequential scans, for the analysis of methionine enkephalin (MW 573) in

glycerol. With L0 present, a) an intense peak representing [M+H]+ was observed in the
initial FAB mass spectrum, b) a successful KflDS-by-FAB mass spectrum validated a, and
last, c) L0 was removed resulting in a decrease in the intensity of [M+H]+.

122

misalignment. A fine mesh grid was placed between the target and L1 without success
and actually resulted in an overall decrease in ion current. In an another experiment,
metal CI volume was mounted in place of L0. It encased the KflDS-by-FAB target
during FAB, but simultaneous alignment of the CI volume with the FAB beam inlet and
with the slit of L1 was difficult to achieve plus an increase in K+ adduct ion formation

was not observed even when installed correctly.

E . E . . 53! r 1

After discussing and solving the analyte/glycerol contamination of the K+ glass
emitter above, potassium contamination of the analyte/matrix liquid through the gas
phase may also occur in theory but experiments and observations have indicated it does
not occur. First, the K+IDS-by-FAB experiment is very sensitive to divided target
geometry and positioning in relation to the center axis of the ion source and LO. Since
the FD/FAB DIP, on which the KflDS-by-FAB target is mounted, is entered from the
front, there is no view of the KflDS-by-FAB experiment and all positioning and
tweaking has to be done before evacuating the ion source chamber as opposed to under
vacuum with a see-through flange and a special DIP to adjust and optimize LO, etc. If
significant amounts of potassium were sputtered into the liquid analyte/matrix during
KflDS-by-FAB, then the target design and geometry, and the presence or absence of L0
would not be important factors in the experiment. Moreover, abundant K+ adduct ion
formation would have always been observed! Next, a wire coated with potassium salts
would emit more K atoms than ions upon heating even though the ionization energy of K
is low. However, the work function of a heated K+ aluminosilicate glass emitter is higher
than the heated wire which results in thermionic emission of approximately 90% ions‘
[52]. Atom emission or matrix retention accounts for missing 10%. When a K+ glass
emitter is bombarded by keV-Xe particles, the K+ ion current is stable and continuous

without evidence of charging. Thus, even without heating, K”r ions migrate through the

123

crystal lattice to the surface for desorption into the gas phase while the respective
electrons migrate to the underlying Re wire or metal surface to maintain matrix charge
neutrality. Clearly, if vapor-phase K or K+ contamination of the analyte/matrix occurred
in each KflDS-by-FAB experiment, K+ adduct ions of the analyte would have been
observed consistently from day one eliminating any need for target development. Last,
the best method for checking the analyte/matrix for potassium contamination is through
experimentation. A normal KflDS-by-FAB experiment was performed by: (l) exposing
the space-divided target for 2 minutes to Xe bombardment with either a bare
analyte/matrix sample surface or a glycerol matrix, (2) removing the DIP and detaching
the K+ glass emitter bead from the target, and (3) reanalyzing the bare surface or liquid
matrix for possible K+ contamination. For the K+IDS-by-FAB experiment, with a bare
analyte/matrix surface target and with LO not installed, the bare target surface initially
emitted 1.5 counts of K+ without a K“ emitter present when exposed to the FAB beam,
and 18 counts of K+ with the K4“ emitter installed and bombarded. After 2 minutes of
exposure to FAB, the K+ emitter was detached, and 1.7 counts of K+ were detected from
the bare surface, a slight increase. With L0 installed, the same experiment resulted in a
0.5 count decrease in K+ detected, since 1 count of K+ was detected initially and only 0.5
counts of K+ were detected after a 2 minute exposure to FAB. Thus, these results suggest
this particular experiment is highly dependent on the exact target area from which K+ is
extracted by the ion optics and, for the most part, K+ does not readily adsorb to the
analyte/matrix surface. With L0 present, the lower K+ KEs are more controlled than at
high KEs resulting in more efficient extraction instead of adsorption to the analyte/matrix
surface as indicated by the decrease in K“ counts. However, in Figure 35, the same type
of quantifying experiment was performed but with the glycerol matrix applied to the
KflDS-by-FAB space-divided target and LO installed in the ion source. The normal
KHDS-by-FAB mass spectrum of glycerol is shown in Figure 35a and the peak

representing [G+K]+ adduct ion at m/z 131 has a relative intensity of 6 counts. The

124

 

 

 

 

 

 

 

     

 

 

  
 
 

 

 

 

 

 

 

 

 

 

 

50
t? : a KflDS-by-FAB of glycerol,
a I Rlofmlzl3lis6wunts
i . [26+H]
g I [G+H]+
I 301 93
n 2
a 20:
3 3 +
. 1 [G+K]
{ 10: K-l- 75 131 [ZG+K]+
Y 1 39 57 J l 223
C . 2 319111;. . . .+.7. . -1 .‘24. :L .
0 50 100 150 200 250
' m/z
R 3: b FAB of residual glycerol from above experiment after
Z the K+ emitter was removed from KflDS-by-FAB target,
1 1
g 2 ‘ Cr+ 56 [644-1]... [2G+H]+
I 1 Fe“ 93
1‘ I K+ [G+K]+
,6, 11 39 131
f 1 [20+K]+
t J 147 223
1’
01 239 1
O 50 100 150 200 250
m/z
20
R 3 C FAB of new glycerol on same target used in above experiment,
Li: R] of m/z 131 is 0.09 counts
1 151 93 185
g , [2G+H]+
1
I 10.
n i
1% j 147
is 5, Cr+ 75 lxlsln-fllk—ns.
1 2
E; : 455I '57 I
04 4». , L.—.—-. :--, . .fi . .....
0 50 100 150 200 250
m/z

Figure 35. Sequential experiments to prove K+ forms adduct ions in the gas phase and K is
not readily deposited into the upper layers of glycerol in KflDS-by-FAB. First, a) K+IDS-
by-FAB mass spectrum of glycerol on KflDS-by-FAB space-divided target which yields 6
counts for [G+K]+, then b) the K+ glass emitter is detached from this target and the
residual glycerol is reanalyzed by FAB to probe K+ contamination which yields 0.6 counts,
and last c) the residual K+ in new glycerol placed on the same target in b is probed.

125

K+IDS-by-FAB target was removed from the mass spectrometer, the K“ emitter bead was
detached from the target, and this same target with residual glycerol was reinserted into
the mass spectrometer for analysis by FAB which resulted in the mass spectrum in Figure
35b. Here, the peak representing the [G+K]+ adduct ion has a relative intensity of 0.6
counts, 10 times less than the R1 of the peak representing [G-l-K]+ in Figure 35a. In
Figure 35c, a new glycerol sample was analyzed showing glycerol contains no potassium
nor does the target. The mass spectra in Figure 38 were normalized to the same level to
make these comparisons valid. Thus, this experiment indicates approximately 10% of the
potassium emitted reaches the matrix, which is not unexpected since the two target
surfaces are close together, but 90% of K+ cationization of desorbed glycerol molecules
occurs in the gas phase. Additional quantifying K+IDS-by-FAB experiments were
conducted by noting the initial and final ion currents on the JEOL HX-l 10 oscilloscope.
Here, the [M+K]+ adduct ions of glycerol and analytes never showed a time dependence
by which the abundances would increase during the course of a run. From these
experiments, it is concluded that only small amounts of potassium are deposited into the
matrix and at least 90% of the observed K+ adduct ion formation in KflDS-by-FAB is a
gas phase process.

Thus, the major thrust of the K+IDS-by-FAB research was to develop an effective
divided target with minimal ion source modification to make the technique simple,
predictable, robust, and reproducible for a wide variety compounds. This approach
ensured a portable platform to other mass spectrometers and a valid technique, with
minimal experimental variables, for the investigation of the ion formation mechanisms in

FAB as discussed below.

126

3. Mechanistic Studies of Ion Formation in FAB using KfiDS—by-FAB

The real utility of K+IDS-by-FAB is detecting the specific types of neutral
molecules (M) desorbed in the FAB experiment. An [M+K]+ adduct ion detected in the
K+IDS-by-FAB mass spectrum indicates M was desorbed into the gas phase providing
the K+ glass emitter and analyte/matrix target halves are not contaminated with each
other. This capability will indicate whether intact neutral analyte molecules or neutral
molecular fragments are desorbed in the FAB experiment, a feat not previously attempted
by the mass spectrometry community and central to further understanding of the highly
debated, not well understood ion formation mechanism(s) in FAB. For a detailed
discussion of the types of neutral molecules desorbed in FAB for a variety of analytes,
the reader is referred to the published KflDS-by-FAB paper in Appendix one.

The KflDS-by-FAB concept for investigating the desorbed neutral molecules in a
FAB experiment is briefly discussed for the analyte cholic acid (MW 408), a bile acid,
dissolved in glycerol. Because this compound is acidic in nature, the [M+H]+ peak at
m/z 409 is not very intense in the FAB spectrum as shown in Figure 36a, and it is better
suited for negative-ion mode FAB mass spectrometry. The peaks at m/z 355 and m/z 373
represent the dominant even-electron (EE) three and two water-loss fragment ions,
[M+H-nH20]+, of cholic acid, respectively, which is not unexpected for a molecule with
four hydroxy sites as shown in the structure in Figure 36a. With K+IDS-by-FAB, the
origin of the ions observed for cholic acid can be put in perspective mechanistically.

In Figure 37, the possible mechanisms are shown for the formation of the [M+H-
3H20]+ fragment ion appearing as a peak at m/z 355 in the cholic acid FAB mass
spectrum. In the first pathway, cholic acid molecules (M) are desorbed intact into the gas
phase by FAB from the glycerol matrix. If M traverses the selvedge region, there is a

high probability that it will be protonated in the protonating environment of the selvedge

127

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a) , [M+H-3H201+
'2' 355 [M+H-2H20]+
a
s
.i>.’
E
:12
350 ' ' ' '460'
b) [M+H-3H O]+
g 355 2
§
§ [M+H-2H20]+
.33
D
m
3 - f ' '500
1er
C) g [Mm-314201“ FAB of 0.01 M
2 355 [M+H-2H201+ KCl in glycerol
:3 373
'3
..>.
a; [M+H-H20]+ [M+H]+ lM+K1+ [M-H+2K]+
°‘ 3.21409 _. it: - 485
360 380‘ ' '400' ' '420' 'F440' ' '4607 'F4éo' ' 151/00
2
d); [M+H-3H20]+ [WK]; FAB of0.1 M
g 355 |_”M+H-2H20]+ 447 KCl in glycerol
2.; ,373
.3 [M+H-H201+[M +Hl+ [M-H+2K]+
0‘ 391 499 411 IL 4815
360 380 '400‘ ' '4‘20' ' ‘440 '460' ' ‘480‘ * '500
m/z
e) g} FAB of glycerol/cholic acid [M-H+2K]+
g? contaminated K+ glass emitter 485
3.
£1
33
“g: [M+K]+
3.; 447
“I..3?.1-..§7.7_
7360' ' '3807 F 400 ‘ '4‘20' ' 440* 460 480 15130
2

Figure 36. Cholic acid (MW 408), in glycerol, a) FAB mass spectrum, b) KflDS-by-FAB
mass spectrum, c) FAB mass spectrum with 0.01 M KC1 salts, d) FAB mass spectrum with

0.1 M KCl salts, e) FAB mass spectrum of analyte/matrix contaminated K+ glass emitter.

128

[(M+H) - 3H20]+(m/z 355)

V3H20

[(M-3H20)+H]+ [M+H]+
(m/z 355) T— (tn/z 409)

" — — _ — “l
I Protonating Environment 1
I. _______
<—K+
[M'3H201mw 354) M(MW408)

   
 

 

Desorption

   

Cholic acid/Glycerol matrix

 

Figure 37. Two possible mechanisms for the formation of the [M+H-3H20]+ fragment
ion in FAB of cholic acid (MW 408).

to form the protonated molecule which appears at a peak at rn/z 409 in the mass
spectrum. Since the average internal energy content in these particular [M+H]+ ions are
high enough to induce the low activation energy water-loss fragmentation pathways, three
waters are lost by three successive gas-phase unimolecular decomposition processes to
yield the [(M+H)-3H20]+ fragment ion peak appearing at m/z 355 in the mass spectrum.
The loss of H20 from hydroxy—containing analytes in CI is a prominent unimolecular
decomposition pathway. A second, equally valid mechanism for forming the [M+H-
3H20]+ fragment ion involves M undergoing keV Xe beam-induced dissociation in the
glycerol matrix to form the degradation product [M-3H20] (MW 354). A cholic acid
molecule is solvated by glycerol molecules in the matrix by hydrogen bonding and the
weaker van der Waals forces. For intact desorption to occur, all these non—covalent
intermolecular bonds between polar functional groups need to be cleaved which is not an

entirely favorable process when the matrix and analyte is extremely polar. Degradation

129

of M would not be unexpected because there is much vibrational and thermal energy in
the matrix during keV-Xe bombardment which could induce the facile l, 2-elimination
cleavages of the polar hydroxy-groups and neighboring hydrogens to form a less polar
analyte molecule that would be easier to desorb. If [M-3H20] is desorbed into the gas
phase, it would be likely protonated in the selvedge region and detected as [(M-
3H20)+H]+ fragment ion at m/z 355 also. Thus, if K+ is injected into the selvedge region
by K+IDS-by-FAB, the types of analyte molecules desorbed, M or [M-3H20] can be
identified from gas-phase K+ adduct formation. However, if the ions appearing in cholic
acid mass spectrum are desorbed preionized, injection of K+ into the selvedge region will
not yield any adduct ion formation since two ions of same polarity will not react.

The K+IDS-by-FAB mass spectrum of cholic acid is shown in Figure 36b. One
new peak is observed in the mass spectrum at m/z 447 representing the [M+K]+ adduct
ion of cholic acid. There is no evidence of a peak at m/z 393 representing the [(M-
3H20)+K]+ adduct ion that would indicate [M-3H20] beam-induced neutral degradation
products are desorbed. Thus, cholic acid molecules are Miami and protonated in
the gas phase by glycerol CI since injection of K+ ions into the selvedge region yielded
an [M+K]+ adduct ion of cholic acid. The [(M+H)-3H20]+ fragment ion is then formed
by unimolecular decomposition of [Mi-H]+ in the gas phase following protonation of M
in the selvedge. For all non-preionized analytes, only desorption of intact analyte
molecules have been observed with KHDS-by-FAB.

As shown in Figure 36a, the cholic acid [M+H]+ ions are quite unstable since the
fragment ions at m/z 355 and 373 are very intense. In Figure 36b, the K+IDS-by-FAB
mass spectrum of cholic acid demonstrates the great stability of the [M+K]+ adduct ion
since there is not an [(M+K)-nH20]+ ion series and the intensity of this peak representing
[M+K]+ is more intense than [M+H]+. The K+ ion forms primarily long range,
electrostatic bonds with molecules because its large size inhibits strong interactions with

analytes. The potassium affinities of most molecules are approximately ten times less

130

than corresponding proton affinities for the same M so the binding energies of molecules
to the K+ ion are much less than the covalent-like proton binding energies [61]. Thus,
the [M+K]+ ions are stable because the K+-M is usually the weakest bond in the [M+K]+
adduct ion so K+ cannot induce much if any fragmentation of M upon adduct formation
because stronger bonds cannot be broken. Additionally, the ionization energy of K+ is
half of most analytes so the charge remains on K+ which limits K+ to inducing chemistry
only by polarization rather than with a charge. The higher intensity of the [M+K]+~
adduct ion peak in comparison with [M+H]+ is consistent with a molecule with a PA less
than glycerol as explained above. Last, while it is expected that the PA and potassium
affinity of the [M-3H20] will be less than M, K+ adduct formation with these less polar
neutral degradation products is still expected to occur. Alkali metals will cationize less
polar, small molecules such as monofunctional alcohols [60], unsaturated hydrocarbons
[177], and hydrocarbons [178] owing to the polarizing ability of K+ on the molecule.

As another follow up study to ensure potassium contamination of the
analyte/matrix was not occurring, the molecular weight region of glycerol and several
analytes such as cholic acid, the cardiac glycoside digoxin (MW 780), the oligosaccharide
stachyose (MW 666), the peptide kassinin (MW 1334), and the peptide bradykinin (MW
1060) are compared in Figure 38 following the analysis of these substances by FAB,
KflDS-by-FAB, and FAB of 0.01 M and 0.1 M KCl salts in glycerol. The appearance of
the [M-H+2K]+ peak in a mass spectrum indicates potassium has contaminated the
glycerol matrix surface. In addition, analyte/matrix was placed on the K+ glass emitter
purposely to demonstrate extreme potassium contamination of the glycerol or glycerol
contamination of the K+ glass emitter as occurred routinely when using the wall-divided
target. For glycerol in Figure 38a, the intensities of the peaks representing the K+ ion.
parallel that of the respective [G+K]+ adduct ion for each K+ cationization method. In
the same Figure for K+IDS-by-FAB of glycerol, nearly the same amount of K+ is

injected into the selvedge region as is from 0.1 M KCl salts in glycerol and as a result,

‘5'38 Glycerol
E 70
= 60
50
40
30

a)

It

iv

fl

Rela
088

39
K+

93
[G+H]+

1 3 l 1 69
[G+K]+ [G-H+2K]+

Cholic Acid

    

447
N+Kl+

409
[M+H]+

485
[M-H+2K]+

Digoxin

9)

05888883888

Relative Intens ty_

781
[M+H]+

819
[M+K]+

857
[M-H+2K]+

 

I FAB

D K+IDS-by-FAB
.0.01 M KCl
00.1 M KC1

K+ Glass Emitter

 

 

 

é

Relative Intensity

0588888388

 

 

667 705 743
[M+I-I]+ [M+K]+ [M-l-l+21(]+
1m o 0
£38 Kassmln
U1
'5 50
0 40
E. 30
.3 20 I
a?
1335 1373 141 1
[M+H]+ [M+K]+ [M-H+2K]+
3388 »_l Bradykinin
'5 80
§ 70
5 60
50
o 40
E 30
2 20
a 18
1060 1098 1 136
[M+H]+ [M+K]+ [M-H+2K]+

Figure 38. A comparison between the molecular weight region peak intensities for the
analytes: a) glycerol, b) cholic acid, c) digoxin, d) stachyose, e) kassinin, and f) bradykinin
dissolved in glycerol and analyzed by FAB, by KHDS-by-FAB, by FAB in 0.01 M KC]
also, by FAB in 0.1 M KC] also, and on the surface of the K+ glass emitter by FAB.

132

both methods produce more intense peaks representing [G-l-K]+ adduct ions than 0.01 M
KC] salts in glycerol. The [G-H-l-ZK]+ ion only appeared in the spectrum obtained from
0.1 M KC] salts in glycerol and not KflDS-by-FAB which indicates the [G+K]+ formed
in K+IDS-by-FAB is a gas phase process. The bar graph for cholic acid, a surface-active
analyte in glycerol, in Figure 38b, is more sensitive to potassium surface contamination
since 0.01 M KC1 salts resulted in the appearance of [M-H+2K]+. In the same Figure,
K+IDS-by-FAB of cholic acid resulted in a more intense [M+K]+ peak and no [M-
H+2K]+ peak which indicates [M+K]+ is formed in the gas-phase. Oddly, stachyose, in
Figure 3.22d, a non-surface-active analyte, meaning it dissolves well in glycerol, also has
a similar RI distribution to cholic acid above because the [M—H+2K]+ peak is observed
for 0.01 KC] salts is high, too. However, in contrast, the [M+K]"’/[M-l-H]+ adduct ion
ratio is higher for stachyose than in the mass spectrum of cholic acid. In Figure 38c, K+
cationization of digoxin, another non-surface-active analyte, by KflDS-by-FAB, KCl
salts, and the K+ glass emitter, are very similar at both low (e.g., K'l'IDS-by-FAB and
0.01 M KC1 salts) and high (e.g., 0.1 M KC1 salts and on the K+ glass emitter) potassium
concentrations. The peaks representing the [M-H+2K]+ ion are very low in comparison
to the [M+K]+ for all the methods which indicates digoxin does not readily coordinate
multiple K+ ions. Thus, the peaks representing [M-l-K]+ ions formed in KHDS-by-FAB
for these three analytes are more intense or equal in comparison to the respective [M+K]+
formed from 0.01 M KC] in glycerol and no [M-H+2K]+ formation is observed in
KflDS-by-FAB, which highly suggests a gas-phase K+ cationization process in K+IDS-
by-FAB. Each of these analytes have low PAs in comparison to glycerol and perhaps the
[M+K]+ peaks are more intense than the [M+H]+ because glycerol cannot protonate these
analytes but gas-phase K+ can. In Figures 38e, the peptide kassinin shows the peak
representing the [M+K]+ ion formed in K+IDS-by-FAB is more intense than 0.01 M KC1
salts in glycerol, and in both cases, there was no [M-H+2K]+ formation. However,

another peptide, bradykinin, in Figure 38f, peaks representing [M+K]'*' and [M-H+2K]+

133

are only formed from 0.1 M KC] in glycerol or from contamination of the K+ bead, but
not K+IDS-by-FAB or the 0.01M KC] solution. The lack of a [M+K]+ ion from K+IDS-
by-FAB for bradykinin may suggest that neutral bradykinin molecules are not desorbed
in great quantities. The behavior of these two peptides in the KflDS-by-FAB experiment
will be discussed in more detail below. Thus, for all these analytes, [M-l-K]+ ions formed
by KflDS-by-FAB undoubtedly appear to be different than the [M+K]+ ions formed by
the addition of small amounts of KC] to the glycerol matrix which further suggests in
K+IDS-by-FAB, K+ forms adduct ions with desorbed molecules in the gas-phase and
potassium does not contaminate the glycerol matrix.

Because glycerol CI is responsible for the protonation of desorbed neutral
molecules in the selvedge region, limiting the amount of glycerol on the target would
reduce the number of glycerol-related species that protonate digoxin which would
ultimately reduce the abundance of the digoxin [M+H]+ ion and the fragment ions
derived from it. In a KHDS-by-FAB experiment, 5 ug of digoxin (MW 780) was added
to and mixed with a very minimal amount glycerol on the target. As a result, a low
intensity peak representing [M+H]+ and no related fragment ions were observed as
shown in Figure 39. However, the peak representing the [M+K]+ adduct ion was more
intense than the [M+H]+ because K+ was the dominant reagent ion. No fragment ions
related to the [M+K]+ ion were observed owing to its great stability.

In FAB, either intact analyte molecules, preionized intact analyte molecules, or
both are desorbed into the gas-phase, and K+IDS-by-FAB can distinguish between these
two analyte forms since K+ will not interact with other positively-charged analytes. For
example, in Figure 40a, the analysis of the salt, benzyltriethylammonium chloride (MW
227), in glycerol by FAB shows an intense peak at m/z 192 indicative of the
benzyltriethylammonium cation. If the benzyltriethylammonium cation was desorbed
directly as a neutral molecule [R4N-H] or a R4N+Cl' ion pair from the glycerol, then the

K+IDS-by-FAB mass spectrum of this same salt in Figure 40b would display a peak

134

 

 

 

.86on .556 Sec
coinage saunas—swat 0: m_ 205 van +213: 9 gangs—8 E 8:25 22: fl :2 8:23
+573: 05 wet—.8058 mama 05 85m 83 32V Exowfi «o 8.58% 39: m<m43
-mOTM £5 E :32? mm :2 “common 28.—«Eon 05 2:83 +M i=2 “comma van—EAEQQEw
05 8.62 8 838 05 :o 3.59:» we E508“ :95 306950 5 warn—haw am 6m 293n—

Sw o2. coo com 8* 8m com oo—

 

:—

+E+OS
ww—

 

+E+E
ma

 

 

 

 

AJJSN'EIJNI HAILV'IEIH

135

 

 

 

 

 

 

 

 

 

 

 

 

R100. CH 192 FAB
f Ia) l 3 [C13szN]+
a 80' (IzHZ C1'
+
i @ CH2—N—CH2CH3 [(C13H22N+C1') C13H22N]+
" 60‘ I 419
e ‘ CH
1 ' ' 2
n 401 CH3 [G+C13H22N]+
f, . [G+H]+ 284
n ‘ 93 [26+H]+ x 5 4. 9"
is 201 75 185 [C H N]+
11 16
1’ C :‘c4151;if.lr. .1132. r. .‘fi‘: ,‘. v . 1L. . 1
100 200 300 400 500
1112
100
R j b [C13H22N]+ KHDS-by-FAB
C
1 ‘ 192
a 80“
E , ' [(C13H22N+C1’) C13H22N]+
v 60- 419
e d
I i
n 40" [G+C13H22NT"
g ; [C7117]+ 284
9] ##‘fi‘4MJ‘7 .A- v . , ~.
g 201 86 1%8Hl4N]+ X5
i . [C5H12N]+ [C11H16N]+
t . 162 1
yo- .4..-;,...Tr....
0 100 200 300 400 500

m/z

Figure 40. a) FAB mass spectrum and b) KfiDS-by-FAB mass spectrum of'
benzyltriethylammonium chloride (MW 227) in glycerol, with L0 present, obtained from a

regular FD/FAB target and KflDS-by-FAB space-divided target, respectively.

136

representing either a [R4N-H+K]+ or a [(R4N+Cl')+K]+ adduct ion, respectively.
However, these particular ions were not observed for K+IDS-by-FAB of
benzyltriethylammonium chloride above, nile blue (353) or tetrabutylamine (MW 242).
In addition, if these analytes and 0.01 M and 0.1 M KC] salts were dissolved in glycerol
and analyzed by FAB, there were still no new K+ adduct ion peaks in the spectra
indicating the desorption of neutral species from salts. This evidence suggests that
preionized analytes such as cations are desorbed intact directly from the matrix in FAB.
This observation is in agreement with the well known fact that desorbed preionized
analytes yield higher ion currents than desorbed neutrals because these cations become
surface-active in positive ion mode, populate the top most glycerol layers for enhanced
desorption and do not require gas-phase ionization. However, this conclusion is
precluded by the fact that ion pairs of salts, R4N+Cl', may be desorbed into the gas phase
and undergo reactive collisions with ions such as [G+H]+ or K+ in the selvedge region
which simply produce the products R4N+, G and KC] or HC] if K+ or [G+H]+ was the
reactant ion, respectively (for the specific Equation, see Appendix one). Last, the K+IDS-
by-FAB analysis of a phosphatidylcholine mixture in Figure 41b shows K+ interacts
favorably with neutrally-charged analytes containing an equal number of highly basic and
acidic charged groups in the gas phase since [Mn+K]+ adduct ions are observed for each
component in this mixture.

Peptides exhibit properties similar to salts and phosphatidylcholine because in
solution, they may contain both ionized basic and acidic sites. However, in FAB, it is
unclear whether peptides retain their charged basic and acidic sites during desorption into
the gas-phase, or only neutral or positively-charged peptides are desorbed into the gas
phase. For the solution chemistry of peptides, fractional composition diagrams in Figure
42 were calculated to describe the acid-base equilibria of these polyprotic analytes based
on the pKa's of each basic and acidic group [179]. Since the pH of the water (a co-

solvent) and glycerol matrix would be approximately between 6 and 8, the dominant form

137

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R u.
760 +
f « 3) [M1+H]+ (CH3)3N-CH2-CH2-9 FAB
:1 - O=P—O'
i - CH2 — CH — CHz—O
v I I
e - 758 9
I ‘ , C=O C=O M1 n, m =16,12
l | n, m = 14, 14
:‘ [M2+I-I]+ ($H2)n (C'H2)m
e 786 788 CH3 (11H M2 mmifl:
n 11. m = .
.s ' 9H
1 . 810 CH3
t 848
y . -. -111 . . ,AJIIEO. . . ,
740 760 780 800 820 840 860 880
mlz
5 b) 760 Imus-o m
1 * [M1+H]+ Y
a .
t
i .
V
c ‘ 758
I .
I‘ ‘ [M2+H]+
: " 1 786788 [M1+K]+
'8 ~ 791298 [M2+K]+
' - 8 0 826
t 824 848850
y - .. , , 1 .
740 760 780 800 820 840 860 880

Figure 41. a) FAB mass spectrum and b) KHDS-by-FAB mass spectrum, obtained with a
regular FD/FAB target and the KflDS-by-FAB space-divided target, respectively, with L0
present, of a glycerophospholipid mixture containing 1-stearoyl-2-palmitoleoy]-3-
phosphatidylcholine and/or l-pa]mitoy]-2-o]eoyl-3-phosphatidy]choline as M] (MW 759),
and l-stearoyl-2-oleoy1-3-phosphatidy]cho]ine and/or l-pa]mitoy]-2-arachidonoyl-3-
phosphatidylcholine as M2 (MW 787), in glycerol.

138

a) W:
1. H-VGVAPG- 0H (MW 498)
At pH=7, dominant form 15 a1

$VGVAPG® .2

2. H-LGG-OH (MW 245)

3. Leu-Enkephalin (MW 555)
H-YGGFL-OH

4. Met-Enkephalin (MW 573)
H-YGGFM-OH

 

 

 

 

 

 

b) 5.1.41.2 I.“
1.Kassinin (MW 1334)
H-DVPKSDQFVGLM-NHZ

At pH=7, dominant form is 012
$DVPKSDQFVGLM-NH2
G 9 G)

 

 

 

 

 

C) Wiles:
1. Bradykinin (MW 1059)
H—RPPGFSPFR-OH

At pH=7, dominant form is 011
egPPGFSPFgG

 

2. Pro-Enkephalin (MW 899) *
H-YGGFMRGL-OH

3. Substance P (MW 1347)
H-RPKPQQFFGLM-NH2

 

 

 

 

 

 

 

 

 

 

 

 

4. Angiotensen [1 (MW 1045) . . . . ° ' . .

H-DRVYIHPF-OH . . . . - .. .. - . .. .. - .

5.GramicidinS(MW1140) o l 2 3 4 5 6 p; 8 9 10 11 l2 13 14

lrFPVOLFPVQ 100 ‘____..._._.
90 .

d) . . . so . . / —-—°o

W: 70 , a

l. Sleep-Inducing Peptide 60 1

(MW 848) as so . . a2

40

H-WAGGDASGE-OH 3o ' ° “3

At pH=7, dominant form is a3 :3 i \ / . —.—a.4

@WAGGDASGgG o. .A , ¢_, ; - . . .

9 o 1 2 3 4 5 6 7 8 9 lo 11 12 13 14

Figure 42. Fractional composition diagrams for polyprotic peptides a) H-VGVAPG-OH, b)
kassinin, c) bradykinin, and d) sleep-inducing peptide in which the prominent forms at
pH=7 and other peptides exhibiting similar behavior are listed on left hand column.

139

of the hexapeptide H-VGVAPG-OH (MW 498) in the matrix, as indicated by the
fractional composition diagram in Figure 42a, would be neutral - protonated on the N-
terminal amine and deprotonated at the C-terminal carboxylate group. Because H-
VGVAPG-OH does not contain any highly acidic or basic side chains, it exists as a
charged but neutral species overall in solution. In Figure 43b, KflDS-by-FAB of H-
VGVAPG-OH shows neutral molecules are present in the gas-phase of a FAB experiment
because a small peak representing [M+K]+ is observed. While in solution, this peptide
simply self-protonates the N-terminal amine from the C-terminal group in solution, but
for desorption, this process is reversed creating a neutral molecule in the gas phase, and it
is protonated again by [G+H]+ in the gas phase as indicated by the fragment ion
structures formed. The peptide kassinin (MW 1334) contains an equal number of
ionizable basic and acidic residues/sites as shown in Figure 42b, and in solution between
pH 6 and 8, it is neutral. K+IDS-by-FAB of this kassinin yields a strong [M-l-K]+ adduct
ion indicating it is desorbed into the gas phase as a neutral (see Figure 33). In Figure 38f
above, K+IDS-by-FAB of the peptide bradykinin (MW 1059) reveals an intense [M+H]+
ion at m/z 1060, but no corresponding [M-l-K]+ adduct ion is observed. Only high
amounts of KC] in glycerol yield a [M+K]+ adduct ions of bradykinin. Bradykinin
possesses two ionizable basic arginine amino acids, and it is believed that it exists in the
glycerol matrix as a overall positively-charged species as shown in Figure 42c. The other
four basic peptides listed in Figure 42c are similar to bradykinin and also do not yield an
[M+K]+ adduct ion in KflDS-by-FAB supporting preionization in the matrix. Thus, if
bradykinin is positively charged in the matrix, it will likely be desorbed as a preformed
ion [M+H]+ and detected at Bill 1060. Various porphyrins also analyzed by K+IDS-by-
FAB resulted in no K+ adduct ion formation in accordance with the observation of
preionized porphyrin species by Musselman et a]. [180]. In Figure 42d, the dominant
form of the sleep-inducing peptide, H—WAGGDASGE-OH (MW 848) in the matrix at pH

7 is negatively-charged overall because it contains two ionizable acidic amino acid

140

 

 

 

 

 

 

 

 

 

15
R . _ _ [M+H]+
i ‘ a) H VGVAPG OH 499
a , FAB
i 1
v 10,
e 1
I l
i‘ l
g 5- b4
5 ‘ 327
i 1 (15 b5 65
y 0' 396 424441
300 350 400 450 500 550 6130
Z
5 1 b) H-VGVAPG-OH 499
1 60* KflDS—by-FAB lM+Hl+
1
g t
I 40‘.
1‘
e ‘ b
n 20. 4
f . 327 [M+K]+
‘ ‘ “5 65 'V 537
y o' 396 441 456
300 350 400 450 500 550 600
m/z
100
: ) “D 500
‘ C H-VGVA PG-OH [M+H]+

on
‘?

KflDS-by-FAB

t<H~°mgn~5H 0<""'"N""GW
O\
O
1 L - -

 

 

 

 

Figure 43. a) FAB mass spectrum of H—VGVAPG-OH (MW 498), b) K+IDS-by-FAB
mass spectrum of H-VGVAPG-OH, and c) KflDS-by-FAB mass spectrum of H-

VGVA‘XDPG-OH (MW 499) in glycerol using either the FD/FAB or KflDS-by-FAB
space-divided target, respectively, with L0 installed.

141

 

 

 

 

 

 

 

 

 

   

    
       

 

60
f . 737 [M+H]+
? . [8G+H]+
i 404
2; ' [9G+H]+
1 , 829
i] 20- C8 [IOG+H]+
e ‘ -w 921
g 719
i
t 1
y l ‘

650 700 750 800 850 900 950
m/z

R * 349 rams. -FAB
f lb) H—WAGGDASGE-OI-I [M+H]+ by
a 1
t .
‘v :
e : [M+K]+
I « 887
{' 1 [8G+H]+
e 4 C8 737
n ; -w -13 [9G+H]+ [lOG+H]+
is . 684 719 a9 829 921
t 803
y

650

750

850

 

900

950
m/z

Figure 44. a) FAB mass spectrum and b) KflDS-by-FAB mass spectrum of the sleep-
inducing peptide, H-WAGGDASGE-OH (MW 848), in glycerol, with L0 installed, using

the FD/FAB and KflDS-by-FAB space-divided targets, respectively.

142

residues. Despite the negatively-charged peptide in solution, in Figure 44b, a routine
K+IDS-by-FAB mass spectrum indicates neutral sleep-inducing peptide molecules are
desorbed intact by the presence of a peak representing [M+K]+ at m/z 887. In positive
ion mode, it is unlikely that an overall negatively-charged peptide will be desorbed intact
and preionized. Thus, the neutral forms of the peptide H-WAGGDASGE-OH are
probably desorbed from the more acidic equilibria instead, which may account for the
poor sensitivity observed for this peptide in FAB. Thus, KflDS-by-FAB experimental
results and the fractional composition diagrams suggests that only peptides with very
basic side chains are preionized in glycerol in positive ion mode, and desorption of these
positively-charged preionized species intact in positive ion mode FAB is probably more
favorable than desorption of the neutral form. However, another explanation of this
phenomenon is peptides with basic side chains may be desorbed as a intact neutral
molecules from the more basic equilibria in glycerol, and because they have very high
PAs, they are more competitively protonated by [G+H]+ than cationized by K“, which
may also be why an [M+K]+ ion is not observed for bradykinin. Perhaps when higher
concentrations of K+ are present in the selvedge from 0.1 M KC1 as shown in Figure 38f
for bradykinin, then K+ cationization becomes more competitive by numbers. Highly
basic peptides have are very low detection limits in the FAB experiment which supports

both intact desorption of preionized species or competitive protonation.

4. Application of NMR for Determining the Charge State of Peptides in

Solution

NMR studies of bradykinin were initiated to determine if bradykinin was
positively charged in water and glycerol at pH 7 and if possible to determine the site of
protonation. Although the sites of protonation and deprotonation in amino acids are well

known, these same sites may be different in a highly functionalized peptide with a

143

molecular weight of over 1000 Da. The Varian VXR series 500 MHz NMR instrument
was used in these studies. First, using 1H NMR, small peptides such as H-VG-OH, H-
GV-OH, H-RG-OH, H-GPA-OH, and H-VGVAPG-OH dissolved in D20 were
spectrally characterized to link the various types of chemical shifts in the one- and two-
dimensional NMR spectra to the respective protons on the peptides. The protons
identified on the N—terminal amino acid residue had very small chemical shift differences
in comparison to the coupling constants so the peaks were overlapped and showed
minimal splitting patterns. However, the rest of the protons on the other amino acid
residues in the peptide generally gave rise to highly complex splitting patterns meaning a
great number of these protons were in unique chemical environments. This observation
may indicate that the N-terminal amine nitrogen is rapidly exchanging protons with the
solvent and it affects the chemical shifts of the local protons in the N-terminal residue. A
rough pH titration with H-VGVAPG-OH showed this peptide to be protonated either on
the N-terminal amine nitrogen or the proline amide nitrogen because the chemical shifts
of protons on carbons surrounding these two nitrogen sites moved upfield as the solution
pH became increasingly basic. Protonation of a site on a molecule alters the electron
density at the site creating a different chemical environment so the protons in the new
chemical environment give rise to new chemical shifts when in resonance. Next,
standard one-dimensional, and two-dimensional correlated spectroscopy (COSY) 1H
NMR spectra were obtained for 10 milligrams of bradykinin in both D20 and 10%
DzO/90% H20 at 30° C. The temperature was set high to minimize the overlap of the or-
hydrogen chemical shifts by the water peak, HDO. Previous work in the literature
identified the chemical shifts of bradykinin at high fields [181] and conducted pH
titrations with bradykinin [182, 183] successfully. A majority of the chemical shifts in
our bradykinin NMR spectra were identified with similar assignments as in the literature
but some of our chemical shifts were different or unique. This may be a consequence of

having the probe above room temperature because the large protonated arginine side

144

chains may swing excitedly and interact with multiple chemical environments creating
new peaks. A pH study of bradykinin in H20 would be the best approach because pD
values from D20 differ slightly from pHs in H20, and the amide nitrogen protons can
only be observed at chemical shifts between 7 and 8 ppm in H20 and not in 100% D20.
The 10% D20 is still needed to reference the chemical shifts. Unfortunately, the use of
H20 produces a large and wide HDO peak in the a-hydrogen chemical shift region at
4.65 ppm. This water peak was never fully eliminated by water suppression pulse
sequence techniques, and in turn, it overlapped with many of the protons on the Ot-
carbons, which best revealed when an amine or amide nitrogen was protonated. The best
water peak suppression technique involved using an inverse detection probe, the "presat"
pulse sequence, and a highly optimized set of higher-order magnetic field shims.

For future NMR research in determining the charge state of peptides in solution,
the latest NMR techniques have to be utilized because high MW compounds produce
difficult spectra to interpret and provide many challenges in the process of obtaining the
spectra. In a peptide, the amine and amide nitrogen protons would be the likely sites of
protonation. Conveniently, these amine and amide proton chemical shifts will move
upfield upon protonation revealing which nitrogen is protonated provided the analysis is
done in water and the water peak is attenuated. The amide chemical shifts can be
assigned in the peptide using two-dimensional COSY techniques once HDO is
attenuated. In addition, HMQC two-dimensional pulse techniques, using the inverse
detection probe, show a two-dimensional plot of the 15N and 1H linkages which may also
be useful in peptide pH studies, but the technique is not overly sensitive because it
essentially detects the 15N satellites of protons. However, it is much more sensitive than
15N NMR [184]. These techniques were attempted using the peptide, bradykinin, but
success was limited because the water peak was never fully attenuated which produced
poor analyte sensitivity and hid many important (it-hydrogen chemical shifts. A

subsequent NMR study with bradykinin dissolved in glycerol would confirm if

145

bradykinin was indeed protonated in this FAB matrix, but this study was never attempted
owing to the difficulties experienced with suppressing the water peak and optimizing the
appearances of the peptide amide protons in water. It is well known that glycerol is
highly viscous and hydroscopic which may potentially interfere with the bradykinin
analysis above. However, even though glycerol is not the ideal solvent, the NMR experts
still believe that peptides dissolved in glycerol can be analyzed with much involved

instrumental tweaking.

5. Conclusions

To date, the identification of the types of analyte molecules desorbed in a FAB
experiment has not been accomplished experimentally from a mechanistic point of view,
and the KflDS-by-FAB technique provides this capability. Thus, FAB really appears to
be a desorption/ionization technique because neutral analyte molecules from the glycerol
matrix are desorbed intact into the glycerol-CI protonating environment, known as the
selvedge region. These gas-phase neutral intact molecules may undergo ion/molecule
reactions with [G-l-H]+ in the selvedge and emerge as protonated analyte molecules.
After protonation, [M+H]"’ ions with excess internal energy undergo unimolecular
decomposition in the free vacuum to form the fragment ions observed in a FAB mass
spectrum. The extent of [M+H]+ fragmentation is proportional to how long the [M-l-H]+
was collisionally stabilized in the selvedge; the longer the residence time, the less
fragmentation will be observed and vice versa. K+IDS-by-FAB further demonstrates that
K+ ions are very useful for probing the types of gas-phase neutral molecules in a high
pressure region. This distinct advantage of K+ may have mechanistic or analytical
applications elsewhere in mass spectrometry such as neutralization-reionization mass
spectrometry (NRMS). The KflDS-by-FAB technique may also have application in

probing the types of analyte equilibria in solution as demonstrated by the peptide studies.

146

More K+IDS-by-FAB work should be done to investigate whether glycerol is truly the
best matrix and whether there truly exists a proton/K+ ion competition in the gas phase
selvedge region. K+IDS-by-FAB should also be used with analytes having unsaturated
sites to see if K+ polarizes these molecules at the double bonds to form adduct ions. The
fact that intact, polar, neutral molecules are desorbed into the gas-phase in FAB may open
up new applications for this technique elsewhere in chemistry concerned with studying
intact, gas-phase molecules with molecular weights exceeding 1000 Da. Characterizing
the two- and three-dimensional structures of gas-phase peptide ions and neutral molecules
desorbed by FAB with another analytical technique would benefit mass spectrometry

immensely to reveal if the fragmentation patterns are linked to the gas-phase structures.

CHAPTER FOUR. THE EVOLUTION OF FRAGMENT IONS
FROM PROTONATED PEPTIDES IN LIQUID
SECONDARY-ION MASS SPECTROMETRY:

MECHANISTIC AND THERMOCHEMICAL
CONSIDERATIONS

I . Introduction

The scientific aspects of mass spectral interpretation lie in our ability to
understand how the ions that are represented in the mass spectrum of a given compound
evolve. With fundamental information on the structure of the nascent molecular ion, and
the chemical processes through which fragment ions are formed, correlations between
spectra and structures can be made, and interpretative skills can be developed. In electron
ionization (EI) mass spectrometry (MS) [2], clastograms and the results from isotopic
labeling experiments have made mass spectral interpretation as much a science as it is an
art. Appearance energy measurements have allowed molecular ion decomposition
mechanisms to be suggested, at least at the appearance energy threshold, for small
molecules. With thermochemical data for fragment ions and possible neutral fragments,
the amount of energy required for each ion decomposition pathway can be estimated.
The average internal energy of the molecular ion determines, in part, the reaction rates for
fragmentation. Few average internal energies of molecular ions following 70-eV EI have
been determined [185] which limits the usefulness of such analyses of thermodynamic
processes. However, it is clear that the major ions represented by the mass spectrum are
formed by cleavage of 1 or 2 bonds in the molecular ion, with little subsequent
unimolecular chemistry. Characterization of metastable ion (MI) decompositions and
products of collisionally-activated dissociation (CAD) have also contributed to

understanding fragmentation processes and pathways. MI decomposition definitely

147

148

establishes a precursor/fragment ion relationship, but care must be exercised because
metastable processes need not be identical to the mechanism that is responsible for the
majority of the fragment ions (at a particular m/z value) in question. If CAD is used to
induce fragmentation of every ion in an EI mass spectrum, a complete MS/MS map is
obtained, and, again, while this map may suggest possibilities for fragmentation
pathways, it need not represent what actually occurs in the ion source to yield the EI mass
spectrum.

Desorption/ionization (D/I) techniques, notably fast—atom bombardment (FAB)
[63] and liquid secondary-ion mass spectrometry (LSIMS) [27], have allowed for the
mass spectral analysis of thermally labile "biomolecules". Unlike with El spectra, the
sophistication of spectral interpretation is not yet available for mass spectra generated by
D/I-techniques. In the past decade, the focus of mass spectrometry has changed from E1
to D/I based methods, from analyte molecular weights of less than 600 to molecular
weights greater than 3000, and from a low to a high density of functional groups in the
analyte molecules being characterized. In addition, further complications, such as the
requirement of presenting the analytes in a matrix for analysis, are now encountered. In
the literature, the interpretation methods used for analyzing FAB or LSIMS spectra of
biomolecules are primarily based on correlating 1er values with proposed ion structures,
and thus with analyte structure. The elucidation of mechanisms requires a more complete
understanding of the correlations between analyte structure and all of the information
contained in the mass spectrum - m/z values MW.

The tools used to develop an understanding of how fragment ions are formed in
E1 have no parallel in FAB or LSIMS. The generation and interpretation of clastograms
for small molecules in EI-MS is fairly straight forward. As the electron energy is
increased, the average energy deposited into the molecules increases, and, as a result, the
rates for dissociation following ionization increases. There is no analog to the clastogram

in FAB or LSIMS. In the positive ion mode of analysis using FAB or LSIMS, if the

149

analyte is a neutral species (not "preionized"), desorbed analyte biomolecules are
protonated in the gas phase, and, subsequently, fragment ions evolve from the excited
protonated molecules, [M+H]+*. The desorption, protonation, and unimolecular
decomposition processes are energy-dependent and not well understood [79], so an
experiment in which the energy or flux of the FAB beam is varied does not give data such
as that contained in an El clastogram. Also, very few mechanisms for fragmentation of
[M+H]+, formed in FAB or LSIMS, in the case of peptides, have been established at the
level of confidence found in the EI-MS literature.

The goal of this work is to develop a "mechanistic context" in which we should
approach the interpretation of D/I mass spectra of peptides, with the goal of developing
an understanding of not only the 1er values of the ions formed, but the relative intensity
(RI) information presented in a mass spectrum, and its relation to the analyte peptide
structure. The focus here is not on fragment ion structures of peptides, which have been
proposed, but on possible complete mechanism, since the probability of a given process
is clearly related to the stabilities of both fragments. The data set considered includes
estimated reaction enthalpy data for various peptide decomposition mechanisms and the
normalized fragment ion RIs from LSIMS, MI decomposition and high energy, single
collision CAD spectra. This set of spectra contains fragment ion m/z and RI data for ions
formed under a variety of conditions. The pentapeptides pentaglycine and methionine
enkephalin are contrasted in this study. This work evaluates fragmentation processes
with a focus on protonated molecule structure, fragment ion structure, fragment neutral
structure, and the thermochemical requirements of the various mechanisms. We will
discuss the possible routes to fragment ions from the protonated peptide, which will
provide a focus for evaluating the information contained in the RI data. A correct
mechanism should be the key to establishing correlations between analyte peptide
structure and fragment ion RIs. However, such reasoning is cyclic and it may be the RI

information that is the key to establishing the mechanism. A single, consistent picture for

150

the spectrum/structure relationship is sought here. In contrast with the situation in El, it
becomes clear that we are truly facing new challenges in D/I spectral interpretation. In
E1, the ionized molecule was first defined, and then an understanding of the major
fragmentation mechanisms leading to intense or unique, informative peaks was
deve10ped. To reach the same level in FAB or LSIMS analysis of peptides, a substantial
number of starting points need to be considered because analyte molecules have many
possible protonation sites and specific secondary structures, both of which appear to

determine subsequent fragmentation chemistry.

2. Experimental Details

Instrumental Parameters.

All mass spectra were obtained on a JEOL HX-110 double-focusing, forward
geometry (EB) mass spectrometer equipped with the JEOL FAB gun, the JEOL SIMS
gun and ion source, and the JEOLlMA-DASOOO data system. For all experiments, the
accelerating voltage was 10 keV, the resolution was 1000 (10% valley definition), and the
scan rate was from m/z 1-1500 min'l. The fast atoms used were 6-keV xenon atoms and
the fast ions used were 6-keV Cs+. MI and CAD product and precursor ion mass spectra
were obtained using linked scanning at constant B/E and B2/E, respectively. Single
collision CAD was achieved by introducing helium into a collision cell (located in front
of the main slit) to reduce the precursor ion signal by 10% [135]. Repetitive scanning
with spectral summing for signal-to-noise enhancement was not done in order to avoid
time-dependent RI data. To ensure RI accuracy, replicate experiments with the same
peptide were compared.

To consider actual CAD processes, the "true" single collision CAD experiment

was desired. A single collision CAD spectrum is routinely obtained by selecting a

151

precursor ion in the "normal" LSIMS spectrum, introducing the He collision gas to reduce
the precursor ion abundance by 10% and scanning the mass analyzer to detect the product
ions. However, in addition to collecting CAD product ions, metastable ions are also
collected which results in a mixed MI/CAD spectrum. The metastable ion contribution to
the CAD spectrum was determined by consecutive scanning in which the metastable ions
were sampled during the first scan, and, during the magnet reset time, the He collision
gas was turned on (10% beam reduction was set prior to experiment) and the MI/CAD
products ions were then sampled during the second scan. LSIMS was used to obtain the
MI and MIICAD mass spectra instead of FAB. FAB resulted in MI spectra with almost
identical RI distributions as MI/CAD spectra preventing meaningful subtraction because
additional CAD products were produced from the highly effective, energizing collisions
between Xe and [M+H]+. Because LSIMS does not require Xe, the additional CAD
fragmentation was eliminated. As a result, the peak intensities and ion types in the MI
spectra using LSIMS were clearly unique from the MI/CAD spectra at 10% beam
reduction, and meaningful subtraction, MI/CAD minus MI, could be performed to obtain

the "true" single collision CAD spectra.

Sample Preparation.

All peptides were purchased from Sigma Chemical (St. Louis, MO) and used
without further purification. The peptide methionine enkephalin was dissolved in water
to a concentration of 2 [lg/11L and pentaglycine was dissolved in 0.1 M HCl to a
concentration of 2 [lg/11L. On the direct insertion probe tip, a small drop of glycerol was
applied to the center of the target with a thin (0.007 in) wire, and l 11L of a peptide
solution was placed in the glycerol and mixed thoroughly. Pentaglycine did not dissolve
completely in 5% acetic acid which resulted in high background noise and low analyte
signal. However, by dissolving pentaglycine in 0.1 M HC], the high background noise

was eliminated and the analyte peaks were clearly revealed with the same RI ratios as

152

when a 5% acetic acid solution was used. Because the low pH may cause hydrolysis,
pentaglycine solutions were prepared immediately before use. The use of plastic vials for
sample storage instead of glass vials were very important for preventing the appearance
of the interfering peaks representing [G+Na]+ and [M+Na]+ adduct ions (G and M
represent glycerol and analyte molecules, respectively) in the mass spectrum because

peptides complex to sodium ions that are available from the walls of the glass vials.

3. Results and Discussion

Peptide Fragment Ion Nomenclature.

The model pentapeptides discussed here are methionine enkephalin and
pentaglycine which contain the amino acid residue sequences H-YGGFM-OH and H-
GGGGG-OH, respectively, using single letter residue designations. The H- and -0H
designate the N-terminal amine hydrogen (not to be confused with the designation for
histidine, H) and C-terminal hydroxy] group, respectively. We will refer to, for example,
pentaglycine as H-GGGGG-OH, rather than GGGGG, to make certain that the
relationship between a designation of a peptide fragment and its mass is always obvious.
For example, if one or two bonds are cleaved in pentaglycine, one may discuss the
fragment GG. What is its mass? This will be apparent because the internal fragment, N-
terminal fragment, and C-terminal fragment, each containing two residues, will have the
unique designations GG, H-GG, and GG-OH, respectively. We are also specifically
indicating the H- and -0H components of the peptide because, to be accurate, the single
letter designations represent peptide residues.

The peptide sequence ions are labeled an, bn, c", xn, yn and 2", using the
nomenclature described by Biemann [186] in which the most common fragment ion

structures are assumed. The subscripted n for N- and C-terminal sequence ions

153

designates the number of residues contained in the fragment. For methionine enkephalin,
the structures for the N- and C-terminal sequence ions, internal ions, immonium ions,
[an+H]+' ions, and side chain ions are shown in Figure 45. The an, bu, xn, and 2,, ions are
simple fragments in which the protonating proton remains on the neutral fragment and no
hydrogen shifts; the cn and yn ions are the result of secondary interactions in which a
remote hydrogen shifts onto the ionic fragment where the protonating proton remains
fixed. For the non-sequence ions, the immonium ions and internal ions involve cleavage
of at least two skeletal bonds, while side chain ions and [an+H]"'° ions are probably the
result of one bond cleavage. The fragment ions not shown in Figure 45 include side
chain loss ions (denoted as -Y if the tyrosine side chain was the neutral radical fragment
lost), neutral loss ions (denoted by —HzO, etc.), and [bn+OH+H]+ ions (i.e., loss of the C-

terminal residue with an ~0H shift to an internal residue).

Types of Mass Spectra from an EB Instrument.

From a double-focusing EB mass spectrometer equipped with a SIMS source,
three different types of mass spectra can obtained: (a) LSIMS, (b) MI, and (c) CAD,
which are shown in Figure 46 for the pentapeptide methionine enkephalin (MW 573).
The LSIMS spectrum in Figure 46a is very similar to previously reported FAB mass
spectra of methionine enkephalin [65, 187]. The single-collision, high-energy CAD mass
spectrum of [M+H]+ is shown in Figure 46c. The distribution of relative peak intensities
are different from reported high-energy CAD spectra at 50% [188] and 75% [189] beam
attenuation, since multiple collision conditions existed in those studies. If the collision
gas, helium, is shut off, then abundant MI decomposition is observed, as shown in Figure
46b. Thus, the single-collision CAD mass spectrum in Figure 46c contains significant
contributions from MI product ions, so by subtracting the MI mass spectrum in Figure
46b from the CAD mass spectrum in Figure 46c results the "true" CAD mass spectrum

shown in Figure 46d. The mass spectrum shown in Figure 46c will be labeled as

154

(R1 = H0C6H4CH2- and R2”; = H) (R4 = C6H5CH2- and R5 = CH3SCH2CH2-)

G)
H NH H NH c
69
HzN/ \CI/ \\(|2H o=c= / \cl/ \(ljfi \0H
m ° "2 ° “5
03+ R
1 0 R3 R4 0
l. l .1 l. l
/ \ \ /
H2N fi/ |fi \gI-Iz H3%/ \C/ \ \OH
0 R2 115
... . a
. o
/ \ / \{I'ICD 691;; NH E
H N c
2 n T 0’ \94 \OH
0 ... l 1
5
(9 R1 [1
H NH c mm
HZN/ \(i/ \ffi $3 (R3=HandR4=C6H5CH2-)
o ., ... fi
- )3 /NH /C®
Immnnllualnn HzN c \CH
(R4 = C6H5CH2-) || |
0 R4
CD ‘14..
CH @
“2%)? fltg )3
HZN fi \CH
0 R4

Sidaflhainjnn
.138

Figure 45. Representative fragment ion structures for the peptide methionine enkephalin
(MW 573).

155

 

é

) .. F H-YGGFM-OH
Ia ‘20 or T ’2 6hVCs+ISIMS
80‘ . 205221 278 ’2 ..MImM/wm

   
 
   
  
 
  
 

A A A

-$

 

 

 

 

 

~<-~a=o~=~ o<-~n—oz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R
C
l
8
i
l
V
C
r
n
t
it :4 01380113
.5 411 [bmn+m+ 4120
' F 556
t c4 -
3, a5 -NH3
y . 443 4 528
400 500 ml:
R 3‘ H-YGGFM-OH 425 574
e c SingleCdlflomnghEoemCADProdncflomud b4 [Mi-3]"
' 1? a1 . Metastable Ions (MI/CAD)
f 120 136 «4 moss
i 397
V
c -Y
I 467
n
t
f. ,4 0135013
is [woman .320
‘ 4'8F3. 556
y 443 M «15 -NH
4 528 3
mlz
5 3d H-YGGFM-OH b4
1 I "True" CAD Product Ions (Ml spectrum
a 1 0: subtroctcdfromMI/CAD spectrum)
¥ 4 04
I .
V 1
C .
1 : .72 'Y
J
3 j 1 ’3 Ib4+on+m+
s 01m [03+H]+ :4
i : 0787* b, “Ma
n 61" cor .
‘ ‘ \M «:2 ]c [‘3 c -M 'Hzo
Yo.c‘,-.-v,l.lel3l95.1%,”, V‘Jflitwvt
o 100 ' 260 300 400 500 m

Figure 46. a) LSIMS, b) MI, c) MI/CAD, and d) "true" CAD mass spectra for methionine enkephalin
(MW 574). See text for description of spectra and peptide sequence ion nomenclature.

156

MI/CAD because the peaks represent products of both MI and CAD processes. In Figure
46b and c, the [M+H]+ peak is on scale and the actual RI axes are shown which indicates
MI and CAD peak R15 are less than 3% of the LSIMS peaks in Figure 46a. In Figure 47,
the LSIMS, MI, MI/CAD and "true" CAD mass spectra are shown for pentaglycine (MW
303). There are no published LSIMS or high-energy CAD mass spectra of pentaglycine.
The four spectra in either Figures 46 or 47 exhibit very similar fragment ion types, but the
distribution of R13 vary dramatically between the four spectra for a given compound. The
unique RI information in LSIMS, MI and CAD spectra reveal the thermodynamics and
kinetics involved in a D/I experiment which is related to the energy content and

unimolecular fragmentation kinetics for the [M+H]+ ions undergoing fragmentation.

M s r .

The fragment ions observed in the LSIMS mass spectrum are the result of fast
unimolecular decompositions of [M+H]+* ions occurring within 1 microsecond or less
after ionization. The [M+H]+ peak in the LSIMS mass spectrum represents the
protonated molecules that are stable on the time scale of the experiment. In the LSIMS
spectrum for methionine enkephalin (Figure 46a), the immonium, internal immonium,
and side chain ions are most abundant comprising 69% of the total ion current (TIC), the
sequence ions (only an, bn, cn, and y") comprise 28% of the TIC, and the remaining 3%
of the TIC represents the internal ions, and the less abundant side chain loss and neutral
loss ions. The sequence ion abundances for methionine enkephalin are ordered [a1] >>
DJ] > [b2, b3, yz] > [02, 04, b4, cz, C3, C4, y3, y4]; b1, and C] ions are absent.

In contrast, the sequence ions account for 92% of the TIC while the neutral loss
ions account for 8% in the LSIMS spectra of the simpler pentapeptide, pentaglycine,
(Figure 47a). The lack of side chains eliminates the possibility of side chain ions and side
chain loss ions and reduces the possibility of immonium and internal ions because

abundant immonium and internal ion formation usually require a large side chain. Here,

157

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

W” a a‘ a. a 304
R
I: a) [M+H]+
a 80.H-GGGGG-0H
! 6 keV Cs+ LSIMS
'v - s Man-1x Ions/Cs" Adducts
e 60*
1
n
t 40-
e b [03+H1+ [cud-IF [a tart,
2 “I 1125 10 c3 c
i 20‘ 30 . a; c 133 b31171“ 147
l 87 172.190
y ' ”5. ‘ i. t 202 2291259 288 .
“Ill II J ill-I1d Ill I .1 1 1. L
50 100 150 200 360 1311/0
Z
"I
R b H-GGGGG-OH b4 N+élofi
1c Metastable Ions (MI) 2 29
f 100%T
1
v 1* ’3
l 286
{1 b2
e 115
n m ”3
i x5 0 yz "2 14
I ‘ c 133 ‘ 47
y K :4 [as-HIP ““3
h 1 . 1 259 1
U 50 160 150 260 250 31012
304
R C H-GGGGG-OH 229 [M+H]+
e Single Collision. High Energy CAD Product ’3 r1, 4
; Ions and Metastable Ions (Ml/CAD) 1 )0
E 100%
I
v 1‘ -H20
e 286
1
n
l
C
2 w
i 247
‘ lu+fll+ a + + -NH_
y 202 1 c4 [2559"]
ii 11 1 _J_ .11 J
200 250 :30
Z
R 0.4 -
e E d H-GGGGG-OH ’3)
1 : "True" CAD Product lom (Ml spectrum 1
a E subtracted from MIICAD spectrum)
1 0.3g
2 D)
V _
e : 115 ’2
: 133 D}
l 0.25 172
n 2 b4
1 g 229 .1120
c : ’4 286
n 0 IE «2 + MW 247
_S : a) y 87[a2+H]* [rfgm 202 4
1 E 30 1:17; a/ c u [a5+fl]* .an
; = J \1 c2 3 219
0 _
6 5'0 160 150 260 2E0 360 mlz

Figure 47. a) LSIMS b) MI, c) Ml/CAD, and (1) "true" CAD mass spectra for pentaglycine (MW 303).

See text for description of spectra and peptide sequence ion nomenclature

158

the sequence ion abundances for pentaglycine are ordered [b2, )7] > [a 1, b3, )3] > [a2, b4,
c2, C3, C4, y1, y4]; b 1, and C} ions are absent. In addition, the [an+H]+' ion series appears

in the LSIMS spectra of both methionine enkephalin and pentaglycine.

mm

Metastable protonated molecules have initial internal energies which result in
sufficiently low decomposition rates (occurring between 1 and 10 microseconds after
ionization) such that fragmentation occurs after acceleration. For the MI mass spectrum
of methionine enkephalin (Figure 46b), the sequence ions (only an, bn, C", and yn)
comprise 72% of the TIC, and the internal, [b4+OH+H]+, side chain loss, and neutral loss
ions comprise the remaining 28% of the TIC. The sequence ion abundances for
metastable methionine enkephalin [M+H]+ ions are ordered [a4, b4] > [b 3, y;, y 3, y4] >
[a], b2, C2, C3, C4, y1, y4]; the b1, and C] ions are absent.

For pentaglycine in Figure 47b, the sequence ions comprise 81% of the TIC while
the remaining 19% is from neutral loss ions. The sequence ion abundances for
metastable methionine enkephalin [M+H]+ ions are ordered [b4, y3] > [b 3, y2, y4] > [b2,
C2, C3, C4]. In both MI spectra, the immonium, side chain, and [an+H]+' ions are absent

which indicates these are higher-energy product ions.

W

In single-collision, high-energy CAD, a single energetic collision between a
helium atom and an [M+H]+ ion adds additional internal energy to the ion which results
in unimolecular decomposition of the energized ion into CAD product ions. These
[M+H]+ ions subjected to CAD are those that have not undergone prompt dissociation in
the ion source, and thus, they are somehow different from the [M+H]+ ions that give the
fragment ions observed in the LSIMS and MI spectra. The differences could be structural

and/or energetic. In the "true" CAD mass spectrum for methionine enkephalin (Figure

159

46d), the distribution of sequence product ions is nearly identical to the MI spectrum in
Figure 46b, while for pentaglycine in Figure 47d, the distribution of sequence product
ions more resembles the LSIMS spectrum in Figure 47a than the MI spectrum in Figure
47b. For both methionine enkephalin and pentaglycine, the appearance of immonium,
internal immonium, side chain, and [an+H]+' ions in the CAD spectra suggests that some

[M+H]+ ions are activated to relatively high internal energies.

A Model for the LSIMS Experiment.

In LSIMS, both neutral peptide molecules (e.g., methionine enkephalin) and
preionized peptides (peptides that contain very basic side chains such as R, K, H appear
to be protonated in the condense phase), are desorbed intact into the gas phase by keV-
Cs+ particle bombardment [124]. The neutral gas-phase peptide molecules are
subsequently protonated in the high-pressure selvedge region by glycerol chemical
ionization (CI) [85, 122], while the preionized peptides require no further chemistry for
mass spectrometric detection. Both the neutral and preionized species contain an
undetermined but sufficient amount of internal energy from the desorption event to
undergo unimolecular fragmentation to some extent. Attempts to estimate the internal
energy content of desorbed intact neutral molecules and preformed ions have been made
previously. Hoogerbrugge et al. [126] determined the internal temperature of sputtered
glycerol molecules to be 190 K which corresponds to an internal energy of 0.044 eV
assuming only 15% of the total degrees of freedom are active. Derwa et al. [127] and
Williams et al. [128] have estimated, using benzylpyridinium salts in glycerol, that the
internal energy content of preformed secondary ions in FAB and LSIMS is between 0 and
4 eV; this energy distribution is most populated between 0 and 2 eV, with a high energy
tail that extends out to 4 eV. Derwa et al. [127] used Hoogerbrugge's internal
temperature value of 190 K to calculate an internal energy of 1.13 eV for desorbed

benzylpyridinium ions assuming all 69 degrees of freedom are active. Using

a_ i

160

Hoogerbrugge's internal temperature for desorbed neutral molecules, methionine
enkephalin and pentaglycine will have approximate average internal energies of 1.8 eV
and 0.91 eV from desorption, respectively, assuming half of the degrees of freedom are
active. If protonated glycerol is the CI reagent ion in the selvedge region that protonates
the analyte molecules, gas-phase peptide molecules may gain an additional 23 kcal/mol
(1 eV) of internal energy, assuming the proton affinity (PA) of glycerol is 209 kcal/mol
[118], and the experimentally determined PA of pentaglycine is 232 kcal/mol [190].
Following ionization of the peptide molecules by protonation or desorption of
preionized peptides, the greater amount of internal energy an [M+H]+ ion possesses, the
greater the chance it will undergo unimolecular decomposition to form fragment ions on
the mass spectrometric time scale [38]. The different types of peptide fragment ions that
appear in a mass spectrum at various m/z values are determined by unimolecular
fragmentation pathways with favorable reaction enthalpies overall, and by unique
structural features of the protonated molecule, which include the type of side chains,
secondary interactions with groups that are many skeletal atoms away, and hydrogen
bonding in the vicinity of the protonation site. The RIs of these fragment ion peaks are
presumably determined by the distribution of initial internal energies in the protonated
molecules in addition to the transition state structure. The amount of initial internal
energy a protonated molecule possesses in excess of the activation energy for the
fragmentation pathway determines the unimolecular decomposition rate constant. The
highly excited [M+I-I]+ ions quickly fragment in the ion source while the "cooler"
[M+H]+ ions either undergo slow, low-energy unimolecular decomposition to form
metastable fragment ions after acceleration or are detected intact. In single-collision,
high-energy CAD, the low energy [M+H]+ ions are energized by single collisions with.
helium atoms to promote fast unimolecular fragmentation. The maximum internal
energy, expressed as center-of—mass collision energy, available to impart into either

pentaglycine or methionine enkephalin from a collision with helium is 130 eV and 69 eV,

161

respectively, at lO-keV lab energy. However, in a high-energy collision, the average
energy deposited into the [M+H]"' ion is only between 1 and 3 eV [135] and, as a result,
both low- and high-energy fragmentation pathways will be accessible similar to LSIMS.
However, single collision CAD is inefficient overall because to a first approximation,

95% or more of the [M+H]+ ions remain intact while the remaining 5% or less fragment.

Protonated Peptide Model.

The first step in correlating spectra with analer structure is defining the initially—
formed protonated molecule. An [M+H]+ ion containing the basic residues arginine (R),
lysine (K) and histidine (H) are expected to be exclusively protonated on these side
chains because they have basic sites with gas-phase PAs greater than 230 kcal/mol [49].
The peptides methionine enkephalin and pentaglycine do not contain these basic residues;
they have a variety of likely protonation sites available. Single and multiple
proton/heteroatom interaction sites on methionine enkephalin, and their estimated PA
values, are illustrated in Figure 48. The PA values were estimated by matching each
different heteroatom functional group(s) in methionine enkephalin with a small molecule
of known PA that contains the exact arrangement of functional group(s) [40, 176]. From
Figure 48, the most probable protonation sites on methionine enkephalin and
pentaglycine are the amide functional groups along the peptide backbone represented by
proton interaction site 2 since the PAs of these sites are highest, extending from 217 to
224 kcal/mol depending on the model compound and side chain presence [40, 191]. In
Figure 48, protonation of the N-terminal amine nitrogen may occur in three ways: (1) the
proton may be attached solely to N at site 3 which is estimated to have a PA of 217
kcal/mol from a protonated primary amine [40]; (2) the amine proton may undergo a
possible hydrogen bond interaction with the carbonyl oxygen, shown by site 5, which is
estimated to have a PA of 212 kcal/mol from the protonated structure of glycine [40, 192,

193]; or (3) the proton may be further stabilized by a secondary hydrogen bond

162

1‘1:

..osobw .«o <m 2 @2858 2a SQ-EOO>.EV gunned—:0 05:250.: co 8% £93 got? .8 m<m 83850 of. .mV 2sz

I llllll 1T:
0 S 2:

+3
...a mu 0 21w o&
x} mm_ __ :2 N30 __ 04/me a wlv com W
\o/\/o\ wo/\:/\/\/\/\.. WM" m
no mo Eu 0 :2 :0 .....n m
__ EM __ _ Ebebwlv as m
NED w .... fl H O ... NED +m w‘lw N.
who ....5 E + «.... m mm
_ +1: 1?: N\|V
O .
£0

163

interaction with the side chain (site 3, but second hydrogen bond not shown) which was
estimated to have PAs extending up to 222 kcallmol from protonated amino acids [40]. If
the N-terminal peptide residue contains a side chain, then it appears that the basic amine
and amide functional groups will be protonated by protonated glycerol with equal
probability. However, if the N-terminal residue is glycine, then the amide nitrogens will
be preferentially protonated. Since the PAs of the amine and amide sites are similar in
magnitude, intramolecular proton transfer and/or hydrogen bonding will undoubtedly
redistribute the population of initial protonation sites through peptide folding and
favorable secondary interactions which will create a new distribution of protonated
pentapeptide isomers [194]. Some [M+H]+ ions may have more complex protonated
forms than the isomers under investigation in Figure 48.

Because the PA of glycerol is higher than the PAs of the C-terminal hydroxyl
group for both pentapeptides and the methionine enkephalin side chains, protonated
glycerol is an ineffective protonating agent. However, the fragment ions of glycerol have
PAs between 140 and 200 kcal/mol, and it is these species that could protonate the basic
side chain sites and C-terrninal carboxylic acid group [176]. If these glycerol-related
fragment ions protonate a basic site in a peptide, then the protonated molecule may gain
up to 3.3 eV of additional int. nal energy. Protonation of the side chains in methionine
enkephalin probably leads to the abundant side chain ions. However, in peptides
containing K, L, and H residues, the exclusive protonation of these basic side chains
appears to result in the formation of sequence ions through remote-site chemistry [195,
196].

Mass spectrometry also cannot clearly indicate which moiety, the amide nitrogen
or the amide oxygen, is protonated in the gas phase. It is well known that, in the
condensed phase, the amide oxygen is more basic than the amide nitrogen and it is
preferentially protonated [197]. In the gas phase, preferential protonation on the amide

oxygen [40, 191, 192, 193, 198, 199, 200] is also theoretically and experimentally proven

164

since the oxygen is still more basic than the nitrogen for simple molecules. However,
when the amide nitrogen is protonated in B-lactams and azetidines, semiempirical
calculations show the amide bond (-C(O)-NH-) becomes more activated than when the
oxygen is protonated [201]. Thus, protonation of the amide nitrogen will cause instability
which results in a competitive and effective simple amide bond cleavage. However,
when the oxygen is protonated, no bond is activated, so stable protonated molecules are
formed which may be detected intact, dissociate by slow metastable processes, or be
selected for excitation by CAD. Using charge-site initiated chemistry, all of the
candidate mechanisms leading to the observed fragment ions begin with nitrogen

protonation.

Peptide Fragmean Mechanisms.

In El, molecular ions with a low density of functional groups and heteroatoms
provided clear starting points for writing charge/radical-site induced decomposition
mechanisms, but the same is not true for protonated peptides containing a high density of
functional groups in LSIMS. In addition to not understanding the LSIMS ion formation
mechanisms well, the complexity of the ionized peptide structures also prevents sequence
ion fragmentation mechanism from being well understood. In contrast to E1, the radical-
induced fragmentation processes clearly need not be considered for even-electron species
in LSIMS experiment. If the amide nitrogen is protonated, unimolecular charge-site
induced decomposition mechanisms can be written to explain the formation of the an, b",
C", y", and 2,. peptide sequence ions. However, if the amide oxygen is protonated, the
subsequent evolution of these sequence ions is not obvious. This suggests that amide
bond activation and dissociation follows N-protonation. If the N-terminal amine is
protonated, again no obvious decomposition mechanisms leading to sequence ions can be
written when a linear form of peptide is considered. However, protonation of the N-

terrninal amine may contribute to the protonation of the amide nitrogens through

165

intramolecular proton transfer steps, since the PAs of both are similar. Below, the
fragmentation mechanisms for forming the various peptide sequence ions in LSIMS that
include the protonation site are discussed. The elucidation of fragmentation mechanisms
is important for establishing the relationship between the peaks representing the fragment
ions in a mass spectrum and the ionized analyte structure to develop spectral

interpretation skills.

124.122.;

Assuming an amide nitrogen is protonated, a bu ion could be formed by inductive
cleavage of the amide bond as shown in Scheme 1. Recently, Wagner et al. [202],
utilizing selective deuterium labeling in an underivatized peptide, concluded that bu ions
are formed by a simple bond cleavage because no deuterium atoms were shifted from the
a-carbon, nitrogens, or side chains during their formation. Thus, for b" formation, it
appears that no hydrogen atoms are shifted and the protonating hydrogen remains on the
neutral fragment. When a peptide is cationized with an alkali ion (Cat+) and the same
amide bond breaks, bn ions of the type [bn-H]Cat+ are formed, in contrast to the
protonated structure reported here [203]. These two situations represent two distinct
types of reactions. The protonated molecule does not resemble a proton loosely bound to
the amide nitrogen — the "proton" forms a covalent bond to the nitrogen, and the charge
resides on the N, not the 1H. What follows is charge-initiated, inductive processes. In
contrast, the charge remains on the alkali ion in the cationized peptide, and the
subsequent chemistry is either remote-site or charge-enhanced. Even though an alkali ion
retains its charge when it complexes with a portion of a peptide, changes in skeletal bond
lengths accompany cationization [61]; thus, the charge can "enhance" the cleavage of
certain bonds. The fact that the form of the bn-type ions that result varies for protonated
and cationized peptides suggests that the chemistry observed here for protonated

molecules is charge-initiated, not remote-site.

Scheme I .

Linear [M+H]+ peptide

Schemell
RI!
C
p,
HzN %$
0
Scheme III
H H R
(le ‘NVc'i (|:H~
\ / ‘0’ \ /
H21( E (EH
0 R"
Linear [M+H]+ peptide
Scheme IV
$1 (II) If".
CH NH N
K \ \ /C\ ;H
H2 (II CH /N\$
0 1'," H H
Linear [M+H]+ peptide
Scheme V
R1 O m

Linear [M+H]+ peptide

1

166

167

211122;;

The formation of an ions requires that the Ca-C(O) bond be cleaved. However,
the structural form of the protonated peptide and the mechanism that lead to this directly
is not apparent. Others, using CI [204], FAB [65] and LSIMS [205], have suggested that
an ions are formed by loss of CO from 12,, ions. The a,, and a1 ion structures and
mechanisms are shown in Schemes I and II, respectively, in which the charge is shifted to
the nitrogen from the a-carbon after loss of CO from a bu ion, consistent with the fact
that the nitrogen-containing fragment has a lower ionization energy than CO. Also, the
formation of [an-H]Cat+ ions has been documented [188, 206] following alkali metal
cationization of peptides. Again, the metal ion remains on the fragment ion and a
hydrogen shifts either from an amide nitrogen or from a B—carbon in a side chain through
a 1, 2-elimination mechanism. In Scheme [[1, the 1, 2-elimination mechanism for an ions
is shown in which the protonating hydrogen atom from the amide nitrogen is shifted onto
the neutral fragment. Deuterium labeling studies are ineffective for studying the origin of
an ions because, if the 1, 2-elirnination mechanism is operative, the hydrogen would
originate from nitrogens which are difficult to selectively label. If an ions evolve from [2,,

ions, then no deuterium shift will be detected.

M

For peptides without basic residues, the RIs of xn and 2,, ions are usually zero in
both positive-ion LSIMS and CAD mass spectra. In Scheme IV, an inductive cleavage of
the N-Ca bond will result in the formation of a 2,, ion. In Scheme V, x" ions may be
formed by a 1, 2-elimination mechanism in which a proton on an amide nitrogen is

shifted to tat-carbon of the neutral, and the Ca-C(O) bond is cleaved.

168

m

Until recently, the decomposition mechanism for the formation of ya ions has
been largely disputed in the literature because the source of the shifting hydrogen was not
established. According to m/z values, a yn ion retains the protonating hydrogen atom and
gains an additional hydrogen. Using SIMS, Hunt et al. [205] proposed that y" ions are
formed by the 1, 2-elimination mechanism shown in Scheme VI in which the shifted
hydrogen originates from the a—carbon. In CI, Kiryushkin et al. [207] proposed a l, 3-
elimination mechanism in which the shifted hydrogen originates from the amide nitrogen
and a three-membered cyclic neutral is formed. In Scheme VII, Kiryushkin's mechanism
was expanded to shift a hydrogen atom from either the N-terminal amine or an N-
terrninal amide nitrogen through secondary interactions, and, depending on the degree of
secondary interaction, a neutral with a three-, six-, nine-, or l2-membered ring is formed.
The favorable secondary interaction is rationalized by a dipole-dipole interaction between
the nitrogen and carbonyl group, and a hydrogen bond between the protonated nitrogen
and interacting nitrogen. Recently, using selective deuterium labeling, Kenny et al. [208]
disclosed that the formation of y" ions involves a hydrogen originating from nitrogen and

not the a-carbon which indicates the 1, 2 elimination mechanism is not operative.

Mala»;

In Scheme VIII, the mechanism for a cyclo-bn ion is shown which involves the
same secondary interaction as yn. The positive charge on the backbone amide nitrogen
migrates to the interacting nitrogen in which a cyclic, protonated amide ion is formed

without hydrogen transfer.

was
The formation of c,, ions in Scheme IX and cyclo-zn ions in Scheme X involves C-

terrninal secondary interactions. In comparison to the N-terminal secondary interaction,

169

Scheme VI

Linear [M+l-l]+ peptide

II
—CH
Secondary interaction

H

eff,
Secondary interaction

|
R
Linear [M+H]+ peptide

Scheme VIII
/
WHN

 

 

Secondary interaction
Secondary interaction

L-

 

Linear [M+H]+ peptide
Linear [M+H]+ peptide

Scheme IX
Scheme X

170

the C-terminal interaction involves a weaker dipole-dipole interaction between the
interacting amide nitrogen or acid oxygen and the a-carbon, and a similar hydrogen bond
between the protonated nitrogen and the interacting nitrogen. Thus, all sequence ions can
be explained mechanistically by protonating the amide nitrogen and using a combination

of charge-site initiated chemistry and secondary interactions.

Evaluation of the LSIMS Relative Intensity Data.

Schemes I through X show simple mechanisms for forming the peptide sequence
fragment ions from a specific protonation site. If secondary interactions were not
required, these mechanisms suggest that, following protonation of an amide N, a certain
distribution of fragment ions would be expected. One possible scenario would yield 30%
bn ions, 40% y" ions, etc., from each amide N-protonation site, with the side chains
having little effect. The RI data for pentaglycine would certainly allow such a model to
be evaluated. The importance of this set of ions in the spectrum, resulting from
protonation of a given amide N, would depend on the extent to which that N is
protonated. Bursey et al. [209] studied oligoglycine fragmentation patterns in low-
energy, single-collision, collision-induced dissociation (CID) mass spectra at various lab
energies and proposed that secondary interactions ultimately influence the types and
abundances of fragment ions formed.

What if mechanisms involving cyclic intermediates/products are dominant? If
these are important, the ion distribution would not have such a simple, primary
dependence on the distribution of initial protonation sites. Protonation would be required
for subsequent chemistry, however, secondary interactions would be critical in allowing
chemistry to occur following protonation. Such reasoning suggested that we organize the
data by focusing on which bond was cleaved to form a given fragment ion, and present
the RI information as shown in Figures 49a and 50a for pentaglycine and methionine

enkephalin, respectively.

Pentaglycme
H-GGGGG-OH

 

 

'5 £9
: u -
a? = a) LSIMS of I C Terrmnal
g E H-GGGGG-OH I N Terminal
3:» a
w s
‘5 a
‘e\° U
1 2 3 4 5 6 7 8 9 10 ll 12 13
Skeletal Bond Position
'5 £3
1:
,3 g b) MI of H-GGGGG-OH
g a I C Terminal
:3 3’» .
m g I N Terrmnal
“6 s
39 U
l 2 3 4 5 6 7 8 9 10 11 12 13
40 Skeletal Bond Position
'5 a .
g g C) Single I C Terminal
3 IE Collision “ N Terminal
2 fl- CAD of
g 3’» H-GGGGG-OH
a: a
>
“5 §
5“ U

 

1 2 3 4 5 6 7 8 9 10 1 1 l 2 1 3
Skeletal Bond Position

Figure 49. Bar graphs showing the percent of skeletal bond cleavage products versus
skeletal bond positions from a) LSIMS, b) MI, c) "true", single collision CAD mass
spectra of the peptide pentaglycine. Structure and bond positions are shown at top.

 

   
 

 

 

 

 

  

 

 

 

 

 

 

 

OH

s: CH3
‘7': l
it I
a; .. in
II: an, CH2 0 0 CH2
38 CH 2% CH2 NH (Ii ('21-! OH
0- +
5?: HZN/ \(f/ \C£2\N§ \(lf \Cfi \Nfi \(E/ H
g a O O ('21-!2 O
O
2 ©

18
'2 g 16 1 . _
g = 14 a) LSIMS ofH-YGGFM-OH . l CTerrmnal _
— 12 _
g3 10 I NTerminal _
a o 3
353;" 6
“5 a 4
s6 2

0

 

1 2 3 4 5 6 7 8 9 10 11 12 13
Skeletal Bond Position

NM
CA

1)) MI ofH-YGGFM-OH

I C Terminal

...
O\

I N Terminal

% of Skeletal Bond
Cleavage Products
ES

 

8
4
o
12345678910111213
20 Skeletal Bond Position
fig 18
g :1 12 C) Single Collision CAD
33 12 ofH-YGGFM-OH acremnai
2 10
i go 8 EDNrerminal
w 6
h- >
cg 421
9
o
P o

 

1 2 3 4 5 6 7 8 9 10 11 12 13
Skeletal Bond Position

Figure 50. Bar graphs showing the percent of skeletal bond cleavage products versus
skeletal bond positions from a) LSIMS, b) MI, c) "true", single collision CAD mass
spectra of the peptide methionine enkephalin. Structure and bond positions shown at top.

173

For the model peptides, pentaglycine and methionine enkephalin, all the R18 of
sequence, irmnonium, internal, side chain ion, side chain loss, neutral loss, [a3+H]+‘, and
[b4+0H+H]+ fragment ions were normalized to 100% so each sequence ion skeletal bond
cleavage can be evaluated in light of all unimolecular decomposition processes.
Sequence ion formation involves the cleavage of one of the three skeletal bonds, Ca-
C(O), C(O)-N, or N-Ca. In Figures 49 and 50, the percentages of the N- and C-terminal
skeletal bond cleavage products vs. skeletal bond position (see peptide structures for
skeletal bond positions) were plotted such that the intensity of a bar represents how often
a skeletal bond was cleaved. For example, in Figure 49a, for LSIMS of pentaglycine, the
cleavage of skeletal bond 1, a Ca-C(O) bond, occurs 9% of the time out of the total
fragmentation processes, and results exclusively in the formation of an N-terrninal a1
fragment ion. In the same Figure, the amide bond at position 5 is cleaved 23% of the
time which results in the formation of both an N-terminal b2 ion 14% of the time and a C-

terminal y 3 sequence ion 9% of the time.

W

In Figure 49a, the skeletal bond cleavage products distribution for LSIMS of
pentaglycine is shown. The N-protonation of amide bonds 2, 5, 8, and 11 leads to an
overall RI fragmentation distribution of 13%:40%:30%:8%, respectively, if the skeletal
bond cleavage products, representing the an, b", C", and yn sequence ions, are summed. It
is immediately clear that peptide fragmentation is not as simple as charge-initiated
chemistry from N-protonation of the amide group since each protonated amide group
would be expected to yield sequence ions in the same ratios with similar RIs.

The cleavage of skeletal bonds 2, 5, 8, and 11 results in the formation of C-
terminal yn and N-terminal bn 31119.11 sequence ions from each bond. The RIs of peaks
representing y 3 and y2 sequence ions, formed from cleavage of skeletal bonds 5 and 8,

respectively, are much greater than the RIs of peaks representing the y4 and y 1 sequence

174

ions from cleavage of skeletal bonds 2 and 11, respectively. If the 1, 2-elimination
mechanism in Scheme VI was operative, all y” peaks would be expected to have similar
RIs because the a-carbon source of a hydrogen atom is local to the N -protonation site for
each amide. The yn mechanism in Scheme VII better accounts for the R15 of y" peaks.
Either the N-terminal amine nitrogen, or an amide nitrogen, in protonated or native form,
can interact through favorable secondary interactions with either of the two central amide
nitrogens, at skeletal bond positions 5 and 8, and induce yn ion formation. The formation
of y4 ions is not as favorable as y 3 and yz ions because a strained three-membered cyclic
neutral product is formed in the former process. The formation of an ions and b" ions in
Schemes I and II are the result of simple cleavages. The RI distribution for these ions
should be similar at each amide bond because the chemistry is local to the protonation
site, but this is not the case. Cleavage of amide skeletal bond 5 yields more b2 ions than
cleavage of skeletal bonds 8 and 11 (b3 and b4 ions, respectively). The a2 ion, from
cleavage of skeletal bond 4, is the only an ion formed as in Scheme I since [a3+H]+' and
[a4+H]+' ion formation account solely for cleavage of skeletal bonds 7 and 10. Using
Scheme 11, the a1 ion formed by cleavage of skeletal bond 1 is really a secondary product
from the absent b1 ion formed by cleavage of skeletal bond 2. Perhaps the combined RIs
of peaks representing an and b" ions show the effects of competitive fragmentation from
yn ion formation, involving secondary interactions, at a particular protonated amide N
site. Also, in some cases, the thermodynamic stabilities of the products may determine
which process is dominant.

Cleavage of skeletal bonds 6, 9, and 12 results in Cu ion formation, which involves
a less favorable C-terminal secondary interaction with a protonated amide nitrogen. The
C2 and C3 ions are more favored than C4 because, again, more stable six-membered cyclic
neutral products can be formed as opposed to a three-membered ring for C4. The cyclo-bz
ion in Scheme VIII may in part account for the large R1 of the b2 peak at skeletal bond

position 5 since a stable six-membered cyclic b2 ion is formed. No xn or 2,, ions are

175

observed for pentaglycine. The 1, 2-elimination mechanism for x" ion formation does not

occur to any extent which further confirms the use of alternative mechanisms.

HIMSELF M M l° . E! H.
In the pentapeptide methionine enkephalin, the side chains influence the charge-

site induced chemistry and secondary interactions in contrast to pentaglycine, probably
through stabilization effects. In Figure 50a, the normalized LSIMS skeletal bond
cleavage product distribution for methionine enkephalin is shown. For methionine
enkephalin, most fragmentation originates from cleavage of each N-protonated amide
bond similar to that for pentaglycine described above. The RI distribution for sequence
ion fragmentation, summed at each amide bond 2, 5, 8, and 11, is 18%:3%:3%:4%,
respectively. The RI distributions resulting from the cleavage of amide bonds 5 and 8 in
methionine enkephalin are nearly identical to the same bond cleavages in pentaglycine.
However, the presence of the tyrosine and methionine side chains in methionine
enkephalin has dramatically enhanced the R15 of peaks representing 61] and y 1 fragment
ions from cleavage amide bonds 2 and 11, respectively. The cleavage of skeletal bond 7

represents [a3-l-H]+‘ solely whereas cleavage of skeletal bond 10 yields the a4 ion.

Estimation of Reaction Enthalpies for the Mechanisms.

For pentaglycine in its linear form, the amide nitrogens are the most probable
protonation sites. Initially, each amide nitrogen may be protonated in an even
distribution, but secondary interactions may redistribute the protons (protons may "hop"
throughout the molecule). The fragmentation mechanisms presented assume that the
amide nitrogen protonation initiates charge-induced chemistry, and hydrogens shift
according to previously reported deuterium labeling studies. If simple bond cleavages

such as homolytic, l, 2-elimination or inductive cleavages were prominent in

pentaglycine without any side chain influences, the RI distribution of skeletal bond

 

176

cleavage products originating from each protonated site would be expected to be very
similar. Such a uniform RI distribution is not observed! Instead, the non-uniform RI
distributions suggest, at each protonated amide N, secondary interactions are in
competition with simple cleavages.

As a next step, the reaction enthalpies for each mechanism in Schemes I-X were
estimated to gain more insight into why one particular fragmentation process dominates
over the other possible pathways at a protonation site and/or why fragmentation from one,
particular protonated site dominates over that of other sites in these two model
pentapeptides, with and without side chains to influence fragmentation. Since the heats
of formation are not known for protonated methionine enkephalin or pentaglycine and
their ionic and neutral fragments, the exact reaction enthalpies cannot be determined.
However, for small analogous neutral molecules, radicals, and ions, the heats of
formation, PAs, and ionization energies (IE) are available [40, 191, 210]. The reaction
enthalpies of peptide fragmentation mechanisms were estimated by matching the
chemical structure of smaller species of known thermochemistry with the chemical
structure of the protonated site and limited surrounding area on the peptide. The outer
regions extending away from the protonation/reaction site were ignored. The same
process was done for the bond cleavage and formation sites in the ionic and neutral
fragments. See Appendix Two for data and methods used to calculate the reaction
enthalpies for sequence ion fragmentation, and Figures 51 and 52 for a graphical

representation of these estimated energies.

i 1e
Homolytic bond cleavage requires the most energy because it results in the.
formation of two radicals. The reaction enthalpies for homolytic bond cleavages of the
peptide skeletal bonds, Ca-C(O), C(O)-N, and N-Ca were estimated to be 87, 98, and 103

kcal/mol, respectively. Peaks representing ions from homolytic cleavages of C(O)-N and

177

.38on E mowgao—o 252 .8283 ovum—on .8 82.35:“. 5:82 Bsfizmm Am Eswfi

:5 .22 .5 = 2.2.... 2%

on .22 5-222 m 62.5.. 22...
ES .22 .2222 m 62.5.. 2.6
:5 .22 .5 2 62:2 2%

ES .22 .2222 a 62:2 .25
ca .22 :2 = 2.22 2.5
on .22 ..2 m 2...... 22m

5 .22 :2 Aomwdmmeu E on .22 5.222 m 23.. 2.5

      

.22 to . ‘Ill a a
a . :ogmmeo 2-2
a .22 :2 onflmzuo
e
2.22 :2on £5
6 a. a .22 ..e aged

 

E .22 :2 22.69%

E .22 .5 £6 32

 

 

ml

 

6226.2.J

coemcoaoi

 

 

.8582 e

a: .22 E soummz

bN $
2 9 .22 .5 9:2qu5
o
c> .22 .5 82:2
an e

a: .22 :2 unmwaovu
3%..“

5.

ex .22 2+ :2 8V .8
mm 6

N

2 mango
N

8. 6&9:

finial—mam

178

.22 +E+Sz ...—Egon 3:88 ..8 £2.58.— Ev 3335:... 5:82 335:3 :05 was $9553 52.0.3888 EBB—oEED .Nm v.53”—

m 0 =2888£ .965on

 

 

 

 

 

 

 

 

 

+ |.
=_U\ max
5% O
.m 530825 gap—ovum
m=V 5 ”DID I 35.50.70 ..
2m 2m: ...2 \\o
WU\$ + \(z “Mylo WW
_ W...“
M / Immly . O
... O OImUdIZIIUKI A: «m
\\ _+ :9 __
-o .... o ..
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII .mlilllllul I:
...m.” \\0 ...”—
om + m z Emu/m I)
S +
a: \ w
\\ W2 o
..m o ..z
OIIII. + :U/
2:. 4.. ..~__ m
\(U U N22 20 U N:
\ / +
Wu 3:2 :02 /U\ mix Wu».

 

179

N-Ca bonds are not observed in LSIMS, MI, or CAD mass spectra. However, homolytic
cleavage of Ca-C(O) is observed which results in the low abundance [an+H]+’ ion series
in LSIMS and single collision CAD spectra, but not in MI spectra—probably because Ca-
C(O) is the weakest skeletal bond to cleave and its energy requirement is within the high
energy tail of [M+H]+ internal energies. The mechanism for [an+H]+‘ ion formation is
shown in Scheme XI. The [an+H]+' ion either exists as a distonic species or more likely
it is probably stabilized from a hydrogen atom transfer from the protonated nitrogen to

the a-carbon because a positive charge on nitrogen is more stable.

We

Simple inductive bond cleavages require less energy than homolytic cleavages
because more stable products are formed, but they require more energy than 1, 2-
eliminations because no new bonds are formed. In Scheme 1, the reaction enthalpy of bn
ions was estimated to be 54 kcal/mol when the amide bond is inductively cleaved. The
high abundance of b" ions in the LSIMS spectra of peptides is good evidence that this
mechanism is operative. If additional energy is available, the Ca-C(O) bond in b" ions
can be cleaved to form an an ion and CO as shown in Scheme 1. If the (IL-carbon is part of
a glycine residue having no side chain, then the formation of an an ion requires an
estimated, exothermic reaction enthalpy of -5 kcal/mol. However, if a stabilizing side
chain is present on the residue, an reaction enthalpies fall to -15 and -20 kcal/mol. A
small activation barrier must exist in the transition from an to bn ions because the
abundances of bn ions are comparable to an ions even though the reaction enthalpies for
forming an ions are exothermic. The a 1 ion shown in Scheme H is a special case for an
ions. From ionization energies, the a1 ion structure is approximately 10 kcal/mol more
stable than the an ion structure because the amide carbonyl group delocalizes the amide
nitrogen lone pair in the an ion structure inhibiting full stabilization of the charge

whereas, in the a] structure, the lone pair on the amine N can fully stabilize the charge.

Scheme X]

R’ o

I HHxaHn /C'H~__>
\(S

HzNA flc\NH

Linear [M+H]+ peptide

Scheme XII

Linear [M+H]+ peptide

180

’I‘ e 'I‘
on /CH~
H21~{ (“2 H2 O=C=NH
O R"
[an+H]+°
i" 0
ll RI"
CH NH
——> / \ / \ £9 + I
V‘HN q CH NH=CH
@ 0 5*"
’1‘. 0
ll
NH
—> /CI\1 / \ Q + NH3
V‘HN % (IEH
0 RN
- R @911 - R'
HN
V‘CH—C— l \
CH n
5§\>"CHR _,mHN/\ /OH + l/CHR
C
II
1H)" C/\ EBOH o”

 

Secondary interaction

 

 

Obn+on+m+)

 

181

Thus, for an a1 ion, the estimated reaction enthalpy is now -30 kcal/mol exothermic if a
stabilizing side chain is present on the a-carbon and -15 kcal/mol exothermic if the
residue is glycine. Formation of the 2,, ion in Scheme IV requires inductive cleavage of
the N-Co, skeletal bond. For 2,, ion formation, reaction enthalpies were estimated to be 62
kcal/mol if a side chain is present on the a-carbon to provide charge stabilization, and 79
kcal/mol if the residue is glycine with no side chain. The high reaction enthalpies
calculated for 2,, ions formation are supported the absence 2,, ions in the LSIMS spectra of
these model peptides. The b,, ions are probably more abundant than 2,, ions because the
b,, ion structure is more stable than 2,, and the C(O)-N bond is weaker than the N-Ca

bond.

The l, 2-elimination mechanism for peptides requires less energy than either
homolytic or inductive bond cleavages even though two bonds are broken, because
energy is released in the formationof a new single bond involving the shifted hydrogen
and the upgrade of a skeletal single bond to a double bond. In Scheme III, assuming the
amide nitrogen is protonated, the estimated reaction enthalpy required to form an a,, ion
by a l, 2-elimination is 42 kcal/mol. However, the reaction enthalpy is reduced to 27
kcal/mol for the formation of an a 1 ion if the terminal amine is protonated, because a 1 ion
structures are more energetically stable as explained above. In Scheme V, if the amide
nitrogen is protonated, the reaction enthalpy for formation of an x,, ion through a l, 2-
elimination mechanism was estimated to be 35 kcal/mol. For the y,, ion formed by a 1, 2-
elimination mechanism in Scheme VI, the reaction enthalpy was estimated to be 38
kcal/mol. While these 1, 2-elimination mechanisms are more favorable than homolytic.
and inductive cleavages, deuterium labeling studies have indicated y,, ions are not formed
by this mechanism. The x,, ions are rarely observed in small peptides without basic

residues and this may be additional evidence that 1, 2-e1imination mechanisms are not

‘I
“/_U

182

operative. No plausible l, 2-elimination mechanisms exist for b,,, c,, and 2,, ions.

MW

Skeletal bond cleavages through secondary interactions involve breaking two
different bonds and forming two new, very similar, bonds which results in very low
reaction enthalpies without large energy deficits. The formation of y,, ions in Scheme VII
first requires the formation of an intermediate through a favorable secondary interaction
so a hydrogen can shift onto the ionic fragment for the formation of a stable cyclic neutral
fragment. Specifically, the y,, ions have very favorable overall reaction enthalpies
because two bonds are cleaved—the protonated amide bond on the peptide backbone and
the interacting N-H bond. Subsequently, energy is returned when new two bonds are
formed-a new N—H bond to the amine on the y,, ion fragment and a new amide bond that
forms the cyclic neutral fragment. Depending on the origin of the shifting hydrogen, the
overall reaction enthalpy for y,, ions was estimated to be —5 kcal/mol exothermic, or
thermoneutral if the hydrogen originates from an amide or amine nitrogen, respectively.
The 5 kcal/mol difference indicates the N—H bond is stronger in an amine nitrogen. For
c,, ions, in Scheme IX, which are formed through C-terminal secondary interactions, the
estimated reaction enthalpies are 5 kcal/mol if a hydrogen shifts from an amide nitrogen
or 10 kcal/mol if the carboxylic acid hydrogen transfers. Again, the 5 kcal/mol difference
indicates that a hydrogen atom is more difficult to remove from C-terminal carboxylic
acid moiety than from the amide nitrogen. Assuming amide hydrogens shift, the overall
reaction enthalpies for c,, ions are 10 kcal/mol greater than y,, ions because the N-Ca
bond is stronger than the amide bond, less energy is returned upon formation of a new N -
CO, bond to form the cyclic neutral product, and the PA of the c,, ion structure, similar to
acetamide, is lower than the PA for the terminal amine structure in y,, ions. In addition,
the y,, ions should be more abundant than c,, ions since the secondary interaction for c,,

ions is less favorable than y,, because the partial positive charge on the (It-carbon is less

183

than that on the carbonyl carbon, and for c,, ions, the presence of side chains and/or on—
hydrogens on the a-carbon imposes more steric hindrance for the interacting C-terminal
amide nitrogens or carboxylic acid oxygen than encountered between the interacting N-
terminal nitrogens and the open carbonyl groups for y,, ions.

For cyclo-bn ion formation in Scheme VIII, a favorable N-terminal secondary
interaction is required, but instead of a hydrogen transfer, the protonated skeletal amide
bond is cleaved and a new amide bond is formed during cyclization to form the ionic
cyclic fragment . The overall reaction enthalpy for cyclo-b,, ion formation is -5 kcal/mol
if the secondary interaction involves an amide nitrogen or O kcal/mol if an amine nitrogen
is involved. The energy difference indicates that the basicity of the amide group is
higher. The formation of the cyclo-2,, ion in Scheme X requires a C-terminal secondary
interaction involving an amide nitrogen in which a subsequent cleavage of a skeletal N-
CO, bond and formation of a new N-Ca bond forms the ionic cyclic fragment ion. This
process requires an overall reaction enthalpy of -5 kcal/mol, which is 10 kcal/mol lower
than that for a c,, ion involving the same amide N interaction, because the PA of the
amide nitrogen in the cyclic fragment ion is greater than that of a terminal amide.
However, if the acid oxygen is involved in the secondary interaction for cyclo-2,, ions, the
reaction enthalpy increases to 20 kcal/mol because the PA of the cyclic C(O)O-Ca ester
site is much less than the PA of the cyclic C(O)N-Ca site formed when an amide nitrogen
interacts. In general, cyclo-bn ions should be more favorable than cyclo—2,, ions because
the secondary interactions in these processes parallel those for y,, and c,, ions above,

respectively.

1 0d n mi 71 i r ti n
The estimated thermodynamic data presented above describe the energetic
requirements of the fragmentation mechanisms in Schemes I-XI, which allowed these

mechanisms to be placed in perspective as illustrated in Figures 51 and 52. All sequence

184

ions have reaction enthalpies below 4 eV (70 kcal/mol), the upper limit of internal
energies for protonated molecule in the LSIMS experiment, which indicates
fragmentation through these mechanisms is thermodynamically accessible. These
reaction enthalpies provide additional insight into the RI information for pentaglycine and
methionine enkephalin in the LSIMS experiment in Figures 49a and 50a. The resulting
y,, and b,, ion formation from amide skeletal bond cleavages is indeed thermodynamically
and kinetically competitive because y,, ions have reaction enthalpies that are slightly
exothermic overall and result in the formation of a stable cyclic neutral intermediate,
while b,, ions evolve from simple bond cleavages, that require approximately 2 eV
without the need for a rate-limiting secondary interaction required for y,, ions. For
pentaglycine and methionine enkephalin in Figures 49a and 50a, no peaks representing a
b; ion are observed at amide skeletal bond position 2, but intense peaks representing the
01 ions are observed at skeletal bond position 1 which supports the idea that b] loses CO
to form a 1. The estimated exothermicity of this b1—>a1 fragmentation process
corresponds well to the exclusive appearance of intense (11 ions in both peptides and, in
fact, the a 1 peak for methionine enkephalin is much more intense than that in the mass
spectrum of pentaglycine, which supports even lower reaction enthalpies through side
chain stabilization. However, 0,, ions are generally less intense or missing in comparison
to d] ions which coincides well with the less exothermic reaction enthalpies for a,,
formation. Further evidence that local side chains lower the reaction enthalpy is observed
for the intense peak at skeletal bond position 10, which represents an a4 ion (in
comparison to the peak at skeletal bond position 11 representing a b4 ion), in Figure 50a
for methionine enkephalin. In LSIMS of pentaglycine, at the same skeletal bond position
10, there is no peak representing the (14 ion, but instead an [a4+H]+'. In Figure 49a for
LSIMS of pentaglycine, the secondary interaction mechanisms required for y,, ion
formation are more thermodynamically favorable, and they account for the two intense

peaks at skeletal bond positions 5 and 8, which represent the )3 and yz ions, and the two

185

less intense y,, peaks at skeletal bond positions 2 and 11 better than those that result from
1, 2-elimination mechanisms. The intense yz peaks in both model peptides at amide
skeletal bond position 8 suggest amide hydrogens rather than amine hydrogens are more
favored to shift, which corresponds well with the calculated y,, reaction enthalpies.
However, the enhanced RI of y2 may also suggest that hydrogens shift from either amide
or amine N sites. The presence of the methionine side chain appears to enhance the y 1
peak at amide skeletal bond position 11 in methionine enkephalin in comparison to the
small y 1 peak for pentaglycine. For the RI distribution of peaks representing c,, ions,
formed from cleavage of N-Co, skeletal bonds 6, 9 and 12, are low for both peptide
models, which indicates b,, and y,, ion formation mechanisms involving amide bond
cleavage(s) and formation(s) are more kinetically favored over cleavage of the N-Ca
bond for c,, ions. The calculated reaction enthalpy for c,, ion formation is slightly
endothermic overall in comparison to that for y,, ions, which further indicates that the C-
terminal secondary interaction is not as favorable as the N-terminal interaction for y,, ion
formation.

Since RIs of peaks in the mass spectrum contain both thermodynamic and kinetic
information for the formation of fragment ions, the estimated reaction enthalpies for the
a,,, b,,, c,,, and y,, ion fragmentation mechanisms allow the purely thermodynamic
attributes of a particular mechanism to be identified and more insight into the complex
and competitive unimolecular decomposition kinetics. Experimentally, these estimated
reaction enthalpies correspond well with the stabilities of the derived fragment ion
structures, and quantitatively rank how favorable the simple bond cleavages and the
rearrangement reactions are with respect to all the fragmentation options of the
protonated peptide as shown in Figures 51 and 52. The trends above for these
pentapeptide models give strong evidence for secondary interactions being in competition
with simple bond cleavages, and the extent to which these postulated mechanisms

produce b,, and y,, ions, respectively, were probed by comparing the pentaglycine RI

186

distributions to those of oligoglycines with three, four and six residues.

Comparison of LSIMS Oligoglycine Peptide Relative Intensity Data.

The extent of secondary interactions was evaluated in LSIMS spectra of triglycine
(MW 189), tetraglycine (MW 246), pentaglycine (MW 303), and hexaglycine (MW 360)
by considering the R18 of b,, and y,, ions peaks in Figures 53a-d, respectively. In
tetraglycine, pentaglycine and hexaglycine, most b,, and y,, ion fragmentation results from
cleavage of the one (i.e., amide skeletal bond 5), two (i.e., amide skeletal bonds 5 and 8
in Figure 49) and three (i.e., amide skeletal bonds 5, 8, and 11) central amide bond(s),
respectively. Triglycine is an exception because it does not have a central amide bond
surrounded by two outer amide bonds. For each oligoglycine, two intense peaks
representing b2 and y2 ions, respectively, are observed independent of the number of
glycine residues. Cleavage of the one central amide bond in tetraglycine in Figure 53b
results in the competitive formation of b2 ions through a simple amide bond cleavage and
yz ions through a favorable N—terminal amine secondary interaction with elimination of a
stable six-membered cyclic neutral product, respectively. For all oligoglycines in Figure
53, the R1 of the b2 peak may be enhanced if the N-terminal amine undergoes a secondary
interaction to either simply protonate or polarize the amide skeletal bond 5 to form
additional b2 ions or cyclo-bz ions, respectively. If only tetraglycine and pentaglycine are
considered in Figures 53b and c, yz ion formation appears to be straightforward because
y2 ions are most likely formed by secondary interactions involving the N-terminal amine
in tetraglycine and the skeletal bond 2 amide nitrogen in pentaglycine, which has a lower
reaction enthalpy. Moreover in hexaglycine in Figure 53d, if formation of six-membered
cyclic neutral products is prominent, a secondary interaction involving amide nitrogen at
skeletal bond 5 would induce cleavage of amide skeletal bond 11 for y; formation and
likely result in a decrease of R15 for amide bond 5 cleavage products, ()2 and y4 .

However, the RI distribution of the peaks representing the b2 and y4 ions for hexaglycine

187

 

 

 

 

 

a) Triglycine l I LSIMS
e Ilhdl

J

 

 

 

j

 

 

 

 

 

Normalized Relative

 

 

db
III-
CI:-

1
I

it.

 

 

uence Ion
7O Seq
0
23': b) Tetraglycine

T.»
a:
'8
.23.
E
e
Z

y3 b2 y2 b3 )7

Sequence Ion

 

 

    
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
  

.L

50 -
g -I
3 40110) Pentaglycine: I LSIMS *-
G
a: g 30 IMI .
'8 s -
:2- : 20
a I- q
2 I'll ,
'5 10q ‘ a
z .l'I-IIFIHIKI

y4 b2 y3 b3 yz b4

y]
Sequence Ion

 

 

 

03%
VIC

1d) Hexaglycine i

w
0

 

 

 

 

 

 

 

 

-—-—-N
omom

 

 

Normalized Relative
Intensity

 

 

OUI

y5 b2 M '93 ya b4 )2 b5 y}
Sequence Ion

Figure 53. Bar graphs comparing normalized RIs of b,, and y,, peaks in the LSIMS and MI
spectra of a) triglycine, b) tetraglycine, c) pentaglycine, and d) hexaglycine.

188

are still as intense as in pentaglycine, which means the b2 and y2 ion formation is
independent of the degree of the secondary interaction that induces their formation. In
Figure 53a for triglycine, the peak representing the )7 ion is also intense similar to those
for the other oligoglycines, which means an N-terminal amine secondary interaction
induces the formation of this yz ion and a three-membered cyclic neutral product. This
process in triglycine is in competition with b; formation if the N-terminal amine induces
a b2 ion through intramolecular protonation and a simple cleavage, or a cyclo-bz through
a polarizing secondary interaction. In Figure 53 for each oligoglycine, y 1 formation
should be more than favorable through secondary interactions, but the peaks representing
the y1 ions are of low intensity, which suggests the formation of protonated glycine is not
a favorable ionic product perhaps owing to its small size and limited stability in the high
internal energy environment of the LSIMS experiment. Thus, the oligoglycine series
shows diglycines, both H-GG$ and @NH3CH2C(O)-G-0H, representing b2 and y2 ions,
respectively, are very stable ionic products because their R15 are intense regardless of
number of residues in the oligoglycine and the degree of the secondary interaction(s).
Under the energetic conditions of LSIMS, the formation of y,, ions and a stable six-
membered cyclic neutral products is not a requirement for all secondary interactions
because evidence shows three-membered rings are also formed. Because a y,, ion is
essentially a shorter, complete N-terminally protonated peptide, the possibility of a
subsequent secondary interactions in y,, ions may induce a secondary fragmentation
processes leading to the smaller ionic products such as 172 and y2 if enough internal

energy remains.

Secondary Fragmentation Processes in LSIMS.
Because the internal energies in LSIMS are estimated to extend out to 4 eV, the
preferential formation of b2 and y2 ions from oligoglycines and the intense (1] peak from

methionine enkephalin shown in Figures 53a-d and 503 above, respectively, may be due

189

in part to secondary fragmentation processes. In El, mechanisms were elucidated for the
most important ions resulting from primary and secondary fragmentation processes, but
the formation of ions such as 0* (carbon) were not considered. For example, in Scheme
XII, a simple inductive cleavage of the ionized amide bond in an a,, ion can lead to a b,,
ion [211]. The reaction enthalpy for this process was estimated to be 51 kcal/mol, similar
to the energy required for the formation of normal b,, ions. In Figure 54a and b, the
fragmentation barrier profile, based on the calculated reaction enthalpies for b,, ions, a,,
ions and a,, to b,, transitions, is shown for methionine enkephalin and pentaglycine,
respectively, to demonstrate the amount of energy needed for secondary fragmentation
processes to occur. In the LSIMS mass spectrum of methionine enkephalin, the R15 of
peaks representing a,, and b,, ions follow this potential energy diagram well because the
RI b4 is less than that of a4 which corresponds to the a4 exothermic pathway enhanced by
the presence of phenylalanine side chain in Figure 54, and the R1 of a1 is very intense
which corresponds to the highly exothermic b] to a] transition enhanced by the tyrosine
side chain. Assuming internal energies of the [M+H]+ ions have an upper limit of 3 eV
(70 kcal/mol) and no reverse activation energies exist for these simple cleavages in
LSIMS, the transition from b4 ions to b3 ions will take 71 kcal/mol, so b4 is an unlikely
precursor to all other ions. However, if b3 is formed by a simple amide bond cleavage,
the transition of b3 ions to b2 ions requires 56 kcal/mol and the transition of b3 to a]
requires only 59 kcal/mol, so secondary fragmentation may account in part for the very
intense a 1 peak in the LSIMS analysis of methionine enkephalin. From the pentaglycine
fragmentation barrier profile in Figure 54b, secondary fragmentation can readily occur
because a transition from b4 to a; requires only 61 kcal/mol. However, without a very
thermodynamically favorable fragmentation process induced by side chains, the R15 of
peaks representing a,, and b,, ions are relatively uniform with the slight exception of b2
and a] which represent the smallest, most stable ionic fragments and the most intense

peaks in the LSIMS spectrum of pentaglycine . In Scheme XIII, the loss of ammonia

190

 

6O -- Ia) Methionine Enkephalin l
i ’ .

Energy (kcal/mol)

 

 

 

 

 

 

 

 

 

 

 

O I.
Peptide b4 04 b3 a3 b 2 a 2 b 1 a1
Sequence Ion

 

60 -r I b) Pentaglycine I

50‘

d

 

Energy (kcal/mol)
U)
o

 

 

 

 

 

 

 

 

 

 

 

Peptide b4 a4 b3 a3 b 2 a2 b 1 a 1
Sequence Ion

Figure 54. Fragmentation barn'er profiles based on estimated reaction enthalpies for a,, and
b,, ions from unimolecular decomposition of a) methionine enkephalin and b) pentaglycine
protonated molecules.

191

from a c,, ion will produce a b,, ion, which was estimated to require 42 kcal/mol. Perhaps
c,, ions are initially formed in high abundance through the favorable secondary
interactions and subsequently lose NH3 to form b,, ions, because c,, ions are generally less
abundant than b,, ions. Thus, for a LSIMS experiment, in some cases there should be
enough internal energy remaining in the primary fragment ions to induce subsequent
secondary fragmentation assuming the activation energies are negligible.

The high abundance of non-sequence, low-mass ions in the LSIMS methionine
enkephalin spectrum such as side chain ions [65, 187], immonium ions [212], and
internal ions [213] indicates secondary and higher processes are occurring to some extent.
By accounting for the formation of these non-sequence ions, 3 more accurate description
of [M-t-H]+ unimolecular decomposition can be obtained for peptides. Assuming the R15
of peaks representing the F, G and M immonium ions, the GF and GGF internal ions, the
GF-28 and GGF-28 internal immonium ions, and the side chain ions are normalized,
along with the sequence ions, and all ions are the result of secondary fragmentation
processes that also originate from a protonated amide nitrogen, the revised RI distribution
for methionine enkephalin fragmentation is 30%:13%:43%:10% from amide bonds 2, 5,
8, and 11, respectively. The primarily difference between this LSIMS fragmentation RI
distribution and the distribution from Figure 50a, which shows only the RI distribution of
sequence fragment ions, is amide bond 8 is cleaved 43% of the time when secondary
fragmentation processes are tallied in comparison to only 3% in Figure 50a. The major
secondary fragmentation process at amide bond 8 is the phenylalanine side chain
stabilizes the formation of the abundant F immonium ion, a nonsequence ion. In
methionine enkephalin, the tyrosine and phenylalanine side chains have much influence
on local fragmentation because the revised distribution indicates that the fragmentation
pathways in the vicinity of the two aromatic side chains are more thermodynamically
favorable because these side chains stabilize the intermediates and final products. Thus,

if a peptide contains side chains, low-energy pathways will exist and then secondary

 

frag

sic'

111'.

sin

192

fragmentation processes may occur more frequently.

Under the high-energy conditions of LSIMS, the fragmentation mechanisms for
side chain ions, immonium ions within the peptide chain and internal ions are not well
understood. For internal ions, MS/MS/MS studies on a hybrid mass spectrometer, using
high/low energy CAD and leucine enkephalin, have shown GGF can be formed from y4
ions and GF can be formed from )3 ions [214]. Because a y4 ion is essentially a smaller
but complete peptide that is protonated on the N-terminal amine, €BNH3CH2C(O)-GFL-
0H, a conceivable mechanism for forming the GGF internal ion from ya, would involve a
subsequent secondary interaction in which the protonated amine nitrogen protonates the
amide nitrogen between F and L to either induce a simple inductive amide bond cleavage
and form a b,, type fragment, H-GGFQ, or form a cyclic internal ion structure identical to
a cyclo-bn ion. Because amine and amide PAs are very similar, intramolecular transfer of
a proton between the amine and amides would be very favorable and would require
approximately zero kcal/mol. In the LSIMS, NH and CAD mass spectra of the peptide H-
GGFL-OI—I, the low intensity peak representing the a3 ion and the dominant peak
representing the b3 ion, H—GGFQ, suggests the H-GGFe fragment is cyclic, which would
have the same structure as the internal ion, GGF. Thus, the estimated reaction enthalpies
for internal ions would parallel those for the formation of b,, and cyclo—bn ions above and
internal immonium ions would parallel a,, ions assuming the overall reaction enthalpies
for forming y,, precursor ions are taken into account. The F immonium ions involving the
internal phenylalanine residue can be formed by a similar mechanism as internal ions if
the protonated N-terminal amide nitrogen on the yz ion would protonate the nitrogen at
amide bond 11 in methionine enkephalin to induce an inductive amide bond cleavage to
form a b 1-type ion and subsequent CO loss which would result in an a1-type ion, or the F
immonium ion [215]. The overall reaction enthalpy for this immonium ion would be
similar to the formation of y2, b1 and a1 ions as discussed above.

To investigate the secondary interactions in more detail, derivatized peptides were

193

synthesized using well known procedures contained in reference 216. An acetylated
peptide is formed by attaching an acetyl group to the N-terrninal amine and, as a result,
the N-terminal amine nitrogen is converted to an amide nitrogen. With an N-acetylated
peptide, the b1 ion now appears in the LSIMS spectra, unlike that of underivatized
peptides, along with (1], that still contains the acetylated group, because
thermodynamically, the b] to a] transition here produces an a1 ion with the less favorable
a,, protonated imide structure instead of the more favorable a1 protonated imine structure.
An a1 ion with the protonated imine structure still exists at m/z 136 for the enkephalin
peptides although it is half its normal size, Which indicates perhaps not all of a1 is from
b] in the underivatized peptide and/or a ketene is eliminated from the acetylated N-
terminus by a 1, 2—elimination mechanism. In addition, acetylation of a peptide produces
a uniform RI distribution of intense peaks representing b,, and y,, ions in LSIMS spectra.
For b,, ions, this increase can be rationalized if the thermodynamically less favorable
protonated imide a1 ion reduces secondary fragmentation processes from b,, ions. For y,,
ions in acetylated methionine enkephalin, the ya; and y 3 ions become as abundant as yz
and y 1 because a hydrogen can be shifted through secondary interactions from the more
thermodynamically favorable acetyl amide nitrogen. Thus, the estimated reaction
enthalpies indicate that secondary fragmentation processes are possible and the RI
distribution of both sequence and nonsequence ions indicates that secondary

fragmentation occurs to some extent, especially if the thermochemistry is very favorable.

Evaluation of the MI Relative Intensity Data.

Metastable [M+H]+ ions fragment after acceleration because their initial internal
energies are too low to induce the required high unimolecular decomposition rates for
fragmentation in the ion source. In contrast to high energy simple cleavages in LSIMS,
the prominent metastable fragmentation mechanisms for both pentaglycine and

methionine enkephalin should consist essentially of low-energy secondary interactions.

194

r r l ' e

In Figure 49b, for metastable decomposition of pentaglycine, the RI distribution
of the skeletal bond cleavage products shows nearly all fragmentation results from
cleavage of the amide skeletal bonds. No a], a,, or [a3+H]+' ions are observed which
verifies that metastable fragment ions only represent primary fragmentation from low
energy [M+H]+ ions and not secondary fragmentation processes. Summing the
fragmentation from cleavage of protonated amide bonds 2, 5, 8, and 11 in Figure 49b
results in the RI distribution of 5%:24%:15%:36%, respectively, which is essentially
opposite from the LSIMS distribution in Figure 49a. From Figure 49b, the abundant y 3
ion, formed from cleavage of skeletal bond 5, strongly suggests that an N-terminal amine
secondary interaction results in the formation of this ionic fragment and the stable six-
membered cyclic neutral product. While formation of any y,, ion is indeed a very
thermodynamically favorable fragmentation process for low-energy [M+H]+ ions, the
thermodynamics for an amide nitrogen secondary interaction and hydrogen shift are more
favorable than the amine nitrogen secondary interaction, and y2 ions should dominate
over )3 ions as in LSIMS. However, for metastable decomposition of protonated
pentaglycine, the peak representing the y 3 ions is greater than that for yz which suggests
the secondary interaction involving the amine requires less energy to fold inward to
complete an interaction than does an amide nitrogen. The origin of the dominant b4 ion
is unclear because it is unlikely that simple inductive bond cleavages are competitive in
very low energy [M+H]+ ions. Instead, the intense metastable peaks representing b4 ions
are probably the result of slow unimolecular decomposition. Protonated pentaglycine in
the gas-phase may exist in a folded state held together by hydrogen bonding and perhaps,
a secondary interaction between the N-terminal amine and amide skeletal bond 11 occurs,
since they would be close in space to form a symmetrical, low-energy flexible 12-

membered cyclo-b4 ion similar to the formation of a six-membered neutral as in Scheme

195

VIII. The low R1 of peaks representing c,, ions at skeletal bond positions 6, 9, and 12,
further suggest that the C-terminal secondary interaction required for c,, ions is not as
thermodynamically favorable as an N-terminal secondary interaction. Thus, the abundant
)3 and b4/cyclo-b4 fragment ions from the metastable decomposition of protonated
pentaglycine indicate low-energy, secondary interactions are the driving force in these

fragmentation mechanisms that produce the very stable ionic and neutral products.

MIMI] M 2!. l .

In Figure 53a—d, the RIs of peaks representing the metastable b,, and y,, ions from
protonated triglycine, tetraglycine and hexaglycine coincide with the metastable ions
from protonated pentaglycine above. In all four cases, the RI distributions differ sharply
from LSIMS. In Figure 53a for triglycine, formation of a b2/cyclo-b2 ion and glycine
neutral product is more favorable than formation of a protonated glycine molecule, a y 1
ion, and a cyclic diglycine neutral product from the same N-terminal amine nitrogen
secondary interaction. In Figure 53b for tetraglycine, the yz ion becomes dominant
perhaps because the larger, more stable protonated diglycine molecule is formed in
addition to the cyclic diglycine neutral product. The RI distribution for protonated
hexaglycine in Figure 53d shows two intense metastable ion peaks representing y4 and b5
ions that parallel the intensity of y 3 and b4 ions for pentaglycine in Figure 53c. For peaks
representing the cyclo-bn ions formed from protonated oligoglycine decomposition,
higher values of n have more intense RIs supporting formation of larger cyclic ions and
small neutral glycine products.

The RI distributions of fragment ions from metastable decomposition of
acetylated triglycine, tetraglycine, and pentaglycine [M-I-H]+ ions were compared to the.
RI distributions of the respective underivatized oligoglycine [M+H]+ ions in Figure 55a-c
to further investigate the importance of N-terminal amine secondary interactions. The

"bulky" acetylated N-terminal amide nitrogen may either inhibit or enhance the

196

a Triglycine I Underivatized

I Acetylated

Normalized Relative
Intensrty

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   
 

 

 

 

 

 

 

 

  

 

 

b1 y2 b2 y1
Sequence Ion
60. -
g 50%-lb) Tetfaglycine . I Underivatized _
'5 ‘ ’
.3 >. 40? . IAcetylated _
a: 2:: :
W -.
2 30.
= g ..
2" 20:
h ..
° :
Z 10,
0' i : : :
b1 y3 b2 y2 b3 y]
Sequence Ion
50: |
4510 Penta i
: yc ne
.2 40: ) g] I
3 352 ,
“and
g g 255 ‘ 1. Underivatized:
N “ =
:5 5 205 Acetylated
e 15;
2 10;
5;
0‘ :

 

 

 

b1 y4 b2 y3 b3 y2 b4 y]
Sequence Ion

Figure 55. Bar graphs comparing normalized RIs of b,, and y,, peaks in MI spectra of the
underivatized and acetylated forms of a) triglycine, b) tetraglycine, and c) pentaglycine.

197

fragmentation of certain b,, and y,, ions in metastable decomposition. For acetylated
triglycine in Figure 55a, the R1 of the peak representing the b2 ion is reduced by two-
thirds, and likewise, for acetylated tetraglycine and pentaglycine, the peaks representing
the yz and y3 ions, respectively, are reduced which initially suggests the N-terminal amine
secondary interaction in the underivatized oligoglycine induced their formation more than
the N-terminal acetylated amide nitrogen. In acetylated triglycine, tetraglycine and
pentaglycine, the R13 of peaks representing y2, y3 and y4 ions increased dramatically
because in one scenario, conditions for a new secondary interaction are created in which a
hydrogen can be shifted to the protonated amide nitrogen at skeletal bond 2 (in the
underivatized peptide) from the acetyl group to form the respective y,, ion, a ketene
molecule and a three-membered cyclic glycine product. However, in the more
thermodynamically favorable scenario, a hydrogen may be shifted from the newly created
acetyl amide nitrogen to form a three-membered cyclic neutral product and the respective
y,, ion. For tetraglycine and pentaglycine in Figures 55b and c, the RIs of peaks
representing the b3 ions also increase because perhaps the secondary interaction involving
the N-terminal acetyl group itself provides a larger, more stable ll-membered cyclo-b 3
ion in comparison to the nine-membered cyclo-b 3 ion formed in underivatized
oligoglycines.

The M1 fragmentation from the oligoglycine series indicates in a broader sense
that secondary interactions are the preferred thermodynamically favorable mechanisms
for producing stable fragments from [M-t-H]+ ions with low internal energies. For
hexaglycine, the neutral products formed in conjunction with the dominant y 3, cyclo-b4
and cyclo—b5 ions are the small, highly stable molecules, cyclic diglycine, diglycine, and
glycine, respectively, which is similar to low—energy CI. So far, in the MI discussion
above, the peptides were assumed to be protonated on the amide skeletal bonds as in
LSIMS, but the N -terminal amine nitrogen in protonated form may also interact with the

amide skeletal bonds in the peptide to induce fragmentation in metastable decomposition.

198

Perhaps protonation of amide nitrogens results in immediate amide bond cleavages as in
LSIMS and protonation of amine nitrogen results in delayed amide bond fragmentation,
on the metastable time scale, after an intramolecular proton transfer through secondary

interactions.

MIRIL [ErM !' . E! H'

The RI distribution of MI skeletal bond cleavage products for methionine
enkephalin in Figure 50b is very similar to the RI distribution for pentaglycine in Figure
49b. However, the presence of the phenylalanine side chain in methionine enkephalin
introduces an additional low-energy pathway to form the a4 ion. Because inductive
cleavages would not be favored in metastable decomposition, perhaps the interacting N-
terrninal amine from a secondary interaction to amide bond 11 catalyzes loss of CO for
the formation of linear 04 or cyclo—a4 ions following cyclo-b4 ion formation through
multiple hydrogen bonding. The unimolecular decomposition of acetylated enkephalin
pentapeptides in LSIMS, MI and CAD results in a very low a4 ion abundance in contrast
to those from analysis of the underivatized enkephalin peptides. This suggests that either
the N-terminal acetyl amide group is too "bulky" to interact effectively and act as a
"catalyst" in the formation of a4 ions, or the acetyl group interacts very well with amide
skeletal bond 11 forming cyclo-b4 ions exclusively. In Figure 50b, cleavage of amide
bonds 2, 5, and 8 in protonated methionine enkephalin metastable decomposition results
in dominant y4, y 3, and y2 ions, respectively. These y,, ions have similar abundances in
contrast to the singly abundant )3 ion for pentaglycine because the tyrosine side chain
likely enhances the stability of the cyclic neutral products formed. When methionine
enkephalin is acetylated, the y4 ion becomes more abundant similar to the oligoglycines
above probably because now a hydrogen can be shifted from the acetyl amide nitrogen
which requires less energy. Internal ions and the phenylalanine immonium ion, typically

considered high energy fragments, also appear in the MI spectra for methionine

199

enkephalin which further suggests that they can are formed from low energy pathways
involving subsequent secondary interactions from y,, precursor ions. In addition, the a1
and y 1 ions which were most abundant in LSIMS, are not observed in the MI distribution
indicating significant energy is required for production of these small ionized fragments.
In addition, if the C-terminal acid group is converted to a methyl ester, the low mass y,,
ions in metastable decomposition increase in abundance which indicates the methyl ester
further stabilizes these smaller fragment ions. Last, if the enkephalin C-terminal acid
group is converted to a terminal amide, C3 and 2,, ions become prominent peaks in the
MI/CAD spectrum because a secondary interaction involving nitrogen is more favorable
than oxygen.

A metastable decomposition map was created in Figure 56 by selecting the
protonated, underivatized methionine enkephalin molecule, each sequence ion and
various non-sequence ions in the LSIMS mass spectrum as both precursor and product
ions for product ion and precursor ion scans (using linked scans at constant B/E and
B2/E), respectively, under metastable decomposition conditions, to investigate secondary
fragmentation processes in more detail. A much simpler but similar CAD map for
leucine enkephalin was presented by Alexander and Boyd [217]. In contrast to El, many
MI product ions are detected from the LSIMS experiment which suggests multiple
protonation sites are slowly decomposing, probably through N —terminal amine secondary
interactions, instead of one dominant charged site. Some secondary fragmentation is
observed on the metastable time scale from primary fragment ions formed immediately
after ionization in the LSIMS experiment. For example, when b4 ions were selected as
the precursor ion for a product ion scan, an intense metastable peak representing the a4
fragment ion, a secondary fragmentation process, was detected. When a4 was selected as
the product ion for a precursor ion scan, an intense peak representing the b4 precursor ion
was detected. If y,, ions are the selected precursor ions, intense peaks representing the

internal ions, immonium ions or lower series y,, ions are detected as shown in the

200

.m0t0m .053 E 2.2 00.20.58
:0 n83 5000 2 38805 0.8 $95.58 05 :5 .53 $528 E8050. 2.65 @0320 05. .mtmm 28 m5 E8250 3 280m “002::

Boa @0588 5.230520 05:250.: mo +E+S= .8 39553 5289502. 0382208 05 9:32? 92: 5:35.:me .om 05$"—

 

 

 

 

 

 

 

 

 

meamem
AI R: N
”3:085me ”V .0
$0M a
me
3 Al m AIIIIINA N01 1* N2
\\ +E+Sz +E+§\
/
/
/ m.
£26 All. 2 \/ a.

 

 

 

\ /. . . \.//..
\ E\
\\ V0\V

wmimumv lllll ”HOD TI Fm

201

metastable map.

Evaluation of the CAD Relative Intensity Data.

Outside the ion source, single collision CAD excites stable [M+H]+ ions from one
high energy interaction with He to induce fast fragmentation. The [M+H]+ internal
energies from CAD are high enough on average to induce simple bond cleavages as in

LSIMS.

MW

In Figure 49c, the "true" single collision CAD distribution of skeletal bond
cleavage products versus skeletal bond positions are shown for pentaglycine from which
the MI products were subtracted out. All sequence ions are identical to those obtained by
LSIMS above except a small fraction of the a2 ion abundance is [a2+H]+' and a4 is
[a4+H]+'. Assuming fragmentation is induced by protonation of the amide nitrogens, the
RI distribution of fragmentation induced by CAD at amide skeletal bond positions 2, 5, 8,
and 11 is 12%:40%:27%: 13%, respectively, in Figure 49c which is nearly identical to the
LSIMS RI distribution for pentaglycine in Figure 49a. The RI distribution of peaks
representing b,, ions in CAD is identical to the distribution of peaks from LSIMS in
Figure 49a but not the distribution of peaks from MI in Figure 49b most likely because
the [M+H]+ internal energies are high enough to competitively induce simple bond
cleavages. In contrast to the distribution of peaks in M1, the RI distribution of peaks
representing b,, ions in CAD may also indicate an "unzipping" process is occurring in
which one precursor ion such as the b4 ion is sustaining formation of the other b3 and b2
product ions, which is energetically feasible from the fragmentation barrier profile in
Figure 54b. Alexander and Boyd [217] showed for leucine enkephalin under extreme
multiple collisions that the distribution of CAD product ions can be made to resemble a

FAB or LSIMS ion distribution indicating either primary fragmentation processes

202

become competitive only at high internal energies, secondary ionization processes can be
induced by high internal energies from subsequent collisions, or a combination. The RI
distribution of peaks representing y,, ions in CAD resemble the MI distribution instead of
that for LSIMS probably because the secondary interactions or rearrangement processes
are strictly rate dependent and the average internal energies in LSIMS may be slightly

higher, which may account for the y,, RI differences in the distributions.

r r i ' In.

In Figure 50c, the "true" single collision CAD product distribution for methionine
enkephalin, with the MI products subtracted out, is shown and it appears to be quite
similar to the MI distribution in Figure 50b above. The addition of three large side chains
to the pentapeptide backbone has doubled the number of degrees of freedom in
methionine enkephalin which means the internal energy imparted into the [M+H]+
becomes more dispersed and the rates of fragmentation decrease to resemble metastable
decomposition. Thus, single collision CAD simply enhances most metastable
fragmentation processes for the larger, neutral protonated peptides. However, the
reappearance of the a,, ions and [a3+H]+’ ions, at skeletal bond position 7, and the
increase in b2, b3, yz, and y 1 ion peak intensities in the CAD distribution indicate a small
population of [M+H]+ ions have high internal energies for fast, higher energy
decomposition. There is no evidence for conversion of primary products to secondary
products in the single collision CAD of protonated methionine enkephalin as in LSIMS.
The protonation sites probably differ initially before fragmentation in LSIMS and CAD
as indicated in the RI distributions in Figures 50a and c, respectively, because in the
LSIMS distribution, the most intense peaks represent the (1] and y 1 fragment ions, but
these peaks are not as intense as those in the CAD distribution. Perhaps, in comparison
to the LSIMS amide nitrogen protonation, the stable [M+H]+ ions that are excited in

CAD are probably protonated on N-terminal amine nitrogen, amide carbonyl oxygens or

203

the side chains inhibiting fast fragmentation processes. CAD simply initiates
fragmentation processes from these existing protonation sites or redistributes the proton
to the amide nitrogens by secondary interactions which subsequently initiates

fragmentation as in LSIMS.

Formation of [b4+0H+H 1+ Ions.

Recently, for peptides, much published work has focused on exact fragmentation
mechanism of [b4+OH+H]+ ions from metastable and CAD [M+H]+ unimolecular
decomposition [218, 219]. In Scheme XIV, assuming the peptide is protonated on the
basic amide carbonyl oxygen rather than on the less basic amide nitrogen, as suggested
by Teesch and Adams [203] for alkali metal cationization, a nucleophilic attack from the
C-terminal acid group on the first inside carbonyl group can occur favorably though a
five-membered ring. However, some current discrepancies in the [b4+OH-t-H]+
mechanism include: where does the proton reside during the reaction?; how is the -OH
group transferred mechanistically?; and what types of neutral products are eliminated? In
all the work to date, no one has tested whether the energetics of the various mechanisms
are reasonable because the elimination of CO and NH=CHR require that three bonds are
broken, and one and half bonds are formed which leaves a very large energy deficit well
above the energy available in MI decomposition. However, as shown in Scheme XIV,
the formation of a three-membered cyclic neutral reduces this large energy deficit since
now two bonds are broken and two similar bonds are formed which results in an

estimated overall reaction enthalpy of only 23 kcal/mol.

204

4. Conclusions

The sequence and non-sequence peptide fragmentation mechanisms were
evaluated using both RI information in LSIMS, MI and single collision CAD mass
spectra and estimated reaction enthalpies, similar to practices in 131. In E1, the first step in
elucidating fragmentation mechanisms was to describe the structures of the ionized
molecules. For E1 of dipeptides, the amine and amide nitrogens are the likely sites of
ionization owing to their low IE [220]. From the R13 of fragment ion peaks in the mass
spectrum, the distribution of fragment ions originating from these two ionization sites
were determined after elucidating the fragmentation mechanisms. E1 was more simplistic
for peptide analysis than LSIMS primarily because only simple peptides could be
analyzed. In LSIMS, peptides with higher molecular weights and a greater number of
functional groups may be protonated, which makes the protonated analyte structure(s)
more difficult to describe. For example, the peptides in LSIMS may be protonated
exclusively at the most basic site, from which all fragmentation originates, or at a basic
site(s) remote from where skeletal bond cleavages occur. However, evidence for these
types of fragmentation mechanisms are not observed in the mass spectra neutral
pentapeptides here, which contain no highly basic residues for preferential protonation.
Instead, the neutral pentapeptides appear to be protonated, with a certain distribution, on
the amide nitrogens, amine nitrogens, amide carbonyls, and to a lesser degree on the side
chains. If all amide nitrogens were initially protonated equally on a pentapeptide with no
side chains to influence fragmentation, such as pentaglycine, then it would be expected
that the distribution of fragment ions from each amide nitrogen would be equivalent.
However, the fragmentation observed from these protonated amide N sites in

pentaglycine is not equal and it appears that secondary interactions are responsible, in

205

part, for either inducing chemistry at each amide nitrogen site and/or redistributing the
initial proton distributions to effect fragmentation. The LSIMS RI fragmentation
distribution for pentaglycine and the estimated reaction enthalpies suggest that both
simple cleavages and secondary interactions are competitive at each protonated amide
nitrogen. Although the most favorable fragmentation pathways are not always observed,
the most stable ionic and neutral products are formed in all cases, whether a simple
cleavage or secondary interaction is required. The M1 RI fragmentation distribution for
pentaglycine in contrast to LSIMS indicates that N-terminal amine secondary interaction
mechanisms are responsible for each of the fragment ions observed in the mass spectra.
The RI distributions observed for methionine enkephalin reveals that each side chain
significantly enhances local fragmentation which indicates side chains stabilize the
fragments formed.

Much work still remains in the elucidation of unimolecular decomposition
mechanisms for peptides in LSIMS and FAB. This work represents one of the first real
attempts to quantify the energetics of the peptide fragmentation mechanisms. However,
the results reported here are preliminary because our knowledge of conformations of
peptides in the gas phase, internal energies from the LSIMS experiment, etc., are limited
at this time preventing us from exploring other mechanistic possibilities. Neutralization-
reionization mass spectrometry (NRMS) is promising for aiding in the identification of
the types of neutral products following unimolecular decomposition of metastable
[M+H]+ ions and energized [M+H]+ ions in CAD, providing soft ionization techniques
such as CI can be implemented for reionization [221]. Gas-phase confirmations of
protonated peptides should be investigated initially using computer molecular modeling
techniques to learn more about the [M-I-H]+ structural relationship to the R1 of peaks in
the mass spectrum and eventually to the fragmentation mechanisms from each protonated
site that involve specific secondary interactions. Experimentally, electrospray ionization

(E81) and [222] laser desorption (LD) are the two softest ionization methods for

206

producing gas-phase molecules and perhaps gas-phase conformations of peptides could
be accurately investigated at first, without fragmentation processes occurring, using
spectrochemical methods. Methods should be developed to accurately estimate the
appearance energies of the peptide fragment ions using hybrid, ion beam mass
spectrometers coupled with desorption/ionization techniques to experimentally confirm
the estimated reaction enthalpies presented here. Laser desorption/chemical ionization
(LD/CI) coupled with FT -MS detection seems promising for creating breakdown curves
to probe the energetics and rates, k(E) functions, of peptide fragmentation mechanisms

[223].

CHAPTER FIVE. FUNDAMENTAL AND ANALYTICAL UTILITY OF
GAS-PHASE METAL ION/BIOMOLECULE REACTIONS

I. Introduction

Product ion structures, organometallic bond strengths, equilibrium constants, and
rate constants of gas-phase ion/molecule reactions between bare metal reagent ions and
small organic molecules with less than two functional groups are typically determined at
or near thermal energies on the millisecond time scale using an ion cyclotron resonance
(ICR) mass spectrometer [224, 225]. Dedicated ion-beam mass spectrometers [226] are
used to obtain appearance energies of organometallic product ions for calculating
respective heats of formation (AHf) values. These fundamental studies with ICR and ion-
beam mass spectrometers allow both the elucidation of reaction mechanisms for gas-
phase organometallic chemistry, and direct correlations to be made between the gas-phase
organometallic chemistry, in the absence of a solvent, and the respective condensed-
phase chemistry. Gas-phase ion/molecule reactions may also be studied in regions of
high pressure in a conventional mass spectrometer such as the enclosed ion source
volume utilized in chemical ionization (CI), the volume (or the selvedge region)
immediately enclosing the area from which molecules are desorbed into the gas-phase in
potassium ion ionization of desorbed species (K+IDS) and fast-atom bombardment (FAB)
experiments, or in a collision cell filled with inert noble gases or reagent gases for
collisional activation dissociation (CAD) or endothermic reactions, respectively. In
comparison to an ICR, the typical mass spectrometer is more restricting because only
reactions occurring faster than a microsecond are observed, the ions generally have

energies much greater than thermal energies, and lower sensitivity is usually observed

207

208

owing to ion transmission losses. However, mass spectrometers are readily available and
provide faster scanning speeds, higher mass ranges and higher resolution than the ICR
mass spectrometer. In the last 10 years, the fourier-transform mass spectrometer (FT-
MS) [227], has evolved into the state-of-the-art instrument because it incorporates all the
inherent advantages of the ICR plus it also incorporates the high mass and resolution
advantages of the sector mass spectrometers and the fast scan speeds of quadrupoles. The
fundamentals of ion/molecule chemistry, advanced instrumentation and
desorption/ionization (D/I) techniques are now available for study of ion/biomolecule
gas-phase chemistry, but this area is still virtually unexplored because the observed
chemistry is still too complex for accurate mechanistic interpretation just yet owing to the
complex interactions of bare metal ions with multiple functional groups of highly
structured and functionalized biomolecules. In this chapter, some fundamental and
analytical advantages of ion/biomolecule reactions are explored and demonstrated using
various metal reagent ions (M+) incorporated into the K+IDS and K+IDS-by-FAB
experiments, more correctly referred to as M+IDS and MfiDS-by-FAB, respectively

The ionization technique, K+IDS, utilizes gas-phase K+ ion/molecule chemistry in
the ion source volume of a mass spectrometer to obtain the molecular weight and some
structural information for analytes of biological origin as explained in Chapter two.
K+IDS is essentially a K” CI experiment in which low-energy, gas-phase K+ ions
(cat)ionize desorbed analyte molecules (M) by simply adding to M without inducing any
fragmentation. Thus, K+ is an excellent probe for determining the types of neutral
molecules present in the gas phase since the respective [M+K]+ adduct ions are detected
for each M without fragmentation. In K+IDS-by-FAB, low-energy K+ ions are injected
into the selvedge region, a glycerol-CI environment, to probe the types of neutral
molecules that are desorbed in the fast-atom bombardment (FAB) experiment as
explained in Chapter three. Metal-containing compounds may be substituted into the

aluminosilicate glass matrix instead of potassium, and M+IDS and M+IDS-by-FAB

209

experiments may be performed analogous to those based on 161138 and K+IDS-by-FAB
to access whether other metal ions, M+, are as analytically useful as K+. In K+IDS and
K+IDS-by-FAB, the K+ reagent ions just "stick" to gas-phase molecules, but by replacing
K“ with other metal reagent ions, the outcome of the ion/molecule chemistry will likely
be complementary to K+. First, the M+ adduct ion may yield unique fragment/product
ions either naturally (exothermic) or when activated (endothermic) which would provide
new, informative structural details for M. Second, if the metal ion has an unique isotopic
distribution, then peaks in the mass spectrum that represent either an M+ adduct ion or a
respective fragment ion will be better identified and confirmed. As an example
application, M+ in a CI environment may test the functionality of unknown molecules by
reacting or adding to those molecules containing only specific functional groups [228].
Recently, chemical ionization of peptides with Fe+ was initially demonstrated which
resulted in the losses of H20, CO and C02 from the [M+Fe]+ ion, and in the cleavage of
amide bonds for the formation of meaningful peptide b,, and y,, fragment ions containing
Fe+ [229]. This fragmentation is contrary to the well known insertion of Fe+ into Co,-
C(O) bonds. Probing gas-phase, highly functionalized molecules such as peptides with
various metal ions should produce fragment/product ions that yield complementary
structural information to those from other mass spectrometric D/I methods which may
help elucidate fragmentation mechanisms and lead to unambiguously identifying the gas-
phase binding sites [230]. These gas-phase binding sites, free of solvent effects, may lead
to identification of the active metal binding sites on biomolecules in condensed-phase

systems aiding in the elucidation of biological mechanisms.

210

2. Reactivities between Metal Ions and Analyte Molecules in M +IDS

The gas-phase ion/molecule chemistry utilizing alkali metal reagent ions is well
characterized because copious amounts of alkali ions can be thermionically emitted into
the gas-phase from aluminosilicate emitters doped with alkali metal substances owing to
the low ionization energies of the alkali metals and the high work function of this emitter
composition [60]. Cs+ and Rb+ only form adduct ions similar to K4" in K+IDS because
the large size of these metal ions prevents sufficient overlap of electron density between
M and the metal ion resulting in a very weak bond. In [M+K]+, the bond is electrostatic
which means the charge resides external to M because the ionization energy (IE) of K is
lower than M. Because the charge is remote from M, charged-induced chemistry cannot
occur but charge-enhanced chemistry can if M is sufficiently activated and polarized by
the metal ion. The metal ions, Na+ and Li+ in Na+IDS [54] and Li+IDS [173],
respectively, form adduct ions with molecules, too, and they also can induce simple
neutral losses such as water loss to form fragment ions containing a metal ion. A
stronger, more covalent-like bond with better overlap of electron density is formed
between these smaller metal ions and M and the IE is greater, which means the charge is
nearer M to possibly initiate charge-induced chemistry. In M+IDS, less common metal
reagent ions, M+, can be thermionically emitted into the gas phase from a heated
aluminosilicate emitter to either form adduct ions, form products from reactive
(exothermic) collisions, or undergo charge exchange with gas-phase biomolecules
depending on the physical properties of the metal ion. In K+IDS and M+IDS,
biomolecules are either thermally desorbed intact and/or thermally decompose through
only the lowest energy pathways depending on the heating rate and biomolecule

structure. In addition, there is no matrix in M+IDS which means there are no matrix

211

effects as in FAB. Thus, for biomolecules, there are only several analyte-related peaks to
interpret in K+IDS and M+IDS mass spectra which makes the extraction of analyte
molecular weight and structural information simple. Therefore, M+IDS is ideal to probe
and characterize the diverse metal ion/biomolecule chemistry under the normal operating
conditions in a typical mass spectrometer, which would be complementary to K+IDS,
K+IDS(M+IDS)-by-FAB, and studies using ICR or FT-MS techniques. Below, the
preparation and characterization of Al+, Ga+, and In+ emitters is described first. Next,
the preliminary results from Al+IDS, Ga+IDS, and In+IDS experiments and their

analytical advantages are presented.

M+IDS Emitter Preparation.

Blewett and Jones [52] initially optimized the molar ratios of an aluminosilicate
matrix doped with Li2C03 to obtain the largest amount of low-energy, thermionically
emitted Li+ ions when the glass emitter was heated to greater than 1000° C. For
preparation of an optimized Li+ emitter, they melted, in powdered form, either Li2CO3 or
LiN03, A1203 and Si02 together in a 1 or 2:1:2 molar ratio to form Li20:A1203:2Si02
glass following either C02 or N02 release, respectively. This glass composition is the
same formulation as the well known ternary system B-eucryptite. They obtained Li+ ion
currents as great at 1 mA, and also discovered that other alkali ion emitters such as 10‘
may be prepared with the same molar ratios as the B-eucryptite Li+ emitter, which was
recently confirmed to give the strongest K+ ion currents [231]. Blewett and Jones [52]
also described the production of Mg“, Ca+, Sr+, Ba+, Al+, Cat“, and In+ ions from oxide
emitters, which provided ion currents less than 1.0 microampere and involved simply
melting the oxides MgO, CaO, SrO, BaO, A1203, Ga203, In203 on a filament wire,
respectively. Successful Cu“ and Ag+ aluminosilicate emitters were prepared using
Cu(NO3)2 or AgNO3 mixed with A1203 and silicic acid in a l or 2:1:2 molar ratio,

respectively [232]. Ni+ and Cr+ borosilicate emitters were successfully prepared in a

212

manner similar to that used for the B-eucryptite aluminosilicate emitters described above
except that boric acid and silicagel were substituted for A1203 and SiOz [233].
Commercially available alkali and alkaline earth ion emitters are based on the B-
eucryptite formula, and are designed to emit copious amounts of ions [234]. They are
used in applications such as the Cs+ emitter pellet in secondary-ion mass spectrometry
(SIMS) guns. The Cs+ emitters for the SIMS gun cost in excess of 300 dollars, but they
can be simply regenerated successfully by incrementally melting small amounts of
CsN03 or CszCO3 into the top of the porous tungsten plug.

The Ga+ and ln+ emitters were prepared in a manner similar to that used for the B-
eucryptite K+ aluminosilicate emitters above instead of melting the respective oxides.
Thus, Ga203 or In203 was mixed with A1203 and SiOz using the molar ratios 1 or 1:1:2,
respectively. All refractory reagents were high purity and crushed with a mortar and
pestle to a powder. An attempted formulation of a C0"' glass emitter with the initial
formula, Co(NO3)2:A1203:2Si02, apparently vaporized upon heating in the Co+IDS
experiment. Al+ emitters were prepared exclusively with the pure, powdered reagent
grade Al(NO3)3-9H20. Acetone was added to the mixed, powdered solids to produce a
"slurry" for controlled formation of the emitter. A Pasteur pipette was used to "drop" or
"layer" the emitter solid from the "slurry" onto the rhenium heater wire loop which was
pre-attached to the direct insertion probe (DIP). After the acetone evaporated, the "dried"
emitter solid was heated briefly in a Bunsen burner flame to firmly attach the applied
solid mass to the Re wire loop, through crystallization or melting, for successful insertion
into the mass spectrometer. If the emitter is heated for extended periods under
atmospheric conditions, the Re wire will begin to oxidize which will drastically shorten
the lifetime of the emitter. Quite often, N02 or C02 gas was released from the
"crystallized" emitter solid upon heating, and this applied solid was splattered off the Re
wire. If this occurred, the emitter powder was first melted to a glass to evolve these

gases, then the glass was recrushed to powder, and reapplied to the Re wire from the

213

acetone slurry. An alternate, more difficult method for emitter preparation was to dip the
Re wire loop into a platinum dish of molten metal-containing aluminosilicate emitter
material, heated with the most inner flame (i.e., the hottest part) of the oxygen-acetylene
torch, and capture a "bead" of molten emitter at the same time being careful not to
oxidize the Re wire or vaporize the Pt dish. The K+IDS, Al+IDS, Ga+IDS, and In+IDS
experiments were conducted on a Hewlett Packard (HP) 5985 GC/MS/DS single
quadrupole mass spectrometer equipped with a direct insertion probe inlet. Thermionic
emission was achieved by passing current through the metal-containing emitter. The K“,
Al+, Ga+, and In+ emitters were part of a current loop in which the emitter rhenium wire
ends were spot-welded to the nickel heater leads on the end of a standard two-filament
K+IDS DIP which were connected to the heating current loop of K+IDS power supply
unit described in Chapters two and three. Before the K+, Al+, Ga+, and In+ emitters
could be used for M+IDS, they had to be "conditioned" first which involved melting the
emitter onto the Re wire. This step was done in the mass spectrometer under vacuum to
prevent oxidation of the Re wire at elevated temperatures and the procedure involved
raising the emitter heating current 0.2 A every 20 sec until 3 A was reached. Heating the
10, Al+, Ga+ and In+ emitters resulted in emission of only the respective ions without
significant co.emission of "impurity" ions with lower ionization energies. Thus, there
was no need to "bake out" any "impurity" ions from these emitters for extended periods.
In the K+IDS, Al+IDS, Ga‘l'IDS, and In‘l'IDS experiments, gaseous analytes were leaked
into the ion source CI volume, into which the emitter was inserted, until the ion source
housing pressure was approximately 1x10‘5 torr, and solid analytes, dissolved in MeOH,
were "dried" onto the opposite Re wire sample holder after the MeOH evaporated .

Liquid analytes were applied directly Re sample holder.

214

M+IDS Emitter Characterization.

The A1+, Cal and In+ ion currents were measured from experiments conducted in
both a vacuum test chamber attached to a vacuum rack as shown in Figure 57a, and in the
ion source of the mass spectrometer. In the vacuum test chamber, an M+ glass emitter,
attached to the current loop of the K+IDS power supply unit via the feed-throughs, was
heated to induce thermionic emission. A realistic portion of the ions were detected by a
faraday metal plate and measured by a picoammeter despite the fact that ions are emitted
in a 360° radius of the bead. In Figure 57b, the In+ ion current and emitter bead
temperature as a function of heating current is plotted for the In203:A1203:28i02 glass
emitter. Microamperes of In+ current are readily obtained between 2.1 and 2.3 A and
above 2.9 A of heating current. In comparison, the K20:A1203:2Si02 and
Ga203zA1203:2Si02 glass emitters yield 10 11A of K+ and 1 ILA of Ga+ above heating
currents of 2.9 A under the same experimental set-up and conditions, respectively. In the
HP mass spectrometer, Int, 63*, Al+ and K+ emitters typically yield 9100, 2200, 8000,
and 3500 absolute A/D signal counts at heating currents of 2, 2.7, 3, and 2.6 A and at
electron multipliers voltages of 1700, 1700, 2000, and 1200 V, respectively. K“ ion
currents are strongest because the ionization energy (IE) of potassium is lowest (4.3 eV)
in comparison to Al+ (6 eV), Ga+(6 eV), and In+ (5.8 eV) which have [55 slightly greater
than lithium (5.4 eV) assuming the work function of the emitter matrices remain constant.
Comparable ion currents of transition metal ions such as Co+ (7.9 eV) are more difficult
to produce thermionically because the IEs are 2 eV higher than those for the other metals

described above.

M+IDS Experimental Results and Discussion.
The gas-phase ion/molecule chemistry between Al+, Ga+, and In+ and simple,

gas-phase analytes was investigated and compared to that of K+. In Figure 58a, the

215

3)

 

Reference 7

   

To Vacuum :7 I, , _
Rack ' ‘ ‘ :sg'i: Power Supply

. ' 7 0-20V'

Heater Current

Power Supply
0-4 A “

 

 

 

 

\f\‘
(Re.

 

    

 

K IDS Power Supply Umt. Ground

 

 

 

 

 

 

 

 

 

 

 

 

                

 

 

 

Picoammeter E:J(Keithley #417)
Ground
M+ or K+ Ions
Vacuum Chamber
1.6E-06 - 1600
g 14E-06 b) . l_.,-/"'—.’-—. 14‘”
2 1.2E-06 .4" "' . 1200 g:
g 1.0E-06 ”An/"M . 1000 j
g ’./' I-
U 8.0E-07 u" . - ' ' ' ' 800 g
E scram . ' - 600 g.
4’: 4.0E-07 ' —D— In+ Ion Current 400 E
mam ' . 200
0.0E+00 - 0
Q——NQVW\O§OQO§N-1NMVW\O_OOGM
o u— — v—n — —n v— v— v— — N N N N N NN NN

Heating Current (A)

Figure 57. a) Schematic diagram of the vacuum chamber, K+IDS power supply unit, and
picoammeter used for testing M+ thermionic emission ion currents from alurrrinosilicate

emitters doped with various metal substances, and b) a plot of In+ emission ion current and
emitter temperature versus heating current for the In203zA1203228i02 emitter.

216

[M+K]+ peak in the mass spectrum represents K+ attachment to gas-phase acetone (MW
58). In this experiment, acetone was leaked into the ion source closed CI volume and
[M+K]+ adduct ions were formed as K+ CI reagent ions were thermionically emitted
from the K+ emitter bead on the DIP inserted into this CI volume. This same experiment
was performed with In+ emitted in the open EI ion volume instead which resulted in a
significant drop in the total ion current (TIC) as shown in the mass spectrum in Figure
58b in comparison to that in Figure 58a. The peak at m/z 173 represents the [M+In]+
adduct ion of acetone. In fact, In+ bound preferentially to the more volatile acetone
component in many solvents such as alcohols than the less volatile solvent itself since
acetone is the commonly used denaturing substance in solvents. Contrary to adduct
formation with K+ and In+, the mass spectrum in Figure 58c indicates Al+ extracts Cl'
from CCl4 (MW 152), which was leaked into the El ion source volume, since CCl3+ is
the dominant peak at m/z 117. The reactive Al+ also provided additional peaks in the
mass spectrum that are only 5% of the base peak; these represent meaningful
fragment/product ions that aid in characterizing this analyte. Under thermal conditions,

the reaction between A1+ and CC14 in Equation 45 occurs because the reaction enthalpy

Al+ + CC14 (g) -) CCI3+ + AlCl (g) (45)

(Aern) is -9 kcal/mol exothermic [235]. If the same reaction is performed with K+ or
In+, no reaction is observed because the estimated AHnms are 49.5 and 12.5 kcal/mol
endothermic, respectively; these reactions are not possible under thermal conditions [40].

Analyses using Ga+IDS and Al+IDS were also performed with the solid fatty
acids, palmitic acid (MW 256) and tridecanoic acid (MW 214), dried on the Re wire
sample holder, and compared to those obtained by K+IDS. In Figure 59b, in the Ga+IDS
mass spectrum of palmitic acid, the peak at m/z 325 represents the [M+Ga]+ adduct ion,

which indicates Ga+ simply forms adduct ions with fatty acids in a manner similar to

217

 

 

‘ a) 39 Metal Ion: K+
K-I- Analyte: Acetone
" Heating Current: 2.6A [M+K]+
Bias Voltage: 0 V
Electron Multiplier: 1200 V 97
- TIC (Counts): 15849

 

 

 

I l
....I... I....I....I....I....I....I....I....I....I....I....I....I. "I
30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- b) Metal Ion: In+ 173
Analyte: Acetone
‘ Heating Current: 2 A [M+In]+
I Bias Voltage: 2 V
Electron Multiplier: 2700 V
. TIC (Counts): 100
llllllllllll lllTTlT
160 162 164 166 168 170 172 174 176 178 180
-— Metal Ion: Al+
l C) Analyte: CC14 CCI3+ Cs+
Heating Current: 3 A 117 133
Bias Voltage: 4 V AIC12+
Electron Multiplier: 2400 V 97
1 TIC (Counts): 13475
. 1 75 x1
x AlCl+ AICHCI+ I.
‘ CC1+ 62
47
AIC+ CC12+
ll 1 J II J

 

 

 

 

 

 

 

 

 

 

I.fi.I....I....I....I....I....Ifi..I.. .....I I.-'... I. .. I
30 40 50 60 70 80 90 100 110 120 130 140

Figure 58. a) K+IDS and b) ln+IDS mass spectra of acetone (MW 58), and c) A1+IDS
mass spectrum of CC14 (MW 152) leaked into HP ion source to 1x10'5 torr base pressure.

218

 

 

. a)
I Metal Ion: K+
' Analyte: Palmitic Acid
- Heating Current: 2.6 A
. Bias Voltage: 0 V
Electron Multiplier: 2500 V
7 TIC (Counts): 851

 

 

 

 

295
[M+K]+

ll

 

 

 

VVVVVVVVVVVV

‘1'.

"" ""l 711‘]

 

 

 

 

 

 

 

 

 

 

 

 

 

1 #1 1 l

l

 

 

 

 

 

b) [M+Ga]+ 325
Metal Ion: Ga+
Analyte: Palmitic Acid
H. Current: 2.9 A [M+Na]+
B. Voltage: 8 V 279
E. Multiplier: 3000 v [M+K]+
TIC (Counts): 1207 295
‘ H... ....I ..5. .Iu....l ‘. "I. .. .r...
240245 250255260265 270 275 280285 290295 300305 3I0 315 320 325 3'30
c) 295 [M+K]+
... Metal Ion: Al+
[M+A"C02] Analyte: Palmitic Acid
239 H. Current: 3 A
[M+Na]+ B. Voltage: 4 v
E. Multiplier: 3000 V
279 TIC (Counts): 794 339

 

 

 

.1

 

 

 

 

 

 

 

XX) 220 240 260 $0 300 320 340 360 380
d) 237 Metal Ion: Al+ l
+ Analyte: Tridecanoic

[NH-Na] H. Current: 3 A Aci .
+ B. Voltage: 4 V g
[M+Al-C021 E. Multiplier: 3ooov I
197 TIC (Counts): 467 I
I

251

279

150160170180190 200 210 220 230 240 250 260 270 280 290

Figure 59. a) K+IDS, b) Ga+IDS, and c) Al+IDS mass spectra of palmitic acid (MW 256),
and d) Al+IDS mass spectrum of tridecanoic acid (MW 214) on the HP mass spectrometer.

219

those formed with K+ as shown in the K+IDS palmitic acid mass spectrum in Figure 59a.
The intense 716a isotopic peak conveniently indicates which peaks in the mass spectrum
represent ions containing gallium. The Al+IDS mass spectra of palmitic acid and
tridecanoic acid are shown in Figures 59c and d, respectively. No [M+Al]+ adduct ion is
observed for either analyte so if this analyte was an unknown, 3 101138 analysis is still
required for molecular weight information. However, a peak representing the [M+Al-
C02]+ ion is observed for each analyte indicating Al+ reacts specifically with the
carboxylic acid functional group. There is no mention of this reaction in the Al+ ICR
literature, probably because carboxylic acid compounds are not very volatile and could
not be studied. Al+ may prove to be a good reagent ion to test analytes for the presence
of carboxylic acid functional groups. The gas-phase chemistry of Al+, Ga"; and In+ has
been characterized previously with compounds having various functional groups [236].
In+ and Ga+ are similar to K+ because they only form adduct ions with M, but Al+ reacts
with the fatty acid analytes tested here and previously with small molecules in the
literature. The observed chemistry is attributed to decreasing bond energies
Al+>Ga+>In+ since the electrostatic interactions and radii increase [237].

Last, in further support of the FAB ion formation mechanism, K+IDS mass
spectra of common FAB liquid matrices glycerol (MW 92), triethanolamine (MW 149),
thioglycerol (MW 108), and nitrobenzyl alcohol (MW 153), shown in Figure 60a-d,
respectively, were obtained from successive experiments to demonstrate that K+ binds
more favorably to the highly basic triethanolamine rather than to the other acidic matrices
as indicated by the TIC values. Thus, K+ affinities likely parallel proton affinities as
observed previously [61]. Additionally, in the 101138 triethanolamine spectrum shown in
Figure 60b, the peaks at m/z 114 and 132 represent either gas-phase unimolecular
decomposition of protonated triethanolamine molecules formed in a "local" selvedge
region, or desorption of protonated degradation molecules of triethanolamine. Further

support for a "local" selvedge region in the K+IDS experiment is the K+-bound cluster

220

 

 

 

 

 

 

 

 

>\

E ' “31m: Glycero 131

8 ': eating Current: 2.9 A [M+K]+
a) 5 .. ias Voltage: 3 V

“:3 4 lectron Multiplier: 1600 V

:33 j C (Counts): 4554

£1""I"" "' " "I 'I""1fi"1'l'

70 75 80 85 9095 10010511011512012511)135140

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

>‘ ,
g : 132 188 AnaTyte: Triethanolamine
g . [M-H20+H]+ [M+K]+ HC: 2.9A
b) 3 - 114 BV: 3v
,5 ; [M-2H20+H]+ EM: 1600V
g . 11c: 52228
D
m I"' '1""I""I " 1"" 1'1
100110 120130140 15016017018019021D210 220230 240250260270 280
2:.
Z” l 147 finflyte: Thioglycero
2 I + C: 2.9A
0) E. - [M+K] BV: 3 V
g . EM: 1600V [2M-H20+K]+
g : TIC: 1742 237
a: ..............................................................
100110m130140 m1m1'701801'9102mz'10m230240250260
b
‘53 r 192 Analyte: Nitrobenzylalcoho
g: j [M+K]+ HC: 2.9A
d) .s .. BV: 3v
0 ‘ EM.16OOV +
g - TIC: 3829 [212?]
7: ‘ . 1.1 .
a: ..................................................

100121)]:10160180200220240260280300320340360

 

 

 

 

 

>5
g - 85 [G+K]"’ Analyte: Glycerol/Cholic Aci 447
2 j Rb+ 131 BC: 2.9 A [M+K]+
e) .5 .. BV: 3 V
2 . EM: 2600V
‘5 j TIC: 18613
T) _. 1.141 _ "tut....rul.1 .1. 1.1.; until-1;... ...‘lnkl Lu LLIIJLALIAAJHIILkaj
of. 1 l 1 1 1 1

 

 

 

 

 

70 110 150 190 230 270 310 350 390 430 470 510 550 590 630

Figure 60. K'l'IDS mass spectra of a) glycerol (MW 92), b) triethanolamine (MW 149), c)
thioglycerol (MW 108), d) nitrobenzyl alcohol (MW 153), and e) cholic acid (MW 408)
mixed in glycerol.

221

ions represented by peaks at m/z 237 and 345 observed in the K+IDS thioglycerol and
nitrobenzyl alcohol spectra in Figure 60c and (1, respectively, although these peaks may
also represent intact desorption of this dimer species. In Figure 60c, the K+IDS mass
spectrum of cholic acid (MW 408) dissolved in glycerol on the Re sample holder is
shown. The appearance of the [M-l-K]+ peaks at m/z 131 and m/z 447 indicate that
glycerol and cholic acid molecules, respectively, are desorbed intact from the glycerol
matrix into the gas phase identical to the FAB experiment as determined by KflDS-by-
FAB. The glycerol matrix essentially protected cholic acid from thermal decomposition
in this experiment because cholic acid readily lost water molecules in a K+IDS

experiment when cholic acid was dried on the Re wire sample holder.

3. Gas-Phase Metal Cationization in FAB Using MflDS-by-FAB

FAB mass spectra of biomolecules tend to be more complex to interpret than
those from K+IDS or M+IDS. Spectral interpretation in FAB may be hindered by sample
and matrix effects because the analyte or mixture of analytes is dissolved in a matrix for
analysis. Also, there are a larger number of fragment ions to interpret in FAB because
more energy is imparted into the biomolecule analytes. However, interpretation of all the
fragment ions is not always possible because most fragmentation mechanisms are not
well known. Often, metal ion cationization of the analytes in FAB can aid spectral
interpretation by overcoming matrix effects and producing more informative,

complementary fragmentation.

M t I a 'x ts.
In the FAB mass spectrum, any analyte peak can be potentially blocked by a

matrix-related peak. Also, the peak at high m/z, that reveals the analyte molecular

222

weight, may represent a protonated molecule, [M+H]+; a radical cation, M'”; a salt
cation, M+; a dehydrated fragment ion, [M+H-H20]+; proton-bound dimer of the analyte,
[2M+H]+; or a proton-bound dimer of the analyte and glycerol, [M+G+H]+ depending on
the chemical nature of the analyte and its intrinsic chemical relationship with the matrix.
For samples of biological origin, large amounts Na+ salts (and/or potassium) are typically
present in the sample, which results in an additional peak representing the sodium adduct
ion of the analyte, [M+Na]+, appearing 22 Da higher than the [M+H]+ peak. The
presence of salts in the matrix sometimes suppresses the other analyte peaks, but this
intense [M+Na]+ adduct ion peak actually can aid in identifying the peak that reveals the
molecular weight of the analyte amidst all other peaks. The analytes in salt-free samples
may be cationized with alkali metal ions with high sensitivity by simply dissolving alkali
salts such as NaI, NaCl, KI or KCl in the matrix at concentrations around 0.1 M. If the
sample is a mixture of analytes, then the alkali metal will cationize each component,
confirming which peaks represent intact analytes. If the sodium contamination or
concentration is very high, then additional, intense peaks appear in the mass spectrum
representing the sodium adduct ion of the sodium salt of glycerol, [M-H+2Na]+;
[M+(NaCl)+Na]+; [M-H+(NaCl)+2Na]+; etc., which may confuse or hinder the
identification of the analyte(s) peak(s), especially if there is a mixture of analytes in the
sample. Assuming the analyte dissolves in the matrix and no condensed-phase reactions
between the analyte and matrix occur, the FAB matrix may still suppress analyte ion
formation. For example, if the analyte molecules are not surface active, but strongly
"matrix-philic", they will not be readily desorbed from the matrix into the gas phase. If
the proton affinity (PA) of matrix is greater than the analyte, then protonated matrix
molecules will not protonate the analyte molecules to a great extent, if at all. Presently, it
is unclear whether analytes are cationized by metal ions in the condensed phase and
subsequently desorbed preionized, or whether gas-phase cationization is the dominant

mechanism.

223

In K+IDS-by-FAB, K+ ions are injected into the selvedge region containing
desorbed neutral molecules (M) and [M+K]+ adduct ions are formed indicating the types
of neutral species present. The KflDS-by-FAB method is not only useful for probing the
neutral molecule species in the selvedge region of the FAB experiment, but it can also be
used for molecular weight confirmation because only one new peak representing [M+K]+
adduct ion appears in the mass spectrum 38 Da higher than the protonated molecule, even
if the [M+H]+ is unstable, eliminating interpretation problems. The [M+K]+ ions formed
in K+IDS-by—FAB are generally less abundant than [M-l-K]+ ions formed by adding 0.1 M
KI salts to the matrix (see Chapter three) because K+ cationization is achieved solely in
the gas-phase. However, gas-phase K+ cationization in KfiDS-by-FAB eliminates any
matrix effects and the higher mass K+ adduct ions of KCl salt cluster ions caused by the
addition of KCl salts to the matrix which yields much simpler spectra to interpret. As
observed in Chapter three, gas-phase K+ is very effective in ionizing analytes with lower
PAs than the matrix which results in [M+K]+ ions that have greater abundance than the
[M+H]+. For a full discussion of the analytical utilities of K+IDS-by-FAB, see the
published paper [124] in Appendix one. The use of metal ions besides K+ in the FAB
experiment either using M+IDS-by-FAB or dissolving metal-containing salts in glycerol
may complement K+ by providing unique isotopic distributions for easy identification of
metal-containing ions as was demonstrated for Ag+ cationization of cholic acid (MW
408) [132]. Ni, Cu, and Ga have only one striking isotopic peak which would provide

easy identification of the cationized ions without loss of sensitivity.

a - 1 nt ment ti
Metastable decomposition and CAD of protonated and cationized analytes formed
in FAB require tandem mass spectrometric techniques for the detection of the fragment
and product ions formed after acceleration outside the ion source, respectively. Tandem

mass spectrometric techniques in conjuction with metastable decomposition and CAD is

224

a method to: isolate the fragmentation from each analyte component in a mixture,
separate matrix peaks that interfere with fragment ion peaks in the FAB mass spectrum,
enhance existing fragment ion abundances (in CAD), and supply unique and/or
complementary, structurally informative fragmentation for an analyte, all of which can
increase the chances of accurately defining the structure of the analyte molecule [238,
239, 240, 241]. The fragmentation differences induced by protons and metal ions are
attributed to a combination of the metal ion binding site location on the analyte, M+-
analyte bond strength, and the changes in conformation the analyte undergoes to host the
interacting metal ion. The metal ion binding sites for peptides probably differ from
proton binding sites if intact ions are formed in either the condensed [188, 242] or gas
phase [203, 206] which would result in different sets of bonds being preferentially
cleaved and account for much of the unique, complementary fragmentation and different
ion abundance distributions in comparison to those observed following protonation.
Assuming the average internal energies of cationized analytes and protonated analytes are
similar and the metal ion and proton binding sites are identical, the intrinsic interaction
between the metal ion and the analyte molecule may initiate different fragmentation
mechanisms than proton-induced chemistry also resulting in complementary
fragment/product ions. The analyte-metal ion bond strength represents the energetic
upper limit for all complementary metal ion-induced bond cleavages and fragmentation
processes observed because if enough internal energy is present to cleave the bare metal
ion from the neutral analyte molecule, then the chemistry stops and complementary
fragmentation requiring more energy will never be observed. If the metal ion-analyte
bond is strong, more structurally informative fragmentation can be potentially obtained
since there will be probably other weaker bonds to cleave. The ionic radius of an alkali
metal ion, in part, dictates how close it can approach the exposed electron density at the
heteroatom site on the analyte molecule before nuclear repulsions occur in order to

achieve good overlap for sufficient bond strengths. Larger metal ions such as Cs+ are

225

unable to achieve as good an overlap as Li+, a smaller ion, and the bond energies between
Cs+ ions and analytes are much weaker and more electrostatic than those for Li+. The
ionization energy dictates, in part, on which species the analyte or metal ion the charge
lies. For K+ interactions with analyte molecules, the ionization energy (IE) of K is 4.4
eV, which is half the value than those for most analytes. Thus, the electron density
remains shifted towards the analyte and the charge remains on potassium. Thus, K+ can
only trigger fragmentation processes by polarizing the local bonds through charge
enhancement which almost borders on remote charge-site chemistry [243]. In contrast,
Li+ has a smaller radius and a higher ionization energy which results in it forming
stronger, more covalent-like bonds with analytes, and the charge, which induces bond
cleavages, is likely closer to the analyte. Experimentally, more fragmentation is observed
with Li+ than 10. In contrast to cationization-induced fragmentation, a proton has an IE
of 13.6 eV, which is higher than that for most analytes, so the charge is transferred onto
the analyte in exchange for electron density which essentially makes the analyte-H bond
covalent. For a protonated heteroatom site with intermediate basicity, the fragmentation
chemistry is believed to be solely charged-site induced, but if the heteroatom site is
highly basic and all protonation occurs here, then remote charge-site chemistry may take
place. The interactions of transition metals, such as Co+ (3d74sl, not 3d3), with analytes
is different than with the alkali metal ions, which have inert, noble gas electron
configurations, because the extra remaining valence shell electrons allow Co+ to behave
as an ionic radical and insert into C-C bonds [224]. This process is thermodynamically
favorable because one C-C bond is cleaved and two Co+-C bonds are formed.
Additionally, the interaction and formation of electrostatic bond(s) between analytes and
metal ions, that have large ionic radii and diffuse electron clouds, most likely results in.
more analyte conformation changes to accommodate the larger metal ion in comparison
to minimal perceivable changes in conformation for proton attachment which is small

point charge. These new conformations will probably have much influence on the new

226

and/or complementary fragmentation observed with metal ions [244].

For any biomolecule analyte, the possible types of complementary fragmentation
can be probed by dissolving different metal substances into the glycerol matrix and
selecting the cationized analyte produced by FAB as the precursor for metastable
decomposition and CAD. With metal ions in the FAB matrix, perhaps the intrinsic
interactions between analyte biomolecules and the metal ions in the condensed-phase can
be sampled with FAB which can possibly lead to the determination of the aqueous metal
binding site(s). In order to determine the metal binding sites, the cationized analyte
biomolecules must yield informative fragmentation by metastable decomposition and
CAD. Once a metal is found that yields suitable complementary fragmentation, the metal
substance can be inserted into the MflDS-by—FAB experiment to ensure the analyte is
only cationized in the gas phase without any matrix effects from having metal ions
dissolved in the matrix. With M+IDS-by-FAB, the intrinsic interactions between gas-
phase biomolecule and metal ions can be sampled and compared to the condensed-phase
M+-analyte interactions in the FAB matrix. Finally, the fragmentation of cationized
analyte molecules in M+IDS-by-FAB will not consistently parallel M+IDS gas-phase
fragmentation chemistry since the cationized molecules in the FAB experiment have
higher average internal energies.

All mass spectra in this section were obtained on a JEOL HX-110 double focusing
forward geometry mass spectrometer. The M+IDS-by-FAB target was placed on the end
of a direct insertion probe. The MflDS-by-FAB experiment was performed identically to
that of K+IDS-by-FAB as described in Chapter three and in Appendix one using the same
divided targets. The preparation of the metal-containing emitter substances and analytical

advantages are described below.

227

MflDS-by-FAB Cationization Results and Discussion.

Since our laboratory has considerable knowledge and continuing interest in
probing gas-phase transition metal chemistry, transition metals were first inserted into the
M+IDS-by-FAB experiment to study ion/biomolecule chemistry. The aluminosilicate
matrix doped with Co(NO3)2 from M+IDS was tried initially by melting it onto one half
of the wall-divided target in Figure l9e, but no Co+ ions were observed. In Figure 61a, b
and c, Co(NO3)2, Cu(NO3)2, and Ni(NO3)2 were melted on the wall-divided target and
Cot“, Cu+, and Ni+ ions represent the dominant peaks, respectively, but the ion currents in
absolute AID counts are very low in comparison to the 11 counts of K+ which was
typically obtained by FAB of the K+ aluminosilicate emitter melted on one half of the
wall-divided target shown in Figure 61d. However, no Zn+ ions were obtained from FAB
of Zn(NO3)2. The number of counts appears to decrease in the order, Cu+ (7.7 eV) >Ni+
(7.6 eV) >Co+ (7.9 eV)>Zn+ (9.4 eV), as the metal ionization energies (IE) increase. Ti
(6.8 eV) and Cr (6.8 eV) have lower IEs than the metals above so the production of Ti+
and Cr"’ ions may be more successful in MflDS-by-FAB, respectively. The compounds
Ti02 or Ti2(SO4)3 and Cr(NO3)3 could be used as sources for these ions, respectively.

Despite the low Co+ ion currents, CoflDS-by-FAB was still demonstrated to
work with glycerol (MW 92) and the peptide H—LGG-OH (MW 245) using the wall-
divided target in Figure 1% without any evidence of condensed phase contamination as
shown in Figures 62a and b, respectively. However, it was later discovered that Co+ does
not ionize these analytes to a great extent owing to a mysterious ion series in the glycerol
Co+IDS-by-FAB mass spectrum at m/z 213, 305, etc. separated by 92 Da. In one
scenario, protonated glycerol reacts with gas-phase intact Co(NO3)2 to produce the
product ion, [G+CoN03]+, appearing at m/z 213 in the mass spectrum as shown in

Equation 46 . Co(NO3)2 can either be desorbed intact or formed by Co“ and NO3’

228

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

81°C 59
5 8013) (30+
E 60‘ [30+ Emitter:
3 Co(NO3)2 CoOH+
'3 40‘ [30+ Counts: 76 CONO"?
'3 20‘ 06 ° 89 .
M G - - - ' v - v i
20 4O 6O 80 100
m/z
>5
"5 1:310) 58 i
9- Mi+ Emitter: Ni+ . E
{3' 60‘ Ni(NO3)2 (171%0)Ni+ 5
'5 40‘ Ni+ Counts: ‘60 NiNO‘fll
’3 20: 0.7 88 .
a: l r 62 ,
o _
20 4O 60 80 113/0
2
£9100 23 + . ;
E 80: e) Na+ a Glass Errutter E
SE) 60' Na+ Counts: 3 :
'3 A1+K+ :
.2 401 27 39 + E
'3 20‘ l [45 58 E
M O'F—-—-" -'A!—V:L?f 2'77 :
0 20 40 60 80 100
m/z
$1003) 23
‘3 . + Li+ Glass Emitter
o 80 L.+ Na
E 60: ; Al+ Li+ Counts: 14
..g. 40' 27 K4"
5 20: 39 Fe+
§ 0 ##5‘6
0 20' ' '4'0' ' '6'0' ' '8'0' ”100
m/z

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2:100 .
'53 ‘ 91+ i
5*: 23b) Cu+ Emitter: 3
T, CU(N03)2 I
.> 40‘ 65 .
g 20‘ 01+ Counts: CuNO‘l";
2 0.8 I 913 ,
G *2? '4‘01 60' '76 100
m/z
.2100
2 80: d) 3%. +Glass Emitter ’;
3 K + i
2 4o: '
'fi 20.
3 ‘ 2.3 u =
00 5'6 100"15o' 200
m/z
>~100
.3 + 185 1
§ 30' f) [633?” [2G+H]+
t: .
E
32
O0 50 100 150 200 250
m/z
g: 100'Fh) [69'3“] 1851
5 80‘ +
E so
3 40'
'3; 20:
00 50 100 150 200
m/z

Figure 61. FAB mass spectra of a) Co(NO3)3, b) Cu(NO3)3, c) Ni(NO3)3, d) K+ glass

emitter bead on wall-divided target in Figure 1%, e) Na+ glass emitter bead on "pop-up"
divided target in Figure 19f, f) Na+IDS-by-FAB mass spectrum of glycerol, g) FAB mass
spectrum of Li+ glass emitter bead on KflDS-by-FAB space-divided target, h) LifiDS-by-

FAB mass spectrum of glycerol.

229

[G+H]+ + Co(NO3)2 (g) -> [G+CoNO3]+ + HNO3 (g) (46)

recombination in the selvedge. Thus, the ion appearing at m/z 150 is [G-H+Co]+ which
is formed by subsequent loss of HNO3. Sunner et al. [97] reported a very similar
mechanism for the reaction of protonated glycerol and KC] in the gas phase. Oddly
enough, if Co(NO3)2 is dissolved in the glycerol matrix, only Co+ ions are observed
indicating that Co(NO3)2 is desorbed primarily as a neutral species because it is surface
active and essentially blocks desorption of glycerol molecules. For intact analyte species
containing Co+, perhaps the Co+ is exchanged from [G-H+Co]+ to the analyte by a
subsequent gas-phase collision to account for the [M-H+Co]+ observed in the H-LGG-
0H mass spectrum in Figure 62b. Although more experimentation would need to be
done to ensure that this is indeed a gas-phase process, it implies that salts such as KNO3
dissolved in glycerol are desorbed intact into the gas phase and react with protonated
glycerol or the protonated analyte to form [G+K]+ and [M-i-K]+ ions, respectively. This
K+ adduct ion formation mechanism may account for the high sensitivity observed in the
practice of adding KI salts to glycerol since more neutral species are desorbed in the FAB
experiment than ions. In Figure 62c, the FAB mass spectrum of Co(NO3)2 and the
analyte digoxin (MW 780) mixed in glycerol is shown and the peak at m/z 838
representing the [M-H-l-Co]+ ion is observed. The low-mass peaks normally in the
protonated digoxin FAB spectra at m/z 97 and 131 are present without Co+ attachment.
The [M-H+Co]+ ions were selected as the precursor ion for a linked scan at constant B/E
under metastable conditions as shown in Figure 62d. The distribution of metastable [M-
H-l-Co]+ fragment ions is similar to those of the metastable [M+H]+ decomposition of
digoxin because the intense peak at m/z 708 and the smaller peak at m/z 579 represent
loss of one and two sugar moieties identified in Figure 62d [176], respectively. However,

the peaks at m/z 602, 664, and 734 are not present in [M+H]"' metastable mass spectra

230

 

é

 
 
    
 

 

 

 

 

 

 

 

  
  

 
 

 

 

 

 

 

 

 

 

 

 

 

 
 
    
   
  

    
   

  

 

                  

1 93 [2G+H]+ 242 CoflDS-by-FAB
£9 804 3) [G+H]+ 185 [ZG-H+C0]+ of glycerol ’.
g 1 on wall target I
a 601 E
T, : [G-H+Co]+ 305 3"” :
2 . + [3 -H+Co] :
33 20-1 294517 J ”(3351] 334 397;
G 0' "50' ' ' 166 150 200 250 300 350 4‘30
2
100, 86 r
>‘ 1 b) CoflDS-by-FAB of H-LGG-OH :
.53 80; a] in glycerol on wall target ;
§ 6,: a
5 : [G+H]+ [ZG-H+Co]+ [M-H+Co]+ :
.L‘a’ 40? 44 :93 [G-H+Co]+ 242 303 .-
35. : G W :
8 2°: :
0* L
0 100 200 300 400
m/z
100
>‘ : c) 131 Co(NO3)2 & digoxin mixed E
553' 80', in glycerol on same target
C l
95’ 50‘: 97 [M-H+Co]+ .-
3 : [ZG-H+Co]+ 338 :
7°? 20: 43 186 337 x5.0 i
a: 1 I 408 i
O: “12* Jul - r“ t .‘ c - - c ‘1 ‘ - fi fi.
0 200 400 600 800 1000
m/z
3 Metastable decomposition 708 338'
€103 d of [Digoxin-H+Co]+ from c) l [M-H+C0]+
g i [(AOS}OSZOH)-H+Co]+
C: 1
7, : [(AOS}OH)-H+Co]+
.2 5:
‘5 1 734 800
3 j 602 664
a: l 404 541579
00 100 200 300 400 500 600 700 800

 

m/z

Figure 62. a) Co+IDS-by-FAB mass spectrum of glycerol and Co(NO3)2 solid separated

on a wall-divided target in Figure 1%, b) CoflDS-by-FAB mass spectrum of H-LGG-OH
(MW 245) in glycerol and Co(NO3)2 separated on a wall-divided target, c) FAB mass

spectrum of digoxin (MW 780) and Co(NO3)2 mixed in glycerol, and d) [digoxin-H+Co]+
metastable decomposition mass spectrum obtained by a linked scan at constant B/E.

231

and they probably represent ring cleavage products either on the aglycone or sugar
moieties .

Na+IDS-by-FAB and LiflDS-by-FAB experiments were also conducted using
Na+ and Li+ aluminosilicate emitters that were prepared identical to the K+ emitters in
K+IDS-by-FAB. In Figure 61c, FAB of the Na+ glass emitter on the "pop-up" target in
Figure 19f yielded only three counts of Na+. The NaflDS-by—FAB mass spectrum of
glycerol in Figure 61f shows Na+ was injected into the selvedge region to produce the
peak at m/z 115 which represents the [G-l-Na]+ without any evidence of contamination.
FAB of the Li+ glass emitter on the space-divided target in Figure 21a yielded 14 counts
of Li+ in Figure 61g owing to the superior geometry of the divided target. The Li+IDS-
by-FAB mass spectrum of glycerol again shows more abundant Li+ adduct ion formation
with glycerol than Na+ adduct formation even though more Na+, from impurities in the
Li+ emitter, were injected into the selvedge region than Li+ owing to the stronger binding

energies of Li+ to alcohols [60].

Metastable and CAD Studies of Cationized Peptides.

There has been much attention in the mass spectrometry community concerning
cationized peptides in recent years because they provide complementary fragmentation in
comparison to that of protonated species. The peptides are usually mixed with alkali
metal salts in an appropriate matrix and subjected to analysis by FAB . Metastable and
CAD analyses of the cationized peptide are the methods of choice for obtaining the
complementary fragmentation because cationized peptides are very stable and they do not
usually undergo unimolecular decomposition on the time scale of the FAB mass
spectrum.

For metastable decomposition and CAD experiments, linked scans at constant B/E
were employed. Helium was the collision gas used in CAD to attenuate the ion beam.

The conversion dynode was adjusted to maximize abundance of C5312+ (m/z 652.5) at an

232

accelerating potential of 1 keV [245] before mass-axis calibration. Ultramark 1621 (PCR
Inc., Gainsville, FL) was used as the mass-axis calibration compound for linked scans at
constant B/E compounds because according to the formula for calculating apparent
metastable fragment ion masses, m*=m22/m1, the lowest linked scan fragment/product
ion mass, m2, obtainable with an [M-l-H]+ precursor ion representing an analyte of
molecular weight 800, m1 is m/z 28 if the lowest Ultramark calibration peak, m*, is at
m/z 1. For comparison, with CsI calibration, the lowest calibration peak is Cs+ at m/z
133 and the lowest fragment/product ion mass obtainable with a linked scan at constant
B/E is m/z 258. The operating pressure of the instrument with FAB Xe gas on was 5x10‘
6 torr.

K+IDS-by-FAB provides the capability to form [M+K]+ ions solely in the gas
phase with the bare K+ metal ion. Using the peptide H-VGVAPG-OH (MW 498) as a
model, metastable and CAD studies were conducted to probe whether the fragmentation
for these gas-phase [M+K]+ ions formed by K+IDS-by-FAB differed from the [M+K]+
ions formed by FAB of both KI salts and analyte dissolved in the glycerol matrix.
Currently, it is unclear if these latter [M+K]+ ions are formed in the condensed or gas
phase. Any differences in fragmentation in the two methods may indicate a different
distribution of K+ binding sites in [M+K]+ ions because [M+K]+ unimolecular
decomposition for both methods is observed under the same metastable or CAD
conditions.

In a FAB mass spectrum, a cationized analyte molecule appears to be quite stable
because little or no fragmentation from this species is normally observed. However,
unique and complementary [M+K]+ metastable unimolecular decomposition is observed
outside the ion source. Thus, the observed metastable fragment ions originate from
[M-l-K]+ ions only having low internal energies because, if an [M+K]+ ion had too much
internal energy, the electrostatic bond between M and K+ would be cleaved and all

fragmentation chemistry would stop. If the internal energies of an [M-l-K]+ ion extended

233

above the range of energies for metastable decomposition into the range for ion source
decomposition, but still below the M-K+ bond energy, the fragmentation rates of these
"energized" [M+K]+ ions may still be low and extend into the metastable time scale
because K4”, which is electrostatically bonded to an analyte, may not be able to promote
fragmentation as promptly as a covalently bound proton.

The metastable mass spectra of the [M-l-K]+ precursor ions of the cationized H-
VGVAPG-OH peptide formed by KflDS-by-FAB and K1 salts in glycerol are shown in
Figures 63a and c, respectively. A comparison of these two metastable spectra shows
peaks representing the same types of ions with very similar intensities indicating that both
[M+K]+ species undergo the same fragmentation mechanisms since both types of
[M+K]+ precursor ions have very similar internal energies and represent a similar
distribution of isomers with K+ bound to different sites. However, there are some minor
exceptions. The K+IDS-by-FAB metastable mass spectrum in Figure 63a shows one
dominant peak at m/z 365 representing the [b4-H+K]+ fragment ion whereas in the
metastable mass spectrum using KI salts in Figure 63c, this same peak is much less
intense. Also, the peak representing the loss of C02 in the K+IDS-by-FAB metastable
mass spectrum is also much more intense than that obtained when using KI salts. When
KI salts are added to glycerol, K1 is probably desorbed intact into the selvedge region and
through multiple, therrnalizing collisions, K+ is attached to glycerol and exchanged
multiple times, losing its kinetic energy, and ultimately attached to a variety of basic sites
in equal numbers to form the isomeric and very stable, low-energy [M+K]+ adduct ions
which may differ slightly from those involved in K+IDS-by-FAB [97]. Here, the bare K+
metal ion may gain up to 2 eV when it is injected in or near the selvedge region from the
gas phase which will limit its collision cross section with the peptide; these high-energy
K+ ions, on average, will then preferentially only stick to the most basic regions with
exposed, high electron densities such as the C-terminus carboxyl group oxygens, or the

proline amide nitrogen and the two neighboring amide carbonyl oxygens in the rigid

234

 

     
    

  
 

 
 

 

 

 

 

 

 

 

    
 
    

 

 

R 521] 317
‘ [b -H+K]+
i: ‘ a) 4365 [M-l-K]+
a l KHDS-by-FAB of H-VGVAPG-OH l
j . . [b5+OH+K]+ 3
E , [M+K]+ metastable decomposrtron -C0 '
v ‘ [y5+H+K]+ 495 j
e 438 e 494 l
I +.
I ‘ yz [04+K] [05-H+K]+ 445 430 -V l
{‘ j 173 H [2(163gH+K]+ 338 43: l
e , 244 H K + b4 U4+H+K1
n . [8+ Eli 327 381 40322
s . ,
i 1
t .
Y
200 300 400 500
m
[a3-H+K]+ gym“ [b H+K1+
Lufmxr, b4
H3C\ {CH3 H3C\ CH3 [05-H+K]+
c 0 [c3+H+K]+ ”2
b) I I 0
cg (I! C H C\H2
/ V \ {C
HzN (If \Céz Nfi II 1 N 2 OH
O 0 A113 0
[y5+H+K]"’ [y4+H+K]+ )3 )7;
R 1 521 5397
f j c) H-VGVAPG-OH & K1 in glycerol '1} .t
a J [M+K]+ metastable decomposition [NH 1
E . [b5+on+K]+
r
v + -V
e Us+H+Kl 494
_ +
1 ”’4 Hm] [a5-H+K]+ rt
:1 ‘ + 445
, yz [a -H+K] +'
e ‘ 173 3266 104;? 365 434 480
n 3 [C3+H+K]+ ; | 408
§ , )‘3 311b4 [y4+H+K]+ 438
1 244 32 381 422 I
‘ b
y . 3
100 200 300 400 500
m/z

Figure 63. The [M+K]+ adduct ions of H-VGVAPG—OH (MW 498) in glycerol formed by
a) KflDS-by-FAB and c) FAB of 0.1 M KI salts in glycerol were selected as the precursor
ion to undergo metastable decomposition. In b), the structure of H-VGVAPG-OH is shown
with the prominent fragmentation sites. The * represents a glycerol contamination peak.

235

molecular region around the proline residue of this peptide. The valine side chains
probably interfere with K+ binding at the N-terminal half of the peptide. These preferred
sites are supported by loss of C02 and the intense [b4—H+K]+, respectively, assuming K+
charged-enhanced (or -polarized) or -induced chemistry. In K‘TDS-by-FAB, perhaps the
attachment of higher kinetic energy K+ ions induces the formation of the abundant [b4-
H+K]+ ions. Charged-enhanced or -induced mechanisms have been proposed for amide
skeletal bond cleavage and [b4-H+K]+ formation [246]. K+ can polarize the amide bond
in such a way to form the intermediate, [b,,-H]---K+---[y,,-H], with the loss of H2 from
hydrogens on the two neighboring a-carbons through a six-membered ring, and because
the K+ affinity of [b,,-H] molecule is greater than the [y,,-H], [b4-H+K]+ ions are formed.
The appearance of the [b5+OH+K]+ rearrangement ion further supports K+ being bonded
to the carbonyl oxygens at the C-terminal half of the peptide because K+ will polarize the
carbonyl group making the carbon more positive enhancing this site for C terminal -OH
nucleophilic attack [188, 203].

In general, the [M+K]+ precursor ions selected for CAD have low initial internal
energies and would not normally fragment before the electric sector of the JEOL HX-l 10
without collisional activation which means these ions were probably thermalized after
their formation in the selvedge region by subsequent third-body collisions. CAD of the
respective precursor [M-rK]+ ions, attenuated by 50%, formed in KHDS-by-FAB and
from KI salts for the same peptide H—VGVAPG-OH are shown in Figure 64a and 0.
CAD essentially increases the internal energy content of the peptide which initiates the
fragmentation that requires higher energy as shown by the intensity increase of the peak
representing the [a 3-H-t-K]+ ion in both CAD spectra in comparison to the metastable
spectra. In the KflDS-by-FAB CAD spectrum in Figure 64a, the peaks representing [b4—o
H+K]+ and loss of C02 are still intense in comparison to those obtained when using KI
salts providing more evidence that 10‘ in KHDS-by-FAB may attach preferentially to the

amide bond between alanine and proline and to the C—terminus, respectively. For low-

w~wwao~=H G<WHm—ox

~<-nm=a~a~ n<-~m-—o;o

236

 

 

 
    

 

 

 
 

  

  
 
  
 

 
   

 

 

 

 

 

 

 
    
 
 
 
 

 

 

 

 

 

 

A
, 493 537]
. a) KflDS-by-FAB of H-VGVAPG-OH {02 war
4 CAD of [M+K]+
I 494
d [a 3-H+K]+ [04-H+K]+ [er+K]+ -YCH4
: 266 I365 [x4-H+K]+ 521
[a5-t-1+1(]+
[y H+Kl+ H+K1+
. ”3*H*K]+ 4+ 31 43!: U;
I (I1 yz [cz+H+K]+ [03mm]+ 422 [1,5,pr
. 72 173 212 b3 232 311 ‘07 64
. 256 294
K+
‘ 39
0 100 200 300 400 500
m/z
01 [62+H+K]+[a3-H+K]+ [a,,-HMO+
_ ‘T “ [b4-H+K]"'
H3C CH3 H3C‘ CH3 “ [05-H+Kl+
b) I 0
C H II
/ v
H2N fi 2
005mm U4+ +x1+0
[x 5-H4-K]+ [I4‘H+K]+ [y3+H+K]+ YZ
. -v 5379
j C) H-VGVAPG—OH & K1 in glycerol 494 [M+I(]"'
_ +
: [23H+K] .CH4
1 521
I [x4-H+K]+ [05_H+K]+
I Iy4+H+K1+ 434
4. yz [b4-H+K]+ [y5+H+K]+
1 173 [a H+K1+ +
‘ 21 2 r 337 365 [x5-H+K]
‘ ”b +
j 3 b3+H+Kl 4074 2 433464
j, 244256 2.2 311 381
100 200 300 400 500
m/z

Figure 64. The [M+K]+ adduct ions of H-VGVAPG-OH (MW 498) in glycerol formed by

a) KfiDS-by-FAB and c) FAB of 0.1 M KI salts in glycerol were selected as the precursor
ion for CAD at 50% beam attenuation. In b), the structure of H-VGVAPG-OH is shown
with the prominent fragmentation sites.

237

energy K+ cationization of H-VGVAPG-OH formed from KI salts, the similar
fragmentation patterns observed in both the metastable and CAD spectra indicate that the
peptide may "wrap" around K+, which is located centrally, interacting intramolecularly
with K+ at multiple, interior heteroatom sites, creating a stable [M+K]+ ion. Thus, the
differences in RIs of some peaks in both metastable and CAD spectra may indicate that
K+ binds slightly different, owing to its initial kinetic energy and/or initial form as a
reactant , in K+IDS-by-FAB and when using KI salts, creating a distribution of different
[M+K]+ isomers that decompose in each case. These comparisons are valid since these
spectra were obtained under the same set of conditions and the [M+I(]+ adduct ions from
each technique had similar ranges of internal energies.

Next, metastable and CAD experiments are presented to demonstrate how
different metal ions hinder or enhance the fragmentation for peptide sequencing using the
model peptide, H—VGVAPG-OH (MW 498). The types of ions and distribution of ion
abundances represented in the metastable mass spectrum of protonated H-VGVAPG-OH
in Figure 65a differs greatly from those obtained during the decomposition of Li+, Na+,
and K+ cationized H-VGVAPG-OH precursor ions shown in Figure 65b, c and Figure
63b, respectively. The mass spectra of cationized species yield more unique, diverse and
useful analytical structural information than the protonated species because the protons
are most likely bonded to the amide nitrogens only inducing specific bond cleavages
providing limited fragmentation. In comparison, the alkali cations are most likely bonded
to H-VGVAPG-OH with weaker yet different binding energies, in the order, Li+ > Na+ >
K+, to sites involving several carbonyl oxygens which may alter the native peptide
conformation and induce multiple types of fragmentation mechanisms through
polarization. For example, the [b5+OH-t-Cat(ion)]+ is more intense in the metastable
spectra of [M+Li]+ and [M+Na]+ than in the metastable spectrum of [M+K]+ because
these metal ions can polarize a carbonyl group more effectively than K+ [203]. The peak

representing the [b5+OI-I+Cat]+ is not in the metastable spectrum of protonated H-

238

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R a H-VGVAPG-OH in glycerol 327 499
f [M+H]+ metastable decomposition b4 [M+H1+
a .
t
i 1
v 1
e i
I
{1 y2
e 173
n )3
is . 244 b5 -CH4
1 1 b3 0 424 483
4 Y4 YS -v
y 229 e 4101 ...1
'100' 200' ' 300' " 400' 'rrn5/(Z)0
R b H-VGVAPG-OH & Lil in glycerol :4 8 489l_505
fr‘ [M+Li]+ metastable decomposition [b5+OH+L‘] [M+Li]+
a . +
E . [C3+H+Li]+ 3135mm]
V . [b3—H+Li]+ [a4+Li]+' Ly5+H+L11+
e yz 406
I 1 73 [b4-H+Li]"'
n v +H+L 01+56 463
ii . L 3 I LV4+H+L11+ * I
n 13
l
7 [a -H+L31]+ b4
i’ j a] 3 J25279 327.33
72
1 1 , “In-1L1. I ..l .
100 W200 300
R C) H-VGVAPG-OH & NaI in glycerol 21 +
it [M+Na]+ metastable decomposition [b 5+OH+Na1+ [M+Na]
505
ii 1 [a4-H+Na]+ [b4-H+Na]+
i; . [63+H+Na]+ U4+H+Nal+ ‘
C .
I _ Lb3+Na1+ [“5'H+Na]+ I
n 418 Lv5+H+Na1+
t Lv3+H+Nal+ ; -V
[e] [X4-H+Na]+ l 478
S - -H N + '
i )1.) [a3 +221032149:}5\)11*[:45_8_H+Na]+ ‘
y 173 3 26'6277295 M] M4]: l ‘
1 . '. 1 '24.? L ILL .jlili." H11. '41 iihl.
100 200 300 400 500
m/z

Figure 65 . Metastable decomposition mass spectra of the a) [M+H]+, b) [M+Li]+, and c)
[M+Na]+ precursor ions of the peptide H-VGVAPG-OH (MW 498) in glycerol, 0.1 M LiI
in glycerol, and 0.1 M NaI in glycerol, respectively. A * is 3 glycerol contamination peak.

239

VGVAPG-OH which suggests that protons are either not readily bonded to this carbonyl
oxygen or do not induce this type of chemistry. In Figures 66a-c and Figure 64b, the
CAD mass spectra of the protonated and Li+, Na+, and K+ cationized H—VGVAPG-OH
precursor ions, attenuated by 50%, are shown, respectively. The [b5+OH+Cat]+ peak
becomes less competitive in comparison to the more intense peaks representing the [a 3-
H+Cat]+ ions and the product ions not containing a metal cation. This latter point is
especially true for CAD of [M-l-Li]+ and it suggests that [M+Li]+ is energized in CAD,
the energy gain cleaves the M-Li+ electrostatic bond, a free, "high kinetic energy" proton
attaches to M to form the [M+H]+ ion as indicated by the peak representing [M+H]+ at
m/z 499 in the CAD spectrum of [M+Li]+ in Figure 66b, and this [M+H]+, if it has ample
internal energy, may undergo unimolecular decomposition to form the regular sequence
ions. In comparison to Li+ and Na+ cationization, the amount of fragmentation falls off
with K+ peptide cationization because the binding energy of K" is much weaker limiting
useful fragmentation.

In conjunction with the mechanistic and thermochemical studies of fragmentation
for protonated peptides presented in Chapter four, fragmentation from metastable
decomposition and CAD of K+ cationized peptides were also analyzed for comparison to
charged-site initiated [M-l-H]+ fragmentation because the unimolecular decomposition of
the [M+K]+ would represent low-energy, remote charged-site fragmentation. The
[M+K]+ of methionine enkephalin, H-YGGFM-OH (MW 573), formed using KI salts in
glycerol, was the selected precursor ion for metastable and single collision CAD mass
spectra shown in Figure 67a and c, respectively. These [M+K]+ spectra represent
sequential scans for the same experimental run; the metastable scan was obtained first
and then He collision gas was administered for the second CAD scan. The [M+K]+ peaks
are on the same scale so the R15 of the peaks could be compared and subtracted. In the
single-collision CAD mass spectrum, no new significant peaks are observed in

comparison to the metastable mass spectrum, and a subtraction of the metastable peak

240

 

 

 

 

 

 

 

 

   

 

 

  

 

 

 

 

  
 

 

 

 

 

 

 

 

 

R a) H-VGVAPG-OH in glycerol 327 499
f CAD of [M+H]+ b4 [M+H]+
E’ ' 3'2
1 173
V
C
I ."3
n 244
E
n b3
5 a, 256 _H o
l 72 b2 “3 4b5c V4§1
t p a; 157 223 C3 “4 Y4 -v
Y j ”I 129 1 I 273' 499 343 4004 51456 I
0 "100" '200""300"" 400' 500
m/z
y2 H-VGVAPG-OH & Lil in glycerol 5 5
7 .
§ 1b) 3 CAD of [M+Li]+ IM+LII+
L J [C3+l~l+l.i]+
t 4 [V +H+Li]+ _ +
i; j a, 3 250 [b3-H+L1]+ _ + [a5-H+Li]+ ”13;”
1 72 -H L. + [04-H+Ll] . +
t; [a3 +2 311] [b4-H+Li]+ Ly5+H+L11 .
n L“4+H+Li]+ [b5+OH+Ll]+
t
e -
n
S
i .
t
Y 1
- 100 200 300 400 500
“12
C H-VGVAPG-OH & NaI 250 + 5123
g I in glycerol [a3‘H+Na] -V [M+Nal+
1 CAD of [M+Na]+ [b4-H+Na]+ [i'4-H+Na]+ 47
a -CH4
E I b4+H+Na1 [ H N 1+ 5(‘5
V _‘ H N + 5 05‘ + a
e [c3+ + a] _ + [64+H+Na] I418 [b5+OH+Na]+
I [X3-H+Na]+ [“4 H+Nal
n 292 32‘ [x4- -H+Na]+ “W
t ILi5+H+Na1+
C . N +
[s] 1 L H+Nbli+H+ a]\ I722”? H+Na]+
\' ' a
i 3 236 \ 307 391 II 448 490
3’ i “I 136152 )2 193 l l 266 I [55913]“ i
4 72 295 78 406
.1, 1 . I ”I 173 I208 I LYII II JLIIL I I'It II 11‘: I111I I!l111.' [J I. III
100 200 300 400 SOOmIz

Figure 66. CAD mass spectra of the a) [M+H]+, b) [M+Li]+, and c) [M+Na]+ precursor
ions of the peptide H—VGVAPG-OH (MW 498) in glycerol, 0.1 M Lil in glycerol, and 0.]
M NaI in glycerol, respectively, at 50% beam attenuation.

241

 

 

 

 

    

 

 

 

 

 

R 51 a) Methionine enkephalin 612]
f I [M+K]+ metastable decomposition £35 [M+l(]+
f 41 [b4+on+x]+ CH3SCH3
1 1 431 550
v 1 -NH3
e 3' 595
I 1
1‘ 2; [a4-H+K]+
c ‘ [C3+H+K]+ 435
n I
-5 “WNW [05-H+K]+
{ 1 x+
y 1 319 [bz-H-rK]; 2Y2?
0 1 . , r 1 , 39 .1 L
0 100 200 300 400 500 600
on m/z
H3

[b4+Ol-I+K]+
[b4-l-1+K]+
[b2-n+x]+ [b3-l-l-l-K]+ [a4-H+K]+ H2

1’) (1:2 ‘1’ 11,1133?“ ...
1121/ 8 wise;

[y4+l-I+K]+ [y3+H+K]+

 

 

 

   
 

 

 

 

 

 

 

 

5
R j C) Methionine enkephalin V -Y 612l+
I: 1 Single collision CAD of [M+K]+ 505 [M+1(]
a 4‘ [b4""0H+K]+ 'CH3SCH3
t ‘ 481 550
l I -NH3
z 3: 595
I : [04-H+K]+
" I 435
t 2‘
g 1 [b3-H+K]+
is 1 11:; [a3+K]+' 44 [05-H+K]""
t
y [b -H+K]+
0 1 - - h. c .2 33.9.12
0 160 Y ' 260 I v ' T 300 400 500 600
m/z

Figure 67. a) Metastable decomposition mass spectrum and c) single-collision, high-energy

CAD mass spectrum of the [M+K]+ of methionine enkephalin (MW 573) formed by FAB
of 0.1 M KI salts in glycerol. The methionine enkephalin structure and prominent

fragmentation sites are shown in b). Both [M+K]+ precursor peaks are on scale.

242

R15 from the CAD peak RIs results in the same peaks with much lower RIs representing
the "true" CAD processes (not shown). The lack of new, structurally informative
fragment/product ions, the presence of peaks representing only small neutral losses, and
minimal increases in peak relative intensities in the CAD mass spectrum suggest that K+
is only weakly bound to methionine enkephalin and unable to induce structurally
significant fragmentation. In contrast to K+ cationization of H—VGVAPG-OH which
showed diverse fragmentation, perhaps the large, bulky aromatic side chains prevent
strong K” bonding to M. The types of fragment/product ions represented by the spectra
indicate K+ is attached to the basic sites in the area of the phenylalanine and methionine
residues. The loss of tyrosine (-Y) can be realized by viewing the biologically active
form of the peptide made from semiempirical calculations [247]. It was estimated that
methionine enkephalin folds inward so the charged N- and C- termini interact and the two
aromatic side chain rings and the methionine side chain extend outward in opposite
directions from the backbone. Essentially, if K+ was bound to multiple sites on the
backbone, it would be in the vicinity of all the side chains. Unfortunately, no exact gas-
phase structures or stereochemical information for methionine enkephalin or similar
peptides are known. By knowing the secondary and tertiary structures of peptides in the
gas phase, the types and relative abundance distributions of fragment/product ions,
observed in the mass spectrum, that undergo low-energy fragmentation processes
involving secondary interactions may be better explained.

Thus, the cationization of biomolecules with M+IDS-by-FAB or metal ions in the
FAB matrix have many analytical advantages in FAB such as molecular weight
confirmation in one and multiple component samples, biomolecule functional group
analysis, and complementary fragmentation to that obtained with protonation. The latter
advantage can aid in the structural elucidation of biomolecules, determination of metal
ion binding sites on the analyte, and determination of the analyte gas-phase conformation.

M+IDS is a good testing ground for metals used in M+IDS-by-FAB and it provides

243

information complementary to gas-phase ion/molecule studies using ICR techniques.
M+IDS, M+IDS-by-FAB and metal-ion containing salts may provide additional
fundamental information to help solve the mechanistic questions for ion formation and
dissociation in FAB. In the quest for obtaining binding site and conformational
information for biomolecules, perhaps a combination of multinuclear NMR studies in
90% water and computer molecular modeling studies would provide starting points for

extracting such information from the observed fragmentation patterns in the mass spectra.

APPENDICES

APPENDIX ONE

244

 

 

Injection of Reagent Ions into the
Selvedge Region in Fast-Atom
Bombardment Mass Spectrometry

Jason C. Rouse and John Allison
Department of Chemistry, Michigan State University, East Lansing, Michigan, USA

 

A divided probe that incorporates a potassium aluminosilicate glass target and an
analyte/glycerol matrix target, spatially separated, was used to inject potassium ions (K‘)
into the high-pressure ”selvedge“ region formed above the analyte/glycerol matrix target
during fast-atom bombardment (FAB); [M + K]* adduct ions that represent the types of
gas-phase neutral molecules present in the selvedge region are observed. Computer model-
ing assisted in designing the divided target and an additional ion optical element for the FAB
ion source to optimize interactions between K ’ ions and the desorbed neutral molecules. The
capability of injecting K" ions into the FAB experiment has utility in both mechanistic
studies and analyses. Experimental results here are consistent with a model for the desorp-
tion / ionization processes in FAB in which some types of neutral analyte molecules are
desorbed intact and are subsequently protonated by glycerol chemical ionization. Unstable
protonated molecules undergo unimolecular decomposition to yield observed fragment ions.
The use of K ‘ cationization of analytes for molecular weight confirmation is demonstrated,
as well as its utility in FAB experiments in which mixtures are encountered. (I Am Soc Mass

Spectrom 1993, 4, 259—269)

 

 

ast-atom bombardment (FAB) [1] is clearly a use-

ful process for generating gas-phase ions from

analytes that were not previously amenable to
mass spectrometric analysis. The processes leading to
the conversion of condensed-phase analyte to gas-phase
ions are generally believed to depend on whether the
analyteexistsinneutralorionicformintl'iematrile].
If, for example, a polar, nonionic analyte (M) is placed
in the matrix, glycerol (G), with a molar G/M ratio of
1000:1 or more, desorption and ionization are proposed
to occur sequentially [3]. The fast-atom beam, imping-
ing the sample on the surface of the target, generates
gas-phase G and M molecules above the sample sur-
face, in the so-called selvedge region. A small fraction
of the molecules may be ionized by subsequent fast-
atom /desorbed molecule collisions or desorbed di-
rectly in ionic form. Because most of the liquid target
is composed of G molecules, most of these ions are
related to G. Thus, essentially glycerol chemical ioniza-
tion (Cl) conditions are created. Initially formed ions
related to glycerol undergo many collisions with G
molecules to generate the [736 + H]’ ion series, as
well as leading to protonation of the analyte, forming
[M + H]‘. Thus, a gas-phase selvedge model explains

 

Address reprint requests to John Allison, Department of Chemistry.
Miclugan State University, East Laming, M] 68824.

C 1993 American Society for Mass Spectrometry
1044—0305/93/5600

the formation of protonated analyte molecules, and it
is consistent with the body of literature on proton
affinities [4]. If the analyte is a salt, such as a tetra-
alkyl ammonium halide, R4N‘X’, which ionizes in
glycerol, then fast atoms induce direct desorption of an
ion such as [R‘Nlfl No further chemistry is needed to
generate an ion representative of the analyte, although
adduct formation may occur as this ion passes through
the selvedge region, to form ions such as [R,N + 61‘.

The analysis of a FAB spectrum is similar to the
analysis of any Cl spectrum, and similar questions
must be addressed. Does the peak at highest mass-to-
charge ratio value (that is not related to the matrix)
represent the protonated analyte [M + H]‘ or the rad-
ical cation M", or are such forms of the analyte unsta-
ble? Does the peak at highest mass-to-charge ratio
value represent a fragment of the protonated molecule
such as the [(M + H) — H201‘ species commonly seen
for alcohols? Such questions become even more exten.
sive because biological analytes may contain salts, and
one may encounter adduct ions such as [M + Na]‘ as
well. Also, when salts are present in high concentra-
tions, ions such as [M - H + 2Na]‘ [5] may be ob-
served. The choice of matrix in using FAB is a second
complication that is obvious when one realizes that the
differences in proton affinities of analyte molecules
and matrix molecules may determine whether any ions

Received October 11, 1991
Revised October 12, 1992
Accepted October 13, 1992

245

260 ROUSE AND ALLlSON

representing the analyte will be formed. Third, there
are many complications when mixtures are (intention-
ally or unintentionally) analyzed by FAB because it is
a ”batch" method. The ratio of ion currents represent-
ing the two components may be a poor representation
of their relative concentrations in the liquid target. In
some cases, this may be due to the preferential proto-
nation of one component in the gas phase, owing to its
relatively high proton affinity. In other cases, the pref-
erential response for a compound may reflect the fact
that it is surface active and thus is desorbed to a
greater extent than other analytes in the matrix.

We demonstrate here that K’ ions can be injected
into the selvedge region of the FAB experiment and
that this capability will allow the neutral molecules
formed by particle bombardment to be identified. Such
a capability not only provides useful information on
the mechanism of FAB, but it also provides useful
analytical information. This work is an outgrowth of
the technique of K‘ ionization of desorbed species
(K ’ IDS) [6] that was developed in this laboratory. The
K‘IDS technique uses K‘ as the CI reagent ion for
identifying molecules desorbed after the rapid heating
of thermally labile analyte molecules. Potassium ions
are very useful for such work because they do not
induce fragmentation on interaction with organic
molecules [7], cannot react by charge transfer owing to
the very low ionization energy of potassium atoms.
and only form adduct ions. Thus, they provide a
straightforward probe of neutral molecules generated
in an ion source. Because we use K‘ here to investi-
gate neutral molecules desorbed by particle bombard-
ment in the FAB experiment, we refer to the technique
as K‘lDS-by-FAB [8].

This work is not designed to prove that the selvedge
region exists, nor are we attempting to define its di-
mensions. Desorbed neutral molecules and desorbed
ions are present in the ion source during a FAB experi-
ment, and we can use gas-phase chemistry to sample
the desorbed neutral molecules. Whether this corre
sponds to sampling the selvedge region or not de-
pends on individual views and definitions of the term
[9-12].

To inject ions into the selvedge region of a normal
FAB experiment, we investigated the use of a divided
FAB target. One-half of the target would contain the
typical glycerol/analyte mixture, and the other half
would contain a material that, when subjected to FAB,
would yield copious amounts of K‘ ions. Although
Munster et al. [13], Miller et al. [14], and Michaud et
al. [15] successfully demonstrated the use of divided
targets to explore gas-phase ion formation mechanisms
in FAB, the interactions of ions derived from one side
of their targets with neutral moleoiles derived from
the other side usually resulted in ion signals of much
lower intensity than those that could be obtained when
both analytes were mixed on a single target. We will
provide some insight into this observation in the fol-
lowing sections.

JAmSoc MassSpectmm 1991‘. 259-269

Experimental

Instrumental Parameters

All mass spectra were obtained on a JEOL FIX-110
double—focusing, forward geometry mass spectrometer
equipped with the JEOL FAB gun, the JEOL JMA-
DASOOO data system, and the combination JEOL field
desorption /FAB (FD/FAB) ion source. For all experi-
ments, the accelerating voltage was 10 keV; the resolu-
tion was 1000 (10% valley definition); and the scan rate
was from m/z 1-1500 per minute. The fast atoms used
were 6—keV xenon atoms. A divided target was con-
structed by highly modifying a JEOL FAB target, as is
described later. It was placed on the end of the JEOL
direct-insertion probe.

Computer Modeling of Ion Optics

The dimensions of the target and ion source optical
elements were modeled using the computer program
SIMION PC/PSZ Version 4.0 [16]. "Electrodes" were
scaled to fit the maximum potential array size of 16,000
points to ensure accuracy. The symmetry used was
planar, nonsymmetrical. The set of electrode voltages
used for each SIMION model was obtained by measur-
ing the actual potentials of the ion optical elements in
the JEOL FD/FAB ion source during a typical K ‘IDS-
by-FAB experiment. The models were refined until the
maximum voltage deviation in the array was 1 X 10“
V. Identical results were obtained with MacSIMION
2.0 [17], in which a more accurate, 330,000—point poten-
tial array was defined.

Sample Preparation

Stachyose, kassinin, cholic acid, digoxin, and
bradykinin (Sigma Chemical, St. Louis, MO); polyethy-
lene glycol 600, benzyltriethylammonium chloride, and
glycerol (Aldrich Chemical Co., Mikwaukee, WI);
thevetin (K & K Laboratories, Cleveland, OH); and
KNO3, A120,, and SiOz (Johnson Matthey Co., Ward
Hill, MA) were used without further purification. Ana-
lytes were dissolved either in methanol (J. T. Baker,
Phillipsburg, NJ) or a 1:1 mixture of methanol and
MilliQ water (Millipore, Bedford, MA). The sample
concentrations ranged from 2 to 10 ug/uL. Spectro-
scopically pure glycerol was coated on either the
K ’ IDS-by-FAB analyte / glycerol target half or the JEOL
FAB target, and 1 uL of sample solution was trans-
ferred to the respective target and mixed with the
glycerol. The FAB spectrum of each compound was
obtained with an alkali metalofree JEOL FAB target.

Results and Discussion

K ‘IDS-by-FAB Target Development

The experimental design for injecting K‘ ions into the
FAB experiment is conceptually very simple, but the

246

J Am Soc Mass Spectrom 1993, 4, 259—260

most straightforward divided target designs that were
investigated gave only limited success. Successful K’
ion injection in the FAB experiment for sampling de-
sorbed neutral molecules is highly dependent on both
the target geometry and the arrangement of, and po-
tentials applied to, the ion source optical elements,
because the region of highest pressure presumably
occupies a small region of space above the sample
target. With the aid of the ion optics computer model-
ing program, SIMION PC /PSZ, a functional K’IDS-
by-FAB target was developed. Our experiences and
insights are summarized here.

The design considerations of a divided target for
K’ lDS-by-FAB required that the FAB beam strike both
a solid surface that would yield K‘ iom and the
matrix/analyte target simultaneously, resulting in the
injection of desorbed K ’ ions into the gaseous selvedge
region above the matrix / analyte target. The [M + K]’
adduct ions formed by this process would then be
extracted and mass analyzed. A common way to gen-
erate gas-phase K * ions with FAB is to bombard a KCl
film. Unfortunately, FAB of KCl yields KnCl; cluster
cations in addition to K’, and the KCl film on the
target needs to be regenerated after each experiment.
One key component of the K’IDS-by-FAB experiment
is the generation of K’ ions. In K’lDS, a K’ alumi-
nosilicate glass emitter with the molar composition
KzozAl203:25i02, generates copious amounts of gas-
phase K‘ ions; surprisingly few neutral molecules are
emitted from such materials (5 10%) when heated
[18]. FAB of the same K’ glass on a target surface at
room temperature also yields continuous, copious
amounts of gas-phase K’ ions, with no evidence of
charging [19]. The FAB mass spectrum of the K' glass
(Figure 1) was obtained by exposing the K’ IDS-by-FAB
target, without a sample or glycerol present, to the
fast-atom beam. The dominant peak represents the
reagent ion K‘. The small Na‘ ion signal at m/z 23 in
Figure 1 arises from impurities in or on (or both) the
K‘ glass emitter and target metal surfaces. The peak at
m/z 52 represents Cr’ sputtered from bare stainless
steel surfaces. We believe that K’ ion generation via
the K‘ glass emitter is the optimal approach because
copious amounts of K’ ions are generated exclusively,

 

(“’030‘3— .("'.’.'

 

l
l
1
l
l

.‘a. .i. ..

on
Figure 1. FAB mass spectrum of the K‘lDS—by-FAB target
without matrix or analyte present.

REAGENT lONS [N THE SELVEDGE IN FAB MS 261

and a single K‘ glass target can be used for many
experiments.

The first K’lDS-by-FAB work used a simple wall-
divided target [20] that is shown in Figure 2a. We
concluded that this wall-divided target was unreliable,
regardless of wall height (0.1-1 mm). Most often, this
target resulted in little [M + K]‘ adduct ion forma-
tion, and either the K ’ IDSby-FAB spectra contained a
peak representing the K’ ion only or pealG represent-
ing the normal FAB matrix and analyte ions, depend-
ing on tuning parameters. In Figure 3, a representative
K‘ IDS-by-FAB mass spectrum for glycerol is shown in
which adduct ion formation occurred. Such a spectrum
is obtained when the K’ glass emitter has been con-
taminated by condensed-phase material. An identical
mass spectrum is obtained if glycerol is intentionally
placed on the K‘ glass emitter and subjected to FAB.
Apparently, the analyte/ matrix can be sputtered onto
the K' emitter by the fast-atom beam if the wall
separating the two targets is too low. Also, the ana-
lyte/matrix can "creep" around and over the wall
because the two target surfaces are physically con-
nected. The abundant [G + KJ’ adduct ion may be
due in part to the desorption of the [G + K]’ ion
directly from the surface of the K‘ glass emitter. The
formation of the [G - H + ZKl’ adduct ion is the
primary indicator that a condensed-phase interaction
between K’ glass and analyte/ matrix has occurred. If
the [G — H + ZKl‘ ion appears in the mass spectrum,
we cannot guarantee that the [G + K]’ adduct ion, as
well as any K’ adducts of analyte molecules observed,
are formed in the gas phase.

To decrease the possibility of cross-contamination,
as demonstrated by the wall-divided target, we de-
signed the divided target shown in Figure 2b. With
this “space-divided" target, the K’ emitter and sam-
ple support are separated in space, but are close enough
so that both targets can be bombarded simultaneously
by fast atoms, and desorbed K’ ions can interact with
the desorbed neutral molecules above the sample tar-
get. The sample stage on a JEOL FAB target was bent
to a 30° angle, and rhenium wire (0.007 in.) connects
the K’ glass emitter to the target post. Unfortunately,
this space-divided target has also performed inconsis-
tently in converting a substantial fraction of K’ ions
into [M + K]* and [G + K]+ adducts; however, it did

‘

.o
e“.
. ' .
a c ., A A
.\ \Ih“ .$ “N3

 

Figure 2. Schematics of (a) the K‘IDS-by-FAB wall-divrded
target and (b) the K’ lDS-by-FAB space-divided target.

 

247

262 ROL’SE .wo ALLISON

 

. 1 ilooxi' {
a I 1 1
It", '.
j .
t 1 [
I": a
3 ' l
n ' . 3'
' ..‘. IMO 1 mm. :
I , _ l m-II' :3
t J l
.' 204 1 IWJ’ j
t ' r I I i l .‘ 1
I ' j _ l ' “ .
a; I ‘ ' m ll L A ll. L A a L L 1 DA A l
O 1. - ‘ an

 

M
Figure 3. K'IDS-by-FAB mass spectrum of glycerol when the
K’ glass emrtter is contaminated wrth matrix/analyte.

solve the cross-contamination problem that plagued
the wall-divided target because [G + K]‘ adduct ions
could be formed without [C - H + 2KJ’ ion forma-
tion.

The ion optics computer modeling program SIMION
was used to gain insight into the ion trajectories from
K’lDS-by-FAB targets to improve performance. Rep-
resentative computational results are presented below.
The utility and limitations of the SIMION investiga-
tions should be clearly stated. "SIMION is a personal
computer program for designing and analyzing
charged particle lenses, ion transport systems, various
types of mass spectrometers and surface probes that
utilize charged particles” [21]. We are using it to gain
insights into the influence of ion source modifications
on the potential surface on which ions move within the
ion source. The results presented here should not be
interpreted literally, but are meant to assist in the
design and analysis of the experiment. Obviously, the
experimental goal is to generate ions from near a solid
surface, and have them"'sweep” through a region of
space close to the matrix surface. The ionic trajectories
are determined by the fields within the source. The
best way to appreciate the topology of the electric
fields is to follow some ion trajectories. This we do in
the figures presented. In these trajectory studies, ions
originate near a surface with an initial kinetic energy
of zero. Clearly this is not an accurate description of
the ions; however, ions with no initial kinetic energy
are most sensitive to the potential surface on which
they move. The other option would be to compute
thousands of trajectories for a variety of possible initial
kinetic energy vectors. Although such a rigorous ap-
proach may be informative, it must be realized that
any trajectories computed cannot be accurate because
SIMION will not allow ion-neutral molecule collisions
to be considered. The point of the experiment is to
inject ions into a high-pressure region of the ion source
where collisions occur. Thus, we use SIMION to sug-
gest physical possibilities. The SIMION results selected
to be shown here are relevant in that they correlate
well with what is observed. They correctly predict
which variations of the K‘IDS-by-FAB experiment
should work; they predict qualitative tuning variations
and sensitivity changes that are observed; and the

J Am Soc Mass Spectrom 1993. 4. 259-269

combination of SIMION and experiments led to an
experimental. design that was effective. Thus, it should
be realized that the SIMION results are shown to
present insights into what could happen, the shape of
the potential energy surface in the ion source, and how
it can be distorted to achieve the desired result—the
sampling of desorbed molecules by K’ attachment.
SIMION was first used to consider the optimization
of the spatial overlap between the K’ ions from the
glass emitter and the neutral molecules desorbed from
the analyte/matrix target. In Figure 4a, a SIMION
model of the normal FAB experiment that incorporates
an ordinary JEOL FAB target placed in the JEOL
FD / FAB ion source is shown. The FD / FAB ion source
electrodes and measured potentials are as follows: FAB
target (repeller), 10,050 V; cathode (L1), 10,010 V; first
focus (L2), 9000 V; second focus (L3), 600 V; deflector
(L4), 30 V; and total ion monitor plate (L5), 0 V. The
L3, L4, and L5 electrodes were included in the SIMION
models, but are not shown in the figures. The FAB
beam is also not shown, but it would originate from
above the figure and strike the target with an average
angle of incidence of 70°. The slits in L1 and L2 are 1
and 2 mm wide, respectively. Here, trajectories for
eleven 100-Da cations, with initial energies of 0 eV, are
shown starting at various locations near the FAB tar-
get. Of the 11 trajectories, only nine ions from the
target area are successfully extracted by the penetrat-
ing electric fields (shown by the contour intervals
labeled Cl-C4). These ions pass through both L1 and
L2, but the beam sharply diverges after 1.2. In Figure
4b, a SIMION model of the K’IDS-by-FAB wall-
divided target is shown. Twenty-three total trajectories
are shown; 11 ions originate on each half of the target.

 

Figure 4. SIMION models of (a) the JEOL FD/FAB target and
ion source; (b) the K’ lDS-by-FAB wall-divided target; and the
K’ lDSbyoFAB space-divided target showing the fate of (c) ana-
lyte/matrix ions and (d) of K’ IOX'LS. Contour lines representing
the extraction field for the potentials 10,040 V (C 1 ); 10,030 V (C2);
10,020 V (C3); 10,010 V (C4); 9500 V (C5); and 9000 V (C6) are
shown. Source elements and potentials are described in the text
for this lO—keV ion source.

248

J Am Soc Mass Spectrom 1993.4, 259-269

and one originates on the wall. Unfortunately, only the
ion that originates from the center of the target was
successfully extracted. Our experimental results with
the K‘IDS-by-FAB wall-divided target coincide with
the SIMION results shown here. The unreliable nature
of the wall-divided target is illustrated by the SIMION
model because the K‘ ions from the K' glass emitter
surface diverge away from the selvedge region above
the opposite target surface, preventing [M + KJ’
adduct formation. Because the contamination area of
this target, discussed previously, is surely centered
around the wall, the SIMION model suggests that ions
from this central region would be easily exnacted. It is
also clear from the SIMION model in Figure 4b why
different tuning parameters were required to optimize
extraction of analyte / glycerol ions, K‘ adduct ions
from the wall area, and K‘ ions in the FD/FAB ion
source. Three different sets of potentials were required
for the optimum sampling of ions that originate from
the three different regions on the target (left, center,
and right). Thus, the SIMION models agree well with
experimental observations.

Figure 4c and d show the SIMION model of the
space-divided K‘IDS-by-FAB target. Figure 4c shows
the trajectories of 19 cations of mass 1000 Da and with
0 eV of initial kinetic energy originating from various
locations near the stainless steel sample target. Twelve
of the 19 ions are successfully extracted from the target
area. These ions traverse both L1 and L2, but the beam
is poorly defined, which is consistent with observed
tuning problems. In Figure 4d, the trajectories are
shown for K’ ions originating from the K’ glass. All
but two trajectories terminate on the source element
L1; however, at least these K’ ions do follow favorable
trajectories, passing close to the presumed selvedge
region. Experimental results with this spacedivided
K‘IDS-by-FAB target are consistent with the K’ ions
trajectories, as shown in Figure 4d. The [M + K]’
adduct ion formation occurred more often with this
design than with the wall-divided target, but the [M +
K]' adduct ion signal was still usually weak in com-
parison to any FAB matrix/analyte ions. A potential
problem with this K’IDS-by-FAB target design is that
a typical K’ ion is promptly accelerated to kinetic
energies of 2 eV and greater in the presumed vicinity
of the selvedge region, resulting in ( 1) short residence
times and (2) kinetic energies too high for adduct
formation [22]. As a result, many of the desorbed
neutral molecules are not sampled by low-kinetic en-
ergy K ‘ ions for the set of ion optical element voltages
used in this SIMION calculation.

In an attempt to correct for the short residence time
and high kinetic energy of the K' ions, an electrode,
L0, with a center slit less than 1 mm wide and a
potential equal to the target potential, was placed
equidistant between the target electrode and the cath-
ode electrode, L1, in the SIMION model to limit the
extent of the extraction field penetration into the target
area. The combination of the space—divided target and

REAGENT IONS IN me SELVEDGE IN FAB MS 263

the electrode L0 resulted in the ability to perform the
K * IDS-bycFAB experiment as originally envisaged, and
the reasons are shown in Figure 5a—d. The SIMION
model in Figure 5a shows the normal FAB target, as
shown in Figure 4a, in the FD/FAB ion source that
incorporates the additional electrode, L0, placed be-
tween the target and L1. Again, 11 ion trajectories from
the FAB target surface are shown. The position and
dimensions of the LO electrode decrease the effects of
the extraction voltages, as shown by the positions of
the contour lines that correspond to the same poten-
tials as those shown in Figure 4a. The presence of L0 in
the SIMION model still results in the extraction of 9 of
the 11 ions, and it produces a well-defined ion beam
compared with the ion beam shown in Figure 4a.
Experimentally, the FAB spectrum is identical whether
[30 is present or not. Figure 5b shows the SIMION
model of the wall-divided K’IDS-by-FAB target cou-
pled with the additional electrode, L0. The presence of
DO in front of the wall-divided target dramatically
increases the number of successfully extracted ions
from 1 to 15 (7 ions from each target half) by decreas-
ing the penetration of the extraction field. This design
was never experimentally tested because SIMION re-
sults show that the K’ ion overlap with the desorbed
neutral molecules above the sample surface may still
be insufficient, and previous experimental results
showed that K+ glass contamination readily occurs.
In Figure 5c and d, the SIMION model of the
space-divided K‘IDS—by-FAB target coupled with the
additional electrode, L0, is shown. In Figure 5c, 18 of
19 matrix/analyte FAB ions from the target area are
accelerated toward the analyzer in a well-defined beam,
in contrast to the SIMION model shown in Figure 4c.
In Figure 5d, all K’ ions are injected into the pre-
sumed selvedge region above the sample target.

 

Figure 5. SIMION models with the additional ion optical ele-
ment L0 present (a) the JEOL FD/FAB target; (b) the K' IDS-by-
FAB wall-divided target; and the K’lDS-by-FAB space-divided
target, with trajectories shown for (c) analyte/ matrix ions and (d)
K' ions. Contour lines representing the potentials 10,040 V (C 1);
10.030 V (C2); 10,020 V (C3); 10,010 V (C4); 9500 V (C5); and 9000
V (C6) are also shown.

 

249

264 ROUSE AND ALLISON

SIMION trajectory calculations show that the L0 elec-
trode increases the residence time and lowers the ki-
netic energy of the K’ ions because the penetrating
extraction fields stop sharply at L0, and essentially a
field-free region is created between L0 and the target.
With the L0 ion optial element present, the typical K ’
ions gain 2 eV of kinetic energy, a realistic upper limit
for K’ adduct formation, in 4.5 as and are present for
approximately 1.9 us in the presumed area of the
"selvedge region." Without L0 installed (Figure 4c and
d), typical K’ trajectories achieve 2 eV in 0.45 [1.5 and
are present for approximately 0.3 as in the “selvedge
region." Thus, there is substantially more opportunity
for each K’ ion to react With a desorbed neutral
molecule with L0 installed. In addition, a high-pres-
sure region would probably be created in the small
volume between the target and L0 electrode that would
provide stabilization for termolecular ion-molecule
adduct formation, if required. Experimentally, there
were no visible traces of sample condensation on L0.
SIMION results clearly indicate that no preferential
tuning problems should occur with the additional L0
ion optical element installed because all K' and ma-
trix/analyte-related ions are focused into a well-
defined beam. Experimental results below confirm
this observation because [M + KJ’ adducts and
analyte / matrix ions appear in the same spectrum.

We incorporated the design shown in Figure 5c and
d into the K‘lDS—by-FAB experiment with successful
results. Figure 6b shows the K’IDS-by-FAB spectrum
of 3 glycerol sample obtained by using the space-di-

 

ii
1

 

 

 

 

 

 

 

a M. ‘
O , I
l
u" M‘
l .
V“. l
. l
I l
0:.-
. 3
2.- W I ,
: < ‘ ”’3
y = Ark A .L
0 I in i. - - ’
.I
I
“ T r
' M'
C
I
. IOi
I
I
'004
O
, . M.
n 00-3 W
; wr
: r l O o
, , 1.1 - - - - - we“! - ~20 .
" u m m sea as m
b '8

Figure 6. (a) FAB mass spectrum and (b) K‘IDSby-FAB mass
spectrum of glycerol (MW 92).

J Am Soc Mass Spectrom 1993. 4. 259-269

vided target with the ion optical element L0 mounted
in the FD/FAB source, showing that [G + K]' adduct
ions can be formed without [G - H 4» 2K]‘ ions being
formed. Thus, the analyte/liquid matrix is not being
sputtered onto the K’ glass. To confirm that these
[G + Kl’ ions are formed in the gas phase, one addi~
tional process needs to be considered. Although glyc-
erol does not contaminate the K’ glass, potassium
could be sputtered into the matrix, yielding the ob-
served adducts from condensed-phase, not gas-phase,
processes.

The following observations and experiments sug-
gest that potassium is not being sputtered from the
thermionic emitter onto the liquid target in this experi-
ment. First, we note that the K’IDS-by-FAB experi-
ment is difficult to perform because it is very sensitive
to the target design and positioning in relation to L0
and the ion source. If significant amounts of potassium
were sputtered from the glass to the liquid matrix and
this led to K’ adducts, the results should be much less
sensitive to experimental design. For example, the de-
signs shown in Figure 4b-d should yield [G + K]‘
and [M + K]‘ ions as well as those shown in Figure
Sb-d; this was not the case. Second, the emission
properties of alkali aluminosilicate glasses suggest that
the particle-induced desorption of K(g) need not occur
to a greater extent than for K'(g) owing to the nature
of these materials and their performance as thermionic
emitters. Third, if potassium is sputtered into the liq-
uid matrix in this experiment, and this is the process
by which K‘ adducts are formed, then a time depen-
dence should be observed. There should be no K’
adducts formed when the FAB experiment begins, and
the adduct ion signals should increase With time as
sputtered potassium accumulates in the matrix. This is
not observed. The K‘ adduct signals are present when
FAB is first initiated, and the adduct ion signals are
constant for a period of several minutes. Fourth, exper-
iments have been performed to test transfer of potas-
sium from the emitter to the sample target. For exam-
ple, if no glycerol is added, and the space-divided
target is bombarded for several minutes, potassium
may be sputtered onto the bare sample surface on
which glycerol is usually deposited. If the K' emitter
is then removed and the bare surface subjected to FAB
analysis, there is no evidence for the accumulation of
potassium on this surface. This is also true when
glycerol is placed on the second target. If the space-di-
vided target is bombarded for a period of time, the K’
emitter removed, and the remaining glycerol analyzed
by FAB, there is no evidence for detectable concentra-
tions of potassium in the glycerol. In some cases, a
very small peak that could represent [G + K]’ could
be observed, but it was barely detectable and very
much smaller than those observed in the K‘lDS-by-
FAB experiment. Also, experiments that involved the
addition of various concentrations of KCl to glycerol
demonstrated that if potassium was sputtered into the
glycerol and this process leads to K‘ adducts, then

 

250

J Am Soc MassSpectrom 1993. 4. 259-269

other ions should be present in these spectra, such as
[G — H + ZKI‘; theseare notobserved.

Probing the Selvedge Region of the FAB Experiment
with K ‘ Ions

Injection of K' ions into the selvedge region of the
FAB experiment will allow the desorbed neutral
molecules to be identified. That is, if K' ions are
injected into the FAB experiment and some [M + K]'
ion is formed, we will interpret this as evidence for the
presence of desorbed M molecules in the selvedge
region. We use this capability here to investigate the
fate of analyte molecules for a variety of analyte com—
pounds. Do they desorb as intact neutral molecules, or
does FAB lead to desorbed molecular fragments as
well?

Glycerol. The ions formed by FAB of glycerol are
shown in Figure 6a. It is well known that a sequence of
pale representing protonated glycerol clusters, [nC
+ H]’,dominatesthespectrumandbeginswith n -1
at m/z 93. In addition, two prominent ”fragment
iom" are represented by peaks at m/z 75 and 57,
which are 18 and 36 Da below m /z 93, respectively.

Consider the ion at m/z 75. There are a variety of
mechanisms by which this ion could be formed. The
ion may be formed as a fragment ion of G", or it may
be a fragment formed from [G + H]’. Because it is an
even-electron ion, it may well be a fragment formed
following protonation and should be written as [(C +
H) - H2015 Another possibility is that FAB imparts
sufficient energy to the sample to induce dehydration.
Thus, FAB could lead to the desorption of G, [G -
H20], and [G - 2(H,O)] into a gaseous, protonating
environment, leading to the ionic species [C + H]‘,
[(G - H20) + H]', and [(G - 2(H20)) + H]’ at m/z
93, 75, and 57, respectively.

When K’ ionsare injected into theselvedge region
above the glycerol target during FAB, an ion series
corresponding to [nG + K]’ is observed that parallels
the [nG + H]’ series (Figure 6b). This indicates that
K’ ions are successfully being injected into a high-
pressure region and are undergoing multiple colli-
sions. This process is also indicated by the ratios of the
peaks representing K’ and its glycerol adducts; there
is fairly extensive conversion of this ”CI reagent ion"
into product ions, which can occur in a high-pressure
environment The relative intensities of both the [nG
+ K]’ series and the [nG + H]‘ series correlate with
each other and are a function of the amount of glycerol
present. The members of both series with small values
of n always dominate, and the low relative intensities
ofthehighermassionsineachsenessuggestthatthey
are formed by sequential clustering processes in a
high-pressure environment [10, 23]. That is, it appears
that FAB induces the desorption of single glycerol

REAGENT IONS [N THE SELVEDGE IN FAB MS 265

molecules and not glycerol clusters. More important,
the [nG+K]‘ ions are the only new ions formed
whenK’ ionsareinjected.Therearenopeaksob—
served corresponding to [(G - nH20) + K]+ ions,
suggesting that the fast-atom beam desorbs glycerol
molecules, predominantly intact, and does not lead to
desorbed degradation products. This is somewhat un-
expected because elimination of water from alcohols is
a very low-energy process, compared with the other
processes that occur, such as desorption (vaporization).
If [G - (H,O)l molecules were desorbed, we would
expect K" adducts to be formed. Although the K’
affinities for [G‘ - (H20)], a diol, and [C - 2(H20)], an
alcohol, would be lower than that for glycerol, a triol.
K+ has been observed to readily form adducts with
small molecules such as alcohols. Thus, we suggest
that the ions observed at m/z 75 and 57 should be
considered fragment ions of [G + H]‘; the elimination
of water occurs following protonation.

Cholic aa'd. Cholic acid [molecular weight (MW) 408],
a bile acid, can be considered an example in which
positive-ion FAB mass spectrometry ”fails." The FAB
mass spectrum of cholic acid in glycerol is shown in
Figure 7a. There is a very small [M + H]’ signal at
m/z 409. The dominant ions are at m/z 355 and 373.
These fragment ions are even-electron ions and corre-
spond to a protonated molecule in which multiple
water molecules have been lost. Again, there are two
extreme situations that could explain the observed

 

 

 

C”..."I- .C-‘.'..

 

 

 

 

 

 

: “3' 1
' as
l
I
t
j W... I
. |
' l
7. M’ .
a “7 1
: mm ...... l
I 4 .anL" n f m-

are ads as. see

b WI

Figure 7. (a) FAB mass spectrum and (b) K’IDSby-FAB mass
spectrum of cholic acid (MW 4%) in glycerol.

 

251

266 ROUSE AND ALLISON

spectrum. First, cholic acid molecules could be de-
sorbed intact by FAB and subsequently protonated in
the gas phase. Because the protonated molecule is
unstable, the [M + H]‘ ions essentially completely
dissociate by losing one, two, and three water
molecules. This would not be unexpected for a
molecule containing multiple — OH groups under Cl
conditions. Also, not unexpected, would be a situation
in which the polar molecule simply is not desorbed
intact on FAB, but instead the dominant analyte-re-
lated molecules that desorb are the less polar, dehy-
drated forms [M - 3H20] (MW 354) and [M — ZHZO]
(MW 372), which then become protonated to yield the
ion currents at m/z 355 and 373. Figure 7b shows the
corresponding K ’ IDS-by-FAB mass spectrum for cholic
acid. When K’ ions are injected into the selvedge
region, only one additional peak (not including those
related to glycerol) appears, at m/z 447, which is the
K' adduct of the intact cholic acid molecule. There is
no peak observed at m /z 393, which would represent
[(M - 3(H20) + Kl’. Thus, the K‘lDS-by-FAB results
suggest that FAB desorbs intact cholic acid molecules
that dissociate completely on protonation. Although
the protonated molecule is unstable, the neutral ana-
lyte molecules remain intact throughout the desorption
process. The results shown in Figure 7b also demon-
strate that K’ adducts of molecules are stable [24]; K‘
forms adducts without inducing fragmentation and is
thus ideal for such experiments. When a cholic
acid /KC1 (low concentration)/ glycerol mixture is sub-
jected toFAB,an[M - H + ZKI‘ ionisformed,but in
K‘IDS—by-FAB of cholic acid, this ion is not formed
(see Figure 7b).

Digoxin. As the size of the analyte molecules in-
creases. the possibility for fragmentation accompany-
ing neutral desorption increases as well. The FAB
spectrum of the cardiac glycoside, digoxin (MW 780),
is shown in Figure 8a. An intense peak at m/z 781
representing the protonated molecule is observed, as is
a rich array of fragment ions. Do all fragment ions
result from the unimolecular fragmentation of the pro-
tonated molecule, or does the molecule decompose on
fast-atom impact to yield an array of smaller, even-
mass neutral molecules, each of which become proto-
nated following dissociation? For example, the peak at
m/z 651 represents the protonated form of digoxin
([(AO$,OS,OS,OH) + H]‘) minus the terminal sugar,
53, due to cleavage of a glycosidic bond, written as
[(AOSjOSZOH) + H]+ using the Light-Kassel-Allison
notation [25]. The aglycone is represented by A, and
the sugars are designated as 5,. Does FAB of the
digoxin / glycerol target lead to desorption of
[AOS,OS,OH] molecules (MW 650) mat become proto-
nated following desorption? Figure 8b shows the
K’lDS-by-FAB mass spectrum for this analyte. Again,
on injection of K’ ions into the FAB experiment, only
one new analyte-related peak is formed, seen at rn/z
819, representing the K’ adduct of the intact digoxin

J Am Soc Mass Spectrom 1993. 4. 259-269

 

 

1‘-.,.‘3- .C-.-.'

 

U

8
V
8
O
O
0

 

‘--...‘8— .Q—-.-..

 

 

 

Figure 8. (a) FAB mas spectrum and (b) K’lDS-by-FAB mass
spectrum of digoxin (MW 780) in glycerol.

molecule. Thus, we conclude that FAB results in the
intact desorption of polar digoxin molecules from the
glycerol matrix; the analyte is protonated following
desorption. The lower mass ions can be interpreted as
true fragment ions of the protonated molecule.

Benzyltriethylammoniurn chloride. When this salt (MW
227) is analyzed by FAB, using glycerol as a matrix,
the resulting spectrum is dominated by the benzyltri-
ethylammonium cation [CanNl’ at m/z 192. Frag-
ments of this cation are also observed. We assume that
this "preformed" cation desorbs directly as an ion,
although one could certainly foresee, through a rear-
rangement reaction, the formation and subsequent de-
sorption of a neutral species, [CanN], which would
then be protonated to yield a peak at m/z 192. An-
other possibility is the intact salt desorbs and becomes
protonated, as suggested by eq 1:

[c,,HnNrcng) + [c + H]' i
[acufian’CI—i + H] —~
[CanNl’ + H0 (1)
When K’IDS—by-FAB is used to probe the selvedge
region in this experiment, no new ions representing
the analyte are formed. Thus, direct desorption of

preformed ions does appear to be the correct mecha-
nism for such molecules.

 

252

I Am Soc Mass Spectrom 1993. l. 259- 209

Peptides. The behavior of peptides in the K’lDS—by-
FAB experiment varies, which suggests that the mech-
anisms for ion formation should depend on the specific
peptide under study. When the peptide kassinin (MW
1334) is analyzed by FAB using glycerol as the matrix,
an abundant protonated molecule is formed. A portion
of the K ’ lDS-by-FAB spectrum of kassinin is shown in
Figure 9. The appearance of the additional peak at
m/: 1373, 38 Da above the protonated molecule. repre-
sents the potassium adduct of the intact, desorbed
peptide. Even for an analyte of this size, FAB appears
to induce the desorption of the analyte intact.

Bradykinin (MW 1059) also forms an abundant pro-
tonated molecule at m/: 1060; however, when the
K‘lDS-by-FAB experiment is performed, and K’ ions
are injected into the selvedge region, no new ions are
formed representing the analyte. Apparently,
bradykinin exists in glycerol in protonated form and
desorbs directly as an ion. This may be due to the very
basic arginine residues in the peptide. Thus, even in a
simple glycerol matrix, without addition of acid or
base, peptides may behave in a variety of ways, de-
pending on the pK,s of the various basic sites in the
molecule.

Analytical Applications of K +lDS-lry—l-‘AB

The obvious analytical utility of K‘IDS-by—FAB, in the
context of FAB analyses, is the molecular weight con-
firmation for analytes that are not preionized in the
matrix. Such capabilities are demonstrated in the cholic
acid example in Figure 7a and b. If this were an
unknown, the 18-Da difference between m/z 355 and
373 would certainly be recognized as a water 1055:
however, one would probably not be able to distin-
guish the peals at m/z 391 and 409 above the back-
ground. Thus, one may assume that the molecular
weight of the analyte was 372. In addition to those
cases in which analytes yield very small peaks in the
molecular weight region of the spectrum, K‘IDS—by-
FAB is also useful in this regard when the mass-to-
charge ratio of the protonated molecule coincides with
that of a matrix-related ion. Similar cationization stud-
ies can be done by adding KCl or KI to the matrix,

 

 

 

a marl
: egg. 2: “WWW
. .
I r
I l‘
'
s g}
| 3: fl“?
0 1373
I I ,
s
H
s
l
i
' Q
1120 1’” 1m 1.“ "C. 1". ‘3“

'8
Figure 9. K‘ lDS-by-FAB mass spectrum of kassinin (MW 1334)
in glycerol.

REAGENT IONS [N THE SELVEDGE [N FAB MS 267

although this can lead to other ions in addition to an
[M + K]‘ ion.

A common occurrence that can ”mask" molecular
weight information in FAB is the formation of ana-
lyte/matrix cluster ions. For example, Figure 8a and b
shows a peak at m/z 873, which is the proton-bound
adduct ion of digoxin and glycerol. Usually, the 92-Da
difference between the peaks at m/z 781 and 873
would be recognized as an adduct ion containing a
single-matrix molecule, although the peak at m/z 873
could represent either the protonated analyte or a
second component with a molecular weight 92 greater
than that of digoxin. Again, the K’ lDS-by-FAB results
clearly indicate the molecular weight of the compound
and assist in identifying the high-mass ions that are
candidates for pseudomolecular ions.

Another typical situation arose in obtaining the FAB
mass spectrum of stachyose (MW 666), an oligosaccha-
ride. The peak representing the protonated molecule is
discernible but of low intensity, comparable to the
intensities of nearby matrix-related ions in the mass
spectrum. Actually, in some experiments with this
analyte, no protonated parent was observed at all. Its
fonnation was dependent on the glycerol/stachyose
ratio used. Also, sodium salt impurities in the sample
led to an [M + Na]‘ peak 22 Da above the protonated
molecule. Figure 10 shows the K’IDS-by-FAB mass
spectrum of stachyose, which clearly allows the molec-
ular weight to be verified and the assignments for the
peaks at m/z 667 and 689 to be confirmed.

K’IDS-by-FAB also provides useful information
when mixtures are the subject of a FAB analysis. The
use of FAB mass spectrometry for the analysis of
mixtures always presents particular challenges, espe-
cially if no chromatographic information is available
for the sample. A mixture that has received much
attention in the desorption/ionization literature is that
encountered in polyethylene and polyproplyene gly-
cols. For example, polyethylene glycol (PEG) 600 is a
mixture of oligomers with the formula
[H(OCH2CH2),,OH]. The distribution of oligomers is
such that the average molecular weight is approxi-
mately 600. One can obtain information on the oligomer
distribution by FAB. If a small amount of PEG is
added to a glycerol matrix, a series of protonated

 

‘-

... ' ”Qfifl W2 "f"
90...!

'1'
M'cs”rs°sl° ‘
,. ... .... I
M. W .
as: a Instr
M M m ass m’
or W 7:?

Jill ...L 4.... ALL

*

 

 

<-I:O-a— o¢—-I-OI

s
'0' us an us rss rss

Figure 10. K’ley-FAB mass spectrum of stachyose (MW
666) in glycerol.

253

268 ROUSEAN'DALLISON

oligomers appears in the resulting FAB mass spec-
trum. Figure 11 shows the K’lDS—by-FAB spectrum of
PEG 600. The K’ adducts of each oligomer clearly
stand out and dominate over all other ions formed.
Again, as in the single-component analyses, FAB ap-
pears to induce the desorption of all of the oligomers
present in the mixture intact.

In our work to understand the fragmentation of
cardiac glycosides [26], we obtained the FAB mass
spectrum of thevetin (MW 858) that is shown in Figure
12a. As expected, a peak representing the protonated
molecule was present at rn/z 859; unexpected was a
second peak 14 Dahigherat m/z 873. It israretofind
high-mass ions separated by 14 Da because the loss of
CH, is rarely observed in unimolecular dissociatiorts.
A more reasonable explanation was that a mixture of
two compounds was present with MWs 858 and 872 in
approximately a 3:1 ratio, respectively. An investiga-
tion revealed that there are two forms of thevetin,
thevetin A and B, whose structures, as indicated on
Figure 12a, differ by 14 Da. Apparently these two
compounds are difficult to separate. The K’lDS-by-
FAB mass spectrum of thevetin (Figure 12b) confirms
thepresenceoftwocompoundsthatdifferinmassby
14 Da because two additional ions are formed on
injection of K’ ions. If one looks more closely at the
spectrum in Figure 12a, the relative abundances of the
isotopic clusters for protonated thevetin A and B are
very different, particularly at m/z 875. In fact, this
same isotopic distribution also appears in the K‘IDS-
by-FAB spectrum in Figure 12b, suggesting that the
sample is at least a three-component mixture. The
third compound may be a reduced variant of thevetin
A

Last, it should be noted that the K’ adduct of the
intact analyte can be selected for collision-activated
dissociation (CAD) or metastable analysis in the
K‘IDS—by-FAB experiment. The product ions of the
K‘lDS—by-FAB precursor [M + KY adduct ions,
forrnedinthegasphase,arethesameasthoseob-
served for the [M + K]+ ions formed when K1 is added
to the peptide/ glycerol matrix. The utility of CAD for
alkali cationized molecules in providing structural in-
formation has been demonstrated [27] and appears to
be a method that can provide complementary informa-

 

"I
~tl eat. .73

 

 

'1‘-.30‘3- a¢--o.

'0' ass sss us no ass
-
Figure 11. K’lDS—by-FAB mass spectrum of polyethylene gly-
col 600 in glycerol.

lArnSoc MassSpectrom I993. $259469

 

 

 

 

 

 

 

 

 

I M. Yin-sass heron

7 ~ ran-ass than,

t m o—or.

i

3 qumQ

' O. on 0

I 4 MM. W

' "I

e

'0‘ J l ” ms

‘ - l H "‘

V .1!” 1. ll: .11 . . l.. . . i. ”+1. I

I“ “0 .0. II. no

‘ m

: i‘MMI’

l ”7

s

I

I

0 maoxr

a l

l l

s

2 {l

l

t .

Y I ii. ““111“ l
“I .2. us

 

Figuelz. (alFABmassspectr-um and (b) K‘lDS—by-FAB mass
spectrumofthevetinA(MW872)andthevetinB(MW858)m
glycerol.

tion to that available from CAD studies of protonated
molecules.

Conclusions

We have demonstrated that K’ ions can be success-
fully injected into the selvedge region of the FAB
experiment to probe the desorbed neutral species pres-
ent. The K’IDS-by-FAB technique is both a mechanis-
tic and an analytical tool. Mechanistically, the desorp-
tion aspects of FAB, for both analyte and matrix
molecules, can be probed. Analytically, K’ cationiza-
tion of analytes is useful for molecular weight confir-
mation.

One assumption central to the K’IDS-by-FAB data
discussed here is that K’ will form adduct ions with a
variety of neutral molecules in the FAB ion source and
that there are no dramatic differences in adduct forma-
tion rates for the molecules under consideration Cer-
tainly, the details of this assumption remain to be
explored. For example, in the experiments with glyc-
erol, K‘ forms a complex with glycerol but not with
dehydrated glycerol, and this is interpreted as suggest-
ing that desorbed, dehydrated glycerol is not a major
product of particle bombardment. This assumes that
K’ will form an adduct with [G — H20]. At this point,
we make two observations. First, K’ and other alkali
ions form adducts with monofunctional molecules such
as alcohols and small molecules such as propene, so
that multiple functional groups are not required [7,

254

I Am Soc Mass Spectrom 1993.4. 259-269

28]. Second, polar groups need not be present in a
molecule for a substantial binding energy to an alkali
ion. Because the interaction is electrostatic, both the
dipole moment and the polarizability must be consid-
ered. ln field desorption /surface ionization experi-
ments, alkali ion adducts with saturated alkanes have
been observed [29], showing that such species can be
stable, even when there is no dipole moment, if the
size and structure of the molecule result in a substan-
tial overall polarizability. Obviously, much work re-
mains to quantitate this aspect of the experiment.

Finally, we note that for the compounds studied to
date, it is intriguing that FAB provides a source of
gas-phase molecules with molecular weights above
1000 Da. This capability may eventually find applica-
tions in fields other than mass spectrometry. For exam-
ple FAB may prove to be a useful tool for the spectro-
scopic characterization of large molecules in the gas
phase in the absence of solvent.

Acknowledgments

This work was supported by the Biotechnology Research Pro-
gram of the Divrsron of Research Resources of the NIH (RR-
00480-20), the Michigan State University Research Excellence
Fund. and the Dow Chemical Company Analytical Sciences
Group. We are grateful to Drs. Karen Light-Wahl, Dan Kassel.
and Cary Schultz for their contributions to this project.

References

1. Barber. M.; Bordoli, R 5.; Sedgwrck, D.; Tyler, A. N. I. Chem.
Soc. Chem. Commun. 1981, 325.

2. FerLselau, C.; Cotter, R I. Chem. Rev. 1”7, 87, 501.

3. Cooks, R. (3.; Busch. K. L. lnt. I. Mass Spectrom. lon Phys.
1983. S3, 111.

4. Sunner, I. A.; Kulatunga, R; Kebarle. P. Anal. Chem. was, 58,
1312.

5. Mallis. L. M.; Russell, D. H. hit. I. Mass Spectrom. lon
Processes 1987. 78, 147.

6. Bombick, D. D.; Allison, I. Anal. Chem. 1987, 59, 458.

18.
19.

21.

REAGENT IONS N THE SELVEDGE iN FAB MS 269

. Allison, I.; Ridge, D. P. I. Am. Chem. Soc. 1979, 101, 4998.
. Rouse, I. C.; Light, K. 1.; Allison, 1. Proceedings of the 38th

ASMS Conference on Mass Spectrometry and Allied Topics; Tuc-
son, AZ, lune 1990; p. 455.
Pachuta, S. I.; Cools, R. G. Chem. Rev. 1987, 87, 647.

. Sunner, I.; Morales, A.; Kebarle, P. Anal. Chem. 1987, 59.

1378.

. Schroder, E.; Munster, 11.; Budzikiewrcz, H. Org. Mass Spec-

trom. 1986, 21, 707.

. Freas, R. 8.; Ross, M. M.; Campana. I. E. I. Am. Chem. Soc.

IQS, 107, 6195.

. Milnster, H; Theobald. F.; Budzikiewrcz, H.; Schrt'rder, E. Int

I. Mass Spectrom. Ion Processes 1987, 79, 73.

. Miller, I. M.; Balasanmugam, K.; Fulcher. A. Org. Mass

Spectrom. mo, 24, 497.

. Michaud, D. P.; Kyranos, I. N.; Brennan. T. F.; Vouros. P.

Anal. Chem. 1990, 62, 1069.

. Dahl. D. A.; Delmore, I. E. SIMION PC/PSZ Version 4.0.

EGG-CS-7233 Revisron 2, April 1988.

. McGilvery, D.; Morrison, R. MaélMlON Version 2.0, Octo-

ber 1991.

Blewett, I. P.; Iones, E. I. Phys. Rev. 1936. 50. 464.
Ackermann, 8.; Allison, I.; Musselman, 8.; Pinkston. D;
Bombick, D.; Dolnikowski. 6.; Tsarbopoulos, A.; Watson, I.
T. Proceedings of the 315i ASMS Conference on Mass Spectrome-
try and Allied Topics; Boston, MA, May 1983; p. 600.

. Kassel, D. B. Ph.D. dissertation, Michigan State University.
1988.
Dahl, D. A.; Delmore, I. E.; Appelhans, A. D. Rev. Scr.
lnstnrm. 1990, 61, 607.

. Bombick, D. D. Ph.D. dissertation, Michigan State Universrty,
1986.

. Honda. F.; Lancaster, C. M.; Fukuda, Y.; Rabalais, I. W. I.
Chem. Phys. 1978, 69. 4931.

. Davidson, W. R.; Kebarle. P. I. Am. Chem. Soc. 1976, 98.
6133.

. Light, K. I.; Kassel, D. 8.; Allison, I. Biomed. Environ. Mass
Spectrom. 1909, 18, 177.

. Light, K I., Allison, I. I. Am. Soc. Mass Spectrom. 1990, 1,
455.

. Teesch, L. M.; Orlando, R. C.; Adams, I. I. Am. Chem. Soc.
1991, 113, 3668.

. Weddle, G. H.; Allison, 1.; Ridge, D. P. I. Am. Chem. Soc.
1977, 99, 105.

. Borchers, F.; Giessmann, U; Rollgen, F. W. Org. Mass Spec-

trom. 1977, 12, 539.

APPENDIX TWO

APPENDIX TWO. THERMOCHEMICAL DATA FOR ESTIMATION
OF REACTION ENTHALPIES FOR PEPTIDE SKELETAL BOND
CLEAVAGES

In Chapter four, the reaction enthalpies (Aern) were estimated for peptide
fragmentation mechanisms by matching the peptide protonation site, the bond cleavage
site(s) and the bond formation site(s) on the [M+H]+ ion, fragment ions and fragment
neutrals with small neutral molecules, radicals and ions with nearly identical chemical
structures and available heats of formation (AHf), proton affinities (PA), and ionization
energies (IE) [40, 191, 210]. The AH,“ and AHf values are listed for each species below
in kcal/mol. In Equation A1, a AHm, of 87 kcal/mol was estimated for the homolytic C(1-
C(0) bond cleavage in Scheme XI using protonated glycine as the model. The AH,“ of y,,
ions formed by the 1, 2-climination mechanism in Scheme VI was estimated to be 38
kcal/mol using N-methyl-acetamidc in Equation A2. The formation of b,, ions in Scheme I
by an inductive cleavage of the amide bond in N-methyl-acetamidc requires a AHm, of
53.5 kcal/mol as in Equation A3. The formation of (11 ions from b; ions in Scheme 11 was
estimated to be -15 kcal/mol exothermic in Equation A6. However, the AHf of
NH2CH2C(O)® in Equation A6 is not available and it was estimated to be 166.3 kcal/mol
from Equations A4 and A5 assuming the AH,“ for homolytic cleavage of the C(O)-O bond
in acetic acid and the ionization of CH3C(O)° in Equation A4 is equal to the identical
processes using glycine in Equation A5. The difference in reaction enthalpies for a l and a,,
ions was estimated in Equations A7 and A8 using 1133 of amines and amides which are 204
and 214 kcal/mol, respectively. In Equation A8, the amide bond carbonyl in a,, ions (see

Scheme 1) delocalizes the electron density of the neighboring amide nitrogen which results

255

256

in the amide N lone pair providing less stabilization for the ionized an structure. In
contrast, the amine N in a1 ions as shown in Scheme II and Equation A7 can provide full
charge stabilization because the lone pair is localized. As a result, the IE for an ions is 10
kcanol higher, owing to electron delocalization, in comparison to that for a1 ions, so the
AH,“ for forming a,, ions with no side chain from b,, ions is -5 kcal/mol, 10 kcal/mol more
than a1 ions. Assuming a hydrogen is shifted from a distant amide nitrogen through a
secondary interaction, the estimation of the overall AH,“ for y,, ions formed is a four-step
process and modeling requires four equations: the protonated amide bond is homolytically
cleaved in Equation A9, a distant hydrogen is cleaved from an amide represented by a
secondary nitrogen in Equation A10, the cyclic neutral product is formed in Equation A1 1,
and the shifted hydrogen is attached to the ionic homolytic product from amide bond
cleavage to form a y,, ion in Equation A12. The overall AH,“ is -5 kcal/mol which was

obtained by summing the individual AHm in Equations A9-A12.

O O
NH3CH2C(O)OH -—) NH3CH20 + 0C(O)OI-I Aern = 86.7 (A1)
61 201 -53.3

O O
CH3C(O)NH2CH3 -—) CH2 = C = O + NH3CH3 Aern = 37.6 (A2)
97 -11.4 146

O
CH3C(O)NH2CH3 —> CH3C(O)O + NHzCH3 AHrmz 53.5 (A3)
97

156 -5.5
CH3C(O)OH —) CH3C(O)O + OOH + 6' AH“: 268.6 (A4)
-103.3 156 9.3
NHzCH2C(O)OH —> NH2CH2C(O)O + 00H + e' AHnm = 268.6 (A5)
-93 X=166.3 9.3
O
NHzCH2C(O)O —> NH2=CH2 + C E O AHnm = - 15 (A6)
166.3 (ESL) 178 ~26.4
$0
CH3CH2NH2 -—> CH3CH2NH2 + 6‘ AH,“ = 204.3 (A7)
-1 1.3 193
$0
CH3C(O)NHCH3 —> CH3C(O)NHCH3 + 6' Aern = 214 (A8)
-56 158

257

O O
CH3C(O)NH2CH3 —> CH3C(O)0 + oNHzCH3 AHm = 98 (A9)
97 -6

201
NH(CH3)2 —) H- + oN(CH3)2 Aern = 91.5 (A10)
-4.4 52.1 35

CH3C(O)o + oN(CH3)2 -) CH3C(O)N(CH3)2 AHm=-85 (A11)
-6 35 -56

O O
H0 + 0NH2CH3 —-> NH3CH3 AB,“ = -107 (A12)
52.1 201 146

In Equations Al-A12, examples of each estimation method for calculating the
AHms are demonstrated. Below, the estimated AHms below utilize the above methods to
estimate the remaining ions formed from homolytic cleavages, 1, 2-elimination cleavages,
inductive cleavages, and secondary interactions in Schemes I-XIV. In Equations A9 and
A13, the AHms of the homolytic amide and N-Ca bond cleavages are estimated to be 98
and 103 kcal/mol, respectively. In Equations A14 and A15, the estimated AHms for the
formation of Jr", and 01 ions by 1, 2—e1imination cleavages in Schemes V and III are 35 and
26.5 kcal/mol, respectively. The AH,“ of a,, ions formed by the 1, 2-elimination
mechanism in Scheme [H was estimated to be 42 kcal/mol, but the AHf of ionized
(CH3C(O)NH=CH2)O in Equation A18 is not known. It was estimated to be 137 kcal/mol
from Equations A16 and A17 assuming the AHm. for removal of hydrogen in methyl ethyl
amine in Equation A16 is equal to the identical processes using ionized N-methyl acetamide
in Equation A17. In Equations A19 and A20, the AH,“ of an a1 ion with a side chain
from a b1 ion was estimated to be -30 kcal/mol using the identical estimation method as for
a1 ions without a side chain shown in Equations A4-A6. In Equations A21 and A22, the
AH,“ of an an ion without a side chain from a b,, ion was estimated to be -5 kcal/mol using
the same method for (11 ions above and Equations A4, A17 and A18. In Equation A23, the
AHm, for conversion of 0,, ions to 1),, ions by amide bond inductive cleavage in Scheme

XII is estimated to be 51 kcal/mol. In Equations A24-A27, the overall AH”m for the

258

[174+OH+H]+ rearrangement ion shown in Scheme XIV was estimated to be 23 kcal/mol by
employing a similar estimation method as used for y,, ions in Equations A9—Al2, except an
00H is shifted instead of H0. In Equations A13, A10, A28, and A29, the overall AHm, of
c,, was estimated to be 5 kcal/mol assuming hydrogen was shifted from an amide nitrogen
as shown in Scheme IX using the same method as for y,, ions in Equations A9-A12. If a
hydrogen is shifted from the carboxylic acid for c,, ion formation, Equations A10 and A28
were substituted with Equations A30 and A31 to give an overall AHm of 10 kcal/mol. In
Equation A32, the AH,“ for conversion of, c,, ions to b,, ions by an inductive cleavage in
Scheme XIII is estimated to be 42 kcal/mol. For y,, ions, if a hydrogen shifts from the
amine nitrogen instead of the amide nitrogen, Equations A10 and All were substituted
with A33 and A26, respectively, to estimate the AH,“ which is 0 chmol. For cyclo-bn
ion formation in Scheme VIII, if a hydrogen was shifted from an interacting protonated
amine nitrogen (not shown), then the AH,“ was estimated to be 0 kcal/mol simply by
negating the AH,“ values in Equations A9, A33, A26, and A12, or if a hydrogen was
shifted from an interacting protonated amide nitrogen, then Equations A9-A12 were
negated to calculate the AH,“ which is -5 kcal/mol. If the skeletal amide nitrogen is
protonated as shown in Scheme VIII for ion cyclo-bn formation, a hydrogen shift is not
required since the amide skeletal bond is cleaved and a new, similar amide bond is formed
with the interacting amine or amide nitrogen. The AHm, for the latter cyclo-bn mechanism
is identical to the former regardless if a hydrogen shifts or not and it can be estimated as
follows. If the protonated amide skeletal bond is cleaved as in Equation A9, the positive
charge can be moved from the protonated amide nitrogen fragment to the cyclic neutral
product in All or A26, depending if an amide or amine N interacts, respectively, providing
(1) the amide nitrogen fragment neutralized by retuming its ionization energy, (2) the amide
or amine hydrogen is removed from the interacting nitrogen as in Equations A10 or A33.
respectively, and (3) this removed hydrogen is ionized and used to protonate the cyclic

neutral product. For 2,; ion formation by N-Ca inductive bond cleavage in Scheme IV, the

259

AHm, was estimated to be 79 kcal/mol in Equation A34 if a side chain was not present on
the a-carbon and 62 kcal/mol in Equation A35 if a side chain was present The overall
AH,“ of cyclo-zn ions was estimated using the same methodology as for cyclo-bn ions
using Equation A13 and the Aerns calculated for c,, ions in Equations A28 or A31, and
A10 or A30 depending if an amide nitrogen or the carboxylic acid oxygen interacted. The
overall AHnns of internal ions and immonium ions were estimated to be 0 kcal/mol
identical to cyclo-bn ions assuming a protonated amine nitrogen interacted and protonated
an amide skeletal bond. For internal immonium ion formation, the loss of CO from internal

ions can be estimated using the AHrms of 0,, ions above.

O O
CH3C(O)NH2CH3 —-) CH2C(O)NH20 + 0CH3 AHm = 102.8 (A13)
97 165 34.8
O O
CH3C(O)NH2CH3 —) CH4 + O=C=NHCH3 AHm, = 35.2 (A14)
97 -17.8 150

O O
NH3CH2C(O)OH -—> NH2=CH2 + HC(O)OH AHrm = 26.5 (A15)
61 178 -90.5

O0 - O
CH3CH2NHCH3 —> H- + CH3CH2NH=CH2 AHm=3l.1 (A16)
177 52.1 156

O0 O
CH3C(O)NHCH3 -9 H- + CH3C(O)NH£H2 AHm, = 31.1 (A17)
158 52.1 X=137

CH3C(O)NH2CH2C(O)OH —) CH3C(O)NH=CH2 + HC(O)OH Aern- — 42 (A18)
4. 5 137 (Est.) -90. 5

NH2CH(CH3)C(O)OH -) NHzCH(CH3)C(O)O + 00H + e' AHrmz 268.6 (A19)
-99 X=160.3 9.3

O
NHzCH(CH3)C(O)O —-) NH2=CHCH3 + C E O AHrm = -29.7 (A20)
160. 3 (Est. ) 157 -26.4

CH3C(O)NHCH2C(O)OH —) CH3C(O)NHCH2C(O)O + 00H + e' Aern = 268.6 (A21)
-l44 X=1 15.3 9.3

O
CH3C(O)NHCH2C(O)O —> CH3C(O)NH=CH2 + C E O AHm, = -4.7 (A22)
115.3 (Est) 137 (Est) -26.4

O
CH3C(O)NH=CH2 —9

260

CH3C(O)O + NH=CH2
137 (ESL) 156 32
O O
CH3C(OH)NHCH3 -) CH3C(OH)- + ONHCH3
97 207 43.6
CH3C(O)OH —) CH3C(O)0 + OOH
- 103.3 -6 9.3
CH3C(O)0 + oNHCH3 —) CH3C(O)NHCH3
-6 43.6 —56

O O
CH3C(OH)- + 00H -) CH3C(OH)OH
207 9. 3 72
CH3- + -N(CH3)2 —) CH3N(CH3)2
34. 8 35 -5.7
O0 O
H- + CH2C(O)NH2 —) CH3C(O)NH3
52. 1 165 103
CH3C(O)OH —) H- + CH3C(O)O¢
-103.3 52.1 -49.6

CH 3- + CH3C(O)OO —) CH3C(O)OCH3
34. 8 -49.6 -98

O
CH3C(O)NH3 —) CH3C(O)O + NH3

103 156 -1 1
NH2CH3 —) H- + ONHCH3
-5 .5 52. 1 43.6

O
CH3CH2NH2CH2CH3 -> CH3CH2NH2 + OCHzCH3

125 - 1 1.3 215.6

O
CH3C(O)NH2CH(CH3)C(O)OCH3 —) CH3C(O)NH2 + CH3CHC(O)OCH3
57 1 15

AHm, = 51 (A23)
AHm= 153.6 (A24)
AHl-m = 106.6 (A25)
AHm = -93.6 (A26)
AH,“ = -144. (A27)
AHm = -75.5 (A28)
AHm, = -114 (A29)
AHm, = 105.8 (A30)
AH,“ = -83.2 (A31)
AHnm = 42 (M2)
AHm, = 101.2 (A33)

Aern = 79.3 (A34)

(A35)
AHnm- - 62

LIST OF REFERENCES

LIST OF REFERENCES

 

1. Watson, J. T. Introduction to Mass Spectrometry, 2nd ed.; Raven: New York, 1985;
Chapters 1-4.

2. McLafferty, F. W.; Turecek, F. Interpretation of Mass Spectra, 4th ed.; University
Science Books: Mill Valley, CA, 1993.

3. Miiller, E. W.; Bahadur, K. Phys. Rev. 1956, 102, 624.
4. Chait, E. M. Anal. Chem. 1972, 44, 77A.

5. Harrison, A. G. Chemical Ionization Mass Spectrometry, 2nd ed.; CRC Press: Boca
Raton, FL, 1992.

6. Einolf, N.; Munson, B. Int. J. Mass Spectrom. Ion Phys. 1972, 9, 141.

7. Huang, E. C.; Wachs, T.; Conboy, J. J .; Henion, J. D. Anal. Chem. 1990, 62, 713A.
8. Busch, K. L.; Cooks, R. G. Science 1982, 218, 247.

9. Beckey, H. D. Int. J. Mass Spectrom. Ion Phys. 1969, 2, 500.

10. Beckey, H. D. Principles of Field Ionization and Field Desorption Mass
Spectrometry; Pergamon: Oxford, 1977.

11. Prokai, L. Field Desorption Mass Spectrometry; Marcel Dekker: New York, 1990;
Vol. 2.

12. Gomer, R. J. Chem. Phys. 1959, 31 , 341.

13. Giessmann, U.; Rollgen, F. W. Org. Mass Spectrom. 1976, II, 1094.

14. Sweeley, C. C.; Soltmann, B.; Holland, J. F. In High Performance Mass
Spectrometry - Chemical Applications; Gross, M. L., Ed.; ACS Symposium Series
70; American Chemical Society: Washington, DC, 1978; Chapter 11.

15. Wong, S. S; Giessmann, U.; Karas, M.; Rollgen, F. W. Int. J. Mass Spectrom. Ion
Processes 1984, 56, 139.

16. Lattimer, R. P.; Schulten, H. R. Anal. Chem. 1989, 6], 1201A.
l7. Baldwin, M. A.; McLafferty, F. W. Org. Mass Spectrom. 1973, 7, 1353.

261

262

 

18. Cotter, R. J. Anal. Chem. 1980, 52, 1589A.

19. Hunt, D. F.; Shabanowitz, J .; Botz, F. K.; Brent, D. A. Anal. Chem. 1977,49, 1160.
20. Soltmann, B.; Sweeley, C. C.; Holland, J. F. Anal. Chem. 1977, 49, 1164.

21. Anderson, W. R.; Frick, W.; Daves, G. D. J. Am Chem. Soc. 1978, 100, 1974.

22. Torgerson, D. F .; Showronski, R. P.; Macfarlane, R. D. Biochem. Biophys. Res.
Commun. 1974, 60, 616.

23. Posthumus, M. A.; Kistemaker, P. G.; Meuzelaar, H. L. C.; Ten Noever de Brauw, M.
C. Anal. Chem. 1978, 50, 985.

24. Benninghoven, A.; Sichtermann, W. K. Anal. Chem. 1978, 50, 1180.

25. Barber, M.; Bordoli, R. 8.; Elliott, G. J.; Sedgwick, R. D.; Tyler, A. N. Anal. Chem.
1982, 54, 645A.

26. Rinehart, K. L. Science 1982, 218, 254.
27. Aberth, W. H.; Straub, K. M.; Burlingame, A. L. Anal. Chem. 1982, 54, 2029.

28. Jardine, 1.; Scanlan, G. F.; Tsarbopoulos, A.; Liberato, D. J. Anal. Chem. 1988, 60,
1086.

29. Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Anal. Chem. 1991, 63, 1193A.
30. Caprioli, R. M. Anal. Chem. 1990, 62, 477A

31. Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Mass
Spectrom. Rev. 1990, 9, 37.

32. Dempster, A. J. Phys. Rev. 1921, 18, 415.

33. Bleakney, W. Phys. Rev. 1929, 34, 157.

34. Nier, A. O. Rev. Sci. Instrum. 1940, 11, 212.

35. N ier, A. O. Int. J. Mass Spectrom. Ion Processes 1990, 100, 1.

36. Nier, A. 0. Science 1955, 121, 737.

37. McLafferty, F. W. Anal. Chem. 1956, 28, 306.

38. Brenton, A. 6.; Morgan, R. P.; Beynon, J. H. Ann. Rev. Phys. Chem. 1979, 30, 51.

39. Harrison, A. G.; Tsang, C. W. In Biochemical Applications of Mass Spectrometry;
Waller, G. R., Ed.; Wiley: New York, 1972; pp 135-156.

40. Lias, S. G.; Bartmess, J. B.; Liebman, J. F.; Holmes, J. L.; Levin, R. D.; Mallard, W.
G. J. Phys. Chem. Ref. Data 1988, 17, Suppl. 1.

263

 

41.

42.

43.

45.

46.
47.
48.
49.
50.
51.
52.
53.
54.
55.

56.
57.

58.

59.
60.
61.
62.

Rosenstock, H. M.; Wallenstein, M. B.; Wahrhaftig, A. L.; Eyring, H. Proc. Nat.
Acad. Sci. 1952, 38, 667.

Cooks, R. G.; Beynon, J. H.; Caprioli, R. M.; Lester, G. R. Metastable Ions; Elsevier:
New York, 1973.

Marcus, R. A. J. Chem. Phys. 1952, 20, 359.

. Robinson, P. J .; Holbrook, K. A. Unimolecular Reactions; Wiley: New York, 1972.

Franklin, J. L.; Dillard, J. G.; Rosenstock, H. M.; Herron, J. T.; Draxl, K; Field, F. H.
Ionization Potentials, Appearance Energies, and Heats of Formation of Gaseous
Positive Ions; National Bureau of Standards: Washington DC, 1969.

Lorquet, J. C. Org. Mass Spectrom. 1981, 16, 478.

Munson, M. S. B.; Field, F. H. J. Am. Chem. Soc. 1966, 88, 2621.

Sunner, J. A.; Kulatunga, R.; Kebarle, P. Anal. Chem. 1986, 58, 1312.

Wu, Z.; Fenselau, C. Rapid Comm. Mass Spectrom. 1992, 6, 403.

Bowen, R. D.; Harrison, A. G. Org. Mass Spectrom. 1981, I6, 159.

Bombick, D. D.;Pinkston, J. D.; Allison, J. Anal. Chem. 1984, 56, 396.

Blewett, J. P.; Jones, E. J. Phys. Rev. 1936, 50, 464.

Reimann, A. L. Thermionic Emission; Wiley: New York, 1934.

Kassel, D. B.; Allison, J. Biomed. Environ. Mass Spectrom. 1988, 17, 221.

Schmelzeisen-Redeker, G.; Giessmann, U.; Rollgen, F. W. Org. Mass Spectrom.
1985, 20, 305.

Bombick, D. D.; Allison, J. Anal. Chem. 1987, 59, 458.

Beuhler, R. J .; Flanigan, B.; Greene, L. J .; Friedman, L. J. Am. Chem. Soc. 1974, 96,
3990.

Wu, H.; Rouse, J. C.; Leali, K. A.; Allison, J. Proceedings of the 40th ASMS
Conference on Mass Spectrometry and Allied Topics, Washington, DC, May 31-
June 5, 1992; p 1795.

Spears, K. G.; Ferguson, E. E. J. Chem. Phys. 1973, 59, 4174.

Allison, J.; Ridge, D. P. J. Am. Chem. Soc. 1979, 101, 4998.

Davidson, W. R.; Kebarle, P. J. Am. Chem. Soc. 1976, 98, 6133.

Rollgen, F. W.; Schulten, H. R. Org. Mass Spectrom. 1975, 10, 660.

264

 

63. Barber, M.; Bordoli, R. S.; Sedgwick, R. D.; Tyler, A. N. J. Chem. Soc., Chem
Commun. 1981, 325.

64. Barber, M.; Bordoli, R. S.; Sedgwick, R. D.; Tyler, A. N. Nature 1981, 293, 270.

65. Barber, M.; Bordoli, R. S.; Garner, G. V.; Gordon, D. B.; Sedgwick, R. D.; Tetler, L.
W.; Tyler, A. N. Biochem. J. 1981, 197, 401.

66. Honig, R. E. Int. J. Mass Spectrom. Ion Processes 1985, 66, 31.
67. Benninghoven, A. Z Phys. 1970, 230, 403.
68. Benninghoven, A.; Jaspers, D.; Sichtermann, W. Appl. Phys. 1976, II, 35.

69. Lew, H. In Methods of Experimental Physics 4A; Hughes, V. W.; Schultz, H. L.,
Eds.; Academic: New York, 1967; p 155.

70. Devienne, F. M.; Roustan, J. C. Org. Mass Spectrom. 1982, I7, 173.

71. Surman, D. J.; Vickerman, J. C. Appl. Surf. Sci. 1981, 9, 108.

72. Martin, S. A.; Costello, C. E.; Biemann, K. Anal. Chem 1982, 54, 2362.
73. Aberth, W. H.; Burlingame, A. L. Anal. Chem. 1988, 60, 1426.

74. Kloppel, K. D.; Weyer, K.; Von Bfinau, G. Int. J. Mass. Spectrom. Ion Phys. 1983,
51, 47.

75. Gilbert, J. R.; Leroi, G. E.; Allison, J. Int. J. Mass Spectrom. Ion Processes 1991,
107, 247.

76. Perel, J. In Desorption Mass Spectrometry: Are SIMS and FAB the Same ?; Lyons, P.
A., Ed.; ACS Symposium Series 291; American Chemical Society: Washington,
DC, 1985; Chapter 8.

77. Aberth, W.; Burlingame, A. L. In Ion Formation from Organic Solids;
Benninghoven, A., Ed.; Springer-Verlag: Berlin, 1983; Vol. 25, p 167.

78. Fenselau, C.; Cotter, R. J. Chem. Rev. 1987, 87, 501.

79. De Pauw, E.; Agnello, A.; Derwa, F. Mass Spectrom. Rev. 1991, 10, 283.
80. De Pauw, E. Mass Spectrom. Rev. 1986, 5, 191.

81. Gower, J. L. Biomed. Mass Spectrom. 1985, 12, 191.

82. Cook, K. D.; Todd, P. J .; Friar, D. H. Biomed. Environ. Mass Spectrom. 1989, 18,
492.

83. Sweetrnan, B. J .; Blair, 1. A. Biomed. Environ. Mass Spectrom. 1988, I 7, 337.
84. Field, F. H. J. Phys. Chem. 1982, 86, 5115.

 

265

 

85. Pachuta, S. J.; Cooks, R. G. Chem. Rev. 1987, 87, 647.

86. Cooks, R. G.; Busch, K. L. Int. J. Mass Spectrom. Ion Physics 1983, 53, 111.

87. Couchouron, M.; Hennequin, J. F. Int. J. Mass Spectrom. Ion Processes 1986, 69, 1.
88. Sigmund, P. Phys. Rev. 1969, 184, 383.

89. Benninghoven, A. Surf. Sci. 1973, 35, 427.

90. Andersen, C. A.; Hinthome, J. R. Anal. Chem. 1973, 45, 1421.

91. Kelly, R. Radiat. Eff. 1977, 32, 91.

92. Lucchese, R. R. J. Chem. Phys. 1987, 86, 443.

93. Kelly, R. Surf Sci. 1979, 90, 280.

94. Mich], J. Int. J. Mass Spectrom. Ion Phys. 1983, 53, 255.

95. Hoogerbrugge, R.; Van der Zande, W. J .; Kistemaker, P. G. Int. J. Mass Spectrom.
Ion Processes 1987, 76, 239.

96. Falcone, G.; Sroubek, Z.; Sindona, G.; Uccella, N. Int. J. Mass Spectrom. Ion
Processes 1988, 83, 223.

97. Sunner, J. A.; Kulatunga, R.; Kebarle, P. Anal. Chem. 1986, 58, 2009.

98. Visentini, J .; Zidarov, D.; Allard, M.; Bertrand, M. J. J. Am. Soc. Mass Spectrom.
1993, 4, 482.

99. Reynolds, J. D.; Cook, K. D.; Burm, J. L. E.; Woods, C. J. Am. Soc. Mass Spectrom.
1992, 3, 113.

100. Dass, C.; Desidcrio, D. M. Anal. Chem. 1988, 60, 2723.

101. Aubagnac, J. L.; Doulut, S.; Rodriguez, M.; Martinez, J. Org. Mass Spectrom.
1992, 27, 645.

102. Detter, L. D.; Hand, 0. W.; Cooks, R. G.; Walton, R. A. Mass Spectrom Rev. 1988,
7, 465.

103. Vékey, K.; Zerilli, L. F. Org. Mass Spectrom. 1991, 26, 939.

104. Sunner, J .; Ikonomou, M. G.; Kebarle, P. Int. J. Mass Spectrom. Ion Processes
1988, 82, 221.

105. Ens, W.; Mao, Y.; Mayer, F.; Standing, K. G. Rapid Comm. Mass Spectrom. 1991,
5, 117.

106. Ligon, W. V. Int. J. Mass Spectrom. Ion Phys. 1983, 52, 183.

266

 

107.

108.
109.
110.

111.
112.
113.
114.

115.
116.
117.

118.
119.

120.
121.
122.
123.
124.

125

126.

Wong, S. S.; Rollgen, F. W.; Manz, I; Przybyski, M. Biomed. Mass Spectrom.
1985, 12, 43.

Vertes, A.; Gijbels, R.; Levine, R. D. Rapid Comm. Mass Spectrom. 1990, 4, 228.
Benninghoven, A. Int. J. Mass Spectrom. Ion Phys. 1983, 53, 85.

Kidwell, D. A.; Ross, M. M.; Colton, R. J. Int. J. Mass Spectrom. Ion Processes
1987, 78, 315.

Sunner, J. J. Am. Soc. Mass Spectrom. 1993, 4, 410.
Freas, R. B.; Ross, M. M.; Campana, J. E. J. Am. Chem. Soc. 1985, 107, 6195.
Schroder, B.; Miinster, H.; Budzikiewicz, H. Org. Mass Spectrom. 1986, 21, 707.

Honda, E; Lancaster, G. M.; Fukuda, Y.; Rabalais, J. W. J. Chem. Phys. 1978, 69,
4931.

Grade, H.; Cooks, R. G. J. Am. Chem. Soc. 1978, 100, 5615.

Day, R. J .; Unger, S. B.; Cooks, R. G. Anal. Chem. 1980, 52, 557A.

Pachuta, S. J .; Cooks, R. G. In Desorption Mass Spectrometry: Are SIMS and FAB
the Same ?; Lyons, P. A., Ed.; ACS Symposium Series 291; American Chemical
Society: Washington, DC, 1985; Chapter 1.

Sunner, J. A.; Kutlatunga, R.; Kebarle, P. Anal. Chem. 1986, 58, 1312.

Sunner, J .; Morales, A.; Kebarle, P. Int. J. Mass Spectrom. Ion Processes 1988, 86,
169

Bojesen, G; Moller, J. Int. J. Mass Spectrom. Ion Processes 1986, 68, 239.
Sunner, J .; Morales, A.; Kebarle, P. Anal. Chem. 1988, 60, 98.

Sunner, J.; Morales, A.; Kebarle, P. Anal. Chem. 1987, 59, 1378.

Vestal, M. L. Mass Spectrom. Rev. 1983, 2, 447.

Rouse, J. C.; Allison, J. J. Am. Soc. Mass Spectrom. 1993, 4, 259.

Ehring, H.; Karas, M.; Hillenkamp, F. Org. Mass Spectrom. 1992, 27, 472.
Hoogerbrugge, R.; Bobeldijk, M.; Kistemaker, P. G.; Los, J. J. Chem. Phys. 1988,

88, 5314.

127.

Derwa, F.; DePauw, E.; Natalis, P. Org. Mass Spectrom. 1991, 26, 117.

128. Williams, D. H.; Naylor, S. J. Chem. Soc., Chem. Commun. 1987, 1408.

129.

Hamdan, M. Rapid Comm. Mass Spectrom. 1992, 6, 224.

267

 

130. Lehmann, W. D.; Kessler, M.; Koning, W. A. Biomed. Mass Spectrom. 1984, I I,

131.
132.
133.
134.
135.

136.

137.
138.
139.
140.
141.
142.

143.

144.

145.

146.
147.

148.

149.
150.
151.

217.

Keough, T. Anal. Chem. 1985, 5 7, 2027.

Musselman, B. D.; Allison, J .; Watson, J. T. Anal. Chem. 1985, 5 7, 2425.
Shiea, J .; Sunner, J. Org. Mass Spectrom. 1991, 26, 38.

Ligon, W. V.; Dorn, S. B. Int. J. Mass Spectrom. Ion Processes 1986, 78, 99.

Hayes, R. N.; Gross, M. L. In Mass Spectrometry; McCloskey, J. A., Ed.; Methods
in Enzymology; Academic: San Diego, CA, 1990; Vol. 193, Chapter 10.

Herman, Z.; Futrcll, J. H.; Friedrich, B. Int. J. Mass Spectrom. Ion Processes 1984,
58, 181.

Levsen, K.; Schwarz, H. Mass Spectrom. Rev. 1983, 2, 77.

Kim, M. 8.; Int. J. Mass Spectrom. Ion Phys. 1983, 50, 189.

Laramée, J. A.; Cameron, D.; Cooks, R. G. J. Am. Chem. Soc. 1981, 103, 12.
Kim, M. S.; McLafferty, F. W. J. Am. Chem. Soc. 1978, 100, 3279.
Neumann, G. M.; Derrick, P. J. Org. Mass Spectrom. 1984, 19, 165.

Griffin, L. L.; McAdoo, D. J. Proceedings of the 40th ASMS Conference on Mass
Spectrometry and Allied Topics, Washington, DC, May 31-June 5, 1992; p 1831.

Dekrey, M. J .; Mabud, M. D. A.; Cooks, R. G.; Syka, J. E. P. Int. J. Mass Spectrom.
Ion Processes 1985, 67, 295.

Bier, M. E.; Amy, J. W.; Cooks, R. G. Int. J. Mass Spectrom. Ion Processes 1987,
77, 31.

Bomse, D. S.; Woodin, R. L.; Beaucharnp, J. L. J. Am. Chem. Soc. 1979, 101,
5503.

Gorman, G. S.; Amster, I. J. Org. Mass Spectrom. 1993, 28, 437.

Jennings, K. R.; Dolnikowski, G. G. In Mass Spectrometry; McCloskey, J. A., Ed.;
Methods in Enzymology; Academic: San Diego, CA, 1990; Vol. 193, Chapter 2.

Mamyrin, B. A.; Karatajev, V. J .; Shmikk, D. V.; Zagulin, V. A. Soviet Phys. JETP
1973, 3 7, 45.

Mattauch, J. Phys. Rev. 1936, 50, 617.
Nier, A. O. J. Am. Soc. Mass Spectrom. 1991, 2, 447.
Miller, P. E.; Denton, M. B. J. Chem. Ed. 1986, 63, 617.

 

268

 

152. Paul, W.; Reinhard, H. P.; von Zahn, U. Z Phys. 1958, 52, 143.
153. Dawson, P. H. Mass Spectrom. Rev. 1986, 5, l.

154. Wiley, W. C.; McLaren, I. H. Rev. Sci. Instrum. 1955, 26, 1150.
155. Cotter, R. J. Anal. Chem. 1992, 64, 1027A.

156. Yost, R. A.; Enke, C. G. Anal. Chem. 1979, 5], 1251A.

157. Bruins, A. P.; Jennings, K. R.; Evans, S. Int. J. Mass Spectrom. Ion Phys. 1978, 26,
395.

158. Jennings, K. R. In High Pen‘ormance Mass Spectrometry - Chemical Applications;
Gross, M. L., Ed.; ACS Symposium Series 70; American Chemical Society:
Washington, DC, 1978; Chapter 1.

159. Yost, R. A.; Boyd, R. K. In Mass Spectrometry; McCloskey, J. A., Ed.; Methods in
Enzymology; Academic: San Diego, CA, 1990; Vol. 193, Chapter 7.

160. Gross, M. L. In Mass Spectrometry; McCloskey, J. A., Ed.; Methods in
Enzymology; Academic: San Diego, CA, 1990; Vol. 193, Chapter 6.

161. Schey, K. L.; Schwartz, J. C; Cooks, R. G. Rapid Commun, Mass Spectrom. 1989,
3, 305.

162. Puzycki, M. A.; McLane, R. D_.; Vlasak, P. R.; Enke, C. G. Proceedings of the 40th
ASMS Conference on Mass Spectrometry and Allied Topics, Washington, DC,
May 31-June 5, 1992; p 1338.

163. Rouse, J. C.; Light, K. J.; Allison, J. Proceedings of the 38th ASMS Conference on
Mass Spectrometry and Allied Topics, Tucson, Arizona, June 3-8, 1990; p 455.

164. Ackermann, B.; Allison, J .; Musselman, B.; Pinkston, D.; Bombick, D.;
Dolnikowski, G.; Tsarbopoulos, A.; Watson, J. T. Proceedings of the 3lst ASMS
Conference on Mass Spectrometry and Allied Topics, Boston, MA, May 1983; p
600.

165. Dolnikowski, G. G.; Watson, J. T.; Allison, J. Anal. Chem. 1984, 56, 197.

166. Munster, H.; Budzikiewicz, H.; Schr'dder, E. Org. Mass Spectrom. 1987, 22, 384.

167. Chiarelli, M. P.; Gross, M. L. Int. J. Mass Spectrom. Ion Processes 1987, 78, 37.

168. Miinster, H.; Theobald, F.; Budzikiewicz, H.; Schroder, E. Int. J. Mass Spectrom.
Ion Processes 1987, 79, 73.

169. Miller, J. M.; Balasanmugam, K.; Fulcher, A. Org. Mass Spectrom. 1989, 24, 497.

170. Michaud, D. P.; Kyranos, J. N.; Brennan, T. F.; Vouros, P. Anal. Chem. 1990, 62,
1069.

 

269

 

171. Kassel, D. B. Ph.D. Thesis, Michigan State University, 1988.

172. Schultz, G. A.; Allison, J. Proceedings of the 37th ASMS Conference on Mass
Spectrometry and Allied Topics, Miami Beach, Florida, May 21-26, 1989; p 527.

173. Light, K. J. Ph.D. Thesis, Michigan State University, 1990.
174. Dahl, D. A.; Delmore, J. E.; Appelhans, A. D. Rev. Sci. Instrum. 1990, 61, 607.

175. Dahl, D. A.; Delmore, J. E. SIMION PC/PSZ Version 4.0, EGG-CS-7233 Revision
2, April 1988.

176. Light, K. J .; Allsion, J. J. Am. Soc. Mass Spectrom. 1990, I, 455.
177. Weddle, G. H.; Allison, J.; Ridge, D. P. J. Am. Chem. Soc. 1977, 99, 105.
178. Borchers, F.; Giessmann, U.; Rollgen, F. W. Org. Mass Spectrom. 1977 , 12, 539.

179. Harris, D. C. Quantitative Chemical Analysis, 2nd Ed.; Freeman: New York, 1987;
p 202.

180. Musselman, B. D.; Watson, J. T.; Chang, C. K. Org. Mass Spectrom. 1986, 21,
215.

181. Denys, L.; Bothner-By, A. A.; Fisher, G. H.; Ryan, J. W. Biochemistry 1982, 21 ,
6531.

182. London, R. B.; Stewart, J. M.; Cann, J. R.; Matwiyoff, N. A. Biochemistry 1978,
I 7, 2270.

183. Lintner, K.; Fermandjian, S.; St. Pierre, 8.; Regoli, D. Biochem. Biophys. Res.
Commun. 1979, 91, 803.

184. Naumenko, V. N.; Koren, A. 0.; Gaponik, P. N. Magn. Reson. Chem. 1992, 30,
558.

185. McLafferty, F. W. Org. Mass Spectrom. 1980, 15, 114.

186. Biemann, K. In Mass Spectrometry; McCloskey, J. A., Ed.; Methods in
Enzymology; Academic: San Diego, CA, 1990; Vol. 193, Chapter 25, Appendix 5.

187. Dass, C.; Desidcrio, D. M. Anal. Biochem. 1987, 163, 52.
188. Grese, R. P.; Cerny, R. L.; Gross, M. L. J. Am. Chem. Soc. 1989, 111, 2835.

189. Bean, M. F.; Carr, 8. A.; Thorne, G. C.; Reilly, M. H.; Gaskell, S. J. Anal. Chem.
1991, 63, 1473.

190. Wu, Z.; Fenselau, C. J. Am. Soc. Mass Spectrom. 1992, 3, 863.
191. Meot-Ner (Mautner), M. J. Am. Chem. Soc. 1984, 106, 278.

 

270

 

192.
193.
194.
195.

196.
197.
198.
199.

200.

201.

202.

203.
204.
205.

206.
207.

208.

209.

210.

211.

212.

Bouchonnet, S.; Hoppilliard, Y. Org. Mass Spectrom. 1992, 27, 71.
Jensen, F. J. Am. Chem. Soc. 1992, 114, 9533.
Naylor, S.; Moneti, G. Biomed. Environ. Mass Spectrom. 1989, I8, 405.

Johnson, R. S.; Martin, S. A.; Biemann, K. Int. J. Mass Spectrom. Ion Processes
1988, 86, 137.

Bunk, D. M.; Macfarlane, R. D. Int. J. Mass Spectrom. Ion Processes 1991, 111, 55.
Fraenkel, G.; Franconi, C. J. Am. Chem. Soc. 1960, 82, 4478.
Cavell, R. G.; Allison, D. A. J. Am. Chem. Soc. 1977, 99, 4203.

Hinton, J. F.; Beeler, A.; Harpool, D.; Briggs, R. W.; Pullman, A. Chem. Phys. Lett.
1977, 47, 411.

Kollman, P.; Rothenburg, S. J. Am. Chem. Soc. 1977, 99, 1333.

Addoud, J .-L. M.; Cafiada, T.; Homan, H.; Notario, R.; Cativiela, C.; Diaz de
Villegas, M. D.; Bordejé, M. G; M6, 0.; Yafiez, M. J. Am. Chem. Soc. 1992, 114,
4728.

Wagner, D. S.; Gage, D. A.; Watson, J. T.; Allison, J. J. Am. Soc. Mass Spectrom.,
submitted for publication.

Teesch, L. M.; Orlando, R. C.; Adams, J. J. Am. Chem. Soc. 1991, 113, 3668.
Tsang, C. W.; Harrison, A. G. J. Am. Chem. Soc. 1976, 98, 1301.

Hunt, D. R; Yates 111, J. R.; Shabanowitz, J .; Winston, S.; Hauer, C. R. Proc. Natl.
Acad. Sci. 1986, 83, 6233.

Teesch, L. M.; Adams, J. J. Am. Chem. Soc. 1991, 113, 812.

Kiryushkin, A. A.; Fales, H. M.; Axenrod, T.; Gilbert, E. J .; Milne, G. W. A. Org.
Mass Spectrom. 1971, 5, 19.

Kenny, P. T. M.; Nomoto, K.; Orlando, R. Rapid Commun. Mass Spectrom. 1992,
6, 95.

Yeh, R. W.; Grimley, J. M.; Bursey, M. M. Biological Mass Spectrom. 1991, 20,
443.

CRC Handbook of Chemistry and Physics, 65th ed.; Weast, R. C., Ed.; CRC Press:
Boca Raton, FL, 1985; p F-188.

Fenselau, G; Murphy, C. M.; Cotter, R. J. J. Am. Soc. Mass Spectrom. 1990, I,
336.

Madden, T.; Welham, K. J .; Baldwin, M. A. Org. Mass Spectrom. 1991, 26, 443.

 

271

 

213
214

215
216

217
218
219

220
221
222

223.
224.

225.

226.
227.
228.

229.
230.
231.

232.
233.

. Thorne, G. C.; Gaskell, S. J. Rapid Comm. Mass Spectrom. 1989, 3, 217.

. Ballard, K. D.; Gaskell, S. J. Int. J. Mass Spectrom. and Ian Processes 1991, 111,
173.

. Renner, D.; Spiteller, G. Biomed. Mass Spectrom. 1986, 13, 405.

. Knapp, D. R. In Mass Spectrometry; McCloskey, J. A., Ed.; Methods in
Enzymology; Academic: San Diego, CA, 1990; Vol. 193, Chapter 15.

. Alexander, A. J .; Boyd, R. K. Int. J. Mass Spectrom. Ion Processes 1989, 90, 211.
. Tang, X.; Ens, W.; Standing, K. G.; Westmore, J. B. Anal. Chem. 1988, 60, 1791.

. Thorne, G. C.; Ballard, K. D.; Gaskell, S. J. J. Am. Soc. Mass Spectrom. 1990, I,
249.

. Svec, H. J.; Junk, G. A. J. Am. Chem. Soc. 1964, 86, 2278.

. Cordero, M. M.; Houser, J. J.; Wesdemiotis, C. Anal. Chem. 1993, 65, 1594.

. Smith, R. D.; Light-Wahl, K. J .; Ogorzalek Loo, R. R.; Winger, B. E.; Loo, J. A.;
Rockwood, A. L. Proceedings of the 40th ASMS Conference on Mass
Spectrometry and Allied Topics, Washington, DC, May 31-June 5, 1992; p 637.

Speir, J. P.; Gorman, G. S.; Comett, D. S.; Amster, I. J. Anal. Chem. 1991, 63, 65.

Allison, J. In Progress in Inorganic Chemistry; Lippard, S. J ., Ed.; Wiley: New
York, 1986; Vol. 34, p 627.

Lehman, T. A.; Bursey, M. M. Ion Cyclotron Resonance Spectrometry; Wiley:
New York, 1976.

O'Connor, P. J. Ph.D. Thesis, Michigan State University, 1991.
Wilkins, C. L.; Gross, M. L. Anal. Chem. 1981, 53, 1661A.

Lombarski, M.; Allison, J. Int. J. Mass Spectrom. Ion Processes 1983, 49, 281.;
Lombarski, M.; Allison, J. Int. J. Mass Spectrom. Ion Processes 1985, 65, 31.

Speir, J. P.; Gorman, G. S.; Amster, I. J. J. Am. Soc. Mass Spectrom. 1993, 4, 106.
Teesch, L. M.; Adams, J. Org. Mass Spectrom. 1992, 2 7, 931.

Hogan, M. J .; Ong, P. P.; Ang, J. L.; Cheang, K. K. Int. J. Mass Spectrom. Ion
Processes 1992, 116, 249.

Holland, P. M.; Castleman, A. W. J. Am. Chem. Soc. 1980, 102, 6175.

Volkening, J .; Heumann, K. G. In Advances in Mass Spectrometry; Todd, J. F. J .,
Ed.; Wiley: New York, 1985; Part B, p 1059.

272

 

234. Heinz, 0.; Reaves, R. T. Rev. Sci. Instr. 1968, 39, 1229.

235. Hodges, R. V.; Armentrout, P. B.; Beauchamp, J. L. Int. J. Mass Spectrom. Ion
Phys. 1979, 29, 375.

236. Chowdhury, A. K.; Wilkins, C. L. Int. J. Mass Spectrom. Ion Processes 1988, 82,
163.

237. Leuchtner, R. B.; Farley, R. W.; Harms, A. C.; Funasaka, H.; Castleman, A. W. Int.
J. Mass Spectrom. Ion Processes 1990, 102, 199.

238. Tomer, K. B. Mass Spectrom. Rev. 1989, 8, 445.

239. Mallis, L. M.; Russell, D. H. Int. J. Mass Spectrom. Ion Processes 1987, 78 147.
240. Leary, J. A.; Williams, T. D.; Bott, G. Rapid Comm. Mass Spectrom. 1989, 3, 192.
241. Tomer, K. B.; Deterding, L. J .; Guenat, C. Biomed. Mass Spectrom. 1991, 20, 121.
242. Grese, R. P.; Gross, M. L. J. Am. Chem. Soc. 1990, 112, 5098.

243. Gross, M. L. Int. J. Mass Spectrom. Ion Processes 1992, 118, 137.

244. Leary, J. A.; Zhou, Z.; Ogden, S.; Williams,T. D. J. Am. Soc. Mass Spectrom.
1990, I. 473.

245. Nakamora, T., JEOL USA, Ltd., Boston, MA, personal communication, 1990.

246. Wagner, D. S.; Allison, J. Proceedings of the 40th ASMS Conference on Mass
Spectrometry and Allied Topics, Washington, DC, May 31-June 5, 1992; p 1399.

247. Balodis, Y. Y.; Nikiforovich, G. V.; Grinsteine, I. V.; Vegner, R. E.; Chipens, G. F.
FEBS Letters, 1978, 86, 239.

"JJJJJJJJJJJJJJ