MSU

LIBRARIES
.-:—.

 

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped be10w.

 

 

 

 

 

 

 

 

 

LIGAND EFFECTS IN THE CHEMISTRY
OF GAS PHASE METAL IONS

By

Richard Martin Stepnowski

A DISSERTATION

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

DOCI‘OR OF PHILOSOPHY

Department of Chemistry

1988

ABSTRACT

LIGAND EFFECTS IN THE CHEMISTRY
OF GAS PHASE METAL IONS

By

Richard Martin Stepnowski

The gas phase chemistry of transition metal ions with organic
molecules has been extensively studied in recent years. The study of
the interactions between metals and molecules can give insight to
basic questions of organometallic chemistry. '

For small compounds such as alkyl halides (RX), reaction
products appear to be formed via three steps: metal ion insertion]
B- H / competitive ligand loss step.

This study involves the reactions of Ni+, NiCO+, NiPF3+ , and

NiCp+ with phenyl, benzyl, and aromatic carbonyl compounds.
Reaction products and pathways were determined by Ion Cyclotron
Resonance (ICR) Mass Spectrometry, using a frequency scanned
detector (FSD).

The phenyl group in some aspects behaves chemically
intermediate to the methyl and ethyl groups. H transfer from the
phenyl group is a high energy process; nonetheless, reactions such

as:
Ni+ + C6H5X —+ NiC6H4+ + HX

are observed. When other pathways are open, such as the
elimination of a molecule from the functional group, these pathways

are preferred.
A major effect of the phenyl group is its inductive effect. The
weak C6H5CH2-X bond plays an important role in in reactions of

substituted benzyl compounds. Charge transfer to C6H5CH2'
following insertion into this bond can occur, due to the low ionization
potential of the benzyl radical.

Much of the chemistry of the NiL+ ions can be explained by the
formation of intermediates with low steric interactions. The NiL+ ions
insert into bonds where steric interactions will be minimized. By
size, Cp > PF3 > CO. This is also the same order as for the metal-

ligand bond energies; the reactivity of the ligated metal ions may be

related to both these factors. The reactions of NiPF3+ follow closely
the reactions of NiCO+, with slightly lower reactivity observed for the
NiPF3+ ion. The slightly lower reactivity of NiPF3+ reflects the steric
crowding for insertion produced by this larger ligand. The reactivity
of NiCp+ indicate that the Cp ligand affects the metal ion differently
than either CO or PF3, more reactive than the other NiL+ species in

the case of the halobenzenes. For aromatic carbonyls, the steric
hindrance of the bulky Cp group appears to prohibit insertion.

ACKNOWLEDGEMENTS

It is difficult to acknowledge all the people who have helped
so much in my time here at MSU; their help has been enormous. Let
me start by thanking Mark Bishop, who represents the best of
teachers. His T'ai Chi classes always gave a pleasant and bizarre
contrast to the world of chemistry. John Wronka gets a special
thanks for building and installing the FSD, which helped so much.
The group always made the lab a fun place to be; thanks, Mike, Mac,
Barb, Pinky, Tony, d.B., Spike, Dan, Karen, Paul, Mark, Jeff, Gary, and
Mary. A toast to the waitresses of the Boom Boom Room and the
Peanut Barrel; there exists no finer place to raise a glass! A round of
applause to the MSU hockey team!

The assistance of John Allison has been immeasureable; he
deserves thanks for giving me the opportunity to work for him, for
his long and patient guidance, and for his sense of humor, which
would always make things look not so bad, just when I thought
things couldn't get worse.

Finally, Mom and Dad, thanks for everything. Let's hope you
don't get too many phone calls like: "Send laywers, guns, and money;
Dad, get me out of this."

Each of us knows one small fraction of the overall plan at any
given moment-his own assingment-and no one is allowed to disclose
his function to another. It’s a precaution against the maniac finding
out the plan. If each man properly brings off his own part, then the
whole scheme will be brought to a successful conclusion. In the
meantime, the plan can’t be either carelessly disclosed or given up
under duress or threat. Each one can only account for a tiny
fragment which would have no meaning to the maniac should he
gain access to it.

Woody Allen
from Death {A Play}

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES

LIST OF SCHEMES

Chapter l-Introduction
References

Chapter 2-Experimental
References

Chapter 3-Reactions of aldehydes with metal ions
References

Chapter 4-Reactions of Co+ with nitriles

The Gas Phase Chemistry of First Row Transition
Metal-and Metal Containing Ions with Alkyl
Nitriles

ABSTRACT

I. Introduction

11. Experimental Section

III. Results and Discussion

The M+(RCN) complex
The Chemistry of M+

Ligand Effects
Relationship to Prior Work
IV. Conclusions
V. References
Chapter 5- Reactions of Ni+ and NiL+ with aromatic
compounds

iii

page
iii

iv

12
13
32
33
42

43
50

50
51
56
59

62
64

67
7O
73
75

77

The Gas Phase Organometallic Chemistry of Ni+, 81
NiCO+, NiPF3+, and Nic5H5+ With Aromatic
Compounds: Chemical Consequences Due to the

Presence of a Phenyl Group in the Organic Molecule,
and Due to the Presence of a Ligand on the Metal

ABSTRACT 8 1
Introduction 82
Experimental Section 8 7
Results and Discussion 8 9
Gas phase chemistry of NiL+ (L: CO, PF3, 100

Cp) with aromatic compounds
NICO+ 1 0 0
NIPF3+ I O 4
NiCp+ I 0 5
Expected Trends in the reactions of NiL+ 106
Increased Reactivity of NiCp+ 10 8
Trends reflecting steric effects 109
Conclusions 1 16
Acknowledgements 118
References 134
Chapter 6-Conclusions 13 7
References 141
Appendix -ICR Operating Manual 142

iv

LIST OF TABLES

Table page
3-1 Reactions of aldehydes with M+ and ML+ 39
4-1 CAD reactions of Co+-(Nitrile) adducts 48
4-2 Reactions of Co+ and CoCO+ with Alkyl Nitriles 5 7
4-3 Reactions of M+ and ML+ with C6H13CN 5 8
4-4 Some Thermochemical considerations (kcal/mol) 60
5-1 Some Relevant Proton Affinities 86
5-2 Reactions of Ni+ with Phenyl Compounds 90
5-3 Reactions of Benzyl Compounds 93
5-4 Reactions of Ni+ with Aromatic Carbonyl Compounds 98
5-5 Reactions of mono-ligated Ni+ with aromatic 101
compounds '

5-6 Reactions of small alkyl molecules with metal ions 119

5-7 Reactions of aromatic compounds with Ni+ and NiL+ 121
5-8 Reactions of dimethyl sulfoxide, cyclopentanone, 132

and cyclohexanone with Ni"' and NiL+

LIST OF FIGURES

Figure

1-1 Interaction of metal ions and alkanes

1-2 Cyclic intermediates allow carbon atoms remote
from the functional group to interact with the
metal ion

2-1 The ICR cell

2-2 Ion motion in the ICR cell

2-3 Block diagram for ion detection using a Marginal
Oscillator Detector

2-4 Circuitry diagram for the Frequency Scanned
Detector

2-5 Block diagram for ion detection using a
Frequency Scanned Detector

2-6 Effect of irradiating voltage on peak intensity

2-7 Effect of irradiating voltage on spectrum baseline

2-8 Effect of double resonance voltage on a selected ion

2-9 Double resonance spectra

2-10 Effect of mismatched peaks

3-1 Bond insertion preferences of the different metal
ions into hexanal and heptanal

vi

page

13
15
16

19

19

22
24
25
28
29
37

Scheme 1-1
Scheme 5-1

LIST OF SCHEMES

vii

Chapter 1

Introduction

In recent years the reactions in the gas phase of transition

metal ions with organic molecules have been extensively studied.1
Besides the fundamental interest in this area, it is hoped that the
study of interactions between metals and molecules can provide an
understanding concerning the question of how catalysts function.

Transition metal ions, Ml“, especially Fe+, Ni+, and Co+, have

been observed to react by the following mechanism2 (Scheme 1-1):

x + M+ [3' H) I "MiH

\I/ Z I--M+HX

y..M’r.HX N
M+~HX +J

Scheme 1-1

The reaction has been proposed to proceed by four distinct steps3.
First, complexation occurs as the metal ion and the molecule form an

initial ion/molecule complex. This step can be followed by insertion
of the metal ion into a bond of the molecule. The third step involves
a 8-H shift to produce two adducts on the metal ion. Finally, by
competitive ligand loss, a neutral molecule is lost, usually the
molecule with the lower proton affinity. This reaction mechanism
has been used to explain the reaction products of metal ions with
many classes of compounds,1 including alkanes,4 alcohols,2 and alkyl
halides.2

Why is it that the transition metal ions have received so much
attention? One reason is the phenomenal reactivity that is observed
in the gas phase. What causes this unusual reactivity? It has been
observed that the reactivity of the first row transition metals with

alkanes follows the trend Fe+ > Co+ > Nil”, with Cr+ and Mn+ being

unreactives. This order also is the same as the order for the
promotional energy of the the ions, that is, the energy required to
achieve an electronic state in which two electrons are not in d

orbitalsS. With two electrons promoted out of d orbitals into s or p
orbitals, the metal ion is capable of forming the two bonds necessary
for bond insertion. Cr+ with a 3d5 configuration and Mn+ with
3d5431 configuration both have high promotional energies, requiring
the excitation of electrons out of a stable half filled d shell. These
ions are correspondingly less reactive than Fe+, Ni+, and Co+.

Other concepts have been used to explain the relative
reactivity of metal ions. Recently, Schilling and Beauchamp6 have
reported the reactions of Pr+, Eu+, and Gd+ with alkanes and alkenes.
Their intent was to determine the role of f-orbitals in the reactivity
of metal ions. They found that Pr+ and Eu+ were quite unreactive.
Their proposed explanation for this unreactive behavior lies in the
first step of the reaction, complexation. A loose collision complex is

formed by the interaction of the ion with a permanent dipole or an

3

ion-induced dipole. The energy gained from this ion/molecule
association allows many ion/molecule reactions to proceed with no
appreciable barrier. This is illustrated in Figure 1-1, taken from

reference 6:

 

Relotwe Energy

 

Figure 1-1 Interaction of metal ions and alkanes

Curve 3 shows the hypothetical interaction between a neutral metal
and a neutral molecule, with an activation barrier Ea for the insertion
process. Curve _c_ shows the same insertion process for a metal ion.
Here the "chemical activation" lowers the effective barrier, allowing
the reaction to occur as if there were no barrier. They hypothesize
that, due to the large size of the Pr+ and Eu+, ions the ion-neutral
separation would be greater. This would reduce the energy of
association, which results from electrostatic interactions. Such a case

is shown in curve b where the energy lowering is not enough to

compensate for the energy barrier, which must occur for the reaction
to proceed. For C-H insertion to occur, the sum of the M+-C and

M+-H bond strengths must be larger than the OH bond being broken.

4

Since almost all of the low-lying electronic states of Pr+ and Eu+ have
either a 4f“'16s1 or 4fn'15d1 valance electronic configuration, one
of the metal 0 bonds must be made with a 4f orbital. They cited
several drawbacks to bonding to f orbitals. First, the 4f orbitals are
small and spatially compact relative to the 5d and 6s orbitals,
making access to these orbitals difficult. Second, the shape of the f
orbitals produces lower electron density along a potential bond axis.
This is similar to the situation with the first row transition metals.
The 3d orbitals are smaller than the 4s orbitals, so it is necessary to
the promote electrons out of the 3d orbitals.

In contrast, Gd+ was observed to be quite reactive. Its ground
state 41‘75d16s1 has two electrons in non-f orbitals, and is thus able
to insert, forming 2 0 bonds without utilizing f electrons.

In another report, Schilling and Beauchamp7 have compared
the reactivity of Cr+ (3d5 ground state) to Mo+ (4d5 ground state).
As stated before, Cr+ is unreactive with alkanes, while Mo+ reacts
with all alkanes, except methane, by dehydrogenation (with a small
reaction cross section). From theoretical considerations, they
concluded that Cr+ forms very weak 0 bonds, causing insertion into
C-H bonds to be endothermic. The larger size of the d orbitals of Mo+
and the resultant decrease in d-d exchange energy lost on the
formation of 0 bonds results in stronger bonds to Mo+, just strong
enough to activate the OH bonds of alkanes. Thus exchange energy
loss in the metal that results from bond formation can have an
important role in the insertion process.

An important consideration is whether the reactant ions are
formed in the ground state. Halle, Armentrout, and Beauchamp8
have reported that excited state Cr+ is reactive with alkanes while,
again, the ground state is not. More recently, Schultz, Elkind, and

Armentrout9 have reported that the 6D ground state of Fe+ is much

less reactive in endothermic reactions with small alkanes than the

first excited state 4F, even though the ground state is less than 0.25
eV lower in energy. These examples confirm the importance of
electronic structure in metal ion reactions.

In addition to studies of bare metal ions, many reactions of
ligated metal ions have been studied. Historically, these reactions
were studied because of the method of producing metal ions.
Electron ionization (E1) of volatile organometallic species such as

Fe(CO)5 produced large amounts of Fe+, and also the ions Fe(CO)x+, x
=1-5, which have been studied.10 These ions also undergo
interesting reactions. Using Fourier Transform Mass Spectrometry
(FTMS), ions can be manipulated to undergo multiple reactions. Using
this method, exotic ML+ species have been produced by other
workers, with L: H, OH, S, and even other metals“,12 These
ligands are presumably bonded to the metal ion by covalent bonds.
Studies of these species show that the ligand plays a direct role in
reactions. For example CoD+ reacts with propene to yield four

products:13

CoD+ + C3H6 —> CoCH3+ + C2H3D (6%) (1 )
—) CoCH2D+ + (32114 (4%) (2)
—-) Coc3H5+ + HD (36%) (3)
—> Coc3H4D+ + H2 (54%) (4)

With species such as MCO+, the ligand does not play such an

active role in the reaction. Recent theoretical studies of MCO+ ions
show that the bonding of the ligand in these species is primarily

electrostatic in nature.14 The reactions of MCO+ resemble that of M+

to a large extent, for no change in the electronic structure of M+

6

occurs upon complexation with the CO ligand.

Obviously, an important consideration in the study of ion/
molecule reactions is the choice of the neutral organic molecule for
the ion to react with. Many different classes of organic compounds
have been used in these studies. Interestingly, small molecules tend
to be unreactive, except for ligand substitution reactions, where an
neutral molecule displaces a ligand on the ion. For example,

acetonitrile and acetaldehyde are unreactive with Co‘I.15’l6
However, reactions such as

Coco+ + CH3CN —+ CoCH3CN+ + co - (5)

occur.

The early experiments in this field suggested that all polar
compounds reacted in the same way, that is by metal ion insertion
into the polar bond, followed by a 8-H shift. This is illustrated in
scheme 1. Reactions of this type were observed for ethyl iodide,
ethanol, diethyl ether, and other compounds},17 Amines were an
exception to this trend.18 It was proposed that the M+-NH2 bond
strength was unusually low, so the overall process would be
thermodynamically unfavored. Thus, the functional group can be
influential in the reaction products formed.

When the functional group is more complex, such as a
carboxylic acid19 or nitro group,20 insertion into bonds within the
functional group is a common pathway, in addition to insertion into
the carbon-functional group bond.

Several studies have examined the question of how
bifunctional organic molecules react with metal ions.21v22
Tsarbopoulos has studied the reactions of a series of chloroalcohols
and bromoalcohols.22 He concluded that, when two functional
groups were present, the product distribution could not be predicted,

7

based on the known chemistry of each functional group. For the
various bifunctional compounds studied, the reaction products were
unique to the combination of functional groups and length of the
alkyl chain between the two functional groups.

Tsarbopoulos and Allison also have reported on the reactions of
molecules with long straight alkyl chain attached to the functional
groups.23 These included chloro- and bromo- alkanes and alcohols,
with two to eight carbons in the alkyl chain. Based on the reactions
of Co with ethanol, insertion into the C-OH bond should be the only
process predicted:

C0+ + C2H50H —) COC2H4+ + H20 (87%) (6)
—-) COH20+ + C2H4 (13%) (7)

However, C-OH insertion does not occur in the reaction of heptanol:

Co+ + C7H150H -—> C0C2H50H+ + C5H10 (3%) (8)
—> C0C4,H8+ + C3H7OH (24%) (9)
—> C0C3H7OH+ + C4H8 (7%) (10)
—> C0C3H50H+ + C4H10 (4%) (1 l)
—> C0C4H6+ + C3H7OH + H2 (14%) (12)
—> C0C3H6+ + C4H90H (14%) (13)
—-) C0C4H90H+ + C3H6 (1%) (l4)
—) COC4H7OH+ + C3H8 (12%) (15)
-) C0C7H150H+ (21%) (16)

As the chain length increases, insertion into internal C-C bonds
becomes preferred over insertion into the C-X bond. A mechanism
was proposed in which the metal ion interacts with carbon atoms
remote from the functional group by the formation of cyclic
intermediates, shown in Figure 1-2.

C\ C/ C\ 1"; “Ex H
\ c/

Figure 1-2 Cyclic intermediates allow carbon atoms remote from
the functional group to interact with the metal ion.

After initial complexation, the alkyl chain can wrap around to
bring remote carbon atoms into close proximity with the metal ion.
This suggests that the initial attraction between the metal ion and
the molecule is a dynamic system. Electrostatic modeling can
suggest that some sites in the molecule are more attractive to the ion
than other sites.3

Branching sites on the alkyl chain can influence where bond
insertion will occur . Ni”r does not insert into the C-CN bond of n-
propyl cyanide, nor does it into insert into the C-CN bond of any of
the n-alkyl cyanides studied.” However, the following reactions

were observed:

Co+ + i-C3H7CN —+ CoHCN+ + c3116 (18)
—) CoC3H6+ + HCN (19)

In this case, the lower bond strength of the C-CN bond for the

9

secondary cyano group may allow insertion at this point.

Another example of how the alkyl group allow the metal to
insert into a normally inaccessible bond is seen with allylamine.21
Reactions of Co+ with primary and secondary amines showed that
insertion into C—NH2 did not occur.18 However, allylamine was

observed to react as follows:

CO+ + C3H5NHZ -> COC3H4NH+ + H2 (20)
—> COC3H4+ + NH3 (21)
-) COC3H5+ + 'NH2 (22)

The strong Co+—allyl bond allowed insertion to occur even though the
Co+-NH2 bond is thought to be unusually weak. One must consider
the strength of the bonds formed as well as the strength of the bonds
broken when proposing why insertion processes occur when they are
not expected.

The study of gas phase metal ion chemistry has progressed
rapidly in the last decade. Our understanding of the factors that
control the reactivity of metal ions has grown. In addition to the
factors that the metal ion have on the reaction, we have seen that
polar molecules themselves can react in many different ways, ways
that perhaps would not have been predicted. These factors include
alkyl chain size and chain branching, as well as the obvious factors
involved with the functional group.

In the projects described in this dissertation, a major goal was
to understand better the role of ligands in the reactions of metal ions.
In the primary project, Ni(CO)4, Ni(PF3)4, and NiCpNO were chosen
as the organometallic sources of the mono-ligated metal ions. Large
amounts of NiCO+, NiPF3+, and NiCp+ are formed on the electron
ionization of these compounds. Reactions of these species would
allow a comparison of the effect of different ligands on reactivity.

10

Aromatic compounds were chosen to react with the metal ions.
There are several reasons why aromatic compounds make a good set
of test compounds. The size of the molecules provides a large cross
section for reaction. The aromatic ring for the most part does not
participate in the reaction. Complications that occur with alkyl
compounds, such as the chain length effect or chain branching, will
not be present. The chemistry is simpler, with usually only one
product being formed. In addition to these practical considerations,
the chemistry of aromatic compounds in these reactions had been
largely unexplored, except for a few isolated experiments.

The experiments described in chapter 5 give information on
ligand effects on metal reactivity. At the same time we would find
how the aromatic group influences reactions, in contrast to alkyl
groups, which have previously been studied by many other workers.

In the process of doing these experiments, it was noticed that
the Ni+ ions created from the three different organometallic
compounds had, in some cases, different reactivity. This raises the
question of whether the metal ions produced by El are formed in
different electronic states.

Two other projects will be reported in this dissertation. The

first, in chapter 3, concerns the reactions of aldehydes with Ni+ and

Co+. Unfortunately, some of the data of these experiments were
unambiguous, due to the resolution of our instrument. Still, some
conclusions about the chain length effect in the reactions of
aldehydes can be made.

A third project, presented in chapter 4, concerned the reactions
of Co+ with straight chain alkyl nitriles. (Nitriles are the same as
cyano compounds, except that there is a difference in nomenclature.
For example, hexyl cyanide is the same molecule as heptanitrile.) In

addition to the reactions of Co+, the reactions of hexyl cyanide were
studied with a variety of other metal ions. A few preliminary

11

experiments of metal— nitrile complexes were also carried out on a
triple quadrupole instrument. The results of these experiments will
be presented in chapter 4, also.

12

References

P‘E‘PPP!‘

>’

10.

ll.

12.
13.

14.
15.
16.
17.
18.
19.

20.

21.

22.
23.

Allison, J. Prog. Inorg. Chem. 1986, 34, 627.
Allison, J.; Ridge, D. P. J. Am. Chem. Soc. 1979, 101, 4998.
Hankinson, D. J.; Allison, J. J. Phys. Chem. 1987, 91, 5307.
Radecki, B. D.; Allison, J. Organometallics 1986, 5, 411.
Babinec, S. J.; Allison, J. J. Am. Chem. Soc. 1984, 106, 7718.
Schilling, J. B.; Beauchamp, J. L. J. Am. Chem. Soc. 1988,
110,15.
Schilling, J. B.; Beauchamp, J. L. Organometallics 1988, 7, 194.
Halle, L. F.; Armentrout, P. B.; Beauchamp, J. L. J. Am. Chem.
Soc.1981, 103, 962.
Schultz, R. H.; Elkind, J. L.; Armentrout, P. B. J. Am. Chem. Soc.
1988, 110, 411.
Foster, M. S.; Beauchamp, J. L. J. Am. Chem. Soc. 1975, 97,
4808.
Jackson, T. C.; Carlin, T. J.; Freiser, B. S. Int. J. Mass Spec. Ion
Proc.l986,72,169.
Jacobson, D. B.; Freiser, B. S. J. Am. Chem. Soc.1985,107,1581.
Jacobson, D. B.; Freiser, B. S. J. Am. Chem. Soc. 1984,
106,3891.
Allison, J.; Mavridis, A.; Harrison, J. F. Polyhedron, in press.
Stepnowski, R. M.; Allison, J. Organometallics, in press.
this work, chapter 3
Huang, S. K.; Allison, J. Organometallics 1983, 2, 883.
Radecki, B. D.; Allison, J. J. Am. Chem. Soc. 1984, 106, 946.
Lombarski, M.; Allison, J. Int. J. Mass Spec. Ion Proc.1985,
65 ,31.
Cassidy, C. J.; Freiser, B.S.; McElvany, S. W.; Allison, J.; J. Am.
Chem. Soc. 1984, 106, 6125.
Lombarski, M.; Allison, J. Int. J. Mass Spec. Ion Phys.
1983,49, 281.
Tsarbopoulos, A.; Allison, J. Organometallics 1984, 3, 86.
Tsarboploulos, A.; Allison, J. J. Am. Chem. Soc. 1985, 107,
5085.

Chapter 2

Experimental

Ion cyclotron resonance (ICR) spectrometry is particularly well
suited for the study of ion/molecule reactions. This relatively new
technique (less than three decades old) detects ions non-
destructively by measuring the frequency of the ion motion in a
magnetic field. The ICR experiment allows ion/molecule reactions to
occur because of the long residence times of the ions; ions are
typically contained in the reaction region for several milliseconds. It
also allows the user to determine if an ion is the product of an
ion/molecule reaction and what reactant ion led to the formation of
the product ion. ICR has been extensively reviewed in the
literature},2 An overview of the technique will be presented here.
An "operating manual" will be presented in Appendix A, that will
give details of the operating procedures used.

The ICR experiment takes place within a cell shown Figure 2-1,

 

 

 

 

 

 

 

 

 

 

 

 

\ \\ Z
_] V Tf—‘l V 1
o l ”—
\ T C '
1 n 7
Source Analyzer Collector

Figure 2-1 The ICR'cell

l3

14

which is contained in a high vacuum chamber. This is placed
between the pole caps of an electromagnet. The cell is divided into
three regions: source, analyzer, and collector. A small positive
potential (~0.5 V) is applied across the trapping plates of the cell to
contain positive ions in the center of the cell. A potential difference
is applied across the drift plates of the source and analyzer regions.
An electric field (E) applied between the plates is perpendicular to
the magnetic field (B). As a result, a force (F) is exerted on the ions,

given by the following equationzz
F=qE+q(va) (1)

where v is the velocity of the ions and q is the charge. This force
causes the ions to move in a direction perpendicular to both the
magnetic and electric field, through the ICR cell with a velocity (vd)
given by:

vd=E/B (2)

Given the typical length of an ICR cell, ~8 cm, and the typical
voltages used on the drift plates, ~0.5 volts, it can be calculated from
the drift velocity that the ions will remain in the cell for several
milliseconds.

Ions thus contained in the cell will precess with a frequency
given by f=qB/m, the cyclotron frequency. The frequency of
precession is proportional to the strength of the magnetic field and
inversely proportional to the mass of the ion, m. If radiofrequency
radiation is applied to the drift plates, at a frequency close to that of
the cyclotron frequency of the ion, the ions gain kinetic energy.

15

On gaining energy, the ions move into orbits with larger radii
(Figure 2-2).

rt Voltoqe
from morgtnol
Vdrttt OSCIHOlOr Vdrllt Vdrtlt

I [I] l

 

 

Source Analyzer Collector

 

 

I __l____ .1:

Figure 2-2. Ion motion in the ICR cell

Three modes of detection have been used to detect ions: fixed
frequency detection with magnetic field scanning; scanned frequency
detection with fixed magnetic field; and simultaneous detection of all
frequencies and interpretation of the complex signal via Fourier
transformation.

The detection technique used in the first commercially
available ICR instrument was to scan the magnetic field, B, bringing
ions into resonance with a fixed frequency oscillator. The parallel
drift plates serve as the capacitive element of an LC tank circuit of a
marginal oscillator. When the magnetic field is scanned such that the
frequency of the ion matches the frequency of the marginal

l6

oscillator, the ion is said to be in resonance and will absorb power
from the rf field produced by the marginal oscillator. It is convenient
to operate the marginal oscillator at a frequency of 153 kHz. At this
frequency, a change in 100 gauss corresponds to a 1 u change in
mass. By scanning the magnetic field, a plot of power absorption
versus B is obtained.

To improve the signal-to-noise ratio, signal modulation is used.
One of the trapping plates is pulsed to a negative voltage , and so
sweeps all the ions out of the cell, at a rate of 20 Hz. The modulated
signal is then analyzed with a lock-in amplifier. Because of the
modulation, slow scan times are necessary, requiring about 20
minutes to record a spectrum. A block diagram of the experimental
setup is shown in Figure 2-3.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

DRIFTa.
- TRAPPING "
VOLTAGES t L— ‘_ MAGNET
CELL
‘—
" MARGINAL ,_
— AMPLIFIER OSCILLATOR F'—
DETECTOR ‘—
MAGNET
. DRIFT MODULATION —'——* POIAER
MODULATION DUDE RECIRON SUPPLY
FEEFEME PULSED DOUBLE RESONANCE W SW'TCH ENERGY&
TOR MODULATION EMISSION
OURFENr
SWEEP
IRRADIATING CONTROL m
PHASE- XY OSCILLATOR
- SENSITIVE .__, REOOFDER ~——-~ f
DETECTOR
Figure 2-3. Block diagram for ion detection using a Marginal

Oscillator Detector

17

We performed experiments with this type of instrument in the
following way. An organometallic compound, such as Co(CO)3NO is

admitted into the ICR cell, at a pressure of about 3x10'6 torr. This
gas is ionized by 70 eV electron ionization, and a mass spectrum
obtained. In a similar manner, the spectrum of an organic molecule,
such as propyl chloride, is obtained. If the two gases were admitted
to the cell at the same time, peaks would appear in the spectrum, if
the total pressure was sufficiently high, that are neither in the
spectrum of Co(CO)3NO nor propyl chloride. One such product ion is

CoHCl+ at m/z 95. How is this ion formed?

To investigate this product, a "double resonance" experiment
can be performed. In this technique, the intensity of a product ion is
constantly monitored. A second radiofrequency signal is then
applied to the ICR cell. This signal, which is of sufficient strength to
eject ions. from the cell, is scanned to irradiate all the possible
precursor ions of the monitored ion. When an ion which is a
precursor to the monitored ion is ejected from the cell, the intensity
of the monitored ion decreases. In the above example, the intensity

of the CoHCl+ ion decreases when Co+ is ejected. This confirms that

CoHCl+ is formed in the reaction:

Co+ + C3H7Cl —> CoHCI+ + c3116 (3)

In this way, a double resonance spectrum is obtained, which
links ionic reactants to products. This experiment reveals much
information about ion chemistry, but is time consuming. Each double
resonance experiment takes about ten minutes to perform, and in a
complex system, many possible products could be formed. The data
for aldehydes and nitriles presented in this dissertation were
collected using a marginal oscillator detector.

18

Because of the long time required to complete an experiment, a
second detection method, frequency scanned detection, is desirable.
Frequency scanned detection (FSD) had been used by Wobschall as
early as 1963, and holds several advantages over marginal oscillator

(MO) detection.3 A spectrum can be taken much faster - in less than
a second. Also, an alternate double resonance scheme can be used, in
which all of the products of a reactant ion are identified. This is a
simpler experiment. For example, if the only information desired
was how one particular ion reacted, the information could be
obtained in one double resonance experiment, rather than the many
experiments it would take by identifying all precursors for all
products, the necessary procedure when using the marginal
oscillator.

The FSD used in our laboratory was designed, built, and
installed by Dr. John Wronka, and his design considerations have
been published.3 This FSD was used to collect the data presented in
Chapter 5, the reactions of metal ions with aromatic compounds. A
block diagram of the circuitry is shown in Figure 2-4. Figure 2-5
shows the block diagram for the complete experimental setup for
using the FSD.

l9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

90' SYNC “HE
[ DELAY
WAVETEK
F— uOOEL l89
x’ SIGNAL 1 BALANCED
SUMMER MODULATOR
DEMooULATosq
DOUBLE
RESONANCE
OSCILLATOR (Th
X - Y Low PASS
1 RECORDER Fl LTE R
RAMP
GEN ERAroq

Figure 2-4 Circuitry diagram for the Frequency Scanned Detector

 

 

 

 

 

 

 

 

 

 

 

 

PRE-
AWLIFIER

 

 

 

 

 

PHASE-
SENSITNE ‘

 

 

 

 

_LLL.

 

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

._s°°_"El

 

 

 

.m

 

DRWTS
TRAPPING CELL
VOLTAGES __‘F___J
;
8.2m
ENERGY 8r
EMISSION
WT
IRRADIATING mm
MacADIOS OSCILLATOR
ADC "‘———
1
UL. _._.
OSCILLO MaclNTOSH

BALANCE
‘_ m JEAPACITOR
- (—
RJNNBRET SIGNAL
SUMMER
SUPPLY
I
RAMP _. IRRADIATING
SCALER OSCILLATOR

 

 

Figure 2-5. Block diagram for ion detection using a Frequency
Scanned Detector

20

The following circuit description is excerpted from the

Frequency Scanned Detector Model #3 Instruction Manual:4 (p15-16,
"circuit theory")

"In the basic configuration developed by Wobschall a
broadband transformer is driven by a voltage controlled
oscillator (VCG) to provide two complementary radio
frequency (rf) signals. These two signals are sent to the ICR
cell and a balance capacitor connected to form a capacitance
bridge. The rf signal sent to the excitation plate of the ICR
cell provides an excitation source for the ions. The
complementary rf signal and the balance capacitor function
to provide a null point at the receiver plate. At the
cyclotron frequency, ions will disturb the balance of the
bridge and a voltage at the cyclotron frequency will be
developed at the null point.

The motion of the ions is a current in a parallel plate
capacitor (the ICR cell). Since current and voltage in a
capacitor are 900 out of phase the voltage developed at the
null point is phase shifted by 90° from the excitation
voltage.

Phase sensitive detection with a reference signal of the
same frequency as the excitation source but phase shifted
by 900 may be used to rectify the power absorption signal...

Frequency scanned operation of the bridge detector can
be greatly simplified if the excitation source already
provides a 900 reference. Fortunately, many commercially
available function generators provide this feature due to
their basic design.

This assumes that there is no phase shift in the devices

21

used in construction of the bridge detector. Every amplifier
has some phase shift with frequency. In fact even the
upper and lower frequency limits of an amplifier imply
phase shift. ’

With proper attention to impedance matching and the
proper choice of components it is possible to build a system
where the phase shift is corrected with a delay line."

After the FSD had been installed, it was necessary to determine
the optimum operating parameters for its use. The most important
parameter which affects the ion signal is the amplitude of the
irradiating voltage. The effect of irradiating voltage on spectra can
be seen in Figure 2-6a-e.

The amplitude of the irradiating voltage was varied and
spectra were recorded. Figure 2-6 shows two peaks, at m/z 105 and
107, from the spectrum of methylbenzyl alcohol. It is clear that the
peak intensity increases with irradiating voltage until a maximum is
reached with an irradiating voltage of ~120 mV. With higher
irradiating voltage, ions absorb too much energy and collide with the
cell walls. Thus the signal decreases. The noise level stays constant

with irradiating voltage.

22

 

 

 

 

e. ZSSmU

 

Figure 2-6 Effect of irradiating voltage on peak intensity

23

Another effect of the irradiating voltage is on the baseline.
Because the capacitance of the balance cell does not match exactly
the capacitance of the ICR cell, the baseline of the spectrum will not
be flat and level. A large irradiating signal will cause this effect to
be more noticeable. This can be observed in Figure 2-7a-f. These
figures show the baseline obtained for different levels of rf voltage
with no ions in the cell. The y- axis represents the dc voltage level.
The scan is from 100 kHz to 440 kHz, a typical range. As the
magnitude of the rf voltage increases, the dc level of the baseline
increases, and also the slope increases, rising towards the high
frequency (low mass) end of the scan.

A parameter of great importance is the amount of double
resonance radiation to use to selectively eject ions from the cell. The
effect of changing the double resonance irradiating voltage on a
selected ion can be seen in Figure 2-8a-f. These spectra show the
m/z 58, 60, and 62 isotopes of Ni+. The magnetic field was set at 16
kG and the double resonance oscillator set to 424 kHz. At low
irradiating voltages, the mass 58 ion is partially ejected. At 220 mV
irradiating voltage, the ion is almost completely ejected. Note that
the abundance of the mass 60 ion, at 410 kHz, is totally unaffected
by the double resonance irradiation. The fact that the isotope is
unaffected can be of great use in some experiments. In these
spectra, a spike appears at the exact frequency that the oscillator is
tuned to. This spike increases with the voltage level but causes no
problem. Occasionally an overtone of the irradiating frequency can
be seen at exactly three times the irradiating frequency as a small
spike, but this is easily recognized. An important consideration in
performing double resonance experiments is that an ion, in the
process of being ejected from the cell, absorbs power and gains
higher translational velocity. Reactions that would not normally
occur may be seen because of the

24

 

 

   

   

g 5 g g 1250 s = 1250
...... 750 750
”uny.uug.nu,.ung.nnng ....... 250 ............ ;~-;; ------ sgsje-eé~ 250
-S;é_fi_-L- ‘;““‘“T~" 3 “”"“ " 5 E
...... _250 _250

a 201nV7 b.551nN7

1250 1250

750 750

250 ............................................... 250

........................................ _250 Hun.”un.nuu.uun.nuun.”U”. _250
c. 120 mV d.17SmV

1750 1750

1250 1250

750 750

......................................... 250 . 250
e 225mV f. 255 mV

Figure 2-7 Effect of irradiating voltage on spectrum baseline

25

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
oooooooooooooooooooooooooooooooooooooooo

 

 

 

 

 

 

Figure 28 Effect of double resonance voltage on a selected ion

26

higher energy of the reactant ion. To minimize this problem, the
lowest irradiating power that causes sufficient ion ejection is used. A
rf level of 220 mV works well.

In order to improve the signal-to-noise ratio and to simplify
instrument operation, a MacADIOS (Macintosh Analog/Digital anut/
Qutput System) data acquisition system, coupled to a Macintosh Plus
computer, is used for signal averaging. The MacADIOS is multi-
purpose system that allows input or output of analog waveforms at
rates of up to 20 kHz. Digital input and output is also allowed.
Communication of acquired data to the Macintosh Plus is via a 500k
bits per second serial interface to the modem port. Software is
provided to allow the user to develop experiments to collect,
manipulate, analyze, and view data. Programming can be done in
three different languages: "MacADIOS Manager"; BASIC; and Aztec C.
MacADIOS Manager is a simple language developed to work with the
MacADIOS. Although the language is easy to use, it has certain
limitations that prohibit the user from performing certain types of
procedures. In these cases, BASIC or C, being more flexible, may
allow the user to perform the experiment. For example, while a
program in MacADIOS Manager allows the user to perform a nested
loop, it does not allow performing successive loops. However, this
type of program is very simple to write in BASIC.

The program has since been translated from BASIC into
MacADIOS Manager language and performs just as well, though it is
not as flexible.

A BASIC program, Scanner, (listed in Appendix A) was used to
collect the data presented in Chapter 5. The basic experiment
consists of first stepping the analog output of the MacADIOS to
provide a -5 to +10 volt ramp in 3150 discrete steps of 4.88
mV/step. This ramp then passes through a potentiometer serving as
a voltage divider. The magnitude of the ramp is here reduced to

27

about -0.5 to +1 volts, and then fed into the voltage controlled
frequency (VCF) input of a Tektronix FG 504 function generator.
The function generator, on receiving a negative voltage, will produce
a lower frequency than the dial is set for. Conversely, a positive
voltage will cause a higher frequency; in this way a frequency scan
in the range 100 kHz to 500 kHz can be produced. Any appropriate
frequency sweep can be produced. The frequency range scanned
will depend on the masses of the ions to be examined. The frequency
output is then attenuated and sent to the signal summer module. At
each step of the voltage ramp the output of the detector is amplified,
low pass filtered, and digitized. Twelve data points are taken, with
2.5 msec between each point. This makes the sampling period long
compared to the time of a 60 Hz wave period. Line frequency, 60
Hertz, is a common source of periodic noise, and this periodic noise
can be hard to remove.

After digitization, the data points are added together and
stored. The ramp is then stepped to the next value. A plot of
detector response versus step number is then produced. By knowing
the starting and stopping frequency, and also the magnetic field
strength, the step number can be converted to m/z. In order to
perform the double resonance experiment the procedure is repeated,
with one important difference. A gate signal is sent to a Wavetek
model 144 HF sweep function generator that is tuned to the
frequency of the ion that is to. be ejected. The experiment then takes
place as usual, except that one ion will not be present in the cell, i.e.,
the selected ion will be constantly ejected from the cell upon
formation. Consequentially, any ions produced by reactions of that
ion will not be present, either. When the spectrum with irradiation
on is subtracted from the normal spectrum, a subtraction spectrum is
produced. The subtraction spectrum should be flat and featureless,
except at the points where the irradiated ion and its products were.

28

At these points, negative dips appear. This is illustrated in Figure
2-9a-c.

glh.

 

 

 

Figure 2-9 Double resonance spectra

Spectrum A show part of the spectrum the 1:1 mixture of
Ni(PF3)4 and cyclopentanone at a total pressure of 6x10’6 torr. The
mass range is from 118 to 83 amu. The parent ion of cyclopentanone
occurs at m/z 84 and is off scale. Product ions that do not occur in
the spectrum of Ni(PF3)4 or cyclopentanone are observed at m/z 86,
112 and 114. In spectrum B, m/z 58, Ni+, has been ejected from the
cell (see Figure 2-8). Notice that the intensities of the products at
m/z 86, 112, and 114 has decreased. Spectrum C shows the
subtraction of spectrum B from C. Negative peaks at m/z 86, 112,

and 114 can be assigned as the products NiCO+, NiC4H6+, and
NiC4H8+. There is a slight positive peak at m/z 84 and 85 in this

29

spectrum. During the course of the experiment, the pressure of
cyclopentanone increased slightly, causing the peaks here to increase
from spectrum A to B.

To acquire data in this way, several programs were evaluated,
but did not produce adequate results. The program tried used the
fast scanning ability of the FSD. The irradiating frequency generator
was triggered to produce a 0.2 sec spectrum and 2000 data points
were accumulated at a 10 kHz digitization rate. This program had
the advantage of gathering data very quickly; however, it was
difficult to discriminate against high frequency noise. The bandpass
of the peaks was narrow and did not allow filtering.

A second attempt used a slower ramp and a 400 Hz digitization
rate. The slower scanning speed allowed filtering to reduce noise. A
drawback to this program was that the scan times produced by the
function generator were not exactly constant. As a result, the
position of the peaks in the irradiated spectrum and the
nonirradiated spectrum would shift relative to each other. When the
two spectra were subtracted from each other, the results could be
very erratic. If the peaks in the spectra mismatched by just a few
data points, the subtraction would produce a signal that looked like a
first derivative of a peak. Figure 2-10 illustrates this effect. When
spectrum A is subtracted from spectrum B, the result is line C.

A B

 

 

Cf\J__

Figure 2-10 Effect of mismatched peaks

30

The program used for these experiments, Scanner, overcomes
the problems in these unsuccessful programs. It uses a 4 minute
scan speed that allows low pass filtering. Controlling the frequency
of the function generator with the MacADIOS produces a very
reproducible frequency scan which minimizes problems caused by
mismatching peaks. A disadvantage is that the slow scan time causes
the experiment to take a long time, over eight minutes. A concern on
taking the data was that the detector may not stabilize sufficiently
before the data is taken. This program utilized time spent in
calculations as delay time, allowing the detector response to stabilize
as much as possible. Care was used that the two spectra, irradiated
and nonirradiated, were gathered in exactly the same way, except for
the gating of the double resonance oscillator. The presence of the
double resonance voltage does not appear to affect the spectrum in
any adverse way.

Several other features are used in collecting data. Adding data
points will increase the signal-to-noise ratio (S/N) by increasing the
signal more than the noise. Subtracting spectra has the result of
adding nothing to the signal but increasing the noise, not a desirable
effect. To avoid from increasing the noise by the mathematical
subtraction, a "moving window" convolution is performed on both
spectra. The spectra are convoluted with a step function. This digital
filtering removes much of the noise of the collected spectrum, but
also has the deleterious effect of decreasing peak intensity. Using
more points in the step function will reduce the noise more, so a
compromise must be reached between the noise reduction and
signal reduction. A three point step function works well.

To decrease noise before digitization, a passive filter with a 0.1
msec time constant is placed before the MacADIOS. This reduces

31

much of the high frequency noise produced in the amplification stage
of the phase sensitive detection module.

The ICR cell was changed in several ways in order to improve
its performance when used with the FSD. The filament had
previously been located in the center of the source region, but was
moved to 1/2 inch from the top of the source. This gives the ions a
longer residence time in the source region, and a greater chance to
react. Another modification was to reduced the distance between the
drift plates from 3/4 to 1/2 inch, changing from a "square cell" cross
section geometry to a "flat cell". This reduced the occurrence of
"trapping", a condition in which the ions remain in the cell for long
periods of time, up to many seconds. The ideal condition is when the
ions drift through the cell in several milliseconds, under single
collision conditions. In trapping, the ions may undergo multiple
collisions and make interpretation difficult. Spectra cannot be
recorded when trapping occurs. Reducing the distance between the
drift plates increased the field felt by the ions and forced them to
drift through the cell. A greater distance between the drift plates
results in poorer control over the ion movement.2

The chemicals used in these experiments were all commercially
available. They were degassed by multiple freeze-pump-thaw cycles
and used without further purification. Any impurities present would
be evident in the spectrum of the compound.

The necessary low pressure in the instrument is produced by a
4 inch diffusion pump with a liquid nitrogen cold trap, and a Ultek
20 US ion pump. Samples are admitted from a triple inlet
(individually pumped by a 2 inch diffusion pump and liquid nitrogen
cold trap) through Varian 951-5106 precision leak valves.
Approximate pressures are measured with a Veeco RG1000
ionization gauge. Pressure can also be measured by the Ultek ion

pump.

32

References

1. Budzikiewicz, H. Mass Spectrometry Reviews 1987, 6, 329.

2. Lehmann, T. A.; Bursey, M. M. " Ion Cyclotron Resonance
Spectrometry" Wiley : New York 197 6

3. Wronka, J; Ridge, D. P. Rev Sci. Instrum. 1982, 53, 491.

4. Wronka, J; Frequency Scanned Detector Model #3 Instruction
Manual

Chapter 3

Reactions of aldehydes with metal ions

This chapter will discuss the reactions of aldehydes with Co+,
Fe+, and Ni+. Aldehydes are an interesting class of compounds for
study because of the presence of the carbonyl functional group.
Aldehydes and ketones can serve as ligands in transition metal
complexes, in the liquid phase. Numerous transition metal catalyzed-
reactions involve aldehydes and ketones as starting products or
intermediates.1 Study of the how the metal reacts with the carbonyl
group may give insight how the metal is influenced by having CO as a
ligand. Aldehydes represent one of the first class of compounds
systematically studied by metal ion/molecule reactions.

In one of the first studies, Corderman and Beauchamp2
observed, for seven aldehydes, the following decarbonylation
reactions:

Nicp+ + RCHO —-> Nicpco+ + RH (1)

-—> NiCpRH+ + CO (2)

Decarbonylation reactions were not seen for formaldehyde, trifluoro-
acetaldehyde, acetyl chloride, acetyl bromide, acetone, and methyl
acetate. They concluded that decarbonylation reactions of NiCp+
were specific for aldehydes, since this reaction did not occur for the
acetone or methyl acetate. Unfortunately, they did not report the
results of Ni+ reacting with these compounds. The experiment was
performed in a conventional ICR mass spectrometer, with NiCp+

generated by 20 eV EI. This also produces Ni+, and its reactions

33

34

should have been observed occurring simultaneously with the
reactions of NiCp+.

Halle, Crowe, Armentrout, and Beauchamp3 performed the next
study of aldehydes using cobalt ions in an ion beam apparatus. In
this experiment the reaction products are formed with the cobalt
ions at a well defined kinetic energy. At low energy the only
reaction observed for formaldehyde and acetaldehyde was:

Co+ + RCHO -—> CoCO+ + RH (3)

More recently, I.K. Gregor4 has reported the reactions of

Cr(CO)3' and Cr(CO)4' with the straight chain aldehydes,
formaldehyde through heptanal, and also isopropanal and isobutanal.
All the aldehydes studied reacted to give alkane elimination. Loss of

CO was also observed, but reactions of 13C labeled acetaldehyde
indicated that the CO lost was a metal bonded ligand rather than
from the decarbonylation reaction. For propanal and larger
aldehydes, loss of H2, as well as loss of CO, occurs.

RCHO+Cr(CO)3‘ -> (alkene)Cr(CO)3' + H2+CO (4)
The compounds larger than butanal can further dehydrogenate:

(alkene)Cr(CO)3' —> (alkadiene)Cr(CO)3' + H2 (5)
The results indicate that when a sufficiently long alkyl chain is

present, the chain may wrap around to interact with the metal
center, resulting in H2 loss.

35

The reactions between the 70 eV EI fragments of Ni(PF3)4 and
Co(CO)3NO with the straight chain aldehydes propanal through
heptanal will be presented in this chapter. The reactions of hexanal
and heptanal with Fe(CO)5 will also be presented. In addition,
reactions of benzaldehyde with Ni+ and Co+ will be presented. The
results of these reactions are summarized in Table 3-1.

A basic question is whether the metal ion inserts into the R-
(CO)H bond, intermediate 1, followed by a a-H shift to the metal ion,
or whether the initial insertion is into the R(CO)-H bond,
intermediate 11, followed by a oc-R group shift.

I. 9 9 +
R-M -C-H R-C-M -H

I II

Thermodynamic estimates of the bond strengths indicate that
the R-CO bond is slightly weaker than the H-CO bond (82.5 kcal vs.

86 kcal).3 Insertion into C-CO bonds is a major pathway in the
reactions of ketones;3t5 however, the cross section for the reaction

between Co+ and formaldehyde is very small.3 This result indicates
that insertion into the CO-H bond is not a favored pathway. Insertion
into the R-CO bond would result in an intermediate in which the
metal ion would be bonded to the polarizable R group, producing a
stabilizing effect.

The results in Table 3-1 are, in some cases, unfortunately
ambiguous. For example, it is impossible to determine the difference
between the loss of C2H4 and CO, both with a nominal mass of 28 u.
The products listed in the table are believed to be the most
reasonable according to known mechanisms. While some of the

36

products may be ambiguous, many of the products can be explained.
For example, consider the reactions of Ni+. With propanal, butanal,
and benzaldehyde, all of the products can be explained by initial
insertion into the R-CHO bond. For example,

Ni+ + C3H7CHO —-) Nico+ + c3118 (17 %) (6)
—> NiC3H8+ + CO (38 %) (7)

For hexanal and heptanal, most of the reactions observed cannot be
explained by insertion into the R-CHO bond.

Ni+ + C6H13CHO .4 NiC3H6O+ + C4H8 (44 %) (8)
—) NiC4H6O+ + C3H8 (22 %) (9 )

A different mechanism must be used to explain why insertion
is favored at the middle of the alkyl chain rather than at the
functional group. If the alkyl chain wraps around to interact with
the metal ion, after the initial complexation with the functional
group, as shown in intermediate 111, then insertion at this location
can be explained.

c\ M+
C“ I: f?

C C
\C’ \H

111

37

The preferred sites of insertion for the metal ions with hexanal
and heptanal are shown in Figure 3-1. For comparison, the preferred
insertion sites for Co+ into hexanol6 is also shown. The C-C bonds are
labeled with numbers in this figure. The bonds that the metal ion
insert into are listed in order of occurrence. For example, in the
reaction of Ni+ with hexanal, most of the products are formed by
insertion into bond Q, with a smaller amount of product formed by
insertion into bond l_)_.

(PABCD .OA ABCDEFG
H-C-C----CCCC H-C-Al-BCCWCCCC HO-C-C-C-C-C-C-C

Fe+z D FeI D, E Co+: C, D, E. A
Ni+z C. D Ni”: D, C
Co+z D, C Co+: D, E

Figure 3-1 Bond insertion preferences of the different
metal ions into hexanal and heptanal.

The preferred site of insertion for aldehydes smaller than
hexanal appears to be adjacent to the carbonyl. Several general
trends are seen in these reactions. The number of products formed
is less than in the reactions of alcohols with similar chain length.
This may reflect that the aldehydes are more specific in the bonds
attacked. The presence of the double bond in the carbonyl may

38

make the intermediate more rigid, making fewer of the C-C bonds in
the alkyl chain susceptible to attack.

Differences in the reactivity of the different metals can also be
seen. Nickel forms more products than cobalt. This is unusual; for in
the nitriles and most other compounds, cobalt is more reactive than

nickel. In general, cobalt is more reactive than nickel, the former

having a lower promotional energy.7

The types of products formed from all the metal ions with the
aldehydes are similar. An exception is that Fe+ here does act
specifically to eliminate H2 from the large aldehydes. Also, the
formation of butadiene is specific for Fe+.

Ni+ also inserts into bonds closer to the functional group than
either Co+ or Fe+. Ni+ and Co+ also are more specific for attacking
bonds, both inserting into only two bonds, while Fe+ inserts into
three bonds. In contrast, these trends are not observed in the
nitriles. For hexyl cyanide, Ni+ inserts into one bond, Fe+ inserts into
two bonds, and Co+ inserts into three bonds. Co+ prefers to insert one
bond closer than the most preferred insertion site for both Fe+ or Ni+.
Clearly, conclusions about the reactivity of the different metal ions
are difficult to make, based only these three examples.

Because many of the products could not be unambiguously
identified, further work on aldehydes was not pursued. The use of
isotope labeled aldehydes would be of great use to identify these
products.

39

ME 3&2

£8 + +Oc=mO_z Lo
93 OO + +2262

N: + OO + +Ommvu_z Lo
a: £8 + 54:82

3va + +092 .8
83 Oemmu + +3322

.8383— OZ n rm .2

 

mam ++ez

GS i=2
:5 OO + 3:82
a: N: + OO + 5:82
:5 ozu + 3322
ES £3 + +092

E3732 Lmfizz

$43 :43" S5 N.
233 i=2
a 3 N: + +0552
:3 £8 + +022 .12

 

.m .z .m .z oz+LoO +028
33 £88
88 OO + too 33 OO + too :3 E88
:88 + N: + +9568 9: OO N + 3:88 53 OO 4 too 580
GS £8
£3 a: + +memOoO 2.3 OO + +wmmOoO 2.: N: + +0588
SEES + +Oc=mOoO 63 MEMO + +88 £3 £8 + +88 +8
omumamuvmmu omummmuvmmu omummumzu he Eeoeom

+42 93 +2 53’ sang—a CO 25:83—

Tm 2an

40

mm: + £2

OO + +c:cu2

02 + +mOU

00 + +mOU

:3 Lou
$3 +c:cOoO

Gog +A:Z

$3 mm: + MW:mO + +Oc:cu2

a3 ”:6 + +Oc:cO2
£5 MW:J + +Oc:mu_z
GS Oe:oO + +552

.M .Z
a: +mOUeU
8.3 Cu + +AOU

RE OO + :58 + +Ow:oOoO

:3 c:NO + +Om:nuou
83 c:mO + +Ow:eOoO
$3 ”:6 + +Oc:oOoO

mm: + £2

23 +22
33 «:8 + +Om:32
83 c:mu + +Oc:mO2
a: 04:3 + 2:82
2 3 Oc:mo + 3:52

rm .2
Gog +AOUOU
GNV 00 + +moU

33 N: + OO + +Oo_:cuou
:3 END + Ou + +Ow:eOoO
RS c:~O + 8 + +9538
33 c:mu + OO + +Oc:mOoO

83 Lee
a: 4:8 + +Ow:vOoO
83 c:NO + +Oo:cOoO
RS e:mO + +Oc:mOoO

 

o:Om:cO

o:OnA~:OVm:O

3:52:00 . Tm SEE.

o:O£~:Ovm:O

+922

+0200

+OUOU

+oU

41

GS 00 + Lo:

:43 OO + N: + 52:52
E: OO + c:~O + +Ow:mOo:
$3 OO + c:mO + +Ow:eOo:
88 OO + MW:mO + +Oc:eoo:

33 the:
£3 0:3 + +Ow:noo:
E: c:mO + Oxide:
GS MW:mO + +Oc:cOo:

83 N: + Emu + 54:62
£3 N: + Oc:mO + 5:64.:

Oamflmgvma

ANS OO + to:

83 OO + N: + +Oo_:cOo:
ES 8 + £8 + +Ow:4Oo:
$8 8 + c:~O + +Oc:eOo:

83 Kc:
RS N: + +Oo_:cOo:
$8 4:8 + +Ow:oOo:
38 e:~O + +Oc:eOo:
:3 N:4 04:8 + +c:cOo:
a: 956 + +c:mOo:

o:Oc®:OVm:O n:

82:38 .Tm 035.

LOUUL

+0:

42

References

1. Huang, Y-H.; Gladysz, J. A. J. Chem. Ed. 1988, 65,298.

2. Corderman, R. R.; Beauchamp, J. L. J. Am. Chem. Soc. 1976,
98,5700.

3. Halle, L. F.; Crowe, W. B.; Armentrout, P. B.; Beauchamp, J. L.
Organometallics. 1984, 3, 1694.

4. Gregor, I. K. Org. Mass Spectrom. 1987, 22,644.

5. Burnier, R. C.; Byrd, G. D.; Freiser, B.S.; J. Am. Chem. Soc.1981,
[03,4360

6. Tsarboploulos, A.; Allison, J. J. Am. Chem. Soc. 1985, 107, 5085

7. Babinec, S. J.; Allison, J. J. Am. Chem. Soc. 1984, 106, 7718.

Chapter 4

Reactions of Co"' with nitriles

The gas phase chemistry of nitriles with Co"' is discussed in this
chapter. This work was performed in 1984-85. As the results were
being prepared for publication, Helmut Schwarz at the Institut fiir
Organische Chemie, Technische Universitat Berlin, published several
reports on the reactions of Fe+ with nitriles.8 Ions formed by E1 on
Fe(CO)5 were allowed to react with nitriles in the source of a ZAB-
HF-3HF triple sector mass spectrometer. The ion corresponding to
Fe(RCN)+ was selected and collided with neutral argon at high energy,
8 keV. Based on the ions produced by the collision ("daughter ions"),
they concluded that Fe+ inserted into only C-H bonds of alkyl nitriles,
because no loss of alkane occurred. In later papers, they did report

alkane loss for large nitriles. In comparing the reactions of Fe+, Co+,

and NY“, they reported that, based solely on the percent of alkene
loss, a different mechanism must be operating for the reactions of

Fe+ than for Co+ and Ni+. .

In our data we see no evidence for different mechanisms for
the reactions of the different metals. It is true that the different
metals do have different product distributions (see Table 4-3).
However, we feel that the products can be explained by the chain
length mechanism which was proposed to explain the reactions of
alcohols and chloroalkanes. This mechanism allows for both C-H and
C-C bond insertion at sites remote from the functional group.
Tendencies for insertion into C-H or C-C may be dependent on the
metal, but not to the extent to indicate that two separate mechanisms

are operative.

43

44

Our study of the reactions of metal ions with nitriles follows
this introduction, in the form of a preprint of a paper which has been
accepted for publication in Organometallics. It includes a discussion
of our results and also comments on the recent work of Schwarz.
Again, his results are not directly comparable to ours, due to the
different experimental technique. His collisional activation results do
parallel our results, however.

One of Schwarz's Claims is that the iron ion adducts that are
formed in the source of his mass spectrometer are only in the form

Fe(RCN)+. Alternatively, the FeRCN+ ion could be in a form such as:
C3H6""Fe+ ..... C3H7CN

To investigate the structure of some of the products formed in the

reactions of Co+ and alkyl nitriles, collisionally activated dissociation
(CAD) experiments were performed on an Extrel Triple Quadrupole
instrument. In these experiments, selected ions collide with an
inert collision gas at low energies, only several eV. If the ion existed

as the adduct M(RCN)+ then a collision might produce only the M+ ion
with loss of the nitrile ligand. However, if the nitrile rearranged,
existing as two ligands bonded to the metal, then a collision might
show more products, such as one of the ligands bonded to the metal.

Co(CO)3NO and a nitrile of interest were introduced into the
chemical ionization (CI) source of the instrument and ionized by 70
eV E1. The metal ions formed react with neutral nitrile as expected.
The "adduct ion" CoP+ (P = parent) was mass selected with the first
quadrupole, and sent into the second quadrupole, the collision region.
The collision chamber was filled with argon at low pressure, 1x10'5
torr. About 50% of the selected ion dissociated at this pressure.
Ionic products were then mass selected with the third set of
quadrupoles. Higher pressure in the collision chamber resulted in a

45

greater percentage of the ions dissociating, but also resulted in
sequential reactions due to multiple collisions. The ion energy was
set at 2 eV. At higher energy, such as 10 eV, the main reaction
observed was CoP+ —> Co+ + P. The product of some of these
reactions are listed in Table 4-1. The table lists the normalized data
of the reactions. A branching ratio (B. R.) for only the reactions
which resulted in the breaking of a carbon-carbon bond is listed.

The products resulting from 2 eV collisions with argon are very
similar to those seen in the ICR experiment. For example, consider
the CAD spectrum of COP+ produced by the reaction of hexyl

cyanide and Co+. Five products are formed; in the ICR experiment,
four of these five products are formed by bimolecular reaction. This
suggests that the mechanism for product formation is the same for
both techniques. The energy gained on collision is may be enough to
cause rearrangement and dissociation of a neutral molecule.

Alternatively, if the COP+ ion is in a form such as:
C3H6“-'Co+ ..... C3H7CN

prior to the collision, the collision may act to break the bond to the
ligand which is held more weakly. In these experiments, it would
be difficult to say whether the products were formed by
rearrangement, simple dissociation, or a combination of both of these
processes. However, the large amount of Co+ produced on collision
indicates that at least some of the ions exist as the simple adduct
with no rearrangement. Collision of a rearranged molecule should
not produce the Co+ ion, unless both ligands are simultaneously lost,
or through multiple collision.

In addition to the CAD spectrum of the COP+ ion, other products
formed in the source could be studied. Table 4-1 lists the CAD

spectrum of m/z 142 and 128 formed ion the Co+/heptyl cyanide

46

system (by loss of ethylene and propylene, respectively).

The CAD spectrum of m/z 128 show that ethylene loss and
ethyl cyanide loss to be the only important reactions. In the ICR
experiment, ethylene loss is the major product in the reaction of Co+
and propyl cyanide. The CAD spectrum of m/z 142 gives branching
ratios similar to those found by the ICR reaction of Co+ and butyl
cyanide. The results produced in the ICR experiment are:

Co+ + C4H9CN —> CoC4H9CN+ (26 %) (1)
—> CoC4H7CN+ + H2 (41 %) (2)
—> CoC2H3CN+ + C2H6 (11%) (3)
—> CoCH3CN+ + C3H6 (21 %) (4)

The products from the CAD reaction of the ion at m/z 142, presumed

to be the reaction product COC4H9CN+ (neglecting the Co+ product)

are:
Coc4H9CN+ —> Coc4H9CN+ (67 %) (5)
—-) CoC4H7CN+ + H2 (13 %) (6)
—> CoCH3CN+ + C3H6 (20 %) (7 )

These results indicate that the the products of the initial reaction

between hexyl cyanide and Co+ have the structure of simple adducts.

Experiments were attempted in which acetonitrile was used as
the collision gas instead of argon. It was hoped that acetonitrile
would react with the selected ion via ligand substitution reactions.
For example, when Co+ reacts with hexyl cyanide to form the
product at m/z 170, some of the product may be in the form

(C4H8)Co+(C2H5CN). We hoped that reactions such as:

4 7
(C4H8)CO+(C2H5CN) + CH3CN —) (CH3CN)CO+(C2H5CN) + C4H8 (8)

and
(C4H8)CO+(C2H5CN) + 2 CH3CN -) (CH3CN)2CO+ + CzHSCN + C4H8 (9)

would be observed. These reactions would give products at m/z 156
and 141, respectively. Reactions of this type were not observed. The
results between using argon and acetonitrile as collision gases were
very similar.

When comparing the results of the two different experimental
techniques, one should realize exactly what factors can influence the
data. In the triple quadrupole experiment, the number and type of
products seen and their relative intensities are dependent on many
parameters, including collision energy and collision chamber
pressure. Conditions in the triple quadrupole must be set so that
enough ions dissociate, but these conditions also tend to increase the
amount of multiple collisions. The multiple collisions make data
interpretation difficult. Because of the many variables involved, it is
difficult to obtain a consistent set of data for these reactions. The
similarities in the results between the two techniques is interesting,
and in many cases CAD spectra could provide complementary data to
the ICR data.

48

Table 4-1

CAD reactions of Co+-(Nitrile) adducts

CAD of hexyl cyanide-Co+, m/z 156, at 1x10‘5 and 8x10'510rr collision

chamber pressure

 

1x10'5 torra 8x10"5 torr
ml; BEE gem 8.813 Per cent BR. NeutraLles
156 100.0 30.3
154 6.6 .35 43.5 .25 H2
128 11.1 .60 94.6 .54 C2H4
100 .9 .05 37.3 .21 2 C2114
59 17.7 100.0 C5H11CN

CAD of heptyl cyanide'Co+, m/z 170 at 11:105 torr collision chamber pressure

 

mjz Per cent B.R ICR B.R. Neutr§l_19_s_5_
170 80.3
168 5.0 .17 .11 H2
142 10.3 .36 .17 C2114
140 1.5 .05 .07 C2H6
128 8.7 .31 .34 C3116
112 2.8 .11 .00 C4H10
59 17.7 C6H13CN

a: pressure of collision gas b: B. R. = Branching Ratio

49

Table 4-1, continued

CAD of m/z 142, (C4H9CN-Co+) from heptyl cyanide-Col“ at 1x10'5 torr, 2eV

collision

 

m / z Per cent B.R ICR B.R. of Neutral loss
butyl-CN
142 100.0 .67 .26
140 20.0 .13 .41 H2
112 0.0 .00 .11 C2H6
100 30.0 .20 .21 C3H5
59 62.0 C4H9CN

CAD of m/z 128, (C3H6CN-Co+) from hexyl cyanide'Co+ at 11110-5 torr, 2eV

collision

 

m /z Per cent B.R ICR B.R. of Neutral loss
mole-CN
128 100.0 .77 .35
100 30.0 .23 .61 C2114
87 0.0 .04 CH3C‘N

59 72.0 C3H7CN

50

The Gas Phase Chemistry of First Row Transition Metal- and Metal-
Containing Ions With Alkyl Nitriles

ABSTRACT

The gas phase ion/molecule reactions of first row transition
metal ions with a series of alkyl nitriles is discussed, with the focus
on the chemistry of Co+. As the length of the alkyl Chain of the
nitrile increases, cleavage of C-C bonds far from the functional group
is observed, which supports an initial "end-on" interaction with the -
CN group. When a single ligand is attached to the metal, the resulting
chemistry appears to be very different from that observed for the
bare metal ion. The steric nature of the ligand effects is discussed.
The observed ion/molecule reactions yield similar results to those
suggested by collisional activation studies of transition metal
ion/alkyl nitrile complexes.

51

I. Introduction

For almost fifteen years, reports have appeared in the literature
concerning the rich chemistry that transition metal ions exhibit with
organic molecules in the gas phase.1 Such studies are begining to
provide fundamental insights into the structure/function

relationships of organometallic chemistry. For the prototypical
reactionZ
Co+ + C3H7Cl -) COC3H6+ + HCl (1 )

the proposed mechanism for formation of the obseryed products2
involves a sequence of events. A metal insertion into the C-Cl bond
is followed by a B-H shift to form the (C3H6)Co+(HCl) intermediate.
This then dissociates in a competitive ligand loss step,3'5 to yield the
observed products.

As larger molecules were studied, an additional mechanistic step
received attention - the complexation step in which the metal ion
and the molecule form an adduct before the "chemistry" (bond
cleavage and bond formation) begins. If this initial complex is
sufficiently long lived, preferred geometries of the complex may be
an important factor in determining the final product distributions -
by determining which bonds in the molecule the metal ion can

access.

We have discussed the chainlength effects in the reactions of Co+
with alcohols and alkyl halidesfi It appears that, if Co+ first
complexes with the polar group, and the chain is sufficiently long,
parts of the molecule that are "remote" to the functional group can
come into close proximity with the metal ion via 5- and 6-membered
rings. These geometries allow for the attack of C-C and C-H bonds far

52

from the functional group. This approach has also been successful in
explaining the collisional activation (CA) spectra of Fe+(A) complexes,
where A is an alkyne7 or nitrile,36-d and may even explain the
preference with which transition metal ions appear to attack skeletal
bonds in saturated hydrocarbons.9

Consider this first step of reaction (1) in which Co"' and propyl
chloride form a complex. One might envisage two extremes
concerning the interactions that would dominate in determining
which geometries would be most favored. One extreme is an
electrostatic interaction. If the Co+/C3H7Cl interaction is dominated
by electrostatics, characteristics of the molecule such as its dipole
moment, charge distribution, and polarizability would control the
situation. The other extreme would require a quantum-mechanical
description of the complex. This would be the case if dative
interactions (Co+<=:ClR) were dominant. We have recently reported
that the metal-ligand bonding in ionic species such as SCCO+ and
CrCO+ is predominantly electrostatic in nature.10 This has also been
suggested“.12 for NiCO+ and CuCO+. One would certainly expect a
Dewar-Chatt type bonding scheme in these systems, and it does
appear to occur in species such as NiOCO.11 However, when the metal
is charged, the total bond energy is dominated by the ion-dipole,
ion-quadrupole, ion-octopole and ion-induced dipole interactionsfi
Because of this, we are pursuing electrostatic models for these
initially-formed metal ion/organic molecule complexes.9.13

In light of the chain length effects observed for the reactions of
metal ions (M+) with alcohols, alkyl halides and alkanes,6 one might
expect other functional groups to behave differently in their
"directing ability" to parts of the molecule far from the functional
group. One group of interest is the cyano group, -CN. This work is

53

the first report on bimolecular ion/molecule reactions of transition
metal ions. with alkyl nitriles; however, Schwarz et al. have reported
a number of experiments in which oolljsiooal aotjvation (CA) was
performed on complexes generated from ion/molecule reactions
involving transition metal-containing ions and alkyl nitrilesSa-d. In
their first report83, a mixture of Fe(CO)5 and a nitrile, CnH2n+1CN (n
= 1-6) was introduced into the ion source of a mass spectrometer and
subjected to electron impact. In their instrument (of "BBB" design),
ions with the composition [FeCnH2n+1CN]+ were selected for the
collisional activation study. Such ions are formed by ligand-
displacement reactions such as

FC(CO)m+ "1' CnH2n+1CN —) FCCnH2n+1CN+ + m CO

(where m = 0,1). These ions undergo 8 kV collisions with helium,
and a mass spectrum of the fragment ions thus formed is obtained.
Their discussion focussed on two dominant processes, in which the
selected ion (M+), once collisionally-activated, forms two species, (M-
H2)+, and (M-C2H4)+. In addition to loss of H2 and ethylene, loss of
other species (e.g., C2H5) occurred to yield minor CA products. It
was proposed that the ions under study were complexes of Fe“ with
the intact nitrile, and that the metal ion initially complexes with the
functional group in an "end-on" fashion. Since labelling studies
suggested that H2 and C2H4 elimination occurs from the same part of
the alkyl chain, it was proposed that the losses of these two neutrals
occur as competing processes, through a common intermediate of the

type:

54

 

H \ H2
I B-H Shift ' elimination

 

C —C C-C cleavage\ C2 "4
’ elimination

 

A second paper,3b which highlights the use of fast atom
bombardment (FAB) to generate ions such as Fe"' from an FeSO4
target, showed that CA of analogous ions from larger nitriles leads to
the loss of C3,4 -containing neutrals to a greater extent than loss of
ethylene. This was discussed in a subsequent paper, 86 in which loss
of H2, alkenes and alkanes for larger nitriles is noted. For larger
nitriles, loss of ethylene and hydrogen could M be considered as
occurring via a common intermediate, since their origins appear to be
different parts of the alkyl chain. Analogous experiments with other
metal ions suggest that Co+ has a "shorter reach", as reflected in
alkane and alkene losses - the results for Co+ were more consistent
with a conventional metal insertion into C-C bonds / B-H shift
mechanism, while insertion into C-H bonds was presented as the first
step in the chemistry of Fe+.

Two possible explanations were cited regarding the "shorter
reach" observation for Co+ chemistry (relative to Fe"). One possibility
is that the M+-NCR bond length is shorter for M = Co+, than for M =

55

Fe+. This would affect the intermediate ring sizes that would be
accessible. (This explanation correlates with our resultle on MCO+
ions. The bond length order r(Fe+-CO) > r(Co+-CO) reflects the fact
that the ligand can get closer to Co"' because it is smaller than Fe+. In
general, first row transition metal ions such as Co+, Ni+ and Cr+ can
get closer to a ligand than can Fe+, because the former have only 3d-
electrons in their valence shell, while Fe+ has a 4s electron. The 48
orbital is more spatially extended than the 3d orbitals, resulting is
size differences as a function of electronic configuration.) A second
possibility may be related to back-bonding, which would affect the
rigidity/geometry about the C-C-N---M+ portion of the initially-
formed intermediate. The role of the metal's donor/acceptor ability
was pursued further.“ The possible distortion of the N-C-C bond to
angles less than 1800 due to a change in hybridization that would
follow (In-3* back-bonding was suggested. Results with a third
transition metal ion, Ni+, were similar to those presented for Co+,
leading to the proposal that back-bonding is least important for Fe+
of the three metals studied. The back-bonding aspect may be 131812;!
to metal-ligand bond lengths; those metals that can approach the
molecule more closely may provide a situation in which improved
orbital overlap for back-bonding occurs.

We present here the products of bimolecular ion/molecule
reactions for a variety of transition metal ions with a series of
nitriles. Reactions for several ligated metal ions are also reported.
These results provide additional insights into the chemistry of metal
centers with polar organic molecules containing one of the more
complex ligands of organic chemistry.

56

11. Experimental Section

All experiments were performed on an ion cyclotron resonance
(ICR) mass spectrometer that has been described elsewhere.14 The
instrument is of conventional design, and was used in the "drift
mode". All samples were degassed by multiple freeze-pump-thaw
cycles and were used without further purification.

Metal ions were made by electron impact on the corresponding
metal carbonyls. Co(CO)3NO was used to form Co"‘ and CoCO+.
Ni(PF3)4 was used to generate NiPF3+. The data presented here
were derived from ICR spectra obtained using 1:1 mixtures (metal-
compoundzorganic) at pressures up to 1.0 x 10'5 torr. From these,
ion/molecule reaction products were identified, and their precursors
determined using ion cyclotron double resonance methods. It should
be noted that there are ion/molecule reaction products for which
unique elemental compositions cannot be assigned. However, we
believe that the reactions as listed in Tables 4-2 and 4-3 represent
the chemistry that is occurring; they are self-consistent. Future
plans of labeling experiments should provide further information on
these complex systems.

57

o_.
5:.
mm.
av.

N_.
vw.
m_.
w..
~_.
mc.
Nc.

v:.
a:.

mu.
ac.
c_.

n_.
v_.
_N.
=_.
c:.

v_.
mm.

(‘1

ms.

m_.
cm.
n_.

cm.
5:.
>_.

=_.
__.
_w.

'01

N_.

an.
m_.

“‘1

3.5 . OO .. .zO...-=~..~-cCoOAI
3.5 . O... . .zcn-=~:~-cuoul
3.; . O; . 2.3-5:.-581

...__ . c; . .zO.-§:=OoOAt
:3 + 20.45.2201

mm. _m. m_. an. cu.
N_. m_. mm. av. em.

3.

2.
Z. .5.
.o.
S... E. a... .2

2..

.a.
w a m M a n

 

.zO..§:=OOOoOAIrle.:5...” . .236;

2055 . .N-c~...-=CoO1|

2.....0
3.5
3...;
5......
SEC
3.3

:5
N:O

+

+

+

+

. zcmasztcuooi
.zO..-§:..-=OoOAII
. z.....-..~....-=OoOAII
. z:...-=~.....-=..eoOAII
.zefcmzmcoooi
.zO..-=~.:~-..OoOA|I
. 2:25: _tcoooi
. 2.1-5: .EOoOAII

N._. .. .zO.-c~...=...oO11

.z....__~...__OoOA11
2;: . .......CoOA1,

3...; . ..z;..o:A|I

8:.52 :22 5.3 +0060 :5 +60 ..o 95:68:

«In. Beak.

 

1112.22.31... .

.

Cnv

58

30.5 +9.5.

“5...:

cm.
:x.

+5...

 

2. ._
w=. J
2... ..
2. 2. a...
an. em. on.
2.
+o2: +0000 3%.. n 2:.
..:...o
2. N:
.2 E... Z.
.2. 2..
2.. 2. 2..
2.
Z.
m... I. .N.
E. 2.. m...
12 luolul .9. . +2

202....0 5.3 +4... 2:. +2 .o 95.88..

MTV 2an

+ 5...... + .zn...__........
+ E... . +zc.__......
+ :10 + Buff...
.. + N: + +2.2.1....
._ . +z..2.........
+202...........

Aull 202...... + 2:
2022.0 its +._$_ .o 252......—

. .20....52 5%....
+ 20:... + +3.52
3.... + .2033...
...~o + +2051...
3.5 + +2.32...
N:... . +2023...
N3.. + .2333...
N... . +202...“...
+2033...

Alli 202...: . .2

59

III. Results and Discussion

Table 4-2 shows the ion/molecule reaction products observed
from bimolecular reactions of Co‘*‘ with CnH2n+1CN, for n = 1-8, and
with 2-cyanopropane. Branching ratios indicating the product
distributions are shown. A number of observations can be made on
these data:

1. Co"' apparently does not insert into the R-CN bond, except in the
case of 2-cyanopropane. It is not suprising that this bond would not
be attacked as easily as C-OH and C-Cl bonds, since it is the strongest
of these bonds.15 Also, more energy is required to eliminate HCN
from RCN than is required to eliminate, e.g., HCl from RC1, as
suggested in Table 4-4. Insertion into the C-CN bond in the
branched nitrile may be favored since, in contrast to its straight
chain analog, the branched compound should have a slightly lower C-
CN bond energy.16 Also, the insertion intermediate is expected to be
somewhat more stable when the C3H7- on the metal is a secondary

propyl group.

60

Table 4-4

Some Thermochemical Considerations (kcal/moi)

ML Alli _(li
D (CH3-X)15 82 91 117
AH (C2H5X —+ C2H4 + HX)15 17 11 33

Proton Affinities (PA) of Some Relevant Compounds24

.L_ M10221
H2 101.3
CH4 132.0
C2H4 162.6
CO 141.9
HCN 171.4

CH3CN 188.4

61

2. In these reactions for CnH2n+1CN, the chemistry obviously changes
as 11 increases. Dihydrogen elimination (following insertion into C-H
bonds) does not occur for n<4; as It varies from 4 to 8, the importance
of the Hz-elimination pathway decreases.

3. Attack of the terminal C-C bond is rarely observed, and is only a
minor product for the largest nitrile studied here.

4. Many of the reaction products appear to be typical for
mechanisms in which insertion into C-H and C-C bonds occurs.
Insertion into various C-C bonds may lead to an intermediate of the
type R'-Co+-R"CN, where both R' and R"CN posess B-H atoms that can
shift - thus elimination of ethylene or ethane, propene or propane,
etc. is observed. We can interpret the data based on the metal
insertion/B-H shift mechanism, however the chemistry could
certainly be more complex. We cannot, for example, distinguish
between loss of ethane and loss of {ethylene + H2}. The patterns of
losses of CnHzn vs. CnH2n+2 we believe are consistent with a simple
mechanism in which the CnH2n+2 may be lost as an intact alkane.
This has been proposed by Schwarz80 for the chemistry involving
Co+, while loss of CnH2n+2 from Fe+ reactions may in fact be {CnHzn +
H2} loss.

5. The reactions as listed follow the proton affinity rule,3 consistent
with retention of the more strongly bound ligand(s). Proton affinities
for molecules representative of those types of species involved in the
reactions are given in Table 4-4.

Table 4-3 shows, for contrast, the reaction products for a
number of other metal ions with a selected nitrile, C6H13CN. As is
typically observed, the Fe+, Co+ and Ni+ ions each form a variety of
products, many of which are common to all three metals. In terms of

62

attack of the skeletal bonds, it is interesting to note that ethylene
loss dominates for Fe+, propene loss dominates for Co+, and the two
processes occur with approximately equal probabilities for Ni+. In
contrast to these three metals, Mn+ and Cr+ are much less reactive,
and do not appear to attack C-C bonds.

The M+(RCN) Complex: The reactions reported here are best put
into perspective by comparison with those observed for other polar
organic compounds. First, a few comments are in order on the initial
interaction between a transition metal ion and an alkyl nitrile.
Schwarz et al.83 have discussed the side-on (I) and end-on
coordination (II) possibilities for the -CN group, and suggest that the
chemistry is consistent with an end-on

3+

C—R M+---- :NEC—R
11

:N

II—

description of the complex. (A similar geometry has been proposed
for protonated molecules such as protonated HCN17.) This same
conclusion is reached if one considers the M+-RCN complex to be
electrostatically bound. If the molecular polarizability is described by
the approximate bond-polarizability model, the C=N bond is more
than twice as polarizable along the bond axis, as it is perpendicular
to the internuclear axis.18 Thus, a greater ion/induced dipole
interaction energy would result in the end-on configuration. The
charge distribution in an RCN molecule would also favor an end-on
complex,19 since the dipole moments of nitriles are large19 (e.g.,
dipole moment of CH3CN = 3.92 Debyes), with the nitrogen at the
negative end of the dipole. Thus, it appears that the most stable

63

configuration of a transition metal ion/alkyl nitrile complex would
posess a linear geometry about the C—C--N—M+ atoms (based on
consideration of features of the functional group alone.) This
suggests another factor that prevents M+ insertion into the C-CN
bond - inaccessibility due to geometric restrictions.

For small nitriles such as acetonitrile, one may expect the end-on

geometry (II) to dominate.

However, for larger alkyl groups, the substantial polarizability of R-
may make the geometry shown in III favored, bringing the metal
ion to a position where it can form smaller ringed intermediates, not
unlike those proposed for saturated polar compounds.6.13

A second possibility is that, upon complexation, sufficient energy
is released to essentially promote the nitrile to a triplet state -
converting the triple bond to a double bond, with the concurrent
formation of an M-C or M-N o-bond yielding a structure such as IV,
again drastically changing the geometric relationship of the metal to
the rest of the molecule. We note that, for the HCN molecule, the first
triplet state (3A") is 4 eV above the ground state.20 If this is typical
for alkyl nitriles, the singlet-triplet gap may be prohibitively large

for this mechanism to be operative.

64

The Chemistry of M+: To discuss the implications of the possible
interaction geometries about the nitrile group, it is useful to consider
the skeletal bonds that are attacked, and the order of preference for
which insertion appears to occur, based on the final product
distributions. We have presented previously the approach used to
determine the order of preference for insertion,6 which is shown
here for the C8H17X compounds (X = CN, OH, Cl) for the Co'*’ ion. A
bond labeled ® is that bond which, upon insertion, leads to the
largest fraction of the observed products. The bond whose cleavage
leads to the next most significant fraction of the products is labeled

@,
C§>C§>CQDC§>C§>C _ C _ C _ C E N
C_C_C@C§>C@C@C_C_ CH
C-C_C§>C_C@C@C_C<‘PC.

If one compares the products and their distributions for the
nitriles and the corresponding alcohols and alkyl halides,6 it is
apparent that the skeletal bond which is cleaved to a greatest extent
(labeled @) is farthest from the functional group in the nitrile cases,
which would be expected from an end-on complexation model.
Larger rings would be required to bring atoms of the alkyl group
close to the metal. On the other hand, the effect is not that dramatic
for the smaller nitriles. In fact, cleavage of many of the skeletal
bonds would not be predicted if a fairly constrained end-on

65

complexation is maintained. There are some differences in reactivity
unique to the nitriles when compared to other polar compounds, but
not as dramatic as might be expected from the initial formation of
the proposed end-on complex.

The data suggest that, while alkyl halides and alcohols may
prefer to interact with parts of the alkyl group via 5— and 6-
membered ring intermediatesG such as (V) and (VI):

in, in, 312— 5112
X ‘ v.1

Larger nitriles appear to preferentially utilize eight membered ring-
intermediates. For example, the fact that H2 elimination is observed
only for nitriles with n=4 or larger may suggest that intermediate
(VII) is important.

66

”26+IQa
+ '-. +
...... H... , 1
H ..... M I... N "“2 c ..- M u. N

.-' \ H' _,.-' E \
H2C c H265 c
\ / \. , ./

H2 ”2

.11 111

Intermediate VIII could explain why certain C-C and C-H bonds are
most often cleaved. For example, structure VIII suggests that, via an
8-membered ring, the e-C comes into close proximity to the metal,
making the two C-C bonds that contain that e-C candidates for
insertion (as well as the e-C-H bonds). Consider the data for C8H17CN
in Table I. Insertion via intermediate VIII into the C(8)-C(e) bond
leads to loss of butene, which represents 24% of the total products;
insertion into the C(e)-C(§) bond leads to loss of propene and propane,
which represent a total of 32% of the products formed. A similar
analysis of the C7H15CN data also suggests that intermediate VIII
can account for a substantial fraction of the products formed (>50%).
However, for the shorter chains, attack of C-C bonds closer to the
functional group accounts for a substantial fraction of the products.
Both (VII) and (VIII) involve eight membered rings. Thus, the end-
on geometry is certainly influencing, but need not dominate the
observed chemistry. Presumably a distribution of cyclic

67

intermediates of various ring sizes may be sampled by the metal,
presenting a variety of "choices". Whether attack of a skeletal bond
will follow or not depends on additional considerations, such as the
subsequent thermodynamics of each insertion process.

Ligand Effects: Tables 4-2 and 4-3 show the effects of the
addition of a carbonyl ligand to the metal center on the products
formed, and their distributions. The effects are dramatic. What is
typically considered to be a "ligand substitution" process

MCO+ + RCN -) MRCN+ + CO (2)

dominates. (For larger reactant molecules, the neutral loss
corresponding to 28 u could also represent ethylene loss, with
retention of the CO ligand - however the binding energy of ethylene
to M+ appears to be slightly larger than that for €04.21) What is most
obvious from Table I is that very little chemistry is apparent for
CoCO+ except with the largest nitriles. Attack of C-C bonds much
farther out on the alkyl chain dominates when a CO is bound to the
metal. The order in which skeletal bonds are preferred for insertion
is dramatically altered. Only in the octyl nitrile is C-C cleavage
evident - and then only involving the two skeletal bonds most
distant from the functional group.

We propose two possible scenarios in which the CO could play a
role leading to this dramatic change in product distributions for a
metal center when this ligand is attached. Both are steric in nature.
In the first case, the CO prevents the metal from getting close to the
more interior C-C bonds, due to steric effects, shown schematically in
IX.

m ‘3
' l

steric . H+
interactions Hz'c ------ .. N
H2C\ /c
H2
I__

In a second scenario, the CO does not interfere, allowing the
metal to "sample" the same preferred complex geometries as it does
when it is ligand-free. However, the relative stabilities of the
insertion intermediates determine which bonds will actually be
attacked. For example, in the case of hexyl cyanide, the Co+ can
apparently attack the central C-C bond to form the insertion

intermediate:

+

3 7

This intermediate must, then, be both geometrically and
energetically accessible. However, when a CO is present, the

analogous intermediate

69

0
u:

C

l+
C H ——Co-——C H CN

37 36

may be less stable due to interactions of the fairly large alkyl groups
with the third ligand, CO. The most favored insertion intermediates
might then become those in which one of the o-bonded groups on the
metal is relatively small (such as H or CH3), consistent with the
reactions observed for CoCO+ with octyl nitrile, and the data for the
other MCO+ ions in Table 4-3. Similar behavior has been observed in
other systems. A dramatic example is in the chemistry of Ni+, NiCO+
and NiPF3+ with heptane.22 Ni+ inserts into a number of skeletal
bonds in heptane, resulting in the elimination of C2, C3 and C4-
containing neutral molecules. Dihydrogen elimination is also

observed. In contrast, NiCO+ forms two products

NiCO+ + n-C7H16 —-> NiC7H16+ + CO (3)
-—) NiC7H14+ + H2 + CO (4)

When the ligand is changed to PF3, no chemistry at all is apparent,
except for ligand substitution.

NiPF3+ + n-C7H16 -—> NiC7H16+ + PF3 (5)

This would be the trend expected for the types of steric effects
proposed, since the cone angle23 for PF3 , which reflects its relative
size as a ligand, is larger than that for CO, suggesting that its steric
effects would be more dramatic.

70

Relationship to Prior Work: Two points are to be made here. The
first is that results presented here and elsewhere suggest that CA
experiments may be difficult to interpret, and need not correlate
with the analogous bimolecular ion/molecule reactions. The second
point is that, while there need not be similarities between these very
different experiments, they do provide similar results in the case of
Co+/RCN chemistry. Schwarz et al.8a suggest that these ions are
formed by ion/molecule reactions of FeCO+ and Fe(CO)2+, and the
mechanism by which the ions are formed has "no implications for the
chemistry of mass-selected Fe(RCN)+ ions".821 The implication is that
an apparent ligand substitution reaction is a literal ligand
substitution, and that all of the chemistry occurs after the energetic
collision with a neutral helium atom. Can it be assumed that
reactions such as

FeCO+ + A —> FeA+ + CO (6)
leave A inteact? If not, what possible forms could FeA+ have?

Such reactions cannot always be considered literally as ligand
substitutions. The observation of such a reaction only indicates that
all of the atoms of the neutral reactant are now associated with the
metal ion. For small molecules such as methanol, it has been
suggested2 that such processes are not simple ligand displacements.
Consider the following:

Feco+ + CH3OH —+ FeCH3OH+ + co (7)
FeCH3OH+ + CH3OH —> FeOHCH3OH+ + CH3 (8)

71

This sequence of reactions has been interpreted as an indication that
the product of reaction (7) has the form CH3-Fe+-OH, suggested by
the subsequent reaction in which the CH3 is displaced. Labeling
experiments show that the methyl group lost in (8) originates
exclusively from the m, negating the occurrence of (8) through a
symmetric intermediate involving two equivalent "methanol ligands"
such as:

CH3OH - - Fe+ - - HOCH3

Similar experiments2 involving MC2H5I+ ions formed by reaction
of MCO+ with ethyl iodide suggest that the ion exists, at least in part,
as M(C2H4)(HI)+. Thus, ligand substitution need not guarantee that

the incoming molecule remains intact.

Ions of the type MCO+ do react with organic molecules, and
induce rearrangement to smaller molecules. This is known. Consider
the molecule ABC, which can be converted into AC and B by a metal
ion. If this also occurs for MCO+, the complex that finally dissociates
to yield the observed products is M+(CO)(AC)(B). The ligands that are
lost are those with the lower metal-ligand binding energies.
Frequently, the ligand lost is a small alkane - which would be
expected since the alkane is a poorer base than the unsaturated
molecule left behind, or CO. When MCO+ reacts with an alkyl
chloride, one may see loss of HCl with retention of CO, again
consistent with the relative proton affinities of the molecules
involved. However, it would also be appropriate in some cases for
M+(CO)(AC)(B) to lose one ligand, CO. Thus, for larger ABC species,
the ion current at an m/z corresponding to MABC+ may contain a

collection of structures such as M+(AC)(B).

72

Suppose MCO+ reacts with a nitrile and inserts into a C-C bond

forming the intermediate

0
111

C

l
c H5 —Co+ —(CH2 )nCN

2

If n 2 2, there are two groups from which a B-H may shift. These
two possibilities yield two intermediates, M+(CO)(C2H4)(H(CH2)nCN)
and M+(CO)(C2H6)(CnH2n-1CN). Based on proton affinities of these
species, one might expect the former to lose ethylene or CO, while the
latter would preferentially lose the alkane. Thus, the "ligand
substitution" species MCn+2H2n+5CN+ may take the form
M+(C2H4)(H(CH2)nCN), but not M+(C2H6)(CnH2n_1CN). This may
explain why collisional activation of some [FeRCN]+ ions leads to loss
of C2H4, but not loss of C2H6. The discussion of the Fe+/CH3O H
system above may explain why CA spectra also show loss of C2H5.
Thus, the question remains as to whether the energetic collision of a
transition metal-containing ion such as [MRCN]+ leads to a "collisional
activation" or to a collision-induced dissociation. The collisional
activation experiment need not directly parallel the reactivity of the
bare metal ion with the neutral substrate, since the ion selected for
study need not contain the intact organic molecule. This depends on
the chemistry by which the selected ion is formed. If there are,
indeed, ligand effects that alter the reactivity of MCO“ relative to
that for M4”, then the distribution of the various MRCN+ geometries
may not resemble the ion/molecule reaction products for M+ with
RCN. In fact, the combination of the ion/molecule reaction results
and the CA data can provide interesting contrasts, and both are
useful for understanding the observed chemistry. The minor

73

processes observed in the CA experiments (such as loss of C2H5) may
provide more information than the dominant products, since such
minor products are not formed in exothermic bimolecular processes.

While such possible complications should be expected, the
results of CA80 of [CoRCN]+ ions are very similar to the chemistry
presented here. There are, of course differences in product
distributions, and in the observation of the least abundant products,
however the correlations are substantial. For example, no reaction
products were observed for Co"' with CH3CN and C2H5CN, and there
was no evidence for chemistry between these species in the
analogous CA experiments. In the case of C3H7CN, loss of C2H4 is
observed in both experiments, however no loss of Hz was observed
in the ICR experiments. In the larger nitriles such as C6H13CN and
C7H15CN, it is interesting to note that loss of methane is reported
from the CA experiments, but not in the ICR experiments. There are
many reasons why this may occur. The ICR experiment may be less
sensitive in detecting minor products. Also, the reaction may be
slightly endothermic - but sufficient energy is available in the
collisional-activation to allow it to occur. However, since the CA
results parallel the ion/molecule reaction results presented here, ions
of the type [CoRCN]+ formed by ligand substitution processes are
apparently simple adduct ions, with the RCN still intact to a great
extent.

IV. Conclusions

The -CN group appears to have unique directing abilities in the
reactions of transition metal ions with alkyl nitriles. An end-on
geometry of the initially-formed complex appears to explain at least
some of the reactivity that follows, in which atoms of the alkyl chain

74

far from the functional group are involved. When a CO is attached to
the metal ion, the reactivity appears to change dramatically. This
may be due to steric interactions at a number of points along the
reaction coordinate. Before the "reactivity" of MCO+ with alkyl
nitriles can be evaluated, further work must be undertaken to

determine the structure(s) of the various MRCN+ product ions.

V.

.4"

9°."P‘E"

10.
11.
12.

13.

14.
15.

75

References

Allison, J. Prog. Inorg. Chem. 1986, 34, 627.

a) Allison,J.; D.P. Ridge, D.P. J. Am. Chem. Soc. 1976, 98, 7445;
b) Allison, J.; Ridge, D.P. J. Am. Chem. Soc. 1979, 101, 4998.
Tsarbopoulos, A.; Allison, J. Organometallics 1984, 3, 86.
Weddle, G.H.; Allison, J.; Ridge, D.P. J. Am. Chem. Soc. 1977, 99,
105.

Foster, M.S.; Beauchamp, J.L. J. Am. Chem. Soc.1975, 97, 4808.
Tsarbopoulos, A.; Allison, J. J. Am. Chem. Soc. 1985, 107, 5085.
Schulze, C.; Schwarz, H. Chimia 1987, 41, 29.

a) Lebrilla, C.B.; Schulze, C.; Schwarz, H. J. Am. Chem. Soc. 1987,
109, 98; b) Drewello, T.; Eckan, K.; Lebrilla, C.B.; Schwarz, H.

Int. J. Mass Spectrom. Ion Proc. 1987, 76, R1 ; c) Lebrilla, C.B.;
Drewello, T.; Schwarz, H. J. Am. Chem. Soc. 1987, 109,

5639; d) Lebrilla, C.B.; Drewello T.; Schwarz, H. Int. J. Mass
Spectrom. Ion Proc. 1987,79, 287.

Hankinson, D.J.; Allison, J. J. Chem. Phys. 1987, 91, 5307.
Allison, J.; Mavridis, A.; Harrison, J.F. Polyhedron, in press.
Bauschlicher, C.J., Jr. J. Chem. Phys. 1986,84, 260.

Merchan, M.; Nebot-Gil, I.; Gonzalez-Luque, R.; Orti, E. J. Chem.
Phys. 1987, 87, 1690.

Hankinson, D.J.; Hooper, E.J.; Allison, J. "Modeling the Electrostatic
Interactions Between Transition Metal Ions and Polar Molecules";
Presented at the 35th ASMS Conference on Mass Spectrometry
and Allied Topics, Denver, Colorado, May 24-29, 1987.

Radecki, B.D.; Allison, J. J. Am. Chem. Soc. 1984, 106, 946.
Thermodynamic information taken/derived from data in: a)
Franklin, J.L.; Dillard, J.G.; Rosenstock, H.M.; Herron, J.T.; Draxl,
K.;Field, F.H. Nat. Stand. Ref. Data Ser., Nat. Bur. Stand. 1969, 26.;
b) Rosenstock, H.M.; Draxl, K.; Steiner, B.W.; Herron, J .T. J. Phys.
Chem. Ref. Data 1977, 6, suppl.1.

16.

17.

18.
19.
20.
21.

22.

23
24

76

Thermochemical data in ref. 15a shows this to be the general
case, while the difference is not large. If

AD = {D(i-C3H7--X) - D(n-C3H7 --X)},
the difference in C-X bond energies, typical values for A D are -2
kcal/mol for X = OH, -4.1 kcal/mol for X = NHz, -3.8 kcal/mol for
X=N02.
See, for example: a) DeFrees, D.J.; McLean, A.D. J. Am.Chem. Soc.
1985, 107, 4350; b) Pearson, P.K.; Schaefer, HR, 111
Astrophys. J. 1974, 192, 33.
Denbigh, K.G. Trans. Faraday Soc. 1940,36, 936.
Pople, J.A.; Gordon, M. J. Am. Chem. Soc. 1967, 89, 4253.
Ditchfield, R.; DelBene, J.; Pople, J.A. J. Am. Chem. Soc.1972, 94,
4806.
Hanratty, M.A.; Beauchamp, J.L.; Illies, A.J.; vanKoppen, P.;
Bowers, M.T. J. Am. Chem. Soc. 1988,110, l.
a) Radecki, B.D. Ph.D. Thesis, Michigan State University, 1985;
b) Radecki, B.D.; Allison, J. presented at the 33rd ASMS
Conference on Mass Spectrometry and Allied Topics, San Diego,
CA, May 26-31, 1985.

. Tolman, C.A. Chem. Rev. 1977, 77, 313.
. Lias, S.G.; Liebman, J.F.; Levin, R.D. J. Phys. Chem. Ref. Data,l984,

I3 , 695-808.

Chapter 5

Reactions of Ni+ and NiL+ with aromatic compounds

Aromatic compounds show very interesting chemistry with
metal ions. There are several reasons why aromatic compounds
show good reactivity. The size of the molecules provides a large
cross section for reaction. The aromatic ring for the most part does
not participate in the reaction. The chemistry is simpler, with
usually only one product being formed. In addition to these practical
considerations, the chemistry of aromatic compounds in these
reactions had been largely unexplored, except for a few isolated
experiments. For these reasons we thought that aromatic compounds
would be a good class of compounds to explore the role of ligands on
the reactivity of metal ions.

Our observations on these reactions follow this introduction, in
the form of a paper which has been submitted to the Journal of the
American Chemical Society. The complete set of data compiled
follows the copy of the paper, listed in Table 5-7. In addition to the
reactions of aromatic compounds, the reactions of cyclopentanone,
cyclohexanone, and dimethyl sulfoxide were also studied. These
reactions are listed in Table 5-8. The results of these experiments
will not be discussed.

While studying the reactions of different NiL+, we observed
that Ni+ ions, produced by 70 eV electron ionization on the
organometallic compounds Ni(CO)4, Ni(PF3)4, and NiCpNO, have
somewhat different reactivity. For example, Ni+ produced from

NiCpNO reacts with iodobenzene to form only the adduct NiC6H51+,

77

78

while Ni+ produced from E1 Ni(CO)4 and Ni(PF3)4 reacts according to:

Ni+ + C6H51 —> NiC6H4+ + HI (1)
—> Nic6H5+ +1 (2)

The ratio of reaction 1 to 2 is 1:1 for Ni+ produced from Ni(CO)4 and
1:2 for Ni+ produced from Ni(PF3)4.

In studies of half of the aromatic compounds (11 out of 22
cases studied) the Ni+ ions from all three sources react identically. In
general, the Ni+ ions produced from Ni(CO)4 and Ni(PF3)4 react very
similarly, with slight differences in the branching ratios. However,
Ni+ from NiCpNO tends to form adducts more often than Ni+ formed
from the other two compounds. For example, Ni+ made from NiCpNO

forms the adduct with benzene, toluene, and aniline while Ni+ from
the other sources is unreactive. For nitrobenzene, ethylbenzene,

benzylamine, acetophenone, and methyl benzoate, Ni+ made from
NiCpNO forms the adduct as well as the products that are formed

from the other two Ni+ ions. With benzyl alcohol, Ni+ from NiCpNO
forms one product, while Ni+ from the other sources forms two. As
stated above, Ni+ made from NiCpNO forms only the adduct with
iodobenzene, while the other Ni+ ions form other products. In these

reactions it seems that Ni+ from NiCpNO is somewhat less reactive
than the other two nickel ions. The one exception to this trend is
phenol, in which Ni+ from NiCpNO is unreactive, while the other two
ions form the adduct.

For the reactions in Table 5-8 (reactions of DMSO,
cyclopenatone, and cyclohexanone), the distinction between the
different metal ions is not as clear. Comparison of the reactivity of

Ni+ from Ni(CO)4 and Ni(PF3)4 with alkanes show that these ions can
78

79

at times exhibit different reactivity.

It seems possible that the Ni+ ions formed by electron
ionization are in different excited states. It has previously been
reported that some portion of the Ni+ ions produced from Ni(CO)4 is
in the excited state.1 This was based on the observation of charge

transfer between Ni+ and Ni(CO)4.
Ni++Ni(CO)4 —9 Ni + Ni(CO)4+ (3)

The ionization potential of Ni is 7.63 eV, from spectroscopic
experiments.2 The ionization potential of Ni(CO)4 is 8.35 eV. For

reaction 3 to occur, Ni+ must have 0.72 eV, or ~16 kcal/mol excess
energy above the ground state.

Recently, Bowers has shown direct evidence for the existence of
excited Ni+ ions produced by E1 on Ni(CO)4.3 In an experimental
technique called high-resolution translational energy loss spectro-
scopy, mass selected ions are collided with a inert collision gas. Ions
that are in the ground states can undergo inelastic collisions, and
convert translational energy into electronic energy. Ions that are in
excited states can undergo superelastic collisions, converting
electronic energy into translational energy. In this experiment,
Bowers was able to identify a peak 1.6 eV above the main peak for

Ni+. This was assigned as the 2F excited state of Ni+, derived from

the 4313d8 configuration.

Unfortunately, data for ions produced from Ni(PF3)4 and
NiCpNO have not been published at this time. Few labs are equipped
to perform these experiments, due to the high resolution necessary.
Another experimental technique that may be able to directly show
the presence of excited states is electron ionization fluorescence
(EIF). Very few labs are equipped to do this relatively simple

80

experiment, in which the visible spectrum produced by the electron
ionization of molecules is recorded.

Excited states of the metal ion may play an important role in
the reactions observed here, and this area will continue to be one of
interest.

81

The Gas Phase Organometallic Chemistry of Ni+, NiCO+, NiPF3+ and
NiC5H5+ With Aromatic Compounds: Chemical Consequences Due to

the Presence of A Phenyl Group in the Organic Molecule, and Due to
the Presence of a Ligand on the Metal

ABSTRACT

The gas phase chemistry of Ni+ and the monoligated nickel
cations NiCO+, NiPF3+ and NiC5H5+ with a series of aromatic
compounds is reported here. The phenyl group is largely unreactive,
and does not appear to prohibit the metal ion from interacting with
other parts of these molecules. For some phenyl compounds, C6H5X,
no reaction is observed for Ni+. When Ni+ does react, products such

as NiC6H4+ and NiC5H5+ are formed - depending on the chemical

composition of the attached functional group (~X). Ni+ reacts with
benzyl compounds by insertion into the C6H5CH2—X bond, frequently
followed by charge transfer to form C7H7+ as a product. The Ni+ ion
is observed to decarbonylate aromatic carbonyl compounds. When a
single ligand is attached to the Ni+, changes in the chemistry are
observed. In many cases, the reactivity decreases as the size of the
ligand increases, suggesting the importance of steric interactions
even for these monoligated species. There are some exceptions,
where NiC5H5+ is most reactive, even though the cyclopentadienyl
ligand is the largest. It is suggested that some charge transfer may
occur in NiC5H5+, resulting in increased positive charge on the metal,

leading to increased reactivity.

82

W

The gas phase chemistry of transition metal ions with organic
molecules has been extensively studied in recent years, using ion
cyclotron resonance (ICR) and ion beam techniques. The reactions of
metal and metal-containing ions have yielded fundamental insights

into systems that activate bonds in organic molecules.1 The organic
molecules that have been studied include alkanes, as well as
molecules that contain the basic structural features (straight chain,
cyclic, branched) and functional groups of organic chemistry
(halogens, alcohols, acids, esters, ketones, ethers, amines, nitro-
compounds and nitriles). Some multifunctional compounds have also

been studied},3 The major reaction mechanisms have been

discussed and reviewed.1

For small compounds such as alkyl halides (RX),4 reaction
products appear to be formed via a series of elementary steps. First,

a transition metal ion such as Fe+ inserts into the C-X bond to yield
the intermediate R-Fe+-X. If the R- group is an ethyl group or larger,
a B-H shifts to yield a complex of the form (CnH2n)Fe+(HX). The
complex dissociates in a competitive ligand loss step.3 For more
complex functional groups (e.g., carboxylic acids), insertion into

bonds within the functional group can also be important.5’6

The role of the alkyl group in such reactions has been studied.
The extent of branching can affect the overall reactivity. For

example, ions such as Co+ do not appear to insert into the C3H7-CN
bond in n-propyl cyanide, but the insertion does occur for iso-propyl

cyanide.7 For straight chain polar organic compounds, the length of
the chain is an important factor in determining the number and

83

types of products. For example, Co+ reacts with ethanol by inserting
into the C-OH bond, which leads to the ultimate products CoH20+ and
CoC2H4+. Insertion into C-C or C-H bonds does not appear to occur.4
However, the reaction products of the reaction of Co+ with heptanol8
suggest that insertion into the C-OH bond is a very minor pathway,
compared to attack of C-C and OH bonds in the alkyl group.
Apparently, the metal may initially interact with the functional
group, however parts of the alkyl group that are "remote" to the
functional group may be brought into close proximity to the metal
ion via cyclic intermediates - making a variety of C-C and OH bonds
candidates for insertion by the metal.

While extensive work has been performed on a variety of
organic compounds, very little has been reported for aromatic
molecules. The (b- group (cyclo-C6H5-) may behave much differently
than other R- groups. Presented here is a survey of the reactions of

Ni+ with a variety of polar compounds that contain a phenyl group.

Several examples of the reactions of aromatic compounds with
gas phase transition metal ions have been published. Corderman and

Beauchamp9 reported that NiCp+ (Cp = cyclo-C5H5) decarbonylates a

variety of carbonyl-containing compounds, reactions (1) and (2).

NiCp+ + RCHO -) Nicpco+ + RH (1)
—> NiCpRH+ + co (2 )

Decarbonylation was observed for a variety of aldehydes including
benzaldehyde, to form benzene and CO.

In 1979, Dietz, Chatellier and Ridge10 reported that Fe+ can
induce the dehydrohalogenation of halobenzenes, reaction (3).

84

Fe+ + C6H5X —> FeC6H4+ + HX X =F,Cl,Br (3)
In the case of iodobenzene, the reactions (4) and (5) were observed:

Fe+ + C6H51 —> FeC6H5+ +1 (85%) (4)
—+ Fe1+ + c6H5- (15%) (5)

Reactions (4) and (5) are analogous to those reported for Fe+ with
methyl iodide. Consideration of the reaction thermodynamics lead to

the following proposed bond dissociation energies: D(Fe+-C6H4) > 66
kcal/mol; D(Fe+-C6H5) > 64 kcal/mol. The FeC6H4+ may be an Fe+-

benzyne complex,10 or may take the form of structure (I):
+Fe\

 

 

(1)

Recently, Chowdhury and Wilkins11 reported the gas phase

chemistry of Au+ with benzene, toluene and ethylbenzene. Most of
the products were due to hydride abstraction. (Note: the Au-H bond

energy. 71 +/- 3 kcal/mo],11 is greater than the Ni-H bond energy,12

60 kcal/mol). For example, Au+ reacts with toluene by (6):

85

(1)-CH3 + Au+ -—> C7H7+ + AuH (6)

It was determined that 90% of the ionic product was in the form of
the benzyl cation, with the remaining 10% having a tropylium
structure. In reactions such as (6), hydride stripping need not
require insertion into a bond. This contrasts with reactions such as

(7), which presumably proceeds4 via an insertion intermediate,
C3H7-Ni+-Clt

Ni+ + C3H7Cl —> C3H7+ + NiCl (7 )

Charge transfer following insertion may occur in (7). While the

ionization energy of the isopropyl radical13 (7.55 eV) is comparable
to that for Ni (7.64 eV), the NiCl moiety has a much higher ionization

energy, 11.4 eV14. Analogous processes may then be expected for

compounds of the type oCHZX, since the ionization energy of ¢CH2° is
7.3 eV.

Based on our present understanding of gas phase organometallic
ion/molecule reactions, would we expect aromatic compounds to
react as their saturated alkyl analogs, or would we expect the phenyl
group to dominate the initial interactions? Gas phase basicity data
may provide some insights into this question. It has been noted that
relative metal ion (M+)-ligand interaction energies correlate with the
ligand's proton affinity (P.A.).15 Larger gas phase basicities indicate
larger M+-L interactions for simple n- and n-donor bases. Table 5-1
lists some relevant proton affinity data.16 The basicities of
functional groups, as indicated by small compounds such as
acetaldehyde, ethanol, etc., are larger than that of benzene,

suggesting that the presence of a phenyl group need not prohibit M+

interactions with other parts of an aromatic molecule.

86

Table 5-1.

Some Relevant Proton Affmitiesa

Meleeule

C6H6
CH3CHO
C2H50H
CH3COOH
(CH3)20
CH3N02
(CH3)2CO
CH3NH2

¢CH3
¢N02
¢CHO
¢0CH3
¢NH2
¢CH2NH2

Eretee Affinity (keellmel)

181.3
186.6
188.3
190.2
192.1
192.5
196.7
214.1

189.3
193.4

200.2
200.3

209.5
216.8

 

a. Data obtained from ref. 16.

87

Reported here is the observed chemistry of Ni+ with a series of
aromatic compounds (phenyl- and benzyl- compounds) containing a
variety of functional groups. These results provide a framework
with which the behavior of the phenyl group may be put into
perspective, relative to other CnHm- groups such as alkyl groups. In
most cases, the phenyl group is unreactive, but affects the
thermochemistry of the system through inductive effects, as will be
discussed.

Also, the reactions of three ligated nickel ions, NiCO+, NiPF3+ and

NiCp+ with the same aromatic compounds are presented. Changes in
reactivity due to the presence of the various ligands reflect a number
of chemical aspects of the ligands; the most important feature
appears to be size (steric effects).

E . IISI'

All experiments were performed on an ion cyclotron resonance
(ICR) mass spectrometer of conventional design which was
constructed at Michigan State University and has been described

elsewhere.17 The ICR cell has been modified from a square to
rectangluar cross-section (0.5 in. x 0.75 in.). All experiments were
performed in drift mode. Mass spectra were obtained with a
frequency-swept detector that was designed, built, and installed by
Dr. John Wronka, and has been described elsewhere.18 Data were
obtained by scanning the frequency swept detector over the range of
frequencies necessary to produce a spectrum, with the magnetic field
of the Varian V-7800 electromagnet held constant at 16 kG. The
aromatic compound and organometallic compound Were admitted to

the cell in approximately a 1:1 ratio, to a total pressure of 6 x 10'6

88

torr. Typically, eleven mass spectra were signal averaged to improve

the signal-to-noise ratio. Data were collected with a MacADIOS® data
acquisition system (GW Instruments, Cambridge, MA) using a

MacIntosh Plus® computer. The data acquisition program was

written in Microsoft® BASIC. After a single resonance mass
spectrum was acquired, a second, fixed frequency oscillator (tuned
to the frequency of a reactant ion) is turned on. The irradiated ion is
thus ejected from the source region of the cell, and another mass
spectrum of the reaction mixture is acquired. Any ions in the
spectrum that were products resulting from the reaction of the
irradiated ion would not appear in the second spectrum. The
spectrum acquired with the double resonance oscillator on is then
subtracted from that obtained with the double resonance oscillator
off. This difference spectrum consists of all ion/molecule reaction
products of the irradiated ion. Double resonance spectra were
obtained for all ions of interest in this way.

All chemicals used in this work were high purity commercial
samples which were used as supplied, except for the liquid samples
which were subjected to multiple freeze-pump-thaw cycles to
remove non-condensable gases. Butylbenzene, benzyl bromide,
benzyl chloride, benzylamine, and phenylacetaldehyde were
obtained from Aldrich Chemical Co. Benzonitrile, acetophenone,
phenol, and benzoyl chloride were obtained from Baker Chemical Co.
Bromobenzene, chlorobenzene, nitrobenzene, benzaldehyde, benzoic
acid, and benzyl alcohol were obtained from Fisher Scientific.
Benzene and benzophenone were obtained from E. M. Science.
Methyl benzyl alcohol was obtained from Eastman Organic Chemicals.
Iodobenzene was obtained from Chem Service. Toluene, aniline,
anisole, and ethylbenzene was obtained from Matheson, Coleman,
and Bell. Nickel tetracarbonyl and tetrakis (trifluorophosphine)

89

nickel were obtained from Alfa Inorganics. Cyclopentadienyl nickel
nitrosyl was obtained from Strem Chemicals.

E II I D' .
The compounds that have been studied are divided up into three

groups. The reactions of Ni+ with the phenyl compounds are listed in

Table 5-2. The reactions of Ni+ with the phenyl compounds are listed
in Table 5-3. Table 5-4 contains the results for aromatic carbonyl
compounds. These tables contain a variety of thermochemical

data19 that are useful in understanding the observed chemistry.

We will first consider the reactant molecules in Table 5-2, the
simple ¢-X molecules. Similar to the reactions reported10 for Fe+, we

do observe the formation of an M+C6H4 product in some cases:
Ni+ + C6H5X —> NiC6H4+ + HX, x = Br, I (8)

To form the HX product, a C-X and a C-H bond must be cleaved.
Presumably the reaction proceeds by initial insertion into the C-X
bond, rather than into the CH bond, since the former is weaker. The
reaction is not observed when X = H, Cl, CN, OH, NH2, OCH3 and N02.
The X=OCH3 and N02 cases will be discussed separately. In general,
the reaction analogous to (8) is not observed when the o-X bond is
strong, and when Aern.( ¢X -> C6H4 + HX) is relatively high. When
this reaction product is not observed, it may be because the initial
insertion is thermodynamically prohibited, or that the overall
reaction is endothermic. In the case of ¢NH2 the barrier may be in

the insertion step, since Ni+ has not been observed to insert into the

ON bond of primary amines.17 Since the overall reaction for X=Cl
would be exothermic (based on the energetics of the process for X =
Br), we may conclude that the initial insertion does not occur.

90

.858; 8 u 11-12306
6.5.3. 3 u foam .3

.0583. 8 u 3:90.30 .0

._oE\_mox 5 53.525 5282.00.52. __m .m

 

m . 02 + enzoo T

Z. 02+oo+mxmoT «029
X... «:2 + foo T Nzze
m c 10910 + foo T

m omxo + x9 T £89
3 Cu: + foo T :3
3 20: + foo T 28
a m .1 + «£0 T _e
n o a: + «zoo T .me
3 _o: + foo T .09
o m «I + foo T :9
.33

: m o
D an

0 mm
a vo—
-- app

-- 5
-- mm
-- mm
-- v o F

«Add «a 3d. «a 3d

02 + +0mzmo_z

£48 02 + co + +mzmo_z
82080 5.8% oz

ONIO + +192

82680 5.88.“. 02
8580 5.8% oz
Agmmv
3.3 _I + +vzoo_z

_ + +mzoo_z

.m: + +5392

umzmmno 558$ 02
8580 888m oz

g

mucsonEoU 3:25 .23 tz .«o 328an

N-m 035.

N02

N22

mzoo
IO
20

ll
X

91

We note that, in the context of typical gas phase organometallic
ion/molecule reactions that have been observed, reaction (8) is
certainly unusual. In reaction (9), 19 kcal/mol is required to convert
ethyl bromide to ethylene and HBr,

Ni+ + CszBr —9 NiC2H4+ + HBI’ (9)

thus the reaction will be exothermic if the Ni+-ethylene bond is
stronger than 19 kcal/mol, which it surely is. In contrast, 67
kcal/mol are required to convert ¢Br to benzyne (C6H4) and HBr.
The reaction is observed only because the binding energy in the
NiC6H4+ product is greater than 67 kcal/mol. This substantial bond

energy may reflect sigma (M-C) bond formation in the product.

Very different reactions are observed for ¢OCH3 and oNOz. In

the case of oOCH3, insertion into the ¢~O bond appears to occur, to

give the o-Ni+-0CH3 intermediate. This is the first case where both
groups on the metal have B-H atoms that can shift. Since the
products {benzene and formaldehyde} are more stable than {benzyne
and methanol} by 61 kcal/mol, the lower energy pathway available
here is selected, and the shift of a H atom 110.11 the methoxy group
dominates the chemistry.

The reactions of ¢N02 are very different from the other reactions
reported, however nitroalkanes also exhibit chemistry with
transition metal ions that is much different than that observed for
alcohols and alkyl halides. In 1984, Cassady, Freiser, McElvany and
Allison reported the reactions of Co"‘ with as series of nitroalkanes.6
In addition to reactions that proceed via insertion into the C-NOZ, C-H

92

and C-C bonds, additional intermediates were apparently formed
such as RO-Co+-NO. Mechanisms were suggested that would allow for
access of such an intermediate.6 In light of this, it is not surprising
that NO elimination from ¢N02 was observed. However the major
product is C5H5Ni+, with loss of NO M. We propose that the

products are formed in the sequence shown in reaction (10).
Ni"' +¢N02—> ¢o-Ni+-No :lSSL> ¢o-Ni+ &> C5H5Ni+ (10)

We note that the C6H50 group has the same skeletal structure as

a cyclic ketone, similar to cyclohexanone. The energy required to

convert C6H50' to C5H5' and CO, 23.2 kcal/mol, is similar to that
required to decarbonylate cyclohexanone to l-pentene and CO, 22.7
kcal/mo].

Finally we note that, for X = OH, OCH3 and NH2, the possibility
exists for insertion into bonds in the functional group (as we see in
oNOz). While an insertion process such as reaction (11) for phenol

(p011 + Ni+ —> ¢o-Ni+-H (11)

may occur, there is no B-H available to shift, and the only option for
this intermediate may be to convert back into reactants.

The discussion will now move to the molecules in Table 5-3, the
benzyl-compounds. In the chemistry with these molecules, insertion
into the o-X bond is not observed, but rather insertion into the ¢CH2-
R bond dominates, reflecting the relative weakness of this bond. This
is graphically seen in variations of the CC bond energies in n-butyl
benzene. The o-C bond is more than 30 kcal/mol stronger than the
¢CH2-C bond!

93

av
or
mm
on

up

m_.

an

ex.

o~1+mxuoe T
94.10318 T «10.10109 5
mfuo + foo T
a: + «.818 T mxmoe 8
mm
Oon + 19 T
a: + 012 T Iofoe
«h A $0.25 ”8283:: «a
me A 3.25 ”Sagan:
.mmzo + foo T 5909 9.
mm A 20125 Eo=§a§ 8
4:0 + foo T £3 E

.—

> _ .aanuea «flea

mnczanoU .355 .«o mcocowom

m-m 035.

mm

om
mm

5

mm
vm
mm

$.23 of + 3101092
£58 090 + +£0.22

m: + +£05.22
«I + +szoe_z

3&8 N: + +9892
$3 :02 + Rio

.92 + +5150
62 + K16
.82me0 cozommm 02

g

mzofofo

fumzo
NIz~+5

94

cm :91.on T 3

m N @140 + :e T
o m ozmo + «:2. T Be
«33

-- mop

mm vm

«an «a am an an

non—cacao .m-m 05a...

82320 cozommm oz

mxooa

ozmo + +£0.22 mxomazov

g

95

This correlates with the observation that only loss of propene is

observed, although when Ni+ reacts with n-butane, ethane loss and
H2 loss is observed. For benzyl chloride, bromide and benzyl alcohol,

the only or major product is C7H7+. (Thermodynamic discussions19
will assume that the product is the benzyl cation). As indicated in
the Introduction, such reactions are reasonable since the relative
ionization energies predict charge transfer to the C7H7 moiety, and
the strength of the bond in the neutral that is formed is substantial.
Listed in Table 5-2 are the lower limits on various Ni-X bonds that
are implied by assuming that the observed reactions are exothermic
or thermoneutral. These are consistent with reported bond
energies”: D(Ni-Cl) = 88 kcal/mol, D(Ni-Br) = 85 kcal/mol . Also,
the relatively weak Ni-H bond,12 60 kcal/mol, would make the
analogous reaction endothermic for toluene.

When X- in o-X has at least 2 skeletal atoms, and both have one
or more H atoms on them, H2 elimination is observed. This suggests
that a site of unsaturation is formed in the -X group — i.e., ethyl
benzene is converted to styrene, benzyl alcohol becomes
benzaldehyde, etc. Presumably, the benzyl C-H bond is the initial
site of insertion in these cases.

Unexpected products are observed in the case of methyl benzyl
alcohol. Water loss is observed, as is typical for aliphatic alcohols,
indicative of insertion into the C-OH bond. An unexpected process is
elimination of formaldehyde. One could conceive of elimination of
CHZO via either an initial insertion into the o-C bond or the C-CH3

bond, as shown in Scheme 5-1:

96

 

 

+ OH methgl- +
I__.) (D- Ni-CII-CH3 SM, d)- Ni-CH3
9H ” +
. +
N1 + (D-e-CH3 -— —>
H OH
OH '
| +. ”(1:
(D_(l:- Nl-CH3 pfifigl- H
H shift
Scheme 5-1

Unfortunately, the possible insertions, followed by a methyl or
phenyl shift, yields a diradical, not formaldehyde. An alternative
mechanism would involve loss of H2 first, followed by loss of CO. It

has been reported that Ni+ will induce the elimination of H2 from

alcohols, presumably via insertion into the O-H bond.4 More
importantly, we report it here for benzyl alcohol. Elimination of H2

yields the corresponding carbonyl-compound as Ni(¢(CO)CH3)+. As

will be discussed shortly (Table 5-4), Ni+ does decarbonylate
acetophenone - thus we suggest a two step pathway, with the neutral
product(s) being {H2 + CO}, not CHZO. This is also supported by the

NiL+ reactions that will be discussed in following sections.

97

The reactions of Ni+ with a series of aromatic carbonyl
compounds are listed in Table 5-4. For all of the compounds except
benzoyl chloride, decarbonylation is observed. A number of reports
have appeared concerning the chemistry of transition metal ions
with carbonyl compounds.5’22’23 The 1976 Corderman and
Beauchamp publication was cited earlier.9 It is interesting to note
that, while CpNi+ was observed to decarbonylate molecules such as
CH3CHO, the analogous reaction was not observed for acetyl chloride
and bromide.9 Freiser et al.22’24 have studied the chemistry of Cu"'
and Fe+ with ketones. Also, cyclic ketones were used in a number of
publications since Fe+ appears to form a metallacycle upon
decarbonylation.22’23’25l26 In 1984, Halle, Crowe, Arlnentrout and
Beauchamp23 reported the chemistry of M+ with several ketones
and aldehydes. They suggest that, in ketones, R1(CO)R2, transition
metal ions appear to insert into both Ri-C bonds. However, in
molecules such as aldehydes, insertion into the C-CO bond is favored
over insertion into the C-H bond as an initial mechanistic step.

We note that reactions similar to reaction (8) for carbonyl
compounds, such as reaction (12),

Ni+ + o-C(O)H -+ NiC6H4+ + CHZO (12)

are get observed. Relatively large amounts of energy are required.
For the above reaction to be exothermic, the Ni+-C6H4 bond energy
would have to be greater than 82 kcal/mol. Thus, such pathways are
thermodynamically prohibited in these compounds. In comparison,
decarbonylation requires little energy. We also note that, for benzoic
acid and methyl benzoate, C02 elimination is a thermodynamically

favored process, but does not occur. For aliphatic organic acids, C02

98

125.8,. 8 1 11100091095 ”6.5.8.. 8. .. 101091095 .9

 

mp
hm

op-
mm
N-
mm

mm

00 + m”109 T
00 + _09 T
0109 + 9190 T
00 + 99 T
N00+ 0109 T
00 + 01009 T
N00 + 19 T

00 + 109 T

00 + mU109 T
0010 + 9100 T
00 + 19 T

010N109 e
.0009 90
9009 8
010809 B
10009 0 o 9
m”10009 «a
0109 mm

mm
5

mm

mop

mop
5

mm

n13 39a 4%.”qu

00 + +0109:
0020000 cozommm oz

00 + +2912

00 + 210092

00 + +1092
00 + +9109_z

00 + +19_z

13.1043de

3:509:00 3:00ch 0308059 5:5 +2 _0 2.09.0039

0% 030%

160090
.0100.

91000

m”1001000

101000
«101000

1600

H
X

99

elimination is not observed,5 so failure to observe the reaction is not

related to the presence of the aromatic group here. Consider the case

of benzoic acid. The decarbonylation may occur via the ¢(CO)-Ni+-OH
intermediate, and would require a (1)-shift. If insertion lead to an

intermediate of the form o-Ni+-COOH, the geometry about the acid

group, due to the sp2 hybridization of the carbonyl C may place the H
too far from the metal for a H shift to occur. However, an OH shift

may occur, again leading to decarbonylation.

In decarbonylation reactions, when both bonds to the carbonyl
group are cleaved, it would be difficult to suggest whether one bond
is attacked preferentially in the initial step. The bond energies in
Table 5-3 suggest that either would be possible. In the case of
benzoic acid, insertion into the C-OH bond may be favored, due to
preferred initial interactions such as those shown in structure (II):

0 .eNi”
ll
a-c—o\
H
(II)

In the case of ¢CH2(CO)H, the very weak ¢CH2-C bond is presumably
the site of attack to form oCHz-Ni+-(CO)H; a H shift leads to

decarbonylation.

The results presented in Tables 5-2, 5-3, and 5—4 provide a good
understanding of how a phenyl group behaves in these gas phase
ion/molecule reactions. In most cases, it behaves as an unreactive
"terminal" group such as H- or CH3-. Actually the bond energies

100

place it between the two, e.g., D(H-CH3) > D(o-CH3) > D(CH3-CH3).

While an H can be removed from a o- group such as in the o-I
reaction, this certainly does not suggest that ¢- behaves as an alkyl
group such as an ethyl group. In fact the energy required to
eliminate HI from methyl iodide is closer to that for o-I than it is for
C2H 51. What we have seen as the major influence of the phenyl
group is in its inductive effects, substantially weakening neighboring
bonds - as seen in the benzyl compounds. We suggest that this effect
may be very useful. Since the phenyl ring does not hamper access of
the metal ion to other parts of the molecule, one may be able to add
a phenyl group to various positions on an alkyl chain, create
"unusually" weak bonds, and determine how such thermodynamic
variations of the molecule affect the chemical reactivity.

W The observed chemistry of NiCO+, NiPF3+

and NiCp+ with the aromatic compounds that are the focus of this
work, is summarized in Table 5-5. First, a few comments are in
order on these ligands, and how their presence may affect the
chemistry of the metal ion in these gas phase reactions.

NiCO“: Of the ligands studied in such gas phase ion/molecule
reactions to date, CO has been among the most popular, since volatile
transition metal carbonyl compounds are readily accessible for mass
spectrometric studies.lv4’15’27 Recent theoretical calculations
suggest that, while the bonding in neutral MCO species (where M is a
first row transition metal) may be described by a Dewar—Chatt type

interaction, this is not the case for the analogous cation. In MCO+ the

metal-ligand bonding is essentially electrostatic in nature.28’29

101

 

9.9va +52

+a.__z 4 + +a_z 3&0 ._ + :22

5.081 92 ._ + £22 ._ + :22

$.80 :32

9.999 5 + :22

+1.22 ._ + +a_z 9.99: ._+ 0~10 ++19_2

5081 92 3 + +a_2 ._ + 222

cofimmm oz 4 + +22 ._ + 122

_1 + +£00.22 5.891 2 ._ + RE

.01 + 21002 ._ + +a_z ._ + RE

.01 + +£00.22 3 + BE 4 + +a_z

+52 3 + :22 1 + :12

+32 +3012 +922
0820011

000000800 0208000 :23 +2 BEE—000.: .90 282003.

3 2.3.

N02.9
012.9

0100.9
10-9
20-9

n +92

102

202009”. 02

._ ++a__2
3.99 1 + N1 ++m10009_z

3.03 3 +122
39.: 10190391092

 

091 ++~10109.__z 3.990 ._+ 0~1++~10109_2 ._+~1++910009_z
3.80 4 +82
5.89192 .. ++a_z 3.11 ._+N1++~10109_z
3&3 N1 9 + 2032 3.98 ._ ++a_z 3&9 ._ ++a_z
3.3091 31210922 3.5 ._+~1++1z109_z 3.98 ._+~1++1z109_z
3.990 ._ + .322 3.03 ._ ++a_z
5.89192 3.98 ._+N1++0109_2 3.98 191301092
3.80 1 + +92

58812 3.98 3292 + +9190 329239190
3.02 + +9190 2.02 + +9190 5.8.0102
+0.22 ._ + +922 .. + $22
+9.02 mega +0.09...

090301..

00:50:00 . Wm 030,—.

MU10291029
MU109100109
MU10910.9
N12910.9
10910-9
5910-9
0910-9
910-9
méfimum

H +I=Z

103

3&8 4++n:2

 

5881 92 ._ +82 3.3 3 + 00 ++m109_z 120009109

20881 e2 5581 92 581 e2 .0699

e881 92 ._++a_z 3++a_z 920009
3.80 5 + 82

+1.__z 3 ++1_z 3.13 ._ + 00 ++m1009_2 010020009
388 ._ + +12

5881 92 1 ++a_z 3&9 ._ + 00 +132 1020009
9.9.98 ._ ++a_z 383 00 + +12

+1.22 3.80 ._ + 00 ++m109_z 3.93 4 + 00 ++m109_z 91020009
2890 1++Ez

00.91922 ._ ++a _ z 3.93 3+00++19_z 12009

$213033

+1.92 +313 +0103 1 LE
09010011

00110100 . Wm 031,—.

104

Thus, the electronic structure of the metal in MCO+ is essentially the
same as that in the bare metal ion. We have recently discussed the

chemistry of MCO+ ions;28 in general they react very similarly to M+.
There are some cases in which the CO can become involved in the

chemistry; that is, MCO+ can react with a molecule AB via

intermediates such as A-M+-(CO)-B. However, the CO is usually
simply a spectator, having little effect on the chemistry. Recent work

suggests that D(Ni+-CO) is approximately 25-30 kcal/mo].29 While

CO has an appreciable electron affinity3O (E.A. (CO) = 31.6 kcal/mol),
little charge transfer occurs, i.e., the charge on the nickel atom is +1.

NiPF3+z Little has been done to date in the gas phase with the PF3
ligand. Reaction (13) has been observed,15c which

CoCO+ + PH3 —> CoPH3+ + CO (13)
indicates that PH3 is more strongly bound than CO to Co+. If, as has

been suggested, the proton affinity of simple ligands reflect their
relative binding energies to transition metal ions, PF3 would be more
weakly bound than PH3 (PA(PF3), 166.5 kcal/mo] < PA(PH3), 188.6
kcal/mol)16. PF3 has a negative electron affinity, so charge transfer
would lead to dissociation. Ligand substitution processes such as
reaction (14)

NiPF3+ + ex 9 Ni¢X+ + PF3 (14)

suggest that PF3 exists as an intact ligand, as opposed to a structure

such as F-Ni+-PF2. From appearance potential measurements31 on

the ions derived from electron ionization of Ni(PF3 )4, it has been
suggested that AHf(N1PF3+) = 35 kcal/mol. Based on AHf(Ni+)= 279
kcal/mol,14 and AHf(PF3) = -219.6 kcal/mol,14 this would lead to a

105

bond strength of D(Ni+-PF3) = 24.6 kcal/mo]. Based on this

information it is difficult to decide whether PF3 would be expected to

be more or less strongly bound to Ni+ than is CO. To order the
relative metal-ligand binding energies, we studied the ion/molecule
reactions in a mixture of Ni(CO)4 and PF3. Reaction (15)

Nico+ + PF3 —> NiPF3+ + C0 (15)

was observed. We suggest that the two bond energies are similar,
with D(Ni+—PF3) > D(Ni+-CO) by a few kcal/mol.

NiCp“: The cyclopentadienyl ligand appears to be much different
concerning its effects on the chemistry of a metal than the other two

ligands studied here. Jacobson and Freiser32a studied the chemistry
of Con+, and report some bond energies that suggest that Con+
forms stronger bonds than does Co+ (for example, D(Co+-C4H6) <
D(CpCo+-C4H6), D(Co+-Cp) < D(CpCo+-Cp)). They report that D(Co+-Cp)
= 85 +/- 10 kcal/mol, which is larger than D(Co+-C6H6), 70 +/- 4
kcal/mo]. This suggests that a bonding mechanism other than a
simple electrostatic one may be operative for the Cp group. It has

been suggested that the Cp group can be actively involved in the
chemistry of the attached metal - molecular fragments can be

"stored" on the Cp ring,32 as well as on the metal ion. For example,
reaction (16)

Con+ + C4H8 —> COCp(CH2)+ + C3H6 (16)

has been discussed32“l by suggesting that the CH2 may not be bound
to the metal, but may be incorporated into the ring to yield a

Co(methyl-cyclopentadienyl)+ product.

106

The Cp ligand appears to be strongly bound to Ni+. In Table 5-5,
this ligand is never displaced by an aromatic compound, while CO
and PF3 are. Also, the molecular ion of CpNiNO reacts with all of the
aromatic compounds studied here in a ligand displacement reaction,
reaction (17),

CpNiNO+ + ¢X —> CpNi¢X+ + NO (17)

with the Cp never being displaced. It is possible that some charge
transfer takes place between the metal ion and the Cp group to yield
a structure of the type Ni+11+5}Cp'5. Charge transfer of this type has
been observed for ligands such as the chlorine atom.33 The electron
A < 42.3 kcal/mol.34 If

affinity of C5H5 has been reported to be
additional positive charge is generated on the metal with this ligand,
it may explain the strength of the Ni+-Cp bond, as well as the
increased bond energies to Co+ when the Cp ligand is present,
relative to those for Co+ alone. The reaction of ¢N02 with Ni+ in Table
5-2 to form NiC5H5+ suggests that D(Ni+-C5H5) > 41 kcal/mol, Me
2535 exists es en intaet ligand in the reectien predeet . We note that
in organometallic chemistry35, the Cp ligand is frequently considered

to take the form Cp'; charge transfer from the metal to Cp would
yield an aromatic ligand (6 lt-electrons).

Expected Trends in the reactions of NiL+: Based on what is
known about these three ligands, some reaction trends may be
proposed. However, first a comment should be made on the "pitfalls"

of over-interpreting the data7 in Table 5-5. Consider the two
reactions:

107

Ni+ + ABC —> NiB+ + AC (18)
Nico+ + ABC —+ NiABC+ + C0 (19)

One may be attempted to suggest that, while Ni+ reacts with ABC,
NiCO+ only undergoes a ligand substitution reaction (i.e., does not
insert into bonds, etc.). However, this may not be the case. If NiCO+
induces the rearrangement of ABC to AC and B, the species that

dissociates to the final products is Ni+(AC)(B)(CO). The order in which
ligands are lost appear to reflect their relative bond energies to the
metal ion. Thus, if AC = H2, it would be preferentially lost relative to
CO, and, if a reaction occurred, it would be apparent. However, if the
products AC and B are the more strongly bound ligands, reaction (19)
will be observed. Thus reaction (19) provides no information (in the
latter case) as to whether rearrangement of ABC to AC and B has
occurred. Thus, it is important, when comparing the chemistry of Ni+
and NiL+, to consider the possible reaction products and their proton
affinities.7 In this way, one may be able to conclude whether or not
the presence of the ligand affects the chemistry, or whether such
information cannot be obtained from the experimental results
presented.

We interpret reactions such as
NiCO++¢Cl —> N1¢Cl++CO

as ligand displacement reactions. If chlorobenzene reacted on the
metal center to form C6H4 and HCl, HCl is more weakly bound than

CO and would be lost. That is, if the HCl elimination occurred, we
would expect to see either NiC6H4+ or NiCOC6H4+ as a NiCO+ product.

Note that NiCp+ does not participate in ligand substitution reactions,

108

presumably reflecting the strong Ni+-Cp bond energy.

Changes in the chemistry of NiL+ as L varies would be expected
to parallel the relative bond energies, which we expect to be:

D(Nl+-c0) < D(Ni+-PF3) < D(Ni+-Cp)

Also, the relative sizes of the three ligands may introduce steric

effects. We have found the "cone angle" concept3° to be useful in
this regard; it suggests that the Cp ligand is the largest, occupying a

"cone" from the origin (metal center) with an angle of 136°, with PF3

next at 104°, and CO the smallest, 95°. These steric effects may
favor the formation of intermediates in which at least one of the
groups attached to the metal is small such as -H. Steric effects may
hinder the insertion process in which two bulky groups are bound to
the metal (in addition to the ligand). In addition to steric effects,
differences in the chemistry of the Ni+ when Cp is attached may
reflect either an increased positive charge on the metal, or reactions
which involve storage of some molecular fragment on the ring.

The trends in reactivity that will be highlighted from the data in
Table 5-5 fall into a few simple categories. There are some reactions
in which NiCp+ appears to be more reactive than the other NiL+, and
even Ni+, which may be due to the increased positive charge on the
metal. We suggest that most of the trends can be related to steric
effects.

WWfi This is observed in the reactions

of the halobenzenes in Table 5-5. Ni+ reacts with bromo- and
iodobenzene to eliminate HX, but not with chlorobenzene. Only

109

ligand substitution reactions appear to be occurring with these
compounds for NiCO+ and NiPF3+. In contrast, NiCp+ eliminates HX
from all three ¢X compounds. The thermochemical data19 in Table

5-1 suggests that the failure of Ni+ to react with ¢Cl may be due to
the relatively strong o-Cl bond (rather than the overall reaction

enthalpy). If the Ni in NiCp+ has a charge of greater than +1.0,
polarization effects may lead to the formation of stronger bonds.

That is, NiCp+ may react with ¢Cl because
[D(CpNi+-Cl) + D(Cp(Cl)Ni+-¢)] > [D(Ni+-C1) + D(ClNi+-¢)].

With the Cp ligand, there are, of course, many other possibilities such
as coupling of the C6H4 group to the Cp ring.

W We propose two types of steric

effects. In the first case, a reaction may be "turned of " due to the
presence of a bulky ligand. That is, while Ni+ may react with ABCH
via the intermediate A-Ni+-BCH, the steric interactions may prevent
access of the analogous intermediate when a bulky ligand is present.
A second steric effect can be seen when the products of Ni+ reflect
attack of several bonds. For the molecule ABCH, the product

distributions for Ni+ may suggest that the intermediate (III) is
formed to a greater extent than intermediate (IV):

A-Ni+-BCH ABC-Ni+-H
(III) (IV)

110

However, with a bulky ligand on the metal, steric interactions
would be less if the OH bond was attacked, making intermediate
(VI) more stable than (V).

L L
l I
A-Ni+-BCH ABC-Ni+-H
(V) (VI)

Consider, for example, the reactions of ¢CH2X, X = Cl and Br. Ni+
reacts by halide abstraction with both compounds. NiPF3+ so does as

well. NiCO+ only reacts with oCHZBr. This may suggest that the
ionization energies (I.E.) may vary, with IE. (NiCO) < I.E. (NiPF3)37,

thus the presence of the CO may hinder the charge transfer step that
leads to the formation of the C7H7+ product. However, NiCp+ my
reacts with the chloride. This may be a case where the energetically
more facile reaction (for the oCHZBr) does not occur because the Br
atom is too large, preventing access of the oCHz-Ni+(Cp)-Br
intermediate. This further suggests that the formation of C7H7+

occurs via an insertion intermediate, not a stripping process.

The decrease in reactivity with increasing ligand size is more
apparent in the data for the aromatic carbonyl compounds in Table
5-4. Here reactivity (i.e., ability to induce decarbonylation) follows
the order

NiCO" > NiPF3+ > NiCp+

decreasing as the size of the ligand increases. Consider first
¢(CO)CH3. If decarbonylation proceeds via intermediates of the type

111

¢-Ni+(L)-(CO)CH3, we would expect decreasing reactivity as the size
of L increases. We note that the cone angles36 of C6H5- and (CO)CH3
are both estimated to be 100°. Thus, 42% of the products correspond
to decarbonylation for NiCO+, 20% for NiPF3+ and 0% for NiCp+. The

ability of the metal in NiCp+ to form stronger bonds apparently
cannot compensate for the steric energies involved. However, this

may not be the case for ¢(CO)H. NiCp+ does decarbonylate
benzaldehyde, suggesting that the reaction proceeds by insertion into
the C-H bond; this will minimize steric interactions since one

molecular fragment on the metal is an -H, with a cone angle36 of
only 75°. Why then doesn't NiCp+ decarbonylate ¢CH2(CO)H? This

may reflect the fact that the weak benzyl-C bond favors insertion at
that point - that is, that decarbonylation of ¢(CH2)(CO)H proceeds
through intermediate (VII), not (VIII).

¢CH2-Ni+-(C0)H ¢CH2(C0)-Ni+-H
(VII) (VIII)

The increased probability that NiL+ will favor intermediates of the
type ABC-Ni(L)+-H is seen in a number of instances. Consider benzyl
alcohol, oCHzOH. Ni+ reacts with this compound to form C7H7+ , and
to eliminate H2, presumably via intermediates (IX) and (X),

respectively.

¢(CH2)-Ni+-OH ¢CH20-Ni+-H
(IX) (X)

In contrast, NiCO+ and NiPF3+ only appear to react via an

intermediate analogous to (X); again the extent of reaction decreases

112

as the size of the ligand increases. In fact, many of the cases where
NiPF3+ and NiCp+ react correspond to those mechanisms in which
insertion into a CH, O-H, or N-H bond occurs. Also note that there is
only one case where NiCO+ does net induce a decarbonylation of an
aromatic carbonyl when Ni+ does. The exception is ¢(CO)¢, in which
the two groups that would be on the metal upon insertion are the

largest. Thus, in this case, even the smallest ligand, CO, appears to
exert steric influences.

In the experiments performed to yield the data in Tables 5-2,
5-3, and 5-4, the reactions of the multi-ligated ions such as
Ni(CO)2’3,4+ were also studied. These are not reported here since
they are generally uninteresting. Typically, only ligand displacement
reactions are observed. There were only two cases in which NiL2+

ions were observed to react:

Ni(PF3)2+ + ¢CHZOH —> Niq)CHO+ + H2 + 2 PF3 (20)
a n d
Ni(CO)2+ + ¢CH(OH)CH3 -) NioCOCH3+ + H2 + 2CO (21)

When two ligands are present, steric effects would certainly be
increased, which would be consistent with decreased "reactivity".
These two cases in which NiL2+ ions react are both systems in which

hydrogen elimination occurs. That is, they proceed via intermediates
with minimal steric energy in the insertion intermediate, since they
place a -H on the metal as one of the two fragments of the molecule.

It was suggested that, in the case of ¢CH(OH)CH3, the loss of CHZO

was actually a two step process - elimination of H2 followed by

113

elimination of CO. jlfhie is managed by the reaetiens ef the ligated

speeiee, where H2 elimination is observed. Following H2 elimination,

cleavage of the Ni+ -L bond appears to occur rather than
decarbonylation of the ketone that is formed as a result of
dehydrogenation.

Finally, we note only one case in which the presence of a ligand

actually leads to n_e_w products. All three NiL+ ions react with benzyl
amine, ¢CH2NH2, to eliminate H2, consistent with our suggestion that
reactions that proceed by insertion into C-H, O-H or N-H bonds are

least affected by steric interactions. However, NiCp+ alone induces
the elimination of a second dihydrogen

Nicp+ + ¢CH2NH2 —+ CpNi¢CN+ + 2H2 (22)

If H2 elimination occurs, sufficient energy may be released when

the resulting ¢CHNH interacts with the Ni+{1+°}center to overcome
the barrier that leads to the elimination of a second H2. Reaction
(23)

¢CH2NH2 —) ¢CN + 2 H2 (23)

114

requires 33 kcal/mol, thus the metal-¢CN binding energy must be

greater than this value. In the discussion of the chemistry of ML+
cations, we suggest that steric interaction energies in some insertion
intermediates may make them energetically less favorable, when L is
large. One may not be able to distinguish between steric effects at
this point in the mechanism, from those in it's immediate precursor -
the initially formed ion/molecule complex. That is, the presence of L
may affect the ability of the metal to get close to some bonds
(structure (XI)) more so than others (structure (XII)), increasing the
stability of XII over XI when L is present:

L
l
Ni+ 1 /L
B—C B—C Ni
A/ \H A/ \H
(XI) (XII)

We have pointed out that there appears to be a number of
correlations in the chemistry of ML+ with the size of L, with cone
angles increasing in the order CO < PF3 < Cp. However, one must ask
whether the reactivity trends reflect steric interactions or some
other feature of the chemistry. Would the steric interactions with
three groups about the metal (as in the insertion intermediate or in
some activated complex that immediately preceeds insertion) be
sufficient to prohibit many reactions? We also noted that D(Ni+-CO) <
D(Ni+-PF3) << D(Ni"‘-Cp). Could the bond energies be the important

factor instead? Many of the trends could be explained as well by

115

assuming that the initial interaction energy upon ion/molecule
complexation is an important factor --required for subsequent
chemistry to occur. Suppose, when Ni+ complexes with some ¢X
compound, 35 kcal/mol is generated, and assists in accessing, in fact
is required for accessing, the insertion intermediate structure. Now
consider the reaction of NiL+. When it forms the initial complex, 35
kcal/mo] is released, and 30 kcal/mol is consumed in rapid cleavage
of the Ni+-L bond. The resulting complex, Ni¢X+ now has only 5
kcal/mol which may be insufficient for further reaction to occur (a
relatively "cold" Ni¢X+ complex), the result being only ligand
displacement. These considerations would predict that the reactivity

of NiL+ would decrease as D(Ni+-L) increases - for those cases in
which ligand displacement reactions occur. We note that Schilling

and Beauchamp37 recently proposed a similar argument to explain

the unreactive behavior of Pr+ and Eu+ with small alkanes - citing
the possible importance of the "chemical activation" that occurs upon
complexation and its relationship to barriers early in the reaction.
This energetic argument has some attractive features, and may seem
more reasonable than the steric argument. However the energetic
argument does not explain the observation that facile reactions occur

for ML+ ions when the initial insertion step involves attack of a CH,
O-H or N-H bond. These questions are providing a framework for
future studies of ligand effects in gas phase organometallic

chemistry. Similar studies on a larger variety of ML+ ions in which
small, relatively strongly bound ligands are used, will allow the
relative importance of various energetic and steric interactions to be
determined.

116
Conclusion

In these gas phase organometallic ion/molecule reactions the
phenyl group behaves chemically, in some ways, intermediate to the
methyl and ethyl groups. H transfer from the phenyl group is a high
energy process, much closer in energy to that required for H transfer
from a methyl group than an ethyl group. The formation of the
strong benzyne-Ni+ bond makes H-abstraction from C6H5- possible
in a few cases. When other pathways are open, such as the
elimination of a small neutral molecule from the functional group,
this latter pathway is preferred.

For comparison, the results of reactions of Ni+ with several
different small alkyl compounds are listed in Table 5-6. In cases
where no data for Ni+ are available, data for Co+ are listed instead.
In comparision to reactions with aromatic compounds, the results are
very interesting. In general, the alkyl compounds form more
products than aromatic compounds. For example, Co+ with
propanoic acid forms 5 different products, decarbonylation products

representing only 25% of the total reaction products.5

Decarbonylation accounts for 100% of the reactions of Ni+ and
benzoic acid. .
The reactions of sec-butyl amine are interesting because a major

product in the reaction with Ni+ is hydride abstraction with charge

transfer:3 8
S-C4H9NH2 + Ni+ —9 C4H10N+ + NiH (60%)

Hydride abstraction is also an important pathway for the reactions of
Co"’ and amines.17 This reaction is not observed for aniline or
benzylamine.

The results in this table show that even relatively simple alkyl

117

molecules can react in many different ways. Aromatic compounds

are much more selective in their reactions.

A major effect of the the presence of a phenyl group in an
organic molecule is its inductive effects. This can be seen in reactions
of substituted benzyl compounds. The weak C6H5CH2-X bond plays
an important role in the observed chemistry. This bond is the only
bond attacked for most of the benzyl compounds studied. Charge
transfer to the C6H5CH2 moiety following insertion into this bond

occurs due to the low ionization potential of this group.

The effect of ligands on metal centers is of major importance in

the understanding of transition metal chemistry in condensed
phases. The reactions of NiPF3+ follow closely the reactions of NiCO+,
with slightly less reactivity observed for the NiPF3+ ion. The metal-
ligand bonding in NiCO+ and NiPF3+ is probably very similar, so this
result is not suprising. The slightly less reactivity of NiPF3+ reflects
the steric crowding involved upon insertion produced by this larger

ligand. Much of the chemistry of the NiL+ ions can be explained by
considering the formation of intermediates which would exhibit low
steric interactions. The reactivity of NiCp+ with the halobenzenes
indicate that this ligand affects the metal ion much differently than
either CO or PF3. Charge transfer to the cyclopentadienyl group to

increase its aromatic character may explain why NiCp+ is more

reactive than the other NiL+ species in these examples. In other
cases the steric hindrance of the bulky Cp group appears to block
reaction pathways.

118

W

The National Science Foundation (grant #CHE 8722111) is
acknowledged for partial support of this work. Also the authors wish
to thank Professor John Wronka for assistance in implementation of
the frequency-swept detector for the MSU ICR, and Professor James
F. Harrison, for helpful discussions and collaborations concerning the
nature of transition metal ion-ligand bonding.

119

Table 5-6

Reactions of small alkyl molecules with metal ions

C3H7CN + Co+ —> CoC2H4+ + CH3CN (.05)
CoCH3CN+ + (22114 (.60)
Coc3H7CN+ (.35)

C2H513 + Co+ —> CoC2H4+ + HI (.78)
CoI+ + €sz (.11)
CoHl+ + C2H4 (.11)

C3H7C13 + 01* —+ C3H7+ + C00 (.72)
CoC3H6+ + HCl (.23)
CoHc1+ + c3116 (.05)

CZHSOHa + Ni+ —) COC2H4+ + H20 (.79)
COH20 + + C2H4 (.16)
COC2H4O + + H2 (.05)

CH3OCH3b+ Ni+ 9 NiCHZO++ CH4 (.61)
NiC2H6O+ (.39)

CZHSCHO + Ni+ -—> NiCO+ + C2H6 (.43)
NiC3H4O+ + H2 (.12)
NiC3H6O+ (.45)

C3H7N02c +

s-C4H9NH2b +

C2H5COCH3C +

C2H5COOH° +

Ni+ ->

Ni+ -—)

Co+ —>

Co+ —)

120

Table 5-6, continued

C3H7+ + NiN02 (.35)

NiHNzO+ + C3H6 (.20)
NiC3H5+ + H2 + N02 (.08)
NiOH+ + C3H6N02 (.06)

NiNO+ + C3H7O (.06)

(plus six minor (<5%) products)

C4H10N+ + NiH (.60)
NiC2H7N+ + C2H4 (.07)
Nic3H7N+ + CH4 (.19)
NiC4H7N+ + 2 H2 (.07)
NiC4H9N+ + H2 (.06)

COC2H4+ + C2H4O (.24)
COC2H4O+ + C2H4 (.52)
C0C3H4O+ + CH4 (.11)
COC4H6O+ + H2 (.05)

COH20+ + C3H4O (.40)
C0C3H4O+ + H20 (.06)
COC2H4+ + C2H202 (.10)
COC2H202+ + C2114 (.19)
COCO+ + C2H6O (.25)

 

a. ref. 4

b. ref 38

c. ref6 (1. ref. 5

Table 5-7

Reactions of aromatic compounds with Ni+ and NiL+

C6H6 + Ni+ (a) —-) N. R.
Ni+ (b) N. R.
Ni+ (c) NiP+
N100+ N”1P+ + CO
Ni(CO)2+ NiP+ + CO
NiCOP+ + CO
NiPF3+ NiP+ + PF3
Ni(PF3)2+ NiLP+ + PF3
NiCp+ NiCpP+
NleNO+ NiCpP+ + NO
C6H5CH3 + Ni+ (a) -—> N. R.
Ni+ (b) N. R.
Ni+ (c) NiP+
NlOO+ NiP+ + CO
Ni(CO)2+ N. R.
NiPF3+ NiP+ + PF3
Ni(PF3)2+ N. R.
NiCp"’ NiCpP+
NiCpNO+ N iCpP+ + NO

 

N. R. = No Reaction
Ni+ (a) = Ni+ produced from Ni(CO)4

Ni+ (b) = Ni+ produced from Ni(PF3)4
Ni+ (c) = Ni+ produced from NiCpNO

C6H50H +

C6H5NH2 +

C6H5COC1 +

Table 5-7, continued

N i+ (a) —9
N i+ (b)
Ni+ (c)
NiCO+
Ni(CO)2+
NiPF3+
Ni(PF3)2+
Nicp+
NiCpNO+

Ni+ (a) —>
Ni+ (b)
Ni+ (c)
NiCO"
Ni(CO)2+
NiPF3+

N i(PF3)2+
NiCp+
NiCPNO+

Ni+(a) ->
Ni+(b)
Ni+(c)
N1CO+
Ni(CO)2+
NiPF3+
Ni(PF3)2+
NiCp+
NicpNo+

NiP+

NiP+

N. R.

NiP+ + 00

N. R.

NiP+ + PF3
NiLP+ + PF3
N. R.

NiCpP+ + NO

N. R.

N. R.
NiP+
NiP+ + CO
N. R.

NiP+ + PF3
NiP+ + 2 PF3

N. R.
N. R.

N. R.

N. R.
N. R.
N. R.
N. R.

N. R.
N. R.

N. R.
NiCpP+ + NO

C6H5CN +

Table 5 -7, continued

Ni+ (a) ->
Ni+ (b)
Ni+ (c)
NiCIfi'
Ni(CO)2+
NiPF3+

N i(PF3)2+
NiCp+
NiCpNO+

Ni+ (a) —)
Ni+ (b)
Ni+ (c)
NiCO+

Ni(CO)2+
NiPF3+
Ni(PF3)2+

NiCp+
NiCpNO+

Ni+ (a) —-)
Ni+ (b)
Ni+ (c)
NiCO+
Ni(CO)2+
NiPF3+

NiP+

NiP+

NiP+

NiP+ + CO
NiC’OP+ + CO
NiP+ + PF3
NiLP+ + PF3
N. R.

NiCpP+ + NO

NiC6H6+ + CO
NiC6H6+ + CO
NiC6H6+ + CO
NiC6H6+ + 2 CO (.29)
NiP+ + L (.71)
NiLP+ + L

NiP+ + L

NiP+ + L (.87)
NiLP+ + L (.13)
NiCpC6H6+ + CO
NiCpP+ + NO

NiP+
NiP+
NiP+

NiP+ + CO
N iLP+ + L
NiP+ + L

C6H5Br +

COHSI +

Table 5-7, continued

Ni(PF3)2+
Nicp+
NiCpNO+

Ni+ (a) —>
N i+ (b)
Ni+ (c)
N1CO+
Ni(CO);),+
NiPF3+
Ni(PF3)2+
NiCp+
NiCpNO+

Ni+ (a) ->

Ni+ (b)

' Ni+ (c)

NiCO+
Ni(CO)2+
NiPF3+
Ni(PF3)2+
NiCp+
NleNO+

NR.

NiCpC6H4+ + HCl
NR.

NiC6H4+ + HBr
NiC6H4+ + 1131'
NiC6H4+ + HBI'
NiP+ + CO
NW++2L
Nw++L

NiPL+ + L
NicpC6H4+ + I-[Br
NiCpP+ + NO

NiC6H4+ + HI (.45)
NiC6H5+ + I (.55)
NiC6H4+ + H] (.37)
NiC6H5+ + I (.63)
NiP+

Nil”r + CO

NR.

NR.

NiPL+ + L
NiCpC6H4+ + HI
NR.

Ni+ (a) —>

Ni+ (b)

Ni+ (c)

NiCO+

Ni(CO)2+
NiPF3+
Ni(PF3)2+

Nicp+

_ NiCpNO+

Ni+ (a) —>

Ni+ (b)
Ni+ (c)

NiCO+

Ni(CO)2+
NiPF3+
Ni(PF3)2+
Nicp+
NicpNo+

125

Table 5-7, continued

NiC5H5+ + C0 + NO (.84)
NiC6H50+ + NO (.16)
NiC5H5+ + C0 + NO (.86)
NiC6H5CO+ + NO (.14)
NiC5H5+ + C0 + NO (.52)
NiC6H50+ + NO (.26)

NiP+ (.21)
NiCOP+ (.28)
NiP+ + CO (.72)
NiCOP++CO

Nil)+ + L

NiP+ + 2 L (.24)
NiPL+ + L (.76)
NiCpP+

NiCpP+ + NO

NiC6H6+ + CH20
NiC6H6+ + C1120

NiC6H6+ + c1120 (.31)

NiP+ (.69)

NiC6H6+ + C1120 + C0 (.17)

NiP+ + CO (.60)
NiCOP+ (.23)
NiP+ + 2 CO
NiP+ + L

NiPL+ + L
NiCpP+

NiCpP“ + NO

Ni+ (a) —)
Ni+ (b)
Ni+ (c)
Nico+

Ni(CO)2+
NiPF3+
Ni(PF3)2+
NiCp+
NiCpNo+

Ni+ (a) —+
Ni+ (b)
Ni+ (c)

NiCO+
Ni(CO)2+

NiPF3+
Ni(PFa.)2’r

Nicp+
NicpNo+

126

Table 5-7, continued

NiC6H5CH3+ + C3H6
NiC6H5CH3+ + C3H6
NiC6H5CH3+ + C3H6
NiC6H5CH3+ + (23116 + co
NiP+ + CO (.56)
NW++2L

NW++L

Nm++2L

NR.
NiCpP+ + NO

NiC6H5CHCH2+ + H2
NiC6H5CHCH2+ + H2
NiC6H5CHCH2+ + H2 ( .63)

NiP+ (.37)

NiCGHSCHCH2+ + H2 + C0 (.41)
NiP+ + C0 (.59)

NiP+ + C0 (.37)

NiCOP+ + C0 (.63)

NiP+ + L

NiLP+ + L (.47)

NiP+ + 2 L (.53)
NR.
NiCpP+ + NO

127

Table 5-7, continued

C6H5CH2NH2 + Ni+ (a) —> NiC6H5CHNH+ + H2

Ni+ (b) NiC6H5C11NH+ + H2

Ni+ (c) NiC6H5CHNH+ + H2 (.66)
MW (.37)

NICO+ NiC6H5CHNH+ + H2 + CO

Ni(C0)2+ NiP+ + 2 C0 (.67)
NiCOP+ + C0 (.32)

Nil>l=3+ NiC6H5CHNH+ + H2 + PF3 (.31)
NiP+ + L (.69)

Ni(PF3)2+ NiP+ + 2 L

Nicp+ NiCpC6H5CHNH+ + H2 (.58)
NiCpC6H5CN+ + 2 H2 (.42)

NicpNo+ NiCpP+ +NO

C6H5CH2CHO+ N1+ (a) —> NiC6H5CH3+ + CO

Ni+ (b) NiC6H5CH3+ +CO

Ni+ (c) NiC6H5CH3++CO

Nico+ NiC6H5CH3+ + 2 C0 (.32)
NiP+ + C0 (.68)

Ni(C0)2+ NiP+ + 2 co

NiPF3+ NiC6H5CH3+ + (:0 + PF3 (.17)
MP” + L (.83)

Ni(PF3)2+ NiP+ + 2 L

NiCp+ NR.

NiCpNO+ NiCpP+ + NO

Ni+ (a) —->

Ni+ (b)

Ni+ (c)
Nico+

Ni(CO)2+
NiPF3+

Ni(PF3)2+

NiCp+
NicpNo+

Ni+ (a) ->

Ni+ (b)
Ni+ (c)

Nico+
Ni(c0)2+
NiPF3+

Ni(PF3)2+

Nicp+
NiCpNO+

128

Table 5-7, continued

C7H7+ + Ni + OH (.75)
NiC6H5CHO+ + H2 (.25)
C7H7+ + Ni + OH (.85)
NiC6H5CHO+ + H2 (.15)
NiC6H5CHO+ + H2
NiC6H5CHO+ + H2 + CO (.37)

NiP+ + CO (.63)
NiP+ + 2 CO

NiC6H5CHO+ + H2 + PF3 (.22)
NiP+ + L (.78)

NiC6H5CHO+ + H2 + 2 L

NR.

NiCpP+ + NO

C7H7+ + Ni + Cl
C7H7+ + Ni + Cl
C7H7+ + Ni + C1
NR.
NR.

C7H7+ + Ni + C1 + PF3
NR.

C7H7+ + Ni + Cp + Cl
NR.

129

Table 5-7, continued

C6H5CH2Br + Ni+ (a) —> C7H7+ + Ni + Br

Ni+ (b) C7H7+ + Ni + Br

Ni+ (c) C7H7+ + Ni + Br

NICO+ C7H7+ + Ni + C0 + Br

Ni(c0)2+ NiP+ + 2 co (.63)
Nic0P+ + C0 (.37)

NiPF3+ C7H7+ + Ni + Br + PF3 (.30)
NiP+ + PF3 (.20)

Ni(PF3)2+ NiP+ + 2 PF3

Nicp+ NR.

NicpNo+ NiCpP+ +NO

C6H5CHOHCH3 + Ni+ (a) —+ NiC6H5CH3+ + CH20 (.69)

NiC6H5CHCH2+ + H20 (.31)

Ni+ (b) NiC6H5CH3+ + CH20 (.52)
NiC6H5CHCH2+ + H20 (.48)

Ni+ (c) NiC6H5CH3+ + CH20 (.24)
NiC6H5CHCH2+ + H20 (.76)

NICO+ NiC6H5COCH3+ + H2 + CO

Ni(C0)2+ NiC6H5COCH3+ + H2 + 2 C0 (.64)
NiP+ + 2 C0 (.36)

NiPF3+ NiC6H5CHCH2+ + H20 + PF3(.70)
NiC6H5COCH3+ + H2 + PF3 (.30)

Ni(PF3)2+ NiP+ + 2 L (.38)

NiPL+ + L (.62)

 

C6H5C0CH3 +

C6H5C00CH3 +

130

Table 5-7, continued

Nicp+
NicpNo+

N i+ (a) —>
Ni+ (b)
Ni+ (c)

NICO+

Ni(CO)2+
Ni(PF3)+

Ni(PF3)2+
Nicp+
NicpNo+
Ni+ (a) —+
Ni+ (b)
Ni+ (c)

NiCO+

Ni(CO)2+
bfiPF3+
Ni(PF3)2+
NiCp+
NicpNo+

NiCpC6H5CHCH2+ + H20 (.28)

NiCpP+ (.72)
NiCpP‘" + NO

NiC6H5CH3+ + C0
NiC6H5CH3+ + co
NiC6H5CH3+ + co (.88)
NiP+ (. 12)

NiC6H5CH3+ + 2 C0 (.42)
NiP+ + CO (.58)

NR.

NiC6H5CH3+ + co + PF3 (.20)
NiP+ + L (.80)

N iPL+ + L

NiCpP+

NleP+ + NO

NiC6H50CH3+ + co
NiC6H50CH3+ + co
NiC6H5CH3+ + C0 (.88)
NiP+ (.12)
NiC6H50CH3+ + 2 co (.45)
NiP+ + CO (.55)

NR.

NiP+ + L

NiPL+ + L

NiCpP+

NiCpP+ + NO

 

Ni+(a) —->
Ni+(b)
Ni+(e)
Nico+

Ni(C0)2+
NiPF3+
Ni(PF3)2+
Nicp+
NicpNo+

Ni+ (a) —>

Ni+ (b)

' Ni+ (c)

Nioo+
Ni(C0)2+
NiPF3+
Ni(PF3)2+
Nicp+
NicpNo+

131

Table 5-7, continued

NiC61150H+ + CO
N1C6H50H+ + CO
NiC6H50H+ + CO
NiC6H50H+ + 2 CO (.32)
NiP+ + CO (.68)

NiCOP+ + CO

NiP+ + PF3

NR.

NR.
NiCpP+ + NO

Ni(C6H5)2+ + CO
Ni(C6H5)2+ + C0
Ni(C6H5)2+ + CO
NiP+ + CO

NiP+ + 2 CO
NiP+ + L

NR.

NR.
NiCpP“ + NO

 

132

Table 5 -8

Reactions of dimethyl sulfoxide, cyclopentanone and
cyclohexanone with Ni+ and NiL+

Ni+ Ni+ Ni+ NiCO+ Ni(C0)2+ NiPF3+ Ni(l>l~‘3)2+ Nicp+ NicpNo+

Dimethyl Sulfoxide (a) (b) (c)

Nis°+ + °2H6 .04 .10 24

NiCH25°+ + CH4 .05 .13 .41 .10

Nicnas°+ + CH3 .43 .47 .18

NiC2H6SO+ (+ L) .47 .29 .16 .90 .74

NiPL+ + L 1.0 .26 1.0

NiCpCH3so+ + N0 + CH3 .46
NiCpP+ (+ N0) .54 1.0
Cyclopentanone

NiC°+ + C4H8 .55 .18 .30

100,116+ + c1120 .45 .30 .16

Nic4ll8+ + co .12 .35

1119+ (+ L) .45 .35 1.0 N. R. 1.0

NiPL+ + L 1.0

Nicp05H6o+ + H2 + No .61
NiCpc4H6+ + c1120 + NO .18
Nicp1>+(+ NO) 21 1.0

 

N. R. = No Reaction

Ni+ (a) = Ni+ produced from Ni(CO) 4
Ni+ (b) = Ni+ produced from Ni(PF3)4
N1“ (c) = Ni+ produced from NiCpNO

Cyclohexanone

mm“ + 051110

Nic3116+ + C2114C0
Nic5H6+ + 013011
NiC5H8+ + 01120
NiC5H10+ + co

NiP+ (+ L)

NiPl.+ (+ L)

NiCpc6H80+ + 112 + NO
Nicp061160+ + 2 H2 + No
NiCpP+ + NO

133

Table 5-8, continued

Ni+
(a)

.56

.24

.20

111* 111*
(b) (C)
.17
.28 23
.19 .15
.06 .09
.08
.30 .45

Nico+ Ni(c0)2+ NiPl=3+ Ni(1>1=3)2+ Nicp+ NicpNo+

1.0

.73
1.0 .27 1.0
.48
.52
1 .0

 

134

References

H

. Allison, J. Prog. Inorg. Chem. 1986, 34, 627.
. Lombarski, M.; Allison, J. Int. J. Mass Spec. Ion Phys.l983, 49,
281.

. Tsarbopoulos, A.; Allison, J. 0rganometallicsl984, 3, 86.

Allison, J.; Ridge, D. P. J. Am. Chem. Soc.1979, 101, 4998.

Lombarski, M.; Allison, J. Int. J. Mass Spec. Ion Proc.l985, 65,31.

Cassidy, C. J .; Freiser, B.S.; McElvany, S. W.; Allison, J.; J. Am. Chem.

Soc.1984, 106, 6125

Stepnowski, R. M.; Allison, J. Organometallicsin press

. Tsarboploulos, A.; Allison, J. J. Am. Chem. Soc.1985, 107, 5085.

9. Corderman, R. R.; Beauchamp, J. L. J. Am. Chem. Soc.1976, 98,
5700

10. Dietz, T. G.; Chatellier, D. 8.; Ridge, D. P. J. Am.
Chem.Soc.l978,100,4905.

11. Chowdhury, A. K.; Wilkins, C. L. J. Am. Chem. Soc.1987, 109,
5336.

12. Squires, R. R. J. Am. Chem. Soc.1985, 107, 4385.

13. Rosenstock H. M.; Draxl, K.; Steiner, B.W.; Herron, J.T. J. Phys.
Chem. Ref. Datal977, 6, suppl. 1.

14. Franklin J. L.; Dillard, J. G.; Rosenstock H. M.; Herron, H. T.; Draxl,
K.; Field F. H. Nat. Stand. Ref. Data Ser., Nat. Bur. Stand.
1969,26

15a. Foster; M. S.; Beauchamp, J. L.; J. Am. Chem. Soc.197l, 93,
4924.

b. Foster; M. S.; Beauchamp, J. L.; J. Am. Chem. Soc.1975, 97,
4808.

c. Weddle, G.H.; Allison, J.; Ridge, D. P. J. Am. Chem. Soc.1977, 99,
105.

16. Lias, S. G.; Liebman, J. F.; Levin, R. D. J. Phys. Chem. Ref.
Datal984,13, 695.

17. Radecki, B. D.; Allison, J. J. Am. Chem. Soc.1984, 106, 946.

18. Wronka, J.; Ridge, D. P. Rev. Sci. lnst.l982, 53, 491.

19. Thermodynamic data were taken and derived from the following

N

999w

we

135

sources. a. ref 14 b. ref 13 c. Weast, R.C.; Astle, M. J.
Handbook of Chemistry and Physics, CRC Publishing Co. 1983.
d. Rosenstock, H.M.; Larkins, J. T.; Walker, J.A. Int. J. Mass
Spec. Ion Phys.l973, II, 309.
In cases where no data were available, heats of formation were
estimated from group equivalent tables given in ref 14.
20. Gaydon, A. G. "Dissociation Energies and Spectra of Diatomic
Molecules" Chapman and Hall, London, 1968 F
21. Burnier, R. C.; Byrd, G. D.; Freiser, B. S. J. Am. Chem. Soc.1981,
103,4360.
22. Halle, L. F.; Crowe, W. B.; Armentrout, P. B.; Beauchamp, J. L.
Organometallics 1984, 3, 1694.
23. Burnier, R. C.; Byrd, G. D.; Freiser, B. S. Anal. Chem. 1980, E
52,1641.
24. Jacobson, D. B.; Freiser, B. S. J. Am. Chem. Soc.1983, 105,7492.
25. Jacobson, D. B.; Freiser, B. S. 0rganomet.l984, 3, 513.
26. a. Dunbar, R.C.; Ennever, J. F.; Fackler, J.P. Inorg. Chem. 1973 12
2735.
b. Meckstroth, W. K.; Ridge, D. P. Int. J. Mass Spec. Ion
Proc.l984,61,149
c. Wronka, J.; Ridge, D. P. J. Am. Chem. Soc.1984, 106,67.
27. Allison, J.; Mavridis, A.; Harrison, J. F. Polyhedron, in press
28. Blomberg, M.; Brandemark, U.; Johansson, J.; Siegbahn, P.;
Wennerberg, J. J. Chem. Phys.l988, 88, 4324.
29. Refeay, K.M. A.; Franklin, J.L. Int. J. Mass Spec. Ion Phys.l976,

 

20.1 9

30. Kiser, R. W.; Krassoi, M. A.; Clark, R. L]. Am. Chem. Soc.1967, 89,
3653.

31a. Jacobson, D. B.; Freiser, B. S. J. Am. Chem. Soc.1985, 107,
7399.

b. Corderman, R. R.; Beauchamp, J. L. Inorg. Chem. 1978 17, 68.
32. Alvarado-Swaisgood, A. B.; Harrison, J. F. J. Phys. Chem. in press
33. Richardson, J. H.; Stephenson, L. M.; Brauman, J.I. J. Chem.
Phys.l973, 59, 5068.

34. Cotton, F. A.; Wilkinson G. Advanced Inorganic Chemistry, John
Wiley & Sons, New York, 1980

35. Tolman, T. A. Chem. Rev. 1977, 77,313.

36. Would we expect IP(NiCO) to be < IP(NiPF3)? We can look at

what happens when Ni is surrounded by 4 ligands:

136

IP(Ni(CO)4) = 8.28 ev,14 IP(Ni(PF3)4) =9.6 ev,30 so it is not

unreasonable!
37. Schilling,J. B.; Beauchamp, J. L. J. Am. Chem. Soc.1988,110,15.
38. Huang, S. K., Ph.D Thesis, Michigan State University, 1983

 

‘ 12"".

Chapter 6

Conclusions

Many areas of gas phase metal ion/molecule reactions have
been studied in recent years. The projects described in this
dissertation add to our knowledge of this field. The reactions of
nitriles and aldehydes show how the functional group influences the
reaction.

In a previous study, the reactions of long alkyl chain
chloroalkanes and alcohols were compared.1 The formation of cyclic
intermediates, allowing metal ion insertion into sites in the alkyl
chain remote from the functional group, dramatically shows that the
alkyl chain can have an active role in the reaction. The alcohols
were observed to form cyclic intermediates to a greater extent than
the chloroalkanes, because the initial interaction between the
molecule and the metal ion is greater for alcohols than chloroalkanes.

With chloroalkanes, Co+ complexes with the Cl, and inserts into the

R-Cl bond, favored by the strong Co+-Cl bond energy.

In the same way that -Cl and -OH have a major role in directing
where bond insertion will occur in the molecule, the -CN and -CH0
groups also have unique directing ability. Fewer products are
formed for aldehydes than the corresponding alcohol, indicating that
the presence of the carbonyl may lead to intermediates that are
more rigid, making fewer C-C bonds susceptible to attack.

Several factors make the reactions of alkyl nitriles interesting.
The strong R-CN bond energy makes this bond inaccessible for metal
ion insertion. The initial interaction between the metal ion and
nitrile appears to lead to an "end-on" complex. The large size of the
end-on complex results in larger cyclic intermediates being formed.

137

 

 

138

Insertion into bonds farther from the functional group (relative to
the alcohols) occurs, with up to eight members in the ring of the
cyclic intermediate.

The addition of a CO ligand to Co"’ had an dramatic effect on the
reactions observed. Very little insertion products were observed for
the CoCO+ ion, and only in the cases of hexyl and octyl cyanide. Two
possibilities are suggested as to why this occurs, both related to
steric effects. The first possibility is that the CO ligand interferes
with insertion into interior C-C bonds due to steric hindrance. The
second possibility is that the ligand does not interfere with insertion,
but the intermediate formed, such as C3H7-Co+(CO)-C3H6CN, may be
less stable due to the steric interactions between the three ligands on
the metal. This implies that insertion intermediates that have
relatively small groups on the metal, i.e., -H or -CH3, will have
higher stability and lead to product formation. These observations
concerning reactions of nitriles with CoCO+ are similar to those made
for metal-ligand reactions with alkanes.2

Aromatic compounds have also been an interesting class of
compounds to study because of the wide variety of reactions
observed. Many were unreactive, such as C6H5Cl and C6H5NH2.
However, some compounds reacted by H transfer from the phenyl
ring: C6H5Br and C6H51. One example of phenyl ring cleavage was
observed was in the reactions of C6H5N02.

The inductive effect of the phenyl ring is the most important
factor in the reactions of the substituted benzyl compounds. The
weak C6H5CH2-X bond is the only bond attacked for the majority of
benzyl compounds. In some cases, charge transfer to C6H5CH2
follows due to the low ionization potential of this group which is
lower than the ionization potential of the metal.

The effect of different ligands on the metal ion can be observed

 

139

in the reactions of NiCO+, NiPF3+, and NiCp+ with aromatic

compounds. The reactions of NiPF3+ follow closely the reactions of
Nico+, with slightly less reactivity for the NiPF3+ ion. The bonding of
NiCO+ and NiPF3+ is probably very similar, so this result is not

surprising. The slightly less reactivity of NiPF3+ reflects the steric
crowding involved for insertion produced by this larger ligand.
Much of the chemistry of the NiL+ ions can be explained by the
formation of intermediates with low steric interactions. It is
important to remember that for these three ligands, the Ni+-L bond
strength increases NiCp+ > NiPF3+ > NiCO+. The size of the ligands, as
measured by their cone angle, also increases in this order. Both
factors should be considered in mechanisms explaining the reactions
of M+-L species.

The reactivity of NiCp+ with the halobenzenes indicate that this
ligand affects the metal ion much differently than either CO or PF3.
Charge transfer to the.cyclopentadienyl group to increase its
aromatic character may explain why this ion is more reactive than
the other NiL+ species in these examples. In other cases the steric
hindrance of the bulky Cp group appears to block reactions.

The study of substituted aromatic compounds with Ni+ and
N iL+ has produced many interesting reactions that perhaps would
not be predicted based on comparison with similar substituted alkyl
compounds. These reactions provide valuable insight into the
question of how transition metal ions react with organic molecules.

Several questions remain unanswered in the reactions of NiL+
with aromatic compounds. In several cases, it was observed that the

NiL2+ (L = C0 or PF3) did not undergo ligand substitution reactions:

140
NiL2++P —) NiLP++L

However, when three ligands were present, the reaction:
NiL3+ + P —-) NiLP+ + 2L

was observed. Unfortunately, the reactions of the NiL3+ ions were
studied only in a few isolated cases. The reactions of these ions may
be useful to determine why ligand substitution reactions sometimes
do not occur.

Another unsettled question is whether Ni+ ions are formed in
different electronic states, depending on the molecule from which
they are formed. Determining the states of the ions may help in
explaining the observed chemistry.

In conclusion, aromatic compounds have provided an
interesting view of the reaction of metal ions with organic molecules.
The small number of products formed in the typical reaction,
combined with the amount of product formed, made these
compounds very easy to study. Because of the variety of reactions
observed, these compounds should prove useful in developing new
ideas about the reactions of metal ions. This, in turn, can lead to a
better understanding of fundamental questions of how organo-
metallic compounds react.

 

141

References

1. Tsarboploulos, A.; Allison, J. J. Am. Chem. Soc. 1985, 107,
5085.
2. Radeki, B. D. Ph.D. Thesis, Michigan State University, 1985

 

APPENDD(

 

APPENDIX

ICR Operating Manual

The following is a short description of the electronic
components that make up the ICR at MSU. The typical operating
parameters are given. The procedure used for obtaining a
spectrum using a frequency swept detector (FSD) is described.
The operation of the instrument has been automated by the use of
a MacIntosh computer; the program used to obtain data with the
computer is also listed.

Magnet The Varian Fieldial magnet control is usually operated at
a constant field of 16 kGauss. Maximum field strength is 23
kGauss. The magnet and power supply is cooled with distilled
water that circulates from a Haskris heat exchanger that is itself
cooled with tap water.

Emission Current Controller This home-built unit (designed and

constructed by Marty Rabb) controls the amount of current that
passes through the ionizing filament. On "EM" mode, the actual
amount of current that makes up the ionizing beam is monitored
and held constant. A voltage proportional to this current is
generated from a Keithley picoammeter (see below), and is input
into the back of this unit. The amount of current is set with the
current knob. On "FIL" mode, the feedback circuit is bypassed.
Filament bias voltage is input on the front panel of this unit via a
BNC connector.

The filament should only be switched on when the magnetic
field is on. The filament on/off switch is on the "little box".

142

 

143

filament current, converts it to a voltage, and passes this signal to
the emission current controller. It is used with the "luA" button
depressed. When the filament is first switched on, the current
through the filament will rise rapidly to about one amp, and the
picoammeter will read .003. If the filament current rises to about
1.5 amps with no. emitted current, this indicates that the filament
is not positioned over the filament hole on the cell, the bias
voltage has not been applied, or there is some problem in
detecting the emitted current. If the current through the filament
remains zero, this indicates that the filament is burned out.

"Th 1i 1 x" Many different electronic signals and voltages
must be interfaced to the cell through a high vacuum feedthrough.
This box sits on top of the high vacuum feedthrough, and provides
a way to apply the voltages necessary to operate the cell. It also
provides the interface for the filament current input and emission

current measurement.

jljektrenix PG 595 Pelee Generater Used in the "locked on" mode,

the pulse generator is used to provide -70 volts to the emission

current control bias input.

fljektrenix m 504 Fenetien Qenereter (FQ) Provides the rf power

that is used to detect ions. It can sweep a frequency range by
setting the stop and start frequencies on the frequency dial,
setting the sweep duration, and having the "free run" button in
the out position. If the free run button is pressed in, the FG will
stay at the start frequency (TRIG SWP mode). A frequency scan
can also be created externally, by applying voltages in the range
of -10 to +10 volts to the "VCF input." This voltage is usually
provided by the MacADIOS. The rf output is sent to the rf
attenuator and also to the 503 counter. When on "free run mode",

 

144

attenuator and also to the 503 counter. When on "free run mode",
the "lin sweep output" is sent to channel X of the oscilloscope. The
"trig output", the 0-5 Volt square wave 90° shifted to the rf
signal, is sent to reference input of the phase sensitive detector.

fljektrenix DC 593 Univetsal Qeantet Measures the frequency of
the If signal from the 504 Function Generator.

RE attenuator Reduces the voltage of the rf signal from the 504
FG. Usually about -40 dB is used. The amount of attenuation
should be set to maximize the signal size. The rf signal is then
sent to the signal summer.

Signal Summer This module splits the rf signal into two 180°
phase-shifted signals. Half of the rf is sent to the excitation plate
of ICR cell, the other half is sent to the balance capacitor. There is
a second signal input, which can be used to add a second rf signal
to the first. The two gate inputs allow both signals to be gated on
or off. The "drift top analyzer" dc bias voltage is input from the
"little box"; the bias voltage is added to the rf signal.

Balanee eapaeiter The balance capacitor is mounted on a BNC "T" to
the vacuum flange along with the preamplifier. The vacuum end
of the "T" is attached to the drift bottom analyzer plate. The
balance capacitor is made of three high quality air capacitors.
They should be adjusted until their capacitance equals exactly the
capacitance of the ICR analyzer plates. The procedure is given in
the FSD instruction manual. Basically they are adjusted until a flat
baseline is reached. This capacitance is very small, about a
picofarad, and cannot be measured with typical capacitance
meters.

 

145

Preamplifier This is mounted on the BNC "T" and provides a 101 x
gain to the ion signal. The drift bottom analyzer dc bias voltage is
input from the little box and is sent to that plate.

Phase sensitve deteeter (PSD) Receives the ion signal from the

preamplifier and compares the signal with the reference signal.
Signals not in phase with the reference signal are rejected. The
reference signal from the 504 FG must be phase shifted to account
for the phase shifts in the components of the bridge detector. This
is done with the digital delay located in the bottom part of the
PSD. The reference signal from the 504 FG has 50!) output
impedance, so the reference impedance switch should be set to
509 and not TTL.

Keithley eleetremeter This electrometer is used as a voltage

amplifier. The main dial should be set to "volts", and 0.1 on the
gain dial. Typically, the unit is operated with power on, but
meter off. This is to prevent damage to the meter needle. The dc
level of the signal baseline can be adjusted with the zero knob.

Qsei_ll_e§_e_ep_e The oscilloscope can be used to monitor the signal as
it is detected. When taking fast sweeps, the spectrum can be seen
using the linear sweep output of the 504 FG as the X channel,
using the oscilloscope in XY mode. The ion signal should be input
into the Y channel.

Buffer A LF 356 op amp is used as a buffer between the amplifier
and the MacADIOS. This is necessary to prevent feedback from
the MacADIOS that occurs when data are analyzed. The signal
buffer is located in the control panel.

146

Eflter A low pass RC filter eliminates most of the high frequency
noise produced by the amplifiers. It processes the signal
immediately before the AINO input of the MacADIOS. The resistor,

R, is 100 1:9, and the capacitor, C, is 104 picofarads.

Wavetek 144 Fanetien generate; This FG is used when doing a

double resonance experiment. The rf is sent to the source of the
ICR cell. The rf can be gated by the MacADIOS at the ”TRIG IN."
The Wavetek FG can also be scanned with an external voltage,
here labeled "VCG IN." This is used in a parent scan double
resonance experiment, with the 504 FG set at a constant
frequency. '

Centre] panel The control panel is a homebuilt panel used to
adjust the bias voltages that go to the 8 adjustable plates of the
ICR cell. The control panel also has timing controls to perform
trapped ICR experiments, but these have never been used. The
"computer" module provides a convenient way to monitor the cell
voltages. Also, a voltage divider is provided to reduce the
magnitude of the frequency controlling voltage ramp from the
MacADIOS.

Obtaining a spectrum: To obtain a spectrum, be sure that all
the components listed above are operational, set to the
appropriate parameters, and connected according to the block
diagram Figure 2-5. Set up the Tektronix FG to sweep an
appropriate range (100-900 kHz) with about 0.1 sec sWeep time
(FREE RUN mode). This sweep should be seen on the oscilloscope
(XY mode). with a sample on an inlet line, open a leak valve
slowly until pressure in the cell is about 5x10'6 torr as read by
the ion gauge. Turn on the filament and adjust the current until

 

 

147

about 30 nAmps of current are emitted. Peaks should now appear
on the oscilloscope. If no peaks appear, the cell probably has to
be "tuned up". Adjust the voltages on the cell plates until peaks
appear. Generally, the trap plates should be ~1 V, the drift bottom
plates ~0 V, and the drift top plates ~-0.5 V. These voltages
should be adjusted so that peak heights are optimized, and may
be quite different than those suggested.

At this point, one must be careful to avoid trapping the ions.
When the ions are trapped, the intensities of low mass ions
decreases as the intensity of the high mass product ions increases
markedly. To test if trapping is occurring, turn off the filament.
The ion peaks should instantly disappear. .If the peaks slowly
decreases, then trapping is occurring. The trapping voltages are
probably too high and should be decreased. Increasing the drift
voltages may help, too.

Other parameters that can be adjusted are the sweep time,
irradiating voltage, and filament current. Sweep time and
irradiating voltage both relate to how much energy is absorbed by
the ions. As sweep time is decreased, the peak intensity
decreases as the ions are absorbing less energy. Also with fast
sweep times, "ringing" occurs. The ion signal is still detected after
the irradiation time, because the ions are in motion for a short
period of time. This ringing is not noticeable at slow scan speeds.
The irradiating voltage can be adjusted with the attenuator. The
voltage should be increased until the height of the highest peak no
longer increases. At the point when the signal no longer increases,
the cell is saturated. Filament current can also be changed to
increase the number of the ions in the cell.

Once a spectrum has been obtained, the next step is to set
up for a double resonance experiment. Turn on the Wavetek PG
and set the frequency to match the frequency of the ion that is to

 

 

148

be ejected. The signal from the Wavetek will appear as a
"squiggle" on the oscilloscope trace. The double resonance rf
should be about 0.5 V peak-to-peak. Carefully adjust the double
resonance frequency until the signal from the Wavetek FG is
positioned exactly over the peak of the ion that is to be ejected.
Now set the Wavetek to gated mode; the signal should disappear.
Set the Tektronix FG to TRIG SWP mode; the scan will stop. At this
point, data is ready to be collected by the computer.

Data collection: Data can be collected from the FSD using the
MacADIOS data acquisition system. Data is stored and
manipulated in arrays called "waves". The program used to collect
double resonance data, "Scanner", is listed below, written in
Microsoft BASIC. The purpose of this program is to collect two
complete spectra. The first spectrum, "irrad", is taken with the
double resonance irradiation on; the other spectrum, "noirrad", is
taken with the same conditions except that the double resonance
irradiation is off. The two spectra are then subtracted so that any
differences can be seen. This result is stored in the subtraction
spectrum, "back".

The MacADIOS has 12 bit analog inputs and outputs and a
+10 V to -10 V range. This means a 20 volt range is divided by
4096 discrete increments; thus the resolution is 20 V/4096 =
4.88mV. Programs must take this into account. Here is the
program followed with line-by-line remarks:

149

SCANNER

1 DIM irrad%(3150),noirrad%(3150),irrado%(3150),noirrado%
(3150),spec%(20),smooth%(5):xl=0:er%=0:smooth%(0)=1:
smooth%(l)=1:smooth%(2)=1:smooth% (3):] :smooth%(4)=l

2 LIBRARY "MacADIOS Calls": CALL mainit

10 FOR i%=-1101 TO 2047

20 CALL ainx(0,0,ll,VARPTR(er%),VARPTR(spec%(0)),0,0,0,0,0,0,0)

3O j%=i%*4.88:CALL aout(j%,5000,0,0):k%=i%+1101

40 CALL integ(VARPTR(spec%(0)),1,l1,VARPTR(x!),-2)

50 irrad%(k%)=x!

60NEXTi%

70 CALL plot(0,20,220,60,VARPTR(irrad%(O)),l000,0,3150,10,2,1,)

110 FOR i%=-1101 TO 2047

120 CALL ainx(0,0,11,VARPTR(er%),VARPTR(spec%(0)),0,0,0,0,00)

130 j%=i%*4.88: CALL aout(j%,0,0,0):k%=i%+1101

140 CALL integ(VARPTR(spec%(0)),1,11,VARPTR(x!),-2)

150 noirrad%(k%)=x!

160 NEXT i%

2 O 0 CALL convolve(VARPTR(irrad%(0)),VARPTR(smooth%(0)),

VARPTR(irrado%(O)),3130,3,0)
210 CALL convolve(VARPTR(noirrad%(0)),VARPTR(smooth%(0)),
VARPTR(noirrado%(0)),3130,3,0)
2 2 0 CALL scale(VARPTR(irrado%(0)),3130,0,1,4,2):CALLscale
(VARPTR(noirrado%(0)),3 130,0,1,4,2)

230 CALL bsave("ICDR","irrad",VARPTR(irrado%(0)),2,6280):
CALL bsave("ICDR","noirrad",VARPTR(noirrado%(0)),2,6280)

240 CALL subi(VARPTR(irrado%(0)),

VARPTR(noirrado%(0)),0,6280,0)

250 CALL bsave("ICDR","back",VARPTR(irrado%(0)),2,6280)

260 BEEP

270 CALL aout(-5372,0,0,0)

150

Remarks:

1 Dimensions all necessary variables,

2 Opens the file holding the MacADIOS routines ("calls").
Starts communication between the MacADIOS and the
Macintosh.

10 Starts the first loop which will collect 3150 data points. The
ramp starts at negative value so that a more of the

allowable output range is used.

20 At each step, 12 points are input into the array "spec%".

30 The counter value i% is multiplied by 4.88, so that the ramp
is incremented by 1 step. This value for j% is output

through aoutO. The ramp is incremented as soon as possible
after collecting data. This allows the maximum time for
the ion signal to stabilize before data is collected. The aout
command also sends a 5 volt signal out aoutl, which is
used to gate the Wavetek FG signal on. Then, 1101 is added
to the counter value; k% will be used to label points in the
array irrad%, which must have positive values.

40 The points in array spec%, are added together (integrated)

and saved as x!, a single precision variable.

50 The current point of the array irrad% is loaded with xl.

60 Returns the loop to line 10.

7O Displays the first collected spectrum so that the user can

see if things are working correctly.

100-160 Repeats the loop 10-60 exactly, with two exceptions. In

line 130, the gate signal to the Wavetek is zero, so no double

resonance rf voltage goes to the cell. Since the ions are not
receiving double resonance, this spectrum is saved as
noirrad%.

200-210 The two collected waves are convoluted by a step

151

function, smooth%. This removes much of the undesired
noise.

220 The convolution step also multiples the signal, so this scaling
step reduces the wave's values to the original,
preconvoluted, values.

230 The waves are saved as irrado% and noirrado%. This allows
the information to be examined using the Macintosh
Manager.

240 Noirrado% is subtracted from the wave irrado% and the
result is stored in the array irrado%.

250 The subtracted spectrum is saved as the array "back".

260 The beep alerts the user that the program is finished
(almost).

270 This resets the ramp to the starting voltage and thus the

Tektronix FG to the starting frequency.

To use this program, the first step is to RUN the program. If
the "tsfr" (transfer) light on the MacADIOS is flashing, and the
frequency read on the Tektronix 503 Counter is increasing, then
the program is successfully running. Halfway through the
program, the computer will display a spectrum, so the user can
see if data is being collected correctly.When it is completed, a
beep will sound. TRANSFER from Microsoft BASIC to MacADIOS
Manager. Once in MacADIOS Manager, then LOAD CONFIGURATION
"plotto". Using the command LOAD FROM BASIC, load "wave" with
irrado, "rave" with noirrado, and "backer" with back. Open the
window View Four and drag "wave" and "back" to this window to
display them. "Wave" should look like the expected mass
spectrum. The baseline may not be flat. "Backer" should be a flat
line. Ions that are the products of the irradiated ion will be seen

152

as a negative dip in "backer".( See Figure 2-9)

This program has also been translated to MacADIOS
Manager language. Although it was not used to gather anybof the
data in this dissertation, it will be listed here, because it should be
more convenient for future experimenters than the BASIC
program. The main convenience is that the experimenter can look
at new data immediately without having to transfer into
MacADIOS Manager to use the display windows there. The
program will not be described because it is essentially the same
program as the BASIC program but with slightly different syntax.

For cycle = 0 to 1

For index = -1101 to 2047
ainx(0,0,11,spec,0,0,0,0,0,0,0)
aout0= (4.88*index)
let rave[index+1101]= integ(spec,l,ll,-2)

next index
aout0=-5372
update[rave]

if cycle = 0 then aoutl = 5000

if cycle = 0 then copywave(rave,wave)

if cycle = 1 then aoutl = 0

if cycle = 1 then convolve(rave,step,ravo,3145,3)
if cycle = 1 then convolve(rave,step,ravo,3145,3)
if cycle = 1 then copywave(ravo,subtracto)

if cycle = 1 then sub(subtracto,wavo,0,3149)

next cycle

A minor inconvenience is that the displayed spectra have no

153

scale telling the viewer what the masses of the ions are. The
experiment produces a frequency scan which is linear. Because of
the inversely proportional relationship between frequency and
mass-to-charge, the spacing between mass peaks is not constant
across a spectrum. But if two of the ions can be identified, the
other masses can be identified by using the following equation,
which uses the mass of the two ions (m1 and m2) and their data
point numbers (n1 and n2, the point at which the center of the

peak occurs.)
mm): {(n-n1)‘[(m2'1 - m1-1)/(n2- up] + m1}’1

The equation finds the range of mass between two data point
numbers. It is convenient to use a calculator to calculate the
masses of the ions. A program for a Hewlett-Packard calculator to
do this calculation is:

RCLZ - RCL3 l/x RCLl 1/x STOS — RCL4 RCL2 - / x RCL5
+ 1/x

To use this program, first enter the program into the calculator.
Then enter the mass of the first ion in memory 1, its data point
number in memory 2, the second ion mass in memory 3, and its
data point number in memory 4. With the program entered, then
enter the data point number an unknown peak, 11, and" press R/S.
The mass of the unknown ion will be calculated. The data point
number for the center of the peaks can be found by using the
cursor in the View Four window.

\ I 1