Trinuclear ruthenium carboxylate complexes as oxidation catalysts

MISSING IMAGE

Material Information

Title:
Trinuclear ruthenium carboxylate complexes as oxidation catalysts
Physical Description:
viii, 139 leaves : ill. ; 28 cm.
Language:
English
Creator:
Davis, Leslie Shannon, 1963-
Publication Date:

Subjects

Subjects / Keywords:
Catalysts   ( lcsh )
Alcohol -- Oxidation   ( lcsh )
Ruthenium carboxylate   ( lcsh )
Oxidation   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1988.
Bibliography:
Includes bibliographical references.
Statement of Responsibility:
by Leslie Shannon Davis.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001127180
oclc - 20082537
notis - AFM4350
sobekcm - AA00004969_00001
System ID:
AA00004969:00001

Full Text













TRINUCLEAR RUTHENIUM CARBOXYLATE
COMPLEXES AS OXIDATION CATALYSTS








By



Leslie Shannon Davis


A DISSERTATION PRESENTED TO THE GRADUATE
SCHOOL OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY








UNIVERSITY OF FLORIDA

1988


eP11 'h-















ACKNOWLEDGEMENTS


An undertaking of this size is rarely accomplished by a single

individual working entirely alone, and this is especially true of

this study. My advisor, Dr. Russell Drago, has been an inspiration,

mentor, and guide throughout this journey. I am indebted to him for

his "idears" and all his encouragement and advice during my sojourn

at Florida. Mrs. Ruth Drago, his kind, gracious wife, opened her

home and welcomed me as family, a gesture I certainly appreciated

and that eased my stay during the past four years. I would also

like to acknowledge Dr. Dave Richardson, Dr. Carl Stoufer, and the

remainder of my committee for their help and support.

The Drago Group as a whole has been an outstanding source of

hope, help and fun during our years together. For all the

camaraderie and aid, I thank each of them. I am especially

grateful, first of all, to my labmates, Alan Goldstein, Tom Cundari,

and Rich Riley, who endured all with happy faces, and were

consistent sources of good humor in the lab. I want to thank Ngai

Wong and Larry Chamusco for their computer and mechanical expertise,

without which much of this work would not have been possible. For

their assistance in various and sundry ways, I thank Jerry Grunwald,

Mark Barnes, and Cindy and Ed Getty. I am also grateful to former







group members Dr. Cindy Bailey and Dr. Iwona Bresinska for the

benefit of their wisdom. To Dr. Carl Bilgrien, the initiator of

this study, I owe a deep debt of gratitude. Very special thanks are

due Mrs. Maribel Lisk for her help, advice, and smiles.

Without the help of many others within the department many

"idears" could not be realized. I am grateful to Dr. Roy King for

his deep understanding of NMR and his willingness to share this

knowledge. The machine shop personnel, Chester, Vernon, and Daley,

were able to make anything I could describe, a talent I am most

grateful for. The creative talents of Rudy and Dick in the glass

shop in deciphering my sketches and still creating what I needed are

greatly appreciated. I also thank Chuck Christ and Paul Sharpe for

their assistance.

The experience of graduate school is not realized entirely in

the laboratory. I am grateful to Fran and Allan Goodman for

illustrating this lesson, and for many, many hours of plain old fun.

I also wish to thank Dr. Linda Lentz and Sasi Kalathoor for their

unswerving encouragement and support.

For first instilling in me an interest in chemistry, I thank

Mrs. Jackie Gay. My love of "things that turn pretty colors" is

entirely due to Dr. Alex Zozulin. I owe Alex an additional debt of

first showing me the joys of research.

My greatest debt is owed my family, without whose love and

support and encouragement I would not have accomplished this feat.

To them--Marcia, Larry and Debbie, and Drew--I dedicate this work.















TABLE OF CONTENTS


agse


ACKNOWLEDGEMENTS .


KEY TO ABBREVIATIONS .


ABSTRACT .


.1ii

. . vi



. . vii


CHAPTERS


I. GENERAL INTRODUCTION .

Catalytic Oxidations ..
Trinuclear Carboxylate Complexes.


II. ALCOHOL OXIDATIONS BY TRINUCLEAR
RUTHENIUM CARBOXYLATE COMPLEXES .


. 14


Introduction . ..
Previous Work . .
Scope of Catalysis ..
Results and Discussion ..
Experimental . ..


.14
.18
.19
.24
.51


III. SYNTHESIS AND CHARACTERIZATION OF A
NOVEL TRINUCLEAR CARBOXYLATE COMPOUND

Background. ........ ...
Characterization. ....... .
Experimental. ..........


. .57


.57
.59
.78









IV. OLEFIN OXIDATIONS BY A NOVEL TRINUCLEAR
RUTHENIUM CARBOXYLATE COMPLEX ....

Introduction . ..
Scope of Catalysis. ....... .
Results and Discussion. ......
Experimental. .... .. .. ...


V. ALKANE OXIDATIONS BY A NOVEL TRINUCLEAR


RUTHENIUM CARBOXYLATE COMPLEX .


Introduction ..
Scope of Catalysis. .
Results and Discussion..
Experimental ..


VI. CONCLUSIONS . .


. 80


. .98


.98
105
105
125


. 126


REFERENCES . . .


BIOGRAPHICAL SKETCH. ............. .


. 129


. 139













KEY TO ABBREVIATIONS


Et20 = diethyl ether

OAc = CH3C02-

pfb = CF3CF2CF2CO2-

prop = CH3CH2CO2'

PPh3 = triphenylphosphine

py = pyridine
tfa = trifluoroacetate













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



TRINUCLEAR RUTHENIUM CARBOXYLATE
COMPLEXES AS OXIDATION CATALYSTS

By

Leslie Shannon Davis


December, 1988


Chairman: Russell S. Drago
Major Department: Chemistry

The family of trinuclear metal carboxylate complexes has been

known to chemists for over 100 years. General studies in terms of

their classical inorganic chemistry, such as ligand exchange or

electron transfer reactions and reactivity, are well documented in

the literature. However, little application of this knowledge has

been attempted.

The series of trinuclear ruthenium carboxylates is very

intriguing'in light of the extensive electrochemistry demonstrated

in these complexes. The ready accessibility to a variety of

oxidation states, combined with the facile exchange of ancillary

ligands, should make these complexes ideal subjects for catalytic

studies.







Previous work has shown a series of trinuclear ruthenium

carboxylates [Ru30(O2CR)6L3]n to be active catalysts for the

oxidation of alcohols to carbonyl-containing products utilizing

dioxygen as the primary oxidant. Continuation of this study has

revealed a mechanism that utilizes the expected synergism between

the metals in this cluster to explain the unique features of this

oxidation.

A previously unknown member of this family, a complex

containing completely fluorinated ligands, has been synthesized and

characterized. Based on the accumulated evidence, this complex has

been formulated as [Ru30(02CCF2CF2CF3)6(Et2O)3](02CCF2CF2CF3). This

complex has been screened as a catalyst for a variety of organic

transformations and has excelled in initiating the free radical

autoxidation of several olefins again using dioxygen as the primary

oxidant. The oxidation studies were extended to alkane oxidations

as well, and were shown to occur by a slightly different mechanism

than that assumed to operate in the industrial, cobalt-catalyzed

oxidation of alkanes like cyclohexane.


viii














CHAPTER I

GENERAL INTRODUCTION


Catalytic Oxidations

The oxidation of organic substrates as a field of interest to

chemists has its origins in the beginnings of the history of

chemistry as a science. Lavoisier, the father of modern chemistry,

demolished the phlogiston theory when he explained the results of

Priestley and Sheele's air experiments.1 Air, he claimed, consisted

of two parts, one of which will support combustion (Priestley's

"fire gas") and one of which will not, and not "phlogiston." 2 He

named the "fire gas" oxygen (for acid former) and formulated the

theory of combustion in the late 1700s. In this origin the modern

field of oxidation chemistry has its roots.

Detailed studies of oxidation processes began in the 1800s.

The degradation of natural rubber was linked to oxygen absorption,

and a great deal of research was aimed at discovering anti-oxidants

for the rubber industry.3,4 The modern theories of autoxidation

processes (as the free radical oxidation of hydrocarbons by 02 is

known), were developed in the early 1900s. The effects of metal

ions on this process were studied during this period by Haber and

Weiss, who formulated the classical mechanism for metal-catalyzed

autoxidation in use today.3,4 (Figure 1.1)

















In2



In. + RH



R, + 02


R02- +


Re + R02


2 R02-


- > 21n,



> Ro



> R02 .


> RO2H


R04R


+ R


R02R


- > nonradical products


Basic autoxidation pathways.3,4


Figure 1.1






3

Autoxidation as a means for producing oxygenated compounds from

hydrocarbons is a highly desired process, although several serious

flaws exist in present processes. Controlling the selectivity of an

autoxidation process, a key element in terms of its usefulness, is

extremely difficult due to the radical nature of the chemistry. A

high activation energy, related to the spin-forbidden reaction

between dioxygen (a triplet state) and organic molecules (a singlet

state) is a barrier as well. Control of the process once initiated

is another disadvantage--the reaction is often hard to stop short of

C02 and H20.

Catalytic oxidations theoretically solve most of these problems

in that the addition of a catalyst should lower the energy barrier,

thus making the reaction easier to start. Product selectivity is

drastically affected by the presence of a catalyst as well. For

these reasons, the "Age of Petroleum" and the "Age of Catalysis" are

inescapably linked.3 Without catalysts to facilitate the conversion

of crude oil to useful products, a petroleum-based economy would not

be possible. Vice versa, without the widespread need for and use of

chemicals and products derived from oil, the study of catalysis

would be relegated to purely academic investigations. Sheldon and

Kochi estimate that today over 90% of the organic chemicals in use

are derived from petroleum, and the majority of petroleum and

petrochemical processes involve the use of catalysts.3 In terms of

the importance of catalytic oxidations, industrial organic

chemicals, including oxygenates from oxidative processes, made up

16.8% of the value of the total chemical industry in 1983.5 Seven








of the top fifty chemicals (by volume) were produced directly from

oxidation processes, and several others were produced from

oxidatively generated intermediates.5

Obviously, catalytic oxidations are industrially valuable.

Serious study and application of homogeneous, liquid-phase oxidation

began in the 1950s.3 Before this time, the majority of industrial

processes used heterogeneous or supported catalysts. However,

homogeneous catalyst systems offer several advantages, especially to

the academician, over their heterogeneous counterparts. Generally

milder reaction conditions (i.e., lower temperature and pressure)

are used in homogeneous processes. Temperatures, mixing rates, and

catalyst concentration are more effectively controlled, and most

importantly, the reaction can feasibly be studied using standard

spectroscopic methods. New, improved surface science techniques

have made the study of heterogeneous catalysts easier, but the

relative perspicuity inherent in homogeneous systems still outweighs

these advances. The major disadvantages of homogeneous systems

industrially are the difficulty in separating products from the

reaction mixture and catalyst recovery. This last deterrent becomes

a major problem when dealing with catalysts containing noble metals

like rhodium or iridium due to their expense.

The advent of the Mid-Century Process (Equation 1-1) and the

discovery of the Wacker process (Equation 1-2) heralded a widespread

interest in homogeneous catalysis as well as organometallic

chemistry as fields of study.6 Emphasis was placed on elucidation










p-CH3-(C6H4)-CH3 > COOH-(C6H4)-COOH Equation 1-1
cat = Co(OAc)2 in HOAc, Br- promoter
200 C
15 30 atm air


CH2=CH2 > CH3CHO Equation 1-2

cat = PdCl2/CuCl2
100 C
10 atm air


of reaction mechanisms and the discovery of new compounds that would

catalyze transformations of organic compounds. Understanding the

chemistry of these processes eventually would lead to improvements

and enhancements of the catalysis. This understanding led to new

growth in both fields, and formed the basis for new expansions of

the chemistry and technology involved in catalytic oxidations.

The disciplines of homogeneous catalysis and organometallic

chemistry are closely related. So much so, in fact, that a major

justification for the study of organometallic complexes has been

their potential use as catalysts. Even though the overwhelming

majority of work in this area has dealt with mono-metallic systems,

the field of multi-metallic catalysts is beginning to emerge as an

area rich in potential for catalytic research. Systems containing

more than one metal have several advantages over their mono-metal

counterparts. Enhanced stability as well as synergistic

interactions between the metals would give multi-metallic systems a

range and versatility unknown in systems containing a single metal.

In theory, the judicious choice of the combination of metals should








lead to a tunablee" catalyst system one where selectivity or

conversion is directly related to the metals involved.

The series of trinuclear metal carboxylate compounds is an

ideal choice for carrying out such studies. Their versatility,

combined with the wealth of knowledge available on the coordination

chemistry of these complexes, make them excellent choices as

subjects for the study of homogeneous catalysis.



Trinuclear Carboxylate Complexes

The family of trinuclear carboxylate complexes has been known

in the chemical literature for over 100 years. Only in the more

recent past have these complexes been extensively studied and

characterized. These studies are extremely interesting in light of

the versatility and uniqueness of multi-metallic systems in general.

The synergistic effects of the presence of two or more metals in

close proximity has been widely studied recently7; several varied

applications of such systems are obvious in biochemistry and enzyme

studies (tryptophan 2,3-dioxygenase, for example, consists of both a-

Cu(II) and an iron porphyrin in the active site)8 as well as

industrial processes involving transition metals on inorganic

supports (SMSI interactions between Ti and Ru and other platinum

metals in Fisher-Tropsch synthesis), and other commercial

applications (oxidations by Co(II) involving Mn(II) as a cocatalyst,

and the widely studied Ziegler-Natta polymerization system which

involves the combination of Zr or Ti and Al as the catalytic

species).








All of the trinuclear metal carboxylate complexes or "basic

carboxylates"9 discovered to date have virtually the same basic

structure (Figure 1.2). (Although not trimers by the strictest

definition, these trinuclear ruthenium carboxylates will be referred

to as "trimers" for the sake of brevity.) The major differences in

these systems occur in metal-metal distances and the planarity of

the M3-O core. The equilateral triangle formed by the metals (as

the apices of the triangular M3-0 core) is bridged above and below

the plane of the triangle by bidentate carboxylate ligands, and each

metal is connected via a central, three-coordinate oxygen atom.

Unlike their dimeric cousins,7 these complexes contain no formal

metal-metal bond. The remainder of the pseudo-octahedral

configuration around each metal atom is completed by the ancillary

ligand L. Obviously a great deal of versatility is inherent in

these complexes not only can the metals used be widely varied, but

the carboxylate bridges and L also increase the permutations

possible. To date, almost all of the first-row transition metalsI0-
14 have been isolated as "basic trinuclear carboxylates" (V,15,16

Cr,17-20 Mn,21-24 Fe,25-29 and Co30-31); others, like Ir,32 Ru,33-37

and Pd, Pt, and Rh38-41 have also been prepared. Titanium42 and

zirconium43 will also form a trinuclear complex slightly distorted

from the traditional basic carboxylate structure involving a central

hydroxy bridge between the metal centers. The carboxylate ligand

can vary from acetate to butyrate for all of these compounds;

partially chlorinated carboxylates as well as fluorinated ones have

also been used. The ligand L is most often a classical coordination









































Figure 1.2 Generalized structure of4 basic trinuclear carboxylates
having the formula [M30(02CR)6L3]n.44








ligand such as pyridine, PPh3, or even H20 or diethyl ether. The

last forms of variation take place in terms of the metals involved.

As these complexes are most commonly isolated, the metals are found

in the +3 oxidation state, causing the cluster as a whole to have a

+1 charge. The other most commonly found form of these basic

trinuclear carboxylates is one with one metal in the +2 oxidation

state, rendering the complex neutral. In these systems, the

assignment of oxidation states is truly a formalism. The iron and

ruthenium complexes in particular can be classified as Robin and Day

Class III compounds, indicating complete delocalization of the metal

electrons. This classification is especially important in the

trinuclear ruthenium carboxylates, the subject of this work.

This versatility has made this family of complexes choice

candidates for a wide range of studies. The mixed-valence, neutral

species (primarily the iron complexes) have been extensively studied

in terms of intramolecular electron transfer reactions.45-48 The

manganese clusters, as well as similar dimeric systems, have been

studied in hopes of elucidating the role of Mn in photosynthesis as

well as for catalysis.49-51 The more classical inorganic chemistry

of these complexes has also been studied, including ligand-exchange

reactions, for example.

Due to the vast information available on these trimers, it is

not unreasonable to expect some studies in terms of their

usefulness. The synergism expected to occur between the metal

centers should manifest distinct differences from their monomeric

analogs. The variety of oxidation states available, combined with








the ready exchange of ligands, makes these complexes ideal choices

for catalysts, especially of homogeneous processes. Finally, the

carboxylate ligands have been shown to be relatively inert to

oxidation processes, as evidenced by the widespread use of metal

acetates (specifically Co(II) and Mn(II)) and acetic acid in

industrial oxidations.52 For these reasons, the family of

trinuclear metal carboxylate complexes would be expected to be good

catalysts for a variety of homogeneous processes.

Only in the last few years have widespread attempts been made

to utilize these complexes as catalysts. The cobalt acetate trimer

(Co30(OAc)6(H20)3+ and others) has been proposed to be one of the

active catalytic species in the oxidation of p-xylene to

terephthalic acid (the Mid-Century/Amoco Process).53 It has also

been shown to oxidize toluene and other hydrocarbons under

relatively mild conditions.13,30,31,53 Others have been used as

catalysts as well. The rhodium acetate complex uses t-butyl

hydroperoxide to selectively oxidize cyclohexene,54 while the iron

acetate trimers have been proposed to catalyze a variety of organic

transformations.55-57 By far the most widely studied (in catalytic

terms) of these trinuclear species is the ruthenium complex.

More literature is available on the ruthenium system in both

catalytic and chemical terms than most of the other trimers. These

complexes were first isolated and characterized33 in 1972; Spencer

and Wilkinson found these trimers to be unique in the family of

basic trinuclear acetates for several reasons. Both mixed-valence

and cationic trimers were readily isolable. These complexes






11

underwent a one-electron non-reversible wave electrochemically, and

readily underwent ligand exchange as well. The most unusual feature

of the ruthenium trimers was the reversible removal of the central

oxygen atom, a reaction unknown for the other trinuclear metal

systems. Later studies by Meyer et al., expanded58-62 the original

electrochemical studies, revealing for the Ru30(OAc)6(pyz)3 (where

pyz = pyrazine) complexes a series of five one-electron reversible

waves. Four of these waves were attributed to the metal center,

corresponding to formal oxidation state changes from Ru(III,III,II)

to Ru(IV,IV,III). Linking these complexes into multinuclear

oligomers revealed systems that would undergo up to ten one- or two-

electron waves, justifying the nickname "electron sponge" for these

ruthenium complexes. A generalized MO scheme,61 shown in Figure

1.3, shows several orbitals of the 7c system of the Ru30 core in a

relatively small energy range. For the cationic, Ru(III,III,III)

complexes, all levels up to El"' are filled; the A2' level is only

partially occupied. The orbitals containing the metal electrons are

virtually indistinguishable, the justification for the Class III

label of delocalization. This depiction also helps explain Meyer's

electrochemistry as well as other spectroscopic properties of these

complexes.

In all likelihood, the electrochemistry revealed for these

complexes prompted the widespread study of the ruthenium complexes

as catalysts. Olefin hydrogenations were first studied63 by

Wilkinson; further studies were carried out both homogeneously and

heterogeneously supported on a carboxylate resin by Rempel and











A2I"


dzx
dx2-y2
dzy


Py
0


0 ----> X



Ru




Figure 1.3 Qualitative molecular orbital description for [Ru30(O2CR)6L3]n
complexes, involving only the 7r system of the Ru30 core. After Wilson et
al.] 1








others.64-66 Ziolkowski, et al,. have studied13,67-69 the kinetics

of cumene hydroperoxide decomposition and the exchange of DMF for

H20 using NMR techniques; they also have reported13 some catalytic

work in alkane oxidations. The ruthenium trimers have also been

involved in the Prins reaction,70 oxidative dehydrogenation of

saturated carbinols,71,72 and dimerization of acrylonitrile.73 In

terms of oxidation catalysis, the ruthenium trimers, in the presence

of hydrogen peroxide, will oxidize substituted phenols to the

corresponding hydroquinone.74 In a mixed solvent system containing

water, carbon tetrachloride, and acetonitrile, the Ru acetate trimer

with periodate will oxidatively cleave alkenes,75 similar to the

traditional chemistry observed for Ru04. They will also catalyze

the isomerization of allylic alcohols.76

A study of the use of the ruthenium trimers as oxidation

catalysts for a variety of organic transformations seemed

potentially interesting, based on their previous use as catalysts

and the large amount of electrochemical potential to be tapped in

these complexes. The use of molecular oxygen as the primary oxidant

has been an ongoing area of research, and the ruthenium trimers have

not previously been shown to be active as catalysts in such a

system. This work involved the continuation of the study of the

oxidation of alcohols by the ruthenium carboxylate trimers,44 as

well as an extension of these catalytic studies to the oxidations of

alkenes and alkanes by a new member of the trinuclear ruthenium

carboxylate family, [Ru30(pfb)6(Et20)3](pfb), which has been

synthesized and characterized.














CHAPTER II


ALCOHOL OXIDATIONS BY TRINUCLEAR
RUTHENIUM CARBOXYLATE COMPLEXES

Introduction

The oxidation of alcohols is a procedure long known and used in

organic chemistry for the production of aldehydes, ketones and

carboxylic acids. Mild reagents, such as Cr03/pyridine or MnO2,

react with alcohols to give primarily the carbonyl product aldehydee

or ketone). Stronger oxidants, like Ru04, continue to oxidize

primary alcohols through an aldehyde intermediate to carboxylic

acids. Other high-valent ruthenium-oxo ions such as Ru042- or Ru04-

will oxidize primary alcohols to carboxylic acids, secondary

alcohols to ketones, and will oxidize unsaturated alcohols without

attacking the double bond.77 Autoxidation of alcohols tends to

produce ketone or acid along with hydrogen peroxide. Shell

commercialized a process for the production of hydrogen peroxide by

the autoxidation of 2-propanol (Eqn 2-1).9


(CH3)2C(H)OH > (CH3)2C=O + H202 (98%) Equation 2-1


The reaction of alcohols with noble metals such as Pd or Pt to

give carbonyl products and a metal hydride species is well-








documented.3,78 A mechanism involving a a-hydride elimination to

give metal hydrides is generally assumed. This mechanism is also

invoked for the Pd(II)-catalyzed oxidation of secondary alcohols to

ketones with oxygen at 25 C.3

A great deal of literature has been published on ruthenium-

catalyzed oxidations of alcohols.79 Besides the general uses of

Ru04, low-valent Ru(II) complexes have been widely studied as

oxidation catalysts with both 02 and milder oxidants such as

iodosobenzene.8 The most widely studied compound of this type,

RuC12(PPh3)3, has been used to oxidatively dehydrogenate alcohols

with oxygen.80 With iodosobenzene, RuCl2(PPh3)3 will selectively

oxidize primary alcohols to aldehydes.8 Sharpless et al. have

found N-oxides combined with RuCl2(PPh3)3 and other ruthenium

compounds will also oxidize alcohols to their respective carbonyl

products.81 In benzene solvent, this complex preferentially

oxidizes long-chain primary alcohols over the corresponding

secondary alcohol.82 Using 02 as the primary oxidant, RuCl2(PPh3)3

oxidizes allyl alcohols to a,j-unsaturated carbonyl complexes in a

variety of relatively poorly coordinating solvents.83 In all of

these oxidations, several general trends arise. All of these

oxidations are shut down in the presence of strong donor solvents

like acetonitrile, indicating coordination of the substrate is

necessary for oxidation to occur. Replacement of a coordinated

nitrile by an alcohol is not highly likely. The mechanism

consistently invoked for these reactions involves the coordination

of alkoxides to a Ru(IV) species with subsequent a-hydride








elimination to give carbonyl product and a Ru(II) hydride. The

hydridic species can be oxidized back to Ru(IV) by the available

oxidant, creating a catalytic cycle.

A wide variety of other ruthenium compounds have also been used

to catalyze the oxidation of alcohols. A ruthenium hydride,

RuH2(PPh3)4, catalyzes the condensation of alcohols to esters and

lactones at elevated temperatures.84 Monomeric ruthenium complexes

containing fluorinated carboxylate ligands have also been shown to

dehydrogenate primary and secondary alcohols via a f-hydride

elimination pathway.85,86 Slightly more active compounds containing

diphosphine ligands have also been prepared and demonstrated to be

catalytic.87 The mixed-valence ruthenium carboxylate dimer

[Ru2(OCR)4C1] has been shown to dehydrogenate methanol to

formaldehyde under relatively mild conditions.88 Ruthenium

complexes as simple as commercially available ruthenium trichloride

have also been shown to be active for both the oxidation of

secondary alcohols and amines with oxygen.89 Ruthenium(III)

solutions will also oxidize allyl alcohol to acrolein.90

Ruthenium complexes containing large, bulky ligands have also

been used for alcohol oxidations. Riley demonstrated a DMSO adduct

of Ru(II), RuX2(DMSO)3L, would catalyze the aerobic oxidation of

thioethers to sulfoxides. This reaction required a reducing

solvent, alcohol, to reduce the Ru(IV) species back to the active

Ru(II) complex, generating a carbonyl product.91

Bidentate imines have also been used with Ru(II) to oxidize

coordinated alcohols in conjunction with 02.92 In these systems, a








Ru(IV) to Ru(II) cycle is again proposed as the pathway of the

oxidation, and a disproportionation step enabling an escape from an

inactive Ru(III) species to active Ru(II) and Ru(IV) complexes is

also invoked. An unusual account of a Ru(III) complex containing

1,3-bis(2-pyridylimino)isoindoline (BPI) ligands is also involved in

the oxidation of alcohols.93 The use of Ru(III) is unusual in that

Ru(III) complexes, generally low-spin t2g5, tend to be

substitutionally inert.94,95 Since the availability of open

coordination spaces is a requirement for a feasible, selective

homogeneous catalyst, Ru(III) complexes would not be expected to be

vary active catalytically. Gagne's system was active for alcohol

oxidations, producing around 60 turnovers (moles of product per

moles of catalyst used) in 24 hours when a strong, noncoordinating

base is present. Secondary alcohols formed ketones which were inert

to further oxidation. Primary alcohols were oxidized initially to

aldehydes (the primary product) which could react further giving

acetals and other products. Again a disproportionation of Ru(III)

to Ru(II) and Ru(IV) is proposed, with a Ru(IV)-coordinated alkoxide

species as the active intermediate. Hydridic ruthenium(II) may be

an intermediate in this reaction as well, arising from the P-hydride

elimination of the Ru(IV)-alkoxide species.

T. J. Meyer has also contributed to this area with his well-

studied ruthenium polypyridyl complexes. Extensive kinetic and

mechanistic studies on alcohol oxidations by these high valent

ruthenium-oxo complexes have been carried out.96-98






18

In light of the extensive, ongoing research into oxidations by

ruthenium complexes in general, and the high potential for catalysis

demonstrated by the trinuclear ruthenium carboxylate complexes,

these particular complexes were chosen to screen as catalysts for

the oxidation of alcohols by molecular oxygen.

Previous Work

Bilgrien discovered that Ru30(prop)6(H20)3+ would catalyze the

selective oxidation of primary and secondary alcohols to the

corresponding carbonyl product using 02 as the primary oxidant.44 A

wide number of alcohols were active in this system, and in all cases

the only product formed was the aldehyde or ketone, with no traces

of carboxylic acid observed. Several different trimeric ruthenium

carboxylate complexes were found to be effective catalysts as well.

These oxidations exhibit a slight rate dependence upon acidity, as

demonstrated by the inhibition of the reaction upon the addition of

acids. On the other hand, bases had a curious effect on the

reaction. Sodium ethoxide enhanced the catalysis, 2,6-lutidine

inhibited the reaction, and NaOH caused precipitation of the

catalyst.

In mechanistic terms, Bilgrien found that for every mole of

carbonyl product produced, a mole of water was formed as well,

implying the four-electron reduction of oxygen to water.44 Hydrogen

peroxide, a likely intermediate in this process, was never detected

in the reaction mixture. These complexes would also oxidize

alcohols with H202 in place of 02 as the primary oxidant. A rough

calculation using the pressure drop of the pressure gauge for the 02








consumption showed that for each mole of 02 consumed, two moles of

product are produced. The rate of the reaction in terms of oxygen

pressure was found to be .25. The catalyst did not seem to

decompose during the reaction, as indicated by both IR and 1H NMR

results. Bilgrien also found that the mixed-valence trimer was

readily oxidized by 02 in alcohol solution to the Ru(III,III,III)

complex, but the reduction of this species by alcohol did not occur.

Bilgrien's mechanism for the ruthenium trimer-catalyzed

oxidation of alcohols is shown in Figure 2.1.44 This scheme invokes

the Ru(III,III,II)-alcohol species as the active intermediate, which

undergoes intramolecular disproportionation to form a Ru(IV,II,II)

ruthenium species. The decomposition of this intermediate could

occur via a number of pathways, the most likely of which involves a

two-step reduction of the alcohol by the trimer. This reduction

would generate the Ru(II,II,II) species, without the central #13-

oxygen first observed by Spencer and Wilkinson,33 which would

readily be oxidized back to the Ru(III,III,II) intermediate in the

presence of 02.

Scope of Catalysis

Apparatus. All pressurized oxidations were carried out in

slightly modified Parr hydrogenation setups (Figure 2.2). This

apparatus has previously been described in detail by Zuzich and

Bilgrien.44,99 For these oxidations, stainless steel pressure

heads, constructed from Swagelok fittings and equipped with standard

sample valves, gas gauges, were also equipped with pressure relief

valves as a safety precaution. This apparatus was directly












RuIII

0
Rul"I RuIII




SRCH20H


RuII


Ru 'I


RuII(RCH2OH)


RuIII(RCH2OH)


Ru'l
H. I

Ru'II RuIV(OCH2R)


RCHO + H20 RCH20H




Figure 2.1 Bilgrien's proposed mechanism for the oxidative
dehydrogenation of alcohols by [Ru30(02CR)6(L)3]n. Other ligands
have been omitted for clarity. "


(H202)


RuII


RulH










-1/4" Silicone
Sceptum
-1/4" Tube to 1/8" NPT
Adapter
On/Off Ball Valve
1/8"NPT


Figure 2.2 Schematic diagramm of a standard pressure head without
safety release valves.









connected to an oxygen tank by copper tubing. The direction and

path of exit gases were controlled by a length of tygon tubing

attached to the exit valve which ran to the back of the hood. A

glass, 250ml, Parr hydrogenation bottle (the reactor vessel) was

attached to this apparatus by a #6 silicone gum rubber stopper and a

metal cage. The bottle is surrounded inside of this cage by an

aluminum sheath, designed to theoretically reduce the amount of

glass shards that would be produced in an explosion.

Sampling techniques have been previously described in detail by

Bilgrien.44 Briefly, a 1-mL gastight string, equipped with a Leur-

lok syringe valve and a 12-inch needle, is inserted through the

septum at the top of the pressure head with the valve closed. The

needle is guided through the ball valve into the reaction mixture.

The syringe valve is opened, a small aliquot withdrawn ( -.2 mL),

the valve closed, and the needle withdrawn. With practice, this

procedure can be accomplished quickly, safely, and with no

observable pressure loss. The aliquots are analyzed using GC,

GC/MS, and GC/IR.

Oxidation procedure. Typical oxidations involved 50 mL of

alcohol (as both solvent and substrate), 1 mL ketone standard, and

10-5 moles of catalyst. Reactions were carried out in a 65 C

silicone oil bath monitored by an Omega 6100 temperature controller

and thermocouple under initial pressures of 40 psig of 02. Stirring

rates of the solutions were controlled by magnetic stirrers beneath

the oil bath; the oil bath was circulated by an overhead stirrer.







23

A slightly modified version of Bilgrien's technique44 was used

for the alcohol oxidations. The pressure bottle was charged with

all components of the reaction except catalyst (i.e., substrate,

standard, and a stirbar), covered with Parafilm, and placed in the

oil bath to equilibrate for 20 30 minutes. The catalyst was added

to the warm solution, the apparatus assembled and pressurized,

placed in the oil bath, and a sample withdrawn. This sample was

denoted as time zero and the start of the reaction. The reaction

was stirred as rapidly as possible to ensure the saturation of the

solution by 02.

Safety precautions. CAUTION! Combinations of warm organic

liquids and dioxygen are potentially explosive. Great care should

be taken, especially during setup and dismantling of oxidation

reactions, to avoid sparking the reaction mixture and causing a

violent explosion. General safety precautions to follow include (1)

let the reaction mixture cool to room temperature before

depressurizing; (2) be sure outlet gases are directed away from any

source of sparks; and (3) become aware of the explosion limits of

solvent, substrates, and the oxidant (whether air or 02) before

beginning an oxidation.

Calculations. Amounts of products formed were determined by GC

in all cases. Calibration curves relating moles of product to

relative peak areas were constructed for all products formed.

Standard procedure involved making up a series of solutions

containing a varying, known amount of product, and a constant amount

of standard in the solvent (alcohol) used. Repeated (at least five)








injections of each of these solutions gave a statistically valid

value for the area percent of the product peak. Knowing the number

of moles of product and standard in each solution gave a mole ratio

of product to standard, which can be plotted against the ratio of

the area percent of the product and standard. The area percent

are obtained electronically from the integration of the peak areas

of the GC chromatogram by an integrator. From the graph of mole

ratio to area percent ratio, the number of moles of product can be

obtained, if the amount of standard added is known.

Results and Discussion

Although Bilgrien's proposed mechanism adequately described the

experimental data,44 further work remained to be done to

substantiate this proposal. Areas to be addressed included the

differences in terms of activity between the mixed-valence and the

cationic trimers, and the fate of the catalyst during the reaction.

Other substrates should be tested, and more kinetic data should be

accumulated as mechanistic support. For these reasons, this

research project was continued.

Bilgrien found the activity of the ruthenium trimer catalysts

varied greatly depending upon the amount of purification of the

complex.44 Liquid chromatography on a four-foot Sephadex LH-20

size-exclusion gel gave the best results, with dramatic effects on

the catalysis, as shown in Figure 2.3. A general activity curve for

the ruthenium propionate trimer, the standard catalyst for most of

the remaining reactions, is shown in Figure 2.4.








150-









100


bI

z
cr


50-



E l l chromatographed
0CXX0 unchromatographed




0 5 10 15
TIME (HRS.)

Figure 2.3 Chromatographed vs. unchromatographed
[Ru30(prop)6(H20)3](prop) in isopropanol oxidations.










1600-


1400-


1200-


1000
Lr

z
- 800


600


400-


200


0
0 50 100 150 2C
TIME (HRS.)

Figure 2.4 Activity curve for [Ru30(prop)6(H20)3]+ catalyzed
isopropanol oxidations.








Other alcohol substrates were screened to further test the

versatility of these trimers as catalysts (Table 2-1). Benzyl

alcohol, as expected, produced only benzaldehyde, and allyl alcohol

was exclusively oxidized to acrolein. Both of these substrates were

oxidized significantly slower than the isopropanol oxidation used as

the common standard for comparison in Bilgrien's work. While

isopropanol oxidations resulted in 147 turnovers in 12 hours, these

substrates only produced 40. Bilgrien also found the rate of

reaction slowed as the substrate varied from primary to secondary

alcohols. These substrates follow this general trend, as the rate

of reaction for both benzyl and allyl alcohol is slower than that

for either primary or secondary alcohols. This reduction in rate

for benzyl alcohol is most probably due to steric bulk and

subsequent hindrance in binding the substrate to the metal center.

The oxidation of allyl alcohol is slower due to a different mode of

substrate binding similar to that proposed by Taqui Khan.90 In the

RuCl3-catalyzed oxidation of allyl alcohol by 02, the substrate is

bound in two sites around the octahedral ruthenium center once by

the double bond and once at the OH moiety. A f-hydride transfer

creates a Ru(III) hydride-alcohol(+) species which is quickly

oxidized by 02 to give acrolein and the regenerated catalyst. No

hydride species is postulated for the trimers, but the relative

slowness of the reaction could be attributed to the inability of the

alcohol to bind at the olefinic site videe infra), and a loss of

stability in the reduced ruthenium-alcohol intermediate.











Table 2-1

Alcohol Substratesa


to/12 hrsb


to/24 hrsb


ethanol


isopropanolc


n-propanolc

n-butanolc

cyclohexanolc

t-butanolc

benzyl alcohol

allyl alcohol

50%
isopropanolf

phenol9


25 No reaction


65

65
100

65

65

65

65

65

65

65


acetaldehyde

acetone
acetone

propanal

butanal

cyclohexanone

No reaction

benzaldehyde

acrolein

acetone


65 No reaction


reaction conditions are as outlined under "Scope of Catalysis"

to = turnovers defined as moles of product/moles of catalyst used

from Bilgrien44

not quantified

reaction run for only 12 hours

auxiliary solvent used was acetonitrile as 50% by volume

solvent used was acetonitrile


substrate


T (oC)


product


313

254
1015

645

d

d


198

147
685

430

d

d



40

42

75









Another congener of the ruthenium carboxylate trimer family,

[Ru30(prop)6(py)3](PF6), was synthesized.. A bar graph comparing all

of the different trimers used in shown in Figure 2.5. The pyridine

adduct is completely unreactive in the oxidation of isopropanol,

indicating that coordination of the substrate in place of the

ancillary ligand L is necessary for catalysis to occur. When

graphed in terms of turnovers, the differences in the cationic and

the mixed-valence trimers becomes even more striking than Bilgrien

reported. The mixed-valence compounds are greater than three times

more active than their cationic counterparts. These differences are

made more enigmatic by the known reaction chemistry of these

complexes. The Ru(III,III,II) trimers are readily oxidized to the

Ru(III,III,III) complexes by 02.33 However, the non-lability of

Ru(III) centers towards substitution is well-documented94,95; the

Ru(III,III,III) trimer would be expected to exchange H20 (or L) for

alcohol ligands very slowly. Assuming coordination of substrate is

necessary for oxidation to occur, the Ru(III,III,III) system should

oxidize alcohols more slowly than the more labile Ru(III,III,II)

counterparts. The exchange of ligands in the Ru(III,III,II) trimer

would be faster, so that even if the oxidation of the trimer from

the Ru(III,III,II) to the Ru(III,III,III) did occur, a molecule of

alcohol would already be present in the coordination sphere of the

catalyst. This would explain some of the differences in the

catalytic activity of these complexes.

A comparison of the ruthenium trimers with other complexes

reported in the literature to oxidize alcohols would be informative














16-

J 700-










o00.




A B C D E


CATALYSTS

Figure 2.5 Comparison of [Ru30(O2CR)6(L)3]n catalysts for
isopropanol oxidations. (A) [Ru30(OAc)6(H20)3]+ (B)
[Ru30(prop)6(H20)3]+ (C) [Ru30(prop)6(py)3](PF6) (D)
[Ru30(prop)6(H20)3] (E) (Ru30(prop)6(PPh3)3].









in terms of gauging the activity of this system. A graphical

comparison is shown in Figure 2.6. As mentioned in the introduction

to this chapter, both RuCl3 and RuCl2(PPh3)3 have been shown to

oxidize alcohols using molecular oxygen as the primary oxidant.80,89

Since the trimers also operate using oxygen, these systems should be

enlightening for comparing relative reactivities of the catalysts.

The trimers are approximately 10 times more active than the other

ruthenium complexes attempted, on the basis of turnovers in 12

hours. Even taking into account that the trimers contain 3 moles of

ruthenium per mole of catalyst, while the others only have one, the

trimers are still over three times more active.

The mechanism of these oxidations could safely be assumed to

not involve autoxidation pathways, due primarily to the selectivity

observed in the reaction. If free radicals were involved in these

oxidations, the further oxidation of aldehydes to carboxylic acids

would be expected. However, acid products are not observed under

our conditions, leading to the assumption the trimers are selective

oxidants. To further justify this claim, typical reactions designed

to prove or disprove free radical chain mechanisms were carried out

(Figure 2.7). The addition of benzoquinone, a free radical trap, to

a typical oxidation has no effect on the reaction. A free radical

initiator, AIBN (azobis(iso-butyronitrile)), was added to the

reaction in place of the catalyst and achieved approximately 10

turnovers in 1 hour and ceased to function. These experiments

emphasize the non-radical nature of these oxidations.








120.00




100.00




80.00

C/)
0,
ULJ

z 60.00
,-
I-


40.00




20.00


0.00


Figure 2.6
oxidations.


1E_11JI Ru30(prop)6(H20)3+
axO0 RuCI2(PPh3)3
VVAAAAA RuC13


5 10
TIME (HRS.)


Comparison of various ruthenium catalysts in isopropanol







150-







100 0


U 00





50


O LLrrnl I Ru30(prop)6(H20)3+
0 CD=/ with benzoquinone
O AAAAAAIBN alone




0 5 10 15
TIME (HRS.)

Figure 2.7 Free radical experiments in isopropanol oxidations.







34

Bilgrien noted that for every mole of product formed, one mole

of water was also produced.44 If the assumption the substrate must

coordinate in order for oxidation to occur is valid, the effects of

adding or removing water in the reaction should prove useful in

determining a mechanism (Figure 2.8). The addition of 5A activated

molecular sieves to the reaction greatly accelerated the rate, while

a reaction run in a 50/50 mixture of isopropanol and water showed a

drop in activity after about five hours. This curvature, indicative

of catalyst deactivation, is not observed in the activity curve

until after 180 hours of reaction time. Seemingly, the presence of

water slowly inactivates the catalyst.

The catalyst does not seem to decompose during catalysis,

according to IH NMR and IR.44 UV-Visible spectroscopy has been very

informative in determining the active species in solution. Since no

induction period is observed for these oxidations, either the

trinuclear carboxylate complexes is the active catalytic species, or

it is a precursor that converts rapidly to the active species in

solution. The lack of an induction period also indicates that the

two different versions of the trimer (Ru(III,III,III) and

Ru(III,III,II)) perhaps perform the oxidation by slightly different

pathways. Bilgrien noted that while the Ru(III,III,II) could be

oxidized to the Ru(III,III,III) in solution, alcohol was not a

strong enough reducing agent to perform the reverse reaction.44

However, a distinct color change is observed when an alcoholic

solution of the catalyst is heated to 43 C under an inert

atmosphere. The changes were monitored via UV as shown in

































1111 Ru3o0(prop).(H2O)3C
0000050% HO20
AAAAA molecular sieves


5 10
TIME (HRS.)


Figure 2.8 Effects of H20 on isopropanol oxidations.


200






150



(n
LJ
z 100

i-




50


15








Figure 2.9. These changes correspond to the conversion of the

Ru(III,III,III) to the Ru(III,III,II) complex as reported by

Wilkinson.33 To effect this change, the solution had to be heated

for 18 hours. However, oxidations were performed at 65 C, so this

conversion may well occur under typical oxidation conditions. If

this conversion were accompanied by production of ketone, the amount

produced (assuming either a stoichiometric conversion either per

mole of catalyst or per mole of ruthenium) was too small to be

detected by GC. This change is reversible; the addition of 02, 30%

H202, or air to the warm alcohol solution immediately oxidizes the

Ru(III,III,II) back to the Ru(III,III,III) with the corresponding

color change. The color change corresponding to this conversion is

not observed under our catalytic conditions; if present, the

Ru(III,III,II) complex would be a transient species at best. These

UV-visible studies indicate a Ru(III,III,II) intermediate created

from a Ru(III,III,III) precursor would be a very slow but possible

process. They give little or no information about the pathway used

by a Ru(III,III,II) precursor, however.

The role of H202 in these oxidations was also pursued further.

Figure 2.10 shows the effects of adding H202 to typical alcohol

oxidations. Hydrogen peroxide is a potent oxidant by itself, as

demonstrated by the upper curves. However, the ruthenium trimer

catalyst will use peroxide in the absence of 02 to oxidize alcohols

to the same carbonyl products. If hydrogen peroxide is an

intermediate in the reduction of 02 as postulated by Bilgrien,44

these graphs indicate the peroxide would be consumed as a co-oxidant

























A
b
S C












400 500 600 700 800
) (nm)






Figure 2.9 UV-Vis studies of [Ru30(prop)6(H20)3](prop) (A)catalyst
in ethanol under N2, 25 "C (B) catalyst in ethanol under N2, 43 C
(C) solution (B) exposed to 02.








350-



300-




250-



200-




150-




100-




50-




(


.u- I


y ~ ~ y w' v v Iu
) 5


I '1 I
10


TIME (HRS.)


I 14


1
15


Figure 2.10 Role of H202 in isopropanol oxidations.


iEE I Ru30(prop)e(H20)3+ under Ar
OCDXDCH202 under Ar
AAAA catalyst and H202 under Ar






L \ '


T






39

in the oxidation reactions. The amount of peroxide formed would, in

all probability, be small and would be consumed as rapidly as it

formed. A low steady-state concentration of peroxide would be one

explanation for the failure to identify peroxide in the reaction

mixture as well.

Determining kinetics in this system was based on the method of

initial rates from initial concentrations.100 The rate expression

was assumed to take the form of Equation 2-2.


dx/dt = kobs [cat]a (P02)b [substrate]c Equation 2-2


In the alcohol oxidations, the substrate alcohol is present in much

higher volume and the conversion of alcohol to product is relatively

small. Therefore, the substrate concentration was assumed to be

relatively constant, giving the rate equation 2-3.


dx/dt = k'obs [cat]a (P02)b Equation 2-3


To obtain the order of the reaction with respect to each remaining

component, one variable was held constant while the other varied.

The rate of the reaction (dx/dt) was assumed to be the slope of the

straight line obtained from a plot of mole of product formed vs.

elapsed time. The appropriate mathematical manipulations gives a

ratio of the rate laws which will yield a value for the reaction

order.(Equations. 2-4,2-5,2-6).








(dx/dt)I = k'obsl [cat]al (PO2)bl Equation 2-4

(dx/dt)i k'obsl [cat]a1 (PO2)bl
--------- -------------------- Equation 2-5
(dx/dt)2 k'obs2 [cat]a2 (P02)b2

Holding one variable constant (for example, (P02)) gives Eqn 2-6.

log (dx/dt)l log (dx/dt)2
a = --------------------------- Equation 2-6
log [cat]l log [cat]2

Several reactions were run where each of the variables was changed

in turn; this data is given in Table 2-2 and graphically in Figures

2.11 to 2.14.

Varying the concentration of the ruthenium catalyst (numbers 1,

4, and 5 in Table 2-2) lead to essentially first-order kinetics

(Figures 2-11 and 2-12). Varying the oxygen pressure was slightly

more demanding in that the total pressure had to be kept at 40 psig

for comparison purposes (numbers 1, 2, and 3 in Table 2-2). The

remainder of the pressure was made up of argon. The reaction order

was found to be approximately .2 in 02, very close to the value of

.25 reported by Bilgrien (Figures 2.13 and 2.14).44 For all

practical purposes, however, the reaction could be considered zero-

order in oxygen, considering the amounts of cumulative error in the

analysis, calibration curves, calculations, and the differences in

the values obtained mathematically and graphically.

A proposed mechanism for these oxidations is shown in Figure

2.15. This scheme differs significantly from that given by Bilgrien

in several areas. The mechanism, beginning with the more labile












Table 2-2

Kinetic Data for Alcohol Oxidations
by Ru30(prop)6(H20)3+


rate law = dx/dt =


[Ru]a
Exp. rx 10-4]


8.89

1.01

9.24

2.80

218.2


k'obs [Ru]a (P02)b


P02
psiQ n/v0


44.0

16.3

27.5

45.5

45.0


.108

.040

.0675

.112

.110


a) Concentration calculated in moles/liter using 50 mL as the total
volume.

b) Initial pressure of reaction in psig

c) N/v calculated from the ideal gas law (PV = nRT) assuming a
volume of 270 mL and 65 C.

d) dx/dt has units of molarity/hour; calculated as explained in text

e) a = 1.20 + .12

f) b = .187 + .07


1.10


dx/dfd
x 10-


8.60

7.16

7.62

2.42

22.3


.185

.259

.119


1.19

1.33







15







Q)
C 10-
0

0
11 1 1 1 18.90 X 0CT4 M
oD OXGG0 O2.80 X 10-4 M
AAAAA 1.82 X 10- M
E 5
E







0 0
0 5 10 15
TIME (HRS.)


varying catalyst concentration.


Figure 2.11 Kinetics:







1.00 -


0.90 -


0.80 -


0.70


S0.60
x
" 0.50
0
1 0.40


0.30


0.20


0.10


0.00


-


2.5


m = 1.04


I I I I I I I
3.0 3.5
log Ru


4.0


Figure 2.12 Kinetics: order in catalyst





















irEE .1079
CeCoI .0400
AA/AAA .0675


TIME (HRS.)


Figure 2.13 Kinetics: varying initial 02 pressure.


12


QD
C
0
-J-
(9



0
E


Atm.
Atm.
Atm.


15








0.50 -


0.45

-4-'
-0
x
-0


0.40


0.35 +-
0.90


m = .140


1.00


1.10 1.20 1.30 1.40
-log (p02) in atm


Figure 2.14 Kinetics: order in 02.


1.50


r 1 1 1 I 1








RuIII
/
0
RuIII


R2CHOH

SLH+
SLOW


L-RuIII


R2CHOH


L-RuIII
\


Ru1 '


HCR2
H


RuIII-L
/
0
RuII


R2CHOH,


L-RuII' RuII-_L
0
RuII
L


FAST


RuIII-L
/


H202


L-RuIII RuIIIL
\ /

RuIII

"O CR2
H


R2C=O


Figure 2.15 Proposed mechanism for the [Ru30(02CR)6(L)3]n -
catalyzed alcohol oxidations. Carboxylate ligands have been omitted
for clarity.


L-RuIII








Ru(III,III,II) species, involves first replacement of the ligand L

by a substrate molecule with concomitant loss of a proton. This

Ru(III,III,II)-alcohol species is postulated to be the active

intermediate in this cycle. Oxidation of this species by 02 (or

later in the cycle, H202) gives a species that can be formulated as

a Ru(III,III,III)-alkoxy radical or a Ru(III,III,IV)-alkoxide

species, depending on the placement of the extra electron. In Robin

and Day Class III systems, this placement is more or less semantics.

Reductive elimination from this species gives carbonyl product and a

coordinatively unsaturated Ru(III,III,II) species. Solvation of

this species by another mole of alcohol regenerates the active

Ru(III,III,II) species.

The Ru(III,III,III) complex is slightly different in that to

reach the active species it must undergo a one-electron reduction

and replace L by a mole of alcohol. Ruthenium(III) species are, in

general, substitutionally inert,94,95 so the replacement step would

be expected to be very slow. The UV-vis studies have demonstrated

the reduction process to be slow as well. These two reasons help

explain the differences between the two congeners.

The kinetics observed experimentally can be verified

mathematically using the mechanism proposed in Figure 2.15. Each

step in the mechanism can be written out and a rate expression

derived for each step (Equations 2-7 through 2-17) using standard

procedures and assuming the steady state approximation is valid for

Equations 2-14, 15, and 16.








kl
- OH2 + ROH \ Ru3'3'2 -
k-1 B


- OR 7 Ru3,3,2 OR


Ru3,3,2
A


Ru3,3,2
B


Ru3,3,2
C


Ru3,3,3


Ru3,3,'2
E


; k4
- OR -----


k5
+ ROH --------->


Ru3,3,'2
E


Ru3,3,2
B


OR + H20
H


-I


Equation 2-7


+ H+ Equation 2-8



+ 022- Equation 2-9



+ R2C=O Equation 2-10


P


- OR
H


dP/dt = k4(D)


Equation 2-11


Equation 2-12


Equation 2-13


dA/dt = -kl(A) + k-_(B)


dB/dt = -k2(B) + k_2(C) k-_(B) + kl(A) = 0
k_2(C) + kl(A) = k2(B) + k_l(B)
k2(B) = k_2(C) + k1(A) kl(B)


dC/dt = -k3(C)(02) + k2(B) k_2(C) = 0
k3(C)(02) = k2(B) k_2(C)


dD/dt = k3(C)(02)
k3(C)(02) =


- k4(D) = 0
k4(D)


Equation
Equation
Equation


2-14
2-14a
2-14b


Equation 2-15
Equation 2-15a


Equation 2-16
Equation 2-16a


Rearrangement and subsequent substitution of Equations 2-14,

15, and 16 into the expression for dP/dt as shown below give the


rate expression for dP/dt in Equation 2-20.


This expression can be


reduced to pseudo first-order in catalyst if k_L(B) is assumed to be


k3
- OR + 02 --k Ru3,3,3 OR
D






49

small. Under our conditions, a large excess of alcohol, the reverse

reaction in Equation 2-7 should only occur to a small extent by Le

Chatelier's Principle, so the assumption seems to be valid.


dP/dt = k4(D) Equation 2-12
= k3(C)(02) (from 2-16a) Equation 2-17
= k2(B) k-2(C) (from 2-15a) Equation 2-18
= [k-2(C) + kl(A) k-i(B)] k-2(C) Equation 2-19
(from 2-14b)

= k1(A) k-l(B) Equation 2-20

Equation 2-21 is the rate expression for the oxygen dependence

obtained from the proposed mechanism. Through appropriate

substitution from Equation 2-16a, this expression takes the form of

Equation 2-22. This equation can be reduced to pseudo zero-order in

oxygen pressure by assuming the concentration of D is constant

throughout the reaction by the steady state approximation.


-d02/dt = k3(C)(02) Equation 2-21
= k4(D) (from 2-16a) Equation 2-22

= k4 Equation 2-23


This mechanism also accounts for the product/02 and

product/water ratios previously observed by Bilgrien. An entire

reaction, consisting of two complete cycles, will produce two moles

of product while reducing one mole of 02 to two moles of water.

Hydrogen peroxide is most probably an intermediate in this

reduction, although never positively identified because it is

consumed as rapidly as it is formed.








A major driving force in this reaction is the large excess of

alcohol available. Ordinarily, the replacement of ligands such as

H20 or PPh3 by the poorly coordinating alcohol would be highly

unlikely. However, with the large excess of alcohol available, the

substitution occurs to a small extent. The low conversion rates

observed in this oxidation (about 2%) are also explained by the

small amount of substitution occurring in these systems.

Interestingly enough, exchange of deuterated methanol for water in a

mixed-metal (Ru2Rh) acetate trimer has been observed in 1H NMR.101

Few detailed NMR studies of these complexes have been

reported,13,66,68 so this exchange may be more extensive than

previously expected. The complete failure of the pyridine adduct to

catalyze the oxidation, even after 24 hours, lends support to the

idea of slow substitution by the alcohol substrate.

None of these theories, however, explain the surprising

activity of the PPh3 adduct. Of all the trimers screened as

catalysts, the mixed-valence Ru30(02CCH2CH3)6(PPh3)3 complex

demonstrated the highest activity. Triphenylphosphine is expected

to be a reasonably strong donor ligand toward Ru(II) (more than

H20), so the substitution by alcohol should be significantly slower

than for the aquo adducts. However, triphenylphosphine is very

easily oxidized to the oxide, a very poor ligand. If all three

phosphine ligands are removed and subsequently oxidized to

triphenylphosphine oxide, the trimer would be essentially naked, and

alcohol coordination would occur rapidly. The presence of

triphenylphosphine oxide was never observed in the reaction mixture;








however, if this hypothesis is true, the quantities of the oxide

would be minute (10-4 to 10-5 moles) and difficult to detect.

The question of nuclearity of the catalyst has yet to be

addressed. The phosphine oxide hypothesis leads to the question of

the number of ruthenium atoms active in the oxidation. In the case

of the triphenylphosphine adduct, theoretically all three atoms

could be involved in the oxidation. A mechanism similar to the

proposal outlined in this chapter could be operating for each metal

center, by virtue of the extensive delocalization over the Ru3-O

core. The synergism and interactions between the metals could

support such reactions, as evidenced by Meyer's oligomers.61 The

spectral data show the catalyst is essentially the same before and

after catalysis, and literature evidence is also available to

support the assumption that the complex remains intact. Considering

the volume of literature available on these complexes with no

reports of decomposition during reaction, it is reasonable to assume

that even under these stringent conditions the cluster retains its

nuclearity. The only physical evidence available is the differences

observed in the catalytic activity of the trimer compared to

monomeric ruthenium systems. The large difference indicates the

chemistry is somehow affected by three metals in close proximity, as

was expected from the outset.

The series of trinuclear ruthenium complexes has not failed in

its promise of producing highly intriguing chemistry. These

complexes have been shown to catalyze the selective oxidation of

alcohol to aldehydes and ketones by dioxygen. These oxidations are






52

presumed to occur via a standard Ru(II)-Ru(IV) cycle, but the cycle

involves the reductive elimination of a Ru(III)-alkoxy radical.

Based on these reactions, these complexes have upheld the potential

promised by their electrochemistry. The unique role of three metal

centers, intimately involved in a chemical transformation, has been

demonstrated, and these complexes manifest unusual catalytic

properties compared to monomeric species. Another enigma is their

catalytic activity, considering that Ru(III) centers are

traditionally inactive species in oxidations. The interactions

between the metals in the trimers can also be supposed to overcome

this trend, and in all probability, actually enhance the catalytic

activity of this system. However, the versatility of these trimers

has not been extensively tested.

Bilgrien found initially these complexes would not oxidize

olefins in alcohol solvent.44 However, changing the solvent to

acetonitrile, widely used in oxidation studies for its inertness,

drastically changed the chemistry. Under 40 psig of 02, cyclohexene

was oxidized to numerous products in the 12 hours. The volume of

products formed generally is indicative of free radical chemistry,

which is antipodally related to selectivity videe infra). The

oxidation of a substrate inert to free radical process, norbornene,

was a complete failure. The lack of success in this area led to

branching out into other trimers containing different ligands.

Experimental

Reagents and equipment. All reagents used were reagent grade

or better and were, for the most part, readily available from






53

Aldrich Chemical Company. All alcohols were passed through a column

of neutral alumina, purity checked by GC, and stored over activated

molecular sieves. If necessary, the substrates were further

purified by standard techniques. Prior to use, the alcohols were

again passed through an alumina column.

GC analysis was performed on a Varian 3300 instrument utilizing

packed, 8-ft, stainless steel columns and both FID and TCD

detectors. Analyses and calibration curves were obtained using 15%

DEGS (diethylene glycol succinate) on Chromosorb W (80/100 mesh). A

Varian 4290 integrator automatically calculated peak areas and

retention times. GCMS was performed on a service basis by Dr. R. W.

King at the University of Florida. All IR spectra were recorded

either as Nujol mulls or KBr pellets on a Nicolet 5DXB spectrometer

and were background corrected. A Perkin-Elmer model 330 UV-visible

spectrometer equipped with a circulating thermal bath was used to

collect UV-vis spectra; all spectra were background corrected.

Elemental analyses were performed on a service basis by the

microanalytical laboratory at the University of Florida.

Synthesis. Trisaquohexakis(propionato)-L3-oxotriruthenium-

(III,III,III) propionate, [Ru30(O2CCH2CH3)6(H20)3](02CCH2CH3), was

prepared as modified by Bilgrien.44 A mixture of 50 mL propionic

acid, 50 mL ethanol, and 1.2 g NaOH were warmed under N2 until the

NaOH dissolved. Two grams of "RuCl3-x(H20)3" were added and the

solution refluxed under nitrogen for four hours until deep green-

black. The solution was cooled to -78 "C for 3-4 hours and filtered








to remove impurities including excess sodium propionate and NaCI.

The filtrate was evaporated on a rotary evaporator and vacuum dried

12 hours at 50 C to give the crude catalyst.

For chromatography on the Sephadex column as described by

Bilgrien, 1 g of crude trimer was dissolved in 100 mL of methanol

and chromatographed in approximately 25 mL fractions. The middle,

blue green fraction was collected, discarding the first and third

"bands," stripped of solvent, and rechromatographed in smaller (5 -

10 mL) fractions This treatment yielded a product whose spectra

matched the reported data. Again, the presence of trace nitrogen in

the elemental analyses of this complex is an enigma. Interestingly,

commercial RuCl3-x(H20) from Aldrich also analyzes for trace

nitrogen, while "pure" RuCl3-x(H20) from Johnson Matthey does not.

Using RuCl3 from Johnson Matthey eliminates the trace nitrogen in

the analyses as shown in the table below. It should be noted that

commercial RuCl3 is an ill-defined, heterogeneous mixture of mono-

and polymeric ruthenium complexes, including oxochloro,

hydroxochloro, and occasional nitrosyl complexes. The average

oxidation state is closer to Ru(IV) than Ru(III), and the main

constituent of RuCl3oxH20 is considered to be a Ru(OH)C13

species.102 Although this does not definitively isolate the source

of the nitrogen in the analyses, this data leads to the conclusion

the nitrogen is most probably inherent in the starting material and

is carried through the reaction.












E


theoretical for
[Ru30(prop)6(H20)3](prop)

Chromatographed
(Aldrich)

crude trimer
(Aldrich)

crude trimer
(Johnson Matthey)

theoretical for
RuCl3-x(H20)

Aldrich
RuCl3-x(H20)

Johnson Matthey
RuCl3-x(H20)


Table 2-3

l mental Analyses

%C %H

28.49 4.68


27.89


29.63


28.33


0.00


0.75


0.24


4.29


4.65


4.34


2.29


2.16


1.62


Tris(pyridine)hexakis(propionato)-A3-oxotriruthenium-

(111,111,111) hexafluorophosphate, [Ru30(02CCH2CH3)6(C5H5N)3](PF6),

was prepared using a modification of Wilkinson's procedure.33 Crude

[Ru30(prop)6(H20)3](prop), (.79 g) was dissolved in 5 mL methanol,

2.5 mL of pyridine was added, and the solution was stirred for 1

hour. A solution of 1 g NaPF6 in 1 mL methanol was added to the

mixture and the resulting solution stored at -40 C for 48 hours.

Dark blue crystals were filtered from the cold solution, washed

three times with diethyl ether, and dried under vacuum at room


%N

0.00


0.50


0.57


0.00


0.00


0.79


0.00






56

temperature for 12 hours. The IR and UV-visible of this complex

matched the reported values. Calculated for

[Ru30(O2CCH2CH3)6(C5H5N)3](PF6): %C = 34.77, %H = 3.95, %N = 3.69;

Found %C = 33.86, %H = 3.86, %N = 3.51.














CHAPTER III


SYNTHESIS AND CHARACTERIZATION OF A NOVEL
TRINUCLEAR CARBOXYLATE COMPOUND


Background

Although trinuclear metal carboxylate complexes have been

widely studied (as mentioned in Chapter I), little variation in the

nature of the bridging carboxylate ligands has been attempted. The

literature reports only two examples where trinuclear carboxylates

have been synthesized using ligands other than alkyl carboxylates -

a ruthenium trimer having dichloroacetate ligands103 and more recent

reports detailing the synthesis of Fe, Cr, and V trifluoro-

acetates.104,105 The lack of such reports, especially for the

ruthenium complexes, most probably stems from Wilkinson's failure to

prepare the trifluoroacetate derivative of the [Ru30(O2CR)6(L)3]n

system.33

As reported in the last chapter, a variety of complexes having

the basic structure Ru30(O2CR)6L3n are catalysts for the selective

oxidation of alcohols employing molecular oxygen as the oxidant.

However, Ru30(prop)6(H20)3+ did not catalyze the reaction of 02 with

norbornene, even after 48 hours of reaction time. This failure

prompted an investigation into routes to a selective olefin

epoxidation catalyst of this general type. Initial attempts

centered around creating a catalyst containing fluorinated

57








carboxylate ligands, thus increasing the acidity of the metal

centers and making the metals more likely to bind an olefin.

Binding the substrate directly to the metal would also provide a

method of selectively oxidizing the substrate to the desired

product.

Previous work in our laboratoryI06 had shown Rh2(OAc)4 to have

a much lower acidity than Rh2(tfa)4, primarily due to the

differences in the electronic nature of the carboxylate ligand.

Doyle and others107-109 extended these observations to the area of

olefin binding. These workers showed that Rh2(tfa)4 will bind

olefins while Rh2(OAc)4 will not, and studied the stability

constants for these reactions. Also, the fluorinated ligands should

be harder to oxidize, making the Ru(III) center a better oxidant. In

light of these discoveries, the exchange of fluorinated for non-

fluorinated carboxylate ligands in the Ru30(prop)6(H20)3+ system

would be an interesting extension of the previous studies.

Heptafluorobutyric acid (pfb acid) was chosen as the exchange

medium due to Doyle's reports108 that the perfluorobutyrate rhodium

dimer bound olefins three times better than the trifluoroacetate

complexes, as well as the fact Wilkinson was unsuccessful33 in his

attempts to perform this exchange with trifluoroacetic acid. The

method used was an adaptation of a previously reported synthetic

route for the conversion of Rh2(OAc)4 to the trifluoroacetate

analogue.106 A typical synthesis involved refluxing crude

Ru30(prop)6(H20)3+ in a 10:1 mixture of pfb acid and pfb anhydride,

stripping away the solvent, and dissolving the residue in diethyl






59

ether. The solution was then filtered and evaporated, leaving dark

black crystals which were dried in vacuo for 12 hours at 50 C.

Characterization

Several spectroscopic methods were used to identify the nature

of this complex. FTIR showed a decided difference between this

complex and the starting propionate trimer (Figures 3.1 and 3.2).

The water absorbance at 3400 cm-I is absent, and the vco stretch has

shifted from 1567 to 1704 cm-'. Other significant differences occur

as well in the CH3 and CF3 regions. As a reference, the vco stretch

for neat perfluorobutryic acid occurs at 1774 cm-.

Proton NMR, shown in Figures 3.3 and 3.4, indicate the absence

of the distinctive peaks representative of the starting material.

The resonances observed are undoubtedly due to a slight impurity,

either in the complex or the solvent since they cannot be attributed

to either coordinated ether or residual starting material. Fluorine

NMR, on the other hand, gives the expected splitting pattern for a

trinuclear carboxylate containing both bound and ionic carboxylates

(Figure 3.5). The resonances are broadened slightly at the base,

indicative of the paramagnetism of the Ru30 core. Again, as a

comparison, the free acid gives rise to three resonances at 82.5,

121, and 128 ppm. The relative insolubility of the complex,

combined with parameters inherent to the program used to transform

the data and the presence of fluorine-containing polymers in the

probe, make precise integration virtually impossible. The best

integrated ratios obtained were 16:3, 11:2, and 10:2, not









































L.
0



4-

A















-4
o.



cv






S'-








LL.
0


4-
0


U-





Q)
01

01,

u-








61


















iD




Of
















/m


LO
LL



03
\ l;


^ "

Y^^Z



c '- I
J3

u^
t ff>
i, 0



^ --'



t .n

, ____ ,________,_________

1 -----------I I '' C

l^. ,7 i~i/" u ^ I7 /-'O 77 9;^ 1 ^ O ZI,


, I


00L,.









































CL











r-.9
co









CLrr
v,
0.l





0-














CL






LA LS..
4-










S--S
=Z










0-



L Im
ro I









>>"





QiS











U_ 1-1





















































































LZO'1S L!
33NVllIrSNVNI%


-o
-u


o D


LL =


LZS 6L


























_Li


n a 2.0 0


Figure 3.3 1H NMR of crude [Ru30(prop)6(H20)3](prop) in CD30D.
* marks residual solvent peaks.









































I I I I I I I I I I I I I
6 4 2 0 PPM






Figure 3.4 1H NMR of [Ru30(pfb)6(Et20)3](pfb) in CD30D. An marks
residual solvent peaks.
































0




ro





4-)
4-








0

cm




Ln
S'
Cr





0.




E-














sa















-i






0
tn






-L










o


-0
-7






-8










-s










-,
OD








significantly different from the 6:1 ratio that would be expected

for a complex having the formula [Ru30(pfb)6(Et20)3](pfb).

The UV-visible spectrum of the perfluorobutyrate complex shows

a similar shift with the absorbances at 610 and 670 nm (of the

original complex) moving to 575 and 760 nm. A new absorbance

appears at 950 nm as well (Figure 3.6). As further evidence for the

existence of the trinuclear species, a titration with pyridine shows

distinct changes in the spectrum upon the addition of three

equivalents of base (Figures 3.7 and 3.8). The shoulder of the

charge-transfer band at 375 shifts to 415 nm with a subsequent

decrease in intensity (E = 2500). The peak at 575 becomes more

distinct as well. These peaks and shoulders, along with the

epsilon values, are given in Table 3.1, as are other spectral data

of interest from IR and NMR spectra.

Molecular weight determinations using the Signer method110 were

quite unsuccessful. Even after 3 weeks of equilibration, a constant

volume for the complex solution was not obtained, indicating the

complex was probably not stable in solution over extended periods of

time. Hovever, FAB mass spectroscopy, a useful technique for

obtaining molecular weights of materials having a high molecular

weight, gave a parent ion peak at 1675 mass units, corresponding to

a protonated Ru30(pfb)6(Et20)3 species. Other significant peaks in

the mass spectrum correspond to the successive loss of coordinated

ether and pfb ligands. (Figures 3.9 and 3.10). Elemental analysis

data further supports the proposed structure of

[Ru30(pfb)6(Et20)3](pfb). Analysis by Galbraith Laboratories gave






69











A
b











A



400 500 600 700 800 900 1000 1100









Figure 3.6 UV-Vis overlay of (A) [Ru30(prop)6(H20)3](prop) and (B)
[Ru30(pfb)6(Et20)3](pfb) in methanol.
































W CD
C a)
0 S-
c ca





.0.0
C Q


4- 4- W
Q. 06 C=

-.-- c-i


CMi CMQ. C
LJ W L4-
00V >



Q. Q. C
00 0
o 0W

4- 4->


0 a







I-' S>J-O0
4--.
o 0
-3






S. -O (2)

M -S-
rrs >
rcr












3 4-0
04 0
.- a, -
4.3 3






















I- 0 > >
= -) *.- .-
CM CD = =
*.- U a- c-
LL- 3 ) )
c, 4. a a


*- Q 0

LI. Cu a







71
















A,
B-N


350


450


550


X (nm)


Figure 3.8 Expansion of the UV-vis titrations with pyridine. 4
[Ru30(pfb)6(Et2O)3](pfb) (B) one equivalent of pyridine (C) two
equivalents of pyridine (C) three equivalents of pyridine.


I-



A
b
s


650
1650













Table 3.1

Spectral Data for
[Ru30(pfb)6(Et2O)3](pfb)


19F NMR (referenced to internal CFC13 at 0 ppm)

80.8 81.2 (triplet)
116.7 117.4 (quartet)
126.7 127.2 (singlet)


FTIR (Nujol mull)

1704(s) 974(m)
1342(m) 936(m)
1224(s) 821(m)
1120(s)


UV-Vis

nm _
375 (sh) 3383
575 (sh) 1574
760 (sh) 1312
950 1444













































4-
0
E
S-

u
0)
0.




E
t,










o co
..-.




c .




S- 0
= CO
Ll_>--






























-o -s
















.- I -
-4 '9
N I


;



















U') 19
r u' n


'-i
,9
















<
<







a
'9



'9
U)


Un

--t



C)















q3





CLI1





-4







0,0
















I I
9




r Ln -


-4


-4.

-4-




Li!)
N
-4


NO

Cu:
Ln:




-4








-4.:
cD







-t4


I I I


'9
'9



































-Q
0.


CvI








































CD
c,




,.













S-

U-
L







di
0.

0.









T3




UJ




CO




3L







77


--






Nf c



nO) G r-.
(- -- (-






c l m







N. G
-4. G4 (S)
(--
















-4..
G L n j ,I-- o











rY
w- 4 -4
nCn














in i
Il-








%C = 21.20, %H = .73, and %F = 47.94, while values calculated for

[Ru30(pfb)6(Et20)1.5](pfb) give %C = 21.24, %H = .78 and %F =

48.46.

All attempts at growing crystals suitable for X-ray analysis

were unsuccessful, probably due to the highly unordered pfb ligands

as well as the ready substitution of water of the ether ligands.

Therefore, no definitive proof for the structure of this complex is

available. However, based on the evidence presented thus far, the

assumption of a complex having the formula [Ru30(pfb)6(Et20)3](pfb)

and the "basic trinuclear acetate" structure is not unreasonable.

Based on this assumption, this complex was screened as a catalyst

for the oxidation of several organic substrates, as will be

discussed in the following chapters.

Experimental

All reagents used were reagent grade or better; the majority

were readily available from Aldrich Chemical Company. FTIR spectra

were collected on a Nicolet 5DXB FT spectrometer either as KBr

pellets or Nujol mulls. Proton and fluorine NMR spectra were

collected on either a Varian XL-200 (at 200MHz) or a Varian VXR-300

(at 300 MHz) FT spectrometers using TMS and CFC13, respectively, as

internal (or external where required) standards at 0 ppm.

Electronic spectra were performed on a PE 331 spectrophotometer and

were background corrected in all cases. Elemental analyses were

performed on a service basis by the University of Florida

microanalytical laboratory or by Galbraith Laboratories (Knoxville,

TN). Mass spectral determinations were carried out at the Middle






79

Atlantic Mass Spectrometry Laboratory at Johns Hopkins University, a

National Science Foundation Shared Instrument Facility.

Synthesis

Tris(etherato)hexakis(heptafluorobutyrato)-(3-oxotriruthenium

(III,III,III) heptafluorobutyrate:

[Ru30(O2CCF2CF2CF3)6(Et2O)3](02CCF2CF2CF3).

Crude Ru30(prop)6(H20)3+ was prepared as described in the previous

chapter. The exchange was carried out by dissolving .5 g of the

crude propionate trimer in a mixture of 10 mL heptafluorobutyric

acid and 1 mL heptafluorobutyric anhydride. The deep green solution

gradually changed to an olive-brown color upon refluxing under N2

for 90 minutes. The mixture was filtered warm and evaporated,

leaving a dark black, gummy solid. This solid was dissolved in 75

mL diethyl ether, filtered and evaporated; this process was repeated

twice. The final solid, obtained as a black powder, was dried in

vacuo at 50 "C for 12 hours. Calculated for

[Ru30(O2CCF2CF2CF3)6(Et20)I.5](02CCF2CF2CF3): % C = 21.24, % H =

0.78, % N = 0.0, % F = 48.46. Found (Galbraith Laboratories) % C =

21.20, % H = 0.73, % F = 47.94. Found (U. of F. Laboratories) % C =

20.91, % H = .12, % N = .11. This complex is highly hygroscopic, in

humid weather becoming quite gummy, and was stored in a desiccator.














CHAPTER IV


OLEFIN OXIDATIONS BY A NOVEL TRINUCLEAR
RUTHENIUM CARBOXYLATE COMPLEX


Introduction

The oxidation of hydrocarbons to a variety of oxygen-containing

organic chemicals is a highly useful industrial transformation as

outlined in Chapter I. Olefinic substrates were investigated

initially by the rubber industry as autoxidation substrates,4 which

led ultimately to the current process for the epoxidation of

ethylene and the Wacker process. Currently, the lack of a feasible

liquid-phase process for the epoxidation of propylene has generated

a great deal of interest in the selective oxidation of olefins.111

Ethylene and propylene are commercially inviting substrates for

study, due to the demand for their respective epoxides for plastics,

solvents, antifreeze, and other chemicals, The epoxidation of

ethylene over a silver-alumina catalyst is a unique, well-studied

heterogeneous system, and will not be further discussed.6,52,112

Propylene, on the other hand, cannot be oxidized to the epoxide

under similar conditions and is currently epoxidized using a

molybdenum-catalyzed process involving alkyl hydroperoxides.6,111

Asymmetric epoxidations have received a great deal of attention

in the literature due to Sharpless' discovery that chiral titanium-

80








isopropoxide complexes catalyze the epoxidation of allylic

alcohols.113-117 Epoxides formed in this fashion are generally

greater than 95% enantiomerically pure. This process has been

licensed by Aldrich Chemical Company and can be used to prepare

intermediates for a host of natural products of interest to the

pharmaceutical industry.118

Oxometal reagents containing most commonly the metals

molybdenum or vanadium, generally in combination with peroxides or

hydroperoxides, have been shown to actively epoxidize olefins.8 A

great deal of controversy concerning the mechanism of the

peroxomolybdenum-catalyzed epoxidation of olefins still exists.

Both Sharpless and Mimoun mechanisms are referred to for these and

similar oxidations.3,8,119,120 Ruthenium compounds, on the other

hand, have long been used to cleave double bond in organic

chemistry. Ruthenium tetroxide in combination with an oxygen

source, is a powerful reagent for cleaving carbon-carbon double

bonds to produce ketones or carboxylic acids.8 The well-studied,

widely-used RuCl2(PPh3)3 has also been shown to selectively oxidize

cyclohexene to the allylic ketone with 02 and styrene to styrene

oxide.119,121 Selective epoxidation by ruthenium compounds is much

harder to achieve, however.

Reports of ruthenium complexes catalyzing a variety of olefin

oxidations using oxygen-atom transfer reagents such as iodosobenzene

instead of 02 abound. Commercially available ruthenium trichloride,

bipyridyl, and periodate in a biphasic solvent selectively oxidized

olefins to epoxides.122 An electrogenerated compound thought to be








[RuV(N40)(O)]2+ (where N40 is bis[2-(2-pyridyl)ethyl][2-oxy-2-(2-

pyridyl)ethyl]amine) is reported to be the active species for the

epoxidation of olefins as well using oxygen-atom transfer

reagents.123

In contrast, only a few ruthenium compounds catalyze the

selective oxidation of olefins with molecular oxygen. The

epoxidation of norbornene was achieved with 02 using several Ru(II)

catalysts.124 This reaction was only about 10% selective to the

epoxide, generating oligomers of norbornene via a ring-opening

process as well as small amounts of norbornanone.

Metalloporphyrins have received a great deal of attention as

researchers try to mimic and understand the activities of biological

systems like cytochrome P-450.125 Iron and manganese prophyrins

have especially been used as probed for this system and will

epoxidize alkenes with oxygen atom transfer reagents.126-130

Recently a ruthenium porphyrin utilizing molecular oxygen as the

oxidant has been prepared by Groves and coworkers.131,132 A

hindered trans-dioxo ruthenium(VI) porphyrin complex, at ambient

temperature and 1 atm 02, will react, albeit slowly, with a variety

of olefins to form epoxides. A similar compound,

[Ru(0)2(dmp)2](PF6)2, where dmp = 2,9-dimethyl-1,10-phenanthroline,

has been shown by Bailey and Drago to epoxidize olefins under

slightly more stringent conditions -- 55 C and 3 atm 02.133 Another

dioxoruthenium(VI) complex containing acetate and pyridine ligands

will oxidize cyclohexene, hexene, and styrene slowly, presumably via

oxygen atom transfer.134








Meyer, in his extensive studies of ruthenium polypyridyl

complexes, has found a Ru(IV)-oxo complex will stoichiometrically or

electrocatalytically oxidize a variety of substrates including

olefins.95,135,136 In general, high-valent ruthenium oxo complexes

have been studied intensively since the proposal that high valent

metal-oxo species are the active intermediates in metalloporphyrin

oxidations.

In light of the recent successes of ruthenium complexes as

olefin epoxidation catalysts, attempting to use the ruthenium

carboxylate trimers to oxidize olefins would be an intriguing

extension of the alcohol oxidation system encountered in Chapter II.

The use of such complexes as catalysts was indeed the primary

justification for the synthesis and characterization of the novel

perfluorobutyrate complex outlined in Chapter III.

Initial tests of the catalytic activity of the new ruthenium

perfluorobutyrate trimer were carried out using n-propanol as the

substrate. Under standard reaction conditions (see Chapter II), no

oxidation occurred after 12 hours of reaction time. The exchange of

alkyl for fluorinated carboxylate ligands was designed to increase

the ability of the metal centers to bind olefins, however.

Attempting to validate the assumption the perfluorobutyrate

complex would catalyze the epoxidation of olefins, a number of

olefins were tested with this complex (Table 4-1). Unfortunately,

the goal of selectively oxidizing olefins was not realized, since

the ruthenium perfluorobutyrate complex actively initiates the free

radical autoxidation of all of the substrates attempted.












Table 4.1

Olefin Substratesa


Substrate


cyclohexene


norbornene




trans-#-
methylstyrene


product


2-cyclohexene-l-ol
2-cyclohexene-l-one
cyclohexene oxide


norbornene oxide:
(exo-2,3-epoxy-
norbornane)


benzaldehyde
acetaldehyde
trans epoxide:
(1R,2R-(+)-l-phenyl-
propylene oxide)


mmoles


1.37
.764
.336


.055
.302



.325
c
trace


turnovers


3 hrs
3 hrs
3 hrs


4 / 24 hrs
22 / 24 hrs



3 / 48 hrs


hexamethyl-
Dewarbenzene


hexamethylbenzene (major)
hexamethylbenzene
oxide (minor)


a) Reaction conditions are slightly different from reaction to
reaction. Specific details can be found under "Scope of Catalysis."

b) Turnovers = moles of product/moles of catalyst used in the
specified time.


c) Not quantified








Scope of Catalysis

The general setup and apparatus used are the same as described

in Chapter II under "Scope of Catalysis." For the olefin

oxidations, the temperature was held constant at 65 *C; all

reactions were performed under 40 psi 02 initial pressure. Except

where otherwise noted, the solvent used was acetonitrile, and in all

cases a minimum 100-fold excess of substrate was used. All

reactions were monitored via GC or CG/MS. The procedure used varied

depending upon the state of the substrate. For norbornene, a solid,

the substrate was dissolved in 50 mL acetonitrile in the pressure

bottle and placed in the oil bath. Upon dissolution of the solid,

the catalyst was added and the apparatus assembled. No preliminary

preparation of norbornene was necessary; however, for some of the

liquid substrates used, pretreatment by washing through a neutral

alumina column to remove peroxides was required. In the case of

cyclohexene, a 20% by volume solution in acetonitrile (10 mL

substrate/40 mL solvent) was used; the other reactions were carried

out using 2 mL substrate in 50 mL solvent. No internal standard was

used in any of these reactions except 2-octanone in the cyclohexene

oxidation. In general, approximately 20 mg of catalyst was used,

corresponding to 10-5 moles. Products were determined by GC using a

DEGS column and FID detector. The amounts of products were

determined from a calibration curve relating moles of products to

relative area percent as described in Chapter II.








Results and Discussion

The oxidation of cyclohexene was attempted first, since it is

one of the easiest substrates to oxidize.3 After an induction

period of one hour, virtually all of the products typical of a free

radical autoxidation process were observed. The major products,

cyclohexene oxide, 2-cyclohexene-l-ol, and 2-cyclohexene-l-one, were

formed in roughly a 4:16:9 molar ratio after 3 hours. Significant

amounts of other products (approximately 10) were also observed but

not quantified. The addition of benzoquinone, a free radical trap,

inhibited the reaction for a finite period (between 6 and 9 hours,

depending on the amount of benzoquinone added), after which the

reaction resumed. Presumably, oxidation of the alkene resumes after

the oxidation of the quinone is complete.

At the other end of the spectrum in terms of oxidizability lies

norbornene. This substrate is widely used to prove the existence of

non-radical pathways in catalytic oxidation studies, since the kinds

of allylic hydrogen abstraction so prevalent in cyclohexene

oxidations are not possible in norbornene. Most of the norbornene

radicals produced are highly unstable, and would be expected to

decompose into alcohol and ketone products, as well as epoxide. An

induction period of 24 hours was observed in the oxidation of

norbornene also, after which primarily norbornene oxide was

produced.(Figure 4.1). This induction period is similar to that

seen in the oxidations by the high-valent oxo-ruthenium complex

Ru(dmp).133 The Ru(pfb) complex was slightly less active than this

previously reported catalyst, producing 22 turnovers in 48 hours as








70



60


50-



> 40-
z
t---
30-



20-



10-




0 50 100 15C
TIME (HRS.)


Figure 4.1 Activity curve for the oxidation of norbornene by
[Ru30(pfb)6(Et2O)3](pfb).






88

compared to 37 turnovers in the same period of time for the Ru(dmp)

catalyst.

Based on this result which seemed to indicate the selective

epoxidation of olefins by the Ru(pfb) catalyst, the oxidation of

trans-f-methylstyrene was attempted. This substrate had previously

been used by Groves to determine both the stereoselectivity and

possible mechanistic pathways for olefin epoxidation by his Ru(VI)

porphyrin.132 The oxidation of trans-f-methylstyrene proceeded very

slowly. Only trace amounts of the trans epoxide were produced after

40 hours of reaction time. The major products of the reaction were

those due to the cleavage of the double bond--benzaldehyde and

acetaldehyde--indicative of a radical process.

Obviously, the perfluorinated ruthenium trimer acts as a potent

free radical initiator. Traylor, et. al, have shown that iron heme

complexes cytochromee P450 analogs) also will catalyze radical-based

oxidations of alkenes.126,127 In order to distinguish between a

free radical chain mechanism and a caged radical pair, the substrate

hexamethylDewarbenzene was chosen. Under Traylor's conditions,

autoxidation processes produce hexamethylbenzene as product, while a

caged radical pathway produces epoxide when m-chloroperbenzoic acid

is used as the oxidant. At 65 C and 3 atm 02, however, this

substrate is extremely reactive. No observable distinction in either

amount of type of product formed could be made between a blank

(using 02 and no catalyst present) and a typical catalytic run.

Another major disadvantage of this substrate is its sensitivity to

light.137-140 For these reasons, hexamethylDewarbenzene has limited








use as a substrate for mechanistic information in catalytic

oxidation studies under these stringent conditions. However, these

results do support the free radical nature of these oxidations.

These reactions show a marked solvent dependency as well (see

Table 4-2). Using norbornene as the substrate, a series of

reactions were run in a variety of solvents. In acetonitrile, 30

turnovers in a 48 hour period were achieved, while no reaction was

observed in benzonitrile, pyridine, or nitrobenzene. Approximately

5 turnovers in 48 hours were achieved in ethanol. These results can

be attributed to the increased solubility of 02 in acetonitrile

compared to the other solvents attempted. The decrease in activity

in ethanol is attributed primarily to its ability to act as a free

radical trap. These experiments, combined with the fact the addition

of AIBN (azo-bis(isobutyronitrile)), a free radical initiator,

decreases the induction period and increases the number of turnovers

achieved (from 22 to 72 in 48 hours) all indicate a free radical

mechanism is involved in the oxidations (Figure 4.2). A caveat is

implicit in these results as well--norbornene is not as inert to

allylic hydrogen abstraction as has been previously assumed.

Observing changes in the catalyst during or after the reaction

would give some insight into the role the catalyst plays in these

reactions. The presence of the fluorinated ligands in the catalyst

enables the fate of the catalyst to be relatively easily monitored

via 19F NMR. Variance in the structure of the compound, changes in

oxidation state, or complete degradation of the catalyst could be

discerned from changes in the resonances of the fluorinated atoms of













Table 4.2

Solvent Dependency
in Norbornene Oxidations


Turnovers


solvent

acetonitrile

pyridine

benzonitrile

ethanolb

nitrobenzene


24 hrs

3.5

0.0

0.0

3.0

0.0


48 hrs

29.8

0.0

0.0

5.0

0.0


a) Turnovers = moles of product/moles of catalyst.

b) No oxidation products from the solvent were observed.









CXXX0 catalyst
A\AAA catalyst
S1 1 1 catalyst


with AIBN
with benzoquinone
alone


TIME (HRS.)


Figure 4.2 Free radical experiments in norbornene oxidations.


100




80




60'




40-




20-








the ligands. Any alteration of the catalyst would justify any

mechanistic considerations as well.

Several attempts were made to identify the nature of the

catalytic species during and after the reaction using norbornene

oxidations. The activity of the catalyst levels off after 150 hours

of reaction time in a typical norbornene oxidation. Analysis of

the spent catalyst indicate the perfluorobutyrate complex decomposes

during the reaction. Fluorine NMR of aliquots of actual reaction

mixtures taken before the reaction, at the end of the induction

period (24 hours), and at 48 hours indicate significant changes in

the catalyst are occurring (Figure 4.3) A new resonance at 119 ppm

appears after 24 hours of reaction time, while the resonance at 117

begins to disappear. The 117 ppm peak has completely vanished at 48

hours, leaving only the 119 ppm and the original 116 ppm peaks in

that region. No integration of these peaks is possible, due to the

low concentrations of catalyst (10-5 M), the low signal-to-noise

ratio, and the baseline roll, an inherent feature of the probe in

fluorine NMR. The vCO stretch in the FTIR shows a similar shift

towards the free acid, shifting approximately 20 wavenumbers to

higher frequency, from 1704 to 1720 cm-'.

The products observed in the oxidation of these olefins, and

the proposed decomposition of the catalyst are consistent with a

typical free radical autoxidation mechanism similar to the standard

Haber-Weiss scheme mentioned in Chapter I (Figure 4.4). This

proposed mechanism involves the one-electron reduction of the

catalyst in the presence of an olefin substrate (bound to a Ru