Hydrocarbon oxidation using molecular oxygen and hydrogen peroxide catalyzed transition metal complexes

MISSING IMAGE

Material Information

Title:
Hydrocarbon oxidation using molecular oxygen and hydrogen peroxide catalyzed transition metal complexes
Physical Description:
iii, 125 leaves : ill. ; 29 cm.
Language:
English
Creator:
Ison, Ana
Publication Date:

Subjects

Subjects / Keywords:
Chemistry thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Chemistry -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2004.
Bibliography:
Includes bibliographical references.
Statement of Responsibility:
by Ana Ison.
General Note:
Printout.
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 - 003103634
System ID:
AA00011364:00001

Full Text











HYDROCARBON OXIDATION USING MOLECULAR OXYGEN AND
HYDROGEN PEROXIDE CATALYZED BY TRANSITION METAL COMPLEXES














By

ANA ISON


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


2004































Dedicated to my husband Elon and daughter Mya, my favorite people in the world.















ACKNOWLEDGMENTS

The past 5 years have been full of joy and sadness, success and defeat,

determination and loss of hope. There are many people who have helped me look past

the hard times. Intentionally or not they were instrumental in pushing me toward

reaching my goal.

I will start by thanking my advisor Dr. David E. Richardson. He has been the ideal

mentor, always skillfully steering me in the right direction, while making me feel like I

was the one figuring everything out. He has always treated me with respect and

understanding, and I will always appreciate that. Most importantly I would likely not be

writing this if not for Dr. Richardson's encouragement. He has always understood why

this process was so difficult for me, but he never let me quit.

Dr. Khalil Abboud has been a great mentor and friend over the years. I very much

appreciate the opportunity to spend 2 years working with him at the x-ray crystallography

lab. Dr. Abboud has always been there to lend a helping hand or a word of

encouragement.

I extend my appreciation to Dr. Michael Gonzalez for giving me the opportunity to

participate in a joint project with the EPA during my last year.

During my time at UF, I have met some really fine people. I will never forget my

Serbian brethren (Ivana Bozidarevic, Nebojca Ciric, Tamara Blagojevic, Ksenija Glusac-

Haskins, Aleksa Jovanovic and many others). Besides helping me perfect my native

language, these folks have been good friends and great company.








Richadson lab members past and present have provided for an interesting work

environment. Dr. Ken Weakley was my southern conservative foe. I hated to see him

leave, but I could not wait to get Rush off the radio! As much as our opinions differed on

many fronts, we got along great. Ken is truly a one-of-a-kind guy, and I appreciate his

help in getting started in the lab. Dr. Deon Bennett has continued to be a good friend

even after leaving the lab. Celeste Regino and "The Guys" (Dan Denevan, Andy Burke

and Mike Mitchell) have definitely brought some good laughs to the lab. I give special

thanks to "The Guys" for putting up with my more moody days.

Most of all, I have to thank my family. My father and mother, Ivan and Duska

Bitanga, I thank for providing for me over the years. My mom especially has always

encouraged me not to give up on my graduate studies. My brother, Tom Bitanga and his

family I thank for their continued support.

My husband Elon has probably been one of the most influential people in my life.

He has been my best friend and my harshest critic. Without him by my side, I would

never have been able to finish this endeavor. Although I didn't always appreciate his

insistence that I need to finish what I started, I have to admit that he was right.

Finally I have a debt of gratitude to pay to my sweet little girl Mya. She was born

into this world just as Elon and I were beginning our graduate careers and she has made it

through along with us. It wasn't always easy for her when mommy and daddy were too

tired to play, or read that extra book. She has helped to keep me sane during my darkest

hours, and always managed to put a smile on my face when I needed it most. I love her

more than anything in this entire world, and I can only hope that she never forgets that.















TABLE OF CONTENTS

page

ACKNOW LEDGM ENTS ................................................................................................. iii

LIST OF TABLES........................................................................................................ viii

LIST OF FIGURES ......................................................................................................... x

ABSTRACT............................................................................................................... xiv

CHAPTER

1 INTRODUCTION ........................................................................................................

M olecular Oxygen as Term inal Oxidant ......................................................................
Free-Radical Autoxidation ....................................................................................
M etal-Catalyzed Peroxide Decomposition.........................................................3...
Electron-Transfer M echanism ...............................................................................
M etal-M ediated M olecular Oxygen Activation ..................................................11
Hydrogen Peroxide as Term inal Oxidant ................................................................ 14
Alkene Epoxidation by Early Transition-Metal Catalysts................................ 14
M etal Porphyrin Catalysts................................................................................ 16
The M n-Salen Assym etric Catalysts ................................................................ 18
The Mn2+/HCO3-/H202 Catalyzed Alkene Epoxidation...................................18
Scope of the Dissertation......................................................................................... 20

2 IRON (II) o,a'-DIIMINE CATALYZED HYDROCARBON AUTOXIDATION ...22

Introduction............................................................................................................ 22
Results and Discussion ............................................................................................ 26
Catalyzed Cum ene Autoxidation...................................................................... 26
Catalyzed Cum yl Peroxide Decomposition...................................................... 30
Proposed M echanism ........................................................................................ 32
Solvent Dependence ......................................................................................... 42
Oxygen Pressure and Temperature Dependence .............................................. 44
M etal Concentration Dependence .................................................................... 45
Ligand Dissociation Studies............................................................................. 47
Precursor Degradation ...................................................................................... 50
Co-oxidation by Reaction Products.................................................................. 51
Active Catalyst Lifetim e.............................................................. ................. 55








Ligand variation studies ................................................................................... 55
Conclusions............................................................................................................ 59

3 HYDROCARBON OXIDATION CATALYZED BY FeL2X2 COMPLEXES.........61

Introduction............................................................................................................ 61
Results and Discussion ............................................................................................ 63
Cum ene Oxidation............................................................................................ 69
Solution-state studies of Fedmp ....................................................................... 71
Conclusions............................................................................................................ 77

4 COMPARISON OF FeL3X2 TO KNOWN AUTOXIDATION CATALYSTS.........79

Introduction............................................................................................................ 79
Results and Discussion ............................................................................................ 79
Fe and Co Acetylacetonates ............................................................................. 79
Fe Complexed by M acrocylic Ligands............................................................. 80
Fe Complexed by Hexa and Tetra Coordinating Pyridyl Ligands ...................81
Poly-Nuclear Iron Complexes.......................................................................... 88
Conclusions............................................................................................................ 90

5 ALKENE EPOXIDATION CATALYZED BY TRANSITION METAL
COMPLEXES USING BICARBONATE-ACTIVATED PEROXIDE (BAP) .........91

Sulfonated Styrene Epoxidation with Mn(III)porphyrin/BAP ................................91
Introduction............................................................................................................ 91
Results and Discussion ............................................................................................ 92
Styrene Epoxidation with Jacobson's Catalyst/BAP............................................... 96
Introduction............................................................................................................ 96
Results and Discussion ............................................................................................ 98
Conclusions......................................................................................................... 103

6 SUM M ARY ............................................................................................................ 104

7 EXPERIM ENTAL.................................................................................................. 107

General......................................................................................................................107
M materials ...................................................................................................................107
Kinetic M odeling .................................................................................................... 108
Oxidation Experim ents .......................................................................................... 108
Peroxide Decomposition Experim ents ................................................................... 108
Sulfonated Styrene Epoxidation ............................................................................. 109
HPLC Analysis ....................................................................................................... 109
Styrene Epoxidation............................................................................................... 109
Synthesis ................................................................................................................... 110

APPENDIX-VARIATIONS IN RATE CONSTANTS................................................1...14








LIST OF REFEREN CES............................................................................................... 119

BIOGRAPHICAL SKETCH ........................................................................................ 125















LIST OF TABLES


Table page

2-1. Catalyzed oxidation of cum ene........................................................................... 28

2-2. Effects of ROOH and H20 on induction period .................................................29

2-3. Experimental and calculated oxygen equivalents accounted for in products
RO O H RO H and RO ......................................................................................... 36

2-4. Simulated and experimental results for uncatalyzed cumene autoxidation...........42

2-5. Solvent effects on cumene oxidation catalyzed by 1..........................................42

2-6. Variation of ligand equivalents........................................................................... 47

2-7. Cumene oxidation catalyzed by Ru analogue of 1..............................................48

2-8. C ounter ion effects.............................................................................................. 50

2-9. Co-oxidation studies ........................................................................................... 52

2-10. Catalysis by products in the absence of 1............................................................54

2-11. Bond dissociation energies. ................................................................................. 55

2-12. Cumene oxidation and peroxide decomposition data of reactions
catalyzed by FeL3X2 complexes. ......................................................................... 58

3-1. Cyclohexane oxidation catalyzed by Fedmp. ......................................................66

3-2. Products from cyclohexane oxidation catalyzed by 1 and Fedmp......................68

3-3. Cumene oxidation catalyzed by Fedmp..............................................................69

3-4. Cumene oxidation catalyzed by Fephen2X2 complexes......................................76

4-1. Cumene oxidation catalyzed by Fe and Co acetyl acetonates ............................80

4-2. Cumene oxidation catalyzed by Fe(II)Pc and Fe(II)cyclam...............................81

4-3. Cumene autoxidation catalyzed by Co and Fe complexes..................................83








4-4. Cumene oxidation catalyzed by A-oxo Fe complexes ........................................89

4-5. Catalytic activity of di-nuclear Fe complexes in cyclohexane oxidation..............90

5-1. Styrene oxide yields using different solvent systems .........................................98

5-2. Experimental expoxide conversion.....................................................................98

5-3. Determination of enantiomeric selectivity........................................................ 101

5-4. Epoxidation data ............................................................................................... 103















LIST OF FIGURES


Figure page

1-1. Radical steps for uncatalyzed hydrocarbon autoxidation .....................................3...

1-2. H aber-W eiss cycle. ...............................................................................................4...

1-3. Proposed catalytic cycle for the activation of molecular oxygen and oxygen
transfer to alkanes in cytochrome P-450............................................................... 7

1-4. Mid-Century process for terephthalic acid synthesis............................................8...

1-5. Electron transfer mechanism in Co(OAc)2 catalyzed mechanism
of p-xylene oxidation. ............................................................................................9...

1-6. Examples of metal-molecular oxygen binding modes........................................ 11

1-7. Selective ethylene epoxidation over a supported silver catalyst......................... 12

1-8. Proposed catalytic cycle for oxygen transfer to P(Ph)3 ......................................13

1-9. Selective oxygen transfer to pendant ligand ....................................................... 14

1-10. Epoxidation of propylene by high valent, early transition metal complexes.........15

1-11. Active oxidant in MTO-catalyzed epoxidation reactions...................................16

1-12. Proposed mechanism of epoxidation co-catalyzed by imidazole .......................17

1-13. Equilibrium formation of peroxycarbonate. ........................................................19

1-14. Mn2+/BAP catalyzed oxidation of alkenes.......................................................... 20

2-1. [Fe(4,7-diphenyl-1,10-phenanthroline)3]2+(CF3SO3)2 (1).................................. 23

2-2. tetra-(pentafluorophenyl)-porphyrin iron(III) chloride (2) .................................25

2-3. Products of metal-catalyzed cumene autoxidation. .............................................26

2-4. Formation of CO and CO2 over time in cumene oxidation catalyzed by 1. ..........27

2-5. Cum ene oxidation catalyzed by 1....................................................................... 28








2-6. Oxygen uptake curves for 1 and 2 catalyzed oxidation......................................29

2-7. M etal-catalyzed peroxide decomposition........................................................... 31

2-8. Plots of In [ROOH ] vs. tim e ............................................................................... 31

2-9. Mechanism of metal catalyzed cumene autoxidation.........................................34

2-10 Disproportionation pathways of alkyl tetroxide. .................................................35

2-11. Simulation of cumene oxidation catalyzed by 1.................................................38

2-12. Simulation of cumyl peroxide decomposition catalyzed by 1............................39

2-13. Simulation of cumene oxidation catalyzed by 2.................................................39

2-14. Simulation of cumyl peroxide decomposition catalyzed by 2............................40

2-15. Simulation of ROOH initiated cumene oxidation catalyzed by 1.......................40

2-16. Simulation of ROOH initiated cumene oxidation catalyzed by 2.......................41

2-17. GC trace of sample after 5h of uncatalyzed reaction.......................................... 41

2-18. Oxygen uptake curves in cumene oxidation catalyzed by 1
in different solvents............................................................................................. 43

2-19. Dependence of rate of oxygen uptake on oxygen pressure.................................44

2-20. Temperature effects on product selectivity for cumene oxidation
catalyzed by 1. ..................................................................................................... 45

2-21. Experimental and calculated dependence of catalyst concentration on rate of
oxygen uptake. .................................................................................................... 46

2-22. Formation of FeL2X2 from 1 in the presence of LiCl.........................................49

2-23. Spectral changes of I at 60 C in the presence of ROOH...................................51

2-24. Product formation and oxygen uptake in cumene oxidation catalyzed
by 1 in the presence of a-methyl styrene............................................................ 53

2-25. Proposed mechanism of a-methyl styrene formation .........................................54

2-26. Comparison of oxygen uptake curve during cumene oxidation catalyzed
by 1 and a curve predicted by simulation using proposed mechanism............... 55

2-27. Numbering scheme for the phen ligand.............................................................. 56








2-28. Oxygen uptake curves during cumene autoxidation catalyzed by FeL3X2
com p lex es. .............................................................................................................58

3-1. Proposed mechanism for the activation of 02 by FeL2X2...................................62

3-2. Exam ples of radical traps.................................................................................... 62

3-3. [cis-Fe(2,9-dimethylphenanthroline)2(H20)2](SO3CF3)2 (Fedmp). ....................63

3-4. Products of the air oxidation of cyclohexane...................................................... 64

3-5. GC trace after 4h of cyclohexane oxidation catalyzed by Fedmp......................65

3-6. Oxygen uptake curves for Fedmp catalyzed cyclohexane oxidation..................66

3-7. GC trace of cyclohexane oxidation in 50/50 cyclohexane/DCB ........................67

3-8. Oxygen uptake curves for cyclohexane oxidation calyzed by 1 and Fedmp.........68

3-9. Oxygen uptake curve during cumene oxidation catalyzed by Fedmp at 60 C.....70

3-10. GC trace of sample taken after 2 h of cumene oxidation catalyzed by Fedmp. ....70

3-11. Top; 2,9-dimethylphenanthroline (dmp), Bottom; Fedmp in de-acetone ............71

3-12. Variable temperature H1-NMR spectrum of Fedmp in de-acetone.....................73

3-13. 'H-NM R spectrum of Fephen2CN2..................................................................... 75

3-14. Oxygen uptake curves during cumene oxidation catalyzed by Fephen2X2
com plexes .............................................................................................................77

4-1. Structures of m acrocyclic ligands....................................................................... 80

4-2. Structures of tpa and tpen ligands....................................................................... 83

4-3. Comparison of oxygen uptake curves during cumene oxidation........................84

4-4. Proposed mechanism of a,a-dimethyl benzyl methyl ether formation...............84

4-5. Cumyl peroxide (ROOH) decomposition catalyzed by Fetpa............................85

4-6. 'H-NM R spectra in acetonitrile-d3...................................................................... 86

4-7. Oxygen uptake curve in cumene oxidation catalyzed by Fe(SO3CF3)2.................87

4-8. Structure of com plex 3 and 4.............................................................................. 88

4-9. Structures of di-nuclear iron complexes. ............................................................. 89








5-1. Equilibrium formation of peroxycarbonate. ........................................................92

5-2. Mn(III) meso-tetrakis(4N-methylpyridinium)porphyrin (MnTMPyP)...............93

5-3. MnTMPyP/BAP catalyzed room temperature epoxidation
of sulfonated styrene. .......................................................................................... 93

5-4. 1H-NMR spectrum after 60 min of reaction. .......................................................94

5-5. HPLC trace of reaction solution after 6 min.......................................................95

5-6. ln[SS] vs. tim e plots............................................................................................ 95

5-7. Fast decomposition of MnTMPyP in the presence of BAP................................ 96

5-8. Proposed mechanism of Mn(III)-salen catalyzed epoxidation ...........................97

5-9. Jacobson's catalyst.............................................................................................. 98

5-10. Pseudocontact shift equation...............................................................................99

5-11. 'H-NMR spectrum of a standard styrene sample (top) and styrene
in the presence of a shift reagent (bottom).........................................................100

A-1. Variation in k, (RH R-) .................................................................................114

A-2. Variation in k2 (ROO- + RH -- ROOH + R-) .................................................... 115

A-3. Variation in k3 (2 ROO- 2 RO- + 02)............................................................115

A-4. Variation k4 (2 ROO- ROOR + 02) ...............................................................116

A-5. Variation in ks (RO- + RH -, ROH + R-).......................................................... 116

A-6. Variation in k6 (RO- + ROOH -- ROH + ROO-) .............................................. 117

A-7. Variation in k7 (RO- -- R'O + Me-) .................................................................. 117

A-8. Variation in kl2 (ROOH + M"- ROO- + H + M")........................................118

A-9. Variation in k13 (ROOH + M" RO-+ HO-+ M1). ....................................... 118














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

HYDROCARBON OXIDATION USING MOLECULAR OXYGEN AND
HYDROGEN PEROXIDE CATALYZED BY TRANSITION METAL COMPLEXES

By

Ana Ison

August 2004

Chair: David E. Richardson
Major Department: Chemistry

The catalytic activity of coordination complex [Fe(4,7-diphenyl-1,10-

phenanthroline)3](SO3CF3)2 (1) has been investigated in the autoxidation of cumene in

the presence of molecular oxygen. Reaction products include 2-phenyl-2-propanol,

cumyl hydroperoxide, acetophenone, and traces of dicumyl peroxide. Parallel

experiments were done comparing the catalytic activity of 1 to that of a fluorinated iron

porphyrin (tetra-pentafluorophenylporphyrin iron(lI1) chloride (2), and the catalytic

activities of 1 and 2 are comparable. Our detailed kinetic studies and UV-visible

spectrophotometry support a radical autoxidation mechanism, in which the catalytic

decomposition of alkyl hydroperoxides generates free-radical autoxidation chain carriers

as described previously for halogenated porphyrin catalysts. The kinetics of the

disproportionation of cumyl hydroperoxide were determined, and the catalytic rate

constants for 1 and 2 were incorporated into a complete kinetic model that fits both

autoxidation and peroxide-decomposition data. The inactivity of the Ru analog of 1 and









other experiments suggest that ligand dissociation is crucial in the activation of 1.

Related FeL3X2, FeL2X2 (L=substituted 1,10-phenanthroline; X=anionic counter-ion),

and FeLnXn (L=tetra,penta,hexa-dentate nitrogen donor ligands) were also investigated

and their catalytic reactivity examined in the context of the metal catalyzed peroxide

decomposition mechanism.

The advantages of using bicarbonate activated peroxide (BAP) versus H202 alone

as the oxidant in the presence of known epoxidation catalysts were investigated. The

room-temperature epoxidation of sulfonated styrene in water was catalyzed by

Mn(III)porphyrin/BAP, and resulted in formation of the epoxide with high conversion

and selectivity. The rate of epoxidation catalyzed by Mn(I)porphyrin/BAP was found

to be comparable to the free Mn2+/BAP system, suggesting catalyst decomposition. The

MnTMPyP degraded quickly under the reaction conditions, releasing free Mn2+ into the

solution. Presumably the resulting epoxide was formed though the well-known free

manganese-catalyzed pathway.

Asymmetric epoxidation of styrene catalyzed by Jacobson's catalyst using BAP

as an oxidant was also studied. Moderate enantioselectivity was achieved in a biphasic

solvent system. Other anionic salts such as AcO- resulted in similar product

enantioselectivity, suggesting that (unlike in the free Mn2+/BAP system),

peroxymonocarbonate (peroxy acid formed in the BAP system) was not essential in

forming the active catalytic species.














CHAPTER 1
INTRODUCTION

Molecular Oxygen as Terminal Oxidant

The use of molecular oxygen as a terminal oxidant in metal catalyzed oxidation of

hydrocarbons has been an area of vigorous and intense research mainly because of the

environmental and economic advantages of using dioxygen in place of peroxides,

peracids or iodoarenes. The ultimate goal is to develop a catalytic system that mimics the

enzyme-mediated activation of molecular oxygen and results in the low temperature,

highly selective oxygen transfer to a substrate. The various mechanisms by which metals

catalyze autoxidation of hydrocarbons can be separated into three broad categories: 1)

metal catalyzed decomposition of intermediate peroxide, 2) direct attack on the substrate

by the metal complex, 3) direct activation of molecular oxygen by the metal complex.1,2

The mechanism is dependent on the solvent system, ligands surrounding the metal and

the substrate. Examples of each mechanism type are described next.

Free-Radical Autoxidation

The importance of molecular oxygen in oxidation was realized as far back as the

18th century with Lavoisier's explanation of combustion.1 Later observations suggested

that the deterioration of many organic materials (such as rubber and natural oils) results

from oxygen absorption. Further studies lead to the recognition that organic peroxides

were the primary products of these oxidative processes, and the free-radical chain theory

of autoxidation that was established in the 1940s 3,4 The liquid-phase autoxidation of

hydrocarbons has been studied extensively since that time, and the well-established








radical mechanism is shown in Figure 1-1. Autoxidation reactions are recognized for

having long induction periods, because of the thermodynamically unfavorable C-H bond

cleavage in the first step. The direct reaction between RH and 02 is spin forbidden.

Molecular oxygen has two unpaired electrons in the ground state and is paramagnetic.

The highest occupied molecular orbitals are two degenerate r* orbitals; therefore,

electron pairing is energetically unfavorable. Molecular oxygen is in a triplet state,

whereas the substrate is in a singlet state. The spin-conservation rule forbids the

interaction of the two species in the ground state. Despite the difference in multiplicity,

hydrocarbons are oxidized in the presence of molecular oxygen mainly due to the

presence of peroxidic impurities.

The initial radical-forming step ki is thought to occur through hydrogen abstraction

from RH by trace peroxidic impurities. The addition of molecular oxygen to the R-

radical is diffusion controlled (>109 M-11s), even below atmospheric pressure. The

radical-propagation step k, involves hydrogen abstraction by the alkyl peroxy radical

(ROO.) to form alkyl hydroperoxide and chain carrying ROO. radical. Termination of

peroxy radicals occurs through a bimolecular reaction, to form secondary oxidation

products. Step kt is an oversimplification of a complex series of steps leading to both

radical and nonradical products. The precise steps are substrate dependent, and can

include the formation of alcohols, ketones, epoxides, carboxylic acids, aldehydes, CO,

CO2 and other products. Once the concentration of alkyl hydroperoxide is high enough,

kp becomes rate limiting. This marks the end of the induction period and the start of

rapid substrate oxidation. Since the primary products such as peroxide, alcohol and








Initiation
RH i W R 2 ROO'

Propagation
ROO + RH kp ROOH + R'
R- + 02 fas ROO'

Termination
2 ROO" kt nonradical products

Figure 1-1. Radical steps for uncatalyzed hydrocarbon autoxidation

ketone are more oxidizable than the substrate, secondary reactions of those products are

common, resulting in poor selectivity. The utility of synthesis of useful products from

saturated hydrocarbons under autoxidizing conditions is limited because of the high

reaction temperatures needed to overcome the long induction period, and poor selectivity.

The addition of metal salts in catalytic amounts was observed to shorten the induction

period and allow for lower reaction temperatures and increased selectivity.1'5 This

discovery prompted interest in developing autoxidation catalysts capable of autoxidizing

hydrocarbons selectively.

Metal-Catalyzed Peroxide Decomposition

Many of the early investigations used low concentrations of Mn, Fe, Co, and Cu

acetates, naphthanates, and phthalocyanines to allow for a shorter induction period, lower

reaction temperatures, and increased selectivity. 1,6-8 Although the addition of such

catalysts does lead to some desired results, the selectivity of the reactions remains poor

unless the conversion is kept low. Nonetheless, metal-catalyzed autoxidation in neat

hydrocarbon is used in major industrial processes, for the synthesis of adipic acid from

cyclohexane. The autoxidation of cyclohexane to form ketone and alcohol (K/A oil) is

done at 150-160C and catalyzed by Co-naphthenate.9 The cobalt catalyst used is soluble









in neat cyclohexane, and does not require the use of co-solvent. High selectivity (ca.

70%) toward K/A oil is maintained by removing the products at a low conversion (ca.

10%). Further direct oxidation of the K/A oil with HNO3 forms adipic acid, a monomer

in the synthesis of nylon 6,6.1,2

The widely accepted mechanism of metal-catalyzed autoxidation carried out neat or

in non-polar solvent, occurs through the metal catalyzed decomposition of alkyl

peroxides. The presence of a metal catalyst leads to an increase in the rate of formation

of chain-carrying radicals.'5'8',10 Therefore, the radical-chain mechanism describing

uncatalyzed autoxidation (Figure 1-1) is unchanged other than the addition of

metal-catalyzed peroxide decomposition steps referred to as the Haber-Weiss mechanism

(Figure 1-2).1

Mill

RO' + OH- ROOH



ROOH : ROO' + H+
M11



ROOH + Mll ROO' + H+ + Mi

ROOH + M" -- RO' + OH- + M"il

Figure 1-2. Haber-Weiss cycle

Many early investigations of metal-catalyzed autoxidation attempted to develop a

more detailed description of the metal-peroxide interaction.1'2,'1 Numerous active

species, such as peroxy metal complexes, substrate adducts and dimers have been









proposed to explain the complex kinetic dependence on metal concentration; however,

none have been unambiguously characterized.11

Another important feature of metal-catalyzed autoxidation reactions is the presence

of a maximum rate of oxidation.1,12 The maximum theoretical rate of oxidation is

dependent on the substrate and is determined by the propagation and termination rate

constants. Equation 2-1 is used to calculate the maximum rate,

kp2 [RH]2
-dO2/dt = 2 2 (2-1)
2kI

where kp and kt are the propagation and termination rate constants, respectively.12

Equation 2-1 holds true in the presence of metal catalysts, as long as they do not catalyze

the propagation and termination steps. Because the major function of the catalyst is to

decompose the intermediate peroxide while the uncatalyzed steps remain unchanged, the

rate of oxidation cannot exceed that inherently governed by the substrate. Hence the rate

of metal-catalyzed autoxidation can be increased only up to a limiting value when the

steps in Figures 1-1 and 1-2 apply.

For a number of years, the most active autoxidation catalyst used in hydrocarbon

media was Co-naphthenate in the industrial oxidation of cyclohexane. More recently

Fe(II) halogenated porphyrin complexes were shown to catalyze the air oxidation of

saturated C-H bonds at room temperature.13-15 Initial interest in using iron porphyrins

arose from the desire to mimic highly selective room-temperature oxidation using

molecular oxygen as exhibited by mono and dioxygenases. The porphyrins especially

have generated much interest because of their function in hemoproteins, and because they









are known to participate in oxygen transport (hemoglobin, myoglobin), electron transport

cytochromee c), and redox chemistry (cytochromes P-450 and peroxidases).1

Cytochrome P-450 in particular catalyzes the activation and insertion of

molecular oxygen into many substrates. The highly selective hydroxylation carried out

by P-450 is proposed to be carried out by the high oxidation state oxoiron(IV)

heme(protoporphyrin IX)'+. The general scheme used to describe the mechanism of P-

450 hydroxylation is shown in Figure 1-3.' The electrons and protons needed by the

enzyme to complete the catalytic cycle are provided by cofactors such as NADP and

ascorbate. Because of the similarity of the porphyrin ligand to that found in P-450, the

initially proposed mechanism of Fe(llI)porphyrin catalyzed air oxidation of alkanes was

that of activation of molecular oxygen to form the high oxidation oxometal species.13

Further investigation by Labinger et al.16'17 showed that a radical peroxide decomposition

mechanism was more suitable to describe the reactivity of the Fe(I[I) halogenated

porphyrins. Although the Fe(flI) halogenated porphyrins have similar characteristics as

enzymatic systems that clearly does not necessarily lead to the same kind of reactivity.

The ligand environment in the enzyme is highly controlled, and the substrate is isolated at

the active site.18 Although radicals are often formed in enzymatic systems, radical-chain

reactions are less favored, compared to reactions catalyzed by simple metal complexes.

Furthermore the cycle described in Figurel-3 is difficult to imagine in the absence of

coupled proton and electron donors.

An interesting example of a catalytic system employing reducing agents as found

in enzymatic oxidations is the Gif family of catalysts. Barton et al.19'20 developed a

system for oxidation and oxidative functionalization of alkanes under mild conditions.









ROH







R

RH
CFe IV

OOH
H20


le-












02+ le- +H


Figure 1-3. Proposed catalytic cycle for the activation of molecular oxygen and oxygen
transfer to alkanes in cytochrome P-450

The reactions use pyridine as solvent in the presence of an organic acid (acetic acid), and

are catalyzed by mainly Fe complexes in the presence of a reductant (Zn, Fe, Cuo).

Simple sources of iron were used such as FeC12, FeC13, or trinuclear oxo-centered

complexes such as [Fe30(O2CCH3)6(py)3]3+.21 The primary reaction products are

ketones; and C-H bond selectivity follows the unusual order of 20 > 3 > 1, based on

adamantane oxidation. There is disagreement whether the mechanism is a free-radical

chain reaction or nonradical metal-centered catalysis.22'23 Product profiles and KIE

values obtained from the catalytic oxidation of a number of substrates are not in full

agreement with a purely radical mechanism or the involvement of a high oxidation

oxometal species.

Oxidation of olefins leads mainly to ketone products and not epoxides as would be

expected in case of a selective metal-centered mechanism.24 The details of the Gif

mechanism are largely a topic of debate, and are not completely understood. The Gif









reactions are an example of the dramatic effects solvent and additives can have on a

mechanism.

Electron-Transfer Mechanism

The Co catalyzed oxidation of p-xylene is an industrially important reaction used in

the production of terephthalic acid, a starting material in polymer synthesis.5 The Mid-

Century process where the reaction is carried out in acetic acid in the presence of

relatively high concentrations of Co(OAc) (~0.1M) is shown in Figure 1-4.

CH3 CO2H


+ 3 02 Co(OAc)2Br- +2 H20
HOAc

CH3 CO2H
p-xylene terephthalic acid

Figure 1-4. Mid-Century process for terephthalic acid synthesis

Strong oxidants such as Co(OAc)3 are able to abstract an electron from the substrate

resulting in the formation of a resonance-stabilized radical cation, followed by loss of a

proton to form a benzyl radical. Under autoxidizing conditions, the benzyl radical is

trapped by oxygen to form a benzylperoxy radical followed by reaction with another

equivalent of the catalyst to form an aldehyde. Further oxidation to the carboxylic acid

occurs easily under the reaction conditions (Figure 1-5).1,10 High concentration of the

catalyst is needed for Co(H) to trap the benzylperoxy radical and circumvent the usual

radical chain reaction of alkylperoxy radicals shown below.

ArCH202- + ArCH3 ArCH200H + ArCH2'









CH3 CH3 CH200

+ CO H+
SCo + 02 C

CH3 CH3 CH3


CH200OO O H OC C. OH

+ co" 02
S-Co" I

CH3 CH3 CO2H

Figure 1-5. Electron transfer mechanism in Co(OAc)2 catalyzed mechanism of p-xylene
oxidation

The rate of electron-transfer reactions is governed by the ionization potential of the

substrate; therefore, the complete oxidation to therephthalic acid is retarded because of

the significantly higher ionization potential of the intermediate p-toluic acid. Additives

such as halide salts are often used to promote the complete oxidation to terephthalic acid

as is the case in the Mid-century process (Figure 1-4).' Reaction of the Co(OAc)2 with

Br results in the formation of bromine atoms that become the chain transfer agents in

place of the Co(Ei) as shown below.

Br' + ArCH3 IN HBr + ArCH2"

The abundance and low cost of butane make it a major feedstock for acetic acid

production. The oxidation of n-butane at 100-1250C in the presence of large amounts of

Co(OAc)2 and promoter in acetic acid results in 80% conversion of n-butane with 83%

selectivity to acetic acid.5 The active catalyst is proposed to be a Co(ll) species

responsible for electron transfer from the substrate. Details of the mechanism are similar

to those of p-xylene oxidation described above, although the mechanism is more









complicated because C-C bond cleavage occurs along with C-H bond cleavage. High

concentration of the Co(II) species is maintained by the presence of promoters such as

ozone, methyl ethyl ketone, and 2-butanone. Ozone is a strong oxidant and the two

ketones easily form radicals that are responsible for oxidizing the Co(II).

Selective single-step oxidation of cyclohexane to form adipic acid has been a

major goal in industry because the acid is an important intermediate in the production of

nylon.5 The selective oxidation of cyclohexane can be achieved under conditions similar

to those described above. Oxidation using high concentrations of Co(II) acetate in acetic

acid at 80-100C leads to 80% conversion of cyclohexane and the formation of adipic

acid with 75% selectivity.25 Such high selectivities are unusual in a free-radical

mechanism; therefore, an electron transfer mechanism was proposed. The reactions

exhibited an induction period that ended with the conversion of Co(II) to Co(III) in

solution. Therefore, direct attack of the metal complex on the substrate is suggested.

Adding radical-forming initiators such as acetaldehyde, cyclohexanone, and AIBN can

shorten the induction period. Another observation supporting an electron-transfer

mechanism is decrease in selectivity upon a decrease in the catalyst concentration.

Presumably the free-radical mechanism becomes predominant, as the catalyst at low

concentrations is unable to trap the alkylperoxy radicals before they enter radical chain

propagation. However, despite the reasonably high selectivity achieved by this system,

the industrial method for adipic acid synthesis is a two-step process the first of which is

based on free-radical chemistry as described in the previous section. Weaker oxidants

such as Mn(II), Pb(IV), Ce(IV), Cu(II), Pd(II) acetates can also be used as catalysts for









the oxidation of aromatic hydrocarbons in the presence of strong acids such as triflic acid

and sulfuric acid.

Metal-Mediated Molecular Oxygen Activation

There are numerous examples of both reversible and irreversible complexation of

molecular oxygen by metal complexes. Metals with two available oxidation states

mainly form short-lived superoxo adducts, and combine to form pt-peroxo complexes

(Figure 1-6). Co(n) salen complexes were shown to form it-peroxo dimers; however,

there are some examples of stable suproxo complexes that did not dimerize.' In the case

where the metal can undergo a 2e oxidation, the preferred binding mode is side on,

forming a peroxo species such as Vaska's complex.26


M' + 02 MI-0-0 M

superoxo peroxo



2
Ph3P..." O Con + 02 Co O-
Ir"
Cl "PPh3 Co O-_O fast Co lllO.' 0oCo"
CO
Vaska's complex m-peroxo

Figure 1-6. Examples of metal-molecular oxygen binding modes

The best known example of molecular oxygen activation by a metal complex

resulting in selective oxygen transfer is the industrially important gas-phase oxidation of

ethylene over a supported silver catalyst, where a silver peroxo complex is implicated in

oxygen transfer.1,2'5 However, the catalyst does not catalyze the selective epoxidation of

higher alkenes because of participation of radical pathways.










H H Ag/A1203 H H
H H 250 C H H

Figure 1-7. Selective ethylene epoxidation over a supported silver catalyst.

The tendency of hydrocarbons to contain trace peroxides and the efficiency of metal

complexes to decompose them offers the greatest obstacle to selective non-radical

oxidation in the presence of molecular oxygen.

A number of heme protein mimics such as Co, Fe, and Mn porphyrin dioxygen

adducts have been isolated and characterized, however none have shown selective

oxygen transfer to hydrocarbons.27'28 The ability of proteins to transfer molecular oxygen

selectively to a substrate stems from their ability to form high-oxidation metal-oxo

complexes, and also from the highly controlled environment surrounding the substrate

that prevents the propagation of radical chains. Such control is difficult to achieve in

simple model systems as is revealed by the lack of synthetic enzyme mimics able to carry

out selective oxidation of hydrocarbons using molecular oxygen as a terminal oxidant.1,2'5

An example of a synthetic Fe porphyrin complex shown to activate molecular oxygen

resulted in oxygen transfer to P(Ph)3. Oxygen transfer from a high oxidation

(porphyrin)FerO complex has been suggested based on experiments where an

independently synthesized (porphyrin)FeVO complex oxidized P(Ph)3 in a stoichiometric

reaction. The proposed catalytic cycle is shown in Figure 1-8.1 Attempts to achieve

similar results using alkanes or alkenes resulted in radical products.

Recent investigations of metal mediated activation of dioxygen lead to interesting

results worth mentioning. Itoh and coworkers29 investigated oxygen transfer by








2 02

2 (Ph)3PO

2 (Ph)3P


0 0.0

2 CFV)




= tetraphenylporphyrin

Figure 1-8. Proposed catalytic cycle for oxygen transfer to P(Ph)3

bis(p-oxo)dicopper(II) complex [Cu2I(L)2(A-oxo)]2+ where

L=N-ethyl-N-[2-(2-pyridyl)ethyl]-2-phenylethylamine. Benzylic hydroxylation of the

ligand side arm occurred with 46% conversion after 20 h at 250C, in acetone under a pure

dioxygen atmosphere. Reactions carried out using 18O2 confirmed that the oxygen found

in the product originated from molecular oxygen. One of the proposed pathways shown

in Figure 1-9 suggests that oxygen transfer could occur through a concerted pathway with

no radical formation. The second possibility is the formation of a short-lived

carbon-centered radical that rebounds at a fast rate recapturing the hydroxy group.

Similar mechanistic pathways have been proposed for oxygenases responsible for

selective hydroxylation in biological systems. Several other examples of ligand

hydroxylation by copper complexes and oxygen have been reported; however, similar

oxygen transfer to uncoordinated alkane or alkene substrates has not been observed.30'31









N,,, / CN
N Cu"O Cu"
N N


SH Ph concerted
H abstraction



N0, /o ,,N N,,,. '0- .,,N
Ill- ". ll Ill
NNC N N.U
---HI

H Ph H Ph




rebound

N,, CuI O Cu.N

N N

H Ph

Figure 1-9. Selective oxygen transfer to pendant ligand

Hydrogen Peroxide as Terminal Oxidant

The search for efficient, selective, and clean oxidation processes for the

epoxidation of alkenes is a high priority for industrial applications and for bridging the

gap between synthetic catalysts and enzymatic systems. In particular, the use of H202

has been explored, because it is an economical oxidant giving only water as a byproduct.

Alkene Epoxidation by Early Transition Metal Catalysts

The significance of catalyzed epoxidation in industry cannot be overstated. One

million tons of propylene is converted to propylene oxide annually catalyzed by

compounds of high-valent early transition metals such as MovI, Wvt, Vv, or Ti"v in the

presence of alkylperoxide. The active catalyst is a metal-peroxo adduct resulting in a








concerted transfer of oxygen to the substrate.32 The metal does not undergo a change in

oxidation state, but behaves as a Lewis acid and withdraws electron density from the 0-0

bond. The electrophilic character of the coordinated peroxide 0-0 bond is increased,

ensuring the concerted transfer of oxygen to the substrate (Figure 1-10).


OM OR +
OR



Figure 1-10. Epoxidation of propylene by high-valent, early transition metal complexes

The early transition metal complexes used are suitable for this chemistry because they are

strong Lewis acids and weak oxidants. One-electron oxidants such as Fe or Co would

lead to radical products and loss of selectivity. Although H202 undergoes the same type

of reaction as alkylperoxides, the usually aqueous solutions of hydrogen peroxide lead to

catalyst inhibition by H20.32

The strong drive to find an effective catalyst using H202 as an oxidant has led to

the discovery of a number of active and selective epoxidation systems.

Titanium(IV)-silicalite (TS-1) supported catalyst uses 30% H202, and is very effective in

epoxidizing linear olefins.33 The hydrophobic nature of the silicalite support prevents

inhibition of the catalyst by water, and absorbs only the hydrophobic substrate. The

major limitation of TS-1; however, is the small pore size of the support, allowing only

straight-chain alkenes to access the micropores of the silicalite. Attempts to increase pore

size and maintain high activity were met with limited success.

First reports by Herrmann et al.34 of methyltrioxorhenium (MTO) catalyzed

epoxidation called for the use of anhydrous H202 in t-butanol. Cyclohexene oxide was









obtained in 90% yield at 10C over a 5 h reaction time. The reaction was further

improved by including heterocyclic bases pyridinee, pyrazole) in a CH2C12 solvent.35'36

High selectivities were obtained with a variety ofolefins using 35% H202 and 0.05%

MTO. The proposed mechanism involves a diperoxorhenium(VII) complex shown in

Figure 1-11. The drawbacks of MTO catalyzed epoxidation are the low stability of MTO

in the presence of H202, the difficult and expensive synthesis of the rhenium complex,

and ring opening of acid-sensitive epoxides under the acidic conditions.


CH3


0


H202
-H20


CH3


0


CH3


H202
-H20


MTO diperoxorhenium(VII)

Figure 1-11. Active oxidant in MTO-catalyzed epoxidation reactions

Metal Porphyrin Catalysts

Similar to cytochrome P-450, simple Fe and Mn porphyrin complexes catalyze the

selective oxygen transfer to alkenes using iodosylbenzene (PhlO), m-chloroperbenzoic

acid (MCPBA), or hypochlorite as oxidants.37-39 However, similar selectivities were not

observed when using oxygen donors containing 0-0 bonds, such as alkyl peroxides and

H202. The mechanistic reason for low selectivity is the propensity for homolytic bond

cleavage of the 0-0 bond resulting in alkoxy radicals as active species, instead of the

high oxidation state (porphyrin)+Felv=0 or (porphyrin)MnV=O.40 Another significant

problem is the presence of competing pathways leading to dismutation of H202, and

catalyst destruction. Further investigation showed that the efficiency and selectivity of

oxygen transfer can be greatly improved by using imidazole as cocatalyst.









The dramatic effect of added imidazole was that epoxide yield increased from 2%

for [Mn(5,10,15,20-tetrakis(2',6'-dichlorophenyl)porphyrin]CI (Mn(TDCPP)CI) in the

absence of imidazole, to as much as 72% of cyclooctene when the cocatalyst was

added.41 The proposed active species in the Mn(TDCPP)CI-H202-imidazole system is

the high oxidation state (porphyrin)MnV=O based on comparisons using PhlO as oxygen

donor. The cis/trans ratios of three different substrates were the same using either H202

or PhlO as oxygen transfer agent, suggesting that the same active species was formed in

both cases. However, product yields were somewhat lower in the

Mn(TDCPP)Cl-H202-imidazole system, presumably because of the parallel peroxide

dismutation pathway preventing the efficient transfer of all the oxygen equivalents to the

substrate. The proposed function of imidazole is both as an axial ligand and base

catalyst. The proposed catalytic cycle is shown in Figure 1-12.41









NH





N
SH-

H20 N


Figure 1-12. Proposed mechanism of epoxidation co-catalyzed by imidazole








The axial coordination of imidazole may aid in the heterolytic bond cleavage of H202 and

prevent the formation of free radicals. Similarly, the base catalyst function of the

cocatalyst favors the formation of the high-oxidation state Mnv=O species by the removal

of H20.

The Mn-Salen Assymetric Catalysts

Since their discovery, chiral Mn(Il)-salen (N,N'-bis(salicylideneamino)ethane)

complexes have proven to be highly effective in the asymmetric epoxidation of virtually

every class of unfunctionalized conjugated olefins.42 The popularity of Mn(III)-salen

complexes stems from facile catalyst preparation, high enantiomeric excess (EE) of the

products, and the use of cheap oxidants. The preferred oxidants are hypochlorite and

iodosylbenzene.43'44 Catalytic conditions can be optimized by including nitrogen

heterocycles such as imidazoles, pyridines, tertiary amine N-oxides, and carboxylate

anions. The additive is thought to function as an axial ligand stabilizing the high

oxidation state Mn(V)=O species.

An additional function of the more lipophyllic N-oxides may be as a phase transfer

agent for HCO- transport to the organic phase in biphasic solvent systems. Additives

such as heterocyclic amines and carboxylates have also improved the effectiveness of

H202 as an oxygen transfer agent. High yields (86%) and EE (92%) were achieved in the

epoxidation of cyanochromene using Jacobsen's catalyst with ammonium acetate and

30% H202.45 Such high EE has not been observed with a wide variety of substrates,

hence further studies must be completed to further improve the Mn-salen/H202 system.

The Mn /HCO3"/H202 Catalyzed Alkene Epoxidation

Alkene epoxidation catalyzed by simple transition metal salts using aqueous H202

and bicarbonate salt was investigated by Richardson et al.46 as well as by Burgess and








coworkers.47 The room temperature epoxidation of sulfonated styrene in aqueous

solution was catalyzed by AM levels of Mn2+ resulting in rate enhancement of at least an

order of magnitude. No epoxide was obtained in the absence of bicarbonate ion, and

trace amounts of epoxide were found when H202 and HCO3- were used after removal of

trace metal ions by chelation. Kinetic studies reveal a mixed-order dependence on

bicarbonate concentration, suggesting that more than one equivalent of bicarbonate is

present in the transition state. Previous studies by Drago et al. and Richardson et al.46'48

have shown equilibrium formation of the peroxycarbonate ion in solutions containing

H202 and HCO3- (Figure 1-13). Other anionic salts such as NaOAc do not catalyze the

epoxidation, nor do they form a peroxy acid species in equilibrium with H202, suggesting

that the active epoxidizing agent can only be formed in the presence of the

peroxycarbonate ion.

O O
H 1o0 H + 0 H 0
0 H O OH -O OrOH + H20


bicarbonate peroxycarbonate

Figure 1-13. Equilibrium formation of peroxycarbonate

Alkene epoxidation carried out in mixed solvents illustrates the scope of the

Mn2+/BAP system (Figure 1-14). More than 30 d-block and f-block metal salts were

screened and only Cr2+ and Fe3+ showed moderate activity, with Mn2+ being the most

active.47 Seventeen alkenes were screened, and results showed that aryl-substituted,

cyclic and trialkyl-substituted alkenes were epoxidized in high yields (63-94%), while

terminal alkenes were unreactive. The large-scale reactions were carried out in DMF or








t-BuOH solvent at room temperature with slow addition of an aqueous solution of

H202/HCO3" over a 16 h period.

R R o
R MnSO4, 5-10eq H202, DMF or t-BuOH RxR
R 0.2M NaHCO3, pH 8.0
R R

Figure 1-14. Mn2+/BAP catalyzed oxidation of alkenes

The observation that most alkenes converted cleanly to the epoxide, and that no

differences in the product profiles were seen if the reaction was carried out in the absence

of air, suggests that HO-, and HOO- are likely not the primary epoxidation species.

Furthermore, EPR studies show signals characteristic of Mn4+ species formed in the

presence of H202/ HCO3-.47 The simplicity, high activity and selectivity of the

Mn2+/BAP system make it a unique and convenient method for the epoxidation of a

variety of alkenes.

Scope of the Dissertation

Chapter 2 deals with the Fe"l diimine catalyzed autoxidation of cumene at moderate

temperatures. This particularly active family of Fe catalysts is an important addition to

the existing literature because most of the highly active autoxidation catalysts, except for

(porphyrin)Fel", are based on Co2+ or Co3+ complexes. The function of the catalyst as a

peroxide decomposition agent is explored and rationalized based on product selectivities,

peroxide decomposition studies as well as catalyst concentration, oxygen pressure and

solvent polarity dependenciese. We present a detailed mechanistic scheme consistent

with the peroxide decomposition mechanism that enables us to model product profiles

over time for the catalyzed autoxidation of cumene at 60 oC. The scheme provides good

fits to our experimental data even when the system is perturbed by adding small mounts









of products at the beginning of the reaction. The identity of the active catalyst was

explored by ligand variation and counter-ion studies.

The interest in further understanding the effect of ligand dissociation on the

formation of the active autoxidation catalyst from Fe(II)diimine (FeL3)2+ complexes, lead

to the studies presented in Chapter 3. A family ofbis-Fe(II)diimine (FeL2X2) complexes

were investigated and their activity in cumene and cyclohexane autoxidation examined.

The complexes were chosen following the observation that loss of one diimine ligand

from FeL32+ is likely the first step toward the active catalyst formation. Effects of X

ligand field strength on catalytic activity were investigated and rationalized.

The catalytic activity of FeL3X2 complexes is compared to known autoxidation and

peroxide decomposition catalysts in a polar solvent. The importance of the ligand in the

formation of the active catalyst is explored in Chapter 4.

In the realm of metal catalyzed alkene epoxidation, the use of an affordable and

clean oxidant such as H202 to attain high selectivity is a desirable achievement. Chapter

5 reports on the application of an H202/HCO3 system as an oxygen transfer agent with

previously known manganese epoxidation catalysts. We discuss the catalyzed

epoxidation of styrene using a water soluble Mn-porphyrin complex and the asymmetric

epoxidation of styrene using Jacobsen's catalyst. The effects of the H202/HCO3- system

on catalyst stability and activity are discussed and compared with previously published

literature.














CHAPTER 2
IRON (II) o,a'-DIIMINE CATALYZED HYDROCARBON AUTOXIDATION

Introduction

The tendency of hydrocarbons to form radicals in the presence of oxygen has been

exploited for the synthesis of important hydrocarbon oxidation products though the use of

simple metal salts (Cr, Mn, Fe, Co, Ni, Cu).' Transition metal salts are thought to

catalyze the reaction by increasing the rate of radical formation by decomposing the

intermediate alkyl hydroperoxide. The formation of terephthalic acid from p-xylene and

the oxidation of cyclohexane to cyclohexanone and cyclohexanol (K/A oil) are two key

industrial processes using this chemistry.1,9

Halogenated iron porphyrin catalysts, first described by Lyons et al.,13'15,49-51 are

among the most active hydrocarbon-soluble catalysts for the autoxidation of neat alkanes

at low temperatures. The complexes catalyze the oxidation of alkanes at room

temperature in the presence of molecular oxygen. The unusually high reactivity of the

iron porphyrin complexes was attributed to the ability of the halogenated ligands to

stabilize the energetically unfavorable Fe(II) porphyrin species by increasing the

electrode potential for the Fe(III/I) couple. It was also shown that catalytic activity

increases as the degree of halogenation of the porphyrin ring increases.51 Although the

complexes were developed in an attempt to mimic the oxidative activity of

cytochrome P-450, it has been established that the mechanism does not mimic that of the

enzyme active site. Instead, these reactions primarily follow the long established radical

autoxidation mechanism, in which catalytic decomposition of alkyl hydroperoxides








generates chain propagating radicals ROO* and RO (Figure 1-1).11,13-17,50-54 The

proposed mechanism is supported by the experimental observation that the rate of

substrate oxidation increases with increasing catalyst efficiency toward

disproportionation of the corresponding alkyl hydroperoxide.51 Lyons et al.15 and

Labinger et al.17 have contributed much of the evidence for the radical mechanism

through their investigations of the room temperature oxidation of isobutene and other

substrates catalyzed by halogenated iron porphyrins. The series of mechanistic steps and

rate constants proposed by

Labinger et al.17 closely model the final product concentrations of catalyzed isobutane

oxidation as reported by Lyons et al.53

We have investigated the mechanism of hydrocarbon autoxidation catalyzed by an

iron(II) tris-diimine coordination complex, [Fe(DPP)3](SO3CF3)2]

(DPP=4,7-diphenyl-1,1 0-phenanthroline) 1 and related complexes, which have catalytic

activity55,56 comparable to that of halogenated iron porphyrins.13-15,50,51


















Figure 2-1. [Fe(4,7-diphenyl-1,10-phenanthroline)3]2+ (CF3SO3")2 (1) X-ray structure of
the FeL32+ cation showing 50% probability ellipsoids55








Such reactivity is notable, since 1 lacks halogenated ligands unlike the highly active iron

porphyrins. Hence, halogenated ligands are not a prerequisite to a highly active

autoxidation catalyst.

Prior work by Richardson et al.5558 and Drago et al. showed 1 to be active in

catalyzing cyclohexane autoxidation. There have been no prior reports of using Fe(L)3X2

(L=diimine) complexes as autoxidation catalysts. The use of Fe(1,10-phenanthroline)32+

in the presence of H202 in the hydroxylation of phenol in acidic aqueous solution has

been reported.59 The proposed oxidants are HO- and the in-situ formed

Fe(l,10-phenanthroline)33+ complex. Related Fe(L)2X2 complexes have been reported

for their activity in oxygenating hydrocarbons in the presence of molecular oxygen, and

the proposed mechanism was that of oxygen activation by the metal complex.60'61

The goal of this work is to show that metal-catalyzed peroxide decomposition is the

major catalytic pathway in the autoxidation of cumene catalyzed by precursor 1.

For our kinetic studies, cumene was chosen as the substrate because of its high

boiling point, its tertiary C-H bond, and the stability of cumyl hydroperoxide. Early

investigations of metal-catalyzed cumene oxidation reported the catalytic properties of

Mn, Co and Cu acetates, acetylacetonates, and Co and Fe phthalocyanines.62-64 In the

case of hydrocarbon-soluble metal salts, solvents such as o-dichlorobenzene or

acetonitrile were employed; however, the less soluble salts required the use of a polar

protic solvent such as acetic acid. Although there has been some disagreement on the

subject in the past, Blanchard et al. showed that when using simple metal salts the role of

the metal was to decompose peroxide and not activate oxygen.64 More recent examples

of cumene autoxidation include catalysis by heterogeneous catalysts such as








iron-aluminum oxides65 and polymer supported66-68 simple Cr, Mn, Fe, Co, and Ni salts.

The studies mentioned suggested the primary mechanism was that of metal-catalyzed

peroxide decomposition and radical-chain propagation.

To provide direct comparisons to the porphyrin catalysts,13-17'50'51,53,54,69

tetra-(pentafluorophenyl)-porphyrin iron(III) chloride (2) was used in parallel

experiments (Figure 2-3). In this investigation, we successfully used a mechanism based

on that proposed by Labinger et al.17 to model the product distribution over time for

cumene autoxidation catalyzed by precursors 1 and 2. A series of FeL3X2 complexes

were synthesized by varying the substituents on the 1,1 0-phenanthroline ligands, and the

effects of those substituents on the activity of the catalysts was studied. Ligand

dissociation studies and spectral investigations were done to elucidate the identity of the

active catalyst. Temperature, solvent, pressure and catalyst concentration dependencies

were also investigated.


Figure 2-2. Tetra-(pentafluorophenyl)-porphyrin iron(lI) chloride (2)









Results and Discussion

Catalyzed Cumene Autoxidation

The major products of catalyzed cumene oxidation are 2-phenyl-2-propanol (ROH),

acetophenone (R'O), cumyl hydroperoxide (ROOH) and small amounts of

dicumylperoxide (ROOR) as shown in Figure 2-3. Additional products detected were

trace amounts of a-methylstyrene, as well as overoxidation products CO, CO2 and H20.

Although not determined quantitatively, the formation of CO and CO2 was confirmed by

GC/TCD. GC traces in Figure 2-4 show the analysis of the headspace vapor inside the

reactor. The buildup of CO and CO2 over time is clearly observed.


H OOH OH




I precatalyst +
heat, 02

RH ROOH ROH R'O
Cumene Cumyl hydroperoxide 2-phenyl-2-propanol Acetophenone








I ROOR
Dicumyl peroxide

Figure 2-3. Products of metal-catalyzed cumene autoxidation

Figure 2-5 shows the time dependent formation of the three major products during

cumene oxidation catalyzed by precursor 1. Slow initial formation of ROOH is followed

by the buildup of ROH and R'O over time as the ROOH concentration reaches a steady









state. Product conversions reported in Table 2-1 show precursors 1 and 2 have

comparable activities in catalyzing substrate oxidation by 02, with differences in the

cumyl hydroperoxide and acetophenone selectivities. The catalytic activities of the two

precursors are comparable even at long reaction times, and after 38h of reaction the

oxygen uptake was -0.14 mol for both 1 and 2. The induction periods are ~35 min for 1

and ~70 min for 2 catalyzed oxidation reactions. The oxygen uptake curves in Figure 2-6

clearly show the differences in the length of induction periods for 1 and 2 catalyzed

reactions. Oxygen uptake begins immediately with addition of small amounts (0.04 M)

of cumyl hydroperoxide (Table 2-2).





100




'5




1n1










Figure 2-4. Formation of CO and C02 over time in cumene oxidation catalyzed by 1









Table 2-1. Catalyzed oxidation of cumene
la 2a Co-napb Uncatalyzeda
% Conversion 15.11.2 15.81.8 10 Trace
% ROOH 14.81.5 4.60.4 Trace Trace
% ROH 73.10.6 78.21.4 80 Nd
% R'O 11.90.9 17.21.3 20 Nd
% ROOR -2 -2 Nd
02 uptake(mol) 0.0390.001 0.0470.005 0.05 Trace
a Reactions done at 600 C and 60 psi of constant 02 pressure; catalyst = 57
1iM; 50 mL cumene; 50 mL benzene; reaction time=5 h; Results and errors
calculated from at least 5 experiments for each catalyst. Nd =not detected.
bT=100 C; 6gl Co-naphthanate (57[tM Co metal); over oxidation products
detected (not characterized).


0.45 ROOH (exp) 0
0 ROH (exp)
0.40 V R'O (exp)

0.35- o

0.30 -
0
0.25 -

u 0.20 -

S0.15-
o
0.10 -

0.05 0

0.00- 1 ,1, .
0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
Time (s)

Figure 2-5. Cumene oxidation catalyzed by 1

Water (0.5 M) also decreases the induction period in the case of 2 only.70 The

observation that ROOH and H20 eliminate the induction period suggests that active

catalyst formation is dependent on the concentration of ROOH or H20 in the case of both

1 and 2. Presumably, the small amount of ROOH produced by the background

(uncatalyzed) reaction is enough to convert the catalyst precursor to an active species.









The uncatalyzed reaction (no metal complex added) produces only trace amounts of

ROOH (Table 2-1).

Co-naphthanate is used industrially as a catalyst for cyclohexane autoxidation and

was used for comparison in the present work. The Co catalyzed reaction yielded only

trace amounts of products under our reaction conditions, and even at T=1000C the

conversion was only 10% (Table 2-1). Co-naphthanate was observed to be highly active

in the autoxdation of cyclohexane at a temperature of 135 C (Appendix). Presumably

the Co complex becomes highly catalytic only at very high temperatures.

0.07

2
0.06 -

0.05 -

S0.04




O 0.02 -

0.01 -

0.00
0 50 100 150 200 250 300 350
Time (min)

Figure 2-6. Oxygen uptake curves for 1 and 2 catalyzed oxidation

Table 2-2. Effects of ROOH and H20 on induction period
% Conversion % ROOH % ROH % R'O
1 + ROOHa 19 15 74 11
2 + ROOHa 20 4 78 18
2 + H20b 25 4 72 24
Cumyl hydroperoxide or water was added to cumene/ benzene initially to
observe the effect on induction period. Induction period was not present in
any of the above experiments. [precursor] = 57 p.M, a [ROOH] = 4x102 M,
reaction time = 3hr 35min. b [H20] = 0.5 M, reaction time = 5hr








Catalyzed Cumyl Peroxide Decomposition

Peroxide decomposition under an inert atmosphere was done to determine the

catalytic activity of 1 and 2 toward decomposition of the peroxide in the absence of

substrate. The peroxide decomposition activity of each complex is expected to parallel

its cumene oxidation activity. The major products of cumyl peroxide decomposition

were ROH, 02 and small amounts of the cleavage product R'O. The stoichiometry of the

peroxide decomposition reaction (not including the cleavage product R'O) follows the

net reaction shown in equation 2-1.

ROOH -> ROH + Y2 02 (2-1)

Experimental evidence for the stoichiometric relationship shown in equation 2-1 is

shown in Figure 2-7. As expected, oxygen was formed in approximately half the amount

of the formed ROH. Plots of In [ROOH] vs. time are linear suggesting a first order

dependence in [ROOH]. Plots of In [ROOH] vs. time give experimental kobs values for

the decomposition by 1 and 2 (both 57 AM) as 4.80.3x10-4 s-1 and 1.50.1xl103 s-,

respectively (Figure 2-8). Under the conditions of the peroxide decomposition

experiments activity begins with no induction since ROOH was been added in high

concentrations in order to monitor its disappearance.

Although complex 2 is approximately 3-fold more active as a peroxide

decomposition agent compared to 1, the two complexes appear to have similar activities

in terms of conversion to ROOH, ROH and R'O after a specified period of time (Table 2-

1). However, the induction period exhibited by 2 is approximately twice as long as that

of 1. Although 2 is a more efficient peroxide decomposition agent, the significantly









shorter induction period of 1 results in the formation of approximately the same amount

of ROOH, ROH and R'O as 2.

Co-naphthanate catalyzed cumyl peroxide decomposition gave only traces of

decomposition products at 600C, and this observation is consistent with the low catalytic

activity toward cumene oxidation at this temperature.


0.035-

0.03
0
D 0
0.025 o

S 0.02
o 0
0.015_
0
0.01


0.005 A
SA A
A A 0

0 50 100 150 200 250
Time (min)

ROOH Oxygen A R'O o ROH

Figure 2-7. Metal-catalyzed peroxide decomposition
0
u_= -3-2_4-----1.43------

-

-2 y =-4.843x10 x +-1.436




y -1.455x10-3 x + -1.114
-5

-6 A Precursor I
Precursor 2
-7
0 500 1000 1500 2000 2500 3000 3500 4000
Time (s)
Figure 2-8. Plots of In [ROOH] vs. time








Proposed Mechanism

As discussed in Chapter 1, the general mechanism of uncatalyzed autoxidation is

shown in Figure 1-1. The termination step (kt) in particular lacks much of the detail

present in a complete radical autoxidation mechanism. Figure 2-9 represents the

mechanism and rate constants used to obtain the numerical fits to our experimental

results. Using a numerical modeling program, the steps and rate constants in Figure 2-9

were used to obtain good fits to the experimentally determined product yields for both

autoxidation and peroxide decomposition experiments. The origins and detailed

descriptions of the specific steps and rate constants are discussed below.

Steps 1-8, 12 and 13 are based on a mechanism proposed by Labinger et al.17 for

the Fe(III) halogenated-porphyrin catalyzed autoxidation of isobutane. Since both

isobutane and cumene have only one tertiary C-H bond we expect the same radical steps

to be applicable here. However, the rate constants used in the Labinger study could not

be applied here since the reported values correspond to reactions at 30C. The rate

constants shown in Figure 2-9 were taken from literature and/or estimated from fits to

both oxidation and peroxide decomposition results.

The initiation step rate constant k, = 10-10 s-1 has not been experimentally measured,

but is a value used most often in literature to describe the very slow initiation step

responsible for the long induction periods.17 It is noteworthy that variation of the k, value

(10-1 -10-7 s-') has little impact on the results suggesting the reaction is not carried by the

initiation step (Appendix-A, Figure A-1). The reported experimental rate constant k2 at

60 oC is in the range of 0.93-1.0 M-'s'1.71'72 Rate constants k3 and k4 are comparable to

the published estimated values of 3.8x105 and 4.1x104 M-s'-1, respectively.72

Experimental values for k5 and k6 have not been reported in the literature; however, the








rate constants used in our model are reasonable based on the values reported for

analogous steps in isobutane (30 oC; ks=8x104 M-s-1; k6=6x106 M's"-) and cyclohexane

(150 C; ks=7x107 M''s-'; k6=8x107 M's"') autoxidation.17 The rate constant used in step

7 was tuned according to the experimental yields of R'O since its value varies for

different systems. All of the steps indicated as 'fast' are diffusion controlled under our

conditions and a rate constant of lx1010 M-'s-1 was used. Rate constants for the catalyst

activation steps k9, k1o, and k,1 were determined by numerical fitting to data from

reactions initiated by adding ROOH or H20 (Table 2-2). Since water did not decrease the

induction period when using 1 as a catalyst, this step was not included in the mechanism.

Finally, the metal catalyzed peroxide decomposition steps k12 and kl3 were obtained by

fitting oxidation and peroxide decomposition data.

The sensitivity of the kinetic curves to changes in the values of rate constants is

illustrated in Appendix-A.

Initiation by the direct reaction of most hydrocarbons with molecular oxygen is

thermodynamically and kinetically unfavorable, and chain initiation (Step 1) in the

absence of added initiators usually occurs through substrate attack by peroxidic

impurities present in the substrate. The addition of oxygen to R* is diffusion controlled

under atmospheric pressure, and the predominant radical species in solution is the

alkylperoxy radical (ROO-). The ROO* radical is relatively stable and approximately as

selective as Br*, and ROO* abstracts the most weakly bound hydrogen from the substrate.

In the case of cumene, the weakest bond is the benzylic tertiary C-H bond.









Initiation
(1) RH k = Ixl0-slo S' R 02, fast a ROO*
Propagation
(2) ROO. + RH k = 1.0 M-sl ROOH + R- 02, fast ROO*
Radical chain and termination steps
(3) 2 ROO k3=2.7 x 105 Ms- 2 RO + 02
(4) 2 ROO- k4 2.2 x 104 M-I S-1 ROOR + 02
(5) RO. + RH k= 9.6 x 104 M-s-1 02, fast R
(6) RO. + ROOH k6-= 1.2 xl06 Ms'- ROH + ROO-
(7) RO. k7 = 8.4 x 104 l R'O + Me- 02, fat MeOO*
fast
(8) MeOO-+ ROO fast ROH + CH20 + 02
Catalyst activation steps
(9) FeDPP3 (1) + ROOH k9=10M-IO 1' (active catalyst)
(10) FeTPPF (2) + ROOH -ko004M s 2' (active catalyst)
(11) FeTPPF (2) + H20 =0004Ms"(active catalyst)
Catalytic steps
(12) ROOH + M." k, ROO + H' + M
(13) ROOH + M" k13 -- RO* + HO + Mll

Figure 2-9. Mechanism of metal catalyzed cumene autoxidation with average rate
constants used in the simulation experiments


The C-H bond cleavage is likely to be rapid if the bond energy of ROO-H is at

least as strong as the broken C-H bond of the substrate. The cumene tertiary C-H bond

energy is 84 kcal mol-' compared to the estimated ROO-H bond strength of 90 kcal/mol.

The primary termination pathway for ROO* radicals is through the formation of the

intermediate tetroxide as shown in Figure 2-10. The tetroxide disproportionate either

though the in-cage formation of ROOR (Step 4) or the diffusion of RO* out of the cage

(Step 3).12
















S O 2 + 02

caged species

Figure 2-10. Disproportionation pathways of alkyl tetroxide

The relative rates of the two pathways are dependent on the substrate and reaction

conditions. In our system, the pathway leading to escape of RO* out of the cage is

slightly favored. Once formed, the RO* radical can enter one of three different pathways,

two leading to the formation of ROH and the third leading to R'O. Steps 5 and 6 involve

hydrogen abstraction from both RH and ROOH resulting in the formation of ROH as well

as the propagating radical ROO-. An alternate pathway for the RO* radical is p-scission

to form Me* radical and R'O. The termination of the MeOO* radical is described by step

8 and results in the formation of ROH, CH20 (formaldehyde) and 02. Although CH20

was not detected as a product, we did observe the formation of CO and CO2 both of

which are oxidation products of CH20 (Figure 2-4). Furthermore, the proposed model

predicts a comparable amount of oxygen equivalents accounted for in products ROOH,

ROH and R'O relative to the amount of oxygen uptake as is shown by experiment.

Table 2-3 shows the comparison between experimental and calculated moles of oxygen

uptake and moles of oxygen accounted for in the three products.

Steps 9-11 were added to the mechanism in view of elimination of the induction

period by ROOH and H20 addition. We propose that prior to catalyzed ROOH

decomposition each complex must be converted to the active form of the catalyst by








reaction with ROOH (or H20 in the case of 273) produced initially by the slow metal free

autoxidation step 2 in Figure 2-9. The rate constants k9, k10, and k 1 were estimated by

fitting the experimental data from ROOH and H20 initiated experiments (Table 2-2).

Table 2-3. Experimental and calculated oxygen equivalents accounted for in products
ROOH, ROH and R'O
1 (exp) a 1 (calc) 2 (exp) a 2 (calc)
02 uptake (mol) 0.0390.01 0.0360.02 0.0470.01 0.0480.03
Moles of 02 in 0.0290.01 0.0280.01 0.0330.06 0.0350.02
products b
% 02 in 742 791 69+5 73+1
products bc
a From oxygen uptake curves at time=5h of experiments described in Table 2-1; b
Products ROOH, ROH and R'O only; c Calculated from the ratio of 02 uptake and moles
of 02 accounted for in products.

According to the mechanism, the primary function of the catalyst is to decompose

the intermediate alkyl hydroperoxide (ROOH), and generate chain-propagating radicals

RO* and ROO* through steps 12 and 13, where one of the steps is diffusion controlled

and the other is rate determining. UV-visible spectra of the catalysts in-situ confirm that

Fe(II) and Fe(III) species are the resting states for 1 and 2, respectively. The weakly

oxidizing catalyst precursor 2 (or its activated form) is slowly reduced by the

hydroperoxide (step 12), while the active metal catalyst derived from 1 is slowly oxidized

by the hydroperoxide (step 13). The rate constants for peroxide decomposition by the

active catalytic complexes derived from 1 (k13 = 1.7 M-s'') and 2 (k12 = 7.0 M-'s-') were

estimated by fitting cumyl hydroperoxide decomposition and cumene oxidation data to

the proposed mechanism in Figure 2-9.

Figures 2-10 and 2-11 show the experimental data for cumene oxidation and

peroxide decomposition catalyzed by 1 and kinetic fits calculated by using the proposed

mechanism. The mechanistic scheme provides a reasonable model for predicting the








experimentally determined formation of ROOH, ROH and R'O. In the case of cumene

oxidation and peroxide decomposition catalyzed by 2, the agreement between the

calculated values and those predicted by the model is also very good

(Figures 2-12 and 2-13). Successful modeling of the differences in the induction period

between 1 and 2 was achieved by using step 9 only in modeling reactions catalyzed by 1,

and steps 10 and 11 in modeling reactions catalyzed by 2. Although 2 is activated by

both ROOH and H20, the proposed rate constants of the two steps are much smaller than

k9. The proposed mechanism also closely models the results of ROOH-initiated cumene

oxidation catalyzed by 1 and 2, as shown in Figures 2-14 and 2-5, respectively. The

yields of the uncatalyzed reaction are also predicted closely by the proposed mechanism.

Simulations done by using the steps and rate constants in Figure 2-9 and by setting the

catalyst concentration to zero, predict the formation of only trace products. Uncatalyzed

reaction simulation results are shown in Table 2-4. Product formation calculated by the

model is consistent with experiment as is shown in the GC trace in Figure 2-17. After 5

hours of reaction under our conditions, in the absence of catalyst, only a trace amount of

ROOH was detected with ROH and R'O at concentrations too low to be determined

quantitatively (Table 2-4).

Although the proposed mechanism includes a fair amount of detail, a number of

additional steps could be added in order to account for other pathways and the many

possible trace products formed in autoxidation reactions. The complex, multi-step,

free-radical chain mechanism of autoxidation, containing multiple initiation, propagation

and chain termination pathways has been modeled in greater detail. For example,






38


Tolman et al.74 proposed a cyclohexane autoxidation scheme comprising of 154

reactions.

The steps and rate constants in Figure 2-9 provide a reasonable model that fits all of

the experimental data. The proposed mechanism provides a reasonable model that fits

both catalyzed cumyl peroxide decomposition and cumene oxidation data using rate

constants 10% of the values reported in Figure 2-9. The 10% variation in the rate

constants is within the experimental error for the oxidation experiments (Table 2-1). In

addition, the values for the rate constants change with different media and we would

expect a variation in the rate constants due to media effects in experiments where high

concentrations of H20 or ROOH were added in comparison to the autoxidation runs.


0.50
-- ROOH (calc) o
0.45 ROH (calc)
0. --- R'O (calc)
S* ROOH (exp)
o ROH (exp)
V R'O (exp)
0.30
0.25 .

| 0.20
o
U 0.15 -
0
0.10
0.05 0 ---
0.00 -'' ---

0 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000
Time (s)

Figure 2-11. Simulation of cumene oxidation catalyzed by 1; Data points correspond to
experimental data while the lines are the calculated results













0.30


0.25


0.20

0
1 0.15


o 0.10
U


0.05 -


0.0---"""~-----i-!-i---------i------
0.00 -
0 500 1000 1500 2000 2500 3000 3500

Time (s)


Figure 2-12. Simulation of cumyl peroxide decomposition catalyzed by 1


o 0.4


S0.3

0
U


0 2500 5000 7500 10000 12500 15000

Time (s)


17500 20000


Figure 2-13. Simulation of cumene oxidation catalyzed by 2








































0 1000 2000 3000


Time (s)


Figure 2-14. Simulation of cumyl peroxide decomposition catalyzed by 2


0.6



0.5



0.4-



! 0.3-
)




o 0.2
U
0 o 7


*
0
0
oV


ROOH (exp)
ROH (exp)
R'O (exp)
ROOH (calc)
ROH (calc)
R'O (calc)


~~~'0


0
J
j


0 0


V


0 2000 4000 6000 8000 10000 12000

Time (s)


Figure 2-15. Simulation of ROOH initiated cumene oxidation catalyzed by 1


4000












0.7

0.6 -

0.5 -

o 0.4 -

0.3 -
0
0.2-

0.1 -

0.0


ROOH (exp)
o ROH (exp)
v R'O (exp) o
ROOH (calc)
ROH (calc)
R'O (calc)




/o
/
/
/




0 2000 4000 6000 8000 10000 12000 14000


Time (s)


Figure 2-16. Simulation of ROOH initiated cumene oxidation catalyzed by 2





Benzene Cumene
200 1.6 mmin 3.2 mmin Internal standard (DCB)


ROH 5


.9 mmin ROOH 0.003 M
11.7 min
RO 6.2 min

/


Figure 2-17. GC trace of sample after 5h of uncatalyzed reaction. Reaction conditions;
60 C, 60 psi 02, 50/50 cumene/benzene









Table 2-4. Simulated and experimental results for uncatalyzed cumene autoxidation
ROOH (M) ROH (M) R'O (M)
Simulation results a 2.4x10-' 1.4x 10 2.7x10-6

Experimental results b 3xl103 Trace Trace

Simulation and experimental results correspond to product concentrations after 5 h of
reaction. a Steps and rate constants in Figure 2-9 used to calculate product
concentrations, with the catalyst concentration set to zero. b Reaction conditions; 60 C,
60 psi 02, 50ml/50ml cumene/benzene.

Solvent Dependence

The effect of changing the solvent polarity was investigated for cumene

autoxidation catalyzed by 1. The reactions were carried out at a slightly higher reaction

temperature of 800C for a reaction time of 2 hours. The four solvents used, span a wide

range of dielectric constants as shown in Table 2-5.

Table 2-5. Solvent effects on cumene oxidation catalyzed by 1
Solvent dielectricc dO2/dt a % Conv % ROOH % ROH % R'O
constant) (M/s)
Benzene (2.3) 1.0x10-4 261 61 66+1 28+1
o-dichlorobenzene 0.9x10-4 17+1 152 621 23+1
(9.9)
t-butyl alcohol 1.3x10-4 25+1 9+2 522 39+1
(12.5)
acetonitrile (37) 1.4x10-4 261 141 471 391
a Rate of oxygen uptake obtained from slope of oxygen uptake curves. 50/50
cumene/co-solvent; 1.2x10-4 M 1; run time=2h; 80 oC; 60 psi 02

The results show a small dependence on solvent dielectric constant with o-

dichlorobenzene resulting in the lowest activity of 1 out of the four solvents used. The

reported dependence of the uncatalyzed free-radical steps of cumene autoxidation on

solvent polarity is not significant.75 For example, the reported rate dependence of

cumene autoxidation in nitromethane (s 36) versus chlorobenzene (s 5.6) gives a ratio of

only 1.26, a very modest increase in the rate when using nitromethane as a solvent while

the difference between the dielectric constants is large.' Radical mechanisms involve










neutral reactants and products ; therefore, there is little charge separation in the transition

state and the rate is expected to be relatively insensitive to the solvent.

The solvent does have an effect on product selectivity as well as the length of

induction period as shown in Figure 2-18. For example, when using t-BuOH and CH3CN

as solvents the amount of oxygen uptake after 2 hours is -0.08mol. However, when

benzene is the co-solvent only 0.06mol of oxygen is consumed while the conversion to

ROOH, ROH and R'O is comparable to reactions done in t-BuOH or CH3CN (Table 2-5).

The results suggest that in benzene, the catalysis is more selective toward the formation

of ROOH, ROH and R'O than to over oxidation products such as CO and CO2. The

increased induction period in CH3CN is also apparent when comparing the oxygen uptake

curves in Figure 2-18.

0.10 -
-- Acetonitrile
...... Benzene
0.08 ---- Dichlorobenzene
-08 t-BuOH
.V

I ..



S 0.04 / /...-






0.00 --
0 20 40 60 80 100 120 140
Time (min)

Figure 2-18. Oxygen uptake curves in cumene oxidation catalyzed by 1 in different
solvents










Oxygen Pressure and Temperature Dependence

The dependence of autoxidation reactions on oxygen pressure is expected to be zero,

even at atmospheric pressure. Considering the mechanism in Figure 2-9, the

diffusion-controlled addition of molecular oxygen to R* is the only step where oxygen is

a reactant. The oxygen pressure dependence in cumene oxidation catalyzed by 1 was

investigated over a pressure range of 30-100 psi. The zero dependence of oxygen

pressure is clearly seen in the oxygen uptake curves shown in Figure 2-19. The observed

behavior provides further support for the radical autoxidation mechanism.

The expected effect of increasing the reaction temperature in autoxidation reactions

is that of increased selectivity toward the secondary oxidation products, in our case ROH,

R'O, etc. The ROOH concentration remains at a relative steady state over the range of

temperatures while the conversion to ROH and R'O increases as expected for an

autoxidation mechanism (Figure 2-20).


0.10
30 psi 02
60 psi 0O
0.08 ----- 100 psi 02


0.06- y


0.04 -


0.02


0.00 ,-,i,,
0 20 40 60 80 100 120 140


Figure 2-19.


Time (min)

Dependence of rate of oxygen uptake on oxygen pressure












0.9 N ROOH
0.8 ROH
S0.7 I"R'O
S0.6
S0.5
0.4
S0.3
S0.
0


60 C 70 C 80 C 90 C
Reaction temperature


Figure 2-20. Temperature effects on product selectivity for cumene oxidation catalyzed
by 1

Metal Concentration Dependence

An important feature of metal catalyzed autoxidation reactions is the maximum rate of

oxidation.1,12 The maximum theoretical rate of oxidation is dependent on the substrate

and is determined by the propagation and termination rate constants. The maximum rate

equation derived from the general autoxidation mechanism (Figure 1-1) is shown in

equation 2-2,


-dO2/dt = 2 2 (2-2)
2 k

where kp and kt are the propagation and termination rate constants respectively.12

Equation 2-2 holds true in the presence of metal catalysts as long as they do not catalyze

the propagation and termination steps. Hence the rate of oxygen uptake during metal

catalyzed autoxidation can be increased only up to a limiting value. The theoretical

limiting rate for cumene at 60 C calculated from Eq. 2-2 is 3.1x10-5 Ms-', and was










obtained using kp= 1.0 M-'s1' and kt = 2.1x105 M-I's'. Constants k, and kt were calculated


from literature activation energy values appropriate for a temperature of 60 C.


The maximum rate was also calculated through simulations using our proposed


mechanism and rate constants shown in Figure 2-9, and yielded a value of 3.8x10-5 Ms"l.


The maximum rate was also determined experimentally by increasing the concentration


of 1 until the rate of oxygen uptake leveled off. The experimental maximum rate is


4.4x10-5 Ms'1. Comparison of rates determined though experiment, our kinetic scheme


and the theoretical rate calculated from Eq. 2-2 is shown in Figure 2-21. Our mechanism


predicts a maximum rate more closely aligned with the theoretical maximum rate of


3.1x10-5 Ms-' calculated from Eq. 2-2. The much larger maximum rate measured


experimentally suggests that at high concentrations the metal may be involved in


catalyzing more than just the peroxide decomposition steps. The metal may participate in


reacting with the alkoxy or alkylperoxy radicals.

5.0e-5

*
I" 4.0e-5
S0
O
3.0e-5

0. 0

S2.0e-5 -
0
t*-
0
S
S I.0Oe-5 -
Experimental rates
0 Simulated rates
0.0 1
0.0 1.0e-4 2.0e-4 3.0e-4 4.0e-4 5.0e-4 6.0e-4 7.0e-4
Concentration of 1 (M)


Figure 2-21. Experimental and calculated dependence of catalyst concentration on rate of
oxygen uptake; 60 oC, 60 psi 02, 50/50 benzene/cumene, Run time=5 h








Ligand Dissociation Studies

The possible involvement of ligand dissociation in the activation of precursor 1 was

investigated. Complex 1 is formed in-situ (indicated by UV-visible spectrum

characterization) when Fe(SO3CF3)2 plus 1, 2, 3, or 6 equivalents of

4,7-diphenyl-1,10-phenanthroline (DPP) are used to catalyze the reaction. Product

conversions increase from 1-3 equivalents along with a parallel increase of the in-situ

concentration of 1 (Table 2-6). Using 6 equivalents of DPP results in a decrease in

conversion (2%) although the in-situ concentration of 1 is high. The results confirm that

1 is not the catalytically active species, but a precursor to the active catalyst. Adding 3

equivalents of DPP to a precursor 1-catalyzed oxidation also leads to inhibition. In order

to ensure that the DPP ligand is not acting as a radical trap, 3 DPP equivalents were

added to catalyst precursor 2, in which case no inhibition of catalysis was observed.

Therefore, adding excess ligand equivalents has no general inhibitory effect but only

effects the formation of the active species in 1 catalyzed reactions (Table 2-6).

Table 2-6. Variation of ligand equivalents
Complex Equiv DPP b % Conv
la 0 15
3 2
2a 0 19
3 23
Fe(Trf)2 1 4
2 8
3 10
6 2
Reaction conditions 60 oC, 60psi 02, reaction time=5 h, 100mL
cumene. a57 piM complex. bNumber of equivalents based on
moles of metal complex (57 tM). Trf= (SO3CF3)1-, DPP=4,7-
diphenyl- 1,10-phenanthroline.

The results suggest that ligand dissociation, presumably to an FeL22+ species, could

be an important step in the activation of 1 since equilibrium DPP dissociation would be








suppressed by the presence of free ligand. The ruthenium(II) analogue of 1 (RuDPP3Cl2),

although soluble, has no catalytic activity in either autoxidation or peroxide

decomposition, presumably because of its lower tendency to dissociate due to strong

Ru-N bonds (Table 2-7).

Table 2-7. Cumene oxidation catalyzed by Ru analogue of 1
Complex % Conversion
FeDPP3Cl2 10
RuDPP3Cl2 Trace
Reaction conditions 60 C, 60psi 02, reaction time=5 h, b50mL/5OmL cumene/ o-
dichlorobenzene, 57 pM complex.

Ligand displacement in iron tris a,a'-diimine (L) complexes such as 1 has been

reported to occur in the presence of a strong nucleophile in solvents having a low

dielectric constant.76

(FeL3)2+ (X2)2- FeL2X2 + L

Although FeL32+ complexes are known to be substitution inert in an aqueous solvent, the

interactions of charged species are expected to be very different in non-polar solvents.

The absence of strong ionic solvation leads to increased tendency for strong ion-pairing

between the metal complex cation and the anion in non-polar solvents.

Fe(1,10-phenanthroline)3(C1)2 was reported to form the FeL2C12 complex in dry

dimethylsulfoxide (DMSO e=4.7) due to the Cl- trapping by the FeL22+ species.76'77 The

formation of the bis complex from Fe(l,10-phenanthroline)3(C104)2 was not observed,

presumably because the C104 ion is not nucleophilic enough to trap the FeL22+ species.

Placing 1 in dry DMSO and adding LiCl results in the formation of a shoulder at

X-640 nm corresponding to the FeL2C12 species (Figure 2-22). Therefore, the expected










effect of a more nucleophilic X" counterion is that it would push the equilibrium to the

right by trapping the FeL22+ species.


0.22

0.20 -

0.18 1

S0.16 -

o 0.14-

0.12 FeL Cl

0.10 -

0.08 -

0.06 -

0.04 i
450 500 550 600 650 700 750
Wavelength (nm)

Figure 2-22. Formation of FeL2X2 from 1 in the presence of LiCi (Solvent DMSO)

We investigated the role of the counterion in the activation of the catalyst by

comparing 1 to Fe(4,7-diphenyl- 1,10-phenanthroline)3C12 (FeDPP3Cl2). In our

experiments both the Fe(DPP)32+ chloride and triflate salts have comparable activities

(Table 2-8). The results suggest that the difference in the coordination ability of the

anion does not appear to have an effect on the formation of the active catalyst. It is likely

that the counterion effect on the equilibrium is so small that under our reaction conditions

it does not translate into a shorter induction period. Also the formation of water may

interfere with the coordinating ability of the more nucleophilic counter-ion due to

solvation effects. Alternatively, the ligand dissociation step may not be rate determining.








Table 2-8. Counter ion effects
Complex % Conversion
1 12


Fe(DPP)3Cl2 10
Reaction conditions 60 oC, 60psi 02, reaction time=5 h, 50mL/50mL cumene/ o-
dichlorobenzene, 57 1pM complex. a DPP=4,7-diphenyl- 1,10-phenanthroline.

Precursor Degradation

Both precursors 1 and 2 exhibit a slow decrease in their signature UV-visible

absorbance during the reaction. During 5 hour autoxidation reaction at 600C the

absorbance of 1 decreases by 38%. Direct monitoring of the Uv-visible absorbance of 1

in the presence of ROOH at 60C was done in order to observe the spectral changes

during reaction of 1 with ROOH. Shown in Figure 2-23 are the spectral changes over a

period of 1 hour of a solution of 1 in benzene (1.5x10-6 M) in the presence of 1.5x104 M

ROOH at 600C. The large unresolved band between 430-540 nm corresponds to metal to

ligand charge transfer (MLCT) transitions typical for Fe-tris diimine complexes. The

large off-scale absorbance at -315 nm is a result of ligand-centered Tr-7r* transitions.

Over a time period of 1 hour, the spectrum shows a decrease in the MLCT band at

Aax = 540 nm. Concurrent formation of two shoulders at 380 nm and 650 nm is also

observed. The shoulder at 650 nm may be a result of a FeL33 species formed through the

oxidation of 1 by ROOH. The formation of an [FeL3]3+ species from a

[Fe(1,10-phenanthroline)3]2+ complex in the presence of H202 has been reported in the

literature, and absorbs in the 600-650 nm region.59 The shoulder may also correspond to

a FeL2X2 species shown to absorb in the same region (Figure 2-22). The identity of the

shoulder formed at 380 nm is equally ambiguous. None of the expected products of

ROOH decomposition (ROH, R'O) absorb in the 380 nm region. However, the

absorbance of the free DPP ligand is slightly red shifted compared to the 7r-Tr* transition









of the coordinated ligand; hence, the shoulder in the region may be the result of free DPP

ligand. Alternatively the shoulder may be a result of a new iron species, likely an

inactive iron complex.



1.4-

1.2 -

S1.0 -A



S 0.6 -

0.4-

0.2 '

0.0
200 400 600 800 1000 1200
Wavelength (rim)

Figure 2-23. Spectral changes of 1 at 60 TC in the presence of ROOH

Attempts to isolate an iron-containing residue from the reaction solution following

cumene oxidation catalyzed by 1 resulted in an oil that could not be characterized

conclusively. Extraction of the reaction solution with NaCl saturated water and

qualitative analysis of the aqueous layer using 1,10-phenanthroline and NCS- as

indicators, suggests the presence of both Fe(II) and Fe(III) in solution.

Co-oxidation by Reaction Products

In the absence of direct evidence of the active catalyst it is important to ensure the

reaction products do not catalyze the reaction. Co-catalysis by oxidation products is

referred to as 'co-oxidation' and has been reported in the literature.' Compounds such as

alcohols and ketones are in general more reactive than the saturated hydrocarbon








substrate and in some cases form radicals easily. Cyclohexanone is susceptible to

hydrogen atom abstraction because of the weak a C-H bond and the formation of a

resonance stabilized radical. Therefore, in the autoxidation of cyclohexane

(BDE 99 kcal mol'1), the formation of cyclohexanone (BDE 94 kcal mol') catalyzes the

reaction. The co-oxidation phenomenon is commonly used to improve product yields, for

example in the co-oxidation of alkenes and aldehydes.1 The acylperoxy radicals formed

from the aldehyde and the resulting peracid are utilized for the epoxidation of olefin and

results in higher yield of epoxide than if aledhydes are not present.

Catalysis of cumene autoxidation by the reaction products was checked in order to

ensure that ROOH, R'O and ROH were not catalyzing the reaction by forming radicals.

Adding R'O in the presence of precursor 1 results in slight inhibition, where adding ROH

results in a minor increase in conversion as shown in Table 2-9. The results of adding

ROH and R'O are minor and their effect on the catalysis is considered relatively

insignificant. Trace product ROOR (4x 102 M) was also added to a 1 catalyzed reaction

under identical conditions (Table 2-9). The concentration of ROOR did not change

during the 5 hours of reaction. The presence of ROOR did not interfere with the

expected formation of ROOH, ROH and R'O, and resulted in an oxygen uptake of 0.04

mol (compare to standard result of 0.039 mol in Table 2-1).

Table 2-9. Co-oxidation studies
ROOH (M) ROH (M) R'O (M)
1 0.08+0.1 0.41+0.09 0.07+0.03
1 + R'O 0.14 0.36 0.10
1 + ROH 0.07 0.55 0.13
ROOH only 0.04 0.01 Nd
Reactions done at 60 oC, 60psi 02, 56AM 1, 50/50 cumene/benzene; 4x102 M
added R'O, ROH or ROOH, reaction time=5 h. Blank reaction under the same
conditions yields trace ROOH only. Nd=none detected.










The trace product a-methyl styrene was also checked for its ability to act as a

co-oxidant. Cumene oxidation catalyzed by 1 in the presence of 0.7 M a-methyl styrene

showed that the styrene is slowly converted to acetophenone (R'O). The observation

explains the low concentrations of a-methyl styrene detected in solution. Figure 2-24

shows the oxygen uptake curve and product formation over time in the presence of added

a-methyl styrene. The results show the absence of co-catalysis by a-methyl styrene, on

the contrary the yield of ROH is slightly inhibited, due to the portion of the ROO-

radicals being diverted to reacting with a-methyl styrene. Oxidation of a-methyl styrene

in the presence of 1 under the same reaction conditions did not result in the formation of

any products suggesting any radicals formed in the reaction must come from cumene.

The proposed mechanism of the formation of acetophenone (R'O) from a-methyl styrene

is shown in Figure 2-25.


0.040
Oxygen uptake
0.035 o ROOH
v ROH
0.030 v RO
a-methyl styrene
0.025 -

0.020 -

0.015 V

0.010 -

0.005- U
v o

0.000 V
0 50 100 150 200 250 300 350
Time (min)

Figure 2-24. Product formation and oxygen uptake in cumene oxidation catalyzed by 1
in the presence of a-methyl styrene; Reaction conditions: 60 C, 60psi 02,
56pM 1, 50/50 cumene/benzene











2 CH2 + + 02


.0 *'CH2

CH2 + ROO 02 CH2 + ROOH


/ 0 CH2

CHC2 O
CH2-- -CH2 + CH20

O 0
CH +ROOH CH3 + ROO
H2 + ROOH



Figure 2-25. Proposed mechanism of a-methyl styrene formation

Uncatalyzed cumene oxidation in the presence of added ROOH produced only

small amounts of additional ROOH (Table 2-9). Adding ROH or R'O also in the absence

of catalyst did not yield any additional products even at an elevated reaction temperature

of 800C, as is evident by the moles of oxygen uptake compared to the blank reaction

shown in Table 2-10.

The lack of co-oxidation in the presence of products is consistent with the bond

energies of their abstractable C-H bonds as is shown in Table 2-11. The results support

the assertion that the catalysis is being carried by an activated precursor species and not

co-catalysis by the reaction products.

Table 2-10. Catalysis by products in the absence of 1
Moles of 02 uptake
ROH only 0.015
R'O only 0.011
Blank 0.018
Reactions done at 80 oC, 60psi 02, 4x102 M ROH or R'O, 50/50 CH3CN/cumene.









Table 2-11. Bond dissociation energies
Bond Dissociation Energy (kcal/mol)7'79
Cumene (D RCH3C-H) 83.2-87.3
Acetophenone (D RCOCH2-H) 90.7-93
2-phenyl-2-propanol (D RCH3CO-H) 102-103

Active Catalyst Lifetime

Active catalyst lifetime can be assessed by comparing the experimental oxygen

uptake curve over a long time period to the oxygen uptake curve predicted by the

proposed mechanistic scheme (Figure 2-10).


0.18
Simulation
0.16 -

0.14 -

0.12 -

0.10

F 0.08 -


0
0.04-

0.02 -

0.00 -- i -i
0 250 500 750 1000 1250 1500 1750 2000 2250 2500
Time (min)

Figure 2-26. Comparison of oxygen uptake curve during cumene oxidation catalyzed by
1 and a curve predicted by simulation using proposed mechanism

The proposed mechanism does not include any steps leading to catalyst destruction

so the predicted oxygen uptake curve does not take into consideration any decrease in the

rate due to catalyst degradation. Comparison of the predicted oxygen uptake curve to

that of the experimental curve shows that experimentally oxygen uptake levels off earlier

than predicted suggesting some catalyst decomposition is occurring. The difference









becomes important at reaction times well beyond the 5 hour reaction time of the standard

runs used in our experiments. Therefore, although active catalyst decomposition does

occur during the reactions it occurs at a slow enough rate not to have a significant effect

on the rate of oxygen uptake during the first 8 hours. The results indicate the active

catalyst is robust and remains active for an extended period of time during the reaction.

Ligand variation studies

In order to investigate ligand effects, substituted 1,10-phenanthroline (phen) ligands were

used to make a family of tris Fe(II) complexes. The phen ligand numbering scheme is

shown in Figure 2-27 and the ligands used are listed in Table 2-12. All of the complexes

are SO03CF3- salts except for [Fe(5-N02 phen)3](C104)2. In our previous discussion we

established that counter-ions did not have a noticeable effect on reactivity and we will

assume the same here. All of the complexes in Table 2-12 had low solubility in the

benzene/cumene solvent system; therefore a 50/50 acetonitrile/cumene solvent system

was used.

6 5

7 4




9 NN
9 10 1 2


1,10-phenanthroline (phen)

Figure 2-27. Numbering scheme for the phen ligand

Electron donating substituents such as methyl groups are known to increase the

ligand binding strength of the phen ligand resulting in a more stable complex and electron

withdrawing substituents have the opposite effect. Considering that ligand dissociation








may be an important first step toward the formation of the active catalyst, we might

expect stronger binding ligands to disfavor ligand dissociation leading to longer induction

periods and lower product conversion during a specified time period. Binding strentght

also has an effect on the redox potentials of FeL3X2 complexes have been shown to

parallel ligand binding strength where electron-withdrawing groups shift the potential to

more positive values. A more positive potential results in a less favorable oxidation of

Fe(II) to Fe(III) and presumably a slower peroxide decomposition step (k13, Figure 2-10).

The results of cumene autoxidation catalyzed by a series of FeL3X2 complexes

are shown in Table 2-12. Oxygen uptake curves during the first 100 minutes reveal a

general trend correlating ligand binding strength with length of induction period (Figure

2-28). Therefore, shorter induction periods lead to higher oxygen uptake and conversion

after a specified time period as expected.

While the induction period is affected by ligand binding strength, the maximum

rate of oxygen uptake for all of the complexes is an average of 1.70.4x104 Ms- .

Therefore, the differences in redox potential do not result in significant variation of the

active catalyst activity. The absence of substituent dependence differs from the reported

redox potential influence on Fe(lI)-halogenated porphyrin activity.51'53'80 For a redox

range of -0.221 V to +0.28 V an approximately 3-fold increase in product formation was

reported for a series of Fe(II)-halogenated porphyrin complexes.5' Although the series

of complexes shown in Table 2-12 span a similar range of redox potentials (-0.4 V) we

do not observe any dependence on the redox potential once the reaction reaches the

maximum rate of oxygen uptake. The reason for the absence of substituent dependence

on catalytic activity of FeL3X2 complexes may offer clues about the identity of the active









catalyst. The active catalyst may not contain the phenanthroline ligand, as it is present in

the FeL3X2 complex. The formation of the active catalyst may include the destruction of

the phenanthroline ligands leaving only portions of the ligands.

Table 2-12. Cumene oxidation and peroxide decomposition data of reactions catalyzed
by FeL3X2 complexes


xa Induction 02 uptake d02/dt
Complex Eo (V) 8 period (min) b (mol) c (104 M s )
Fe(5-NO2 phen)32+ 1.18 0 0.050 1.8
Fe(phen)32+ 0.99 25 0.049 1.7
1 0.92 40 0.034 1.2
Fe(4-Me phen)32+ 0.92 170 0.039 1.8
Fe(4,7-Me2phen)32+ 0.86 200 0.029 1.7
Fe(3,4,7,8Me4phen)32+ 0.81 180 0.046 1.9
aAll of the complexes are SO3CF3_ salts except for Fe(5-NO2 phen)3(C104)2.
bInduction period defined as the time when oxygen uptake curve becomes linear.
CReactions done at 600 C and 60 psi of constant 02 pressure; pre-catalyst = 57 pM; 50
mL cumene; 50 mL CH3CN; reaction time=5 h. cReactions done in a Parr reactor at 60
C under argon; pre-catalyst = 57 (JM; Abbreviations: bpy=2,2'-bipyridine, phen=l,10-
phenanthroline.


0.06
Fe(5-NO2 phen)32
0.05 Fe(phen),32
.. Fe(4-Me phen)3 ..
0.04 -- Fe(4,7-Me2 phen)32+
-" -- Fe(3,4,7,8-Me2 phen)32+ "
3 .. -
0.03


>1 0.02 /;-" /


0.01 ^ >"/ t

0.00
0 50 100 150 200 250 300 350
Time (min)

Figure 2-28. Oxygen uptake curves during cumene autoxidation catalyzed by FeL3X2
complexes








In such a case the redox potential of the active species would not be altered by

substituents present on the phenanthroline ligand in the precursor complex.

Fe(3,4,7,8Me4phen)32+ deviates from the pattern of the other five complexes not

only in induction period but the abrupt increase in the rate of oxygen uptake between

140-160 minutes (Figure 2-28). The peculiar behavior of Fe(3,4,7,8Me4phen)32+ could be

related to the ligand being substituted in four positions instead on only one or two.

Conclusions

In summary, 1 is an active autoxidation catalyst precursor comparable in reactivity

to the highly active halogenated iron porphyrins. Although the peroxide decomposition

activity of precursor 2 (k12 = 7.0 M-1s') is higher than that of 1 (k13 = 1.7 M-s's), the later

exhibits a shorter induction period resulting in comparable activities. The proposed

mechanism suggests that the primary function of 1 is to provide an active catalyst that

decomposes the intermediate hydroperoxide and generates chain-propagating radicals.

The rate dependence on oxygen pressure, temperature, solvent polarity, and catalyst

concentration was also investigated and is consistent with a free-radical mechanism.

Results imply 1 is not the active catalyst, and suggest that ligand dissociation is an

important step toward catalyst precursor activation. Uv-visible investigations of spectral

changes during the reaction of 1 with ROOH suggest the formation of new Fe species

in-situ; however, the classification of any such species as the active catalyst is not certain.

Co-oxidation by reaction products was ruled out and hence we conclude that catalysis

through peroxide decomposition is indeed metal-mediated.

A series of FeL3X2 complexes with substituted 1,10-phenanthroline ligands were

synthesized and a correlation between ligand binding strength with length of induction

period was observed. The correlation supports the proposed ligand dissociation step








implicated in catalyst activation. Increasing the binding strength of the ligand results in a

stronger metal ligand interaction and a complex less susceptible to ligand dissociation.

The absence of substituent dependence on catalytic activity of FeL3X2 complexes

suggests that the phenanthroline ligands may not remain in tact in the active catalyst.

Simple iron-phenanthroline precursor 1 discussed here is comparable in activity to

the halogenated iron porphyrins and is among the most active known hydrocarbon

soluble, low temperature, autoxidation catalysts. Related FeL3X2 complexes with

substituted 1,10-phenanthroline ligands are also highly active autoxidation precursors

only when using a polar co-solvent such as acetonitrile to ensure solubility.













CHAPTER 3
HYDROCARBON OXIDATION CATALYZED BY FEL2X2 COMPLEXES

Introduction

The investigation of cumene autoxidation catalyzed by 1 presented in Chapter 2

brings up interesting questions about the identity of the active catalyst. The observed

dependence of iron/ligand ratio on catalytic activity suggests that ligand dissociation is

important in the formation of the active species. In an effort to explore this question, we

investigated a series of Fe(II) bis a,a'-diimine (FeL2X2) complexes. In this chapter we

will compare the catalytic activity and mechanism of catalysis by FeX2L2 complexes with

the previously studied FeL3X2 precursors.

There have been prior reports of FeL2X2 (L=2,2'-bipyridine) complexes used as

catalysts for the air oxidation of hydrocarbons.61'82 In these studies, the proposed

mechanism was that of metal-mediated activation of oxygen leading to a high oxidation

state metal-oxo species followed by direct oxygen transfer to the substrate instead of the

radical-based peroxide decomposition mechanism.60'61'82

Sawyer et al. reported the catalytic activation of 02 by iron(II) bis diimine

complexes for the direct oxygenation of cyclohexene.60'82 Fe(2,2'-bipyridine)22+ was

used as a catalyst. The suggested mechanism involves the formation of a high oxidation

state iron complex and is shown in Figure 3-1. Radical pathways were ruled out since no

cyclohexene hydroperoxide was detected.









Fe1(bpy)22+ + c-C6HI2 [L2Fe"(c-C6Hio)] 02 L2FeI
C6HIo0

Figure 3-1. Proposed mechanism for the activation of 02 by FeL2X2

The presence of 2,6-ditertbutyl-4-methylphenol (BHT), a known alkoxy-radical

trap, was shown to completely inhibit the catalysis, which would suggest that radicals are

present. The proposed function of BHT in this case was that of inhibition by

coordination to the metal to form an inactive complex as shown below.60

Ar-OH (BHT) + (bpy)2Fe"'-OH (bpy)2FeIII-OAr + H20

Other radical traps such as a-tocopherol and quercetin were also used and resulted in

complete inhibition of oxidation. Although the structures of the radical traps are

drastically different, reaction inhibition by all three radical traps was proposed to occur

by the same mechanism shown above. The structures of the radical traps are shown in

Figure 3-2.

HO




a-tocopherol (Vitamin-E)
OHOH
OH
HO O OH t-Bu t-Bu
WYOH
OH 0

3,5,7,3',4'-pentahydroxyflavone 2,6-ditertbutyl-4-methylphenol
Quercetin (BHT)

Figure 3-2. Examples of radical traps.

A related iron(II) complex was reported as an active catalyst for the air oxidation of

cyclohexane. The complex [cis-Fe(2,9-dimethyl-1,10-phenanthroline)2(H20)2](SO3CF3)2








(Fedmp) (Figure 3-3) was suggested to activate molecular oxygen through the same

mechanism as was proposed by Sawyer et al.61 for Fe(2,2'bipyridine)22+.

2+

SN,
SF'"-OH (-SO3CF3)2

N KOH2




Figure 3-3. [cis-Fe(2,9-dimethylphenanthroline)2(H20)2](SO3CF3)2 (Fedmp)

The high temperature (125C) and pressure (550 psi air) at which the oxidation was

done, as well as the formation of typical autoxidation products such as cyclohexanol,

cyclohexanone, adipic acid etc., suggests the possibility that Fedmp catalyzes the reaction

through the free-radical peroxide decomposition mechanism. Our investigation of the air

oxidation of cyclohexane and cumene catalyzed by Fedmp and related complexes will be

discussed and compared to the mechanism and activity of 1.

Results and Discussion

cis-[Fe(2,9-dimethyl-1,10-phenanthroline)2(H20)2](SO3CF3) (Fedmp) was reported

to catalyze the air oxidation of cyclohexane in the presence of small amounts of

cyclohexanone as initiator. The expected yield is 20% conversion to cyclohexanol,

cyclohexanone, adipic acid and small amounts of short chain acids. The reported

products and reaction conditions are shown in Figure 3-4. Because of the limitations of

our high-pressure reactor we were forced to modify the total pressure of the reaction.

The total pressure was dropped to 100psi, this provided an oxygen partial pressure of-60








psi compared to the 1 10psi partial pressure of oxygen when 550 psi air is used as

reported.

H OHO
OC02H
+ + .CO2H


Cyclohexane Cyclohexanol Cycloxexanone Adipic acid


Reaction conditions: 125 oC, 550 psi air, 1.5x10"5 mol Fedmp, 125ml CH3CN,
5ml cyclohexane, 8.2 mL cyclohexanone, reaction time 4 h.

Figure 3-4. Products of the air oxidation of cyclohexane

After 4 h of reaction under 100 psi total pressure (- 60 psi 02) and otherwise

identical conditions as described in Figure 3-4, only traces of cyclohexanol and

cyclohexanone were detected by GC/FID (Figure 3-5, Table 3-1). Increasing the Fedmp

concentration to 10x the reported value results in slightly higher conversion, as shown by

the oxygen uptake curves in Figure 3-6. Even at higher Fedmp concentration, the yield is

only -1% and does not compare well with the reported 20% conversion. Running the

reaction in neat cyclohexane also gave a low product yield of only 3%; however, this may

be because of the insolubility of Fedmp in the neat substrate.

Eventually, we were able to run the oxidation under 550 psi air; however, the

resulting product yields were almost identical to reactions run at a lower pressure

(Table 3-1). The apparent lack of dependence on oxygen pressure is an indicator of a

free-radical autoxidation mechanism as discussed in Chapter 2. Furthermore, the blank

reaction yields approximately the same amount of product as the catalyzed reaction

(Table 3-1). All of the detected products are likely a result of the co-oxidation by the










added cyclohexanone and not metal catalysis. As discussed in Chapter 2, cyclohexanone

(BDE 94 kcal mol-') has a weak a C-H bond and forms radicals more easily than

cyclohexane (BDE 99 kcal mol'1).

The poor agreement of our results and the reported observations led us to question

the accuracy of the GC analysis. Upon closer inspection we observed poor

reproducibility of the GC analysis and poor mass balance between the amount of

substrate consumed and products detected. The low reproducibility of the GC analysis

was possibly because of the low miscibility of cyclohexane in CH3CN, resulting in a non-

homogeneous solution. An attempt to solve the miscibility problem by diluting the

sample in a solvent that was miscible with both components has not improved the

reproducibility to a great extent or changed the results.


1.75-oo



1.50-



1.25-



1.00-



0.75-



0.50-



0.2-






Figure 3-5. GC trace after 4h of cyclohexane oxidation catalyzed by Fedmp









Table 3-1. Cyclohexane oxidation catalyzed by Fedmp
Cyclohexanone (M) Cyclohexanol (M) Run time
100 psi 02 0.01 0.005 4 h
100 psi 02 0.03 0.007 24 h
550 psi air 0.01 0.003 4 h
Blank 100 psi 02 0.01 0.004 4 h
125 C, 1.5x10' mol Fedmp, 125ml CH3CN, 5ml Cyclohexane 5ml, Cyclohexanone
0.008 ml.


8.0e-3

7.0e-3 -

6.0e-3 -

5.0e-3 -

4.0e-3 -

3.0e-3 -

2.0e-3 -

1.0e-3 -


Fedmp 1.5x105 mol
-- Fedmp 1.5x104 mol









.0

4-


0 25 50 75 100 125 150 175 200 225 250
Time (min)


Figure 3-6. Oxygen uptake curves for Fedmp catalyzed cyclohexane oxidation

In view of the apparent low activity and the miscibility problem we chose to

investigate alternative reaction conditions. In our previous studies of cyclohexane

autoxidation catalyzed by 1, we employed a 50/50 o-dichlorobenzene

(DCB)/cyclohexane solvent system.58 Although Fedmp was not fully soluble in the

solvent system at room temperature, the solution was homogeneous at the end of the

reaction suggesting full solubility at the reaction temperature. Products of Fedmp

catalyzed reaction are cyclohexanone, cyclohexanol, adipic acid and three


V.V









uncharacterized minor products Figure 3-7. The same major products were observed in

our studies of cyclohexane oxidation catalyzed by 1. Comparison of the product yields

using the two complexes is shown in Table 3-2. Complex 1 is clearly more active and

yields more products at lower complex concentrations and shorter reaction time. The

drastic difference in the catalytic activity between 1 and Fedmp can be appreciated by

examining the oxygen uptake curves during cyclohexane oxidation. Fedmp exhibits a

long induction period and slow oxygen uptake while 1 has practically no induction and

consumes almost double the oxygen in only 30 minutes Figure 3-8. Following the

30-minute time period the oxygen uptake curve of 1 levels off. Prior research in the

Richardson labs has shown that oxygen uptake ceases because of buildup of CO and CO2

in the reactor headspace causing the internal pressure to exceed the set regulator pressure

of the incoming 02.58








1.25-





0. 0-
07


I I








Figure 3-7. GC trace of cyclohexane oxidation in 50/50 cyclohexane/DCB
Figure 3-7. GC trace of cyclohexane oxidation in 50/50 cyclohexane/DCB








Table 3-2. Products from cyclohexane oxidation catalyzed by 1 and Fedmp
Amt. Cat. Cy-one (mol) Cy-ol (mol) Adipic acid Run time
(mol) (mol)
Fedmp 6.2x10' 1.3x103 1.0x102 1.8x10-4 4h
1 3.5x10-5 1.0x10-2 1.2x10-2 1.6xl0"3 2 h
Reaction done in 50/50 DCB/cyclohexane by volume, 135 C, and 50psi oxygen


0 25 50 75 100 125 150 175 200 225 250
Time (min)

Figure 3-8. Oxygen uptake curves for cyclohexane oxidation calyzed by 1 and Fedmp.
Reaction conditions same as in Table 3-2

Considering the lack of dependence on oxygen pressure and the presence of an

induction period, the most likely mechanism of catalysis is that of metal catalyzed

peroxide decomposition and not metal mediated activation of oxygen leading to a high

oxidation state metal-oxo species followed by direct oxygen transfer to the substrate.








Cumene Oxidation

In order to compare the mechanism and activity of Fedmp to that of 1 we turned

to cumene oxidation. In the following experiments we use Fedmp in cumene oxidation

and cumyl peroxide decomposition.

Under standard conditions used in Chapter 2 (Table 3-3) at a temperature of 600C

the reaction is extremely slow with no detectable products after 5 hours of reaction.

Oxygen uptake starts only after a long induction period of 300 min (5 h) as shown in

Figure 3-9. After 32 hours of reaction time the predominant product is ROOH and

suggests that Fedmp is a poor peroxide decomposition catalyst. The supposition is

supported by the observation that peroxide decomposition catalyzed by Fedmp at 60C

under an argon atmosphere yielded no detectable decomposition of the peroxide over a

period of 4 hours. Increasing the reaction temperature to 800C results in the formation of

detectable products following a 2 hour reaction as shown in the GC trace in Figure 3-10.

The low peroxide decomposition activity of Fedmp parallels its low cumene

oxidation activity. The results are consistent with a metal catalyzed peroxide

decomposition mechanism.

Table 3-3. Cumene oxidation catalyzed by Fedmp
Temp (oC) % Conv. % ROOH % ROH % R'O Run time
Fedmp 60 26 81 16 3 32 h
Fedmp 80 5 80 18 2 2 h
1 80 24 15 47 38 2 h
1.2x10'- mol complex, 50/50 CH3CN/cumene, 60 oC, 60 psi 02













0.08

0.07 -

0.06 -

0.05 -


II 0.04 -

e 0.03 -

0.02 -

0.01 -

0.00 -
0 250 500 750 1000 1250 1500 1750 2000

Time (min)


Figure 3-9. Oxygen uptake curve during cumene oxidation catalyzed by Fedmp at 60C


Minutes
Figure 3-10. GC trace of sample taken after 2 h of cumene oxidation catalyzed by Fedmp











Solution-state studies of Fedmp

The low activity of Fedmp in cumene oxidation and cumyl peroxide decomposition

prompted us to investigate the solution-state structure of the complex. Upon closer

solution-state analysis of Fedmp it became clear that the cis configuration might not be

retained in solution. Although CHN analysis supports a 2:1 ligand metal ratio, the

H'-NMR analysis does not support the presence of the cis configuration in solution. An

H'-NMR spectrum of the dmp ligand and Fedmp complex in de-acetone is shown in

Figure 3-11. If the solution geometry was indeed cis we would expect the dmp ligand to

lose its C2 symmetry axis and result in 2 methyl peaks and 6 aromatic hydrogen peaks. If

we compare the Fedmp spectrum to that of the symmetrical free dmp ligand we see very

little difference between the two except for the downfield shift of the Fedmp peaks.

Hc Hc He -CH,
Hb -K Hb
Ha Ha a


H3C N ---CH,




8.0 7.0 6.0 5.0 4.0 3.0 2.0

Fedmp
He -CH,
Residual
Ha Hb Solvent
peak




t I I I I | r ,I ,. I i I I I I I

ppJ 1) 8.0 7.0 6.0 5.0 4.0 3.0 2.0

Figure 3-11. Top; 2,9-dimethylphenanthroline (dmp), Bottom; Fedmp in de-acetone








Variable temperature studies were done to rule out fast exchange between the two

ligands, which would make them appear symmetrical. Spectra down to -600C do not

reveal any splitting of the aromatic or methyl peaks suggesting that there is no fast ligand

exchange occurring in solution (Figure 3-12). On the contrary all of the peaks are

sharpened as the temperature is lowered. The probability that the complex adopts a trans

geometry is low because of the steric interactions of the methyl groups. In fact there is

no reported trans complex of the FeL2X2 (L=diimine) type, even in the absence of

substituents in the 2 and 9 positions. The only know example of a related trans complex

is that of Ru(bipy)2(OH2)22+, which is formed through photoisomerization of the cis

complex.83 We conclude that although in the solid state Fedmp exists as the cis

complex, once it is dissolved the complex loses this geometry and has a symmetrical

structure with respect to the dmp ligand. Disproportionation of Fephen2X2 complexes

has been previously reported in the literature, where the more stable tri-substituted

complex is formed.84

2 FeL2X2 -> FeL3X2 + FeX2 + L

Although a Fedmp3X2 complex has never been isolated, presumably because of the

inability of three 2,9-dimethyl- 1,10-phenanthroline ligands to approach the metal closely

for good overlap, it is possible that in solution the three ligands aggregate around one

Fe2+ to stabilize the ion. Instability of Fedmp2X2 complexes in solution has been

previously reported, and most of the literature characterization was done in the solid

state.8586 We rationalized based on ligand field strength of the X- ligands that a more

stable analogue of Fedmp would be Fedmp2NCS2; however, once synthesized it too had

a symmetrical H1-NMR spectrum.












25 "C


ppm (f1) 8.50 8.00


-CH3


25 C



-10 *C



-60 OC

ppm(A9 200 250


Residual
solvent
peak


A2.30 220 2.10 2
2.30 220 2.10 200 1.90


Figure 3-12. Variable temperature H'-NMR spectrum of Fedmp in de-acetone. Top;
aromatic region, Bottom; methyl protons and residual solvent peaks








The dmp ligand does not provide as strong a ligand field as other a,a -diimines that

are not substituted in the 2 and 9 positions. On the basis of electron-donating character of

the methyl substituents, the dmp ligand is expected to be a stronger field ligand compared

to 1,10-phenanthroline (phen). Comparison of x-ray structures of Fedmp2NCS2 and

Fephen2NCS2 reveal the reasons for the weaker binding of the dmp ligand to iron. The

Fe-Ndiimine distances were shown to be on average longer in the case of dmp (2.27 A)

compared to average bond lengths of 2.205 A for Fephen2NCS2.87'88 The longer bond

length suggests a weaker interaction between the metal and dmp ligand. The reason for

the longer bonds can be attributed to steric crowding of two methyl groups. Each of the

methyl groups experiences intramolecular contacts with the neighboring dmp ligand. In

order to avoid the close contacts with the neighboring ligand the methyl groups are forced

to twist away from the interaction. The distortions caused by the crowding are evident

from the planarity of the ligand, where the largest deviation from the plane of the ligand

is 0.162 for Fedmp2NCS2 compared to only 0.039 A in Fephen2NCS2. The result of the

ligands bending away from each other is that the iron atom deviates 1.004 A from the

plane of the dmp ligands. This compares to an approximate planar configuration of the

metal in Fephen2NCS2 where the deviation is only 0.077 A.88

In view of the poor dmp ligand orbital overlap with the metal we chose to explore

FeL2X2 complexes where L= 1,10-phenanthroline (phen), in hopes of retaining the cis

configuration in solution.

We prepared a series of cis-Fephen2X2 (X=CN-, NCS-, Cl) complexes in order to

compare them to the activity of 1 and evaluate the effect of changing the X ligand.









The complexes in the series were characterized by CHN and IR the results of which

support the assignment of a cis geometry. The complexes maintain the cis configuration

in solution as is clearly demonstrated by the H1-NMR spectrum of Fephen2CN2

(Figure 3-13). The cis configuration is expected to lead to loss ofphen ligand symmetry

resulting in 8 different aromatic proton resonances instead of only 4 present in the

uncoordinated ligand.


K)
0
0
-m
lcJfl)


K) K)J
8 0
8 50
8.50


8.00


Figure 3-13. 'H-NMR spectrum of Fephen2CN2

'H-NMR spectra of Fephen2C12 and Fephen2NCS2 were not obtained because of poor

solubility in standard NMR solvents. The cis arrangement in solution was verified by

Uv-visible spectroscopy (Chapter 6).

In previous studies of cumene autoxidation catalyzed by 1, results suggest that

ligand dissociation is an important step in catalyst activation. In the case of Fephen2X2


He Hd

Hf Hc


Hg _Hb
N N~
Hh Ha








complexes one of the diimine ligands is replaced by two X' ligands. Such a change in the

ligand environment should have a notable effect on the formation of the active catalyst as

well as the robustness of the complex. Presumably the weaker X- ligand will favor

dissociation and the formation of the active species.

The relative reactivities of the three Fephen2X2 complexes are in agreement with

the ligand field strength of the X ligands, CN < NCS < Cl (Table 3-4). The complex

bearing CN- ligands is by far the least active and has the longest induction period

(Figure 3-14). The observation is not entirely surprising since CN- is a stronger field

ligand than even 1,10-phenanthroline.

Table 3-4. Cumene oxidation catalyzed by Fephen2X2 complexes
Complex %Conv %ROOH %ROH %RO
Fephen2CN2a 7 1 451 392 161
Fephen2NCS2 a 151 485 323 213
Fephen2Cl2 b 181 204 592 223
Fedmp a 51 791 181 11
la 262 141 471 391
50/50 cosolvent/cumene, 1.2x105 mol complex, 80 C, 60psi 02, run
time=2 h; Product yields and errors calculated from 2 or more
experiments; a cosolvent CH3CN; bcosolvent o-dichlorobenzene

The difference in product conversion between Fephen2NCS2 and Fephen2Cl2 is small, but

the difference in moles of oxygen consumed is much larger as shown in Figure 3-14. The

reason for the apparent discrepancy is that Fephen2Cl2 is much less selective and

produces more secondary reaction products such as CO and CO2.

The activity differences between Fephen2NCS2 and Fephen2Cl2 cannot be

interpreted entirely based on ligand field strength since the autoxidation reactions were

done in different solvents. Unlike the NCS- and CN complexes the Fephen2Cl2 complex

disproportionate to the FeL3Cl2 complex in CH3CN and the reactions had to be done in

o-dichlorobenzene.










We might have expected that Fephen2NCS2 and Fephen2Cl2 would be more active

than 1 based on the ligand field strength order ofphen > NCS- > Cl ; however, this is not

the case. Although bleaching of the FeL2X2 complexes is observed to occur within the

first 30 minutes, the rate of oxygen uptake does not show evidence of slowing down

because of degradation of the active catalyst (Figure 3-14). The low activity is likely a

direct result of replacing one phenanthroline ligand with two X ligands. Presumably the

X ligands are not able to provide the appropriate environment for the formation of a

highly active catalyst.


0.07
Fephen2CI2
0.06 .. Fephen2NCS2
--- Fephen2CN2
S 0.05 -
o

-. 0.04 .-

0.03 -

O 0.02 -

0.01 ..

0.00 "-- ."--
0 20 40 60 80 100 120
Time (min)

Figure 3-14. Oxygen uptake curves during cumene oxidation catalyzed by Fephen2X2
complexes

Conclusions

The Fedpm catalyst was shown to be a poor catalyst for the oxidation of both

cumene and cyclohexane under autoxidizing conditions. Solution state studies suggest

that the cis configuration of Fedmp is not retained in solution. Further evidence suggests









that the dmp ligand does not bind to the metal as strongly as other diimine ligands not

substituted in the 2,9 positions. The weak ligand strength of the dmp ligand may be the

reason for the poor catalytic activity. Product profiles, the presence of an induction

period and zero dependence on oxygen pressure all suggest that Fedmp catalyzes the

reaction by decomposition of the intermediate peroxide and not through molecular

oxygen activation and direct transfer of the oxygen to the substrate as previously

suggested.

Fephen2X2 complexes are less catalytically active in cumene autoxidation

compared to the previously studied complex 1. The low activity does not appear to be a

result of active catalyst decomposition since the rate of oxygen uptake does not slow

down during the reaction. Presumably the X- ligands are not as good as

1,10-phenanthroline ligands at providing the needed ligand environment for the

formation of the active catalyst.














CHAPTER 4
COMPARISON OF FEL3X2 TO KNOWN AUTOXIDATION CATALYSTS

Introduction

The uniquely high activity of 1 as an autoxidation catalyst in hydrocarbon media is

a direct result of the lipophilic nature of the ligand, which solubilizes the metal in the

non-polar reaction medium thereby facilitating catalysis. We were interested in

comparing the activity of 1 to well-known autoxidation and peroxide decomposition

catalysts not soluble in hydrophobic solvents such as benzene. In order to assess the

relative activity of 1 compared to other autoxidation catalysts the following discussion

focuses on reactions done in acetonitrile as a co-solvent.

Results and Discussion

Fe and Co Acetylacetonates

Many of the early studies of autoxidation were with Co, Cu and Fe acetates,

decanoates and acetylacetonates (acac=2,4-pentanedionate). One of the more active

autoxidation catalysts is Co(II)acac2.8 Under our conditions the activity of the Co(II)

complex is approximately half that of 1 (Table 4-1). In the case of Co(lII)acac3 the

reaction yields only trace ROOH and we observe no color change of the reaction solution

indicating the Co(lI)acac3 remains unaltered throughout the reaction. When using the

analogous Fe(II)acac2 or Fe(III)acac3 the conversion to products is very low in both

cases.








Table 4-1. Cumene oxidation catalyzed by Fe and Co acetyl acetonates
Complex % Convy % ROOH % ROH % R'O
Fe(II)acac2 1 94 6 0
Fe(III)acac3 trace trace 0 0
Co(II)acac2 12 63 25 12
Co(III)acac3 trace trace 0 0
1 24 15 47 38
Blank 0.5 >99 trace trace
50/50 cumene/CH3CN; 1.2x10 M metal complex; run time=2h; 80 oC; 60 psi 02

Fe Complexed by Macrocylic Ligands

Fe(II)phthalocyanine (Fe(II)Pc) is considered analogous to iron porphyrin

complexes because of the similarity of the phthalocyanine ligand (Figure 4-1) to

porphyrin.89 The interest in using Fe(II)Pc complexes in the presence of oxygen atom

donors such as iodosyl benzene and alkyl peroxides stems from the desire to obtain

synthetic mimics of the enzyme P-450 as discussed in Chapter 1. Reports of using

Fe(II)Pc supported on zeolite or carbon black as a catalyst in cyclohexane oxidation in

the presence of t-BuOOH suggest a non free-radical mechanism.90'91 Fe(II)Pc is also

reported to have high peroxide decomposition activity, and has been used as an

autoxidation catalyst.8




N N-- N
NNH HN
N N
-NH HN- N




1,4,8,11 -tetraazacyclotetradecane (cyclam) phthalocyanine (Pc)

Figure 4-1. Structures of macrocyclic ligands








Under our reaction conditions, Fe(II)Pc catalyzed oxidation of cumene resulted in

moderate activity (14 % conv.); however, still considerably lower than 1 (Table 4-2).

Fe(II) 1,4,8,11-tetaazacyclotetradecane (Fe(II)cyclam) complexes have been used in the

catalytic epoxidation of olefins both in the presence of iodosyl benzene and H202. The

observed high selectivity to the epoxide suggests the absence of radical species.92

However, when using alkyl peroxides the high selectivity to the epoxide is lost because

of homolytic cleavage of the peroxide leading to the formation of free radicals.

Cumene oxidation catalyzed by Fe(II)cyclam was observed to give very low

product yields as shown in Table 4-2. Oxidative attack on the methylinic carbons of the

ligand may contribute to active catalyst destruction and the low activity.

Table 4-2. Cumene oxidation catalyzed by Fe(II)Pc and Fe(II)cyclam
Complex % Convy % ROOH % ROH % R'O
Fe(II)Pc 14 49 28 23
Fe(II)cyclam 3 56 25 19
1 24 15 47 38
blank 0.5 >99 Trace trace
50/50 cumene/CH3CN; 1.2x104 M metal complex; run time=2h; 80 oC; 60 psi 02

Fe Complexed by Hexa and Tetra Coordinating Pyridyl Ligands

In view of the high activity of pyridyl ligands such as 1,10-phenanthroline we

investigated other examples of pyridyl-based ligands. Some of the more common

examples are 2-pyridylmethyl ligands such as tris(2-pyridylmethyl) amine (tpa) and

N,N,N',N'-tetrakis(2-pyridylmethyl)-aminopyridyl (tpen) both of which were chosen for

our study and are shown in Figure 4-2. Iron tpen and tpa complexes have been used

extensively as mimics of non-heme iron oxygen activating enzymes.93'94 The formation

of iron (1II) peroxo species from both Fetpen and Fetpa complexes in the presence of

H202 or t-BuOOH has recently been reported.93'95'96 Peroxoiron(nII) complexes are








increasingly being considered as potential intermediates in oxidations catalyzed by non-

heme iron centers in biology. Numerous claims of selective oxygen transfer to alkanes

and alkenes particularly by mono and di-nuclear Fetpa complexes in the presence of alkyl

peroxides have been reported.97 Such claims have been challenged by the view that the

peroxoiron(III) species slowly undergoes homolytic scission to form free radicals

responsible for the observed products.98'99 We chose two complexes of the type

discussed above and investigated the peroxide decomposition activity of

Fe(III)tpen(C104) and (Fe(II)tpa(CH3CN)2](SO3CF3)2 as well as their activity in cumene

autoxidation. The catalytic activity in the air oxidation of ethylbenzene of a large group

of iron complexes bearing penta, hexa and tetradentate pyridyl type ligands has recently

been reported.100 (Fe(II)tpa(CH3CN)2](SO3CF3)2 and an iron complex bearing a ligand

closely related to tpen were included in the report.

Cumene oxidation catalyzed by Fe(tpen)(C104)3 resulted in high conversions especially

in the early minutes of the reaction. The initial rate of the reaction was approximately

twice that seen for the 1 catalyzed reaction. The rate of oxygen uptake was 1.4xl0-3 Ms"-1

for Fe(tpen)(C104)3 (Fetpen) compared to the maximum rate of 8.0x10-4 Ms-1 for the 1

catalyzed autoxidation. Although no induction period was observed for Fe(tpen), the

oxygen uptake slowed down after ~40 min indicating catalyst destruction. A comparison

of the oxygen uptake curves is shown in Figure 4-3. The absence of induction period

suggests that Fetpen is an extremely efficient peroxide decomposition catalyst.








N








tris(2-pyridylmethyl)amine (TPA) N,N,N',N'-tetrakis(2-pyridylmethyl)-aminopyridyl (TPEN)

Figure 4-2. Structures of tpa and tpen ligands


Table 4-3. Cumene autoxidation catalyzed by Co and Fe complexes
Complex % Conv % ROOH % ROH % R'O
Fe(III)tpen(C104) 19 20 31 49
[Fe(II)tpa(CH3CN)2] 34 4 45 51
(S03CF3)2
[Fe(II)tpa(CH3CN)2] 10 14 67 19
(S03CF3)2 at 30 C (5 h)
Fe(SO3CF3)2 2 64 9 26
Fe(Pyridine)4(Cl)2 0.7 >99 trace trace
1 24 15 47 38
Fe(phen)3(SO3CF3)2 36 10 46 44
blank 0.5 >99 trace trace
50/50 cumene/CH3CN; 1.2x104 M metal complex; run time=2h; 80 oC;
60 psi 02

Cumene oxidation catalyzed by [Fe(II)tpa(CH3CN)2](SO3CF3)2 (Fetpa) results in

the highest activity seen for any of the complexes investigated in this study. In addition

to the three major products listed in Table 4-3, a considerable amount of

at-dimethyl benzyl methyl ether is also formed. The formation of the ether likely occurs

through the reaction of the RO- radical with MeOO- as shown in Figure 4-4. The

proposed pathway is reasonable since the Of-cleavage of RO- results in the formation of

R'O and MeOO- (Figure 2-10, Step 7). Assuming most of the R'O formed occurs though






84


step 7, the high selectivity toward R'O dictates the formation of large concentration of

the MeOO- radical as well.


1.2e-1

1le- [Fe(II)tpa(CH3CN)2](SO3CF3)2
Fe(III)tpen(C104)3
9.0e-2 -
0
S7.5e-2 -
o
1 6.0e-2 ... "

60 4.5e-2 -

3.0e-2 -

1 5 e 2 ,


0 10 20 30 40 50 60 70 80 90 100 110 120 130
Time (min)

Figure 4-3. Comparison of oxygen uptake curves during cumene oxidation


0o


+ MeOO *


S____ 01


+ 02


a,a-dimethyl
benzyl methyl ether


Figure 4-4. Proposed mechanism of aa-dimethyl benzyl methyl ether formation


Unlike Fetpen, Fetpa does not suffer from loss of activity and continues catalyzing

the oxidation although at a somewhat slower rate after -25 minutes (Figure 4-3). Fetpa is

an active catalyst even at 30 oC, where 1 has no activity at all (Table 4-3). Cumyl

peroxide decomposition catalyzed by Fetpa is extremely rapid even at room temperature.









Approximately half of the ROOH is decomposed during the first minute, followed by a

considerably slower rate of decomposition. The observed decrease in ROOH

concentration over time is shown in Figure 4-5. Upon the addition of the complex to a

solution of 0.1M ROOH a short-lived blue species is observed.

0.14

0.12

S0.10 -

0.08 -
o *

-2 0.06 -

S0.04 -

0.02 -

0.00 - -,
0 100 200 300 400 500 600 700 800 900 1000 1100 1200
Time (s)
Figure 4-5. Cumyl peroxide (ROOH) decomposition catalyzed by Fetpa


The blue color is consistent with a previously reported blue Fe(mI)peroxo species

(ax~ 600 nm) formed when reacting Fetpa and t-BuOOH.93 Following the rapid

disappearance of the blue intermediate, a reddish brown solution remained. The blue

Fe(lII)peroxo species likely decomposes through the homolytic decomposition of the

FeO-OR bond. The formation of alkoxy radicals from alkyl hydroperoxides exposed to

Fetpa complexes has been reported.98'99

In order to determine whether the species formed after decomposition of the blue

Fe(UI)peroxo complex is catalytically active, we added one equivalent of cumyl peroxide

to a solution of Fetpa and used that solution as the catalyst in cumene oxidation. To our