COMPUTATIONAL STUDIES OF ANTI-TUMOR DRUG TIRAPAZAMINE AND REACTIONS AND REARRANGEMENTS OF AND

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

By

Jin Liu, B.S.

*****

The Ohio State University 2005

Dissertation Committee: Approved by Professor Matthew S. Platz, Advisor

Professor Christopher M. Hadad ______Professor Sherwin J. Singer Advisor Graduate Program in Chemistry Professor Anne B. McCoy

ABSTRACT

The two possible mechanisms of action of the anti-cancer drug tirapazamine have been investigated using density functional theory (DFT). Tirapazamine is a novel anti- cancer drug which is inactive in -rich, healthy cells but is active in oxygen-poor, cancerous cells and induces DNA strand scission. We calculate that the mechanism proceeds via a series of electron-transfer/proton-transfer/fragmentation steps to eventually produce hydroxyl in a series of energetically favorable reactions. It has been proposed that the generated may then react with the sugar residues of the DNA backbone moiety to produce sugar-centered radicals. A direct mechanism involving -atom abstraction from the sugar residues by a tirapazamine-centered radical anion or its is not energetically feasible. However, the hydroxyl radical can remove a hydrogen atom from the amino group of tirapazamine to form an iminyl radical. This radical can remove a hydrogen atom of a DNA sugar in an exothermic process. The calculations predict that tirapazamine, acting as a surrogate for molecular oxygen, can then react exothermically with the sugar-centered radical to oxygenate the radical center and thereby induce the DNA strand break. Related reactions of some tirapazamine analogs were studied, and promising new drug candidates were discussed.

ii

DFT, CCSD(T), and CBS-QB3 calculations were performed to understand the

chemical and reactivity differences between acetylnitrene (CH3C(=O)N) and

methoxycarbonylnitrene (CH3OC(=O)N) as well as sulfonylnitrenes. Acetyl azide is

calculated to decompose by concerted migration of the along with extrusion. Methoxycarbonyl azide, on the other hand, does have a preference for stepwise Curtius rearrangement via the free . Methanesulfonyl azide prefers to decompose to form singlet methanesulfonylnitrene rather than to extrude nitrogen in concert with sulfurylimine formation (pseudo Curtius Rearrangement). The bimolecular reactions of acetylnitrene, methoxycarbonylnitrene and methanesulfonylnitrene with , and were calculated and found to have enthalpic barriers that are near zero and free energy barriers are controlled by entropy.

Similar calculations were performed on diazoketones, diazoesters and

diazoalkanes. We find that conformations in which the migrating group and the diazo

moiety have an anti disposition extrude nitrogen and undergo concerted Wolff

Rearrangement to and avoid the formation of the free . When these groups

have a syn disposition, carbenes are formed, and these intermediates subsequently

rearrange. Diazoacetone favors a concerted thermal decomposition and methyl

diazoacetate prefer stepwise carbene formation. These preferences are explained with the

use of isodesmic equations, and upon consideration of the geometries of the transition

states and the free carbenes.

iii

The reactions with oxygen of triplet carbenes and triplet nitrenes were calculated

by DFT, CCSD(T) and CASPT2 methods. The fact that triplet carbenes react faster with oxygen than do triplet nitrenes has been explained on the basis of the strength of the

bonds being formed.

iv

To my parents and Peng

v

ACKNOWLEDGMENTS

Standing at a new milestone of my , I look back to my journey on the road of

science. I can see so many people leading me, supporting me, and encouraging me on this

road.

I would first thank two of my advisors, Dr. Matthew S. Platz and Dr. Christopher

M. Hadad. It is their intellectual support and patient guidance that made this dissertation possible. They set a lifetime’s worth of example for me to follow, as a scientist and as a person.

I appreciate that many faculty members in physical division gave me a

tremendous education. They led me into the world of physical chemistry through their

classes and personal discussions. I would give my special thanks to Dr. Singer and Dr.

McCoy as my committee members.

I would like to mention all group members in Platz and Hadad group, for their

friendship, encouragement, and discussions. I will cherish the wonderful moments we

shared. I also would like to thank The Ohio Supercomputer Center for technical support

and NSF-funded Environmental Molecular Science Institute at OSU for partial financial support.

vi

My parents, Qiaoyun Liu and Binyan Liu, know nothing about chemistry. What

they know is, love. Their love from my mother land, the other side of the Earth, support

me to be a strong person during my five years studies in this foreign country. They are not only my parents, but also my mentors and best friends. I love you both, mom and dad, with all my heart.

Finally, I would thank the most important person in my life, Peng Tao. As a lab

colleague, your intelligent insight helped me tremendously during my completion of this

dissertation; as a friend, you are always with me during my countless tough time; as a life partner, your love and support are incredible. I think I am ready to accept your request and to become the luckiest person in the world, spending the rest of my journey with you.

Yes, I do.

vii

VITA

May 19, 1978...... Born – Xi’an, China

1996 - 2000...... B. S. Chemistry Beijing University, China

2000 – 2005...... Graduate Teaching and Research Associate, The Ohio State University.

PUBLICATIONS

Research Publications

1. Liu, J.; Mandel, S.; Hadad, C. M.; Platz, M. S. “A Comparison of Acetyl and Methoxycarbonylnitrenes by Computational methods and a Laser Flash Photolysis Study of Benzoylnitrene.” J. Org. Chem. 2004, 69, 8583-8593.

2. Liu, J.; Hadad, C. M.; Platz, M. S. “The Reaction of Triplet Nitrene with Oxygen: A Computational Study.” Org. Lett. 2005, 7, 549-552.

3. Mandel, S.; Liu, J.; Hadad, C. M.; Platz, M. S. “A Study of Singlet and Triplet 2,6 – Difluorophenylnitrene by Time Resolved Infrared ” J. Phys. Chem. A 2005, 109, 2816-2821.

FIELDS OF STUDY

Major Field: Chemistry

viii

TABLE OF CONTENTS

P a g e

Abstract...... ii

Dedication...... iv

Acknowledgments ...... v

Vita ...... viii

List of Tables...... xii

List of Figures ...... xv

Chapters:

1. Introduction………………………………………………………………………1

2. Mechanism of action of Tirapazamine…………………………………….…….15 2.1 Introduction………………………………………………………………….15 2.2 Computational methods……………………………………………………...20 2.3 Results………………………………………………………………………..23 2.3.1 of the triapazamine radical anion…………………...23 2.3.2 Hydrogen atom transfer reactions of TO·⎯, TOH·, TOH·’ with Amino Ribose……………………………………………………28 2.3.3 N-O bond cleavage reactions…………………………………….32 2.3.4 Oxygen transfer from tirapazamine to the desoxyribose ring…....34 2.4 Conclusions…………………………………………………………………..36 2.5 References for Chapter 2…………………………………………………….46

3. Analogs of tirapazamine…………………………………………………………48 3.1 Introduction…………………………………………………………………..48 3.2 Computational methods……………………………………………………...49

ix

3.3 Results………………………………………………………………………..49 3.3.1 Effects of phenyl ring substitution and benzannulation………….49 3.3.2 Lumiflavin N-oxide……………………………………………...61 3.3.3 The action of T and T’…………………………………………………...62 3.4 Conclusions…………………………………………………………………..70 3.5 References for Chapter 3…………………………………………………….71

4. Radical additions to aromatic N-oxides………………………………………….72 4.1 Introduction…………………………………………………………………..72 4.2 Computational methods……………………………………………………...72 4.3 Results………………………………………………………………………..73 4.3.1 Reactions of with or pyridine…………...73 4.3.2 Reaction of methyl radical with pyridine N-oxide………………74 4.3.3 Oxygen transfer from pyrazine, 1,4-dioxide to methyl radical…..78 4.3.4 Effect of phenyl ring……………………………………………..81 4.3.5 Effect of substitution……………………………………………..82 4.3.6 Silicon centered radicals…………………………………………84 4.4 Conclusions…………………………………………………………………..85 4.5 References for Chapter 4…………………………………………………….86

5. A comparison of acetyl and methoxycarbonylnitrenes…………………………..87 5.1 Introduction………………………………………………………………87 5.2 Computational methods………………………………………………….90 5.3 Results……………………………………………………………………92 5.3.1 Singlet-triplet energy splittings…………………………………..92 5.3.2 Curtius rearrangement……………………………………………98 5.3.3 Intramolecular C-H insertion…………………………………...107 5.3.4 Bimolecular reactions…………………………………………..108 5.4 Conclusions……………………………………………………………..114 5.5 References for Chapter 5……………………………………………….115

6. Sulfonylnitrenes and azides…………………………………………………….118 6.1 Introduction…………………………………………………………….118 6.2 Computational methods………………………………………………...120 6.3 Results…………………………………………………………………..121 6.3.1 Methylsulfonylnitrene singlet-triplet energy separation………..121 6.3.2 Intramolecular C-H insertion reactions of singlet alkylsulfonylnitrenes……………………………………………124 6.3.3 Pseudo-Curtius rearrangements………………………………...126

x

6.3.4 Bimolecular reactions of singlet methylsulfonylnitrene………..129 6.3.5 Bimolecular reactions of triplet methylsulfonylnitrene………...131 6.3.6 Is thermal rearrangement concerted with nitrogen extrusion…..132 6.3.7 Other modes of thermal decomposition of methanesulfonyl azide…………………………………………………………….134 6.3.8 Oxygen, sulfur and nitrogen substituent effects………………..134 6.4 Conclusions……………………………………………………………..142 6.5 References for Chapter 6……………………………………………….143

7. The reaction of triplet nitrenes with oxygen……………………………………146 7.1 Introduction……………………………………………………………..146 7.2 Computational methods………………………………………………...147 7.3 Results…………………………………………………………………..147 7.4 Conclusions……………………………………………………………..157 7.5 References for Chapter 7……………………………………………….157

8. Wolff rearrangement of diazo esters and ketones………………………………160 8.1 Introduction…………………………………………………………………160 8.2 Computational methods…………………………………………………….161 8.3 Results………………………………………………………………………162 8.3.1 Conformational isomerism……………………………………...162 8.3.2 Rearrangements of diazoketones and diazoesters………………164 8.3.3 Concerted 1,2 hydrogen migration and nitrogen extrusion rearrangement of 1-diazopropane………………………………173 8.3.4 Discussion………………………………………………………176 8.4 Conclusions…………………………………………………………………182 8.5 References for Chapter 8…………………………………………………...183

9. Miscellaneous…………………………………………………………………..186 9.1 Singlet and triplet 2,6 – difluorophenylnitrene…………………………186 9.2 Phthalic acids and their complexes with iron……….………………….194 9.3 Naphthyl and biphenyl azides………………………………………….207 9.4 References for Chapter 9……………………………………………….216

Numerals Bibliography……………………………………………………………………………218

xi

LIST OF TABLES

Table Page

2.1 The effect of hydration on the O-H bond cleavage in H2O and the N-O bond cleavage in NH2OH (ΔH (kcal/mol)) at the B3LYP/6-311+G**//B3LYP/6- 31G* and CBS-QB3 levels of theory...... …… . . . . .22

2.2 The free energy of reaction (298K, kcal/mol) of tirapazamine radical anion (TO·⎯) with ion in both the gas phase and the dielectric continuum of water at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 25

+ + 2.3 The free energy of reaction (298K, kcal/mol) of tirapazamine with H3O or NH4 at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 27

2.4 Bond dissociation energies of (R)-2-amino-(S)-4-hydroxy-(R)-5-(hydroxymethyl) tetrahydrofuran at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory... . .28

2.5 Reactions of TO·⎯, TOH· and TOH·’ with an amino-substituted ribose ring at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory ...... …… . . . . .32

3.1 Free energies (kcal/mol, 298K) of protonation of monocyclic di-N-oxides by ammonium ion in aqueous phase at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 51

3.2 Free energies (kcal/mol, 298K) of protonation of benzannulated di-N-oxides by ammonium ion in aqueous phase at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 51

3.3 A comparison of the free energy (298K, kcal/mol) of protonation on the oxygen atoms of the N=N-O and C=N-O units at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 54

3.4 Free energies (298K, kcal/mol) for homolysis reactions of monocyclic di-N- oxides in the gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 57

xii

3.5 Free energies (298K, kcal/mol) for homolysis reactions of benzannulated di-N- oxides in the gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 58

3.6 Free energies (298K, kcal/mol) for homolysis reactions of benzannulated di-N- oxides in gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... …… . . . . .59

3.7 Free energies (298K, kcal/mol) for homolysis reactions of benzannulated di-N- oxides in gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory ...... …… . . . . .60

5.1 Singlet-Triplet energy gap [∆H (298K) (kcal/mol)] of species 9, 10 and 11 by DFT, CCSD(T) and CBS-QB3 methods...... 93

6.1 Decomposition modes of sulfonyl azides. (As calculated at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory. Energies are in kcal/mol and are enthalpies and free energies at 298 K.)...... 141

7.1 DFT, CASPT2 and CBS-QB3 calculated energy separations ΔH (298K) (kcal/mol). The CBS-QB3 numbers are in parentheses, and the B3LYP/6- 311+G*//B3LYP/6-31G* values are in brackets...... 148

‡ 7.2 Activation barriers (ΔH ) for the direct reaction of O2 with various nitrenes . . 153

7.3 Charge on the O2 unit in the transition state for O2 reaction with various nitrenes at the B3LYP level...... 155

8.1 Singlet-triplet energy gap [ΔG(298K) (kcal/mol)] for R-C-H at the B3LYP/6- 311+G**//B3LYP/6-31G* and CBS-QB3 levels of theory...... 164

8.2 The relative energies [ΔG(298K) (kcal/mol)] for the gas-phase stepwise and concerted Wolff Rearrangement of RCHN2 at the B3LYP/6-311+G**//B3LYP/6- 31G* level of theory...... 177

9.1 values and relative energies calculated in various media with different levels of theory...... 190

xiii 9.2 TD-DFT calculated absorption maxima (λ, nm) and oscillator strengths (f) for 79a, 79b, 79c, 80a, 80b, 80c, at B3LYP/6-31+G** level of theoy ...... 208

xiv LIST OF FIGURES

Figure Page

1.1 Illustration of the principal differences between the vasculature of normal and tumor tissues...... 2

1.2 of DNA ...... 3

1.3 Mechanism of action of an hypoxia selective drug ...... 4

1.4 Hypoxia activated prodrugs ...... 4

1.5 The structure of tirapazamine...... 5

2.1 The relative free energy change ΔG (298K, kcal/mol) for the homolytic cleavage of TOH· at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 46

2.2 The relative free energy change ΔG (298K, kcal/mol) for the homolytic cleavage of TOH·’ at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 33

2.3 The change of enthalpy [ΔH (298K, kcal/mol)] for oxygen transfer reactions from tirapazamine to CH3OCH2 radical in gas phase, aqueous phase and in at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 35

2.4 Change of enthalpy [ΔH (298K, kcal/mol)] for oxygen transfer reactions between tirapazamine and a DNA sugar radical at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 36

2.5 The mechanism proposed by Denny and co-workers ...... 36

2.6 The relative free energy change ΔG (kcal/mol) of the stepwise reactions to form benzotriazinyl radical at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 38

xv 2.7 The relative free energy change ΔG (kcal/mol) of two competitive reactions of hydroxyl radical at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 40

2.8 One-Electron Reduction of Benzotriazine 1-Oxides by -Hydroxyalkyl Radicals as proposed by Denny and co-workers...... 41

2.9 The relative free energy change ΔG (298K, kcal/mol) of the reaction between the benzotriazinyl radical and a model DNA sugar moiety at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 42

2.10 The relative free energy change ΔG (298K, kcal/mol) of the reaction between the benzotriazinyl radical and a model DNA sugar moiety at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 43

2.11 The structure of a DNA sugar moiety and as part of a DNA chain...... 44

3.1 The structures of riboflavin, riboflavin N-oxide, and lumiflavin N-oxide...... 61

3.2 Free energies [ΔG (kcal/mol, 298K)] of reaction for lumiflavin N-oxide in the gas phase (top) and in the aqueous phase (bottom), as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 62

3.3 Free energies [ΔG (kcal/mol, 298K)] of reactions for desoxytirapazamine T in the gas phase (top) and in the aqueous phase (bottom), as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 63

3.4 Free energies [ΔG (kcal/mol, 298K)] of reactions for desoxytirapazamine T in the gas phase (top) and in the aqueous phase (bottom), as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 65

3.5 The change of enthalpies (298K, kcal/mol) of reactions of tirapazamine and CH3OCH2 radical at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 66

3.6 The change of enthalpies (298K, kcal/mol) of reactions of desoxytirapazamine T and CH3OCH2 radical at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 68

xvi 3.7 The change of enthalpies (298K, kcal/mol) of reactions of desoxytirapazamine T’ and CH3OCH2 radical at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 69

4.1 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to benzene at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 73

4.2 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to all possible positions of pyridine at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 74

4.3 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to all possible positions of pyridine N-oxide at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 75

4.4 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to pyridine N-oxide and the oxygen transfer reaction at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 76

4.5 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of the C1-radical derived from an amino-substituted deoxyribose model to pyridine N-oxide and the oxygen transfer reaction at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 78

4.6 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to pyrazine, di-N-oxide and the oxygen transfer reaction at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 80

4.7 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to possible positions of 1,4-di-N-oxide-quinoxaline and phenazine N,N'-dioxide at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 82

4.8 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to possible positions of 2,4,6-trimethylpyridine-N-oxide and 2,4,6- trichloropyridine-N-oxide at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory ...... 83

xvii 4.9 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to all possible positions of tetrachloropyrazine di-N-oxide and tetramethylpyrazine di-N-oxide at the B3LYP/6-311+G**//B3LYP/6- 31G* level of theory...... 83

4.10 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of trimethylsilyl radical to possible positions of pyridine-N-oxide at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 84

4.11 The change in enthalpy [ΔH (kcal/mol, 298K)] for the fragmentation reaction of the N-O bond in the silyl adduct at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 85

5.1 The geometries of 9-11 in their singlet and triplet states. Distances are shown in Å, and angles are in degrees...... 95

5.2 The energy surface [ΔG(298K) (kcal/mol)] for Curtius rearrangement of 9 (top) and 10 (bottom) at the B3LYP/6-311+G**//B3LYP/6-31G* level in acetonitrile, cyclohexane, and in the gas phase as well as CCSD(T)/aug- cc-pVDZ//B3LYP/6-31G* and CBS-QB3 levels. Distances are shown in Å, and angles are in degrees...... 99

5.3 The energy surface [ΔG(298K) (kcal/mol)] and isodesmic reactions for Curtius rearrangement of 13 and an isomeric analog 12 at the B3LYP/6- 311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees...... 102

5.4 The energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Curtius rearrangement of (a) acetyl azide and (b) methoxycarbonyl azide at the B3LYP/6-311+G**//B3LYP/6-31G* (red, top) and CBS-QB3 (black, bottom) levels of theory. Distances are shown in Å, and angles are in degrees...... 105

5.5 The energy surface [ΔG (298K) (kcal/mol)] for the reaction of singlet 9 with propane at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS- QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees...... 108

5.6 The energy surface [ΔG (298K) (kcal/mol)] for the reaction of singlet 9 with ethylene at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS- xviii QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees...... 109

5.7 The energy surface [ΔG (298K) (kcal/mol)] for the reaction of singlet 1 and 2 with methanol at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees...... 109

5.8 A plot of the log kobs for the decay of benzoylnitrene in versus 1/T (in K)...... 112

5.9 A plot of the log of the absolute rate constant for reaction of benzoylnitrene with 1-hexene in CF2ClCFCl2 versus 1/T (in K)...... 113

6.1 The CBS-QB3 optimized geometries of singlet (34S) and triplet (34T) methylsulfonylnitrene. Bond distances are in Angstroms, and bond angles are in degrees...... 123

6.2 The energy surface (ΔG, kcal/mol) for intramolecular cyclization of 1- butylsulfonylnitrene at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom, underlined) levels in the gas phase. Enthalpies of activation are shown in parentheses...... 125

6.3 The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of methylsulfonylnitrene at the B3LYP/6-311+G**//B3LYP/6-31G* level in the gas phase...... 128

6.4 The energy surface (ΔG, kcal/mol) for bimolecular reactions of singlet methylsulfonylnitrene (1S) at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom, underlined) levels in the gas phase. . . . . 130

6.5 The energy surface [ΔG, kcal/mol (top) and ΔH, kcal/mol (bottom)] for stepwise versus concerted decomposition of methanesulfonyl azide (39) at the B3LYP/6-311+G**//B3LYP/6-31G* level in the gas phase...... 133

6.6 The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of methoxysulfonylnitrene (40) at the B3LYP/6-311+G**//B3LYP/6-31G* level in the gas phase...... 136

xix 6.7 The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of thiomethoxysulfonylnitrene (43) at the B3LYP/6-311+G**//B3LYP/6- 31G* level in the gas phase...... 137

6.8 The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of N,N-dimethylaminosulfonylnitrene (46) at the B3LYP/6- 311+G**//B3LYP/6-31G* level in the gas phase...... 138

6.9 The energy surface [ΔG, kcal/mol (top) and ΔH, kcal/mol (bottom)] for stepwise versus concerted decomposition of methoxysulfonyl azide (49) at the B3LYP/6-311+G**//B3LYP/6-31G* levels in the gas phase...... 152

7.1 The reactions of nitrenes with oxygen, ΔG (298K) (kcal/mol), calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level in acetonitrile (PCM)...... 152

7.2 The rearrangement of an alkoxynitrene oxide to a nitrate, ΔG (298K) (kcal/mol), calculated at the CBS-QB3 (top) and B3LYP/6- 311+G**//B3LYP/6-31G* (bottom) level...... 156

8.1 The energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Wolff Rearrangement of diazoacetone (54) in the gas phase at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (middle, in parentheses) levels of theory, and at the PCM level for acetonitrile (PCM) at the single-point B3LYP/6-311+G**(PCM)//B3LYP/6-31G*(gas) level of theory (bottom, in brackets)...... 166

8.2 The energy surface [ΔG(298K) (kcal/mol)] in the gas phase at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom, in parentheses) levels of theory...... 168

8.3 The energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Wolff Rearrangement of methyl diazoacetate (57) at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory in gas phase (top) and in acetonitrile (PCM, bottom)...... 170

8.4 The gas-phase energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Wolff rearrangement of methoxydiazoacetone (60) and ethyl diazoacetate (61) at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory. The numbers in parentheses are relative free energies between xx vertical isomeric pairs in the Figure 8.of isodesmic reactions. These refer to pairs of compounds displayed vertically, in which diazo compounds are compared with their diazo isomers, and isomeric transition states and isomeric carbenes are compared...... 172

8.5 The gas-phase energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted migration and nitrogen extrusion of 1-diazopropane (66) at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory...... 175

8.6 Isodesmic reactions [ΔG(298K) (kcal/mol)] at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory...... 177

9.1 The comparison of IR frequencies calculated by B3LYP and CASSCF...... 191

9.2 The IR spectra of (a) (K), (b) triplet nitrene (3N), and (c) singlet nitrene (1N), calculated at the B3LYP/6-31G* level of theory using the PCM model for acetonitrile. An open shell (UB3LYP) description was used for the singlet and triplet nitrene...... 192

9.3 The ATR-FTIR spectrum for aqueous phthalic acid at different pH value...... 196

9.4 Comparison of ATR-FTIR spectrum of phthalic acid at pH 6.5 (top) with dianionic form of phthalic acid in the gas phase (middle) and the aqueous phase (bottom)...... 198

9.5 Comparison of ATR-FTIR spectrum of phthalic acid at pH 4.5 (top) with mono-anionic form of phthalic acid in the gas phase (middle) and the aqueous phase (bottom)...... 199

9.6 Comparison of ATR-FTIR spectrum of phthalic acid at pH 4.5 (top) with neutral phthalic acid in the gas phase (middle) and the aqueous phase (bottom)...... 201

9.7 The ATR-FTIR spectrum for phthalic acid adsorbed on hematiteat different pH value...... 202

9.8 Second derivative of spectrum of adsorbed phthalic acid...... 203

xxi 9.9 The structure of inner-sphere complex model and outer-sphere complex model between phthalic acid and Fe...... 204

9.10 Comparison of ATR-FTIR spectrum of phthalic acid absorbed on the hematite at pH 3.47 (top) with outer-sphere complex model (middle) and inner-sphere complex model (bottom)...... 205

9.11 The orbitals involved in vertical transitions of 1-naphthyl azide 79a. The

relative orbital energies are in kcal/mol...... 210

9.12 The orbitals involved in vertical transition of 2-naphthyl azide 79b. The relative orbital energies are in kcal/mol ...... 211

9.13 The orbitals involved in vertical transition of 79c. The relative orbital energies are in kcal/mol ...... 212

9.14 The orbitals involved in vertical transition of ortho-biphenyl azide 80a. The relative orbital energies are in kcal/mol...... 213

9.14 The orbitals involved in vertical transition of para-biphenyl azide 80b. The relative orbital energies are in kcal/mol...... 214

9.14 The orbitals involved in vertical transition of biphenyl 80c. The relative orbital energies are in kcal/mol...... 215

.

xxii

CHAPTER 1

INTRODUCTION

This dissertation includes two major parts: the first part is a study of the action of

the anti-tumor drug tirapazamine and its analogs in chapters 1-4. The second section

concerns studies of reactions of nitrenes and carbenes described in chapters 5-8.

1.1 Tirapazamine

In cancer chemotherapy, there is a long standing problem: serious side effects often occur upon treatment due to collateral damage to healthy cells. When cancer cells are killed, many normal cells are also damaged by the administered drug. How can medical science selectively damage cancer cells without simulataneously damaging normal cells? Unfortunately, at the biochemical level, cancer cells closely resemble normal cells. However, there is one characteristic feature of tumor cells: hypoxia. In normal cells, the blood vessels are well organized, which provides normal cells sufficient oxygen, as illustrated in Figure 1.1 (a). However, because of the rapid growth of a solid tumor, the solid tumor contains significant regions where the blood vessels are chaotic, as shown in Figure 1.1 (b). These unorganized vessels poorly oxygenate tumor cells which

1 result in a low oxygen supply in some region of the solid tumor.1 This low oxygen

condition is called hypoxia.

(a) Normal tissue (b) tumor tissue

Figure 1.1 Illustration of the principal differences between the vasculature of normal and tumor tissues.2

In radiotherapy, after a DNA radical has been generated, there are two competitive reactions.3 When the oxygen concentration is high, ,

·- such as radical anion (O2 ), are formed, and the DNA strands break as a result

of oxidative chemistry.4 This collective damage induces cell death. However, under

hypoxic condition, the DNA radical can be reductively repaired by thiols.5 This process

is illustrated in Figure 1.2. Therefore, hypoxia is not a desirable feature for the selective

killing of a rapidly growing tumor in the presence of healthy cells during radiotherapy.

2

DNA-H Radiation RSH

DNA-H DNA

O2 DNA-OO•

DNA break Cell death

Figure 1.2. of DNA.

An anti-tumor agent may selectively assist damage to DNA under hypoxic conditions.6 In principle, a non-toxic agent may be selectively activated under hypoxic conditions and subsequently may selectively damage DNA. One advantage of this kind of drug is that in the presence of a large oxygen concentration, the activated drug would be neutralized and thus will not damage the normal cells. The mechanism is illustrated in

Figure 1.3.

3 In Cancer cell ( hypoxic cell)

Enzymatic +e- D D Cancer Cell DNA Damage Pro-Drug Activated Drug O2 O2

In Normal cell ( Oxic cell)

Figure 1.3. Mechanism of action of a hypoxia selective drug

There are several classes of hypoxia selective anti-tumor agents, including

quinones7, nitro-aromatics8, aliphatic N-oxides9, and aromatic N-oxides,9 as shown in

Figure 1.4. Tirapazamine is an aromatic N-oxide which is currently in phase III clinical trials. Several other hypoxia-activated prodrugs, including AQ4N, NLCQ-1 and dinitrobenzamide mustards, are in preclinical or early clinical development.

O N O OH O HN CH2OCONH2 O N H2N OMe Br N NO2 N N N Me N NMe H N NH2 OH O HN O OH N O O quinones nitroaromatics aliphatic N-oxides aromatic N-oxides

Figure 1.4. Hypoxia activated prodrugs.

The structure of tirapazamine (TO) is shown in Figure 1.5.

4

O N N

N NH2 O

Figure 1.5. The structure of tirapazamine.

This aromatic di-N-oxide was discovered by Brown et al. in 1985.10 Since then,

the medicinal properties of tirapazamine have been well established, although the

mechanism of action of tirapazamine still remains unclear. The purpose of this research

was to use computational methods to understand the mechanism of action of tirapazamine

and use this information to design a second generation of drugs, with potentially

enhanced properties.

It is generally believed that tirapazamine is activated enzymatically to form a

radical anion (TO•–) which upon protonation can form two possible neutral, isomeric radicals TOH• and TOH•’.11 It is hypothesized that these reactive intermediates (TO·⎯,

TOH·, TOH·’) can be consumed by reaction with oxygen faster than reactions which damage DNA, hence their greater toxicity in the absence of oxygen.12 These intermediates may damage DNA by two distinct mechanisms. It is possible that TO•–,

TOH•, and or TOH•’ may react directly with sugar moieties of DNA by hydrogen-atom abstraction. These reactions would ultimately induce DNA strand breaks.13 In a second

pathway, TOH• and TOH•’ radicals may fragment to form a desoxy-tirapazamine (T or

5 T’) and hydroxyl radical, the latter being a potent intermediate which is known to damage

DNA. These two mechanisms of action of tirapazamine are illustrated in Scheme 1.1.

6

Scheme 1.1

O O N - N N +e N

N NH2 N NH2 O O TO TO OH + +H N N or O OH N NH2 N N sugar (SH) OH N or N S + N NH N NH 2 2 OH OH O O N H N TOH TOH' N or N N NH NH2 2 N H O OH

O N N N N N NH2 N NH2 O T T'

+ +

HO HO

7

To better appreciate these mechanisms, we will first attempt to discover the

possible source of protonation. The pKa of the radical anion of tirapazamine is 6.3.14 Our calculations predicted that the reactions of tirapazamine with either water or the DNA bases are endothermic in an aqueous solvent continuum. The calculations indicate that an alternative source of a proton can be provided by an ammonium ion, which is often a counter ion of cellular DNA. Then we have computed the changes in the free energies

and the reaction barriers of these reactions of tirapazamine in the gas phase and in a

solvent dielectric field.

We will then predict that hydrogen-atom transfer reactions of TO•–, TOH•, and

TOH•’ with a model system of an amino-substituted ribose ring are too endoergic to be

significant, but that fragmentation reactions of TO•–, TOH•, and TOH•’ are energetically feasible. According to this mechanism, the radical anion TO•– must be protonated to form a species capable of generating hydroxyl radical.

Hydroxyl radical reacts with a DNA sugar moiety to form a DNA-centered

radical. Under aerobic conditions, the deoxyribose radicals may react with molecular oxygen and the resulting peroxy radicals ultimately decompose to induce DNA strand breaks.3 In the absence of molecular oxygen, DNA radicals generated by radiation can be

chemically repaired via reaction with cellular thiols.4

Tirapazamine is more toxic in the absence than in the presence of oxygen.

Presumably the radical anion of tirapazamine is scavenged by oxygen, thus the hydroxyl

radical is produced more efficiently under hypoxic conditions. Nevertheless, the DNA

8 strand cleavage reactions require a source of oxygen.15 Therefore, the optimal condition

required for drug action (hypoxia) seems to be in direct opposition to the conditions

required for a radical-based agent to effectively promote DNA strand cleavage.

This seeming paradox was resolved by Daniels and Gates who proposed a dual

role of tirapazamine in DNA-cleavage processes.11 The drug is proposed to not only produce hydroxyl radical, but, because it is an N-oxide , it will serve to transfer the oxygen to the radical center, and thereby oxygenate DNA radicals. The resulting

alkoxy sugar radicals could then induce DNA strand breaks. In this way, tirapazamine

may serve as a substitute for molecular oxygen, as shown in Scheme 1.2.

9

Scheme 1.2

CH2 CH O B radiation 2 O B H (HO )

O N CH N 2 O CH2 + B O B N O N NH O N 2 N O H2N

O CH2 O CH2 CH2 N O O O B N O -B + O N NH H H 2 DNA Strand Break

10

We calculated the oxidization of the C1’ radical of the deoxyribose ring with

tirapazamine, and the results demonstrate that oxygen transfer from tirapazamine to a

DNA radical is thermodynamically feasible. Significantly, mono-N-oxide derivatives T

and T’ are less active than the di-N-oxide tirapazamine. The reasons for their low

activities will be proposed.

Studies of the mechanism of action of tirapazamine may lead to the design of a

better drug. We will also discuss the role of the phenyl ring and substituents around the

tirapazamine ring in providing direction to the synthesis and an evaluation of analogs.

The addition of radicals to different positions of an aromatic N-oxide will be discussed as

well.

Another proposed mechanism of action of tirapazamine has been reported by

Denny and co-workers.16 In this mechanism, the radical generated from tirapazamine that

damages DNA is not the hydroxyl radical, but is the benzotriazinyl radical. The DNA

oxidation mechanism is also different from that presented in the preceeding section.

The mechanism of drug action of tirapzamine and its analogs will be discussed in

Chapters 2-4.

1.2 Reactions and rearrangement of nitrenes and carbenes

Both nitrenes and carbenes are important reactive intermediates, with two bonds

less than most stable compounds of carbon and nitrogen, respectively. Because of their

11 similarity, many of our studies of the reactions of nitrenes are compared to the corresponding reactions of carbenes.

Upon activation with heat or light, both acyl and alkoxycarbonyl azides extrude

molecular nitrogen and form carbonyl nitrenes. These nitrenes can undergo

intramolecular rearrangement (Curtius Rearrangement) to form isocyanates as shown

below.17

O O O Δ, hν C

R XN3 R XN RX N

Δ, hν

RX = or aryl or RX = O-alkyl or O-aryl

There is evidence that migration of RX can proceed in concert with nitrogen

extrusion in some cases, without the intervention of a free nitrene. There is also evidence

that the stepwise or concerted nature of the process varies with the mode of azide

decomposition.18 In this study, computational methods have been applied to understand

the differences between the Curtius rearrangements of acetyl and methoxycarbonyl

nitrenes and azides, as well as their reactions with solvents. Related investigations have

been performed for sulfonyl nitrenes and azides. Chapters 5 and 6 provide the details of

these studies.

Thermolysis of diazoketones and esters generally leads to via Wolff

Rearrangement (WR). In reactive solvents, it is occasionally possible to intercept

acylcarbene intermediates. This has led to the realization that WR can proceed either in

12 concert with nitrogen extrusion or stepwise via carbene intermediates.19 The concerted

and stepwise reactions of Wolff rearrangement of diazo esters and ketones have been

studied using computational methods. We have also considered Wolff-like

rearrangements of alkyl diazomethanes.

Triplet aromatic carbenes react with oxygen at rates that approach diffusion

control.20 Triplet aromatic nitrenes react with oxygen much more slowly than the

analogous carbene processes.21 To investigate the origin of the different reactivity of

triplet carbenes and nitrenes with oxygen, Density Functional Theoretical (DFT)22 and ab initio molecular orbital calculations23 were performed.

1.3 References for Chapter 1

1. Vaupel, P.; Kallinowski, F.; Okunieff, P. Cancer Res. 1989, 49, 6449.

2. Brown, J. M.; Wilson, W. R. Nature Rev. 2004, 4, 437.

3. Greenberg, M. M. Chem. Res. Toxicol. 1998, 11, 1235.

4. Breen, A. P.; Murphy, J. A. Free Rad. Biol. Med. 1995, 18, 1033.

5. Bump, E. A.; Brown, J. M. Pharmac. Ther. 1990, 47, 117.

6. Brown, J. M. Cancer Res. 1999, 59, 5863.

7. Tomasz, M. Chem. Biol. 1995, 2, 575.

8. Stratford, I. J.; Workman, P. Anticancer Drug Des., 1998, 13, 519.

9. Denny, W. A. Eur. J. Med. Chem., 2001, 36, 577.

10. Brown. J. M. Br. J. Cancer. 1993, 67, 1163.

11. Daniels, J. S.; Gates, K.S. J. Am. Chem. Soc. 1996, 118, 3380.

13

12. Elwell, J.H.; Siim, B.G.; Evans, J.W.; Brown, J.M. Biochem. Pharmacl. 1997, 54, 249.

13. Brown, J. M.; Wang, L.-H. Anti-Cancer Drug Design 1998, 13, 529.

14. Wardman, P.; Priyadarsini, K. I.; Dennis, M. F.; Everett, S. A.; Naylor, M. A.; Patel, K. B.; Stratford, I. J.; Stratford, M. R. L.; Tracy, M. British Journal of Cancer, Supplement. 1996, 74, S70.

15. Laderoute, K.L.; Wardman, P.; Rauth, M. Biochem. Pharmacl. 1988, 37, 1487

16. Shinde, S. S.; Anderson, R. F.; Hay, M. P.; Gamage, S. A.; Denny, W. A.; J. Am. Chem. Soc.; 2004, 126, 7865.

17. (a) Wallis, E.S. Org. Reactions. 1946, 3, 267. (b) Bauer, E. Angew. Chem. Int. Ed. Engl. 1974, 13, 376.

18. (a) Lwowski, W.; Tisue, G. T. J. Am. Chem. Soc. 1965, 82, 4022. (b) Tisue, G. T.; Linke, S.; Lwowski, W. J. Am. Chem. Soc. 1967, 89, 6303. (c) Linke, S.; Tisue, G. T.; Lwowski, W. J. Am. Chem. Soc. 1967, 89, 6308.

19. Kirmse, W. Eur. J. Org. Chem. 2002, 14, 2193.

20. Bucher, G.; Scaiano, J. C.; Platz, M. S. Kinetics of Carbene Reactions in Solution. Landolt-Bornstein, Group II, Volume 18, Subvolume E2 Springer, Berlin, Germany, (1998), p.141.

21. (a) Gritsan, N. P.; Pritchina, E. S. J. Inf. Record. Mater. 1989, 17), 391 (b) Liang, T.-Y., Schuster, G. B. J. Am. Chem. Soc. 1987, 109, 7803 (c) Pritchina, E. A.; Gritsan, N. P. J. Photochem. Photobiol., A: Chemistry 1988, 43, 165. (d) Gritsan, N. P.; Pritchina, E. A. Russ. Chem. Rev. 1992, 61, 910.

22. Ziegler, T. Chem. Rev. 1991, 91, 651.

23. Jensen, F. Introduction to Computational Chemistry; Wiley: Chichester, 1998.

14

CHAPTER 2

MECHANISM OF ACTION OF TIRAPAZAMINE

2.1. Introduction

Tirapazamine (TO) is currently undergoing phase III clinical trials for the

treatment of certain oxygen-deficient (hypoxic) solid tumors.1,2 The selectivity of the drug results from its greater toxicity in an oxygen-deficient environment. Various mechanisms of action of tirapazamine (TO) have been proposed (Scheme 2.1).3 The drug

is believed to be activated enzymatically to form a radical anion (TO·⎯) which upon protonation forms neutral, isomeric radicals TOH· and TOH·’.4 It is hypothesized that

these reactive intermediates (TO·⎯, TOH·, TOH·’) can be consumed by reaction with oxygen faster than reactions which damage DNA, hence their greater toxicity in the absence of oxygen.5

It has been proposed that TO·⎯, TOH·, and or TOH·’ react directly with sugar moieties of DNA by hydrogen-atom abstraction.6 These reactions would produce sugar-

centered radicals which are known to lead to single strand breaks of DNA. Alternatively, it has been proposed that TOH· and TOH·’ radicals may fragment to form

15 desoxytirapazamine (T) and hydroxyl radicals which can subsequently damage nucleic

acids.4,5

To better appreciate these mechanisms, we have computed the energies of TO·⎯,

TOH·, TOH·’ in the gas phase and in a solvent dielectric field representative of water, along with enthalpies and enthalpies of activation of the reactions of Scheme 2.1. The calculations predict that the reactions of TO·⎯, TOH·, and TOH·’ with an amino- substituted ribose are too endothermic to be important and that TOH· and TOH·’ will instead both fragment rapidly to form hydroxyl radical.

16

Scheme 2.1

O O N - N N +e N

N NH2 N NH2 O O TO TO OH + +H N N or O OH N NH2 N N sugar (SH) OH N or N S + N NH N NH 2 2 OH OH O O N H N TOH TOH' N or N N NH NH2 2 N H O OH

O N N N N N NH2 N NH2 O T T'

+ +

HO HO

17

After hydroxyl radical is generated, it can react with a DNA sugar moiety to form

a DNA-centered radical. Under aerobic conditions, the deoxyribose radicals may react

with molecular oxygen and the resulting peroxyl radicals will ultimately decompose to

induce DNA strand breaks.5 In the absence of molecular oxygen, DNA-centered radicals

generated by radiation can be chemically repaired via reaction with cellular thiols.7

Tirapazamine is more toxic in the absence than in the presence of oxygen.

Presumably the radical anion of tirapazamine is scavenged by oxygen, thus hydroxyl radical is produced more efficiently under hypoxic conditions. Nevertheless, the DNA strand cleavage reactions require a source of oxygen. Therefore, the optimal condition required for drug action (hypoxia) seems to be in direct opposition to the conditions required for a radical-based agent to effectively promote DNA strand cleavage.

This seeming paradox was resolved by Daniels and Gates who proposed a dual role of tirapazamine in DNA cleavage processes.4 The drug is proposed to not only produce hydroxyl radical, but, since tirapazamine has an N-oxide , it may

oxygenate DNA radicals, and the resulting radicals could induce DNA strand breaks. In

this way, tirapazamine may serve as a substitute for molecular oxygen, as shown in

Scheme 2.2.

18

Scheme 2.2

CH2 O CH2 B radiation O BH2 H (HO )

O N CH N 2 O CH2 + B O B N O N NH O N 2 N O O O H2N

O CH2 O CH2 CH2 N O O O B N O -B + O O H N NH2 O H O DNA Strand Break

19 Due to the large size of tirapazamine and the DNA sugar moiety, we initially

tested this proposal using small model systems. We utilized methoxymethyl radicals to

model the deoxyribose ring. Bolstered by these results, we calculated the oxidation of the

C1’ position of the deoxyribose ring with tirapazamine, and the results demonstrate that

oxygen transfer from tirapazamine to a DNA radical is thermodynamically feasible.

2.2 Computational Methods

Density functional theory (DFT)8 methods have been successfully applied to the study of aromatic radicals.9 Many theoretical procedures, such as unrestricted Hartree-

Fock theory, fail for aryl radicals due to high spin contamination. However, DFT

methods avoid this issue, and the unrestricted B3LYP method, in particular, has been

well calibrated to provide qualitative and quantitative agreement with experiment for a

wide variety of thermochemical and kinetic issues for aryl radicals.10

All geometry optimizations, frequency calculations and single-point energy

calculations were performed using Gaussian 9811 at the Ohio Supercomputer Center. The geometries were optimized at the B3LYP/6-31G* level of theory. Vibrational frequencies were calculated for each stationary point to verify each to be a minimum energy structure and to provide zero-point vibrational energies, which were scaled by the factor of

0.9806.12 Single-point energy calculations of these species at the B3LYP/6-311+G**

level were performed using the corresponding B3LYP/6-31G* geometries. The zero-

point vibrational energies, thermal and entropic corrections were obtained from the

20 B3LYP/6-31G* vibrational frequency calculations. Six Cartesian d functions were used

with all basis sets.

All of the experimental studies of tirapazamine were performed in the aqueous

phase. However, many tirapazamine calculations are relevant to the gas phase. To better understand the experimental results, solvation effects must be considered.

A number of programs are available for solvation studies using continuum

dielectric or reaction field methods. Tomasi, Barone and coworkers have developed the

polarizable continuum model (PCM)13 which has demonstrated utility for the study of

reactive intermediates. This method provides a means to examine the thermodynamics

and kinetics of the important steps in the possible mechanisms of reaction of

tirapazamine. However, because of hydrogen bonding between solute and solvent,

continuum dielectric methods may not provide the complete picture. Inclusion of explicit

solvent in the first solvation shell, followed by immersion of the solute-solvent complex in a continuum dielectric field, may provide a satisfactory answer. To evaluate this possibility, we have calculated the effect of of hydration on the N-O bond cleavage in H2NOH; a process that is similar to the N-O bond scission in tirapazamine.

We first calculated the cleavage of water to hydrogen atom and hydroxyl radical.

Upon addition of a second water molecule to the system, we found that the bond

dissociation enthalpy reduced from 114.9 kcal/mol to 113.4 kcal/mol, only a 1.5 kcal/mol

change. Using H2NOH as a model molecule, as a function of the number of water

molecules, there is only a small change in the value of the bond dissociation enthalpy

(BDE). It can be seen that when the first water molecule is added, the BDE becomes

21 larger. However, when the second water molecule is added, the BDE decreases. All of

these changes, however, are smaller than 3 kcal/mol.

All calculations of the gas-phase bond dissociation energies were performed with

the B3LYP/6-311+G**//B3LYP/6-31G* and CBS-QB314 methods. CBS-QB3 is

considered to be one of the most accurate calculations available, although it is rather time

consuming. Our calculations (Table 2.1) show that the differences between these two

methods are no more than 4.5 kcal/mol when corrections with the polarizable continuum

model (PCM) for solvation are applied. Therefore, the B3LYP/6-311+G**//B3LYP/6-

31G* level should be a reliable method for these reactions. Thus, the N-O BDE of the

neutral tirapazamine radicals TOH· and TOH·’ were calculated using Density Functional

Theory.

Reaction Gas phase Aqueous phase by PCM B3LYP CBS-QB3 B3LYP CBS-QB3 H2O Æ H + OH 114.9 119.1 119.1 120.0 (H2O)2 Æ (H2O)OH + H 113.4 120.4 122.4 123.4

H2NOH Æ H2N + OH 59.1 65.0 57.4 61.9 (H2NOH)(H2O) Æ H2N(H2O) + OH 61.2 69.5 59.0 63.0 (H2NOH)(H2O)2 ÆH2N(H2O)2 + OH 58.2 65.4 56.0 59.9

Table 2.1. The effect of water hydration on the O-H bond cleavage in H2O and the N-O bond cleavage in NH2OH (ΔH (kcal/mol)) at the B3LYP/6-311+G**//B3LYP/6-31G* and CBS-QB3 levels of theory.

22 2.3 Results

2.3.1 Protonation of the Tirapazamine Radical Anion

Tirapazamine requires enzymatic, one-electron reduction process to generate the

tirapzamine radical anion. The radical anion must be protonated to form a species capable

of generating hydroxyl radical. It is possible that the proton originates from a DNA base,

A, G, C, or T. But calculations predict that the reactions of tirapazamine with bases of

DNA are endothermic. These results can be questioned because only a single base and

not a base in a DNA environment is considered. However, the bases are not likely to be the source of the proton because in the DNA double helix structure, the base is inside the duplex, while the tirapazamine radical anion is outside. The possibility of the reaction of a DNA base with tirapazamine is very low.

An alternative possible source of proton is an ammonium ion. In a cell, there are a

great number of primary and secondary ammonium ions coordinated to DNA. When a

tirapazamine radical anion molecule is formed in a cell and approaches the DNA, there is

a likelihood that the tirapazamine radical anion will react with ammonium ion. We

calculated the reaction of tirapazamine radical anion with ammonium ion in the gas

phase, with one water molecule in the first solvation shell and in a continuum model for

water using the PCM approximation, and we found that all of these reactions are

endothermic

In the gas phase, the proton transfer reaction between tirapzamine radical anion and ammonium ion is very exothermic, –120.3 kcal/mol, because charge is destroyed in

23 the gas phase. The calculations in the presence of the dielectric continuum of water give us a reasonable result, specifically –3.8 kcal/mol.

24

Reaction ΔG (kcal/mol) gas Aqueous O O N N N N +NNH4 + H3 -120.3 -3.8

N NH2 N NH2 O OH

O OH

N N N N +NNH4 + H3 -122.3 2.5 N NH2 N NH2 O O O (H2O) O (H2O) N N N N +NNH4 (H2O) + H3 (H2O) -105.9 -3.6

N NH2 N NH2 O OH

O (H2O) OH N N N N +NNH4 (H2O) + H3 (H2O) -107.8 -0.7 N NH 2 N NH2 O O (H2O)

Table 2.2. The free energy of reaction (298K, kcal/mol) of tirapazamine radical anion (TO·⎯) with ammonium ion in both the gas phase and the dielectric continuum of water at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

25 It is believed that tirapazamine is reduced enzymatically before protonation.4 Is it possible for tirapazamine to abstract a proton before an electron? Further study of

+ tirapazamine suggested that both an ammonium ion and H3O are not sufficiently acidic

to protonate neutral tirapazamine, as shown in Table 2.3. Therefore, in a functioning

cancer cell, sufficiently strong acidic sources are not present in order to protonate

tirapazamine prior to enzymatic reduction to form the radical anion.

26

Reaction ΔG (kcal/mol) gas Aqueous O O N N N N +NNH4 + H3 -16.2 9.4

N NH2 N NH2 O OH O OH N N N N +NNH4 + H3 -12.1 18.2

N NH2 N NH2 O O O O N N N N +HH3O + 2O -55.4 3.5

N NH2 N NH2 O OH O OH N N N N + H3O + H2O -51.3 12.3

N NH2 N NH2 O O

+ Table 2.3. The free energy of reaction (298K, kcal/mol) of tirapazamine with H3O or + NH4 at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory

27 2.3.2 Hydrogen-Atom Transfer Reactions of TO·⎯, TOH·, TOH·’ with Amino-Ribose

We have calculated the gas phase enthalpies of reaction of an amino-substituted

ribose ring with reactive intermediates derived from tirapazamine. There are four

positions, C1’, C3’, C4’, C5’, of the amino-substituted ribose ring which can donate a

hydrogen atom, as shown in Table 2.4.

BDE (kcal/mol)

HO C1’-H 89.1 5 H2C O 1 NH2 H 4 C3’-H 92.8 H 3 2 HO H C4’-H 90.5 C5’-H 92.8

Table 2.4. Bond dissociation energies of (R)-2-amino-(S)-4-hydroxy-(R)-5- (hydroxymethyl) tetrahydrofuran at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

Calculations of the bond dissociation energies reveal that the BDE of C1’-H is the smallest, which suggests that the sugar molecule prefers to lose hydrogen radical from this position. To abstract a hydrogen atom from the C4’ position is ~1.4 kcal/mol more endothermic than C1’ position, while the bond dissociation energies of C3’-H and C5’-H are the same, ~3.7 kcal/mol larger than the C1’-H of the sugar. The results are given in

Table 2.5 in which the products of hydrogen transfer from only C1’ are pictured. The

28 thermodynamics of hydrogen transfer from the other sugar positions are readily calculable using Table 2.4. The accuracy of these calculations will be discussed in a later section.

The differences between these BDEs are smaller than 3.7 kcal/mol, which

suggests that there will be little selectivity in the reaction of a highly reactive radical with

amino ribose. Table 2.5 presents the thermodynamics of the reactions of the neutral

radicals and radical anion derived from tirapazamine with the model sugar ring. We

considered whether the neutral radical or radical anion of tirapazamine directly abstracts

a hydrogen atom from the amino-substituted ribose ring. Rows (3), (4), (5), (6) contain

the two kinds of neutral radicals of tirapazamine, TOH·, and TOH·’, produced upon the

addition of a hydrogen atom to an oxygen, carbon or nitrogen position of tirapazamine.

29 Reactions ΔH (kcal/mol) O HO HO O CH2 N CH2 N O N O N NH2 NH2 H H H H N NH (1) N NH2 2 46.4 O HO O HO H H TO

O O HO HO CH2 N O N CH2 NH N O NH 2 NH2 H H 53.3 H (2) N NH2 N NH2 HO H HO O H O TO O OH HO HO N CH2 N CH2 N O O NH2 N NH2 H H 27.4 H N NH (3) 2 N NH2 HO H HO H OH OH TOH

O O HO HO N CH2 N CH2 N O O NH2 N NH2 (4) H H H 36.4 H N NH 2 N NH2 HO H HO OH OH H TOH

OH HO OH HO CH2 CH2 N O N O (5) N NH2 N NH2 H H H 29.3 N NH2 N NH2 HO H O OH HO H TOH ' OH HO OH HO CH2 CH2 N O NH N O (6) N 2 NH NH2 H H H 32.1 N NH2 N NH2 HO O H HO H O TOH '

Table 2.5. Reactions of TO·⎯, TOH· and TOH·’ with an amino-substituted ribose ring at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

30 The tirapazamine radical anion, produced enzymatically, may be protonated at

multiple sites to form a neutral radical. It is commonly assumed based on product

analysis that the radical anion intermediate of tirapazamine is protonated on the oxygen

of the C=N-O unit.4 However, using the B3LYP/6-311+G**//B3LYP/6-31G* level of

theory, we have calculated the energies of two isomeric, neutral protonated radicals. We found that the TOH·’ radical is ~2.1 kcal/mol more stable than that of the TOH· radical in the gas phase. Upon removal of the phenyl ring, the protonation at the N=N-O site is still preferred. The thermodynamic preference is greater than 3 kcal/mol in the gas phase.

Therefore, if products of homolytic cleavage reactions are derived from protonation of the C=N-O unit, as seems likely from product studies, we predict that kinetic control must dominate over thermodynamic control via a Curtin-Hammett equilibrium. More details will be discussed in Chapter 3.

O OH N N N + TO N + TO

N NH2 N NH2 OH O TOH TOH' ΔH = -2.1 kcal/mol

It is clear that all conceivable hydrogen transfer reactions of TO·⎯, TOH· and

TOH·’ with an amino-substituted ribose are all endothermic by more than 20 kcal/mol in the gas phase. The calculations predict that neither TO·⎯, TOH·, nor TOH·’ are the species responsible for damaging DNA.

31

2.3.3 N-O Bond Cleavage Reactions

As it seems unlikely that TO·⎯, TOH·, or TOH·’ will react directly with a DNA sugar, the fragmentation of the neutral radicals to form a desoxytirapazamine (T or T’)

and hydroxyl radical was considered, as shown in Scheme 2.1.

We calculated the barrier to N-O bond scission of TOH· and TOH·’. Transition

states were located, and in these gas-phase calculations, loose (hydrogen-bonded)

complexes, the initial products formed prior to infinite separation of hydroxyl radical and

the product, were found. Therefore, although protonation of the O of the N=N-O is more

stable than protonation of the O of the C=N-O, in the gas phase, the homolysis to

generate hydroxyl radical is more facile for N-O bond scission from the protonated C=N-

O. In other words, while protonation on the N=N-O unit is modestly favored, there is a

very large kinetic preference for N-O bond scission from the protonated N-O unit.

The N-O bond cleavage reactions of tirapazamine radical TOH and TOH’ are

shown in Figures 2.1 and 2.2, respectively.

32

O O N O N N N N N + OH N NH2 N NH2 OH N NH2 OH

In gas phase 0.0 5.0 -10.4

In aqueous phase 0.0 11.8 -5.9

Figure 2.1. The relative free energy change ΔG (298K, kcal/mol) for the homolytic cleavage of TOH· at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

OH OH N N N N N N + OH N NH N NH N NH 2 2 2 O O O

In gas phase 0.0 10.5 4.0

In aqueous phase 0.0 8.4 0.0

Figure 2.2. The relative free energy change ΔG (298K, kcal/mol) for the homolytic cleavage of TOH·’ at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

33 In the gas phase, the thermodynamically more stable radical TOH·’ cleaves more

slowly (ΔH‡= 10.5 kcal/mol) than does TOH· (5.0 kcal/mol). Hydrogen bonding with

one or two water molecules to H2N-OH had little influence on the N-O bond dissociation

energy. Thus these interactions were not considered on the N-O cleavage of the TOH·

and TOH·’ radicals.

2.3.4 Oxygen transfer from tirapazamine to deoxyribose ring

After the hydroxyl radical generated, it may abstract a hydrogen atom from DNA

and produce a DNA centered radical. The DNA radical needs to be oxidized in order to

induce a DNA strand break. It has been proposed by Gates and co-workers4 that the N- oxide tirapazamine itself could provide the needed oxygen source. Therefore, we studied the reaction of methoxymethyl radical, a model of a DNA radical, with tirapazamine to form the radical addition intermediate and its subsequent fragmentation reaction in the gas phase, water and acetonitrile. We predict that the initial reaction to form the radical- tirapazamine adduct at the N-1 oxygen is exoergic by more than 20 kcal/mol and the subsequent fragmentation reaction is also exothermic by more than 30 kcal/mol in the gas phase, water or acetonitrile (Figure 2.3).

34 O O O N N N 12N N CH3OCH2 + N + CH3OCH2O N NH2 N NH N 2 NH2 O O CH2OCH3

Reactions Gas Water Acetonitrile 1 -23.5 -28.2 -31.3 2 -36.3 -43.0 -39.4

Figure 2.3. The change of enthalpy [ΔH (298K, kcal/mol)] for oxygen transfer reactions from tirapazamine to CH3OCH2 radical in gas phase, aqueous phase and in acetonitrile at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

It has been noted that the C1’ position of a DNA sugar moiety is the position most

thermodynamically able to donate a hydrogen atom. Unsurprisingly, we calculate that the

oxygen-transfer reaction from tirapazamine to the C1’ position of the model amino-

substituted deoxyribose radical is also extremely exothermic. Comparing these results,

we predict that oxygen transfer from tirapazamine to either methoxymethyl radical or a

DNA sugar radical is exothermic by about –37 kcal/mol. The reaction is illustrated in

Figure 2.4.

35 O HO O O O N HO NH2 N N N 12N N O + + NH2 N HO NH2 N NH N O 2 NH2 O O HO 2N OH OH Reactions Gas OH 1 -23.3 2 -33.3

Figure 2.4. Change of enthalpy [ΔH (298K, kcal/mol)] for oxygen transfer reactions between tirapazamine and a DNA sugar radical at the B3LYP/6-311+G**//B3LYP/6- 31G* level of theory.

2.3.5. The Denny Mechanism

In 2003, Denny and coworkers proposed that instead of hydroxyl radical, a benzotriazinyl radical is formed and is responsible for DNA strand breaks,15 as illustrated

in Figure 2.5. To test this proposed mechanism, I studied the key reactions of this

mechanism using computational methods and discovered that the proposed reaction

sequence is very exoergic.

O N N + O OH N N NH2 N X

N NH2 O OH N N + H2O N NH

Figure 2.5. The mechanism proposed by Denny and co-workers. 36

Denny proposed a concerted loss of water (the bottom step of Figure 2.5).

However, a concerted transition state is not found by theory. This suggests that the N-O bond is broken first, and that a complex is formed. Then in a subsequent step, the complex will either dissociate to form free hydroxyl radical and desoxytirapazamine, or overcome a barrier to form the benzotriazinyl radical and water, as shown in Figure 2.6.

Therefore, the hydroxyl radical lifetime can be limited by its immediate capture by the amino substituent of desoxytirapazamine T itself. We did not find a direct pathway leading from T and hydroxyl radical to the benzotriazinyl radical and water. Complex formation was always observed when these reagents approached each other.

37

O N O N N H N N N O N NH2 H N OH N O H N NH 5.0 4.3 HO Complex 2 0.0 1.4

O N N -8.1

N NH2 O -10.4 OH N N O N NH 2 N N + OH HO Complex 1 N NH2 -24.9

O N N + H2O N NH

Figure 2.6. The relative free energy change ΔG (kcal/mol) of the stepwise reactions to form benzotriazinyl radical at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

38

Additional calculations suggested that the benzotriazinyl radical can react

favorably with the DNA sugar itself and generate a DNA centered radical, although the

reaction is less exoergic than the reaction of hydroxyl radical with the DNA sugar moiety. The results are shown in Figure 2.7. The reaction of hydroxyl radical with a DNA sugar moiety is 14.5 kcal/mol more exoergic than the reaction between the benzotriazinyl radical with DNA sugar moiety, and proceeds with a small free energy barrier of only 1.9 kcal/mol.

39 O HO O HO O N NH2 N O NH N N 2 + H +

N NH HO N NH2 HO

0.0 -12.8

HO HO HO O NH 2 O O NH2 NH2 OH + H + H O H HO 2 HO O HO H

0.0 1.9 -27.3

Figure 2.7. The relative free energy change ΔG (kcal/mol) of two competitive reactions of hydroxyl radicalradicalwhich form DNA centered radicals, at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

40 As to how the benzotriazinyl radical damages DNA, Denny and co-workers

proposed a chain reaction, as shown in Figure 2.8.16 This mechanism suggests that the

benzotriazinyl radical abstracts a hydrogen atom from the model DNA sugar moiety to

generate an -hydroxyalkyl radical. This is followed by the formation of an adduct of - hydroxyalkyl radical joined to the N4 position of the benzotriazine-1,4-dioxides. This adduct then breaks down to form the tirapazamine radical proposed by Daniels and Gates.

After losing a water molecule, a new benzotriazinyl radical and an aldehyde is formed.

This interesting mechanism is limited to -hydroxyalkyl radicals because ethers and phosphorylated radicals of this type, as present in DNA, can not undergo the final concerted step which involves a hydrogen transfer.

O O N N N R N + H HO N NH2 N NH2 O RCH O O H

O R R - H N CH2 O N HO N NH O 2 O N N - H2O N N

N NH N NH2 OH

Figure 2.8. One-Electron Reduction of Benzotriazine 1-Oxides by -Hydroxyalkyl Radicals as proposed by Denny and co-workers.16

41 To test this mechanism, first we calculated the reaction of benzotriazinyl radical

with a DNA sugar moiety. The results are shown in Figure 2.9. This reaction forms the

C1 radical is exoergic by 12.8 kcal/mol, which is 3.7 kcal/mol more exoergic than the

formation of the C5 radical, as proposed in Denny’s mechanism.

O HO O HO O N NH2 N O NH N N 2 + H +

N NH HO N NH2 HO

0.0 -12.8

HO HO O CH2 O CH O N NH2 N O NH N N 2 + H + H N NH HO N NH2 HO

0.0 -9.1

Figure 2.9. The relative free energy change ΔG (298K, kcal/mol) of the reaction between the benzotriazinyl radical and a model DNA sugar moiety at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

Calculations of the addition of 1-methyl-1-hydroxyethyl radical and the

CH3OCH2 radical, models of a sugar moiety, to the tirapazamine radical anion have been performed. Addition to the N4 position does not lead to an adduct which is a potential energy minimum (addition to N4 position leads to rearrangement to the more stable adduct to oxygen position), thus the addition step is not a feasible process according to 42 our calculations. The change of enthalpies of the reaction of tirapazamine radical anion with the (CH3)2COH radical or CH3OCH2 radical (in parenthesis) are shown below.

-30.8 (-25.3)

O N X (7.4) N

N NH X 2 O (1.3) -27.8 (-23.5)

However, a more attractive option for these reactants is hydrogen transfer from the -hydroxyalkyl radical to tirapazamine as shown in Figure 2.10.

O O OH N O N N N + + H C CH H C CH 3 3 N NH 3 3 2 N NH2 O OH 0.0 -23.4

Figure 2.10. The relative free energy change ΔG (298K, kcal/mol) of the reaction between tirapazamine and a model DNA sugar radical [(CH3)2COH radical] at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

43 The radical addition step in Denny’s mechanism is not thermodynamically

favored but the overall mechanism is feasible according to our calculations. The energy

of the adduct formed at nitrogen is too high, however, for this to be a plausible

mechanism. Furthermore, the sugar model used in this mechanism and in Denny’s

experiment are -hydroxyalkyl radicals, as shown in Figure 2.11 (a). The hydrogen atom of the hydroxyl group plays a key role in this mechanism. However, in native DNA, the requisite hydrogen atom does not exist, as shown in Figure 2.11(b). Therefore, Denny’s mechanism is possible but unlikely for a DNA sugar moiety, and certainly can not be the mechanism of action of tirapazamine in vivo.

O O P O O Base O

O O HO P OH O O O RCH2OH = Base O HO

O O P O O

(a) (b)

Figure 2.11 The structure of a DNA sugar moiety and as part of a DNA chain.

44 2.4 Conclusions

We have investigated possible mechanisms of action of tirapazamine using

density functional theory (DFT), and demonstrated that the mechanism proposed by

Daniels and Gates (Scheme 2.1) is energetically feasible. In this mechanism, a series of

electron transfer/proton transfer/fragmentation reactions produces hydroxyl radical, a

well-known DNA cleavage agent.

The calculations found that the reaction of tirapazamine radical anion with

+ ammonium ion or with an NH4 -H2O cluster is exceedingly exothermic in the gas phase

because charge is dissipated. The reaction of tirapazamine radical anion with ammonium

ion in an aqueous continuum is slightly (–3.8 kcal/mol) exothermic. Thus, a primary or

secondary ammonium ion is predicted to be the key proton source to generate the

important reactive intermediate from the pro-drug, tirapazamine.

The reaction barrier and free energy change of the reactions to generate hydroxyl

radical have been calculated. The N-O bond dissociation of TOH has a lower barrier and is more exothermic than that of TOH’. Our calculations predict that TOH is the active intermediate to release the DNA cleavage agent, hydroxyl radical.

A second mechanism involving hydrogen-atom abstraction from the sugar moiety

by a tirapazamine-centered radical is not energetically feasible according to the

calculations. Therefore, we predict that a DNA radical produced by reaction with a

triazinyl radical, reacts with tirapazamine to form TO and a DNA sugar oxyradical.

Tirapazamine is predicted to transfer oxygen to radicals as opposed to TO and TO’ which

preferentially adds carbon-centered radicals. The calculations verified that the oxygen-

45 transfer reaction from tirapazamine to a DNA radical is energetically feasible and is a very unique feature of the drug.

Another mechanism proposed by Denny and co-workers has also been investigated. This mechanism proposes that benzotriazinyl radical is responsible for the

DNA strand break. Calculations have been performed to study this mechanism.We find that although some steps of the mechanism are not thermodynamically favorable, the overall mechanism is possible for the DNA sugar model used in their mechanism.

However, for the native in vivo system, the mechanism is not significant due to a lack of the key hydrogen atom in the DNA strand. It is more likely that both hydroxyl and benzotriazinyl radicals can generate the necessary DNA-centered radical, and then tirapzamine transfer its own oxygen to a DNA radical to cause the DNA strand break.

2.5 References for Chapter 2

1. Brown, J. M. Br. J. Cancer 1993, 67, 1163.

2. Brown, J. M.; Siim, B. G. Seminars Rad. Oncol. 1996, 6, 22.

3. Patterson, A. V.; Saunders, M. P.; Chinje, E. C.; Paterson, L.H.; Statford, I. J. Anti-Cancer Drug Design 1998, 13, 541.

4. Daniels, J. S.; Gates, K.S. J. Am. Chem. Soc. 1996, 118, 3380.

5. Daniels, J. S.; Gates, K. S.; Tronche, C.; Greenberg, M. M. Chem. Res. Toxicol. 1998, 11, 1254.

6. (a) Wilson, W. R. Tumor hypoxia: challenges for cancer chemotherapy; Wilson, W. R., Ed.; Kluwer Academic: Lancaster, 1992. (b) Denny, W. A. The Lancet Oncol. 2000, 1, 25.

46

7. Wardman, P.; Priyadarsini, K. I.; Dennis, M. F.; Everett, S. A.; Naylor, M. A.; Patel, K. B.; Stratford, I. J.; Stratford, M. R. L.; Tracy, M. British Journal of Cancer, Supplement. 1996, 74(27), S70-S74.

8. Ziegler, T. Chem Rev. 1991, 91, 651.

9. Barckholtz, C.; Barckholtz, T. A.; Hadad, C. M. J. Am. Chem. Soc. 2001, 121, 491.

10. (a) Becke, A. D. J. Chem. Phys. 1993, 98, 1372. (b) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B. 1988, 37, 785.

11. Frisch, M. J. et al. Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

12. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

13. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

14. Montgomery, J. A., Jr.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys. 1999, 110, 2822.

15. Anderson, R. F.; Shinde, S. S.; Hay, M. P.; Gamage, S. A.; Denny, W. A. J. Am. Chem. Soc. 2003, 125, 748.

16. Shinde, S. S.; Anderson, R. F.; Hay, M. P.; Gamage, S. A.; Denny, W. A.; J. Am. Chem. Soc. 2004, 126, 7865.

47

CHAPTER 3

ANALOGS OF TIRAPAZAMINE

3.1 Introduction

In Chapter 2, different possible mechanisms of action of tirapazamine were studied. According to our calculations, the mechanism proceeds via a series of electron- transfer/proton-transfer/fragmentation steps to eventually produce hydroxyl radical is energetically favored. The hydroxyl radical generated in this process may then react with the DNA sugar moiety to produce a sugar-centered radical. The second role played by tirapazamine is to oxygenate the radical generated on the DNA sugar and thereby induce the DNA strand break in an oxygen deficient environment.

Study of the mechanism of action of tirapazamine may lead to the design of a better drug. In this chapter, we will discuss a series of the analogs of tirapazamine. The role of the phenyl ring and substituents around the tirapazamine ring during the protonation and hydroxyl radical generation process will be discussed to provide direction to the synthesis and evaluation of analogs. The origin of the inactivity of two tirapazamine analogs will also be investigated.

48 3.2 Computational Methods

Density Functional Theory (DFT)1 has been applied in this study. All of the

structures of interest were optimized and their vibrational frequencies calculated at the

B3LYP/6-31G* level of theory. The zero-point vibrational energy and the thermal and

entropic correction were obtained from the vibrational frequency calculations. Singlet-

point energy calculations were performed at the B3LYP/6-311+G** level of theory, using six Cartesian d functions for all basis sets. The Polarizable Continuum Model

(PCM)2 has been applied in some cases for the solvent effect. All calculations were

performed using Gaussian 983 at the Ohio Supercomputer Center.

3.3 Results

3.3.1 Effects of phenyl ring substitution and benzannulation

We previously found that the protonation of tirapazamine by ammonium ion in

aqueous phase is exothermic by –3.8 kcal/mol according to our previous calculations. It

is necessary to know if drug analogs will react in the same manner. Therefore, the

protonation of tirapazamine analogs with ammonium ion have been calculated.

49 O OH O OH N N N N N N N N

N X N X N X N X OH O OH O

12 3 4

where X= NH2, F, CF3, CH3, OCH3, H, CN, CHO

Protonation reactions of monocyclic di-N-oxides and benzannulated di-N-oxides

radical ions to form neutral radical groups 1-4 in the aqueous phase are shown in Table

3.1-3.2. For the monocyclic ring systems, the H, OCH3, NH2, CH3, and F substituents

make the protonation process exoergic, while the systems with substituents CHO, CF3

and CN, the protonation is a slightly endoergic process. However, when hydrogen is

added to the oxygen of the N=N-O unit of the compound with the CHO substituent, the reaction is endoergic by 15.4 kcal/mol. For bicyclic ring systems, the trend is rather similar.

50

O OH + O N NH4 NH3 N N N N N or N X N X N X O O Entry X OH 1 H 0.0 -31.5 -30.4

2 OCH3 0.0 -5.8 -1.9

3 NH2 0.0 -4.3 -4.2

4 CH3 0.0 -3.2 -2.7 5 F 0.0 -1.5 1.1 6 CHO 0.0 0.5 15.4

7 CF3 0.0 0.6 0.9 8 CN 0.0 4.6 5.7

Table 3.1 Free energies (kcal/mol, 298K) of protonation of monocyclic di-N-oxides by ammonium ion in aqueous phase at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

51

O + O OH N NH NH N 4 3 N N N or N N X N X N X O OH O Entry X

1 NH2 0.0 -3.8 2.5

2 OCH3 0.0 -3.0 3.4

3 CH3 0.0 -2.8 -1.6 4 H 0.0 -2.4 -0.1 5 CHO 0.0 -0.3 2.4 6 F 0.0 0.8 2.2 7 CN 0.0 3.3 3.4

8 CF3 0.0 3.4 2.8

Table 3.2. Free energies (kcal/mol, 298K) of protonation of benzannulated di-N- oxides by ammonium ion in aqueous phase at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

52

As shown in Table 3.3, the gas-phase protonation of the N=N-O oxygen atom is slightly preferred for tirapazamine and its analogs in which the amino group is replaced with methyl, hydrogen, CF3 and cyano substituents. This tendency is also observed in the monocyclic system with amino, methyl and hydrogen. Fluoro, methoxy and formyl substituents favor protonation at the C=N-O oxygen atom due to hydrogen-bonding interactions. In the aqueous phase, protonation of the monocyclic and bicyclic analogs is preferred at the C=N-O oxygen with one exception (replacement of the NH2 group of tirapazamine with a CF3 group). Thus, in the aqueous phase, protonation is preferred at the oxygen atom that forms the neutral radical which fragments most readily to generate hydroxyl radical.

53

O OH O OH N N N N N N N N N X N X N X N X OH O OH O

Gas phase Aqueous phase Gas phase Aqueous phase

NH2 -2.1 6.3 -3.1 0.1 CH3 -2.1 1.2 -2.2 0.5 H -1.1 2.3 -1.2 1.1

CF3 -0.6 -0.6 0.3 0.3 CN -0.5 0.1 0.4 1.1 F 1.3 1.4 2.2 2.6

OCH3 6.6 6.4 5.1 3.9 CHO 7.4 2.7 8.3 14.9

Table 3.3. A comparison of the free energy (298K, kcal/mol) of protonation on the oxygen atoms of the N=N-O and C=N-O units at the B3LYP/6-311+G**//B3LYP/6- 31G* level of theory.

54

Tirapazamine contains an NH2 substituent on the aromatic ring (Table 3.6, entry

3), and that functional group is expected to influence the energetics of extrusion of

hydroxyl radical. Therefore, we compared the reaction barriers and changes of free

energies of the fragmentation process for compounds 1 - 4 with a series of substituents,

ranging from electron-withdrawing to electron-donating substituents such as F, CF3, CN,

CHO, H, NH2, CH3, and OCH3, respectively. Tables 3.4 and 3.5 list the energetics of the

reactions of monocyclic ring compounds, which contain compounds from group 1 and group 2, while Tables 3.6 and 3.7 detail the reactions of bicyclic systems, which contain compounds from groups 3 and 4. In general, the reactions of group 1 are the most exoergic and have the lowest reaction barriers, followed by group 3, then group 2. The reactions of group 4 are endoergic and have the highest reaction barrier. It is obvious that the presence of a benzannulated phenyl ring increases the reaction barrier to extrusion of hydroxyl radical and makes the reactions less exoergic. It is equally clear that protonation of the oxygen of a C=N-O unit leads to reaction barriers that are signigicantly smaller and reactions that are clearly more exoergic than protonation of the oxygen atom of N=N-O unit.

Boyd and coworkers4 have used monocyclic ring systems as models for

tirapazamine. The present results indicate that the monocyclic systems are not good

quantitative models of tirapazamine. The phenyl ring is not an inert bystander and

increases the reaction barrier by about 5 kcal/mol, and the free energy change of reaction

is ~6 kcal/mol higher than in the simple monocyclic ring system.

55 After protonation of the oxygen atom of C=N-O, the reaction barriers are about 6

kcal/mol lower than fragmentation reactions following protonation the oxygen of N=N-O

unit of the 1,2,4-triaza-di-N-oxides. The free energy changes are about 15 kcal/mol lower

in the former system.

For most reactions, NH2 is the substituent that produces the lowest barrier to formation of hydroxyl radical in groups 1-4. The formyl group is the substituent that

raises the barrier to fragmentation to the greatest extent.

56

O O O N N N N N N + OH N X N X N X Entry X OH OH

1 NH2 0.0 0.7 -16.9 2 F 0.0 1.1 -15.4

3 CF3 0.0 1.6 -13.9

4 CH3 0.0 1.6 -13.1

5 OCH3 0.0 1.7 -13.1 6 H 0.0 2.1 -11.9 7 CN 0.0 2.7 -11.1 8 CHO 0.0 8.2 -3.2

Table 3.4. Free energies (298K, kcal/mol) for homolysis reactions of monocyclic di- N-oxides in the gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

57

OH OH N N N N N N + X OH N X N N X O O O Entry X

1 NH2 0.0 6.0 -2.1 2 H 0.0 7.0 -1.0

3 CH3 0.0 7.4 -0.5 4 F 0.0 8.5 0.4 5 CN 0.0 8.9 0.3 6 CHO 0.0 9.3 1.7

7 CF3 0.0 9.4 1.1

8 OCH3 0.0 4.8 -3.0

Table 3.5. Free energies (298K, kcal/mol) for homolysis reactions of monocyclic di- N-oxides in gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

58

O O N O N N N N N N X + OH N X OH N X OH Entry X 1 F 0 3.8 -11.4

2 NH2 0 5.0 -10.4

3 CF3 0 5.1 -9.4

4 OCH3 0 5.4 -7.4

5 CH3 0 6.1 -7.1 6 CN 0 6.4 -5.9 7 H 0 6.8 -6.5 8 CHO 0 12.3 -25.4

Table 3.6. Free energies (298K, kcal/mol) for homolysis reactions of benzannulated di- N-oxides in the gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

59

OH OH N N N N N N + OH N X N X N X O OO Entry X

1 NH2 0.0 10.5 4 2 H 0.0 12.3 5.9

3 CH3 0.0 12.6 6.2 4 F 0.0 13.5 6.8

5 CF3 0.0 14.3 7.4 6 CN 0.0 15.8 6.3 7 CHO 0.0 16.3 7.9

8 OCH3 0.0 12.2 6.1

Table 3.7. Free energies (298K, kcal/mol) for homolysis reactions of benzannulated di- N-oxides in gas phase, as calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

60

3.3.2 Lumiflavin N-oxide Riboflavin N-oxide appears to have some advantages over tirapazamine as an anti-tumor agent. It enjoys greater water solubility because of the ribityl sugar side

chain; It is an analog of riboflavin (vitamin B2) and as such may be recognized as a

nutrient. After hydroxyl radical is released, it forms riboflavin (vitamin B2), which is an

innocuous substance that is generally regarded as safe. Therefore, calculations of

lumiflavin N-oxide, a simple model of riboflavin N-oxide, were performed. The

structures of riboflavin, riboflavin N-oxide, and lumiflavin N-oxide are illustrated in

Figure 3.1.

OH OH HO HO OH OH

OH OH CH3 H3C N N O H3C N N O H3C N N O NH NH NH H3C N H3C N H3C N O O O OO Riboflavin Riboflavin N-oxide Lumiflavin N-oxide

Figure 3.1. The structures of riboflavin, riboflavin N-oxide, and lumiflavin N-oxide.

Calculations of reactions of lumiflavin N-oxide have been performed. The results are illustrated in Figure 3.2. The electron and proton transfer processes are feasible because of the exoergicity. However, the reaction to release hydroxyl radical is endoergic

61 by 9.5 kcal/mol in the gas phase, but is near zero in the PCM model for the aqueous

phase. Therefore, according to our calculations, hydroxyl radical generation by

lumiflavin N-oxide would be much slower than that of tirapazamine itself.

CH CH + CH 3 3 NH4 NH 3 H C N N O H C N N O 3 H C N N O 3 +e 2 3 NH NH NH H3C N H3C N H3C N OO OO OH O

51.1 0.0 -121.8 91.1 0.0 - 2.7

CH3 CH3 CH3 H3C N N O H3C N N O H3C N N O NH NH H C N H C N NH + OH 3 3 H C N OH O O 3 OH O

0.0 15.9 9.5 0.0 13.2 0.1

Figure 3.2. Free energies [ΔG (kcal/mol, 298K)] of reaction for lumiflavin N-oxide in the gas phase (top) and in the aqueous phase (bottom), as calculated at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

3.3.3 The action of T and T’

Tirapazamine contains two N-oxide moieties and is biologically active. Desoxy-

tirapazamines, T and T’, which contain only a single N-oxide group are not biologically

active. To better understand these observations, we calculated the reduction of T and T’

and the protonation of these radical anions by ammonium ion.

62 In the case of T, the calculations reveal that unsurprisingly, the reduction by a free electron is exoergic in both the gas phase and water, as shown in Figure 3.3. Compound

T is less easily reduced than tirapazamine by 6.9 kcal/mol. However, the reaction of T radical anion with ammonium ion is endoergic by 1.7 kcal/mol in the aqueous phase, as compared to tirapazamine (TO) where the analogous proton-transfer process is exoergic by 3.8 kcal/mol. Ammonium ions are common counterions to the negatively charged

DNA, but will not protonate the radical anion of T – this in part may explain the lack of activity of T, as protonation of the radical anion is required to trigger the release of hydroxyl radical from desoxytirapazamine T.

O O OH N NH + - N 4 NH3 N N +e N N N NH2 N NH 2 N NH2 T T TH 31.3 0.0 -125.0 81.5 0.0 1.7

OH OH N N N N N N + OH N NH2 N NH2 N NH2 TH TS(TH) dT

0.0 10.5 3.9 0.0 10.7 0.6

Figure 3.3. Free energies [ΔG (kcal/mol, 298K)] of reactions for desoxytirapazamine T in the gas phase (top) and in the aqueous phase (bottom), as calculated at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

63

The calculations predict that the reduction of T’ is also very exoergic in both the gas phase and in aqueous phase (Figure 3.4). Tirapazamine is more easily reduced than

T’ by 4.9 kcal/mol in gas phase. However, the protonation of the T’ radical anion by ammonium ion and subsequent dissociation to release hydroxyl radical are both exoergic in the gas phase and in aqueous phase. This is in contrast to tirapazamine in processes wherein the fragmentation of TOH is endothermic. Thus, we predict that T’ will be less readily reduced than tirapazamine, but that after reduction, it will be rapidly decomposed.

There are two possible reasons for the inactivity of T’. One is its less favorable reduction potential. More significantly, the T’ radical anion is significantly more basic than the tirapazamine radical anion. The drugs are reduced enzymatically to form radical anions which must migrate to the cellular DNA. The radical anion derived from T’ may be protonated long before it encounters nucleic acid and release hydroxyl radical outside the vicinity of the nucleic acid.

64 NH + N - N 4 NH3 N N +e N N

N NH2 N NH 2 N NH2 O O OH T T' TH' 33.3 0.0 -119.9 80.5 0.0 -2.7

N N N N N N + OH N NH2 N NH2 OH OH N NH2

T'H TS(T'H) dT'

0.0 3.8 -11.6 0.0 9.2 -9.2

Figure 3.4. Free energies [ΔG (kcal/mol, 298K)] of reactions for desoxytirapazamine T in the gas phase (top) and in the aqueous phase (bottom), as calculated at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

The third reason for the inactivity of T’ lies in the process that results in

oxygenation of the DNA radical. Calculations of the reaction of tirapazamine radical

anion TO, desoxytirapazamine T and T’ with methoxymethyl radical have been

performed. Here methoxymethyl radical has been utilized as a simple model of a DNA

sugar radical. As shown in Figure 3.5, methoxymethyl radical can undergo numerous

exothermic reactions with tirapazamine. The most exothermic reaction, however, is addition to oxygen as opposed to carbon or nitrogen. This confirms that tirapazamine could selectively oxygenate DNA sugar radicals and subsequently induce strand cleavage.

65

H H H

C O C H

H O H C H

C N C

N C H C H O C C H O H H N C H C H N H H H O H C C N H H C O H H H H O C C N

C N H

C C N C C C H H C N N C C C H H -10.7 N H H C N O

H H -8.2 -25.3 O

O H H C C

H H H + O H H O H O H H N C H C H H N C C N -8.8 C N H C O C -17.6 H C C H N N C H C C H C H N H O C N C C C C C H C H H C H H C N C C N H O H H N N H C H H O

H

H O

-5.4 -11.6 H H H H O C H -23.5 C H C N H O H C C N

C C C H H O H N + N O C H H

H H H C N C O C N H H C C N O C C N

C H C C H C C C H C H H C N H N H N C N

H H O H O

H C O H

H C H H

Figure 3.5. The change of enthalpies (298K, kcal/mol) of reactions of tirapazamine and CH3OCH2 radical at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

66 As shown in Figures 3.6 and 3.7, desoxytirapazamines T and T’ also have numerous exothermic reaction channels with methoxymethyl radical. However, oxygen

addtion, unlike the case with tirapazamine, is not the most exothermic process. The most

favorable reaction is addition to nitrogen. Thus neither T nor T’ will selectively

oxygenate DNA sugars and as a result are less effective than tirapazamine in promoting

DNA strand cleavage.

67

H

H H C O C H H H O C H C

N C H C H H N C O H H C H H H C O H C C C N C H N H H C C C N H O O H H N H

C C C C H H N H C C H C N N N

H H C C C H H C N N

H -8.5 H -7.5 -19.2

O H H H

O C O H C H H C H -4.6 H -21.8 H C N C O H H H C N H H C C N C C N

C H H C C N C C C H O N O H C H H N C H C N H C H H H H C N H H H

C C N

C C C H

H N C N 7.7 -10.8 H H H H H H C C H

O -17.1 H H O

C + H N

C C O N H

C C N C C H H

H C N N N C C

H H H C C H C C H H N O C N

H O H C H

H H

C N H N C C

H C C C N N H C H

H H C H H C O H H

Figure 3.6. The change of enthalpies (298K, kcal/mol) of reactions of desoxytirapazamine T and CH3OCH2 radical at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

68

H

N C H N C C

C H C C N N H C

H

H O

-18.3 H C O H

H C H

H O H H C C

H H H

+ H

H H H C H H

C N C O H N H C C H -26.7 C N N C C

C C C H H C N N C H C C N H H H C N O

H

H O

-34.4

H

H C O

H

H C H H

C N H

C C N

C C C H N H C N

H H O

Figure 3.7. The change of enthalpies (298K, kcal/mol) of reactions of desoxytirapazamine T’ and CH3OCH2 radical at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

69 3.4 Conclusions

The properties of monocyclic and bicyclic substituted 1,2,4-triazole-1,4-di-N- oxides were considered using density functional theory calculations. Substituents influence the site of protonation of the di-N-oxide radical anions. Amino, methyl and hydrogen groups at the C3 position direct protonation to the N=N-O oxygen of the radical anions. Fluorine, methoxy and formyl groups direct protonation to the C=N-O oxygen of the radical anion to take advantage of hydrogen-bonding interactions. In all cases, fragmentation of XĊR-NROH radicals is more rapid than that of RXŃ-NROH radicals. Methoxy and fluorine substituents do not raise the barrier to fragmentation of the neutral radicals, relative to amino, thus they may be more active compounds than tirapazamine. The formyl group raises the barrier to fragmentation of the corresponding neutral radicals, and molecules substituted with this group should have low activity.

Barriers to fragmentation of the neutral radicals of monocyclic systems are on average ~3 kcal/mol lower than the bicyclic systems.

The two inactive mono-N-oxide derivatives of tirapazamine were also studied by

DFT calculations and the results are in good agreement with experiment. The neutral hydro-radical derived from N-1 oxide is not as easy to protonate and to fragment.

Although the N-4 oxide neutral hydro-radical is predicted to fragment rapidly, the carbon-centered DNA sugar radical addition to a nitrogen atom of the N-oxide is predicted to be more competitive than addition to oxygen. Thus the N-4 oxide can not selectively oxygenate the DNA radical.

70

3.5 References for Chapter 3

1. Ziegler, T. Chem Rev. 1991, 91, 651.

2. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

3. Frisch, M. J. et al. Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

4. Ban, F.; Gauld, J.W.; Boyd, R.J. J. Am. Chem. Soc. 2001, 123, 7320.

71

CHAPTER 4

RADICAL ADDITIONS TO AROMATIC N-OXIDES

4.1 Introduction

When tirapazamine reacts with carbon-centered DNA radicals, the most favorable

position is oxygen. However, analogs of tirapazamine, the desoxytirapazamines prefer

radical addition to nitrogen, instead of oxygen. In this chapter, we will discuss what is

the most favorable position for other aromatic N-oxides, such as pyridine N-oxide,

pyrazine-1,4-di-N-oxide, phenazine N,N'-dioxide, 2,4,6-trimethylpyridine-N-oxide, 2,4,6-

trichloropyridine-N-oxide, tetrachloropyrazine di-N-oxide, and tetramethylpyrazine di-N-

oxide.

4.2. Computational Methods

Density Functional Theory (DFT)1 has been applied in this study. All of the

geometries have been fully optimized at the B3LYP/6-31G* level. Vibrational

frequencies have been calculated at the same level to obtain the zero-point vibrational energy correction as well as thermal and entropic corrections to the free energy. Singlet-

72 point energy calculations were performed at the B3LYP/6-311+G** level of theory, using six Cartesian d functions for all basis sets. The calculations described in this chapter are all gas-phase calculations, because we found in the previous chapter, that solvent effects are not significant in radical addition reactions. All of the calculations were performed with Gaussian 982 at the Ohio supercomputer center.

4.3 Results

4.3.1 Reactions of methyl radical with benzene or pyridine

The reactions of methyl radical, a simple model of a DNA sugar radical, with

benzene were calculated. The reaction is exothermic by 6.9 kcal/mol, with a barrier of

12.1 kcal/mol, as shown in Figure 4.1. To our knowledge, the experimental enthalpy of this reaction is not known, but the experimental activation barrier of 8.9 is in fair agreement with our calculations.3

CH3 H H CH3 + CH3

0.0 12.1 -6.9

Figure 4.1 The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl

radical to benzene at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

73

Further calculations of methyl radical with pyridine have been performed. The enthalpies of reaction of the addition of methyl to all possible positions of pyridine are shown in Figure 4.2. The addition to all positions are exothermic, and the most favorable position is nitrogen., which is exothermic by 11.3 kcal/mol.

-11.3

N -8.4

-7.7

-6.6

Figure 4.2. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to all possible positions of pyridine at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

4.3.2 Reaction of methyl radical with pyridine N-oxide

The simplest aromatic N-oxide model is pyridine N-oxide. We studied the reactions of pyridine N-oxide and methyl radical by computational methods. The results are illustrated in Figure 4.3.

74 -7.3

O N -22.6

-6.4

-20.8

Figure 4.3. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to all possible positions of pyridine N-oxide at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

It can be seen that methyl radical reacts both rapidly and exothermically with pyridine N-oxide on either oxygen or carbon, with the latter intermediate being the more stable species. The reaction barrier for the addition of methyl radical to oxygen is ~9.8 kcal/mol, and the barrier to addition to a carbon position is about 1.7 kcal/mol lower.

This is not surprising as carbon addition at the ortho or para position produces a stabilized radical. The additions of methyl radical to benzene and to the ortho and para positions of pyridine are –6.9, -8.4 and –6.6 kcal/mol, respectively. The addition of methyl to the nitrogen atom of pyridine is exothermic by 11.3 kcal/mol. These reactions are approximately 10 kcal/mol less exothermic than addition of methyl to the para carbon of pyridine-N-oxide.

Addition of methyl radical to carbon will form nitroxyl radical 5 (Figure 4.4) and does not lead to oxygenation of the methyl radical. Thus, reaction of a DNA sugar radical with pyridine N-oxide will lead to adduct formation instead.

75

O O O N N N + CH3 H H C H C H 3 3 5 0.0 8.1 -20.8

CH3 O N

9.8

CH3 CH3 O N N O N + OCH3

6 -29.1 -7.3 -7.1

Figure 4.4. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to pyridine N-oxide and the oxygen transfer reaction at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

76 This is not the case when methyl radical adds to the oxygen atom of pyridine-N- oxide to form 6. The dissociation of the N-O bond of intermediate 6 has a rather small reaction barrier, only 0.2 kcal/mol, and the fragmentation to form pyridine and methoxyl radical is exoergic by about 30 kcal/mol. This ultra short lifetime of radical 6 is consistent with literature reports.4 N-alkoxy radicals can be generated by photoreduction of N- alkoxy pyridine salts. The N-alkoxy radicals fragment in ns or ps (depending on the substituent) to form pyridines and alkoxy radicals. Thus, although net oxygen transfer is more favorable than methyl radical addition to carbon, DFT calculations indicate that it will be only a minor process with pyridine-N-oxide.

So far we used methyl radical as a simple model for the DNA sugar moiety.

However, to demonstrate that methyl radical is a good model and that the steric effect of

DNA sugar radical is not significant in these addition reactions, the addition reactions of a DNA sugar moiety with pyridine N-oxide have been performed. The results are shown in Figure 4.5. The addition to the carbon position is less exothermic because of the steric effect, but the radical still prefers addition to carbon. Therefore, the simple methyl radical model predicts the same trend followed with the DNA sugar radical.

77 HO O NH2 O HO N O HO N O NH2 -8.3 + HO HO 0.0 O NH2

HO N O

-14.0

Figure 4.5. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of the C1-radical derived from an amino-substituted deoxyribose model to pyridine N-oxide and the oxygen transfer reaction at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

4.3.3 Oxygen transfer from pyrazine-1,4-di-N-oxide to methyl radical

Tirapazamine is a di-N-oxide, thus the mono-N-oxide pyridine N-oxide might not be the ideal model. We calculated the oxygen transfer reaction from pyrazine-1,4-di-N- oxide to methyl radical. Our results demonstrate that the 1,4-di-N-oxide is also a powerful oxygen donor. The results are shown in Figure 4.6. In fact, addition of a methyl radical to pyrazine-di-N-oxide is predicted to be faster than to pyridine-N-oxide. The addition to carbon of the pyrazine-di-N-oxide is only 0.8 kcal/mol more exothermic than that of pyridine N-oxide, but the addition to oxygen is 11.4 kcal/mol more exothermic. It suggests that the di-N-oxide tends to increase the reactivity of the oxygen position.

78 However, addition to carbon is still predicted to be faster than to the oxygen of pyrazine- di-N-oxide. Thus, this molecule should fail to selectively oxygenate a DNA sugar radical.

79

O O O H N N N H + CH3 CH3 CH3 N N N O O O 8 0.0 5.8 -23.4

CH3 O N

N O 6.1

CH3 CH3 O O N N N + OCH3 N N N O 7 O O -32.5 -18.7 -14.3

Figure 4.6. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to pyrazine, di-N-oxide and the oxygen transfer reaction at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

80 The reaction of methyl radical with pyrazine-1,4-di-N-oxide, and the subsequent fragmentation of radical 7 are still quite exoergic, although the reaction barrier to fragmentation of 7 is somewhat higher than the corresponding reaction barrier for pyridine-N-oxide-methyl radical adduct 6.

The reaction of methyl radical with pyrazine-1,4-di-N-oxide, and the subsequent fragmentation of radical 7 are still quite exoergic, although the reaction barrier to fragmentation of 7 is somewhat higher than the corresponding reaction barrier for pyridine-N-oxide-methyl radical adduct 6.

4.3.4 Effect of phenyl ring

In Chapter 3, we found that the phenyl ring in tirapazamine is not a spectator.

Here, again, we studied the effect of a phenyl ring. The addition reaction of methyl radical with 1,4-di-N-oxide-quinoxaline and phenazine N,N'-dioxide are shown in Figure

4.7. The phenyl ring makes reactions of methyl radical with 1,4-di-N-oxide-quinoxaline more exothermic, but the most favorable position of reaction is still carbon. In the presence of two phenyl rings, oxygen becomes the preferred position for phenazine N,N'- dioxide.

81 -22.0 -29.5

-21.8 O -27.0 O N N -19.5

N N O O

Figure 4.7. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to possible positions of 1,4-di-N-oxide-quinoxaline and phenazine N,N'-dioxide at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

4.3.5 Effect of substituent

To understand the effect of substituents for the preference of addition of methyl radical to either an oxygen position or carbon position, we studied the reactions of methyl radical with 2,4,6-trimethylpyridine-N-oxide, 2,4,6-trichloropyridine-N-oxide as well as tetrachloropyrazine di-N-oxide and tetramethylpyrazine di-N-oxide. The results are illustrated in Figures 4.8 and 4.9. For 2,4,6-trimethylpyridine-N-oxide, 2,4,6- trichloropyridine-N-oxide and tetramethylpyrazine di-N-oxide, the most favorable position is the oxygen position, rather than the ortho or para carbon position. However, for the tetrachloropyrazine di-N-oxide, the carbon position is still preferred to the oxygen position. Both chlorine and methyl substituents make the addition to the oxygen position more exothermic, thus increasing the opportunity for the oxygen addition over nitrogen addition.

82 -43.3 -25.7

O O Cl N Cl H3C N CH3 -25.5 -17.9 Cl -13.2 CH3 -16.3

Figure 4.8. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to possible positions of 2,4,6-trimethylpyridine-N-oxide and 2,4,6- trichloropyridine-N-oxide at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

-26.2 -22.8

O -32.9 O -22.2 Cl N Cl H C N 3 CH3

Cl N Cl H3C N CH3 O O

Figure 4.9. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of methyl radical to all possible positions of tetrachloropyrazine di-N-oxide and tetramethylpyrazine di-N-oxide at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

83 4.3.6 Silicon Centered Radicals

Finally, silicon centered radicals are predicted to add preferentially to the oxygen atom of pyridine-N-oxide.

O CH N 3 Si + H3C CH3 0.0

O Si(CH3)3 N O N

H Si(CH3)3

-5.7 -24.4

Figure 4.10. The change in enthalpies [ΔH (kcal/mol, 298K)] for the addition of trimethylsilyl radical to possible positions of pyridine-N-oxide at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

This is a consequence of the significantly greater strength of a Si-O bond relative to a Si-C bond. The fragmentation of the oxygen adduct is very exothermic by –24.4 kcal/mol, as shown in Figure 4.11.

84 + (CH3)3Si NO (CH3)3Si O N ΔH = -24.4 kcal/mol

Figure 4.11. The change in enthalpy [ΔH (kcal/mol, 298K)] for the fragmentation reaction of the N-O bond in the silyl adduct at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

4.4 Conclusions

Methyl radicals add exothermically to benzene and pyridine. It adds even more exothermically to the carbon of pyridine-N-oxide to form a nitroxyl radical. The addition of methyl radical to the oxygen atom of pyridine-N-oxide is exothermic, and the resulting adduct then fragments readily to form pyridine and an alkoxy radical. Nevertheless, this overall sequence is less favorable than addition to the carbon atom of pyridine-N-oxide.

Pyrazine di-N-oxide increases the exothermicity of the addition to the oxygen position, although the most favorable addition is still to the carbon position. Thus pyridine-N-oxide and pyrazine di-N-oxide should react with DNA sugar radicals to form adducts and not by oxygen transfer, with subsequent strand cleavage of DNA.

In Chapter 3, we found that desoxytirapazamines T and T’ have numerous exothermic reaction channels with methoxymethyl radical. But oxygen addition, unlike the case with tirapazamine, is not the most exothermic process. This could be explained by the fact that pyrazine di-N-oxide increases the exothermicity of the pathway for addition to the oxygen position relative to that for pyridine N-oxide. As an aromatic di-

N-oxide, tirapazamine has a greater tendency to add radicals to an oxygen atom than the mono-N-oxide desoxytirapazamines T or T’. 85 The preference does not change with presence of one phenyl ring. As shown in

Chapter 3, however, the most exothermic reaction of tirapazamine is addition to oxygen and not carbon. This is probably due to the effect of nitrogen on the ring as well as the amino group as a substituent. With two phenyl rings, the most favorable position becomes the oxygen atom due to the fact that no addition can proceed on the di-N-oxide ring. Thus we predict that phenazine N,N'-dioxide will selectively oxygenate DNA sugar radicals and induce strand cleavage.

Substitutent effects were also investigated. Both chlorine and methyl substituents could increase the opportunity for oxygen addition over nitrogen addition. This is probably due to a steric effect.

4.5 References for Chapter 4

1. Ziegler, T. Chem Rev. 1991, 91, 651.

2. Frisch, M. J. et al. Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

3. Holt, P. M.; Kerr, J. A. Int J Chem Kinet. 1977, 9, 185.

4. Lorance, E. D.; Kramer, W. H.; Gould, I. R. J. Am. Chem. Soc. 2002, 124,15225.

86

CHAPTER 5

A COMPARISON OF ACETYL AND METHOXYCARBONYLNITRENES

This chapter is reproduced in part with permission from the Journal of the Organic Chemistry volume 69, page 8583. Copyright 2005, American Chemical Society.

5.1 Introduction

Carbonyl substituted azides are readily synthesized and have been studied by physical and organic chemists for several decades.1 Upon activation with heat or light, both acyl and alkoxycarbonyl azides extrude molecular nitrogen and form carbonyl nitrenes. These nitrenes can undergo intramolecular rearrangement (Curtius

Rearrangement) to form isocyanates.2

O O O Δ, hν C

R XN3 R XN RX N

Δ, hν

RX = alkyl or aryl or RX = O-alkyl or O-aryl

87 There is evidence that migration of RX can proceed in concert with nitrogen extrusion in some cases, without the intervention of a free nitrene. There is also evidence that the stepwise or concerted nature of the process varies with the mode of azide decomposition.3

Carbonyl-substituted nitrenes are efficiently intercepted with a variety of reagents, and cis and trans substituted traps are particularly diagnostic trapping agents.

Extending the Skell-Woodworth Rules from carbenes4 to nitrenes allows deduction of the multiplicity of the reactive intermediate that is captured. Product studies with acylnitrenes and cis revealed retention of configuration in the aziridine product.5

O O O

1 R N3 hν R N R N

This result is consistent with trapping of the singlet state of the acylnitrene.

Different results are obtained with alkoxycarbonyl azides, which form mixtures of aziridines upon photolysis, in the presence of alkenes.6

88 O O O

N 1N RO 3 hv RO RO N

O O O + 3 RO N RO N RO N

This result is consistent with the interception of both the singlet and triplet states of an alkoxycarbonylnitrene.

The conclusions deduced from chemical trapping studies are also consistent with matrix spectroscopic experiments. The triplet ESR spectra of alkoxycarbonylnitrenes are observed upon low temperature photolysis of alkoxycarbonyl azides,7,8 confirming that alkoxycarbonylnitrenes have triplet ground states. However, triplet acylnitrenes have never been detected by ESR spectroscopy when acyl azides are irradiated at cryogenic temperatures. Photolysis of benzoyl azide and related azides fails to produce ESR spectra characteristic of a benzoylnitrene. Taken together, the chemical trapping studies and matrix spectroscopic studies indicate that acylnitrenes have singlet ground states.

Imidogen (NH),9 alkyl7,10 and arylnitrenes7,11 have triplet ground state multiplicities. Recent density functional theory (DFT) and high level ab initio calculations, however, indicate that other acylnitrenes have closed-shell singlet ground states because of a bonding interaction between the oxygen and nitrogen atoms of the intermediate.12, 13

89 O O R R N N

The position of the carbonyl vibration of singlet benzoylnitrene, observed by matrix and time-resolved infrared (TRIR) spectroscopy, is consistent with the prediction of DFT calculations.13 A recent report by Pritchina, Gritsan, and Bally have examined the chemical preferences of formylnitrene (HC(=O)N) and hydroxycarbonylnitrene

(HOC(=O)N.13b

To better understand the differences between acetylnitrene (CH3C(=O)N) and methoxycarbonylnitrene (CH3OC(=O)N), we have performed calculations to predict their ground state multiplicities and reactivity at consistent levels of theory, and we have attempted to directly compare alkylcarbonylnitrenes and alkoxycarbonylnitrenes with the use of isodesmic equations. Herein, we are pleased to report our results. Some of these predictions were tested by Mandel’s Laser Flash Photolysis studies of benzoylnitrene.31

5.2 Computational Methods

Density functional theory (DFT)14 and ab initio methods15 were applied in this study. The geometries were completely optimized at the B3LYP/6-31G* level. Analytical vibrational frequencies were calculated at the same level for each stationary point to verify a minimum energy structure and to provide zero-point vibrational energy

90 corrections, which were scaled by a factor of 0.9805.16 Each transition state was verified to connect to the respective reactant and product by careful optimization (opt=calcfc or calcall) after displacement (typically 10%) along the reaction path for the normal coordinate of the imaginary vibrational frequency. In some cases, intrinsic reaction coordinate (IRC)17 calculations were also performed. The zero-point vibrational energies as well as thermal and entropic contributions to the free energies were obtained from the

B3LYP/6-31G* frequency calculations; the thermal and entropic corrections used the unscaled vibrational frequencies. Single-point energy calculations of all stationary points were performed at the B3LYP/6-311+G** level using the corresponding B3LYP/6-31G* geometries. Six Cartesian d functions were used for these calculations.

In some cases, CCSD(T)18 singlet-point energies were evaluated using the

B3LYP/6-31G* geometries, with the 6-311+G** (6d) basis set as well as Dunning’s correlation consistent basis set, aug-cc-pVDZ.19 For further comparison, Complete Basis

Set (CBS) methods20 have also been applied to increase the accuracy of the results. The energies provided are CBS-QB3 enthalpies and free energies at 298.15 K. All stationary points, including transition states, were re-characterized at the CBS-QB3 level, complete with full geometry optimization at the required level. To study the influence of solvent, polarizable continuum model (PCM)21 calculations were used for the Curtius rearrangement with acetonitrile and cyclohexane as solvents. Population analyses were performed at the B3LYP/6-311+G**//B3LYP/6-31G* level using the natural population analysis (NPA)30 method of Weinhold and co-workers.

All DFT, CCSD(T) and CBS-QB3 calculations were performed with Gaussian

9822 at the Ohio Supercomputer Center.

91

5.3 Results

5.3.1 Singlet-Triplet Energy Splittings

The singlet-triplet energy splitting of nitrenes 9 and 10 as well as carbene 11 were computed using B3LYP, CCSD(T), and CBS-QB3 methodologies. In particular, the

CBS-QB3 method relies on a B3LYP geometry, a CCSD(T) energy as an estimate for electron correlation and an extrapolation to the infinite basis set limit. As such, it is the most accurate method utilized in this study. The results are given in Table 5.1.

92

O O O

CH3 N N CH CH CH3O 3 11 9 10

9 10 11 Level of Theory B3LYP/6-311+G**//B3LYP/6-31G* 4.9 13.9 3.7

CCSD(T)/6-311+G**//B3LYP/6-31G* 1.9 12.2 4.4

CCSD(T)/aug-cc-pVDZ//B3LYP/6-31G* 1.6 12.0 3.6

CBS-QB3 -4.0 5.7 1.3

a A negative sign denotes a singlet ground state.

Table 5.1. Singlet-Triplet energy gap [∆H (298K) (kcal/mol)] of species 9, 10 and 11 by DFT, CCSD(T) and CBS-QB3 methods.a

93

B3LYP and CCSD(T) methods predict that compounds 9-11 will all have triplet ground states as opposed to CBS-QB3 theory which indicates that while 10 and 11 have triplet ground state multiplicities, acetylnitrene 9 is predicted to have a singlet ground state. This conclusion is in agreement with G2 calculations by Faustov et al.12c B3LYP predicts the triplet nitrenes to be too stable, relative to CBS-QB3 by approximately 8-9 kcal/mol. In the case of the carbene, the discrepancy is only 2.4 kcal/mol. Scott et al. came to similar conclusions in their studies of carbonyl carbenes.23

The geometries of 9-11 in their singlet and triplet states are shown in Figure 5.1.

The bond distance between the oxygen and nitrogen atoms in the singlet state is about

0.46 Å shorter than that in the triplet state for acetylnitrene, and the O-C-N bond angle in the singlet state is also smaller than that in the triplet state. Therefore, as noted in earlier studies,13 a bonding interaction between the oxygen and nitrogen atoms provides an explanation for the singlet ground state of acetylnitrene 9. Methoxycarbonylnitrene 10 and carbene 11 have weaker bonding interactions between oxygen and nitrogen or between oxygen and carbon in their singlet states as demonstrated by the longer bond distances and larger bond angles (Figure 5.1). Our results are reminiscent of the results obtained by Pritchina et al.13 in which B3LYP/6-31G* and CCSD methods, coupled with a complete basis set extrapolation, predicted that H(C=O)N has a triplet ground state, but

CCSD(T) with a complete basis set extrapolation predicted that this same species has a singlet ground state.

94 O O

1.77 H 1.31 H 1.29 1.84 O o 86.8 o 1.27 C C 91.9 H 1.93 H 1.48 C N o 1.26 1.31 N 93.3 1.28 1.49 C H C 1.45 H O C H 1.38 C H H H

19 110 111

O O 2.28 1.24 2.23 O 1.24 o 2.28 116.8o C 117.0 H H 1.22 H 1.51 H C o 1.51 C C 120.6 H 1.38 N 1.34 C C 1.43 C H 1.44 O 1.40 N H H H H

3 3 3 9 10 11

Figure 5.1. The geometries of 9-11 in their singlet and triplet states. Distances are shown in Å, and angles are in degrees.

To better understand why the ground state multiplicities of 9 and 10 differ, several isodesmic reactions were calculated with DFT methods. To verify the accuracy of DFT methods in this case, CBS-QB3 methods were also applied to the isodesmic reactions (7) and (9). Compared to the CBS-QB3 method, the B3LYP results differ by less than 1 kcal/mol, and indicate that DFT is a reliable method for this purpose. Upon consideration of reaction 7, it can be seen that singlet methoxycarbonylnitrene 13 is 13.1 kcal/mol more stable than acylnitrene 12 in the singlet state. On comparing the triplet state, methoxycarbonylnitrene 313 is even more stable, relative to 312, by 17.1 kcal/mol

(reaction III) than the singlet state analogs. The difference in the singlet-triplet energy splitting of 9 and 10 is explained by isodesmic equations 8 and 10. Conjugation of

95 oxygen with a C(=NH)OH group is stabilizing by 15.7 kcal/mol (reaction (8)) as in the carbonylnitrene singlet state, but conjugation of oxygen to a C(=O)NH· radical (reaction

(10)), as in the triplet nitrene, is even more stabilizing (by 18.1 kcal/mol). Thus the methoxycarbonylnitrene has a triplet ground state, and the acetylnitrene does not.

96

O -13.9 O (7) C N C N CH OCH CH CH O 3 2 -13.1 (CBS-QB3) 3 2 1 112 13

HN -15.7 HN

(8) C C CH3OCH2 OH CH3CH2O OH II-left II-right O O -16.9

C C (9) 3 -17.1 (CBS-QB3) CH CH O 3 CH3OCH2 N 3 2 N 312 313

O -18.1 O

(10) C C CH3OCH2 NH CH3CH2O NH

IV-left IV-right

ΔH (kcal/mol) B3LYP/6-311+G**//B3LYP/6-31G*

97 5.3.2 Curtius Rearrangement

The reaction surfaces for the Curtius rearrangement of nitrenes 1 and 2, as computed by DFT and CBS-QB3, are shown in Figure 5.2, and these results are in agreement with previous calculations12c on the activation barrier for rearrangement of 1.

The Curtius rearrangement of 1 is substantially more exoergic than that of 2 because a relatively weak N-O bond is formed in the isocyanate product, but surprisingly, rearrangement of 1 has a larger free energy of activation. A substantial solvent effect is not predicted by the polarizable continuum model (PCM) calculations using either polar solvent acetonitrile or the non-polar solvent cyclohexane for the Curtius Rearrangements.

98

O O 1.31 1.77 1.25 2.25 1.21 1.18 o N C O o 125.3 H 86.8 C 1.44 H C 1.54 1.48 H N C H 1.26 1.28 C H C N H H

H H 9 9-ts 9-p

0.0 14.8 -72.1 DFT (CH3CN) 0.0 14.2 -72.9 DFT (cyclohexane) 0.0 13.9 -73.2 DFT (gas) 0.0 20.3 -63.4 CCSD(T) (gas) 0.0 21.8 -67.4 CBS-QB3 (gas)

O O H H 1.29 1.84 91.9o 1.23 2.25 C H o H C C 126 1.31 N 1.23 H 1.27 1.38 N H O o 1.18 C 1.45 107 C 1.43 O C O N 1.41 H 1.45 O H H 10 10-ts 10-p

0.0 10.9 -26.9 DFT (CH3CN) 0.0 10.2 -27.6 DFT (cyclohexane) 0.0 9.0 -29.5 DFT (gas) 0.0 13.3 -21.8 CCSD(T) (gas) 0.0 14.1 -24.4 CBS-QB3 (gas)

Figure 5.2. The energy surface [ΔG(298K) (kcal/mol)] for Curtius rearrangement of 9 (top) and 10 (bottom) at the B3LYP/6-311+G**//B3LYP/6-31G* level in acetonitrile, cyclohexane, and in the gas phase as well as CCSD(T)/aug-cc-pVDZ//B3LYP/6-31G* and CBS-QB3 levels. Distances are shown in Å, and angles are in degrees.

99 The relative free energies of the two reactions can be understood upon consideration of the following isodesmic equations.

N C O NCO CH3OCH2 CH3CH2O ΔG = 36.8 kcal/mol 14 15

CH3OCH2 NH2 CH3CH2O NH2 ΔG = 27.8 kcal/mol

14' 15'

A C-O bond is simply much stronger than an N-O bond, which makes the product of the

Curtius rearrangement of 9 much more stable than that of 10.

To understand the difference in the respective activation barriers, the isodesmic reactions shown in Figure 5.3 were calculated for the configurational isomers: ethoxycarbonylnitrene 13 and methoxymethylacylnitrene 12. The transition state for

Curtius rearrangement of 13 is 20-30 kcal/mol more stable relative to the transition state for Curtius Rearrangement of its isomeric methoxymethylacylnitrene 12. The oxygen atom present in 13 is responsible for an interaction that stabilizes the transition state for the Curtius rearrangement. This interaction is not present in the transition state of the isomeric acylnitrene 12. The extra oxygen atom stabilizes the transition state of Curtius rearrangement more than it stabilizes nitrene 5, thus methoxycarbonylnitrene should rearrange faster than acetylnitrene. Thus we predict that Curtius rearrangement of singlet acetylnitrene 9 will be much slower than that of singlet methoxycarbonylnitrene 10, despite the greater exoergicity of the former process. Furthermore, Curtius rearrangement

100 of 9 should not be competitive with bimolecular reactions of this nitrene at ambient temperature. This is in good agreement with the reports of Lwowski and co-workers.3,5

101

O O

H H H C N C C H N O H C N O C C O H O C H H H C H H

H C H H

H 13 13-ts 13-p

0.0 9.2 -29.8 0.0 8.2 -20.1

13.8 21.3 -36.8 12.9 30.9 -38.7

O O

H C H H C N H H H C N C H C H C

H O N

O O

H H C H C C H

H H O 12 12-ts 12-p

0.0 16.7 -80.3 0.0 26.2 -71.8

Figure 5.3. The energy surface [ΔG(298K) (kcal/mol)] and isodesmic reactions for Curtius rearrangement of 13 and an isomeric analog 12 at the B3LYP/6- 311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees.

102 We also considered a concerted pathway for Curtius rearrangement from the precursor azides for isocyanate formation (Figure 5.4). CBS-QB3 calculations find two stable conformations of acetyl azide in a manner reminiscent of diazo carbonyl compounds.24 To the best of our knowledge, experimental studies of the conformational isomerism of methoxycarbonyl azide or acetylazide have not been reported; however, experimental and computational studies of fluorocarbonyl azide (F(C=O)N3) have been reported25 and the conformer with syn relationship between the carbonyl and the azide group is favored as we predict for acetylazide. In the gas phase, for acetyl azide, the less stable form (ΔG=4.5 kcal/mol at the CBS-QB3 level) positions the azide group anti to the carbonyl group. The energy difference is slightly reduced in cyclohexane (4.1 kcal/mol) and in acetonitrile (3.2 kcal/mol). CBS-QB3 theory predicts that this conformer will extrude nitrogen (ΔG≠~32 kcal/mol) to form acetylnitrene which can subsequently isomerize to methylisocyanate in the absence of trapping agents.

103

O O o N N 32.8 104 C C + H O H 1.50 N 27.2 31 130o H C N 32.2 C o 110 1.80 140o H 1.59 C H H 27.1 o N H H o 104 N 9-ts 25.9 2.56 N 1.10 C 96 N 1.1 N O 1.81 27.3 28-ts H C H 2.18

N C 27-ts N

H H N 27-28-ts 10.8

10.8 12.0 5.6 4.8 O 4.5 0.0

H C N 0.0 + N N H C O O

o o H 9 31 118 123 1.52 C H 1.50 C o N H 115 N o N C o 117 111 N 1.12 C o 118 1.24 H H 1.24 H H 2.53 N 2.42 1.13 N 28 27

N C O H -61.1 N C O H -61.1 C + N N N N C H -61.9 + H H -61.9 H 9-p 31 9-p 31

(a) Curtius rearrangement of acetyl azide

104 H O o O H 107 C C O o H o N H o O 126 C 142 1.10 H N C + N N o N 1.91 36.4 C O 90 N C O N 2.36 H 1.87 H H 31 32.0 N H 1.10 36.7 1.96 10-ts 27.1 29-ts N 31.6 O H 30-ts 28.6 H C C

H O N N 18.0 N 29-30-ts 14.5 8.6 O

H H + N N 8.9 C C N H O 31 10 1.4 1.5 0.0 O 0.0 H H o 122 O C C o H H N o H O 112 126 1.13 N C C N 1.25 o H O 107 N 2.30 N 1.25 1.13 N 2.22 -11.4 -11.4 29 30

H H -9.9 H H -9.9 N N N C N C + C N N O C O + O H O H 10-p 31 2-p 31

(b) Curtius rearrangement of methoxycarbonyl azide

Figure 5.4. The energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Curtius rearrangement of (a) acetyl azide and (b) methoxycarbonyl azide at the B3LYP/6- 311+G**//B3LYP/6-31G* (red, top) and CBS-QB3 (black, bottom) levels of theory. Distances are shown in Å, and angles are in degrees.

105 The lower energy conformer (by 4.5 kcal/mol) of acetyl azide has a syn relationship between the carbonyl and the azide groups. CBS-QB3 theory predicts that acetyl azide will decompose by concerted migration of the methyl group along with nitrogen extrusion and that the free energy of activation for this concerted process is only

27 kcal/mol. Thus we predict that a free nitrene is not produced upon of acetyl azide. This explains the failure of Lwowski and co-workers to intercept pivaloylnitrene upon pyrolysis of pivaloyl azide.3 In this Curtin-Hammett situation, the relative barriers between concerted and stepwise Curtius rearrangement may, of course, be very sensitive to substituents on the carbonyl group. Indeed, for methoxycarbonyl azide (Figure 5.4b), the stepwise process that generates the free nitrene is favored over the concerted Curtius rearrangement. This is in good agreement with experimental studies of alkoxycarbonyl azides7,8 which do form trappable alkoxycarbonylnitrenes upon thermolysis. Essentially the alkoxy group stabilizes the alkoxycarbonylnitrene and destabilizes the alkoxyisocyanate. These effects are felt by the transition states which lead to these products. Thus, the alkoxycarbonylnitrene is formed in the pyrolysis of the alkoxycarbonyl azide.

This computational result is well precedented as Kaplan, Meloy and Mitchell26 have experimentally demonstrated similar behavior in carbonyl-substituted diazo compounds. Pyrolysis of syn diazo carbonyls leads to Wolff rearrangement in concert with loss of nitrogen, but heating compounds with anti-disposed carbonyl and diazo groups produces the trappable carbene.

The barriers predicted for loss of nitrogen are consistent with the activation energies to thermal decompostion of a number of azides.27 The barriers to decomposition

106 of phenylacetyl azide and benzoyl azide in toluene are 21.5 and 27.1 kcal/mol, respectively.

5.3.3 Intramolecular C-H Insertion

The potential energy surfaces for the intramolecular C-H insertion reactions of 32 and 33 were calculated using DFT methods. The insertion reaction of 33 is about 1.8 kcal/mol more exoergic than that of 32, and it has a smaller free energy of activation barrier by about 0.7 kcal/mol (see supporting information). In each case, the free energy of activation is positive, but the enthalpy of activation is slightly negative. The barriers of these reactions are entirely entropic as found previously for the reaction of dichlorocarbene with alkenes.28,29

O O H C N 2 N H H2C C CH3 H2 32

O O O N ONH H2C C CH3 H2 33

107 5.3.4 Bimolecular Reactions

Energy surfaces for the reactions of singlets 1 and 2 with propane, ethylene, and methanol were calculated by both DFT and CBS-QB3 methods. The results are given in

Figure 5.5-7. Both methods predict that all of these reactions with singlet acetylnitrene 9 and methoxycarbonylnitrene 10 are extremely exoergic. The reactions of 9 with propane are the most exoergic, -68.6 kcal/mol. The alkene cycloaddition reactions are exoergic by about -55 kcal/mol, and the O-H insertion reactions are the least exoergic, with a change of free energy of approximately -40 kcal/mol. Compound 10 reacts more exoergically than 9 in all three reactions. The changes in activation enthalpies are given in Figure 5.5-

7 as well as the changes in free energies.

O

O O H C 1.77 H 1.31 N C H C o H H H 86.8 C H C C 1.10 H H N 1.48 H N 1.53 H H 1.26 o H H C + C 113 C H H H H H C C H H H H H C H C H H H H C C

H H

H H 9 19-ts 19-p 0.0 13.8 -70.5 0.0 18.6 -68.6

Figure 5.5. The energy surface [ΔG (298K) (kcal/mol)] for the reaction of singlet 9 with propane at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees.

108 O O 1.29 1.86 1.77 H H 1.31 O H C 2.30 o 1.49 N H 86.8 1.33 1.22 H C C o H 1.48 C C H 1.28 122 C N + 2.76 C H C 1.26 1.52 N H 2.76 H H 1.40 H C 1.50 H H C 1.46 C H H H H C H H

H 9 22-ts 22-p 0.0 9.5 -54.4 0.0 8.8 -55.4 Figure 5.6. The energy surface [ΔG (298K) (kcal/mol)] for the reaction of singlet 9 with ethylene at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees.

O O O 1.31 1.77 C o C 86.8 H H H H C N H 1.48 H N N 108o 0.97 C C 1.26 H H C H H O H H C H H O + O H 1.42 C H H H H C 9 H H 0.0 17.7 -35.7 0.0 21.3 -34.1

O

O H H H H H C C O H C H C H o N H 1.84 0.97 N H O 1.29 108 O o H O H C 91.9 C C H 1.31 N + 1.42 O H 1.27 1.45 O O H H H H H C C

10 H H

0.0 11.3 -42.9 -44.0 0.0 11.8

Figure 5.7. The energy surface [ΔG (298K) (kcal/mol)] for the reaction of singlet 1 and 2 with methanol at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom) levels of theory. Distances are shown in Å, and angles are in degrees.

109 Indeed, transition state structures could be located for the propane and ethylene reactions with acetylnitrene 9 wherein the target reactant approaches anti to the carbonyl oxygen of the nitrene. A transition state for attack syn to the carbonyl oxygen could be located, but it does not lead to an insertion product (see appendix).13b Instead, the syn transition state leads to the Curtius rearrangement as verified by an IRC calculation. In essence, syn attack corresponds to a concerted Curtius rearrangement with a small effect of solvation by an associated propane or ethylene molecule. However, for methoxycarbonylnitrene 10, transition states for true insertion into the C-H bond of propane and into the C=C bond of ethylene could not be located (see supporting information). Instead, the reactions of 10 with propane and ethylene are barrierless processes and proceed directly to products. Once again, approach of the substrate to the nitrene occurs anti to the carbonyl oxygen.

Two unique transition states for 9 and for 10 could be located for reactions with methanol. These transition states, once again, differ by the orientation of the substrate relative to the carbonyl oxygen of the nitrene, and in general, anti attack is preferred (see

Figure 5.7 and supporting information). The O-H insertion barrier for methoxycarbonylnitrene 10 is lower than that of compound 9 by ~5 kcal/mol. Activation enthalpies are close to zero and are even negative in certain cases. This is reminiscent of the experimental findings of Moss and Turro28 and the theoretical work of Houk and

Rondan29 that demonstrated that the free energy barriers to bimolecular reactions of halocarbenes are dominated by unfavorable entropic changes. In each case calculated here, the bimolecular reactions look like nucleophilic attack on nitrogen.

110 The calculations predict that bimolecular reactions of singlet acetylnitrene and methoxycarbonylnitrene will be faster than the Curtius Rearrangements of these species, particularly at low temperatures where the unfavorable T∆S component of the free energy barrier is small.

These predictions were tested by Mandel and coworkers who studied benzoylnitrene by Laser Flash Photolysis techniques. Mandel et al. reported that benzoylnitrene reacts with pentane and with tetramethylethylene with barriers of -3.2 and -0.06 kcal/mol. These results are consistent with our calculations of the bimolecular reactions of acetylnitrene.

The lifetime of benzoylnitrene in pentane decreases as the temperature decreases.

An Arrhenius treatment of the temperature dependence of the pseudo-first-order decay of benzoylnitrene is shown in Figure 5.8. The apparent activation energy for the disappearance of benzoylnitrene is –3.2 kcal/mol.

111 6.5

6.4

6.3

6.2

6.1

log kobs 6 y = 698.14x + 3.3939 5.9 R2 = 0.9924

5.8

5.7 0.0034 0.0036 0.0038 0.004 0.0042 0.0044 1/T(K)

Figure 5.8. A plot of the log kobs for the decay of benzoylnitrene in pentane versus 1/T (in

K).

An Arrhenius treatment of the data is given in Figure 5.9. The activation energy for the reaction of singlet benzoylnitrene with 1-hexene is –0.06 ± 0.001 kcal/mol.

112 6.416 6.414

6.412 y = 12.817x + 6.3564 6.41 R2 = 0.9449 6.408 6.406

logkobs 6.404 6.402 6.4 6.398 6.396 0.0032 0.0037 0.0042 0.0047 1/T (K)

Figure 5.9. A plot of the log of the absolute rate constant for reaction of benzoylnitrene with 1-hexene in CF2ClCFCl2 versus 1/T (in K).

113 5.4 Conclusions

CBS-QB3 calculations correctly predict that methoxycarbonylnitrene has a triplet ground state, while acetylnitrene has a singlet ground state. B3LYP/6-

311+G**//B3LYP/6-31G*, CCSD(T)/6-311+G**//B3LYP/6-31G* and CCSD(T)/aug- cc-pDVZ//B3LYP/6-31G* calculations incorrectly predict that acetylnitrene has a triplet ground state. Isodesmic reactions reveal that the oxygen atom stabilizes both the singlet and triplet states of the carbonyl nitrenes, but stabilizes the triplet more than the singlet state. Thus, methoxycarbonylnitrene has a triplet ground state. Our calculations are consistent with that of earlier studies12,13 which indicated a significant bonding interactions in the singlet state of acetylnitrene. We find a comparable (but slightly weaker) interaction in the singlet state of methoxycarbonylnitrene.

The Curtius rearrangement of singlet acetylnitrene is much more exoergic than that of methoxycarbonylnitrene. Nevertheless, the Curtius rearrangement of the methoxycarbonylnitrene is predicted to be faster than that of acetylnitrene because the oxygen atom stabilizes the transition state of the Curtius rearrangement of methoxycarbonylnitrene.

Bimolecular reactions of singlet acetylnitrene and methoxycarbonylnitrene with propane, ethylene and methanol have enthalpic barriers that are close to zero and free energy barriers that are completely entropic in nature as found previously for halocarbene singlet states.20,21 Bimolecular reactions of the singlet states of acetylnitrene and methoxycarbonylnitrene will be much faster than the Curtius rearrangement, particularly at low temperatures.

114 These predictions were tested by Mandel et al.31 in a laser flash photolysis (LFP) study of benzoylnitrene. Absolute rate constants of reaction of benzoylnitrene with several substrates were determined at ambient temperature. The pseudo-first-order rate constant of decay of benzoylnitrene in pentane increases as the temperature decreases.

The apparent activation energy is –3.20 ± 0.02 kcal/mol. The activation energy for the reaction of benzoylnitrene with 1-hexene is essentially zero. The LFP data demonstrate that entropic factors control the bimolecular reactivity of benzoylnitrene as predicted by theory.

5.5. References for Chapter 5

1. Lwowski, W. in “Azides and Nitrenes” Scriven, E.F.V, ed. Academic Press, New York, N.Y. 1984, 205

2. (a) Wallis, E.S. Org. Reactions. 1946, 3, 267. (b) Bauer, E. Angew. Chem. Int. Ed. Engl. 1974, 13, 376.

3. (a) Lwowski, W.; Tisue, G. T. J. Am. Chem. Soc. 1965, 82, 4022. (b) Tisue, G. T.; Linke, S.; Lwowski, W. J. Am. Chem. Soc. 1967, 89, 6303. (c) Linke, S.; Tisue, G. T.; Lwowski, W. J. Am. Chem. Soc. 1967, 89, 6308.

4. Skell, P. S.; Woodworth, R. C. J. Am. Chem. Soc. 1956, 78, 4496.

5. Autrey, T.; Schuster, G. B. J. Am. Chem. Soc. 1987, 109, 5814.

6. (a) Lwowski, W.; Mattingly, T. W., Jr. J. Am. Chem. Soc. 1965, 87, 1947. (b) Lwowski, W.; Woerner, F. P. J. Am. Chem. Soc. 1965, 87, 5491. (c) McConaghy, J. S.; Lwowski, W. J. Am. Chem. Soc. 1965, 87, 5490.

7. Wasserman, E. Prog. Phys. Org. Chem., 1971, 8, 319.

8. Sigman, M. E.; Autrey, T.; Schuster, G. B. J. Am. Chem. Soc. 1988, 110, 4297.

9. (a) Fairchild, P. W.; Smith, G. P.; Crosly, D. R.; Jeffries, J. B. Chem. Phys. Lett. 1984, 107, 181. (b) Engelking, P. C.; Lineberger, W. C. J. Chem. Phys. 1976, 65, 4323. 115

10. (a) Travers, M. J.; Cowles, D. C.; Clifford, E. P.; Ellison, G. B.; Engelking, P. C. J. Phys. Chem. 1999, 111, 5349. (b) Kemnitz, C. R.; Ellison, G. B.; Karney, W. L.; Borden, W. T. J. Am. Chem. Soc. 2000, 122, 1098. (c) Barash, L.; Wasserman, E.; Jager, W. A. J. Chem. Phys. 1967, 89, 3931. For a correction, see Ferrante, R. F. J. Chem. Phys. 1987, 86, 25. (d) Radziszewski, J. G.; Downing, J. W.; Wentrup, C.; Kaszynski, P.; Jawdosiuk, M.; Kovacic, P.; Michl, J. J. Am. Chem. Soc. 1985, 107, 2799.

11. (a) Hrovat, D. A.; Waali, E. E.; Borden, W. T. J. Am. Chem. Soc. 1992, 114, 8698. See also Castell, O.; García, V. M.; Bo, C.; Caballolo, R. J. Comput. Chem. 1996, 17, 42-48. (b) Gritsan, N. P.; Zhu, Z.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc. 1999, 121, 1202.

12. (a) Gritsan, N. P.; Pritchina, E. A. Mendeleev Comm. 2001, 94. (b) Shapley, W. A.; Bacskay, G. B. J. Phys. Chem. A 1999, 103, 6625. (c) Faustov, V. I.; Baskir, E. G.; Biryukov, A. A. Russ. Chem. Bull. Int. Ed. 2003, 52, 2328.

13. (a) Pritchina, E. A.; Gritsan, N. P.; Maltsev, A.; Bally, T.; Autrey, T.; Liu, Y.; Wang, Y.; Toscano, J. P. Phys. Chem. Chem. Phys. 2003, 5, 1010. (b) Pritchina, E. A.; Gritsan, N. P.; Bally, T. Russ. Chem. Bull. Int. Ed. 2004, 53, 0000.

14. Ziegler, T. Chem. Rev. 1991, 91, 651.

15. Jensen, F. Introduction to Computational Chemistry; Wiley: Chichester, 1998.

16. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

17. (a) Fukui, K. Acc. Chem. Res. 1981, 14, 363; (b) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523.

18. (a) Cizek, J. Adv. Chem. Phys. 1969, 14, 35. (b) Barlett, R. J. J. Phys. Chem. 1989, 93, 1963. (c) Raghavachari, K.; Trucks, G. W.; Pople, J. A.; Head-Gordon, M. Chem. Phys. Lett. 1989, 157, 479.

19. Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007.

20. Montgomery, J. A., Jr.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys. 1999, 110, 2822.

21. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

22. Frisch, M. J. et al. Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

23. Scott, A. P.; Platz, M. S.; Radom, L. J. Am. Chem. Soc. 2001, 123, 6069.

116

24. Regitz, M. “Diazo Compounds: Properties and Synthesis” Academic Press, New York, N. Y., 1986.

25. Mack, H; Della Vedova, C. O.; Willner, H. J.Mol. Struct.1993, 291, 197

26. (a) Kaplan, F.; Meloy, G. K. J. Am. Chem. Soc. 1966, 88, 950. (b) Kaplan, F.; Mitchell, M. L. Tetrahedren Lett. 1979, 759.

27. (a) Abramovitch, R. A.; Kyba, E. P. “The Chemistry of the Azido Group.” S. Patai, ed., Interscience, New York, NY 1971. (b) L’Abbe, G.; Chem. Rev., 1969, 69, 345

28. Turro, N. J.; Butcher, J. A.; Moss, R. A.; Guo, W.; Munial, R. C.; Fedorynski, M. J. Am. Chem. Soc. 1980, 102, 7576.

29. Rondan, N. G.; Houk, K. N.; Moss, R. A. J. Am. Chem. Soc. 1980, 102, 1770.

30. Reed, A. E.; Weinhold, F.; Curtiss, L. A. Chem. Rev. 1988, 88, 899.

31. Liu, J.; Mandel, S.; Hadad, C. M.; Platz, M. S. J. Org. Chem. 2004, 69, 8583.

117

CHAPTER 6

SULFONYLNITRENES AND AZIDES

6.1 Introduction

Thermolysis of sulfonyl azides leads to extrusion of molecular nitrogen.1 It is clear from chemical trapping studies and the formation of C-H insertion products that singlet nitrene intermediates are produced under these conditions, where R is alkyl2 or aryl3 and SH is a solvent trap.

O O Δ SH O RSN H 3 RSN RSN O -N2 S O O

An isomeric reactive intermediate, a sulfurylimine, formed by formal rearrangement of the nitrene can be intercepted with methanol upon photolysis of benzenesulfonyl azide.3a

O OCH3 O O S hv CH3OH O RSN3 RNS RN O O H

118 When benzenesulfonyl azide is pyrolyzed in the vapor phase at 650 oC, azobenzene is formed.4 This implies that the azide forms sulfurylaniline upon pyrolysis at elevated temperatures and that this species extrudes to form phenylnitrene which subsequently dimerizes.4

O Δ O O S N3 S N NS -N O 2 O O

NN N+SO2

A trace amout of azobenzene is formed when benzenesulfonyl azide is decomposed in refluxing cyclohexane.5 Stronger evidence for the pseudo Curtius

Rearrangement was reported by Abramovitch and Holcomb who decomposed mesitylenesulfonyl azide in dodecane and found a 20.7% yield of 2,4,6 trimethylaniline formed presumably from hydrolysis of a sulfurylaniline.6

The photochemistry of aromatic sulfonyl azides has been studied by chemical and physical methods by Maloney and co-workers.7 Maloney and co-workers detected triplet p-toluenesulfonylnitrene at 77K by UV-Vis and ESR spectroscopy8 upon photolysis of p- toluene azide in EPA glass. The triplet nitrene was also detected in solution phase by laser flash photolysis techniques. The photochemistry of sulfonyl azides immobilized in inert gas, cryogenic matrices has been communicated by Sheridan et al.9 119 One can ask the question whether thermally induced nitrogen extrusion and rearrangement to sulfurylimine proceeds stepwise or in concert. In the previous section we explored the question of stepwise vs concerted rearrangement for the Curtius10 and

Wolff11 rearrangement.

For the sulfonyl azides, there are additional complexities as any generated sulfur dioxide can decompose sulfonamide reaction products, and in the solution phase, it has been noted that a radical scavenger2a,b inhibits the formation of sulfur dioxide upon pyrolysis of alkylsulfonyl azides. Thus, sulfur dioxide may be formed by more than one mechanistic pathway. The thermolysis of sulfonyl azides is clearly quite complicated mechanistically. These questions have stimulated this computational study of sulfonyl azides and their corresponding nitrenes.

6.2 Computational Methods

Density functional theory (DFT)12 and CBS-QB313 methods have been applied in this study. For the DFT calculations, all of the geometries of interest were completely optimized at the B3LYP/6-31G* level. At the same level, analytical vibrational frequencies were calculated to verify a minimum energy structure as well as to provide the zero point vibrational energy corrections, which were scaled by a factor of 0.9806.14

The thermal and entropic contributions to the free energies were also obtained from the

B3LYP/6-31G* vibrational frequency calculations. Single-point energy calculations of all stationary points were performed at the B3LYP/6-311+G** level using the corresponding B3LYP/6-31G* geometries. Six Cartesian d functions were used for these

120 calculations. UV spectra were computed with time-dependent DFT15 methods at the

B3LYP/6-31+G**//B3LYP/6-31G* level. In some cases, CASSCF/6-31G*16 and

CASPT2/6-31G*17 calculations were performed using the MOLCAS18 program.

Complete Basis Set (CBS)13 methods have been applied to increase the accuracy of the results. Some of the stationary points, including transition states, were fully characterized at the CBS-QB3 level. The energies provided are CBS-QB3 enthalpies and free energies at 298.15 K. All DFT and CBS-QB3 calculations were performed with

Gaussian 9819 at the Ohio Supercomputer Center. Unless noted otherwise, energies are presented as free energies at 298 K.

6.3 Results

6.3.1 Methylsulfonylnitrene Singlet-Triplet energy separation

Density Functional Theory (DFT) calculations (B3LYP/6-311+G**//B3LYP/6-

31G*) predict that triplet methanesulfonyl nitrene (34T) is 17.8 kcal/mol more stable than the corresponding singlet state (34S) in the gas phase (Figure 6.1). CBS-QB3 calculations predict that the energy separation is only 8.9 kcal/mol. The CBS-QB3 value is considered more reliable because the CBS-QB3 method is based on a B3LYP geometry, CCSD(T) single-point energy for electron correlation, and an extrapolation to the infinite basis set limit. ESR spectroscopy has demonstrated that phenylsulfonylnitrene has a triplet ground state,8 which is consistent with our calculations. To our knowledge, triplet methylsulfonylnitrene has not been detected by matrix ESR spectroscopy.

121 The experimentally determined singlet-triplet energy gaps of (NH),20

21 22 methylnitrene (CH3N) and phenylnitrene (36.0, 31.2, and 18.0 kcal/mol, respectively) are much larger than that of methylsulfonylnitrene 34T. The calculated singlet-triplet gap of 34T is similar to that calculated for methoxycarbonylnitrene (CH3OC(=O)N, 5.7 kcal/mol). 10

The singlet state of phenylnitrene is open shell and has a quinoidal geometry.23

N N

The singlet states of acetylnitrene (CH3C(=O)N) and methoxycarbonylnitrene

(CH3OC(=O)N) are both closed shell to take advantage of a bonding interaction between the acyl oxygen and nitrogen.24

O O 1.84 CH3O 1 CH3O N N

This interaction is so stabilizing in acetylnitrene that the singlet is the ground state by 4.0 kcal/mol. The singlet state preference for acetylnitrene is supported by chemical- trapping, time-resolved infrared, time-resolved UV, and matrix isolation spectroscopic studies. 10,24

The O-C-N bond angles in singlet acetyl- and methoxycarbonylnitrene are 87o and 92o, respectively.10 There is a corresponding N-O bonding interaction in singlet methylsulfonylnitrene. The calculated CBS-QB3 geometries of 34S and 34T are given in

122 Figure 6.1. We were unable to obtain an open-shell wave function at the B3LYP level, so we estimated the energy difference between the closed-shell and open-shell singlet states at the CASSCF and CASPT2 levels. The closed-shell singlet state of 34S is 26.0 and 21.1 kcal/mol lower in energy than the open-shell singlet state at the (bottom-of-the-well)

CASSCF and CASPT2 levels of theory, respectively.

O O O 1.43 1.47 H O S 2.53 1.79 S o 107o 71.9 1.81 H 1.80 C 1.56 1.70 N C H N H H H 34S 34T

Figure 6.1. The CBS-QB3 optimized geometries of singlet (34S) and triplet (34T) methylsulfonylnitrene. Bond distances are in Angstroms, and bond angles are in degrees.

Because singlet sulfonylnitrenes have closed-shell electronic structures, we expect they will rapidly intersystem cross to the lower energy triplet state by an allowed spin- orbit25 coupling process. The lifetime of singlet ethoxycarbonylnitrene, which also has a closed-shell singlet electronic structure is less than 10 ns.26 We expect the lifetime for singlet methylsulfonylnitrene to be comparable. Singlet nitrene 34S is predicted by TD-

DFT27 calculations to have weak absorption bands at 517 and 218 nm and appears to be a poor candidate for time-resolved UV-Vis spectroscopy (see supporting information).

Triplet 34T is predicted to absorb strongly at 309 nm and should be detectable by this technique.

123 Both 34S and 34T are predicted to have prominent vibrational bands around 1260 cm-1 due to a S-O asymmetric stretching mode and will be difficult to distinguish (see supporting information). Triplet nitrene 34T is predicted to have a unique band at 1070 cm-1 due to an S-O symmetric stretching mode. This mode is absent in 34S and could be of use in TRIR spectroscopy in detecting 34T in solution.

An isomeric equation indicates that the N-O bonding interaction in acetylnitrene is more substantial than for methylsulfonylnitrene and energetically favors the carbonylnitrene by ~18 kcal/mol. The S-N and S-O bonds of the sulfonylnitrene are longer than the C-N and C-O bonds of acetylnitrene. Consequently, the N-O bond is longer and weaker in methylsulfonylnitrene 34S than in acetylnitrene.

N

H H H H O H H C C C O N C H C ΔG=18.8 (17.8) kcal/mol C H S S H H C O C O O O H H H B3LYP (CBS-QB3)

H

6.3.2 Intramolecular C-H insertion reactions of singlet alkylsulfonylnitrenes

Intramolecular cyclization reactions of sulfonylnitrenes are well known.1-3

Unsurprisingly, DFT and CBS-QB3 caclulations predict that the cyclization of singlet 1- butylsulfonylnitrene is facile (Figure 6.2).

124

O

O

S

N H

C H H

H H C C

H C H H

H TS(35-36) 2.7 (ΔH =0.7 )

0.0 4.1 (ΔH =2.4 ) 0.0

O

O N S

H

C O H -88.2 H H H O H C S -89.4 C H N C H H H C H H H C H

35 H C C H H H 36

Figure 6.2. The energy surface (ΔG, kcal/mol) for intramolecular cyclization of 1- butylsulfonylnitrene at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom, underlined) levels in the gas phase. Enthalpies of activation are shown in parentheses.

125 Our calculations predict that long chain alkylsulfonylnitrene singlet states will have very short lifetimes in solution (τ < 10 ns) because of their rapid cyclization.

6.3.3 Pseudo-Curtius Rearrangements

Singlet methylsulfonylnitrene can suffer two highly exoergic rearrangement reactions. The first process, formation of the sulfurylimine (eq 11), is known thermally for phenyl- and mesitylsulfonylnitrene at elevated temperatures and upon photochemical activation of arylsulfonyl azides.3-6

O O S O CH N S (11) CH3 N 3 O 34S 37

It is also possible to cleave the formal S-O single bond as shown in equation (12)

O O O S S O CH (12) 3 N CH3 N 34S 38

Although reaction (12) is predicted to be faster than reaction (11) (see Figure 6.3), it is actually slightly less exoergic. Fragmentation of 38 to methyl radical, SO and NO is endoergic by 20.4 kcal/mol (Figure 6.3). Both rearrangement reactions (11) and (12) are slow relative to an intramolecular C-H insertion reaction of a long-chain, singlet alkylsulfonylnitrene and are slow relative to bimolecular reactions of singlet

126 methylsulfonylnitrene (vide infra) and should not consume relaxed nitrene 34S in solution in hydrocarbon solvents at temperatures near ambient.

It is not clear whether or not singlet methylnitrene (H3C-N) exists in a shallow potential energy minimum or isomerizes to methyleneimine (H2C=NH) in a barrierless process.28 Triplet methylnitrene can be studied more readily than the corresponding singlet by computational methods. Cleavage of 37 to SO2 and triplet methyl nitrene is endoergic by 6.8 kcal/mol. As the singlet-triplet energy separation of methylnitrene is

21 31.2 kcal/mol, the fragmentation of 37 to singlet methylnitrene and SO2 (reaction 13) is deduced from the experimental and computational data to be very endoergic (38.0 kcal/mol) and probably is not an important reaction in solution at temperatures near ambient. It is possible that hydrogen migration and fragmentation of singlet sulfonylnitrene 34S are concerted (reaction 14). However, our calculations indicate that the barrier to this process is 48.6 kcal/mol, and this reaction is not likely to proceed in solution.

O + (13) CH3 N S CH3 N SO2 O

O + (14) CH3 N S CH2 NH SO2 O

127

O

O S H

C N H H

TS(34S-37)

25.1

O 21.4 S H

H N C O

O O H S

H TS(34S-38)

C N

H H

O -14.1 0.0 S -34.1 O O H CH3 + SO + NO N C H

O H S H 38 C N H -59.5 H

34S -54.5 3 CH3- N + SO2 -62.7

O

H

S -102.8 H C

H H C=NH + SO N O 2 2 37

Figure 6.3. The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of methylsulfonylnitrene at the B3LYP/6-311+G**//B3LYP/6-31G* level in the gas phase.

128

6.3.4 Bimolecular reactions of singlet methylsulfonylnitrene

The reactions of 34S with propane, ethylene and methanol were calculated. The enthalpies of activation are close to zero and are slightly negative in the first two reactions due to the thermal correction to 298 K. Thus, we predict negative activation energies in these reactions as was recently observed in the reactions of benzoylnitrene with and alkenes.10 The free energy barriers are positive and entirely entropic in origin. The free energy barriers are 2-6 kcal/mol lower than the corresponding barriers for acetylnitrene.10

129

H O H H O H H H O C C O C S H H H H H H H H S N C C C O C H C S + H H H H H H C H H H H O N C C C N H H H H H H H H

0.0 9.3 (ΔH = -0.3 kcal/mol) -81.6

0.0 6.7 -87.2

O O O H O H H S H O H H S H C C S C + C H H O C H H N C N C N H H C H H H C H H H H H

0.0 7.2 (ΔH = -0.7 kcal/mol) -69.2

0.0 5.4 -74.5

H O O O O H H H C H S O H H S H O S N O + H C O O N C N H H H C C H H H H H C H H

H H

0.0 11.8 (ΔH = 0.2 kcal/mol) -45.3

0.0 11.5 -46.3 Figure 6.4. The energy surface (ΔG, kcal/mol) for bimolecular reactions of singlet methylsulfonylnitrene (1S) at the B3LYP/6-311+G**//B3LYP/6-31G* (top) and CBS- QB3 (bottom, underlined) levels in the gas phase.

130 6.3.5 Bimolecular reactions of triplet methylsulfonylnitrene 1T

Characteristic hydrogen-atom abstraction and addition reactions of triplet methylsulfonylnitrene 34T were calculated. Both reactions are exoergic by over 20 kcal/mol at the B3LYP/6-311+G**//B3LYP/6-31G* level, and therefore should proceed rapidly in solution at ambient temperature.

O O CH3 CH3 3 H C + H3CSN + 2 H3CSNH CH ΔG = -2.6 kcal/mol CH CH O 3 O 3

O O 3 H CCH ΔG = -7.2 kcal/mol H3CSN + 2 2 H3CSN CH2 O O CH 2

The reactions of triplet 34T with propane are calculated to be more exoergic than the corresponding reactions of triplet methyl and phenylnitrene and for this reason may proceed more rapidly. This is consistent with the laser flash photolysis studies of

Maloney and co-workers. 7

3 H3C N + CH3CH2CH3 H CNH + 3 CH3CHCH3 ΔG = 10.2 kcal/mol

3 N + CH3CH2CH3 NH + CH3CHCH3 ΔG = 6.3 kcal/mol

O O + CH CH CH + CH O 3 3 2 3 CH3CHCH3 ΔG = 1.6 kcal/mol 3 N CH3O NH

131 6.3.6. Is thermal rearrangement concerted with nitrogen extrusion?

As discussed previously carbonyl azides can exist in two conformations.29 When the carbonyl and azide groups are syn to each other, Curtius rearrangement is concerted with nitrogen extrusion.

O N N CH NCO+ N CH3 N 3 2

When these groups are held anti, nitrene formation is favored.

O O + N CH3 N CH3 N 2 N N

Conformational interconversion is rapid, of course. Our previously mentioned calculations predicted that acetyl azide preferentially fragments via the syn conformation.10 The Curtius Rearrangement proceeds in concert with nitrogen extrusion, and we predicted that free nitrenes are not formed upon thermolysis of acetyl azide,10 in good agreement with experiment.30

Different behavior is predicted for methanesulfonyl azide. Theory predicts that stepwise decomposition of methanesulfonyl azide is clearly preferred. This explains why pyrolysis of sulfonyl azides leads to insertion reactions which are characteristic of the intermediacy of a singlet nitrene2,3 while pyrolysis of acyl azides leads efficiently to isocyanate.30 Singlet nitrene insertion products are formed in poor yields on photolysis of many aryl sulfonyl azides.3 This observation implies that the concerted migration is more prevalent upon photochemical activation than thermolysis.

132 O O

S 1.95 N N o 75.6 O H N O C H S H 2.12

H C N TS(39-37) H H 1.90 N

45.0 N (46.5) TS(39-34S) 36.7 (37.4)

21.5 (31.4)

0.0 O

(0.0) O H S + N N

C N O H

H O S H 34S

C H N N N H

39 -41.2 (-30.6)

O

H + N N S H C

H N O 37

Figure 6.5. The energy surface [ΔG, kcal/mol (top) and ΔH, kcal/mol (bottom)] for stepwise versus concerted decomposition of methanesulfonyl azide (39) at the B3LYP/6- 311+G**//B3LYP/6-31G* level in the gas phase.

133 6.3.7 Other modes of thermal decomposition of methanesulfonyl azide

Because radical inhibitors quench the formation of sulfur dioxide,2 we considered other mechanisms of decomposition of methanesulfonyl azide. The weakest bond (least positive ΔH) of methanesulfonyl azide is the N=N bond whose fragmentation leads to nitrene. However, the sulfur-nitrogen bond is also a weak bond and may fragment in competition with nitrene formation. This could initiate radical chain decomposition of methanesulfonyl azide in the absence of a radical . Our calculations predict that reaction (15), homolytic cleavage of the S-N bond, is competitive with nitrogen extrusion and sulfonylnitrene formation at the B3LYP/6-311+G**//B3LYP/6-31G* level.

ΔG (kcal/mol) ΔΗ (kcal/mol) O

CH3 S + N3 27.2 36.0 (15) O O S N O CH3 + 3 46.4 59.5 (16) O CH3 S N3 O CH3 ++SO2 N3 30.0 50.6 (17) O

CH3 S N + N2 O 21.9 31.4 (18)

6.3.8 Oxygen, sulfur and nitrogen substituent effects

Our calculations indicate that methanesulfonyl azide is a good thermal source of sulfonylnitrenes in agreement with experiment.2,3 To find other efficient sources of nitrenes, we decided to modify the structure of the azide and evaluated the methoxy (40), thiomethoxy (43) and N,N-dimethylamino (46) substituted sulfonyl azides. 134 As shown in Figures 6-8, DFT calculations predict that nitrenes 7, 43 and 46 all have triplet ground state multiplicities. The DFT singlet-triplet energy separations are all slightly larger than those of methylsulfonylnitrene 34.

All of these nitrenes isomerize to sulfurylimines over reasonable barriers. The relaxed sulfonylnitrenes 40S and 46S will likely undergo bimolecular reactions in solution faster than they will isomerize. Isomerization of thiomethoxysulfonylnitrene 43S to a sulfurylimine is predicted to be substantially faster than 40S or 46S, and this pathway may operate in solution.

The Pseudo-Curtius Rearrangement of methoxysulfonylnitrene 40S is substantially less exoergic (-35.1 kcal/mol) than that of methylsulfonylnitrene 34S (-62.7 kcal/mol), thiomethoxy analog 43S (-57.1 kcal/mol) or N,N-dimethylamino compound

46S whose isomerization product immediately fragments to form a highly stable, ground state singlet amino nitrene (47S).

H3C O N NS H3C N N +SO H3C O 2 H3C

We considered the possibility that the decomposition of methoxysulfonyl azide 48 might prefer stepwise decomposition to form nitrene 40S as compared to concerted rearrangement and nitrogen extrusion as compared to form 49 (Figure 6.9). Stepwise decomposition to form methoxysulfonylnitrene is preferred to decomposition but the preference is less pronounced than for methanesulfonyl azide. Sulfurylimine formation is more likely upon pyrolysis of methoxysulfonyl azide than methanesulfonyl azide.

135 O

S 25.4 O N O

C H H H

TS(40S-42) O 16.2 O S

O N H

C H

H TS(40S-41)

O

O O 0.0 H S

C N O 1 H S H O N

O 41

C H H -35.1 40S H -21.3 -56.2

3 O -63.0 CH3 + SO + NO H 2

S O O C

H H O H S N H C O

O N 40T H 42

Figure 6.6. The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of methoxysulfonylnitrene (40) at the B3LYP/6-311+G**//B3LYP/6-31G* level in the gas phase.

136

O

H S 13.7 N CH

H O S

O O TS(43S-45) 9.0 S

H N S

C H

H

TS(43S-44)

O 1

S O N 0.0 S O S S

H C H N C O H -44.0 H H H 43S 44

-57.1 CH3 + S2O + NO -19.7

O 3 H -65.5 C H S H O H O S H N C H O 43T S N S 45

Figure 6.7. The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of thiomethoxysulfonylnitrene (43) at the B3LYP/6-311+G**//B3LYP/6-31G* level in the gas phase.

137

O

H S

N C N H O H 25.2

C H H H

TS(46S-48) O H

H C 12.7 S O H N

N

H C

H H TS(46S-47)

O H 1 H C 0.0 S H O N

N

HC O H S H H

N 46S C H N O H -34.8 C -21.6 H H

H O 3 H 48 H CH3 ++CH3NSO NO C S -57.3 N O H

H N C H

H S H N O

-73.8 C N O 46T H H

C H ΔG (kcal/mol) H B3LYP/6-311+G** //B3LYP/6-31G* 47

Figure 6.8. The energy surface (ΔG, kcal/mol) for pseudo-Curtius rearrangement of N,N- dimethylaminosulfonylnitrene (46) at the B3LYP/6-311+G**//B3LYP/6-31G* level in the gas phase.

138

O O

O S O

N S N

N O N H H H O C C N

H N H H

TS(49-40) 38.6 TS(49-40S) (39.2) 34.3 (34.7)

21.8 (31.3)

O

S + N N O N

0.0 O (0.0) C H H

H O

H 40S O H S C

O H N

N

-13.3 N (-3.1) 49

O

O H S + N N C N O H

H 41

Figure 6.9. The energy surface [ΔG, kcal/mol (top) and ΔH, kcal/mol (bottom)] for stepwise versus concerted decomposition of methoxysulfonyl azide (49) at the B3LYP/6- 311+G**//B3LYP/6-31G* levels in the gas phase.

139 Other bond scissions of methoxy (49), thiomethoxy (50), and N,N- dimethylaminosulfonyl (51) azides were considered (see Table 6.1). In the case of the methoxy and dimethylamino substituted azides 49 and 51, nitrogen extrusion is clearly favored relative to the other possible bond fissions, although these azides will also be prone to free radical chain decomposition. Sulfur substituted azides 50 has a very weak

S-N bond, and it will be more prone to radical fragmentation reactions even in the presence of radical scavengers. Thus, our calculations predict that pyrolysis of methoxy and dimethylamino sulfonyl azides 49 and 51 (two infrequently studied types of azides) should efficiently produce the corresponding nitrenes.

140

O

XSN+N2 NN O O O SN XSN3 XS + N3 O XS O O

X + S N3 O

ΔH ΔG X (kcal/mol) (kcal/mol) N---N S---N X---S N---N S---N X---S

CH3 (39) 31.4 36 59.5 21.5 27.2 46.4

OCH3(49) 31.3 36.6 53.2 21.8 27.9 39.5

SCH3(50) 28.3 29.1 37.9 18.4 19.7 25.2

N(CH3)2(51) 30.8 35.3 49.9 21.1 26.5 35.2

Table 6.1. Decomposition modes of sulfonyl azides. (As calculated at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory. Energies are in kcal/mol and are enthalpies and free energies at 298 K.)

141 6.4 Conclusions

CBS-QB3 calculations predict that triplet methylsulfonylnitrene lies 8.9 kcal/mol below the lowest singlet state. The singlet-triplet separation is substantially smaller than that of imidogen (36 kcal/mol),20 methylnitrene (31.2 kcal/mol),21 and phenylnitrene (18 kcal/mol).22 Unlike pheneylnitrene which has an open-shell singlet state, the most stable singlet of methylsulfonylnitrene has a closed-shell singlet structure to take advantage of nitrogen-oxygen bonding. This is very reminiscent of the structure of acetylnitrene.10,24

However, acetylnitrene has a singlet ground state due to stronger N-O bonding than is present in methylsulfonylnitrene.

Methylsulfonylnitrene singlet is predicted to rapidly insert into C-H and O-H bonds and to add to alkenes. These reactions are expected to have negative or zero activation energy. These reactions are also expected to be more rapid than the isomerization of singlet methylsulfonylnitrene to methylsulfurylimine.

Acyl azides10 thermally extrude nitrogen in concert with alkyl migration to form isocyanates without the intervention of a nitrene intermediate. Methanesulfonyl azide prefers stepwise thermal decomposition to form singlet nitrene rather than concerted pseudo-Curtius Rearrangement to form methylsulfurylimine.

The N=N bond of methanesulfonyl azide is the weakest bond in the molecule.

However, the S-N bond is only 5.7 kcal/mol stronger than the N=N bond. Thus, the S-N bond may cleave at a rate competitive with nitrogen extrusion and initiate a radical chain process leading to the elimination of sulfur dioxide.

142 Triplet methylsulfonylnitrene has more favorable spectroscopic properties (309 nm, 1069 cm-1) than the singlet nitrene and should be more readily detected by time resolved spectroscopic methods.

We hypothesize, based on these calculations, that methoxy and dimethylamino substituted sulfonyl azides may be efficient thermal sources of nitrenes.

6.5 References for Chapter 6

1. (a) Abramovitch, R. A.; Davis, B. A. Chem. Rev. 1964, 64, 149. (b) Lwowski, W. in Azides and Nitrenes, Scriven, E. F. V., ed.; Academic Press, New York, N. Y. 1984, pp205-206.

2. (a) Breslow, D. S.; Edwards, E. I.; Linsay, E. C.; Omura, H. J. Am. Chem. Soc. 1976, 98, 4268. (b) Breslow D. S.; Sloan, M. F.; Newburg, N. R.; Renfrow, W. B. J. Am. Chem. Soc. 1969, 91, 2273. (c) Sloan, M. F.; Renfrow, W. B.; Breslow, D. S. Tetrahedron Lett. 1964, 2905.

3. (a) Lwowski, W.; Scheiffele, E. J. Am. Chem. Soc. 1963, 87, 4359. (b) Abramovitch, R. A.; Azogu, C. I.; Mcmaster, I. T. J. Am. Chem. Soc. 1969, 91, 1219. (c) Ayyangar, N. R.; Bambal, R. B.; Nikalje, D. D.; Srinivasan, K. V. Can. J. Chem. 1985, 63, 887. (d) Abramovitch, R. A.; Knavs, G. N. J. Org. Chem. 1975, 40, 883. (e) Horner, L.; Christmann, A. Chem. Ber. 1963, 96, 388.

4. Reichle, W-T. Inorg. Chem. 1964, 3, 402.

5. Balabanov, G. P.; Yu, I.; Dergunov; Gal’perin, V. A. J. Org. Chem. (USSR) 1966, 2, 1797

6. Abranovitch, R. A.; Holcomb, W. D. Chem. Comm. 1969,1298.

7. Garay, J-C.; Maloney, V.; Marlow, M.; Small, P. J. Phys. Chem. 1996, 100, 5788.

8. (a) Smolinsky, G.; Wasserman, E.; Yager, W. A. J. Am. Chem. Soc. 1962, 84, 3320. (b) Moriarty, R. M.; Rahman, M.; King, G. J. J. Am. Chem. Soc. 1966, 88, 842. (c) Wasserman, E. Prog. Phys. Org. Chem. 1971, 8, 319.

9. Sheridan, R. S.; Rempala, P. A nitrene to nitrene rearrangement: Photochemistry of benzenesulfonyl azide. Book of Abstracts, 217th ACS National Meeting, Anaheim, Calif., March 21-25 (1999).

143 10. Liu, J.; Mandel, S.; Hadad, C. M.; Platz, M. S. J. Org. Chem. 2004, 69, 8583.

11. Liu, J.; Popik, V.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc., submitted.

12. (a) Labanowski, J. W.; Andzelm, J. Density Functional Methods in Chemistry; Springer: New York, 1991. (b) Parr, R. G.; Yang, W. Density Functional Theory in Atoms and Molecules; Oxford University Press: New York, 1989.

13. Montgomery, J. A., Jr.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys. 1999, 110, 2822.

14. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

15. Stratmann, R. E.; Scuseria, G. E.; Frisch, M. J. J. Chem. Phys. 1998, 109, 8218.

16. Roos, B. O. Int. J. Quantum Chem. Symp. 1987, 15, 175.

17. (a) Anderson, K.; Malmqvist, P.-Å.; Roos, B. O.; Sadlej, A. J.; Wolinski, K. J. Phys. Chem. 1990, 94, 5483. (b) Anderson, K.; Malmqvist, P.-Å.; Roos, B. O. J. Chem. Phys. 1992, 96, 1218. (c) Anderson, K.; Roos, B. O. Int. J. Quantum Chem. 1993, 45, 591.

18. Andersson, K. et al. MOLCAS Version 5; Lund University: Sweden, 2000.

19. Frisch, M. J. et al Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

20. Gilles, A.; Mesanet, J.; Vermeil, C. Chem. Phys. Lett. 1974, 23, 346

21. Travers, M. J.; Cowles, D. C.; Clifford, E. P.; Ellison, G. B.; Engelking, P. C. J. Chem. Phys. 1999, 111, 5349.

22. Travers, M. J.; Cowles, D. C.; Clifford, E. P.; Ellison, G. B. J. Am. Chem. Soc. 1992, 114, 8699.

23. Borden, W. T.; Gritsan, N. P.; Hadad, C. M.; Karney, W. L.; Kemnitz, C. R.; Platz, M. S. Acc. Chem. Res. 2000, 33, 765.

24. Autrey, T.; Schuster, G. B. J. Am. Chem. Soc. 1987, 109, 5814. (b) Wasserman, E. Prog. Phys. Org. Chem. 1971, 8, 319. (c) Pritchina, E. A.; Gritsan, N. P.; Maltsev, A.; Bally, T.; Autrey, T.; Liu, Y.; Wang, Y.; Toscano, J. P. Phys. Chem. Chem. Phys. 2003, 5, 1010. (d) Pritchina, E. A.; Gritsan, N. P.; Bally, T. Russ. Chem. Bull. Int. Ed. 2004, 53, 0000. (e) Hayashi, Y.; Swern, D. J. Am. Chem. Soc. 1973, 95, 5205. (f) Puttner, R.; Hafner, K. Tetrahedron Lett. 1964, 3119. (g) Semenov,

144 V. P.; Studenikov, A. N.; Bespalov, A. D.; Ogloblin, K. A. J. Org. Chem. USSR ( Engl. Transl.) 1977, 12, 2052. (h) Inagaki, M.; Shingaki, T.; Nagai, T. Chem. Lett. 1981, 1419. (i) Inagaki M.; Shingaki, T; Nagai, T. Chem. Lett. 1982, 9.

25. (a) Salem, L.; Rowland, C. Angew. Chem. Int. Ed. Engl., 1972, 11, 92. (b) Michl, J. J. Am. Chem. Soc. 1996, 118, 3568 (c) Zimmerman, H. E.; Kutataladze, A. G. J. Org. Chem. 1996, 61, 6926. (d) Kita, F.; Nau, W. M.; Adam, W.; Wirz, J. J. Am. Chem. Soc. 1999, 121, 9265. (e) Johnson, W. T. G.; Sullivan, M. B.; Cramer, C. J. Int. J. Quant. Chem. 2001, 85, 492.

26. Buron, C.; Platz, M. S. Org. Lett. 2003, 5, 3383.

27. Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys. 1998, 108, 4439.

28. Kemnitz, C. R.; Ellison, G. B.; Karnoy, W. L.; Borden, W. T. J. Am. Chem. Soc. 2000, 122, 1098.

29. Mack, H; Della Vedova, C. O.; Willner, H. J. Mol. Struct. 1993, 291, 197.

30. (a) Lwowskii, W.; Tisue, G. T. J. Am. Chem. Soc. 1965, 82, 4022. (b) Tisue, G. T.; Linke, S.; Lwowskii, W. J. Am. Chem. Soc. 1967, 89, 6303. (c) Linke, S.; Tisue, G. T.; Lwowskii, W. J. Am. Chem. Soc. 1967, 89, 6308.

145

CHAPTER 7

THE REACTION OF TRIPLET NITRENES WITH OXYGEN

This chapter is reproduced with permission from Organic Letters, volume 7, page 549. Copyright 2005, American Chemical Society.

7.1 Introduction

Triplet aromatic carbenes react with oxygen at rates that approach diffusion control.1 Triplet aromatic nitrenes react with oxygen much more slowly than the analogous carbene processes.2 Gritsan and Pritchina have reported that triplet para- aminophenylnitrene reacts with oxygen with a rate constant of (4.5±1.2)×106 M-1s-1 in and that the corresponding rate constant for the para nitro analog is (0.8±0.1)×106

M-1s-1 under the same conditions.2a Similar results were obtained with para-tolylnitrene.

Liang and Schuster2b reported that triplet para–nitrophenylnitrene reacts with oxygen with a rate constant smaller than 2×105 M-1s-1 in acetonitrile. To investigate the origin of the different reactivity of triplet carbenes and nitrenes with oxygen, Density Functional

Theoretical (DFT) and ab initio molecular orbital calculations were performed.3,4

146 7.2 Computational Methods

For the DFT calculations, all structures of interest are completely optimized at the

B3LYP/6-31G* level. Analytical vibrational frequencies were calculated at the same level for each stationary point to verify a minimum energy structure and to provide zero- point vibrational energy corrections, which were scaled by a factor of 0.9806.5 Each transition state was verified to connect to the respective reactant and product by careful optimization (opt=calcfc or calcall) after displacement (typically 10%) along the reaction path for the normal coordinate of the imaginary vibrational frequency. The zero-point vibrational energies as well as the thermal and entropic contributions to the free energies were taken from the B3LYP/6-31G* frequency calculations. Single-point energy calculations for all stationary points were performed at the B3LYP/6-311+G** level using the corresponding B3LYP/6-31G* geometries. Six Cartesian d functions were used in these calculations. In some cases, CASPT2 single-point energies were calculated with the B3LYP/6-31G* geometries, and CBS-QB36 calculations were performed. The energies provided are enthalpies and free energies at 298.15 K. In some cases, solvation calculations were performed with the Polarizable Continuum Model (PCM)7 using acetonitrile as the solvent. All of the calculations were performed using Gaussian 98 at the Ohio Supercomputer Center.

7.3 Results

Calculations have been reported which predict that carbene8 and nitrene oxides have singlet ground states. Zelentsov et al. using CASSCF9 methods predicted that the singlet-triplet energy gap of C6H5NOO is -13.6 kcal/mol. Our DFT, CBS-QB3 and

147 CASPT2 calculations also indicate that CH3-NOO, CH3-O-N-O-O and C6H5NOO, have singlet ground states, as shown in Table 7.1, in agreement with previous reports.

CH3-N-O-O Ph-N-O-O CH3O-N-O-O

open-shell singleta 51.8 53.6 48.3 20.1 14.8 16.0 triplet (24.1) (16.4) (18.5) [13.7] [8.5] [10.0] 0 0 0 closed-shell singlet (0) (0) (0) [0] [0] [0] a CASPT2, (CBS-QB3), [B3LYP/6-311+G**//B3LYP/6-31G*]

Table 7.1. DFT, CASPT2 and CBS-QB3 calculated energy separations ΔH (298K) (kcal/mol). The CBS-QB3 numbers are in parentheses, and the B3LYP/6- 311+G*//B3LYP/6-31G* values are in brackets.

In order to assess the difference between carbenes and nitrenes, CBS-QB3 calculations were used to calculate the ΔH of reactions (19)-(24) (kcal/mol) (Scheme

7.1). The reaction of triplet phenylcarbene with oxygen is exceedingly exoergic (-57.1 kcal/mol), but the analogous reaction of triplet phenylnitrene is much less exoergic (-6.2 kcal/mol) at 298 K. This explains the very different rates of reaction of these intermediates with oxygen. The C-O bond formed in the reaction of a triplet carbene with oxygen is much stronger than the bond formed between N and O in the analogous nitrene reaction, hence the faster rate of the former process.

148

Scheme 7.1. CBS-QB3 and DFT calculations of carbenes and nitrenes with oxygen (298K, kcal/mol). The B3LYP/6-311+G**//B3LYP/6-31G* numbers are in parentheses.

ΔH ΔG H 3 -69.9 (-60.4) -57.1 (-48.0) (19) CH + O2 C O O H 3 -73.9 (-64.7) -60.9 (-52.6) (20) CH3 CH + O2 CH3 C O O 3 (21) N + O2 N O -20.2 (-5.7) -6.2 (+6.4) O

(22) 3 CH3 N + O2 CH3 N O -24.7 (-10.5) -13.1 (+1.2) O

3 (23) CH3O N + O2 CH3ONO -15.8 (-2.8) -4.8 (+9.0) O

3 (-2.4) -2.7 (+9.3) (24) CH2O- N + O2 CH2ONOO -15.3

To better understand these results, a series of reactions (25)-(31) were calculated

(Scheme 7.2).

149

Scheme 7.2. CBS-QB3 and DFT calculations of reactions (298K, kcal/mol). The B3LYP/6-311+G**//B3LYP/6-31G* numbers are in parentheses.

ΔH ΔG

3 3 (25) N-CH2-CH2-CH3 NH2-CH2-CH2- CH 23.4 (25.5) 23.5 (24.9)

3 3 (26) HO-CH2-CH2- N CH3-CH2-O- N 9.9 (9.1) 10.6 (8.9)

3 3 (27) HOCH2 N CH2O N 25.4 (25.0) 26.3 (24.7)

3 3 (28) HOCH2 N OCH2 N 19.8 (20.5) 20.9 (19.5)

(29) O-O-N-CH2-CH2-CH3 NH2-CH2-CH2-CH-O-O -27.9 (-32.0) -28.4 (-31.7)

3 3 (30) O-O-N-CH2-CH2- CH N-CH2-CH2-CH-O-O -62.0 (-54.6) -60.3 (-53.2)

HO-CH -CH -N-O-O CH -CH -O-N-O-O (31) 2 2 3 2 24.3 (20.4) 24.4 (20.0)

150 As shown in equation (25), a triplet alkylnitrene is considerably more stable than an isomeric triplet alkyl carbene. This effect was deduced by experiment and confirmed by theory for phenylnitrene and isomeric pyridylcarbenes.10-12

Equation (26) clearly indicates that triplet nitreno- is more stable than triplet ethoxynitrene. The dominant effect in this equation is the weakness of the N-O bond of the nitrene relative to the OH bond of the alcohol. This same effect is also evident in equations (27) and (28) which reveal that a phenyl group is a far more stabilizing triplet nitrene substituent than either a benzyloxy or alkyl group.

Reaction (29) reveals that carbene oxides are much more stable than the isomeric nitrene oxides. As N-H and C-H bond dissociation energies are rather similar,13 it is clear that the bond formed in the carbene reaction is much stronger than the corresponding bond in the nitrene oxidation process. This explains the increased exothermicity and rate of the carbene reaction with oxygen, relative to triplet phenylnitrene.

A triplet nitrene-carbene oxide is much more stable (equation 30) than a nitrene oxide-triplet carbene because a triplet nitrene is more stable than a triplet carbene and a nitrene oxide is less stable than a carbene oxide. The two effects are nearly additive.

Equation (31) demonstrates that the nitrene-oxide ethanol is more stable than ethoxynitrene oxide. The alkoxy group destabilizes a nitrene oxide more than it destabilizes a triplet nitrene. This makes the reaction of the triplet alkoxy nitrene with oxygen slightly less favorable than the corresponding reaction of a triplet arylnitrene.

Srinivasan et al.14 have generated benzyloxynitrene in acetonitrile solution and used TRIR spectroscopy to demonstrate that this nitrene reacts with oxygen with a large rate constant of 109-10 M-1s-1. This result can not be explained by a difference in reaction

151 thermodynamics of triplet phenyl versus triplet methoxynitrene (see Scheme 7.1, reaction

21 versus reaction 24) which are rather similar. We, therefore, investigated an electron transfer route to form the nitrene oxides. Our calculations predict that oxynitrenes are much more readily ionized than arylnitrenes (Figure 7.1). Triplet benzyloxynitrene is more readily ionized than its hydroxymethylphenylnitrene analog by ~30 kcal/mol in acetonitrile.

- CH2O-N +e 115.4 -113.1 + O2 3 CH2O- N 2.3 + CH2ONOO O2 - HOCH N +e 2 145.6 -142.9 + O2

2.8 HOCH 3N HOCH NOO 2 + O 2 2

Figure 7.1. The reactions of nitrenes with oxygen, ΔG (298K) (kcal/mol), calculated at the B3LYP/6-311+G**//B3LYP/6-31G* level in acetonitrile (PCM).

Thus we speculate that oxynitrenes react faster with oxygen than do arylnitrenes because of partial electron transfer in the transition state followed by rapid collapse of the

“ion pair” to form the nitrene oxide.

RNO 3 N OO R NR+ O2 O

Based on our speculation that the oxynitrenes may react faster due to an electron transfer component, we searched for transition states for the conversion of the nitrenes to

152 the nitrene oxides at the B3LYP level in both the gas phase and with the PCM model for solvation (using acetonitrile). There is, of course, some ambiguity about what the spin state should be for the transition states, as the nitrene and molecular oxygen are triplet states, but the nitrene oxide product is a singlet. All attempts to locate transition states on the singlet surface (using both closed- and open-shell methods) were unsuccessful.

However, using a triplet state description for the transition state provide a number of transition states for the reactions of methylnitrene, methoxynitrene, benzyloxynitrene and hydroxymethylphenylnitrene in the gas phase and with the PCM model for acetonitrile.

The enthalpic activation barriers are provided in Table 7.2 for the four different nitrenes that were investigated. For the smaller nitrenes in the gas phase, the CBS-QB3 level was used to verify that the B3LYP energetic trends were qualitatively reliable (and they are).

‡ a Table 7.2. Activation barriers (ΔH ) for the direct reaction of O2 with various nitrenes.

b b reaction CBS-QB3 (gas) B3LYP (gas) B3LYP (CH3CN) CH3-N 10.9 12.9 12.3 CH3O-N 11.6 14.3 11.3 HOCH2Ph-N c 13.2 11.8 PhCH2O-N c 11.3 8.7 a In kcal/mol at 298K, relative to the energies of infinitely separated reactants being set at zero in energy. In each case, the transition state was treated as a triplet with an unrestricted wave function. b Energies at the B3LYP/6-311+G** level based on fully optimized geometries at the B3LYP/6-31G* level in either the gas phase or with the PCM model for acetonitrile. c Not calculated.

153 From Table 7.2, we predict that methylnitrene will react slower than methoxynitrene with molecular oxygen due to the relative activation barriers in the gas phase. At the CBS-QB3 and B3LYP levels, the preference for CH3N reactivity over that of CH3ON is 0.7 and 1.4 kcal/mol, respectively; however, the trend is inverted on going to acetonitrile where CH3ON is predicted to have a 1.0 kcal/mol lower barrier for reaction at the B3LYP level. For the larger nitrenes (HOCH2Ph-N and PhCH2O-N), both the gas phase and PCM levels show that the oxynitrene is predicted to have a strong preference for reaction with O2. Indeed, the difference in activation barriers is 3.1 kcal/mol for the oxynitrene and the predicted (theoretical) relative rate difference would be 1.9×102 by traditional transition state theory. This theoretical rate preference for the oxynitrene compares favorably to the experimental 103-104 rate enhancement noted by Toscano and co-workers.14

We further investigated the electronic wave functions for the different transition states, and indeed, there is more electron transfer to the O2 unit for the oxynitrenes as evaluated by the Natural Population Analysis15 method (see Table 7.3). There is a significant amount of negative charge that accumulates on the O2 fragment in the transition state (for the gas and PCM calculations), and the effect is more pronounced for the oxynitrenes. For example, the alkyl- and phenylnitrenes have about –0.08 e on the O2 fragment in the transition state; however, the oxynitrenes have about –0.15 e in their analogous transition states. The effect is amplified at the PCM level.

154 Table 7.3. Charge on the O2 unit in the transition state for O2 reaction with various nitrenes at the B3LYP level.a

gas phase PCM (CH3CN) reaction 6-31G* 6-311+G** 6-31G* 6-311+G**

CH3-N –0.068 –0.078 –0.072 –0.086

CH3O-N –0.072 –0.078 –0.158 –0.181

HOCH2Ph-N –0.071 b –0.094 b

PhCH2O-N –0.145 –0.160 –0.183 –0.212

a In electrons, using the respective B3LYP/6-31G* geometry which was fully optimized in either the gas phase or with the PCM model for acetonitrile. b Attempts to localize the orbitals with the NPA method failed with this basis set.

Finally, the rearrangement of an alkoxynitrene oxide to a nitrate is predicted to be exceedingly exoergic and to involve a dioxaziridine intermediate (Figure 7.2).

155 O CH N 3 O O CH N O 3 O O 32.7 20.9 (33.2) (21.6) 5.2 0.0 (8.6) O (0.0) CH3 N CH N O O O 3 O O -68.0 (-70.2) O CH N 3 O O

Figure 7.2. The rearrangement of an alkoxynitrene oxide to a nitrate, ΔG (298K) (kcal/mol), calculated at the CBS-QB3 (top) and B3LYP/6-311+G**//B3LYP/6-31G* (bottom) level.

Such a pathway was proposed by Toscano and co-workers.14 The dioxaziridine intermediates can be formed by photolysis of nitrene oxides.16 Our results mirror those of

Makareeva et al.16

Scheme 7.3. CBS-QB3 and DFT calculations of triplet phenylnitrene with nitrene oxide and dioxaziridine intermediate (298K, kcal/mol). The B3LYP/6-311+G**//B3LYP/6- 31G* numbers are in parentheses.

ΔH ΔG

3N + N O 2 N O -84.2 (-80.2) -81.9 (-79.5) O 52

O -92.9 (-94.1) -91.5 (-93.5) 3N + N 2 N O O 53

156 Both the nitrene oxide and the dioxaziridine intermediate are predicted to react extremely exothermically with triplet phenylnitrene (Scheme 7.3). In this sense, aromatic nitrene oxidation is autocatalytic. The reaction of a triplet arylnitrene with oxygen produces intermediates, such as 52 and 53, which more rapidly oxidize additional nitrene substrates than does molecular dioxygen.

7.4 Conclusions

In summary triplet phenylcarbene reacts more rapidly with oxygen than does triplet phenylnitrene because the former reaction is much more exoergic than the latter.

The C-O bond formed in the carbene reaction is much stronger than the N-O bond formed in the analogous nitrene process. We propose that triplet methoxynitrene will react with oxygen faster than triplet phenylnitrene due to charge separation in the transition state. This electron-transfer mechanism rationalizes the rapid reaction of molecular oxygen with triplet benzyloxynitrene, whereas phenylnitrene reacts slowly.

The oxidation of phenylnitrene is autocatalytic. The nitrene oxides and diazidines react more exothermically with triplet phenylnitrene than does oxygen.

7.5 References for Chapter 7

1. Bucher, G.; Scaiano, J. C.; Platz, M. S. Kinetics of Carbene Reactions in Solution. Landolt-Bornstein, Group II, Volume 18, Subvolume E2 Springer, Berlin, Germany, (1998), p.141.

2. (a) Gritsan, N. P.; Pritchina, E. S. J. Inf. Record. Mater. 1989, 17), 391 (b) Liang, T.-Y., Schuster, G. B. J. Am. Chem. Soc. 1987, 109, 7803 (c) Pritchina, E. A.; Gritsan, N. P. J. Photochem. Photobiol., A: Chemistry 1988, 43, 165. (d) Gritsan, N. P.; Pritchina, E. A. Russ. Chem. Rev. 1992, 61, 910.

157 3. Frisch, M. J. et al. Gaussian 98, Gaussian, Inc.: Pittsburgh, 1998.

4. Ziegler, T. Chem. Rev. 1991, 91, 651

5. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

6. Montgomery, J. A., Jr.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys. 1999, 110, 2822.

7. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

8. (a) Qi, B.; Su, K.; Wang, Y.; Wen, Z.; Tang, X. Wuli Huaxue Xuebao 1998, 14, 1033. (b) Subhan, W.; Rempala, P.; Sheridan, R. S. J. Am. Chem. Soc. 1998, 120, 11528 (c) Gutbrod, R.; Kraka, E.; Schindler, R. N.; Cremer, D. J. Am. Chem. Soc. 1997, 119, 7330. (d) Warner, P. M. J. Org. Chem. 1996, 61, 7192. (e) Cremer, D.; Gauss, J.; Kraka, E.; Stanton, J. F.; Bartlett, R. J. Chem. Phys. Lett. 1993, 209, 547. (f) Bach, R. D.; Andres, J. L.; Owensby, A. L.; Schlegel, H. B.; McDouall, J. J. W. J. Am. Chem. Soc. 1992, 114, 7207. (g) Cremer, D.; Schmidt, T.; Sander, W.; Bischof, P. J. Org. Chem 1989, 54, 2515. (h) Gauss, J.; Cremer, D. Chem. Phys. Lett. 1987, 133, 420. (i) Cremer, D. J. Am. Chem. Soc. 1979, 101, 7199.

9. Zelentsov, S. V.; Zelentsova, N. V.; Shchepalov, A. A. High Energy Chemistry. 2002, 36, 326.

10. (a) Crow, W. D.; Wentrup, C. Tetrahedron Lett. 1968, 6149. (b) Wentrup, C. J. Chem. Soc., Chem. Commun. 1969, 1386. (c) Kuzaj, M.; Lüerssen, H.; Wentrup, C. Angew. Chem. Int. Ed. Engl. 1986, 25, 480. (d) Wentrup, C. Topics Current Chem. 1976, 62, 173.

11. Platz, M. S. Acc. Chem. Res. 1995, 28, 487.

12. (a) Karney, W. L.; Borden, W. T. Adv. Carbene Chem. Vol3, p205, Elsevier, New York, NY 2001. (b) Kemnitz, C. R.; Karney, W. L.; Borden, W. T. J. Am. Chem. Soc. 1998, 120, 3499.

13. (a) Barckholtz, C.; Barckholtz, T. A.; Hadad, C. M.; J. J. Am. Chem. Soc. 1999, 121, 491. (b) http://webbook.nist.gov/chemistry/

14. Srinivasan, A.; Kebede, N.; Saavedra, J. E.; Nikolaitchik, A. V.; Brady, D. A.; Yourd, E.; Davies, K. M.; Keefer, L. K.; Toscano, J. P. J. Am. Chem. Soc. 2001, 123, 5465.

15. Reed, A. E.; Weinhold, F.; Curtiss, L. A. Chem. Rev. 1988, 88, 899.

158 16. Makareeva, E. N.; Lozovskaya, E. L.; Zelentsov, S. V. High Energy Chem. 2001, 35, 177.

159

CHAPTER 8

WOLFF REARRANGEMENT OF DIAZO ESTERS AND KETONES

Reproduced with permission from the Journal of American Chemical Society, submitted for publication. Unpublished work copyright 2005 American Chemical Society.

8.1 Introduction

Thermolysis of diazoketones and esters generally leads to ketenes via Wolff

Rearrangement (WR). In reactive solvents, it is occasionally possible to intercept acylcarbene intermediates. This has led to the realization that WR can proceed either in concert with nitrogen extrusion or stepwise via carbene intermediates.1,2

A few trends have been noted. First, it has been demonstrated by spectroscopic methods that diazo carbonyl compounds exist in two planar conformations with either a syn or anti disposition of the carbonyl and diazo groups.3

O O N C N C H RC RC H N N

syn anti

160 Kaplan, Meloy and Mitchell demonstrated that conformation plays a critical role in the WR process.4 Syn diazo carbonyl compounds undergo concerted thermal WR, whereas the anti conformer is the more efficient source of carbene. It is also often found that diazo esters are more efficient sources of acylcarbenes than are the analogous diazo ketones.1,2

To understand the origins of these effects, a number of density functional theoretical (DFT)5 and ab initio molecular orbital calculations6 were performed. We have also considered Wolff-like rearrangements of alkyl diazomethanes.

8.2 Computational Methods

Density functional theory (DFT)5 and ab initio molecular orbital calculations6 have been applied in this study. The geometries were completely optimized at the

B3LYP/6-31G* level. Analytical vibrational frequencies were calculated at the same level for each stationary point to verify a minimum energy structure and to provide zero- point vibrational energy corrections, which were scaled by a factor of 0.9806.7 Each transition state was verified to connect to the respective reactant and product by careful optimization (opt=calcfc or calcall) after displacement (typically 10%) along the reaction path for the normal coordinate of the imaginary vibrational frequency. The zero-point vibrational energies as well as thermal and entropic contributions to the free energies were obtained from the B3LYP/6-31G* frequency calculations; the thermal and entropic corrections used the unscaled vibrational frequencies. Single-point energy calculations of all stationary points were performed at the B3LYP/6-311+G** level using the

161 corresponding B3LYP/6-31G* geometries. Six Cartesian d functions were used for these calculations.

Complete Basis Set (CBS) methods8 have also been applied to increase the accuracy of the results. The energies provided are CBS-QB3 enthalpies and free energies at 298.15 K. All stationary points, including transition states, were re-characterized at the

CBS-QB3 level with full geometry optimization at the required level.

To study the influence of solvent, polarizable continuum model (PCM)9 calculations were used with acetonitrile and cyclohexane as solvents.

All DFT and CBS-QB3 calculations were performed with Gaussian 9810 at the

Ohio Supercomputer Center.

8.3 Results

8.3.1 Conformational Isomerism

B3LYP calculations indicate that diazoacetone (54) prefers a syn alignment of the carbonyl and diazo groups in the gas phase by 1.6 kcal/mol.

O O N C N C H ΔG (298K) = 1.6 kcal/mol CH3 C CH3 C H N N 54s 54a

162 Solvent has little effect on this equilibrium. The calculated energy difference is 1.7 kcal/mol in cyclohexane and is 1.6 kcal/mol in acetonitrile using the PCM method.

Methyl diazoacetate (57) is also predicted to prefer a syn alignment but by only

0.2 kcal/mol in the gas phase, 0.2 kcal/mol in cyclohexane and by 0.1 kcal/mol in acetonitrile.

O O N C N C H ΔG= 0.2 kcal/mol CH3O C CH3O C H N N 57s 57a

On this basis, one can very simply predict that as diazoketones have more of the syn orientation at equilibrium than their ester analogs and following Kaplan, Meloy and

Mitchell,4 that (in the absence of additional information) diazoketones will be more prone to concerted Wolff Rearrangement than diazo esters.

It is known that for compounds RC(=O)CHN2, the syn conformer is preferred to

3 the anti form. NMR measurements of HC(=O)CHN2 reveal a syn/anti ratio of 7/3, suggesting an experimental free energy difference of 0.5 kcal/mol. This experimental energy difference compares favorably to the calculated gas-phase values of 0.6 and 1.1 kcal/mol at the B3LYP and CBS-QB3 levels, respectively.

Although the triplet state is the preferred ground state for all of the carbenes in this study, the singlet state of the carbene is expected to be involved in the Wolff

Rearrangement (WR) process. Therefore, all of the carbene structures in the potential

163 energy surfaces (see subsequent figures) for the WR refer to the singlet states of carbenes. The singlet-triplet energy gaps are shown in Table 8.1. The B3LYP level appears to over-estimate the singlet-triplet energy gap, as has been noted by Radom and coworkers.11

Table 8.1. Singlet-triplet energy gap [ΔG(298K) (kcal/mol)] for R-C-H at the B3LYP/6- 311+G**//B3LYP/6-31G* and CBS-QB3 levels of theory.a

R B3LYP/6-311+G**//B3LYP/6-31G* CBS-QB3 O

CH 3C 4.2 0.1 O

CH 3OC 5.8 4.4 O

CH 3OCH 2C 5.3 2.8 O

CH 3CH 2OC 5.5 3.9

CH 3CH 2 3.0 - a A positive value reflects a triplet ground state.

8.3.2 Rearrangements of Diazoketones and Diazoesters

The energy surface for gas-phase syn-anti isomerism and nitrogen extrusion of diazoacetone (54) was calculated at both the B3LYP and CBS-QB3 theoretical levels.

The results are shown in Figure 8.1. Theory predicts that nitrogen extrusion will proceed most rapidly when the carbonyl and diazo groups are aligned syn to each other as in 54s.

The preference for concerted (TS(54-56)) over stepwise (TS(54-55)) WR is 1.1 (CBS-

QB3) to 1.5 kcal/mol (B3LYP). In perfect agreement with Kaplan, Meloy and Mitchell,4 theory predicts that decomposition of syn diazoacetone leads to Wolff Rearrangement in 164 concert with nitrogen extrusion, whereas stepwise decomposition of the anti conformer is slower and produces carbene. Consideration of solvent (via PCM calculations in acetonitrile) does not alter this conclusion as concerted WR is still favored by 1.6 kcal/mol. Wolff Rearrangement of the free carbene must overcome a barrier of 3.9

(B3LYP) to 8.3 kcal/mol (CBS-QB3). This is in reasonable agreement with previous calculations of Radom, Schaefer and co-workers for the formyl analog.11 In general, the

CBS-QB3 and B3LYP results are similar and differ only in the accuracy of the computation of the free carbene (55).

165

O

H C H C C H O

H N

C N

N H H C C O N H H

C H + N N H TS(54-55) C TS(54-56) C H H 34.7 TS(55-56) O 33.2

(34.7) H C N 28.1 C C N [36.1] H (33.6) H 24.2 H (26.4) [34.5] [29.5] (18.1) TS(54s-54a) [26.4] 14.4

O

H + N N (13.0) C

C H C [15.0] H H 1.6 55 0.0 (1.6) [1.6] (0.0) [0.0]

O O

N

C C H N H H H C H C C C

H H H N N 54s -41.8 54a -41.8

H (-45.4) (-45.4) H

H C [-38.3] H H C [-38.3] H N N N N C C O + C C O +

H 56 H 56

Figure 8.1. The energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Wolff Rearrangement of diazoacetone (54) in the gas phase at the B3LYP/6- 311+G**//B3LYP/6-31G* (top) and CBS-QB3 (middle, in parentheses) levels of theory, and at the PCM level for acetonitrile (PCM) at the single-point B3LYP/6- 311+G**(PCM)//B3LYP/6-31G*(gas) level of theory (bottom, in brackets).

166 A computational study of formylcarbene by Radom, Schaefer, and co-workers11 has evaluated the role of a three-membered ring oxirene structure as an intermediate or transition state on the potential energy surface for carbonylcarbenes.

O O O CC C CH H C 3 CH C H CH3 CH 3

Our calculations (Figure 8.2) at the B3LYP and CBS-QB3 levels show that the oxirene is a transition state that interconverts acetylcarbene and formyl(methyl)carbene.

167

O H H C C C

O O H H H H

C C C C H 8.3 H H H C C H H (14.6) 4.3 3.9 (8.3) (1.2)

0.0 0.8 (0.0) (5.3)

O

H O H C C H H C C C H C H H H

-65.0 -65.0 (-63.5) (-63.5) H

H H H C H H C

C C O

C C O

H H

Figure 8.2. The energy surface [ΔG(298K) (kcal/mol)] in the gas phase at the B3LYP/6- 311+G**//B3LYP/6-31G* (top) and CBS-QB3 (bottom, in parentheses) levels of theory.

168

Since the B3LYP and the more expensive CBS-QB3 calculations gave similar results, we will use the faster B3LYP method throughout the rest of this study. The analogous surface for decomposition of methyl diazoacetate 57 is given in Figure 8.3.

Now the preference is decidedly (3.6 kcal/mol) in favor of stepwise (TS(57-58)) over concerted (TS(57-59)) decomposition via the conformer with an anti orientation of the carbonyl and diazo groups. The calculations are consistent with the observation that diazoesters are often better thermal sources of carbenes than are diazoketones.1,2

169

O O

H H N H C C C N

C H H O C O C O H H H

H

C H N N N C + H O C TS(57-59) H N TS(57-58) TS(58-59) 39.5 35.9 34.1 [41.2] [35.2] [37.2]

O H

H 27.5 C C H O C N N

H [28.9]

O TS(57s-57a) H H

C C N N H C + H 13.1 O 58 [12.9]

0.2 0.0 [0.3] [0.0] O O

H H H N C H H C C N C C O C O H H

N -15.5 H

N -15.5 57s H [-13.5] 57a H H H C C [-13.5] O O H H O N O N N C N C + C + C

H H 59 59

Figure 8.3. The energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Wolff Rearrangement of methyl diazoacetate (57) at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory in gas phase (top) and in acetonitrile (PCM, bottom).

170

To understand the origin of these effects, analogous calculations were performed on an isomeric pair of diazo ketones, 60 and 61, so that energetic differences between reactants, transition states and products could be easily elucidated.

concerted CH3OCH2CH C O + N2 O 64 C CH3OCH2 CHN2 O stepwise + 60 C N2 CH3OCH2 CH 64

concerted CH3CH2OCH C O + N2 O 12 C CH3CH2O CHN2 O stepwise + 61 C N2 CH3CH2O CH 63

The results for methoxydiazoacetone (60) and ethyl diazoacetate (61) are shown in Figure 8.4.

171 O O N

H O C H C C H N C C H O C H H H C H H C H H N H

N TS(60-62) TS(60-64) 35.9 (25.9) 34.7 25.2 (19.8) (21.9) O

C H O C H

C C

H H H 62 H +

N N

O

N O H H

C N H H C C C H H O C C H O C C H H H

H H TS(61-65)

N N TS(61-63) 38.8 33.8 27.1 1.9 0.0 (25.6) H O O (23.8). H O C H C H O C N H H C O C H C C N C C O H H H H C H H 63 C C N H H H H H

+ N H

60a 60s H H C N N O C H O C

H 0.2 0.0 C

H 65

O H + H

C C O H H H H N N O C N C C C N -16.1 H H H O N C C

H H H N 61a 61s -46.5 (-6.6)

H H

H H C C O C H O C

H 64 + N N

Figure 8.4. The gas-phase energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted Wolff rearrangement of methoxydiazoacetone (60) and ethyl diazoacetate (61) at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory. The numbers in parentheses are relative free energies between vertical isomeric pairs in the Figure 8.of isodesmic reactions. These refer to pairs of compounds displayed vertically, in which diazo compounds are compared with their diazo isomers, and isomeric transition states and isomeric carbenes are compared.

172

The syn conformer of ethyl diazoacetate 61s is 23.8 kcal/mol more stable 61.than the analogous conformer of methoxydiazoacetone 60s. Carboethoxycarbene 63 is more stable 8.than the isomeric formyl ether carbene 62 by 21.9 kcal/mol. These results are due to resonance within the ester group itself12 as shown in equation (32), as calculated at the

B3LYP/6-311+G**//B3LYP/6-31G* level of theory.

O O C C H COH C H ΔG = - 26.5 kcal/mol (32) 3 2 H3CH2CO H

Ester resonance dramatically influences the potential energy surface of the concerted Wolff Rearrangement reaction (Figure 8.4). Concerted Wolff Rearrangement of ethyl diazoacetate 61 to form nitrogen and ethoxyketene 65 is exoergic by 16.1 kcal/mol. In contrast, concerted loss (via TS(60-64)) of nitrogen and Wolff

Rearrangement of methoxydiazoacetone 60s to methoxymethylketene 64 is much more exoergic by 46.5 kcal/mol. In the latter process, ester resonance is not sacrificed, hence the greater exoergicity. Ethyl diazoacetate 61 prefers stepwise decomposition to form a carbene ester because the loss of ester resonance in the transition state TS(61-65) leading directly to ethoxyketene 65 and nitrogen raises the energy of the transition state.

8.3.3 Concerted 1,2-Hydrogen Migration and Nitrogen Extrusion Rearrangement of 1-Diazopropane

Concerted migration of carbon with nitrogen extrusion is well documented with acyl diazo compounds.13 There has been speculation that the analogous migration of

173 carbon or hydrogen may transpire in the photochemistry of simple alkyl diazo compounds.14 However, a computational study by Shevlin and McKee of tert- butyldiazomethane argued against this possibility in the thermolysis of this particular diazo compound.15

CH CH CH concerted 3 2 + N 68 2

CH3CH2CHN2

CH CH CH + 66 stepwise 3 2 N2 67

As shown in Figure 8.5, 1-diazopropane 66 exists in two conformers which resemble those of the diazo ketones.

In the lower energy conformation 66a, a methyl group is oriented anti to the diazo group and is poised to migrate. Note that in the absence of the carbonyl group, the designations of syn and anti are reversed. We found a transition state (TS(66-68)) for this concerted methyl migration with a free energy of activation barrier of 31.7 kcal/mol.

However, we find that stepwise decomposition of 1-diazopropane (66) via TS(66-67) to form 1-propylidene (ethylcarbene, 67) is substantially (5.5 kcal/mol) preferred to concerted nitrogen extrusion ((TSMe(66-68)) and methyl migration (Figure 8.5). The lowest energy pathway, however, involves 1,2-migration of hydrogen in concert with nitrogen extrusion (TSH(66-68), 24.9 kcal/mol). Our calculations predict that 1- propylidene (ethylcarbene, 68) is not formed during the pyrolysis of 1-diazopropane.

174 H H H H H C C C N N H H C C H C H H H H

H N H N C H TSMe(66-68) C C H TS(66-67) H H 31.7 24.7 26.2 N 24.9 N H H H H H H 20.4 19.6 C H C C C C H C H H H H H H H C C + N N

H 1.5 C 0.0 TSH(66-68) H H 67 H H N H

C C H N

H C C N H C H H H N H C H

H 66s 66a

-54.3 -54.3

H H H H C C H H H N N H N N + + C C C C

H H H H 68 68

Figure 8.5. The gas-phase energy surface [ΔG(298K) (kcal/mol)] for stepwise and concerted migration and nitrogen extrusion of 1-diazopropane (66) at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

175 8.3.4. Discussion

The results were compiled in Table 8.2 and analyzed with the aid of isodesmic reactions (Figure 8.6).

176

Table 8.2. The relative energies [ΔG(298K) (kcal/mol)] for the gas-phase stepwise and concerted Wolff Rearrangement of RCHN2 at the B3LYP/6-311+G**//B3LYP/6-31G* level of theory

R syn TS(syn-anti) anti TS_concerted TS_stepwise carbene TS_carbene ketene O

CH 3C 0 14.4 1.6 33.2 34.7 24.5 28.1 -41.8 O

CH 3OC 0 13.1 0.2 39.5 35.9 27.5 34.1 -15.5 O a CH 3OCH 2C 0 - 1.9 34.7 35.9 25.2 - -46.5 O

CH 3CH 2OC 0 - 0.2 38.8 33.8 27.1 - -16.1

CH 3CH 2 0 - 1.5 31.7 26.2 20.4 - -54.3 a Not calculated.

O 27.0 O C C CH CH CH O CH + N2 3 2 2 CH3CH2CH2O CH 69 N2 70 - 9.6 - 0.7 O O C 18.1 C + N2 CH3O CH2CH2CH CH3O CH2CH2CH N 71 2 72 12.2 15.5 O 23.5 O C C CH OCH CH CH + N2 3 2 2 CH3OCH2CH2 CH N 73 2 74

Figure 8.6. Isodesmic reactions [ΔG(298K) (kcal/mol)] at the B3LYP/6- 311+G**//B3LYP/6-31G* level of theory.

177

As expected, esters are always more stable than the isomeric ketone-ethers.

However, there is very little difference (0.7 kcal/mol, see Figure 8.6) in the ability of an ester or a simple alkyl group to stabilize a singlet carbene. Similar results were found with the keto carbenes (3.2 kcal/mol, equation 33).

O O C C ΔG= - 3.2 kcal/mol (33) CH3 CH2CH2CH CH3CH2CH2 CH 75 76

In comparison, the diazo group strongly enjoys conjugation with an ester moiety by a large amount, 9.6 kcal/mol (Figure 8.6). Similar effects are observed with ketones

(8.1 kcal/mol, equation 34).

O O C C CH3 CH2CH2CH ΔG= 8.1 kcal/mol (34) CH3CH2CH2 CH - N2 N2 77 78

The net result is that the carbonyl group differentially stabilizes a diazo group more than the corresponding singlet carbene. This explains why the barriers to decomposition of diazoalkanes are uniformly lower than those of the analogous diazo carbonyl compounds. Loss of resonance between the carbonyl and diazo moieties is present in both modes of decomposition of diazoacetone and is absent, of course, in both modes of decomposition of diazopropane. Thus, carbonyl-diazo resonance, or the lack of

178 it, does not explain why diazoacetone and 1-diazopropane prefer concerted acetyl and hydrogen migration and why 1-diazopropane prefers stepwise decomposition to form 1- propylidene (ethylcarbene) to concerted methyl migration. (However, we should add that

H-migration in concert with nitrogen extrusion is more favored overall.) The explanation for the differences in methyl migration is found in the four transition states.

Diazoacetone prefers planar conformations to maximize resonance between the carbonyl and diazo groups. Singlet carbonyl-substituted carbenes have a very different preference. The carbonyl group and carbene carbon define a plane, but the hydrogen atom attached to the singlet carbene carbon is nearly orthogonal to this plane.11 In this geometry, the filled orbital of the singlet carbene, not the empty p orbital, conjugates with the π system of the carbonyl group. This alignment allows the singlet carbene to benefit from enolate anion type resonance. Triplet carbonyl carbenes, in contrast, are predicted to be planar.11

-73.8o O 1.50 1.42 O H o H O H H3C 131 C H C 2.15 + N2 3 N H3C 2.65 N N N

The transition state for carbene formation must balance the tendency to remain planar to maximize carbonyl-diazo conjugation and the desire of the carbene for a non- planar geometry. The balance struck by the TS has a H-C-C-O torsional angle of 73.8°.

179 The TS is carbene-like, and there is substantial deconjugation of the carbonyl and diazo groups.

In the concerted Wolff Rearrangement, the torsion angle between the CO and CH bonds is 69°. Consequently, there is slightly more ketone-diazo conjugation in the TS for the concerted loss of nitrogen than in the TS to stepwise formation of carbene. This effect is buttressed by the greater exothermicity of WR relative to carbene formation. Thus the barrier to Wolff Rearrangement is slightly lower in energy than the barrier to stepwise formation of carbene (by 1.1-1.5 kcal/mol, Figure 8.1).

N O O N N 1.60 1.39 N H o OCC + N H3C H H3C 96.6 C 2.19 2 CH3 2.23 H

In the TS [(TS(66-67), Figure 8.5] for thermal formation of singlet ethylcarbene from 1-diazopropane, an α C-H bond eclipses the π system of the diazo group in the reactant and eclipses the empty p orbital of the carbene. The H-C-H angle at the developing carbene carbon is only 83.9°. This maximizes a hyperconjugative interaction of the C-H bond with the empty p orbital of the carbene. 16 The TS is carbene-like with an α C-H bond interacting with the developing p orbital of the carbene.

180 H H H 83.9o H 1.53 119o H H H H C 3.47 H C H CC 3 + N2 N 2.57 H C H C 3 N N 3 N TS(66-67) ΔG = 26.2 kcal/mol

In the TS for concerted 1,2 hydrogen migration and nitrogen extrusion [TSH(66-

68)], the migrating hydrogen eclipses the developing empty p orbital of the carbon that becomes a carbene center.

H1.20 H N H N H N N 1.52 1.47 1.47 H H CC 129o H3C 3.37 + N2 H H CH3 H3C H 2.65

TSH(66-68) ΔG = 24.9 kcal/mol

In the TS for concerted loss of nitrogen and methyl migration [TSMe(66-68)], it is the methyl group that interacts with the developing empty p orbital of the carbene-like center.

181 H H N H N H N N 1.59 1.47 H H H CC 91.0o N H C 2.60 + 2 H 3 H CH3 H3C 2.18

TSMe(66-68) ΔG = 31.7 kcal/mol

The TS to concerted formation also looks much more like a carbene, with a methyl group interacting with the empty p orbital of the carbene, than it resembles propene. Concerted propene formation from 1-diazopropane is more exothermic than the formation of methyl ketene from diazoacetone, yet unlike diazoacetone, ethylcarbene formation is more facile than that of propene by a rearrangement involving carbon. (Note: hydrogen migration (TSH(66-68)) in concert with nitrogen extrusion is more favorable than stepwise carbene formation.) This result implies that an α C-H bond affords better stabilization of a carbene than does an α C-CH3 bond. To test this possibility, we calculated the energies of the two relevant conformations of ethylcarbene. Indeed ethylcarbene prefers to stabilize the alkylcarbene center by a C-H bond by 3.5 kcal/mol relative to a C-CH3 bond.

84.2o H H H H 1.45 H 1.58 1.47 1.52 119.0o 97.6o H ΔG= 3.5 kcal/mol H3C H3C 2.57 2.29

182 This preference explains most of the 5.5 kcal/mol benefit of the TS[(TS(66-67)] to carbene formation to the TS[TSMe(66-68)] leading to propene formation by concerted methyl migration. Both modes of decomposition of 1-diazopropane have a carbene-like

TS. The methyl migration TS has the higher energy carbene conformation, and thus, this process is slower than the process which forms the carbene in its preferred conformation.

The remaining 2.0 kcal/mol is due to the steric cost of bringing a methyl group so close

(C-C-C angle of 91.0°) to the diazo carbon.

8.4 Conclusions

The energetic benefit of ester resonance stabilizes ethyl diazoacetate and singlet carboethoxycarbene by 25.6 and 21.9 kcal/mol, respectively, relative to their keto-ether isomers. Ester resonance is lost during the Wolff Rearrangement of ethyl diazoacetate, thus the conversion of this diazoester to ethoxyketone is 30.4 kcal/mol less exoergic than the conversion of methoxymethyldiazoacetone to methoxymethylketene. Diazoesters prefer extrusion of nitrogen with the formation of carboethoxycarbene to preserve ester resonance, but diazoketones (where this is not a consideration) prefer to extrude nitrogen in concert with Wolff Rearrangement.

1-Diazopropane prefers thermal decomposition to form propene and nitrogen.

Loss of nitrogen proceeds in concert with hydrogen migration to form propene. The methyl migration is computed to proceed stepwise. The transition states of both the stepwise and concerted processes are carbene-like. Ethylcarbene prefers the conformation in which an α C-H bond interacts with the empty p orbital of the carbene. The analogous

183 conformation in which the α C-CH3 bond interacts with the empty p orbital of the carbene is 3.5 kcal/mol higher in energy. Thermal decomposition of 1-diazopropane prefers concerted 1,2-hydrogen migration and nitrogen extrusion because the TS resembles the more stable conformation of ethylcarbene, in which a hydrogen atom is perfectly poised for 1,2-migration. The calculations are restricted to the ground state surface and do not predict the outcomes of photochemical experiments.

8.5 References for Chapter 8

1. Kirmse, W. Eur. J. Org. Chem. 2002, 14, 2193.

2. Toscano, J. P. "Laser flash photolysis studies of carbonyl carbenes" in Advances

in Carbene Chemistry, Brinker, U. H., Ed, 1998, volume 2, pp. 215-244.

3. Regitz, M.; Maas, G. Diazo Compounds; Academic Press, New York, N.Y., 1986,

16-23 and references therein.

4. (a) Kaplan, F.; Meloy, G. K. J. Am. Chem. Soc. 1966, 88, 950. (b) Kaplan, F.;

Mitchell, M. L. Tetrahedron Lett. 1979, 759.

5. (a) Labanowski, J. W.; Andzelm, J. Density Functional Methods in

Chemistry; Springer: New York, 1991. (b) Parr, R. G.; Yang, W.

Density Functional Theory in Atoms and Molecules; Oxford University

Press: New York, 1989.

6. Jensen, F. Introduction to Computational Chemistry; Wiley: Chichester, 1998.

7. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

184 8. Montgomery, J. A., Jr.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem.

Phys. 1999, 110, 2822.

9. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

10. Frisch, M. J. et al. Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

11. (a) Scott, A. P.; Nobes, R. H.; Schaefer, H. F., III; Radom, L. J. Am. Chem. Soc.

1994, 116, 10159. (b) Vacek, G.; Galbraith, J. M.; Yamaguchi, Y.; Schaefer, H.

F., III; Nobes, R. H.; Scott, A. P.; Radom, L. J. Phys. Chem. 1994, 98, 8660. (c)

Scott, A. P.; Platz, M. S.; Radom, L. J. Am. Chem. Soc. 2001, 123, 6069.

12. Wiberg, K. B.; Hadad, C. M.; Rablen, P. R.; Cioslowski, J. J. Am. Chem. Soc.

1992, 114, 8644.

13. (a) Roth, H. D. Acc. Chem. Res. 1977, 10, 85. (b) DoMinh, T.; Strausz, O. P.;

Gunning, H. E. J. Am. Chem. Soc. 1969, 91, 129. (c) Tomioka, H.; Okuno, H.;

Izawa, Y. J. Org. Chem. 1980, 45, 5278. (d) Rando, R. R. J. Am. Chem. Soc.

1970, 92, 6706. (e) Wulfman, D. S.; Poling, B.; McDaniel, R. S., Jr. Tetrahedron

Lett. 1975, 4519.

14. Platz, M. S. "Issues and challenges in the chemistry of alkylcarbenes" in

Advances in Carbene Chemistry, Brinker, U. H., Ed, 1998, volume 2, pp. 133-

174.

15. Armstrong, B. M.; McKee, M. L.; Shevlin, P. B. J. Am. Chem. Soc. 1995, 117,

3685.

16. Sulzbach, H. M.; Platz, M. S.; Schaefer, H. F. III; Hadad, C. M. J. Am. Chem.

Soc. 1997, 119, 5682-5689

185

CHAPTER 9

MISCELLANEOUS

9.1 Singlet and Triplet 2,6 – Difluorophenylnitrene

This section is reproduced in part with permission from the Journal of the Physical Chemistry volume 109, page 2816. Copyright 2005, American Chemical Society.

9.1.1 Introduction

Photolysis of aryl azides in solution at ambient temperature releases singlet arylnitrenes (1N), which can either rearrange to ultimately form ketenimines (K) or relax to form triplet arylnitrenes (3N) as shown in Scheme 1 for 2,6-difluorophenyl azide (A).1

The singlet nitrene is the key intermediate whose partitioning determines whether stable products are ultimately formed from the ketenimine or the lower energy triplet nitrene intermediate.

Most singlet aryl nitrenes will have lifetimes too short to monitor by nanosecond time-resolved vibrational spectroscopy.1 As singlet 2,6-difluorophenylnitrene is the longest lived singlet aryl nitrene known in solution,2 it appears to be the most attractive singlet arylnitrene for study by this technique, thereby motivating this study of the photochemistry of 2,6-difluorophenyl azide by Time-Resolved Infrared (TRIR)

186 spectroscopy. Computational methods have been applied to assist spectroscopic assignments.

187

Scheme 9.1

F O

N S CH3

CH3 I F O S H3C CH3

F F F F N 1 N N3 N

F F F A AZ F 1N K

kisc

HNEt2 F F 3N N

F NEt2 3N F H

F F

N N

F F

188

9.1.2 Computational Methods

Density functional theory (DFT) was applied in this study.3 The geometries were completely optimized at the B3LYP/6-31G* level. Analytical vibrational frequencies were calculated at the same level for each stationary point to verify a minimum energy structure. As noted below, vibrational frequencies were scaled by 0.9613.4 To study the influence of solvent, polarizable continuum model (PCM)5 calculations were performed with acetonitrile as the solvent.

CASSCF methods6 provide accurate results for singlet open-shell nitrenes.1,2

However, both Gaussian 987 and Gaussian 038 can compute vibrational frequencies at the CASSCF level, the IR intensities are not provided; therefore, the utility of this method for predicting the IR spectra is limited. Due to this limitation, we resorted to using open- shell DFT calculations as pioneered by Cramer et al. for treating the open-shell singlet nitrene.9 In these calculations, the value for the triplet nitrene was 2.05, and for the open-shell singlet, the value was 0.91 (Table 9.1). Therefore, the open-shell singlet nitrene appears to almost be an equal mix of singlet and triplet wave function. The unscaled vibrational frequencies obtained by DFT and CASSCF have been compared, and DFT is found to generate frequencies comparable to those predicted by CASSCF.

Thus, DFT calculated vibrational (IR) frequencies and intensities are used for the open- shell singlet nitrene.

189

Molecule S2 Energy Energy Energy difference difference difference (kcal/mol), (kcal/mol), (kcal/mol), DFT, gas DFT, PCM, CASSCF, gas phase acetonitrile phase 1N 0.911 0 0 0 3N 2.046 -12.6 -11.9 -18.7

Table 9.1. values and relative energies calculated in various media with different levels of theory.

Karney and Borden10 demonstrated that CASSCF methods accurately predict the barriers to singlet aryl nitrene rearrangements. Theory has also accurately predicted the singlet –triplet energy separation of phenylnitrene.11

All DFT and CASSCF calculations were performed at the Ohio Supercomputer

Center using the Gaussian suite of programs.7,8

9.1.3 Results

Density Functional Theory (DFT) methods were used to predict the vibrational spectra of ketenimine K, singlet (1N), and triplet 2,6-difluorophenylnitrene (3N). The vibrational spectrum of the singlet nitrene (1N) was also predicted by CASSCF theory and was in excellent agreement with DFT calculations (Figure 9.1). The spectra predicted by theory are shown in Figure 9.2.

190 3500 y = 0.9367x + 16.201 2 3000 R = 0.9991

2500

2000

1500

1000

500

0 0 500 1000 1500 2000 2500 3000 3500 4000

IR by CASSCF

3500 y = 0.9459x + 8.9993 R2 = 0.9994 3000

2500

N 2000

F C F 15 0 0 C C

10 0 0 C C H H C

500 H

0 0 500 1000 1500 2000 2500 3000 3500 4000 IR by CASSCF

triplet

3500

3000 y = 0.9255x + 49.369 2 2500 R = 0.9973 N

C 2000 F F C C 1500 C C

H IR by B3LYP IR H 1000 C

500 H

0 0 500 1000 1500 2000 2500 3000 3500 4000 IR by CASSCF

Open shell singlet

Figure 9.1. The comparison of IR frequencies calculated by B3LYP and CASSCF.

191 35 1217 30 25 1513 20 1202 1396 15 Intensity 762 1271 1833 10 1165 838 1316 1566 5 650 1027 0 0 500 1000 1500 2000 1/cm

(a) ketenimine (K)

35 30 1433 25 20 1565 15 975 Intensity 10 1228 747 1297 5 0 0 500 1000 1500 2000 1/cm

(b) triplet 2,6-difluorophenyl nitrene (3N)

40 1435 35 30 1583 25 20 15 1307 Intensity 980 1238 10 763 5 1133 1358 1524 819 0 0 500 1000 1500 2000 1/cm

(c) singlet 2,6-difluorophenyl nitrene (1N)

Figure 9.2. The IR spectra of (a) ketenimine (K), (b) triplet nitrene (3N), and (c) singlet nitrene (1N), calculated at the B3LYP/6-31G* level of theory using the PCM model for acetonitrile. An open shell (UB3LYP) description was used for the singlet and triplet nitrene.

192 9.1.4 Experimental results obtained by Sarah Mandel

Time Resolved Infrared (TRIR) spectroscopy was used for the first time to study the solution phase photochemistry of an aryl azide and to detect the reactive intermediates generated. Laser flash photolysis (LFP, 266 nm) of 2,6-difluorophenyl azide released open-shell singlet 2,6-difluorophenylnitrene (1N, 1404 cm-1) which was detected in cyclohexane-d12 at 283 K and in CD3CN at 253 and 273 K. At ambient temperature, it was possible to detect the products of singlet nitrene decay, which are the isomeric ketenimine (K) at 1576 cm-1 and the lower energy triplet nitrene at 1444 cm-1.

Diethylamine and dimethylsulfoxide can scavenge the singlet nitrene k= 4.2 x 107 M-1s-

1 and k= 7.2 x 107 M-1s-1, respectively and reduce the yield of triplet nitrene and ketenimine. TRIR spectroscopy was used to measure the absolute rate constant for the reaction of ketenimine with diethylamine (9.0 x 104 M-1s-1) and of triplet nitrene with azide precursor (6.9 x 106 M-1s-1). The spectroscopic assignments were consistent with

DFT calculations.

9.1.4 Conclusions

The solution phase photochemistry of 2,6-difluorophenyl azide was studied by

Time Resolved Infrared (TRIR) Spectroscopy. The spectroscopic assignments were consistent with DFT calculations.

193

9.2. Phthalic acids and their complexes with iron

9.2.1. Introduction

In soil, complexation between organic acids and minerals plays an important role.

The metal-organic complex formation can decrease the biodegradability of natural organic matter (NOM),12 which is significant in global carbon cycling.13 The coating of organic acids may also inhibit dissolution and weathering,14 as well as affect the surface electrostatic properties of clays.15

The focus of this study is to understand the bonding mode and structure of phthalic acid, a surrogate for natural organic matter, at the hematite/water interface. The structure of phthalic acid is show below.

Attenuated total reflectance Fourier-transform infrared (ATR-FTIR) spectroscopy16 was employed to obtain vabrational frequencies of absorption of phthalic acid at the hematite/water interface. To better understand the IR spectra obtained, quantum mechanical calculations have been applied.

194 9.2.2 Computational Methods

Density Functional Theory (DFT)17 has been applied in this study. All of the structures have been fully optimized at the B3LYP level of theory, with 6-31+G* basis set for carbon, oxygen and hydrogen atoms, and Stuttgart SDD basis set18 for iron atoms.

All of IR spectra have been obtained by vibrational frequency calculations at the same level of theory, and the calculated frequencies have been scaled by a factor of 0.9613.19

Polarizable continuum model (PCM)20 calculations were performed to study the influence of water.

9.2.3 Results

9.2.3.1 The calculations for phthalic acid

The different forms of phthalic acid including dianion, monoanion, and neutral phthalic acid have been calculated and compared with the ATR-FTIR spectroscopic spectrum. The ATR-FTIR spectrum for aqueous phthalic acid at different pH value has been done by YuSik Huang, as illustrated in Figure 9.3. At high pH, phthatic acid has two C-O stretching peaks at around 1550 and 1400, respectively. At low pH, the two C-O peaks are replaced by C=O peak and C-OH, at 1709 and 1294, respectively.

195

Figure 9.3. The ATR-FTIR spectrum for aqueous phthalic acid at different pH value.

196

We compared the ATR-FTIR spectrum of phthalic acid at pH 6.5 with calculated vibrational frequencies of dianionic phthalic acid in the gas and aqueous phases, as shown in Figure 9.4. The match of two C-O peaks suggests that at pH 6.5, the dianionic form of phthalic acid is the dominant form. At pH 4.5, the comparison of the ATR-FTIR spectrum with calculated vibrational frequencies of the monoanionic form of phthalic acid in the gas and aqueous phases, as shown in Figure 9.5, suggests that at pH 4.5, it is more like a mixture of different forms, rather than dominated by only the mono anionic form of phthalic acid.

197

Figure 9.4 Comparison of ATR-FTIR spectrum of phthalic acid at pH 6.5 (top) with dianionic form of phthalic acid in the gas phase (middle) and the aqueous phase (bottom).

198

Figure 9.5 Comparison of ATR-FTIR spectrum of phthalic acid at pH 4.5 (top) with mono-anionic form of phthalic acid in the gas phase (middle) and the aqueous phase (bottom).

199 For neutral phthalic acid, there are three possible conformers as shown below.

H

O H H O O H H

C C C H H H O C C H C C C O C C C C O

C C C C C O C O H O H C C C H C C H C H

H H H O O O H

To compare with experimental frequencies, a Boltzmann-weighed, a theoretically derived spectrum was utilized. The weighting of each structure was determinined via a

Boltzmann average18 as shown below.

Here, Gi is the free energy of structure i relative to the structure with the lowest

free energy set as zero; gi is the structure degeneracy; kB B is Boltzmann’s constant and T is temperature. The comparison of ATR-FTIR spectrum of phthalic acid at pH 2.5 with neutral phthalic acid at gas phase and aqueous phase, as shown in Figure 9.6, reveals that at pH 2.5, the dominant species is the neutral phthalic acid. Gas phase and aqueous phase calculations have no significant difference.

200

Figure 9.6 Comparison of ATR-FTIR spectrum of phthalic acid at pH 4.5 (top) with neutral phthalic acid in the gas phase (middle) and the aqueous phase (bottom).

201 9.2.3.2 Phthalic acid – iron complex

The ATR-FTIR spectrum for Phthalic Acid adsorbed on hematite at different pH value has been obtained by YuSik Huang, as illustrated in Figure 9.7.

Figure 9.7 The ATR-FTIR spectrum for phthalic acid adsorbed on hematite at different pH value

At different pH values, the peaks are similar to the peaks of pure phthalic acid at pH 6.5, exhibiting only small variations with pH. We have revealed that at pH 6.5, the dominant species of phthalic acid is the dianionic form of phthalic acid. Thus we believe that the phthalic acid – iron complex should be formed between dianionic phthalic acid

202 and iron. After interpreting the IR results, two complexes were proposed: a dominant outer-sphere complex at the 1407 cm-1, and a minor inner-sphere complex at 1417 cm-1, as illustrated in Figure 9.8.

Figure 9.8 Second derivative of spectrum of adsorbed phthalic acid.

Two surface models, inner-sphere model and outer-sphere model, have been calculated to help provide assignment of the IR spectrum. The inner-sphere model has the iron directly connected to the phthalic acid, while the outer-sphere model has water in between the carboxylic acid and the iron. The structures of these two models are shown in Figure 9.9. The inner sphere model is on the left, and the outer-sphere model is on the 203 right.

H O H O H H O H H O H H H H O O

O H H C H O H C H H H O C H C H O Fe O O H C C H H H O O O Fe O C H C C C O H H H O H O C C O H H H C H O O C H H Fe H O C O C O Fe O H H H H O O O H H O H O H H H H H H H O H O O H

(a) “Inner-sphere complex” model (b) “outer-sphere complex” model

Figure 9.9. The structure of inner-sphere complex model and outer-sphere complex model between phthalic acid and Fe.

The IR spectrum of these two models have been calculated. The comparison of

ATR-FTIR spectrum of phthalic acid absorbed on the hematite at pH 3.47 with the spectrum calculated by DFT for the inner-sphere and outer-sphere complex models are shown in Figure 9.10.

204

Figure 9.10 Comparison of ATR-FTIR spectrum of phthalic acid absorbed on the hematite at pH 3.47 (top) with outer-sphere complex model (middle) and inner-sphere complex model (bottom).

205 The calculated complex has more peaks than the experimental IR spectrum because explict water molecules have been used to probe the aqueous solvation effect.

The extra peaks are the water vibrations. The experimental IR spectrum has eliminated these peaks as background. In comparison of the bands around 1400 cm-1, the outer- sphere and inner-sphere model are at 1387 and 1412 cm-1, respectively, which is in fair agreement with the experimental prediction of 1407 and 1417 cm-1.

9.2.4 Conclusions

Attenuated total reflectance Fourier-transform infrared (ATR-FTIR) spectroscopy was employed to obtain vibrational frequencies of phthalic acid, a surrogate for natural organic matter (NOM), adsorbed at the hematite/water interface. To better understand the

IR spectra obtained, quantum mechanical calculations have been applied. Dianionic, monoanionic and neutral phthalic acids and their complexes to iron have been modeled by Density Functional Theory (DFT). Polarizable Continuum Model (PCM) calculations were performed to study the influence of aqueous solvation. The structure and vibrational frequencies of two surface complexes, outer-sphere and inner sphere variants, were calculated. The computational results are compared to experimental IR frequency data, and these results have confirmed that spectrum assignment of the outer- sphere complex and inner-sphere complex.

206 9.3 Naphthyl and biphenyl azides

Thermolysis or photolysis of the naphthyl azides21 and biphenyl azides22 could generate naphthylnitrenes and biphenyl nitrenes, respectively. To better understand the excited states of the naphthyl and biphenyl azides, time-dependent density functional theory (TD-DFT)23 calculations were performed to predict the vertical transition of 1- naphthyl azide (79a), 2-naphthyl azide (79b), naphthalene (79c), as well as ortho- biphenyl azide (80a), para-biphenyl azide (80b), and biphenyl (80c). All structures were optimized at the B3LYP/6-31G* level of theory, and the TD-DFT calculations were performed at the B3LYP/6-31+G** level of theory. All of the calculations were performed with Gaussian 038 at the Ohio Supercomputer Center.

N3 N3

79a 79b 79c

N3

N3

80a 80b 80c

The calculated absorption maxima (λmax) (above 250 nm) and oscillator strengths of these molecules are given in Table 9.2.

207

Compound Excited state No. Dominant contribution λmax f 79a 1 44-46 352 0.0002 2 44-45 316 0.1807 3 44-47 290 0.0053 4 44-48 261 0.0029 79b 1 44-46 347 0.0001 2 44-45 310 0.0294 3 44-45 294 0.1326 4 43-46 271 0.0001 5 43-45 252 0.6290 79c 1 34-35 284 0.0609 80a 1 51-52 334 0.0007 2 51-54 280 0.0372 3 51-53 274 0.1861 4 50-52 263 0.0792 5 51-55 251 0.0046 80b 1 51-53 345 0.0002 2 51-52 290 0.6822 3 51-54 272 0.0466 4 51-55 259 0.0009 80c 1 41-42 255 0.4473

Table 9.2 TD-DFT calculated absorption maxima (λ, nm) and oscillator strengths (f) for 79a, 79b, 79c, 80a, 80b, 80c at B3LYP/6-31+G** level of theory.

208 The orbitals involved in the vertical transitions of 1-naphthyl azide are given in

Figure 9.11. All of the transitions are π −> π* transitions. The strongest absorption is at

316 nm, with the transition from HOMO orbital to LUMO orbital. It is also noted that the

LUMO+1 orbital only involves π orbitals on N=N=N units, which may be responsible for the extrusion of molecular nitrogen and the generation of 1-naphthylnitrene. In the case of 2-naphthyl azide, it has a similar LUMO+1 orbital, as shown in Figure 9.12. The orbitals involved in transitions of ortho- and para-biphenyl azides are shown in Figure

9.13-14. Unsprisingly, all of the transitions are π −> π* transitions. But for ortho- biphenyl azide, the LUMO orbital is the orbital only involes π character on N=N=N units.

209

Figure 9.11. The orbitals involved in vertical transitions of 1-naphthyl azide 79a. The relative orbital energies are in kcal/mol.

210

Figure 9.12. The orbitals involved in vertical transitions of 2-naphthyl azide 79b. The relative orbital energies are in kcal/mol.

211

Figure 9.13. The orbitals involved in vertical transition of naphthlene 79c. The relative orbital energies are in kcal/mol.

212

Figure 9.14. The orbitals involved in vertical transitions of ortho-biphenyl azide 80a. The relative orbital energies are in kcal/mol.

213

Figure 9.15. The orbitals involved in vertical transitions of para-biphenyl azide 80b. The relative orbital energies are in kcal/mol.

214

Figure 9.16. The orbitals involved in vertical transitions of biphenyl azide 80b. The relative orbital energies are in kcal/mol.

215 9.4 References for Chapter 9

1. Gritsan, N.P.; Platz, M.S. Adv. Phys. Org. Chem. 2001, 36, 255.

2. Gritsan, N.P.; Gudmundsdottir, A.D.; Tigelaar, D.; Zhu, Z.; Karney, W.L.; Hadad, C.M.; Platz, M.S. J. Am. Chem. Soc. 2001, 123(9), 1951.

3. a) Becke, A. D. J. Chem. Phys. 1993, 98, 5658. b) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785-789. c) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys Lett. 1989, 157, 200.

4. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

5. (a) Miertus, S.; Scrocco, E.; Tomasi, J. Chem Phys. 1981, 55, 117. (b) Miertus, S.; Tomasi, J. Chem. Phys.1982, 65, 239. (c) Barone, V.; Cossi, M.; Tomasi, J. J. Chem. Phys. 1997, 107, 3210. (d) Barone, V.; Cossi, M.; Tomasi, J. J. Comput. Chem. 1988, 19, 505. (e) Tomasi, J.; Persico, M. Chem. Rev. 1995, 95, 2027.

6. Roos, B. O. Int. J. Quant. Chem. Symp. 1987, 15, 175.

7. Frisch, M. J. et al. Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

8. Frisch, M. J. et al. Gaussian 03; Gaussian, Inc.: Pittsburgh, 2003.

9. (a) Cramer, C. J.; Dulles, F. J.; Giesen, D. J.; Almlof, J. Chem. Phys. Lett. 1995, 255, 165. (b) Cramer, C. J. J. Am. Chem. Soc. 1998, 120, 6261.

10. (a) Karney, W.L.; Borden, W.T. J. Am. Chem. Soc. 1997, 119, 1378. (b) Karney, W.L.; Borden, W.T. J. Am. Chem. Soc. 1997, 119, 5059.

11. (a) Kim, S.-J. I.; Hamilton, T.P.; Schaefer, H. F. III; J. Am. Chem. Soc. 1992, 114, 5349. (b) Hrovat, D.A.; Waal, E.E.; Borden, W.T.; J. Am. Chem. Soc. 1992, 114, 8698. (c) Travers, M.J.; Cowles, D.C.; Clifford. E.P.; Ellison, G.B. J. Am. Chem. Soc. 1992, 114, 8699.

12. Keil, R. G.; Hedges, J. I. Chem. Geol. 1993, 107, 385.

13. Torn, M. S.; Trumbore, S. E.; Chadwick, O. A.; Vitousek, P. M.; Hendricks, D. M. Nature, 1997, 389, 170.

14. Ochs, M. Chem. Geol. 1996, 132, 119.

15. Pignatello, J. J. Soil Sci. Soc. Am. 1989, 22, 45.

216

16. Harrick, N. J. Internal Reflection Spectroscopy. New York, NY, John Wiley & Sons, Inc. 1967.

17. Ziegler, T. Chem. Rev. 1991, 91, 651.

18. Fuentealba, P.; Preuss, H.; Stoll, H.; Szentpaly, L. V. Chem. Phys. Lett. 1989, 89, 418.

19. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

20. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

21. (a) E. F. V. Scriven in “Reactive Intermediates” R. A. Abramovitch, Ed., Plenum: New York, 1982, Vol. 2, Chapt. 1. b) P. A. S. Smith, in “Azides and Nitrenes - Reactivity and Utility” E. F. V. Scriven, Ed., Academic Press Inc.:New York, 1984, Chapt. 3.

22. P. A. S. Smith, B. B. Brown, J. Am. Chem. Soc. 1951, 73, 2435.

23. Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys. 1998, 108, 4439.

217

BIBLIOGRAPHY

1. (a) Abramovitch, R. A.; Davis, B. A. Chem. Rev. 1964, 64, 149. (b) Lwowski, W. in Azides and Nitrenes, Scriven, E. F. V., ed.; Academic Press, New York, N. Y. 1984, pp205-206.

2. (a) Abramovitch, R. A.; Kyba, E. P. “The Chemistry of the Azido Group.” S. Patai, ed., Interscience, New York, NY 1971. (b) L’Abbe, G.; Chem. Rev., 1969, 69, 345

3. (a) Anderson, K.; Malmqvist, P.-Å.; Roos, B. O.; Sadlej, A. J.; Wolinski, K. J. Phys. Chem. 1990, 94, 5483. (b) Anderson, K.; Malmqvist, P.-Å.; Roos, B. O. J. Chem. Phys. 1992, 96, 1218. (c) Anderson, K.; Roos, B. O. Int. J. Quantum Chem. 1993, 45, 591.

4. (a) Barckholtz, C.; Barckholtz, T. A.; Hadad, C. M.; J. J. Am. Chem. Soc. 1999, 121, 491. (b) http://webbook.nist.gov/chemistry/

5. (a) Becke, A. D. J. Chem. Phys. 1993, 98, 1372. (b) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B. 1988, 37, 785.

6. (a) Breslow, D. S.; Edwards, E. I.; Linsay, E. C.; Omura, H. J. Am. Chem. Soc. 1976, 98, 4268. (b) Breslow D. S.; Sloan, M. F.; Newburg, N. R.; Renfrow, W. B. J. Am. Chem. Soc. 1969, 91, 2273. (c) Sloan, M. F.; Renfrow, W. B.; Breslow, D. S. Tetrahedron Lett. 1964, 2905.

7. (a) Cizek, J. Adv. Chem. Phys. 1969, 14, 35. (b) Barlett, R. J. J. Phys. Chem. 1989, 93, 1963. (c) Raghavachari, K.; Trucks, G. W.; Pople, J. A.; Head-Gordon, M. Chem. Phys. Lett. 1989, 157, 479.

8. (a) Crow, W. D.; Wentrup, C. Tetrahedron Lett. 1968, 6149. (b) Wentrup, C. J. Chem. Soc., Chem. Commun. 1969, 1386. (c) Kuzaj, M.; Lüerssen, H.; Wentrup, C. Angew. Chem. Int. Ed. Engl. 1986, 25, 480. (d) Wentrup, C. Topics Current Chem. 1976, 62, 173.

9. (a) Fairchild, P. W.; Smith, G. P.; Crosly, D. R.; Jeffries, J. B. Chem. Phys. Lett. 1984, 107, 181. (b) Engelking, P. C.; Lineberger, W. C. J. Chem. Phys. 1976, 65, 4323.

218 10. (a) Fukui, K. Acc. Chem. Res. 1981, 14, 363; (b) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523.

11. (a) Gritsan, N. P.; Pritchina, E. A. Mendeleev Comm. 2001, 94. (b) Shapley, W. A.; Bacskay, G. B. J. Phys. Chem. A 1999, 103, 6625. (c) Faustov, V. I.; Baskir, E. G.; Biryukov, A. A. Russ. Chem. Bull. Int. Ed. 2003, 52, 2328.

12. (a) Gritsan, N. P.; Pritchina, E. S. J. Inf. Record. Mater. 1989, 17), 391 (b) Liang, T.-Y., Schuster, G. B. J. Am. Chem. Soc. 1987, 109, 7803 (c) Pritchina, E. A.; Gritsan, N. P. J. Photochem. Photobiol., A: Chemistry 1988, 43, 165. (d) Gritsan, N. P.; Pritchina, E. A. Russ. Chem. Rev. 1992, 61, 910.

13. (a) Hrovat, D. A.; Waali, E. E.; Borden, W. T. J. Am. Chem. Soc. 1992, 114, 8698. See also Castell, O.; García, V. M.; Bo, C.; Caballolo, R. J. Comput. Chem. 1996, 17, 42-48. (b) Gritsan, N. P.; Zhu, Z.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc. 1999, 121, 1202.

14. (a) Kaplan, F.; Meloy, G. K. J. Am. Chem. Soc. 1966, 88, 950. (b) Kaplan, F.; Mitchell, M. L. Tetrahedren Lett. 1979, 759.

15. (a) Karney, W.L.; Borden, W.T. J. Am. Chem. Soc. 1997, 119, 1378. (b) Karney, W.L.; Borden, W.T. J. Am. Chem. Soc. 1997, 119, 5059.

16. (a) Kim, S.-J. I.; Hamilton, T.P.; Schaefer, H. F. III; J. Am. Chem. Soc. 1992, 114, 5349. (b) Hrovat, D.A.; Waal, E.E.; Borden, W.T.; J. Am. Chem. Soc. 1992, 114, 8698. (c) Travers, M.J.; Cowles, D.C.; Clifford. E.P.; Ellison, G.B. J. Am. Chem. Soc. 1992, 114, 8699.

17. (a) Labanowski, J. W.; Andzelm, J. Density Functional Methods in Chemistry; Springer: New York, 1991. (b) Parr, R. G.; Yang, W. Density Functional Theory in Atoms and Molecules; Oxford University Press: New York, 1989.

18. (a) Lwowski, W.; Mattingly, T. W., Jr. J. Am. Chem. Soc. 1965, 87, 1947. (b) Lwowski, W.; Woerner, F. P. J. Am. Chem. Soc. 1965, 87, 5491. (c) McConaghy, J. S.; Lwowski, W. J. Am. Chem. Soc. 1965, 87, 5490.

19. (a) Lwowski, W.; Scheiffele, E. J. Am. Chem. Soc. 1963, 87, 4359. (b) Abramovitch, R. A.; Azogu, C. I.; Mcmaster, I. T. J. Am. Chem. Soc. 1969, 91, 1219. (c) Ayyangar, N. R.; Bambal, R. B.; Nikalje, D. D.; Srinivasan, K. V. Can. J. Chem. 1985, 63, 887. (d) Abramovitch, R. A.; Knavs, G. N. J. Org. Chem. 1975, 40, 883. (e) Horner, L.; Christmann, A. Chem. Ber. 1963, 96, 388.

20. (a) Lwowski, W.; Tisue, G. T. J. Am. Chem. Soc. 1965, 82, 4022. (b) Tisue, G. T.; Linke, S.; Lwowski, W. J. Am. Chem. Soc. 1967, 89, 6303. (c) Linke, S.; Tisue, G. T.; Lwowski, W. J. Am. Chem. Soc. 1967, 89, 6308.

219 21. (a) Miertus, S.; Scrocco, E.; Tomasi, J. Chem Phys. 1981, 55, 117. (b) Miertus, S.; Tomasi, J. Chem. Phys.1982, 65, 239. (c) Barone, V.; Cossi, M.; Tomasi, J. J. Chem. Phys. 1997, 107, 3210. (d) Barone, V.; Cossi, M.; Tomasi, J. J. Comput. Chem. 1988, 19, 505. (e) Tomasi, J.; Persico, M. Chem. Rev. 1995, 95, 2027.

22. (a) Pritchina, E. A.; Gritsan, N. P.; Maltsev, A.; Bally, T.; Autrey, T.; Liu, Y.; Wang, Y.; Toscano, J. P. Phys. Chem. Chem. Phys. 2003, 5, 1010. (b) Pritchina, E. A.; Gritsan, N. P.; Bally, T. Russ. Chem. Bull. Int. Ed. 2004, 53, 0000.

23. (a) Qi, B.; Su, K.; Wang, Y.; Wen, Z.; Tang, X. Wuli Huaxue Xuebao 1998, 14, 1033. (b) Subhan, W.; Rempala, P.; Sheridan, R. S. J. Am. Chem. Soc. 1998, 120, 11528 (c) Gutbrod, R.; Kraka, E.; Schindler, R. N.; Cremer, D. J. Am. Chem. Soc. 1997, 119, 7330. (d) Warner, P. M. J. Org. Chem. 1996, 61, 7192. (e) Cremer, D.; Gauss, J.; Kraka, E.; Stanton, J. F.; Bartlett, R. J. Chem. Phys. Lett. 1993, 209, 547. (f) Bach, R. D.; Andres, J. L.; Owensby, A. L.; Schlegel, H. B.; McDouall, J. J. W. J. Am. Chem. Soc. 1992, 114, 7207. (g) Cremer, D.; Schmidt, T.; Sander, W.; Bischof, P. J. Org. Chem 1989, 54, 2515. (h) Gauss, J.; Cremer, D. Chem. Phys. Lett. 1987, 133, 420. (i) Cremer, D. J. Am. Chem. Soc. 1979, 101, 7199.

24. (a) Roth, H. D. Acc. Chem. Res. 1977, 10, 85. (b) DoMinh, T.; Strausz, O. P.; Gunning, H. E. J. Am. Chem. Soc. 1969, 91, 129. (c) Tomioka, H.; Okuno, H.; Izawa, Y. J. Org. Chem. 1980, 45, 5278. (d) Rando, R. R. J. Am. Chem. Soc. 1970, 92, 6706. (e) Wulfman, D. S.; Poling, B.; McDaniel, R. S., Jr. Tetrahedron Lett. 1975, 4519.

25. (a) Salem, L.; Rowland, C. Angew. Chem. Int. Ed. Engl., 1972, 11, 92. (b) Michl, J. J. Am. Chem. Soc. 1996, 118, 3568 (c) Zimmerman, H. E.; Kutataladze, A. G. J. Org. Chem. 1996, 61, 6926. (d) Kita, F.; Nau, W. M.; Adam, W.; Wirz, J. J. Am. Chem. Soc. 1999, 121, 9265. (e) Johnson, W. T. G.; Sullivan, M. B.; Cramer, C. J. Int. J. Quant. Chem. 2001, 85, 492.

26. (a) Scott, A. P.; Nobes, R. H.; Schaefer, H. F., III; Radom, L. J. Am. Chem. Soc. 1994, 116, 10159. (b) Vacek, G.; Galbraith, J. M.; Yamaguchi, Y.; Schaefer, H. F., III; Nobes, R. H.; Scott, A. P.; Radom, L. J. Phys. Chem. 1994, 98, 8660. (c) Scott, A. P.; Platz, M. S.; Radom, L. J. Am. Chem. Soc. 2001, 123, 6069.

27. (a) Smolinsky, G.; Wasserman, E.; Yager, W. A. J. Am. Chem. Soc. 1962, 84, 3320. (b) Moriarty, R. M.; Rahman, M.; King, G. J. J. Am. Chem. Soc. 1966, 88, 842. (c) Wasserman, E. Prog. Phys. Org. Chem. 1971, 8, 319.

28. (a) Travers, M. J.; Cowles, D. C.; Clifford, E. P.; Ellison, G. B.; Engelking, P. C. J. Phys. Chem. 1999, 111, 5349. (b) Kemnitz, C. R.; Ellison, G. B.; Karney, W. L.; Borden, W. T. J. Am. Chem. Soc. 2000, 122, 1098. (c) Barash, L.; Wasserman, E.; Jager, W. A. J. Chem. Phys. 1967, 89, 3931. For a correction, see Ferrante, R. F. J. Chem. Phys. 1987, 86, 25. (d) Radziszewski, J. G.; Downing, J. W.;

220 Wentrup, C.; Kaszynski, P.; Jawdosiuk, M.; Kovacic, P.; Michl, J. J. Am. Chem. Soc. 1985, 107, 2799.

29. (a) Wallis, E.S. Org. Reactions. 1946, 3, 267. (b) Bauer, E. Angew. Chem. Int. Ed. Engl. 1974, 13, 376.

30. (a) Wilson, W. R. Tumor hypoxia: challenges for cancer chemotherapy; Wilson, W. R., Ed.; Kluwer Academic: Lancaster, 1992. (b) Denny, W. A. The Lancet Oncol. 2000, 1, 25.

31. (a) Karney, W. L.; Borden, W. T. Adv. Carbene Chem. Vol3, p205, Elsevier, New York, NY 2001. (b) Kemnitz, C. R.; Karney, W. L.; Borden, W. T. J. Am. Chem. Soc. 1998, 120, 3499.

32. (a)Cramer, C. J.; Dulles, F. J.; Giesen, D. J.; Almlof, J. Chem. Phys. Lett. 1995, 255, 165. (b) Cramer, C. J. J. Am. Chem. Soc. 1998, 120, 6261.

33. (a) Labanowski, J. W.; Andzelm, J. Density Functional Methods in Chemistry; Springer: New York, 1991. (b) Parr, R. G.; Yang, W. Density Functional Theory in Atoms and Molecules; Oxford University Press: New York, 1989.

34. Sheridan, R. S.; Rempala, P. A nitrene to nitrene rearrangement: Photochemistry of benzenesulfonyl azide. Book of Abstracts, 217th ACS National Meeting, Anaheim, Calif., March 21-25 (1999).

35. a) Becke, A. D. J. Chem. Phys. 1993, 98, 5658. b) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785-789. c) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys Lett. 1989, 157, 200.

36. Abranovitch, R. A.; Holcomb, W. D. Chem. Comm. 1969,1298.

37. Andersson, K. et al. MOLCAS Version 5; Lund University: Sweden, 2000.

38. Armstrong, B. M.; McKee, M. L.; Shevlin, P. B. J. Am. Chem. Soc. 1995, 117, 3685.

39. Autrey, T.; Schuster, G. B. J. Am. Chem. Soc. 1987, 109, 5814.

40. Autrey, T.; Schuster, G. B. J. Am. Chem. Soc. 1987, 109, 5814. (b) Wasserman, E. Prog. Phys. Org. Chem. 1971, 8, 319. (c) Pritchina, E. A.; Gritsan, N. P.; Maltsev, A.; Bally, T.; Autrey, T.; Liu, Y.; Wang, Y.; Toscano, J. P. Phys. Chem. Chem. Phys. 2003, 5, 1010. (d) Pritchina, E. A.; Gritsan, N. P.; Bally, T. Russ. Chem. Bull. Int. Ed. 2004, 53, 0000. (e) Hayashi, Y.; Swern, D. J. Am. Chem. Soc. 1973, 95, 5205. (f) Puttner, R.; Hafner, K. Tetrahedron Lett. 1964, 3119. (g) Semenov, V. P.; Studenikov, A. N.; Bespalov, A. D.; Ogloblin, K. A. J. Org. Chem. USSR ( Engl. Transl.) 1977, 12, 2052. (h) Inagaki, M.; Shingaki, T.; Nagai, T. Chem. Lett. 1981, 1419. (i) Inagaki M.; Shingaki, T; Nagai, T. Chem. Lett. 1982, 9. 221 41. Balabanov, G. P.; Yu, I.; Dergunov; Gal’perin, V. A. J. Org. Chem. (USSR) 1966, 2, 1797

42. Ban, F.; Gauld, J.W.; Boyd, R.J. J. Am. Chem. Soc. 2001, 123, 7320.

43. Barckholtz, C.; Barckholtz, T. A.; Hadad, C. M. J. Am. Chem. Soc. 2001, 121, 491.

44. Borden, W. T.; Gritsan, N. P.; Hadad, C. M.; Karney, W. L.; Kemnitz, C. R.; Platz, M. S. Acc. Chem. Res. 2000, 33, 765.

45. Breen, A. P.; Murphy, J. A. Free Rad. Biol. Med. 1995, 18, 1033.

46. Brown, J. M. Br. J. Cancer 1993, 67, 1163.

47. Brown, J. M. Cancer Res. 1999, 59, 5863.

48. Brown, J. M.; Siim, B. G. Seminars Rad. Oncol. 1996, 6, 22.

49. Brown, J. M.; Wang, L.-H. Anti-Cancer Drug Design 1998, 13, 529.

50. Brown, J. M.; Wilson, W. R. Nature Rev. 2004, 4, 437.

51. Brown. J. M. Br. J. Cancer. 1993, 67, 1163.

52. Bucher, G.; Scaiano, J. C.; Platz, M. S. Kinetics of Carbene Reactions in Solution. Landolt-Bornstein, Group II, Volume 18, Subvolume E2 Springer, Berlin, Germany, (1998), p.141.

53. Bump, E. A.; Brown, J. M. Pharmac. Ther. 1990, 47, 117.

54. Buron, C.; Platz, M. S. Org. Lett. 2003, 5, 3383.

55. Casida, M. E.; Jamorski, C.; Casida, K. C.; Salahub, D. R. J. Chem. Phys. 1998, 108, 4439.

56. Daniels, J. S.; Gates, K. S.; Tronche, C.; Greenberg, M. M. Chem. Res. Toxicol. 1998, 11, 1254.

57. Daniels, J. S.; Gates, K.S. J. Am. Chem. Soc. 1996, 118, 3380.

58. Denny, W. A. Eur. J. Med. Chem., 2001, 36, 577.

59. Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007.

222 60. Elwell, J.H.; Siim, B.G.; Evans, J.W.; Brown, J.M. Biochem. Pharmacl. 1997, 54, 249.

61. Frisch, M. J. et al Gaussian 98; Gaussian, Inc.: Pittsburgh, 1998.

62. Frisch, M. J. et al. Gaussian 03; Gaussian, Inc.: Pittsburgh, 2003.

63. Garay, J-C.; Maloney, V.; Marlow, M.; Small, P. J. Phys. Chem. 1996, 100, 5788.

64. Gilles, A.; Mesanet, J.; Vermeil, C. Chem. Phys. Lett. 1974, 23, 346

65. Greenberg, M. M. Chem. Res. Toxicol. 1998, 11, 1235.

66. Gritsan, N.P.; Gudmundsdottir, A.D.; Tigelaar, D.; Zhu, Z.; Karney, W.L.; Hadad, C.M.; Platz, M.S. J. Am. Chem. Soc. 2001, 123(9), 1951.

67. Gritsan, N.P.; Platz, M.S. Adv. Phys. Org. Chem. 2001, 36, 255.

68. Harrick, N. J. Internal Reflection Spectroscopy. New York, NY, John Wiley & Sons, Inc. 1967.

69. Holt, P. M.; Kerr, J. A. Int J Chem Kinet 1977, 9, 185.

70. Jensen, F. Introduction to Computational Chemistry; Wiley: Chichester, 1998..

71. Keil, R. G.; Hedges, J. I. Chem. Geol. 1993, 107, 385.

72. Kemnitz, C. R.; Ellison, G. B.; Karnoy, W. L.; Borden, W. T. J. Am. Chem. Soc. 2000, 122, 1098.

73. Kirmse, W. Eur. J. Org. Chem. 2002, 14, 2193.

74. Laderoute, K.L.; Wardman, P.; Rauth, M. Biochem. Pharmacl. 1988, 37, 1487

75. Liu, J.; Mandel, S.; Hadad, C. M.; Platz, M. S. J. Org. Chem. 2004, 69, 8583.

76. Liu, J.; Popik, V.; Hadad, C. M.; Platz, M. S. J. Am. Chem. Soc., submitted.

77. Lorance, E. D.; Kramer, W. H.; Gould, I. R. J. Am. Chem. Soc. 2002, 124,15225

78. Lwowski, W. in “Azides and Nitrenes” Scriven, E.F.V, ed. Academic Press, New York, N.Y. 1984, 205

79. Mack, H; Della Vedova, C. O.; Willner, H. J. Mol. Struct. 1993, 291, 197.

223 80. Makareeva, E. N.; Lozovskaya, E. L.; Zelentsov, S. V. High Energy Chem. 2001, 35, 177.

81. Montgomery, J. A., Jr.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys. 1999, 110, 2822.

82. Ochs, M. Chem. Geol. 1996, 132, 119.

83. P. A. S. Smith, B. B. Brown, J. Am. Chem. Soc. 1951, 73, 2435.

84. Patterson, A. V.; Saunders, M. P.; Chinje, E. C.; Paterson, L.H.; Statford, I. J. Anti-Cancer Drug Design 1998, 13, 541.

85. Pignatello, J. J. Soil Sci. Soc. Am. 1989, 22, 45.

86. Platz, M. S. "Issues and challenges in the chemistry of alkylcarbenes" in Advances in Carbene Chemistry, Brinker, U. H., Ed, 1998, volume 2, pp. 133- 174.

87. Platz, M. S. Acc. Chem. Res. 1995, 28, 487.

88. Reed, A. E.; Weinhold, F.; Curtiss, L. A. Chem. Rev. 1988, 88, 899.

89. Regitz, M. Diazo Compounds: Properties and Synthesis, Academic Press, New York, N. Y., 1986.

90. Regitz, M.; Maas, G. Diazo Compounds; Academic Press, New York, N.Y., 1986,

16-23 and references therein.

91. Reichle, W-T. Inorg. Chem. 1964, 3, 402.

92. Rondan, N. G.; Houk, K. N.; Moss, R. A. J. Am. Chem. Soc. 1980, 102, 1770.

93. Roos, B. O. Int. J. Quant. Chem. Symp. 1987, 15, 175.

94. Scott, A. P.; Platz, M. S.; Radom, L. J. Am. Chem. Soc. 2001, 123, 6069.

95. Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

96. Shinde, S. S.; Anderson, R. F.; Hay, M. P.; Gamage, S. A.; Denny, W. A.; J. Am. Chem. Soc.; 2004, 126, 7865.

97. Sigman, M. E.; Autrey, T.; Schuster, G. B. J. Am. Chem. Soc. 1988, 110, 4297.

98. Skell, P. S.; Woodworth, R. C. J. Am. Chem. Soc. 1956, 78, 4496.

224 99. Srinivasan, A.; Kebede, N.; Saavedra, J. E.; Nikolaitchik, A. V.; Brady, D. A.; Yourd, E.; Davies, K. M.; Keefer, L. K.; Toscano, J. P. J. Am. Chem. Soc. 2001, 123, 5465. 100. Stratford, I. J.; Workman, P. Anticancer Drug Des., 1998, 13, 519.

101. Stratmann, R. E.; Scuseria, G. E.; Frisch, M. J. J. Chem. Phys. 1998, 109, 8218.

102. Sulzbach, H. M.; Platz, M. S.; Schaefer, H. F. III; Hadad, C. M. J. Am. Chem. Soc. 1997, 119, 5682-5689

103. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.

104. Tomasz, M. Chem. Biol. 1995, 2, 575.

105. Torn, M. S.; Trumbore, S. E.; Chadwick, O. A.; Vitousek, P. M.; Hendricks, D. M. Nature, 1997, 389, 170.

106. Toscano, J. P. "Laser flash photolysis studies of carbonyl carbenes" in Advances in Carbene Chemistry, Brinker, U. H., Ed, 1998, volume 2, pp. 215-244.

107. Travers, M. J.; Cowles, D. C.; Clifford, E. P.; Ellison, G. B. J. Am. Chem. Soc. 1992, 114, 8699.

108. Travers, M. J.; Cowles, D. C.; Clifford, E. P.; Ellison, G. B.; Engelking, P. C. J. Chem. Phys. 1999, 111, 5349.

109. Turro, N. J.; Butcher, J. A.; Moss, R. A.; Guo, W.; Munial, R. C.; Fedorynski, M. J. Am. Chem. Soc. 1980, 102, 7576.

110. Vaupel, P.; Kallinowski, F.; Okunieff, P. Cancer Res. 1989, 49, 6449.

111. Wardman, P.; Priyadarsini, K. I.; Dennis, M. F.; Everett, S. A.; Naylor, M. A.; Patel, K. B.; Stratford, I. J.; Stratford, M. R. L.; Tracy, M. British Journal of Cancer, Supplement. 1996, 74, S70.

112. Wardman, P.; Priyadarsini, K. I.; Dennis, M. F.; Everett, S. A.; Naylor, M. A.; Patel, K. B.; Stratford, I. J.; Stratford, M. R. L.; Tracy, M. British Journal of Cancer, Supplement. 1996, 74(27), S70-S74.

113. Wasserman, E. Prog. Phys. Org. Chem., 1971, 8, 319.

114. Wiberg, K. B.; Hadad, C. M.; Rablen, P. R.; Cioslowski, J. J. Am. Chem. Soc. 1992, 114, 8644.

115. Zelentsov, S. V.; Zelentsova, N. V.; Shchepalov, A. A. High Energy Chemistry. 2002, 36, 326.

225 116. Ziegler, T. Chem. Rev. 1991, 91, 651.

226