An Experimental and Computational Comparison of the Reactions of N2O and CO2 in Several Inorganic Systems

Please do not remove this page

Davis, Jack Vickery https://scholarship.miami.edu/discovery/delivery/01UOML_INST:ResearchRepository/12386108050002976?l#13386108040002976

Davis, J. V. (2021). An Experimental and Computational Comparison of the Reactions of N2O and CO2 in Several Inorganic Systems [University of Miami]. https://scholarship.miami.edu/discovery/fulldisplay/alma991031605661102976/01UOML_INST:ResearchR epository

Open Downloaded On 2021/10/02 02:14:36 -0400

Please do not remove this page UNIVERSITY OF MIAMI

AN EXPERIMENTAL AND COMPUTATIONAL COMPARISON OF THE REACTIONS OF N2O AND CO2 IN SEVERAL INORGANIC SYSTEMS

Jack Vickery Davis

A DISSERTATION

Submitted to the Faculty of the University of Miami in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Coral Gables, Florida

August 2021

UNIVERSITY OF MIAMI

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

AN EXPERIMENTAL AND COMPUTATIONAL COMPARISON OF THE REACTIONS OF N2O AND CO2 IN SEVERAL INORGANIC SYSTEMS

Jack Vickery Davis

Approved:

______Carl D Hoff, Ph.D. Roger M. Leblanc, Ph.D. Professor of Chemistry Professor of Chemistry

______Burjor K. Captain, Ph.D. Guillermo Prado, Ph.D. Associate Professor of Chemistry Dean of the Graduate School

______Manuel Temprado, Ph.D. Associate Professor of Chemistry University of Alcalá

DAVIS, JACK VICKERY (Ph.D., Chemistry) An Experimental and Computational Comparison of the (August 2021) Reactions of N2O and CO2 in Several Inorganic Systems

Abstract of a dissertation at the University of Miami.

Dissertation supervised by Professor Carl D. Hoff. No. of pages in text. (129)

A series of four inorganic systems were examined to gain insight into the complex

bonding patterns of nitriles, CO2 and N2O and their potential to be used as a chemical

feedstock. In the first system, a thorough computational comparison of nitriles (RCN)

t t binding to V(N[ Bu]Ar)3 (VL3) and Mo(N[ Bu]Ar)3 (MoL3) (Ar=3,5-Me2C6H3) revealed

the importance of the ligand environment and electronic state. In the first complex,

kinetic measurements of various nitriles (RCN) binding to VL3 revealed two distinct

binding mechanisms dependent on R. Unencumbered aryl nitriles such as C6H5CN and

2,6-F2C6H3CN bound by an associative mechanism drawn from thermodynamically

favorable π-interactions between the arene of the nitrile and anilide ligand. This is

reflected in their experimentally determined activation parameters, having a relatively

low ∆H‡ (~3 kcal/mol) but a more negative ∆S‡ (-27 cal/mol K). Alkyl nitriles such as

MeCN and AdCN cannot make these favorable interactions and bind with a relatively

higher ∆H‡ (7 kcal/mol) but a lower ∆S‡ (-10 cal/mol K). The net of these effects resulted

in nearly identical ∆G‡ (9 kcal/mol) at -40°C.

The reactivity of (R3Sn)2O (R= Ph, Cy) toward N2O and CO2 was examined in a

detailed experimental and computational study. N2O binding to (R3Sn)2O to form trans-

(R3Sn)2N2O2 is predicted to be unfavorable at all temperatures for both Ph and Cy derivatives. The rate and activation parameters of the reverse reaction of N2O extrusion

from trans-(R3Sn)2N2O2 was studied through thermal decomposition of the hyponitrite

complexes in an FTIR cell. CO2 was found to reversibly bind to (R3Sn)2O for both Ph

and Cy at room temperature to form their respective (R3Sn)2CO3. Both of these reactions

were studied computationally and a mechanism for both is proposed.

Kinetic studies on the carboxylative cyclization of a propargylamine by Au(IPr)X

were performed to gain insights into the mechanism of CO2 incorporation by a gold catalyst. A variety of experiments were run to determine optimal reaction conditions, mechanistic details and activation parameters. An important vinyl gold intermediate was synthesized and characterized crystallographically. The relative reaction rates of different gold compounds [Au(IPr)X] X= (Cl, vinyl) and [Au2(L)X2] were compared, with the

digold achieving the highest reaction rates.

The mechanochemical milling of Na2O and N2O in the presence of other additives is

reported. Mechanochemical milling of these samples produces cis-Na2N2O2 in moderate yields at room temperature in few hours. Further milling leads to production of NaNO3 at

yields approaching 50% after 50 hours. This work represents the first total oxidation of

nitrous oxide to nitrate and may significant industrial relevance.

Acknowledgements

The work herein would not have been possible without the endless help from my family. I thank my wife Cristina for constantly pushing me forward and supporting me in more ways than I can count. I thank my parents for always being there for me and pushing me since I was a little boy and for always supporting me in all my life choices.

I thank my mentor and advisor Dr. Carl Hoff for his constant support over the past five years. Without his mentorship and guidance this work would not have been possible.

Through the early struggles of reactions that did not go as planned to our current more successful work I learned many lessons that will help me throughout my career and have shaped me into being a more resilient and well-rounded individual.

I thank Dr. Burjor Captain for his useful suggestions in obtaining quality crystals and his work in solving our crystal structures and I thank Dr. Manuel Temprado for large theoretical and computational contributions and his computational chemistry mentorship.

I also like to thank the students who have helped me along the way. From my first two years I thank Leo for his mentorship in computational chemistry and making beautiful figures. I thank Mohan for his endless help in our weeks long reactions and many crystallographic analysis attempts. I thank Oswaldo for his help in our mixer mill studies and for forcing me to learn how to teach again. Finally, I thank everyone at the

University of Miami Chemistry Department for support and guidance when I needed it.

iii

TABLE OF CONTENTS

Page

LIST OF FIGURES ...... vi

LIST OF SCHEMES...... xi

LIST OF TABLES ...... xii

Chapter 1: The Use of Catalysts to Produce Cleaner, More Efficient Reactions

1.1 Introduction and Purpose ...... 1

1.2 The Use of Transition Metals in Small Molecule Activation……...………… 2

1.3 Reactions of N2, N2O and CO2……………………………………………. ... 4

Chapter 2: The Mechanism of Binding of Nitriles to VL3 and ML3 Complexes

2.1 Background ...... 8

2.2 Results and Discussion ...... 10

2.3 Conclusion ...... 32

2.4 Experimental Details ...... 33

Chapter 3: Comparative Pathways for Elimination of N2O or CO2 from Trialkyltin Hyponitrites and Carbonates

3.1 Background ...... 35

3.2 Results and Discussion ...... 37

3.3 Conclusion ...... 56

3.4 Experimental Details ...... 57

Chapter 4: The Mechanism of Carboxylative Cyclization of Propargylamine by N- Heterocyclic Carbene Complexes of Au(1)

4.1 Background ...... 68

4.2 Results and Discussion ...... 73

4.3 Conclusion ...... 102

4.4 Experimental Details ...... 102

Chapter 5: Ball Milling Reaction of Na2O and Na2O2 with N2O

5.1 Background ...... 106

5.2 Results and Discussion ...... 108

5.3 Conclusion ...... 115

5.4 Experimental Details ...... 116

References…………… ...... 120

List of Figures

Figure 1.1. Projected growth in atmospheric CO2 fraction (left) and the projected increase in global surface temperature in the 21st century (right)...... 1

Figure 1.2 The total oxidation of N2O to NaNO3 using sodium peroxide and more N2O as oxidants is calculated to be favorable by 155.7 kcal/mol at room temperature in the gas phase...... 5

Figure 2.1 Summary of reactions and structures of key complexes in binding and splitting of dinitrogen by complex MoL3...... 10

Figure 2.2 Solid-state structure of DFBN-1 (left) and Me2NCN-1 (right) with thermal ellipsoids at 50% probability...... 12

Figure 2.3 Time-resolved spectra of DFBN (1 mM) binding to 1 (0.3 mM) at −44 °C, acquired over 2 s...... 13

Figure 2.4 Single wavelength kinetic traces of aromatic RCN (1 mM) binding to 1 (0.3 mM) at −44°C (DFBN and PhCN at λ = 687 nm; MesCN at λ = 705 nm)...... 14

Figure 2.5 Second order rate plot for DFBN binding to 1 at various concentrations (1-10 mM) over a temperature range of −62 °C to −35 °C with [1]0 = 0.3 mM...... 14

Figure 2.6 Eyring plots for reaction of 1 with several nitriles...... 16

Figure 2.7 Optimized structures of the two most stable configurations of 1 with the anilide ligands adopting a three-down (configuration A, left) or a one-down, two-up arrangement (configuration B, right)...... 17

Figure 2.8 Optimized structures of 1 interacting with a benzene molecule. Hydrogen atoms omitted for clarity...... 18

Figure 2.9 Optimized structure of the transition state for binding MeCN to 1 in the A configuration. Hydrogen atoms omitted for clarity...... 22

Figure 2.10 Shown is the increase in relative energy as V-N distance is elongated for DFBN, AdCN, MeCN and MesCN binding to 1...... 23

Figure 2.11 Optimized configurations at fixed V···Nnitrile distances of (A): 2.08 (B): 2.58, (C): 3.08 Å showing a linear dissociation of the nitrile ligand from AdCN-1...... 24

vi Figure 2.12 Optimized configurations at fixed V···Nnitrile distances of (A): 2.05 (B): 2.55, (C): 3.05 Å...... 24

Figure 2.13 Energy (z axis, kcal·mol−1), V–Nnitrile distance (x axis, Å), and V–N–C angle (degrees) for dissociation of AdCN (left) and DFBN (right)...... 25

Figure 2.14 Plot of the center point distance between the top three C atoms of the arene of the anilide ligand and the bottom three C atoms of the arene of DFBN as a function of V···N distance...... 26

Figure 2.15 Plot of ∆S‡ (cal·mol−1·K−1) versus ∆H‡ (kcal·mol−1) for binding of nitriles to 1 (left) and 2 (right). All data are for nitriles except for AdNC (green diamond)...... 28

9 7 8 Figure 3.1 Structures of the IPr (left), IPr-CO2 (center) and IPr-N2O (right) highlighting the differing architectures for the adducts...... 36

Figure 3.2 An ORTEP of the molecular structure of 1Cy showing 40% probability thermal ellipsoids. Complete crystallographic data and files are available in section 4.4...... 38

Figure 3.3 Comparison of the first order reaction profiles for thermal decomposition of compounds 1Ph and 1Cy at 77.9 °C...... 40

Figure 3.4 Eyring plots for thermal decomposition of compounds 1Ph and 1Cy...... 41

Figure 3.5 Computed thermodynamic profile (kcal·mol−1) for stoichiometric reduction of Ph two moles of 1 promoted by •Cr(CO)3C5Me5...... 43

Figure 3.6 FTIR spectra obtained for the reaction of compound 1Ph with 2.5 equivalents of HCr(Cp*)(CO)3 in methylene chloride at room temperature. The steady growth of N2O is shown at 2219 cm-1...... 45

Figure 3.7 (a) FTIR data showing carbonate bands near 1605 and 1329 cm−1 that increase Cy Cy with decreasing temperature as 2 is converted into 3 (left). (b) Plot of ln(Keq) versus Cy Cy 1/T for binding of CO2 to 2 and produce 3 in methylene chloride solution (right). ....46

Figure 3.8 Schematic Gibbs energy and enthalpy diagram for elimination of N2O from 1Me computed at the B3LYP-D3(BJ)/Def2-TZVP level of theory...... 48

Ph Figure 3.9 Optimized structure of the transition state (TS2 ) for N2O elimination from 1Ph...... 50

vii Me Figure 3.10 Schematic Gibbs energy and enthalpy diagram for addition of CO2 to 2 and subsequent isomerization to the final product 3Me computed at the B3LYP- D3(BJ)/Def2-TZVP level of theory...... 51

−1 Figure 3.11 Reaction profile (kcal·mol ) for the unfavorable addition of N2O to H2O and Me3SnOSnMe3...... 54

Ph Figure 3.12 Reaction between 1 and •Cr(CO)3C5Me5...... 60

-1 -1 Figure 3.13 CO2 emission (2338cm ) in the reaction outperforms N2O (2219cm ) by an order of magnitude. Peaks attributed to the proposed compound CrNNOSn (2005, 1931, 1672 cm-1)...... 61

Ph Figure 3.14 Shown are NMR of 1 in CD2Cl2 (blue), after 20 minutes of reaction with 1.2mg of •Cr (red), after 40 minutes of reaction (green) and the following day (purple)...... 62

Figure 3.16 Schematic Gibbs energy and enthalpy diagram for an alternative isomerization reaction to yield Int1a from 1Mea...... 67

Figure 4.1 Proposed mechanism for conversion of [Au(IMe)Cl] to an intermediate gold carbamate complex followed by conversion to Au(IMe)(Vinyl) in one of two rate determining processes during catalysis...... 69

Figure 4.2 Proposed mechanism for protonation of IKa with three potential adducts between PPA and an added acid: A) PPA-HCl, B) PPA-CO2 , and C) PPA-CA...... 72

Figure 4.3 An ORTEP showing the molecular structure of 1 at 40% thermal ellipsoid probability...... 73

Figure 4.5 Rate of product formation under catalytic conditions [PPA]0 = 0.256 M, CO2 pressure = 2 atm (absolute) in EtOH at 21°C using [1] 0 = 0.00686M (red squares) compared to [2] 0 = 0.00702M (black squares)...... 76

Figure 4.6 Initial rates of carboxylation of [1]0 = 0.0072 M, [PPA]0 = 0.052 M, T = 21°C in anhydrous EtOH at pressure of 1, 2, and 3 atm absolute CO2 pressure...... 77

viii Figure 4.7 FTIR spectroscopic data after ≈30 minutes reaction time for reaction of 1 (0.007M) and PPA (.052 M) in EtOH at 21 °C and 1.1, 2.0, and 3.0 atm pressure (absolute)...... 79

Figure 4.8 Concentrations of 1 versus time added as a solid to solutions of PPA in EtOH under 15 psi (2 atm absolute pressure) CO2 at 21°C...... 80

Figure 4.9 Second order plot of 1 /[Product] versus time (min) for reaction of 1 with PPA-HCl with each reagent at 0.007M concentration in EtOH at 21°C and 2 atm pressure (absolute) CO2...... 83

Figure 4.10 First order plots for rate of production of the 2-oxazolidinone product in reaction of 1 with CO2 in EtOH at 21 °C and absolute pressures of 2 and 3 atm...... 84

Figure 4.11 Spectra showing the fall and rise of a broad band centered at 1667 cm−1 assigned to the proposed ethoxycarbonate complex [Au(IPr)(OCO2Et)]...... 86

Figure 4.12 Plots of initial rate of product production at 40 (black), 21 (red) and 0.6 (blue) °C under constant CO2 pressure of 2 atm (absolute), and standard initial concentrations of [PPA] 0.054 M and 1 = 0.007...... 90

Figure 4.13 Eyring plot for the generation of product data shown in Figure 4.12 under standard low PPA concentration conditions: [PPA] = 0.052 M, [1] = 0.007 M, in EtOH under 2 atm pressure (absolute) of CO2...... 91

Figure 4.14 Gibbs free energy diagram computed at the PBE0- t D3(BJ)(PCM,CH3OH)/Def2-TZVP level of theory for proton transfer from PPA•HCl to 1t and liberation of the oxazolidinone product ...... 92

Figure 4.15 Optimized structures computed for the transition states for reaction of 1t with CAt (left) and with PPA•HClt (right) at the PBE0-D3(BJ)(PCM,MeOH)/Def2-TZVP level of theory ...... 93

Figure 4.16 Gibbs free energy diagram computed at the PBE0- D3(BJ)(PCM,CH3OH)/Def2-TZVP level of theory for CO2 insertion into the [Au(IMe)(OMe)] complex...... 96

Figure 4.17 Optimized structures of [Au(IMe)(OCO2Me)] (left) compared to previously reported computed structure for Au(IMe)OC=ON(Me)(CH2C≡CCH3) (right)...... 97

Figure 4.18 Plots of the rate of production (M) of oxazolidinone product versus time (Min) at the 0.0065 M level of [1] at 21°C in EtOH...... 100

ix Figure 5.1 Comparison of FTIR data for ball milling results of Na2O/KBr under 3 atm 14 15 N2O (red) and N2O (black). The spectra shown were obtained after 6 hours of ball milling...... 109

Figure 5.2 KBr pellet FTIR spectra as a function of time for ball milling of Na2O (0.5g) in KBr (1.5g) under 30 psi, 2.0 ATM N2O...... 110

Figure 5.3 Initial rates of production of nitrate as a function of time (hours) for KF, KBr, KCl, and KI...... 111

Figure 5.4 Mole fraction in total Na distribution for the first 20.5 hours of ball milling 0.5 g Na2O in 1.5 g KBr in standard conditions...... 113

Figure 5.5 Initial rates of nitrate production as a function of time for samples of 1.5g KBr and 0.5g Na2O2 (blue), 0.25g Na2O2 + 0.25g Na2O, and 0.5g Na2O...... 114

Figure 5.6 Pellet spectra obtained after about 8 hours of milling corrected for nitrate concentration...... 115

x List of Schemes

Scheme 2.1 Computed thermodynamic data for interconversion between the A and B configurations of L-1 (L = N2 (top), PhCN (bottom)). Enthalpy and Gibbs energy values in kcal·mol−1 and entropy in cal·mol−1·K−1...... 19

Scheme 2.2 Thermochemical values for the reaction of the most stable structure from Figure 2.8 and DFBN containing a π-stacking interaction with a benzene molecule to yield complex DFBN-1 and the most stable tilted T-shape benzene dimer...... 21

Scheme 3.1 Binding and dissociation reactions studied in this work...... 37

Ph Scheme 3.2 Proposed mechanism for N2O elimination from 1 promoted by •Cr(Cp*)(CO)3...... 42

Scheme 3.3 Isomerization reactions for the 1Me hyponitrite...... 48

Scheme 4.1 Pd-catalyzed cyclization of amines with carbon monoxide ...... 68

Scheme 4.2 [Au]-catalyzed carboxylative cyclization of PPA ...... 69

Scheme 4.3 Synthesis of a series of dinuclear gold(I) complexes ...... 70

Scheme 4.4 The CA/CS Equilibrium ...... 71

xi List of Tables

Table 2.1 Bimolecular rate constants (kon) and activation parameters measured for coordination of various nitriles to 1...... 15

Table 2.2 Computed Thermochemical data for Ligand Binding to 1 and 2. ∆H and ∆G values in kcal·mol−1 and ∆S in cal·mol−1·K−1...... 20

Table 2.3 . Comparison of experimental and computational thermochemical data for RCN-1 or RNC-1 adduct formation from ligand addition to 1 and 2...... 21

Cy Table 3.1 Selected intramolecular distances and angles for 1 ...... 38

Table 3.2 Experimental activation parameters obtained for thermal decomposition of compounds 1Ph and 1Cy...... 41

Table 3.3 Crystallographic data for compounds 1Cy ...... 63

Table 3.4 Selected intramolecular distances and angles for compound trans- (Cy3Sn)2N2O2...... 64

Table 3.5 Selected intramolecular distances and angles for compound trans- (Ph3Sn)2N2O2 ...... 65

Table 3.6 Selected intramolecular distances and angles for compound [(OEP)Fe]2(μ- N2O2) ...... 65

Table 4.1 Crystallographic Data for Compound 1...... 73

Table 4.2 Selected intramolecular distances and angles for compounds 1, IKA-Me and IKA-Ph...... 74

Table 4.3 Data used for calculation of k2 for rate of cleavage of IKa by HCarb of 0.54 ± 0.7 M-1 min-1...... 81

Table 4.4 Structural parameters computed for the transition states for protonation of complex 1t. a Values corresponding to an O…H distance. b Values corresponding to a N…H distance...... 94

14 Table 5.1 Tabulated peak assignments for balling milling reactions of Na2O and N2O or 15 N2O...... 109

xii Chapter 1 The Use of Catalysts to Produce Cleaner, More Efficient Reactions 1.1 Introduction and Purpose

From the smelting of metal to the production of glass and now more modern

processes of oil refining, plastic production or fertilizer synthesis chemical manufacture

has been at the forefront of societal innovation throughout human history. These

innovations have allowed an increase in world population from just 600 million in the

year 1700 to almost 8 billion today. These chemical innovations have afforded quality of

life improvements throughout the world such as better access to food and water, faster

transportation, stronger infrastructure, and lower costs of living.

These gains have not come without a cost; accumulation of plastics in the ocean has developed into the single greatest threat to large marine mammals and has permeated the entire oceanic ecosystem.[1] Plastic waste in the ocean is projected to surpass the mass of

Figure 1.1. Shown is the projected growth in atmospheric CO2 fraction (left) and the projected increase in global surface temperature in the 21st century (right).

1 2

[2] all fish by 2050. Carbon dioxide (CO2) emissions from the burning of fossil fuels have

steadily climbed in every year recorded and as a result global temperature is forecast to

increase 3°C during the 21st century.[3]

Industrial production of fertilizer has led to enormous increases in farmland

productivity but as a result has introduced problems with algae blooms and more recently

nitrous oxide (N2O) emissions. N2O or laughing gas is a potent greenhouse gas,

absorbing 298 times more heat compared with CO2 over a 100 year time frame. N2O is a known ozone depleting substance (ODS) and with the global adoption of the Montreal

Protocol, it has become the largest contributor to ozone depletion.[4]

Worldwide efforts to mitigate the effects of climate change are currently being

assessed on the world stage however large-scale adoption of greenhouse gas

sequestration has yet to be adopted. It seems most likely that the best adoption of climate

change mitigation will come through economically feasible alternatives. The purpose of

the research described herein is to investigate the use of small waste molecules such as

CO2 and N2O as feedstock in the chemical industry to reduce the anthropogenic impact of manufacturing on the environment.

1.2 The Use of Transition Metals in Small Molecule Activation

Improvements in chemical manufacture have often come with the development of a

catalyst. Catalysts are additives in a chemical reaction which promote increases in

reaction rate and in certain circumstances, lowering of required reaction temperature and

or pressure. Both of which promote both a faster and more energetically efficient

reaction. Today almost all major commercial chemical processes are performed with the 3

use of a catalyst. Perhaps the most relevant to the work herein is the catalytic production

+ - of the fertilizer ammonium nitrate, [NH4] [NO3] .

Ammonium nitrate is created by the simple 1:1 acid-base reaction of ammonia with

nitric acid. Production of ammonia and nitric acid is more complicated; nitric acid is

produced through a multistep oxidation of ammonia called the Ostwald process. A

reaction scheme of this process can be seen in reactions 1, 2 and 3. [5]

4 ( ) + 5 ( ) 4 ( ) + 6 ( ) ( = 216.3 / ) (1)

𝑁𝑁𝑁𝑁3 𝑔𝑔 𝑂𝑂2 𝑔𝑔 → 𝑁𝑁𝑁𝑁 𝑔𝑔 𝐻𝐻2𝑂𝑂 𝑔𝑔 ∆𝐻𝐻 − 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 𝑚𝑚𝑚𝑚𝑙𝑙 2 ( ) + ( ) 2 ( ) ( = 28.0 / ) (2)

𝑁𝑁𝑁𝑁 𝑔𝑔 𝑂𝑂2 𝑔𝑔 → 𝑁𝑁𝑂𝑂2 𝑔𝑔 ∆𝐻𝐻 − 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 𝑚𝑚𝑚𝑚𝑚𝑚 4 ( ) + ( ) + 2 ( ) 4 ( ) ( = 83.2 / ) (3)

𝑁𝑁𝑁𝑁2 𝑔𝑔 𝑂𝑂2 𝑔𝑔 𝐻𝐻2𝑂𝑂 𝑙𝑙 → 𝐻𝐻𝑁𝑁𝑂𝑂3 𝑎𝑎𝑎𝑎 ∆𝐻𝐻 − 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 𝑚𝑚𝑚𝑚𝑚𝑚 The first step, oxidation of ammonia to nitric oxide occurs at between 4-10 atmospheres of pressure and between 600-800°C. These conditions and the use of a platinum and rhodium catalyst allow for yields of 98% in step 1, with the remaining 2% lost as N2 and N2O. The nitric oxide is then further oxidized and hydrolyzed in steps 2

and 3 to yield nitric acid. The overall reaction for this and the resulting ammonium nitrate

can be seen in reaction 4.

( ) + 2 ( ) ( ) + (4)

𝑁𝑁𝑁𝑁3 𝑔𝑔 𝑂𝑂2 𝑔𝑔 → 𝐻𝐻𝐻𝐻𝑂𝑂3 𝑎𝑎𝑎𝑎 𝐻𝐻2𝑂𝑂 As ammonia is the only nitrogen source for ammonium nitrate it is important to

understand how ammonia is made as well. Ammonia is produced via the Haber process;

where nitrogen and hydrogen gas are combined at high temperature and pressure

(250-350 atm and 450-550°C) in the presence of an iron catalyst, usually ferrite.[6] This

reaction can be seen in reaction 5. 4

( ) + 3 ( ) 2 ( ) ( = 22.0 / ) (5)

𝑁𝑁2 𝑔𝑔 𝐻𝐻2 𝑔𝑔 → 𝑁𝑁𝑁𝑁3 𝑔𝑔 ∆𝐻𝐻 − 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 𝑚𝑚𝑚𝑚𝑚𝑚 Hydrogen (H2) is another industrial feedstock produced through the steam

reformation of natural gas. Steam reformation produces almost all the hydrogen in the

United States and involves the high temperature and pressure reaction of methane with

steam on the surface of a nickel catalyst.[7] This can be seen in reaction 6.

( ) + ( ) ( ) + 3 ( = +49.2 / ) (6)

𝐶𝐶𝐶𝐶4 𝑔𝑔 𝐻𝐻2𝑂𝑂 𝑔𝑔 → 𝐶𝐶𝐶𝐶 𝑔𝑔 𝐻𝐻2 ∆𝐻𝐻 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 𝑚𝑚𝑚𝑚𝑚𝑚 Combining reactions 4, 5, and 6 shows that the production of ammonium nitrate is achieved through air, water and natural gas.

( ) + ( ) + 2 ( ) ( ) + ( ) ( = 300.3 / ) (7)

𝐶𝐶𝐶𝐶4 𝑔𝑔 𝑁𝑁2 𝑔𝑔 𝑂𝑂2 𝑔𝑔 → 𝐶𝐶𝐶𝐶 𝑔𝑔 𝑁𝑁𝐻𝐻4𝑁𝑁𝑁𝑁3 𝑠𝑠 ∆𝐻𝐻 − 𝑘𝑘𝑘𝑘𝑘𝑘𝑘𝑘 𝑚𝑚𝑚𝑚𝑚𝑚 The seemingly simple production of ammonium nitrate requires three major steps

with different catalysts and conditions: steam reforming of natural gas, reduction of N2 to

ammonia and oxidation of ammonia to nitric acid. None of the steps would be possible

without the use of transition metal catalysts.

1.3 Reactions of N2, N2O and CO2

It is important to remember the scale of industrial synthesis; in 2017 global

ammonium nitrate production was estimated at 21.6 million tones.[8] The two main

processes, the Haber Process and the Ostwald Process are over 100 years old and despite

optimizations, inefficiencies still exist. As mentioned in section 1.1, N2O emissions are a

significant contribution to climate change and are produced during fertilizer synthesis. If

a 98% yield in the oxidation of ammonia to nitric oxide is assumed and half of the waste 5

produced is N2, that leaves more than 200,000 tonnes of N2O produced per year from fertilizer synthesis alone.

Current N2O abatement protocol call for the decomposition or reduction of N2O

with NH3 to N2 and either O2 or H2O. The work of our group has been to find ways to instead take waste N2O and convert it into useful products, mainly nitrate. Despite over

100 years of optimization in the Ostwald Process very little research has gone into the oxidation of nitrous oxide. N2O can be thought of as a partially oxidized version of the

o very stable N2; as such it has a positive ∆Hf of +19.6 kcal/mol. This high energy is what makes N2O interesting as a potential industrial reagent; N2 has already been activated

and oxidation is heavily exothermic. A computed potential energy diagram for the total

oxidation of N2O to nitrate by sodium peroxide and N2O can be seen in figure 1.2.

Figure 1.2 The total oxidation of N2O to NaNO3 using sodium peroxide and more N2O as oxidants is calculated to be favorable by 155.7 kcal/mol at room temperature in the gas phase; this is discussed in detail in Chapter 5. 6

The reactions of N2O are dominated by N2O acting as an oxidant with loss of N2.

An exception to this is the capture of N2O into metal oxides (X2O) to form hyponitrites,

X2N2O2. Even examples of this are extremely rare; first reported by Jansen and Feldmann

in 1996, sodium oxide, Na2O was found to react with N2O gas at high temperature to

[9] form microcrystalline cis Na2N2O2. The work in Chapters 4 and 5 aim to undercover

the reactivity of hyponitrites; how they are formed and how to oxidize them further to

nitrates.[10,11]

In contrast with N2O, CO2 is heavily researched, and a multitude of reactions of a

metal oxide complex with CO2 leading to the corresponding carbonate are known. Metal

carbonates are generally formed in a reversible reaction of the metal oxide with CO2

directly. An example of this is calcium carbonate or limestone; when limestone is heated

[12] to 1000°C, CO2 is released, and calcium oxide or lime is formed. CO2 and N2O share

the same electronic structure; they are said to be isoelectronic however their chemical

reactivity is significant different. Hyponitrites are not known to react reversibly, often

when a hyponitrite is heated N2O is permanently lost and to my knowledge no study has ever been conducted directly comparing carbonates and hyponitrites. Chapter 3 reports on the utilization of CO2 as a feedstock in tangent with a gold catalyst to produce a

biologically relevant compound. Utilizing the knowledge gained from this reaction,

Chapter 4 reports on a direct comparison of trialkyltin carbonates and hyponitrites.[13,14]

The goal of most work relating to both the Haber and Ostwald Processes is to

decrease the activation requirements for the production of ammonia and therefore the

production of nitric acid. Currently natural gas is used to produce hydrogen, which is then 7

reacted at very high temperature and pressure. The work performed on the oxidation of

N2O to nitrate could theoretically be expanded to direct oxidation of N2 to nitrate, bypassing the Haber Process and the natural gas requirement all together. No study has shown significant conversion of N2 or N2O to nitrate so far and a lot regarding the

chemistry of N2 and N2O remains unknown. Chapter 2 reports on a computational study

regarding the mechanisms of binding for nitriles to both vanadium trisanalide and

molybdenum trisanalide complexes. Nitriles serve as a weak ligand capable of mimicking

the reactivity of N2 and as such revealed the importance of spin state and ligand

environment in binding among transition metal complexes.[15] Chapter 2 The Mechanism of Binding of Nitriles to VL3 and ML3 Complexes

2.1 Background

The difference in reactivity of the FeMo- and FeV-cofactors of

nitrogenase has been extensively studied.[16] A recent structural determination has shown

that binding of CO to the FeV-cofactor occurs at two Fe centers and is remarkably similar

to that discovered earlier for the FeMo-cofactor.[17] However, the structure of the initially

formed dinitrogen adduct and the mechanism of its formation are not completely known.

Following the initial discovery of dinitrogen binding to transition metals by Allen and

Senoff in 1965,[18] a range of binding motifs and degrees of activation have been

discovered. The most common remains end-on coordination followed by bridging

dinuclear. Recent discoveries appear to indicate that side on/end on binding at multi

nuclear sites achieve greater degrees of activation.[19]

The closest analogs to end-bound dinitrogen are nitrile ligands. As a rule, nitriles

have several advantages over dinitrogen in terms of conventional metal-ligand bond

−1 formation. First, the enthalpy of protonation of N2 (118 kcal·mol ) is only 61% that of

PhCN (194 kcal·mol−1) in the gas phase[20] and it is generally held that nitriles are better

σ donors than dinitrogen. In addition, they are also generally accepted as being better π

acceptors (and donors) than dinitrogen itself.[21] For multi-site binding, however,

dinitrogen appears to provide a broader range of hapticities and structural

architectures.[19] An additional facet of nitrile binding is that they are “tunable” ligands and varying the substituent R of RCN can allow probing the steric and electronic factors of interaction with a vacant metal site that is not possible with N2. It should be kept in

8 9

mind that both, dinitrogen and nitriles, are “weak ligands” and for that reason small

factors which may be ignored as trivial for stronger ligands may prove critical in

determining complex stability.

The binding of dinitrogen and nitriles to the sterically crowded Kubas complexes

i [22] (M(PR3)2(CO)3, M = Cr, Mo, W, R = Pr, Cy) has been extensively studied and

provides a basis for comparison of the relative kinetic and thermodynamic affinities for

ligand binding of nitriles compared to dinitrogen for these low valent diamagnetic

complexes. For conventional binding a preference of ≈5 kcal·mol−1 for end-on nitrile versus end-on dinitrogen binding is typical.[23] However, there are few broad range

studies of the physical aspects of binding of nitrile ligands to higher valent paramagnetic

complexes. The work of Kovacs on nitrile hydratases provides kinetic and structural data

for an Fe(III) mimic of that enzyme.[24]

The current work provides insight into the nature of the binding site at

t [25] t [26] V(N[ Bu]Ar)3 (Ar=3,5-Me2C6H3), 1 and Mo(N[ Bu]Ar)3, 2. Neither of these

complexes forms an observable mononuclear N2 bound complex. In the case of 2,

reaction with dinitrogen, at room temperature and below, leads to formation of the N2-

bridged dinuclear which reacts further to form two equivalents of the metal nitride as

[26, 27] summarized in Figure 2.1. The rate of N2 uptake can be accelerated by addition of

bases or through redox cycles[28] but efforts to observe an initial adduct between 2 and

dinitrogen have been unsuccessful. None of the complexes shown in Figure 2.1 have

been isolated or detected for binding of dinitrogen to 1. However, multitude of examples

can be found in the literature with vanadium complexes promoting the fixation and, in

[29] some cases, activation of N2. For instance, binding of dinitrogen to an analogous 10 mononuclear three-coordinate vanadium(II) complex have been extensively explored by

Mindiola and coworkers.[29d]

N N V N ? N + N2 N t Ar[tBu]N M N[ Bu]Ar t Ar N[ Bu]Ar

- 1 N 1 2- (M = V) N2 2 (M = Mo) Ar[tBu]N t tBu Ar[tBu]N Mo N[ Bu]Ar t N Bu N + N2 N Ar t 2 N Mo 2 Ar[tBu]N Mo N[ Bu]Ar t N t Ar N Bu - N2 N[ t Bu]Ar Ar Ar[tBu]N Mo N[ Bu]Ar t N[ Bu]Ar 2

Figure 2.1 Summary of reactions and structures of key complexes in binding and splitting of dinitrogen by complex 2. [26,27] The mononuclear dinitrogen complex has not been detected. In this chapter, I report a detailed computational comparison of nitrile binding to 1 and 2. The results obtained for the V(III) species are compared to those previously reported for the analogous Mo(III) complex 2.[30] Despite experimental work revealing rather similar rate and activation parameters, computational investigation unmasked a complex set of factors that must be considered in the reaction mechanism.

2.2 Results and Discussion

2.2.1 Experimental Results

Experimental results were obtained through previous work done in collaboration with Elena Rybak-Akimova at Tufts University and Christopher C. Cummins at MIT. 11

The extent of my contributions are located in the computational section, however

experimental results are provided for clarity.

1 t 2.2.2 Preparation of η -RCN-V(N[ Bu]Ar)3 Complexes

Addition of nitriles to dark green-brown solutions of 1 in diethyl ether or toluene

results in an immediate color change to deep purple. The adduct with benzonitrile, PhCN-

t V(N[ Bu]Ar)3 (PhCN-1) can be isolated as purple crystals by recrystallization from n-

hexane at −35 °C, but repeated attempts to obtain crystals suitable for X-ray diffraction

studies were met without success. Reaction of 1 and 2,6-F2C6H3CN (DFBN) also resulted

in rapid formation of a purple color but, in this case, single crystals of DFBN-1 could be

grown from concentrated diethyl ether solutions. Reaction of 1 with Me2NCN in diethyl

ether results in formation of a deep blue/purple solution from which the product

Me2NCN-1 can be isolated as a cerulean blue solid by precipitation from n-pentane.

Crystals of Me2NCN-1 suitable for X-ray diffraction studies could be grown from concentrated solutions in a 1:1 mixture of toluene: diethyl ether at –35 °C. The 1H NMR spectra of isolated and purified nitrile complexes were collected in C6D6. All complexes gave rise to paramagnetically shifted and broadened resonances. No evidence was

2 obtained for formation of an isomeric η -complex of Me2NCN-1 even in low temperature

NMR studies at –80 °C in toluene-d8.

2.2.3 Crystal Structures of DFBN-1 and Me2NCN-1

The solid-state structure of DFBN-1 (Figure 2.2) contains a vanadium metal

center in a distorted trigonal pyramidal coordination geometry (Nanilide–V–Nnitrile

96.10(7)° avg., Nanilide–V–Nanilide 118.89(7)° avg.). The V–Nnitrile interatomic distance

(2.044(2) Å) is longer than the V–Nanilide distances (1.943(2) Å avg.). The anilide ligands 12

adopt a one-down, two-up arrangement, where one of the three aryl rings points

downward away from the nitrile and two point upward towards it. This conformation

likely arises from steric effects between the coordinated nitrile and the tert-butyl groups

of the amide ligands. Furthermore, there is a deviation from linearity in the angle V–N–C

(161.7(2)°) which is attributed to a π-stacking interaction between the F2Arene group on

the nitrile and the Me2Arene group of one of the anilides bound to vanadium, a behavior

t [30f] also previously noticed by us in the complex Ph(H)CN-Mo(N[ Bu]Ar)3.

Figure 2.2 Solid-state structure of DFBN-1 (left) and Me2NCN-1 (right) with thermal ellipsoids at 50% probability. Hydrogen atoms have been omitted for clarity.[15]

The solid-state structure of Me2NCN-1 determined by X-ray crystallography

(Figure 2.2) reveals an η1 binding mode for the cyanamide ligand, in contrast the adduct

2 of Me2NCN with 2 has been shown in previous work to yield a stable η -bound

derivative.[31] The anilide ligands adopt a crystallographically imposed three-fold symmetric arrangement, which is markedly different than that observed in the structure of

DFBN-1. The V–Nnitrile interatomic distance of 2.037(3) Å is comparable with the V– 13

Nnitrile distance of 2.044(2) Å observed in DFBN-1. The reported data are comparable to

related V(III) structures in the literature.[32]

2.2.4 Stopped Flow Kinetic Studies of Nitrile Binding to 1

The rapid binding kinetics of several nitriles (aromatic nitriles DFBN, PhCN,

MesCN, and aliphatic nitriles Me2NCN, AdCN, and MeCN) to 1 were investigated using stopped-flow methodology with spectrophotometric registration. The growth of visible absorption bands was observed in all reactions and nitrile binding was very clean and well-behaved. For example, the time-resolved spectra for DFBN binding (Figure 2.3)

reveals significant build up at λ = 525 nm and λ = 687 nm. Similar spectral changes were observed for other nitriles.

Figure 2.3 Time-resolved spectra of DFBN (1 mM) binding to 1 (0.3 mM) at −44 °C, acquired over 2 s. Selected traces shown for clarity. The initially recorded spectrum is shown in black and the final spectrum in red. Single-wavelength measurements were necessary to quantify rapid nitrile binding

to 1 at variable concentrations and temperatures (typically, −62 °C to −35 °C).

Representative kinetic traces recorded for aromatic nitrile binding in single wavelength

mode are shown in Figure 2.4. 14

Figure 2.4 Single wavelength kinetic traces of aromatic RCN (1 mM) binding to 1 (0.3 mM) at −44°C (DFBN and PhCN at λ = 687 nm; MesCN at λ = 705 nm).

Varying the concentration of the nitriles resulted in a linear increase in k1obs (Figure 2.5).

Figure 2.5 Second order rate plot for DFBN binding to 1 at various concentrations (1-10 mM) over a temperature range of −62 °C to −35 °C with [1]0 = 0.3 mM. This observed behavior corresponds to the reaction of reversible nitrile binding to 1 as described by equation 1.

+ RCN kon − V(N[tBu]Ar) RCN V(N[tBu]Ar) 3 k 3 (1) off

The dependence of k1obs = kon[RCN] + koff shown in Figure 2.5 reveals a linear relationship with the slope equal to kon and an intercept corresponding to koff. At lower temperatures, essentially zero intercept indicated that the equilibrium was completely shifted toward product formation. At higher temperatures such as −35 °C for DFBN, as shown in Figure 2.5, the reverse reaction, nitrile dissociation, became noticeable as indicated by an increasing intercept corresponding to koff. The accuracy in koff values (and 15

the derived values of Keq) is relatively low, as these values are determined from the intercepts of the dependencies of kobs on the concentrations of nitrile as shown in Figure

2.5. Kinetic estimates of Keq were not warranted. The slope of the graph indicative of kon is used for kinetic analysis of nitrile binding. Rate constant data are summarized in Table

2.1. Activation parameters for nitrile binding were calculated from Eyring plots seen in figure 2.6.

Table 2.1 Bimolecular rate constants (kon) and activation parameters measured for coordination of various nitriles to 1. For comparison purposes, data reported[30] for complex 2 between brackets.

‡ ‡ ‡ RCN kon(−40 °C) ∆H ∆S ∆G (−40 °C) (M−1·s−1) (kcal·mol−1) (cal·mol−1·K−1) (kcal·mol−1) PhCN (8.1 ± 0.2)·103 2.9 ± 0.6 −28 ± 3 9.4 [468 ± 22] [5.2 ± 0.2] [−24 ± 1] [10.8] DFBN (7.5 ± 0.2)·103 3.3 ± 0.2 −26 ± 1 9.4 [316 ± 3] [4.7 ± 0.4] [−26 ± 2] [10.8] MesCN (6.6 ± 0.3)·103 6.4 ± 0.3 −13 ± 2 9.4 [193 ± 14] [5.0 ± 0.3] [−26 ± 1] [11.1] 3 Me2NCN (15.5 ± 0.7)·10 5.6 ± 0.3 −15 ± 1 9.0 [708]a,b [6.4 ± 0.4]b [−18 ± 2]b [10.6]b MeCN (10.1 ± 0.5)·103 7.2 ± 0.3 −9 ± 2 9.2

AdCN (5.1 ± 0.5)·103 6.7 ± 0.8 −12 ± 4 9.6 [97]a [5 ± 1] [−28 ± 5] [11.5] AdNC (10.1 ± 0.9)·103 4.6 ± 0.3 −20 ± 1 9.2 [16·103]a [5.5 ± 0.5] [−15 ± 4] [9.0] a b Rate constants at −40 °C were extra- or interpolated from Eyring plots; kon and activation parameters represent formation of the end-on (η1) adduct. 16

Figure 2.6 Eyring plots for reaction of 1 with several nitriles.

Surprisingly the values of ∆G‡ at T = −40 °C for binding to 1 are approximately

constant at 9.3 ± 0.3 kcal·mol−1 as are those for 2 at 11.0 ± 0.5 kcal·mol−1 with the

exception of binding of AdNC to 2 which exhibits a lower value of 9.0 kcal·mol−1. This is the only ligand studied for which the rate of binding to 2 is faster than to 1. This is primarily associated with a less unfavorable entropy of activation.

2.2.5 DFT-Optimized Structures of 1

The optimized structures of the two most thermodynamically stable

conformations computed for 1 in the gas phase are shown in Figure 2.7. Configuration A is a “3 anilides-down” configuration analogous to that adopted by 2.[26] Configuration B

is a “2 anilides up-1 down” and contains a stabilizing interaction between V and an arene

group pendant on one of the anilide ligands analogous to the solid state structure of the

[33] related compound V(N[Ad]Ar)3 (Ar = 3,5-Me2C6H3). A computed difference of ≈ −4

kcal·mol−1 in enthalpy between both structures favoring configuration B was previously

reported,[30] however, that difference is lowered to −0.6 kcal·mol−1 when London dispersion corrections are included in the calculations.[34] 17

The difference in computed ∆S values is small but favors A over B and at room

temperature, the computed Gibbs energy differences are negligible. Moreover, the

interconversion process between both conformations has been computed to have a small

barrier in the order of 7 kcal·mol–1. Thus, computational data would predict the

establishment of a fast equilibrium between both configurations in the gas phase at room

temperature and below.

2.2.6 Optimized Structures for Interaction of C6H6 with Complex 1

The crystal structure of complex DFBN-1 in Figure 2.2 prompted computational

investigation of a plausible stabilization of 1 by interaction with the solvent. Two

different structures were optimized using the B configuration of 1 including explicitly a

benzene molecule to simulate the toluene solvent and are shown in Figure 2.8. 18

Figure 2.8 Optimized structures of 1 interacting with a benzene molecule. Hydrogen atoms omitted for clarity. The structure on the left contains a π-stacking stabilizing interaction analogous to that observed in the crystal structure of DFBN-1 shown in Figure 2.2 and the alternative structure on the right exhibits two C-H··· π interactions between benzene and the two arenes located in the same side of the plane defined by the three anilide nitrogens. In both cases, the interaction of the benzene molecule with the V center is not significant since the shorter distances between the two moieties are V···H = 2.92, V···C = 3.83 Å (Figure

2.7 left) and V···H = 3.45, V···C = 3.94 Å (Figure 2.8 right). These structures are stable with respect to enthalpy of reaction by −1 kcal·mol−1 (Figure 2.8 left) and −2.2

kcal·mol−1 (Figure 2.8 right) related to the B configuration of 1 shown in Figure 2.7. The

entropy change in conversion of the intercalated structure on the left to the structure on

the right was computed and found to be −0.6 cal·mol−1·K−1 indicating that the later

structure would be predicted to be slightly more stable with respect to Gibbs energy at

25°C.

2.2.6 Computed Enthalpies of A and B Interconversion of Complexes of 1

Since the incoming ligand can bind to 1 in configurations A or B (see Figure 2.7),

both configurations were computed for all adducts. Representative thermochemical data

for interconversion is shown in Scheme 2.1 for N2 and PhCN.

N N N ∆H = 0.3 N N V N ∆S = 12.1 N N V N ∆ = -3.4 G25 ºC N

C ∆H = -3.3 C ∆S = 0.6 N N = -3.4 ∆G N V N 25 ºC N V N N N

Configuration A Configuration B

Scheme 2.1 Computed thermodynamic data for interconversion between the A and B configurations of L-1 −1 (L = N2 (top), PhCN (bottom)). Enthalpy and Gibbs energy values in kcal·mol ·and entropy in cal·mol−1·K−1. The Gibbs energy changes for the A → B conversion are both −3.4 kcal·mol−1 at

25 °C, however, the isomerization is slightly endothermic for the N2 complex and

exothermic for PhCN. The difference with respect to entropy of interconversion is the

opposite, and favors N2 over PhCN. This is attributed to more favorable overlap and

some bonding interactions between the π* orbital on the nitrile and the anilide arenes.

This is less significant for N2 as is the overall intramolecular dispersion interaction which

also occurs. The formation of these stabilizing interactions, however, is entropically

disfavored. 20

2.2.7 Computed Thermochemical Data for Ligand Binding to 1 and 2

A summary of computed thermochemical data for ligand binding to 1 and 2 is

shown in Table 2.2. As it can be seen, the binding of alkyl nitrile MeCN to 1 is in the

range of 2-5 kcal·mol−1 less exothermic than the binding of aryl nitriles and this difference is even larger for nitrile binding to 2 (ΔΔH = 5-12 kcal·mol−1). Moreover, the

computed enthalpy of binding of the aryl nitriles studied to 1 span a range of only 2.5

kcal·mol−1, however, the nitrile binding to 2 is more sensitive to the substituent effect of

the incoming ligand and that range is expanded to 7.2 kcal·mol−1. In keeping with the larger range in enthalpies of nitrile binding to 2 compared to 1, greater discrimination in qualitative binding studies is observed for 2 than for 1.[26,30]

Table 2.2 Computed Thermochemical data for Ligand Binding to 1 and 2. ∆H and ∆G values in kcal·mol−1 and ∆S in cal·mol−1·K−1.a 1 2

L ΔH ΔS ΔG(25 °C) ΔH ΔS ΔG(25 °C)

N2 −1.4 −47.7 12.8 −3.8 −50.2 11.2 MeCN −14.2 −40.1 −2.2 −5.5 −51.9 10.0 b b b Me2NCN −18.3 −46.1 −4.6 −23.7 −63.0 −4.9 PhCN −17.7 −47.1 −3.7 −13.5 −57.1 3.5 MesCN −19.4 −57.6 −2.2 −13.1 −65.1 6.3 DFBN −16.4 −52.3 −0.8 −15.5 −61.7 2.9

C6F5CN −16.3 −55.9 0.4 −17.8 −65.2 1.7

4-CF3C6H4CN −17.5 −52.9 −1.7 −16.1 −60.7 2.0

4-Me2NC6H4CN −18.8 −45.9 −5.1 −10.6 −56.3 6.2 AdNC −25.8 −51.7 −10.4 −32.4 −63.5 −13.5 a Values reported with respect to the A configuration (see Figure 2.7) for complex 1 and b 2 [31] 2; Me2NCN forms an η species when bound. 21

Due to the weak computed binding for many ligands, as well as experimental difficulties,

limited comparison can be made between experiment and theory as summarized in Table

2.3.

ΔHcalc ΔHexpt ΔΔHexpt-calc DFBN-1 −16.4 −10.4±0.8 6.0

Me2NCN-2 −23.7 −22.0±1.0 1.7 PhCN-2 −13.5 −14.5±1.5 −1.0 AdNC-1 −25.8 −17.1±0.7 8.7 AdNC-2 −32.4 −29.1±0.4 3.3 Table 2.3 . Comparison of experimental and computational thermochemical data for RCN-1 or RNC-1 adduct formation from ligand addition to 1 and 2. Values in kcal·mol−1.

A reasonable correlation is seen between calculated and experimental data for 2.

In the case of complexes of 1, the computed data for ∆H reported with respect to the A

configuration are more exothermic by ≈ 7 kcal·mol−1. There is a stabilization of 1 by

interaction with the solvent and since it is known that fluorinated arenes exhibit enhanced

intermolecular π-stacking interactions,[35] thermochemical values for the reaction in

Scheme 2.2 were derived also from DFT calculations to simulate better what is occurring

in toluene solution in the binding of DFBN to 1.

F F C N V N N N N V N N ∆H = -10.5 kcal·mol-1

+ +

F N H F

As it can be seen in Scheme 2.2, the derived value for the enthalpy of binding is −10.5

kcal·mol−1 in perfect agreement with that obtained experimentally (−10.4 ± 0.8

kcal·mol−1).

2.2.8 Computed Transition State Structure for Binding MeCN to 1 in the A

Configuration

The transition state for approach of MeCN to complex 1 in the A configuration

was found and its optimized structure is shown in Figure 2.9. The activation parameters

derived (ΔH‡ = 2.5 kcal·mol−1 and ΔS‡ = −40.7 cal·mol−1·K−1) were in significant disagreement with the experimental data from Table 2.1 (ΔH‡ = 7.2 ± 0.3 kcal·mol−1 and

ΔS‡ = −9 ± 2 cal·mol−1·K−1). Attempts to locate the transition state for nitrile binding to 1

in the B configuration were not successful.

Figure 2.9 Optimized structure of the transition state for binding MeCN to 1 in the A configuration. Hydrogen atoms omitted for clarity. Selected distances (Å) and angles (degrees): V—N1 = 3.358; N1—C1 = 1.156 Å; V—N1—C1 = 146.8⁰. 23

2.2.9 Computed Paths for Nitrile Dissociation from RCN-1 Complexes in

the B Configuration

Search for the transition state structure for binding to B was made by use of the

principle of microscopic reversibility. Starting with the established structure shown in

Figure 2.2 for DFBN-1 in the B configuration, a relaxed scan was performed along the

V–N distance in increments of 0.1 Å between 2 and 3 Å and in increments of 0.2 Å

beyond that. The same procedure was repeated for AdCN, MeCN and MesCN ligands. A

plot of relative energy vs. V···N distance can be seen in Figure 2.10. Overall, the four

nitriles have a similar rise in energy as the bond is lengthened; closer inspection reveals a

V shaped pattern for the aliphatic nitriles MeCN and AdCN while the energy growth is

more parabolic for the aryl nitriles.

Figure 2.10 Shown is the increase in relative energy as V-N distance is elongated for DFBN, AdCN, MeCN and MesCN binding to 1. Selected structures at fixed V···N distances are shown in Figures 2.11 and 2.12

for the release of AdCN and DFBN from AdCN-1 and DFBN-1 respectively. 24

Figure 2.11 Optimized configurations at fixed V···Nnitrile distances of (A): 2.08 (B): 2.58, (C): 3.08 Å showing a linear dissociation of the nitrile ligand from AdCN-1.

Figure 2.12 Optimized configurations at fixed V···Nnitrile distances of (A): 2.05 (B): 2.55, (C): 3.05 Å showing a more angular trajectory for DFBN dissociation from DFBN-1. A plot of Energy (z axis) versus V–N distance (x axis) and V–N–C angle (y axis) yields insight into the different pathways computed for AdCN and DFBN as shown in

Figure 2.13. 25

Figure 2.13 Energy (z axis, kcal·mol−1), V–Nnitrile distance (x axis, Å), and V–N–C angle (degrees) for dissociation of AdCN (left) and DFBN (right). There is a clear difference in the computed minimum energy pathways for nitrile

dissociation shown in Figure 2.13 for AdCN-1 and DFBN-1. In the case of AdCN, the V–

N bond lengthens and the V–N–C angle remains almost linear along the reaction

coordinate. In contrast, the computed pathway for DFBN shows that the V–N–C angle

which is somewhat angular to begin with, becomes increasingly bent during the

dissociation process due to the establishment of an enthalpically favorable, but

entropically unfavorable π-stacking interaction between the DFBN and one of the arenes

pendent to one of the anilide ligands. The arene-arene distance is computed to remain nearly constant as shown in Figure 2.14. As the V-nitrile length increases by 2.5 Å, the arene-arene distance is computed to increase by 0.25 Å, approximately one tenth the

change.

The kinetic data reported for binding of nitriles to 1 and 2 do not differ that greatly, but the processes, absolute energies, and computed mechanisms do. The detailed structural, kinetic, thermodynamic, and computational studies for nitrile and isonitrile ligand binding apply only to these specific V(III) and Mo(III) complexes, nevertheless, the principles gleaned from this work give insight into the more difficult task of dinitrogen binding. Interpretation of the data on the reversible nitrile binding are not definitive and provide a partial understanding of the binding process in solution for these defined paramagnetic complexes with a complex geometrical arrangement. The study of the current complexes 1 and 2 in this work, although challenging, is nevertheless orders of magnitude simpler than modeling the elusive binding to V and Mo nitrogenase cofactors. As pointed out in a cogent review by Sellmann[37] the search for “principles 27

rather than blueprints” may be what ultimately solves the quest for understanding

nitrogen fixation.

Three sources of error caution against over interpretation of the data in this work.

The first is in the experimental errors in the activation parameters which are on the order

of ∆H‡ = ± 0.5 kcal·mol−1 and ∆S‡ = ± 3 cal·mol−1·K−1. The second is computational

errors where relative errors comparing similar systems are similar in magnitude to

experimental errors, and absolute errors, for example in enthalpies of reaction are on the

order of ± 2.5 kcal·mol−1. The third source of error is uncertainties in intimate solvation

energies of the complexes studied. This problem dates back to the work of Langford and

Tong.[38] In spite of these difficulties, a consistent picture emerges for the ligand binding mechanisms for 1 and 2.

2.2.11 Nitrile Binding to 1

The kinetic data for ligand binding to 1 at −40 °C collected in Table 2.1 show that

there is a nearly constant ∆G‡(−40 °C) = 9.3 ± 0.3 kcal·mol−1 for all ligands studied. A

similar conclusion may be reached for binding of nitriles to complex 2: ∆G‡(−40 °C) =

11.0 ± 0.5 kcal·mol−1. This simple observation masks a more complex behavior, and it

does not carry over to its ∆H‡ and ∆S‡ components as shown in Figure 2.15.

The data for binding to 1, including the isonitrile AdNC, all fall on a line, often

referred to as isokinetic behavior. If the point for the isonitrile AdNC is removed in

Figure 2.15 (it is sufficiently different than a nitrile that this is warranted), the remaining

“points” are clustered at two extremes with a large gap in the middle. These may be

separated into two limiting categories: a first class (PhCN and DFBN) with low enthalpic

barriers but very unfavorable entropies of activation (∆H‡ ≈ 3 kcal·mol−1 and ∆S‡ ≈ −27 28

cal·mol−1·K−1) and a second class (MesCN, AdCN and MeCN) with higher ∆H‡ values

but less unfavorable entropies of activation (∆H‡ ≈ 6.8 kcal·mol−1 and ∆S‡ ≈ −11

cal·mol−1·K−1).

Figure 2.15 Plot of ∆S‡ (cal·mol−1·K−1) versus ∆H‡ (kcal·mol−1) for binding of nitriles to 1 (left) and 2 (right). All data are for nitriles except for AdNC (green diamond).

A mechanistic dichotomy of this type is common where an associative and a

competing dissociative pathway are present for ligand substitution.[38] The ligands which

have a lower value for ΔH‡ and more negative entropy of activation would be consistent

with an associative mechanism while the ligands with a more positive ∆H‡ and a less

negative ∆S‡ would be consistent with a mechanism with higher dissociative character.

The steric and electronic profiles of both class of ligands showed no clear

characteristic to differentiate them as stronger or weaker donors or as more or less

sterically encumbered. The common feature of the first class of ligands, PhCN and

DFBN, is their ability to form π-stacking interaction like that observed in the structure of

DFBN-1 shown in Figure 2.2. Nevertheless, this kind of interaction is not restricted to the structure of the final adduct but can be established along the entire nitrile addition path. 29

Consequently, this stabilizing interaction at the transition state for the ligands capable of establishing this interaction would serve to reduce its energy when compared to those ligands for which this stabilization is weaker or not possible. Optimized structures as a function of the V···Nnitrile distance for nitrile addition to 1 (Figures 2.11 and 2.12) for

AdCN and DFBN are representative of each previously mentioned class of ligands. The data provides a convincing picture supporting the retention of the π-stacking interaction in the nitrile dissociation from DFBN-1, with the arene-arene distance computed to remain nearly constant as DFBN dissociates (See Figure 2.14).

The differing scenarios for ligand binding to 1 can be now readily explained in a simple proposed mechanism. In both classes of ligands, approach to the transition state is proposed to involve solvent displacement to clear the path for nitrile approach. For PhCN and DFBN ligands, as discussed previously, a π-stacking interaction is established during the binding process that leads to a reduction of the enthalpy of activation while disfavors the process entropically, in keeping with the position of DFBN and PhCN in Figure 2.15

(lower activation entalpies and more negative activation entropies). The rest of ligands in

Figure 2.15 (MesCN, AdCN and MeCN) establish weaker dispersion contacts than the π- stacking interactions exhibited in the binding of PhCN and DFBN to 1. Accordingly, the transition state is located at higher energy values but with a less unfavorable entropy of activation, in agreement with the kinetic values determined for this class of ligands and their position in Figure 2.15.

There is an advantage at low temperatures for the class of ligands able to establish a stabilizing π-stacking interaction during the binding event. They have more unfavorable entropies of activation and the slope of the Eyring plots in figure 2.6 shows that whereas 30

there is nearly an intersection of all the points at −40 °C, at lower temperatures, these

nitriles at faster rates than do the other ligands. At temperatures higher than −40 °C, the

later ligands may be projected to bind more rapidly. It seems probable to the authors that

the former ligands could cross-over to the alternative pathway without the involvement of

π-stacking interactions at higher temperatures. The rates of binding at −40 °C are fast,

and binding is no longer quantitative and the contribution of koff to kobs adds additional

difficulties. For that reason no experimental data was acquired to test this hypothesis. The

important conclusion is that at very low temperatures the associative binding through

initial establishment of a π-stacking interactions or other stabilizing dispersion interaction

may aid in both the thermodynamic and kinetic aspects of nitrile binding for suitable

nitriles.

2.2.12 Binding of Nitriles to 2

The data for ligand binding to 2 shown in Figure 2.15, all fit in a tight box near

∆H‡ = 5.0 kcal·mol−1 and ∆S‡ = −26 cal·mol−1·K−1 with the exception of the isonitrile

AdNC. The fact that the activation parameters for ligand binding to 1 and 2 can be

displayed on a graph with the same units on the axis is surprising. It is also misleading

since the creation of a vacant site at 2 is not only different to 1 but it involves much

greater energy changes. The doublet state of complex 2 is ≈ 40 kcal·mol−1 higher in

energy than the quartet state and this large gap must be lowered to achieve the

intersystem crossing needed to clear the dz2 orbital for binding the nitrile. The fact that all the nitrile ligands have similar enthalpies and entropies of activation and appear in a tight box in Figure 2.15 indicates that, in terms of altering the energy gap between the quartet and doublet states, the R group of the R-C≡N does not change things much for the nitriles 31

studied here. The probable reason for that is that there is an essentially common barrier to

intersystem crossing for nitrile ligands. There is overall good agreement between

experimental and computational results. Computational data for MeCN binding to 2 in its

A configuration show the MeCP to occur at a short Mo—N1 distance of 2.276 Å in

which the N1≡C1 bond of 1.165 is similar to the value of 1.157 for MeCN. The

computed Mo—N1—C1 angle at the MECP is 154.1⁰. The computational energy for the

MECP of 6.0 or 3.2 kcal·mol−1 respectively for the A and B configurations of 2 are in excellent agreement with the enthalpy of activation previously measured for other nitriles

(ΔH‡ = 4.7-6.4 kcal·mol−1, see Table 2.1 and Figure 2.15). The kinetics of binding of

MeCN can not be studied experimentally since the addition of MeCN to 2 results in rapid reductive nitrile coupling yielding the corresponding diiminato species [μ-

t [30d] NC(Me)C(Me)N][Mo(N[ Bu]Ar)3]2.

2.2.13 Inferences for Binding of Dinitrogen to 1 and 2

The computed thermodynamic properties for binding dinitrogen to 1 and 2 as shown in Table 2.2 indicate that mononuclear complexes N2-1 and N2-2 might be accessible as intermediates if kinetic barriers were low, but that it would take extremely low temperatures for them to be observable. Further, the data in Table 2.2 appears to overestimate enthalpies of binding to 1 by about 7 kcal·mol−1 making binding to 1 even

less likely to form in detectable amounts. It might not be capable of displacing solvent.

The situation for binding to 2 is thermodynamically more favorable.

Regarding reaction kinetics, however, there would be a clear advantage for

binding to 1 rather than 2 in terms of populating a high energy intermediate complex. In

terms of binding to 1, dinitrogen would be a ligand with weak dispersion interactions to 32

complex 1 similar to the second class of ligands as MeCN in Figure 2.15 and the barrier to binding N2 would not be prohibitive. Most likely it would be similar to the

thermodynamic barrier. Thus, binding would be under thermodynamic control.

The view regarding binding of N2 to complex 2 is more difficult to assess. It was

considered that reaction of two moles of 2 with dinitrogen to form the µ-N2 bridged

dinuclear adduct was formed from generation of small amounts of a thermodynamically

uphill mononuclear adduct of N2 which is then rapidly trapped by a second mole of 2 to

form 2-NN-2.[26-28] The thermochemical data reported here, though uphill by ≈ 12

−1 kcal·mol to form hypothetical 2-N2 while unfavorable for isolation are not prohibitive

for formation of an intermediate complex. Additional computational and experimental

investigation on these and related systems are in progress.

2.3 Conclusion

Stopped flow kinetics revealed the Gibb’s free energy of activation for the

binding of nitriles to both 1 and 2 was between 9-11 kcal/mol for each nitrile studied.

This seemingly innocuous result covered up the significantly different activation

mechanisms between 1 and 2. The reactivity of 1 was dominated by the R group of RCN;

aryl nitriles bound by an associative mechanism assisted by the anilide arene while

aliphatic nitriles bound by a dissociative mechanism. For 2, the rate limiting step was

dominated by intersystem crossing and conversion of the quartet state to the doublet state

to free the dz2 orbital for bonding. This resulted in little change in activation parameters

with changing nitrile R groups. For both systems it is important to remember that rarely is

a transition metal binding site truly open in solution. Computational investigation and

thermochemical data strongly suggest the presence of a solvent molecule in the open site 33

of 1 prior to nitrile binding. This solvent displacement has a major effect on activation

energy, especially computation gas phase calculations. For 2, the high spin state forbids

solvent binding by requiring a spin transition.

2.4 Experimental Details

2.4.1 Computational Details

Unless stated otherwise, DFT calculations were carried out using the Gaussian

09[39] or Gaussian 16[40] suite of programs. Geometry optimizations were performed without any symmetry restrictions using the PBE0 functional,[41] the D3(BJ) empirical

dispersion correction[42] and the Def2-SV(P)[43] basis set. All stationary points were

optimized in the gas phase by computing analytical energy gradients. The obtained

stationary points were characterized by performing energy second derivatives, confirming

them as minima or transition states by the number of negative eigenvalues of the Hessian

matrix of the energy. To further refine the energies, single-point calculations in toluene

solution using the IEF-PCM model[44] on the previously optimized gas phase structures

were finally performed using the larger Def2-TZVP[43] basis set. Likewise, analogous

calculations were also carried out with the B3LYP[45] density functional for some selected

species. Computed electronic energies at the PBE0-D3(BJ)-PCM/Def2-TZVP level were

corrected for zero-point energy, thermal energy and entropic effects calculated at the

PBE0-D3(BJ)-PCM/Def2-SV(P) level to determine ΔH0(298 K) and ΔG0(298 K) values.

Minimum energy crossing points (MECP)[46] were obtained with ORCA 4.2.[47]

2.4.2 Procedure for Calculating Nitrile Dissociation

Computed paths for nitrile dissociation were found by systematically extending and freezing the V-NCR bond length. Upon optimization of the bond structure, the V-N bond was extended 0.1 Å, the distance was frozen, and the structure was optimized again.

This procedure was repeated until the full length required for dissociation was achieved.

The computational method for these optimizations was analogous to the one described above. Chapter 3 Comparative Pathways for Elimination of N2O or CO2 from Trialkyltin Hyponitrites and Carbonates

3.1 Background

Hyponitrites are an important class of intermediates involved in the metal

mediated two electron reduction of nitric oxide to nitrous oxide by nitric oxide

reductases[48] as shown in reaction 1.

+ - 2 •NO + 2H + 2e 2HNO H2N2O2 H2O + N2O (1)

The rate of N2O elimination, the last step of equation. 1 has been extensively

studied for the more stable trans isomer as shown in reaction 2 and depends on a number

of factors including pH, the presence of radicals, and electrophiles.[49]

trans-H2N2O2 N2O + H2O (2)

Likewise, the reactions of organic esters of hyponitrite have been extensively studied photochemically[50] where they are a source of organic radicals (reaction 3).

trans-RO-N=N-OR + hν 2 •OR + N2 (3)

While nitrous oxide elimination from hyponitrites is common, examples of N2O

addition to metal oxides to form hyponitrite complexes are rare. In this regard, Feldmann

[9] o and Jansen reported formation of pure crystalline cis-Na2N2O2 at 360 C as shown in

reaction 4.

Na2O(s) + N2O(g) cis-Na2N2O2 (4)

While the examples of N2O addition to complexes forming hyponitrites is limited,

the analogous reactions with isoelectronic CO2 leading to carbonates are more common and have been much more explored.[51] The fact that a much larger number of carbonate

complexes exist in compared to hyponitrite complexes implies greater kinetic access and

also more thermodynamic stability for carbonates. The structures of N-heterocyclic

[52] [53] carbene (NHC) adducts of CO2 and N2O are both known and they are shown in

Figure 3.1. These Lewis acid-Lewis base pairs may serve as models for possible initial

binding in the formation of the corresponding hyponitrites or carbonates.

[54] [52] [53] Figure 3.1 Structures of the IPr (left), IPr-CO2 (center) and IPr-N2O (right) highlighting the differing architectures for the adducts. Somewhat surprisingly, available data show similar approximate enthalpies of

binding and stability[55] for the different binding motifs displayed by the NHC adducts of

N2O and CO2. In addition, Roessler and coworkers have recently isolated and structurally

2 0 characterized η complexes of both CO2 and N2O to a (NHC)2Ni fragment with similar

computed enthalpies of binding.[56] 37

Information regarding the mechanism and energetics of hyponitrite formation may be obtained by study of the mechanism of nitrous oxide elimination using the principle of microscopic reversibility. The work reported here compares irreversible elimination of nitrous oxide from trans-[(R3Sn)2(μ-N2O2)] (R = phenyl and cyclohexyl) hyponitrite complexes to reversible carbon dioxide capture of R3Sn-O-SnR3 compounds as shown in Scheme 3.1.

SnR3 O N N O N2O R3Sn Ph 1 : R=Ph CO O 2 R 1Cy: R=Cy 3Sn SnR3 2Ph: R=Ph CO O O 2 2Cy: R=Cy R3Sn SnR3 O 3Ph: R=Ph 3Cy: R=Cy

Scheme 3.1 Binding and dissociation reactions studied in this work. 3.2 Results and Discussion

Cy 3.2.1 Synthesis and Crystal Structure of Trans-[(Cy3Sn)2(μ-N2O2)], 1

The reaction of two equivalents of Cy3SnCl with trans-Ag2N2O2 resulted in the

Cy formation of trans-[(Cy3Sn)2(μ-N2O2)], 1 , and AgCl. Slow evaporation of the reaction mixture after the removal of AgCl resulted in the formation of colorless crystals in 70.9% yield. The complex was characterized by a combination of 119Sn NMR and single crystal

X-ray diffraction analysis. An ORTEP showing the molecular structure of compound 1Cy is shown in Figure 3.2. Selected structural data are shown in Table 3.1. 38

Figure 3.2 An ORTEP of the molecular structure of 1Cy showing 40% probability thermal ellipsoids. Complete crystallographic data and files are available in section 4.4.

Cy a Table 3.1 Selected intramolecular distances and angles for 1 .

Atoms Distance (Å) Atoms Angle (deg) N(1)-N(1*) 1.226(7) N(1*)-N(1)-O(1) 112.2(4) N(1)-O(1) 1.367(4) N(1)-O(1)-Sn(1) 114.0(2) O(1)-Sn(1) 2.073(3) O(1)-Sn(1)-C(1) 107.56(14) Sn(1)-N(1) 2.912 O(1)-Sn(1)-C(7) 105.99(17) Sn(1)-N(1*) 2.782 O(1)-Sn(1)-C(13) 96.70(13) Sn(1)-C(1) 2.169(4) Sn(1)-C(1)-C(2) 114.8(3) Sn(1)-C(7) 2.173(4) Sn(1)-C(1)-C(6) 110.0(3) Sn(1)-C(13) 2.157(4) C(1)-C(2) 1.513(5) a Estimated standard deviations (ESD) in the least significant figure are given in parentheses. 39

Analysis of crystal structures for 1Cy and that previously reported[57,58] for 1Ph

revealed that both compounds have very similar bond distances and angles. Measurement

of Sn-N distances showed that the Sn(1)-N(1*) distances are shorter than the Sn(1)-N(1)

distances for both compounds. These Sn-N distances are long for bonding interactions as

Sn can certainly be 5-coordinate. However, it has been reported that the Sn-N bond

distances for 5-coordinate Sn complexes obtained by the reactions of triorganotin

chlorides with 2-mercaptopyrimidines[59] are in the range of Sn-N distances obtained for

trans-bisstannyl hyponitrites and may imply some Sn–N bonding interactions for 5

coordinate Sn.[60]

Cy Ph Comparison of crystal structures of 1 and 1 with that of [(OEP)Fe]2(μ-N2O2) showed a significant difference in the metal atom position. Sn atoms in 1Cy and 1Ph are

bent towards the N=N bond so that both the Sn atoms are positioned beside the N=N

[61] bond, whereas the Fe atoms in [(OEP)Fe]2(μ-N2O2) are positioned away from the N=N

bond which could be due to the geometry of the OEP ligand.

3.2.2 NMR Studies of Thermal Decomposition of 1Ph and 1Cy

Thermal treatment of solutions of trans-[(R3Sn)2(μ-N2O2)] species lead to slow

elimination of N2O and the formation of the corresponding distannoxanes. The attempts

to study thermal decomposition of 1Ph and 2Ph by variable temperature 119Sn NMR were

unsuccessful due to the broadening of peaks assigned to 1Ph and 2Ph at higher

temperatures. Therefore, NMR samples were heated in a temperature-controlled water

bath for a certain time and then the samples were removed and allowed to equilibrate at room temperature before 119Sn NMR were run. The rate of decomposition of 1Ph was

determined by the ratio of integrated peak area for 2Ph to the total integrated peak areas 40

for tin including both complexes. The thermal decomposition rate for compound 1Ph at 70

°C observed by 119Sn NMR is discussed in section 3.4. The attempts to study thermal

decomposition of 1Cy by 119Sn NMR were unsuccessful even at room temperature due to

the peak broadening caused by rapid ring flipping of cyclohexyl rings.

3.2.3 FTIR Studies of Thermal Decomposition of 1Ph and 1Cy

FTIR provided a more efficient and accurate method to determine the rates of

thermal decomposition for 1Ph and 1PCy by measuring the absorbance of the FTIR band assigned to nitrous oxide at 2219 cm−1 in benzene solutions. Representative data for the

thermal decomposition rates for these compounds at 77.9 °C observed by FTIR is shown

in Figure 3.3.

Figure 3.3 Comparison of the first order reaction profiles for thermal decomposition of compounds 1Ph and 1Cy at 77.9 °C. 41

Eyring plots for thermal decomposition for compounds 1Ph and 1Cy are shown in

Figure 3.4. Analysis of the Eyring plots allowed determination of the experimental activation parameters shown in Table 3.2 for thermal decomposition of compounds 1Ph

and 1Cy to produce the corresponding distannoxane with the elimination of nitrous oxide

as shown in Scheme 3.1. Results indicate that the compound 1Cy has an enthalpy of

activation, approximately 7 kcal·mol−1 higher than the compound 1Ph but a less

unfavorable entropy of activation.

Figure 3.4 Eyring plots for thermal decomposition of compounds 1Ph and 1Cy.

1Cy 1Ph ∆H‡ (kcal·mol−1) 22.7 ± 2.5 15.8 ± 2.0 ∆S‡ (cal·mol−1·K−1) −12.5 ± 6 −28.5 ± 5 Table 3.2 Experimental activation parameters obtained for thermal decomposition of compounds 1Ph and 1Cy. 42

Ph 3.2.4 Reaction of 1 with •Cr(Cp*)(CO)3 (Low Radical Loading)

Ph Reactions of 1 with •Cr(Cp*)(CO)3 (Cp* = C5Me5) using a low loading (15%

Ph mol) of •Cr(Cp*)(CO)3 relative to compound 1 at room temperature resulted in near complete loss of N2O in less than an hour, compared to several days for an uninitiated reaction. A plausible mechanistic scheme consistent with our experimental observations and calculations (vide infra) is shown in Scheme 3.2.

Ph Ph (a) Ph Ph + Ph O + Ph Cr N Sn Ph Cr Sn Ph O N Sn OC CO Sn N OC N O O Activation CO Ph Ph Ph Ph OC OC Cr 1Ph CrSn

N2O (b)

Ph Ph Ph Ph O Ph (c) (e) Ph N Sn O Ph Ph O Cr N Sn N O Sn Sn Ph OC O Ph N O Sn CO Ph Cr CrSn Ph Ph Ph + Cr Ph OC Termination

(c) (d) CO2 Ph Ph Chain Reaction Ph Ph Ph Ph O Ph O N Ph Sn Sn Ph Sn Sn Ph N O OC Cr Sn Ph Ph Ph Ph N O Ph Ph OC N 2 CrNNOSn

Ph Scheme 3.2 Proposed mechanism for N2O elimination from 1 promoted by •Cr(Cp*)(CO)3. Cyclopentadiene methyl groups are omitted for clarity.

Step (a) in Scheme 3.2 shows the uphill production of the •ON=N-OSnPh3 radical

[62] and the known Cr(Cp*)(CO)3(SnPh3) compound. The oxygen centered radical produced in this way may either combine with a second equivalent of •Cr(Cp*)(CO)3 in step (c) or lose N2O to generate •OSnPh3 in step (b) and enter into the radical chain reaction shown in step (d). Several termination steps are possible such as that shown in

Ph reaction (e). The enhanced rate of loss of N2O from 1 promoted by the presence of 43

•Cr(Cp*)(CO)3 is not reproducible and not predictable. The same behavior has been

observed in other hyponitrite reactions2 and is often associated with radical processes

where chain termination steps may occur with several reagents.

DFT calculations were performed to analyze the thermodynamics of this transformation and the computed data obtained is schematically shown in Figure 3.5 supporting the viability of the mechanism proposed in Scheme 3.2.

+40 Ph3SnCr Ph3SnONNO Sn) N O +20 (Ph3 2 2 2 ∆H=+16.5 N2O ∆S=-9.3 2 Ph3SnCr ∆ 1 G=+19.3 Ph3SnO (Ph3Sn)2N2O2 Cr ∆H=+11.9 0 ∆S=+22.9 2 (Ph3Sn)2N2O2 ∆G=+5.1 3 N2O Ph3SnCr (Ph3Sn)2O -20 Ph SnONNO 3 2 N ∆ 2O kcal/mol H=-19.2 Ph SnCr ∆S=+24.9 4 3 G Ph3SnO

∆ ∆G=-26.6 -40 Initiation (Ph3Sn)2O ∆G= +5.1 kcal/mol ∆H=-23.8 1 2 ∆S=+57.1 Propagation ∆G=-40.8 ∆G= -45.9 kcal/mol -60 3 4 Termination 5 ∆G= -50.9 kcal/mol 5 Cr -80 2 N2O 2 (Ph3Sn)2O ∆H=-71.4 -100 ∆S=68.4 ∆G=-91.7 Reaction Coordinate

Figure 3.5 Computed thermodynamic profile (kcal·mol−1) for stoichiometric reduction of two moles of 1Ph promoted by •Cr(CO)3C5Me5. Steps 1 and 2 serve to generate a •OSnPh3 radical which may participate in a chain reaction, steps 3 and 4, and end in termination step 5. 44

Ph 3.2.5 Reaction of 1 with •Cr(Cp*)(CO)3 (High Radical Loading).

Ph Addition of a large excess (400% mol) of •Cr(Cp*)(CO)3 to compound 1 at

room temperature resulted in the formation of 10 times more carbon dioxide than nitrous

oxide. In addition, an intermediate complex proposed as “Cr(Cp*)(CO)2(NNO-SnPh3)” based on its experimental and computed IR spectrum is observed as discussed in

more detail in section 3.4. This complex resembles the highly stable nitrosyl

[18] complex Cr(Cp*)(CO)2(NO) and its formation is proposed to occur by step (c) in

Scheme 3.2. Attempts to isolate and crystallographically characterize the proposed Cr(Cp*)(CO)2(NNO-SnPh3) complex have so far been unsuccessful.

Ph 3.2.6 Reaction of 1 with Protic Acids, HAc, HSPh, and HCr(Cp*)(CO)3

Reaction of 1Ph with protic acids in benzene or methylene chloride was found to occur rapidly as show in equation 5 for X = OAc, SPh, Cr(Cp*)(CO)3.

trans-(Ph3Sn)2N2O2 + 2HX 2Ph3Sn-X + trans-H2N2O2 2Ph3SnX + H2O + N2O (5)

FTIR data for the reaction of 1Ph with 2.5 equivalents of the weak acid

HCr(Cp*)(CO)3 in methylene chloride at room temperature are shown in Figure 3.6. 45

1.4

1.2

1.0

0.8

0.6 Absorbance

0.4

0.2

0.0

2200 2100 2000 1900 1800 Wavenumber ( cm-1 )

Ph Figure 3.6 FTIR spectra obtained for the reaction of compound 1 with 2.5 equivalents of HCr(Cp*)(CO)3 -1 in methylene chloride at room temperature. The steady growth of N2O is shown at 2219 cm . Data showing the growth and decay of the band assigned to trans-H2N2O2 as well as the growth of a band assigned to H2O are available in section 3.4. The overall reaction is rapid with over 90% of 1Ph consumed within 10 minutes of

mixing the reagents. The decomposition of hyponitrous acid to N2O and H2O occurs at a

slightly slower rate.

Ph 3.2.7 Effect of Added O2, CO2, N2O, PPh3 the on Decomposition of 1

Little change in the rate of conversion of 1Ph to stannoxane and nitrous oxide was

observed when the reaction was performed under an atmosphere of O2, CO2, or N2O 46

rather than Argon. Added PPh3 to the reaction solution showed no change in rate or

product distribution.

Ph Cy 3.2.8 Reversible Binding of CO2 to 2 and 2

Ph Cy Exposure of toluene solutions of 2 or 2 to CO2 at room temperature caused

immediate precipitation of a voluminous white precipitate in keeping with other reports

on reaction of stannoxanes and carbon dioxide.[63] Use of methylene chloride as solvent

led to production of clear solutions which could be studied by NMR and IR as described

in section 4.4. Temperature dependent studies for 2Ph and 2Cy in methylene chloride

solutions showed characteristic bands for the carbonate group[64] as shown in Figure 3.7a

for the band at 1605 cm−1 assigned to one of two carbonate bands in the complex 3Cy as a

function of CO2 pressure. There is an increase in intensity as the temperature is decreased

which yielded equilibrium data.

Figure 3.7 (a) FTIR data showing carbonate bands near 1605 and 1329 cm−1 that increase with decreasing Cy Cy Cy temperature as 2 is converted into 3 (left). (b) Plot of ln(Keq) versus 1/T for binding of CO2 to 2 and produce 3Cy in methylene chloride solution (right).

o A plot of ln(Keq) versus 1/T is shown in Figure 3.7b and yields ∆H = −8.7 ± 0.6

−1 o −1 −1 o −1 kcal·mol , ∆S = −17.1 ± 2.0 cal·mol ·K and ∆G 298K = −3.6 ± 1.2 kcal·mol for the 47

Cy Cy formation of 3 by reaction of 2 and CO2 (see Scheme 3.1) with all species in

methylene chloride solution.

Attempts to determine thermodynamic data for formation of 3Ph, which showed overall similar behavior to 3Cy were frustrated due to the broadening of bands in the

infrared spectrum which were temperature dependent. This was assigned to associative

interactions in the less sterically encumbered 3Ph compared to 3Cy.

3.2.9 DFT Computational Studies of the Decomposition of Trans-bisstannyl Hyponitrites

DFT calculations were performed to reveal the mechanism of the reactions

observed experimentally (see Scheme 3.1). In this section, the elimination of nitrous

oxide from trans-[(R3Sn)2(μ-N2O2)] hyponitrite complexes is analyzed using the simpler

Me trans-[(Me3Sn)2(μ-N2O2)] hyponitrite 1 in which the substituent groups of the tin atoms for the compounds experimentally studied, 1Ph (R = Ph) and 1Cy (R = Cy), have been

replaced by methyl groups.

As stated in a previous section, in the crystalline structures of both 1Ph and 1Cy the tin group is bent towards the N=N moiety while in that of the solid structure of

[61] [(OEP)Fe]2(μ-N2O2) is located away from it. Accordingly, the thermodynamic and

kinetic parameters for the isomerization between both structures for the model 1Me

compound (1Mea and 1Mec) were also computed and the results are shown in Scheme 3.3. 48

∆ ∆ Me3Sn H = 3.3 H = 2.9 ∆H = 4.5 N O ∆H = 5.3 N O O N SnMe3 Me3Sn N O O N O N SnMe3 SnMe3 SnMe3

1Mea 1Meb 1Mec

Scheme 3.3 Isomerization reactions for the 1Me hyponitrite.

As it can be seen in Scheme 3.3, the most thermodynamic stable isomer (1Mea)

exhibits the same configuration as that observed in the solid-state structure of 1Cy (see

Figure 3.2) and 1Ph.[58] Nonetheless, the interconversion between all conformational

isomers shown in Scheme 3.3 by rotation around the N-O bonds is facile as inferred by

the activation enthalpies computed.

The computed Gibbs energy (and enthalpy) diagram for elimination of N2O from

1Me is shown in Figure 3.8.

Me Figure 3.8 Schematic Gibbs energy and enthalpy diagram for elimination of N2O from 1 computed at the B3LYP[45]-D3(BJ)[42]/Def2-TZVP[43] level of theory (See Section 3.4 for further details). The isomerization steps for interconversion between 1Mea and 1Mec (Scheme 3.3) not included in the figure. 49

An alternate mechanism with slightly higher barriers calculated for

interconversion between 1Mea and the key intermediate Int1a is shown in Section 3.4. In

Figure 3.8, the lowest energy TS leading to product goes through migration of the Me3Sn

group from the terminal O atom in the trans position through TS1 leading to the Int1a intermediate. A larger barrier exists for it to go forward to products than for it to go backwards to starting material. The computed enthalpy of activation of extrusion of nitrous oxide from the most stable isomer of 1Me (1Mea) of 24.2 kcal·mol−1 is in good

agreement with the experimental value for 1Cy of 22.7 ± 2.5 kcal·mol−1 at 298 K while the activation entropy calculated for 1Me (−3.7 cal·mol−1·K−1) is slightly less negative

than that measured for 1Cy (−12.5 ± 6 cal·mol−1·K−1). However, the enthalpy of activation

Ph −1 experimentally determined for N2O elimination from 1 of 15.8 ± 2.0 kcal·mol is about

7 kcal·mol−1 lower and the entropy of activation is about 16 kcal·mol−1 less unfavorable

(see Table 3.2). In this regard, calculations for the real complex 1Ph were also performed

to understand the discrepancy observed between the activation parameters experimentally

measured and shown in Table 3.2 for 1Ph and 1Cy. The values computed, ∆H‡(1Ph) = 19.0

kcal·mol−1 and ∆S‡(1Ph) = −13.2 cal·mol−1·K−1, while higher than those experimentally

determined are 6.3 kcal·mol−1 and 9.5 cal·mol−1·K−1 lower than that previously calculated

for the elimination of nitrous oxide from 1Me shown in Figure 3.8. Those differences are

in line with the experimental activation parameters collected in Table 3.2 (∆∆H‡(1Cy−1Ph)

= 6.9 kcal·mol−1 and (∆∆S‡(1Cy−1Ph) = 16 cal·mol−1·K−1) and are due to the

establishment of stabilizing interactions between the phenyl groups linked to different tin

atoms that are possible for the transition state but not for the 1Ph hyponitrite since the two

SnPh3 moieties are far from each other. The optimized structure of the transition state 50

Ph Ph (TS2 ) for N2O elimination from 1 is shown in Figure 3.9. The π-stacking interactions

that can be observed in the structure serve to stabilize the transition state and,

accordingly, to decrease the enthalpy of activation while the interactions make the

structure less flexible and, consequently, a more unfavorable entropy of activation is

obtained.

Ph Ph Figure 3.9 Optimized structure of the transition state (TS2 ) for N2O elimination from 1 .

Finally, according to the principle of microscopic reversibility, the enthalpy and

Me Gibbs energy of activation computed for N2O addition to Me3Sn-O-SnMe3, 2 , to yield the 1Me hyponitrite (Figure 3.8) of 53.7 and 65.6 kcal·mol−1 respectively are too high to

occur at room temperature.

Me 3.2.10 Computational Study of Binding of CO2 to 2

As discussed in the previous section, the carbon dioxide capture of bis-stannoxane

Me compounds was also studied for the simpler Me3Sn-O-SnMe3, 2 , compound. The

Cy Ph binding and release of CO2 is facile for both 2 and 2 as observed experimentally and 51 computed data for the simpler 2Me model (Figure 3.10) are in agreement with that. The values in Figure 3.10 give insight into the process of initial CO2 addition followed by sequential rearrangement to the most stable 3Me carbonate product. The computed value

o −1 Me of ΔH = −9.3 kcal·mol shown in Figure 3.10 for CO2 capture by 2 is in excellent

Cy o agreement with that determined experimentally for CO2 addition to 2 of ∆H = −8.7 ±

0.6 kcal·mol−1.

Me Figure 3.10 Schematic Gibbs energy and enthalpy diagram for addition of CO2 to 2 and subsequent isomerization to the final product 3Me computed at the B3LYP[45]-D3(BJ)[42]/Def2-TZVP[43] level of theory. 3.2.1 Discussion

Metal complexes of hyponitrites are much less common than the ubiquitous

[10] carbonate analogs. For example, it is reported in chapter 5, generation of cis-Na2N2O2 by the ball milling reaction shown in reaction 6.

o Na2O(s) + N2O(g) cis-Na2N2O2(s) {ball milling, KBr, 38 C, 2 atm N2O pressure} (6) 52

The only prior report of direct reaction was that of Feldmann and Jansen at 360 oC in a

tube furnace preparation. These observations indicate net favorable thermodynamics for

the insertion process of N2O. The analogous reaction to form sodium carbonate is also

thermodynamically favorable as shown in equation 7.[65]

o −1 Na2O(s) + CO2(g) Na2CO3(s) ∆G 298K = −66 kcal·mol (7)

Reaction 6 is one of the few reported reactions in which nitrous oxide inserts into a metal

[66] o −1 oxide directly. Computational studies yield a value of ∆G 298K ≈ −26 kcal·mol for

equation 6. Subtracting equation 6 from equation 7 yields equation 8 for which the Gibbs

energy change can be estimated using the thermodynamic value for reaction 7 and that

computed for reaction 6 and as shown in equation 8.

o −1 cis-Na2N2O2(s) + CO2(g) Na2CO3(s) + N2O(g) ∆G 298K ≈−40 kcal·mol (8)

The current work reports studies of binding of N2O and CO2 at R3Sn-O-SnR3. This will

allow comparison of the main group metal Sn to the alkali metal Na. Work in progress is

aimed at kinetic and thermodynamic studies of Pt containing hyponitrites and carbonates.

It is hoped that trends may emerge suggesting how best to approach new reaction

chemistry of nitrous oxide. Considerable work in this area has been reported by

Severin[67] and Hayton[68]

The Gibbs energy value estimated for displacement of N2O by CO2 from cis-

Na2N2O2 as shown in equation 8 may be compared to the computed value for the

analogous reaction for the tin hyponitrite 1Me studied in this work as shown in equation 9.

trans-Me3SnO-N=N-OSnMe3 + CO2(g) Me3SnOCO2SnMe3 + N2O(g) (9) 53

The calculated Gibbs energy change for addition of N2O to Me3Sn-O-SnMe3

Me o −1 (2 ) is ∆G 298K = +40.3 kcal·mol as it can be seen in Figure 3.8. The computed

o Me ∆G 298K value for CO2 insertion into the Sn-O bond of 2 to make the monomeric

Me o −1 carbonate complex 3 , as shown in Figure 3.10, is ∆G 298K = −1.8 kcal·mol .

Confidence in that value is gained by the derived experimental value from the graph in

o −1 Cy Figure 3.7b of ∆G 298K = −3.6 ± 1.2 kcal·mol for the analogous reaction with 3 .

o −1 Utilizing these data yield an estimate of ∆G 298K = −44 kcal·mol for displacement of

bound N2O by CO2 in reaction 9. This is surprisingly close to the estimate made for

o −1 reaction 8 of ∆G 298K = −40 kcal·mol . The relatively close agreement assigned to

conversion of a hyponitrite to a carbonate for Na2O and Me3SnOSnMe3 appears to imply

that the dominant factor determining relative stability may lie in the hypothetical O2−

[69] transfer reaction shown in reaction 10 and relative Lux acidity of N2O and CO2 and

that this is closely related to resonance delocalization and charge stabilization in the

carbonate ion being greater than that of the hyponitrite ion.

2− 2− N2O2 (g) + O=C=O(g) N=N=O(g) + CO3 (g) (10)

2− 2− Neither N2O2 (g) nor CO3 (g) are stable in the absence of stabilizing counterions.

Nevertheless if addition of 2{X+} to the two anions is thermodynamically comparable, the difference in internal energy due presumably to greater resonance stabilization in the carbonate versus hyponitrite dianion may be dominant. The type of acid base reaction shown in reaction 10 and expected to not be thermoneutral, is different than simple binding reactions of CO2 and N2O which, as discussed earlier, may be similar in reaction energy. Additional experimental work to test that hypothesis is planned. 54

In benzene solution in sealed NMR tube studies, the first order decay of the

hyponitrite complexes and their impervious response to added PPh3, O2, CO2, and N2O

was surprising. This is in keeping with the computed mechanism that involves migration

of the Sn group from O to N as shown in Figure 3.8. It is of interest to compare the

Me mechanism computed in this work for N2O elimination from the hyponitrite 1 to the

well studied mechanism for dissociation of trans-H2N2O2 hyponitrous acid computed

previously by Morokuma[70] and more recently by Thynell.[71] Figure 3.11 shows a

comparison of the mechanism for the reverse reaction: N2O addition to H2O (previously

[71] computed) or to Me3SnOSnMe3 (computed in this work, see Figure 3.8) to yield trans-

Me H2N2O2 or hyponitrite 1 respectively.

−1 Figure 3.11 Reaction profile (kcal·mol ) for the unfavorable addition of N2O to H2O and Me3SnOSnMe3. [71] Data for reaction of H2O and N2O leading to trans-H2N2O2 are from Thynell. 55

N2O capture by either water or Me3SnOSnMe3 to form trans-hyponitrous acid or

Me o the trans-hyponitrite 1 complex are both highly endergonic processes (∆G 298K ≈ +53 and +40 kcal·mol−1 respectively). The transition states and intermediates for both tin and hyponitrous acid are very similar with no readily discernible difference. The Gibbs energy barrier however is substantially higher for water than it is for tin (+120 vs +65 kcal·mol−1). The ability of tin to become hypervalent is the most probable reason for this.

In either TS, tin can obtain substantial donation from both the N and O lone pairs while migrating due to its larger size and vacant 5d orbitals. In contrast, H migration requires substantially weakening of the O-H bond leading contributing to its higher activation energy.

The rapid reaction of protic acids to eliminate H2N2O2 and which subsequently decomposes to H2O and N2O in an acid catalyzed reaction is a characteristic reaction of hyponitrites.[48,49] Reaction of 1Ph or 1Cy with the transition metal carbonyl hydride

HCr(Cp*)(CO)3 is rapid and produces Cr(Cp*)(CO)3(SnPh3) and H2N2O2 in high yields in organic solvents. Since tin hyponitrites are stable, anhydrous, and store well under dry argon, combined with the fact that the chromium hydride may be prepared in high purity and that Cr(Cp*)(CO)3(SnR3), CrSn, products are non-volatile, the reaction of tin hyponitrites with HCr(Cp*)(CO)3 presents a convenient method to prepare and vacuum transfer anhydrous hyponitrous acid which may be of synthetic utility.

It is well known that organic hyponitrites are decomposed by radicals and either alkyl or alkoxy radicals may be produced during this process. The tin hyponitrites, since they are large and also contain vacant 5d orbitals are a ready radical receptor. It was

Ph hoped that reaction of 1 with two equivalents of •Cr(Cp*)(CO)3, Cr•, would proceed as 56

shown in reaction 11 since radical promoted elimination of nitric oxide from hyponitrous,

as opposed to the more common elimination of nitrous oxide, would be of potential use in

catalytic oxidation of nitrous oxide.

1Ph + 2 Cr• 2 CrSn + ON=NO 2 CrSn + 2 •NO (11)

However, a more complex chemistry resulted as discussed earlier (see Scheme

3.2). At low chromium radical concentrations, elimination of N2O from the tin

hyponitrites is moderately accelerated as proposed via generation of the Ph3SnO• radical.

At higher Cr• concentration, an unstable intermediate complex proposed as

“Cr(Cp*)(CO)3(ONNO-SnPh3)” was consistent with evolution of CO2 and formation of

Cr(Cp*)(CO)2(NNO-SnPh3).

3.3 Conclusion

Ph Cy Experimental studies of the rate of loss of N2O from 1 and 1 are in good

Me agreement with computational studies for 1 . The reaction is impervious to added N2O,

CO2, O2, and PPh3 which is in keeping with the concerted migration pathway computed

Me [72] for 1 and closely resembles published work for H2N2O2(g) with larger computed barriers found for hyponitrous acid than for 1Me. The lower activation energy for loss of

nitrous oxide from the tin hyponitrites compared to hyponitrous acid is attributed to more

facile mobility of the tin group compared to the proton. The computed estimate for

o ∆G 298K for displacement of N2O from a hyponitrite by CO2 to form a carbonate was

Me −1 −1 found to be similar for 1 (−44 kcal·mol ) and cis-Na2N2O2 ( − 42 kcal·mol ). This

may imply that while the absolute value of hyponitrite binding may vary considerably,

the relative stability of hyponitrite and carbonate binding may be similar. 57

Additional experimental and computational work is in progress with sodium oxide and other metal oxides with high Lux basicity[69] to further map kinetic and thermodynamic

factors controlling the reactivity of hyponitrites.

3.4 Experimental Details

3.4.1 General Experimental Procedure

Unless otherwise indicated, all reactions were performed under an atmosphere of

Argon. Reagent grade solvents were dried by standard procedures and were freshly distilled prior to use. 1H and 119Sn NMR spectra were recorded on Bruker 400 spectrometer operating at 399.995 MHz and 149.145 MHz respectively. FTIR spectra

TM were recorded on a PerkinElmer Frontier FTIR/ NIR spectrometer. Ph3SnCl (98%)

was purchased from Sigma-Aldrich and used without further purification. Cy3SnCl

(95%),

(Ph3Sn)2O (95%) and (Cy3Sn)2O (95%) were purchased from Gelest, Inc. and used

without further purification. Benzene-D6 (D, 99.5%) was purchased from Cambridge

Isotopes Laboratory, degassed and stored over 4Å molecular sieves. Dichloromethane-

D2 (D, 99.8%) ampoules were purchased from Cambridge Isotopes Laboratory and used

without further purification. Research grade carbon dioxide (99.999%) were purchased

[73] from Airgas South and dried by passing through a drierite column. Ag2N2O2 was

prepared according to the literature and dried over vacuum followed by P2O5 and stored

at -35°C away from light. 58

3.4.2 Synthesis of (Cy3Sn)2N2O2

Cy3SnCl (100 mg, 0.25 mmol) was dissolved in methylene chloride (5 mL).

Ag2N2O2 (40.0 mg, 0.145 mmol) was added and the slurry was stirred for approximately

two hours in the dark where reaction completion was confirmed by NMR. After filtration

to remove AgCl, the colorless solution was evaporated to dryness to afford a white

powder. Recrystallization by slow evaporation of a methylene chloride/ heptane solvent

mixture at room temperature resulted in colorless crystals. Yield 70 mg (70.9%). Spectral

1 119 data: H NMR (CD2Cl2, RT, in ppm) δ = 1.20 – 2.00 (m, 66H, cy), Sn NMR

(CD2Cl2, RT, in ppm) δ = 22.24.

3.4.3 Synthesis of (Ph3Sn)2N2O2

[12,13] (Ph3Sn)2N2O2 was prepared according to a modified literature procedure.

Ph3SnCl (100.0 mg, 0.26 mmol) was dissolved in methylene chloride (5 mL). Ag2N2O2

(40.0 mg, 0.145 mmol) was added and the slurry was stirred for approximately two hours

in the dark where reaction completion was confirmed by NMR. After filtration to remove

AgCl, the colorless solution was evaporated to dryness to afford a white powder.

Recrystallization by slow evaporation of a methylene chloride/ heptane solvent mixture at

room temperature resulted in colorless crystals. Yield 74 mg (75.1%). Spectral data: 1H

119 NMR (CD2Cl2, RT, in ppm) δ = 7.61 (d, 12H, ph), 7.43 (m, 18H, ph), Sn NMR

-1 (CD2Cl2, RT, in ppm) δ = -89.81, Neat ATR: ν(NO): 996.06 cm .

3.4.4 Synthesis of (Cy3Sn)2CO3

(Cy3Sn)2O (18 mg, 0.024 mmol) was dissolved in methylene chloride-D2 (0.75

mL) and the solution was filtered into a J-Young NMR tube to remove residual 59

Cy3SnOH. CO2 was added to the tube at a pressure of 15 psi at room temperature. NMR

analysis showed nearly full conversion to carbonate at room temperature. Spectral data:

1 119 -1 H NMR, Sn NMR (CD2Cl2, RT, in ppm) δ = 0.92, Liquid cell FTIR ν(C=O) (cm in

CH2Cl2): 1606.6 (br), 1582.6 (sh), 1329.59 (br).

3.4.5 Synthesis of (Ph3Sn)2CO3

(Ph3Sn)2O (14.0 mg, 0.02 mmol) was dissolved in methylene chloride-D2 (0.75

mL) and the solution was filtered into a J-Young NMR tube. CO2 was added to the tube

at a pressure of 15 psi at room temperature. Cooling to -20 °C for 16 hours afforded

colorless crystals, removal of the crystals from either CO2 atmosphere or solution caused their decomposition. NMR analysis showed nearly full conversion to carbonate at room

1 119 temperature. Spectral data: H NMR, Sn NMR (CD2Cl2, RT, in ppm) δ = -125.8,

-1 Liquid Cell FTIR ν(C=O) (cm in CH2Cl2): 1520.8 (s), 1539.3 (sh), 1377.5 (s).

3.4.6 Kinetic Study of the Thermal Decomposition of Tin Hyponitrites

119 Initial investigations of the decomposition of (Ph3Sn)2N2O2 by Sn NMR resulted in noisy data and identified the need for direct, real time measurements. These

initial NMR studies were successful in showing the reaction to be clean even at elevated

temperatures (80°C). The clean nature of the decomposition allowed for rate

determinations by direct measurement of the N2O IR bands by the procedure below.

In the glovebox approximately 10mg of 1 or 2 was dissolved in 1mL of dried benzene. Upon stirring for 10 minutes the solution was filtered and added to a jacketed

high pressure FTIR cell (CaF2) connected to a water bath and allowed to equilibrate to

the desired temperature for 10 minutes. Spectra were recorded in regular intervals in 60 accordance with the expected reaction rate with a minimum of 8 minutes separating scans to avoid excess radiative heating. Elimination of N2O was detected by growth in the known peak at 2219 cm-1.

Ph 3.4.7 Reaction of HCr(CO)3C5Me5 with 1

A solution of 40mg of HCr(CO)3C5Me5 in 4mL of dichloromethane was prepared in the glove box under Argon atmosphere. The solution was added to 40mg of 1Ph solid, mixed thoroughly and added to a solution FTIR cell with CaF2 windows. The reaction proceeded as shown in figure 3.12 below.

Figure 3.12 The first spectrum (black) represents a solution of only CrH. Subsequent spectra began 2 minutes after mixing and were taken every 2 minutes until ending at 34 minutes. There is a steady decrease in CrH with growth in CrSn. Two isosbestic points are observed at 1982 and 1894 cm-1.

Ph 3.4.8 Reaction of •Cr(CO)3C5Me5 with 1 (Excess •Cr)

A solution of 50.5mg of •Cr(CO)3C5Me5 in 5mL of dichloromethane was prepared in the glove box under Argon atmosphere. The solution was added to 35.5mg of 1Ph solid, 61

mixed thoroughly and added to a solution FTIR cell with CaF2 windows. The reaction

proceeded as shown in figure 3.14 below.

-1 -1 Figure 3.13 CO2 emission (2338cm ) in the reaction outperforms N2O (2219cm ) by an order of magnitude. Peaks attributed to the proposed compound CrNNOSn (2005, 1931, 1672 cm-1) reach a maximum at about 30 minutes and remain stable after that. Band assignments have been marked for the previously reported compounds SnCr (1970 and 1884) and Cr≡Cr (1905, 1848).

Ph 3.4.9 Reaction of •Cr(CO)3C5Me5 with 1 (Substoichiometric •Cr)

Ph 10.8mg of 1 was dissolved using an ampoule of 0.72mL of CD2Cl2 in the glove box

under Argon atmosphere. 1.2mg of •Cr(CO)3C5Me5 was added to the solution; it was

mixed thoroughly and added to an NMR tube. The reaction proceeded as shown in figures 3.15 below. 62

Ph Figure 3.14 NMR of 1 in CD2Cl2 (blue), after 20 minutes of reaction with 1.2mg of • Cr (red), after 40 minutes of reaction (green) and the following day (purple). By the first scan about 2/3 of 1Ph has been converted to 2Ph with the reaction slowing afterwards. An expanded view is shown to show the absence of observable SnCr around +50ppm or any other tin products.

Cy 3.4.10 Determination of Keq for Reversible Binding of CO2 to 2

Cy A solution of 103 mg 2 in 5 ml CH2Cl2 that had been dried over molecular sieves was prepared in the glove box and transferred to a Hamilton gas tight syringe with a valve fitting. This was then transferred to thermostatted reactor system that has been described in detail previously.[13] Following standard techniques variable temperature infrared data were obtained as shown in Figure 3.7a. Temperature dependent studies for

2Cy in methylene chloride yield the van t’Hoff plot shown in Figure 3.7b from which thermodynamic parameters ∆Ho = −8.7 ± 0.6 kcal·mol−1, ∆So = −17.1 ± 2.0

−1 −1 o −1 cal·mol ·K and ∆G 298K = −3.6 ± 1.2 kcal·mol were derived.

3.4.11 Crystallographic Analyses

Colorless single crystals of 1Cy suitable for diffraction analysis were grown by slow

evaporation of a methylene chloride/heptane solvent mixture at room temperature. The

data crystals were glued onto the end of a thin glass fiber for tda a collection at room

temperature. X-ray int ensity data were measured by using a Bruker SMART APEX2 63

CCD-based diffractometer using Mo Kα radiation (λ = 0.71073 Å).[74] The raw data frames were integrated with the SAINT+ program by using a narrow-frame integration algorithm.[74] Corrections for Lorentz and polarization effects were also applied with

SAINT+. An empirical absorption correction based on the multiple measurement of equivalent reflections was applied using the program SADABS. All structures were solved by a combination of direct methods and difference Fourier syntheses, and refined by full-matrix least-squares on F2, by using the SHELXTL software package.[75,76] All non-hydrogen atoms were refined with anisotropic displacement parameters. Hydrogen atoms were placed in geometrically idealized positions and included as standard riding atoms during the least squares refinements. Crystal data, data collection parameters, and results of the analyses are listed in Table 3.3.

Table 3.3 Crystallographic data for 1Cy

Compound 1Cy

Empirical formula Sn2O2N2C36H66 Formula weight 796.28 Crystal system Triclinic Lattice parameters a (Å) 9.8269(4) b (Å) 10.0645(4) c (Å) 10.8581(5) α (deg) 85.261(1) β (deg) 86.119(1) γ (deg) 65.515(1) 3 V (Å ) 973.32(7) Space group P1 (# 2) Z value 1 64

3 ρcalc (g / cm ) 1.359 -1 µ (Mo Kα) (mm ) 1.313 Temperature (K) 293

2Θmax (°) 56.90 No. Obs. ( I > 2σ(I)) 4445 No. Parameters 190 Goodness of fit GOF* 1.084 Max. shift in cycle 0.001 Residuals*: R1; wR2 0.0392; 0.1085 Absorption Correction, Multi-scan Max/min 0.7457/0.5883 Largest peak in Final Diff. Map (e- / Å3) 1.687

2 2 1/2 *R = Σhkl(Fobs-Fcalc)/ΣhklFobs; Rw = [Σhklw(Fobs-Fcalc) /ΣhklwFobs ] , 2 w = 1/σ (Fobs); GOF = [Σhklw

a Table 3.4 Selected intramolecular distances and angles for compound trans-(Cy3Sn)2N2O2

a Estimated standard deviations (ESD) in the least significant figure are given in parentheses 65

a Table 3.5 Selected intramolecular distances and angles for compound trans-(Ph3Sn)2N2O2

Atoms Distance (Å) Atoms Angle (deg) N(1)-N(1*) 1.235(5) N(1*)-N(1)-O(1) 111.1(3) N(1)-O(1) 1.372(3) N(1)-O(1)-Sn(1) 115.64(17) O(1)-Sn(1) 2.063(2) O(1)-Sn(1)-C(1) 90.21(10) Sn(1)-N(1) 2.931 O(1)-Sn(1)-C(7) 109.94(11) Sn(1)-N(1*) 2.799 O(1)-Sn(1)-C(13) 116.50(10) Sn(1)-C(1) 2.124(3) Sn(1)-C(1)-C(2) 119.0(2) Sn(1)-C(7) 2.128(3) Sn(1)-C(1)-C(6) 122.7(2) Sn(1)-C(13) 2.132(3) C(1)-C(2) 1.398(4)

a Estimated standard deviations (ESD) in the least significant figure are given in parentheses

[61] Table 3.6 Selected intramolecular distances and angles for compound [(OEP)Fe]2(μ-N2O2)

Atoms Distance (Å) Atoms Angle (deg) N(1A)-N(1A*) 1.250(3) N(1A*)-N(1A)-O(1A) 108.5(2) N(1A)-O(1A) 1.375(2) N(1A)-O(1A)-Fe(1) 118.56(12) O(1A)-Fe(1) 1.8891(15) O(1A)-Fe(1)-N(1) 103.81(7) Fe(1)-N(1A) 2.818 O(1A)-Fe(1)-N(2) 100.98(7) Fe(1)-N(1A*) 3.905 O(1A)-Fe(1)-N(3) 98.65(7) Fe(1)-N(1) 2.0492(18) O(1A)-Fe(1)-N(4) 101.51(7) Fe(1)-N(2) 2.0637(18) Fe(1)-N(3) 2.0618(17) Fe(1)-N(4) 2.0611(18)

a Estimated standard deviations (ESD) in the least significant figure are given in parentheses 66

As discussed in section 3.2.1, analysis of crystal structures for trans-

(Cy3Sn)2N2O2 and trans-(Ph3Sn)2N2O2 revealed that both compounds have very similar bond distances and angles. Measurement of Sn-N distances showed that the Sn(1)-N(1*) distances are shorter than the Sn(1)-N(1) distances for both compounds. These Sn-N distances are long for bonding interactions as Sn can certainly be 5-coordinate.

However, it has been reported that the Sn-N bond distances for 5-coordinate Sn complexes obtained by the reactions of triorganotin chlorides with 2- mercaptopyrimidines are in the range of Sn-N distances obtained for trans-bisstannyl hyponitrites and may imply some Sn–N bonding interactions for 5 coordinate Sn.[77]

Comparison of crystal structures of trans-(Cy3Sn)2N2O2 and trans-(Ph3Sn)2N2O2 with [(OEP)Fe]2(μ-N2O2) showed a significant difference in the metal atom position. Sn atoms in trans-(Cy3Sn)2N2O2 and trans-(Ph3Sn)2N2O2 are bent towards the N=N bond so that both the Sn atoms are positioned beside the N=N bond, whereas the Fe atoms in

[(OEP)Fe]2(μ-N2O2) are positioned away from the N=N bond which could be due to the geometry of the OEP ligand. This structural difference also implies the possibility of forming a 5-coordinated Sn intermediate in the decomposition of trans-bisstannyl hyponitrites to the corresponding distannoxanes with the elimination of nitrous oxide.

4.4.12 Computational Section

Electronic structure calculations were carried out using the B3LYP functional[45] and the

D3 version of Grimme’s dispersion with Becke-Johnson damping (D3(BJ)).[42] Unless stated otherwise, optimizations were performed by computing analytical energy gradients 67 with the Def2-TZVP basis sets.[43] The obtained minima were characterized by performing energy second derivatives, confirming them as minima by the absence of negative eigenvalues of the Hessian matrix of the energy. Transition states were characterized by a single imaginary frequency, whose normal mode corresponded to the expected motion. In the case of the calculations for the species containing phenyl groups in the tin substituents such as 1Ph or TS2Ph and those related to the reaction with

•Cr(Cp*)(CO)3 geometry optimizations and frequency calculations were performed with the Def2-SVP basis set[43] and then, to further refine the energies obtained, singlepoint

[43] calculations were performed using the larger Def2-TZVP basis set. To determine H298K and G298K values, computed electronic energies were corrected for zero-point energy, thermal energy and entropic effects estimated from the normal mode analysis. All calculations were performed with the Gaussian 16 suite of programs.[40]

Figure 3.16 Schematic Gibbs energy and enthalpy diagram for an alternative isomerization reaction to yield Int1a from 1Me Chapter 4 The Mechanism of Carboxylative Cyclization of Propargylamine by N-heterocyclic Carbene Complexes of Au(1) 4.1 Introduction

Catalytic functionalization of amines with carbon monoxide has a long history

dating back to the work of Heiber and Hein as described in the classic monographs by

Wender and Pino [78]. Advances in that area have been made, for example in the recent

preparation of the novel cyclic imine carboxamide shown in Scheme 4.1, reported by

Kollar et al [79].

Scheme 4.1 Pd-catalyzed cyclization of amines with carbon monoxide [79]. Renewed interest in utilization of green chemical methods for carbon dioxide

capture has focused on cycloaddition reactions with reactive acetylene substrates [80].

Compared to carbon monoxide, direct incorporation of carbon dioxide into organic

synthesis presents a challenge due to the thermodynamic stability of carbon dioxide.

Ikariya and co-workers have reported insertion of carbon dioxide into the N-H bond followed by carboxylative cyclization of propargylamine (PPA) as shown in Scheme 4.2

[81].

68 69

Scheme 4.2 [Au]-catalyzed carboxylative cyclization of PPA [81]. Recently, we have explored the activity of a series of dinuclear gold(I) complexes bearing bis(N-heterocyclic carbene) in the Ikariya transformation ( Figure 4.1) [82]. This series of gold catalysts is easily assembled using the recently discovered weak base approach as shown in Scheme 4.3 [82].

Scheme 4.3 Synthesis of a series of dinuclear gold(I) complexes [14].

These complexes were deployed in the carboxylative cyclization of PPA with the eight-methylene bridged (n = 8) NHCs bearing 2,6-diisopropylphenyl substituent (for X

= Cl, Br) shown to give both higher yields and faster reaction rates than the mononuclear

[Au(IPr)Cl] (2) analogue [14]. The rate of this remarkable reaction is not fast for any of the catalysts reported to date. At substrate to catalyst ratios of 50 to 100, achieving near quantitative yields typically requires some 24 hours corresponding to turnover frequencies on the order of 4 hours. The fact that the catalytic activity of the linker complexes varies as a function of the spacing between the two Au centers suggests that they do not operate independently of each other. To gain insight into design of more efficient catalysts for this reaction we have reported kinetic and computational studies comparing the optimal linker complex to its mononuclear analogue [14]. This work stressed the role of carbamic acid (CA) and carbamate salt (CS) of PPA as the substrates of the reaction rather than PPA itself. This is in keeping with recent work by Kortunov and co-workers concerning the equilibrium processes shown in Scheme 4.4 [83]. 71

Scheme 4.4 The CA/CS Equilibrium

In the presence of excess CO2, as is typical in catalytic studies, the concentration

of free PPA is minimal and the total substrate is distributed between CA and CS. In spite

of earlier reports by Calderazzo [84] and Toste and Bergman [85] of stable L-Au-O-

C(=O)NR1R2 complexes, as well as the established carbonylation reactions of secondary

amines, the role of Au-carbamate complexes has apparently not been suggested as a

potential resting state for gold in the catalytic cycle [86]. The mechanism proposed in

Figure 4.1 [14] where the energetics are presented indicates one possible pathway to formation of cyclic Au-oxazolidinone intermediate 1 during the induction period starting with 2 and converting it to a CA derivative and then to 1. The authors view this as the first of two major processes which both contribute to the combined turnover frequency.

The second major process is the protodeauration to cleave the Au-vinyl bond to

yield the oxazolidinone product and regenerate the catalyst. Several reaction partners for

1 were considered as shown in Figure 4.2 [14].

The three acids studied computationally may be viewed as adducts between the

+ − basic PPA and a parent acid. In path A of Figure 4.2, [(R)(CH3)NH2] Cl the HCl salt of

PPA which is formed during the induction period shown in Figure 4.1 is a by-product of the conversion of 2 to [Au(IPr)O(O=C)NR1R2]. In the absence of an additional source of

chloride this reaction partner for 2 will be present in low concentration since its 72

maximum concentration will be that of its parent 2. In path B, the CA of PPA may be viewed as the Lewis acid adduct PPA-CO2.

In the present contribution, we report kinetic and computational studies related to the second major process along the catalytic pathway shown in Figure 4.2. In addition, we report pressure dependent studies of cleavage of 1 by CO2 as well as using n-butyl

benzyl amine (BBA), the saturated analogue of PPA. The goal of this work is to gain

insight into the energy barriers present in the second major process of this remarkable

transformation.

Figure 4.2 Proposed mechanism for protonation of IKa with three potential adducts between PPA and an added acid: A) PPA-HCl, B) PPA-CO2, and C) PPA-CA. 73

4.2 Results and Discussion

4.2.1. Synthesis and Structure of 1

The [Au(IPr)(vinyl)] complex, 1, where IPr = N,N’-bis-[2,6 (diisopropyl)phenyl]

imidazol-2-ylidene, vinyl = 1,3 oxazolidinone, was prepared by using a method

analogous to that described by Ikariya and coworkers and described in the experimental

section below. The structure in the solid state is shown in Figure 4.3

Table 4.1 Crystallographic Data for Compound 1. Figure 4.3 An ORTEP showing the molecular structure of 1 at 40% thermal ellipsoid probability.

As can be seen in Table 4.2, there are no significant structural difference between

compound 1 and the previously reported complexes IKa-Me and IKa-Ph [81]. The

compound IKa-Me is similar to 1, but has a methyl group bonded to N(3) in place of the

benzyl group. The compound IKa-Ph also has a methyl group bonded to N(3) in place of

the benzyl group, but in addition has a phenyl group bonded to C(29) in place of the 74

methyl group. More similarities lie between 1 and IKa-Me as both contain methyl groups

at the alkenyl carbon C(29). The C(1)-Au(1)-C(29) bond angle in 1 is 173.30° indicating

that the geometry around gold is essentially linear. In IKa-Me the angle about the gold

atom is 172.19°, which is similar to that found in 1, while this angle in IKa-Ph is closer to linear at 177.63°. The reason for the difference can be attributed to the more symmetric vinyl group (on both sides of the C atom bound to Au) in IKa-Ph. In 1 and IKa-Me the groups are not very symmetric with a methyl group on one side and a bulkier

oxazolidinone group on the other. The C(29)-C(30) alkenyl bond length is 1.310Å and

the C(30)-C(29)-Au(1) angle is 116.7° which are similar in comparison to IKa-Me and

IKa-Ph. One may also surmise that due to the benzyl group in 1 the coordination of the

vinyl group is altered such that the angle between the planes of the two five membered

rings is 32.98°. In IKa-Me and IKa-Ph this angle between the two planes are closer to co-

planar at 10.80° and 12.66°, respectively.

Table 4.2 Selected intramolecular distances and angles for compounds 1, IKA-Me and IKA-Ph.

4.2.2. Comparison and Catalytic Activity of 1 and 2

A plot of the concentration of product formed as a function of time under the

standard low concentration conditions used is shown in Figure 4.4.

Figure 4.4 Rate of product formation under catalytic conditions [PPA]0 = 0.052 M, CO2 pressure = 2 atm (absolute) in EtOH at 21°C using [1]0=0.00615M (red squares) compared to [2]0=0.0072M. At this low concentration of PPA, the relative initial rates clearly show an advantage for 1 over 2. In addition, there is a much slower acceleration in the initial rate

with time allowing the reaction to go essentially to completion within 1500 minutes

utilizing 1 whereas the use of 2 does not lead to complete conversion even after 6000

minutes.

A five-fold increase in initial concentration of PPA to 0.256 M decreases the

difference in performance between 1 and 2 as shown in Figure 4.5 where the initial rates 76

of reaction are virtually identical. The only difference occurs in the second day of

reaction where the rate of reaction for 2 begins to slow more dramatically than that of 1.

0.256M PPA level because there is now nearly complete conversion of 2 into 1 during the

induction period. The chloride ion may increase the rate of acid cleavage as discussed later. The dramatic deceleration in Figure 4.4 is readily explained by recombination of an increased fraction of 1 undergoing protonation by PPA-HCl rather than HCarb leading to

an increase in 2. As reported earlier, towards the end of the reaction X-Ray quality 77

crystals of 2 precipitate from solution. While the initial rates of reaction at high PPA

concentrations are similar, the long-term rate of conversion to product may decelerate

more rapidly for 2 and this occurs more dramatically at high PPA concentrations as

discussed later.

5.2.3. Role of CO2 Pressure on Initial Reaction Rates at Low PPA Concentrations

Investigation of the rate of protonation of IKa in the absence of substrate by CO2

in EtOH prompted investigation of whether CO2 pressure plays a role on the reaction

rate. The initial rates of product production at 1, 2, and 3 atm CO2 pressure (absolute)

were investigated as shown in Figure 4.6.

Figure 4.6 Initial rates of carboxylation of [1]0 = 0.0072 M, [PPA]0 = 0.052 M, T = 21°C in anhydrous EtOH at pressure of 1, 2, and 3 atm absolute CO2 pressure. 78

The increase in rate with the CO2 pressure is attributed to a combination of two

factors. The first is, as discussed in the following section, that CO2 alone will slowly

protonate 1 in alcohol solution. The second is that an increase in pressure will also serve

to shift the CA/CS equilibrium to favor CA. This is illustrated in Figure 4.7 where FTIR

spectra from the reactions in Figure 4.6 under active conditions at ≈30 minutes total

reaction time are shown-all for initial PPA concentrations of 0.052 ±0.002 M

concentration.

It is clear from these data that there is a stronger band due to CA as pressure is

increased. This is due to conversion of CS to CA by CO2 . The broad band due to CS

occurs below 1600 cm−1 is not as intense as the band due to CA. The large change in rate

in Figure 4.6 in changing from 1.1 to 2 atm and the smaller change in rate on going from

2 atm to 3 atm correspond roughly to the changes in intensity of the CA bands in Figure

4.7. This indicates that the increase in CA concentration is the dominant factor in the rate

increase. It should be noted that as the CA concentration increases, the CS concentration

decreases. If CS were a faster acid cleavage reactant than CA, the reverse effect would be expected. 79

Figure 4.7 FTIR spectroscopic data after ≈30 minutes reaction time for reaction of 1 (0.007M) and PPA (.052 M) in EtOH at 21 °C and 1.1, 2.0, and 3.0 atm pressure (absolute). The band near 1657 cm−1 is due to 1 and it is imbedded in bands due primarily to CA on either side of it. 5.2.4. Derived Estimate for Rate of Reaction of 1 and CA

In contrast to other reagents where the rate of acid cleavage may be measured

directly, the rate of the isolated step for cleavage of 1 by the CA of PPA cannot be measured directly since it occurs together with continuous product formation. However the rate of product formation corresponds directly in this system to d[P]/dt = rate = k2[1][HA]. Under conditions where the dominant source of acid cleavage is CA

(particularly at a fixed low CO2 pressure and the absence of chloride or other acid

source)this relationship may be tested to see if a reliable estimate for k2 for cleavage of 1

by CA may be generated from an average over several CA levels of {rate/[HA][1]}. The

fact that we are able to measure [1] with reasonable accuracy at low PPA concentrations

prompted us to examine this reaction step more closely. All data in this section were 80

measured at 21 °C at 15 psi CO2 (2 atm pressure absolute). Data for the [1] were obtained by computer subtraction of experimental data. Experimental data for [1] after mixing solid 1 with solutions of 0.020 M, 0.52 M, and 0.255 M are shown in Figure 4.8. It should be noted that the low values at t = 0 are due to the fact that the first spectra were taken after shaking for ≈1 minute and that only after ≈5 minutes stirring does 1 completely dissolve. The stock solution levels were on the order of 0.0065 M and the hypothetical reaction curve if everything went into solution immediately would start with a decrease from that point. There is a relatively rapid loss of more than half the concentration of 1 which corresponds to an increase in reaction rate with increasing substrate concentration. Flow studies using solutions of this reaction are planned to better analyze the initial rate of approach to the near steady-state period.

Figure 4.8 Concentrations of 1 versus time added as a solid to solutions of PPA in EtOH under 15 psi (2 atm absolute pressure) CO2 at 21°C. 81

The concentration of CA in these solutions during the steady state period also remains roughly constant due to the buffering capacity of PPA and removal of CA results in its partial regeneration from uptake of CO2 by CS. For that reason, we analyze at this time only data at the standard 15 psi (2 atm absolute) pressure of CO2. These data are summarized in Table 4.3, and lead to an average rate constant of 0.54 ±0.07 M−1 min−1. It should be kept in mind that this is a preliminary estimate and the rate is expected to vary as a function of factors that alter the activity of CA.

-1 -1 Table 4.3 Data used for calculation of k2 for rate of cleavage of IKa by HCarb of 0.54 ± 0.7 M min .

4.2.5. Reaction of 1 and PPA-HCl

The most thermodynamically potent reagent for cleavage to product of 1 is the

PPA-HCl salt. In an earlier proposed mechanism for activation of 2 during the induction period we have proposed reaction (1) to occur:

In order to investigate this kinetically the PPA-HCl salt was prepared and the rate of reaction (2) investigated: 82

The reaction was monitored under settings designed to mimic catalytic conditions.

Since the PPA-HCl concentration cannot exceed that of added 2 unless external chloride

is added, reaction (2) was studied under the dilute solution conditions [1]0 = [PPA-HCl] =

0.007 M. This places [PPA-HCl] at a disadvantage in the second order reaction since both reagents are at the mM level. In spite of that the reaction occurred at a rate comparable to

that of the other reagents tested. A plot of 1 /[IKa] versus time was found to be linear in

agreement under these conditions with a second order rate law as shown in Figure 4.9 .

Analysis of the second order rate constant based on the growth of the 2-

oxazolidinone as shown in Figure 4.9 yielded a rate constant of 9 M−1 min−1. Similar

analysis based on the rate of decay of 1 versus time yielded a value of 10.2 M−1 min−1.

The average value of 9.6 M−1 min−1 is adopted for this reaction. 83

Figure 4.9 Second order plot of 1 /[Product] versus time (min) for reaction of 1 with PPA-HCl with each reagent at 0.007M concentration in EtOH at 21°C and 2 atm pressure (absolute) CO2.

4.2.6. Rate of Cleavage by CO2 /EtOH

It was reported by Ikariya and co-workers that CO2 in MeOH is capable of

reaction with the [Au(IPr)(vinyl)] intermediate complex to yield product. We have

investigated this quantitatively as a function of CO2 pressures at pressures of 2 and 3 atm

absolute pressure of CO2. Based on spectroscopic and computational data the reaction is

proposed to form an ethoxy carbonate complex following the stoichiometry shown in

Equation (3): 84

As discussed later, a rapidly established equilibrium involving reversible addition of carbon dioxide is present. Kinetic plots obey pseudo first order rate laws over three half-lives at constant pressures as shown in Figure 4.10. Increasing the CO2 pressure from 2 to 3 atm absolute pressure increased the observed rate of reaction proportionally.

Figure 4.10 First order plots for rate of production of the 2-oxazolidinone product in reaction of 1 with CO2 in EtOH at 21 °C and absolute pressures of 2 and 3 atm.

−1 −1 The rate law d[P]/dt = kobsd [pCO2][1] has a value for kobsd = 0.0049 atm min based on an average value for the two experiments shown in Figure 4.10. The solubility

[87] −1 −1 of CO2 in EtOH at 1 atm and 25 °C is 0.14 M . This allows kobsd = 0.0049 atm min to 0.035 M−1 min−1. The reaction was also studied under strictly anhydrous conditions using EtOH which had been stored in an inert atmosphere glove box under active molecular sieves for several days. The same conditions were used in catalytic reactions. 85

In the event trace water is present it would be present at the same level in catalytic

reactions in which a carefully dried solvent was used. The presence of trace water at

some level is unavoidable and its possible role on the rate of protonation is beyond the

scope of the current investigation.

4.2.7. Reaction of 1 and CO2 at High Pressure

The linear dependence of the rate of reaction of 1 and CO2 at 2 and 3 atm

prompted qualitative investigation of the rate at 60 atm. As described above, 0.007 M

solution of 1 in EtOH was loaded under argon into a high-pressure reactor. The reactor

was then pressured at 900 psi CO2 for an absolute pressure of ≈60 atm. The reactor was filled and shaken for ≈10 minutes at which time the pressure was released. During the

release of CO2 the solution and reactor were chilled to low temperature. After repeated

shaking to degassing cycles (3 times), the solution and venting CO2 the reactor was

opened and the FTIR cell was loaded. A spectrum recorded at ≈12 minutes total time

showed approximately 70% conversion to product. This corresponds to a rate constant of

≈0.1 atm−1 min−1 at 60 atm. This corresponds to an ≈10 fold increase over the rate at 2

atm. That is less than the 60/2 = 30-fold increase in pressure indicating a non-linear

increase in rate which diminishes at higher pressures. The overall observed rate constant

at 60 atm CO2 atm does however make it competitive with concentrations of CA on the

order of 0.2 M.

4.2.8. Reversible Equilibrium for CO2 Binding

In ethanol solutions of 1 at 2, 3, and 60 atm CO2 the same broad band centered at

about 1663 cm−1 is observed. Shown in Figure 4.11 are infrared spectra of the solution 86 following reaction at 2 atm the following day. After recording an initial IR spectrum and saving it, the CO2 present in the gas over the solution was purged with Ar once and an IR spectrum, was recorded. This was then repeated with two replacement cycles of the gas over the solution with Ar. Following this, the Ar was evacuated and replaced with CO2.

−1 The bands shown in the FTIR region 1500-2600 cm show only CO2 and the oxazolidinone product and a broad band near 1667 cm−1.

Figure 4.11 Spectra showing the fall and rise of a broad band centered at 1667 cm−1 assigned to the proposed ethoxycarbonate complex [Au(IPr)(OCO2Et)]. The same band can be found at the end of reactions in which PPA at low concentration levels is reacted with CO2 using 1 as catalyst. It is not formed if 2 is used as catalyst. In spite of several attempts to grow crystals suitable for structural determination those efforts have so far been unsuccessful. 87

4.2.9. Reaction of the Carbamic Acid of (n-bu)(benzyl)amine (BBA) and 1

The saturated analog of PPA, BBA was investigated under reaction conditions

identical to those used in studies of reactions with 1. In contrast with PPA cleavage by

the carbamic acid of BBA does not lead to further reaction since it does not contain the

reactive alkyne unit. Exposure of BBA to CO2 yielded clean production of the

corresponding carbamic acid derivative as shown in reaction 4:

It was anticipated that this would provide a good model reaction for the rate of

acid cleavage of 1 by the carbamic acid of PPA. The rate of reaction (5) was investigated

under the standard conditions of [BBA] = 0.052 M and [1] = 0.0072 M had a rate

−1 constant of kobsd = 0.01 min under 2 atm pressure (absolute) of CO2.

The fact that the kobsd value for reaction of 1 with BBA carbamic acid was nearly identical to the data reported in a previous section for reaction of CO2 alone suggested the

possibility that the reaction could occur by reactions (3) and (6). 88

Detailed examination of the FT-IR data did not show any sign of intermediate

[Au(IPr)(OCO2Et)]. This implies that either the direct reaction with CA of BBA occurs or that if a mechanism such as shown in (3) and (6) occur then reaction (6) is sufficiently

rapid as to not allow buildup to detectable levels of [Au(IPr)(OCO2Et)]. It does not

appear likely that the reaction with CO2 alone would be halted by addition of BBA, and

so it seems reasonable to assume that any [Au(IPr)(OCO2Et)] formed would be rapidly converted into [Au(IPr)(OCO2BBA)] by reaction (6). The most surprising result in this

section was the much slower rate of reaction of the CA of BBA compared to the CA of

PPA. Additional investigations of these reactions as a function of CO2 pressure are

planned. More detailed analysis of this question is beyond the scope of the present work.

4.2.10. Variable Temperature Study of Catalysis in Dilute Conditions

The rate of product formation as well as the rate of approach to steady state

conditions was studied under dilute conditions where the rate determining step is

postulated to be the cleavage of IKa by acid to products. The two major processes involve

reactions shown in Figure 4.1 (formation of 1) and Figure 4.2 (protodeauration of 1) can

be written in simplified form in Eqs. (7) and (8) below. 89

A key to solving the kinetics is accounting for the total Au catalyst in Equation

(9).

The [AuCarb] cannot be measured accurately at high substrate loadings due to overlapping bands due to CA. In contrast to earlier work in reactions in which 2 was used as catalyst, the [Missing Au] cannot be taken to be zero at low concentrations of PPA. At high concentrations of PPA this is no doubt approximately true, but at high concentrations of PPA the concentration of 1 cannot be determined.

The approach in this work, based on utilization of 1, may allow the assumption

that [Missing Au] is low. The discovery of the existence of [Au(IPr)(OCO2Et)] as a new

category for [Missing Au] may limit that as discussed later. Nevertheless, this provides

an avenue to estimate the temperature dependence and activation parameters for the

catalytic process. Provided the [Missing Au] is low, Equation (10) simplifies to

postulated approximate Equation (11)

The rate equation under these conditions only would then be Equation (11): 90

Under the condition that k1 >> k2 [CA], Equation (11) reduces to rate = d[P]/dt =

k2 [CA][T] which can be rearranged to: k2 ≈rate/[CA][T]. It should be noted that in this mechanism, we and others, propose that k2 = rate/[CA][ 1 ]. We have performed variable

pressure studies and also determined [1] at 21 °C where we have measured [CA] as well

as [1] as described above. For preliminary analysis, we have assumed here that [1]

≈½[Total Au] and [CA] ≈½[PPA]. Refinement of that assumption would require

assessment of the variable temperature pressure dependence of PPA under CO2 as a

function of pressure, as well as measurements of [1] as a function of temperature. For the purpose of an approximate estimate of the activation parameters we have collected rate data as a function of T shown in Figure 4.12 below. An Eyring plot of these data is shown in Figure 4.13.

These data lead to determination of activation parameters ∆H‡ = + 3.8 kcal/mol,

and ∆S‡ = -55 cal/mol K. In view of the rather low correlation coefficient considerable

error is assigned to these measurements. It should be emphasized these data are for k2

≈rate/[½PPA][1t=0] and not for k2; they are approximate values only. The data are,

however, in good agreement with computational values discussed later.

4.2.11. Computational Study of Proton Transfer to 1

We have previously reported a DFT study with a plausible mechanism for the

carboxylative cyclization of propargylamines using 2 as catalyst to yield complex 1 as

shown in Figure 4.1 [14]. Likewise, the thermodynamics of the protonolysis of the Au-

vinyl bond present in 1 with several proton sources was also examined in that article as

shown schematically in Figure 4.2. However, the mechanism and kinetics of this later

step were not studied. In this regard, to gain a better understanding of the protonation 92

processes that have been studied experimentally and discussed above, DFT calculations

were performed in the current work. Calculations were performed using truncated species

in which the IPr NHC ligand present in 1 was substituted by the simpler IMe ligand, the

benzyl group present in the propargyl amine employed experimentally was also replaced

by a methyl group and methanol was used as solvent. In the current work, the truncated

complexes computed will be labeled with a t superscript.

The computed pathway for protonation of 1t with PPA•HClt and subsequent

liberation of the final oxazolidinone product is shown in Figure 4.14. In accord with

experimental observations, this process is exergonic and has a relatively low kinetic

barrier (∆H‡ = + 1.6 kcal/mol and ∆S‡ = −48 cal/mol K). The Gibbs energy barrier

computed in this work of 15.9 kcal/mol is in reasonable agreement to that previously

computed for the reaction of 1 with PPA’•HCl (PPA’ = 1-methylamino-2-butyne) by Lin

and coworkers of 12.3 kcal/mol [88].

Figure 4.14 Gibbs free energy diagram computed at the PBE0-D3(BJ)(PCM,CH3OH)/Def2-TZVP level of theory for proton transfer from PPA•HClt to 1t and liberation of the oxazolidinone product. R = CH2C≡CCH3. 93

The optimized structure of the first transition state computed for the protonation

of the truncated complex 1t by PPA•HClt is shown in Figure 4.15. The structure of the

transition state located for the protonation of 1t by the model carbamic acid, CAt, is also

shown in Figure 4.15 for comparison.

Figure 4.15 Optimized structures computed for the transition states for reaction of 1 t with CAt (left) and with PPA•HClt (right) at the PBE0-D3(BJ)(PCM,MeOH)/Def2-TZVP level of theory. Hydrogen atoms other than the one being transferred omitted for clarity. The structures of the two transition states located are similar as can be seen in

Table 4.4 and involve only proton transfer to the vinylic carbon bound to the metal center

yielding the oxazolidinone product coordinated by the alkene moiety to the Au(I) center

as shown in Figure 4.14. 94

Table 4.4 Structural parameters computed for the transition states for protonation of complex 1t. a Values corresponding to an O…H distance. b Values corresponding to a N…H distance. Between brackets values corresponding to the optimized structure of the transition state computed previously by Lin and coworkers [88]. Finally, the last step of the reaction is ligand displacement of the oxazolidinone

product by the chloride anion through a Y- shaped transition state (see Figure 4.14). The

computed activation parameters in MeOH solution for reaction of CAt and 1t are ∆H‡ =

+1.6 kcal/mol and ∆S‡ = −47 cal/mol deg. These values are essentially identical to those

computed for the analogous reaction with PPA•HClt as the proton source (see above).

Moreover, these data may be compared to the experimental activation parameters for the

composite production of product under low CA conditions of ∆H‡ = +3.8 kcal/mol and

∆S‡ = −55 cal/mol deg (See Figure 4.13). The agreement between experimental and

computational values is good and both support a process with a very low activation

barrier and kinetically controlled by a high activation entropy. At the low PPA 95

concentrations studied experimentally, the rate of acid cleavage is low and is no doubt the dominant factor controlling the temperature dependence. The difference in entropic barriers may be due to less steric hindrance in the truncated system compared to the

actual system. For example, the IPr ligand is significantly more sterically demanding than

the truncated IMe used in our calculations. Examination of the structure in Figure 4.3

shows clearly how the Au center is obscured by the bulky ligand and the entropic price

for clearing a way to the active site would be expected to be larger for IPr compared to

the computed complex. The activation enthalpy computed for protonation of 1t by

methanol is ∆H‡ = 20 kcal/mol, much higher than those reported for the more potent

acids CA t and PPA•HClt. However, this process is favored by the fact that the proton source is the solvent and it is present in solution in large concentrations and increasing

considerably the rate of reaction. This is in agreement with the experimental observation

that the oxazolidinone product is also formed in alcoholic solutions in the presence of

t CO2 and in the absence of other added acids. The protonation of 1 by MeOH and

subsequent product loss yields the [Au(IMe)OMe] complex in a slightly endothermic

process ( ∆H0 = 1.4 kcal/mol).

4.2.12. Computational Study of the Reversible CO2 Insertion into the [Au(IMe)(OMe) Complex

In order to provide additional evidence for the reaction product of 1 and CO2 in

the absence of other coordinating products which yielded the oxazolidinone product in quantitative yield, the proposed formation of a highly labile complex formulated as

[Au(IPr)(OCO2Et)] based on its infrared spectrum and establishment of a rapid

decarbonylation/carbonylation equilibrium in ethanol solution ( vide supra ), DFT 96

calculations were also carried out to explore the formation of the simpler model complex

[Au(IMe)(OCO2Me)] as shown in Figure 4.16.

Figure 4.16 Gibbs free energy diagram computed at the PBE0-D3(BJ)(PCM,CH3OH)/Def2-TZVP level of theory for CO2 insertion into the [Au(IMe)(OMe)] complex. Once the [Au(IMe)(OMe)] complex is formed after the slightly endergonic

protonation of complex 1t with a molecule of methanol solvent as discussed in the

previous section, the insertion of CO2 into the Au-O bond of the methoxy complex is

thermodynamically favorable and kinetically facile as seen in the schematic Gibbs energy

diagram shown in Figure 4.16 . Since several isomers are possible depending on the

oxygen directly bound to the Au metal, a number of isomerizations are required to form

the most stable minimum [Au(IMe)(OCO2Me)] complex whose optimized structure is

shown in Figure 4.17 and can be compared with the previously reported structure of the

corresponding carbamate species [14]. 97

Figure 4.17 Optimized structures of [Au(IMe)(OCO2Me)] (left) compared to previously reported computed [14] structure for Au(IMe)OC=ON(Me)(CH2C≡CCH3) (right) . There are similarities between the two structures. The Au-O bond lengths for the

coordinated O atom are 2.046 and 2.041 Å, and the Au-O-C bond angles are 119.4 and

118.5° respectively for the carbonate and carbamate species. The distances from the Au

metallic center to the non-bonded oxygens are 3.154 and 3.089 Å indicating a weak

Au···O interaction with the acyl group of the pendant carbonate or carbamate ligand,

respectively. The computed asymmetric stretching νO=C=O IR frequencies scaled by a

0.96 factor for both compounds shown in Figure 4.17 appear at 1642 and 1589 cm−1 for

the carbonate and carbamate species, respectively. These values are in qualitative

agreement to the IR bands observed experimentally of 1667 and 1642 cm−1 assigned to the ethoxy carbonate [Au(IPr)(OCO2Et)] complex and the carbamate

[Au(IPr)(OC=ON(nBu)(CH2Ph))] complex. Finally, the CO2 loss, the reverse reaction to that shown in Figure 4.16, has a low computed Gibbs activation barrier of 14.5 kcal/mol in agreement with the reversible nature of this transformation after several evacuation and refill cycles with Ar and CO2 as discussed in a previous section. 98

4.2.13. Discussion

This work reports detailed studies of the rate of acid cleavage of 1 as part of an

effort to determine the transition states and activation barriers to the two major processes

in carboxylation cyclization of PPA as shown in Figs. 1 and 2. Ikariya and coworkers[81] have reported earlier structure/reactivity studies for two vinylgold complexes structurally similar to the PPA derivative shown in Figure 4.3. An advantage to the use of FT-IR spectroscopy to probe reaction kinetics is that it allows, at least for low substrate

loadings, simultaneous monitoring of both product formation and concentration level of

the key intermediate vinyl gold complex 1. Experimental work aimed at determining the

reaction barrier to formation of this complex, as shown in Figure 4.1, will require first

determining the barriers to reaction 2. These two reactions are coupled and it is

postulated that each contributes to the net turnover time in the catalytic cycle. At low acid

loadings, the rate determining step is generally held to be the reaction with acid

(protodeauratation) to yield products. The present work focuses on those conditions. As

originally envisioned and shown in Figure 4.2, the three major reagents for acid cleavage,

are PPA-HCl, PPA-CO2 (CA), and PPA-CA (CS). The most potent thermodynamically is

PPA-HCl but its kinetic role is not fully understood. The source of PPA-HCl is 2 and its

concentration is generally rather low. Some beneficial aspects to having low chloride

concentration levels have been proposed in the literature, and among these is its potential

role in acid cleavage.

The second acid reagent CA has the advantage that it is present in much greater

concentrations that PPA-HCl. My earlier work in this topic highlighted to previously

overlooked role of CA and CS reagents in this catalytic process [14]. The ratio of these 99

reagents changes with temperature, CO2 partial pressure and concentration. While nucleophilic displacement reactions involving CS would be expected to be rapid, to the best of our knowledge, no information is available about their possible role in experimental acid cleavage (protodeauration) reactions.

Variable pressure studies reported here showed an increase in the rate of both the catalytic reaction as well as the protodeauration step. Synthetic studies have led to the conclusion that in terms of reaction yield over a fixed period, the partial pressure of CO2 is unimportant. This work clearly shows, particularly at low substrate loading, that rates are accelerated at higher pressures. Studies of carboxylation under pressure at low PPA loadings led to discovery of an intermediate assigned as [Au(IPr)(OCO2Et)] based on its

infrared spectrum and establishment of a rapid on/off reaction as carbon dioxide is added

or removed, and supported by computational calculations. This led to study of the role of

CO2 in alcohol solutions as a cleavage reagent. This had been qualitatively reported by

Ikariya earlier but only with respect to formation of organic products. The pressure

dependent studies reported here show that at high pressures CO2 is a kinetically

competent reagent for cleavage to product.

The current work presents rate measurements for acid cleavage of a different

vinyl gold complex in a different solvent. Second order rate constants derived for this

reaction with 1 at 21 °C and 15 psi CO2 (2 atm pressure absolute) and catalyst loadings

−1 −1 −1 −1 on the 0.0065 M level are: PPA-HCl (9.6 M min ) > PPA-CO2 (0.54 M min ) >

−1 −1 −1 −1 CO2 /EtOH (0.035 M min ) > BBA-CO2 (0.01 M min ). These data span a range of

three orders of magnitude. As expected, the fastest reaction occurred with PPA-HCl. A 100

plot under “real-world” conditions of the rate of product production for these is shown in

Figure 4.18.

Figure 4.18 Plots of the rate of production (M) of oxazolidinone product versus time (Min) at the 0.0065 M level of [1] at 21°C in EtOH. Unless state otherwise pCO2 = 15 psi = 2 atm (absolute). The most important catalyst is clearly the carbamic acid CA in spite that it has a

rate constant approximately 20 times lower than that of PPA-HCl. The principal reason is

that PPA-HCl never exceeds the low concentration of 2. The reaction between two species at the 0.0035 M level each (total 0.007 M) will lose the competition to PPA at higher concentrations. This is shown in Figure 4.18 for 0.252 M PPA where the rate of reaction is steeper than PPA-HCl even at its maximum rate. For the 0.052 M reaction however, PPA-HCl is competitive even at the reduced level. 101

One of the most interesting results in this work was the acceleration of the rate of

acid cleavage of 1 by CO2 alone as pressure was increased from 2 to 3 to 60 atm. The purple curve in Figure 4.18 shows that CO2 cleavage is relatively slow. However, it has

the advantage that it can be amplified by increasing the pressure. At 60 atm CO2 pressure

the reaction was competitive with any of the reagents shown in Figure 4.18. The

chemistry of CO2 in pressure expanded alcohol solvents has been stressed by Liotta and

co-workers [89]. This aspect is under further investigation in our laboratories.

The other major surprise, observed at all pressures when 1 reacted only with CO2

[90] was the formation of the proposed complex [Au(IPr)(OCO2Et)]. Sadighi has reported

the preparation of [Au(SIPr)(OtBu)], and also the insertion of CO2 into [Ag(IPr)(R)].

There are other suggested reactions involving CO2 converting a gold alkoxide into a

metastable alkoxycarbonate which may present a better leaving group in chemical

conversions. In spite of that there are no reported structures of the proposed complex.

Ongoing efforts in our lab continue towards the isolation and full characterization of this

up to now elusive complex. Computational data indicate that the complex may exist in an

unfavorable equilibrium with the carbamate complex as shown in Equation (12)

Indeed, this very possibility is what may upset our goal to determine the

[Au(IPr)(Carb)] concentration. In the infrared spectra, we are only able to accurately

resolve the bands at low carbamate loadings; otherwise computer subtraction of the

strong substrate bands mask the bands assigned to the vinyl gold and gold carbamate

complexes present in much lower concentrations. 102

4.3 Conclusion

This work is aimed at better understanding the barriers to the cyclocarboxylation

reaction of alkynes by monuclear and dinuclear complexes. As reported earlier, the

different rates of reaction for the differing linker complexes span a range of behaviors [14].

In some cases the yields are higher than for the benchmark complex [Au(IPr)Cl] ( 2 ), but in others they are not. The origins of the favorable and unfavorable interactions may only be rationalized for the dinuclear complexes once the reactivity at the single metal center model complex are known. Progress has been made regarding understanding the second major process the protonation of 1 to yield product. Additional work in progress is aimed at better understanding the first major process-regenerating 1 from the reaction products of protonation.

4.4 Experimental Details

4.4.1 General Data

Experimental work, materials, and methods in this work were performed using

techniques which have been described in detail previously [14]. Representative kinetic

measurements as well as the high pressure study of reaction of 1 and CO2 are described in

detail below.

4.4.2 Kinetic Studies of Carboxylation of PPA by 1

In a typical kinetic study, a solution containing 42.5 mg PPA in 5 mL of EtOH

that had been dried and stored over molecular sieves was prepared and loaded into a

Hamilton gas tight syringe in the glovebox. The reactor was loaded under a CO2 103 atmosphere with an inner glass tube containing 25 mg of 1 . The reactor was purged and under a flow of CO2 the 5 mL solution of PPA was added without dissolution of the catalyst. The reactor was stirred for approximately 30 minutes under CO2 atmosphere.

During that time IR spectra were taken until the carbamic acid/carbamate equilibrium process for PPA was complete. IR data were collected by transfer and return of an aliquot of the solution under constant CO2 pressure to a Harrick medium pressure FTIR cell with

CaF2 windows through stainless steel and heavy wall Teflon tubing. The catalytic reaction was initiated by shaking the tube vigorously and mixing the catalyst with the already equilibrated solution of the carbamic acid and carbamate salt of PPA.

4.4.3 High Pressure Study of Carboxylation of PPA

In the glovebox a 0.007 M solution of 1 in degassed anhydrous EtOH was prepared and 2 mL loaded into a syringe. The syringe was taken from the glovebox and under a flow of argon connected to a special pressure reactor and loaded under a flow of argon. The reactor consisted of a 30 mL stainless steel bomb with pipe thread fittings at the top and bottom. At the bottom a ¼ inch diameter stainless steel test tube ≈6 inches long was attached via a ¼ inch Swagelok fitting. The top was connected by a pipe thread to Swagelok adapter to a ¼ inch stainless steel cross. One arm of the cross had a high- pressure valve and was connected to a Schlenk line. The apparatus was evacuated and filled with argon several times and then loaded under an outflow of argon. The apparatus was also connected through flexible high-pressure tubing to a gas cylinder of research grade CO2. The cylinder pressure of 900 psi CO2, 61.2 atm, was placed above 1.2 atm Ar for a partial pressure in the reactor of ≈60 atm. 104

[CAUTION: This reaction was done making use of a safety shield and taking

extreme caution. Carbon dioxide dissolves readily in EtOH and its use under high

pressure requires special precautions.] The reactor was filled and shaken for ≈10 minutes.

During the shaking procedure the valve to the main tank was closed. Following the

uptake of CO2 the reactor was refilled to 900 psi each time. The reactor was vigorously shaken to ensure saturation of the EtOH with CO2. It was shaken in such a way as to

ensure that the EtOH left the ¼”diameter test tube and went up into the larger diameter (

≈1.25”diameter 30 mL bomb). The timer was started upon initial exposure of the solution

to CO2 at pressure. After ≈10 minutes the valve to the CO2 cylinder was closed, and the

pressure within the bomb was released through a valve at the top. In order to degas the

solution, it was shaken again so that the solution would lose CO2 in the larger 30 mL

bomb. This procedure was done rapidly three times. During this time the Joule cooling

effect lowered the temperature of the test tube to 0 °C or below. The Swagelok fitting to

the ¼tube containing the solution was disconnected and a Hamilton gas tight syringe

fitted with a long narrow-gauge needle was used to remove the solution to fill the FTIR

cell. The first spectrum was run at about 11 min and showed approximately 70%

conversion to oxazolidinone product. The presence of a strong band attributed to IPrAu-

OCO2Et was found in the same position as lower pressure experiments.

4.4.4 Crystallographic Analysis

Single crystals of 1 suitable for diffraction analysis were grown by slow

evaporation of a benzene/octane solvent mixture at room temperature. The data crystals

for 1 was glued onto the end of a thin glass fiber. X-ray intensity data were measured by

using a Bruker SMART APEX2 CCD-based diffractometer using Mo K α radiation ( λ= 105

0.71073 ˚A) [74]. The raw data frames were integrated with the SAINT+ program by using a narrow-frame integration algorithm [74]. Corrections for Lorentz and polarization effects were also applied with SAINT+. An empirical absorption correction based on the

multiple measurement of equivalent reflections was applied using the program SADABS.

The structure was solved by a combination of direct methods and difference Fourier

syntheses, and refined by full-matrix least-squares on F2, by using the SHELXTL

software package [75,76]. Crystal data, data collection parameters, and results of the

analyses are listed in Table 4.1. Compound 1 crystallized in the triclinic crystal system.

The space group P1 was assumed and confirmed by the successful solution and

refinement of the structure. All non-hydrogen atoms were refined with anisotropic

displacement parameters. Hydrogen atoms were placed in geometrically idealized

positions and included as standard riding atoms during the least squares refinements.

Crystallographic data (cif file) for the structural analysis has been uploaded with this

report and has also been deposited with the Cambridge Crystallographic Data Centre,

CCDC 2020043 for compound 1.

4.4.5 Computational Section

Electronic structure calculations were carried out using the PBE0 functional[41],

the D3 version of Grimme’s dispersion with Becke-Johnson damping (D3(BJ))[42] and

using the Def2-TZVP basis set[43] along with the corresponding ECP for Au[108]. All

stationary points were optimized in methanol solution with the polarizable continuum

model (PCM) using the integral equation formalism[44]. Transition states were

characterized by a single imaginary frequency, whose normal mode corresponded to the

expected motion. All calculations were performed with the Gaussian 16 suite[40]. Chapter 5 Ball Milling Reaction of Na2O and Na2O2 with N2O 5.1 Background

Abatement of ·NO, ·NO2 and N2O from industrial exhaust gases is an important

transformation, particularly for the high-volume Ostwald process which combusts

[91] ammonia to produce nitric acid. The two radicals ·NO and ·NO2 are much more

[92] reactive than N2O. In the BASF DeNOx process they are reduced to dinitrogen and water as shown in equations (1) and (2).

4 ·NO + 4NH3 + O2 4N2 + 6H2O (1)

6 ·NO2 + 8NH3 7N2 + 12H2O (2)

The harsh pollution arising from ·NO, ·NO2 and their reaction products with air

has been apparent f or more than a century. The need to also remove waste N2O has

increased after it became recognized as a significant at mospheric greenhouse gas three

hundred times more harmful than carbon dioxide.[93] The BASF DeNOx process destroys

it by high temperature thermal decomposition of a ceria-based catalysts covered with

platinum gauze[92] as shown in equation (3).[94]

0 2N2O 2N2 + O2 ∆G 298 = -52 kcal/mol (3)

In addition to destruction of valuable NH3 in the DeNOx process, it also wastes

the chemical potential present of a set of reactive species {·NO, · NO2, N2O} which could

be further oxidized rather than reduced. Nearly a hundred years ago Lewis and

Randall[95] concluded that oxidation of dinitrogen to nitric acid had a free energy change

near zero, and that at ambient conditions an equilibrium solution of 0.1 M HNO3 could be

produced:

0 2N2(g) + 5O2(g) + 2H2O(l) 4HNO3(aq) (0.1M DG 298 = 0 kcal/mol (4)

106 107

A catalytic process for equation (4) could result in elimination of the Ostwald

process entirely. It presents a formidable challenge. For that reason, we have begun investigation of reactivity of nitrous oxide with metal oxides and peroxides as shown in

equation (5) and equation (6):

0 Na2O(s) + N2O(g) + 2O2(g) 2NaNO3(s) DG 298 = -110 kcal/mol (5)

0 Na2O2(s) + N2O(g) + 1.5O2(g) 2NaNO3(s) DG 298 = -86 kcal/mol (6)

To our knowledge direct conversion of nitrous oxide to nitrate has not been

reported. A path to reaction (5) might proceed through addition of nitrous oxide to a

metal oxide to form a hyponitrite complex. The reverse of this reaction has been

extensively studied due to its importance in enzymatic reduction of nitric oxide.[96] There

are surprisingly few reports of addition of nitrous oxide directly to a metal oxo compound

to form a hyponitrite.[97] A notable exception is the report of Jansen and Feldmann[9] that

N2O and Na2O react in a tube furnace to produce cis-Na2N2O2 as shown in equation (7).

° Na2O(s) + N2O(g) cis-Na2N2O2(s) (360 C, 2 h) (7)

[9] The other reported preparation of cis-Na2N2O2 is indirect and involves reaction

of ·NO with Na in liquid NH3. The structurally characterized complex of Jansen and

Feldmann was found to have greater stability than amorphous cis-Na2N2O2 and began

[98] decomposition to sodium orthonitrite (which may be viewed as NaNO2 trapped in a

° matrix of Na2O) at temperatures above 360 C.

In this chapter, studies regarding the ball milling of sodium oxide and sodium

peroxide with nitrous oxide to produce cis-Na2N2O2 are reported. Further study has

revealed that the addition of additives such as the potassium halide salts, KCl, KBr and

KI can increase the yield of cis-Na2N2O2 and promoted further oxidation to 108

NaNO3.Through tandem FTIR analysis and computational studies a reaction scheme is

proposed.

5.2 Results and Discussion

Initial attempts at ball milling Na2O and N2O met with limited success. FTIR

spectra recorded through air exclusion ATR measurements revealed the characteristic

bands of cis-Na2N2O2 in moderate yields. The very moisture sensitive nature of the

mixture of unreacted sodium oxide and cis-Na2N2O2 always led to slow hydration and

decomposition releasing N2O gas. Pellet spectra taken in KBr left no doubt that cis-

Na2N2O2 was formed, matching very closely with bands reported by Jansen and

Feldmann[9]. Production of nitrate was not observed.

In attempts to increase yield, the ball milling reactions were repeated with

additional reagents such as silica gel, zeolites, fluorocarbon oil, silicone oil, CaO, and

CaF2 which were added to the mixer cells. The reactions did not occur at room

temperature. Heating to 80°C and then 140 °C also failed to yield any detectable product.

At temperatures of 140 °C, the fluorocarbon oil under these conditions was found to react

with the sodium oxide forming a cross linked polymer.[100] In all exploratory reactions,

the authors have used small quantities of active ingredients and exercised prudent safety

measures which are essential in the high energy ball milling of reactive materials.[101]

Eventually it was decided to mill a mixture of anhydrous KBr (1.5g) and Na2O (0.5g).

This resulted in a large increase in cis-Na2N2O2 yield and with further milling resulting in

the steady production of sodium nitrate (NaNO3). Representative spectra are shown in

Figure 5.1. 109

14 Figure 5.1 Comparison of FTIR data for ball milling results of Na2O/KBr under 3 atm N2O (red) and 15 N2O (black). The spectra shown were obtained after 6 hours of ball milling. Production of nitrate was proven by direct addition methods to the KBr pellet spectra and quantified by dissolution of the samples in distilled water as discussed in

15 section 5.4. Ball milling of the KBr/Na2O sample matrix under N2O resulted in isotopic shifts in excellent agreement with literature values for Na2N2O2 and NaNO3 and is summarized in table 5.1.

14 15 Table 5.1 Tabulated peak assignments for balling milling reactions of Na2O and N2O or N2O.

14 -1 15 -1 Peak Assignment Na2 N2O2 (cm ) Na2 N2O2 (cm )

N2O 2229.9 2158.4

NaNO3 1384.6 1352.4

Na2N2O2 N-O 1330.4, 1319.9 1287.9, 1277.2

Na2N2O2 N=N 1099.8, 1071.5 1072.0, 1043.7

Na2N2O2 N=N-O 886.8, 855.0 873.8, 854.5 110

Cis-Na2N2O2 peak values match very closely with those reported by Jansen and

Feldmann[9]. Note that in all FTIR data in KBr it is not possible to assign the counterion such that this and other salts could be referred to as {Na, K}N2O2 but this is not done for simplicity.

FTIR spectroscopic data for the initial rate of reaction are shown in Figure 5.2 for

Na2O/KBr pellet spectra. There is a steady growth in the peak assigned to nitrate. The only source of oxygen in these reactions was nitrous oxide. The net stoichiometry of nitrate formation under these conditions corresponds to reaction (8).[102]

0 Na2O + 5N2O 2NaNO3 + 4N2 ∆G 298 = -210 kcal/mol (8)

Figure 5.2 KBr pellet FTIR spectra as a function of time for ball milling of Na2O (0.5g) in KBr (1.5g) under 30 psi, 2.0 ATM N2O. The KBr pellet data give a picture of the relative ratios of product and how they change with time, but more quantitative data were obtained by studies of hydrolysis reactions of weighed solid samples. The initial rate of nitrate production derived from 111

aqueous ATR spectra of hydrolyzed samples as a function of time are shown for KF,

KCl, KBr and KI in Figure 5.3.

Figure 5.3 Initial rates of production of nitrate as a function of time (hours) for KF, KBr, KCl, and KI. The rates of reaction appear to be roughly in the order KBr = KCl = KI » KF. The

spectra for the KF reaction also appear to show lower amounts of cis-Na2N2O2 than the other alkali metal halides. Ball milling the reaction and then stopping it under gas pressure does not lead to continued product buildup in our experience to date. It is only under active ball milling conditions that nitrate is produced. The process of removing a sample in the glove box results in inevitable contact with water which produces NaOH which does not react with N2O under these conditions.

The extinction coefficient for cis-Na2N2O2 is not known, and due to its extreme

moisture sensitivity only estimates can be made. Hydrolysis in acid solution releases all

bound N2O, including that which is encapsulated in the matrix. Even in the absence of

added Na2O, KBr alone under milling conditions was shown to lead to pellets with 112

14 encapsulated nitrous oxide and these may be seen in Figure 5.1 for both N2O and

15 [103] N2O. Grassian and coworkers have reported infrared data for surface bound N2O on

higher valent oxides as have others[104]. Encapsulated carbon dioxide has also been

detected in KBr pellets[105] and may also arise from in situ chemical generation of carbon

dioxide. Our results on ball milling of only KBr and N2O and getting a small signal of

encapsulated N2O are attributed to a minor trapping which may occur as part of the ball

milling procedure. Long term study of pellets containing cis-Na2N2O2, show that there is a parallel decrease in bands due to hyponitrite together with an increase in bands of

encapsulated nitrous oxide suggested that slow migration of trace water through the pellet

matrix is responsible.

In solution it is known that the cis-Na2N2O2 rapidly loses N2O however the trans

isomer does not appreciably decompose in basic solution.[106] Study of hydrolysis and the

volume of N2O evolved as detected by gas phase FTIR studies has allowed us to piece

together a rough reaction profile showing the time evolution of products in this reaction.

A first picture of these results is shown in Figure 5.4.

As shown in Figure 5.4 there is a slow production of NaNO3 that occurs

throughout the reaction. Surprisingly the N2O evolution studies showed that after an

induction period of about four hours there is a fairly rapid rise in concentration of cis

Na2N2O2 to nearly 0.6 mole fraction of all distributed sodium oxide starting material.

This then decreases and remains roughly constant. To our surprise, the hydrolysis product

showed consistent low levels of trans-Na2N2O2 < 0.1 mole fraction present in solution.

An exact match was obtained by comparison of an authentic sample. There is no

evidence in the pellet FTIR data for formation in the pellet itself of trans-Na2N2O2. It is 113

possible this arises from the cis isomer hydrolysis, but it seems more likely that it may

arise from hydrolysis of an unidentified intermediate material. It should also be added

that iodometric titration for the presence of peroxide in the solution shows levels less than

4% in these reactions. In most cases there are no signs of significant Fe concentration in

the salt matrix or corrosion of the mixing chamber. In the presence of not sufficiently dry

material, the presence of moisture has led to some corrosion and detection of Fe in the

salt matrix. The role of trace Fe catalysis of the oxidation cannot be excluded, but provided all materials are rigorously dry there is no significant evidence for that.

Figure 5.4 Mole fraction in total Na distribution for the first 20.5 hours of ball milling 0.5 g Na2O in 1.5 g KBr in standard conditions. The crude kinetic data shown in Figure 5.4 are also complicated since this is not a

simple reaction in solution. The apparent slow rise in the initial formation of cis-Na2N2O2

may be because it requires several hours to reduce the particle size down to the micron or

nanometer scale. Once the near nano-meter scale is reached and the surface area and activity peak, the true rate of binding of N2O to activated Na2O may be faster than 114

apparent. The tentative conclusion can be made at this point that formation of cis-

Na2N2O2 is faster than its oxidation. Extended ball milling for time periods of 50 hours

results in yields approaching 50 percent. To improve the oxidation step and yield of the

overall reaction, ball milling studies were repeated by replacing Na2O with increasing amounts of Na2O2. A comparison of the reactions can be seen below in Figure 5.5.

Figure 5.5 Initial rates of nitrate production as a function of time for samples of 1.5g KBr and 0.5g Na2O2 (blue), 0.25g Na2O2 + 0.25g Na2O, and 0.5g Na2O.

Addition of Na2O2 resulted in an increase in nitrate production of approximately

20% either in combination with Na2O or fully replacing it. The increase in nitrate

production results even while cis Na2N2O2 levels are lower. Pellet spectra recorded at 8

hours reveal very little cis Na2N2O2 in the peroxide reaction; whether this is a result of

increased cis Na2N2O2 consumptions or evidence of another reaction pathway are under

investigation. A comparison of KBr/Na2O and KBr/Na2O2 pellets can be seen in Figure

5.6. 115

Figure 5.6 Pellet spectra obtained after about 8 hours of milling corrected for nitrate concentration. The 1.5g KBr, 0.5g Na2O spectrum (black) shows much higher absorbance for cis Na2N2O2 compared to 1.5g KBr, 0.5g Na2O2 (red). 5.3 Conclusion

Despite a relatively slow rate of conversion and only moderate yields, this represents the first conversion of nitrous oxide to nitrate at low temperatures and

pressures of N2O. In addition, there is now more ready access to investigations of the

o reactivity of cis-Na2N2O2 which may be prepared in situ at 38 C and possibly lower

temperatures. The RETSCH mm500 is equipped to study such low temperature reactions

and such work is planned. Work in progress[107] is aimed at understanding the current

mechanism and designing an improved one with increased rate and product yield from

the full set of gases N2O, ·NO, ·NO2, and O2. The fact that N2O can be converted at

ambient conditions to sodium nitrate, makes the search for achievement of a Lewis-

Randall catalyst based on reaction (5) seem less impossible. 116

5.4 Experimental Details

5.4.1 General Data

All manipulations were performed in a Vacuum/Atmospheres glove box equipped

with a Braun gas purifier system. Open trays of indicating drierite were kept along the

back of the glove box to reduce moisture content. Nitrous oxide, oxygen, and other gases

used were obtained from Airgas, of research grade, and used as received. Isotopically

15 15 labelled N2O was obtained from Cambridge Isotopes and was 99% N and used as

received. KF, KBr, KCl, KI salts were obtained from commercial vendor and were ACS

certified purity. Salts were typically dried in an oven at 250°C in an oven for several

hours and then taken into the glove box and stored in sealed dry containers. Sodium

Oxide (86.5% Na2O, 13.5% Na2O2) and Sodium Peroxide (95%) were obtained from

Alfa Aesar and used without purification. Mixer mill experiments were performed in a

RETSCH mm500 mixer mill in a 50 ml stainless steel grinding jar with a top with two ports for gas addition/evacuation. Infrared data were obtained on a Perkin Elmer Frontier

FTIR spectrophotometer with MCT detector and ATR Diamond Cell Accessory. Gas

Phase FTIR were obtained in a 10 cm International Crystal Laboratories glass cell with two stopcocks and equipped with either NaCl or CaF2 windows.

5.4.2 Representative Ball Milling Experiment

In a typical experiment, the top and bottom and mixer cell pieces for the RETSCH

mm500 cell were heated in an oven to 145°C. Rubber o-ring and gasket materials were

heated in a lower temperature oven at 85°C. These are all taken quickly to the

antechamber of the glove box and evacuated and filled with nitrogen three times. Once 117

under nitrogen the mixer cell was allowed to cool to room temperature. A 0.5g sample of

Na2O was weighed and transferred to the mixer cell. To the mixer cell, 1.5g of KCl was

then added along with one 25 mm stainless steel ball and was then sealed in the glove

box. The cell was then transferred to the mixer mill and was evacuated and filled with

rd N2O in two cycles. On the 3 cycle, the cell was filled with 30 psi (~3 atm absolute pressure) N2O. The main tank valve to the N2O was then closed. This allowed the manifold and fitting and mixing jar to be under a near constant pressure since it could be replenished by a limited amount of gas on the high-pressure side of the manifold and the main tank valve of the gas cylinder. Under no circumstances have we done mixing reactions where the main tank valve was left open. The possibility of a hose coming loose under the active mixing conditions must be kept in mind at all times even with tubing secured by hose clamps. A typical single cycle was to run for 5 minutes mixing at 30 Hz, followed by a rest period cycle of 2 minutes. A set of 12 of these cycles provides one hour of active mixing with 24 minutes of rest to allow for adequate cooling. It is highly recommended that close attention be paid to the manufacturer’s instructions and that the powerful forces during high mixing speed not be underestimated. Due to the hazardous nature of some of the reactants, at no time was the instrument left running without the physical presence of at least one person.

5.4.3 Method for Obtaining High Quality KBr Pellet Data

The high hygroscopic nature of the reagents used precluded traditional pellet

methods by rapidly opacifying the surface. For this reason, a sandwich method was

developed to slow the rate of hydration and allowed for more consistent data. In the

glovebox, 100mg of dried KBr was added to a previously dried pellet press. The KBr was 118

spread to evenly coat the bottom of the press. A layer containing 2mg of sample and

98mg of KBr, mixed thoroughly, was then added on top of the KBr in the pellet press. A

final 100mg of KBr was added on top of the sample/ KBr mixture and the pellet was

pressed at 5t for approximately 5 minutes and an FTIR spectrum was taken. The layering

of the KBr was essential in slowing both the rate of product decomposition and

decreasing the amount of baseline uncertainty.

5.4.3 Measurement of Aqueous Nitrate in Solution

Quantitative measurement of NaNO3 was performed through hydrolysis with

distilled water and comparison to a calibration curve. Samples of approximately 150mg

were mixed with 1mL of distilled water then quickly analyzed by FTIR using the ATR

attachment. It is important to limit exposure to air during this time as idle samples can

absorb CO2 to produce Na2CO3 and interfere with measurement. Creation of a calibration

curve using ACS certified KNO3 at concentrations of 10mg/mL to 100mg/mL afforded a

good linear correlation and accurate nitrate detection in samples.

5.4.4 Measurement of Gas Evolution from Solid Samples

In order to accurately determine the volume of N2O absorbed by the sample the following method was developed. In the glove box, 100mg of sample was weighed into a small metal weigh boat and placed into a 10cm gas cell with CaF2 windows. The gas cell

was then evacuated and scanned by FTIR to ensure evacuation. While evacuated, 1mL of

distilled water was added to the gas cell through a septum onto the sample. Reaction with

water was immediate and results in a large amount of gas evolution. The cell was then 119

scanned to detect the amount of N2O and compared a calibration curve produced though

serial addition of N2O to the cell under vacuum without sample.

5.4.5 Quantitative Determination of Peroxide Content of Sample

Using standard iodometric titration procedures employing sodium thiosulfate

several samples were analyzed for the presence of Na2O2. An example of this procedure is as follows. A sample of approximately 150mg is weighed in the glovebox. To this sample, 2mL of distilled water is added and stirred until N2O evolution ceases at which

point 3mL of 60/40 mixture of glacial acetic acid and hexane was added. Note, never add

the acetic acid/ hexane mixture directly to samples containing Na2O or Na2O2, the

extremely exothermic acid/base reaction is potentially hazardous. Saturated NaI

(~0.1mL) was added to the DI water/acetic acid/hexane mixture under stirring,

production of a red color indicates peroxide. Titration of the solution to a colorless

endpoint using 0.1M Na2S2O3 allows for accurate determination of peroxide content in

samples as verified by comparison with known of amount of Na2O2. References

1. Karleskint, G., Introduction to Marine Biology. Cengage Learning: 2009.

2. Jang, Y. C.; Lee, J.; Hong, S.; Choi, H.; Shim, W.; Hong, S., Journal of the Korean Society for Marine Environment & Energy 2015, 18, 263-273.

3. Meehl, G. A., Stocker, T.F., Collins, W.D., Friendlingstein P., Gaye A.T., Gregory J. M., Kitoh A., Knittu R., Murphy J. M., Noda A., Raper S. C. B., Watterson I. G., Weaver A. J., Zhao Z. C., Global Climate Projections. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change Cambridge University Press: Cambridge, United Kingdom and New York, NY, USA., 2007.

4. Tian, H. Q.; Xu, R. T.; Canadell, J. G.; Thompson, R. L.; Winiwarter, W.; Suntharalingam, P.; Davidson, E. A.; Ciais, P.; Jackson, R. B.; Janssens- Maenhout, G.; Prather, M. J.; Regnier, P.; Pan, N. Q.; Pan, S. F.; Peters, G. P.; Shi, H.; Tubiello, F. N.; Zaehle, S.; Zhou, F.; Arneth, A.; Battaglia, G.; Berthet, S.; Bopp, L.; Bouwman, A. F.; Buitenhuis, E. T.; Chang, J. F.; Chipperfield, M. P.; Dangal, S. R. S.; Dlugokencky, E.; Elkins, J. W.; Eyre, B. D.; Fu, B. J.; Hall, B.; Ito, A.; Joos, F.; Krummel, P. B.; Landolfi, A.; Laruelle, G. G.; Lauerwald, R.; Li, W.; Lienert, S.; Maavara, T.; MacLeod, M.; Millet, D. B.; Olin, S.; Patra, P. K.; Prinn, R. G.; Raymond, P. A.; Ruiz, D. J.; van der Werf, G. R.; Vuichard, N.; Wang, J. J.; Weiss, R. F.; Wells, K. C.; Wilson, C.; Yang, J.; Yao, Y. Z., Nature 2020, 586 (7828), 248.

5. Jones, A. V.; Clemmet, M.; Highton, A.; Golding, E., Access to Chemistry. Royal Society of Chemistry: 1999 250.

6. Holleman, A. F., Wiberg, E., Inorganic Chemistry. Academic Press: 2001 662- 665.

7. Reimert, R.; Marschner, F.; Renner, H.-J.; Boll, W.; Supp, E.; Brejc, M.; Liebner, W.; Schaub, G., Gas Production, 2. Processes. Ullmann's Encyclopedia of Industrial Chemistry 2011.

8. Ammonium Nitrate Production by Country. Knoema.com (accessed 10 May).

9. (a) Feldmann, C.; Jansen, M. Z. Anorg. Allg. Chem. 1997, 623, 1803-1809. (b) Feldmann, C.; Jansen, M. Angew. Chem. Int. Ed. 1996, 35, 1728-1730.

10. Davis, J. V.; Guio, O.; Captain, B.; Hoff, C. D. ACS Omega, submitted.

120 121

11. Davis, J. V.; Gomage, M. M.; Guio, O.; Captain, B.; Temprado, M.; Hoff, C. D. Inorganic Chemistry,Submitted.

12. Lazell, E. W., Hydrated Lime: History, Manufacture and uses in Plaster, Mortar, Concrete; A Manual for the Architect, Engineer, Contractor and Builder. Jackson-Remlinger Printing Company: 1915.

13. Davis, J. V.; Gamage, M. M.; Guio, O.; Bayrakdar, T. A. C. A.; Nolan, S. P.; Temprado, M.; Captain, B.; Hoff, C. D., Journal of Organometallic Chemistry 2021, 934.

14. T.A .C.A . Bayrakdar, F. Nahra, J.V. Davis, M.M. Gamage, B. Captain, M. Temprado, M. Marazzi, M. Saab, K.V. Hecke, D. Ormerod, C.D. Hoff, S.P. Nolan, Organometallics, 2020, 39, 15, 2907–2916.

15. Palluccio T. D.; Germain M. E.; Silvia J. S.; Serafim L. F.; Davis J. V.; Hoff C. D.; Temprado M.; Cummins C. C.; Rybak-Akimova E. V., Submitting to Inorganic Chemistry.

16. a) Eady, R. R., Chem. Rev. 1996, 96 (7), 3013-3030; b) Fisher, K.; Dilworth, M. J.; Newton, W. E., Biochemistry 2006, 45 (13), 4190-4198; c) Crans, D. C.; Smee, J. J.; Gaidamauskas, E.; Yang, L., Chem. Rev. 2004, 104 (2), 849-902;

17. a) Rohde, M.; Grunau, K.; Einsle, O. Angew. Chem. Int. Ed. 2020, 59, 23626- 23630; b) Spatzal, T.; Perez, K. A.; Einsle, O.; Howard, J. B.; Rees. D. C. Science, 2014, 345 (6204), 1620-1623.

18. Allen, A. D., Senoff, C. V., Chem. Commun. 1965, 621-622.

19. a) Burford, R. J.; Fryzuk, M. D. Nat. Rev. Chem. 2017, 1, 0026.; b) Chalkley, M. J.; Drover, M. W.; Peters, J. C., Chem. Rev. 2020, 120 (12), 5582-5636; c) Kim, S.; Loose, F.; Chirik, P. J., Chem. Rev. 2020, 120 (12), 5637-5681; d) Singh, D.; Buratto, W. R.; Torres, J. F.; Murray, L. J., Chem. Rev. 2020, 120 (12), 5517- 5581; e) Sorsche, D.; Miehlich, M. E.; Searles, K.; Gouget, G.; Zolnhofer, E. M.; Fortier, S.; Chen, C.-H.; Gau, M.; Carroll, P. J.; Murray, C. B.; Caulton, K. G.; Khusniyarov, M. M.; Meyer, K.; Mindiola, D. J. J. Am. Chem. Soc. 2020, 142, 18, 8147–8159.

20. Hunter, E. P.; Lias, S. G., J. Phys. Chem. Ref. Data, 1998, 27 (3), 413-656.

21. Achord, P.; Fujita, E.; Muckerman, J. T.; Scott, B.; Fortman, G. C.; Temprado, M.; Cai, X.; Captain, B.; Isrow, D.; Weir, J. J.; McDonough, J. E.; Hoff, C. D., Inorg. Chem. 2009, 48 (16), 7891-904. 122

22. Zhang, K.; Gonzalez, A. A.; Mukerjee, S. L.; Chou, S. J.; Hoff, C. D.; Kubat- Martin, K. A.; Barnhart, D.; Kubas, G. J., J. Am. Chem. Soc. 1991, 113 (24), 9170-9176.

23. a) Gonzalez, A. A.; Hoff, C. D., Inorg. Chem. 1989, 28 (23), 4295-4297; b) Hoff, C. D., Prog. Inorg. Chem. 1992, 503-561.

24. Shearer, J.; Jackson, H. L.; Schweitzer, D.; Rittenberg, D. K.; Leavy, T. M.; Kaminsky, W.; Scarrow, R. C.; Kovacs, J. A., J. Am. Chem. Soc. 2002, 124 (38), 11417-11428.

25. Fickes, M. G. Ph.D Thesis, Massachusetts Institute of Technology, 1998.

26. a) Laplaza, C. E.; Cummins, C. C., Science 1995, 268 (5212), 861-863; b) Laplaza, C. E.; Johnson, M. J. A.; Peters, J. C.; Odom, A. L.; Kim, E.; Cummins, C. C.; George, G. N.; Pickering, I. J., J. Am. Chem. Soc. 1996, 118 (36), 8623- 8638.

27. Curley, J. J.; Cook, T. R.; Reece, S. Y.; Müller, P.; Cummins, C. C. J. Am. Chem. Soc. 2008, 130 (29) 9394–9405.

28. Tsai, Y. -C; Cummins, C. C. Inorg. Chim. Acta, 2003, 345, 63-69.

29. See for example: a) Tanabe, Y.; Nishibayashi, Y. Coord. Chem. Rev. 2019, 381, 135–150; b) Sekiguchi, Y.; Arashiba, K.; Tanaka, H.; Eizawa, A.; Nakajima, K.; Yoshizawa, K.; Nishibayashi, Y. Angew. Chem. Int. Ed. 2018, 57 (29), 9064- 9068; c) Kokubo, Y.; Yamamoto, C.; Tsuzuki, K.; Nagai, T.; Katayama, A.; Ohta, T.; Ogura, T.; Wasada-Tsutsui, Y.; Kajita, Y.; Kugimiya, S.; Masuda, H. Inorg. Chem. 2018, 57, 11884−11894; d) Tran, B. L.; Pinter, B.; Nichols, A. J.; Chen, C.-H.; Krzystek, J.; Ozarowski, A.; Telser, J.; Baik, M.-H.; Meyer, K.; Mindiola, D. J. J. Am. Chem. Soc. 2012, 134, 13035-13045; e) Groysman, S.; Villagran, D.; Freedman, D. E.; Nocera, D. G. Chem. Commun. 2011, 47 (37), 10242-10244; f) Janas, Z.; Sobota, P. Coord. Chem. Rev. 2005, 249 (21-22), 2144-2155; g) Clentsmith, G. K. B.; Bates, V. M. E.; Hitchcock, P. B.; Cloke, F. G. N. J. Am. Chem. Soc. 1999, 121 (44), 10444-10445; h) Song, J. I.; Berno, P.; Gambarotta, S. J. Am. Chem. Soc. 1994, 116 (15), 6927-6928; i) Buijink, J. K. F.; Meetsma, A.; Teuben, J. H. Organometallics, 1993, 12 (6), 2004-2005. j) Edema, J. J. H.; Meetsma, A.; Gambarotta, S. J. Am. Chem. Soc. 1989, 111 (17), 6878-6880.

30. a) While data here has been reported earlier for most ligands, it has been repeated and checked in this work and some data revised slightly; b) Cherry, J. P. F.; Johnson, A. R.; Baraldo, L. M.; Tsai, Y. C.; Cummins, C. C.; Kryatov, S. V.; Rybak-Akimova, E. V.; Capps, K. B.; Hoff, C. D.; Haar, C. M.; Nolan, S. P. J. Am. Chem. Soc., 2001, 123, 7271-7286; c) Mendiratta, A.; Cummins, C. C.; Kryatov,O. P.; Rybak-Akimova, E. V.; McDonough, J. E.; Hoff, C. D.; Inorg. Chem. 2003, 42, 8621-8623; d) Stephens, F. H.; Figueroa, J. S.; Cummins, C. C.; 123

Kryatova, O.; Kryatov, S. V.; Rybak-Akimova, E. V.; McDonough, J. E.; Hoff, C. D.; Organometallics, 2004, 23, 3126-3138; e) Mendiatta, A.; Cummins, C. C.; Kryatova, O. P.; Rybak-Akimova, E. V.; McDonough, J. E.; Hoff, C. D.; J. Am. Chem. Soc. 2006, 128, 4881-4891; f) Temprado, M.; McDonough, J. E.; Mendiratta, A.; Tsai, Y.-C.; Fortman, G. C.; Cummins, C. C.; Rybak-Akimova, E. V.; Hoff, C. D. Inorg. Chem. 2008, 47, 9380-9389; g) Germain, M. E.; Temprado, M.; Castonguay, A.; Kryatova, O. P.; Rybak-Akimova, E. V.; Curley, J. J.; Mendiratta, A.; Tsai, Y.-C.; Cummins, C. C.; Prabhakar, R.; McDonough, J. E.; Hoff, C. D. J. Am. Chem. Soc. 2009, 131 (42) 15412–15423; h) Palluccio, T.; Rybak-Akimova, E. V.; Cai, X.; Majumdar, S.; Temprado, M.; Silvia, J. S.; Cozzolino, A. F.; Tofan, D.; Cummins, C. C.; Captain, B.; Hoff, C. D., J. Am. Chem. Soc., 2013, 135, 11357-11372; i) Majumdar, S.; Stauber, J. M.; Palluccio, T. D.; Cai, X.; Velian, A.; Rybak-Akimova, E. V.; Temprado, M.; Captain, B.; Cummins, C. C.; Hoff, C. D., Inorg. Chem. 2014, 53 (20), 11185-11196.

31. Tsai, Y. -C.; Stephens, F. H.; Meyer, K.; Mendiratta, A.; Gheorghiu, M. D.; Cummins, C. C. Organometallics, 2003, 22 (14) 2902-2913;

32. a) Gerlach, C. P.; Arnold, J. J. Chem. Soc. Dalton Trans. 1997, (24), 4795-4806; b) Brussee, E. A. C.; Meetsma, A.; Hessen, B.; Teuben, J. H., Chem. Commun. 2000, (6), 497-498.

33. Ruppa, K. B. P.; Desmangles, N.; Gambarotta, S.; Yap, G.; Rheingold, A. L., Inorg. Chem. 1997, 36 (6), 1194-1197.

34. Bursch, M.; Caldeweyher, E.; Hansen, A.; Neugebauer, H.; Ehlert, S.; Grimme, S. 2019, 52, 258-266; b) Liptrot, D. J.; Power, P. P. Nat. Rev. Chem. 2017, 1, 004, 1- 12.

35. Mottishaw, J. D.; Sun, H. J. Phys. Chem. A, 2013, 117, 7970−7979.

36. a) Schweizer, W. B.; Dunitz, J. D. J. Chem. Theory Comput. 2006, 2, 288–291; b) Bludský, O.; Rubeš, M.; Soldán, P.; Nachtigall, P. J. Chem. Phys. 2008, 128, 114102-1–114102-8; c) Pitonak, M.; Neogrady, P.; Rezac, J.; Jurecka, P.; Urban, M.; Hobza, P. J. Chem. Theory Comput. 2008, 4, 1829–1834; d) Rezac, J.; Hobza, P. J. Chem. Theory Comput. 2008, 4, 1835–1840.

37. Sellmann, D.; Sutter, J., Acc. Chem. Res. 1997, 30 (11), 460-469.

38. a) Langford, C. H.; Gray, H. B. Ligand Substitution Processes. W. A. Benjamin, Inc., 1966, New York; b) Langford, C. H.; Tong, J. P. K., Acc. Chem. Res. 1977, 10 (7), 258-264.

39. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; 124

Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A. Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Keith, T.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, Revision D.01; Gaussian, Inc.: Wallingford CT, 2013.

40. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Petersson, G. A.; Nakatsuji, H.; Li, X.; Caricato, M.; Marenich, A. V.; Bloino, J.; Janesko, B. G.; Gomperts, R.; Mennucci, B.; Hratchian, H. P.; Ortiz, J. V.; Izmaylov, A. F.; Sonnenberg, J. L.; Williams-Young, D.; Ding, F.; Lipparini, F.; Egidi, F.; Goings, J.; Peng, B.; Petrone, A.; Henderson, T.; Ranasinghe, D.; Zakrzewski, V. G.; Gao, J.; Rega, N.; Zheng, G.; Liang, W.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Throssell, K.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M. J.; Heyd, J. J.; Brothers, E. N.; Kudin, K. N.; Staroverov, V. N.; Keith, T. A.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A. P.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Millam, J. M.; Klene, M.; Adamo, C.; Cammi, R.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Farkas, O.; Foresman, J. B.; Fox, D. J. Gaussian 16, Revision B.01; Gaussian, Inc.: Wallingford, CT, 2016.

41. Adamo, C.; Barone, V. J. Chem. Phys. 1999,110, 6158-6170.

42. Grimme, S.; Ehrlich, S.; Goerigk, L. J. Comp. Chem. 2011, 32, 1456-1465.

43. Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297-3305.

44. a) Miertuš, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55, 117-129; b) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999-3094; c) Scalmani, G.; Frisch, M. J. J. Chem. Phys. 2010, 132, 114110.

45. a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652; b) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B, 1988, 37, 785-789.

46. a) Harvey, J. N.; Aschi, M.; Schwarz, H.; Koch, W. Theor. Chem. Acc. 1998, 99, 95–99; b) Smith, K. M.; Poli, R.; Harvey, J. N. New. J. Chem. 2000, 24, 77–80; c) Harvey, J. N.; Poli, R.; Smith, K. M. Coord. Chem. Rev. 2003,361–347 ,238 ; d) Poli, R.; Harvey, J. N. Chem. Soc. Rev. 2003, 32,1–8 ; e) Carreón-Macedo, J. L.; Harvey, J. N. J. Am. Chem. Soc. 2004, 126, 5789–5797. 125

47. a) Neese, F. WIREs Comput. Mol. Sci. 2012, 2, 73–78; b) Neese, F. WIREs Comput. Mol. Sci. 2018, 8, e1327; c) Neese, F.; Wennmohs, F.; Becker, U.; Riplinger, C. J. Chem. Phys. 2020, 152, 224108. 48. Van Stappen, C.; Lehnert, N. Inorg. Chem. 2018, 57, 4252-4269.

49. (a) Hughes, M. N.; Stedman, G. 231. J. Chem. Soc. 1963, 1239-1243. (b) Buchholz, J. R.; Powell, R. E J. Am. Chem. 1963, 85, 509-511. (c) Buchholz, J. R.; Powell, R. E., J. Am. Chem. 1965, 87, 2350-2353. (d) Poskrebyshev, G. A.; Shafirovich, V.; Lymar, S. V., J. Am. Chem. 2004, 126, 891-899.

50. (a) Mendenhall, G. D.; Stewart, L. C.; Scaiano, J. C., J. Am. Chem. 1982, 104, 5109-5114. (b) Ogle, C. A.; Martin, S. W.; Dziobak, M. P.; Urban, M. W.; Mendenhall, G. D., J. Org. Chem. 1983, 48, 3728-3733.

51. Sumida, K.; Rogow, D. L.; Mason, J. A.; McDonald, T. M.; Bloch, E. D.; Herm, Z. R.; Bae, T. H.; Long, J. R. Chem. Rev. 2012, 112, 724-781.

52. Duong, H. A.; Tekavec, T. N.; Arif, A. M.; Louie, J. Chem. Commun. 2004, 112- 113.

53. Tskhovrebov, A. G.; Solari, E.; Wodrich, M. D.; Scopelliti, R.; Severin, K. Angew. Chem. Int. Ed. 2012, 51, 232-234.

54. Arduengo, A. J.; Krafczyk, R.; Schmutzler, R.; Craig, H. A.; Goerlich, J. R.; Marshall, W. J.; Tetrahedron 1999, 55, 14523-14534.

55. Temprado, M.; Majumdar, S.; Cai, X.; Captain, B.; Hoff, C. D. Struct. Chem. 2013, 24, 2059-2068.

56. Lombardi, B. M. P.; Gendy, C.; Gelfand, B. S.; Bernard, G. M.; Wasylishen, R. E.; Tuononen, H. M.; Roesler, R. Angew. Chem. Int. Ed. 2021, 60, 7077-7081.

57. Beck, W.; Engelmann, H.; Smedal, H. S. Z. Anorg. Allg. Chem. 1968, 357, 134- 138.

58. Mayer, T.; Beck, W.; Böttcher, H.-C. Z. Anorg. Allg. Chem. 2011, 637, 345-347.

59. Sato, H. Bull. Chem.Soc. Jpn. 1967, 40, 410-410.

60. (a) Ma, C.; Tian, G.; Zhang, R. J. Organomet. Chem. 2006, 691, 2014-2022. (b) Wakamatsu, K.; Orita, A.; Otera, J. Organometallics 2010, 29, 1290-1295.

61. Berto, T. C.; Xu, N.; Lee, S. R.; McNeil, A. J.; Alp, E. E.; Zhao, J.; Richter-Addo, G. B.; Lehnert, N. Inorg. Chem. 2014, 53, 6398-6414. 126

62. Capps, K. B.; Bauer, A.; Sukcharoenphon, K.; Hoff, C. D. Inorg. Chem. 1999, 38, 6206-6211.

63. Plasseraud, L. CO2 Inorganics 2020, 8, 31.

64. Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds, Part A: Theory and Applications in Inorganic Chemistry, 6 ed, Wiley-New York, 2009.

65. Thermochemical data are derived from https://webbook.nist.gov/chemistry/ and related online sources.

66. Temprado, M.; Davis, J. V.; Guio, O.; Hoff, C. D. unpublished results

67. Severin, K. Chem. Soc. Rev. 2015, 44, 6375-6386.

68. Wright, A. M.; Hayton, T. W., Inorg. Chem. 2015, 54, 9330-9341.

69. Weller, J. M.; Chan, C. K., ACS Appl. Energy Mater. 2020, 3, 6466-6475.

70. Mebel, A. M.; Morokuma, K.; Lin, M. C.; Melius, C. F. J. Phys. Chem. 1995, 99, 1900-1908.

71. Zhang, K.; Thynell, S. T., J. Phys. Chem. A 2017, 121, 4505-4516.

72. Davis, J. V.; Guio, O.; Gamage, M. M.; Captain, B.; Temprado, M.; Hoff, C. D. work in progress.

73. Oza, T. M.; Thaker, R. H., J Am Chem Soc 1955, 77 (19), 4976-4980.

74. Apex2 Version 2.2-0 SAINT+ Version 7.46A. Bruker Anal. X-ray Syst. Inc.: Madison, Wisconsin, USA, 2007.

75. Sheldrick, G. M., Acta Crystallogr. Sect. 2008, (64), 112–122.

76. Sheldrick, G. M., SHELXTL Version 6.1. Bruker Anal. X-ray Syst. Inc.: Madison, Wisconsin, USA, 2000.

77. Ma, C.; Tiana, G.; Zhang, R. J. Organomet. Chem. 2006, 691, 2014–2022.

78. I. Wender , P. Pino, Organic Syntheses Via Metal Carbonyls , Volumes 1 & 2, Ed., Wiley–Interscience, New York, NY, 1977.

79. L. Damas, R.M.B. Carrilho, S.C.C. Nunes, A .A .C.C. Pais, L. Kollar, M. Pineiro, M.M. Pereira, R. Soc. Open Sci. 5 (2018) 181140. 127

80. (a) T. Mitsudo, Y. Hori, Y. Yamakawa, Y. Watanabe, Tetrahedron Lett. 28 (1987) 4 417–4 418. (b) S. Yoshida, K. Fukui, S. Kikuchi, T. Yamada, Chem. Lett. 38 (2009) 786–787. (c) V.A. Peshkov, O.P. Pereshivko, A. A. Nechaev, A. A. Peshkov, E.V. Van der Eycken, Chem. Soc. Rev. 47 (2018) 3861–3898.

81. (a) S. Hase , Y. Kayaki , T. Ikariya , Organometallics 32 (2013) 5285–5288. (b) S. Hase , Y. Kayaki , T. Ikariya , ACS Catal. 5 (2015) 5135–5140.

82. (a) A . Collado, A . Gomez-Suarez, A .R. Martin, A .M.Z. Slawin, S.P. Nolan, Chem. Comm. 49 (2013) 5541–5543. (b) O. Santoro, A. Collado, A.M.Z. Slawin, S.P. Nolan, C.S.J. Cazin, Chem. Comm. 49 (2013) 10483–10485. (c) C.M. Zinser, F. Nahra, M. Brill, R.E. Meadows, D.B. Cordes, A.M.Z. Slawin, S.P. Nolan, C.S.J. Cazin, Chem. Comm. 53 (2017) 7990–7993. (d) N.V. Tzouras, F. Nahra, L. Falivene, L. Cavallo, M. Saab, K. Van Hecke, A. Collado, C.J. Collett, A.D. Smith, C.S.J. Cazin, S.P. Nolan, Chem. Eur. J. 26 (2020) 4515–4519. (e) S.G. Guillet, V.A. Voloshkin, M. Saab, M. Belis, K. Van Hecke, F. Nahra, S.P. Nolan, Chem. Comm. 56 (2020) 5953–5956. (f) T. Scattolin, N.V. Tzouras, L. Falivene, L. Cavallo, S.P. Nolan, Dalton Trans 49 (2020) 9694–9700.

83. (a) P.V. Kortunov, M. Siskin, M. Paccagnini, H. Thomann, Energy Fuels 30 (2016) 1223–1236. (b) P.V. Kortunov, M. Siskin, L.S. Baugh, D.C. Calabro, Energy Fuels 29 (2015) 5940–5966.

84. D.B. Dell’Amico, F. Calderazzo, L. Labella, F. Marchetti, G. Pampaloni, Chem. Rev. 103 (2003) 3857–3897.

85. M.W. Johnson, S.L. Shevick, F.D. Toste, R.G. Bergman, Chem. Sci. 4 (2013) 1023–1027.

86. (a) A. Zhdanko, M.E. Maier, Chem. Eur. J. 20 (2014) 1918–1930. (b) A. Zhdanko, M.E. Maier, Chem. Eur. J. 19 (2013) 3932–3942.

87. E. Wilhelm, R. Battino, Chem. Rev. 73 (1973) 1–9.

88. R.M. Yuan, Z.Y. Lin, ACS Catal. 5 (2015) 2866–2872.

89. J.L. Gohres, A.T. Marin, J. Lu, C.L. Liotta, C.A. Eckert, Ind. Eng. Chem. Res. 48 (2009) 1302–1306.

90. (a) D.S. Laitar , P. Muller , T.G. Gray , P. Sadighi , Organometallics 24 (2005) 4503–4505. (b) B.K. Tate, A.J. Jordan, J. Bacsa, J.P. Sadighi, Organometallics 26 (2017) 964–974.

91. Ostwald, W. Jan. 09, 1902, “Improvements in the Manufacture of Nitric Acid and Nitrogen Oxides”, Patent GB190200698A. 128

92. (a) “BASF-The Right Choice for Nitric Acid Plant” http://www.catalysts.basf.com/p02/USWeb Internet/catalysts/en_GB/function/conversions:/publish/content/microsites/catalys ts/prods-inds/process-catalysts/BF-9834_US_N2O_and_DeNOX_Technote.pdf. (b) Gao, F.; Kwak, J. H.; Szanyi, J.; Peden, C. H. F., Top. Catal. 2013, 56 (15-17), 1441-1459.

93. Ravishankara, A. R.; Daniel, J. S.; Portmann, R. W. Science 2009, 326, 123-125.

94. Thermochemical data are derived from https://webbook.nist.gov/chemistry/ and related online sources.

95. Lewis, G. N.; Randall, M. Thermodynamics-The Free Energy of Chemical Substances. McGraw Hill, New York 1923.

96. Blomberg, M. R. A. Inorg. Chem. 2020, 59 (16), 11542-11553.

97. (a) See ref. 73 (b) Severin, K. Chem. Soc. Rev. 2015, 44 (17), 6375-6386.

98. Liu, H. Z.; Klein, W.; Sani, A.; Jansen, M. Phys. Chem. Chem. Phys. 2004, 6 (5), 881-883.

99. (a) Bilke, M.; Losch, P.; Vozniuk, O.; Bodach, A.; Schuth, F. J. Am. Chem. Soc. 2019, 141 (28), 11212-11218. (b) Porcheddu, A.; Colacino, E.; De Luca, L.; Delogu, F. ACS Catal. 2020, 10 (15), 8344-8394.

100. For example, ball milling of SnO2 with Teflon has been shown to reported to form Sn-F compounds as detected by NMR: Senna, M.; Turianicova, E.; Zorkovska, A.; Makreski, P.; Kanuchova, M.; Scholz, G.; Balaz, M.; Balaz, P.; Sepelak, V.; Hahn, H. J. Nanopart. Res. 2015, 17 (9).

101. Caution is indicated in all ball milling experiments, particularly with a potentially powerful oxidant such as nitrous oxide. Care must be taken that no readily combustible material is available to nitrous oxide under these conditions. Examination of the cell itself following each experiment is also called for, since oxidation or nitridation may also occur, particularly under active milling conditions at higher temperatures.

102. This is the net reaction, it is possible that N2O converts to O2 or a peroxide, but the ultimate source of O2 comes from N2O itself. Work under mixed N2O/O2 atmosphere is in progress.

103. Miller, T. M.; Grassian, V. H. Catal. Lett. 1997, 46 (3-4), 213-221.

104. (a) Meizyte, G.; Green, A. E.; Gentleman, A. S.; Schaller, S.; Schollkopf, W.; Fielicke, A.; Mackenzie, S. R. Phys. Chem. Chem. Phys. 2020, 22 (33), 18606- 129

18613. (b) Kokalj, A.; Kobal, I.; Matsushima, T. J. Phys. Chem. B 2003, 107 (12), 2741-2747. (c) Hamilton, S. M.; Hopkins, W. S.; Harding, D. J.; Walsh, T. R.; Haertelt, M.; Kerpal, C.; Gruene, P.; Meijer, G.; Fielicke, A.; Mackenzie, S. R. J. Phys. Chem. A 2011, 115 (12), 2489-2497.

105. (a) Keresztury, G.; Incze, M.; Soti, F.; Imre, L.. Spectrochim Acta A 1980, 36 (11), 1007-1008. (b) Hadjiivanov, K. I.; Panayotov, D. A.; Mihaylov, M. Y.; Ivanova, E. Z.; Chakarova, K. K.; Andonova, S. M.; Drenchev, N. L. Chem. Rev. 2021, 121 (3), 1286-1424.

106. https://www.sciencedirect.com/topics/chemistry/hyponitrous-acid

107. Blackwelder, P.; Temprado, M.; Davis, J. V.; Guio, O.; Captain, B.; Hoff, C. D. work in progress.

108. Andrae, D.; Häußermann, U.; Dolg, M.; Stoll, H.; Preuß, H. Theor. Chim. Acta 1990, 77, 123-141.