Reactions and Properties of Molybdenum Bis(Dithiolene) Complexes Based on Bis(Trifluoromethyl) Dithiolene and Labile Ligands

Reactions and Properties of Molybdenum Bis(dithiolene) Complexes Based on Bis(trifluoromethyl) Dithiolene and Labile Ligands

Neilson Nguyen

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy (Ph.D.)

Graduate Department of Chemistry University of Toronto

Reactions and Properties of Molybdenum Bis(dithiolene) Complexes Based on Bis(trifluoromethyl) Dithiolene and Labile Ligands

Neilson Nguyen

Doctor of Philosophy (Ph.D.)

Graduate Department of Chemistry University of Toronto

2014

Abstract

Metal dithiolenes [M(S2C2R2)n] have been studied for decades since their discovery due to their interesting spectroscopic, redox, biological and catalytic properties.

S2C2(CF3)2-containing complexes have been studied but there is still much left unexplored due to the difficulty of obtaining the precursors and synthesizing the ligand. We present a synthetic method that uses easily obtained precursors and we built a relatively inexpensive apparatus to safely isomerize and react the gaseous intermediates.

Molybdenum disulfide is used extensively in the petrochemical industry as catalyst for hydrodesulfurization of petroleum resources. We use dithiolenes to create homogenous structural model complexes of the proposed active sites of the molybdenum disulfide catalyst. We also experimentally determine competitive binding affinities for dihydrothiophene and tetrahydrothiophene, and explore some basic catalytic properties.

Dithiolenes undergo reactions with alkenes to form new bonds. We present a new dithiolene reaction where it is attacked by a nucleophile (triphenylphosphine) to create a zwitterionic ligand as well as open an active site on a previously coordinatively saturated molybdenum tris(dithiolene). This technique is used to create a structural model complex for DMSO reductase and produce a pre-catalyst for that same reaction.

After the determination that the actual catalyst for previously observed activity was a molybdenum bis-dithiolene complex, kinetic determination experiments were performed to elucidate the mechanism. Kinetic investigations suggest the binding of the phosphine oxide created from the use of triphenylphosphine as the oxygen acceptor competes with DMSO in the binding to the active site of the molybdenum bis(dithiolene). Additionally, the removal of oxygen from DMSO using the catalyst appears to involve a polar transition state.

iii

Acknowledgments

Big thanks to my supervisor, Professor Ulrich Fekl for encouragement, motivation, support and understanding throughout my work.

Special thanks to crystallographer Dr. Alan J. Lough (St. George campus). Receiving crystal structures were some of the best days during my research.

Thanks to my parents, Dang Nguyen and Mai Ly, who supported my decisions and financed my education.

Contributions

Neilson Nguyen – Synthesis and NMR characterization of all compounds, data fitting (Chapter 5)

Dr. Alan J. Lough – X-ray crystallography of all obtained structures

Professor Ulrich Fekl – Computational modeling (Chapter 3)

Dr. Antonio De Crisci – Computational modeling (Chapter 3)

Dave Armstrong – Computational modeling (Chapter 4)

Table of Contents

Acknowledgments...... iv

Contributions...... v

Table of Contents...... vi

List of Tables ...... ix

List of Figures...... x

List of Equations...... xv

Chapter 1: Introduction...... 1

1.1 Synthesis of Dithiolenes ...... 3

1.2 Applications and Reactions of dithiolenes...... 5

1.3 Scope of Thesis...... 9

Chapter 2: Convenient Lab-scale Synthesis of Hexafluorobutyne and Bis(trifluoromethyl)dithiete (tfd) ...... 12

2.1 Abstract...... 12

2.2 Introduction...... 12

2.2.1 Synthesis of hexafluorobutyne...... 13

2.3 Synthesis of hexafluorocyclobutene ...... 23

2.4 Synthesis of hexafluorobutyne...... 24

2.5 Synthesis of bis(trifluoromethyl) dithiete S2C2(CF3)2 ...... 28

2.6 Results and Discussion...... 30

References...... 32

Chapter 3: Molybdenum Dithiolene Complexes as Structural Models for the Active Sites of Molybdenum(IV) Sulfide Hydrodesulfurization Catalysts...... 34

3.1 Abstract...... 34

3.2 Introduction...... 34

3.3 Experimental...... 38 vi

3.3.1 Synthesis of Mo(tfd)2(tht)2...... 39

3.3.2 Synthesis of 2,5-Dihydrothiophene ...... 39

3.3.3 Synthesis of Mo(tfd)2(dht)2...... 39

3.3.4 Attempted synthesis of Mo(tfd)2(thiophene)2 ...... 40

3.3.5 Thermal decomposition of Mo(tfd)2(dht)2 ...... 40

3.3.6 Non-catalytic dehydrogenation of dht with Mo(tfd)2(dht)2 ...... 40

3.3.7 Hydrogen reactivity...... 41

3.3.8 Isomerization of 1,4-cyclohexadiene...... 41

3.3.9 Reaction with carbon monoxide to produce Mo(tfd)2(dht)(CO) ...... 41

3.3.10 Determination of Keq (preferred binding of tht over dht) ...... 42

3.4 Results and Discussion...... 42

3.5 Reactivity ...... 53

3.6 Conclusions...... 56

3.7 Appendix...... 58

References...... 64

Chapter 4: A Structural Model for DMSO Reductase from Covalent Addition of Phosphine to Dithiolene in a Molybdenum Tris(dithiolene)...... 68

4.1 Abstract...... 68

4.2 Introduction...... 68

4.3 Experimental...... 69

4.3.1 Synthesis of Compound 3 - Mo(tfd)2(SC6H4SPPh3)(PPh3) * 0.5 C6H6 ...... 69

4.3.2 Synthesis of Mo(tfd)2(SC6H4SPPh3)(PPh3) * CHCl3...... 70

4.3.3 Observation of Compound 2 - Mo(tfd)2(SC6H4SPPh3) ...... 72

4.3.4 DFT-Optimized Structure of 2 and its Analogues ...... 72

4.3.5 Synthesis of Compound 3’ - Mo(tfd)2(SC6H4SP(p-tolyl)3)(P(p-tolyl)3)...... 73

4.3.6 Equilibrium constant determination of 3’ by dilution...... 73

vii

4.3.7 Equilibrium constant determination for formation of 3 by UV-Vis titration...... 75

4.4 Results and Discussion...... 78

4.5 Appendix...... 83

Chapter 5: Mechanistic elucidation of the oxotransferase activity for Mo(tfd)2(tht)2...... 93

5.1 Abstract...... 93

5.2 Introduction...... 93

5.3 Experimental...... 96

5.3.1 Reaction of dimethyl sulfoxide with 3...... 100

5.3.2 Catalytic reduction of dimethyl sulfoxide with 3 and triphenylphosphine...... 101

5.3.3 Catalytic reduction of dimethyl sulfoxide with 3’ and P(p-tolyl)3 over time. .... 101

5.3.4 Catalytic reduction of dimethyl sulfoxide with Mo(tfd)2(DMS)2 and P(p- tolyl)3 over time...... 102

5.3.5 Reaction of Mo(tfd)2(tht)2 with excess DMSO...... 103

2- 5.3.6 Reaction of MoO(tfd)2 and an oxidant with DMSO and phosphine ...... 103

5.4 Results and Discussion...... 104

5.5 Conclusion ...... 119

5.6 References...... 121

Chapter 6: Summary and Conclusion...... 122

6.1 Chapter 2...... 122

6.2 Chapter 3...... 122

6.3 Chapter 4...... 123

6.4 Chapter 5...... 123

6.5 Final Remarks ...... 123

viii

List of Tables

Table 2.1: Cost analysis of various procedures to synthesize hexafluorobutyne...... 22

Table 3.1: Crystal data and structure refinement for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2...... 45

Table 3.2: Selected structural data for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2...... 47

Table 3.3: (Continued) Selected structural data for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2...... 48

Table 3.4: A comparative sampling of bite angles (°) in bis-dithiolenes with various ligands (mnt

= S2C2(CN)2)...... 51

Table 4.1: Crystal data and structure refinement for compound 3 as benzene solvate and as chloroform solvate...... 71

Table 4.2: DFT optimized coordinates for 2...... 90

Table 5.1: Model fitting and simulation results. Kinetic models were based on schemes 5.1-5.3. Possible simplifications were tested by setting various parameters to zero and fitting the remaining ones. (n is the number of data points in the model, p is number of parameters fitted, AICc is the Akaike Information Criterion for small sets (lower is better))...... 113

Table 5.2: Model fitting and simulation results on data taken from reactions performed in C6D6. Kinetic models were based on scheme 5.4. Possible simplifications were tested by setting various parameters to zero and fitting the remaining ones. Only the final four models were fitted to verify consistency with the observations in CDCl3...... 115

List of Figures

Figure 1.1: A prototypical dithiolene complex ...... 1

Figure 1.2: Relationships and nomenclature of dithiolene precursors...... 2

Figure 1.3: Molybdopterin Cofactor...... 6

Figure 1.4: X-ray crystal structure of the active site in nitrate reductase from Cupriavidus necator. The two ligands in the bottom left and bottom right are the molybdopterin cofactor. The ligand in the upper right is a cysteine residue on the surrounding protein. The protein backbone has been omitted for clarity...... 7

Figure 2.1: Hexafluorocyclobutene isomerization apparatus ...... 26

Figure 2.2: Hexafluorocyclobutene isomerization apparatus (assembled) ...... 27

Figure 2.3: ‘tfd’ synthesis apparatus...... 28

Figure 2.4: 'tfd' synthesis apparatus (assembled)...... 29

Figure 3.1: Views of exposed molybdenum sites in hexagonal MoS2; a) standard orientation, where the hexagonal layers are oriented horizontally; b) rotated to highlight the geometry of the edge sites. The picture was generated with ORTEP, using MoS2 coordinates from the literature...... 36

Figure 3.2: Top: Structure of Mo(tfd)2(dht)2. Bottom: Structure of Mo(tfd)2(tht)2. Non-hydrogen atoms are displayed using 30 % thermal ellipsoids. Hydrogen atom positions are calculated..... 44

Figure 3.3: Bond angles used for analysis; a) example for SMStrans angle. b) example for SSS (nonbonding) angle...... 49

Figure 3.4: View of the sulfur environment of Mo(tfd)2(dht)2, using van-der-Waals radii for sulfur atoms; A: view perpendicular to S1-S3-S6 “close-packed” plane; B: view onto S5-S6 edge...... 52

Figure 3.5: Determination of the equilibrium constant for the binding of tht versus dht at 22 °C in

CDCl3, according to the equation: Mo-dht + tht ⇌ Mo-tht + dht. The model, predicting a linear dependence of the ratio [Mo-tht]/[Mo-dht] versus the ratio of free thioethers ([tht]/[dht]) fits the data very well (R2 = 0.9991). No cooperativity is observed, and the two labile sites behave as independent in this process. The slope yields Keq = 6.5 ± 0.5 (tht binds more strongly, by a factor of close to seven, than dht). See Experimental Section for details...... 54

1 Figure 3.6: H NMR spectrum of Mo(tfd)2(tht)2 (400 MHz, CDCl3) δ 2.28 (m, 8H, tht), 3.54 (m, 8H, tht)...... 58

19 Figure 3.7: F NMR spectrum of Mo(tfd)2(tht)2 (376 MHz, CDCl3) δ –54.9 (s, 12F, (CF3)2 x 2) ...... 59

13 1 Figure 3.8: C{ H} NMR spectrum of Mo(tfd)2(tht)2 (100 MHz, CDCl3) δ 30.109 (s, Cβ, tht), δ

42.31 (s, Cα, tht)...... 60

Figure 3.9: 1H NMR spectrum of Mo(tfd)2(dht)2 (400 MHz, CDCl3) δ 4.35 (m, 8H, (CH2) x 4), 5.99 (m, 4H, (CH) x 4)...... 61

19 Figure 3.10: F NMR spectrum of Mo(tfd)2(dht)2 (376 MHz, CDCl3) δ –54.96 (s, 12F, (CF3)2 x 2) ...... 62

13 Figure 3.11: C NMR spectrum of Mo(tfd)2(dht)2 (100 MHz, CDCl3) δ 48.78 (s, CH2), δ 126.62 (s, CH)...... 63

Figure 4.1: DFT Optimized Structure of Mo(tfd)2(SC6H4SPPh3) ...... 72

1 Figure 4.2: H NMR spectra of 3’ in CD2Cl2 for various initial concentrations at 29 °C. The region shown exclusively encompasses the signals for the methyl group on all P(p-tolyl)3 1 containing species. H NMR (400 MHz, CD2Cl2) δ 2.34 (s, “c”, free P(p-tolyl)3), 2.45 (s, “a”, 2’),

2.42 (s, “b”, bdt-bound P(p-tolyl)3 of 3’), 2.23 (s, “d”, Mo-bound P(p-tolyl)3 of 3’)...... 74

The slope yields the dissociation constant Keq = 6(5) x10-5 M. The slight deviation from linearity is likely due to a compounding of errors due to each dilution being used to make the subsequent dilution...... 75

Figure 4.4: UV-vis spectra (solvent = CH2Cl2) following the titration of Mo(tfd)2(bdt) (0.106 o mM initially) with PPh3 (0 → 5.69 equiv. with 12.3 mM) (T ≈ 29 C)...... 76

Figure 4.5: Determined concentration of components during titration of Mo(tfd)2(bdt) with PPh3. Region past 4 equiv. omitted for clarity...... 77

Figure 4.6: Deconvoluted spectra for the absorbing components fitted to the multiple equilibrium model created using (1) and (2)...... 78

Figure 4.7: Molecular structure of 3, from X-ray crystallography on the benzene solvate (30 % probability ellipsoids). H and F atoms omitted, for clarity. Selected distances and angles: Mo1- S1, 2.358(3); Mo1-S2, 2.364(2); Mo1-S3, 2.335(2); Mo1-S4, 2.369(3); Mo1-S5, 2.416(2); Mo1- P1, 2.564(2); S5-Mo1-P1, 76.24(7); S1-Mo1-S2, 80.72(8); S3-Mo1-S4, 80.74(9)...... 79

Figure 4.8: Representative structure of the active site for the DMSO reductase family of enzymes. (X = S or O, serine residue may be replaced with cysteine)...... 80

1 Figure 4.9: H NMR spectrum of 3 (400 MHz, CD2Cl2) δ = 7-7.8 (m, 34H, (CH) x 34)...... 83

19 Figure 4.10: F NMR spectrum of 3 in equilibrium with 2 in CD2Cl2. (376 MHz, CD2Cl2) δ – 54.94 (s, 3), –55.09 (s, 2) ...... 84

31 Figure 4.11: P NMR of 3 in equilibrium with 2. (162 MHz, CD2Cl2) δ 43.54 (s, 3), 54.14 (s, 3) δ 47.58 (s, 2)...... 85

1 Figure 4.12: H NMR spectrum of 3’ in equilibrium with 2’. (400 MHz, CD2Cl2) δ 2.35 (s, free

P(p-tolyl)3), 2.46 (s, 2’), 2.43 (s, bdt-bound P(p-tolyl)3 of 3’), 2.23 (s, Mo-bound P(p-tolyl)3 of 3’), 7-7.7 (m, aromatic protons of 3’ and 2’)...... 86

19 Figure 4.13: F NMR spectrum of 3’ in equilibrium with 2’ in CD2Cl2. (376 MHz, CD2Cl2) δ – 54.87 (s, 3’), –55.07 (s, 2’) ...... 87

31 Figure 4.14: P NMR of spectrum 3’ (162 MHz, CD2Cl2) δ 42.70 (s, 3’), 54.63 (s, 3’)...... 88 xii

Figure 4.15: Molecular structure of 3, from X-ray crystallography on the chloroform solvate. Thermal ellipsoids correspond to 30 % probability. Hydrogen and fluorine atoms (on the tfd ligands) are omitted, for clarity. Selected distances and angles: Mo1-S1, 2.333(3); Mo1-S2, 2.356(3); Mo1-S3, 2.371(3); Mo1-S4, 2.342(3); Mo1-S5, 2.407(3); Mo1-P1, 2.560(3); S5-Mo1- P1, 75.19(9); S1-Mo1-S2, 81.43(10); S3-Mo1-S4, 81.36(10)...... 89

Figure 5.1: Typical NMR spectrum of a kinetic experiment in CDCl3. A is assigned as P(p- tolyl)3, B is assigned as OP(p-tolyl)3, C is assigned as DMSO, D is assigned as DMS. Not shown is the aromatic region that contains P(p-tolyl)3 and OP(p-tolyl)3...... 97

Figure 5.2: Typical results from the catalyzed reaction of DMSO and P(p-tolyl)3 using o Mo(tfd)2(tht)2 at a concentration of 0.328 mM at 25 C...... 98

Figure 5.3: Typical NMR spectrum of a kinetic experiment in C6D6. A is assigned as P(p-tolyl)3,

B is assigned as OP(p-tolyl)3, C is assigned as DMSO, D is assigned as DMS. Not shown is the aromatic region that contains P(p-tolyl)3 and OP(p-tolyl)3...... 100

Figure 5.4: Catalytic reduction of DMSO with 3’ and P(p-tolyl)3 at 25 °C...... 102

Figure 5.5: Catalytic reduction of DMSO with P(p-tolyl)3 using Mo(tfd)2(DMS)2 at 25°C. The increased concentration of DMS relative to OP(p-tolyl)3 for the stoichiometric reaction is due to the DMS already present in the Mo(tfd)2(DMS)2 catalyst...... 103

Figure 5.6: Simulated reaction progression based on models 1-4. Models 1 & 2 almost perfectly overlap and so are difficult to distinguish. This is also true for models 3 & 4...... 107

Figure 5.7: Simulated reaction progression based on models 5-8. Models 5 & 6 overlap significantly. This is also true for models 7 & 8. The actual values (the black symbols) cannot be seen as they are perfectly overlapped by model 8...... 110

Figure 5.8: Simulated reaction progression based on models 5-8 based on kinetics observed in benzene-d6. The other reagents and products could not be accurately measured due to significant overlap in their NMR signals. For clarity only 11 data points are shown between 0 and 15000 seconds and 17 between 15000 and 83000 seconds. The actual values (the black symbols) cannot be seen as they are perfectly overlapped by models 7 and 8...... 116

xiii

Figure 5.9: Determination of reaction order with respect to DMSO by method of initial rates and plotting the logarithm of initial rate with respect to DMS versus concentration of DMSO...... 118

xiv

List of Equations eq. 1.1...... 3 eq. 1.2...... 4 eq. 1.3...... 4 eq. 1.4...... 4 eq. 1.5...... 5 eq. 1.6...... 5 eq. 1.7...... 8 eq. 1.8...... 8 eq. 2.1...... 12 eq. 2.2...... 13 eq. 2.3...... 14 eq. 2.4...... 14 eq. 2.5...... 15 eq. 2.6...... 15 eq. 2.7...... 15 eq. 2.8...... 16 eq. 2.9...... 16 eq. 2.10...... 17 eq. 2.11...... 17

xv eq. 2.12...... 17 eq. 2.13...... 18 eq. 2.14...... 18 eq. 2.15...... 19

xvi 1

Chapter 1: Introduction

Dithiolene complexes share the common feature of containing one or more ligands that are coordinated to the metal through two sulfur atoms that themselves are bonded to two sp2 carbon atoms.

R S

S R

Figure 1.1: A prototypical dithiolene complex

These bidentate dithiolene ligands behave as 4-electron σ-donors through the lone pairs on the sulfur atoms but can also behave as π-donors. The π-orbitals of the sulfur atoms are perpendicular to the plane of the dithiolene ligand and are oriented to effectively donate into the metal’s d-orbitals. Since the highest occupied molecular orbital on the ligand is C-S anti-bonding and C-C bonding, donating this electron density into the metal gives the C-S bonds some double bond character while removing some double bond character from the C-C bonds. This creates a planar metallocycle with a highly delocalized π-system.1 A consequence of donating into the metal’s d-orbital is that the ligand can effectively compete with metal bound oxo (M=O) or sulfo (M=S) ligands for the metal’s empty π-orbitals. This weakens the bonding of those other ligands making them more labile and thus facilitating oxotransfer and hydrodesulfurization reactions.

R S R S R S +e- +e-

-e- -e- R S R S R S

1,2-dithioketone radical anion 1,2-enedithiolate

-2H+ +2H+

R R SH S

S R R SH

1,2-dithiete 1,2-enedithiol

Figure 1.2: Relationships and nomenclature of dithiolene precursors

The dithiolene ligand can exist in different redox forms as shown in Figure 1.2. Ligands such as this are referred to as “non-innocent” as they can contribute directly to the redox properties of the metal complexes they are a part of.2 Now because the ligand contributes to the overall redox state of the complex it can be difficult to assign a formal oxidation state to the metal. But as Figure 1.2 shows, the oxidation state of the ligand changes what bonding is present. As the ligand becomes more oxidized, as in the case of dithioketones, the C-C backbone acquires more single bond character while the C-S bonds acquire more double bond character. The opposite is true as the ligand becomes more reduced, as in the case of enedithiolates. Knowing this, X-ray crystallographic data of dithiolene complexes can be used to reveal the bond lengths of the ligands and help assign oxidation states to them and thus assign oxidation states to the metal.3 Additionally, X-ray absorption spectroscopy of the metal can provide more direct evidence of its oxidation state. Qualitative predictions can be made when comparing different dithiolenes. For example, electron-withdrawing dithiolene ligands tend to be reduced, having more electron density there, than electron-donating dithiolenes. Measurements using EPR spectroscopy on

3 paramagnetic tris(dithiolenes) have demonstrate a continuous shift in spin density from the metal to the ligand as the ligands became more electron withdrawing.4

In terms of geometry, homoleptic bis(dithiolene) complexes are generally square planar while tris(dithiolene) complexes prefer a trigonal prismatic structure. This preference for trigonal prismatic geometry over the much more common octahedral geometry is a distinguishing characteristic of tris(dithiolenes). Indeed, the first molecular trigonal prismatic complex was a rhenium tris(dithiolene).5 Trigonal prismatic molecules are of interest, as their structure resembles the internal structure of some solid-state compounds like MoS2. Thus, model complexes with trigonal prismatic structure are highly sought after as they mimic the essential geometry of such solid-state catalysts like MoS2.

1.1 Synthesis of Dithiolenes

The synthesis of dithiolenes is a vast and very diverse field. A thorough treatment of all the synthetic methods available is far beyond the scope of this thesis so only those methods directly applicable to it shall be discussed. For the interested reader a far more thorough treatment can be found in the works of T. B. Rauchfuss6 and R. H. Holm et. al.7

One of the simplest means of synthesizing dithiolenes is direct reaction of the metal with a 8 dithiete as in Ni(tfd)2 (eq. 1.1) . (tfd= S2C2(CF3)2, “trifluoromethyl dithiete” when not bound to a metal, “trifluoromethyl dithiolene” when bound).

F3C F3C CF3 S S S toluene 2 + Ni Ni eq. 1.1 S r.t. 7d S S F3C CF3 F3C tfd Ni(tfd) 2

9 Neutral metal containing species can also be used (eq. 1.2) as in the case of Mo(tfd)3.

CF3

F3C F3C

S S S + 6 CO eq. 1.2 3 + Mo(CO)6 reflux Mo S S S S F3C S CF3 F3C F3C CF3

Dithietes may also be reacted with metal compounds to yield dithiolenes as in the case of 2- 2- 9 Mo(tfd)2O being synthesized (eq. 1.3) through the direct reaction of MoO2S2 with tfd.

O 2- S 2- tfd Mo S eq. 1.3 Mo S S O F3C S CF3 S O F3C CF3

While the use of dithietes is quite versatile, one drawback is that the appropriate dithiete itself may not exist. Bis(nitrile) dithiete (S2C2(CN)2) is unstable and cannot be isolated. But the 2- 9 reduced dithiolate, maleonitrile dithiolate [S2C2(CN)2] can be synthesized quite easily. If the dithiolate is available then dithiolenes can also be synthesized by the action of reduced 2- 2- dithiolates (e.g. [S2C2(CN)2] , [S2C6H4] ) on metal salts. For example, Mo(tfd)2(bdt) is 2- 10 synthesized (eq. 1.4) by the action of S2C6H6 (bdt-H2) on Mo(tfd)2O .

O 2- 2- SH S S Mo S eq. 1.4 S SH Mo S F3C S CF3 S S S S -H2O F3C CF3 CF F3C CF 3 3 F3C

The resulting dithiolene can then be oxidized to yield a stable neutral complex.

Interestingly, some dithiolenes are known to bind alkenes.11,10 But rather than binding at the metal as is the case with most well known organometallic complexes, the alkenes are instead bonded to the ligands (eq. 1.5).

S S S S eq. 1.5 Mo Mo S S S S S S S S

F C CF3 F C CF3 3 CF3 3 CF F3C 3 F3C

An important feature is the newly formed dihydrobenzodithiin is more labile than the original dithiete ligand. Alkene binding can then be used to conveniently exchange a dithiete with another ligand (eq. 1.6).10

NC CN

S S Na2S2C2(CN)2 S S

Mo Mo eq. 1.6 S S S S S S S S

CF3 CF3 F3C F3C

This greatly expands the number of ligands that can be used and the combinations of complexes that can be produced. For example, the reaction in eq. 1.6 produces the first tris(dithiolene) complex that contains three different dithiolenes.

1.2 Applications and Reactions of dithiolenes

The earliest work in dithiolenes began in the 1930s with the synthesis of toluene-3,4-dithiol and 1-chlorobenzene-3,4-dithiol. These dithiolenes reacted with various metals such as cadmium, mercury, tin and zinc to form intensely colored complexes that served as the basis for colorimetric analytical determination.12

Dithiolenes are also found in nature and serve in the active site of a specialized group of molybdenum and tungsten containing metalloenzymes. Known examples are the DMSO reductase family enzymes that include formate dehydrogenase, nitrate reductase, DMSO reductase, and sulfite oxidase. Common to all of these enzymes is the molybdopterin cofactor (Figure 1.3).13,14 Xanthine and sulfite oxidase family enzymes have one molybdopterin ligand in the active site while DMSO reductase family enzymes have two. A representative X-ray structure of one of these active sites can be seen in Figure 1.4. These enzymes catalyze the transfer of oxygen on various substrates including xanthine, sulfoxides, sulfites, nitrites, nitrogen oxides, formats and aldehydes. Protons serve as the oxygen sink or water can serve as the oxygen source depending on the nature of the enzyme. As electrons are also produced or used in either case, the non-innocent nature of the dithiolene ligand could facilitate these reactions.

2+ O S Mo

H S N N

H2N N N O H H

OPO3

Figure 1.3: Molybdopterin Cofactor

Figure 1.4: X-ray crystal structure of the active site in nitrate reductase from Cupriavidus necator.15 The two ligands in the bottom left and bottom right are the molybdopterin cofactor. The ligand in the upper right is a cysteine residue on the surrounding protein. The protein backbone has been omitted for clarity.

An ongoing area of research is the study of dithiolene-based metalloenzymes through the creation and/or simulation of model complexes.16,17,18 Models are required because molybdopterin itself is unstable when removed from the protein, losing the metal and irreversibly losing functionality due to oxidation.3

In addition to modeling and replicating biological catalysts, a number of novel dithiolene catalysts have been researched. Cobalt dithiolenes based on bis(nitrile) dithiolate ligands and nickel dithiolenes based on chlorobenzene dithiolate ligands catalyze water splitting reactions.19,20 Nickel and molybdenum dithiolenes based on bis(trifluoromethyl)dithiolate

8 ligands have been shown to isomerize quadricyclane to norbornadiene.21 Nickel bis(trifluoromethyl)dithiolene catalyzes the oligmerization of ethylene when activated with methylaluminoxane.22

As mentioned earlier, alkenes can bind to some dithiolenes as shown in eq. 1.5. Interestingly, alkenes can bind to dithiolenes in both interligand and intraligand fashion as shown in eq. 1.7 and eq. 1.8.23,24

F C CF 3 3 F3C CF3 S S S S eq. 1.7 Ni Ni

S S S S F C CF 3 3 F3C CF3

F C CF 3 3 F3C CF3 S S S S

Ni Ni eq. 1.8

S S S S F C CF 3 3 F3C CF3

Under very pure conditions the major product is the intraligand adduct and this is predicted by symmetry arguments.25 But in the presence of a monoanionic dithiolene complex, the interligand product is the favoured product. This is due to the formation of a dimetallic complex.23

Alkene addition has the effect of limiting the allowed oxidation states of the ligand. The 2- oxidation is no longer allowed and thus the electrons must be assigned to the metal. In effect the alkene addition on the ligand reduces the metal center. In heteroleptic dithiolene complexes, in particular tris(dithiolenes), the alkene adds to the most electron-donating ligand.10 At first this seems counterintuitive as alkene addition is an electrophilic reaction on the part of the complex, but in heteroleptic dithiolenes the more electron withdrawing ligands pull electron density away from the more electron donating ligands, this oxidizes them from dithiolates into dithioketones and makes them more susceptible to alkene addition via a [4+2] cycloaddition reaction. The utility of this is that complexes can be made to add alkenes predictably at desired locations by tuning the relative electron donating and withdrawing capabilities of the ligands.

1.3 Scope of Thesis

In this work the reactions of molybdenum bis(dithiolenes) containing tfd ligands are studied. In particular it is used as a structural model complex for hydrodesulfurization catalysts, and as a functional model for oxotransferase catalysts due to its similarities to such systems. In addition, the property of ligand non-innocence is explored through finding novel reactions the ligands can undergo.

In chapter 2 the synthesis of the difficult to obtain chemicals hexafluorobutyne and bis(trifluoromethyl)dithiete (tfd) is presented using easily obtained precursors and improvised apparatus to safely isomerize and react the gaseous intermediates.

In chapter 3 we use molybdenum dithiolenes to create structural model complexes of the proposed active sites of the molybdenum disulfide catalyst as well as determine competitive binding affinities for dihydrothiophene and tetrahydrothiophene, and explore some basic catalytic properties. The entirety of this work is published in the European Journal of Inorganic Chemistry.26

In chapter 4 we present a new dithiolene reaction where it is attacked by a nucleophile (triphenylphosphine) to create a zwitterionic ligand as well as open an active site on a previously coordinatively saturated molybdenum tris(dithiolene). This technique is used to create a structural model complex for DMSO reductase and produce a pre-catalyst for that same reaction. The entirety of this work is published in Inorganic Chemistry.27

In chapter 5 we perform kinetic experiments on a molybdenum bis(dithiolene) to elucidate the mechanism of its oxotransferase-like activity. Kinetic investigations suggest the binding of the phosphine oxide created from the use of triphenylphosphine as the oxygen acceptor competes with DMSO in the binding to the active site of the molybdenum bis(dithiolene). Additionally, the removal of oxygen from DMSO using the catalyst appears to involve a polar transition state.

References

1 Bertini, I; Gray, H. B.; Stiefel, E. I.; Valentine, J.S. Biological Inorganic Chemistry Structure and Reactivity University Science Books: Sausalito Ca, 2007, pp. 527-529

2 Kaim, W.; Schwederski, B. Coord. Chem. Rev. 2010, 254, 1580–1588.

3 Hine, F. J.; Taylor, A. J.; Garner, C. D. Coord. Chem. Rev. 2010, 254, 1570-1579.

4 Fekl, U.; Sarkar, B.; Kaim, W.; Zimmer-De Iuliis, M.; Nguyen, N. Inorg. Chem. 2011, 50, 8685.

5 Eisenberg, R.; Ibers, J. A. J. Am. Chem. Soc. 1965, 87, 3776.

6 Rauchfuss T.B. Prog. Inorg. Chem. 2004, 52, 1-55

7 Holm, R. H.; Solomon, E. I.; Majumdar, A.; Tenderholt, A. Coord. Chem. Rev. 2011, 255, 993- 1015.

8 Wang, K.; Stiefel, E. I.; Patil, A. O.; Zushma, S.; US Patent #6,743,960

9 Davison, A.; Holm, R. H. Inorg. Synth. 1967, 10,8.

10 Harrison, D. J.; Lough, A. J.; Nguyen, N.; Fekl, U. Angew. Chem. Int. Ed. Engl. 2007, 46, 7644–7.

11 Wang, K.; Stiefel, E. I. Science 2001, 291, 106.

12 Miller, C. C.; Lowe, A. J. J. Chem. Soc. 1940, 1258-1263.

13 Mendel, R. R. J. Biol. Chem. 2013, 288, 13165–13172.

14 Romão, M. J.; Archer, M.; Moura, I.; Moura, J. J. G.; LeGall, J.; Engh, R.; Schneider, M.; Hof, P.; Huber, R. Science 1995, 270, 1170–1176.

15 Coelho, C.; González, P. J.; Moura, J. G.; Moura, I.; Trincão, J.; João Romão, M. J. Mol. Biol. 2011, 408, 932-948.

16 Donahue, J. P.; Goldsmith, C. R.; Nadiminti, U.; Holm, R. H. J. Am. Chem. Soc. 1998, 120, 12869–12881.

17 Majumdar, A.; Pal, K.; Sarkar, S. Dalton T. 2009, 1927–1938.

18 Kail, B. W.; Pérez, L. M.; Zarić, S. D.; Millar, A. J.; Young, C. G.; Hall, M. B.; Basu, P. Chem. Eur. J. 2006, 12, 7501–9.

19 McNamara, W. R.; Han, Z.; Yin, C.-J. (Madeline); Brennessel, W. W.; Holland, P. L.; Eisenberg, R. PNAS 2012, 109 , 15594–15599.

20 Hontzopoulos, E.; Knostantatos, J.; Vrachnou-Astra, E.; Katakis, D. J. Mol. Catal. 1985, 31, 327–333.

21 King, R. B.; Ikai, S. J. Mol. Catal. 1978, 4, 361–373.

22 Wang, K.; Patil, A. O.; Zushma, S.; McConnachie, J. M. J. Inorg. Biochem. 2007, 101, 1883– 1890.

23 Dang, L.; Shibl, M. F.; Yang, X.; Harrison, D. J.; Alak, A.; Lough, A. J.; Fekl, U.; Brothers, E. N.; Hall, M. B. Inorg. Chem. 2013, 52, 3711.

24 Harrison, D. J. ; Nguyen, N. ; Lough, A. J. ; Fekl, U. J. Am. Chem. Soc. 2006, 128, 11026.

25 Fan, Y.; Hall, M. B. J. Am. Chem. Soc. 2002, 124, 12076

26 N. Nguyen, D. J. Harrison, A. J. Lough, A. G. De Crisci, U. Fekl, European Journal of Inorganic Chemistry 2010, 2010, 3577–3585.

27 N. Nguyen, A. J. Lough, U. Fekl, Inorganic Chemistry 2012, 51, 6446–6448.

Chapter 2: Convenient Lab-scale Synthesis of Hexafluorobutyne and Bis(trifluoromethyl)dithiete (tfd) 2.1 Abstract

The synthesis of bis(trifluoromethyl)dithiete (S2C2(CF3)2) from the commercially available precursor 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane (C4Cl2F6) is presented here. This set of procedures involves modified and improved (for small-scale research lab preparations) literature procedures. An important intermediate is the useful but difficult to procure compound 1,1,1- 4,4,4-hexafluorobut-2-yne (henceforth abbreviated as ‘hexafluorobutyne’). Although the individual chemical steps needed are known, the set of existing publications is very fragmented and not optimised for the modern research lab. Providing a useful and up-to-date procedure in one place seems particularly important in light of recent restrictions on international shipping of the key intermediate, hexafluorobutyne, which is a toxic gas.

2.2 Introduction

Bis(trifluoromethyl)dithiete (S2C2(CF3)2 or ‘tfd’) is a very useful ligand for dithiolene chemistry.1 This is because it is extremely electron-withdrawing, relatively stable in its charge- neutral form (useful for generating metal complexes in a highly oxidized state), and possesses an NMR “handle” through its fluorine atoms. However it is not commercially available and the synthesis is non-trivial. The most direct synthesis is the direct combination of sulfur and 1,1,1,4,4,4-hexafluorobut-2-yne (henceforth abbreviated as ‘hexafluorobutyne’) at 450 oC (eq. 2.1).2 The high temperature is required to crack the polymeric and/or crystalline sulfur and produce the necessary S2 molecules for the [2+2] cycloaddition reaction.

CF3 F3C 450 oC S + S2 eq. 2.1 S

F3C CF3

This is a difficult step for most research labs, since 450 oC is not easily accessible with most laboratory heating mantles, and open flames are nowadays avoided, in particular when highly

13 flammable compounds are being used. Hexafluorobutyne is highly flammable and also very toxic. Another difficulty is obtaining the hexafluorobutyne itself. Gases tend to be more expensive than liquids or solids due to the added cost of the tank or lecture bottle and special handling needed to produce specialty gases. In some countries (eg. Canada) the chemical is not produced domestically and export controls on gaseous fluorinated substances make it extremely difficult to obtain it internationally. Synthesis from a freely available source is preferable.

2.2.1 Synthesis of hexafluorobutyne

There are a number of synthetic routes to hexafluorobutyne with varying degrees of cost and feasibility. They shall be presented here grouped in terms of synthetic similarity.

2.2.1.1 Liquid phase synthesis

The easiest methods for a laboratory chemist are simple liquid phase reactions where the liquid and possibly solid reagents are mixed in a single container and the gaseous products are evolved. The only special equipment is a gas trap to collect the gases.

A fast and high-yielding preparation of hexafluorobutyne from hexachlorobutadiene, requiring the high-boiling fluorinated solvent perfluoroperhydrophenanthrene (eq. 2.2) was reported in 1997.3

CF3 F O O Cl2C Cl S KF + F F + eq. 2.2 190oC

Cl2C Cl CF3

18-Crown-6 may be substituted for sulfolane with comparable yields.4 We found the solvent to be prohibitively expensive but this can change with time (and location), such that this preparation may become extremely attractive in the future.

Wong et al. produced hexafluorobutyne by dechlorination of 2,3-dichloro-1,1,1,4,4,4- hexafluorobutene through the reaction with zinc in acetic anhydride at 140 oC for 10 hours (eq. 2.3).5

CF3 Cl CF3

Ac2O + Zn ZnCl2 eq. 2.3 140oC +

Cl CF3 CF3

The previous reaction also works with 2,3-dibromo-1,1,1,4,4,4-hexafluorobutene and ethanol solvent.6 This is a potentially very attractive method (as will be discussed below), although it is not the method we chose for our work.

2.2.1.2 “Single gas phase reagent” syntheses

More complicated than simple liquid phase synthesis is a gas phase reaction where one of the reagents or intermediate products is a gas. This requires additional equipment to handle the gas and react it with additional liquid or solid phase reagents.

Chambers et al. produced hexafluorobutyne through the dehydrohalogenation of 1,1,1,3,4,4,4- heptafluorobut-2-ene using molecular sieves over the course of four weeks at room temperature (eq. 2.4).7

CF3 F3C molecular sieves eq. 2.4 4 wk

F3C F CF3

The 1,1,1,3,4,4,4-heptafluorobut-2-ene was obtained by the reaction of hexachlorobutadiene with KF in refluxing N-methyl-2-pyrrolidone (eq. 2.5).8

F3C Cl2C Cl KF, NMP eq. 2.5 reflux

F C F Cl2C Cl 3

While the reactions in eq. 2.4 and eq. 2.5 are viable, the timescale of four weeks is rather long.

Jia et al. produced hexafluorobutyne by reacting hexafluorobutadiene with AlOClF (eq. 2.6).9

CF3 F2C F AlOClF eq. 2.6

F2C F CF3

However, supplies of hexafluorobutadiene are equally as difficult to procure as hexafluorobutyne.

Pötter et al. showed (E)-(perfluorobut-2-ene-2,3-diyl)bis(trifluoro-λ4-sulfane) decomposes at room temperature to hexafluorobutyne and other unspecified decomposition products (eq. 2.7).10

CF3 F3S CF3 R.T. + Decomp. eq. 2.7

F3S CF3 CF3

However the precursor itself is not commercially available and also not trivial to synthesize as it uses a high-pressure autoclave and requires SF4, a difficult to handle gaseous fluorinating agent.10

The pathway to hexafluorobutyne preferred by us involves the high temperature cycloreversion of hexafluorocyclobutene over KF at 590 oC (eq. 2.8).11

F CF F 3

F 590 oC eq. 2.8 F KF

F F CF3

Despite the high reaction temperature, performing this cycloreversion reaction is very straightforward and requires only inexpensive extra equipment (below). Hexafluorocyclobutene itself was easily obtained by dechlorination of 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane using zinc and ethanol (eq. 2.9). This dechlorination was disclosed in the literature but mentioned in one sentence only.12 It certainly will be useful that we are giving details on this dechlorination in the Experimental Section.

F F F F Cl F EtOH F Zn eq. 2.9 + + ZnCl2 Cl F F

F F F F

Fortunately, the starting material for the dechlorination (1,2-dichloro-1,2,3,3,4,4- hexafluorocyclobutane) is a liquid (not a gas) and easily available from several commercial sources (for example from Synquest Labs and Matrix Scientific).

2.2.1.3 “Two gas phase reagent” syntheses

Even more complicated to perform is reactions that have two reagents in the gas phase. Such reactions are rather commonly performed in the chemical industry; in the laboratory, however, safely controlling the flow and mixing ratio of two gasses requires additional equipment. If a process using a single gas could be used, then the need for such equipment can be avoided.

Nappa dehalogenated 2-chloro-1,1,1,3,4,4,4-heptafluorobutene with hydrogen over a nickel, copper, chromium and calcium fluoride catalyst at 400oC (eq. 2.10).13

CF3 Cl CF3

Ni,Cu,Cr,CaF2 eq. 2.10 o H2, 400 C

F CF3 CF3

The yield of hexafluorobutyne for this process was 19.2% at best.

Poss et al. used a similar process to dehydrohalogenate 1,1,1,2,2,3,4,4,4-nonafluorobutane (eq. 2.11).14

CF3 F3C CHF Ni,Cu,Cr eq. 2.11 CF o 2 H2, 400 C

F3C CF3

Instead of hydrogen, carbon monoxide can also be used as the halogen acceptor in a slightly different process (eq. 2.12).15

CF3 Cl CF3

CuCl2, KCl CO + + COCl2 eq. 2.12 300oC

Cl CF3 CF3

This process also works with 2-butyne at 275 oC.

2.2.1.4 Exotic methods

These methods use extremely non-trivial reagents and/or reaction conditions. While interesting from an academic perspective they are wholly impractical when other more straightforward methods are available. They are included here for completeness.

Thoreson et al. showed dichloro-2,3-dichloro-1,1,1,4,4,4-hexafluorobutene could also be dechlorinated by reaction with an iron(0) complex (eq. 2.13).16

+Cl- tBu tBu CF Si 3 Si Cl CF3

P P tol. P P P P + + eq. 2.13 Fe R.T. Fe P P P P Cl CF3 Cl CF3

Apart from proceeding at room temperature this method provides no advantage over the previously mentioned process in eq. 2.3, as the complex itself is not commercially available and much more difficult to obtain than zinc metal.

Kobayashi et al. photolyzed 1,2,4,6,7-pentakis(trifluoromethyl)-3,5-diazatricyclo[4.1.0.02,7]hept- 3-ene into 2,4,5-tris(trifluoromethyl)-4,5-dihydro-1H-imidazole and hexafluorobutyne (eq. 2.14).17

CF3

F3C CF3 F3C N N hv eq. 2.14 CF3 CF + F3C 3 Et2O N NH H F3C CF3 F3C

However, the precursor itself was made from exotic precursors that are not commercially available18 making this process unattractive in the laboratory setting.

Kobayashi et al. also found that photolysis of 1,2,3,8-tetrakis(trifluoromethyl)cyclooctatetraene also liberated hexafluorobutyne (eq. 2.15).19

CF3

CF3 CF3 CF3 hv + eq. 2.15

CF3 CF3 CF3

CF3

But again, the precursor is also not commercially available and non-trivial to synthesize.19 In addition there were numerous side products and the literature implies that the hexafluorobutyne was a minor product. This process is thus not a real option, from a cost and efficiency standpoint.

2.2.1.5 Selection

When selecting a particular synthetic approach cost, yield and overall ease are important factors to consider. The exotic methods are rejected outright for requiring difficult to obtain precursors. The two gas phase reagent approaches are promising for large-scale preparations but require too much startup capital for smaller runs if the necessary gas mixing equipment is not already available. Of the single gas phase reagent approaches, the processes in eq. 2.4 and eq. 2.5 are rejected for taking too long (about four weeks) while eq. 2.6 and eq. 2.7 are rejected for requiring difficult to procure or synthesize precursors. This leaves eq. 2.8 with eq. 2.9 for further consideration. Both the liquid phase approaches appear very convenient in terms of reagents and reaction conditions and are also considered.

2.2.1.6 Cost analysis

Based on prices of chemicals and equipment in December 2013, a relative cost analysis can be performed. For this analysis common costs such as gas traps, hot plates and cryogens, which are common to all procedures, will be excluded and only unique costs will be assessed to determine

20 the relative differences between them. Additionally, specialized equipment used in the subsequent synthesis to make tfd will be assumed to be available as such equipment had to have been acquired anyway to make tfd. Thus for the isomerization of hexafluorocyclobutene into hexafluorobutyne the cost of the heating tape, also needed for making tfd, is not considered.

To perform this analysis the costs for chemicals were obtained from suppliers commonly available to research laboratories, in this case Alfa Aesar, Sigma Alrich, Synquest Labs and Chemglass. As chemicals have different price points for different quantities (10 g of chemical may have different cost per gram than 100 g of chemical) the quantities were selected to approximately yield ~100 g of hexafluorobutyne as a point of comparison. Once the quantities were selected the cost per gram of each chemical (the “unit cost”) was determined. Each procedure was reviewed and the quantity of chemical used in each one was multiplied by its unit cost and then summed up to obtain the total cost for a typical run in that procedure. The total cost was then divided by the yield to obtain the unit cost of hexafluorobutyne allowing for cost comparisons between each procedure. The results of the analysis can be found in Table 2.1.

The procedure specified in eq. 2.2 is very costly at $37.14 per gram of hexafluorobutyne but the bulk of the cost is the perfluorophenanthrene solvent. This can be fractionally distilled and recycled for subsequent runs. If this route is taken, and assuming 90% recovery of the solvent, then the unit cost of hexafluorobutyne using this procedure is $6.23 a large improvement. Nonetheless the initial cost of over $3000 may be unattractive despite the synthetic ease of the procedure.

The procedure specified in eq. 2.3 is synthetically just as easy as the procedure in eq. 2.2 but far less costly at $5.37/g. It is suggested that this procedure should be seriously considered if the reagents are easily available at acceptable cost.

The procedure specified by eq. 2.8 and eq. 2.9 is the one used in this work and shall be described below. It has the lowest unit cost overall at $4.86/g. The quartz tube specified as part of the cost per run as the tubes are sacrificial and become badly etched, compromising their integrity and rendering them unsafe if used more than a few times. Fortunately they are inexpensive compared to the other reagents. As mentioned previously, heating tape is not considered part of the cost as it is used in the next step for making tfd anyway. Alternatively, a laboratory tube furnace may be used instead of heating tape for similar cost benefits.

A question that an alert reader will ask is why our group did not use the alternative process of producing hexafluorobutyne from a 2,3-dihalo-1,1,1,4,4,4-hexafluorobutene reagent (eq. 2.3) as it appears a great deal more convenient for a marginal increase in cost. The reason is because at the time when our group was searching for a feasible method (2007) the precursors required for that process were not available in economically viable quantities, but 1,2-dichloro-1,2,3,3,4,4- hexafluorocyclobutane was. Since the required equipment and expertise to react hexafluorobutyne with sulfur at 450 oC (eq. 2.1) could be directly applied to the cycloreversion of hexafluorocyclobutene using KF at 590 oC (eq. 2.8), it was decided the minor cost of an extra quartz tube and KF was preferable to the much greater cost of the 2,3-dihalo-1,1,1,4,4,4- hexafluorobutene precursor. However at the time of this writing (2013) the availability of 2,3- dihalo-1,1,1,4,4,4-hexafluorobutene reagents has increased considerably and the costs have also decreased significantly. While we find it important to mention the preparation shown in eq. 2.3, we have worked successfully and repeatedly, with the preparation shown in eq. 2.8 and eq. 2.9, which will therefore be described in detail.

Cost Analysis Consumables List Density Quantity (g) Cost ($) Unit cost $/g Vendor 1,2-Dichloro-1,2,3,3,4,4-hexafluorocyclobutane 100 $175.00 $1.7500Synquest 2,3-Dichloro-1,1,1,4,4,4-hexafluoro-2-butene 100 $225.00 $2.2500Synquest Acetic Anhydride 1.082 1000 $45.00 $0.0450Alfa Aesar Ethanol 0.789 500 $25.00 $0.0500Alfa Aesar Hexachlorobutadiene 250$111.00 $0.4440Alfa Aesar KF 500$112.00 $0.2240Alfa Aesar KF (spray dried with 0.5% SiO2) 50 $68.90 $1.3780Sigma Aldrich Perfluoroperhydrophenanthrene 2.03 100 $154.00 $1.5400Alfa Aesar Quartz Tube 1 $11.20 $11.2000Chemglass Sulfolane 1.261 1000 $70.00 $0.0700Alfa Aesar Zinc 500$55.20 $0.1104Alfa Aesar

Analysis Mass (g) Volume (mL) Cost eq. 2.2 procedure Hexachlorobutadiene 260 $115.44 Perfluoroperhydrophenanthrene 2030 1000 $3,126.20 Sulfolane 378.3 300 $26.48 KF 500 $112.00 Yield of hexafluorobutyne 91 Total $3,380.12 Cost $/g $37.14 eq. 2.2 procedure w/ 90% solvent recovery Hexachlorobutadiene 260 $115.44 Perfluoroperhydrophenanthrene 2030 1000 $312.62 Sulfolane 378.3 300 $26.48 KF 500 $112.00 Yield of hexafluorobutyne 91 Total $566.54 Cost $/g $6.23 eq. 2.3 procedure 2,3-Dichloro-1,1,1,4,4,4-hexafluoro-2-butene 63 $141.75 Zinc 55 $6.07 Acetic Anhydride 116.856 108 $5.26 Yield of hexafluorobutyne 28.5 Total $153.08 Cost $/g $5.37 eq. 2.8 with eq. 2.9 procedure 1,2-Dichloro-1,2,3,3,4,4-hexafluorocyclobutane 100 $175.00 Ethanol 157.8 200 $7.89 Zinc 60 $6.62 KF (spray dried with 0.5% SiO2) 6 $8.27 Quartz Tube 1 $11.20 Yield of hexafluorobutyne 43.04 Total $208.98 Cost $/g $4.86

Table 2.1: Cost analysis of various procedures to synthesize hexafluorobutyne.

2.3 Synthesis of hexafluorocyclobutene

100 g (429 mmol) of 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane (C4Cl2F6) are dissolved in 200 mL of anhydrous ethanol and reacted with 60 g of zinc metal in a 500 mL round bottom flask with a reflux condenser and liquid nitrogen cooled cold trap attached successively. The reaction is initiated by gently warming the mixture with a heat gun to reflux. Once started, the dechlorination is exothermic and self-sustaining, and the water cooled reflux condenser is needed to recycle any ethanol and 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane vapor. The hexafluorocyclobutene (C4F6) is a colorless gas that is collected with a liquid nitrogen cooled cold trap. The hexafluorocyclobutene is re-purified by first warming from cryogenic temperature to 0 oC with an ice bath and then allowing it to return to room temperature, as the gas evaporates it is again collected with another liquid nitrogen cooled cold-trap, leaving behind ethanol. The product is vacuum transferred into a Pyrex reaction vessel equipped with a Teflon valve (“Pyrex bomb” or “Schlenk bomb”). Yield 53.8 g (332 mmol or 77%). At room temperature, the storage vessel will be under slightly elevated pressure (vapour pressure of hexafluorocyclobutene = 20 psi at ambient temperature).

In terms of safety this procedure should be carried out in a fume hood at all times. The reaction of zinc and 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane can exothermically accelerate beyond control (“thermal runaway”) if not monitored. If the reaction rate increases to the point that the recondensing ethanol is in danger of saturating the condensor and escaping into the cold trap then the reaction can be slowed by immersing the round bottom flask in ice water. The reaction may need to be reheated to restart it if this is done. When moving onto the next step of warming the hexafluorocyclobutene it is important to note the frozen hexafluorocyclobutene will also have collected some liquid air and may have frozen in such a way as to plug the escape of these gases. It is therefore necessary to carefully inspect the trap while it is cold and to melt the plugs first as well as clear a channel for any liquefied air to escape. Some loss of product may occur but this is necessary to avoid a possible explosion of the trap if the pressure exceeds its maximum rating. As the temperature is raised to 0 oC, considerable outgassing of the liquefied and dissolved air will occur, violently “boiling” the hexafluorocyclobutene. To prevent loss of product the warming should be slow enough to keep this “boiling” to a minimum as it outgases. During the step to vacuum transfer the hexafluorocyclobutene into a pyrex storage vessel it should be emphasized that the vapor pressure of hexafluorocyclobutene is greater than

24 atmospheric vapor pressure at room temperature. It must be kept cold (~0 oC) at all times when vacuum transferring or the positive pressure will blow the container off the vacuum manifold it is attached to. Only when it is safely sealed in a storage vessel that can handle such pressure should it be allowed to reach room temperature. Additionally, air may still be entrained in the hexafluorocyclobutene so it is recommended it is “degassed” via three freeze- pump-thaw cycles before finally being allowed to reach room temperature.

2.4 Synthesis of hexafluorobutyne

First a tube furnace must be constructed; if a tube furnace is already available then this step may be omitted and the quartz tube directly inserted into the tube furnace. A tube furnace is constructed (Figure 2.1 and Figure 2.2) by wrapping a quartz tube (60 cm length, 8 mm outer diameter, 6 mm inner diameter) with fiberglass-insulated heating tape (Omega Engineering product STH051-040) and wrapped with aluminum foil. A type-K thermocouple (with associated measuring device) is inserted between the tape and the tube, to measure temperature.

Once a tube furnace has been obtained, the quartz tube is preloaded with 30 cm of rigorously dried potassium fluoride (approx 6g of Sigma Aldrich ‘spray dried’ with 0.5% SiO2) and packed at both ends with glass wool. The KF is placed in such a way as to be at the end of the gas stream, this allows the gas to be preheated by the tube itself before entering the KF. Inside the tube, the hexafluorocyclobutene will be thermally cracked to hexafluorobutadiene and will immediately isomerize to hexafluorobutyne on the KF. With the closed storage vessel containing the hexafluorocyclobutene attached, the tube is first evacuated and then heated at full operating temperature of 590 oC for one hour, in order to further dry the KF contents. The quartz tube is sacrificial, as KF at this temperature will badly etch the tube. After the drying procedure is finished, the hexafluorocyclobutene (C4F6) is slowly fed into the tube furnace at low pressure, by opening the Teflon valve of the storage vessel (Pyrex bomb). The product is condensed at the other side using a liquid nitrogen-cooled gas trap and will be vacuum transferred again into a “Pyrex bomb”. Yield is 60%-80%. As the pressure of hexafluorobutyne at ambient temperature is about 7 atm, we strongly recommend to keep the vessel at liquid nitrogen temperature and use the hexafluorobutyne immediately to prepare tfd. We did once confirm an 80% yield of crude

25 hexafluorobutyne by weighing the receptacle. Generally, we had good results when using the hexafluorobutyne directly without weighing and determining the final yield of tfd at the end.

In terms of safety, KF is toxic and gloves should be worn when handling it. If using the heating- tape based tube furnace rather than commercial unit it should be stated that the furnace and all associated electrical connections should not be touched while it is plugged in, even if it is not powered. It should only be handled when it is completely unplugged. While the heating tape is insulated this should not be exclusively relied upon for electrical safety. Additional safety can be achieved by inserting a ground-fault interrupter in the power line to the heating tape.

As stated previously, hexafluorobutyne has a gas pressure of 7 atm at room temperature, which exceeds the reliable pressure range of most pyrex reaction vessels. It is therefore imperative that it is always kept cold (<-20 oC) while being collected and stored. Otherwise there is a real danger of the vessel exploding. Only if it can be transferred into a steel lecture bottle can it be allowed to rise to room temperature. When ready to be used for tfd synthesis it may be kept in an ice bath to raise the pressure to above atmospheric pressure (approx 3.2 atm), but still within safe ranges for most pyrex vessels. The whole procedure should be performed in a fume hood.

Figure 2.1: Hexafluorocyclobutene isomerization apparatus

Vacuum system valve

Quartz tube Rubber packed with KF tubing

Digital thermometer Heating tape (wrapped in Hexafluorobutyne aluminum foil) (immersed in liquid nitrogen)

Hexafluorocylobutene (immersed in ice water)

Figure 2.2: Hexafluorocyclobutene isomerization apparatus (assembled)

2.5 Synthesis of bis(trifluoromethyl) dithiete S2C2(CF3)2

A high-temperature heating mantle was constructed by wrapping heating tape around the bottom of a three necked round bottom flask and securing the tape with stiff iron wire. The flask was then filled with 35 g of sulfur and the three-necked round bottom flask was outfitted with glassware as shown in Figure 2.3 and Figure 2.4.

Figure 2.3: ‘tfd’ synthesis apparatus.

Figure 2.4: 'tfd' synthesis apparatus (assembled)

The Pyrex bomb containing hexafluorobutyne is controlled by valve “A”. With this valve still closed, the system is purged with nitrogen. The high temperature heating tape is being used now to heat the sulfur until reflux. The correct temperature is determined by simply observing recondensing yellow sulfur on the inner walls of the round bottom. When valve “A” is carefully opened, the gas starts flowing into the round bottom reaction vessel “B”. We found that a useful flow rate is obtained when the hexafluorobutyne is cooled no longer by liquid nitrogen but by ice water. As the gas flows, the hexafluorobutyne reacts with the sulfur and very rapidly produces bis(trifluoromethyl) dithete (tfd), visible in the form of orange vapor. This orange vapor exits the round bottom flask through still head “C”, in part contaminated with sulfur vapors and unreacted hexafluorobutyne. Sulfur starts to very quickly condense, as the gases rise through the still head,

30 and most of it drips back into the reaction vessel. However some sulfur will still travel to still head “D”, where it can be trapped with the attached round bottom flask. The remaining gaseous compounds, however, travel into condenser “E”, where tfd condenses and is collected in the Schlenk bomb receptacle “F”, while unreacted hexafluorobutyne exits through a bubbler. After the reaction is complete (typically ~3h) the tfd can still contain small amounts of elemental sulfur, in which case it is purified by vacuum transfer into another Pyrex bomb. The tfd ligand can be further purified by fractional distillation (bp 97 oC), although we did not find this necessary.

19 Typical yield between 60%-80% based on used hexafluorobutyne. F NMR (376 MHz, CDCl3)

δ –62.84 (s, 6F, (CF3) x 2) referenced to external trifluoroacetic acid capillary at δ –78.5.

Overall yield from 100 g of 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane is typically ~40 g of

S2C2(CF3)2 or about ~41% overall.

For safety, if the heating-tape based heating mantle is used, then metal parts of the apparatus should not be touched while the device is plugged in. A ground fault interrupter is recommended for additional safety. There is a small explosion danger from the positive pressure in the hexafluorobutyne pyrex bomb and it should therefore be strategically placed to minimize injury to the experimenter if this should happen. It must be kept in an ice bath or colder to keep the pressure down to safer levels. The refluxing sulfur presents a strong flammable hazard so all ground glass joints must be thoroughly inspected and sealed. Plastic Keck clips cannot be used, as they will melt at these temperatures. Wire clips and metal joint clamps are recommended to hold together the apparatus. The tfd itself is toxic and should be handled in a fume hood.

2.6 Results and Discussion

Overall yield from 100 g of 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane is typically ~40 g of

S2C2(CF3)2 or about ~41% overall.

The purified tfd ligand obtained this way can be further purified by fractional distillation. The tfd is unstable with time but remains highly active for several weeks. It slowly dimerizes and

31 deposits elemental sulfur. It is believed that the final decomposition product is 2,3,5,6- tetrakis(trifluoromethyl)-1,4-dithiine.2

F C 3 F3C S S CF 3 F3C S CF3 S 2 + S2 S

F C F3C S CF3 3 F3C SS CF 3

Scheme 2.1

While the dimer can be thermally cracked to restore active tfd monomer,2 the decay to 2,3,5,6- tetrakis(trifluoromethyl)-1,4-dithiine is irreversible and represents a severe hindrance to reproducible research with it. However once the tfd had been bound to a metal (ex. Ni(tfd)2,

MoO(tfd)2) it is often stable indefinitely.

We noticed unpurified tfd becoming unusable due to degradation after three months. It is therefore strongly recommended that tfd be used within the first three months after its creation, preferably in the first month. If longer storage is required then purification by fractional distillation is recommended. A single distillation increases the usable lifetime up to six months, but it is still recommended to use the tfd in the first three months. Highly purified tfd (triply distilled) has greater stability and samples have not deposited elemental sulfur after two years of air-free storage. It is believed that side products produced during the reaction of hexafluorobutyne and sulfur are the cause of the instability, initiating and/or catalyzing the dimerization and degradation of tfd. Any decision regarding whether to purify or not should be made just after the tfd is synthesized, as unpurified degraded samples cannot be restored.

In conclusion, we have synthesized tfd from easily obtained 1,2-dichloro-1,2,3,3,4,4- hexafluorocyclobutane. First, dehalogenation with zinc yields hexafluorocyclobutene. Cycloreversion is performed over KF to yield hexafluorobutyne. Finally, reaction with refluxing sulphur yields tfd. The enabling technology was “heating tape” allowing the inexpensive and straightforward construction of tube furnaces and heating mantles required to safely and accurately produce and control the temperatures needed.

References

1 Dithiolene Chemistry: Synthesis, Properties, and Applications, E. I. Stiefel, Ed; Progress in Inorganic Chemistry; John Wiley and Sons: Hoboken, NJ, 2004; Vol 52.

2 Krespan, C. G. J. Am. Chem. Soc. 1961, 83, 3434–3437.

3 Chambers, R. D.; Edwards, A.R. J. Chem. Soc., Perkin Trans. 1 1997, 3623-3627.

4 Chambers, R. D.; Edwards, A. R. Solvents for use in fluorination reactions. WIPO Patent No. 1998055429, December 10, 1998.

5 Wong, H. N. C.; Xing, Y.; Zhou, Y.; Gong, Q.; Zhang, C. Synthesis 1984, 9, 787-790

6 Haszeldine, R. N. J. Chem. Soc. 1952, 2504–2513.

7 Chambers, R. D.; Roche, A. J. J. Fluorine Chem. 1996, 79, 121–124.

8 Maynard, J. T. J. Org. Chem. 1963, 28, 112–115.

9 Jia, X.; Zhou, X.; Quan, H.; Tamura, M.; Sekiya, A. J. Fluorine Chem. 2011, 132, 1188–1193.

10 Pötter, B.; Seppelt, K.; Simon, A.; Peters, E. M.; Hettich, B. J. Am. Chem. Soc. 1985, 107, 980–985.

11 Chambers, R. D.; Jones, C. G. P.; Taylor, G.; Powell, R. L. J. Chem. Soc. Chem. Comm. 1979, 964.

12 Henne, A. L.; Ruh, R. P. J. Am. Chem. Soc. 1947, 69, 279-281.

13 Nappa, M. J. Synthesis of 2-chloro-1,1,1,3,3,4,4,4-heptafluoro-2-butene and hexafluoro-2- butyne. U.S. Patent 20090156869, June 18, 2009.

14 Poss, A. J.; Nalewajek, D.; Van Michael, D. P.; Nair, H. K. Process for the production of fluorinated alkenes. WIPO Patent No. 2011146802, November 24, 2011.

15 Poss, A. J.; Nalewajek, D.; Nair, H. K.; Van Michael, D. P.; Singh, R. R. Process for preparation of hexafluoro-2-butyne and cis-hexafluoro-2-butene. WIPO Patent No. 2011146812, November 24, 2011.

16 Thoreson, K. A.; McNeill, K. Dalton T. 2011, 40, 1646–1648.

17 Kobayashi, Y.; Nakano, T.; Nakajima, M.; Kumadaki, I. Tetrahedron Lett. 1981, 22, 1369– 1370.

18 Kobayashi, Y.; Nakano, T.; Nakajima, M.; Kumadaki, I. Tetrahedron Lett. 1981, 22, 1113- 1114.

19 Kobayashi, Y.; Ando, A.; Kawada, K.; Kumadaki, I. J. Chem. Soc. Chem. Comm. 1981, 1289– 1291.

Chapter 3: Molybdenum Dithiolene Complexes as Structural Models for the Active Sites of Molybdenum(IV) Sulfide Hydrodesulfurization Catalysts

3.1 Abstract

The removal of sulfur (as H2S) from organosulfur species in petroleum feedstocks (hydrodesulfurization, HDS) is carried out on an enormous scale using heterogeneous catalysts based on MoS2 (usually doped with Co). Partially hydrogenated thiophenes are postulated intermediates in the MoS2-catalyzed hydrodesulfurization of thiophene. The present contribution describes new molecular models for the proposed active sites in HDS catalysis. The models are derived from a mixed-ligand (push-pull) molybdenum tris(dithiolene) (Mo(tfd)2(bdt); tfd=S2C2(CF3)2, bdt=S2C6H4): selective intraligand alkyne binding converts the bdt group to a labile Mo-chelating benzodithiin, which can be substituted with a variety of weak donor ligands.

Complexes Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 (dht = 2,5-dihydrothiophene; tht = tetrahydrothiophene) were synthesized and crystallographically characterized. The Mo(tfd)2 substructures closely resemble the presumed active site in MoS2 HDS catalysts. Coordination geometries at molybdenum are approximately trigonal prismatic, with the metal bearing two strongly bound dithiolene (tfd) groups and two comparatively weakly bound thioether ligands (dht or tht). Competitive binding experiments establish that tht binds more strongly to the Mo center than dht (Keq = 6.5 ± 0.5). Preliminary reactivity studies reveal that Mo(tfd)2(dht)2 decomposes to Mo(tfd)3, thiophene and unidentified species upon heating. Also, Mo(tfd)2(tht)2 induces the isomerization of 1,4-cyclohexadiene to 1,3-cyclohexadiene at elevated temperature.

3.2 Introduction

As high-grade petroleum reserves dwindle, lower and lower grades of petroleum resources must be brought into service to meet continually growing energy demands. Environmental concerns require the sulfur content of low-grade petroleum be reduced before use.1 Hydrodesulfurization

(HDS) processes – the removal of sulfur from organosulfur species as H2S – utilize catalysts

based on molybdenum(IV) sulfide (MoS2; most commonly encountered as the hexagonal molybdenite, usually modified with cobalt for HDS applications) and hydrogen gas as a feedstock. As greater and greater efficiency is being demanded of hydrodesulfurization techniques, the search for better catalysts has intensified. Of special interest is the desulfurization of thiophenes, which are less reactive in HDS than non-cyclic thioethers and much less reactive than thiols.2,3 Thus, removing sulfur from thiophenes and related ‘stubborn’ (refractory) compounds (deep desulfurization) is a formidable challenge. While some noble metal catalysts show excellent activity in deep desulfurization,4 the high activity of those metals is offset by their cost. Developing increasingly active Mo-based catalysts is thus a worthwhile goal, and we focus here on molecular coordination compounds of molybdenum. Heterogeneous MoS2-based HDS catalysis has been probed with experimental surface techniques as well as with computational (DFT) studies on cluster models.5,6,7 However, small molecular (soluble) models mimicking the active sites in molybdenite HDS catalysts are rare.8 Small molecule models should be useful for understanding the mechanism of HDS catalysis and could lead to the development of better catalysts.6 It is thought that exposed (coordinatively unsaturated) edges of 9 MoS2 sheets are the active sites and that the internal centers are inactive. A fragment of the molybdenite structure is shown in Figure 3.1, which shows internal sites, edge sites and key metric parameters.

Figure 3.1: Views of exposed molybdenum sites in hexagonal MoS2; a) standard orientation, where the hexagonal layers are oriented horizontally; b) rotated to highlight the geometry10 of the 11 12 edge sites. The picture was generated with ORTEP , using MoS2 coordinates from the literature.

As shown in Figure 3.1, the molybdenum edge sites possess pyramidal structures, with molybdenum at the apex of a square pyramid and four sulfur atoms forming the base of the pyramid. Current evidence suggests that thiophene coordinates to the molybdenum edge sites. The coordinated thiophene then undergoes hydrogenation to 2,5-dihydrothiophene and, finally, desulfurization (via cycloreversion) of the organic moiety produces 1,3-butadiene and a terminal 13 metal sulfide (Scheme 3.1). Reduction of the metal sulfide with H2 releases H2S and regenerates the unsaturated Mo center.

S S cyclo- H2 S Mo reversion S S Mo S S S Mo S S S S S S S

5 Scheme 3.1: Desulfurization of thiophene at a MoS2 edge site. Thiophene initially binds in a η fashion to the active site14. Upon hydrogenation the resulting dht binds in a η1 fashion through the sulfur atom whereby it undergoes cycloreversion to liberate butadiene.

Accurate structural models of MoS2-based HDS catalysts should have the square pyramidal substructure (with Mo at the apex and four sulfur-donors forming the basal plane) seen for the edge sites in molybdenite. Additionally, the model should incorporate vacant (or labile) sites to bind organosulfur substrates (e.g., thiophene derivatives). While many examples of molybdenum 15,16,17,18,19 (or tungsten20) complexes bearing four sulfur donors are known, they usually contain additional strongly bound ligands, such as thiolates or phosphines, which cannot be displaced by thioether groups or other weakly-coordinating sulfur donors. However, they do show catalytic activity toward nitrate and formate reduction, and in that area are showing promise as models for enzymes. On the other hand, the few truly four-coordinate MoS4 complexes (e.g., Mo(SR)4, R = 2,4,6-triisopropylbenzene) are extremely electrophilic toward small donor molecules, forming a variety of MoS4L complexes, where L = (e.g.) alkyne, MeCN, t 21 BuNC, CO. However, these four coordinate MoS4 complexes have tetrahedral rather than pyramidal geometry at the metal, making them less desirable as model complexes for the molybdenite edge sites.

22 20 Monosulfided molybdenum or tungsten species with dithiolene ligands (‘SM(S2C2R2)’) have also been produced, and they may be regarded to be excellent models for the intermediate obtained after butadiene loss (right-hand side of Scheme 3.1). Schrauzer et. al. produced thioether complexes on tungsten based dithiolenes by methylation of one of the dithiete 23 ligands. Mo(CO)2(S2C2Me2)2 complexes are also known and offer both the square pyramidal 20,24 MoS4 structure as well as labile CO groups. While CO is a labile ligand in such complexes (as we also confirmed experimentally, see below), we are providing here an alternative approach to opening up two labile sites at ‘S4Mo’.

The first coordination sphere of neutral molybdenum tris(dithiolene)s resembles the environment of the internal (non-edge-sites) molybdenum atoms in molybenite – in both cases, the metal geometry is approximately trigonal prismatic with six sulfur-donor ligands. Our group has previously reported that mixed-ligand molybdenum tris(dithiolene)s (Mo(tfd)2(bdt) and

Mo(bdt)2(tfd) with tfd = S2C2(CF3)2, bdt = S2C6H4) react with ethylene using the sulfur atoms of one bdt ligand (i.e., intraligand alkene addition) to form a metal-chelating dihydrobenzodithiin moiety.25 The weakly bound dihydrobenzodithiin can be substituted with a variety of nucleophiles, allowing access to new molybdenum bis- and tris(dithiolene) complexes. In this paper we extend this versatile method to create molybdenum bis(dithiolene)s with thioether ligands (partially hydrogenated thiophenes), which provide structural models for postulated intermediates in MoS2-based HDS catalysis.

3.3 Experimental

25 nd Mo(tfd)2(bdt) was prepared by literature methods. Grubb’s 2 generation catalyst, tetrahydrothiophene (tht) (99%) and bis(trimethylsilyl)acetylene (btmsa) (99%) were obtained from Sigma Aldrich. The tetrahydrothiophene was redistilled and dried over molecular sieves (3 Å) before use. The bis(trimethylsilyl)acetylene was used directly. Solvents were dried on a MBraun Solvent Purification System (MB-SPS). NMR data were collected on a Bruker Avance 1 13 1 III 400 MHz spectrometer. H and C{ H} spectra were referenced to the CDCl3 solvent peaks, 1 (7.26 ppm and 77.23 ppm respectively). H spectra were also referenced to the C6D6 solvent peaks where applicable (7.16 ppm). 19F spectra were referenced to an external capillary of neat trifluoroacetic acid (δ, ppm, -78.5). Elemental analysis was performed by Chemisar Laboratories,

Guelph, Ontario, Canada. X-ray crystallographic data for Mo(tfd)2(tht)2 and Mo(tfd)2(dht)2 are available as files CCDC-762433 and CCDC-762434, respectively, as supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

3.3.1 Synthesis of Mo(tfd)2(tht)2

In air: 51.3 mg (74.5 μmol) of Mo(tfd)2(bdt) were dissolved in 10 mL of dry hexanes along with 63.4 μL (719 μmol) of tetrahydrothiophene and 80.4 μL (360 μmol) of btmsa and left to stand. Brown crystals slowly came out of solution over ca. 1 h, and the solution turned a red-purple color. After 18 h, the crystals were recovered by filtration and washed three times with 5 mL portions of hexane. The crystals were dried in vacuo to give Mo(tfd)2(tht)2. (39 mg, 54 μmol, 72%). Crystals were stored under inert atmosphere and were found to be stable at room temperature. X-ray quality crystals were obtained directly from the hexane reaction mixture. Representative NMR spectra can be found in Figure 3.6, Figure 3.7 and Figure 3.8 in the 1 19 appendix. H NMR (400 MHz, CDCl3) δ 2.28 (m, 8H, tht), 3.54 (m, 8H, tht). F NMR (376 13 1 MHz, CDCl3) δ –54.9 (s, 12F, (CF3)2 x 2) . C{ H} NMR (100 MHz, CDCl3) δ 30.109 (s, Cβ, tht), δ 42.31 (s, Cα, tht). Anal. Calc. for C16H16F12S6Mo (724.59): C 26.52, H 2.23, S 26.55; Found: C. 26.82, H 1.97, S 26.23.

3.3.2 Synthesis of 2,5-Dihydrothiophene

75 mg (88.3 μmol) of Grubb’s 2nd Generation Catalyst, 1,3-bis-(2,4,6-trimethylphenyl)-2- (imidazolidinylidene)(dichlorophenylmethylene)(tricyclohexylphosphine) ruthenium, were added to 75 mL of 0.1 M solution of diallyl sulfide in dry chloroform (7.5 mmol). Solution was stirred under a stream of argon for 24 h. Volatiles were vacuum transferred to a new container and heated to 70 °C until the volume was reduced to ca. 1 mL. The solution was vacuum-distilled a second time to a new container and stored under nitrogen. NMR spectroscopy showed the sample to consist of ca. 70% 2,5-dihydrothiophene and ca. 30% chloroform. Volumes used were adjusted for actual 2,5-dihydrothiophene content.

3.3.3 Synthesis of Mo(tfd)2(dht)2

The procedure for Mo(tfd)2(tht)2 was repeated but using 52.4 mg (76.1 μmol) of Mo(tfd)2(bdt), 50 μL of the 2,5-dihydrothiophene solution, and 50 μL (224 μmol) of btmsa. The crystals were dried in vacuo to give Mo(tfd)2(dht)2. (24.4 mg, 33.8 μmol, or 44%). Crystals were stored under

40 inert atmosphere and were stable at room temperature. X-ray quality crystals were obtained directly from the hexane solution. Representative NMR spectra can be found in Figure 3.9, 1 Figure 3.10, and Figure 3.11 in the appendix. H NMR (400 MHz, CDCl3) δ 4.35 (m, 8H, (CH2) 19 x 4), 5.99 (m, 4H, (CH) x 4). F NMR (376 MHz, CDCl3) δ –54.96 (s, 12F, (CF3)2 x 2) . 13 1 C{ H} NMR (100 MHz, CDCl3) δ 48.78 (s, CH2), δ 126.62 (s, CH). Anal. Calc. for

C16H16F12S6Mo (720.56): C 26.67, H 1.68, S 26.70; Found: C. 26.89, H 1.49, S 26.04.

3.3.4 Attempted synthesis of Mo(tfd)2(thiophene)2

2 mg of Mo(tfd)2(bdt) were placed into a NMR tube with 500 μL of C6D6, ca. 10 μL of btmsa, ca. 5 μL of thiophene and ca. 3 μL of 3,5-bis(trifluoromethyl)bromobenzene as an internal standard. After two hours only unreacted Mo(tfd)2(bdt) and btmsa along with a smaller quantity of Mo(tfd)2[bdt(btmsa)] adduct (20% yield relative to unreacted Mo(tfd)2(bdt)) were identified by NMR spectroscopy and integration. No new products could be detected. The procedure for

Mo(tfd)2(tht)2 was also repeated but substituting tht with thiophene. After 24 hours the solution remained green and no crystals were observed.

3.3.5 Thermal decomposition of Mo(tfd)2(dht)2

2.5 mg of Mo(tfd)2(dht)2 (3.5 μmol) were placed in a NMR tube with 600 μL of CDCl3 and ca. 5

μL of 3,5-bis(trifluoromethyl)bromobenzene as an internal standard and heated to 60 °C for 20 h.

Thiophene (12% yield relative to starting Mo(tfd)2(dht)2) was identified in the products using NMR spectroscopy and integration.

3.3.6 Non-catalytic dehydrogenation of dht with Mo(tfd)2(dht)2

0.7 mg of Mo(tfd)2(dht)2 (0.967 μmol) were placed in a NMR tube with 600 μL of a 0.062 M solution of dht in CDCl3 along with about 5 μL of 3,5-bis(trifluoromethyl)bromobenzene as an internal standard. The tube was heated to 60 °C for 24 h and then to 120°C for 5 hours.

Thiophene was identified by NMR spectroscopy in 43% yield based on the Mo(tfd)2(dht)2 reagent.

3.3.7 Hydrogen reactivity

Both the thermal decomposition and the non-catalytic dehydrogenation experiments were repeated but before heating the NMR tubes were thoroughly degassed by freeze-pump-thaw cycling. Hydrogen at atmospheric pressure was admitted to the tubes (ca. 1 mL) and then sealed. Heating proceeded as described previously. No new products or significant changes in product distribution were observed by NMR spectroscopy.

3.3.8 Isomerization of 1,4-cyclohexadiene

0.5 mg of Mo(tfd)2(tht)2 (0.69 μmol) and 5 μL of 1,4-cyclohexadiene (52.9 μmol) were added to 1 500 μL of C6D12 in a NMR tube and sealed. The contents were heated to 110 °C for 24 h. H NMR spectroscopy revealed isomerization to 1,3-cyclohexadiene (35% by NMR integration).

3.3.9 Reaction with carbon monoxide to produce Mo(tfd)2(dht)(CO)

2 mg of Mo(tfd)2(dht)2 (2.77 μmol) were added to 400 μL of C6D6 in a NMR tube along with ca. 5 μL of 3,5-bis(trifluoromethyl)bromobenzene as an internal standard. Carbon monoxide gas was gently bubbled into the solution for a total of 4 h with occasional refilling of solvent to make up for evaporative losses. The yellowish-brown solution turned bright orange and a new complex 19 was identified by F NMR spectroscopy and integration (88% yield relative to Mo(tfd)2(dht)2). 1H NMR spectroscopy and integration showed an equal amount of free dht to bound dht on this complex (excluding dht bound to known Mo(tfd)2(dht)2). The sample was degassed by freeze- pump-thaw cycling, and 300 μL of a 0.067 M solution of dht in C6D6 (20 μmol) were added, followed by shaking. After 2 h, NMR integration revealed 85% Mo(tfd)2(dht)2 and 15% 1 Mo(tfd)2(dht)(CO), indicating that the equilibrium shifted back to starting materials. H NMR

19 (400 MHz, C6D6) δ 2.66 (m, 8H, (CH2) x 4), 4.60 (m, 4H, (CH) x 4). F NMR (376 MHz, C6D6)

δ –54.92 (s, 12F, (CF3)2 x 2)

3.3.10 Determination of Keq (preferred binding of tht over dht)

A stock solution of dht in CDCl3 was made by adding 200 μL of diallyl sulfide to 16 mL of

CDCl3 and 10 mg of Grubb’s catalyst. After 24 h the reaction was checked for completion by NMR spectroscopy and the volatiles were purified by vacuum transfer. Stock solution concentration was approximately 0.097 M. The method used for equilibrium constant determination is based on relative ratios (by NMR integration) and not on absolute amounts. 2 mg of Mo(tfd)2(tht)2 were added to 400 μL of CDCl3 in a NMR tube. NMR spectra were taken at

22 °C. After each spectrum, a small amount of dht in CDCl3 was added. The ratios of the various species were determined by NMR integration: Mo-bound tht (“Mo-tht”), Mo-bound dht (“Mo- dht”), free dht and free tht were independently observed and quantified by NMR integration. An excellent fit was obtained for the simple model Mo-dht + tht ⇌ Mo-tht + dht, demonstrating that the two labile sites at Mo have very similar preference for binding tht versus dht, i. e. no cooperativity is observed. The equilibrium constant (for the reaction as written, Mo-tht as the product of the reaction) is equal to ([Mo-tht]/[Mo-dht])([tht]/[dht])-1, and a straight line was indeed obtained from plotting the ratio [Mo-tht]/[Mo-dht] against the ratio [tht]/[dht]) (Figure

3.5). The slope yields Keq = 6.5(5); tht binds more strongly than dht by that factor. A similar experiment was performed with thiophene (NMR, 20 °C). No substitution was observed at any concentration including neat thiophene.

3.4 Results and Discussion

The syntheses of our structural models exploited the ligand-based reactivity of Mo(tfd)2(bdt): treatment of the tris(dithiolene) with bis(trimethylsilyl)acetylene (btmsa)26 gave

Mo(tfd)2[bdt(btmsa)], which was subjected to ligand substitution with excess dht or tht (dht =

2,5-dihydrothiophene; tht = tetrahydrothiophene), yielding Mo(tfd)2(dht)2 or Mo(tfd)2(tht)2 respectively, upon loss of the metal-coordinated benzodithiin (Scheme 3.2, see Experimental

Section for details). The resulting complexes were characterized by multinuclear NMR spectroscopy, elemental analysis and X-ray crystallography. However this approach failed to produce any complexes with thiophene ligand.

Mo(tfd)2(bdt)

S S

Mo S S R = SiMe3 S S F3C CF3 CF3 F3C excess RC CR (btmsa) excess R R S S S S S Mo Mo S S S S S S S R S S F3C CF3 F3C CF3 CF3 F3C S R CF3 F3C

Mo(tfd)2[bdt(btmsa)] Mo(tfd)2(dht)2

Scheme 3.2: Syntheses of Mo(tfd)2(dht)2 from Mo(tfd)2(bdt). An analogous process may be used for synthesis of Mo(tfd)2(tht)2 by substituting dht with tht.

Figure 3.2: Top: Structure of Mo(tfd)2(dht)2. Bottom: Structure of Mo(tfd)2(tht)2. Non-hydrogen atoms are displayed using 30 % thermal ellipsoids. Hydrogen atom positions are calculated.

Mo(tfd)2(dht)2 Mo(tfd)2(tht)2

Empirical formula C16 H12 F12 Mo1 S6 C16 H16 F12 Mo1 S6 Formula mass 720.56 724.59 Crystal size [mm] 0.08 x 0.06 x 0.02 0.24 x 0.16 x 0.10 Crystal system triclinic Monoclinic

Space group P-1 P21/c a [Å] 8.3184(3) 12.0316(5) b [Å] 12.1554(8) 25.0104(6) c [Å] 12.9245(9) 8.4960(4) α [°] 76.092(3) 90 β [°] 90.074(4) 109.275(1) γ [°] 69.878(3) 90 V [Å3] 1186.03(12) 2413.27(16) Z 2 4 T [K] 150(2) 150(1) λ [Å] 0.71073 0.71073 F(000) 708 1432 Data/restraints/parameters 4826 / 0 / 317 5440 / 0 / 316 -3 dcalcd. [mg m ] 2.018 1.994 μ [mm-1] 1.181 1.161 Reflections (collected) 7732 13879 Reflections (unique) 4826 5440 GOF 1.164 1.041 R1 [I >2σ(I)] 0.0553 0.0516 wR2 [all data] 0.1465 0.1358

Table 3.1: Crystal data and structure refinement for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2.

Figure 3.2 shows the crystallographically determined structures of Mo(tfd)2(dht)2 and

Mo(tfd)2(tht)2. Crystallographic data are summarized in Table 3.1. The molecular structures of

Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 (Figure 3.2) are, expectedly, quite similar. In each case, the molybdenum centers are coordinated by six sulfur atoms. Two adjacent (cis) coordination sites are occupied by thioether ligands (dht or tht); these labile sites will possibly allow access to

mechanisms utilizing two coordination sites on the same Mo atom. In Mo(tfd)2(dht)2, the π- bonds of the dht ligands do not interact with the metal (i.e., dht binds in an η1-S fashion), consistent with experimental (NEXAFS) data for dht adsorbed on a sulfided molybdenum 27 19 surface. NMR spectroscopy ( F) indicates apparent C2v symmetry for Mo(tfd)2(dht)2 and

Mo(tfd)2(tht)2, indicating interconversion between ring conformers in solution. The crystal structure determinations for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 have yielded detailed information on the sulfur environment in such species, which is very consistent across the two structures. Key structural data are summarized in Table 3.2. The Mo-S(tfd) bond lengths (Table 3.2) for

Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 (averages 2.335(1) Å and 2.333(1) Å, respectively) are 25 28 marginally shorter than the analogous Mo-S(tfd) bonds in Mo(bdt)2(tfd) and Mo(tfd)3 (averages 2.367(1) Å and 2.355(4) Å, respectively). Particularly striking is the fact that the thioether ligands are apparently very weakly bonded, with Mo-S bond lengths that are much longer, by almost 0.2 Å, than the Mo-S(tfd) bonds.

Mo(tfd)2(dht)2 Mo(tfd)2(tht)2 Bond Distances (Å) C-C(thioether) 1.347(12), C10-C11 1.481(8), C10-C11 1.326(11), C14-C15 1.502(7), C14-C15 Mo-S(tfd) 2.336(2), Mo1-S1 2.331(1), Mo1-S1 2.329(2), Mo1-S2 2.336(1), Mo1-S2 2.337(2), Mo1-S3 2.324(1), Mo1-S3 2.337(2), Mo1-S4 2.343(1), Mo1-S4 Mo-S(thioether) 2.520(2), Mo1-S5 2.513(1), Mo1-S5 2.520(2), Mo1-S6 2.523(1), Mo1-S6

Bond Angles (degrees) S-Mo-S for tfd 81.28(6), S1-Mo1-S2 81.50(5), S1-Mo1-S2 81.76(6), S3-Mo1-S4 81.55(5), S3-Mo1-S4 S-Mo-S for thioether 75.20(6), S5-Mo1-S6 73.85(4), S5-Mo1-S6 S-Mo-S for trans S 142.11(6), S1-Mo1-S4 139.92(5) S3 Mo1 S2 127.77(6), S1-Mo1-S5 138.83(5) S1-Mo1-S4 137.21(6), S3-Mo1-S2 130.58(5) S3-Mo1-S5 136.69(6), S3-Mo1-S5 133.30(5) S1-Mo1-S5 134.27(6), S6-Mo1-S2 128.73(5) S2-Mo1-S6 129.85(6), S6-Mo1-S4 135.44(5) S4-Mo1-S6

Table 3.2: Selected structural data for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2.

Mo(tfd)2(dht)2 Mo(tfd)2(tht)2 Non-bonded distances (Å) S-S in trigonal face inter-tfd 3.170(2), S1-S3 3.117(2), S1-S3 3.132(2), S2-S4 3.167(2), S2-S4 tfd-thioether 3.239(2), S3-S6 3.277(2), S3-S6 3.233(2), S1-S6 3.151(2), S1-S6 3.150(2) S2-S5 3.268(2) S2-S5 3.242(2) S4-S5 3.232(2) S4-S5 S-S along prism edge inter-tfd 3.038(2), S1-S2 3.047(2), S1-S2 3.059(2), S3-S4 3.048(2), S3-S4 inter-thioether 3.075(2), S5-S6 3.026(2), S5-S6

Non-bonded angles (trigonal face, degrees) S-S-S 60.77(6), S3-S1-S6 63.04(5), S3-S1-S6 60.58(6), S1-S3-S6 58.99(5), S1-S3-S6 58.65, (6) S1-S6-S3 57.97(5), S1-S6-S3 62.14(6), S4-S2-S5 60.29(5), S4-S2-S5 59.21(6), S2-S4-S5 61.4(5), S2-S4-S5 58.64(6), S2-S5-S4 58.3(5), S2-S5-S4

S-S-S to S-S-S interplanar 0.59(11), S1-S3-S6 to S2-S4- 0.57(8), S1-S3-S6 to S2-S4- angle S5 S5

Table 3.3: (Continued) Selected structural data for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2.

There are few examples of structurally characterized Mo-thioether complexes available for comparison. The closest one is the ethylene adduct [Mo(tfd)2{bdt(CH2CH2)}], with an average distance of 2.523 Å to the sulfurs on the dihydrobenzodithiin ligand.25 The closest non- dithiolene-based Mo(IV) system is a octahedral molybdenum ‘S4-crown’ thioether complex, with slightly shorter Mo-S(thioether) distances (2.469 - 2.498 Å).29

Computed Mo-S distances for dht and tht bound to a Mo3S9 catalyst model fragment are 2.53 Å 30 and 2.50 Å respectively. Thus, Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 can be regarded structurally extremely close to the active sites they are modeling .

SSS (ca. 60o) a) R2S SR2 SMStrans b) o S S (ca. 136 ) S Mo S S S S Mo F3C CF CF3 S S 3 F3C S

Figure 3.3: Bond angles used for analysis; a) example for SMStrans angle. b) example for SSS (nonbonding) angle.

In both compounds, the coordination geometry at molybdenum is very close to trigonal prismatic. Figure 3.3 shows the bond angles used to corroborate this statement. Using the three largest SMStrans angles (Figure 3.3a) to position the structures on the scale ranging from perfect 31 octahedron (0%) to perfect trigonal prism (100%), Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 were found to have 92% and 94% trigonal prismatic character, respectively.32 The largest deviations33 in SMStrans angles were ca. 14° (for Mo(tfd)2(dht)2) and ca. 11° (for Mo(tfd)2(tht)2). The triangular faces of the trigonal prisms contain SSS (formally non-bonded) angles very close to 60° (Figure 3.3b), with the largest deviation being ca. 3° for in both cases (Table 3.2). This observation is surprising insofar as the large difference between Mo-S(tfd) and Mo-S(thioether) bond lengths could be expected to lead to a more distorted MoS6 substructure.

Trigonal prismatic geometry, which maximizes ligand-metal and interligand π interactions,34 is quite common for high-valent (oxidized) metal tris(dithiolene) (e.g., neutral group six tris(dithiolene)s). Reduced analogues, on the other hand, usually exhibit distortion toward octahedral geometries, thereby relieving ligand-ligand repulsion in these comparatively electron- rich systems. To illustrate, the metal geometry in neutral Mo(tfd)3 is very close to trigonal 28 32 2- prismatic (99 % using the SMStrans angle criterion, see above) , while dianionic [Mo(tfd)3]

50 has considerably more octahedral character28 (70% trigonal prismatic).33 In general, trigonal prismatic structures are more likely for complexes with low d electron counts (i.e., dn, n ≤ 2).35 36 This has been explicitly discussed for MoS2 using crystal field theory. Analogous effects have 37 been shown computationally and experimentally for Mo(butadiene)3 complexes and n computationally for M(CH3)6 complexes (M= Ti, Zr, Hf, V, Nb, Ta, Cr, Mo, W, Tc, Re, Ru, Os; 38 n= -2,-1,0,+1) . The two complexes synthesized for the present study – Mo(tfd)2(dht)2 and 2 Mo(tfd)2(tht)2 – are charge-neutral and formally Mo(IV) (d ) species, with two ene-dithiolate donors and two neutral thioether ligands. Similarly, neutral group VI metal tris(dithiolene)s, 2 such as Mo(tfd)3, can be reasonably formulated as a d species, with two dianionic dithiolene ligands and one neutral, weakly donating dithioketone ligand.39 The trigonal prismatic geometries observed at the metals in Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 can, therefore, be rationalized on the basis of low d electron counts and neutral charge states for both complexes, with analogy to group VI metal tris(dithiolene)s. The observed trigonal prismatic geometries mimic the environment of the Mo centers in molybdenum(IV) sulfide.

The S-Mo-S angles involving the thioether ligands are very acute (75.2° for Mo(tfd)2(dht)2 and

73.9° for Mo(tfd)2(tht)2). The S-Mo-S angles in the dithiolene chelate rings are much wider (ca. 82 o) for both complexes. Interestingly, there are many examples in the literature of Mo (or W) bis(dithiolene) complexes with various non-dithiolene ligands that also exhibit this acute non- dithiolene ligand angle and wider dithiolene ligand bite angles. Relevant examples are summarized in Table 3.4. Despite a large variation in non-dithiolene ligands (which are of variable steric demand), the non-dithiolene and the dithiolene bite angles remain remarkably consistent.

Substance ic t (degrees) (%) 33 32 r on-dithiolene bite angle ithiolene bite angle eviation ef. rigonal prisma haracte (average) N D T c D R

[Et4N]2[Mo2(SCH2CH2OH)2(mnt)4] 68 83 82 25 [16]

[Et4N]2[Mo2(SPh)2(mnt)4] 67 82 96 12 [16]

[Et4N]2[W2(SCH2Ph)2(mnt)4] 68 83 95 10 [16]

[Et4N]2[W2(SCH2CH3)2(mnt)4] 67 82 87 21 [16]

[Et4N]2[W2(SCH2CH2OH)2(mnt)4] 68 83 92 18 [16]

[Et4N]2[W2(SPh)2(mnt)4] 66 82 93 14 [16]

[Et4N]2[Mo2(SCH2Ph)2(mnt)4] 67 83 91 20 [16]

[Et4N]2[Mo2(SCH2CH3)2(mnt)4] 68 82 86 22 [16]

[Et4N]2[Mo(NCS)2(mnt)2] 66 86 88 23 [17]

[Ph4P]2[W(SPh)2(mnt)2]·0.5 (CH3)2CHOH 72 82 98 7 [16]

[Et4N][Mo(PPh3)(NCS)(mnt)2] 76 83 83 21 [17]

[Et4N][Mo(PPh3)(SC6H4-4-Me)(mnt)2] 77 82 93 10 [17]

[Ph3PNPPh3][Mo(PPh3)(SC6H4-2-COOH)(mnt)2] 76 82 73 29 [17] . [Et4N][Mo(SPh)(PPh3)(mnt)2] CH2Cl2 77 82 90 17 [18] . [PPh4][Mo(PPh3)(SCH2CH3) (mnt)2] CH2Cl2 76 83 74 28 [19]

[PPh4][Mo(PPh3)(SCH2Ph)(mnt)2] 77 81 94 14 [19]

[Et4N][Mo(PPh3)(Br)(mnt)2] 78 83 81 23 [19]

[Mo(tfd)2{bdt(CH2CH2)}] 72 82 86 20 [25]

Mo(tfd)2(dht)2 75 82 92 14 this work

Mo(tfd)2(tht)2 74 82 94 11 this work

Table 3.4: A comparative sampling of bite angles (°) in bis-dithiolenes with various ligands (mnt

= S2C2(CN)2)

This narrow non-dithiolene ligand angle is intriguing. A similar observation was made by Holm 40 and co-workers when narrow C-M-C bond angles for M(CO)2(S2C2Me2)2 were observed: 83.5° for M = Mo and 84.1° for M = W. These bond angles narrowed even more upon stepwise reduction (monoanion, dianion), down to 72.3° for the dianions. CO is, of course, a very special ligand due to its π-acceptor properties, and Fomitchev, Lim and Holm40 concluded that the observed effect is “suggestive of a weak bonding interaction between the carbonyl ligands”.

While the narrow bite angle described above is a deviation from perfect trigonal prismatic geometry, the ‘trigonal’ planes in Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 are almost perfectly parallel.

If the S1-S3-S6 and S2-S4-S5 faces of both complexes are taken to define planes, the angle between the two planes is 0.59° and 0.57° for Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2, respectively (i.e., essentially parallel as in a true triangular prism). In each triangular face, the sulfurs are essentially closest packed. A diagram showing van-der-Waals spheres for the sulfur atoms of

Mo(tfd)2(dht)2 is shown in Figure 3.4 (the structure of the tht complex leads to a virtually identical picture). Visually, the impression of a very symmetrical arrangement of closest packed sulfur atoms (Figure 3.4, A) is striking, which is quantitatively re-enforced by the very consistent non-bonded distances (Table 3.2).

Figure 3.4: View of the sulfur environment of Mo(tfd)2(dht)2, using van-der-Waals radii for sulfur atoms; A: view perpendicular to S1-S3-S6 “close-packed” plane; B: view onto S5-S6 edge.

While the preference for trigonal prismatic (instead of octahedral) geometry is electronic in origin (as is the acuteness of one S-Mo-S angle), the resulting structure is surprisingly consistent with a closest-packing model in which the two closest-packed ‘S3’ triangles are stacked on top of another and in which the central Mo is dislocated from the center toward the more donating sulfurs. In both Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2, the sulfur atoms are found in a geometric arrangement very similar to what is observed for molybdenum inside solid MoS2, including overlapping Van der Waals radii for adjacent sulfur centers.

3.5 Reactivity

The new complexes showed some interesting and promising reactivity at slightly elevated temperature (60-120 °C), which, unfortunately, was accompanied by decomposition.

Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 were stable for days at room temperature in non-polar/non- coordinating solvents (e.g., C6D6), but rapidly reacted with coordinating solvents such as THF and acetonitrile (color change to green or red respectively observed immediately upon dissolution ). Both complexes were somewhat unstable in chloroform, decaying within a few days at room temperature, and within 1 hour at 60 °C. The decomposition products were 19 1 identified by F/ H NMR as Mo(tfd)3 and free thioethers (dht or tht). Presumably molybdenum metal was also deposited, as faint darkening of the NMR tubes could be observed; this film was insoluble in organic solvents but dissolved in inorganic acids. The complexes were considerably more stable in C6D6: only minor decomposition was observed after heating either complex to 60°C for 1 day.

The complexes were sufficiently stable (in CDCl3) to determine the relative binding constant of dht versus tht with respect to coordination to the Mo(tfd)2 fragment. We found that tht binds more strongly by a factor of 6.5(5) at a temperature of ca. 22 °C (Figure 3.5). The lesser binding affinity for dht, relative to tht, is likely due to electron-withdrawing alkene group in the dht case, which can be expected to decrease the σ-donor ability of the thioether ligand. Substitution of dht with thiophene was not observed at any concentration level of thiophene in CDCl3 or with neat thiophene. These results can be compared to a recent computational paper by Joshi et al. that 41 calculates the adsorption energies of various sulfur-containing molecules onto MoS2. The

54 lowest binding energies for an unsaturated molybdenum edge structure with tht and dht were found to be –1.67 eV and –1.60 eV respectively, corresponding to an energy difference of –6.75 kJ/mol. Our experimentally determined equilibrium constant corresponds to a difference in binding energy of 4.6(2) kJ/mol (for tht versus dht), quite consistent with the calculated value. Joshi et al. also calculated the binding energy of thiophene to be –1.04 eV, compared with –1.60 eV for dht (difference: –0.56 eV or –54 kJ/mol). This energy difference would yield an equilibrium constant (at 22 °C) of 3.7 x109 in favor of dht binding and thus explains the inability of thiophene to displace dht.

Figure 3.5: Determination of the equilibrium constant for the binding of tht versus dht at 22 °C in CDCl3, according to the equation: Mo-dht + tht ⇌ Mo-tht + dht. The model, predicting a

55 linear dependence of the ratio [Mo-tht]/[Mo-dht] versus the ratio of free thioethers ([tht]/[dht]) fits the data very well (R2 = 0.9991). No cooperativity is observed, and the two labile sites behave as independent in this process. The slope yields Keq = 6.5 ± 0.5 (tht binds more strongly, by a factor of close to seven, than dht). See Experimental Section for details.

When Mo(tfd)2(dht)2 was heated in chloroform-d (60 °C, 1 h), thiophene (12%, relative to original concentration) was identified by 1H NMR spectroscopy as one of the decomposition products. Mo(tfd)3 was also observed, as well as a smaller percentage of unidentified decomposition products (by 19F NMR), and a faint darkening of the nmr tube, presumably molybdenum metal (as mentioned earlier). It therefore appears that the dht ligands in the

Mo(tfd)2(dht)2 complex were dehydrogenated to thiophene, although free hydrogen gas was not observed in the 1H NMR spectrum. These reactions were not catalytic when excess dht was present. At the temperatures accessible for this system (before decomposition occurs) we have not been able to reproducibly demonstrate desulfurization. Butadiene, which would be expected to form upon desulfurization of dht (Scheme 3.1), was not observed in any of the reactions we screened. Performing the same reaction after an atmosphere of hydrogen was sealed into the tubes produced no change in the product type or distribution. All hydrogen reactions where done under one atmosphere of pressure; greater pressures may be needed to observe any reactivity. Many proposed HDS mechanisms invoke the action of the surrounding sulfur atoms42,43 and/or 44,45 multiple metal centers in MoS2 to facilitate the hydrodesulfurization of thiophene. Such pathways may not be accessible at the relatively low temperatures we tried (60-120 °C, as necessitated by the temperature-sensitivity of our model complexes; cf. 300-400 °C for industrial HDS processes2). It will be the goal of future studies (with more temperature-stable analogs) to investigate whether a HDS cycle can be achieved with a single molybdenum center or whether the participation of more than one metal is required.45 Nonetheless, the observation of the “reverse” reaction to hydrodesulfurization (dehydrogenation of dht to thiophene), shows that on the principle of microreversibility Mo(tfd)2(dht)2 should be capable of the forward reaction if not for its instability. The pursuit of more stable analogues based on its structure is promising.

We observed the isomerization of 1,4-cyclohexadiene to 1,3-cyclohexadiene at 110 °C in cyclohexane-d12 with Mo(tfd)2(tht)2 present (presumably this reaction would also occur with

Mo(tfd)2(dht)2 but this was not tested). Performing the same reaction using Mo(tfd)2(bdt) did not yield any isomerized diene.

Finally, we would like to address the question of how the lability of carbon monoxide (which binds weakly to molybdenum bis(dithiolene)s, see Introduction) relates to the lability of thioethers. When a slow stream of carbon monoxide is bubbled through a solution of 1 19 Mo(tfd)2(dht)2 in C6D6 for a total of 4 h, a new complex is observed by H and F NMR spectroscopy (ca. 70-90% conversion, as determined by 19F NMR, using 3,5- bis(trifluoromethyl)bromobenzene internal standard). The new species is assigned as

Mo(tfd)2(dht)(CO): NMR integration of dht protons yields that only one dht molecule is bound per molybdenum. In addition, the amount of free dht observed matches the dht bound to the new complex, in accord with expectations from mass balance. When a large excess of dht was added, the equilibrium shifted back to Mo(tfd)2(dht)2. Thus, carbon monoxide and thioethers have similar binding strength when coordination to Mo(tfd)2 fragments is involved. Some metal bis(dithiolene) complexes containing carbonyls are already known,20,24 and it may become useful knowledge for future syntheses that CO and thioethers can be of comparable lability in such systems.

3.6 Conclusions

The molecular compounds Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 were synthesized and characterized as structural models for the hydrodesulfurization catalyst molybdenum disulfide. X-ray crystallography shows that they are excellent structural models for the proposed active site of molybdenum disulfide. They both bind thioethers and dht binds in η1 fashion (rather than η3) that is consistent with how dht is observed to bind to MoS2 catalysts. They both have strongly trigonal prismatic character with the sulfurs in a closest-packing arrangement (also consistent with MoS2) despite large differences between the dithiolene and thioether bond lengths. Equilibrium studies shows the binding affinities of the tht and dht thioethers in the complexes to be consistent with computational models for thioether binding to MoS2, raising confidence in validity of those models. Preliminary investigations show some intriguing reactivity consistent with transfer dehydrogenation, yet temperature-sensitivity hampers full exploration of this avenue. Nonetheless the demonstration of dehydrogenation may suggest that hydrogenation should be possible if the principle of microreversibility is applied. It may be anticipated that

57 molecular compounds for actual HDS catalysis will have to be designed to be more robust at high-temperature conditions, which will likely necessitate departure from a close structural model. Additionally, it may be necessary to consider models containing more than one metal site (e.g., homo- or heterobimetallic complexes of Mo). Alternatively, tfd may not be the most appropriate ligand for these systems due to its electron-withdrawing nature. As HDS is an inherently reductive process more electron-donating dithiolenes may be necessary. Future developments include substituting such electron-donating dithiolenes for tfd and assessing their effects. Nonetheless, the observation of alkene isomerization does show the possibility of at least one type of catalysis. The lability of the ligands, as demonstrated by the equilibrium studies with tht, dht and CO, indicate the metal center is easily accessible under mild conditions. Taken together these two observations hint at the possibility of using this complex for other types of catalysis, not just for HDS.

3.7 Appendix

1 Figure 3.6: H NMR spectrum of Mo(tfd)2(tht)2 (400 MHz, CDCl3) δ 2.28 (m, 8H, tht), 3.54 (m, 8H, tht). b

S S a Mo S S S S F3C CF3 CF3 F3C

19 Figure 3.7: F NMR spectrum of Mo(tfd)2(tht)2 (376 MHz, CDCl3) δ –54.9 (s, 12F, (CF3)2 x 2) .

S S

Mo S S S S F3C CF3 CF3 F3C

13 1 Figure 3.8: C{ H} NMR spectrum of Mo(tfd)2(tht)2 (100 MHz, CDCl3) δ 30.109 (s, Cβ, tht), δ

42.31 (s, Cα, tht). b

S S a Mo S S S S F3C CF3 CF3 F3C

Figure 3.9: 1H NMR spectrum of Mo(tfd)2(dht)2 (400 MHz, CDCl3) δ 4.35 (m, 8H, (CH2) x 4), 5.99 (m, 4H, (CH) x 4).

S S a Mo S S S S F3C CF3 CF3 F3C

19 Figure 3.10: F NMR spectrum of Mo(tfd)2(dht)2 (376 MHz, CDCl3) δ –54.96 (s, 12F, (CF3)2 x 2) . S S

Mo S S S S F3C CF3 CF3 F3C

13 Figure 3.11: C NMR spectrum of Mo(tfd)2(dht)2 (100 MHz, CDCl3) δ 48.78 (s, CH2), δ 126.62 (s, CH). b

S S a Mo S S S S F3C CF3 CF3 F3C

References

1 T. V. Choudhary, S. Parrott, B. Johnson, Environ. Sci. Technol. 2008, 42, 1944-1947.

2 C. Song, Catal. Today 2003, 86, 211-263.

3 I.V. Babich, J.A. Moulijn, Fuel 2003, 82, 607-631.

4 Y. Sun, R. Prins, Angew. Chem. Int. Ed. 2008, 47, 8478-8481.

5 B. Hinnemann, P.G. Moses, J.K. Nørskov: J. Phys.: Condens. Matter 2008, 20, 064236- 064244.

6 P. Raybaud, Appl. Catal. A: Gen. 2007, 322, 76-91.

7 M. Sun, J. Adjaye, A.E. Nelson, Appl. Catal. A: Gen. 2004, 263, 131-143.

8 M.-G. Choi, L. M. Daniels, R. J. Angelici Inorg. Chem. 1991, 30, 3647-3651.

9 P. G. Moses, B. Hinnemann, H. Topsøe, J. K. Nørskov J. Catal. 2007, 248, 188-203.

10 This simple picture is based on the structure of bulk material and neglects surface reconstruction.

11 ORTEP-3 for Windows, version 1.08 (based on Johnson, C. K.; Burnett, M. N. ORTEP-3 Version 1.0.3): L. J. Farrugia, J. Appl. Cryst. 1997, 30, 565.

12 a) B. Schönfeld, J. J. Huang, S. C. Moss, Acta Cryst. B 1983, 39, 404-407. b) R. G. Dickinson, L. Pauling, J. Am. Chem. Soc. 1923, 45, 1466-1471.

13 D. L. Sullivan, J. G. Ekerdt J. Catal. 1998, 178, 226-233.

14 Zonnevylle, M. C.; Hoffmann, R.; Harris, S. Surf. Sci. 1988, 199, 320-360

15 D. Sellmann, B. Hadawi, F. Knoch, M. Moll, Inorg. Chem. 1995, 34, 5963-5972.

16 A. Majumdar, K. Pal, K. Nagarajan, S. Sarkar, Inorg. Chem. 2007, 46, 6136-6147.

17 A. Majumdar, K. Pal, S. Sarkar, Dalton Trans. 2009, 1927-1938.

18 A. Majumdar, K. Pal, S. Sarkar, J. Am. Chem. Soc. 2006, 128, 4196-4197.

19 A. Majumdar, K. Pal, S. Sarkar, Inorg. Chem. 2008, 47, 3393-3401.

20 S. Groysman, R. H. Holm, Inorg. Chem. 2007, 46, 4090-4102.

21 E. Roland, E. C.Walborsky, J. C. Dewan, R. R. Schrock, J. Am. Chem. Soc. 1985, 107, 5795- 5797.

22 H. Sugimoto, T. Sakurai, H. Miyake, K. Tanaka, H. Tsukube, Inorg. Chem. 2005, 44, 6927– 6929.

23 G. N. Schrauzer, H. N. Rabinowitz, J. Am. Chem. Soc. 1969, 91, 6522-6524.

24 B. S. Lim, J. P. Donahue, R. H. Holm, Inorg. Chem. 2000, 39, 263–273.

25 D. J. Harrison, A. J. Lough, N. Nguyen , U. Fekl, Angew. Chem. Int. Ed. 2007, 46, 7644-7647.

26 A variety of alkenes may also be used to labilize the bdt ligand. See reference 25 and D. J. Harrison, U. Fekl, Chem. Commun. 2009, 7572-7574.

27 H. Xu, P. Uvdal, C. M. Friend, J. Stöhr, Surface Science 1993, 289, L599-L603.

28 K. Wang, J. M. McConnachie, E. I. Stiefel, Inorg. Chem. 1999, 38, 4334-4341.

29 T. Yoshida, T. Adachi, K. Matsumura, K. Kawazu, K. Baba Chem. Lett. 1991, 20, 1067-1070.

30 X.-Q. Yao, Y.-W. Li, H. Jiao, J. Mol. Struct.: Theochem. 2005, 726, 81-92.

31 C. L. Beswick, J. M. Schulman, E. I. Stiefel Prog. Inorg. Chem. 2004, 52, 55-110.

32 The degree of trigonal prismatic character by the SMStrans criterion is determined by first taking the individual “corner” sulfur atoms on one triangular face of a trigonal prism and then measuring the S-M-S angles to all the sulfur atoms on the opposite triangular face. Nine angles are obtained this way. The average of the three largest angles is used to determine the degree of octahedral vs. trigonal prismatic character using SMStrans. The SMStrans values for a perfect octahedron and a perfect trigonal prism are 180o and 136o, respectively. Small chelate bite angles may constrain SMStrans from reaching the octahedral limit, so a correction is applied (εcorr). εcorr is o found by εcorr = 90 + SMSintrda where SMSintra is the average of the three smallest angles S-M-S angles that correspond to the sulfurs on the same chelating ligand (chelate ‘bite’). Finally the degree of prismatic character on the scale ranging from perfect octahedron (0%) to perfect o o trigonal prism (100 % TP) is found by using % TP = [1 - (SMStrans -136 )/(ε corr -136 )] x 100 %. This criterion is paraphrased from [31] but inversely scaled so that trigonal prismatic character is 100 % while octahedral is 0 %.

33 For some sense of distortion, the largest deviation is found by removing the SMSintra angles from the previous set of nine angles found in [32] and taking the difference between the smallest and largest angles in the remaining six.

34 M. L. Kirk, R. L. McNaughton, M. E. Helton Prog. Inorg. Chem. 2004, 52, 111-212.

35 For complexes with purely σ donor ligands, the t1u orbitals (in an octahedral field) experience net stabilization upon distortion to trigonal prismatic geometry, while the t2g orbitals (in an octahedral field) are destabilized. See: S. K. Kang, H. Tang, T. A. Albright, J. Am. Chem. Soc. 1993, 115, 1971-1981. Thus, for complexes with higher d electron counts (dn, n > 2), the stabilization of the t1u orbitals is off-set by the destabilization of electrons in the t2g orbitals, usually favoring octahedral geometry.

36 R. Huisman, R. De Jonge, C. Haas, F. Jellinek, J. Solid State Chem. 1971, 3, 56-66.

37 M. Kaupp, T. Kopf, A. Murso, D. Stalke, C. Strohmann, J. R. Hanks, F. G. N. Cloke, P. B. Hitchcock, Organometallics 2002, 21, 5021-5028.

38 M. Kaupp, Chem. Eur. J. 1998, 4, 1678-1686.

39 A variety of resonance forms can be envisioned for neutral group VI metal tris(dithiolene)s: the most reasonable contributors, in light of computational and spectroscopic evidence, have metal oxidation states of IV or V (singlet diradical character for the latter case). For early MO calculations on metal tris(dithiolene)s, see: a) E. I. Stiefel, R. Eisenberg, R. Rosenberg, H. B.Gray, J. Am. Chem. Soc. 1966, 88, 2956-2966. For more recent discussion concerning the electronic structures of metal tris(dithiolene)s, see: b) K. Ray, T. Petrenko, K. Wieghardt, F. Neese, Dalton Trans. 2007, 1552-1566. c) R. R. Kapre, E. Bothe, T. Weyhermüller, S. DeBeer George, K. Wieghardt, Inorg. Chem. 2007, 46, 5642-5650. d) A. L. Tenderholt, R. K. Szilagyi, R. H. Holm, K. O. Hodgson, B. Hedman, and E. I. Solomon Inorg. Chem. 2008, 47, 6382–6392.

40 D. V. Fomitchev, B. S. Lim, R. H. Holm, Inorg. Chem. 2001, 40, 645-654.

41 Y. V. Joshi, P. Ghosh, P. S. Venkataraman, W. N. Delgass, K. T. Thomson, J. Phys. Chem. C 2009, 113, 9698-9709.

42 H. Kwart, G.C.A. Schuit, B.C. Gates, J. Catal. 1980, 61, 128-134.

43 K.F. McCarty, G.L. Schrader, J. Catal. 1987, 103, 261-269.

44 R. A. Sánchez-Delgado (Ed.), “Organometallic Modeling of the Hydrodesulfurization and Hydrodenitrogenation Reactions”, Catalysis by Metal Complexes. 2004, vol. 24, p. 95-137.

45 T.M. Brunier, M.G.B. Drew, P.C.H. Mitchell, J. Chem. Soc. Faraday Trans. 1992, 88, 3225- 3232.

Chapter 4: A Structural Model for DMSO Reductase from Covalent Addition of Phosphine to Dithiolene in a Molybdenum Tris(dithiolene)

4.1 Abstract

Triphenylphosphine (PPh3) reversibly adds to the bdt ligand in the molybdenum tris(dithiolene) complex Mo(tfd)2(bdt) (tfd=S2C2(CF3)2; bdt=S2C6H4), turning chelating bdt into the monodentate zwitterionic ligand SC6H4SPPh3. A second PPh3 molecule fills the newly created open site in the crystallographically characterized product, Mo(tfd)2(SC6H4SPPh3)(PPh3). The complex is a structural model for DMSO (dimethyl sulfoxide) reductase.

4.2 Introduction

When transition metal complexes undergo reactions, the supporting ligands are normally “spectator ligands”. They control geometry and the electronic structure of the metal, whereas oxidation/reduction as well as bond-breaking and bond-making occur at the metal. In comparably rare cases, where frontier orbitals at the ligands are more accessible than metal orbitals, “non-innocent” behavior of the ligand is observed.

Redox chemistry that involves electron transfer by reducing the ligand has been known for 1 decades for dithiolene (S2C2R2) complexes of transition metals. Bond-making and bond- breaking reactivity at dithiolene ligands appears much less understood than electron transfer involving the ligands. While the sulfur centers of dithiolenes can be expected to be attacked by electrophiles (e. g., alkylating agents),2 and nucleophiles have been reported to attack other ligands (for example, nitrides),3 it is still very rare that a dithiolene ligand is attacked by a nucleophile other than an alkene.4,5 Alkene additions to the ligand6 have been reported for square-planar metal bis(dithiolene) complexes since the late sixties for strained alkenes,7 and since 2001 for simple unstrained alkenes.8 Then, in 2007 metal tris(dithiolene) complexes were also found to be reactive to simple unstrained alkenes.9 However, alkenes are often considered amphiphilic and are not clear-cut examples of nucleophiles. We now report the addition of the nucleophile triphenylphosphine (PPh3) to the bdt ligand in the molybdenum tris(dithiolene)

Mo(tfd)2(bdt) (tfd=S2C2(CF3)2, bdt=S2C6H4), creating the zwitterionic ligand SC6H4SPPh3. The product complex is a structural DMSO reductase model.

4.3 Experimental

Experiments were conducted under inert (N2 or Ar) atmosphere using standard glove-box (M

Braun UniLab) or Schlenk-type techniques except where noted. Mo(tfd)2(bdt) was prepared by literature procedures.9 All other chemicals were purchased from Sigma Aldrich or Alfa Aesar and used as received. Dichloromethane and hexanes were dried/deoxygenated using an M Braun solvent purification system (MB-SPS). Benzene (C6H6), dimethylsulfoxide (DMSO) and dichloromethane-d2 (CD2Cl2) were dried/degassed over molecular sieves (3A) before use. UV- vis spectra were obtained on a Cary 14 spectrophotometer. All NMR spectra were obtained on a Bruker 400 MHz spectrometer. Residual solvent proton peaks were used as reference: 1H (δ, 19 ppm, CD2Cl2, 5.32). F NMR data were referenced to external trifluoroacetic acid (δ, ppm, - 78.5). 31P NMR data were referenced to external phosphoric acid (δ, ppm, 0). 13C NMR could not be obtained due to poor solubility of compounds. Elemental analyses were conducted at Guelph (Chemisar) Laboratories, Guelph, ON, Canada.

Computational Details:

All calculations were performed using the GAMESS(US) Version 12 January R3. All molecular structures were optimized in the gas phase using the restricted DFT (B3LYP hybrid functional)10 method for spin singlet states, using chemically reasonable initial guesses for each structure. The following basis sets were used: 6-31G for H, 6-31G* for C, S, P, and F11 and LANL2DZ12 with effective core potential for Mo. The calculations of the structures containing P(C6H5)3 utilized the same basis set, but included Grimme's third-generation empirical dispersion correction.13

4.3.1 Synthesis of Compound 3 - Mo(tfd)2(SC6H4SPPh3)(PPh3) * 0.5 C6H6

Mo(tfd)2(bdt) (29.9 mg, 43.4 mmol) was dissolved in 2.8 mL of benzene to give a dark green solution. Triphenylphosphine (42.2 mg, 161 mmol) was dissolved separately in 1.6 mL of benzene. The solutions were mixed and the mixture immediately turned dark purple. The solution was left to stand under an inert atmosphere. After 1 day, magenta crystals of 3 had

70 formed and the supernatant turned dark red. The crystals were filtered off and washed 3 times with hexane. Crystals were dried under vacuum and stored under an inert atmosphere. X-ray quality crystals were obtained without applying vacuum. (38.9 mg, 31 mmol, 71% based on 1 19 Mo(tfd)2(bdt)). H NMR (400 MHz, CD2Cl2) δ = 7-7.8 (m, 34H, (CH) x 34), F NMR (376 31 MHz, CD2Cl2) δ –54.94 (s, 12F, (CF3)2 x 2), P NMR (162 MHz, CD2Cl2) δ 43.54 (s, 1P, PPh3),

54.14 (s, 1P, PPh3). Anal. Calc. for C53H37F12MoP2S6 (1252.07): C 50.84, H 2.98, P 4.95, S 15.37; Found: C. 50.59, H 2.62, P 4.33 S 14.87.

Representative NMR spectra for this compound can be found in Figure 4.9 -Figure 4.11 in the appendix. Crystal data are summarized in Table 4.1.

4.3.2 Synthesis of Mo(tfd)2(SC6H4SPPh3)(PPh3) * CHCl3

The procedure to synthesize compound 3 was repeated but replacing the benzene solvent with chloroform. X-ray quality crystals were obtained directly from the solution. Crystal data are summarized in Table 4.1.

Mo(tfd)2(SC6H4SPPh3)(PPh3) * Mo(tfd)2(SC6H4SPPh3)(PPh3) * 0.5 C6H6 CHCl3

Empirical formula C50H34F12MoP2S6, 0.5(C6H6) C50H34F12MoP2S6, CHCl3

Formula mass 1252.07 1332.38

Crystal size [mm] 0.22 x 0.07 x 0.02 0.18 x 0.06 x 0.06

Crystal system triclinic monoclinic

Space group P-1 P21/n a [Å] 11.8783(12) 10.4327(3) b [Å] 13.9884(17) 24.0859(13) c [Å] 17.2680(15) 21.8431(12)

α [°] 73.317(7) 90.00

β [°] 88.340(7) 90.154(3)

γ [°] 76.867(5) 90.00

V [Å3] 2674.5(5) 5488.7(5)

Z 2 4

T [K] 150(1) 150(1)

λ [Å] 0.71073 0.71073 F(000) 1262 2672 Data/restraints/parameters 8593 / 168 / 708 9635 / 0 / 676 -3 dcalcd. [mg m ] 1.555 1.612 μ [mm-1] 0.618 0.749 Reflections (collected) 21197 32783 Reflections (unique) 8593 9635 GOF 1.009 0.982 R1 [I >2σ(I)] 0.0795 0.0784 wR2 [all data] 0.1949 0.2759

Table 4.1: Crystal data and structure refinement for compound 3 as benzene solvate and as chloroform solvate.

4.3.3 Observation of Compound 2 - Mo(tfd)2(SC6H4SPPh3)

400 μL of a 1.22 mM solution of Mo(tfd)2(bdt) (0.488 μmol) were placed into a J.Young NMR tube along with 60 μL of a 12.6 mM solution of PPh3 (0.741 μmol). The molar ratio was approximately 1.52 mole equivalents of PPh3 to Mo(tfd)2(bdt). The solution was allowed to equilibrate for 10 minutes. New peaks in addition to the known peaks for 3 were observed: 1H 19 NMR (400 MHz, CD2Cl2) δ 7-7.8 (m, 19H, (CH) x 19), F NMR (376 MHz, CD2Cl2) δ –55.09 31 (s, 12F, (CF3)2 x 2), P NMR (162 MHz, CD2Cl2) δ 47.58 (s, 1P, PPh3).

Representative NMR spectra for this compound can be found in Figure 4.10 and Figure 4.11 in the appendix.

4.3.4 DFT-Optimized Structure of 2 and its Analogues

DFT computations were performed of P(C6H5)3 binding to Mo(tfd)2(bdt) (see Figure 4.1 below). Binding of the phosphine to the bdt ligand resulted in the formation of a stable 5-coordinate, square pyramidal structure. DFT optimized coordinates of 2 can be found in Table 4.2 in the appendix. Analogous binding of PH3 and P(CH3)3 to one of the tfd ligands resulted in spontaneous reformation of the original six-coordinate structure.

Figure 4.1: DFT Optimized Structure of Mo(tfd)2(SC6H4SPPh3)

4.3.5 Synthesis of Compound 3’ - Mo(tfd)2(SC6H4SP(p-tolyl)3)(P(p-tolyl)3)

Mo(tfd)2(bdt) (11.2 mg, 16.3 μmol) was dissolved in 2.7 mL of benzene to give a dark green solution. Tris(p-tolyl)phosphine (66.4 mg, 218 μmol) was dissolved separately in 1.1 mL of benzene. The solutions were mixed and immediately turned dark purple. The solution was left to stand under an inert atmosphere. After 3 days, magenta crystals of 3’ had formed and the supernatant turned dark red. The crystals were filtered off and washed 2 times with benzene. Crystals were dried under vacuum and stored under an inert atmosphere. (11 mg, 8.5 μmol, 52% 1 based on Mo(tfd)2(bdt)). H NMR (400 MHz, CD2Cl2) δ 2.23 (s, 9H, (CH3) x 3), 2.42 (s, 9H, 19 (CH3) x 3), 7-7.7 (m, 28H, (CH) x 28). F NMR (376 MHz, CD2Cl2) δ –54.87 (s, 12F, (CF3)2 x 31 2) . ), P NMR (162 MHz, CD2Cl2) δ 42.70 (s, 1P, PPh3), 54.63 (s, 1P, PPh3).

Representative NMR spectra of this compound can be found in Figure 4.12 -Figure 4.14 in the appendix.

4.3.6 Equilibrium constant determination of 3’ by dilution.

3’ (1.7 mg, 1.31 μmol) was dissolved in 500 μL of CD2Cl2. The first dilution was prepared by withdrawing a 250 μL sample of the stock solution and injecting it into a new NMR tube, then diluting the stock solution with 250 μL of CD2Cl2. This was repeated 4 times, with each new NMR tube having half the concentration of the prior one. Each NMR sample was further diluted 1 with an additional 100 μL of CD2Cl2. H NMR spectra were recorded at 29°C and line fitting was performed to deconvolute and integrate the overlapping signals for free P(p-tolyl)3, 2’ 1 Mo(tfd)2(SC6H4SP(p-tolyl)3) and 3’. The concentrations were calculated by taking the H integration ratios of the products and the calculated combined concentration of both metal species and then solving for their individual concentrations. An excellent fit was obtained for the simple model 3’ ⇌ 2’ + P(p-tolyl)3. The equilibrium constant for the reaction as written is equal -1 to [2’][P(p-tolyl)3][3’] and a straight line was indeed obtained for plotting [2’][P(p-tolyl)3] against [3’] with the slope being 6(5) x10-5 M (see below Figure 4.3).

0.12 mM

0.23 mM

0.47 mM

0.94 mM

1.87 mM

2’), 2.42 (s, “b”, bdt-bound P(p-tolyl)3 of 3’), 2.23 (s, “d”, Mo-bound P(p-tolyl)3 of 3’).

1.E-07 9.E-08 8.E-08 7.E-08 6.E-08 5.E-08 4.E-08

[P(p-tolyl)3][2'] 3.E-08 2.E-08 1.E-08 0.E+00 0.E+00 2.E-04 4.E-04 6.E-04 8.E-04 1.E-03 1.E-03 1.E-03 2.E-03 2.E-03 [3']

Figure 4.3: Determination of the equilibrium constant for the dissociation of tris(p- tolyl)phosphine from 3' at 29°C through 1H NMR spectroscopy and integration. The model predicting the linear dependence of [P(p-tolyl)3][2'] fits the data reasonably well (R2 =0.9902). The slope yields the dissociation constant Keq = 6(5) x10-5 M. The slight deviation from linearity is likely due to a compounding of errors due to each dilution being used to make the subsequent dilution.

4.3.7 Equilibrium constant determination for formation of 3 by UV-Vis titration.

Triphenylphosphine (16.1 mg, 61.3 μmol) was dissolved in 5 mL of dichloromethane to create a standard solution (12.3 mM in PPh3). 250 μL of the Mo(tfd)2(bdt) standard solution (1.38 mM) was taken and placed in a vial (15 mm OD diameter) and diluted with 3 mL of dichloromethane

(0.106 mM in Mo(tfd)2(bdt). UV-vis spectra (scan range 400-720 nm) were collected while injecting 5 μL of the triphenylphosphine solution after every scan (T ≈ 29°C) for a total addition of 160 μL or 5.69 mole equivalents of triphenylphosphine. The resulting spectra were processed with ReactLab™ EQUILIBRIA14 and fitted to the multiple equilibrium model (1) and (2).

Mo(tfd)2(bdt) + PPh3 ⇌ 2 (1)

2 + PPh3 ⇌ 3 (2)

The equilibrium constant for (1) was found to be 4(3) x 104 M-1. The equilibrium constant for (2) was also 4(3) x 104 M-1. Taking their reciprocal values gives dissociation constants of 2(1) x 10-5 M and 2(1) x 10-5 M respectively.

Titration of Mo(tfd)2(bdt) (initially 0.1 mM) with PPh3 at various concentrations

1.8

1.6

1.4

1.2

0.8

0.6

0.4 Absorbance (arbitrary units) 0.2

0 400 450 500 550 600 650 700 Wavelength (nm)

Figure 4.4: UV-vis spectra (solvent = CH2Cl2) following the titration of Mo(tfd)2(bdt) (0.106 o mM initially) with PPh3 (0 → 5.69 equiv. with 12.3 mM) (T ≈ 29 C).

Component distribution

0.0002

0.00015

0.0001

Concentration (M) 0.00005

0 00.511.522.533.54 Equivalents of PPh3 added

PPh3 Mo(tfd)2(bdt) 2 3

Figure 4.5: Determined concentration of components during titration of Mo(tfd)2(bdt) with

PPh3. Region past 4 equiv. omitted for clarity.

Deconvoluted Spectra

16000

14000

12000

10000

8000

6000

4000

Absorbance (arbitrary units) 2000

0 400 450 500 550 600 650 700 Wavelength (nm)

Mo(tfd)2(bdt) 2 3

Figure 4.6: Deconvoluted spectra for the absorbing components fitted to the multiple equilibrium model created using (1) and (2).

4.4 Results and Discussion

9 When the known compound Mo(tfd)2(bdt) (1) was treated with excess (4 eq or more) triphenylphosphine, we observed the formation of a new stable complex. As will be detailed below, an intermediate species (2) is involved in the formation of the stable product (3) from 1. A sharp singlet in the 19F-NMR spectrum is observed for 3, at –54.94 ppm, indicating that all

CF3 groups on the tfd ligands are equivalent either by symmetry, or as discussed below, by rapid pseudo-rotation. Two sharp 31P singlet peaks are seen at very different shifts, at 54.14 ppm and 43.54 ppm. Magenta-colored crystals of 3 were obtained from two different solvents, benzene and chloroform, by directly reacting concentrated solutions of 1 with an excess (ca. 3.7 eq) PPh3. Figure 4.7 shows the structure of 3, where two phosphines have added in a very unusual way:

One has attacked the bdt ligand, whereas the other one coordinates to molybdenum, such that compound 3 is Mo(tfd)2(SC6H4SPPh3)(PPh3). Apart from the Mo-bound phosphine, which is labile (see below), the first coordination sphere of molybdenum in 3 (thiolate ligand in conjunction with two dithiolenes) is very similar to what is seen for the DMSO reductase family of molybdenum oxotransferases15,16 (Figure 1.4, Figure 4.8) or for nitrate reductase from desulfovibrio desulfuricans17 and also for a model complex for nitrate reductase from Sarkar’s IV 18 group, [Et4N][Mo (PPh3)(p-Me-SPh)(mnt)2]. The Mo-thiolate and the Mo-P bond lengths are very similar across the two complexes, at 2.416(2) vs. 2.390(2) (this work vs. Sarkar’s complex)19 for Mo-thiolate and 2.564 (2)/ 2.586(2) for Mo-P. The Mo-dithiolene bond distances

80 are also very similar, the average being 2.357(15) vs. 2.364(13). The same is true for thiolate- Mo-P angles, at 76.24(7) vs. 77.16(6).

ser O X

Mo S S S S

Figure 4.8: Representative structure of the active site for the DMSO reductase family of enzymes. (X = S or O, serine residue may be replaced with cysteine)

The molecule shows no symmetry (C1) in the crystal, but flexibility in the dangling SC6H4SPPh3 will lead to at least Cs symmetry in solution. Furthermore, trigonal-prismatic Mo(IV) complexes have access to a low-barrier “twisting” process in which two distinct positions can rapidly 19 exchange. This leads to pseudo-C2v symmetry in solution as observed in the NMR spectra of six-coordinate bis(dithiolene) complexes having two different non-ditholene ligands.20,21 The positions occupied by SC6H4SPPh3 and by PPh3 are thus expected to exchange, and one averaged 19 CF3 environment is indeed seen in the F NMR of 3, at –54.94 ppm.

The connectivity of atoms is unchanged in solution: one phosphine is bonded to the bdt ligand and the other one is coordinated to molybdenum, as supported by 31P NMR: the singlet at 43.54 ppm corresponds to the sulfur-bonded phosphorus, its shift is very similar to the 44.1 ppm shift of S-coordinated, monodentate SC6H4SPPh3 in a gold dithiolene complex, reported as a synthetic 5 by-product in 11% yield. While the formation of SC6H4SPPh3 has precedent, its high-yielding and reversible formation is unprecedented. The second singlet at 54.14 ppm corresponds to the Mo-coordinated phosphorus, consistent with literature shifts (49.68 ppm).18

The formation of 3 could proceed via 2 (Scheme 4.1). When 1 was treated with between one and 19 two equivalents of PPh3, an additional species, 2, is indeed observed, with a singlet in the F 31 NMR at –55.09 ppm and a singlet in the P NMR at 47.58 ppm. Addition of excess PPh3 (> 2 eq) leads to disappearance of these signals and the appearance of the signals for 3.

The DFT optimized structures of 2 show that tfd bonded phosphine is unstable and reverts to

Mo(tfd)2(bdt) and free phosphine while bdt bonded phosphine optimizes to a stable 5-coordinate structure. Thus, computational modeling suggests 2 is a five-coordinate complex with square based pyramidal geometry (Figure 4.1). More support for the equilibrium in Scheme 4.1 is obtained with a triarylphosphine having a methyl as a spectroscopic ‘handle’: When tris(p- tolyl)phosphine is added to 1, the products 2’ and 3’ (Scheme 4.1) can be clearly distinguished not only in their 31P and 19F spectra but also in their 1H NMR spectra. 2’ shows a single sharp singlet at 2.45 ppm (1H), while 3’ shows a sharp singlet at 2.42 ppm and a broad singlet at 2.23 ppm. This is consistent with 2’ having only one attached triarylphosphine while 3’ has two. The broad singlet at 2.23 ppm is assigned as the Mo-coordinated phosphine. It is believed that this broadening is due to the bulky aryl groups on the phosphine being sterically retarded from free rotation, this would in turn lead to the methyl groups being inequivalent on the NMR time scale

-5 leading to a broadening of the signals. K for 3 ⇌ 2 + PPh3 was determined to be 2(1) x 10 M at 29 °C, (UV-vis). When a sufficiently dilute solution of 3 is equilibrated for ca. 30 min. at room temperature, some 2 can be seen by NMR. Further dilution shifts the equilibrium more toward 2. → -5 Similarly, we observe 3’ ← 2’ + P(p-tolyl)3, where K= 6(5) x 10 M (29 °C). 3 is stable for at least a few days in the presence of excess phosphine, while 2 slowly and irreversibly decays. Decay may be due to dimerization.22

0 S-PR3 S-PR3 PR3 S S S S PR3 PR3 Mo Mo Mo S S S S S S S S S S S S F C F3C F3C 3 CF CF3 CF CF3 CF CF3 F3C 3 F3C 3 F3C 3 1,1' 2,2' 3,3' 1, 2, 3: R = Ph 1', 2', 3': R= p-tolyl

Scheme 4.1

In addition to using triphenylphosphine and tris(p-tolyl)phosphine as nucleophiles, tris(pentafluorophenyl)phosphine was tested. It produced an insoluble precipitate with 1 that was

82 presumably the tris(pentafluorophenyl)phosphine analogue of 3, but further characterization by NMR was difficult due to its insolubility. Tricyclohexylphopshine was also tested and produced insoluble precipitate as well, but further characterization was also hindered due to the instability of tricyclohexylphopshine in chlorinated solvents as well as the insolubility of the resulting complex. Tris(o-tolyl)phosphine was also used as a nucleophile but no reaction was observed. Neither an analogue of 3 or 2 was created. It is likely that the ortho positioning of the methyl groups makes the phosphine too bulky to successfully attack 1. Thiols and hydroxides were also tested, but these only reduced 1 to the dianion. Overall these experiments show that to observe interesting reactivity the nucleophiles must not be too reducing and sterically unhindered.

Tantalizingly, DMS was observed when 3 was reacted with DMSO, the reactivity expected for a DMSO reductase model complex. The kinetics and mechanism of this reaction shall be explored in detail in chapter 5.

We conclude that nucleophilic addition to a non-innocent ligand can open a coordinatively saturated complex and thus enhance reactivity. In Mo(tfd)2(bdt), a phosphine can directly add to the bdt ligand creating a zwitterionic SC6H4SPPh3 monodentate ligand, giving a complex that is a structural DMSO reductase model. Future work will expand the generality of the new logical but initially counter-intuitive approach to make an open site: via addition of a nucleophile, but to the ligand.

4.5 Appendix

6000

5500

5000

4500

4000

3500

3000

2500

2000

1500

1000

500

.5 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 f1 (ppm)

1 Figure 4.9: H NMR spectrum of 3 (400 MHz, CD2Cl2) δ = 7-7.8 (m, 34H, (CH) x 34).

19 Figure 4.10: F NMR spectrum of 3 in equilibrium with 2 in CD2Cl2. (376 MHz, CD2Cl2) δ – 54.94 (s, 3), –55.09 (s, 2)

31 Figure 4.11: P NMR of 3 in equilibrium with 2. (162 MHz, CD2Cl2) δ 43.54 (s, 3), 54.14 (s, 3) δ 47.58 (s, 2).

1 Figure 4.12: H NMR spectrum of 3’ in equilibrium with 2’. (400 MHz, CD2Cl2) δ 2.35 (s, free

P(p-tolyl)3), 2.46 (s, 2’), 2.43 (s, bdt-bound P(p-tolyl)3 of 3’), 2.23 (s, Mo-bound P(p-tolyl)3 of 3’), 7-7.7 (m, aromatic protons of 3’ and 2’).

19 Figure 4.13: F NMR spectrum of 3’ in equilibrium with 2’ in CD2Cl2. (376 MHz, CD2Cl2) δ – 54.87 (s, 3’), –55.07 (s, 2’)

1.60E+09

52.63 42.70 1.50E+09

1.40E+09

1.30E+09

1.20E+09

1.10E+09

1.00E+09

9.00E+08

8.00E+08

7.00E+08

6.00E+08

5.00E+08

4.00E+08

3.00E+08

2.00E+08

1.00E+08

0.00E+00

-1.00E+08 62 61 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 f1 (ppm)

31 Figure 4.14: P NMR of spectrum 3’ (162 MHz, CD2Cl2) δ 42.70 (s, 3’), 54.63 (s, 3’).

C -3.245674407 2.938562349 2.146001796 S -0.577142945 6.039020951 -0.012191945 C -4.225866389 3.774510061 1.724023602 P 2.802575636 5.483336425 1.586866229 C -3.266944095 2.195141631 3.465699170 C 1.491212237 5.687316331 2.812607475 C -5.477907173 4.088725519 2.519410006 C 0.537271925 4.673557508 2.987552618 F -3.669749378 2.993659661 4.485232378 C 1.478985756 6.827578221 3.634391056 F -2.023702404 1.767171312 3.809128277 C -0.430206990 4.804772088 3.981839419 F -4.062308415 1.111162599 3.450338887 C 0.504171130 6.948524579 4.623038012 F -5.232772596 5.038101305 3.455424483 C -0.448577519 5.940681068 4.794280541 F -6.466960752 4.559569464 1.735528731 H 0.513964639 3.808751624 2.333627645 F -5.970971337 3.007517933 3.161433660 H 2.210295300 7.618254836 3.497073401 S -1.829247324 2.640066310 1.133701971 H -1.171242117 4.025108221 4.111083662 S -4.069733912 4.636951446 0.185214940 H 0.485522869 7.833148009 5.252619917 Mo -1.916887101 4.203837975 -0.633516006 H -1.212169234 6.042991000 5.559742787 C -1.453637291 4.153081942 -3.908278006 C 3.323579424 7.104838289 0.980889065 C -0.575792768 3.203240387 -3.502292538 C 4.676875207 7.372004547 0.723683523 C -1.510171358 4.728532290 -5.308125866 C 2.341825865 8.064245239 0.678779730 C 0.441440765 2.534228493 -4.403096453 C 5.043091276 8.597756841 0.170658815 F -0.274514184 4.878388213 -5.844868445 C 2.720687793 9.283871817 0.122717731 F -2.065942794 5.962365658 -5.304069993 C 4.067493779 9.550303010 -0.133196980 F -2.232567424 3.969752915 -6.155248611 H 5.438999561 6.636895344 0.960079943 F 1.542347531 3.310049406 -4.577750286 H 1.295735176 7.861344415 0.880795311 F 0.874182962 1.366215592 -3.885953145 H 6.090832945 8.808396140 -0.021759942 F -0.052258518 2.258014432 -5.628736689 H 1.960329822 10.021977337 -0.113165216 S -2.654820891 4.799243013 -2.778852714 H 4.357850857 10.502063366 -0.568676026 S -0.533046922 2.702777379 -1.805667996 C 4.193395997 4.618493737 2.373211472 C 0.568107734 6.052072074 -1.396317917 C 4.280232469 4.569117763 3.773495839 C 1.812804664 5.389937925 -1.341473731 C 5.195790656 4.026754396 1.585240296 C 0.243960685 6.757036013 -2.561969754 C 5.369641112 3.941122979 4.376260046 H -0.727965923 7.233984662 -2.628684299 C 6.281919955 3.405501759 2.198486177 C 2.703944433 5.455228756 -2.418972775 C 6.369427905 3.362672948 3.592174013 C 1.122806463 6.799215632 -3.643373774 H 3.500839016 5.008155928 4.387230683 H 3.651548758 4.929663269 -2.361462817 H 5.117128808 4.037182512 0.502743939 H 0.823611393 7.311695812 -4.552364067 H 5.431545558 3.898188167 5.459474523 C 2.358447986 6.157112462 -3.571617638 H 7.052655394 2.946040510 1.587118770 H 3.039326675 6.176625351 -4.416507671 H 7.213245831 2.870037447 4.066586161 S 2.237941131 4.241581003 -0.025130767

Table 4.2: DFT optimized coordinates for 2

References

1 Kirk, M. L.; McNaughton, R. L.; Helton, M. E.; Prog. Inorg. Chem. 2004, 52, 111.

2 Schrauzer, G. N.; Zhang, C.; Schlemper, E. O. Inorg. Chem. 1990, 29, 3371.

3 Bakir, M.; White, P.S.; Dovletoglou, A. ; Meyer, T.J.; Inorg. Chem. 1991, 30, 2835.

4 Nomura, M; Fujita-Takayama, C; Sugiyama, T; Kajitani, M; J. Organomet. Chem. 2011, 696, 4018.

5 Cerrada, E.; Fernández, E. J.; Jones, P. G.; Laguna, A.; Laguna, M.; Terrobat, R.; Organometallics 1995, 14, 5537.

6 For addition of alkenes to P,S-chelate ligands: Ouch, K.; Mashuta, M. S.; Grapperhaus, C. A. Inorg. Chem. 2011, 50, 9904.

7 (a) Schrauzer, G. N.; Mayweg, V. P. J. Am. Chem. Soc. 1965, 87, 1483. (b) Wing, R. M.; Tustin, G. W.; Okamura, W. H. J. Am. Chem. Soc. 1970, 92, 1935.

8 (a) Wang, K.; Stiefel, E. I.; Science 2001, 291, 106. (b) Harrison, D. J. ; Nguyen, N. ; Lough, A. J. ; Fekl, U. J. Am. Chem. Soc. 2006, 128, 11026. (c) Dang, L; Shibl, M. F.; Yang, X.; Alak, A.; Harrison, D. J.; Fekl, U.; Brothers, E. N.; Hall, M. B. J. Am. Chem. Soc. 2012, 134, 4481.

9 Harrison, D. J.; Lough, A. J.; Nguyen, N.; Fekl, U.; Angew. Chem. Int. Ed. 2007, 46, 7644.

10 a) A. D. Becke, J. Chem. Phys., 1993, 98, 5648-5642. b) P. J. Stephens, F. J. Devlin, C. F. Chablowski, M. J. Frisch, J. Phys. Chem., 1994, 98, 11623-11627. c) R. H. Hertwig, W. Koch, Chem. Phys. Lett., 1997, 268, 345-351.

11 a) M. M. Francl, W. J. Petro, W. J. Hehre, J. S. Binkley, M. S. Gordon, D. J. DeFrees, J. A. Pople, J. Chem. Phys., 1982, 77, 3654. b) W. J. Hehre, R. Ditchfield, J. A. Pople, J. Chem. Phys., 1972, 56, 2257.

12 a) P. J. Hay, W. R. Wadt, J. Chem. Phys., 1985, 82, 270. b) P. J. Hay, W. R. Wadt, J. Chem. Phys., 1985, 82, 284. c) P. J. Hay, W. R. Wadt, J. Chem. Phys., 1985, 82, 299.

13 a) S. Grimme, J. Comput. Chem., 2006, 27, 1787-1799. b) S. Grimme, Wiley Interdiscip. Rev. -Comput. Mol. Sci., 2011, 1, 211-228.

14 ReactLab EQUILIBRIA, version 1.1; software for global analysis; Jplus Consulting; Karawara, Australia; 2011

15 Hine, F. J.; Taylor, A. J.; David Garner, C.; Coord. Chem. Rev. 2010, 254, 1570.

16 Majumdar, A.; Sarkar, S.; Coord. Chem. Rev. 2011, 255, 1039.

17 Dias, J.M.; Than, M.E.; Humm, A.; Huber, R.; Bourenkov, G.P.; Bartunik, H.D.; Bursakov, S.; Calvete, J.; Caldeira, J.; Carneiro, C.; Moura, J.J.G.; Moura, I.; Romão, M.J.; Structure 1999, 7, 65.

18 Majumdar, A.; Pal, K.; Sarkar, S.; Dalton Trans. 2009, 1927.

19 Argyropoulous, D.; Mitsopoulou, C.A.; Katakis, D.; Inorg. Chem. 1996, 35, 5549.

20 Donahue, J.P.; Goldsmith, C.R.; Nadiminti, U.; Holm, R.H.; J. Am. Chem. Soc. 1998, 120, 12869.

21 Sung, K.M.; Holm R.H.; Inorg. Chem. 2000, 39, 1275

22 Enemark, J. H.; Cooney, J. J. A.; Wang, J.-J.; Holm, R. H.; Chem. Rev. 2004, 104, 1175.

Chapter 5: Mechanistic elucidation of the oxotransferase activity for Mo(tfd)2(tht)2 5.1 Abstract

The catalytic oxygen atom transfer reactivity of Mo(tfd)2(SC6H4SP(p-tolyl)3)(P(p-tolyl)3), 3, from DMSO to P(p-tolyl)3 forming DMS and OP(p-tolyl)3 was investigated. It was determined that 3 was actually a precatalyst to a Mo(tfd)2 species and that Mo(tfd)2(tht)2 was a better catalyst. To elucidate the mechanism, NMR spectroscopy was used to track the conversion of reactants to products in real time so that the reaction kinetics may be analyzed. Results suggest that the OP(p-tolyl)3 produced remains bonded to the catalyst and must be substituted with DMSO in a fast but reversible reaction in the proposed catalytic cycle. Experiments in different solvents suggest the removal of oxygen from DMSO involves a polar transition state. Finally, the mechanism suggests the DMSO reduction step proceeds by a first order reaction. This is the only known DMSO reductase model complex that proceeds by a first order reaction.

5.2 Introduction

Oxotransferases are a group of metalloenzymes responsible for the transfer of oxygen and subsequent redox activities as mentioned previously in Chapter 1.1 A few common ones include DMSO reductase that reduces DMSO to DMS and then produces water from reacting the oxygen with protons and electrons. Another is sulfite oxidase where oxygen derived from water is added to sulfite ions liberating protons and electrons. Since the electrons are eventually transferred to the electron transport chain, sulfite oxidase is thus very important for the metabolism of sulfur compounds.

Successful synthesis of oxotransferase cofactors is useful for the development of medical products to treat related illnesses.2 Successfully mimicry of oxotransferase enzymes with model complexes is useful in understanding them for applications in industrial processes such as the epoxidation of alkenes and oxygen reduction catalysts in fuel cells.

A key feature of many of these enzymes is the use of molybdopterin ligands that are essentially dithiolene ligands (Figure 1.3: Molybdopterin Cofactor). Dithiolene ligands are examples of “non-innocent” ligands in that the ligands themselves can have multiple oxidation states and thus

94 contribute to the overall charge of the complex. This also gives more direct ability to manipulate the redox properties of the complex by changing the properties of the ligands.

While most biological oxotransferases use protons along with electrons as the oxygen acceptors, phosphines have become popular among researchers, as using protons has been difficult due to decomposition of most complexes resulting in conversion to the tris(dithiolene).3 Phosphines are also popular due to their high solubility in organic solvents. However, if phosphines are used, then the question of how the phosphines and their products (phosphine oxides) themselves are affecting the mechanism arises.

There is a tremendous amount of literature about the kinetics of various oxotransferase model complexes and their mechanisms and Holm et al. have compiled extensive comparative data.4 When transferring oxygen from a substrate, for most complexes the following kinetic scheme applies:

k Scheme 5.1 MIV + XO MVIO + X

where M is the metal center of the complex in question (Mo or W) and XO is the substrate

(DMSO, R3NO, etc.). This scheme applies if the reaction is first order in both reactants to give an overall second order rate law. A lesser-known kinetic behavior involves a pre-equilibrium step followed by a rate determining first order reaction:

K k Scheme 5.2 IV IV VI M + XO M [XO] M O + X

This behavior can be identified if the rate becomes zero order with respect to XO at higher XO concentrations. This pathway is very rare and only a few examples are known among non- dithiolene based model complexes with only one known dithiolene-based complex (a sulfite oxidase).5 There are no known examples among DMSO reductase model complexes. Finding more examples of the pathway in Scheme 5.2 for dithiolene based complexes could help in the understanding of the transition state for such oxotransfer reactions.

As mentioned previously in chapter 4, compound 3 showed possible oxotransferase activity when reacted with DMSO. The kinetics and mechanism of this reaction shall be studied here.

A difficulty with studying oxotransferase reactions of simple substrates such as DMSO and aryl phosphines is tracking them. Ideally some sort of real-time method is preferable but UV-vis and IR spectrometry cannot easily distinguish DMSO and aryl phosphines. In addition, the metal complexes are strongly colored and their decomposition as the reaction progresses introduces its own systematic errors. Quenching the reaction as it progresses and using other analytical techniques such as titration can work, but also introduces additional errors.

NMR spectroscopy is ideal for this application as the substrates can easily be distinguished due to the much greater information density of a typical high-field NMR spectrum. Additionally the substrates can be predictably engineered to give maximum analytical signal and resolution. In this case the aryl phosphine, triphenylphosphine, provided a strong but broad aromatic signal, so to improve resolution and ease analysis further, a different phosphine, P(p-tolyl)3, was selected, as its methyl groups provide a strong narrow aliphatic signal and did not alter the structure to significantly alter the reactivity.

Of paramount importance in kinetics is the ability to track the reaction progress in real time. The only caveat of NMR spectroscopy is that the time scale of a typical NMR spectrum, acquisition is on the order of minutes, therefore the reaction conditions should be designed to take place on the order of hours so that a complete spectrum can be obtained without the reaction significantly changing while it is doing so.

If this challenge can be met, then a modern high-field NMR spectrometer can be programmed to sequentially acquire spectra and modern software can simultaneously track and integrate almost every species to give a great deal more data for analysis of kinetics. Older methods like UV-Vis often cannot track as many species, and usually the concentrations of some species must be inferred mathematically from other species. This can be extremely misleading if there are unknown side reactions occurring. NMR allows direct measurement and tracking of almost all species and thus allows much more accurate data to be used. If there were additional reactions occurring they would be much easier to recognize.

5.3 Experimental

Mo(tfd)2(tht)2 was prepared as described previously in chapter 3. Mo(tfd)2(DMS)2 was prepared completely analogously to Mo(tfd)2(tht)2. Compound 3 was prepared as described previously in chapter 4. P(p-tolyl)3, DMSO and dichlorobromomethane were obtained from Sigma Aldrich.

DMSO and CDCl3 were dried over molecular sieves (3 Å) before use. C6D6 was dried over sodium/benzophenone and vacuum transferred before use. NMR data were collected on a Bruker 1 Avance III 400 MHz spectrometer. H spectra were referenced to the CDCl3 solvent peak (7.26 1 ppm). H spectra were also referenced to the C6D6 solvent peaks where applicable (7.16 ppm).

To simplify handling of the sub-milligram quantities of compounds, standard solutions of the compounds were made. 100 μL of DMSO were diluted to 5 mL with CDCl3 and 320 mg of P(p- tolyl)3 were dissolved in CDCl3 and diluted to 5mL. 10 mg of the Mo(tfd)2(tht)2 catalyst and 100 mg of P(p-tolyl)3 were dissolved in CDCl3 and diluted to 5 mL. These particular quantities were selected so that equal quantities of each standard solution would give approximately 1% catalyst loading for the reaction of DMSO and P(p-tolyl)3. It was found that at these particular concentrations the reaction would reach 50% completion in approximately 1-3 hours. This time scale is selected so that NMR spectroscopy could be optimally used to track the progress of the reaction without distorting the data points due to long acquisition times of the spectra. P(p-tolyl)3 was added to the Mo(tfd)2(tht)2 stock solution as a stabilizer as it was found a solution of pure

Mo(tfd)2(tht)2 would decay into Mo(tfd)3 and tht within two days at room temperature with a subsequent large loss of catalytic activity. The addition of P(p-tolyl)3 stabilized the stock solution for much longer periods and kinetic experiments performed a week apart showed little loss of activity.

Unless otherwise explicitly stated and defined, a typical kinetic experiment is performed as follows: 50 μL of the Mo(tfd)2(tht)2 catalyst solution and 50 μL of the P(p-tolyl)3 solution were added to 250 μL of CDCl3 in a J. Young NMR tube under air-free conditions. 50 μL of the DMSO solution were injected and shaken to mix. This instant was designated the start time, and NMR spectra were taken at regular intervals at a controlled temperature of 25 oC for a total of 3 to 12 hours. A typical 1H NMR spectrum is shown in Figure 5.1. Concentrations were varied by varying the quantity of stock solution for the corresponding reactant. The CDCl3 was readjusted to ensure the total volume would always be 400 μL. NMR spectroscopy allowed for

simultaneous monitoring of all reactants and products (DMSO, DMS, P(p-tolyl)3 and OP(p- tolyl)3). A graph of a typical reaction run looks like Figure 5.2. For clarity, this figure and all subsequent figures (unless otherwise noted) show only 9 data points between 0 and 6500 seconds and only 16 are shown between 6500 and 36500 seconds. In total there are normally 132 points per trace.

1300

1200

D 1100

1000

900 B

800

700

600 A 500 C

400

300

200

100

-100

2.70 2.65 2.60 2.55 2.50 2.45 2.40 2.35 2.30 2.25 2.20 2.15 2.10 2.05 f1 (ppm)

Catalyzed reaction of DMSO and P(p-tolyl)3

0.035

0.03

0.025

DMSO 0.02 DMS P(p-tolyl)3 0.015 OP(p-tolyl)3

Concentration [M] 0.01

0.005

0 0 10000 20000 30000 40000 Time (s)

Figure 5.2: Typical results from the catalyzed reaction of DMSO and P(p-tolyl)3 using o Mo(tfd)2(tht)2 at a concentration of 0.328 mM at 25 C.

Several trials were conducted varying the initial concentrations of DMSO. The initial rate of each reaction was found by fitting a 6th order polynomial equation to the integration of the DMS signal versus time, then taking the coefficient of the linear term. To find the reaction order the logarithm of the initial rate versus the logarithm of the concentration was plotted (Figure 5.9).

For further insight into the mechanisms, kinetic models need to be constructed and tested against experimental data. While this can be done by hand, computers and the appropriate software have revolutionized this task, and the program used in this work was Dynafit 3. The program takes arbitrarily defined kinetic models and converts them each to a set of differential equations. It then takes kinetic data based on the concentrations of the reagents and products at various time increments and fits the kinetic model by first solving the differential equations numerically to produce a simulated plot of the concentrations with respect to time. It then compares this plot to the real data and calculates the difference, the “error”. Then it iteratively adjusts the parameters, the rate constants and initial concentrations, to minimize that (absolute) deviation between the simulated and real data. With enough iterations, the simulated and real data converge to the best possible fit that can then be outputted along with associated standard errors and confidence intervals. The program is quite flexible allowing different parameters to be optimized or held constant.6

To perform the analysis for this system, the concentrations of the reactants and products (DMSO,

DMS, P(p-tolyl)3 and OP(p-tolyl)3), as determined by NMR integration, were simultaneously recorded during a typical kinetic run and fed into the program. Various models were constructed and run in the program. The findings will be discussed below, under “results and discussion”.

A special note should be made that occasionally old samples (>3 weeks) of the Mo(tfd)2(tht)2 catalyst solution (also containing P(p-tolyl)3) exhibited dramatically increased catalytic activity with the reactions going to completion in a few minutes than the typical 3 hours for 50% completion. The cause was eventually traced to the formation of another catalyst, dichloromethylenetri(p-tolyl)phosphorane. Presumably this was formed through the reaction of the CDCl3 solvent with P(p-tolyl)3, itself catalyzed by the aged Mo(tfd)2(tht)2. The phosphorane independently catalyzed the reaction of DMSO with P(p-tolyl)3 by a different pathway. This type of reactivity has already been observed in carbon tetrachloride.7 The question arises whether the phosphorane is exclusively responsible for the catalytic activity observed. Experiments conducted in C6D6 also demonstrated catalytic activity comparable to experiments performed in

CDCl3 with freshly prepared reagents. Additionally, experiments to determine if MoO(tfd)2 plays a role in the catalytic cycle were exclusively performed in C6D6 and also demonstrated catalytic activity (Experiment 5.3.6). If the phosphorane were exclusively responsible for the activity, then none of these experiments should have shown consistent activity. To ensure the kinetics of the

100

Mo(tfd)2(tht)2 were being properly studied and not some effect by this phosphorane, only newer standard solutions of the catalyst were used and additional experiments were performed in C6D6 to corroborate the data. C6D6 was not used exclusively due to partial overlap of some of the signals of the reagents and products on the NMR spectra as show in Figure 5.3.

900

A 850

800

C 750

700

650

600

550

500

450

400

350

300

250

200

150 D B 100

-50

2.18 2.16 2.14 2.12 2.10 2.08 2.06 2.04 2.02 2.00 1.98 1.96 1.94 1.92 1.90 1.88 1.86 1.84 1.82 1.80 1.78 1.76 1.74 1.72 1.70 1.68 1.66 1.64 f1 (ppm)

Figure 5.3: Typical NMR spectrum of a kinetic experiment in C6D6. A is assigned as P(p-tolyl)3,

B is assigned as OP(p-tolyl)3, C is assigned as DMSO, D is assigned as DMS. Not shown is the aromatic region that contains P(p-tolyl)3 and OP(p-tolyl)3.

5.3.1 Reaction of dimethyl sulfoxide with 3

In a J.Young NMR tube 3 (1.1 mg, 0.89 μmol) was dissolved in 370 μL of CD2Cl2 and DMSO was injected (1 μL, 14 μmol). The tube was sealed and allowed to stand at room temperature. The pink solution turned green after 140 minutes. 1H NMR integration revealed dimethyl sulfide (DMS) present in 57% yield relative to the triphenylphosphine originally present on 3.

101

5.3.2 Catalytic reduction of dimethyl sulfoxide with 3 and triphenylphosphine.

Triphenylphosphine (20.3 mg, 77 μmol) was placed in a vial along with 500 μL of CD2Cl2. This solution was divided equally into two J.Young NMR tubes. A standard solution of Mo(tfd)2(bdt) was made by dissolving Mo(tfd)2(bdt) (4.75 mg, 6.90 μmol) in 5 mL of CD2Cl2 to give a dark green solution (1.38 mM in Mo(tfd)2(bdt)). In one tube (referred to as the “catalyst” tube) 250 μL of the Mo(tfd)2(bdt) solution was injected into one NMR tube to create 3 in situ (0.345 μmol).

250 μL of CD2Cl2 was injected into the other (“control”) tube. DMSO (10 μL, 141 μmol) was injected into both tubes at the same time. Both tubes were sealed and 1H NMR spectra were recorded daily until no further increases in DMS concentration in the catalyst tube was observed (~2 days). The small 1H integration of the DMS in the control tube (~ 1.88 μmol) was subtracted from the 1H integration of DMS in the catalyst tube (~ 31.0 μmol) to give the catalytic DMS production. Approx 84 mole equivalents of DMS (relative to the starting complex) was observed giving a turnover number of about ~80.

5.3.3 Catalytic reduction of dimethyl sulfoxide with 3’ and P(p-tolyl)3 over time.

P(p-tolyl)3 (1.2 mg, 3.94 μmol) was placed into a J. Young NMR tube along with 3’ (1.8 mg,

1.39 μmol) and 400 μL of CD2Cl2 to give a magenta solution. ~2 μL of 1,3,5- tris(trifluoromethyl)benzene was added as an internal standard. DMSO (5 μL, 70.5 μmol) was injected directly into the tube and shaken. The tube was sealed and 1H NMR spectra were recorded every ~5 minutes for a total of 12 hours at a temperature of 25°C. The concentrations were determined by 1H integration and plotted against time (see below Figure 5.4).

102

Concentration vs. Time

0.01 0.009 0.008 0.007 0.006 0.005 0.004 0.003

Concentration (M) Concentration 0.002 0.001 0 024681012 Time (hours)

OP(p-tolyl)3 P(p-tolyl)3 DMS

Figure 5.4: Catalytic reduction of DMSO with 3’ and P(p-tolyl)3 at 25 °C.

5.3.4 Catalytic reduction of dimethyl sulfoxide with Mo(tfd)2(DMS)2 and P(p- tolyl)3 over time.

P(p-tolyl)3 (2.3 mg, 7.55 μmol) was placed into a J. Young NMR tube along with

Mo(tfd)2(DMS)2 (0.8 mg, 1.19 μmol) and 400 μL of CD2Cl2 to give a yellow-orange solution. ~2 μL of 1,3,5-tris(trifluoromethyl)benzene was added as an internal standard. DMSO (5 μL, 70.5 μmol) was injected directly into the tube and shaken. The tube quickly became a purple-magenta colour. The tube was sealed and 1H NMR spectra were recorded every ~5 minutes for a total of 3 hours at a temperature of 25 °C. The concentrations were determined by 1H integration and plotted against time (see below Figure 5.5).

103

Concentration vs. Time

0.03

0.025

0.02

0.015

0.01 Concentration (M) Concentration

0.005

0 00.511.522.53 Time (hours)

OP(p-tolyl)3 P(p-tolyl)3 DMS

5.3.5 Reaction of Mo(tfd)2(tht)2 with excess DMSO

1.6 mg (2.2 μmole) of Mo(tfd)2(tht)2 were dissolved in 400 μL of CDCl3 and reacted with 1.21 μL (14 μmole) of DMSO. The solution was allowed to stand for 3 days during which the solution went from amber to red to blue. 1H NMR spectra were taken and integration revealed that 19 dimethylsulfide was produced in 79 % yield with respect to Mo(tfd)2(tht)2. F NMR spectra revealed new peaks that do not correspond to the catalyst. These are believed to be decomposition products.

2- 5.3.6 Reaction of MoO(tfd)2 and an oxidant with DMSO and phosphine

4.0 mg (4.8 μmole) of [MoO(tfd)2][NEt4]2, 5.1 mg (18.7 μmole) of ferrocenium tetrafluoroborate and 15.4 mg (50.8 μmole) of P(p-tolyl)3 were dissolved in 400μL of C6D6. 3 μL (3.3 mg or 42 μmole) of DMSO were added and the solution was left to stand for 1h. 1H NMR spectra showed

DMS produced in 13 % yield relative to P(p-tolyl)3. Identically prepared experiments omitting either [MoO(tfd)2][NEt4]2 or ferrocenium tetrafluoroborate showed no production of DMS over the same time period.

104

5.4 Results and Discussion

Inspired by the close structural similarity of compound 3 (made in chapter 4) to oxotransferase enzymes, we tested for similar activity (Scheme 5.3). In a stoichiometric oxotransfer reaction (without extra phosphine) addition of 16 eq of DMSO produced ~1 eq of dimethyl sulfide 1 31 (relative to 3), along with Ph3PO ( H, P). The complex decomposed slowly, as is expected given the instability of 2. However the reaction was successfully made catalytic with the addition of extra phosphine before the addition of DMSO. We typically observed at least 80 turnovers in presence of 112 equivalents of PPh3 and 408 equivalents of DMSO over the course of ~2 days.

DMSO + P(p-tolyl)3 Æ DMS + OP(p-tolyl)3

Scheme 5.3

Preliminary kinetic studies revealed the rate of oxygen-transfer to increase with time rather than decrease. This induction period (Figure 5.4) indicates that the catalytically active species might actually be a decomposition product of 2 or 3. A possible pathway is the loss of the bdt-derived ligand to give a Mo(tfd)2 complex with two labile solvent molecules. Indeed, in a control experiment, the reaction of Mo(tfd)2(DMS)2 (where DMS, dimethyl sulfide, is very labile) with

6.3 eq of P(p-tolyl)3 and 59 eq of DMSO yielded nearly quantitative conversion of P(p-tolyl)3 to

OP(p-tolyl)3 with equivalent generation of DMS, where the catalytic rate was high from the outset, without induction period. It is clear from the plot (Figure 5.5) that Mo(tfd)2(DMS)2 is much more active as a catalyst and even though almost twice the amount of P(p-tolyl)3 reactant was used the reaction still proceeded to 50% completion in less than 30 minutes while 3’ required over 8 hours. The shape of the curve also indicates the rate of change is decreasing with time and suggests the catalytically active species is remaining constant or decreasing in concentration. Since the DMS in Mo(tfd)2(DMS)2 is much more labile than the SC6H4SP(p- tolyl)3 in 3’, it may be that 3’ slowly decays into the catalytically active species while

Mo(tfd)2(DMS)2 has much more direct access to the catalytic cycle. It is not believed that

Mo(tfd)2(DMS)2 is itself the catalyst as the concentration plot in Figure 5.4 using 3’ shows the free DMS and OP(p-tolyl)3 products to form and be detected essentially simultaneously, if

105

Mo(tfd)2(DMS)2 was formed from 3’ then the observed free DMS would appear to decrease relative to OP(p-tolyl)3 as some of it would have been incorporated into Mo(tfd)2(DMS)2. The catalytically active species may be Mo(tfd)2 coordinated with two solvent molecules, but further study will be needed.

With the success of DMSO as an oxygen donor for the oxotransferase reaction, other oxygen donors were attempted. Trimethylamine oxide and tetra-n-butylammonium nitrate were tested but upon reaction with 3 and excess phosphine the reaction mixture turned blue, suggesting decomposition of 3, and subsequent NMR spectra did not show catalytic oxygen removal. Atmospheric oxygen itself was an effective oxygen donor and catalytic production of phosphine oxide was observed. Sulfite (in the form of aqueous sodium sulfite) was tested as an oxygen acceptor and preliminary experiments suggest oxygen transfer, but this was not studied in sufficient detail. Future experiments should expand upon this work.

To explore the kinetics of the catalyzed reduction of DMSO with P(p-tolyl)3 in detail,

Mo(tfd)2(tht)2 was selected as the catalyst for all subsequent experiments as the separate NMR signals for tht from the product DMS allow for easy confirmation and measurement of catalyst loading. Additionally, the distinct tht signals would not interfere with measuring DMS yields, as any DMS measured must be exclusively from DMSO reduction.

The oxygen transfer from DMSO to P(p-tolyl)3 should yield exactly equal amounts of DMS and

OP(p-tolyl)3 without side products, and indeed it can be seen in Figure 5.2 that there exists near perfect match for the loss of reactants and formation of products, which are both formed in near perfect unity. No other species in significant quantity (>1%) was observable by NMR spectroscopy. The catalyst loading is ~1%.

It is possible that the reaction cycle involves a simple single oxygen atom transfer from DMSO onto Mo(tfd)2(tht)2 and a subsequent removal by P(p-tolyl)3. An alternative scheme is the formation a molybdenum dioxo complex through the reaction of the Mo(tfd)2(tht)2 with two DMSO molecules. The resulting complex already has precedence in the literature with nitrile 8 dithiolate (S2C2(CN)2) ligands. This would imply that simply adding DMSO directly to

Mo(tfd)2(tht)2 without P(p-tolyl)3 should generate two equivalents of DMS. However experiments have shown that the yield is 79% or less than one equivalent. Thus the single oxygen atom transfer scheme is a better fit to the results.

106

To determine the kinetic parameters, kinetic data were obtained as described in the experimental and fed in Dynafit 3. A couple of models were constructed and tested as follows:

DMS DMSO

k 2tht + DMS 1 DMSO

O Mo(tfd) 2(tht)2 ktht Mo(tfd)2 Mo(tfd)2

kd k metal decomp 2 products P(p-tolyl) 3 OP(p-tolyl)3

Scheme 5.4: Simple two-step model with decomposition and tht substitution

Scheme 5.4 was constructed to be simple yet reflect the most likely role of the catalyst as an oxygen atom transfer reagent. First, the Mo(tfd)2(tht)2 reacts with the DMSO and transfers an oxygen atom, producing MoO(tfd)2 and liberating two tht and a DMS molecule with a rate of ktht. The MoO(tfd)2 then reacts with a phosphine that removes an oxygen to give a phosphine oxide and Mo(tfd)2 with a rate of k2. While for clarity it is shown only four-coordinate, it is likely coordinated by additional solvent molecules. The Mo(tfd)2 then reacts with another DMSO to remove an oxygen and again produce MoO(tfd)2 and DMS with a rate of k1. The cycle repeats with another phosphine. As mentioned earlier it was observed the catalyst decomposed as the reaction progressed. It is believed the decomposition step (kd) starts from the MoO(tfd)2 intermediate based on the results from adding DMSO to Mo(tfd)2(tht)2. DMS is produced and the decomposition products from that reaction share a few NMR signals with the decomposition products from the catalytic reaction of Mo(tfd)2(tht)2 on DMSO and P(p-tolyl)3.

As it was unknown which steps are relevant, various models were tested omitting or retaining certain steps, paying close attention to the quality of the fit but not unnecessarily increasing the number of free parameters (see below).

107

Simulated Reaction Progression models 1-4

15 concentration (mM) concentration

0 0 5000 10000 15000 20000 25000 30000 35000 40000 Time (s)

Actual OP(p-tolyl)3 Actual DMSO Actual P(p-tolyl)3 Actual DMS Fitted DMSO for model 1 Fitted OP(p-tolyl)3 for model 1 Fitted P(p-tolyl)3 for model 1 Fitted DMS for model 1 Fitted DMSO for model 2 Fitted OP(p-tolyl)3 for model 2 Fitted P(p-tolyl)3 for model 2 Fitted DMS for model 2 Fitted DMSO for model 3 Fitted OP(p-tolyl)3 for model 3 Fitted P(p-tolyl)3 for model 3 Fitted DMS for model 3 Fitted DMSO for model 4 Fitted OP(p-tolyl)3 for model 4 Fitted P(p-tolyl)3 for model 4 Fitted DMS for model 4

Figure 5.6: Simulated reaction progression based on models 1-4. Models 1 & 2 almost perfectly overlap and so are difficult to distinguish. This is also true for models 3 & 4.

In model 1, only k1 and k2 were fitted. It was found that mathematically k1 and k2 are codependent so if they are both optimized freely, then the error on each would be extremely -3 -1 -1 large. Thus k1 was deliberately set to 1 x 10 M s and k2 was fitted to the data to give 3.9(1) x 10-4 M-1 s-1. It should also be noted that the substitution of tht with DMSO was assumed to go extremely fast so the simulation was started assuming total conversion to the MoO(tfd)2 species

108

had already occurred. When the fit was performed it was found k2 fitted to a low error but again on visual inspection of the plot (Figure 5.6 model 1) the model poorly represented the data.

In model 2, the substitution of tht with DMSO was included as ktht and was found to be 6(218) x 5 -1 -1. 10 M s Once again it was also found that mathematically k1 and k2 were codependent and k1 -3 -1 -1 -4 -1 -1 had to be deliberately set to 1 x 10 M s giving a k2 of 3.9(1) x 10 M s once fitted, identical to model 1. The simulated plot (Figure 5.6 model 2) is also in poor agreement with the actual data and the error on ktht substitution step was extremely large at a standard error of 218 x 105 on a value of 6 x 105, even greater than the fitted value itself. Both plots for model 1 and 2 are almost exactly identical. Their similarity calls in into question the necessity of the ktht substitution step.

-4 -1 -1 -3 -1 -1 -4 -1 Model 3 fits k1, k2 and kd to give 6(2) x 10 M s , 2(2) x 10 M s and 5(7) x 10 s respectively. Modeling this system produces a kinetic trace showing fairly good visual agreement with experimental data. However the rate constants showed very high error.

Nonetheless this is considered an improvement since k1 and k2 are no longer mathematically codependent and can be solved from the existing data.

In model 4, the tht substitution step and the decomposition step were combined to see if a better 5 -1 -1 -4 -1 -1 fit for the constants could be obtained, giving ktht = 9(300) x 10 M s , k1 = 6(2) x 10 M s , -3 -1 -1 -4 -1 k2 = 2(3) x 10 M s , kd = 5(8) x 10 s . However, once again the error on the tht substitution step was excessively high and the other constants were not improved with the addition of the extra step.

Normally, Mo(tfd)2(tht)2 is an amber color in CDCl3 and when additional alkyl sulfides are present (DMS and tht). When DMSO or OP(p-tolyl)3 is added, the complex turns a red color, presumably through the formation of a Mo(tfd)2(DMSO)x or Mo(tfd)2(OP(p-tolyl)3)x species. In the case of DMSO this red color persists for a few minutes before turning blue. When P(p-tolyl)3 is reacted with Mo(tfd)2(tht)2 there is no color change. When both DMSO and P(p-tolyl)3 are reacted with Mo(tfd)2(tht)2 and the catalytic oxotransfer is started the solution remains a red color while the catalyst is active. This possibly indicates the resting state involves

Mo(tfd)2(DMSO)x or Mo(tfd)2(OP(p-tolyl)3)x species. The similar color between the two species also raises the question if there may be an equilibrium between the two. It can be speculated that perhaps the oxotransfer step of oxygen onto P(p-tolyl)3 first produces a Mo(tfd)2(OP(p-tolyl)3)x

109

species that must equilibrate with DMSO to produce Mo(tfd)2(DMSO) before the next step in the cycle can be begin. With this in mind, Scheme 5.5 was constructed.

DMSO 2tht DM DMSO S

Mo(tfd) 2(tht) 2 k1 ktht Mo(tfd) 2

OP(p-tolyl) 3 OP(p-tolyl) 3 O metal (k )(k) K kd 3 -3 3 decomp DMSO DMSO Mo(tfd)2 products

P(p-tolyl) 3 OP(p-tolyl) 3

Mo(tfd) 2

Scheme 5.5: Two-step with equilibrium with decomposition and tht substitution

This scheme is very similar to Scheme 5.4 but instead of Mo(tfd)2 as an intermediate it is replaced with the equilibrium reaction of Mo(tfd)2(DMSO)x and Mo(tfd)2(OP(p-tolyl)3)x. First, the catalytic cycle is entered by DMSO substituting for tht on Mo(tfd)2(tht)2 to get

Mo(tfd)2(DMSO)x with a rate constant of ktht. Eventually the oxygen transfer occurs and DMS is liberated and MoO(tfd)2 is produced with a rate of k1. This reacts with P(p-tolyl)3 at a rate of k2 but instead of OP(p-tolyl)3 being liberated immediately it remains coordinated with Mo(tfd)2 to give Mo(tfd)2(OP(p-tolyl)3)x. DMSO substitutes the OP(p-tolyl)3 on the complex with a rate constant of k3 but as OP(p-tolyl)3 builds up the reverse reaction can occur with a rate constant of k-3. Together these two constants are the equilibrium constant that shall be denoted as K3. The decomposition of MoO(tfd)2 is represented by kd.

110

Simulated Reaction Progression models 5-8

concentration (mM) concentration 10

0 0 5000 10000 15000 20000 25000 30000 35000 40000 Time (s)

Actual OP(p-tolyl)3 Actual DMSO Actual P(p-tolyl)3 Actual DMS Fitted DMSO for model 5 Fitted OP(p-tolyl)3 for model 5 Fitted P(p-tolyl)3 for model 5 Fitted DMS for model 5 Fitted DMSO for model 6 Fitted OP(p-tolyl)3 for model 6 Fitted P(p-tolyl)3 for model 6 Fitted DMS for model 6 Fitted DMSO for model 7 Fitted OP(p-tolyl)3 for model 7 Fitted P(p-tolyl)3 for model 7 Fitted DMS for model 7 Fitted DMSO for model 8 Fitted OP(p-tolyl)3 for model 8 Fitted P(p-tolyl)3 for model 8 Fitted DMS for model 8

-2 -1 -1 -1 In model 5 just the rate constants k1 and k2 were fitted to give 6.2(7) x 10 s and 2(7) M s -2 -1 respectively. The equilibrium constant K3 was found to be 4.5(6) x 10 s by modeling it as a reversible reaction with the rate constants k3 and k-3. The constant k-3 was defined as 1 while k3 was fitted to the data. The equilibrium constant was found by dividing k3 by k-3. This equilibration was found to be extremely fast relative to the time scale of the experiment as K3 was always a consistent value even if various k-3 values were tested. Otherwise, K3 would change in value and error as different k-3 values are used. This result gives some evidence to the assertion that there is an equilibration step between Mo(tfd)2(DMSO)x and Mo(tfd)2(OP(p- tolyl)3)x species.

111

When comparing to the existing models, the error on the constants had not been significantly improved and visual inspection of the plot also does not show significant improvement of the simulation.

In model 6, the substitution of tht (ktht) was included in attempt to obtain a better fit. The values -1 -1 -2 -1 -1 -1 - obtained were ktht = 800(9000) M s , k1 = 3.6(7) x 10 s , k2 = 2(22) x 10 M s 1 and K3 = 9(2) x 10-2. However this model produced even larger errors in the rate constants without much improvement in the simulated plot when compared to model 5.

In model 7, decomposition (kd) rather than tht substitution was fitted. The values obtained were -2 -1 -3 -1 -1 -3 -1 k1 = 4(1) x 10 s , k2 = 6(9) x 10 M s , K3 = 0.13(3) and kd = 1(2) x 10 s . The error on the rate constants are still greater than the constants themselves but on visual inspection the fit to the data is extremely close, almost perfectly matching the experimental data.

Finally, model 8 incorporates both decomposition and tht substitutions. The values obtained were -4 -1 -1 -2 -1 -3 -1 -1 ktht = 8(5) x 10 M s , k1 = 4.3(4) x 10 s , k2 = 6(3) x 10 M s , K3 = 0.13(1) and kd = 1.0(5) x 10-3 s-1. The error on the rate constants is now smaller than the constants themselves, raising confidence in their accuracy while visual inspection of the fit shows it to be in excellent agreement with the data.

Interestingly, the low equilibrium constant of K3 at 0.13(1) indicates that the catalyst is inhibited by the build up of the OP(p-tolyl)3 product driving the equilibrium toward Mo(tfd)2(OP(p- tolyl)3)x species. This accounts for the varying rate from the simpler Scheme 5.4. As the products build up the rate changes due to the inhibition.

DMS may also participate in the equilibrium but since it is formed at exactly the same concentration as OP(p-tolyl)3 it is mathematically impossible to distinguish them if they start at the same initial concentration. Future experiments are needed where DMS and OP(p-tolyl)3 are added initially at varying concentrations to distinguish them.

Nonetheless, Model 8 is best supported by the existing data. More quantitative support for this model comes from analysis using Akaike’s rigorous quantitative method as implemented in Dynafit. The results of this analysis are show in the AICc column in Table 5.1. The lower the AICc, the better the model is at describing the data. While using more parameters to a fit a model

112 will generally yield better fits over models with fewer parameters this can give the illusion that more parameters are necessary, when in fact the results are only an artifact of the analysis also incorporating parameters that are not physically or chemically meaningful. Akaike’s method discriminates between models by taking into account the extra parameters used and penalizing models with more parameters over ones with fewer. Models having different numbers of parameters can then be compared more equally. The weight for a model the analysis assigns gives a strong indication of how well it fits the data while at the same time minimizing extraneous parameters.9 Model 8 has the lowest AICc, indicating that it explains the data far better than any other model. Model 7 is still close though and this has some merit due to the uncertainty in ktht. The rate for ktht is not expected to be very accurate if the reaction it corresponds to (the conversion of Mo(tfd)2(tht)2 to Mo(tfd)2(DMSO)) is very fast, then it may reach equilibrium before the first NMR spectra are taken. Indeed a significant color change is directly observed when the final reactant, DMSO, is added. This color change reaches equilibrium before the sample can be placed into the NMR spectrometer. In such a case, model 7 and 8 would be indistinguishable since the NMR data would start after most of the Mo(tfd)2(tht)2 had converted, rendering the ktht rate unobservable. Future experiments would incorporate some way to measure ktht more directly and by a faster technique. Since a color change is already observed, a UV-vis method to find ktht could be used in addition to using NMR spectroscopy for the other rate constants.

113

# Model Results n p AICc -3 -1 -1 1 Scheme 5.4 k1: 1 x 10 M s -4 -1 -1 Fitting parameters: k2: 3.9(1) x 10 M s 528 1 1439.86 k1, k2 5 -1 -1 2 Scheme 5.4 ktht: 6(218) x 10 M s -3 -1 -1 Fitting parameters: k1: 1 x 10 M s 528 2 1434.5 -4 -1 -1 ktht, k1, k2 k2: 3.9(1) x 10 M s -4 -1 -1 3 Scheme 5.4 k1: 6(2) x 10 M s -3 -1 -1 Fitting parameters: k2: 2(2) x 10 M s 528 3 -101.862 -4 -1 k1, k2, kd kd: 5(7) x 10 s 5 -1 -1 4 ktht: 9(300) x 10 M s Scheme 5.4 -4 -1 -1 k1: 6(2) x 10 M s Fitting parameters: -3 -1 -1 528 4 -152.628 k2: 2(3) x 10 M s ktht, k1, k2, kd -4 -1 kd: 5(8) x 10 s -2 -1 Scheme 5.5 k1: 6.2(7) x 10 s -1 -1 5 Fitting parameters: k2: 2(7) M s 528 3 86.4351 -2 -1 k1, k2, K3 K3: 4.5(6) x 10 s -1 -1 ktht: 800(9000) M s Scheme 5.5 -2 -1 6 k1: 3.6(7) x 10 s Fitting parameters: -1 -1 -1 528 4 270.636 k2: 2(22) x 10 M s ktht, k1, k2, K3 -2 K3: 9(2) x 10 k : 4(1) x 10-2 s-1 Scheme 5.5 1 7 k : 6(9) x 10-3 M-1 s-1 Fitting parameters: 2 528 4 -1850.06 K3: 0.13(3) k1, k2, K3, kd -3 -1 kd: 1(2) x 10 s -4 -1 -1 ktht: 8(5) x 10 M s -2 -1 8 Scheme 5.5 k1: 4.3(4) x 10 s -3 -1 -1 Fitting parameters: k2: 6(3) x 10 M s 528 5 -1937.54 ktht, k1, k2, K3, kd K3: 0.13(1) -3 -1 kd: 1.0(5) x 10 s

114

Additional evidence for MoO(tfd)2 being one of the intermediates comes from using MoO(tfd)2 2- to enter the catalytic cycle. Neutral MoO(tfd)2 has not been isolated but MoO(tfd)2 can be synthesized from known methods.10 This substance can be oxidized to neutral MoO(tfd)2 in situ. 2- MoO(tfd)2 has no catalytic activity but upon addition of an oxidant (Magic Blue, ferrocenium tetrafluoroborate) catalytic activity (production of OP(p-tolyl)3 and DMS with concomitant reduction of DMSO and P(p-tolyl)3) was observed.

To verify the next step, the oxygen transfer from MoO(tfd)2 to P(p-tolyl)3 was observed when 2- MoO(tfd)2 is oxidized in the presence of P(p-tolyl)3. A signal corresponding to known OP(p- tolyl)3 was observed by NMR spectroscopy. Since the reaction was performed under air-free conditions the only source of oxygen is presumably from the MoO(tfd)2.

The catalyst exhibits decomposition as the reaction progresses. It starts as a brown orange and turns green to blue. This is accompanied by an observed decrease in activity. A possible explanation is dimerization of the complex to create oxo-bridged species.5 Attempts to crystallize the decomposition products for characterization were unsuccessful.

It is interesting to note that both the decomposition reaction (kd) and the equilibrium reaction

(K3) reduce the overall rate of the reaction as time progresses. The decomposition reaction does so by directly removing the catalyst, while the equilibrium reaction inhibits the catalysis with the product (OP(p-tolyl)3). Both effects increase as time progresses. The most straightforward way to distinguish their effects is to directly vary the initial concentration of OP(p-tolyl)3 and observe how the kinetics are affected.

As a final check to ensure the kinetic data observed is valid, and not the result of a possible reaction between the solvent and the reactants, a completely different solvent, C6D6 was substituted and the exact same analysis was performed (Table 5.2 and Figure 5.8).

115

# Model Results n p RSS AICc -2 -1 5 Scheme 5.5 k1: 2.5(6) x 10 s -1 -1 Fitting parameters: k2: 2(20) M s 266 3 199.475 -68.4034 -2 -1 k1, k2, K3 K3: 2.5(6) x 10 s -1 -1 6 ktht: 10(100) M s Scheme 5.5 -2 -1 k1: 2.5(6) x 10 s Fitting parameters: -1 -1 -1 266 4 191.933 -76.5789 k2: 8(11) x 10 M s ktht, k1, k2, K3 -2 K3: 2.6(7) x 10 7 k : 2.7(4) x 10-2 s-1 Scheme 5.5 1 k : 7(7) x 10-3 M-1 s-1 Fitting parameters: 2 266 4 4.41175 -1080.16 K3: 0.037(5) k1, k2, K3, kd -3 -1 kd: 1(1) x 10 s -1 -1 8 ktht: 0.09(20) M s -2 -1 Scheme 5.5 k1: 2.5(4) x 10 s -3 -1 -1 Fitting parameters: k2: 6(7) x 10 M s 266 5 4.62781 -1065.35 ktht, k1, k2, K3, kd K3: 0.043(6) -3 -1 kd: 1(1) x 10 s

116

Simulated Reaction Progression models 5-8 (data collected from kinetics observed in benzene-d6)

15 concentration (mM) concentration 10

0 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 Time (s)

Actual P(p-tolyl)3 Fitted P(p-tolyl)3 for model 5 Fitted P(p-tolyl)3 for model 6 Fitted P(p-tolyl)3 for model 7 Fitted P(p-tolyl)3 for model 8

In examining the fitted models in Table 5.2, it can be seen many of the rate constants are in reasonable agreement (within an order of magnitude) with the rate constants observed in CDCl3.

Noteworthy differences include the rate for ktht and the much greater error for many of the other constants. As mentioned previously, this is likely due to the possibility that ktht is very large and that particular reaction finishing before it can be observed. Indeed, model 7 that does not include that step now has the lowest AICc value; it better describes the data using as few parameters as possible. But the AICc values for both are relatively close compared to other models. Further experimentation will be needed to truly distinguish them.

117

The greater errors in the other rate constants can be attributed to fewer analytical signals being measured. This was because the NMR signals for the other reagents and products overlapped with each other. With fewer signals there is less data to accurately derive constants from.

Nonetheless the reactions in C6D6 do provide some interesting insights when we examine either of the two best models (7 & 8) and compare them across solvents. The difference in equilibrium constant K3 (0.043(6) for C6D6 and 0.13(1) for CDCl3) can easily be attributed to the different solvents used. Interestingly, the color of the solution as it reacts is not red as it is in chloroform, but light amber green in benzene. This may indicate possible solvent participation in the mechanism and may also contribute to the differing equilibrium constant. The rate constant k2, that is believed to correspond to the attack of the phosphine onto the MoO(tfd)2 complex, is -3 -1 -1 -3 -1 -1 almost the same across both solvents (6(7) x 10 M s for C6D6 and 6(3) x 10 M s for

CDCl3), possibly indicating a solvent-independent transition state for this reaction. Of great interest is k1 that is believed to correspond to the removal of oxygen from DMSO. In CDCl3 it is -2 -1 -2 -1 approximately 4.3 x 10 s while in C6D6 it is approximately 2.5 x 10 s . The lower value in

C6D6 likely indicates a polar transition state that is less favored in the relatively less polar solvent as opposed to CDCl3.

As mentioned in the introduction Holm et al. have made an extensive and organized compilation of kinetic data for many oxotransferase complexes. Interestingly most of those complexes follow second order kinetics for the oxygen transfer from DMSO but it can be seen that k1 is first order. This observation is further cooroborated when plotting the logarithm of rates versus concentrations of DMSO (Figure 5.9), it can be seen the line has a slope of –0.0558, but the standard deviation on the slope was found to be 0.0584. With such a large error it may be that the slope is zero, thus the concentration of DMSO appears to have no effect on the rate and makes the DMSO reduction step first order.

118

Logarithms of rate versus concentration of DMSO

-12.2 -4.3 -4.1 -3.9 -3.7 -3.5 -3.3 -3.1 -2.9 -2.7 -2.5

-12.22

-12.24 y = -0.0558x - 12.446

-12.26

-12.28 Rate Ln([M]/s) -12.3

-12.32

-12.34 Ln Concentration, ln([M])

[DMSO] Linear ([DMSO])

Figure 5.9: Determination of reaction order with respect to DMSO by method of initial rates and plotting the logarithm of initial rate with respect to DMS versus concentration of DMSO.

There is no known dithiolene-based DMSO reductase model complex that follows first order kinetics for the DMSO reduction step. Mo(tfd)2(tht)2 appears to be the first such complex. Thus, there is nothing to directly compare against. However, it may be that the K3 equilibrium step with phosphine oxide is dominating the cycle and interfering with an accurate assessment of the effects of DMSO. To observe only the rate of DMSO reduction by Mo(tfd)2(tht)2 a future experiment would be performed without phosphine. A difficulty in this is that the reaction has been observed to proceed very quickly and may not be observable by NMR spectroscopy. Different techniques will be needed.

119

Alternatively, the observed rate may indeed be representative of the underlying chemistry between DMSO and Mo(tfd)2(tht)2. If so, it is not known what particular factor makes

Mo(tfd)2(tht)2 so different kinetically. All of the complexes reviewed by Holm were dianions or monoanions, while Mo(tfd)2(tht)2 is neutral. It may be that all the complexes studied by Holm do proceed by Scheme 5.2 but the extra available electron in Holm’s complexes makes the M[XO] reduction step proceed faster than the initial formation step, so in Holm’s studies the rate determining step is second order and that is what is observed, while in Mo(tfd)2(tht)2 the reduction step is slow enough to be rate determining so that is observed. Further experimentation will be needed.

5.5 Conclusion

There is still a great deal of work to be done to fully explore the kinetics of this reaction. A higher field NMR instrument may allow resolution of the overlapping reagent and product signals and allow the kinetics to be more accurately probed in C6D6. Alternatively different sulfoxides and phosphines could be used with signals that do not overlap in C6D6. More accurate measurement of ktht can be performed using a separate UV-Vis technique. Additional experiments varying the starting concentrations of the products, DMS and OP(p-tolyl)3, will help determine if they play some role in the catalytic cycle.

Additional future work would continue attempts to crystallize the decomposition products to determine if dimerization is indeed the cause of the decrease in activity later on in the reaction. Another approach would be to prevent dimerization by creating a sterically hindered complex as has been successfully demonstrated with other oxotransferase model complexes based on dithiocarbamate ligands.11 However doing so with a dithiolene based system while still maintaining similar electronic features presents unique synthetic challenges.

If indeed the proposed mechanism is correct where phosphine oxide is strongly binding and must first be substituted by DMSO then crystallization and of the phosphine oxide bound metal complex would provide additional evidence of the existence of this species in the catalytic cycle. Alternatively, different oxygen acceptors may be used whose oxides have different binding

120 affinities. If the model is correct then those oxides that have lower binding affinity should increase the overall rate of reaction.

121

5.6 References

1 Hine, F. J.; Taylor, A. J.; Garner, C. D. Coord. Chem. Rev. 2010, 254, 1570–1579.

2 Veldman, A.; Santamaria-Araujo, J. A.; Sollazzo, S.; Pitt, J.; Gianello, R.; Yaplito-Lee, J.; Wong, F.; Ramsden, C. A.; Reiss, J.; Cook, I.; Fairweather, J.; Schwarz, G. Pediatrics 2010, 125, e1249–54.

3 Das, S. K.; Biswas, D.; Maiti, R.; Sarkar, S. J. Am. Chem. Soc. 1996, 118, 1387-1397.

4 Holm, R. H.; Solomon, E. I.; Majumdar, A.; Tenderholt, A. Coord. Chem. Rev. 2011, 255, 993- 1015.

5 Enemark, J. H.; Cooney, J. J. A.; Wang, J.-J.; Holm, R. H.; Chem. Rev. 2004, 104, 1175.

6 Kuzmic, P. Anal. Biochem. 1996, 237, 260.

7 Castrillón, J. P. A.; Szmant, H. H. J. Org. Chem. 1965, 30, 1338–1339.

8 Das, S. K; Chaudhury, P. K.; Biswas, D.; Sarkar, S. J. Am. Chem. Soc., 1994, 116, 9061–9070

9 Burnham, K. P.; Anderson, D. R. Model Selection and Multimodel Inference, 2nd ed.;Springer: New York, 2002

10 Davison, A.; Holm, R. H. Inorg. Synth. 1967, 10,8

11 Moradi-Shoeili, Z.; Boghaei, D. M. Spectrochim. Acta A 2012, 88, 210–215.

122

Chapter 6: Summary and Conclusion

Molybdenum bis(dithiolene) complexes have shown a great deal of interesting chemistry from new catalytic properties to new ligand based reactions on dithiolenes.

6.1 Chapter 2

A straightforward lab scale synthesis of tfd was presented using commercially available reagents and equipment. 1,2-dichloro-1,2,3,3,4,4-hexafluorocyclobutane was dechlorinated with zinc in ethanol to produce 1,2,3,3,4,4-hexafluorocyclobutene, isomerized at 590oC over KF to produce hexafluorobutyne and then reacted with sulfur in a gas phase reaction at 450oC to produce bis(trifluoromethyl)dithiete (‘tfd’). The key pieces of enabling equipment were the use of liquid nitrogen cooled apparatus to handle the gaseous intermediates, and the use of heating tape to safely produce the high temperatures required that were outside of most standard laboratory heating devices. Future work should look into alternative liquid-phase reactions and ascertain the reliability and quality of reducing 2,3-dichloro-1,1,1,4,4,4-hexafluorobutene with zinc to produce the required hexafluorobutyne.

6.2 Chapter 3

The molecular compounds Mo(tfd)2(dht)2 and Mo(tfd)2(tht)2 were synthesized and shown by X- ray crystallography to be excellent structural models for the proposed active site of molybdenum disulfide, a catalyst used in hydrodesulfurization, matching the configuation of the sulfur atoms around the metal center, and the ability to bind thioethers in the same binding mode. Preliminary investigations show some intriguing reactivity consistent with transfer dehydrogenation, yet temperature-sensitivity hampers full exploration of this avenue. The new complexes are useful, however, to test equilibria at room temperature, as demonstrated with a dht/tht competition on a sulfur-ligated Mo(IV) center. Future model complexes will need to be designed for greater temperature stability and it may be necessary to move away from close structural models as well as consider multimetallic model complexes.

123

6.3 Chapter 4

We conclude that nucleophilic addition to a non-innocent ligand can open a coordinatively saturated complex and thus enhance reactivity. In Mo(tfd)2(bdt), a phosphine can directly add to the bdt ligand creating a zwitterionic SC6H4SPPh3 monodentate ligand, giving a complex that is a structural DMSO reductase model and a pre-catalyst to a functional DMSO-reducing system. Future work will expand on the generality of the new logical but initially counter-intuitive approach to make an open site: via addition of a nucleophile, but to the ligand.

6.4 Chapter 5

Research into the mechanism of oxo transfer catalyzed by Mo(tfd)2(tht)2 is still ongoing. Additional future work would include attempts to crystallize the decomposition products to determine if dimerization is the cause for the decrease in activity as the reaction progresses. Crystallization of the possible phosphine oxide complex would add evidence to the validity of the proposed mechanism. And different oxygen acceptors are predicted to increase the reaction rate if their oxides bind less strongly to the catalyst. As the first known example of a DMSO reductase model complex that obeys first order kinetics for the DMSO reduction step, further research will be needed to determine why it is so different from all existing model complexes.

6.5 Final Remarks

The chemistry of metal dithiolenes is rich and fascinating, and just one specific complex, molybdenum bis(trifluoromethyl) dithiolene “Mo(tfd)2X” and its related compounds, has yielded noteworthy contributions to the field. The dithiolene has been shown to be a very versatile catalyst and has demonstrated dehydrogenation, alkene isomerization and oxotransferase activity. Intriguingly, we have found coordinatively saturated complexes like Mo(tfd)2(bdt) can be opened up by nucleophilic attack onto the ligand, a reaction not seen before in a dithiolene. This opens up a whole new approach for activating dithiolenes and creating new ligands. Central to all these reactions is the non-innocence of the dithiolene ligand that makes them possible.