Assembly of Multifunctional Materials Using Molecular Cluster Building Blocks

Home , Cyanometalate, Dimanganese decacarbonyl, Octahedral cluster

Assembly of Multifunctional Materials Using

Molecular Cluster Building Blocks

Bonnie Choi

Submitted in partial fulfillment of the

requirements for the degree of Doctor of Philosophy

in the Graduate School of Arts and Sciences

Columbia University

2018

Bonnie Choi

Abstract

Assembly of Multifunctional Materials Using Molecular Cluster Building Blocks

Bonnie Choi

This thesis explores the synthesis, properties, and potential applications of molecular clusters and the hierarchical solids that form when complementary clusters are combined. Chapter

1 introduces the diverse set of molecular clusters that I employ as nanoscale building blocks in the assembly of multifunctional materials. The core structure of the molecular clusters is closely related to the superconducting Chevrel phases. In discrete clusters, however, the core is passivated by organic ligands, which add stability and important functionalities. The molecular clusters have rich physical and chemical properties of their own, and I present some of the techniques used to investigate their intrinsic electronic properties. Finally, I review some of the modes by which the molecular clusters interact with another to assemble into hierarchical solids. The structural tunability and complexity embedded in the molecular clusters will enable the design of modular, well-defined, multifunctional materials with desirable electronic and magnetic properties.

Chapter 2 details the synthesis and characterization of a family of manganese telluride molecular clusters. By varying the ligands that decorate the surface of the inorganic core, I show that the core structures can be tuned. The study of molecular clusters provides insight into how extended solids form. As such, I make structural comparisons of the clusters to known solid-state compounds. Being structurally varied and chemically flexible, the clusters reported in this chapter present an exciting new class of building blocks for the assembly of solid-state compounds.

In Chapters 3-4, I present a nanoscale approach to investigate the electronic behaviors of individual molecular clusters. By using a scanning tunneling microscope-based break-junction

technique and density-functional theory calculations, I study the effects of the junction environment and the redox properties of the molecular clusters on the conductance of single-cluster junction. Importantly, current blockade effect is observed at room temperature in the single-cluster junctions, allowing for the conductance to be turned on or off by varying the bias potential.

Chapters 5-7 explore the synthesis and properties of the hierarchical solids comprised of molecular cluster building blocks. Chapter 5 unveils an approach to create a three-dimensional

(3D) coordination network of molecular clusters by using a bifunctional cyanide ligand. The cyanide ligand is appended to the metal sites of the cluster through the carbon terminus, leaving the nitrogen end available for coordination by a divalent metal cation. Whereas the molecular cluster itself is paramagnetic across a temperature range of 3-300 K, the 3D coordination compound shows a ferromagnetic transition at ~25 K. In Chapter 6, I describe the importance of a molecular recognition feature on the molecular cluster that contributes to the assembly of a layered, van der Waals solid. The bulk material contains monolayers of fullerene and can be mechanically exfoliated to thinner layers, providing a key templated strategy to isolate free monolayers of fullerene. Lastly, Chapter 7 details layered, van der Waals solids of rhenium and molybdenum synthesized using traditional solid-state reactions. Because the neighboring cluster units are covalently bound together, the inter-cluster coupling is much stronger in the plane of these materials than that of the self-assembled solid described in Chapter 6. The strong two- dimensional (2D) character in these layered materials allows for the exfoliation of bulk crystals into robust, low-defect monolayers. The surfaces of these monolayers are covered with substitutionally labile ligands, which is an atypical yet valuable feature among 2D materials. I demonstrate that the electronic properties of the monolayers can be tuned by exchanging the surface ligands.

TABLE OF CONTENTS

List of Figures ...... v

List of Tables...... x

Acknowledgements...... xi

Chapter 1. Chevrel Phase Molecular Clusters and Cluster-Assembled Materials ...... 1 1.1. Introduction ...... 1 1.2. Dimensional Reduction of Chevrel Phases ...... 2 1.3. Solution-Phase Synthesis of Molecular Clusters ...... 3 1.3.1. Cluster Core ...... 4 1.3.2. Ligand Shell ...... 5 1.4. Electronic Properties of Molecular Clusters ...... 5 1.4.1. Background of the STM-BJ Technique ...... 6 1.4.2. Environment Effect ...... 7 1.4.3. Linker Group ...... 7 1.4.4. Coupling between Electrode and Cluster ...... 8 1.5. Assembly of Molecular Clusters into Hierarchical Solids ...... 9 1.5.1. Charge Transfer ...... 9 1.5.2. Molecular Recognition...... 10 1.5.3. Metal Coordination ...... 10 1.5.4. Increasing the Inter-Cluster Coupling ...... 11 1.6. Structural Characterization ...... 12 1.7. Redox Properties ...... 12 1.8. Collective Properties ...... 12 1.9. Summary and Outlook ...... 13 1.10. References ...... 13

Chapter 2. Structural, Dimensional, and Interfacial Diversity in Manganese Telluride Molecular Clusters ...... 21 2.1. Preface...... 21 2.2. Introduction ...... 21 2.3. Synthetic Design ...... 22 2.4. Dissociation of Phosphine Ligands ...... 28 2.5. Aggregated Growth into Solid-State Materials...... 32 2.6. Comparison to Bulk MnTe Phases ...... 33 2.7. Conclusions ...... 35 2.8. General Synthesis Information ...... 35 i

2.9. Synthetic Procedures and Characterization of Compounds ...... 36 2.10. Instrumentation ...... 43 2.11. Structural Determination ...... 45 2.12. Powder X-Ray Diffraction ...... 57 2.13. References ...... 64

Chapter 3. Environmental Effect on Single-Cluster Junctions ...... 70 3.1. Preface...... 70 3.2. Introduction ...... 70 3.3. Synthetic Design ...... 71 3.4. STM-BJ Measurements and Histogram Analysis ...... 72 3.5. Conductance Decay Constants for the Ligand and Cluster Series ...... 73 3.6. DFT Calculations ...... 74 3.7. Tight-Binding Model ...... 76 3.8. Conclusions ...... 80 3.9. General Synthesis Information ...... 81 3.10. Synthetic Procedures and Characterization of Compounds ...... 81 3.11. Instrumentation ...... 87 3.12. Additional Conductance Data ...... 87 3.13. Additional Details on the Tight Binding Model ...... 90 3.14. Details on the DFT Calculations ...... 91 3.15. References ...... 92

Chapter 4. Room Temperature Single-Cluster Current Blockade ...... 95 4.1. Preface...... 95 4.2. Introduction ...... 95 4.3. Synthetic Design ...... 97 4.4. STM-BJ Measurements ...... 98 4.5. Histogram Analysis and I-V Plots ...... 99 4.6. Two-Step Incoherent Transport Mechanism ...... 106 4.7. DFT Calculations and Sequential Tunneling ...... 107 4.8. Current Blockade in Mo-Based Single-Cluster Junction ...... 108 4.9. Conclusions ...... 112 4.10. General Synthesis Information ...... 113 4.11. Synthetic Procedures and Characterization of Compounds ...... 113 4.12. Instrumentation ...... 116 4.13. Structural Determination ...... 117 4.14. STM-BJ Experimental Details ...... 121 4.15. Details on the DFT Calculations ...... 122 ii

4.16. References ...... 122

Chapter 5. Three-Dimensional Network of Cyanide-Linked Molecular Clusters ...... 126 5.1. Introduction ...... 126 5.2. Synthesis and Properties of Cyanide-Ligated Fe6S8 Molecular Clusters ...... 127 5.3. Assembly into 3D Coordination Network ...... 129 5.4. Magnetic Properties of Cyanide-Bridged Fe6S8 Clusters ...... 132 5.5. Conclusions ...... 134 5.6. General Synthesis Information ...... 135 5.7. Synthetic Procedures and Characterization of Compounds ...... 135 5.8. Instrumentation ...... 136 5.9. Structural Determination ...... 138 5.10. References ...... 138

Chapter 6. Self-Assembly of a Layered van der Waals Material Using a Structure-Directing Cluster...…...... 142 6.1. Preface...... 142 6.2. Introduction ...... 143 6.3. Synthesis of Structure-Directing Cluster ...... 144 6.4. Self-Assembly into a Layered Solid and Mechanical Exfoliation ...... 146 6.5. Molecular Recognition of the Ligands and Fullerenes ...... 148 6.6. Determination of Charge on the Layered Solid ...... 151 6.7. Optical and Electronic Properties of the van der Waals Crystal ...... 152 6.8. Photopolymerization of the Template ...... 153 6.9. Conclusions ...... 156 6.10. General Synthesis Information ...... 156 6.11. Synthetic Procedures and Characterization of the Compounds ...... 157 6.12. Instrumentation ...... 158 6.13. Structural Determination ...... 161 6.14. X-Ray Structural Analysis of Cluster Oxidation State ...... 164 6.15. Powder X-Ray Diffraction ...... 164 6.16. Details on the Raman Spectroscopy ...... 167 6.17. Optical Absorption Measurement ...... 169 6.18. References ...... 171

Chapter 7. Layered Chevrel phases of Re and Mo ...... 176 7.1. Introduction ...... 176 7.2. Synthesis and Characterization of Bulk Crystals ...... 177 7.3. Mechanical Exfoliation of Crystals ...... 179 iii

7.4. Chemical Intercalation and Exfoliation of Re6Se8Cl2 ...... 180 7.5. Surface Functionalization of Monolayers ...... 184 7.6. Conclusions ...... 188 7.7. General Synthesis Information ...... 188 7.8. Synthetic Procedures ...... 189 7.9. Instrumentation ...... 190 7.10. Additional Details on the Raman Data ...... 192 7.11. Additional Details on the Electronic Absorption Data ...... 193 7.12. Additional Details on EDS...... 193 7.13. Structural Determination ...... 194 7.14. References ...... 195

LIST OF FIGURES

Figure 1.1. Structure of Chevrel phases ...... 1

Figure 1.2. Family of molecular clusters with a M6E6 core ...... 4

Figure 1.3. Schematic of a single-cluster junction in the STM-BJ set-up ...... 6

i Figure 2.1. Molecular structure of Mn4Te4(P Pr3)4 ...... 23

Figure 2.2. Schematic of manganese telluride molecular clusters synthesis ...... 25

Figure 2.3. Molecular structures of Mn4Te4(PEt3)4 and Mn4Te4(PEt3)4 ...... 26

Figure 2.4. Molecular structure of Mn6Te6(PEt3)6 ...... 27

i i Figure 2.5. Molecular structures of Mn4Te4( Pr2NHC)4 and Mn8Te8( Pr2NHC)6 ...... 30

1 31 i Figure 2.6. H NMR and P NMR spectra of Mn4Te4(P Pr3)4, product of the reaction of i i i Pr2NHC with Mn4Te4(P Pr3)4, and Mn4Te4( Pr2NHC)4 as directly synthesized...... 31

Figure 2.7. PXRD of α-MnTe upon pyrolysis of the orange aggregate product ...... 33

Figure 2.8. Pictorial comparison of Mn6Te6 cluster and wurtzite lattice type...... 34

Figure 2.9. Electronic absorption spectrum of Mn4Te4(PEt3)4 ...... 37

i Figure 2.10. Electronic absorption spectrum of Mn4Te4(P Pr3)4 ...... 38

Figure 2.11. Electronic absorption spectrum of Mn4Te4(PCy3)4 ...... 39

Figure 2.12. Electronic absorption spectrum of Mn6Te6(PMe3)6...... 40

Figure 2.13. Electronic absorption spectrum of Mn6Te6(PEt3)6 ...... 41

i Figure 2.14. Electronic absorption spectrum of Mn4Te4( Pr2NHC)4 ...... 42

i Figure 2.15. Electronic absorption spectrum of Mn8Te8( Pr2NHC)6 ...... 43

Figure 2.16. Reconstructed precession frames of the 0kl and 1kl reciprocal layers ...... 48

Figure 2.17. Diffraction image showing streaks and diffuse scattering ...... 52

Figure 2.18. Synthetic precession image (h0l section; h right, l up)...... 52

Figure 2.19. Simulated PXRD of Mn4Te4(PEt3)4 ...... 58

Figure 2.20. PXRD of Mn4Te4(PEt3)4...... 58 v

Figure 2.21. Simulated PXRD of Mn6Te6(PEt3)6 ...... 59

Figure 2.22. PXRD of Mn6Te6(PEt3)6 ...... 59

i Figure 2.23. Simulated PXRD of Mn4Te4(P Pr3)4 ...... 60

i Figure 2.24. PXRD of Mn4Te4(P Pr3)4...... 60

Figure 2.25. Simulated PXRD of Mn4Te4(PCy3)4 ...... 61

Figure 2.26. PXRD of Mn4Te4(PCy3)4...... 61

i Figure 2.27. Simulated PXRD of Mn8Te8( Pr2NHC)6 ...... 62

i Figure 2.28. PXRD of Mn8Te8( Pr2NHC)6...... 62

Figure 2.29. Simulated PXRD of Mn6Te6(PMe3)6 ...... 63

Figure 2.30. PXRD of Mn6Te6(PMe3)6...... 63

Figure 3.1. Structure of the cluster core Co6Se8 and conducting ligand series Ln ...... 71

Figure 3.2. Logarithmically-binned conductance histograms for 1-3 and L1-L3 ...... 73

Figure 3.3. Computational studies of conducting ligands Ln and model clusters (PMe3)5Co6Se8(Ln)...... 75

Figure 3.4. Sample transmission functions using the tight-binding model ...... 77

Figure 3.5. 2D conductance vs. displacement histograms for 1-3 and L1-L3 in BrN ...... 88

Figure 3.6. 2D conductance vs. displacement histograms for 1-3 and L1-L3 in TCB ...... 89

Figure 3.7. Logarithmically-binned conductance histograms and 2D conductance vs. displacement histograms for 1 collected in TCB under ambient conditions and under an inert atmosphere of Ar...... 90

Figure 3.8. 2D plot showing the conductance ratio of 1 to L1 ...... 91

Figure 4.1. Structure of the [Co6S8] core and molecular connector L used to wire the cluster into a junction. Schematic of the STM-BJ measurement in an ionic environment ...... 98

Figure 4.2. Logarithmically binned 1D conductance histograms for Co6S8L6 and 2D conductance vs. displacement histogram ...... 99

Figure 4.3. 1D conductance histograms of ligand L measured at different tip biases in PC ..... 100

Figure 4.4. Semi-logarithmic plot of the current vs. tip bias for Co6S8L6 and Co6Se8L6 measured in PC. Sample I-V trace for Co6S8L6. 2D I-V histogram for Co6S8L6. Normalized differential conductance...... 101

Figure 4.5. 1D conductance histogram of Co6S8L6 measured at different gate voltages using a 3- electrode setup ...... 102

Figure 4.6. 1D and 2D conductance histograms of Co6Se8L6 ...... 103

Figure 4.7. CV of Co6Se8L6 ...... 104

Figure 4.8. Sample traces of I-V measurement for Co6S8L6...... 105

Figure 4.9. I-V 2D histogram for Co6Se8L6 and normalized differential conductance ...... 105

Figure 4.10. HOMO and HOMO-13 orbitals of model compound (PMe3)4Co6S8L2 with energies relative to vacuum, as calculated from DFT. Schematic of the proposed sequential tunneling transport mechanism ...... 107

Figure 4.11. Molecular structure of Mo6S8L6 ...... 109

Figure 4.12. Logarithmically binned 1D conductance histograms for Mo6S8L6 and 2D conductance vs. displacement histogram ...... 110

Figure 4.13. Current vs. tip bias for Mo6S8L6. CV of Mo6S8L6. 2D I-V histogram and normalized differential conductance ...... 111

5- Figure 5.1. Structure of [Fe6S8(CN)6] ...... 128

Figure 5.2. CV of 4 ...... 129

Figure 5.3. Temperature dependence of the reciprocal molar magnetic susceptibility and effective magnetic moment of 4 at 500 Oe ...... 129

Figure 5.4. HAADF image of 5’ and EDS mapping of Mn, Fe, and S ...... 130

Figure 5.5. Molecular structure of 5 ...... 131

Figure 5.6. PXRD pattern of 5’ ...... 132

Figure 5.7. Temperature dependence of the reciprocal molar magnetic susceptibility and effective magnetic moment of 5’ at 10 Oe ...... 133

Figure 5.8. Temperature dependence of the ZFC-FC magnetization of 5’ with an external field of 100 Oe and magnetization of 5’ as a function of the applied field above and below the transition temperature...... 134

Figure 6.1. Schematic of a nanoscale director organizing a monolayer of fullerenes ...... 144 vii

Figure 6.2. Electronic absorption spectra of PEt2phen and Co6Se8(PEt2phen)6 ...... 145

Figure 6.3. CV of Co6Se8(PEt2phen)6 ...... 146

Figure 6.4. AFM topographic images of as-grown single crystal and mechanically exfoliated crystal of [Co6Se8(PEt2phen)6][C60]5...... 147

Figure 6.5. Optical images of mechanically exfoliated crystals of [Co6Se8(PEt2phen)6][C60]5 . 148

Figure 6.6. Optical microscope images of as-grown single crystal and mechanically exfoliated crystals of [Co6Se8(PEt2phen)6][C60]5 ...... 148

Figure 6.7. Molecular structure of the nanoscale director Co6Se8(PEt2phen)6 before and after self-assembly...... 149

Figure 6.8. Molecular structure of [Co6Se8(PEt2phen)6][C60]5...... 150

Figure 6.9. Representative Raman spectra of [Co6Se8(PEt2phen)6][C60]5 ...... 151

Figure 6.10. Geometries of Co6Se8(PR3)6 clusters in 0, 1+ and 2+ oxidation states...... 152

Figure 6.11. Log plot of conductance (G) vs. 1/T for [Co6Se8(PEt2phen)6][C60]5 and mid-IR absorption spectrum of [Co6Se8(PEt2phen)6][C60]5...... 152

Figure 6.12. Photopolymerization of [Co6Se8(PEt2phen)6][C60]5 ...... 154

Figure 6.13. TGA of Co6Se8(PEt2phen)6 and [Co6Se8(PEt2phen)6][C60]5 ...... 155

Figure 6.14. PXRD of Co6Se8(PEt2phen)6 ...... 166

Figure 6.15. PXRD of [Co6Se8(PEt2phen)6][C60]5 ...... 167

Figure 6.16. Electronic absorption spectra of [Co6Se8(PEt2phen)6][C60]5 at different temperatures ...... 171

Figure 7.1. Molecular structure of Re6Se8Cl2. Optical microscope image of as grown Re6Se8Cl2 crystals and AFM image of mechanically exfoliated flakes ...... 177

Figure 7.2. PXRD pattern of Re6Se8Cl2 ...... 178

Figure 7.3. Molecular structure of Mo6S3Br6. Optical microscope image of as grown Mo6S3Br6 crystals and AFM image of mechanically exfoliated flakes ...... 179

Figure 7.4. PXRD pattern of Mo6S3Br6 ...... 179

Figure 7.5. Chemical intercalation/liquid exfoliation of Re6Se8Cl2...... 181

viii

Figure 7.6. AFM image of MLs dropcast on a sapphire substrate. High resolution HAADF TEM micrograph of a ML and corresponding SAED pattern. EDS mapping of Cl, Re, and Se ...... 182

Figure 7.7. Optical microscope image of exfoliated nanosheets ...... 183

Figure 7.8. AFM image of exfoliated MLs and corresponding step height histogram ...... 183

Figure 7.9. Surface functionalization reaction of MLs supported on a substrate ...... 185

Figure 7.10. AFM image of MLs after reaction with TMSCN and corresponding step height histogram...... 186

Figure 7.11. Raman spectrum in the 100-400 cm–1 region of the MLs before and after treatment with TMSCN ...... 187

Figure 7.12. Macroscopic and microscopic electronic absorption spectra of MLs before and after reaction with TMSCN...... 187

LIST OF TABLES

Table 2.1. Mean bond distances and angles for [MnTe]n (n = 4, 6, 8) clusters ...... 27

Table 2.2. Selected crystallographic data for manganese telluride molecular clusters...... 55

Table 4.1. Selected crystallographic data for Co- and Mo-based single-cluster junction ...... 120

Table 5.1. Selected crystallographic data for 4 and 5 ...... 138

Table 6.1. Selected crystallographic data for Co6Se8(PEt2phen)6 and [Co6Se8(PEt2phen)6][C60]5 ...... 163

Table 6.2. Selected Raman spectroscopy data for [Co6Se8(PEt2phen)6][C60]5 ...... 169

Table 7.1. Chemical compositions of Re6Se8Cl2 and alkali-doped Re6Se8Cl2 as determined by EDS ...... 194

Table 7.2. Selected crystallographic data for Re6Se8Cl2 ...... 194

ACKNOWLEDGEMENTS

I would first like to thank all the teachers and mentors who have inspired and encouraged me to dive into the field of chemistry. Karen Monda at Ramsey High School urged me to really reflect on how chemical concepts work. She was the first to teach me to think outside of the box and to become an independent academic. Paul Kropp and his intermediate organic chemistry class at UNC Chapel Hill taught me how applicable organic chemistry is to real life and how exciting learning about chemistry can be. I am grateful to Michael Crimmins for allowing me to work as an undergraduate researcher for two years. I especially want to thank my postdoctoral mentor Dan

Carper who taught me the basics of chemical research and everything I needed to know about working in a laboratory setting. The skills I learned from him over the two years at UNC Chapel

Hill proved invaluable throughout graduate school.

I am deeply grateful to my graduate research advisor, Xavier Roy. His lab was just starting up (Ari Turkiewicz was the only lab member then) when I joined his group the summer before graduate school started. Xavier has taught me how to be a scientist who can balance different projects simultaneously, collaborate with other scientists in different fields, and communicate science through writing manuscripts and grant applications and through numerous outreach activities. From the start, he was heavily interested in my projects and was always encouraging and positive when things were not going so well. I am incredibly lucky to have had Colin Nuckolls nearby (literally, as he was on the floor below our lab). He provided me a lot of great ideas about research and life, and he was always extremely generous with me using the resources in his lab.

Mike Steigerwald has been essential to my development as a scientist, and he has been 100 % supportive and caring of my personal and professional growth. He attended every group meeting to give feedback on my research progress. His innate curiosity and enthusiasm about science could xi always turn around a difficult day (or weeks/ months) of work. The fondest memories I will have of graduate school will be stopping by his office to chat about anything – from science to travel destinations around the world.

I am grateful to Latha Venkataraman who was not only my collaborator but also my role model in science. She was always patient and talked me through the details of single-molecule junctions and the various inconsistencies in the language between physics and chemistry. I admire her positive spirit and confidence when dealing with reviewer feedback. I want to thank Brian

Capozzi and Giacomo Lovat in Latha’s lab for their contributions. They were extremely accommodating when handling the tricky and most often air-sensitive molecular clusters.

Xiaoyang Zhu has been an essential mentor to me over the last several years. He had a lot of faith in me to grow larger crystals for spectroscopic measurements, which I finally succeeded in my last two years of graduate school. The one unfortunate aspect was not being able to make something that wasn’t black or dark-brownish black. I would like to thank members of his lab including

Kihong Lee, Xinjue Zhong, Tuan Trinh, and Fang Liu, who were great to work with and were patient when waiting for me to deliver on my promise of larger crystals. Of course, I must thank

Fufy here as well, who provided me with a lot of laughter, cuddles, and emotional support.

I am lucky to have connected with Patrick Batail from Université d’Angers who offered me the idea of working with layered Chevrel phases of molybdenum and rhenium. Because of his sincere optimism and love of science, I always left feeling invigorated and refueled to continue with my research after our conversations. He was incredibly supportive and proud of what I had accomplished, and he gave me a lot of great ideas to pursue. I also want to thank Andrew Crowther from Barnard College for his willingness to look at the Raman features of my materials, Ann

McDermott for helpful discussions about solid-state NMR of molecular clusters, and Dave

xii

Reichman and Andrew Millis for fruitful discussions about how theory can help better understand the clusters and cluster-assembled materials. I want to thank the students and post-docs in their labs for their contributions: Maria Paley in Andrew Crowther’s lab, Yunyao Xu in Ann’s lab, Jia

Chen from Andrew Millis’s lab, and Martina Lessio in Dave’s lab. I also want to thank Jaeeun Yu from Colin’s lab who contributed to the exfoliation of layered van der Waals materials.

Throughout graduate school, I came across a lot of molecular structures to characterize via single crystal X-ray diffraction. The structural details were essential to my projects, and I cannot thank Dan Paley enough for teaching me how to use the instrument, collect and process the data.

These structures were often very difficult to refine, and he was always patient enough to sit with me to solve the structures. My projects would have gone nowhere without his help on other characterization instruments (especially SQUID magnetometer, which managed to break every month or so). I want to thank Raul Hernandez Sanchez from Colin’s lab who brought in a lot of expertise in cluster chemistry and served as an auxiliary post-doc to the Roy lab. I am grateful to have learned about magnetism and data processing from him.

I want to thank Ryan Hastie, Alison Doyle, Dani Farrell, Carlos Garcia, Vivian Polanco,

Chris Cecilio, Daniel Lugo, Jay Kirschenbaum, and the rest of the Columbia Chemistry staff for their administrative support throughout graduate school. I want to thank Dario Vasquez, Nava

Sternberg, and the rest of the MRSEC staff for their support. I am also grateful to the National

Science Foundation for funding my graduate studies.

I want to thank my colleagues and friends for their support throughout the years. Andrew

Pinkard and Evan O’Brien were my classmates, labmates, and officemates. They provided me with a lot of great ideas for my research and were always willing to talk things through. I am lucky to have had Anouck Champsaur from Colin’s lab so close by; she sincerely cared about my projects

xiii and gave key feedback on a lot of materials. Anastasia Voevodin and Evan Doud were part of Roy

Group Generation 2, and I am entirely reassured about leaving the lab to them. Of course, it is not only for lab- or research-related reasons I am thankful to them. The support and care I received from them throughout graduate school have been incredible. I can always count on them for good laughs, a glass of wine, or a pep-talk to turn bad days into good ones. I would also like to thank the rest of the Roy lab members for their contributions. I want to thank the former and current members of SHP who somehow manages to make Saturday mornings fun and Dessert Club for making every Friday a treat. I am grateful to the rest of my friends from chemistry and non- chemistry for always having my back. I also want to thank all the furry friends I got to dogsit through DogVacay/Rover.

I am entirely indebted to my family and Sam’s family for their unconditional love and support throughout the years. My parents, Yoon Choi and Kyungsun Sung, made tremendous sacrifices (which I don’t think I could ever do myself) to move across the world for my brother,

Jason, and me. We were not familiar with the language, knew no one, and had very little. But, we persevered. The gratitude I feel for my parents is beyond what words can describe. I hope you guys are proud of me, and I thank you for your endless courage and love.

Finally, I would not be where I am today without Sam. He has been by my side through it all – as my best friend and partner. He has always been the first person to celebrate my accomplishments and the first person to console my struggles. Everyone who has had a chance to get to know Sam can vouch for his buoyant and optimistic energy that diffuses across the entire room. I am exceptionally blessed to be loved unconditionally by Sam, even through the times when

I don’t think I deserve it. Thank you for always being here for me and cheering me on. I cannot wait to see what unfolds in the next chapters of our lives.

xiv

To my parents

Chapter 1. Chevrel Phase Molecular Clusters and Cluster- Assembled Materials

1.1. Introduction

A well-studied class of solid-state compounds includes the Chevrel phases with an

1 archetypal chemical formula, MxMo6E8 (M = transition metal; E = chalcogen). The Chevrel phases constitute a remarkable number of materials with fascinating physical and chemical properties. Since they have first gained attention for their superconducting properties,2-5 in particular high transition temperatures and critical fields, other potential applications have emerged, including batteries,6-8 catalysis,9, 10 and thermoelectric technologies.11-13

Figure 1.1. Structure of Chevrel phases, MxMo6E8. Molybdenum, purple; chalcogen, green; metal cation, orange.

The basic building block shown in Figure 1.1 is common to the Chevrel phases. An octahedron of Mo atoms is concentric with a cube of chalcogen atoms (i.e., S, Se, Te).

Compositional variations are possible with the substitution of Mo atoms for other transition metal atoms and the chalcogens for halide atoms. In some structures, metal cations occupy the sites within the crystalline lattice that extends three dimensionally. The most common method of their 1 preparation is to mix stoichiometric amounts of the elements in a fused silica tube and to heat the mixture at high temperatures (e.g., 800 - 1000 °C). Because they are formed using high temperature solid-state reactions, the structures are most likely the thermodynamically favored phases. By using a diffusion technique at intermediate temperatures (e.g., 400 - 500 °C), different metal ions can be inserted into the channels of Chevrel phase solids.14

1.2. Dimensional Reduction of Chevrel Phases

A significant body of work has been dedicated to understanding the structural and reactivity relationships between the molecular components and their extended arrays. This work inspired the excision of cluster core from the Chevrel phase solids into frameworks of reduced dimensionality and connectivity, which include two-dimensional sheets, one-dimensional chains, and even discrete molecular clusters. However, the top-down approach to obtaining molecular fragments of

Chevrel phases often involve harsh or intricate reaction conditions. For example, the excision of

15, 16 [Mo6Se8] cluster core from the parent Mo6Se8 solid is carried out in molten cyanide salts.

Metal halides can be integrated in the high-temperature solid-state reactions of Re6E8, to isolate

17, 18 solids of [Re6E8] with reduced dimensionalities and even discrete molecular clusters. Other methods to obtain discrete molecular clusters include ligand exchange in the cluster core of octahedral metal halide solids19, 20 or reductive dimerization of trinuclear fragments.21 Nonetheless, these reactions are often low yielding or contain a mixture of unknown compounds as side- products.

1.3. Solution-Phase Synthesis of Molecular Clusters

An alternative method of making discrete cluster fragments of Chevrel phases involves solution-phase assembly with organometallic reagents and stabilizing organic ligands. This bottom-up approach uses much milder conditions to prepare monodisperse, atomically precise molecular clusters. Crucially, the development of solution-phase assembly has expanded the family of Chevrel phase molecular clusters to include 3d transition metal-containing cores, as studied extensively by Holm,22, 23 Coucouvanis,24, 25 Saito,26 Fenske,27, 28 Cecconi,29, 30

Steigerwald,31-33 and others. The molecular clusters are not only attractive as large molecules with interesting chemical and physical properties,23, 34, 35 but they also offer practical advantages for the synthesis of solid-state compounds. The use of precursor compounds allows for much lower reaction temperatures to obtain bulk extended solids and for the preparation of new phases that may be inaccessible by traditional solid-state routes. Furthermore, comparing the structures of the molecular clusters with those of extended solids has provided insight into the chemistry of molecules-to-solids process.31-33, 36

The strategies to synthesize polynuclear metal chalcogenide complexes are briefly described here. One method involves self-assembly of metal cations and chalcogen anions in the presence of stabilizing ligands. Among many sources of chalcogen atoms, a common reagent is

29, 30 H2E. Transition metal sulfide clusters have been synthesized using H2S, including the iron sulfide clusters Fe6S8L6 (L = stabilizing ligand) described in Chapter 5. The use of H2Se and H2Te are rare due to the toxicity and difficulty in handling these reagents. An alternative method combines a low-valent organometallic complex and a functionally low-valent chalcogen. In this respect, trialkylphosphine chalcogenides (R3PE), which are milder reagents to handle than H2E, have been demonstrated as soluble and stable organometallic sources of chalcogen.37, 38 Despite

3 their stability, phosphine chalcogenides of selenium and tellurium can transfer zero-valent chalcogen to metals due to the poor E-P orbital overlap and more favorable M-E bonding. Because of the greater strength of S-P bond, metal sulfide clusters are not easily synthesized using phosphine sulfides. By controlling the ratio of the reagents, clusters with different core structures or with mixed ligands can be isolated.33, 39

1.3.1. Cluster Core

The field of polynuclear metal chalcogenide chemistry is expansive, as highlighted by comprehensive reviews over the past few decades.40, 41 Here, I focus on transition metal chalcogenide molecular clusters of a single extended family: ligand-passivated, roughly fourteen- atom spherical cages (Figure 1.2).21, 27, 41, 42

Figure 1.2. Family of roughly fourteen-atom-cage molecular clusters with a M6E6 core. Adapted from reference 45.

With M = V, Cr, Fe, Co and Ni; E = S, Se and Te; and L = two-electron donor ligand, this

31, 32, 43 44 family includes the Chevrel type cage M6E8L6, the inverse structure M8E6L8, as well as

28 33 their respective chalcogen- and metal-filled analogs, M6E9L6 and M9E6L8. Each member of this family may be viewed as an exoskeleton of six metal atoms and six chalcogens that assemble to

45 form a stack of two chair-type six-membered rings analogous to our Mn6Te6 prismane cluster that will be described in more detail in Chapter 2. Supplementing the hexagonal core with two 4

44 metals produces the M8E6 cluster (e.g., Ni8Se6 ). The inverse structure or Chevrel type M6E8 is obtained by shifting the metal atoms towards the center of each hexagonal ring and complementing

31 43 the slightly deformed prismane core with two chalcogen atoms (e.g., Cr6Te8, Fe6Te8, and

37 Co6Te8 ). In addition to these basic structures, the M8E6 and M6E8 cores can be filled with an

33 28 additional metal atom (e.g., Ni9Te6 ) and with an oxygen atom (e.g.,V6Se8O ), respectively.

1.3.2. Ligand Shell

The inorganic cores of the molecular clusters are passivated by a ligand shell that provides both stability and solubility to the molecular clusters. While the ligands prevent communication

(e.g., electronic or magnetic) between neighboring clusters, they embed important functionalities.

As will be detailed in Chapters 3-4, decorating the cluster core with bifunctional ligands allows for fundamental electronic studies of the molecular clusters. In other cases, the ligands assist in the self-assembly of molecular clusters into hierarchical solids, as will be detailed in Chapters 5-

6. Finally, the ligands that stabilize the surfaces of cluster-based materials can be exchanged for other types of ligands to modify the materials’ properties, and this feature will be highlighted in

Chapter 7.

1.4. Electronic Properties of Molecular Clusters

The molecular clusters that we employ in the assembly of multifunctional materials have rich physical chemistry of their own. We explore the electronic properties of individual molecular clusters by connecting them to nanoscopic gold electrodes and measuring the conductance of a single-cluster unit. To do so, we use a scanning tunneling microscope-based break junction (STM-

BJ) technique (Figure 1.3). The motivation to study single-cluster electric circuits is to help

5 elucidate the structure-function relationships, which will eventually enable the rational design of molecular clusters with desirable properties.

Figure 1.3. Schematic of a single-cluster junction in the STM-BJ set-up. The cluster core is decorated with bifunctional ligands that attach the cluster to the nanoscopic gold electrodes. Conductance is measured across the junction when the cluster is in contact with both electrodes (i.e., substrate and tip).

1.4.1. Background of the STM-BJ Technique

The ultimate limit of miniaturization of electronics is to use discrete molecules to fabricate the electric components. A major hurdle in the development of molecular electronics was the mechanical difficulty in connecting the electrodes to the molecules at the nanoscale. In 2003, Tao et al. addressed this key issue by using a scanning tunneling microscope (STM) tip to connect an individual molecule and create a reliable single-molecule electric circuit.46 This technique involves bringing the STM tip into contact with the molecule in solution and then gradually retracting the tip. The conductance across the electrode-molecule-tip is measured until the tip is stretched beyond the length of the molecule and the junction is ruptured. The process of formation, elongation, and rupture of the junction is repeated thousands of times to obtain statistical data.47 From these measurements, one-dimensional (1D) and two-dimensional (2D) conductance-displacement histograms can be created.

1.4.2. Environment Effect

The environment (i.e., solvent molecules) around the molecule connected to the junction has various effects on the conductance.48, 49 Such a solvent-induced effect is attributed to the solvent’s ability to tune the work function of the electrodes.48 It is also possible to control the junction’s electrostatic environment by exposing very different areas of the electrodes to an ionic

2 solution. The STM tip is insulated with Apiezon wax such that its area is reduced to ~1 µm whereas the gold substrate (and electrode) retains an area greater than 1 cm 2. This asymmetry in size affords control of the electrostatic environment around the tip and substrate such that bias polarity across the junction can be modulated. We assign this environment effect to the formation of a much denser double layer near the smaller area tip compared to the substrate.50

1.4.3. Linker Group

The linker group attaches the molecular cluster to the electrodes mechanically and electronically. Gold is primarily used as the electrode material because it remains inert throughout most electronic processes. As such, we focus on aurophilic ligands. The way the linker group interacts with the electrode dictates whether the molecule transports holes or electrons. For phenyl ligands terminated with an alkylthiol (-SR) group at the para- position, the dominant conducting molecular orbitals are in the highest-occupied molecular orbitals (HOMO).51 In Chapters 3-4, we employ ligands that contain a thiomethyl group at the para- position of the phenyl ring to connect the molecular cluster to the gold electrodes. The other terminus of the ligand contains a phosphine group that coordinates to the metal atoms via the M-P bond.

1.4.4. Coupling between Electrode and Cluster

By varying the structure of the linker group, the electronic coupling between the cluster core and the electrodes can be dramatically changed. Whereas the meta-substituted linker alone shows a conductance peak across the electrode-linker-tip junction, the molecular cluster ligated with the same linker group shows no conductance peak, indicating that there is no electronic pathway in which a carrier can travel from one electrode to the other through the cluster system.52

By synthetically engineering the linker groups to have both strong and weak coupling to the electrodes, molecular switches have been made, where the conductance can be turned on or off by switching the conductance pathway through the molecule.53, 54

While there have been many synthetic efforts to strengthen the coupling between the molecule and the electrodes to reach higher conductance values,55-57 poor coupling can also be beneficial and elicit interesting physical phenomena. One of the features that can arise from poor coupling is the Coulomb blockade effect.58 This effect refers to the increased resistance at small bias voltages in a single-electron device. Instead of following classical Ohm’s law, the resistance in single-electron features discrete steps, as the electrons in the device create a strong Coulombic repulsion against a flow of electrons under certain threshold voltages. Though rare, the Coulomb blockade effect has been demonstrated in specific organic molecules,59 single ion bonded to polypyridyl ligands,60, 61 and quantum dots.62, 63

Through density-functional theory (DFT) calculations, we find that the HOMO of our

Chevrel phase molecular clusters (specifically, Co6E8L6) is localized on the inorganic core and is uncoupled to the aurophilic linker group. Meanwhile, the ligand-based HOMO of the cluster that would couple charge across the cluster junction are much like the HOMO of the ligands alone. We hypothesize that the weak coupling across the electrode-ligand-cluster core-ligand-

8 tip contributes to the Coulomb blockade-like effect in our single-cluster junction, and this feature will be the focus of Chapter 4.

1.5. Assembly of Molecular Clusters into Hierarchical Solids

Through judicious choice of nanoscale building blocks, we can create new materials with tailored chemical and physical properties.64 While nanocrystals have been employed as building blocks to construct superlattice structures,65, 66 they lack atomic definition and contain defect sites due to the nature of their synthesis that results in polydisperse structures and complex surfaces. On the other hand, molecular clusters are attractive nanoscale building blocks because they are atomically precise, monodisperse, and compositionally diverse. The use of molecular clusters in the bottom-up assembly of solids will enable the rational design of well-defined materials with interesting electronic and magnetic properties. Among the numerous strategies to obtain cluster- based materials, I present three relevant modes: (1) charge transfer, (2) molecular recognition, and

(3) metal coordination.

1.5.1. Charge Transfer

When electron-rich metal chalcogenide molecular clusters combine with complementary electron-accepting molecular clusters such as fullerenes67-69 or metal oxide clusters,70 there is an electron transfer between the electrically neutral clusters in solution to produce solid combinations.

This inter-cluster electrostatic interaction helps create the binary superlattice.71 Some of the superlattices are structurally analogous to the traditional ionic compounds such as CsCl,70 NaCl,67

67 and CdI2. Unlike with traditional atoms where modification of the individual atoms is not conceivable, the ligand shell of the molecular clusters can be easily tuned to induce subtle changes

9 in the packing of the binary superlattice.70 Since we can make the structural analogy to traditional binary atomic solids, we consider the cluster-assembled superlattices as “superatomic” solids67 in which the “superatoms” are defined as discrete molecular clusters with collective atom-like properties.72

1.5.2. Molecular Recognition

Another method to create organized hierarchical solids is to use building blocks with preorganized nanostructures that can direct the assembly of functional solids. This concept is derived from supramolecular principles, which focus on non-covalent interactions between molecules (e.g., hydrogen-bonding, van der Waals, weak electrostatic, and π-π).73 The organic phosphine ligands that we employ in the synthesis of molecular clusters are highly tunable, and we can affix specific molecular recognition moieties to guide the clusters into programmable structures. The forces holding the clusters together in the layered material discussed in Chapter 6 are weak van der Waals interactions and π-π interactions. Hydrogen-bonding74, 75 and electrostatic templating76 of complementary building blocks into ordered arrays have also been demonstrated with Chevrel phase molecular clusters.

1.5.3. Metal Coordination

As mentioned in Section 1.3.2, the ligand shell on the molecular cluster typically prevents electronic and magnetic communication between the neighboring clusters in the solid- and solution-phases. One of the ways to increase inter-cluster communication is to use a bifunctional ligand that can bridge clusters together through a metal cation to form an extended array. This strategy has been used extensively to create metal-organic frameworks.77 Cluster-containing metal

10 framework solids include expanded Prussian blue analogs. The cyanide ligands capping the octahedral metal sites in the cluster can coordinate to a divalent metal cation M’ through the M-

C≡N-M’-N≡C-M linkage to create extended porous solids.78, 79 Other cyanide-bridged organic- inorganic hybrid compounds include Kagome-type structures with fascinating magnetic

80 properties. More recently, it has been demonstrated that the ligands on the [Co6Se8] molecular clusters can be functionalized with carboxylate groups to react with Zn2+ salts to form zinc- carboxylate linkages and yield a 2D or 3D crystalline network.74

1.5.4. Increasing the Inter-Cluster Coupling

The cluster-based assemblies presented thus far do not involve direct covalent bonding between the neighboring cluster units. To increase inter-cluster coupling, we can append appropriate functional ligands on the molecular clusters to differentially and directionally connect the clusters.39, 81 Another appealing approach to improve inter-cluster coupling is to fuse the cores in a controllable fashion. There are several examples of fused dimers containing two stellated

82-85 octahedral [M6E8] cores linked together through two μ4-E bridging ligands. The dimerization occurs upon the loss of terminal ligands; however, the mechanism by which this process occurs remains elusive. The inter-cluster bonding mode is similar to those in the parent Chevrel phases.

By investigating the chemistry of dimerization further, we can potentially extend the core fusion strategy to create different types of polymerized clusters. The new cluster polymers would represent nanoscale fragments of the parent Chevrel phase framework.

1.6. Structural Characterization

In most cases, we can obtain single crystals of the cluster-assembled solids that are suitable for single crystal X-ray diffraction (SCXRD). When fine polycrystalline solids are obtained, we use synchrotron X-ray diffraction to characterize the superlattice structure.68 The phase purity of the solids is often determined by comparing the powder X-ray diffraction (PXRD) pattern of bulk solids with the simulated pattern from the crystal structure. When fullerenes are used as a complementary building block in the assembly, the degree of charge transfer can be easily monitored by Raman spectroscopy. We can also compare the changes in the bond lengths of M-L and M-M within the molecular cluster to deduce the degree of charge transfer.

1.7. Redox Properties

A key advantage of using molecular clusters in the assembly of hierarchical solids is the rich electrochemical properties the constituent molecular clusters impart on the resulting solids.

The majority of the Chevrel phase molecular clusters we utilize support reversible, multi- sequential electron transfer. In the assembled solids, the redox properties of the cluster components are retained,74, 86 which may prove invaluable for their applications in catalysis and batteries.

1.8. Collective Properties

Because of their atomic precision, the cluster-assembled solids exhibit collective behaviors, and this feature inspires the further development of bottom-up assembly hierarchical materials using molecular cluster building blocks. Some of the materials properties that can be tuned using different molecular clusters are: electronic transport, thermal transport, and magnetic ordering. For example, we perform electrical resistivity measurements on cluster-assembled materials to

12 determine their electronic properties. Depending on the constituent clusters, the activation energy of the materials can vary from 100 to 850 meV.67, 69 We also study the structure-function relationships in a series of fullerene-containing co-crystals of Co6E8L6 using thermal transport measurements.87 Lastly, we investigate long-range cooperative magnetic properties in cluster- assembled materials using superconducting quantum interference device (SQUID) and muon-spin relaxation. By varying the inter-cluster interactions, we can adjust the onset of the ferromagnetic transitions.68

1.9. Summary and Outlook

The molecular approach of using nanoscale building blocks to assemble solids presents a promising path to creating multifunctional materials with tunable properties. Owing to the expansive collection of molecular clusters, we can design and synthesize a wide variety of solid- state materials. The collective properties that arise in these materials are especially attractive for their potential applications. There are many remaining challenges to address so that we can apply the cluster-assembled solids to real-life applications. How can we predictably control the structures and properties of the assembled materials? How do the new properties compare to the conventional materials? What can we do to increase the stability of the materials? To confront these challenges, we can combine our knowledge and expertise of chemistry, physics, theory, and materials science.

With the numerous opportunities at hand, it seems like there is also plenty of room at the top.

1.10. References

1. Chevrel, R.; Sergent, M.; Prigent, J., Sur de Nouvelles Phases Sulfurées Ternaires du Molybdène. J. Solid State Chem. 1971, 3, 515-519.

2. Chevrel, R.; Hirrien, M.; Sergent, M., Superconducting Chevrel Phases: Prospects and Perspectives. Polyhedron 1986, 5, 87-94. 3. Peña, O., Chevrel Phases: Past, Present and Future. Physica C 2015, 514, 95-112. 4. Fischer, Ø ., Chevrel Phases: Superconducting and Normal State Properties. Appl. Phys. 1978, 16, 1-28. 5. Andersen, O. K.; Klose, W.; Nohl, H., Electronic Structure of Chevrel-Phase High-Critical- Field Superconductors. Phys. Rev. B 1978, 17, 1209-1237. 6. Aurbach, D.; Lu, Z.; Schechter, A.; Gofer, Y.; Gizbar, H.; Turgeman, R.; Cohen, Y.; Moshkovich, M.; Levi, E., Prototype Systems for Rechargeable Magnesium Batteries. Nature 2000, 407, 724-727.

7. Levi, E.; Gofer, Y.; Vestfreed, Y.; Lancry, E.; Aurbach, D., Cu2Mo6S8 Chevrel Phase, A Promising Cathode Material for New Rechargeable Mg Batteries: A Mechanically Induced Chemical Reaction. Chem. Mater. 2002, 14, 2767-2773.

8. Levi, E.; Gershinsky, G.; Aurbach, D.; Isnard, O.; Ceder, G., New Insight on the Unusually High Ionic Mobility in Chevrel Phases. Chem. Mater. 2009, 21, 1390-1399. 9. Fischer, C.; Alonso-Vante, N.; Fiechter, S.; Tributsch, H., Electrocatalytic Properties of Mixed Transition Metal Tellurides (Chevrel-Phases) for Oxygen Reduction. J. Appl. Electrochem. 1995, 25, 1004-1008. 10. McCarty, K. F.; Anderegg, J. W.; Schrader, G. L., Hydrodesulfurization Catalysis by Chevrel Phase Compounds. J. Catalysis 1985, 93, 375-387. 11. Kaiser, A. B., Electron-Phonon Interaction and Thermopower Nonlinearities in Chevrel- Phase Compounds. Phys. Rev. B 1987, 35, 4677-4681. 12. Caillat, T.; Fleurial, J. P., Thermoelectric Properties of the Semiconducting Chevrel Phase Mo2Re4Se8. J. Phys. Chem. Solids 1998, 59, 1139-1144. 13. Caillat, T.; Fleurial, J. P.; Snyder, G. J., Potential of Chevrel Phases for Thermoelectric Applications. Solid State Sci. 1999, 1, 535-544. 14. Rabiller, P.; Rabiller-Baudry, M.; Even-Boudjada, S.; Burel, L.; Chevrel, R.; Sergent, M.; Decroux, M.; Cors, J.; Maufras, J. L., Recent Progress in Chevrel Phase Syntheses: A New Low Temperature Synthesis of the Superconducting Lead Compound. MRS Bull. 1994, 29, 567-574.

15. Mironov Y. V.; Virovets A. V.; Naumov N. G.; Ikorskii V. N.; Fedorov V. E., Excision of the {Mo6Se8} Cluster Core from a Chevrel Phase: Synthesis and Properties of the First 7− Molybdenum Octahedral Cluster Selenocyanide Anions [Mo6Se8(CN)6] and 6− [Mo6Se8(CN)6] . Chem. Eur. J. 2000, 6, 1361-1365.

16. Magliocchi, C.; Xie, X.; Hughbanks, T., A Cyanide-Bridged Chain of Mo6Se8 Clusters: A Product of Cyanide-Melt Cluster Synthesis. Inorg. Chem. 2000, 39, 5000-5001. 17. Yaghi, O. M.; Scott, M. J.; Holm, R. H., Rhenium-Selenium-Chlorine Solid Phases: Cluster Excision and Core Substitution Reactions of Molecular Species. Inorg. Chem. 1992, 31, 4778-4784. 18. Long, J. R.; McCarty, L. S.; Holm, R. H., A Solid-State Route to Molecular Clusters: 2+ Access to the Solution Chemistry of [Re6Q8] (Q = S, Se) Core-Containing Clusters via Dimensional Reduction. J. Am. Chem. Soc. 1996, 118, 4603-4616.

19. Hilsenbeck, S. J.; Young, V. G.; McCarley, R. E., Synthesis, Structure, and Characterization of N-Ligated Mo6S8L6 Cluster Complexes. Molecular Precursors to Chevrel Phases. Inorg. Chem. 1994, 33, 1822-1832.

20. Zhang, X.; McCarley, R. E., High-Yield Synthesis of the W6S8 Cluster Unit as the Pyridine Complex W6S8(py)6 and Attempts To Prepare Tungsten Analogs of the Chevrel Phases. Inorg. Chem. 1995, 34, 2678-2683.

21. Saito, T.; Yamamoto, N.; Yamagata, T.; Imoto, H., Synthesis of [Mo6S8(PEt3)6] by Reductive Dimerization of a Trinuclear Molybdenum Chloro Sulfido Cluster Complex Coordinated with Triethylphosphine and Methanol: A Molecular Model for Superconducting Chevrel Phases. J. Am. Chem. Soc. 1988, 110, 1646-1647. 22. Snyder, B. S.; Holm, R. H., Hexakis(triethylphosphine)hexathioxohexairon+: Extension of + Stabilization of the Basket Core Topology to the [Fe6S6] Oxidation Level. Inorg. Chem. 1990, 29, 274-279. 23. Goddard, C. A.; Long, J. R.; Holm, R. H., Synthesis and Characterization of Four n+ Consecutive Members of the Five-Member [Fe6S8(PEt3)6] (n = 0−4) Cluster Electron Transfer Series. Inorg. Chem. 1996, 35, 4347-4354.

24. Coucouvanis, D., Use of Preassembled Iron/Sulfur and Iron/Molybdenum/Sulfur Clusters in the Stepwise Synthesis of Potential Analogs for the Fe/Mo/S Site in Nitrogenase. Acc. Chem. Res. 1991, 24, 1-8. 25. Al-Ahmad, S. A.; Salifoglou, A.; Kanatzidis, M. G.; Dunham, W. R.; Coucouvanis, D., Octanuclear Heterometallic Clusters with Rhombic Dodecahedral Cores. The Synthesis, n- Structural Characterization, and Properties of the [Fe6S6(p-RPhO)6[M(CO3]2] clusters (M = Mo, n = 3, R = Me, OMe, N(Me2); M = W, n = 3, R = Me; M = Mo, n = 4, R = Me, OMe, COMe). Precursors for Synthetic Analogs for the Fe/Mo/S Site in Nitrogenase. Inorg. Chem. 1990, 29, 927-938. 26. Tsuge, K.; Imoto, H.; Saito, T., Syntheses, Structures, and Molecular-Orbital Calculations of Chromium Chevrel-Type Cluster Complexes [Cr6E8(PR3)6] (E = S, PR3 = PEt3, PMe3; E = Se, PR3 = PEt3, PMe3, PMe2Ph). Bull. Chem. Soc. Jpn. 1996, 69, 627-636.

27. Fenske, D.; Ohmer, J.; Hachgenei, J., New Co and Ni Clusters with Se and PPh3 as Ligands: [Co4(μ3‐Se)4(PPh3)4], [Co6(μ3‐Se)8(PPh3)6], [Co9(μ4‐Se)3(μ3‐Se)8(PPh3)6], and [Ni34(μ5‐ Se)2(μ4‐Se)20(PPh3)10]. Angew. Chem. Int. Edit. 2003, 24, 993-995. 28. Fenske, D.; Grissinger, M. A.; Loos, M.; Magull, J., Neue Oktaedrische Metallcluster des Zr und V Die Kristallstrukturen von [Cp6Zr6S9] und [V6Se8O(PMe3)6]. Z. Anorg. Allg. Chem. 2004, 598, 121-128. 29. Agresti, A.; Bacci, M.; Cecconi, F.; Ghilardi, C. A.; Midollini, S., Transition-Metal Complexes with Sulfur Atoms as Ligands. 7. Synthesis, Properties, Structure, and Molecular Orbital Calculations of the Paramagnetic Cluster [Fe6(μ3-S)8(PEt3)6](BPh4)2. Inorg. Chem. 1985, 24, 689-695. 30. Cecconi, F.; Ghilardi, C. A.; Midollini, S., A Novel Paramagnetic Octahedral Iron Cluster: Synthesis and X-Ray Structural Characterization of [Fe6(μ3-S)8(PEt3)6](BPh4)2. J. Chem. Soc. Chem. Comm. 1981, 640-641. 31. Hessen, B.; Siegrist, T.; Palstra, T.; Tanzler, S. M.; Steigerwald, M. L., Hexakis(triethylphosphine)octatelluridohexachromium and a Molecule-Based Synthesis of Chromium Telluride, Cr3Te4. Inorg. Chem. 1993, 32, 5165-5169. 32. Steigerwald, M. L.; Siegrist, T.; Stuczynski, S. M., Initial Stages in the Molecule-Based Growth of the Solid-State Compound Cobalt Telluride (CoTe). Inorg. Chem. 1991, 30, 4940-4945. 33. Brennan, J. G.; Siegrist, T.; Stuczynski, S. M.; Steigerwald, M. L., The Transition from Molecules to Solids: Molecular Syntheses of Ni9Te6(PEt3)8, Ni20Te18(PEt3)12 and NiTe. J. Am. Chem. Soc. 1989, 111, 9240-9241. 34. Arratia-Pérez, R.; Hernández-Acevedo, L. A., The Hexanuclear Rhenium Cluster Ions 4− Re6S8X6 (X = Cl, Br, I): Are These Clusters Luminescent? J. Chem. Phys. 1999, 110, 2529-2532. 35. Gray, T. G.; Rudzinski, C. M.; Nocera, D. G.; Holm, R. H., Highly Emissive Hexanuclear 2+ 2+ Rhenium(III) Clusters Containing the Cubic Cores [Re6S8] and [Re6Se8] . Inorg. Chem. 1999, 38, 5932-5933.

36. Lee, S. C.; Holm, R. H., Nonmolecular Metal Chalcogenide/Halide Solids and Their Molecular Cluster Analogues. Angew. Chem. Int. Edit. 2003, 29, 840-856. 37. Stuczynski, S. M.; Kwon, Y. U.; Steigerwald, M. L., The Use of Phosphine Chalcogenides in the Preparation of Cobalt Chalcogenides. J. Organomet. Chem. 1993, 449, 167-172. 38. Steigerwald, M. L., Clusters as Small Solids. Polyhedron 1994, 13, 1245-1252. 39. Champsaur, A. M.; Velian, A.; Paley, D. W.; Choi, B.; Roy, X.; Steigerwald, M. L.; Nuckolls, C., Building Diatomic and Triatomic Superatom Molecules. Nano Lett. 2016, 16, 5273-5277.

40. Dance, I.; Fisher, K., Metal Chalcogenide Cluster Chemistry. Prog. Inorg. Chem. 2007. 41. Degroot, M. W.; Corrigan, J. F., Comprehensive Coordination Chemistry II: Oxford, 2003.

42. Diana, E.; Gervasio, G.; Rossetti, R.; Valdemarin, F.; Bor, G.; Stanghellini, P. L., Co6(μ3- S)8(CO)6·3S8 - Structure, Bonding, and Vibrational Analysis of an Exceptionally Electron- Rich Carbonyl Cluster. Inorg. Chem. 1991, 30, 294-299. 43. Steigerwald, M. L.; Siegrist, T.; Gyorgy, E. M.; Hessen, B.; Kwon, Y. U.; Tanzler, S. M., Effect of Diverse Ligands on the Course of a Molecules-to-Solids Process and Properties of its Intermediates. Inorg. Chem. 1994, 33, 3389-3395. 44. Fenske, D.; Krautscheid, H.; Muller, M., New Intermediate Steps in the Synthesis of Larger Nickel Clusters. Angew. Chem. Int. Edit. 1992, 31, 321-323. 45. Choi, B.; Paley, D. W.; Siegrist, T.; Steigerwald, M. L.; Roy, X., Ligand Control of Manganese Telluride Molecular Cluster Core Nuclearity. Inorg. Chem. 2015, 54, 8348- 8355. 46. Xu, B.; Tao, N. J., Measurement of Single-Molecule Resistance by Repeated Formation of Molecular Junctions. Science 2003, 301, 1221-1223. 47. Venkataraman, L.; Klare, J. E.; Nuckolls, C.; Hybertsen, M. S.; Steigerwald, M. L., Dependence of Single-Molecule Junction Conductance on Molecular Conformation. Nature 2006, 442, 904-907. 48. Fatemi, V.; Kamenetska, M.; Neaton, J. B.; Venkataraman, L., Environmental Control of Single-Molecule Junction Transport. Nano Lett. 2011, 11, 1988-1992. 49. Luka-Guth, K.; Hambsch, S.; Bloch, A.; Ehrenreich, P.; Briechle, B. M.; Kilibarda, F.; Sendler, T.; Sysoiev, D.; Huhn, T.; Erbe, A.; Scheer, E., Role of Solvents in the Electronic Transport Properties of Single-Molecule Junctions. Beilstein J. Nanotechnol. 2016, 7, 1055-1067. 50. Capozzi, B.; Xia, J.; Adak, O.; Dell, E. J.; Liu, Z.-F.; Taylor, J. C.; Neaton, J. B.; Campos, L. M.; Venkataraman, L., Single-Molecule Diodes with High Rectification Ratios through Environmental Control. Nat. Nanotechnol. 2015, 10, 522-527. 51. Su, T. A.; Neupane, M.; Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C., Chemical Principles of Single-Molecule Electronics. Nat. Rev. Mater. 2016, 1, 16002. 52. Roy, X.; Schenck C. L.; Ahn, S.; Lalancette R. A.; Venkataraman, L.; Nuckolls, C.; Steigerwald M. L., Quantum Soldering of Individual Quantum Dots. Angew. Chem. Int. Edit. 2012, 51, 12473-12476. 53. Su, T. A.; Li, H.; Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C., Stereoelectronic Switching in Single-Molecule Junctions. Nat. Chem. 2015, 7, 215-220.

54. Su, T. A.; Li, H.; Zhang, V.; Neupane, M.; Batra, A.; Klausen, R. S.; Kumar, B.; Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C., Single-Molecule Conductance in Atomically Precise Germanium Wires. J. Am. Chem. Soc. 2015, 137, 12400-12405. 55. Batra, A.; Kladnik, G.; Gorjizadeh, N.; Meisner, J.; Steigerwald, M.; Nuckolls, C.; Quek, S. Y.; Cvetko, D.; Morgante, A.; Venkataraman, L., Trimethyltin-Mediated Covalent Gold- Carbon Bond Formation. J. Am. Chem. Soc. 2014, 136, 12556-12559. 56. Cheng, Z. L.; Skouta, R.; Vazquez, H.; Widawsky, J. R.; Schneebeli, S.; Chen, W.; Hybertsen, M. S.; Breslow, R.; Venkataraman, L., In Situ Formation of Highly Conducting Covalent Au-C Contacts for Single-Molecule Junctions. Nat. Nanotechnol. 2011, 6, 353- 357. 57. Chen, W.; Widawsky, J. R.; Vázquez, H.; Schneebeli, S. T.; Hybertsen, M. S.; Breslow, R.; Venkataraman, L., Highly Conducting π-Conjugated Molecular Junctions Covalently Bonded to Gold Electrodes. J. Am. Chem. Soc. 2011, 133, 17160-17163. 58. Ploog K. H., Single Charge Tunneling, Coulomb Blockade Phenomena in Nanostructures. Adv. Mater. 2004, 5, 227-227. 59. Danilov, A.; Kubatkin, S.; Kafanov, S.; Hedegård, P.; Stuhr-Hansen, N.; Moth-Poulsen, K.; Bjørnholm, T., Electronic Transport in Single Molecule Junctions: Control of the Molecule-Electrode Coupling through Intramolecular Tunneling Barriers. Nano Lett. 2008, 8, 1-5. 60. Park, J.; Pasupathy, A. N.; Goldsmith, J. I.; Chang, C.; Yaish, Y.; Petta, J. R.; Rinkoski, M.; Sethna, J. P.; Abruña, H. D.; McEuen, P. L.; Ralph, D. C., Coulomb Blockade and the Kondo Effect in Single-Atom Transistors. Nature 2002, 417, 722-725. 61. Liang, W.; Shores, M. P.; Bockrath, M.; Long, J. R.; Park, H., Kondo Resonance in a Single- Molecule Transistor. Nature 2002, 417, 725-729. 62. Andres, R. P.; Bein, T.; Dorogi, M.; Feng, S.; Henderson, J. I.; Kubiak, C. P.; Mahoney, W.; Osifchin, R. G.; Reifenberger, R., Coulomb Staircase at Room Temperature in a Self- Assembled Molecular Nanostructure. Science 1996, 272, 1323-1325. 63. Klein, D. L.; Roth, R.; Lim, A. K. L.; Alivisatos, A. P.; McEuen, P. L., A Single-Electron Transistor Made from a Cadmium Selenide Nanocrystal. Nature 1997, 389, 699-701. 64. Claridge, S. A.; Castleman, A. W.; Khanna, S. N.; Murray, C. B.; Sen, A.; Weiss, P. S., Cluster-Assembled Materials. ACS Nano 2009, 3, 244-255. 65. Shevchenko, E. V.; Talapin, D. V.; Kotov, N. A.; O'Brien, S.; Murray, C. B., Structural Diversity in Binary Nanoparticle Superlattices. Nature 2006, 439, 55-59. 66. Collier, C. P.; Vossmeyer, T.; Heath, J. R., Nanocrystal Superlattices. Annu. Rev. Phys. Chem. 1998, 49, 371-404.

67. Roy, X.; Lee, C. H.; Crowther, A. C.; Schenck, C. L.; Besara, T.; Lalancette, R. A.; Siegrist, T.; Stephens, P. W.; Brus, L. E.; Kim, P.; Steigerwald, M. L.; Nuckolls, C., Nanoscale Atoms in Solid-State Chemistry. Science 2013, 341, 157-160. 68. Lee, C. H.; Liu, L.; Bejger, C.; Turkiewicz, A.; Goko, T.; Arguello, C. J.; Frandsen, B. A.; Cheung, S. C.; Medina, T.; Munsie, T. J. S.; D'Ortenzio, R.; Luke, G. M.; Besara, T.; Lalancette, R. A.; Siegrist, T.; Stephens, P. W.; Crowther, A. C.; Brus, L. E.; Matsuo, Y.; Nakamura, E.; Uemura, Y. J.; Kim, P.; Nuckolls, C.; Steigerwald, M. L.; Roy, X., Ferromagnetic Ordering in Superatomic Solids. J. Am. Chem. Soc. 2014, 136, 16926-16931. 69. O'Brien, E. S.; Trinh, M. T.; Kann, R. L.; Chen, J.; Elbaz, G. A.; Masurkar, A.; Atallah, T. L.; Paley, M. V.; Patel, N.; Paley, D. W.; Kymissis, I.; Crowther, A. C.; Millis, A. J.; Reichman, D. R.; Zhu, X. Y.; Roy, X., Single-Crystal-to-Single-Crystal Intercalation of a Low-Bandgap Superatomic Crystal. Nat. Chem. 2017, 9, 1170-1174. 70. Turkiewicz, A.; Paley, D. W.; Besara, T.; Elbaz, G.; Pinkard, A.; Siegrist, T.; Roy, X., Assembling Hierarchical Cluster Solids with Atomic Precision. J. Am. Chem. Soc. 2014, 136, 15873-15876. 71. Pinkard, A.; Champsaur, A. M.; Roy, X., Molecular Clusters: Nanoscale Building Blocks for Solid-State Materials. Acc. Chem. Res. 2018, 51, 919-929. 72. Castleman, A. W.; Khanna, S. N., Clusters, Superatoms, and Building Blocks of New Materials. J. Phys. Chem. C 2009, 113, 2664-2675. 73. Descalzo A. B.; Martínez‐Máñez, R.; Sancenón, F.; Hoffmann, K.; Rurack, K., The Supramolecular Chemistry of Organic-Inorganic Hybrid Materials. Angew. Chem. Int. Edit. 2006, 45, 5924-5948. 74. Champsaur, A. M.; Yu, J.; Roy, X.; Paley, D. W.; Steigerwald, M. L.; Nuckolls, C.; Bejger, C. M., Two-Dimensional Nanosheets from Redox-Active Superatoms. ACS Cent. Sci. 2017, 3, 1050-1055. 75. Selby, H. D.; Roland, B. K.; Carducci, M. D.; Zheng, Z., Hydrogen-Bonded Extended 2+ Arrays of the [Re6(μ3-Se)8] Core-Containing Clusters. Inorg. Chem. 2003, 42, 1656-1662. 76. Champsaur, A. M.; Mézière, C.; Allain, M.; Paley, D. W.; Steigerwald, M. L.; Nuckolls, C.; Batail, P., Weaving Nanoscale Cloth through Electrostatic Templating. J. Am. Chem. Soc. 2017, 139, 11718-11721. 77. James, S. L., Metal-Organic Frameworks. Chem. Soc. Rev. 2003, 32, 276-288. 78. Shores, M. P.; Beauvais, L. G.; Long, J. R., Cluster-Expanded Prussian Blue Analogues. J. Am. Chem. Soc. 1999, 121, 775-779. 79. Bennett, M. V.; Beauvais, L. G.; Shores, M. P.; Long, J. R., Expanded Prussian Blue 3-/4- Analogues Incorporating [Re6Se8(CN)6] Clusters: Adjusting Porosity via Charge Balance. J. Am. Chem. Soc. 2001, 123, 8022-8032.

80. Baudron, S. A.; Batail, P.; Coulon, C.; Clérac, R.; Canadell, E.; Laukhin, V.; Melzi, R.; Wzietek, P.; Jérome, D.; Auban-Senzier, P.; Ravy, S., (EDT-TTF-CONH2)6[Re6Se8(CN)6], a Metallic Kagome-Type Organic-Inorganic Hybrid Compound: Electronic Instability, Molecular Motion, and Charge Localization. J. Am. Chem. Soc. 2005, 127, 11785-11797. 81. Selby, H. D.; Roland, B. K.; Zheng, Z., Ligand-Bridged Oligomeric and Supramolecular Arrays of the Hexanuclear Rhenium Selenide Clusters-Exploratory Synthesis, Structural Characterization, and Property Investigation. Acc. Chem. Res. 2003, 36, 933-944. 82. Cecconi, F.; Ghilardi, C. A.; Midollini, S.; Orlandini, A., Dimerization of the Stellated Octahedral Unit Co6S8P6: Synthesis and X-Ray Crystal Structure of [Co12(μ3-S)14(μ4- S)2(PEt3)10][TCNQ]2, where TCNQ = Tetracyanoquinodimethane. Inorg. Chim. Acta 1993, 214, 13-15. 83. Kamiguchi, S.; Imoto, H.; Saito, T.; Chihara, T., Synthesis, Structure, FAB Mass Spectrum, and Magnetic Property of a Dodecanuclear Cluster Complex of Chromium with Hydrido Ligands [Cr12S16(H)2(PEt3)10]. Solid State Sci. 1999, 1, 497-508.

84. Zheng, Z.; Long, J. R.; Holm, R. H., A Basis Set of Re6Se8 Cluster Building Blocks and Demonstration of Their Linking Capability: Directed Synthesis of an Re12Se16 Dicluster. J. Am. Chem. Soc. 1997, 119, 2163-2171.

85. Mizutani, J.; Amari, S.; Imoto, H.; Saito, T., Reaction of [Mo6S8(PEt3)6] and NOBF4: Structure and Molecular Orbital Calculation of an Octahedral Cluster Complex [Mo6S8(NO)(PEt3)5]. J. Chem. Soc. Dalton Trans. 1998, 819-824. 86. Voevodin, A.; Campos, L. M.; Roy, X., Multifunctional Vesicles from a Self-assembled Cluster-Containing Diblock Copolymer. J. Am. Chem. Soc. 2018, ASAP. 87. Ong, W.-L.; O’Brien, E. S.; Dougherty, P. S. M.; Paley, D. W.; Fred Higgs Iii, C.; McGaughey, A. J. H.; Malen, J. A.; Roy, X., Orientational Order Controls Crystalline and Amorphous Thermal Transport in Superatomic Crystals. Nat. Mater. 2016, 16, 83-88.

Chapter 2. Structural, Dimensional, and Interfacial Diversity in Manganese Telluride Molecular Clusters

2.1. Preface

This chapter is based on a manuscript entitled “Ligand Control of Manganese Telluride

Molecular Cluster Core Nuclearity” by Bonnie Choi, Daniel W. Paley, Theo Siegrist, Michael L.

Steigerwald, and Xavier Roy, published in Inorganic Chemistry.1 It is reprinted with permission from the American Chemical Society. I synthesized all compounds. I characterized the molecular clusters using SCXRD with essential input from Dr. Daniel W. Paley and Dr. Theo Siegrist.

2.2. Introduction

Our interest in synthesizing and studying manganese-based molecular clusters is motivated by the reports of remarkable magnetic2-4 and catalytic5, 6 properties in manganese oxide clusters, and by the promising uses of MnTe in the preparation of dilute magnetic semiconducting alloys,7,

8 in the fabrication of X-ray and γ-ray detectors9, 10 and as magnetic impurities in nanocrystals.11,

12 While the chemistry of manganese oxide molecular clusters is versatile and rich,13-15 there is only a small handful of known manganese clusters containing heavier chalcogens (i.e., S, Se, and

Te), and none that are passivated with L-type two-electron donor ligands. Noteworthy examples

2- 16 4- 17 10- 18 from literature include [Mn4(SPh)10] , [Mn4Se4(SeSiMe3)4] , [Mn4Sn4Se17] , and [Mn4-

i 4- 19, 20 Te4(E Pr)4] (E = S, Se, Te).

In this chapter, we report the synthesis, structural diversity, and chemical behaviors of a family of manganese telluride molecular clusters whose charge-neutral cores are passivated by two-electron donor ligands. We describe three different core structures: a cubane-type Mn4Te4, a prismane Mn6Te6, and a dicubane Mn8Te8. Various trialkylphosphines and N-heterocyclic 21 carbenes (NHCs) are employed as surface ligands and it is shown that the formation of the different

i cluster core structures is controlled by the choice of ligand. Bulky ligands such as L = P Pr3 or

PCy3 generate a Mn4Te4L4 cubane cluster, while the smaller L = PMe3 produces the prismane

Mn6Te6L6. The PEt3 ligand is intermediate in size, and its use yields both cubane and prismane clusters.

These manganese telluride molecular clusters are labile, and the capping phosphines can be replaced by stronger ligands while the internal core structure of the cluster remains intact. The interplay of structural diversity, and ligand versatility and lability makes these clusters potentially useful building blocks for the assembly of larger aggregates and extended structures. In this respect, we demonstrate the simplest prototype of these solid-forming reactions: the direct

i i coupling of two Mn4Te4( Pr2NHC)4 units to form the dicubane Mn8Te8( Pr2NHC)6. Lastly, we discuss the core structure of these molecular clusters as recognizable building units for the zinc blende and the hypothetical wurtzite lattices of solid-state compound MnTe.

2.3. Synthetic Design

The hard, high-spin Mn(II) ion interacts very weakly with soft organophosphine ligands21 and as a result, ionic MnX2 (X = Cl, Br or I) precursors proved to be a poor choice for the synthesis of phosphine-capped manganese telluride clusters. For example, the reaction of MnCl2, bis(tert-

i butyldimethylsilyl)telluride and a large excess of P Pr3 resulted in the formation of an insoluble, amorphous solid that, when pyrolyzed, yields bulk α-MnTe. In light of these results, we turned to an alternative chemical approach in which we react a low-valent manganese organometallic complex with a source of functionally “low-valent” tellurium to give the targeted molecular

22, 23 clusters. Trialkylphosphine tellurides (R3PTe) have been used as reactive sources of “low-

22 valent” tellurium to form metal-tellurium bonds with several low-valent transition metal

24 25 26 complexes, including Co2(CO)8, Cr(DMPD)2, and Fe(COT)2 (DMPD = 2,4- dimethylpentadienyl, COT = cyclooctatetraene). By controlling the reaction conditions, stable polynuclear metal chalcogenide clusters have been intercepted in the molecules-to-solids process.

i Figure 2.1. Molecular structure of Mn4Te4(P Pr3)4 showing thermal ellipsoids at 50% probability. Manganese, magenta; tellurium, teal; phosphorus, orange; carbon, black. Hydrogen atoms and a toluene of solvation are removed to clarify the view.

4 27 i We found that the 17-electron complex Mn(η -butadiene)2PMe3 reacts with Pr3PTe in

i refluxing tetrahydrofuran (THF) in the presence of a large excess of P Pr3 (10 equivalents) to form a brown solution from which dark orange crystals grew at -35 °C over 24 h. SCXRD reveals that this compound is a polynuclear manganese telluride complex with a cubane-type core,

i Mn4Te4(P Pr3)4 (Figure 2.1). The core is a tetrahedron of Mn atoms with each face capped with a

Te atom forming a larger tetrahedron; four phosphine ligands passivate this inorganic core,

i 28 forming again a tetrahedron. As in the closely related cluster Fe4Te4(P Pr3)4, the cubane core of

i Mn4Te4(P Pr3)4 is severely distorted; the Mn2Te2 rhombs are non-planar with mean Te-Mn-Mn-Te torsion angle of 156.6°. Other important mean bond distances and angles are listed in Table 2.1.

4 Because the preparation and isolation of Mn(η -butadiene)2PMe3 can be both time- consuming and challenging, we sought a more convenient precursor for the preparation of this molecular cluster. Analogous to the previously reported use of the complex (Et3P)2Cr(allyl)2 as a

25 source of reactive, low-valent chromium, we treated a suspension of anhydrous MnCl2 in THF

i with one equivalent of P Pr3 and two equivalents of allylmagnesium chloride at -78 °C. After ~2 h, we obtained a dark brown mixture, which we believe contains a manganese diallyl compound

i or a reactive equivalent thereof, to which we added one equivalent of Pr3PTe. The reaction was warmed to room temperature and was stirred for ~5 h. Then, the mixture was concentrated to

i dryness and recrystallized from toluene to afford dark orange crystals of Mn4Te4(P Pr3)4 identical

4 to the compound prepared from Mn(η -butadiene)2PMe3. In contrast to our original approach

(Method A), this one-pot synthesis (Method B) does not require a large excess of phosphine and employs commercially available precursors (Figure 2.2).

Figure 2.2. Schematic of the synthesis and reactivity of manganese telluride molecular clusters.

We prepared cubane-type clusters, Mn4Te4L4, capped with various trialkylphosphines (L

i = PEt3, P Pr3, PCy3) using both approaches (Figure 2.3). The cores of Mn4Te4(PCy3)4 and

i Mn4Te4(PEt3)4 are similar to that of Mn4Te4(P Pr3)4, but the Mn-Mn, Mn-Te, and Mn-P distances

i in Mn4Te4(PEt3)4 are shorter than those in Mn4Te4(P Pr3)4 and Mn4Te4(PCy3)4, reflecting the

29, 30 smaller cone angle of PEt3 (Table 2.1). This cubane geometry is pervasive throughout molecular cluster chemistry, and we compare Mn4Te4L4 to several related compounds. The only

4- reported manganese telluride cubane cluster to date is the ionic cluster [Mn4Te4(EC3H7)4] (E =

19, 20 S, Se, Te), which is capped with chalcogenolate ligands. As in our neutral Mn4Te4L4 clusters,

4- the core of [Mn4Te4(EC3H7)4] is significantly distorted from a perfect cube, and the Mn-Te distances are within the typical range for such bonds. The mean Mn-Mn separation in all Mn4Te4L4

(~3.1 Å ) is significantly longer than the nearest-neighbor distance of 2.73 Å in elemental Mn and

31, 32 lies between that of traditional singly-bonded manganese dimers (e.g., ~2.9 Å in Mn2(CO)10)

4- 20 and that of [Mn4Te4(EC3H7)4] (~3.3 Å ), which exhibits no Mn-Mn interactions. We can also view the Mn4Te4L4 cluster as an illustration of the topological self-duality of the tetrahedron: the cubane core is the intersection of two concentric tetrahedra, Mn4 and Te4, that have different edge lengths (~3.1 Å and ~4.4 Å respectively).

Figure 2.3. Molecular structures of (a) Mn4Te4(PEt3)4 and (b) Mn4Te4(PEt3)4 showing thermal ellipsoids at 50% probability. Color scheme is the same as in Figure 2.1. Hydrogen atoms and molecules of solvation are removed to clarify the view.

4 We found that the reaction of Mn(η -butadiene)2PMe3 with Et3PTe in the presence of a large excess of PEt3 produces a mixture of two clusters of different nuclearity: Mn4Te4(PEt3)4 and

Mn6Te6(PEt3)6. These compounds were separated based on their difference in solubility; during the course of the reaction, Mn6Te6(PEt3)6 precipitates from refluxing THF as a brown solid. We filter this solid and recrystallize it from dichloromethane at -35 °C. Figure 2.4 shows the molecular structure of Mn6Te6(PEt3)6 as determined by SCXRD. The cluster has a cylindrical prismane core built from two face-sharing hexagonal Mn3Te3 puckered rings with alternating Mn and Te atoms at the vertices; each Mn atom is further coordinated by one PEt3. The Mn2Te2 rhombs that form the lateral surfaces of the cylinder are non-planar but to a lesser degree than in the cubane 26 structures; the mean Te-Mn-Mn-Te torsion angle is 164.3° in Mn6Te6(PEt3)6 while it is 155.2° in

Mn4Te4(PEt3)4. The Mn-Mn, Mn-Te and Mn-P distances in Mn6Te6(PEt3)6 are similar to those of

Mn4Te4(PEt3)4 (Table 2.1).

a Table 2.1. Mean bond distances (Å ) and angles (°) for [MnTe]n (n = 4, 6, 8) clusters. Distances (Å ) Angles (°) Compound Mn-Mn Mn-Te Mn-L (L = P, C) Mn-Te-Mn Te-Mn-Mn-Teb c Mn4Te4(PEt3)4 3.029 2.712 2.479 67.9 155.2 i Mn4Te4(P Pr3)4 3.118(18) 2.738(9) 2.533(5) 69.4(5) 156.7(13)

Mn4Te4(PCy3)4 3.10(6) 2.732(17) 2.527(17) 69.1(14) 156(3)

Mn4Te4(Me2NHC)4 3.22(5) 2.741(18) 2.185(14) 72.1(13) 159(3)

Mn6Te6(PMe3)6 3.07(3) 2.714(16) 2.512(17) 68.9(6) (rhomb) 162.6(4) 99(3) (hexagon)

Mn6Te6(PEt3)6 3.010(17) 2.720(7) 2.540(12) 69.5(3) (rhomb) 164.3(6) 101(2) (hexagon) i d d Mn8Te8( Pr2NHC)6 3.27(12) 2.77(4) 2.175(6) 72(3) 159(6) 69.3c (bridge) 180.0c (bridge) (a) We list average geometric parameters and their standard deviations. Crystallographic esd’s of individual measurements are typically much smaller; they are 0.01-0.02 Å for Mn-C distances, 0.001-0.005 Å for other distances, and 0.01-0.1 ° for angles. (b) The Te-Mn-Mn-Te torsion angle is the dihedral between two fused Mn2Te rings. The angle would be 180º for a perfectly planar rhomb and ~109.5º for Te-Te >> Mn-Mn. (c) No standard deviation is listed when a measurement has only one independent value. (d) The bridging Mn-Mn and Mn-Te distances, respectively 3.1715(18) and 2.7513(10) Å , are omitted from these averages.

Figure 2.4. Molecular structure of Mn6Te6(PEt3)6 (a, side-view; b, top-view) showing thermal ellipsoids at 50% probability. Color scheme is the same as in Figure 2.1. Hydrogen atoms and a dichloromethane of solvation were removed to clarify the view.

Our findings thus far suggest that the core structure of these manganese telluride clusters

i is controlled by the size of the capping ligand. Phosphines with a large cone angle (P Pr3 or PCy3) form clusters with a small radius of curvature, Mn4Te4L4. The intermediate-sized phosphine PEt3 produces a mixture of Mn4Te4(PEt3)4 and Mn6Te6(PEt3)6. In addition, we have found that the prismane Mn6Te6(PEt3)6 is metastable, reorganizing into the cubane-type Mn4Te4(PEt3)4 when left in solution for ~1 week at -35 °C. Our hypothesis would predict that smaller capping ligands favor the formation of clusters with larger radii of curvature, and we find this to be the case with the smallest trialkylphosphine ligand, PMe3. This ligand gives the prismane Mn6Te6(PMe3)6. We used

PXRD to confirm, more importantly, that using PMe3 forms Mn6Te6(PMe3)6 exclusively, and we do not see any evidence of the cubane cluster or of the conversion of the prismane to the related, and as yet hypothetical, Mn4Te4(PMe3)4 cubane cluster. The prismane core is assembled from three non-planar Mn2Te2 rhombs that are analogous to the ones in Mn4Te4L4. This Mn6Te6 structural

2+/+ 33-36 unit is reminiscent of the [Fe6S6] cluster but to the best of our knowledge, Mn6Te6L6 is the first example of a prismane chalcogenide cluster with a fully reduced core (i.e., an inorganic core which has no formal charge). It is also the only prismane cluster containing either Mn or Te.

2.4. Dissociation of Phosphine Ligands

4- In contrast to [Mn4Te4(EC3H7)4] , which does not participate in ligand-substitution reactions, the Mn4Te4L4 clusters are labile, and in solution, ligand dissociation leads to the precipitation of an insoluble orange solid. The stability of the cubane compounds Mn4Te4(PR3)4 is strongly influenced by the choice of capping ligand, solvent, and temperature. While

Mn4Te4(PEt3)4 decomposes within minutes when dissolved in dichloromethane at room

i temperature, Mn4Te4(P Pr3)4 and Mn4Te4(PCy3)4 remain unchanged in solution for ~16 h in non-

28 coordinating solvents such as dichloromethane or toluene. The same cluster solutions are stable indefinitely when kept at -35 °C, and the addition of free phosphine increases their lifetime at room temperature. All Mn4Te4(PR3)4 compounds react immediately in coordinating solvents such as

THF to form larger insoluble cluster aggregates.

Lability and ligand substitution are essential features of cluster-based catalytic systems37 but they often make solution-based studies and applications challenging. The coordinating characteristics of N-heterocyclic carbenes (NHCs)38 are similar to those of phosphines.39-42

Importantly, they are generally stronger -donor ligands than phosphines and form more stable bonds to metals.43 Despite the ready availability of NHCs, very few examples of NHC complexes of Mn(II) are known to date.44-46 We explored whether a “carbene telluride” (NHCTe), analogous to a trialkylphosphine telluride, could transfer a tellurium atom to a low-valent manganese to form a manganese-tellurium bond while liberating an NHC that could bind to the metal to form an NHC- capped manganese telluride molecular cluster. To examine this reaction, we prepared the precursor i 47 i Pr2NHCTe from the corresponding carbene Pr2NHC (1,3-diisopropyl-4,5-dimethylimidazol-2- ylidene)48 and elemental Te, and used this carbene telluride in our cluster synthesis. We employed i Pr2NHCTe in both synthetic approaches (Methods A and B) and obtained the cubane-type cluster

i Mn4Te4( Pr2NHC)4 (Figure 2.5). In contrast to phosphine-capped clusters whose syntheses require

i an excess of free phosphine, only a stoichiometric amount of Pr2NHCTe is required to form

i Mn4Te4( Pr2NHC)4. Using the same synthetic approach we prepared Mn4Te4(Me2NHC)4, where

48 Me2NHC = 1,3,4,5-tetramethylimidazol-2-ylidene.

i i Figure 2.5. Molecular structures of (a) Mn4Te4( Pr2NHC)4 and (b) Mn8Te8( Pr2NHC)6 showing thermal ellipsoids at 50% probability. Color scheme is the same as in Figure 2.1. Nitrogen, blue. Hydrogen atoms and toluene molecules of solvation are removed to clarify the view.

i SCXRD shows that the cores of Mn4Te4(Me2NHC)4 and Mn4Te4( Pr2NHC)4 are similar to

i that of Mn4Te4(PR3)4. Orientational disorder within the single crystals of Mn4Te4( Pr2NHC)4 precluded detailed metrical analysis of this complex, but Mn4Te4(Me2NHC)4 gave a higher quality single crystal structure, and we list the mean bond distances and angles of this cluster in Table 2.1.

i Due to the instability of the Me2NHC in solution, we use Pr2NHC in subsequent substitution reactions (see below). The Mn-Te distances in the Me2NHC-capped cluster are similar to those in the phosphine-capped clusters, although the Mn-Mn distances are slightly longer (Table 2.1). The average Mn-C(NHC) bond length is 2.185 Å , suggesting a Mn-C(NHC) single bond with

44 negligible MnC(NHC) back-bonding contributions. In this respect, the behavior of Mn4Te4L4

i t 49, 50 is similar to that of the cubane Fe4S4L4 (L = P Pr3, P Bu3, PCy3). We demonstrate that

i i i Mn4Te4(P Pr3)4 reacts with Pr2NHC to afford Mn4Te4( Pr2NHC)4. Similarly, NHC ligands have been used on Co4S4 and Fe4S4 clusters to access oxidation states that are unstable with phosphine ligands.51, 52

1 i i i Figure 2.6. H NMR spectra of (a) Mn4Te4(P Pr3)4, (b) product of the reaction of Pr2NHC with Mn4Te4(P Pr3)4, and i 31 (c) Mn4Te4( Pr2NHC)4 as directly synthesized, in C7D8 at 240 K. P NMR spectra are shown in the inset of (a) and (b). In (c), minor solvent impurities are visible (◊, THF; ♦, hexanes). The peaks are color-coded according to the i i i i reaction scheme above (orange, Mn4Te4(P Pr3)4; red, Pr2NHC; blue, Mn4Te4( Pr2NHC)4; green, P Pr3). We consistently i i observe free P Pr3 peaks in Mn4Te4(P Pr3)4.

The stronger binding of the NHC to the tetrahedral Mn(II) manifests itself in the increased

i solution-phase stability of Mn4Te4( Pr2NHC)4 vis-à-vis that of Mn4Te4(PR3)4. We further demonstrate the affinity of the cubane core for NHC in a ligand-substitution reaction. When we

i i treat a d8-toluene solution of Mn4Te4(P Pr3)4 with 4.5 equivalents of Pr2NHC at room temperature

1 i the solution turns from red to orange, and H NMR spectroscopy verifies that Pr2NHC replaces

i 1 i the erstwhile cluster-bound P Pr3 (Figure 2.6). The H NMR spectrum of Mn4Te4(P Pr3)4 (Figure

2.6a) exhibits two broad resonances centered at 3.5 and 6.8 ppm (corresponding to methyl and 31

i methine groups of P Pr3 bound to Mn, respectively). We always observe a small amount of free

i 1 31 P Pr3 at 1.0 and 1.6 ppm in the H NMR and at -17.5 ppm in the P NMR. Upon addition of a

i i slight excess of Pr2NHC, the peaks due to free P Pr3 increase in intensity, and the peaks at 3.5 and

6.8 ppm completely disappear (Figure 2.6b). Two new broad peaks emerge at 5.4 and 7.5 ppm,

i corresponding to the cluster-bound Pr2NHC isopropyl methyl and olefinic methyl groups. These

i new broad resonances at 5.4 and 7.5 ppm match those of Mn4Te4( Pr2NHC)4 synthesized from i i 31 Pr2NHCTe and low-valent manganese (Figure 2.6c). After the addition of Pr2NHC, the P NMR

i spectrum shows the disappearance of a broad peak at -11 ppm corresponding to P Pr3 bound to the cluster. These results indicate that the stronger NHC ligand replaces the weaker phosphine on our molecular cluster. Similarly, NHC ligands have been used on Co4S4 and Fe4S4 clusters to access oxidation states that are unstable with phosphine ligands.51, 52

2.5. Aggregated Growth into Solid-State Materials

While the NHC ligands can replace phosphines, they themselves remain labile, and in solution their dissociation from the cluster leads to the dimerization of the cubane. After several

i weeks at -35 °C a toluene solution of Mn4Te4( Pr2NHC)4 deposits the dicubane cluster

i Mn8Te8( Pr2NHC)6, which forms subsequent to the loss of one of the NHC caps from each of two cluster monomers. The structure of the dicubane cluster (Figure 2.5b) is composed of two

0 [Mn4Te4] cores directly linked via two Mn-Te bonds. The bridging Mn2Te2 rhomb is strictly planar, featuring distances Mn-Mn (3.172 Å ), Mn-Te (2.751 Å ) and angle Mn-Te-Mn (69.3°).

When the phosphine-capped clusters lose sufficient amounts of phosphine, they aggregate to form an insoluble orange precipitate that is highly air-sensitive and amorphous as determined by PXRD. We performed pyrolysis on the solid that forms when THF is added to a toluene solution

i of Mn4Te4(P Pr3)4. The orange solid was weighed and sealed under vacuum in a 20 cm-long Pyrex tube. The sample was slowly heated to 300 °C in a tube furnace, keeping the empty end of the sample tube at room temperature. The orange solid turned black at ~200 °C, and a clear liquid condensed at the cold end. After 12 h at 300 ºC, the tube was removed from the furnace, cooled, and opened in the glovebox. The PXRD pattern of the black powder (80-90 % of the initial mass) corresponds to that of hexagonal α-MnTe with NiAs lattice (Figure 2.7), and the condensed liquid

i 1 31 contained free P Pr3 as identified by H and P NMR. From the mass of the solid pre- and post- pyrolysis, we calculate that the composition of the aggregation product ranges from

i i (MnTe)(P Pr3)0.25 to (MnTe)(P Pr3)0.17.

Figure 2.7. PXRD of α-MnTe upon pyrolysis of the orange aggregate product.

2.6. Comparison to Bulk MnTe Phases

We consider our manganese telluride molecular clusters as reconstructed fragments of the associated solid-state compound, MnTe. Though the properties of the molecular clusters show fundamental differences with those of their corresponding bulk compound, the study of the formation of molecular clusters is a valuable model for understanding the bulk structure and vice

33 versa. MnTe typically crystallizes in its most energetically favored hexagonal NiAs lattice type53 but it can also adopt a metastable zinc blende structure when grown as films by molecular beam epitaxy at low temperature54, 55 or when alloyed with II-VI semiconductors.56-58 Consistent with these observations, theoretical calculations have suggested that the energy difference between the zinc blende and NiAs structures is small59 while the wurtzite lattice, which has been postulated but has never been experimentally observed,60 is significantly higher in energy.

Figure 2.8. Pictorial comparison of Mn6Te6 cluster and a fragment of solid state wurtzite lattice type.

Based on these experimental and theoretical studies, we examine the cubic Mn4Te4L4 and hexagonal Mn6Te6L6 clusters as fragments of the zinc blende and wurtzite structures, respectively.

Similar M4E4 and M6E6 clusters (M = Zn, Cd and E = S, Se, Te) have been postulated as a base for the growth of zinc blende and wurtzite lattices,61, 62 and both of these species have been observed experimentally by mass spectrometry of ZnS clusters obtained by laser ablation of bulk

ZnS.63 Figure 2.8 illustrates the relationship between the prismane core and the wurtzite structure.

A set of simple transformations can be used to generate a hexagonal fragment of the wurtzite lattice from the core of Mn6Te6L6. At first sight, the link between the Mn4Te4 cubane and the zinc blende structure is less apparent. The Mn4Te4 core structure has a cubic symmetry, suggesting that it can form rock salt or zinc blende lattices. These structures are related structurally by a simple shift of cations from an octahedral environment in the rock salt to a tetrahedral environment in the zinc blende, but we note that the rock salt structure of MnTe is postulated to be much higher in energy 34 than either zinc blende or wurtzite structures.59 Our work highlights the importance of the capping ligands on the internal structure of the cluster, but it also suggests that in the molecular cluster regime, the cubic zinc blende and hexagonal wurtzite structures are relatively close in energy.

These results provide insight into the structural reconstruction taking place during the molecule- to-solid transformation.

2.7. Conclusions

We have described the synthesis and characterization of a family of manganese telluride molecular clusters whose neutral inorganic cores are stabilized by two-electron donor capping ligands. We prepare these clusters by treating a functionally low-valent organometallic complex of manganese with a source of functionally low-valent tellurium. Using this approach, we control both the core structure of the clusters and the ligands that decorate their surfaces. We report three structural types of manganese telluride: the cubane Mn4Te4, the prismane Mn6Te6 and the dicubane

Mn8Te8. Our polynuclear compounds are labile and this feature enables ligand substitution reactions that preserve the internal core structure of the cluster. While ligand dissociation leads to rapid core aggregation, the use of strong NHC capping ligands, slows down this reaction and allows us to isolate a stable intermediate of the aggregation process. Being structurally varied and chemically flexible, the clusters reported in this work form an exciting new class of building blocks for the assembly of solid-state compounds.

2.8. General Synthesis Information

All reactions and sample preparations were carried out under inert atmosphere using standard Schlenk techniques or in a nitrogen-filled glovebox. Anhydrous manganese(II) chloride,

35 trimethylphosphine, triethylphosphine, triisopropylphosphine and tellurium powder were purchased from Strem Chemicals. Allyl magnesium chloride (1.7 M solution in THF), tricyclohexylphosphine, and all other reagents and solvents were purchased from Sigma Aldrich.

Dry and deoxygenated solvents were prepared by elution through a dual-column solvent system

4 27 48 (MBraun SPS). Mn(ƞ -butadiene)2PMe3, Me2NHC (1,3,4,5-tetramethylimidazol-2-ylidene), i 48 3 i 47 Pr2NHC (1,3-diisopropyl-4,5-dimethylimidazol-2-ylidene), Me2NHCTe, Pr2NHCTe, and

i 22, 23 R3PTe (R = Et, Pr, and Cy) were prepared according to published protocols.

2.9. Synthetic Procedures and Characterization of Compounds

Mn4Te4(PEt3)4

4 Method A. To a solution of Mn(ƞ -butadiene)2PMe3 (200 mg, 0.84 mmol) in 3 mL of THF was added Et3PTe (205 mg, 0.84 mmol) and PEt3 (990 mg, 8.36 mmol). The reaction mixture was heated at reflux for 3 h. Once cooled to room temperature, the mixture was filtered through a 0.2

μm syringe filter and concentrated in vacuo. The solution was cooled to -35 °C and dark orange crystals grew over 3 d. The supernatant solution was decanted and the recovered dark orange crystals were dried in vacuo. Yield: 42 mg (17 %)

Method B. A suspension of anhydrous MnCl2 (400 mg, 3.17 mmol) in 50 mL of THF was treated with PEt3 (370 mg, 3.17 mmol). The mixture was cooled to -78 °C and allyl magnesium chloride (3.7 mL, 1.7 M in THF, 6.34 mmol) was added dropwise. After stirring at -78 °C for 2 h, a solution of freshly prepared Et3PTe (410 mg, 3.47 mmol of PEt3 and 400 mg, 3.13 mmol of Te in 18 mL of THF) was added to the reaction mixture, which was subsequently slowly warmed to room temperature. After stirring for 4 h, the reddish, brown mixture was evaporated to dryness in vacuo. The product was dissolved in 30 mL of toluene and the mixture was filtered through a

36 medium frit over Celite. The filtrate was concentrated in vacuo, and filtered through a 0.2 μm syringe filter. Slow pentane vapor diffusion at -35 °C yielded dark orange crystals over 1 week.

The supernatant solution was decanted and the recovered dark orange crystals were dried in vacuo.

Yield: 200 mg (21 %).

Anal. Calcd for Mn4Te4P4C24H60: C, 23.97; H, 5.03. Found: C, 23.62; H, 4.72.

Figure 2.9. Electronic absorption spectrum of Mn4Te4(PEt3)4 in dichloromethane (0.59 μM). The solution was stabilized by the addition of 75 μL of PEt3.

i Mn4Te4(P Pr3)4

4 Method A. To a solution of Mn(ƞ -butadiene)2PMe3 (150 mg, 0.62 mmol) in 7 mL of THF

i i was added Pr3PTe (180 mg, 0.62 mmol) and P Pr3 (1000 mg, 6.24 mmol). The reaction was heated at reflux for 2 h. Once cooled to room temperature, the mixture was filtered through a 0.2 μm syringe filter and concentrated in vacuo. The solution was cooled to -35 °C and dark orange crystals grew over 24 h. The supernatant solution was decanted and the recovered dark orange crystals were dried in vacuo. Yield: 100 mg (46 %)

Method B. A suspension of anhydrous MnCl2 (600 mg, 4.77 mmol) in 50 mL of THF was

i treated with P Pr3 (760 mg, 4.77 mmol). The mixture was cooled to -78 °C and allyl magnesium chloride (5.6 mL, 1.7 M in THF, 9.53 mmol) was added dropwise. After stirring at -78 °C for 2 h,

i i a solution of freshly prepared Pr3PTe (760 mg, 4.77 mmol of P Pr3 and 610 mg, 4.77 mmol of Te in 20 mL of THF) was added to the reaction mixture, which was subsequently warmed slowly to room temperature. After stirring for 5 h, the reddish, brown reaction mixture was evaporated to dryness in vacuo. The product was dissolved in 40 mL of toluene and the mixture was filtered through a medium frit over Celite. The filtrate was concentrated in vacuo, and filtered through a

0.2 μm syringe filter. The solution was cooled to -35 °C and dark orange crystals grew over 24 h.

The supernatant solution was decanted and the recovered dark orange crystals were dried in vacuo.

Yield: 760 mg (44 %).

Anal. Calcd for Mn4Te4P4C36H84: C, 31.54; H, 6.18. Found: C, 29.90; H, 5.91.

i Figure 2.10. Electronic absorption spectrum of Mn4Te4(P Pr3)4 in dichloromethane (1.35 μM).

Mn4Te4(PCy3)4

4 Method A. To a solution of Mn(ƞ -butadiene)2PMe3 (100 mg, 0.42 mmol) in 3 mL of THF was added Te (53 mg, 0.42 mmol) and PCy3 (176 mg, 0.63 mmol). The reaction was heated at reflux for 6 h. Once cooled to room temperature, the mixture was filtered through a 0.2 μm syringe filter and concentrated in vacuo. Slow pentane vapor diffusion at -35 °C yielded dark orange crystals over 1 week. The supernatant solution was decanted, the recovered dark orange crystals were washed with 2 mL of pentane and dried in vacuo. Yield: 30 mg (15 %)

Method B. To a suspension of anhydrous MnCl2 (250 mg, 2.00 mmol) in 30 mL of THF was added PCy3 (560 mg, 2.00 mmol). The mixture was cooled to -78 °C and allyl magnesium chloride (2.3 mL, 1.7 M in THF, 4.00 mmol) was added dropwise. After stirring at -78 °C for 2 h, a solution of freshly prepared Cy3PTe (560 mg, 2.00 mmol of PCy3 and 253 mg, 2.00 mmol of Te in 10 mL of THF) was added to the reaction mixture, which was subsequently slowly warmed to room temperature. After stirring for 2 h, the reddish, brown reaction mixture was evaporated to dryness in vacuo. The product was dissolved in 20 mL of toluene and the mixture was filtered through a medium frit over Celite. The filtrate was concentrated in vacuo, and filtered through a

0.2 μm syringe filter. The solution was cooled to -35 °C and brown crystals grew over 1 week. The supernatant solution was decanted and the recovered dark orange crystals were washed with 3 mL of pentanes and dried in vacuo. Yield: 500 mg (52 %). Anal. Calcd for Mn4Te4P4C72H132: C, 46.70;

H, 7.18. Found: C, 46.42; H, 7.18.

Figure 2.11. Electronic absorption spectrum of Mn4Te4(PCy3)4 in dichloromethane (1.56 μM).

Mn6Te6(PMe3)6

Tellurium (200 mg, 1.56 mmol) and PMe3 (2.0 mL, 19.40 mmol) were stirred in 5 mL of

4 THF for 1 h. The complex Mn(ƞ -butadiene)2PMe3 (355 mg, 1.48 mmol) was added as a solid to the resulting green-grey suspension and the reaction mixture was heated at reflux for 2 h. Once 39 cooled to room temperature, the mixture was filtered through a 0.2 μm syringe filter, and the dark brown precipitate was dissolved in 3 mL of dichloromethane. The solution was filtered through a

0.2 μm syringe filter. The solution was cooled to -35 °C and brown crystals grew over 3 d. The supernatant solution was decanted and the recovered brown crystals were dried in vacuo. Yield:

190 mg (32 %). Anal. Calcd for Mn6Te6P6C18H54: C, 13.93; H, 3.51. Found: C, 12.02; H, 2.85.

Figure 2.12. Electronic absorption spectrum of Mn6Te6(PMe3)6 in dichloromethane (2.71 μM). The solution was stabilized by the addition of 50 μL of PMe3.

Mn6Te6(PEt3)6

4 To a solution of Mn(ƞ -butadiene)2PMe3 (200 mg, 0.84 mmol) in 3 mL of THF was added

Et3PTe (205 mg, 0.84 mmol) and PEt3 (990 mg, 8.36 mmol). The reaction mixture was heated at reflux for 3 h. Once cooled to room temperature, the mixture was filtered through a 0.2 μm syringe filter. The brown precipitate was dissolved in 2 mL of dichloromethane and filtered through a 0.2

μm syringe filter. The solution was cooled to -35 °C and brown crystals grew over 3 d. The supernatant solution was decanted and the recovered brown crystals were dried in vacuo. Yield:

86 mg (23 %).

Anal. Calcd for Mn6Te6P6C36H90: C, 23.97; H, 5.03. Found: C, 23.04; H, 4.83.

Figure 2.13. Electronic absorption spectrum of Mn6Te6(PEt3)6 in dichloromethane (0.42 μM). The solution was stabilized by the addition of 80 μL of PEt3.

Mn4Te4(Me2NHC)4

4 To a solution of Mn(ƞ -butadiene)2PMe3 (75 mg, 0.31 mmol) in 3 mL of THF was added

Te (40 mg, 0.31 mmol) and Me2NHC (38 mg, 0.31 mmol). The reaction mixture was heated at reflux for 2 h. Once cooled to room temperature, the mixture was filtered through a 0.2 μm syringe filter. The brown precipitate was dissolved in 2 mL of dichloromethane and filtered through a 0.2

μm syringe filter. Slow pentane vapor diffusion at -35 °C yielded dark orange crystals over 5 weeks.

The supernatant solution was decanted and the recovered orange crystals were dried in vacuo.

Yield: 46 mg (45 %).

Anal. Calcd for Mn4Te4N8C28H48: C, 27.41; H, 3.94; N, 9.13. Found: C, 27.45; H, 3.85; N, 8.92.

i Mn4Te4( Pr2NHC)4

4 Method A. To a solution of Mn(ƞ -butadiene)2PMe3 (239 mg, 1.00 mmol) in 7 mL of THF

i was added Pr2NHCTe (308 mg, 1.00 mmol). The reaction mixture was heated at reflux for 3 h.

Once cooled to room temperature, the mixture was filtered through a 0.2 μm syringe filter. Slow

41 pentane vapor diffusion at -35 °C yielded dark orange crystals over 3 d. The supernatant solution was decanted and the recovered orange crystals were dried in vacuo. Yield: 326 mg (89 %)

Method B. A suspension of anhydrous MnCl2 (300 mg, 2.38 mmol) in 25 mL of THF was cooled to -78 °C and allyl magnesium chloride (2.8 mL, 1.7 M in THF, 4.76 mmol) was added

i dropwise. After stirring at -78 °C for 1 h, a solution of Pr2NHCTe (733 mg, 2.38 mmol) in 10 mL of THF was added to the reaction mixture and warmed slowly to room temperature. After stirring for 2 h, the orange mixture was evaporated to dryness in vacuo. The product was dissolved in 10 mL of toluene and the mixture was filtered through a medium frit over Celite. The filtrate was concentrated in vacuo, and filtered through a 0.2 μm syringe filter. Slow pentane vapor diffusion at -35 °C yielded dark orange crystals over 3 d. The supernatant solution was decanted and the recovered orange crystals were dried in vacuo. Yield: 652 mg (71 %)

Anal. Calcd for Mn4Te4N8C44H80: C, 36.41; H, 5.56; N, 7.72. Found: C, 33.76; H, 5.38; N, 6.97.

i Figure 2.14. Electronic absorption spectrum of Mn4Te4( Pr2NHC)4 in dichloromethane (1.71 μM).

i Mn8Te8( Pr2NHC)6

A suspension of anhydrous MnCl2 (540 mg, 4.31 mmol) in 40 mL of THF was cooled to -

78 °C and allyl magnesium chloride (5.0 mL, 1.7 M in THF, 8.63 mmol) was added dropwise.

i After stirring at -78 °C for 1.5 h, a freshly prepared solution of Pr2NHCTe (780 mg, 4.31 mmol

i of Pr2NHC and 550 mg, 4.31 of Te in 20 mL of THF) was added to the reaction mixture and warmed slowly to room temperature. After stirring for 2 h, the orange mixture was evaporated to dryness in vacuo. The product was dissolved in 20 mL of toluene and the mixture was filtered through a medium frit over Celite. The filtrate was concentrated in vacuo, and filtered through a

0.2 μm syringe filter. The solution was cooled to -35 °C and orange crystals grew over 4 weeks.

The supernatant solution was decanted and the recovered orange crystals were dried in vacuo.

Yield: 660 mg (43 %).

Anal. Calcd for Mn8Te8N12C66H120: C, 31.18; H, 4.76; N, 6.61. Found: C, 31.87; H, 4.81; N, 6.78.

We note that when we rapidly crystallize the same product using pentane vapor diffusion at -35 °C,

i Mn4Te4( Pr2NHC)4 forms.

i Figure 2.15. Electronic absorption spectrum of Mn8Te8( Pr2NHC)6 in dichloromethane (0.86 μM).

2.10. Instrumentation

NMR

All 1H and 31P NMR were recorded on a Bruker DRX400 (400 MHz) NMR spectrometer.

Elemental Analysis

All elemental analyses were performed by Robertson Microlit Laboratories with glovebox

43 handling. We note that the ligands on the clusters are labile and the more volatile ones can dissociate from the core when the samples have been evacuated. As a result, we observe lower organic content than expected in some samples containing volatile ligands. For example, the difference between theoretical and found organic content is the greatest for the molecular cluster

Mn6Te6(PMe3)6 which contains the most volatile phosphine PMe3.

Electronic Absorption

Electronic absorption measurements were performed on a Shimadzu UV-1800 spectrophotometer. All samples were dissolved in dichloromethane, loaded in a quartz cuvette in the glovebox and sealed under nitrogen. All spectra were taken following a recording of the background spectrum of the solvent.

X-ray Diffraction

Single crystal X-ray diffraction data were collected on an Agilent SuperNova diffractometer using mirror-monochromated Cu Kα or Mo Kα radiation. The crystals were mounted using a MiTeGen MicroMount cooled to 100 K with an Oxford-Diffraction Cryojet system. Data collection, integration, scaling (ABSPACK) and absorption correction (face-indexed Gaussian integration64 or numeric analytical methods65) were performed in CrysAlisPro.66 Structure solution was performed using ShelXS,67 ShelXT,68 or SuperFlip.69 Subsequent refinement was performed by full-matrix least-squares on F2 in ShelXL.67 Olex270 was used for viewing and to prepare CIF files. PLATON71 was used for SQUEEZE,72 ADDSYM73 and TwinRotMat. Many disordered solvent molecules were modeled as rigid fragments from the Idealized Molecular Geometry

Library.74 Details of crystallographic data and parameters for data collection and refinement are in

Table 2.2.

Powder X-ray diffraction data were collected on a PANalytical X’Pert3 Powder diffractometer. Small crystals were evenly dispersed, uncrushed, on a zero-background Si plate and covered with a polycarbonate dome, as the crushing of the crystals led to the destruction of the crystalline lattice as indicated by significant broadening and degradation of the diffraction pattern upon sample grinding. The polycarbonate dome has a broad amorphous peak from 18 to

25 º, and this background was subtracted using HighScore Plus.

2.11. Structural Determination

Mn4Te4(PEt3)4

A THF solution of Mn4Te4(PEt3)4 was cooled to afford dark orange crystals upon standing

3 d at -35 °C. A suitable crystal (.14 x .11 x .06 mm) was mounted with the aid of STP oil treatment and cooled to 100 K on the diffractometer. Complete data (99.5%) was collected to 0.72 Å . 65473 reflections were collected (1060 unique, 1043 observed) with R(int) = 3.9 % and R(sigma) = 0.8 % after absorption correction (Tmax = 0.883, Tmin = 0.733).

The lattice was cubic I, <|E2-1|> was 0.768, and the apparent Laue group was m-3m.

Therefore, a solution was initially attempted in I-43m. This gave an acceptable refinement with the phosphine alkyl groups disordered over the mirror plane. To resolve the disorder, a solution was attempted in I23. The twin instruction TWIN 0 1 0 1 0 0 0 0 -1 -4 was applied to test for combined merohedral and racemic twinning. Two twin domains were obtained, related by the twin law 0 -1 0/-1 0 0/0 0 1; i.e. a mirror plane perpendicular to 110.

When the twinning had been identified, the refinement in I23 proceeded routinely.

Hydrogen atoms were placed in calculated positions and refined with riding coordinates and ADPs.

The final refinement (1060 data, 30 parameters, 0 restraints) converged with R1 (Fo > 4σ(Fo)) =

- -3 1.2 %, wR2 = 2.9 %, S = 1.19. The largest Fourier features were 0.27 and -0.17 e Å .

i Mn4Te4(P Pr3)4

i A toluene solution of Mn4Te4(P Pr3)4 was cooled to afford dark orange crystals upon standing 24 h at -35 °C. A suitable crystal (.15 x .07 x .05 mm) was mounted with the aid of STP oil treatment and cooled to 100 K on the diffractometer. Complete data (99.8 %) was collected to

0.815 Å . 22304 reflections were collected (5667 unique, 5436 observed) with R(int) = 3.3 % and

R(sigma) = 2.8 % after absorption correction (Tmax = 0.432, Tmin = 0.141).

The structure was solved routinely in C2/c. A toluene molecule was located on an inversion center and refined as a rigid fragment from the IMGL with ADPs restrained by RIGU. All other non-H atoms were freely refined. Hydrogen atoms were placed in calculated positions and refined with riding coordinates and ADPs. The final refinement (5667 data, 278 parameters, 45 restraints) converged with R1 (Fo > 4σ(Fo)) = 2.3 %, wR2 = 5.5 %, S = 1.08. The largest Fourier features were

0.54 and -0.60 e- Å-3.

Mn4Te4(PCy3)4

A toluene solution of Mn4Te4(PCy3)4 was cooled to afford black crystals upon standing 7 d at -35 °C. A suitable crystal (.06 x .03 x .02 mm) was mounted with the aid of STP oil treatment and cooled to 100 K on the diffractometer. Complete data (99.9 %) was collected to 0.833 Å; 98.0 % completeness was collected to 0.815 Å. 36452 reflections were collected (16652 unique, 12884

46 observed) with R(int) = 4.8 % and R(sigma) = 7.0 % after absorption correction (Tmax = 0.751,

Tmin = 0.463).

The structure was solved routinely in P-1 and all non-H atoms were located in several successive difference maps. A toluene molecule in a general position showed slight disorder and was refined as a rigid fragment from the IMGL with ADPs restrained by RIGU. Other non-H atoms were freely refined. Hydrogen atoms were placed in calculated positions and refined with riding coordinates and ADPs. The final refinement (16652 data, 806 parameters, 45 restraints) converged with R1 (Fo > 4σ(Fo)) = 4.4 %, wR2 = 11.6 %, S = 1.03. The largest Fourier features were 1.79 and -1.21 e- Å-3 and occurred near Te atoms.

Mn6Te6(PMe3)6

A dichloromethane solution of Mn6Te6(PMe3)6 was cooled to afford orange crystals upon standing 3 d at -35 ºC. A suitable crystal (.06 x .04 x .04 mm) was mounted rapidly and cooled to

100 K on the diffractometer. Complete data (99.8 %) was collected to 0.815 Å . 65179 reflections were collected (13443 unique, 11510 observed) with R(int) = 5.7 % and R(sigma) = 4.4 % after absorption correction (Tmax = 0.878, Tmin = 0.824).

The lattice was metrically near body-centered tetragonal (a = b = 16.08 Å, c = 41.38 A, α

= β = 90 °, γ = 90.6 °) but R(int) statistics showed the true lattice was face-centered orthorhombic

(a = 22.62, b = 22.86, c = 41.38). The space group was assigned as Fdd2 based on the systematic absences. A solution in P1 and subsequent ADDSYM analysis confirmed this assignment. The structure was solved in Fdd2 using ShelXS. The refinement initially appeared routine with two half-clusters and a dichloromethane in the asymmetric unit. However, the Fourier map revealed a minor (<10 % occupied) orientation of each independent cluster. Additionally, the structure has a

47 local (non-crystallographic) mirror plane perpendicular to the c axis, which in combination with the --2 axis makes the structure pseudocentrosymmetric.

Close inspection revealed that the unit cell contains four layers stacked ABCD along the polar c axis, with layers AB related by d glides and layers AC related by the F centering. The minor component is generated by a translation of the entire asymmetric unit by (½ 00), which is a stacking fault corresponding to ABAB stacking. This form should be nearly isoenergetic because of the non-crystallographic mirror plane. The presence of stacking faults perpendicular to c is confirmed by synthetic precession frames which show that reflections with odd indices are systematically weak and streaked along the l reciprocal axis (Figure 2.16).

Figure 2.16. Reconstructed precession frames of the 0kl and 1kl reciprocal layers with k right and l up. Some apparent systematic absence violations in the 0kl layer are caused by overlap with adjacent layers.

The pseudosymmetry and whole-molecule disorder were treated with the following refinement strategy: Each crystallographically independent fragment was placed in a residue with chemically equivalent atoms named equivalently. The atomic positions in each residue were restrained with SAME. Furthermore, all P-C and 1,3 C-C distances were made equivalent with

SADI. The asymmetric unit contains one Te-Mn-PR3 unit in the pseudo-mirror plane, one above,

48 and one below. The atoms above and below the pseudo-plane were chosen so that pairs of atoms were related by the pseudo-inversion center. These pairs of atoms were constrained with EADP.

Furthermore, the ADPs of the minor component were constrained to match their equivalents (by translation ½ 00) in the major component. Finally, all ADPs were further restrained with RIGU and a short-range SIMU for overlapping atoms.

The structure was tested for pseudomerohedral twinning with TwinRotMat and additionally the twin laws 0-10/100/001 (fourfold rotation around the c axis) and 010/100/00-1

(twofold rotation around 110) were tested, but no twin law was found.

All hydrogen atoms were placed in calculated positions and refined with riding coordinates and

ADPs. The Flack parameter was refined to 0.57(6). The final refinement (13443 data, 393 parameters, 2030 restraints) converged with R1 (Fo > 4σ(Fo)) = 4.1 %, wR2 = 8.7 %, S = 1.05. The largest Fourier features were 1.31 and -0.91 e- Å -3.

Mn6Te6(PEt3)6

A dichloromethane solution of Mn6Te6(PEt3)6 was cooled to afford black crystals upon standing 3 d at -35 °C. A suitable crystal (.16 x .09 x .05 mm) was mounted with the aid of STP oil treatment and cooled to 100 K on the diffractometer. Complete data (99.5 %) was collected to

0.815 Å . 18021 reflections were collected (6276 unique, 6039 observed) with R(int) = 3.9 % and

R(sigma) = 3.7 % after absorption correction (Tmax = 0.471, Tmin = 0.075).

2 By analysis of the systematic absences, the space group was Cmc21 or Cmca. E -1 was

0.945 but the a glide absences were weakly violated with = 3.0 vs. 19.0 for all reflections.

Refinement in Cmc21 proceeded successfully while solution in Cmca gave an apparent disorder of

49 the atoms labeled P52 and P72, which are equivalent in Cmca and inequivalent in Cmc21. Analysis in PLATON ADDSYM confirmed that Cmc21 was the correct choice.

All Mn, Te and P atoms were located easily. P52 and P72 lie on a mirror plane and the alkyl groups on these phosphines were disordered over the mirror plane. Due to this disorder and the pseudo-centrosymmetry discussed above, these alkyl groups were located with difficulty using extensive distance restraints for the intermediate refinements. In the final refinement, the disordered ethyl groups were restrained with SAME and RIGU instructions. All other non-H atoms were refined freely. Hydrogen atoms were placed in calculated positions and refined with riding coordinates and ADPs.

The structure contained a void with an apparent THF molecule disordered over a mirror plane. Explicit refinement of this solvent was unsuccessful. Analysis in PLATON SQUEEZE gave a void of 157 Å 3 containing 46.5 electrons (THF = 40 e-). This was modeled as a diffuse contribution to the overall scattering. The final refinement (6276 data, 309 parameters, 193 restraints) converged with R1 (Fo > 4σ(Fo)) = 3.0 %, wR2 = 7.9 %, S = 1.08. The largest Fourier features were 1.08 and -0.70 e- Å -3.

Mn4Te4(Me2NHC)4

A dichloromethane solution of Mn4Te4(Me2NHC)4 was cooled to -35°C with slow vapor diffusion of pentanes to afford orange crystals after 5 weeks. A suitable crystal (.10 x .04 x .02 mm) was mounted rapidly and cooled to 100 K on the diffractometer. After data collection, inspection of diffraction images and a plot of frame scales revealed that substantial radiation damage had occurred after several hours. Accordingly, the run list was truncated to give a minimal complete data set and crystal decay (B factor only) was refined in ABSPACK. After this procedure

50 and face-indexed absorption correction (Tmax 0.615, Tmin 0.159), the data set was 99.8 % complete to 0.833 Å in mmm with R(int) 7.0 % and R(sigma) 7.8 %. 23035 reflections were collected with

8095 unique and 6691 observed.

The structure was solved readily in ShelXS and all non-H atoms were located routinely.

Since the diffraction was somewhat weak and dominated by the heavy atoms, some restraints were required for C and N positions and ADPs. The four Me2NHC ligands were made similar using

SAME and all anisotropic ADPs were restrained with RIGU. A cocrystallized molecule of dichloromethane was restrained with DFIX for 1,2 and 1,3 distances.

All hydrogen atoms were placed in calculated positions and refined with riding coordinates and ADPs. The Flack parameter was determined as -.011(14) by the Parsons method. The final refinement (8095 data, 440 parameters, 568 restraints) converged with R1 (Fo > 4σ(Fo)) = 8.5 %,

- -3 wR2 = 24.0 %, S = 1.06. The largest Fourier features were 2.27 and -2.26 e Å and occurred near

Te atoms.

i Mn4Te4( Pr2NHC)4

i A toluene solution of Mn4Te4( Pr2NHC)4 was cooled to -35 °C with slow vapor diffusion of hexanes to afford orange block-like crystals after 3 d. A suitable crystal (.19 x .12 x .09 mm) was mounted quickly under air with the aid of STP oil treatment and cooled to 100 K on the diffractometer. Complete data were collected to 0.833 Å and 91.7 % completeness to 0.72 Å .

252569 reflections were collected (33207 unique, 26451 observed) with R(int) 4.5 % and R(sigma)

3.3 % after absorption correction (Tmax .840, Tmin .703).

The diffraction showed evidence of extensive disorder; many reflections were streaked and some diffuse scattering was present (Figure 2.17). The crystal used for the full data set was the

51 best of many that were examined from several crystallization attempts. Many crystals showed substantial radiation damage during the data collection. There were weak, diffuse superstructure reflections corresponding to a doubled a-axis (Figure 2.18), but when the data set was integrated and solved on this larger cell the refinement was unstable and the difference map was unsatisfactory.

Figure 2.17. Diffraction image showing streaks and diffuse scattering.

Figure 2.18. Synthetic precession image (h0l section; h right, l up) showing weak, diffuse superstructure reflections.

The space group was assigned as P-1 based on the E2-1 statistic. The structure solved easily in P-1 with two clusters in the asymmetric unit. (A solution in P1 was also evaluated but the refinement was not improved and the Flack parameter was 0.5.) Each independent cluster appeared to be disordered over two positions. One disorder could be resolved in two parts and the other disorder caused relatively large difference peaks near the Te atoms of the cluster. The three independent cluster geometries were restrained with SAME for the intermediate refinements but their atomic positions were unrestrained in the final refinement.

i Six of the eight independent Pr2NHC ligands were disordered over two or three positions.

The more well-behaved ligands were used as FRAG fragments to introduce the starting positions of the more severely disordered ligands. The geometries of all ligands were restrained with SAME and FLAT and all Mn-C distances were made equivalent using SADI. Several pairs of ligands had unreasonably short intra- or intermolecular contacts and these were treated by introducing anti- bumping restraints where necessary. One position of a disordered ligand (RESI NHC 8) was located near an inversion center so that it overlapped with its own symmetry equivalent; this ligand was placed in PART -1 because both symmetry-equivalent positions cannot be present in the same unit cell. (This apparent disorder could be interpreted in terms of a supercell or a pseudocentrosymmetric structure, but these possibilities did not result in satisfactory refinements as noted above.)

There were two toluene molecules in the asymmetric unit, one disordered over two positions and one in three positions. These were refined as rigid fragments with DFT-optimized geometry from the Idealized Molecular Geometry Library.74 All ADPs were stabilized with RIGU and SIMU.

Hydrogen atoms were placed in calculated positions and refined with riding coordinates and ADPs. The final refinement (33207 data, 2256 parameters, 8971 restraints) converged with R1

- (Fo > 4σ(Fo)) = 4.7 %, wR2 = 9.8 %, S = 1.02. The largest Fourier features were 2.31 and -1.23 e

Å -3 and occurred near Te atoms.

i Mn8Te8( Pr2NHC)6

i A toluene solution of Mn8Te8( Pr2NHC)6 was cooled to afford orange crystals upon standing in solution for 4 weeks at -35 °C. The crystals were covered with oil inside of the glovebox and mounted rapidly under air. A suitable crystal (.07 x .04 x .02 mm) was mounted with the aid of STP oil treatment and cooled to 100 K on the diffractometer. Complete data (99.6 %) was collected to 0.833 Å ; 97.3 % completeness was collected to 0.815 Å . 30272 reflections were collected (10212 unique, 7940 observed) with R(int) = 4.1 % and R(sigma) = 5.4 % after absorption correction (Tmax = 0.677, Tmin = 0.268).

The structure solved readily in P-1. The cluster sits on an inversion center with half a molecule in the asymmetric unit. There is one disordered toluene on a general position and one on an inversion center. These were refined with a SAME restraint on their geometry, a short-range

SIMU restraint for overlapping atoms, and RIGU restraints on all atoms. All other atoms were freely refined. Hydrogen atoms were placed geometrically and refined with riding coordinates and

ADPs.

The final refinement (10212 data, 635 parameters, 696 restraints) converged with R1 (Fo >

- -3 4(Fo)) = 4.4 %, wR2 = 11.1 %, S = 1.02. The largest Fourier features were 2.00 and -1.09 e Å and occurred near Te atoms.

Table 2.2. Selected crystallographic data for manganese telluride molecular clusters.

i Compound Mn4Te4(PEt3)4 Mn4Te4(P Pr3)4 Mn4Te4(PCy3)4 Mn6Te6(PMe3)6 C H Mn P Te , C H Mn P Te , Formula 36 84 4 4 4 72 132 4 4 4 C24H60Mn4P4Te4 C7H8 C7H8 C18H56Mn6P6Te6, CH2Cl2 MW 1202.76 1463.2 1943.94 1636.59 Space group I23 C2/c P-1 Fdd2 a (Å ) 13.17415(5) 23.6635(3) 14.2055(4) 22.5903(5) b (Å ) 13.17415(5) 10.83596(10) 15.2008(6) 22.8156(5) c (Å ) 13.17415(5) 23.9044(3) 20.8007(6) 41.3011(7) α (°) 90 90 83.467(3) 90 β (°) 90 107.9866(12) 84.482(2) 90 γ (°) 90 90 76.400(3) 90 V (Å 3) 2286.48(2) 5829.92(11) 4326.1(2) 21287.0(7) Z 2 4 2 16 -3 ρcalc (g cm ) 1.747 1.667 1.492 2.043 T (K) 100 100 100 100 λ (Å) 0.71073 1.54184 1.54184 0.71073

2θmin, 2θmax 7.58, 59.07 12.66, 142.9 8.33, 143.6 6.42, 59.53 Nref 65473 22304 36452 65179 R(int), R(σ) 0.0397, 0.0076 0.0330, 0.0283 0.0476, 0.0700 0.0566, 0.0442 μ(mm-1) 3.733 23.592 16.047 4.897

Tmax, Tmin 0.883, 0.733 0.432, 0.141 0.751, 0.463 0.878, 0.824 Data 1060 5667 16652 13443 Restraints 0 45 45 2030 Parameters 30 278 806 393

R1(obs) 0.0118 0.0225 0.0444 0.0406

wR2(all) 0.0284 0.055 0.1157 0.0868 S 1.194 1.073 1.032 1.047 Peak, hole (e- Å -3) 0.27, -0.17 0.54, -0.60 1.79, -1.21 1.31, -0.91 Flack 0.56(6)

i i Compound Mn6Te6(PEt3)6 Mn4Te4(Me2NHC)4 Mn4Te4( Pr2NHC)4 Mn8Te8( Pr2NHC)6 C H Mn N Te , C H Mn N Te , C H Mn N Te , Formula 28 48 4 8 4 44 80 4 8 4 66 120 8 12 8 C36H90Mn6P6Te6 CH2Cl2 C7H8 3C7H8 MW 1804.13 1311.83 1543.45 2818.45 Space group Cmc21 P212121 P-1 P-1 a (Å ) 20.8798(6) 9.5916(4) 13.84818(18) 14.1776(4) b (Å ) 14.5239(4) 19.4968(15) 19.8259(3) 14.6358(7) c (Å ) 22.4484(7) 23.8118(14) 24.1907(4) 14.9029(8) α (°) 90 90 93.2204(11) 61.056(5) β (°) 90 90 103.8643(12) 83.162(4) γ (°) 90 90 97.4031(11) 87.953(3) V (Å 3) 6807.6(4) 4452.9(5) 6368.18(15) 2686.0(2) Z 4 4 4 1 -3 ρcalc (g cm ) 1.76 1.957 1.61 1.742 T (K) 100 100 100 100 λ (Å) 1.54184 1.54184 0.71073 1.54184

2θmin, 2θmax 10.83, 143.3 8.70, 147.4 6.63, 59.41 8.97, 143.0 Nref 18021 23035 252629 30272 R(int), R(σ) 0.0390, 0.0371 0.0699, 0.0784 0.0455, 0.0327 0.0413, 0.0538 μ(mm-1) 30.153 30.623 2.608 24.527

Tmax, Tmin 0.471, 0.075 0.615, 0.159 0.840, 0.703 0.677, 0.268 Data 6276 8095 33207 10212 Restraints 193 568 8971 696 Parameters 309 440 2256 635

R1(obs) 0.0299 0.0854 0.0466 0.0443 wR2(all) 0.0787 0.2403 0.0977 0.111 S 1.066 1.065 1.115 1.024 Peak, hole (e- Å -3) 1.08, -0.70 2.27, -2.26 2.31, -1.23 2.00, -1.09 Flack 0.493(11) -0.011(14)

2.12. Powder X-Ray Diffraction

The collected diffraction patterns are in good agreement with the simulated powder patterns generated from the SCXRD data and there are no indications of amorphous content, confirming the purity of the crystalline phase. We note that the simulated patterns were generated from SCXRD run at 100 K, while the PXRD patterns were collected at room temperature, and that some of the solvates are removed from the crystal lattice under ambient conditions. These details account for the differences between the experimental PXRD patterns and the simulated ones. We

i were unable to obtain PXRD patterns for Mn4Te4(Me2NHC)4 and Mn4Te4( Pr2NHC)4 as the crystalline materials degrade quickly at room temperature under radiation. It is worth noting that we also observed substantial radiation damage during SCXRD data collection at 100 K.

Figure 2.19. Simulated PXRD of Mn4Te4(PEt3)4 as predicted from the SCXRD structure.

Figure 2.20. PXRD of Mn4Te4(PEt3)4.

Figure 2.21. Simulated PXRD of Mn6Te6(PEt3)6 as predicted from the SCXRD structure.

Figure 2.22. PXRD of Mn6Te6(PEt3)6. The peak at 9º suggests that a small amount of Mn4Te4(PEt3)4 is present in the sample.

i Figure 2.23. Simulated PXRD of Mn4Te4(P Pr3)4 as predicted from the SCXRD structure.

i Figure 2.24. PXRD of Mn4Te4(P Pr3)4.

Figure 2.25. Simulated PXRD of Mn4Te4(PCy3)4 as predicted from the SCXRD structure.

Figure 2.26. PXRD of Mn4Te4(PCy3)4.

i Figure 2.27. Simulated PXRD of Mn8Te8( Pr2NHC)6 as predicted from the SCXRD structure.

i Figure 2.28. PXRD of Mn8Te8( Pr2NHC)6.

Figure 2.29. Simulated PXRD of Mn6Te6(PMe3)6 as predicted from the SCXRD structure.

Figure 2.30. PXRD of Mn6Te6(PMe3)6.

2.13. References

1. Choi, B.; Paley, D. W.; Siegrist, T.; Steigerwald, M. L.; Roy, X., Ligand Control of Manganese Telluride Molecular Cluster Core Nuclearity. Inorg. Chem. 2015, 54, 8348- 8355. 2. King, P.; Wernsdorfer, W.; Abboud, K. A.; Christou, G., Single-Molecule Magnets: A Reductive Aggregation Route to New Types of Mn12 Complexes. Inorg. Chem. 2005, 44, 8659-8669. 3. Lecren, L.; Roubeau, O.; Li, Y. G.; Le Goff, X. F.; Miyasaka, H.; Richard, F.; Wernsdorfer, W.; Coulon, C.; Clerac, R., One-Dimensional Coordination Polymers of Antiferromagnetically-Coupled [Mn4] Single-Molecule Magnets. Dalton Trans. 2008, 755-766. 4. Christou, G.; Gatteschi, D.; Hendrickson, D. N.; Sessoli, R., Single-Molecule Magnets. MRS Bull. 2000, 25, 66-71. 5. Yano, J.; Kern, J.; Sauer, K.; Latimer, M. J.; Pushkar, Y.; Biesiadka, J.; Loll, B.; Saenger, W.; Messinger, J.; Zouni, A.; Yachandra, V. K., Where Water is Oxidized to Dioxygen: Structure of the Photosynthetic Mn4Ca Cluster. Science 2006, 314, 821-825. 6. Tsui, E. Y.; Kanady, J. S.; Agapie, T., Synthetic Cluster Models of Biological and Heterogeneous Manganese Catalysts for O2 Evolution. Inorg. Chem. 2013, 52, 13833- 13848.

7. Talwar, D. N.; Holliday, K., Localized Excitations in Diluted Magnetic Cd1-xMnxTe Semiconductor Ternary Alloys and CdTe/Cd1-xMnxTe Superlattices. Physica B 1999, 263, 540-545. 8. Furdyna, J. K., Diluted Magnetic Semiconductors. J. Appl. Phys. 1988, 64, R29-R64. 9. Burger, A.; Chattopadhyay, K.; Chen, H.; Ndap, J. O.; Ma, X. Y.; Trivedi, S.; Kutcher, S. W.; Chen, R. J.; Rosemeier, R. D., Crystal Growth, Fabrication and Evaluation of Cadmium Manganese Telluride Gamma Ray Detectors. J. Cryst. Growth 1999, 198, 872- 876. 10. Hossain, A.; Yakimovich, V.; Bolotnikov, A. E.; Bolton, K.; Camarda, G. S.; Cui, Y.; Franc, J.; Gul, R.; Kim, K. H.; Pittman, H.; Yang, G.; Herpst, R.; James, R. B., Development of Cadmium Magnesium Telluride (Cd1-xMgxTe) for Room Temperature X- and Gamma-Ray Detectors. J. Cryst. Growth 2013, 379, 34-40. 11. Mikulec, F. V.; Kuno, M.; Bennati, M.; Hall, D. A.; Griffin, R. G.; Bawendi, M. G., Organometallic Synthesis and Spectroscopic Characterization of Manganese-Doped CdSe Nanocrystals. J. Am. Chem. Soc. 2000, 122, 2532-2540. 12. Weaver, A. L.; Gamelin, D. R., Photoluminescence Brightening via Electrochemical Trap Passivation in ZnSe and Mn2+-Doped ZnSe Quantum Dots. J. Am. Chem. Soc. 2012, 134, 6819-6825. 64

13. Holynska, M.; Clerac, R.; Dehnen, S., A Salicylaldoxime [Mn12] Complex with an II III Unprecedented [(Mn4Mn8 )-Mn (μ-OCH3)4] Core. Eur. J. Inorg. Chem. 2012, 5500- 5505. 14. Sessoli, R.; Tsai, H. L.; Schake, A. R.; Wang, S. Y.; Vincent, J. B.; Folting, K.; Gatteschi, D.; Christou, G.; Hendrickson, D. N., High-Spin Molecules - [Mn12O12(O2CR)16(H2O)4]. J. Am. Chem. Soc. 1993, 115, 1804-1816. 15. Holynska, M.; Frank, N.; Pichon, C.; Jeon, I. R.; Clerac, R.; Dehnen, S., A Defect Supertetrahedron Naphthoxime-Based [Mn9(III)] Single-Molecule Magnet. Inorg. Chem. 2013, 52, 7317-7319. 16. Costa, T.; Dorfman, J.; Hagen, K. S.; Holm, R. H., Synthesis and Stereochemistry of Mononuclear, Binuclear, and Tetranuclear Manganese Thiolates. Inorg. Chem. 1983, 22, 4091-4099. 17. Khadka, C. B.; Macdonald, D. G.; Lan, Y. H.; Powell, A. K.; Fenske, D.; Corrigan, J. F., Trimethylsilylchalcogenolates of Co(II) and Mn(II): From Mononuclear Coordination Complexes to Clusters Containing -ESiMe3 Moieties (E = S, Se). Inorg. Chem. 2010, 49, 7289-7297.

10- + 18. Santner, S.; Dehnen, S., [M4Sn4Se17] Cluster Anions (M = Mn, Zn, Cd) in a Cs Environment and as Ternary Precursors for Ionothermal Treatment. Inorg. Chem. 2015, 54, 1188-1190.

2- 19. Stephan, H. O.; Chen, C. N.; Henkel, G.; Griesar, K.; Haase, W., [Fe4Te4(SPr(I))4] and 3- [Fe4Te4(SPr(I))4] , Novel Fe4Te4 Cluster Complexes in Different Metal Oxidation-States 4- and [Mn4Te4(Tepr(I))4] , the 1st Chalcogenide Cubane Cluster Containing Divalent Metal- Ions. J. Chem. Soc. Chem. Comm. 1993, 886-888.

4- 4- 20. Stephan, H. O.; Henkel, G., [Mn4Te4(SC3H7)4] and [Mn4Te4(SeC3H7)4] - Novel Heterocubane Clusters of Manganese. Angew. Chem. Int. Edit. 1994, 33, 2322-2324. 21. Hosseiny, A.; Mackie, A. G.; Mcauliffe, C. A.; Minten, K., The Coordination Chemistry of Manganese: The Preparation and Spectroscopic Characterization of Some Manganese(II) Complexes of Triphenylphosphine of Unusual Stoichiometry, Mn(PPh3)X2. Inorg. Chem. 1981, 49, 99-105. 22. Zingaro, R. A.; Steeves, B. H.; Irgolic, K., Phosphine Tellurides. J. Organomet. Chem. 1965, 4, 320-323. 23. Steigerwald, M. L.; Sprinkle, C. R., Application of Phosphine Tellurides to the Preparation of Group-II-VI (2-16) Semiconductor-Materials. Organometallics 1988, 7, 245-246. 24. Steigerwald, M. L.; Siegrist, T.; Stuczynski, S. M., Initial-Stages in the Molecule-Based Growth of the Solid-State Compound CoTe. Inorg. Chem. 1991, 30, 4940-4945.

25. Hessen, B.; Siegrist, T.; Palstra, T.; Tanzler, S. M.; Steigerwald, M. L., Cr6Te8(PEt3)6 and a Molecule-Based Synthesis of Cr3Te4. Inorg. Chem. 1993, 32, 5165-5169. 65

26. Steigerwald, M. L.; Siegrist, T.; Stuczynski, S. M.; Kwon, Y. U., (Et3P)4Fe4Te4 - An Intermediate between Molecular Reagents and Solid-State Products. J. Am. Chem. Soc. 1992, 114, 3155-3156. 27. Harlow, R. L.; Krusic, P. J.; Mckinney, R. J.; Wreford, S. S., Preparation and Properties of Paramagnetic Bis(Eta-Butadiene)(Lewis Base)Manganese Complexes. Organometallics 1982, 1, 1506-1513. 28. Steigerwald, M. L.; Siegrist, T.; Gyorgy, E. M.; Hessen, B.; Kwon, Y. U.; Tanzler, S. M., Effect of Diverse Ligands on the Course of a Molecules-to-Solids Process and Properties of Its Intermediates. Inorg. Chem. 1994, 33, 3389-3395. 29. Tolman, C. A., Steric Effects of Phosphorus Ligands in Organometallic Chemistry and Homogeneous Catalysis, Chem. Rev. 1975, 77, 313-348. 30. Eichhofer, A.; Fenske, D., Syntheses and Structures of New Copper(I) Selenide Clusters [Cu32Se16(PPh3)12], [Cu52Se26(PPh3)16] and [Cu72Se36(PPh3)20]. J. Chem. Soc. Dalton Trans. 1998, 2969-2972. 31. Dahl, L. F.; Ishishi, E.; Rundle, R. E., Polynuclear Metal Carbonyls: 1. Structures of Mn2(CO)10 and Re2(CO)10. J. Chem. Phys. 1957, 26, 1750-1751.

32. Dahl, L. F.; Rundle, R. E., Crystal Structure of Dimanganese Decacarbonyl, Mn2(CO)10. Acta Crystallogr. 1963, 16, 419-426.

+ 33. Snyder, B. S.; Holm, R. H., [Fe6S6(PEt3)6] - Extension of Stabilization of the Basket Core + Topology to the [Fe6S6] Oxidation Level. Inorg. Chem. 1990, 29, 274-279. 34. Coucouvanis, D.; Salifoglou, A.; Kanatzidis, M. G.; Simopoulos, A.; Kostikas, A., Synthesis, Structural Characterization, and Electronic Structures of the 3- 3- [Fe6S6Cl6(Mo(CO)3)2] , [Fe6S6Br6(Mo(CO)3)2] Clusters. J. Am. Chem. Soc. 1987, 109, 3807-3808. 35. Kanatzidis, M. G.; Dunham, W. R.; Hagen, W. R.; Coucouvanis, D., A New Iron Sulfide 3+ Cluster Containing the Prismane [Fe6(μ-S)6] Core - Synthesis, Structure, and Properties of [Et4N]3[Fe6S6Cl6]. J. Chem. Soc. Chem. Comm. 1984, 356-358.

3+ 36. Kanatzidis, M. G.; Salifoglou, A.; Coucouvanis, D., Chemistry of [Fe6S6] Prismatic Cages - Synthesis, Structural Characterization, and Electronic-Structures of the - - [Et4N]3[Fe6S6I6] Clusters (L = P-CH3C6H4O ,Br ). Inorg. Chem. 1986, 25, 2460-2468. 37. Rao, P. V.; Holm, R. H., Synthetic Analogues of the Active Sites of Iron-Sulfur Proteins. Chem. Rev. 2004, 104, 527-559. 38. Hopkinson, M. N.; Richter, C.; Schedler, M.; Glorius, F., An Overview of N-Heterocyclic Carbenes. Nature 2014, 510, 485-496. 39. Bourissou, D.; Guerret, O.; Gabbai, F. P.; Bertrand, G., Stable Carbenes. Chem. Rev. 2000, 100, 39-91. 66

40. Chianese, A. R.; Kovacevic, A.; Zeglis, B. M.; Faller, J. W.; Crabtree, R. H., Abnormal C5-Bound N-Heterocyclic Carbenes: Extremely Strong Electron Donor Ligands and their Iridium(I) and Iridium(III) Complexes. Organometallics 2004, 23, 2461-2468. 41. Kausamo, A.; Tuononen, H. M.; Krahulic, K. E.; Roesler, R., N-Heterocyclic Carbenes with Inorganic Backbones: Electronic Structures and Ligand Properties. Inorg. Chem. 2008, 47, 1145-1154. 42. Ofele, K.; Herrmann, W. A.; Mihalios, D.; Elison, M.; Herdtweck, E.; Scherer, W.; Mink, J., Multiple Bonds between Main-Group Elements and Transition-Metals: 126. Heterocyclic Carbenes as Phosphane-Analog Ligands in Metal-Complexes. J. Organomet. Chem. 1993, 459, 177-184. 43. Huang, J. K.; Schanz, H. J.; Stevens, E. D.; Nolan, S. P., Stereoelectronic Effects 5 Characterizing Nucleophilic Carbene Ligands Bound to the Cp*RuCl (Cp* = η -C5Me5) Moiety: A Structural and Thermochemical Investigation. Organometallics 1999, 18, 2370- 2375. 44. Chai, J. F.; Zhu, H. P.; Most, K.; Roesky, H. W.; Vidovic, D.; Schmidt, H. G.; Noltemeyer, M., Synthesis and Reaction of Mn-II Iodides Bearing the β-Diketiminate Ligand: The First Divalent Manganese N-Heterocyclic Carbene Complexes. Eur. J. Inorg. Chem. 2003, 4332-4337. 45. Chai, J. F.; Zhu, H. P.; Ma, Q. J.; Roesky, H. W.; Schmidt, H. G.; Noltemeyer, M., Synthesis and Structural Characterization of Three-Coordinate Mn-II, Fe-II, and Zn-II 2- Complexes Containing a Bulky Diamide Ligand [DippN(CH2)3NDipp] (Dipp = 2,6- i Pr2C6H3). Eur. J. Inorg. Chem. 2004, 4807-4811. 46. Hock, S. J.; Schaper, L. A.; Herrmann, W. A.; Kuhn, F. E., Group 7 Transition Metal Complexes with N-Heterocyclic Carbenes. Chem. Soc. Rev. 2013, 42, 5073-5089. 47. Kuhn, N.; Henkel, G.; Kratz, T., Imidazole Chemistry: 3. 2-Telluroimidazolines - Stable Tellurocarbonyl Compounds. Chem. Ber. 1993, 126, 2047-2049. 48. Kuhn, N.; Kratz, T., Synthesis of Imidazol-2-ylidenes by Reduction of Imidazole-2(3H)- thiones. Synthesis 1993, 561-562. 49. Goh, C.; Segal, B. M.; Huang, J. S.; Long, J. R.; Holm, R. H., Polycubane Clusters: 1+,0 0 Synthesis of [Fe4S4(PR3)4] (R = Bu-t, Cy, Pr-i) and [Fe4S4] Core Aggregation upon Loss of Phosphine. J. Am. Chem. Soc. 1996, 118, 11844-11853. 50. Zhou, H. C.; Holm, R. H., Synthesis and Reactions of Cubane-Type Iron-Sulfur-Phosphine Clusters, Including Soluble Clusters of Nuclearities 8 and 16. Inorg. Chem. 2003, 42, 2478- 2478. 51. Deng, L.; Holm, R. H., Stabilization of Fully Reduced Iron-Sulfur Clusters by Carbene 0 Ligation: The [FenSn] Oxidation Levels (n = 4, 8). J. Am. Chem. Soc. 2008, 130, 9878- 9886.

52. Deng, L.; Bill, E.; Wieghardt, K.; Holm, R. H., Cubane-Type Co4S4 Clusters: Synthesis, Redox Series, and Magnetic Ground States. J. Am. Chem. Soc. 2009, 131, 11213-11221. 53. Sugiura, H.; Sawaoka, A.; Saito, S.; Inoue, K., Effect of Pressure on the Neel Temperature and the Lattice-Parameter of MnTe. J. Phys. Chem. Solids 1979, 40, 701-705. 54. Buschert, J. R.; Peiris, F. C.; Samarth, N.; Luo, H.; Furdyna, J. K., Unusual Elastic- Constants of Cubic MnTe in Strained-Layer Superlattices Measured by X-Ray Diffraction. Phys. Rev. B 1994, 49, 4619-4622. 55. Janik, E.; Dynowska, E.; BakMisiuk, J.; Leszczynski, M.; Szuszkiewicz, W.; Wojtowicz, T.; Karczewski, G.; Zakrzewski, A. K.; Kossut, J., Structural Properties of Cubic MnTe Layers Grown by MBE. Thin Solid Films 1995, 267, 74-78. 56. Lee, Y. R.; Ramdas, A. K.; Aggarwal, R. L., Energy-Gap, Excitonic, and Internal Mn2+ Optical-Transition in Mn-Based II-VI Diluted Magnetic Semiconductors. Phys. Rev. B 1988, 38, 10600-10610. 57. Munekata, H.; Ohno, H.; Vonmolnar, S.; Segmuller, A.; Chang, L. L.; Esaki, L., Diluted Magnetic III-V Semiconductors. Phys. Rev. Lett. 1989, 63, 1849-1852. 58. Dietl, T., A Ten-Year Perspective on Dilute Magnetic Semiconductors and Oxides. Nat. Mater. 2010, 9, 965-974. 59. Krause, M.; Bechstedt, F., Structural and Magnetic Properties of MnTe Phases from Ab Initio Calculations. J. Supercond. Nov. Magn. 2013, 26, 1963-1972. 60. Vanyarkho, V. G.; Zlomanov, V. P.; Novoselova, A. V., Physicochemical Study of Manganese Telluride. Neorg. Mater. 1970, 6, 1257-1259. 61. La Porta, F. A.; Andres, J.; Li, M. S.; Sambrano, J. R.; Varela, J. A.; Longo, E., Zinc Blende versus Wurtzite ZnS Nanoparticles: Control of the Phase and Optical Properties by Tetrabutylammonium Hydroxide. Phys. Chem. Chem. Phys. 2014, 16, 20127-20137.

4- 4- 62. Choy, A.; Craig, D.; Dance, I.; Scudder, M., [S4Zn10(SPh)16] , [S4Cd10(SpP)16] , a Molecular Fragment of the Sphalerite Ms Lattice - Structural Congruence of Metal Sulfides and Metal Thiolates. J. Chem. Soc. Chem. Comm. 1982, 1246-1247. 63. Burnin, A.; Sanville, E.; Bel Bruno, J. J., Experimental and Computational Study of the + ZnnSn and ZnnSn Clusters. J. Phys. Chem. A 2005, 109, 5026-5034. 64. Blanc, E.; Schwarzenbach, D.; Flack, H. D., The Evaluation of Transmission Factors and Their 1st Derivatives with Respect to Crystal Shape Parameters. J. Appl. Crystallogr. 1991, 24, 1035-1041. 65. Clark, R. C.; Reid, J. S., The Analytical Calculation of Absorption in Multifaceted Crystals. Acta Crystallogr. A 1995, 51, 887-897.

66. Version 1.171.37.35, Oxford Diffraction / Agilent Technologies UK Ltd, Yarnton, England. 2014. 67. Sheldrick, G. M., A Short History of SHELX. Acta Crystallogr. A 2008, 64, 112-122. 68. Sheldrick, G. M., Crystal Structure Refinement with SHELXL. Acta Crystallogr. C 2015, 71, 3-8. 69. Palatinus, L.; Chapuis, G., SUPERFLIP - A Computer Program for the Solution of Crystal Structures by Charge Flipping in Arbitrary Dimensions. J. Appl. Crystallogr. 2007, 40, 786-790. 70. Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.; Puschmann, H., OLEX2: A Complete Structure Solution, Refinement and Analysis Program. J. Appl. Crystallogr. 2009, 42, 339-341. 71. Spek, A. L., Structure Validation in Chemical Crystallography. Acta Crystallogr. D 2009, 65, 148-155. 72. Vandersluis, P.; Spek, A. L., Bypass - An Effective Method for the Refinement of Crystal- Structures Containing Disordered Solvent Regions. Acta Crystallogr. A 1990, 46, 194-201. 73. Lepage, Y., Missym-1.1 - A Flexible New Release. J. Appl. Crystallogr. 1988, 21, 983- 984. 74. Guzei, I. A., An Idealized Molecular Geometry Library for Refinement of Poorly Behaved Molecular Fragments with Constraints. J. Appl. Crystallogr. 2014, 47, 806-809.

Chapter 3. Environmental Effect on Single-Cluster Junctions

3.1. Preface

This chapter is based on a manuscript entitled “Solvent-Dependent Conductance Decay

Constant in Single Cluster Junctions” by Bonnie Choi, Brian Capozzi, Seokhoon Ahn,

Ari Turkiewicz, Giacomo Lovat, Colin Nuckolls, Michael L. Steigerwald,

Latha Venkataraman, and Xavier Roy published in Chemical Science.1 I synthesized all compounds. Dr. Brian Capozzi in Professor Latha Venkataraman’s group performed all single- molecule conductance measurements.

3.2. Introduction

Controlling charge transport through molecular electrodes is critical to the realization of nanoscale electronic devices.2, 3 While numerous organic molecules have been studied as connecting wires for single-molecule junction studies,4-11 very little is known about the effect of metal complexes in these types of junctions.12-15 We recently reported that we could incorporate electron-rich molecular clusters in single-molecule electrical circuits.16 In order to determine how transport through such systems depends on molecular length, we have connected the same electron-rich cluster, Co6Se8, to conducting ligands of varying lengths.

We have found that the inclusion of the cluster in the molecular circuit reduces the effect of ligand length on conductance decay with apparent molecular size. Moreover, we have found that the decay constant is impacted greatly by changing the solvent from 1,2,4-trichlorobenzene

(TCB) to 1-bromonaphthalene (BrN). Specifically, the decay constant of the cluster is 0.04 Å -1 in

BrN, while it is 0.12 Å -1 in TCB. We hypothesize that the unusually low decay constant of the

70 clusters arising in BrN indicates that the cluster energy levels are close to the metal Fermi level

(EF). This work demonstrates, for the first time, a molecular system where the tunneling decay constant can be modified by altering the environment around the molecule.

3.3. Synthetic Design

The single-cluster circuits that we have designed, assembled, and studied consist of an

17, 18 atomically defined Co6Se8 molecular cluster (Figure 3.1a) wired between nanoscale electrodes. The wiring is formed from bifunctional, conjugated ligands (Figure 3.1b) that bind specifically and directionally to the electrode and to the cluster. We employ an atomically defined segment of polyacetylene19 that has a phosphine group on one terminus that coordinates to a cobalt atom on the clusters and a thiomethyl group on the other terminus that attaches to the Au electrode.20, 21 The mono-, di-, and triene ligands are L1, L2, and L3 and the corresponding clusters are 1, 2, and 3, respectively. Figure 3.1c shows the molecular structure of 1 as determined by

SCXRD.16

Figure 3.1. (a) Structure of the cluster core Co6Se8. (b) Chemical structure of the conducting ligand series Ln (n = 1, 2, 3). (c) Molecular structure of 1 as characterized by SCXRD. Carbon, black; cobalt, blue; selenium, green; phosphorus, orange; sulfur, yellow.

3.4. STM-BJ Measurements and Histogram Analysis

We measured the conductance of both the individual molecular clusters (1-3) and the free conducting ligands (L1-L3) using a scanning tunneling microscope based break-junction (STM-

BJ) technique.22 In this technique, an Au STM tip and substrate are repeatedly brought into and out of contact to form and break Au-Au point contacts in solutions of the target compounds. During this process, a bias voltage is applied across the junction while current is measured to determine the conductance (G = I/V) of the junction. The measurements are repeated thousands of times, and the data is analyzed to reveal statistically significant results. The data is processed by compiling thousands of individual conductance traces into one-dimensional (1D), logarithmically-binned conductance histograms.23 We further generate two-dimensional (2D) histograms of the conductance versus displacement by aligning each conductance trace after the point contact rupture (at a conductance of 0.5 G0) and overlaying all conductance traces.

In order to characterize transport through the molecular clusters and the free ligands, we measured the conductance of 1-3 and L1-L3 in two different solvents, BrN and TCB. These solvents were chosen taking into consideration the solubility of both the ligand and the cluster systems as well as for their varied affinity to gold electrodes.23 Figure 3.2a,b contains the 1D histograms for the measurements in BrN, and Figure 3.2d,e shows the same for the measurements in TCB. The 2D histograms for 1 and L1 in each solvent are an inset in the respective figures. The

2D histograms show a significant difference in length of the molecular feature for 1 and for L1.

Moreover, the cluster junction lengths measured from the 2D histograms correlate with the molecular lengths of the cluster with the ligands fully extended (measured in BrN: 9 Å and 21 Å , and expected from SCXRD: 13 Å and 32 Å , for L1 and 1 respectively). Despite the additional complexity of the cluster system, we conclude that we are indeed probing transport through Au-

72 ligand-cluster-ligand-Au junctions based on this large difference in the observed lengths.

Furthermore, the histograms in Figure 3.2a,d shows shoulders, with an increasing prominence for the longer-ligand systems. Comparing the conductance of these shoulders with the ligand conductance in Figure 3.2b,e, we attribute these shoulders to free ligands, that is, ligands that have detached from the clusters.

Figure 3.2. Logarithmically-binned conductance histograms for 1-3 in (a) BrN and (d) TCB and L1-L3 in (b) BrN and (e) TCB. Insets in (a, d) and (b, e) are 2D histograms for 1 and L1 respectively. (c) and (f) Conductance peak values of the single-molecule junctions for 1-3 and L1-L3 as a function of total number of “ene” units (e.g., 1 has 2 units while L1 has 1 unit) for measurements in BrN and TCB, respectively shown on a semi-log plot along with least- square linear fits. Marker size reflects the error in the Gaussian fits to the conductance histogram peaks.

3.5. Conductance Decay Constants for the Ligand and Cluster Series

We fit the peaks of all conductance histograms for both solvents with a Gaussian function and plot the peak conductance values versus the number of C=C units or “enes” in each molecule

(Figure 3.2c,f). In both solvents, and for both free ligand and cluster, we observe that the conductance decreases exponentially with increasing molecular length following the relationship

G ~ e-βn, where n is the number of “ene” units in the backbone and β is the decay constant. We report the decay constant per Angstrom using a length of 2.48 Å per “ene” unit. The decay constant for the free ligand series is essentially independent of the solvent (β = 0.15 Å -1 in TCB and 0.17

Å -1 BrN). The unexpected result is the factor of 3 difference in the decay constants of the cluster series in different solvents as can be seen comparing Figures 4.2c and 4.2f. In TCB, the β of the cluster system is 0.12 Å -1, and in BrN it is 0.04 Å -1. We note that the difference between the decay constant of the ligand and that of the cluster is greater in BrN than in TCB. Furthermore, the absolute values of the conductance of the cluster series are significantly higher when measured in

BrN than in TCB, with the conductance of 3 being almost an order of magnitude higher in BrN compared to TCB. Such a solvent-induced effect on the conductance has been observed in other systems, and this has been attributed to the solvent’s ability to modulate the electrode work function.24, 25

3.6. DFT Calculations

Regardless of solvent, the effect of C=C chain-length on conductance is less pronounced in 1-3 than in L1-L3. Furthermore, the conductance and the decay of L1-L3 are insensitive to the choice of solvent, while the solvent significantly influences those of 1-3. To understand the origin of these trends, we modeled compounds 1-3 with the simplified clusters (Me3P)5Co6Se8(L1),

(Me3P)5Co6Se8(L2), and (Me3P)5Co6Se8(L3), respectively, and compared the electronic structures of these model clusters to that of the corresponding ligands alone. Since the ligands used in this study are generally highest occupied molecular orbital (HOMO)-conducting,26, 27 we examine the occupied orbitals of L1-L3 and 1-3 that contain the sulfur p lone pair and the C=C -bonds.

Figure 3.3. Computational studies of conducting ligands Ln (left) and model clusters (PMe3)5Co6Se8(Ln) (right) for (a) L1 and (PMe3)5Co6Se8(L1), (b) L2 and (PMe3)5Co6Se8(L2), and (c) L3 and (PMe3)5Co6Se8(L3). The orbitals associated with sulfur pπ lone pairs are shown. HOMO levels are shown for L1-L3, and HOMO-8, HOMO-7 and HOMO-4 levels are shown for (Me3P)5Co6Se8(L1), (Me3P)5Co6Se8(L2) and (Me3P)5Co6Se8(L3), respectively.

We find that for each of the clusters, the highest energy occupied orbitals are localized on the inorganic core and uncoupled to the ligand and the Au-binding thiomethyl group. Since these high-energy occupied orbitals are effectively isolated from the environment by the ligands, it is unlikely that they are important for conduction across the molecularly wired cluster. As shown in

Figure 3.3, the highest occupied orbital that has weight on the ligand is HOMO-8 in

(Me3P)5Co6Se8(L1), HOMO-7 in (Me3P)5Co6Se8(L2), and HOMO-4 in (Me3P)5Co6Se8(L3).

These highest ligand-based orbitals, that can couple charge across the molecularly wired cluster junctions, are essentially combinations of the p lone pair on sulfur and the C=C -bonds, and these orbitals are much like the HOMOs of the ligands alone. In each case, the energies of the two related orbitals are essentially identical (e.g., the energy of HOMO-4 in 3 differs from the energy of HOMO in L3 by less than 0.03 eV). Therefore, while the ligand-based orbitals provide a conduit through which the carriers can move from one electrode through the cluster to the other electrode, the variation in the relevant energies should be the same in the 1-3 set as in the L1-L3 set. From a 75 simple model for tunneling transport,28 the barrier for transport is defined by the conducting orbital energy. Thus, if the conducting orbitals' energies are the same in the ligand and the cluster, then the decay constants should be the same in both systems. However, this is not what is experimentally observed.

3.7. Tight-Binding Model

To understand our results, we consider several possible mechanisms of charge transport through these junctions. Charge transfer can occur via a coherent off-resonance process through an orbital lying on the ligand-cluster-ligand assembly that is coupled to both electrodes. In that situation, the conductance depends on at least two related factors: the energy of this conducting

29 orbital relative to the metal EF, and the coupling between this orbital and both electrodes. As the length of the molecule increases, the HOMO-LUMO gap narrows, and if conductance were just related to energy level alignment, one would naively expect conductance to actually increase.

However, transport through the junction is also related to how well the conducting orbital overlaps with the leads, and since the orbital is more delocalized over a longer conjugated molecule, this overlap decreases with increasing length. The conductance thus typically decays exponentially with increasing molecular length. Specifically, as the conjugated backbone gets longer, the molecular orbital is delocalized over a larger molecule, and since the orbital is normalized, a smaller fraction of its amplitude resides on the sulfur atoms; therefore, the bonding (coupling) between the molecule and the electrodes decreases.

Figure 3.4. Sample transmission functions using the tight-binding model for (a) ligand and (b) cluster series. Insets in (a) and (b) are schematic diagrams of the models used. Parameters employed are Γ = -0.15 eV, δ = -1.5 eV, ε = - 3.5 eV, τ = -1 eV, and E0 = -1.9 eV. (c) Conductance values versus number of repeat “ene” units, taken from the transmission functions shown. The data points in (c) are color-coded as the transmission functions in (a) and (b). Data points are fit with a line, and both lines show a similar decay of about 0.5 Å -1. (d) 2D plot showing the ratio of decay constants βLigand/βCluster, obtained by keeping Γ, ε, and δ constant, while varying E0 and τ (the parameters for the cluster site). It is not possible to obtain a ligand decay constant that is more than 1.2 times that of the cluster decay constant.

If we assume that the conducting orbitals of the cluster and of the ligand for a given length are similar in both character and energy, we can develop a simple tight-binding model of the molecular junctions to examine how the additional electronic structure of the cluster could impact transport through the system. Our tight-binding model is schematically presented in the insets of

Figure 3.4a,b for the ligand and the cluster respectively. For the conducting ligands, we assign a single energy level, ε, for each unit, and allow nearest neighbors to be coupled by δ. The terminal units are coupled to the Au electrodes using an imaginary self-energy, iΓ/2. We apply a similar model for the cluster, adding an additional energy level, E0, between two ligands and coupling this site to its nearest neighbor ligand states with τ. We computed the transmission functions for these model systems using a Green’s function approach.30, 31 77

Sample computed transmission functions are shown in Figure 3.4a,b using the same values for ε, δ and Γ for the ligand and the cluster series. The transmission functions display resonances at energy values corresponding to the molecular orbitals of the system where the probability of an electron being transmitted through the system is unity. The transmission function for L1 contains one resonance at energy ε, while longer molecules have resonances equal to the number of sites in the corresponding model. As the length of the molecule increases, the frontier resonance moves closer to EF but also narrows, which is a consequence of the frontier orbital being delocalized over a longer molecular backbone. Upon comparing the transmission functions for the ligands with those of the clusters, we see that the clusters contain resonances that are closer to EF than their ligand counterparts, but with narrower resonances. This observation leads to a lower transmission at EF to which the experimentally measured values should correspond; more importantly, it also leads to a conductance that is more sensitive to the exact location of the EF.

In Figure 3.4c, we show the conductances that are determined from the tight-binding model for each molecule versus the number of “ene” units in the molecule (using the same model parameter values for both series). From the fit to these values, it is clear that the predicted decay constants are essentially the same for the ligand and cluster series. We use one set of E0 and τ values to calculate the representative transmission/conductance functions shown in Figure 3.4a,b.

Regardless of what value is assigned to E0 and τ, we find that this model predicts very similar decay constants for the two systems (Figure 3.4d).

Our tight-binding calculations suggest that the ligand and cluster series should have the same β values, unless the energy alignment of the cluster resonance is altered relative to the electrode Fermi level in this model. We have three sets of observations that support this hypothesis:

1) β of the cluster in BrN is significantly lower than in TCB, 2) β values measured in both solvents

78 are the same for the ligand series, and 3) β of the cluster is lower than that of the ligand in both solvents. The steeper transmission curves of the cluster series in Figure 3.4 indicate that the resonance energies are closer to EF. Within this coherent transport model, we can see that a small change in EF will result in a large shift in β for the cluster relative to the ligand. For instance,

-1 changing EF by -0.5 eV shifts the β value to 0.1 Å for the clusters while a similar change in EF for the ligand changes β to 0.3 Å -1. These results, when viewed in light of the known ability of

24 solvent-binding to produce changes in EF, point to BrN shifting EF closer to resonance relative to TCB. This effect is compounded by the sensitivity of the metal. The free ligand and the cluster have very different characteristics (e.g., size, steric hindrance, redox behavior, dielectric constant polarizability and binding ability) that will result in different shifts in EF.

The discussion above presupposes coherent charge transport through a single orbital spanning the whole junction. An alternative mechanism involves a direct through-space charge transfer from the electrode to an unoccupied molecular level on the cluster through a resonant transfer process.15 In this picture, the cluster does not have to be chemically attached to the electrodes to form a conducting junction and the charge transfer efficiency depends on the core- electrode spacing. We discount this mechanism based on a previously published study in which we demonstrated that our clusters form molecular junctions by bonding their terminal thiomethyl groups to the Au electrodes.16 By varying the substitution pattern or removing the aurophilic functionality, we showed that we can modulate or completely shut down the conductivity of these molecular junctions, suggesting that there is an orbital pathway for the transport of charge in these cluster systems, and refuting the idea of direct through-space charge transfer mediated by an orbital localized on the core.

We also consider a hopping mechanism for charge transfer, a process generally mediated by a thermally induced conformational change.32, 29 We first rule out the possibility that such a conformational change can occur in the ligand4 and/or in the cluster core as the rigid structure of our clusters lacks internal bonds with rotational degrees of freedom that could be accessed through an activated process. On the other hand, the ligand connection points (i.e., the C-S-Au and the C-

P-metal torsional angle) have the rotational degrees of freedom for such a conformational change.33 In this mechanism, the charge would tunnel from the source electrode across the ligand to the cluster and then would transfer to the drain electrode through a second coherent tunneling process. Such a transport process requires that the cluster can reversibly change its oxidation state with each charge transfer. Since the applied bias in these measurements is not small (~ 0.5 V) and the cluster core Co6Se8 is redox active, it is plausible that such a hopping process is at play.

However, as the length of the ligands increases, the probability of tunneling into the cluster should decrease, irrespective of solvent. Within our experimental constrains, it is thus difficult to rationalize our observation that β changes with solvent using this mechanism.

3.8. Conclusions

We measured the charge transport through molecular clusters with ligands of different lengths and showed that the conductance decay depends on the solvent used for these measurements. Our results illustrate a novel effect that allows the environment to alter the conductance decay constants due to the proximity of the cluster resonance energy to the metal

Fermi level. This study opens up the possibility to carry out conductance measurements using

7 clusters designed to modulate EF through a gating effect. While the conducting ligands alone are limited to a one-dimensional system, the three-dimensional architecture of the metal chalcogenide

80 cluster allows us to envision novel electronic devices where a molecular cluster is contacted by electrodes at multiple locations.

3.9. General Synthesis Information

Unless otherwise noted, all reactions were carried out under nitrogen using standard

Schlenk techniques or in a nitrogen-filled glovebox. Chlorodiethylphosphine was purchased from

Acros Organics. Selenium powder and dicobalt octacarbonyl were obtained from Strem Chemicals.

Potassium tert-butoxide, 4-bromobenzaldehyde, and all other reagents and solvents were purchased from Sigma Aldrich. Dry and deoxygenated solvents were prepared by elution through a dual column solvent system (MBraun SPS).

3.10. Synthetic Procedures and Characterization of Compounds

Compound 4

(1,3-Dioxolan-2-ylmethyl)triphenylphosphonium bromide (11.60 g, 27.0 mmol) was dissolved in 120 mL of THF. To the solution, lithium methoxide (1.51 g, 39.7 mmol) suspended in 15 mL of THF was added and rinsed with 5 mL of THF and 5 mL of methanol. A reflux condenser was attached and the suspension was heated to reflux for 30 min. 4-

Bromobenzaldehyde (2.00 g, 10.8 mmol) dissolved in 25 mL of THF was added dropwise to the refluxing suspension and was further heated to reflux for ~12 h. The mixture was cooled to RT.

In air, 100 mL of 10 % HCl solution was added and stirred for 1 h. The mixture was poured into

225 mL of dichloromethane. The organic phase was extracted and the aqueous phase was washed

81 with dichloromethane (3 x 30 mL). The combined organic phase was washed with saturated aqueous NaHCO3 solution and brine, dried with MgSO4, and evaporated to dryness. The crude product was purified by column chromatography. Yield: 1.85 g (81 %).

The structure of compound 4 was confirmed by 1H NMR as published in literature.34

Compound 5

This preparation was analogous to that of compound 4, using 27.1 mmol of (1,3-dioxolan-

2-ylmethyl)triphenylphosphonium bromide, 29.0 mmol of lithium methoxide, and 10.8 mmol of compound 4. Yield: 2.24 g (87 %).

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = 6.23-6.29 (1H, m), 6.95-7.07 (2H, m),

7.24-7.30 (1H, m), 7.38-7.42 (2H, m), 7.52-7.55 (1H, m), 9.61 (1H, d).

Compound 6

4-Bromobenzaldehyde (0.89 g, 4.8 mmol) was dissolved in 40 mL of THF and cooled to

0 °C. Dimethyl-4-thiomethylbenzyl phosphonate35 (1.18 g, 4.8 mmol) was added and the solution was stirred for 30 min. A solution of potassium tert-butoxide (0.62 g, 5.5 mmol) in 10 mL of THF was added dropwise to the cold solution. The reaction was stirred and warmed gradually to RT over ~12 h. In air, 50 mL of water was added and the mixture was poured into 50 mL of

82 dichloromethane. The organic phase was extracted and the aqueous phase was washed with dichloromethane (2 x 10 mL). The combined organic phase was washed with water then brine, dried with MgSO4 and evaporated to dryness. The white solid was recrystallized at -30 °C from a mixture of toluene and n-hexanes. Yield: 1.35 g (92 %).

1 H NMR (300 MHz, [d2-dichloromethane], 298 K): δ = 2.50 (3H, s), 7.00 (2H, m), 7.21-7.25 (2H, m), 7.32-7.48 (6H, m).

Compound 7

This preparation was analogous to that of compound 6, using 4.1 mmol of aldehyde 4, 4.1 mmol of dimethyl-4-thiomethylbenzyl phosphonate, and 4.5 mmol of potassium tert-butoxide.

The white solid was recrystallized at -30 °C from dichloromethane and washed with n-hexanes.

Yield: 803 mg (60 %).

1 H NMR (300 MHz, [d2-dichloromethane], 298 K): δ = 2.51 (3H, s), 6.61 (2H, m), 6.91 (2H, m),

7.21-7.47 (8H, m).

Compound 8

This preparation was analogous to that of compound 6, using 1.1 mmol of compound 5,

1.1 mmol of dimethyl-4-thiomethylbenzyl phosphonate, and 1.2 mmol of potassium tert-butoxide. 83

The yellow solid was recrystallized at -30 °C from dichloromethane and washed with n-hexanes.

Yield: 270 mg (67%).

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = 2.49 (3H, s), 6.48-6.61 (4H, m), 6.84-6.94

(2H, m), 7.19-7.46 (8H, m).

Conducting ligand L1

Compound 6 (1.16 g, 3.8 mmol) was dissolved in 50 mL of THF and cooled to -78 °C. n-

Butyllithium (1.6 M in hexanes, 2.6 mL, 4.2 mmol) was added dropwise and the reaction was stirred for 45 min. Chlorodiethylphosphine (0.57 g, 4.6 mmol) in 10 mL of THF was added dropwise to the solution and the reaction was warmed gradually to RT over ~12 h. The solvent was removed in vacuo and 20 mL of toluene was added to the crude product. The mixture was filtered through a fine frit and the solvent was once again removed in vacuo. The white solid was recrystallized at -30 °C from a mixture of toluene and n-hexanes. Yield: 1.00 g (84 %).

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = 1.02 (6H, m), 1.70 (4H, m), 2.50 (3H, s),

7.10 (2H, m), 7.25 (2H, m), 7.45-7.51 (6H, m).

31 P NMR (162 MHz, [d2-dichloromethane], 298 K): δ = -15.

Conducting ligand L2

This preparation was analogous to that of conducting ligand L1, using 1.2 mmol of compound 7, 1.3 mmol of n-butyllithium, and 1.4 mmol of chlorodiethylphosphine. The yellow solid was recrystallized at -30 °C from THF and washed with n-hexanes. Yield: 178 mg (40 %).

1 H NMR (300 MHz, [d2-dichloromethane], 298 K): δ = 0.98 (6H, m), 1.67 (4H, m), 2.47 (3H, s),

6.65 (2H, m), 6.96 (2H, m), 7.19 (2H, m), 7.35-7.42 (6H, m).

31 P NMR (162 MHz, [d2-dichloromethane], 298 K): δ = -15.

Conducting ligand L3

A solution of tetramethylethylenediamine (0.1 mL, 0.67 mmol) and n-butyllithium (1.7 M in THF, 0.4 mL, 0.67 mmol) in 10 mL of THF was stirred at -78 °C for 30 min. The solution was cannula transferred to compound 8 (200 mg, 0.56 mmol) dissolved in 40 mL of THF at -78 °C and stirred for 1.5 h. Chlorodiethylphosphine (210 mg, 1.68 mmol) in 3 mL of THF was added dropwise, and the mixture was warmed gradually to RT over ~12 h. The solvent was removed in vacuo and 5 mL of THF was added to the crude product. The mixture was filtered through a fine frit and recrystallized at -30 °C from THF. Yield: 92 mg (44 %).

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = 1.04 (6H, m), 1.71 (4H, m), 2.51 (3H, s),

6.52-6.62 (4H, m), 6.84-6.96 (2H, m), 7.21-7.47 (8H, m).

31 P NMR (162 MHz, [d2-dichloromethane], 298 K): δ = -16.

General synthesis of Co6Se8(Ln)6 clusters 1-3:

Co6Se8(L1)6 (1)

Conducting ligand L1 (223 mg, 0.71 mmol) was dissolved in 40 mL of toluene. Selenium powder

(56 mg, 0.71 mmol) was added and the suspension was stirred until the solid dissolved. Dicobalt octacarbonyl (56 mg, 0.16 mmol), dissolved in 5 mL of toluene, was added to the solution and the reaction was heated to reflux for ~12 h. The hot mixture was filtered through a fine frit. The dark brown solution was cooled to RT and concentrated in vacuo, and the product was precipitated with diethyl ether. Yield: 102 mg (65 %). The crystal structure of 1 has been published in a recent report16, and crystallographic data is available from the Cambridge Crystallographic Data Centre

(deposition number 894790).

1 H NMR (400 MHz, [d8-tetrahydrofuran], 298 K): δ = 0.89 (36H, m), 2.06 (24H, m), 2.48 (18H, s), 7.13 (12H, s), 7.20 (12H, m), 7.32-7.43 (36H, m).

31 P NMR (162 MHz, [d8-tetrahydrofuran], 298 K): δ = 58 (broad).

Co6Se8(L2)6 (2)

This preparation was analogous to that of 1, using 0.88 mmol of conducting ligand L2, 0.88 mmol of selenium powder, and 0.40 mmol of dicobalt octacarbonyl. Yield: 137 mg (66 %).

1 H NMR (400 MHz, [d8-tetrahydrofuran], 298 K): δ = 0.90 (36H, m), 2.04 (24H, m), 2.47 (18H, s), 7.04-7.33 (72H, m).

We were not able to measure the 31P NMR spectrum of compound 2 because of its low solubility.

Co6Se8(L3)6 (3)

This preparation was analogous to that of 1, using 0.24 mmol of conducting ligand L3, 0.24 mmol of selenium powder, and 0.06 mmol of dicobalt octacarbonyl. The dark brown product precipitated upon cooling the hot filtrate to RT. Yield: 8 mg (14 %).

1 H NMR (400 MHz, [d8-tetrahydrofuran], 298 K): δ = 0.89 (36H, m), 2.07 (24H, m), 2.46 (18H, s), 6.50-7.38 (84H, m).

We were not able to measure the 31P NMR spectrum of compound 3 due to its low solubility.

3.11. Instrumentation

All 1H and 31P NMR were recorded on a Bruker DRX300 (300 MHz) or Bruker DRX400

(400 MHz) NMR spectrometer.

3.12. Additional Conductance Data

Figure 3.5 shows the 2D conductance versus displacement histograms for ligands L1-L3 and clusters 1-3 in BrN. Both 1D and 2D conductance histograms for L1 and L2 are constructed from 5000 individual conductance traces, while those of L3 are constructed from 6000 traces; all

87 data was collected at an applied voltage of 500mV. Histograms for 1 were constructed from 4000 traces, for 2 were constructed from 3000 traces and for 3 were constructed from 2000 traces. In solution, the clusters tend to decompose over time, so we were unable to obtain as much data as we collected for the ligands. Cluster traces were collected at an applied voltage of 375mV and solution concentration of 5-10 μM.

Figure 3.5. 2D conductance versus displacement histograms for 1-3 and L1-L3 collected in BrN. Histograms were created by aligning individual conductance traces at 0.5 G0 and then overlaying all traces to generate the 2D image. Molecular plateau lengths for the clusters are roughly twice as long as their ligand only counterparts.

We have also carried out conductance measurements on 1-3 and L1-L3 in TCB. 2D conductance versus displacement histograms for ligands L1-L3 and clusters 1-3 in TCB are shown

88 in Figure 3.6. Conductance histograms are constructed from 3000 traces for 1, 1000 traces for 2,

4000 traces for 3, and 10000 traces for L1-L3.

Figure 3.6. 2D conductance versus displacement histograms for 1-3 and L1-L3 collected in TCB. Histograms were created by aligning individual conductance traces at 0.5 G0 and then overlaying all traces to generate the 2D image. Molecular plateau lengths for the clusters are roughly twice as long as their ligand only counterparts.

We measured the conductance of 1 under an inert atmosphere of Ar gas to examine whether the presence of oxygen or water in the cluster solution impacts the measurement (Figure 3.7). We observe no change in the conductance of 1.

Figure 3.7. (a) Logarithmically-binned conductance histograms, and (b) and (c) 2D conductance versus displacement histograms for 1 collected in TCB under ambient conditions and under an inert atmosphere of Ar. The peak at ~10-3 G0 comes from uncoordinated L1. Histograms were created by aligning individual conductance traces at 0.5 G0 and then overlaying all traces to generate the 2D image.

3.13. Additional Details on the Tight Binding Model

We use a tight binding model to determine a Hamiltonian matrix and then use a non- equilibrium Green’s Function formalism in order to qualitatively model transmission through the molecular junctions. As described in the main text, we use an n-site model to represent the molecule, and couple only nearest neighbor sites. For the ligands, the model consists of 1, 2 and 3 sites (for L1-L3) of energy ε, with nearest neighbors coupled by δ. The cluster model consists of two n-length ligands with an additional site in between the ligands. This site has an energy E0 and is coupled to its nearest neighbors by τ. In all cases, the molecule is coupled to the leads by an imaginary, energy independent, self-energy term -iΓ/2. In order to compute molecular transmission functions, we turn to the non-equilibrium Green’s Function formalism. The retarded Green’s function for the molecular junction is defined as G(E) = [EI-H]-1, and transmission is then given as T(E)= Tr(ΓLGΓRG), which is computed numerically. Here, ΓL and ΓR are coupling matrices coupling the molecule to the left and right leads, respectively. Values for Γ, ε, and δ are chosen such that the ligand conductance values are similar to the experimental values.

When adding the cluster site, we explore a parameter space of -1.5 to -0.1 eV for τ and -

3.0 to -1.6 eV for E0 to see the impact that these values have on the decay constant for the two systems. This is demonstrated in Figure 3.4d, where it is clear that it is not possible to obtain significantly different conductance decay constant () values using this model. As mentioned in the text, we also show in Figure 3.8 that it is possible to obtain higher conductances for a given cluster as compared to its corresponding ligand.

Figure 3.8. 2D plot showing the conductance ratio of 1 to L1 (G1/GL1) as obtained from our tight binding model. We kept Γ, ε, and δ constant while varying E0 and τ. We find that, given certain parameters, we can wind up with a system in which the cluster is more conducting than its corresponding ligand.

3.14. Details on the DFT Calculations

All calculations were done using Jaguar (Schrodinger, Inc., New York, NY, 2014). All calculations were based on density functional theory (DFT), and used the B3LYP functional. The

6-31G** basis sets were used throughout, with the LACVP potentials/basis sets used for the heavy atoms.

For the strictly organic molecules (L1, L3, and L3) the geometries were fully optimized.

For the cluster-comntaining systems the geometries were optimized in the sense that we found

91 local energetic minima, however owing to the complicated chemical structures there are undoubtedly many similar, energetically essentially degenerate local minima in each case.

For each molecule we studied we have included the final total energy and the final optimized geometry. For the strictly organic systems we have also included diagrams indicating the atomic numbering; similar diagrams for the cluster-containing systems are too cluttered to be useful.

3.15. References

1. Choi, B.; Capozzi, B.; Ahn, S.; Turkiewicz, A.; Lovat, G.; Nuckolls, C.; Steigerwald, M. L.; Venkataraman, L.; Roy, X., Solvent-Dependent Conductance Decay Constants in Single Cluster Junctions. Chem. Sci. 2016, 7, 2701-2705. 2. Nitzan, A.; Ratner, M. A., Electron Transport in Molecular Wire Junctions. Science 2003, 300, 1384-1389. 3. Joachim, C.; Gimzewski, J. K.; Aviram, A., Electronics Using Hybrid-Molecular and Mono-Molecular Devices. Nature 2000, 408, 541-548. 4. Meisner, J. S.; Kamenetska, M.; Krikorian, M.; Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C., A Single-Molecule Potentiometer. Nano Lett. 2011, 11, 1575-1579. 5. Chen, F.; Li, X. L.; Hihath, J.; Huang, Z. F.; Tao, N. J., Effect of Anchoring Groups on Single-Molecule Conductance: Comparative Study of Thiol-, Amine-, and Carboxylic- Acid-Terminated Molecules. J. Am. Chem. Soc. 2006, 128, 15874-15881. 6. Li, X. L.; He, J.; Hihath, J.; Xu, B. Q.; Lindsay, S. M.; Tao, N. J., Conductance of Single Alkanedithiols: Conduction Mechanism and Effect of Molecule-Electrode Contacts. J. Am. Chem. Soc. 2006, 128, 2135-2141. 7. Solomon, G. C.; Andrews, D. Q.; Hansen, T.; Goldsmith, R. H.; Wasielewski, M. R.; Van Duyne, R. P.; Ratner, M. A., Understanding Quantum Interference in Coherent Molecular Conduction. J. Chem. Phys. 2008, 129, 054701. 8. Scullion, L. E.; Leary, E.; Higgins, S. J.; Nichols, R. J., Single-Molecule Conductance Determinations on HS(CH2)4O(CH2)4SH and HS(CH2)2O(CH2)2O(CH2)2SH, and Comparison with Alkanedithiols of the Same Length. J. Phys. Condens. Mat. 2012, 24, 164211. 9. Wang, C. S.; Batsanov, A. S.; Bryce, M. R.; Martin, S.; Nichols, R. J.; Higgins, S. J.; Garcia-Suarez, V. M.; Lambert, C. J., Oligoyne Single Molecule Wires. J. Am. Chem. Soc. 2009, 131, 15647-15654.

10. Martin, S.; Grace, I.; Bryce, M. R.; Wang, C. S.; Jitchati, R.; Batsanov, A. S.; Higgins, S. J.; Lambert, C. J.; Nichols, R. J., Identifying Diversity in Nanoscale Electrical Break Junctions. J. Am. Chem. Soc. 2010, 132, 9157-9164. 11. Yamada, R.; Kumazawa, H.; Noutoshi, T.; Tanaka, S.; Tada, H., Electrical Conductance of Oligothiophene Molecular Wires. Nano Lett. 2008, 8, 1237-1240. 12. Boardman, B. M.; Widawsky, J. R.; Park, Y. S.; Schenck, C. L.; Venkataraman, L.; Steigerwald, M. L.; Nuckolls, C., Conductance of Single Cobalt Chalcogenide Cluster Junctions. J. Am. Chem. Soc. 2011, 133, 8455-8457. 13. Yelin, T.; Vardimon, R.; Kuritz, N.; Korytar, R.; Bagrets, A.; Evers, F.; Kronik, L.; Tal, O., Atomically Wired Molecular Junctions: Connecting a Single Organic Molecule by Chains of Metal Atoms. Nano Lett. 2013, 13, 1956-1961. 14. Leary, E.; Van Zalinge, H.; Higgins, S. J.; Nichols, R. J.; de Biani, F. F.; Leoni, P.; Marchetti, L.; Zanello, P., A Molecular Wire Incorporating a Robust Hexanuclear Platinum Cluster. Phys. Chem. Chem. Phys. 2009, 11, 5198-5202. 15. Park, J.; Pasupathy, A. N.; Goldsmith, J. I.; Chang, C.; Yaish, Y.; Petta, J. R.; Rinkoski, M.; Sethna, J. P.; Abruna, H. D.; McEuen, P. L.; Ralph, D. C., Coulomb Blockade and the Kondo Effect in Single-Atom Transistors. Nature 2002, 417, 722-725. 16. Roy, X.; Schenck, C. L.; Ahn, S.; Lalancette, R. A.; Venkataraman, L.; Nuckolls, C.; Steigerwald, M. L., Quantum Soldering of Individual Quantum Dots. Angew. Chem. Int. Edit. 2012, 51, 12473-12476. 17. Stuczynski, S. M.; Kwon, Y. U.; Steigerwald, M. L., The Use of Phosphine Chalcogenides in the Preparation of Cobalt Chalcogenides. J. Organomet. Chem. 1993, 449, 167-172. 18. Crawford, N. R. M.; Hee, A. G.; Long, J. R., Cluster Synthesis via Ligand-Arrested Solid Growth: Triethylphosphine-Capped Fragments of Binary Metal Chalcogenides. J. Am. Chem. Soc. 2002, 124, 14842-14843. 19. Meisner, J. S.; Ahn, S.; Aradhya, S. V.; Krikorian, M.; Parameswaran, R.; Steigerwald, M.; Venkataraman, L.; Nuckolls, C., Importance of Direct Metal-π Coupling in Electronic Transport through Conjugated Single-Molecule Junctions. J. Am. Chem. Soc. 2012, 134, 20440-20445. 20. Park, Y. S.; Whalley, A. C.; Kamenetska, M.; Steigerwald, M. L.; Hybertsen, M. S.; Nuckolls, C.; Venkataraman, L., Contact Chemistry and Single-Molecule Conductance: A Comparison of Phosphines, Methyl Sulfides, and Amines. J. Am. Chem. Soc. 2007, 129, 15768-15679. 21. Arroyo, C. R.; Leary, E.; Castellanos-Gomez, A.; Rubio-Bollinger, G.; Gonzalez, M. T.; Agrait, N., Influence of Binding Groups on Molecular Junction Formation. J. Am. Chem. Soc. 2011, 133, 14313-14319. 22. Xu, B. Q.; Tao, N. J. J., Measurement of Single-Molecule Resistance by Repeated Formation of Molecular Junctions. Science 2003, 301, 1221-1223.

23. Gonzalez, M. T.; Wu, S. M.; Huber, R.; van der Molen, S. J.; Schonenberger, C.; Calame, M., Electrical Conductance of Molecular Junctions by a Robust Statistical Analysis. Nano Lett. 2006, 6, 2238-2242. 24. Fatemi, V.; Kamenetska, M.; Neaton, J. B.; Venkataraman, L., Environmental Control of Single-Molecule Junction Transport. Nano Lett. 2011, 11, 1988-1992. 25. Nakashima, S.; Takahashi, Y.; Kiguchi, M., Effect of the Environment on the Electrical Conductance of the Single Benzene-1,4-diamine Molecule Junction. Beilstein J. Nanotechnol. 2011, 2, 755-759. 26. Capozzi, B.; Chen, Q. S.; Darancet, P.; Kotiuga, M.; Buzzeo, M.; Neaton, J. B.; Nuckolls, C.; Venkataraman, L., Tunable Charge Transport in Single-Molecule Junctions via Electrolytic Gating. Nano Lett. 2014, 14, 1400-1404. 27. Meisner, J. S.; Ahn, S.; Aradhya, S. V.; Krikorian, M.; Parameswaran, R.; Steigerwald, M.; Venkataraman, L.; Nuckolls, C., Importance of Direct Metal-π Coupling in Electronic Transport Through Conjugated Single-Molecule Junctions. J. Am. Chem. Soc. 2012, 134, 20440-20445. 28. Simmons, J. G., Generalized Formula for Electric Tunnel Effect between Similar Electrodes Separated by a Thin Insulating Film. J. App. Phys. 1963, 34, 1793-1803. 29. Datta, S., Quantum Transport - Atom to Transistor. Cambridge University Press: 2005. 30. Datta, S., Electronic Transport in Mesoscopic Systems. Cambridge University Press: 1995. 31. Nitzan, A., Electron Transmission through Molecules and Molecular Interfaces. Ann. Rev. Phys. Chem. 2001, 52, 681-750. 32. Choi, S. H.; Kim, B.; Frisbie, C. D., Electrical Resistance of Long Conjugated Molecular Wires. Science 2008, 320, 1482-1486. 33. Park, Y. S.; Widawsky, J. R.; Kamenetska, M.; Steigerwald, M. L.; Hybertsen, M. S.; Nuckolls, C.; Venkataraman, L., Frustrated Rotations in Single-Molecule Junctions. J. Am. Chem. Soc. 2009, 131, 10820-10821. 34. Meisner, J. S.; Sedbrook, D. F.; Krikorian, M.; Chen, J.; Sattler, A.; Carnes, M. E.; Murray, C. B.; Steigerwald, M.; Nuckolls, C., Functionalizing Molecular Wires: A Tunable Class of α,ɷ-Diphenyl-μ,ν-dicyano-oligoenes. Chem. Sci. 2012, 3, 1007-1014. 35. Xue, C. H.; Luo, F. T., Efficient and Rapid Synthesis of Oligo(p-phenylenevinylene) via Iterative Coherent Approach. J. Org. Chem. 2003, 68, 4417-4421.

Chapter 4. Room Temperature Single-Cluster Current Blockade

4.1. Preface

This chapter is largely based on a manuscript entitled “Room-Temperature Current

Blockade in Atomically Defined Single-Cluster Junctions” by Giacomo Lovat, Bonnie Choi,

Daniel W. Paley, Michael L. Steigerwald, Latha Venkataraman, and Xavier Roy published in

Nature Nanotechnology.1 I synthesized all compounds. I characterized the molecular clusters using

SCXRD with essential input from Dr. Daniel W. Paley. Dr. Giacomo Lovat in Professor Latha

Venkataraman’s group performed all single-molecule conductance measurements.

4.2. Introduction

Fabricating nanoscopic devices capable of manipulating and processing single units of charge is an essential step towards creating functional devices where quantum effects dominate transport characteristics. The archetypal single-electron transistor comprises a small conducting or semiconducting island separated from two metallic reservoirs by insulating barriers.2-6 By enabling the transfer of a well-defined number of charge carriers between the island and the reservoirs, such a device may enable discrete single-electron operations.7-10

The transfer of a well-defined number of electrons between the island and the reservoirs is possible when random charge fluctuations due to thermal and quantum effects are suppressed. This effect imposes two constraints on the nanoscale system: (1) the energy required to add one electron to the island (i.e., the charging energy (e2/2C) where C is the capacitance of the island) must exceed the thermal energy kBT; and (2) the conductance of the tunnel junctions (i.e., the insulating barrier

2 between the island and the reservoirs) must be smaller than the conductance quantum G0 = 2e /h.

11 In single-electron circuits, quantum dots have emerged as ideal candidates since their small sizes 95 make the capacitive charging energy much larger than the thermal energy. For example, Coulomb blockade and Coulomb staircase effects of single-electron tunneling has been observed at room temperature for evacuated tunnel junctions where a scanning tunneling microscope tip is positioned above metallic or semiconducting nanoparticles. In these cases, the charging energy is on the order of 10-100 meV.2, 12-14 However, the intrinsic variations in size and shape of nanoparticles synthesized with conventional nanofabrication techniques result in poor reproducibility of their transport characteristics.

We describe a single-molecule junction comprising a redox-active, atomically precise cobalt chalcogenide cluster wired between two nanoscopic electrodes.15, 16 We observe current blockade at room temperature in thousands of single-cluster junctions. Below a threshold voltage, charge transfer across the junction is suppressed. The device is turned on when the temporary occupation of the core states by a transiting carrier is energetically enabled, resulting in a sequential tunneling process and an increase in current by a factor of ~600. We perform in situ and ex situ cyclic voltammetry as well as density functional theory calculations to unveil a two-step process mediated by an orbital localized on the core of the cluster in which charge carriers reside before tunneling to the collector reservoir. As the bias window of the junction is wide enough to include one of the cluster frontier orbitals, the current blockade is lifted and charge carriers can tunnel sequentially across the junction.

We also describe an analogous single-cluster junction formed with Mo6S8L6 in which we can access different charge states at room temperature. In the case of Mo6S8L6, we can access both the negatively charged and positively charged species by varying the tip bias across the cluster junction. The asymmetric conductance behaviors in the Mo-based cluster system further support our two-step incoherent tunneling mechanism. The high structural tunability within this class of

96 molecular clusters will enable the design of molecular-scale electronic systems with multiple functionalities.

4.3. Synthetic Design

The single-cluster junctions consist of an octahedral core of cobalt atoms surrounded by a cube of chalcogen (sulfur or selenium). This cluster-type was introduced in Chapter 1 and subsequently used to study the solvent effect on the conductance in single-cluster junction in

Chapter 3. Here, we wire an organic bifunctional ligand (L = diethyl-4- thiomethylphenylphosphine) between the cluster core and the Au electrodes: the phosphorus end of the ligand attaches to the core via Co-P bond, and the aurophilic thiomethyl group connects to the Au electrodes via Au-S bond. Figure 4.1a shows the core structure of Co6S8L6, as determined by SCXRD. These compounds are stable in multiple charged states and we structurally characterize the neutral, monocationic, and dicationic species for each core composition (See

Table 4.1 for structural details). SCXRD data show only small variations in inter-atomic distances among the different charge states, indicating that reorganization energies upon changing the charge state are small. Electrochemical characterization and DFT calculations of similar clusters indicate that their energy-frontier orbitals are fully localized on the inorganic core and essentially decoupled from the ligands, making this system an ideal playground to explore the effects of charge quantization on single-molecule transport.15, 16 The atomic precision of these clusters overcomes the main shortcoming of devices built from quantum dots,2, 4, 14 which have intrinsic size, shape and composition variations resulting in poor reproducibility of transport characteristics.

Figure 4.1. (a) Top: Molecular structure of the [Co6S8] core as determined by SCXRD. Cobalt, blue; sulfur, yellow. The [Co6E8] core is a magnetic singlet (S = 0) in the neutral state, a doublet (S = ½ ) in the 1+ state and a triplet (S = 1) in the 2+ state. Bottom: Structure of the molecular connector L used to wire the cluster into a junction. (b) Schematic of the STM-BJ measurement in an ionic environment. The Au tip is coated with an insulating wax before immersion in a solution of the target clusters in PC. As a potential difference is established between the tip and the substrate, dissolved ions accumulate more densely on the small tip area left uncovered by the wax.

4.4. STM-BJ Measurements

We measure, reliably and reproducibly, the conductance and current-voltage (I-V) characteristics of single cluster junctions using the STM-BJ technique with Au metal electrodes.17

The clusters are wired into junctions using bifunctional ligands that bind specifically and directionally to the core and the electrode.15, 16 The phosphine group on one terminus of the ligand attaches to the core via a Co-P coordination bond, and the aurophilic thiomethyl group on the other terminus of L connects to the Au electrodes via an Au-S bond. Measurements are carried out in propylene carbonate (PC), a polar solvent that can dissolve ionic compounds to create an ionic environment around the junction.18 The Au tip is coated with an insulating layer to suppress background ionic current.19 Figure 4.1b illustrates the asymmetric double-layer that builds up in the junction as a result of the small area of the coated tip (surface area ~1 µm2) relative to the large- area, uncoated substrate (~1 cm2). This asymmetrical ionic environment results in the energy levels

98 of the cluster core being pinned to the substrate chemical potential, thereby allowing the tip chemical potential to be modulated relative to the molecular orbitals by the applied bias.18, 20 In contrast to previous electrochemical gating systems that require both reference and counter electrodes, the junctions can be gated using only two electrodes.

4.5. Histogram Analysis and I-V Plots

Figure 4.2. Logarithmically binned 1D conductance histograms for Co6S8L6 at (a) negative and (b) positive tip bias, respectively. Each histogram is constructed from 1,000 traces measured at a fixed tip bias using 100 bins/decade. (c) 2D conductance-displacement histogram for Co6S8L6 constructed from 1,000 traces measured at -0.72 V tip bias. The histogram is created by aligning all the traces at the point where the conductance crosses 0.5 G0, and then overlaying them using 100 bins/decade on the conductance axis and 625 bins/nm on the displacement axis.

Conductance results for the Co6S8L6 system are presented first. The experiment typically begins by introducing the monocationic [Co6S8L6][BF4], but we find that the charge state of the cluster trapped between the tip and the substrate is dictated by the potential across the junction.

Figure 4.2a,b shows logarithmically binned one-dimensional (1D) conductance histograms at

21 different tip biases. Clear peaks are visible at integer multiples of the conductance quantum (G0

= 2e2/h = 77.5 µS), associated with the breaking of an Au-Au junction, and at a bias-dependent value for the molecular cluster junction. Specifically, the molecular conductance peak shifts to higher values as the magnitude of the bias is increased. Figure 4.2c shows a two-dimensional (2D) histogram of the conductance of Co6S8L6. This allows a direct determination of the extent of the 99 junction elongation, which has been shown to correlate with the molecular length.22 The junction elongation for this cluster is ~1 nm. By accounting for the Au relaxation gap (~0.6-0.8 nm upon breaking of Au contacts23), we find that this elongation correlates well with the molecular length of a cluster with ligands fully extended, as measured from SCXRD (~1.8 nm). We conclude that we are indeed probing transport through Au-L-Co6S8-L-Au junctions. If the cluster were to decompose during the measurements, we would observe a clear peak due to junctions formed with

L alone (Figure 4.3).16

Figure 4.3. 1D conductance histograms of ligand L measured at different tip biases in PC using STM-BJ technique. The inset presents the current as a function of tip bias and shows a clear linear trend.

100

Figure 4.4. (a) Semi-logarithmic plot of the current versus tip bias for Co6S8L6 (red trace) and Co6Se8L6 (blue trace) measured in PC. The inset shows the current blockade region at low tip bias on a linear scale. (b) Lower panel: in situ CV recorded for a solution of Co6S8L6 in PC with 0.1 M TBAPF6 supporting electrolyte using a sweep rate of 100 mV/s. The voltammogram shows two current steps at reduction (V < 0) and oxidation (V > 0) tip bias separated by ~1 V. Upper panel: CV recorded ex situ in a standard 3-electrode electrochemical cell using a glassy carbon working electrode in dichloromethane solution of 0.1 M TBAPF6 supporting electrolyte and using a sweep rate of 100 mV/s. The data is calibrated using the ferrocene/ferrocenium redox couple. (c) Sample I-V trace for Co6S8L6. (d) 2D I-V histogram for Co6S8L6. The current axis is logarithmically binned. The superimposed yellow curve represents an average of the current over all traces per voltage bin. Note that the data between -0.2 V and 0.2 V includes some additional contributions due to capacitive currents as the voltage is swept at a rate of 50 V/s. Pink circles overlaid on the plot are I-V values extracted from the conductance measurements in Figure 4.4a. (e) Normalized differential conductance, dI/dV, extracted from the average I-V curve (red trace) and from the conductance measurements (pink dots).

We determine the most probable junction current for each tip bias from the conductance histograms and plot these as a function of tip bias in Figure 4.4a (red trace). The current through the junction is very low over a bias window that extends from approximately +360 mV to -450 mV. Beyond these voltage thresholds, the current through the junction increases sharply, and at a bias of -810 mV the current is a factor of ~560 higher than that at -180 mV. To understand these

101 results, we perform in situ cyclic voltammetry (CV) measurements on Co6S8L6 using the STM 2- electrode setup as a nanoscale electrochemical cell. The voltammogram shown in Figure 4.4b

(lower panel) reveals three distinct charge states separated by two reversible redox peaks, consistent with the electrochemical behavior of Co6S8L6 measured ex situ using standard CV technique with macroscopic electrodes. At low tip bias, the species at the electrode surface is

+ + [Co6S8L6] . A large negative applied voltage reduces [Co6S8L6] to Co6S8L6; conversely, a large

+ 2+ positive applied voltage oxidizes [Co6S8L6] to [Co6S8L6] . The potential of the redox peaks in the in situ voltammogram matches well with the threshold voltages observed in the single-cluster junction measurements, suggesting that the increase in the junction current occurs when the charge state on the cluster is altered. Independently, we perform traditional ionically gated STM-BJ measurements in which a third electrode is used to apply a gate voltage as detailed previously.24

By varying the gate voltage, the charge state of the cluster can be changed and thus the conductance can be turned on or off (Figure 4.5). These results are consistent with the in situ CV measurements.

Figure 4.5. (a) 1D conductance histogram of Co6S8L6 measured at different gate voltages using a 3-electrode setup. Depending on the gate voltage, conductance is turned on or off (e.g., trace at -1.0 V when conductance is turned on and at 0.0 V when conductance is turned off). (b) 2D conductance histogram of Co6S8L6 at a gate voltage of -1.0 V.

102

Figure 4.6. 1D conductance histograms of Co6Se8L6 measured at (a) negative and (b) tip biases, respectively, in PC using STM-BJ technique. (c) 2D histogram measured at -450 mV.

We turn to the cobalt selenide system, Co6Se8L6, to further understand the interplay between the cluster energy levels and the junction current. The tip bias dependence of the junction current extracted from 1D histograms (Figure 4.6) is shown in Figure 4.4a (blue trace). The shape of the current versus tip bias curve for Co6Se8L6 is very similar to that for Co6S8L6; there is a low

-5 current region (conductance ~10 G0) at low tip bias and then a sharp linear increase in current beyond a threshold. This behavior is analogous to that observed in quantum dots at low temperature,14, 25 and it is commonly attributed to Coulomb blockade. The tunneling conductance across the Co6Se8L6 junction in the blockade regime is like that for Co6S8L6 and the on/off current ratio is over 600 for Co6Se8L6. A clear difference between the two systems is that the current blockade for Co6Se8L6 is lifted at a smaller positive voltage than for Co6S8L6. Voltammetric data explain this behavior: the CV of Co6Se8L6 (Figure 4.7) shows that the 1+/2+ redox couple is shifted to lower voltage when compared to Co6S8L6.

103

Figure 4.7. CV of Co6Se8L6 in 0.1 M TBAPF6 in dichloromethane with a 100 mV/s scan rate.

To corroborate these findings, we measured current-voltage (I-V) curves for over 3000 single-cluster Co6S8L6 junctions over a range of -1.3 to +1.3 V. These measurements allow us to probe a wider voltage range than that accessible through the conductance histogram method. A sample I-V trace is shown in Figure 4.4c (additional individual traces are included in Figure 4.8).

All I-V curves were overlaid on a logarithmic scale to generate a 2D I-V histogram (Figure 4.4d).

An average I-V curve (yellow trace) is obtained from this map and used to determine an average normalized differential conductance, dI/dV (Figure 4.4e). The average I-V curve shows a crossover from a low-conductance to a high-conductance regime when we sweep the voltage for both bias polarities. The boundaries of the blockade regime are clearly marked by two peaks at -

0.6 and +0.7 V in the dI/dV trace. The I-V values extracted from the conductance histograms

(shown in Figure 4.4a and included as pink circles in Figure 4.4d,e) are in reasonable agreement with the I-V curve. The difference is ascribed to the high scan rate (60 V/s) employed in the I-V experiment,26 which will significantly alter the double-layer density around the tip. Similar behaviors are observed for the Co6Se8L6 junctions (Figure 4.9).

104

Figure 4.8. Sample traces of I-V measurement for Co6S8L6.

Figure 4.9. (a) I-V 2D histogram for Co6Se8L6. The current axis is logarithmically binned. The superimposed white curve represents an average of the current over all traces per voltage bin. Note that the data between -0.2 V and 0.2 V includes some additional contributions due to capacitive currents as the voltage is swept at a rate of 50 V/s. (b) Normalized differential conductance, dI/dV, extracted from the average I-V curve.

105

4.6. Two-Step Incoherent Transport Mechanism

Our results can be understood in terms of a two-step incoherent transport mechanism whereby carriers first tunnel through the connector from one electrode to a level localized on the inorganic core and then into the other electrode. We have four observations that strongly support this sequential tunneling process. First, we can rule out a simple off-resonance/in-resonance mechanism whereby a resonant conductance level is gradually approached as the tip bias is varied.

Indeed, the dramatic increase in current observed for our system is significantly higher than what has been measured in single-molecule diodes.18, 27, 28 Considering that the molecular orbitals on the cluster core are at best weakly coupled to the ligand levels15, 16 (see the discussion of DFT calculations in Section 4.7) and that two peaks are observed in the dI/dV plots (which would imply that two different ligand levels come in-resonance), our results preclude a coherent resonant transport mechanism. Second, the CV measurements (both in situ and ex situ) clearly show that different charge states are accessible within the bias window of the junctions. Third, we can reject a coherent transport mechanism that is simply gated as the cluster charge state is altered.29 Were coherent transport (with gating effects) occurring, we would expect that increasing the cluster

+ 2+ charge state, from Co6S8L6 to [Co6S8L6] to [Co6S8L6] , would produce a monotonic change in

+ conductance. Instead, Figure 4.2a,b shows that the conductance is low for [Co6S8L6] and

2+ increases at both negative and positive tip biases for Co6S8L6 and [Co6S8L6] , respectively. Finally, we can rule out that the sharp increase in the current comes from “outer sphere” electron transfer between the electrodes and the cluster within the junction or clusters around the junction (i.e., electrochemical current). Measured at a bias beyond the threshold voltage, the 2D histogram in

Figure 4.2c shows that the current does not change with distance until the junction breaks. At the

-6 breaking point, the conductance suddenly drops to the instrument noise floor (~10 G0), indicating

106 that the cluster needs to be connected to the electrodes in order to conduct and that the electrochemical current is at least two orders of magnitude smaller than the junction current when a coated tip is used.

Figure 4.10. (a) HOMO and HOMO-13 orbitals of model compound (PMe3)4Co6S8L2 with energies relative to vacuum, as calculated from DFT. The energy level diagram shows several orbitals energetically close to the HOMO. These orbitals are quite similar to the HOMO we present explicitly. In Co6S8L6, there is a set of essentially identical, L-based S-centered orbitals; we show just one for simplicity. Note that the HOMO-LUMO gap is around 3 eV. (b) Schematic of the proposed sequential tunneling transport mechanism. Green and red markers represent cluster levels available and unavailable for sequential tunneling, respectively. Solid and hollow markers represent occupied and unoccupied cluster levels, respectively. Bottom panel illustrates the absolute current (blue) and dI/dV (green) that would correspond to each charge state of the cluster as a function of tip bias.

4.7. DFT Calculations and Sequential Tunneling

Using DFT, we model Co6S8L6 with a simplified cluster, (PMe3)4Co6S8L2, and show that the cluster core energy levels are weakly coupled to the electrodes. Because the thiomethyl- terminated ligand L is known to conduct through its highest occupied molecular orbital (HOMO),30 we examine the occupied levels of the [Co6S8] system. The HOMO is localized on the inorganic core and the highest occupied orbital that has weight on L is HOMO-13, which is essentially a combination of the thiomethyl S pπ lone pair and the C π-bonds. Figure 4.10a shows the HOMO

107 and HOMO-13 orbitals of (PMe3)4Co6S8L2 and their relative position on the energy scale. Our

DFT calculations clearly show that the [Co6S8]-based HOMO and the L-based HOMO-13 are at best weakly coupled, and that orbitals of energy higher than HOMO-13 are all localized on the inorganic core, sequestered from the environment by the ligand shell. These calculations support our hypothesis that a coherent transport through a single orbital cannot explain our data.

Additionally, any redox process will involve orbitals that are localized on the cluster core. From these observations, we propose a core-centered sequential tunneling process that is mediated by the ligand.

Figure 4.10b is a schematic illustrating this sequential tunneling mechanism that leads to a current blockade. At small bias, the cluster in the junction is in the 1+ charged state and it has no cluster-based energy-frontier levels within the bias window. Consequently, transport occurs via a tunneling mechanism, likely through a single orbital that has weight on the L-based S and the core; this is the blockade regime. By increasing the tip bias in the negative direction, the bias window is opened to include an unoccupied level (0/1+ redox) and sequential tunneling can begin. The blockade is lifted and a sharp increase in current is observed. Similarly, as the tip bias is increased in the positive direction, the bias window is opened to include an occupied level (1+/2+ redox) and sequential tunneling is allowed. By extending the bias window further, we expect an additional steep rise in current as an additional occupied level (2+/3+ redox) is accessed. However, this third process is currently inaccessible with our experimental constraints.

4.8. Current Blockade in Mo-Based Single-Cluster Junction

Similar to the Co-based single-cluster junctions, we found that Mo-based single-cluster junction exhibits single-electron tunneling albeit with different accessible charge states. Figure

108

4.11 shows the molecular structure of Mo6S8L6 used to form the single-cluster junction between two Au electrodes. The synthesis of Mo6S8L6, entails a quantitative conversion from pyridine-

31 bound Mo6S8 to L-bound Mo6S8 in tetrahydrofuran when the mixture of Mo6S8(py)6 and L is heated to reflux temperature. As the ligand exchange proceeds, Mo6S8(py)6, which is insoluble in tetrahydrofuran, is converted to Mo6S8L6 that dissolves readily. The reaction mixture is filtered and the solvent in removed in vacuo to yield Mo6S8L6. This 20-valence electron molecular cluster can be characterized by NMR. Sharp peaks in the 1H NMR and 31P NMR spectra suggest an

31, 32 absence of unpaired electrons in the neutral Mo6S8L6 molecular cluster.

Figure 4.11. Molecular structure of Mo6S8L6 as determined by SCXRD. Molybdenum, magenta; sulfur, yellow; phosphorus, orange; carbon, black.

We employ the same experimental conditions as in the Co-based system. Figure 4.12a,b shows the logarithmically binned 1D conductance histograms at positive and negative tip biases, respectively, and Figure 4.12c shows a 2D histogram of the conductance of Mo6S8L6 at +400 mV, which shows the extent of junction elongation (~1 nm). We note that the one-electron reduced

33 Mo6S8 molecular cluster is sensitive to oxidation and is most likely contributing to the poorer conductance stability in the STM-BJ measurements.

109

We next determined the most probable junction current for each tip bias from the conductance histograms and plotted these as a function of tip bias (Figure 4.13a). The current across the junction is very low over a bias window that extends from +0.2 to -0.7 V. This range is the current blockade regime. The current through the junction increases beyond these threshold

0 voltages. At low tip bias, the species at the electrode surface is [Mo6S8L6] . A large negative tip

0 1- 0 bias reduces [Mo6S8L6] to [Mo6S8L6] while a large positive tip bias oxidizes [Mo6S8L6] to

1+ [Mo6S8L6] . Ex situ CV performed with macroscopic electrodes in a dichloromethane solution of

Mo6S8L6, shows reversible one-electron reduction and oxidation waves at -1.37 V and -0.13 V (V vs. Fc/Fc+ redox couple), respectively (Figure 4.13b). The spacing between the two redox events

34 (1.24 V) in Mo6S8L6 agrees well with that of known Mo6S8(PEt3)6 (1.27 eV).

Figure 4.12. Logarithmically binned 1D conductance histograms for Mo6S8L6 at (a) positive and (b) negative tip bias, respectively. Each histogram is constructed from at least 1,000 traces measured at a fixed tip bias using 100 bins dec- 1 . (c) 2D conductance-displacement histogram for Mo6S8L6 constructed from 1,000 traces measured at +400 mV tip bias. The histogram is made by aligning all the traces at the point where the conductance crosses 0.5 G0 and then overlaying them using 100 bins dec-1 on the conductance axis and 635 bins nm-1 on the displacement axis.

110

Figure 4.13. (a) Current versus tip bias for Mo6S8L6 measured in PC. (b) CV of Mo6S8L6 recorded in dichloromethane with 0.1 M TBAPF6 supporting electrolyte, using a standard three-electrode electrochemical cell with a glassy carbon working electrode. Sweep rate: 100 mV s-1. The data is calibrated using the ferrocene/ferrocenium redox couple. (c) Two-dimensional I-V histogram. The current is logarithmically binned, and the tip bias axis is linear. The superimposed yellow curve is an average of the current over all traces per voltage bin. (d) Normalized differential conductance dI/dV extracted from the average I-V curve from (c).

In addition to measuring the conductance of Mo6S8L6 at various tip biases, we also obtained

I-V curves for over 1000 cluster junctions over the range -0.8 to +0.8 V. Figure 4.3c shows a 2D

I-V histogram of all overlaid I-V curves on a logarithmic scale. We note that the data between -

0.2 and +0.2 V include some additional contributions due to the capacitive currents. We use the average I-V curve (yellow trace in Figure 4.13c) to generate the normalized differential conductance, dI/dV, plot (Figure 4.13d). Based on the normalized dI/dV trace, the current blockade is lifted at the onset of +0.2 and -0.6 V.

111

Analogous to the Co6E8L6 study, the increase in junction current occurs when the charge state on the cluster is modified. In the case of Mo6S8L6, we can access both the negatively charged and positively charged species by varying the tip bias across the cluster junction. Figure 4.13d shows a clear asymmetry about 0 V in contrast to what was observed for Co6E8L6 systems. This observation augments our previous claim that the transport across these single-cluster junctions is not simply a resonant transfer through a single orbital on the molecular system. The shallowest resonance in the dI/dV spectrum corresponds to the energy level closest to the Fermi energy of the leads, and the spacing of the resonances reflects the location of the molecular energy levels relative to the Fermi energy. For the Mo6S8L6 system, the energy levels are most likely asymmetrically spaced from the Fermi of the leads, thus showing a clear asymmetry about 0 V.

4.9. Conclusions

We have studied single-electron transport in a single redox-active molecular cluster wired between two nanoscopic electrodes. The atomic precision, the small size of the core and the weak electronic coupling to the organic connectors allow us to observe current blockade at room temperature in thousands of single-cluster junctions. The dominant transport mechanism in these devices is likely a two-step process akin to electron tunneling in a double-barrier tunnel junction.

The ligands act as insulating barriers interposed between the electrodes and the molecular levels of the inorganic core. Current flows through the junction only when the temporary occupation of the core states by a transiting carrier, i.e. the “charging” of the cluster, is energetically permitted.

The on/off ratio of these transistors is much larger than that of typical electrochemically-gated organic molecules,7, 27, 35 an effect attributed in part to the small electron transfer reorganization energies of the system.36 By taking advantage of the rich chemistry of molecular clusters,37 these

112 results provide an optimistic path to designing room-temperature molecular-scale transistors with multiple states and functionalities.

4.10. General Synthesis Information

All reactions and sample preparations were carried out under inert atmosphere using standard Schlenk techniques or in a nitrogen-filled glovebox unless otherwise noted.

Chlorodiethylphosphine was purchased from Acros Organics. Dicobalt octacarbonyl and selenium powder were purchased from Strem Chemicals. All other reagents and solvents were purchased from Sigma Aldrich. Dry and deoxygenated solvents were prepared by elution through a dual- column solvent system (MBraun SPS). Diethyl-4-(methylthio)phenyl phosphine (L), Co6Se8L6,

16, 31 and Mo6S8(py)6 were prepared according to published protocols.

4.11. Synthetic Procedures and Characterization of Compounds

Co6S8L6

The Co-based molecular cluster was prepared using a modified procedure for the synthesis

38 of [Co6S8(PEt3)6][BPh4] by Cecconi et al. The ligand L (1.00 g, 4.71 mmol) was added to a degassed solution of Co(BF4)2•6H2O (530 mg, 1.57 mmol) in acetone/ethanol (1:1; 15 mL:15 mL).

The mixture turned from pink to green-brown. Hydrogen sulfide was bubbled through the solution for 10 mins at room temperature. During that time, the reaction turned dark brown. The mixture was stirred for 16 h under nitrogen. The black precipitate that formed was collected on a glass frit, washed with a small amount of acetone/ethanol and dried in vacuo. 1H NMR shows that the solid contains only Co6S8L6. The product was recrystallized by slow evaporation of a concentrated dichloromethane solution under inert atmosphere. Yield: 25 mg (5 %)

113

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = 0.87 (6H, m), 1.95 (4H, m), 2.49 (3H, s),

7.14 (2H, d), 7.30 (2H, t).

[Co6S8L6][BF4]

When we expose the acetone/ethanol filtrate from the Co6S8L6 synthesis to air for several days, black block-like crystals of [Co6S8L6][BF4] form in the solution. The product was recrystallized by diffusing ethanol into a concentrated solution of [Co6S8L6][BF4] in dichloromethane. The crystals were collected, rinsed with ethanol and dried in vacuo. Yield: 180 mg (35 %).

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = -2.68 (4H, br), -0.08 (6H, br s), 2.48 (3H, s), 6.98 (2H, d), 7.33 (2H, m).

[Co6S8L6][BF4]2

A solution of AgBF4 (5 mg, 24 µmol) in 0.2 mL of acetonitrile was added dropwise to a solution of Co6S8L6 (21 mg, 11 µmol) in 3 mL of dichloromethane. The vial was wrapped in foil to protect the reaction from light and stirred 16 h. The solvent was removed in vacuo and 5 mL of dichloromethane was added to the crude product. The mixture was filtered through a 0.2 µm syringe filter and the solvent was removed in vacuo. The solid was dissolved in 0.5 mL of acetonitrile and the resulting dark brown solution was layered over toluene (10 mL). Dark brown crystals of [Co6S8L6][BF4]2 were obtained after 5 days at -35 °C. The crystals were collected, rinsed with toluene and hexanes, and dried in vacuo. Yield: 16 mg (70 %).

1 H NMR (400 MHz, [CD3CN], 298 K): δ = -4.47 (4H, br), -1.28 (6H, br s), 2.34 (3H, s), 6.78 (2H, d), 8.02 (2H, m).

114

[Co6Se8L6][BF4]

A solution of ferrocenium tetrafluoroborate (9 mg, 44 µmol) in 4 mL of acetonitrile was

1 added dropwise to a solution of Co6Se8L6 (72 mg, 44 µmol) in 4 mL of tetrahydrofuran. The reaction was stirred 16 h. The solvent was removed in vacuo and the solid was rinsed with toluene.

The product was dissolved in 2 mL of acetonitrile and the solution was filtered through a 0.2 µm syringe filter. The mixture was concentrated in vacuo and layered over toluene (5 mL). Dark brown crystals of [Co6Se8L6][BF4] were obtained after 5 days at -35 °C. The crystals were collected, rinsed with toluene and hexanes, and dried in vacuo. Yield: 30 mg (40 %)

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = -1.44 (4H, br), 0.00 (6H, br s), 2.51 (3H, s), 6.90 (2H, d), 7.16 (2H, m).

[Co6Se8L6][BF4]2

A solution of ferrocenium tetrafluoroborate (15 mg, 53 µmol) in 4 mL of acetonitrile was

1 added dropwise to a solution of Co6Se8L6 (60 mg, 26 µmol) in 4 mL of tetrahydrofuran. The reaction was stirred 16 h. The solvent was removed in vacuo and the solid was rinsed with toluene.

The product was dissolved in 2 mL of acetonitrile and the resulting solution was filtered through a 0.2 µm syringe filter. The mixture was concentrated in vacuo and layered over toluene (5 mL).

Dark brown crystals of [Co6Se8L6][BF4]2 were obtained after 2 days. The crystals were collected, rinsed with toluene and hexanes, and dried in vacuo. Yield: 32 mg (50 %)

1 H NMR (400 MHz, [d2-dichloromethane], 298 K): δ = -5.26 (4H, br), -0.76 (6H, br s), 2.43 (3H, s), 6.58 (2H, d), 7.74 (2H, m).

115

Mo6S8L6

The ligand L (220 mg, 1.05 mmol) was added to a suspension of Mo6S8(py)6 (210 mg, 0.16 mmol) in 5 mL of tetrahydrofuran, and the mixture was heated at reflux for 16 h. The mixture was cooled and filtered through a 0.2 μm PTFE syringe filter. The brown filtrate was concentrated in vacuo. Slow diethyl ether vapor diffusion yielded dark brown crystals. The supernatant solution was decanted and the recovered dark orange crystals were dried in vacuo. Yield: 290 mg (86 %)

1 H NMR (400 MHz, d6-benzene, 298 K): δ = 1.17 (6H, m), 1.97 (3H, s), 2.21 (4H, m), 7.15 (2H, d), 7.78 (2H, t)

31 P NMR (162 MHz, d6-benzene, 298K): δ = 21 (s)

4.12. Instrumentation

NMR

All 1H NMR and 31P NMR were recorded on a Bruker DRX400 (400 MHz) spectrometer.

X-ray Diffraction

Single crystal X-ray diffraction data were collected on an Agilent SuperNova diffractometer using mirror-monochromated Cu Kα or Mo Kα radiation. Data collection, integration, scaling (ABSPACK) and absorption correction (face-indexed Gaussian integration39 or numeric analytical methods40) were performed in CrysAlisPro.41 Structure solution was performed using ShelXT.42 Subsequent refinement was performed by full-matrix least-squares on

F2 in ShelXL.42 Olex243 was used for viewing and to prepare CIF files. Many disordered molecules

(e.g., BF4 anions) were modeled as rigid fragments from the Idealized Molecular Geometry

116

Library.44 ORTEP graphics were prepared in CrystalMaker.45 Details of crystallographic data and parameters for data collection and refinement are in Table 4.1.

Crystals were mounted on MiTeGen mounts with the aid of Parabar oil and cooled to 100

K on the diffractometer for screening and data collection. A minimum of 1 hemisphere of data to

0.8 Å resolution was collected for all compounds.

Ex situ Cyclic Voltammetry

Electrochemistry was carried out in dichloromethane or acetonitrile with tetra-n- butylammonium hexafluorophosphate (TBAPF6) supporting electrolyte (0.1 M) using a Princeton

Applied Research PARSTAT 2263 potentiostat inside an argon-filled glovebox. Measurements were carried out with a glassy carbon working electrode, a platinum counter electrode, and a silver pseudo-reference electrode. Measurements were calibrated using the ferrocene/ferrocenium redox couple.

4.13. Structural Determination

Structure solution and space group assignment were performed in ShelXT with no difficulty. In the final refinements, non-H atoms were refined anisotropically with no restraints unless noted; C-H hydrogens were placed in calculated positions and refined with riding isotropic

ADPs and coordinates. The non-routine details of the refinements are given below:

Co6S8L6

One C6H4SMe group was disordered over two positions, which were located with the aid of rigid phenyl fragments (AFIX 66) and subsequently refined with SAME and RIGU restraints.

All other details of the refinement were routine.

117

[Co6S8L6][BF4] and [Co6Se8L6][BF4]

These structures are isostructural and were solved and refined routinely in P-1.

[Co6S8L6][BF4]2

There are two independent clusters, one with 1/6 molecule in the asymmetric unit and the other with 1/3 molecule in the asymmetric unit. These were each refined routinely; one disordered ethyl group on the phosphine was modeled in two independent positions with SAME and RIGU restraints to stabilize its geometry and ADPs.

There are three independent BF4 molecules, all disordered over different symmetry elements. One BF4 sits on a twofold axis (Wyckoff site f) with a whole molecule in the asymmetric unit; this was placed in PART -1 with occupancy ½ . Another BF4 is on a threefold axis (Wyckoff site d) with 1/3 molecule in the asymmetric unit; this molecule is additionally disordered in a 90:10 ratio by a non-crystallographic inversion center. The final BF4 is on Wyckoff site a (site symmetry

32) and fulfills the symmetry of the threefold axis but not the perpendicular twofold axis. The necessary special position constraints were set up by hand: for the B atom and the F sitting on the threefold axis, x = y = 0; U11 = U22; U23 = U13 = 0; U12 = ½ * U11.

All 1,2 B-F distances and 1,3 F-F distances were made equivalent with a floating DFIX restraint. The ADPs of two of the disordered BF4 anions were restrained with RIGU. ADPs of overlapping atoms were stabilized with a short-range SIMU restraint.

118

[Co6Se8L6][BF4]2

This crystal is isostructural with [Co6S8L6][BF4]2 and the coordinates from the previous solution were used directly instead of solving de novo. All details of the refinement were identical, except that the disorders of the L ethyl group and the BF4 on Wyckoff site d were not observed.

Mo6S8L6

This structure was solved and refined routinely in P-1.

119

Table 4.1. Selected crystallographic data for Co- and Mo-based single-cluster junction. Compound Co6S8L6 [Co6S8L6][BF4] [Co6S8L6][BF4]2 Mo6S8L6

Formula C66H102Co6S14P6 C66H102Co6S14P6BF4 C66H102Co6S14P6B2F8 C66H102Mo6S14P6 MW 1883.71 1970.52 2057.33 2105.77 Space group I2/a P-1 P-3c1 P-1 a (Å ) 21.3182(2) 12.9224(3) 21.0803(3) 12.4685(4) b (Å ) 16.8028(1) 13.9345(3) 21.0803(3) 14.3077(5) c (Å ) 23.1835(3) 14.8239(3) 33.4515(5) 14.7115(6) α (°) 90 114.2610(18) 90 102.643(3) β (°) 103.849(1) 94.1899(17) 90 112.344(3) γ (°) 90 114.823(2) 120 109.881(3) V (Å 3) 8063.05(14) 2109.23(9) 12873.6(4) 2091.89(14) Z 4 1 6 1 -3 ρcalc (g cm ) 1.552 1.551 1.592 1.672 T (K) 100 100 100 100 λ (Å) 1.54184 1.54184 1.54184 0.71073

2θmin, 2θmax 6.564, 146.002 6.85, 145.968 7.168, 146.452 6.972, 58.650 Nref 75385 68304 164775 44853 R(int), R(σ) 0.0608, 0.0318 0.0554, 0.0265 0.1668, 0.0516 0.0513, 0.0508 μ(mm-1) 14.251 13.716 13.578 1.373

Tmax, Tmin 0.981, 0.970 0.759, 0.493 0.963, 0.919 0.881, 0.957 Data 8036 8409 8592 10135 Restraints 197 0 115 0 Parameters 498 466 535 424

R1(obs) 0.0291 0.0391 0.0614 0.0326

wR2(all) 0.0666 0.1050 0.1693 0.0582 S 1.024 1.026 1.006 1.053 Peak, hole 0.53, -0.46 0.95, -0.94 1.03, -0.83 0.64, -0.52 (e- Å -3) CCDC # 1557952 1557951 1557954

120

4.14. STM-BJ Experimental Details

We measured the conductance of single-molecule junctions formed with two gold electrodes using a home-built modified Scanning Tunneling Microscope (STM). We used 0.25 mm diameter cut gold wire (99.95%, Alfa Aesar) as the STM tips and 100 nm gold-coated

(99.999%, Alfa Aesar) steel pucks as substrates. A commercially available single-axis piezoelectric positioner (Nano-P15, Mad City Labs) was used to achieve sub-angstrom level control of the tip-substrate distance. The STM was controlled using a custom written program in

IgorPro (Wavemetrics, Inc.) and operated in ambient conditions at room temperature. The gold substrates were cleaned using a UV/ozone cleaning lamp for 20 minutes prior to use. For each measurement, 1000 traces were first collected prior to adding molecular solutions to check the cleanliness of the gold surface. Solutions of the target molecules at ~10 μM concentration in propylene carbonate (Alfa Aesar, 99% purity) were added to the substrate for STM break-junction measurements at applied biases ranging from +/-0.1 to +/-1 V. After the formation of each Au-Au junction, the piezoelectric motor moved the substrate at a speed of 20 nm/s to break the junction.

The current and voltage data were acquired at 40 kHz. For each of the compounds, Co6Se8L6 and

Co6S8L6, we collected 1,000 traces at each bias to generate 1D and 2D conductance histograms without data selection.

For the three-electrode measurements, a Pt wire was inserted into the PC solution (in addition to the tip and substrate) and the voltage at the Pt wire was controlled using a separate output channel of the data acquisition card, thereby working as a gate electrode.10 Measurements were done with an added electrolyte (100 mM tetrabutylammonium perchlorate). The tip/substrate bias was set to 90 mV and the potential on the Pt wire was varied during the measurement.

Conductance histograms of 1000 traces measured at different gate potentials were recorded.

121

4.15. Details on the DFT Calculations

All calculations were done using Jaguar (Schrodinger, Inc., New York, NY, 2014). All calculations were based on density functional theory and used the B3LYP functional. The 6-31G** basis sets were used throughout, with the LACVP potentials/basis sets used for the heavy atoms.

The geometry of trans-Co6Se8[P(CH3)3]4L2, where L = P(CH3)2(C6H4SCH3) was optimized.

4.16. References

1. Lovat, G.; Choi, B.; Paley, D. W.; Steigerwald, M. L.; Venkataraman, L.; Roy, X., Room- Temperature Current Blockade in Atomically Defined Single-Cluster Junctions. Nat. Nano. 2017, 12, 1050-1054. 2. Andres, R. P.; Bein, T.; Dorogi, M.; Feng, S.; Henderson, J. I.; Kubiak, C. P.; Mahoney, W.; Osifchin, R. G.; Reifenberger, R., ''Coulomb Staircase'' at Room Temperature in a Self- Assembled Molecular Nanostructure. Science 1996, 272, 1323-1325. 3. Chen, S. W.; Ingram, R. S.; Hostetler, M. J.; Pietron, J. J.; Murray, R. W.; Schaaff, T. G.; Khoury, J. T.; Alvarez, M. M.; Whetten, R. L., Gold Nanoelectrodes of Varied Size: Transition to Molecule-like Charging. Science 1998, 280, 2098-2101. 4. Klein, D. L.; Roth, R.; Lim, A. K. L.; Alivisatos, A. P.; McEuen, P. L., A Single-Electron Transistor Made from a Cadmium Selenide Nanocrystal. Nature 1997, 389, 699-701. 5. Liang, W. J.; Shores, M. P.; Bockrath, M.; Long, J. R.; Park, H., Kondo Resonance in a Single-Molecule Transistor. Nature 2002, 417, 725-729. 6. Park, J.; Pasupathy, A. N.; Goldsmith, J. I.; Chang, C.; Yaish, Y.; Petta, J. R.; Rinkoski, M.; Sethna, J. P.; Abruna, H. D.; McEuen, P. L.; Ralph, D. C., Coulomb Blockade and the Kondo Effect in Single-Atom Transistors. Nature 2002, 417, 722-725. 7. Albrecht, T.; Guckian, A.; Ulstrup, J.; Vos, J. G., Transistor-like Behavior of Transition Metal Complexes. Nano Lett. 2005, 5, 1451-1455. 8. Nitzan, A.; Ratner, M. A., Electron Transport in Molecular Wire Junctions. Science 2003, 300, 1384-1389. 9. Perrin, M. L.; Burzuri, E.; van der Zant, H. S. J., Single-Molecule Transistors. Chem. Soc. Rev. 2015, 44, 902-919. 10. Su, T. A.; Neupane, M.; Steigerwald, M. L.; Venkataraman, L.; Nuckolls, C., Chemical Principles of Single-Molecule Electronics. Nat. Rev. Mater. 2016, 1, 16002.

122

11. Averin, D.; Likharev, K., Single Electronics: A Correlated Transfer of Single Electrons and Cooper Pairs in Systems of Small Tunnel Junctions. Mesoscopic Phenomena in Solids 1991, 30, 173-271. 12. Dorogi, M.; Gomez, J.; Osifchin, R.; Andres, R. P.; Reifenberger, R., Room-Temperature Coulomb-Blockade from a Self-Assembled Molecular Nanostructure. Phys. Rev. B 1995, 52, 9071-9077. 13. Graf, H.; Vancea, J.; Hoffmann, H., Single-Electron Tunneling at Room Temperature in Cobalt Nanoparticles. Appl. Phys. Lett. 2002, 80, 1264-1266. 14. Kano, S.; Tada, T.; Majima, Y., Nanoparticle Characterization Based on STM and STS. Chem. Soc. Rev. 2015, 44, 970-987. 15. Choi, B.; Capozzi, B.; Ahn, S.; Turkiewicz, A.; Lovat, G.; Nuckolls, C.; Steigerwald, M. L.; Venkataraman, L.; Roy, X., Solvent-Dependent Conductance Decay Constants in Single Cluster Junctions. Chem. Sci. 2016, 7, 2701-2705. 16. Roy, X.; Schenck, C. L.; Ahn, S.; Lalancette, R. A.; Venkataraman, L.; Nuckolls, C.; Steigerwald, M. L., Quantum Soldering of Individual Quantum Dots. Angew. Chem. Int. Edit. 2012, 51, 12473-12476. 17. Xu, B. Q.; Tao, N. J. J., Measurement of Single-Molecule Resistance by Repeated Formation of Molecular Junctions. Science 2003, 301, 1221-1223. 18. Capozzi, B.; Xia, J.; Adak, O.; Dell, E. J.; Liu, Z.-F.; Taylor, J. C.; Neaton, J. B.; Campos, L. M.; Venkataraman, L., Single-Molecule Diodes with High Rectification Ratios through Environmental Control. Nat. Nanotechnol. 2015, 10, 522-527. 19. Nagahara, L. A.; Thundat, T.; Lindsay, S. M., Preparation and Characterization of STM Tips for Electrochemical Studies. Rev. Sci. Instrum. 1989, 60, 3128-3130. 20. Capozzi, B.; Low, J. Z.; Xia, J. L.; Liu, Z. F.; Neaton, J. B.; Campos, L. M.; Venkataraman, L., Mapping the Transmission Functions of Single-Molecule Junctions. Nano Lett. 2016, 16, 3949-3954. 21. Gonzalez, M. T.; Wu, S. M.; Huber, R.; van der Molen, S. J.; Schonenberger, C.; Calame, M., Electrical Conductance of Molecular Junctions by a Robust Statistical Analysis. Nano Lett. 2006, 6, 2238-2242. 22. Kamenetska, M.; Koentopp, M.; Whalley, A.; Park, Y. S.; Steigerwald, M.; Nuckolls, C.; Hybertsen, M.; Venkataraman, L., Formation and Evolution of Single-Molecule Junctions. Phys. Rev. Lett. 2009, 102, 126803. 23. Yanson, A. I.; Bollinger, G. R.; van den Brom, H. E.; Agrait, N.; van Ruitenbeek, J. M., Formation and Manipulation of a Metallic Wire of Single Gold Atoms. Nature 1998, 395, 783-785.

123

24. Capozzi, B.; Chen, Q.; Darancet, P.; Kotiuga, M.; Buzzeo, M.; Neaton, J. B.; Nuckolls, C.; Venkataraman, L., Tunable Charge Transport in Single-Molecule Junctions via Electrolytic Gating. Nano Lett. 2014, 14, 1400-1404. 25. Kouwenhoven, L. P.; Marcus, C. M.; Mceuen, P. L.; Tarucha, S.; Westervelt, R. M.; Wingreen, N. S., Electron Transport in Quantum Dots. Nato. Adv. Sci. 1997, 345, 105-214. 26. Bard, A. J.; Faulkner, L.F., Electrochemical Methods: Theory and Applications. Wiley: United States, 2001. 27. Osorio, H. M.; Catarelli, S.; Cea, P.; Gluyas, J. B. G.; Hartl, F.; Higgins, S. J.; Leary, E.; Low, P. J.; Martin, S.; Nichols, R. J.; Tory, J.; Ulstrup, J.; Vezzoli, A.; Milan, D. C.; Zeng, Q., Electrochemical Single-Molecule Transistors with Optimized Gate Coupling. J. Am. Chem. Soc. 2015, 137, 14319-14328. 28. Kay, N. J.; Higgins, S. J.; Jeppesen, J. O.; Leary, E.; Lycoops, J.; Ulstrup, J.; Nichols, R. J., Single-Molecule Electrochemical Gating in Ionic Liquids. J. Am. Chem. Soc. 2012, 134, 16817-26. 29. Schwarz, F.; Kastlunger, G.; Lissel, F.; Egler-Lucas, C.; Semenov, S. N.; Venkatesan, K.; Berke, H.; Stadler, R.; Lörtscher, E., Field-Induced Conductance Switching by Charge- State Alternation in Organometallic Single-Molecule Junctions. Nat. Nanotechnol. 2016, 11, 170-176. 30. Batra, A.; Darancet, P. T.; Chen, Q.; Meisner, J.; Widawsky, J. R.; Neaton, J. B.; Nuckolls, C.; Venkataraman, L., Tuning Rectification in Single-Molecular Diodes. Nano Lett. 2013, 13 (12), 6233-6237. 31. Hilsenbeck, S. J.; Young, V. G.; Mccarley, R. E., Synthesis, Structure, and Characterization of N-Ligated Mo6S8L6 Cluster Complexes - Molecular Precursors to Chevrel Phases. Inorg. Chem. 1994, 33, 1822-1832. 32. Jin, S.; Popp, F.; Boettcher, S. W.; Yuan, M.; Oertel, C. M.; DiSalvo, F. J., Synthesis, Characterization and Properties of Mo6S8(4-tert-butylpyridine)6 and Related M6S8L6 Cluster Complexes (M = Mo, W). J. Chem. Soc. Dalton Trans. 2002, 3096-3100. 33. Saito, T.; Yamamoto, N.; Nagase, T.; Tsuboi, T.; Kobayashi, K.; Yamagata, T.; Imoto, H.; Unoura, K., Molecular-Models of the Superconducting Chevrel Phases - Syntheses and Structures of [Mo6S8(PEt3)6] and [PPN][Mo6X8(PEt3)6] (X = S, Se, PPN = (Ph3P)2N). Inorg. Chem. 1990, 29, 764-770.

34. Saito, T.; Yamamoto, N.; Yamagata, T.; Imoto, H., Synthesis of [Mo6S8(PEt3)6] by Reductive Dimerization of a Trinuclear Molybdenum Chloro Sulfido Cluster Complex Coordinated with Triethylphosphine and Methanol - A Molecular-Model for Superconducting Chevrel Phases. J. Am. Chem. Soc. 1988, 110, 1646-1647. 35. Li, Z.; Li, H.; Chen, S.; Froehlich, T.; Yi, C.; Schönenberger, C.; Calame, M.; Decurtins, S.; Liu, S.-X.; Borguet, E., Regulating a Benzodifuran Single Molecule Redox Switch via

124

Electrochemical Gating and Optimization of Molecule/Electrode Coupling. J. Am. Chem. Soc. 2014, 136, 8867-8870. 36. Wang, X. B.; Niu, S. Q.; Yang, X.; Ibrahim, S. K.; Pickett, C. J.; Ichiye, T.; Wang, L. S., Probing the Intrinsic Electronic Structure of the Cubane [4Fe-4S] Cluster: Nature's Favorite Cluster for Electron Transfer and Storage. J. Am. Chem. Soc. 2003, 125, 14072- 14081. 37. Degroot, M. W.; Corrigan, J. F., Comprehensive Coordination Chemistry II: Oxford, 2003. 38. Cecconi, F.; Ghilardi, C. A.; Midollini, S., Synthesis and Structural Characterization of a Paramagnetic Octahedral Cobalt-Sulfur Cluster, [Co6(μ3-S)8(PEt3)6]BPh4. Inorg. Chem. 1982, 64, L47-L48. 39. Blanc, E.; Schwarzenbach, D.; Flack, H. D., The Evaluation of Transmission Factors and Their 1st Derivatives with Respect to Crystal Shape Parameters. J. Appl. Crystallogr. 1991, 24, 1035-1041. 40. Clark, R. C.; Reid, J. S., The Analytical Calculation of Absorption in Multifaceted Crystals. Acta Crystallogr. A 1995, 51, 887-897. 41. Version 1.171.37.35, Oxford Diffraction / Agilent Technologies UK Ltd, Yarnton, England. 2014. 42. Sheldrick, G. M., SHELXT - Integrated Space Group and Crystal Structure Determination. Acta Crystallogr. 2015, 71, 3-8. 43. Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.; Puschmann, H., OLEX2: A Complete Structure Solution, Refinement and Analysis Program. J. Appl. Crystallogr. 2009, 42, 339-341. 44. Guzei, I. A., An Idealized Molecular Geometry Library for Refinement of Poorly Behaved Molecular Fragments with Constraints. J. Appl. Crystallogr. 2014, 47, 806-809. 45. CrystalMaker Software Ltd, Oxford, England (www.crystalmaker.com).

125

Chapter 5. Three-Dimensional Network of Cyanide-Linked Molecular Clusters

5.1. Introduction

Cyanometalate anions have been used extensively as building blocks for the assembly of magnetic coordination compounds due to their ability to form strong bonds with transition metal cations and to mediate strong magnetic interactions. Cyanide bridged assemblies span a wide variety of hybrid organic-inorganic materials, including single molecule magnets,1-3 single chain magnets,4, 5 magnetic sponges,6 switchable spin-crossover7-9 and charge-transfer systems.10, 11 The use of cyanide groups to link distinct molecular components is invaluable to the field of molecular magnetism because it makes it possible to make materials that are not only magnetically interesting but can also merge the intrinsic physical properties of the molecular building blocks, such as electric conductivity,12 redox activity,13, 14 luminescence,15 and chirality.16, 17 This molecular approach allows for a rational design of intricate magnetic materials through effective manipulation of dimensionality, topology, and structural versatility of the molecular components.

Cyanide complexes of octahedral metal clusters face-capped by chalcogens or halides with the composition M6E8 (M = transition metal; E = S, Se, Te) or M6X12 (X = halide) have been used as building blocks to construct three-dimensional (3D) coordination networks.18-21 Notable examples of such assembly include cluster-expanded Prussian blue analogues.22-25 Thus far, no significant magnetic ordering has been observed in the cyanide-linked networks of M6E8/M6X12 molecular clusters. Even at low temperatures, the paramagnetic centers in the assemblies are well- separated by diamagnetic clusters. Assemblies containing paramagnetic molecular clusters have also not coupled magnetically with the surrounding paramagnetic transition metal ions, as determined by magnetic susceptibility measurements.20, 23, 25

126

Of the many known Chevrel phase molecular clusters, those with Fe6S8 core have

26 especially pronounced paramagnetism. For example, in [Fe6S8(PEt3)6](PF6), the ground state has

S = 7/2 due to the ferromagnetic coupling between low-spin Fe(III) and one intermediate-spin

Fe(II), and oxidized [Fe6S8(PEt3)6](PF6)2 has S = 4. The paramagnetic ground states in these iron sulfide molecular clusters most likely arise from their closely spaced set of electroactive orbitals.27,

28 Unlike with 4d/5d transition metal chalcogenide clusters, the orbitals in which the unpaired electrons reside in iron sulfide clusters may overlap more closely with the σ- or π-type orbitals of the cyanide ligands used in magnetic superexchange coupling. In this section, we describe the synthesis and properties of a cyanide-ligated Fe6S8 molecular cluster, [NEt4]5[Fe6S8(CN)6], and its reaction with Mn(II) ions to assemble into a 3D extended coordination compound that shows ferromagnetic ordering below ~25 K.

5.2. Synthesis and Properties of Cyanide-Ligated Fe6S8 Molecular Clusters

Exchange of the phosphine ligands with stronger σ-bonding ligands such as N-heterocyclic carbenes and cyanides have been reported for iron sulfide clusters with different core compositions.29, 30 We first explored whether we can employ similar ligand exchange techniques on the octahedral cluster to obtain cyanide-ligated Fe6S8. We found that [NEt4]5[Fe6S8(CN)6] (4) can be prepared in quantitative yield via a ligand exchange reaction with [Fe6S8(PEt3)6][BF4]2 and tetraethylammonium cyanide (NEt4CN) in acetonitrile. Whereas the phosphine-bound Fe6S8 cluster is highly soluble in acetonitrile, the cyanide substituted cluster 4 is sparingly soluble.

Therefore, as the substitution reaction proceeds, 4 precipitates out of acetonitrile solution. The molecular structure of 4 is shown in Figure 5.1 and selected crystallographic data are presented in

5- Table 5.1. The compound contains a discrete octahedral cluster anion [Fe6S8(CN)6] , which

127 consists of an octahedron of Fe atoms capped by a cube of S atoms and six terminal cyanide ligands.

The Fe-Fe distances range from 2.608(1) and 2.651(5) Å and average to 2.63(1) Å , and the average

Fe-S bond distance is 2.25(1) Å . These values agree well with those reported for phosphine-ligated

31 Fe6S8 molecular clusters.

5- Figure 5.1. Structure of [Fe6S8(CN)6] in compound 4, showing thermal ellipsoids at 50% probability. Iron, green; sulfur, yellow; carbon, black; nitrogen, light blue.

We reliably obtain only 4 even when we vary the added equivalencies of the NEt4CN,

2+ indicating that the ligand substitution proceeds with one-electron reduction of the [Fe6S8] core and complete substitution of the phosphine ligands. Once isolated, 4 is only soluble in highly polar organic solvents such as methanol and propylene carbonate. Cyclic voltammetry (CV) of

5+ [Fe6S8(CN)6] in propylene carbonate shows two reversible redox events, in line with the electrochemical behaviors observed in iron sulfide molecular clusters (Figure 5.2).26, 27 The magnetic susceptibility measurements for 4 establish room temperature S = 5/2 (μeff = 6.0 μB) ground state, which can be rationalized by ferromagnetic coupling between five low-spin Fe(III)

(S = 1/2) and one low-spin Fe(II) (S = 0) (Figure 5.3).

128

Figure 5.2. CV of 4 in 0.1 M TBAPF6 in propylene carbonate with a 100 mV/s scan rate.

Figure 5.3. Temperature dependence of the (a) reciprocal molar magnetic susceptibility and (b) effective magnetic moment of 4 at 500 Oe. The red line in (a) is a fit to the data at 3-300 K to the Curie-Weiss law.

5.3. Assembly into 3D Coordination Network

We used 4 as a building block for the synthesis of cyanide-bridged 3D coordination compound. Upon adding a solution of Mn(acetate)2 in methanol to 4 dissolved in propylene carbonate, black powder precipitates out of the mixture immediately. When the diffusion of the two layers is fast, polycrystalline materials form that have uniform dispersion of Fe, Mn, and S

129 throughout the crystal in mass % ratio: 44, 22, 34, respectively (Figure 5.4). We use this polycrystalline material for the investigation of the compound’s magnetic properties.

Figure 5.4. (a) HAADF image of 5’ and (b-d) EDS mapping of Mn, Fe, and S in the polycrystalline material, respectively.

When the diffusion of the two layers is significantly slowed down, with the addition of a buffer layer of 1:1 propylene carbonate and methanol, small black crystals are obtained, which were suitable for single crystal X-ray diffraction (SCXRD). X-ray analysis reveals that the

5- compound is a 3D network of [Fe6S8(CN)6] and Mn(II) units interconnected by cyanide ligands to form Fe-C≡N-Mn linkages (Figure 5.5a). Overall, the compound has a chemical formula of

(NEt4)Mn3(OAc)3(HOMe)4(H2O)Fe6S8(CN)6 (5). Within the interstitial spaces of the structure lie one NEt4 cation and a methanol of solvation.

In contrast to the cluster-expanded Prussian blue analogs that have a general formula

M’4[M6E8(CN)6]3·xH2O, we observe that our coordination network is structurally more complex.

First, there are three different environments for Mn(II) (Figure 5.5b). Mn1 contains only one Fe-

C≡N-Mn linkage with a Fe6S8 cluster, and five of its octahedral coordination sites are occupied by one methanol, one water, one acetate, and one doubly coordinated acetate molecule. Mn2 contains two Fe-C≡N-Mn linkages separated by ~97° (N-Mn-N), and four of its remaining coordination sites are occupied by two methanol and two acetate molecules. The acetate molecules coordinated to Mn2 are also coordinated to Mn1 of neighboring Fe6S8 clusters. Mn3 contains three Fe-C≡N-

130

Mn linkages in fac configuration to the Mn(II) site. The remaining coordination sites of Mn3 are occupied by one methanol and one doubly coordinated acetate molecule. In all three types of Mn sites, methanol serves as a coordinating solvent through the oxygen. The complexity of this 3D

+ structure may be attributed to several factors. The large [NEt4] cations supporting the cluster anion may not be able to fit into the interstitial spaces of a cage-like shape as in other cluster-based

Prussian blue analogs.25 Another possibility is charge-balance. It has been shown in other cluster- based assemblies with cyanide bridges that the charge of the molecular cluster affects the overall structure of the 3D network compound.20 Lastly, the coordinating abilities of acetate in the Mn(II) precursor as well as solvent may halt the growth of more extended Fe-C≡N-Mn linkages. Efforts to address these concerns are currently underway.

Figure 5.5. (a) Molecular structure of 5 along the b-axis. Color scheme as in Figure 5.1. Manganese, blue; oxygen, red. NEt4 cation, methanol of solvation, and hydrogen atoms have been omitted for clarity. (b) Asymmetric unit of 5 showing three different Mn(II) sites.

131

Figure 5.6. PXRD pattern of 5 simulated from SCXRD (light blue) and measured experimentally (black).

We note that the powder X-ray diffraction (PXRD) pattern of the polycrystalline material and the simulated PXRD pattern from the SCXRD data are markedly different (Figure 5.6). As aforementioned, we use the polycrystalline sample to characterize other physical properties of the coordination compound, which we distinguish as 5’. We are currently investigating the structure of the polycrystalline compound 5’ with synchrotron X-ray diffraction.

5.4. Magnetic Properties of Cyanide-Bridged Fe6S8 Clusters

Above 25 K, 5’ exhibits Curie-Weiss behavior, following the relationship χM(T) = [C/(T –

θ)] + χD + χTIC, where C is the Curie constant, θ is the Weiss constant, and χD and χTIC are the diamagnetic and temperature-independent contributions, respectively (Figure 5.7). We obtained a

-1 -1 -1 good fit to the data, with C = 7.1 emu K Oe (mol f.u.) , θ = 25.6 K, χD = -0.0009 emu Oe (mol

-1 -1 -1 f.u.) , and χTIC = -0.002 emu Oe (mol f.u.) . The positive Weiss constant indicates ferromagnetic interactions between neighboring magnetic units. Our analysis indicates that 5’ has a room temperature μeff of 8.0 μB/f.u. which is smaller than the value for uncoupled three Mn(II) (each S

132

5- = 5/2) and one [Fe6S8(CN)6] (S = 5/2). The μeff increases abruptly below 25 K and reaches a maximum (μeff = 83.0 μB) at 21 K, which is much higher than the ferromagnetic spin state of the individual components; this observation suggests long-range ferromagnetic ordering.

Figure 5.7. Temperature dependence of the (a) reciprocal molar magnetic susceptibility and (b) effective magnetic moment of 5’ at 10 Oe. The red line in (a) is a fit to the data at 30-300 K to the Curie-Weiss law.

Figure 5.8a shows the temperature dependence of zero-field cooled (ZFC) and field cooled

(FC) magnetization for 5’ at 100 Oe. We observe an abrupt change in magnetization at ~25 K. In the ZFC measurement, we cooled the sample to 3 K in the absence of an applied magnetic field, then applied 100 Oe, and observed the magnetization of the sample as we raised the temperature.

In the FC measurement, we applied the same field at a temperature well above 25 K and observed the magnetization as we reduced the temperature. As the strength of the magnetic field increases, the transition softens.

133

Figure 5.8. (a) Temperature dependence of the ZFC-FC magnetization of 5’ with an external field of 100 Oe. (b) Magnetization of 5 as a function of the applied field above (100 K) and below (20 K, 1.7 K) the transition temperature. The inset shows the low-field magnetization demonstrating hysteresis below the transition temperature.

The magnetic response of 5’ to an external field changes dramatically upon cooling below the transition temperature. Figure 5.8b shows the field-dependent magnetization curve at three different temperatures. At 100 K, the magnetization scales linearly with the applied magnetic field.

At 1.7 K, the magnetization curves are sigmoidal with a coercive field of ~200 Oe. The hysteresis is maintained at 20 K and supports long-range ferromagnetic ordering below the transition temperature (~25 K).

5.5. Conclusions

We observe long-range ferromagnetic ordering in a 3D coordination compound assembled from cyanide-ligated iron sulfide molecular clusters and high-spin transition metal cations. The strong cooperative magnetic response in this material likely arises from strong coupling between the paramagnetic centers that are bridged by cyanide. By varying the transition metal ion precursors, we may be able to tune the magnetic properties of cluster-linked networks. These results present a promising opportunity to design magnetic materials with multiple functionalities.

134

5.6. General Synthesis Information

All reactions and sample preparations were carried out under inert atmosphere in a nitrogen-filled glovebox unless otherwise noted. The compound [Fe6S8(PEt3)6][BF4]2 was synthesized according to a reported procedure.31 All reagents were purchased from Aldrich and used as received. Acetonitrile and propylene carbonate were dried with 4 Å molecular sieves, degassed, and distilled under reduced pressure. All other dry and deoxygenated solvents were prepared by elution through a dual-column solvent system (Mbraun SPS).

5.7. Synthetic Procedures and Characterization of Compounds

[NEt4]5[Fe6S8(CN)6] (4). To a solution of [Fe6S8(PEt3)6][BF4]2 (500 mg, 0.3 mmol) in 10 mL of acetonitrile was added tetraethylammonium cyanide (320 mg, 2.0 mmol). The reaction mixture was stirred to dissolve tetraethylammonium cyanide and was left undisturbed at room temperature.

Crystals of [NEt4]5[Fe6S8(CN)6] grew overnight. The supernatant solution was decanted and the dark brown crystals were washed with acetonitrile and ether and dried in vacuo. Yield: 460 mg

-1 (87%) IR (ATR): νCN = 2076 cm

(NEt4)Mn3(OAc)3(HOMe)4(H2O)Fe6S8(CN)6 (5, 5’). To a solution of 4 (100 mg, 0.06 mmol) in

2 mL of propylene carbonate was added slowly a solution of Mn(acetate)2 (110 mg, 0.6 mmol) in

2 mL of methanol. Upon the addition of the Mn(acetate)2 solution, black powder precipitated (5’).

The supernatant solution was decanted and the black precipitate was washed with methanol and dried in vacuo. Yield: 60 mg (68 %). Dark brown rectangular crystals of 5 suitable for SCXRD were obtained by slow diffusion of the Mn(acetate)2 solution to a concentrated solution of 4 over

-1 1 week. IR (ATR): νCN = 2080 cm (broad)

135

5.8. Instrumentation

Infrared Spectroscopy

Infrared spectra were recorded on a Perkin Elmer Spectrum-400 FTIR spectrometer using a PIKE ATR attachment.

X-ray Diffraction

Single crystal X-ray diffraction data were collected on an Agilent SuperNova diffractometer using mirror-monochromated Cu Kα radiation. The crystals were mounted using a

MiTeGen MicroMount cooled to 100 or 140 K with an Oxford-Diffraction Cryojet system. Data collection, integration, scaling (ABSPACK) and absorption correction (face-indexed Gaussian integration32 or numeric analytical methods33) were performed in CrysAlisPro.34 Structure solution was performed using ShelXS,35 ShelXT,36 or SuperFlip.37 Subsequent refinement was performed by full-matrix least-squares on F2 in ShelXL.35 Olex238 was used for viewing and to prepare CIF files. PLATON39 was used for SQUEEZE,40 ADDSYM41 and TwinRotMat. Many disordered solvent molecules were modeled as rigid fragments from the Idealized Molecular Geometry

Library.42 The most pertinent crystallographic data for all compounds are listed in Table 5.1.

Power X-ray diffraction data were collected on a PANalytical X’Pert3 Powder diffractometer using Cu Kα radiation. Small crystals were evenly dispersed, uncrushed, on a zero- background Si plate.

Cyclic Voltammetry

Electrochemistry for 4 was carried out in THF with tetra-n-butylammonium hexafluorophosphate (TBAPF6) supporting electrolyte (0.1 M) using a CH Instruments

136

Electrochemical Analyzer potentiostat inside a nitrogen-filled glovebox. Measurements were carried out with a glassy carbon working electrode, a platinum counter electrode, and a platinum pseudo-reference electrode. Measurements were calibrated using ferrocene/ferrocenium redox couple.

Magnetic Susceptibility Measurements

Magnetic susceptibility of the compounds was measured using a Cryogenics S700X

SQUID magnetometer. Loose polycrystalline sample was placed into a gel capsule, which was placed inside a plastic straw. Temperature sweeps of 4 and 5’ were performed for temperature range 3-300 K at various fields. 5’ was measured under both zero-field cooled (ZFC) and field cooled (FC) conditions. In both cases, the magnetization was measured in the temperature range

3-100 K at an applied field of 10, 50, 100, 500 Oe. In addition, field sweeps at applied fields between -50 kG and 50 kG were measured at 1.7, 20, 100K. The magnetism data were corrected for the small diamagnetic contribution of the gel capsule containing the sample and the straw and the diamagnetic contribution of the compounds.

137

5.9. Structural Determination

Structure solution and space group assignment were performed in ShelXT.

Table 5.1. Selected crystallographic data for 4 and 5. Compound 4 5 Formula C54H112N15Fe6S8 C25H45O12N7Mn3Fe6S8 MW 1563.2 1392.1 Space Group C 2/c P 21/c a (Å ) 15.1051(7) 19.572(5) b (Å ) 22.5628(10) 13.2164(18) c (Å ) 21.4841(10) 19.025(5) α (°) 90 90 β (°) 91.631(4) 90.84(2) γ, (°) 90 90 V (Å 3) 7319.1(6) 4920.7(19) Z 4 4 -3 ρcalc (g cm ) 1.419 1.879 T (K) 100 140 λ (Å) 1.54184 1.54184 2θmin, 2θmax 7.044, 146.108 12.128, 118.866 Nref 23919 33080 R(int), R(σ) 0.0738, 0.0800 0.3876, 0.2863 μ(mm-1) 11.766 23.468 Size (mm) 0.155 x 0.036 x 0.023 0.044 x 0.027 x 0.019 Tmax, Tmin 0.763, 0.632 0.714, 0.557 Data 7298 5052 Restraints 293 22 Parameters 575 558 R1(obs) 0.0505 0.1047 wR2(all) 0.1229 0.2610 S 1.026 0.983 Peak, hole (e- Å -3) 0.91, -0.80 0.82, -0.74

5.10. References

1. Song, Y.; Zhang, P.; Ren, X. M.; Shen, X. F.; Li, Y. Z.; You, X. Z., Octacyanometallate- II Based Single-Molecule Magnets: (Co9M6V)M (M = W, Mo). J. Am. Chem. Soc. 2005, 127, 3708-3709. 2. Aromí, G.; Brechin, E. K., Synthesis of 3d Metallic Single-Molecule Magnets. In Single- Molecule Magnets and Related Phenomena, Winpenny, R., Ed. Springer Berlin Heidelberg: 2006. 3. Shatruk, M.; Avendano, C.; Dunbar Kim, R., Cyanide‐Bridged Complexes of Transition Metals: A Molecular Magnetism Perspective. Prog. Inorg. Chem. 2009. 4. Lescouëzec, R.; Vaissermann, J.; Ruiz‐Pérez, C.; Lloret, F.; Carrasco, R.; Julve, M.; Verdaguer, M.; Dromzee, Y.; Gatteschi, D.; Wernsdorfer, W., Cyanide‐Bridged Iron(III)–

138

Cobalt(II) Double Zigzag Ferromagnetic Chains: Two New Molecular Magnetic Nanowires. Angew. Chem. Int. Edit. 2003, 42, 1483-1486. 5. Lescouezec, R.; Toma, L. M.; Vaissermann, J.; Verdaguer, M.; Delgado, F. S.; Ruiz-Perez, C.; Lloret, F.; Julve, M., Design of Single Chain Magnets through Cyanide-Bearing Six- Coordinate Complexes. Coordin. Chem. Rev. 2005, 249, 2691-2729. 6. Pinkowicz, D.; Podgajny, R.; Sieklucka, B., Molecular Magnetic Sponges. Molecular Magnetic Materials 2016. 7. Niel, V.; Martinez-Agudo, J. M.; Muñoz, M. C.; Gaspar, A. B.; Real, J. A., Cooperative Spin Crossover Behavior in Cyanide-Bridged Fe(II)−M(II) Bimetallic 3D Hofmann-like Networks (M = Ni, Pd, and Pt). Inorg. Chem. 2001, 40, 3838-3839. 8. Shatruk, M.; Dragulescu-Andrasi, A.; Chambers, K. E.; Stoian, S. A.; Bominaar, E. L.; Achim, C.; Dunbar, K. R., Properties of Prussian Blue Materials Manifested in Molecular Complexes: Observation of Cyanide Linkage Isomerism and Spin-Crossover Behavior in Pentanuclear Cyanide Clusters. J. Am. Chem. Soc. 2007, 129, 6104-6116. 9. Volatron, F.; Catala, L.; Riviere, E.; Gloter, A.; Stephan, O.; Mallah, T., Spin-Crossover Coordination Nanoparticles. Inorg. Chem. 2008, 47, 6584-6586. 10. Berlinguette, C. P.; Dragulescu-Andrasi, A.; Sieber, A.; Galán-Mascarós, J. R.; Güdel, H.- U.; Achim, C.; Dunbar, K. R., A Charge-Transfer-Induced Spin Transition in the Discrete Cyanide-Bridged Complex {[Co(tmphen)2]3[Fe(CN)6]2}. J. Am. Chem. Soc. 2004, 126, 6222-6223. 11. Enoki, T.; Miyazaki, A., Magnetic TTF-Based Charge-Transfer Complexes. Chem. Rev. 2004, 104, 5449-5477. 12. Coronado, E.; Galan-Mascaros, J. R.; Gomez-Garcia, C. J.; Laukhin, V., Coexistence of Ferromagnetism and Metallic Conductivity in a Molecule-Based Layered Compound. Nature 2000, 408, 447-449. 13. Fortier, S.; Le Roy, J. J.; Chen, C.-H.; Vieru, V.; Murugesu, M.; Chibotaru, L. F.; Mindiola, D. J.; Caulton, K. G., A Dinuclear Cobalt Complex Featuring Unprecedented Anodic and Cathodic Redox Switches for Single-Molecule Magnet Activity. J. Am. Chem. Soc. 2013, 135, 14670-14678. 14. Freedman, D. E.; Jenkins, D. M.; Iavarone, A. T.; Long, J. R., A Redox-Switchable Single- Molecule Magnet Incorporating [Re(CN)7]3. J. Am. Chem. Soc. 2008, 130, 2884-2885. 15. Chelebaeva, E.; Larionova, J.; Guari, Y.; Ferreira, R. A. S.; Carlos, L. D.; Paz, F. A. A.; Trifonov, A.; Guérin, C., Luminescent and Magnetic Cyano-Bridged Coordination Polymers Containing 4d−4f Ions: Toward Multifunctional Materials. Inorg. Chem. 2009, 48, 5983- 5995.

139

16. Kaneko, W.; Kitagawa, S.; Ohba, M., Chiral Cyanide-Bridged MnIIMnIII Ferrimagnets, II III [Mn (HL)(H2O)][Mn (CN)6]·2H2O (L = S- or R-1,2-diaminopropane): Syntheses, Structures, and Magnetic Behaviors. J. Am. Chem. Soc. 2007, 129, 248-249. 17. Inoue, K.; Imai, H.; Ghalsasi Prasanna, S.; Kikuchi, K.; Ohba, M.; Ōkawa, H.; Yakhmi, J. V., A Three‐Dimensional Ferrimagnet with a High Magnetic Transition Temperature (TC) of 53 K Based on a Chiral Molecule. Angew. Chem. Int. Edit. 2001, 40, 4242-4245.

18. Beauvais, L. G.; Shores, M. P.; Long, J. R., Cyano-Bridged Re6Q8 (Q = S, Se) Cluster- Cobalt(II) Framework Materials: Versatile Solid Chemical Sensors. J. Am. Chem. Soc. 2000, 122, 2763-2772.

19. Magliocchi, C.; Xie, X.; Hughbanks, T., A Cyanide-Bridged Chain of Mo6Se8 Clusters: A Product of Cyanide-Melt Cluster Synthesis. Inorg. Chem. 2000, 39, 5000-5001. 20. Jin, S.; DiSalvo, F. J., 3-D Coordination Network Structures Constructed from 6- [W6S8(CN)6] Anions. Chem. Mater. 2002, 14, 3448-3457. 21. Brylev, K. A.; Mironov, Y. V.; Naumov, N. G.; Fedorov, V. E.; Ibers, J. A., New Compounds from Tellurocyanide Rhenium Cluster Anions and 3d-Transition Metal Cations Coordinated with Ethylenediamine. Inorg. Chem. 2004, 43, 4833-4838. 22. Shores, M. P.; Beauvais, L. G.; Long, J. R., Cluster-Expanded Prussian Blue Analogues. J. Am. Chem. Soc. 1999, 121, 775-779. 23. Bennett, M. V.; Beauvais, L. G.; Shores, M. P.; Long, J. R., Expanded Prussian Blue 3-/4- Analogues Incorporating [Re6Se8(CN)6] Clusters: Adjusting Porosity via Charge Balance. J. Am. Chem. Soc. 2001, 123, 8022-8032. 24. Beauvais, L. G.; Long, J. R., Synthesis and Characterization of Prussian Blue Analogues 2+ Incorporating the Edge-Bridged Octahedral [Zr6BCl12] Cluster Core. Inorg. Chem. 2006, 45, 236-243. 25. Yan, B.; Zhou, H.; Lachgar, A., Octahedral Niobium Chloride Clusters as Building Blocks of Templated Prussian Blue Framework Analogues. Inorg. Chem. 2003, 42, 8818-8822. 26. Cecconi, F.; Ghilardi, C. A.; Midollini, S.; Orlandini, A.; Zanello, P., Redox Behaviour of 2+ the Iron-Sulphur Cluster [Fe6(μ3-S)8(PEt3)6] . Synthesis and Crystal Structure of the New + - Paramagnetic Monopositive Species [Fe6(μ3-S)8(PEt3)6] as its [PF6] Salt. J. Chem. Soc. Dalton Trans. 1987, 831-835. 27. Goddard, C. A.; Long, J. R.; Holm, R. H., Synthesis and Characterization of Four n+ Consecutive Members of the Five-Member [Fe6S8(PEt3)6] (n = 0−4) Cluster Electron Transfer Series. Inorg. Chem. 1996, 35, 4347-4354. 28. Bencini, A.; Ghilardi, C. A.; Midollini, S.; Orlandini, A.; Russo, U.; Uytterhoeven, M. G.; Zanchini, C., Crystal and Molecular Structure, Magnetic Properties and Scattered-Wave Description of the Electronic Structure of the Hexanuclear Octahedral Clusters [Fe6(μ3- S)8(PEt3)6][PF6] (n = 1 or 2). J. Chem. Soc. Dalton Trans. 1995, 963-974. 140

3− 29. Scott Thomas, A.; Zhou, H. C., The First All‐Cyanide Fe4S4 Cluster: [Fe4S4(CN)4] . Angew. Chem. Int. Edit. 2004, 43, 5628-5631. 30. Deng, L.; Holm, R. H., Stabilization of Fully Reduced Iron−Sulfur Clusters by Carbene 0 Ligation: The [FenSn] Oxidation Levels (n = 4, 8). J. Am. Chem. Soc. 2008, 130, 9878-9886. 31. Agresti, A.; Bacci, M.; Cecconi, F.; A. Ghilardi, C.; Midollini, S., Transition-Metal Complexes with Sulfur Atoms as Ligands. 7. Synthesis, Properties, Structure, and Molecular Orbital Calculations of the Paramagnetic Cluster [Fe6(μ3-S)8(PEt3)6](BPh4)2. Inorg. Chem. 1985, 24, 689-695. 32. Blanc, E.; Schwarzenbach, D.; Flack, H. D., The Evaluation of Transmission Factors and Their 1st Derivatives with Respect to Crystal Shape Parameters. J. Appl. Crystallogr. 1991, 24, 1035-1041. 33. Clark, R. C.; Reid, J. S., The Analytical Calculation of Absorption in Multifaceted Crystals. Acta Crystallogr. A 1995, 51, 887-897. 34. Version 1.171.37.35., Oxford Diffraction / Agilent Technologies UK Ltd, Yarnton, England. 2014. 35. Sheldrick, G. M., A Short History of SHELX. Acta Crystallogr. A 2008, 64, 112-122. 36. Sheldrick, G. M., Crystal Structure Refinement with SHELXL. Acta Crystallogr. 2015, 71, 3-8. 37. Palatinus, L.; Chapuis, G., SUPERFLIP - A Computer Program for the Solution of Crystal Structures by Charge Flipping in Arbitrary Dimensions. J. Appl. Crystallogr. 2007, 40, 786-790. 38. Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.; Puschmann, H., OLEX2: A Complete Structure Solution, Refinement and Analysis Program. J. Appl. Crystallogr. 2009, 42, 339-341. 39. Spek, A. L., Structure Validation in Chemical Crystallography. Acta Crystallogr. D 2009, 65, 148-155. 40. Vandersluis, P.; Spek, A. L., Bypass - An Effective Method for the Refinement of Crystal Structures Containing Disordered Solvent Regions. Acta Crystallogr. A 1990, 46, 194-201. 41. Lepage, Y., Missym-1.1 - A Flexible New Release. J. Appl. Crystallogr. 1988, 21, 983- 984. 42. Guzei, I. A., An Idealized Molecular Geometry Library for Refinement of Poorly Behaved Molecular Fragments with Constraints. J. Appl. Crystallogr. 2014, 47, 806-809.

141

Chapter 6. Self-Assembly of a Layered van der Waals Material Using a Structure-Directing Cluster

6.1. Preface

This chapter is based on a manuscript entitled “van der Waals Solids from Self-Assembled

Nanoscale Building Blocks” by Bonnie Choi, Jaeeun Yu, Daniel W. Paley, M. Tuan Trinh, Maria

V. Paley, Jessica M. Karch, Andrew C. Crowther, Chul-Ho Lee, Roger A. Lalancette, Xiaoyang

Zhu, Philip Kim, Michael L. Steigerwald, Colin Nuckolls, Xavier Roy in Nano Letters.1 It is reprinted with permission from the American Chemical Society. I synthesized all compounds with assistance from Jessica M. Karch. Dr. Daniel W. Paley characterized the molecular clusters using

SCXRD with essential input Professor Roger A. Lalancette. Dr. Jaeeun Yu from Professor Colin

Nuckoll’s group performed the mechanical exfoliation and atomic microscopy. Dr. M. Tuan Trinh from Professor Xiaoyang Zhu’s group performed the optical absorption measurements. Dr. Chul-

Ho Lee from Professors Philip Kim and Colin Nuckolls’ groups fabricated a device to measure the conductivity of the bulk single crystal. Maria V. Paley from Andrew C. Crowther’s group performed Raman spectroscopy of the bulk crystal.

In addition, this chapter contains materials from a manuscript entitled “Two-Dimensional

Fullerene Assembly from an Exfoliated van der Waals Template” by Kihong Lee, Bonnie Choi,

Ilan Jen-La Plante, Maria V. Paley, Xinjue Zhong, Andrew C. Crowther, Jonathan S. Owen,

Xiaoyang Zhu, and Xavier Roy in Angewandte Chemie International Edition.2 I synthesized the bulk crystals used in the study and assisted Kihong Lee in characterizing some of the materials.

142

6.2. Introduction

The development of facile methods for isolation and manipulation of 2D materials has fueled an explosion of interest in layered van der Waals structures.3-6 These solids exhibit strong in-plane bonding and comparatively weak interactions between the layers, allowing them to be exfoliated. Fullerenes are attractive molecular building blocks for assembling functional

7, 8 materials, yet C60 crystallizes into a three-dimensional solid held together by isotropic van der

9 Waals interactions. As a result, C60 cannot be exfoliated into discrete layers, unlike other

10 11 nanostructured forms of carbon such as graphene and carbon nanotubes. Ionic solids of C60 with layered structures have been reported,12, 13 but these solids also cannot be exfoliated because the components are held together by electrostatic interactions. While theory suggests that neutral

14 C60 could form mechanically stable, free-standing monolayers, these structures do not assemble spontaneously from solution and have only been prepared as epitaxial layers through carefully controlled vapor growth or on chemically modified surfaces.15 As with many other 2D materials,

16-19 unique emergent properties may arise in low-dimensional C60 arrays.

Here, we describe a layered van der Waals solid self-assembled from a structure-directing building block and C60 fullerene. The resulting crystalline solid contains a corrugated monolayer of neutral fullerenes and can be mechanically exfoliated. The absorption spectrum of the bulk solid shows an optical gap of 390 ± 40 meV that is consistent with thermal activation energy obtained from electrical transport measurement. We find that the dimensional confinement of fullerenes significantly modulates the optical and electronic properties compared to the bulk C60 crystal, which has a bandgap of 2.3 eV.20

While the mechanical exfoliation of bulk crystals yields thinner layers of the material, it is difficult to isolate much thinner layers within the 2D regime due to the rather weak noncovalent

143 forces holding the building blocks together within the plane of the crystal. We resolve this issue by photopolymerizing the bulk crystals then using a solvent-induced exfoliation technique to obtain a single sandwich layer. We show that the bulk layered van der Waals solid can be used as a template to create a 2D fullerene-based material. Photopolymerization enhances the intralayer mechanical strength to enable exfoliation beyond what was possible with mechanical exfoliation of the pristine bulk crystal. Furthermore, we show that the covalent structure can be depolymerized using thermal treatment, demonstrating the reversibility and stability of the system.

6.3. Synthesis of Structure-Directing Cluster

We employ a structurally adaptable transition metal chalcogenide cluster that associates with C60 and thereby directs the spontaneous assembly of C60 into neutral, corrugated monolayers.

The director- and fullerene-layers form a 2D van der Waals solid whose intralayer bonding is strong enough and whose interlayer bonding is weak enough that the material may be mechanically exfoliated (Figure 6.1).

Figure 6.1. Schematic of a nanoscale director organizing a monolayer of fullerenes. Additional fullerenes that are not held in the ligand sphere have been omitted from the schematic. We hypothesize random distribution of fullerenes along the exfoliated layers. Optical microscope images show a crystal of the layered van der Waals material before and after exfoliation.

144

The reaction of the PEt2phen (PEt2phen = diethyl(9-ethynylphenanthrene)phosphine) ligand, elemental Se, and Co2(CO)8 in refluxing toluene provides the nanoscale director,

Co6Se8(PEt2phen)6. Single crystal X-ray diffraction (SCXRD) confirms that the inorganic core of

21 Co6Se8(PEt2phen)6 is isostructural with the parent cluster Co6Se8(PEt3)6. The similarity between the two compounds is further supported by electronic absorption spectroscopy where the spectrum

21 (Figure 6.2) shows three longer-wavelength absorptions characteristic of the Co6Se8 core, in addition to the absorptions due to the phenanthrene groups in the UV region. The electrochemical properties of Co6Se8(PEt2phen)6 (Figure 6.3) also show three [Co6Se8]-centered reversible oxidations but our director is less reducing than Co6Se8(PEt3)6 due to the diminished electron- donating ability of the PEt2phen ligand compared to PEt3.

Figure 6.2. Electronic absorption spectra of PEt2phen (13 μM) and Co6Se8(PEt2phen)6 (0.72 μM).

145

Figure 6.3. CV of Co6Se8(PEt2phen)6 in 0.1 M TBAPF6 in tetrahydrofuran with a 100 mV/s scan rate.

6.4. Self-Assembly into a Layered Solid and Mechanical Exfoliation

22 We have previously reported that Co6Se8(PEt3)6 forms hierarchical solids with C60; however, while Co6Se8(PEt3)6 is functionally spherical, Co6Se8(PEt2phen)6 can be anisotropic. The ligands on Co6Se8(PEt2phen)6 can reorganize to a thermodynamically stable unit with a shape that is complementary to C60. When we combine Co6Se8(PEt2phen)6 and five equivalents of C60 in toluene, thin, flat, and hexagonal brown crystals of [Co6Se8(PEt2phen)6][C60]5 form. Mechanical evaluation in tandem with crystallography show that this material is not only structurally two- dimensional but also behaves physically like a layered solid.

146

Figure 6.4. AFM topographic images of (a) as-grown single crystal and (b) mechanically exfoliated crystal of [Co6Se8(PEt2phen)6][C60]5. The optical microscope image of the respective crystals is shown in the inset. The thickness of the exfoliated crystal is 139 nm.

Similar to more traditional 2D van der Waals materials, [Co6Se8(PEt2phen)6][C60]5 crystals can be exfoliated. Figure 6.4a shows the atomic force microscope (AFM) image of a

[Co6Se8(PEt2phen)6][C60]5 single crystal grown from solution. We observe a step-and-terrace structure on the surface of the [Co6Se8(PEt2phen)6][C60]5 crystals with ~3 nm steps. We mechanically exfoliated single crystals of [Co6Se8(PEt2phen)6][C60]5, thereby isolating multilayer samples that we transfer onto Si wafers (Figure 6.4b). Under an optical microscope, exfoliated crystals display various colors (Figure 6.5) that we interpret as interference fringes, suggesting that flakes of different thickness are produced using this technique. We produce samples that are as thin as ~130 nm and have smooth surfaces (average roughness Rq ~0.6 nm). This surface

23 roughness is comparable to self-assembled C60 monolayers on Au, and C60 films deposited and annealed on graphite.24 In some exfoliated samples, we also observe small, ~3 nm-high islands on the surfaces (Figure 6.6).

147

Figure 6.5. Optical images of mechanically exfoliated crystals of [Co6Se8(PEt2phen)6][C60]5.

Figure 6.6. Optical microscope images of (a) as-grown single crystal and (c) mechanically exfoliated crystals of [Co6Se8(PEt2phen)6][C60]5, and respective AFM topographic images (b and d). The thickness of the exfoliated crystal is 190 nm as determined by AFM.

6.5. Molecular Recognition of the Ligands and Fullerenes

Flexible molecular recognition elements in the ligand sphere create this self-assembled van der Waals material. Fullerenes are known to bind to molecular receptors such as corannulenes,25 calix[n]arenes,26, 27 cyclic paraphenyleneacetylenes,28 and contorted hexabenzocoronenes29, 30 that have concave surfaces. We reasoned that the octahedral geometry of the Co6Se8 cluster provides a simple yet versatile platform upon which we could build a flexible unit that can recognize C60.

While the phenanthrene ring on each phosphine is flat and provides only weak van der Waals interactions with C60, three ethynylphenanthrene groups - one from each of three adjacent phosphines - can rotate in concert to form a “buckybowl” (Figure 6.7). The disposition of the

148 ligands is different in isolated Co6Se8(PEt2phen)6 (Figure 6.7a) than in the co-crystal of

[Co6Se8(PEt2phen)6][C60]5 (Figure 6.7b). In the presence of C60, the phenanthrenes decorating the

Co6Se8 core extend and organize themselves to form two antipodal buckybowls, each capable of interacting with a guest C60. This supramolecular "dumbbell" is the template that forms a basic unit in the extended structure. An unanticipated consequence of the buckybowl formation is the exposure of the equator of the Co6Se8 core, leaving a void that is filled by additional fullerenes

(Figure 6.8a).

Figure 6.7. Molecular structure of the nanoscale director built from Co6Se8(PEt2phen)6 (a) before and (b) after self- assembly. The PEt2phen ligands on each cluster reorganize to form two buckybowls that host C60s in the solid-state.

149

Figure 6.8. (a) Edge-on view of the molecular structure of [Co6Se8(PEt2phen)6][C60]5 looking down the ab plane. Layer CF consists of [Co6Se8(PEt2phen)6][C60]2, and Layer FL is a corrugated monolayer of C60. Cobalt, blue; selenium, light green; phosphorus, orange; carbon, black. The fullerenes in Layer CF are shown in light blue; fullerenes that are sitting in the buckybowls within Layer FL are grey; fullerenes that are not associated with the buckybowls within FL are green. Hydrogen atoms and toluene molecules of solvation were removed to clarify the view. (b) Space-filled view of a single Layer CF looking down the c-axis. (c) Space-filled view of a single Layer FL looking down the c-axis.

Figure 6.8a shows the extended molecular structure of the dumbbell with the other three equivalents of C60 in [Co6Se8(PEt2phen)6][C60]5. The structure is composed of two layers

(identified as Layers CF and FL) stacked along the hexagonal c-axis. Layer CF is trigonal with stoichiometry [Co6Se8(PEt2phen)6][C60]2 in which the molecular cluster is surrounded by a flattened octahedron of fullerenes that are 3.3 Å above and below the mean plane of the layer

(Figure 6.8b). Layer FL is a slightly corrugated trigonal array of fullerenes (Figure 6.8c). For each cluster, there are five C60s that occupy three crystallographically distinguishable positions: two symmetry-related C60s (grey) are sitting in the buckybowls, and these are a part of the Layer

150

FL. Two symmetry-related C60s (light blue) are located in Layer CF, close to inorganic cluster; the fifth C60 (green), positioned within Layer FL, is not sitting in a buckybowl but is in contact with two interstitial coplanar toluene solvent molecules. Within Layer FL, the C60s are in close contact having centroid-to-centroid distance ranging from 9.65 to 9.94 Å . The step height in the layers

(Figure 6.4) is consistent with the height of a monolayer (Layer CF + Layer FL) calculated from the crystal structure (1/3 of the c-axis length, 2.64 nm). Therefore, when we exfoliate the flakes, they unzip along the 2D sheets of fullerenes.

6.6. Determination of Charge on the Layered Solid

Figure 6.9. Representative Raman spectra for high and low power measurements of the [Co6Se8(PEt2phen)6][C60]5 sample performed on the sample crystal location. The low power measurement was performed first. Peak 1, the Hg(8) mode of C60, is included in the fit because it overlaps slightly with the Ag(2) pentagonal pinch mode. The black curve and red curves are for the high power density (950 W/cm2) and low power density (40 W/cm2) respectively. Fits are the overlaid in the same color, and fits of individual peaks are shown in blue.

Micro-Raman spectroscopy of the layered van der Waals material indicates that no charge is transferred between the constituent clusters in [Co6Se8(PEt2phen)6][C60]5 (Figure 6.9) and that the structure is held together entirely through van der Waals interactions. SQUID magnetometry 151 reveals that the material is diamagnetic and supports the observation that there is no measurable charge transfer. We have collected structural data on a large library of Co6Se8(PR3)6 clusters in known oxidation states (0, 1+, and 2+; Figure 6.10), and we can assess the oxidation state of the cluster by comparing its Co-P and Co-Cotrans distances to this database. This analysis also indicates that the cluster bears no charge in the layered solid.

Figure 6.10. Geometries of Co6Se8(PR3)6 clusters in 0, 1+ and 2+ oxidation states. Open points represent PAr3-capped clusters, which showed a significant effect of steric crowding. See discussion in Section 7.14 for the 3 “neutral” data points grouped among the 1+ structures.

6.7. Optical and Electronic Properties of the van der Waals Crystal

Figure 6.11. (a) Log plot of conductance (G) vs. 1/T for [Co6Se8(PEt2phen)6][C60]5. The two-probe conductance measurements were done on a single crystal. Using the Arrhenius fit, solid red line, we obtained an activation energy of 400 meV. (b) Mid-IR absorption spectrum of [Co6Se8(PEt2phen)6][C60]5. The red dashed curve was obtained from a fit using band-to-band transition for semiconductors (blue) with an additional Gaussian peak (green) near the band- edge. The obtained bandgap is 390 ± 40 meV. 152

We performed two-probe electrical resistivity measurements on single crystals of

[Co6Se8(PEt2phen)6][C60]5. As shown in Figure 6.11a, the layered solid exhibits in-plane activated electrical transport behavior and we observe an exponential decrease of the conductance (G) with decreasing temperature. This thermally activated transport displays an Arrhenius behavior with an activation energy Ea of ~400 meV. This energy barrier is higher than the transport barrier of the

22 parent, ionic solid [Co6Se8(PEt3)6][C60]2 (Ea ~150 meV).

Figure 6.11b shows the optical absorption spectrum for [Co6Se8(PEt2phen)6][C60]5 using

Fourier Transform Infrared (FT-IR) spectroscopy. The measurement was performed in microscopic transmission mode on a single flake of the crystal at room temperature. The measurement was performed in microscopic transmission mode on a single crystal flake at room temperature. The spectrum is characteristic of that of a direct bandgap semiconductor.28 The enhanced absorbance near the threshold is similar to that of an excitonic resonance.29 While the exact origin of this resonant feature near the threshold/bandgap is not known, we use a Gaussian in addition to the bulk part of the Elliot’s theory which describes band-to-band optical transition for direct bandgap semiconductors to fit the experimental data.29 The fit yields a bulk bandgap of

390 ± 40 meV, in excellent agreement with the activation energy obtained from electrical transport measurement. The origin of the resonant enhancement near the bandgap (peak at ~ 350 meV, green curve in Figure 6.11b) is not known and remains a subject for future mechanistic study.

6.8. Photopolymerization of the Template

While the interlayer bonding is still weaker than the intralayer bonding in the layered material, [Co6Se8(PEt2phen)6][C60]5, to enable mechanical exfoliation of the crystals into thinner layers, the intralayer interactions are still too weak to allow for successive exfoliations into free-

153 standing monolayers. As the crystals are exfoliated, they break apart laterally to yield smaller-area sheets. To overcome this issue, we use photopolymerization techniques to covalently link the fullerenes in the FL layer and solvent-induced exfoliation to dissolve the non-polymerized layers to isolate molecular-thin 2D materials (Figure 6.12a,b).2 Upon photo-conversion, the irradiated sample remains crystalline as shown in PXRD (Figure 6.12c); however, the solubility changes dramatically. While pristine crystals of [Co6Se8(PEt2phen)6][C60]5 disassemble into their individual components in toluene, the irradiated crystals of [Co6Se8(PEt2phen)6][C60]5 do not completely dissolve, and we use a solvent-induced exfoliation method to isolate purely the irradiated materials (Figure 6.12a). The conversion of the FL layer in which the fullerenes are held together by molecular recognition and weak van der Waals interactions into covalently polymerized FL layer reduces the sample’s solubility in toluene while augmenting the in-plane mechanical strength of the crystal.

Figure 6.12. (a) Proposed structure for the 2D nanosheet. The grey fullerenes within the FL layer are polymerized and the blue fullerenes in the CF layer are not. (b) Optical image of a photopolymerized crystal during the solvent- induced exfoliation process, leaving the bottom-most layer on the substrate (1) as the parent crystal is displaced (2). (c) PXRD of as-grown [Co6Se8(PEt2phen)6][C60]5 bulk crystals (black) and crystals which were photopolymerized then washed with toluene (green). (d) AFM image of an exfoliated nanosheet transferred to a SiO2/Si substrate. (e) Height profile of the highlighted light green portion in (d). Adapted from reference 2.

154

Using AFM, transmission electron microscopy, and energy-dispersive X-ray spectroscopy, we characterize the polymerized material as one FL/CF/FL sandwich layer. As shown in Figure

6.13d,e, the thickness of the nanosheets are ~5 nm. From the SCXRD of the pristine

[Co6Se8(PEt2phen)6][C60]5, the sandwich structure should have a thickness of ~3.7 nm. We attribute the differences in thicknesses to the buckling of the building blocks upon polymerization and thus expansion along the c-axis. Furthermore, we demonstrate that the 2D nanosheets can be depolymerized using thermal treatment. Thermogravimetric analysis (TGA) on both the structure- directing cluster Co6Se8(PEt2phen)6 and the layered material [Co6Se8(PEt2phen)6][C60]5 show that both compounds are stable up to 520 K (Figure 6.13). When the exfoliated nanosheets supported on a substrate are heated to 450 K for 1 h to induce depolymerization of the fullerenes, the thickness of the flake decreases from 4.9 to 3.3 nm. The thickness upon thermal treatment agrees well with the expected height of the FL/CF/FL sandwich structure in the bulk

[Co6Se8(PEt2phen)6][C60]5 crystal from SCXRD. Breaking the covalent bonds between neighboring fullerenes in the FL layer presumably allows the structure to relax back to its original packing.

Figure 6.13. TGA plots of (a) Co6Se8(PEt2phen)6 and (b) [Co6Se8(PEt2phen)6][C60]5.

155

6.9. Conclusions

We describe here a 2D van der Waals solid self-assembled from a nanoscale building block and C60. The resulting crystalline solid features a corrugated monolayer of fullerenes that are separated by a layer comprising a combination of the structure-directing inorganic cluster and C60.

The crystal can be mechanically exfoliated into thinner sheets with smooth surfaces. Because we see a consistent step size of ~3 nm, we believe that the exfoliation results in “unzipping” of the fullerene monolayer. The absorption spectrum of the bulk crystal shows what is consistent with an excitonic feature at room temperature and bulk optical gap of 390 ± 40 meV that matches well with the thermal activation energy obtained from transport measurement. The bulk crystals can be photopolymerized and exfoliated to yield 2D nanosheets of two fullerene monolayers sandwiched by the structure-directing template layer. These properties are decidedly different from those of the bulk fullerene crystals and thus make this system exciting for future study of dimensionally confined neutral fullerenes. Furthermore, we aim to generalize the use of structure-directing molecular clusters to reorganize canonical materials in new and exciting ways.

6.10. General Synthesis Information

All reactions and sample preparations were carried out under inert atmosphere using standard Schlenk techniques or in a nitrogen-filled glovebox. Chlorodiethylphosphine was purchased from Acros Organics. Dicobalt octacarbonyl and selenium powder were purchased from

Strem Chemicals. C60 buckminsterfullerene was purchased from Bucky USA. All other reagents and solvents were purchased from Sigma Aldrich. Dry and deoxygenated solvents were prepared by elution through a dual-column solvent system (MBraun SPS). 9-Ethynylphenanthrene was prepared according to published protocol.31

156

6.11. Synthetic Procedures and Characterization of the Compounds

Diethyl(9-ethynlphenanthrene)phosphine (“PEt2phen”)

A solution of 9-ethynylphenanthrene31 (1.0 g, 4.9 mmol) in THF (50 mL) was cooled to -

78 ºC and n-butyllithium (3.7 mL, 5.9 mmol, 1.6 M in hexanes) was added dropwise. The reaction mixture was stirred for 1 h at -78 ºC. A solution of chlorodiethylphosphine (920 mg, 7.4 mmol) in

THF (10 mL) was added dropwise to the solution and the reaction was warmed gradually to room temperature. After stirring for ~12 h, the solvent was removed in vacuo and 20 mL of toluene was added to the crude product. The mixture was filtered through a medium frit over celite. The filtrate was evaporated to dryness in vacuo to afford a yellow oil. Yield: 1.1 g (80 %)

1 H NMR (400 MHz, [d8-toluene], 298 K): δ = 1.18-1.26 (6H, m), 1.50-1.74 (4H, m), 7.26-7.46

(5H, m), 7.85 (1H, s), 8.27-8.35 (2H, m), 8.63-8.65 (1H, m)

31 P NMR (162 MHz, [d8-toluene], 298 K): -39

Co6Se8(PEt2phen)6

To a solution of PEt2phen (1.07 g, 3.68 mmol) in toluene (50 mL) was added selenium powder (290 mg, 3.68 mmol) and the suspension was stirred for 1 h at room temperature. Dicobalt octacarbonyl (315 mg, 0.92 mmol) dissolved in toluene (15 mL) was added and the mixture was heated to reflux for ~16 h. The hot mixture was filtered through a fine frit. The dark brown solution was cooled to room temperature and concentrated to 10 mL in vacuo. 80 mL of diethyl ether was added to precipitate the product. The crude product was isolated by filtration through a fine frit 157 and dissolved in 20 mL of toluene. Hexanes diffusion at room temperature yielded dark brown crystals. The supernatant solution was decanted and the recovered dark brown crystals were washed with hexanes (2 x 20 mL) and dried in vacuo. Yield: 745 mg (89 %) based on amount of dicobalt octacarbonyl used.

1 H NMR (400 MHz, [d8-toluene], 298 K): δ = 1.23-1.31 (6H, m), 2.21-2.38 (4H, m), 7.12-7.16

(1H, m), 7.23-7.27 (2H, m), 7.31-7.34 (1H, m), 7.46-7.50 (1H, m), 7.89 (1H, s), 8.16-8.21 (2H, t),

9.02 (1H, d)

31 P NMR (162 MHz, [d8-toluene], 298 K): 38 (broad)

[Co6Se8(PEt2phen)6][C60]5

The cluster Co6Se8(PEt2phen)6 (42 mg, 15 μmol) and C60 (55 mg, 77 μmol) were dissolved in 30 mL of toluene. The solution filtered through a 0.2 μm syringe filter. Dark brown crystals were obtained after 8 weeks upon slow evaporation of toluene. The solid was collected, rinsed with hexanes and ether, and dried in vacuo. Yield: 20 mg (20 %)

Alternatively, dark brown crystals were obtained after 2 weeks upon diffusion of hexanes vapor into a co-solution of Co6Se8(PEt2phen)6 (18 mg, 6 μmol) and (21 mg, 29 μmol) in 8 mL of toluene. The solid was collected, rinsed with hexanes and ether, and dried in vacuuo. Yield: 29 mg

(77%)

6.12. Instrumentation

NMR

All 1H and 31P NMR were recorded on a Bruker DRX400 (400 MHz) NMR spectrometer.

158

Electronic Absorption

Solution-phase electronic absorption spectra were taken on a Shimadzu UV-1800 spectrophotometer. All samples were dissolved in THF and loaded in a quartz cuvette in the glovebox and sealed under nitrogen. All spectra were taken following a recording of the background spectrum of the solvent.

X-ray Diffraction

Single crystal X-ray diffraction data were collected on an Agilent SuperNova diffractometer using mirror-monochromated Cu Kα radiation. The crystals were mounted using a

MiTeGen MicroMount cooled to 100 K with an Oxford-Diffraction Cryojet system. Data collection, integration, scaling (ABSPACK) and absorption correction (face-indexed Gaussian integration32 or numeric analytical methods33) were performed in CrysAlisPro.34 Structure solution was performed using ShelXS,35 ShelXT,36 or SuperFlip.37 Subsequent refinement was performed by full-matrix least-squares on F2 in ShelXL.35 Olex238 was used for viewing and to prepare CIF files. PLATON39 was used for SQUEEZE,40 ADDSYM41 and TwinRotMat. Many disordered solvent molecules were modeled as rigid fragments from the Idealized Molecular Geometry

Library.42 Details of crystallographic data and parameters for data collection and refinement are in

Table 6.1.

Powder X-ray diffraction data were collected on a PANalytical X’Pert3 Powder diffractometer. We used an Anton Paar air-free sample holder equipped with a polycarbonate dome to seal the sample under a nitrogen atmosphere. This dome has a broad amorphous peak from 18 to 25º, and this background was subtracted using HighScore Plus.

159

Cyclic Voltammetry

Electrochemistry for Co6Se8(Et2phen)6 was carried out in THF with tetra-n- butylammonium hexafluorophosphate (TBAPF6) supporting electrolyte (0.1 M) using a CH

Instruments Electrochemical Analyzer potentiostat inside a nitrogen-filled glovebox.

Measurements were carried out with a glassy carbon working electrode, a platinum counter electrode, and a platinum pseudo-reference electrode. Measurements were calibrated using ferrocene/ferrocenium redox couple.

AFM

AFM images were acquired in ScanAsyst mode by a Bruker Dimension Icon system under ambient conditions.

TGA

Thermogravimetric analysis was performed on a Q500 TA Instruments Thermogravimetric

Analyzer. Crystalline materials were dispersed on a platinum pan and loaded into the furnace under nitrogen flow. The samples were heated at 15 °C/min to 900 °C. Co6Se8(PEt2phen)6 crystallizes with one toluene molecule per formula unit. This solvent molecule is lost from at ~130 °C (Figure

6.13a). The cluster Co6Se8(PEt2phen)6 begins to decompose at ~260 °C. Similarly, the layered compound [Co6Se8(PEt2phen)6][C60]5 crystallizes with two toluene molecules per formula unit, which are lost at ~200 °C (Figure 6.13b). The solid [Co6Se8(PEt2phen)6][C60]5 begins to decompose at ~260 °C.

160

6.13. Structural Determination

Co6Se8(PEt2phen)6

Crystals were grown by vapor diffusion of hexanes into a concentrated toluene solution, which afforded black blocks. Part of a crystal (.14 x .04 x .02 mm) was separated carefully, mounted with STP oil treatment, and cooled to 100 K on the diffractometer. Complete data (98.9%) were collected to 0.815 Å . 36565 reflections were collected (11067 unique, 10365 observed) with

R(int) 3.2 % and R(sigma) 3.1 % after Gaussian absorption correction (Tmax .821, Tmin .498).

The structure solved readily in P-1 using ShelXS with ½ cluster in the asymmetric unit. All non-H, non-solvent atoms were located in Fourier maps and refined anisotropically with no restraints. A toluene molecule was located on an inversion center and refined as a rigid fragment from the IMGL. C-H hydrogens were placed in calculated positions and refined with riding coordinates and ADPs.

The final refinement (11067 data, 0 restraints, 686 parameters) converged with R1 (Fo >

- -3 4σ(Fo)) = 2.7 %, wR2 = 7.0 %, S = 1.04. The largest Fourier features were 0.75 and -0.69 e Å .

[Co6Se8(PEt2phen)6][C60]5

Crystals were prepared by slow evaporation of a toluene solution of the precursors. Many crystals were aggregates of individual flakes stuck together on their flat (001) faces. A single crystal (.08 x .05 x .01 mm) was selected, mounted with STP oil treatment and cooled to 100 K on the diffractometer. The diffraction pattern was weak but showed sharp Bragg peaks out to the resolution limit. Since the thin crystal was expected to need accurate absorption correction, a highly redundant data set was collected, approximately 2x the full sphere (21.8x redundancy in -

3m1). Frames were collected at 12.5 s exposures for detector theta = 35 ° and 62.5 s for theta =

161

110 °, which resulted in a 4 day data collection. 101472 reflections were collected (4657 unique,

4030 observed) with R(int) 10.0 % and R(sigma) 3.1 % after absorption correction (Tmax .948,

Tmin .748).

Merging R statistics and suggested the space group was R-3 or R-3m. Superflip proposed R-3m, which gave a well-resolved Co6Se8P6 core, but the density for the phosphine alkyl groups and the C60 molecules was hard to interpret. A refinement in R-3 (merohedrally twinned) gave a more interpretable difference map, but the phosphine was disordered by clockwise/counterclockwise orientation around the c axis. Inspection of this disordered structure showed that it was actually a special position disorder in R-3m. The phosphine geometry was introduced to the R-3m solution as a FRAG fragment in part -1 with DFIX restraints on bonded

C-C distances.

The C60 molecules were apparent in the difference map but specific atomic positions were difficult to locate. Each fullerene was introduced as a rigid fragment from the Idealized Molecular

Geometry Library with a negative part number and an occupancy determined from its Wyckoff position. Subsequent refinement and examination of difference maps showed that each fullerene was disordered over several independent positions; two were modeled in two independent positions and one was modeled in three independent positions. None of the fullerenes was in a position that fit its site symmetry. Due to the extensive disorder and the rigid nature of the C60 molecule, a group isotropic ADP was refined for each fullerene.

A disordered toluene molecule was located in the cluster layer and refined as a rigid fragment in 3 independent positions.

In the final refinement it was possible to release all DFIX restraints on the phosphine ligand.

All ADPs were restrained by RIGU; the disordered toluene was additionally restrained by SIMU.

162

All H atoms were placed in calculated positions and refined with riding coordinates and isotropic

ADPs.

The final refinement (4657 data, 407 parameters, 636 restraints) converged with R1 (Fo >

- -3 4σ(Fo)) = 4.1 %, wR2 = 9.7 %, S = 1.03. The largest Fourier features were 0.61 and -0.46 e Å .

Table 6.1. Selected crystallographic data for Co6Se8(PEt2phen)6 and [Co6Se8(PEt2phen)6][C60]5.

Compound Co6Se8(PEt2phen)6 • C7H8 [Co6Se8(PEt2phen)6][C60]5•2 C7H8

Formula C127H122Co6P6Se8 C434H130P6Co6Se8 MW 2819.32 6514.45 Space group P-1 R-3m a (Å ) 13.1106(2) 16.7992(2) b (Å ) 15.1825(3) 16.7992(2) c (Å ) 16.7649(3) 79.0867(11) α (°) 107.4345(16) 90 β (°) 103.1693(15) 90 γ (°) 107.2058(17) 120 V (Å 3) 2852.06(10) 19329.0(6) Z 1 3 -3 ρcalc (g cm ) 1.641 1.679 T (K) 100 100 λ (Å) 1.54184 1.54184

2θmin, 2θmax 9.814, 143.7 8.258, 143.1 Nref 36565 101472 R(int), R(σ) 0.0316, 0.0312 0.0998, 0.0309 μ(mm-1) 10.77 5.196 Size (mm) 0.14 x 0.03 x 0.02 0.08 x 0.06 x 0.01

Tmax, Tmin 0.821, 0.498 0.968, 0.748 Data 11067 4657 Restraints 0 636 Parameters 686 407

R1(obs) 0.0274 0.0412

wR2(all) 0.0700 0.0967 S 1.037 1.028 Peak, hole (e- Å -3) 0.75, -0.69 0.62, -0.46

163

6.14. X-Ray Structural Analysis of Cluster Oxidation State

We analyzed a set of 35 published and unpublished crystal structures of Co6Se8(PR3)6 clusters in unambiguous neutral, 1+ and 2+ oxidation states. A manuscript is in preparation, but the key findings are summarized in Figure 6.10. One-electron oxidation causes a lengthening of approximately 0.02-0.04 Å in the cluster’s average Co-P bond length. This change is predicted in

n+ 43 a DFT study of the related series Co6Te8(PH3)6 (n = 0-2). A second one-electron oxidation causes further Co-P lengthening and a substantial (0.1 Å ) contraction of the average trans- Co-Co distance.

Several features are noteworthy: Co6Se8(PAr3)6 clusters are an exceptional case due to the high steric demand of the ligand. PAr3-capped clusters are denoted by open points. Three reported structures of neutral clusters revealed a geometry agreeing closely with the majority of 1+ clusters.

All three of these unusual examples had been treated with PLATON SQUEEZE to model disordered solvent. Therefore, we surmise that these 3 structures were probably inadvertently characterized as salts with unknown anions.

[Co6Se8(PEt2phen)6][C60]5 contains a cluster with geometry almost identical to the isolated building block. Specifically, Co6Se8(PEt2phen)6 in its co-crystal with C60 has its average Co-P distance lengthened by 0.009(4) Å and its average Co-Cotrans distance contracted by 0.0025(19) Å .

This demonstrates that the fullerene co-crystal is assembled through van der Waals forces with no charge transfer from cluster to C60.

6.15. Powder X-Ray Diffraction

The collected diffraction patterns are in good agreement with the simulated powder patterns generated from the SCXRD data and there are no indications of amorphous content,

164 confirming the purity of the crystalline phase. Small crystals were evenly dispersed, uncrushed, on a zero-background Si plate and covered with a polycarbonate dome, as the crushing of the crystals led to the destruction of the crystalline lattice as indicated by significant broadening and degradation of the diffraction pattern upon sample grinding. We note that the material

[Co6Se8(PEt2phen)6][C60]5 grows as thin plates that tend to pack with a preferred orientation on the substrate. We note also that the simulated patterns were generated from SCXRD run at 100 K, while the PXRD patterns were collected at room temperature, and that some of the solvates are removed from the crystal lattice under ambient conditions. Taken together, these details account for the differences between the experimental PXRD patterns and the simulated ones.

165

Figure 6.14. PXRD of Co6Se8(PEt2phen)6 simulated from SCXRD data (top) and experimental (bottom).

166

Figure 6.15. PXRD of [Co6Se8(PEt2phen)6][C60]5 simulated from SCXRD data (top) and experimental (bottom). The red asterisks denote the presence of a small fraction of C60 impurity.

6.16. Details on the Raman Spectroscopy

For Raman measurements of [Co6Se8(PEt2phen)6][C60]5, a 532 nm diode-pumped solid- state laser reflects off a 532 nm dichroic beam-splitter in a Nikon Ti/U inverted microscope. A

40x, 0.6 N.A. objective focuses the light to a 1 μm2 spot size. Scattered light passes through a 75- micron pinhole into a Princeton Instruments 0.3-meter Acton Spectrometer with a 1200 g/mm grating and onto a Princeton Instruments PIXIS 400 CCD array detector. The instrument has a 5

167 cm-1 resolution, which is determined from the 546 nm Hg emission line. Exposure times were 300 or 600 seconds.

-1 The Ag(2) pentagonal pinch mode, located at 1469 cm in solid C60, is a useful probe for characterizing metal chalcogenide cluster-C60 materials, because this vibration shifts to lower

-1 44 energy by 6 cm per electron transferred to C60. However, photopolymerization can shift the

-1 -1 44 Ag(2) mode to 1463 cm and 1459 cm , and fullerene triplet state excitation can produce peaks at 1465 cm-1 and 1458 cm-1,44 so these effects must also be considered in such a fullerene-rich material. Since these two effects are absent at low power densities, we perform high and low power measurements on the same crystal location to unambiguously assign the origin of the Ag(2) peak shift in our spectra. At each power, consecutive spectra were measured to assess sample stability over time. The higher power density was ~950 W/cm2. Our low power density was ~40

W/cm2. For these measurements, several spots on sample crystals were subjected to high power first and low power second while others were exposed to the lower power first. The Ag(2) modes were fit using Voigt functions in which the Gaussian width was set to the instrument resolution.

High power measurements produced peaks at 1469 ± 1 cm-1 and 1458.8 ± 0.9 cm-1, while low power measurements showed peaks at 1470.3 ± 0.9 cm-1 and 1464.3 ± 1.9 cm-1. The errors of the peak locations were determined from the standard deviations of fit results for spectra measured on different crystals and different days. Peak positions remained the same for high and low power measurements, regardless of whether the spot on the crystal was first subjected to a high or low power. Full fitting parameters can be found in Table 6.2 for the spectrum in Figure 6.9, with the uncertainties in the table reflecting fit quality. This observation that the peak positions are reversible with laser power, combined with the expectation that polymerization is an irreversible

-1 process leads to the conclusion that the high-power peak at 1458.8 cm is caused by C60 excitation

168 into the triplet state. The low power peak at 1464.3 cm-1 also indicates the presence of the triplet state, which can persist even for power densities of 40 W/cm2.44

Thus, we assign peaks shifts in these samples to C60 triplet state excitation, not charge- transfer. We observe no charge-transfer to the fullerenes from the molecular cluster in these materials. Interestingly, at each power there is a peak at 1470 cm-1, which is likely originating from fullerenes more closely associated with the cluster. The fullerene layer is more likely to exhibit peak shifts to lower energy due to triplet state excitation because the layer more closely resembles solid C60.

Table 6.2. Selected Raman spectroscopy data for [Co6Se8(PEt2phen]6[C60]5. High Power Low Power Peak 1, Peak 2, Peak 3, Peak 1, Peak 2, Peak 3, Hg(8) Ag(2) Ag(2) Hg(8) Ag(2) Ag(2) Energy (cm-1) 1427.0 ± 1458.8 ± 0.4 1468.0 ± 0.2 1427.2 ± 1464.2 ± 0.5 1470.5 ± 0.3 0.3 0.6 Area 20000 ± 22000 ± 25000 ± 8000 ± 11000 ± 10000 ± 1000 3000 2000 1000 2000 2000 Lorentzian 17.6 ± 1.2 10.4 ± 1.0 7.35 ± 0.60 15.8 ± 2.2 6.1 ± 1.0 3.06 ± 0.77 FWHM (cm-1)

6.17. Optical Absorption Measurement

To measure mid-IR absorption spectrum, we used Fourier Transform Infrared (FT-IR) spectroscopy (Thermo Scientific Inc.). Each measurement was carried out on a single crystal flake using transmission microscopic mode. The crystals were transferred onto a ZnSe substrate placing inside a cell with the KBr windows. During measurement and storage, samples were kept free from oxygen and moisture. The absorption spectrum (Figure 6.11) was obtained by averaging 5 different measurements and subtracting the vibronic absorptions from the ligands (Figure 6.14).

169

Figure 6.14. (a) Mid-IR optical absorption spectra of [Co6Se8(PEt2phen)6][C60]5 before (green) and after (red) ligand absorption subtraction. The result was an average from 5 different measurements on a single crystal flake from (b).

The absorption coefficient is described by Elliot’s theory for Wannier excitons as: 45

휇2 2퐸 ћ휔 − 퐸푏 ∞ ћ휔 − 퐸 1 1 훼(ћ휔) = 퐶푉 [∑ 푏 푠푒푐ℎ ( 푗 )] + ∫ 푠푒푐ℎ ( ) 푑퐸 3 8휇푏 ћ휔 푗 훤 퐸 훤 퐸푏 푗 푔 −2휋√ 1 − 3 (퐸 − 퐸푔) 퐸−퐸푔 1 − 푒 ћ The first and second terms describe the Wannier exciton and bulk band-to-band transition,

2 b respectively. μ cv is transition dipole moment, ћω is photon energy, Eb is binding energy, E j = Eg

- Eb/j2 are the energies of transitions to bound states, j is the exciton index, μ is exciton reduced mass.

We used the bulk part of the Elliot’s theory with an additional Gaussian peak to fit the data:

ћ휔−퐸 2 ∞ −( 1) ћ휔 − 퐸 1 1 훼(ћ휔) = 퐴푒 푊 + ∫ 푠푒푐ℎ ( ) 푑퐸 8휇푏 퐸 훤 퐸푏 푔 −2휋√ 1 − 3 (퐸 − 퐸푔) 퐸−퐸푔 ћ 1 − 푒

Where the first term is a Gaussian function with a centered energy E1 and a width W.

The joint valence-band energy-momentum dispersion is given by:

ћ2푘2 퐸 (푘) − 퐸 (푘) = − 푏푘4 + 퐸 푐 푉 2휇 푔

170

The result from Figure 6.11 was obtained using parameters: Eb = 1 meV, Eg = 390 meV, Γ = 44 meV, and 8μb/ћ3 = 0.7

0.5 300 K 260 K 200 K 0.4 160 K 90 K 77 K 0.3

0.2 Absorption

0.1

0.0 0.3 0.4 0.5 0.6 0.7 0.8 Energy (eV)

Figure 6.16. Absorption spectra of [Co6Se8(PEt2phen)6][C60]5 at different temperatures. The results were obtained using FT-IR via microscopic transmission mode. The crystals were transferred onto a ZnSe substrate placing inside a cryostat with two ZnSe windows. The blue-solid curve is a result from filtering out high oscillation frequencies at 77 K. The oscillation is resulted from Fabry-Perot interference pattern caused by two parallel surfaces in a thin crystal.

6.18. References

1. Choi, B.; Yu, J.; Paley, D. W.; Trinh, M. T.; Paley, M. V.; Karch, J. M.; Crowther, A. C.; Lee, C. H.; Lalancette, R. A.; Zhu, X. Y.; Kim, P.; Steigerwald, M. L.; Nuckolls, C.; Roy, X., van der Waals Solids from Self-Assembled Nanoscale Building Blocks. Nano Lett. 2016, 16, 1445-1449. 2. Lee, K.; Choi, B.; La Plante, I. J.; Paley, M.; Zhong, X.; Crowther, A. C.; Owen, J.; Zhu, X.; Roy, X., Two‐Dimensional Fullerene Assembly from an Exfoliated van der Waals Template. Angew. Chem. Int. Edit. 2018, ASAP. 3. Geim, A. K.; Novoselov, K. S., The Rise of Graphene. Nat. Mater. 2007, 6, 183-191. 4. Butler, S. Z.; Hollen, S. M.; Cao, L. Y.; Cui, Y.; Gupta, J. A.; Gutierrez, H. R.; Heinz, T. F.; Hong, S. S.; Huang, J. X.; Ismach, A. F.; Johnston-Halperin, E.; Kuno, M.; Plashnitsa, V. V.; Robinson, R. D.; Ruoff, R. S.; Salahuddin, S.; Shan, J.; Shi, L.; Spencer, M. G.; Terrones, M.; Windl, W.; Goldberger, J. E., Progress, Challenges, and Opportunities in Two-Dimensional Materials Beyond Graphene. ACS Nano 2013, 7, 2898-2926. 5. Xu, M. S.; Liang, T.; Shi, M. M.; Chen, H. Z., Graphene-Like Two-Dimensional Materials. Chem. Rev. 2013, 113, 3766-3798. 171

6. Coleman, J. N.; Lotya, M.; O'Neill, A.; Bergin, S. D.; King, P. J.; Khan, U.; Young, K.; Gaucher, A.; De, S.; Smith, R. J.; Shvets, I. V.; Arora, S. K.; Stanton, G.; Kim, H. Y.; Lee, K.; Kim, G. T.; Duesberg, G. S.; Hallam, T.; Boland, J. J.; Wang, J. J.; Donegan, J. F.; Grunlan, J. C.; Moriarty, G.; Shmeliov, A.; Nicholls, R. J.; Perkins, J. M.; Grieveson, E. M.; Theuwissen, K.; McComb, D. W.; Nellist, P. D.; Nicolosi, V., Two-Dimensional Nanosheets Produced by Liquid Exfoliation of Layered Materials. Science 2011, 331, 568- 571. 7. Shinar, J., Optical and Electronic Properties of Fullerenes and Fullerene-Based Materials. Marcel Dekker Inc.: New York, 2000. 8. Diederich, F.; Gomez-Lopez, M., Supramolecular Fullerene Chemistry. Chem. Soc. Rev. 1 999, 28, 263-277.

9. Kroto, H. W.; Allaf, A. W.; Balm, S. P., C60 - Buckminsterfullerene. Chem. Rev. 1991, 91, 1213-1235. 10. Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K., Two-Dimensional Atomic Crystals. P. Natl. Acad. Sci. 2005, 102, 10451- 10453. 11. Coleman, J. N., Liquid-Phase Exfoliation of Nanotubes and Graphene. Adv. Funct. Mater. 2009, 19, 3680-3695. 12. Konarev, D. V.; Khasanov, S. S.; Otsuka, A.; Maesato, M.; Saito, G.; Lyubovskaya, R. N., A Two-Dimensional Organic Metal Based on Fullerene. Angew. Chem. Int. Edit. 2010, 49, 4829-4832. 13. Konarev, D. V.; Khasanov, S. S.; Otsuka, A.; Ishikawa, M.; Yamochi, H.; Saito, G.; Lyubovskaya, R. N., Formation of Hexagonal Fullerene Layers from Neutral and + ·- Negatively Charged Fullerenes in {(Ph3P)3Au }2(C60 2)(C60)·C6H4Cl2 Containing Gold Cations with the C-3v Symmetry. Inorg. Chem. 2014, 53, 6850-6855.

14. Reddy, C. D.; Yu, Z. G.; Zhang, Y. W., Two-Dimensional van der Waals C60 Molecular Crystal. Sci. Rep. 2015, 5, 12221. 15. Sanchez, L.; Otero, R.; Gallego, J. M.; Miranda, R.; Martin, N., Ordering Fullerenes at the Nanometer Scale on Solid Surfaces. Chem. Rev. 2009, 109, 2081-2091. 16. Hou, J. G.; Yang, J. L.; Wang, H. Q.; Li, Q. X.; Zeng, C. G.; Yuan, L. F.; Wang, B.; Chen, D. M.; Zhu, Q. S., Surface Science - Topology of Two-Dimensional C60 Domains. Nature 2001, 409, 304-305. 17. Bonifazi, D.; Enger, O.; Diederich, F., Supramolecular [60]Fullerene Chemistry on Surfaces. Chem. Soc. Rev. 2007, 36, 390-414. 18. Hornbaker, D. J.; Kahng, S. J.; Misra, S.; Smith, B. W.; Johnson, A. T.; Mele, E. J.; Luzzi, D. E.; Yazdani, A., Mapping the One-Dimensional Electronic States of Nanotube Peapod Structures. Science 2002, 295, 828-831. 172

19. Feng, M.; Zhao, J.; Huang, T.; Zhu, X. Y.; Petek, H., The Electronic Properties of Superatom States of Hollow Molecules. Acc. Chem. Res. 2011, 44, 360-368. 20. Lof, R. W.; Vanveenendaal, M. A.; Koopmans, B.; Jonkman, H. T.; Sawatzky, G. A., Band-Gap, Excitons, and Coulomb Interaction in Solid C60. Phys. Rev. Lett. 1992, 68, 3924-3927. 21. Roy, X.; Schenck, C. L.; Ahn, S.; Lalancette, R. A.; Venkataraman, L.; Nuckolls, C.; Steigerwald, M. L., Quantum Soldering of Individual Quantum Dots. Angew. Chem. Int. Edit. 2012, 51, 12473-12476. 22. Roy, X.; Lee, C. H.; Crowther, A. C.; Schenck, C. L.; Besara, T.; Lalancette, R. A.; Siegrist, T.; Stephens, P. W.; Brus, L. E.; Kim, P.; Steigerwald, M. L.; Nuckolls, C., Nanoscale Atoms in Solid-State Chemistry. Science 2013, 341, 157-160. 23. Lee, S. W.; Shon, Y. S.; Lee, T. R.; Perry, S. S., Structural Characterization and Frictional Properties of C60-Terminated Self-Assembled Monolayers on Au(111). Thin Solid Films 2000, 358, 152-158. 24. Jeong, Y. J.; Yun, D. J.; Jang, J.; Park, S.; An, T. K.; Kim, L. H.; Kim, S. H.; Park, C. E., Solution-Processed n-Type Fullerene Field-Effect Transistors Prepared Using CVD- Grown Graphene Electrodes: Improving Performance with Thermal Annealing. Phys. Chem. Chem. Phys. 2015, 17, 6635-6643. 25. Xiao, W.; Passerone, D.; Ruffieux, P.; Ait-Mansour, K.; Groning, O.; Tosatti, E.; Siegel, J. S.; Fasel, R., C60/Corannulene on Cu(110): A Surface-Supported Bistable Buckybowl- Buckyball Host-Guest System. J. Am. Chem. Soc. 2008, 130, 4767-4771. 26. Pan, G. B.; Liu, J. M.; Zhang, H. M.; Wan, L. J.; Zheng, Q. Y.; Bai, C. L., Configurations of a Calix[8]arene and a C60/Calix[8]arene Complex on a Au(111) Surface. Angew. Chem. Int. Edit. 2003, 42, 2747-2751. 27. Perez, E. M.; Martin, N., Curves Ahead: Molecular Receptors for Fullerenes Based on Concave-Convex Complementarity. Chem Soc Rev 2008, 37, 1512-1519. 28. Kawase, T.; Tanaka, K.; Seirai, Y.; Shiono, N.; Oda, M., Complexation of Carbon Nanorings with Fullerenes: Supramolecular Dynamics and Structural Tuning for a Fullerene Sensor. Angew. Chem. Int. Edit. 2003, 42, 5597-5600. 29. Tremblay, N. J.; Gorodetsky, A. A.; Cox, M. P.; Schiros, T.; Kim, B.; Steiner, R.; Bullard, Z.; Sattler, A.; So, W. Y.; Itoh, Y.; Toney, M. F.; Ogasawara, H.; Ramirez, A. P.; Kymissis, I.; Steigerwald, M. L.; Nuckolls, C., Photovoltaic Universal Joints: Ball-and-Socket Interfaces in Molecular Photovoltaic Cells. Chem. Phys. Chem. 2010, 11, 799-803. 30. Ball, M.; Zhong, Y.; Wu, Y.; Schenck, C.; Ng, F.; Steigerwald, M.; Xiao, S. X.; Nuckolls, C., Contorted Polycyclic Aromatics. Acc. Chem. Res. 2015, 48, 267-276. 31. Grunder, S.; Torres, D. M.; Marquardt, C.; Blaszczyk, A.; Krupke, R.; Mayor, M., Synthesis and Optical Properties of Molecular Rods Comprising a Central Core- 173

Substituted Naphthalenediimide Chromophore for Carbon Nanotube Junctions. Eur. J. Org. Chem. 2011, 478-496. 32. Blanc, E.; Schwarzenbach, D.; Flack, H. D., The Evaluation of Transmission Factors and Their 1st Derivatives with Respect to Crystal Shape Parameters. J. Appl. Crystallogr. 1991, 24, 1035-1041. 33. Clark, R. C.; Reid, J. S., The Analytical Calculation of Absorption in Multifaceted Crystals. Acta Crystallogr. A 1995, 51, 887-897. 34. Version 1.171.37.35, Oxford Diffraction / Agilent Technologies UK Ltd, Yarnton, England. 2014. 35. Sheldrick, G. M., A Short History of SHELX. Acta Crystallogr. A 2008, 64, 112-122. 36. Sheldrick, G. M., Crystal Structure Refinement with SHELXL. Acta Crystallogr. C 2015, 71, 3-8. 37. Palatinus, L.; Chapuis, G., SUPERFLIP - A Computer Program for the Solution of Crystal Structures by Charge Flipping in Arbitrary Dimensions. J. Appl. Crystallogr. 2007, 40, 786-790. 38. Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.; Puschmann, H., OLEX2: A Complete Structure Solution, Refinement and Analysis Program. J. Appl. Crystallogr. 2009, 42, 339-341. 39. Spek, A. L., Structure Validation in Chemical Crystallography. Acta Crystallogr. D 2009, 65, 148-155. 40. Vandersluis, P.; Spek, A. L., Bypass - An Effective Method for the Refinement of Crystal Structures Containing Disordered Solvent Regions. Acta Crystallogr. A 1990, 46, 194-201. 41. Lepage, Y., Missym-1.1 - A Flexible New Release. J. Appl. Crystallogr. 1988, 21, 983- 984. 42. Guzei, I. A., An Idealized Molecular Geometry Library for Refinement of Poorly Behaved Molecular Fragments with Constraints. J. Appl. Crystallogr. 2014, 47, 806-809. 43. Palstra, T. T. M.; Steigerwald, M. L.; Ramirez, A. P.; Kwon, Y. U.; Stuczynski, S. M.; Schneemeyer, L. F.; Waszczak, J. V.; Zaanen, J., Electron Correlations on a Mesoscopic Scale - Magnetic-Properties of Transition-Metal Telluride Cluster Compounds. Phys. Rev. Lett. 1993, 71, 1768-1771.

44. Kuzmany, H.; Matus, M.; Burger, B.; Winter, J., Raman-Scattering in C60 Fullerenes and Fullerides. Adv. Mater. 1994, 6, 731-745. 45. Saba, M.; Cadelano, M.; Marongiu, D.; Chen, F. P.; Sarritzu, V.; Sestu, N.; Figus, C.; Aresti, M.; Piras, R.; Lehmann, A. G.; Cannas, C.; Musinu, A.; Quochi, F.; Mura, A.;

174

Bongiovanni, G., Correlated Electron-Hole Plasma in Organometal Perovskites. Nat. Commun. 2014, 5, 5049.

175

Chapter 7. Layered Chevrel phases of Re and Mo

7.1. Introduction

The excitement generated by 2D materials has been fueled by the endless possibilities they offer for fundamental research,1, 2 and for novel technologies in nanoelectronics,3, 4 catalysis,5-8 sensing,9 and energy production and storage.10, 11 The realization of many of these promises, however, hinges on the development of effective strategies to control their surface chemistry, which plays a prominent role in determining their physical and chemical properties. This remains a major challenge for nearly all existing 2D materials. Graphene is unique in this respect: its surface can be functionalized, opening up many new technological frontiers, but this process invariably disrupts its network of C=C bonds and introduces defects in both its atomic and electronic structures.12 Transition metal dichalcogenides (TMDCs), by contrast, have non-reactive surfaces: their atomic structure is constituted of chemically inaccessible metal atoms, sandwiched between inert layers of bridging chalcogens.13 As a result, the few existing surface functionalization methods are almost exclusively focused on MoS2 and are severely limited because they typically yield low surface coverage density, use harsh reaction conditions, break metal-chalcogen bonds, and/or rely on weak non-covalent interactions.14-17

In this section, we describe a single-layer 2D rhenium selenochloride semiconductor with a unique hierarchical structure and a surface that can be chemically modified through ligand substitution. We demonstrate the first example of such tunability by installing cyanide ligands on the surface of the nanosheets and show that the structural integrity is maintained. We also introduce a layered molybdenum sulfurbromide semiconductor that contains a similar hierarchical structure as the rhenium selenochloride material and briefly describe its prospects for use as a low- dimensional material. 176

7.2. Synthesis and Characterization of Bulk Crystals

Figure 7.1. (a) Edge-on view of Re6Se8Cl2. (b) Top-view of a single layer. Color code: Rhenium, blue; selenium, red; chlorine, green. (c) Optical microscope image of as grown Re6Se8Cl2 crystals. (d) AFM image of mechanically exfoliated flakes with a thickness of 60-150 nm.

Figure 7.1a shows the layered van der Waals structure of the 2D semiconductor Re6Se8Cl2, as determined by single crystal X-ray diffraction (SCXRD). This compound is a 2D structural

18, 19 analogue of the 3D superconductor Chevrel phase. In the basal plane of the Re6Se8Cl2 structure, neighboring clusters assemble into an array of equilateral parallelograms, with each cluster being connected to each of its four in-plane neighbors by two Re-Se linkages. (Figure 7.1b). Each cluster is made of a Re6 octahedron enclosed in a Se8 cube and is capped by two terminal Cl ligands in the trans positions. The weak van der Waals contacts between the layers and strong in-plane bonding give this material a robust 2D character. First reported by Sergent and coworkers,20 microcrystalline Re6Se8Cl2 is synthesized by heating a stoichiometric mixture of Re, Se, and ReCl5 to 1100 ºC in a fused silica tube sealed under vacuum. Millimeter-sized single crystals, shown in

Figure 7.1c, are grown via chemical vapor transport in a temperature gradient of 970-920 ºC using

177

ReCl5 as the transporting agent. We use SCXRD to determine the structure of the compound and powder X-ray diffraction (PXRD) to confirm that the sample is phase pure (Figure 7.2).

Figure 7.2. PXRD pattern of Re6Se8Cl2 simulated from SCXRD data (green) and measured experimentally (black).

Another layered van der Waals material with an analogous Chevrel phase structure is

21 Mo6S3Br6, which was first reported by Sergent and co-workers. Figure 7.3 shows the molecular structure of Mo6S3Br6. The cohesion of the layers is achieved by van der Waals contacts between the Br atoms, which gives this material a strong 2D character (Figure 7.3a). In contrast to

Re6Se8Cl2, this material contains an additional degree of anisotropy within the plane of the crystal.

One layer is built from chains along the c-axis in which the cluster units are slightly tilted to accommodate two inner sulfur atoms in the apical positions of the clusters, and along the b-axis, the chains are linked by Mo-S bonds (Figure 7.3b). Crystalline Mo6S3Br6 is synthesized by heating a mixture of Mo, S, Mo6Br12, and Nb to 1175 ºC in a fused silica tube sealed under vacuum.

Millimeter-sized single crystals, shown in Figure 7.3c, are grown when the temperature is reduced to room temperature slowly. Nb reacts with the reaction mixture at high temperatures to presumably make NbBr5 that acts as a transporting agent for the Mo6S3Br6 crystal growth. We

178 obtain orange, needle-like crystals of NbBr5 along the edges of the quartz tube that we later remove with water and acetone. We use PXRD to confirm that the sample is phase pure (Figure 7.4).

Figure 7.3. (a) Edge-on view of Mo6S3Br6. (b) Top-view of a single layer. Color code: Molybdenum, purple; sulfur, yellow; bromine, green. (c) Optical microscope image of as grown Mo6S3Br6 crystals. (d) AFM image of mechanically exfoliated flakes with a thickness of 3-8 nm.

Figure 7.4. PXRD pattern of Mo6S3Br6 simulated from SCXRD data (pink) and measured experimentally (black).

7.3. Mechanical Exfoliation of Crystals

Consistent with their layered van der Waals nature, crystals of Re6Se8Cl2 and Mo6S3Br6 can be mechanically exfoliated. Figures 7.1d and 7.3d display atomic force microscopy (AFM)

179 images of mechanically exfoliated flakes. As expected, the van der Waals interactions between the

Br atoms holding the layers together in Mo6S3Br6 is weaker than those between Cl atoms in

Re6Se8Cl2, and we observe that it is significantly easier to mechanically exfoliate the Mo6S3Br6 to thinner layers. The thinnest flakes of Mo6S3Br6 we have produced are ~2 nm in height, corresponding to about two layers. Physical characterizations of the 2D nanosheets of Mo6S3Br6 are currently underway. On the other hand, the thinnest flakes we have achieved for Re6Se8Cl2 using the mechanical exfoliation techniques are ~10 nm in height, and thus this mechanical process does not produce Re6Se8Cl2 monolayers or sheets in the 2D regime. To achieve this, we have devised a technique for reactive exfoliation. Below, we focus on the chemical intercalation/exfoliation technique for Re6Se8Cl2.

7.4. Chemical Intercalation and Exfoliation of Re6Se8Cl2

22, 23 As is the case with many TMDCs, alkali metals can be intercalated into Re6Se8Cl2.

+ Results for the intercalation of Li ions are presented in Figure 7.5a. Crystals of Re6Se8Cl2 are suspended in a ~0.1 M solution of n-butyllithium in hexanes at -40 °C. After 24 h, the suspension is decanted, and the solid is washed with hexanes and dried under vacuum. The intercalation process is monitored by PXRD (Figure 7.5b). For pristine Re6Se8Cl2, the interlayer spacing, calculated from the position of the (001) diffraction peak, is d001 ~8.1 Å . After intercalation, d001

~8.4 Å , indicating a ~0.3 Å increase of the interlayer spacing. This is consistent with the insertion

+ of Li ions between the Re6Se8Cl2 layers and is the first report of intercalation in such a compound.

Elemental analysis using energy-dispersive X-ray spectroscopy (EDS) reveals that the Re6Se8Cl2 composition remains unchanged (Table 7.1). Inductively coupled plasma optical emission spectrometry (ICP-OES) is used to determine the Li composition in the intercalated compound,

180

Li1.6Re6Se8Cl2. The intercalation is reversible: when we treat bulk Li1.6Re6Se8Cl2 crystals with Br2, the (001) peak shifts back to its original position and the peaks in the range 13-16º reappear, as anticipated for the deintercalated structure.

Figure 7.5. (a) Schematic of the chemical intercalation/liquid exfoliation process. Colors scheme same as in Figure 7.1; Lithium, orange. (b) PXRD pattern of Re6Se8Cl2 (black); Li1.6Re6Se8Cl2 (red); and deintercalated Re6Se8Cl2 (blue). (c) Solutions obtained when pristine Re6Se8Cl2 (left), and Li1.6Re6Se8Cl2 are soaked in NMF. Tyndall scattering using a green laser beam indicates the solutes are nanoparticles. When pristine Re6Se8Cl2 crystals are soaked in NMF for 8 months, the solvent is colorless and shows no Tyndall scattering. (d) Electronic absorption spectra of a Re6Se8Cl2 crystal (black) and a solution of MLs in NMF (red).

The intercalation compound Li1.6Re6Se8Cl2 dissolves readily in N-methylformamide

(NMF). Dissolution results in exfoliation of the ionic structure, forming a dark brown solution

+ containing solvated Re6Se8Cl2 monolayers (MLs), which are presumably associated with Li cations (Figure 7.5c). For reference, pristine Re6Se8Cl2 does not dissolve in NMF, even after soaking for more than 8 months. To prepare a solution of MLs, Li1.6Re6Se8Cl2 crystals are suspended in NMF; a dark brown solution is obtained after a few hours, which is centrifuged to remove undissolved solid. Sonication, which can damage the structure, is unnecessary, and the

181 solution is stable for months when kept in an inert atmosphere. Tyndall scattering demonstrates that the MLs are at least on the scale of tens of nanometers (Figure 7.5c). The electronic spectrum of the solution, featuring an absorption onset at ~1.47 eV (Figure 7.5d, red line), agrees well with

24 that of the parent compound Re6Se8Cl2 (Figure 7.5d, black line).

Figure 7.6. (a) AFM image of MLs dropcast on a sapphire substrate. (b) AFM image of a few MLs and height profile along the black line. (c) High resolution HAADF TEM micrograph of a ML and (d) corresponding SAED pattern. (e) High resolution HAADF TEM micrograph showing a tear in the edge of a ML. (f-h) EDS mapping of Cl, Re, and Se in the ML, respectively.

To verify the structure of the MLs, an aliquot of the solution is dropcast onto a sapphire substrate, and the solvent is allowed to evaporate. A representative AFM image of the deposited material (Figure 7.6a) reveals a high coverage density of plate-like micron-sized nanosheets

(Figure 7.7). A closer examination of these nanosheets (Figure 7.6b) shows that the thickness of the first layer is ~1.7 nm. Some MLs are superimposed, and the step height histograms (Figure

7.8) of the AFM images indicates that subsequent layers are ~1.5 nm thick, corresponding to the height of one ML. We attribute the difference between the height of the ML measured by AFM and that calculated from the SCXRD data (~1 nm) to the AFM tip-surface interactions, which

182 inherently overestimate the step height, as well as to the presence of cations and solvent molecules on the top surface of the ML and between the ML and the substrate. Comparable differences in height have been reported for other 2D materials prepared by liquid exfoliation.25-27

Figure 7.7. Optical microscope image of exfoliated nanosheets deposited on a SiO2/Si substrate. High density coverage of transparent blue rectangular areas corresponds to thin, exfoliated MLs.

Figure 7.8. (a) AFM image of a few exfoliated MLs on a sapphire substrate, and (b) corresponding step height histogram. (c) Large area AFM image of MLs on a sapphire substrate, and (d) corresponding step height histogram.

183

The structure of the ML is investigated by transmission electron microscopy (TEM).

Figure 7.6c,d presents a high resolution TEM micrograph and corresponding selected area electron diffraction (SAED) pattern of a dropcast ML, respectively. Figure 7.6c reveals the ordered oblique array of individual Re6 octahedra, which appear as white dots in the high resolution high angle annular dark field (HAADF) micrograph. The diffraction peaks in the SAED pattern correspond to the (010) and (100) reflections when the electron beam is aligned along the c-axis.

The interplanar spacings calculated from the SAED pattern (d010 = 6.4 Å and d100 = 6.3 Å ) match well with those obtained from PXRD and SCXRD (Figure 7.2). Figure 7.6e presents a high resolution HAADF TEM micrograph of a tear at the edge of a ML. EDS elemental mapping of an exfoliated flake shows uniform distribution of Re, Se, and Cl on the nanosheet (Figure 7.6f-h).

The Re:Se:Cl mass ratio, 62:35:3, matches well with the composition of the parent compound

(theoretical: 61:35:4).

7.5. Surface Functionalization of Monolayers

From the crystal structure of Re6Se8Cl2, we can infer that Cl atoms cap the surface of the

2D semiconductor. This is corroborated by recent scanning tunneling microscopy (STM)

24 measurements performed by our team on mechanically exfoliated Re6Se8Cl2 crystals. Previous studies on isolated Re6Q8L6 (Q = S, Se, Te) molecular clusters have shown that the clusters can be decorated with a wide range of ligands using ligand substitution chemistry.28, 29 This observation is crucial to the prospect of functionalizing the surface of Re6Se8Cl2 MLs.

184

Figure 7.9. (a) Schematic of the surface functionalization reaction of MLs supported on a substrate. Color as in Figure 7.1a; Carbon, grey; nitrogen, aqua. (b) Raman spectra (2000-2400 cm–1 region) of MLs on sapphire substrate (green), MLs after reaction with TMSCN (red), and pristine sapphire substrate exposed to TMSCN in air (blue). The signal from the sapphire background has been subtracted in all the spectra. (c) AFM image and (d) TEM micrograph (Inset: corresponding SAED pattern) of MLs after surface functionalization.

As a first demonstration, we install cyanide groups on the surface of the MLs (Figure 7.9a).

The MLs, supported on a sapphire substrate, are reacted in toluene with an excess of trimethylsilyl cyanide (TMSCN), which is expected to substitute –CN for –Cl.30, 31 After 24 h at 80 ºC, the sample is thoroughly rinsed with toluene and acetone. Comparing the micro-Raman spectra of the MLs before and after the reaction attests to the surface functionalization. The Raman spectrum of the pristine MLs (Figure 7.9b, green) contains no peaks in the spectral region 2000-2400 cm–1, where we expect the CN stretching vibrational modes. The MLs after substitution show two broad peaks at ~2140 and ~2230 cm–1 (Figure 7.9b, red). The main peak at 2140 cm–1 is a clear signature of

2+ cyanide attachment to the MLs as it agrees well with reported CN frequencies for [Re6Se8] molecular clusters decorated with –CN, in which  back-donation is negligible.32, 33 TMSCN, which has a strong, sharp peak at 2190 cm–1, is not detected on the surface. The second broad peak

185 at 2230 cm–1 closely resembles the one observed in the Raman spectrum of a pristine sapphire substrate exposed to TMSCN in air (Figure 7.9b, blue).

Figure 7.10. (a) AFM image of a few MLs on a sapphire substrate after reaction with TMSCN and (b) corresponding step height histogram. (c) Large area AFM image of MLs on a sapphire substrate after reaction with TMSCN and (d) corresponding step height histogram.

Crucially, the surface functionalization reaction preserves the [Re6Se8] structure of the ML.

Figure 7.7c shows an AFM image of a few MLs after the TMSCN treatment. The step height histogram (Figure 7.10) confirms that the thickness of the MLs is similar to that of the pristine

MLs. The ML Raman modes in the 100-400 cm–1 spectral region, which are characteristic of the

[Re6Se8] bonding, are essentially unchanged before and after functionalization (Figure 7.11).

Carrying out the reaction on MLs supported on a TEM grid enables direct imaging and diffraction of the resulting nanosheets (Figure 7.9d). The SAED pattern of the functionalized 2D

186 semiconductor matches with that of the pristine ML, confirming the retention of the crystalline

[Re6Se8] structure. Furthermore, we note that the surface functionalization of MLs with –CN ligands results in a 30-40 meV redshift of the lowest energy absorption peak (Figure 7.12). Larger shifts should be achievable with more strongly interacting ligands.

Figure 7.11. Raman spectrum in the 100-400 cm–1 region of the MLs before and after treatment with TMSCN.

Figure 7.12. (a) Macroscopic and (b) microscopic electronic absorption spectra of MLs deposited on sapphire substrate before (blue) and after (orange) reaction with TMSCN. 187

7.6. Conclusions

The ability to chemically functionalize atomically thin semiconductors opens up a host of new opportunities for 2D materials, both as a way to tune their physical properties and as a platform to build novel structures and devices. Traditional 2D materials are limited in this respect because their covalent modification necessitates breaking intralayer bonds. The unique hierarchical structures of the Re6Se8Cl2 monolayers enables a new approach to functionalize the surface of this

2D material through ligand substitution. Importantly, the substitution reaction does not affect the intralayer bonding. These results chart a clear path to attaching a wide range of functional ligands using the traditional methods of coordination chemistry.

7.7. General Synthesis Information

All reactions and sample preparations were carried out under inert atmosphere in a nitrogen-filled glovebox, unless otherwise noted. Re (99.99%), Se (99.999%), ReCl5 (99.9%), Mo

(99.95 %), and S (99.5%) were purchased from Alfa Aesar. Nb (99.8%) was purchased from Strem

Chemicals. All other reagents and solvents were purchased from Sigma Aldrich. Dry and deoxygenated solvents were prepared by elution through a dual-column solvent system (MBraun

SPS). N-methylformamide (NMF) was dried using calcium hydride and distilled under vacuum.

34 Mo6Br12·2H2O was prepared according to published protocol and dried under vacuum at 275 °C for 5 days to yield dehydrated Mo6Br12.

188

7.8. Synthetic Procedures

Re6Se8Cl2

Re (330 mg, 1.80 mmol), Se (190 mg, 2.41 mmol), and ReCl5 (200 mg, 0.55 mmol) were ground with a mortar and pestle and pressed into a pellet. The pellet was sealed in a quartz tube of approximately 30 cm in length under vacuum, heated to 1100 °C at 1 °C/min in a tube furnace, and held at 1100 °C for 3 days. When cooled to RT, polycrystalline Re6Se8Cl2 was obtained. Large millimeter-sized single crystals were grown via chemical vapor transport in a temperature gradient of 970-920 °C using ReCl5 as the transporting agent. Excess transport agent was removed from the part of the quartz tube where single crystals have grown by using a gradient of 300-20 °C to condense the liquid to the cooler end.

Mo6S3Br6

Mo (70 mg, 0.72 mmol), S (25 mg, 0.78 mmol), Mo6Br12 (33 mg, 0.88 mmol), and Nb (6 mg, 0.06 mmol) were ground with a mortar and pestle and pressed into a pellet. The pellet was sealed in a quartz tube of approximately 20 cm in length under vacuum, heated to 1175 °C at

1 °C/min in a box furnace, held at 1175 °C for 3 days, and cooled to RT at 0.5 °C/min When cooled to RT, large millimeter-sized, rod-shaped crystals were grown in place of the pellet. Orange, needle-like crystals of NbBr5 had grown along the sides of the quartz tube, which were removed by rinsing the tube with water and acetone in air.

Li1.6Re6Se8Cl2

Microcrystalline powder of Re6Se8Cl2 was suspended in 10 mL of hexanes and cooled to -

40 °C. To the suspension, n-butyllithium (2.5 M in hexanes) was added to obtain a ~0.1 M solution.

189

The reaction was held at -40 °C for 24 h. The suspension was decanted, washed with hexanes, and dried under vacuum for 12 h.

7.9. Instrumentation

Electronic Absorption

Solution-phase electronic absorption spectrum was taken on a Cary 60 UV-Vis spectrophotometer (Agilent Technologies). Li1.6Re6Se8Cl2 was dissolved in NMF. The dark brown solution was centrifuged, diluted with NMF and loaded in a 1 cm quartz cuvette sealed under nitrogen. The spectrum was taken following a recording of the background spectrum. The macroscopic absorption spectra of MLs on sapphire substrates were acquired by measuring the transmittance with a Shimadzu UV-1800 spectrophotometer. For the microscopic absorption of

MLs, samples were mounted onto Montana Instruments Cryo-Optic X-Plane, and the reflectance contrast was measured using tungsten-halogen broadband light source, spectrally resolved with

Princeton Instruments IsoPlane-320 and PyLoN-IR. Both set of measurements were performed in room temperature and air.

ICP-OES

Lithium content in the intercalated crystals of Re6Se8Cl2 was determined using ICP-OES performed by Robertson Microlit Laboratories (Ledgewood, NJ).

X-ray Diffraction

Single crystal X-ray diffraction data was collected on an Agilent SuperNova diffractometer using mirror-monochromated Mo Kα radiation. The crystal was mounted using a MiTeGen

190

MicroMount cooled to 100 K with an Oxford-Diffraction Cryojet system. Data collection, integration, scaling (ABSPACK) and absorption correction (face-indexed Gaussian integration35 or numeric analytical methods36 were performed in CrysAlisPro.37 Structure solution was performed using ShelXS,38 ShelXT,39 or SuperFlip.40 Subsequent refinement was performed by full-matrix least-squares on F2 in ShelXL.38 Olex241 was used for viewing and to prepare CIF files.

Details of crystallographic data and parameters for data collection and refinement are in Table 7.2.

Powder X-ray diffraction data were collected on a PANalytical X’Pert3 Powder diffractometer. Small crystals were evenly dispersed, uncrushed, on a zero-background Si plate and covered with a Kapton polyimide thin film for air-sensitive intercalated samples. The Kapton film has a broad amorphous peak from 15 to 30º, and this background was subtracted using

HighScore Plus.

AFM

AFM images and height profiles were acquired in PeakForce QNM in tapping mode using a Bruker Dimension FastScan AFM under ambient conditions.

Energy Dispersive X-Ray Spectroscopy

EDS of the bulk powder samples were performed on FEI Helios NanoLab 660 SEM. The samples were deposited on carbon tape and evacuated in vacuo. EDS mapping of the MLs was performed on FEI Titan Themis 200 kV TEM available at the Imaging Suite in Advanced Science

Research Center at the City University of New York. The MLs dissolved in NMF were dropcast onto a copper grid with carbon backing and the solvent was evaporated in vacuo for 12 h.

191

Transmission Electron Microscopy

High resolution TEM and SAED were performed on FEI Talos F200X. The MLs dissolved in NMF were dropcast onto a copper grid with carbon backing and the solvent was evaporated in vacuo for 12 h. We note that the error in electron diffraction from TEM is < 1%.

Raman Spectroscopy

Raman spectra of the MLs were acquired on a Renishaw InVia micro-Raman Instrument.

The MLs dissolved in NMF were dropcast onto sapphire substrates and the solvent was evaporated in vacuo for 12 h at 80 °C. The substrate was rinsed thoroughly with toluene and acetone. The samples were pre-exposed to laser irradiation for 10-30 mins before measurements to minimize drastic changes in baseline, as residual solvents evaporate from irradiation during measurements.

A 532 nm diode laser and 2400 g/mm grating was used to collect Raman scattered light. The laser light was focused onto MLs with 50x objective, and the spot size was ~5 μm in diameter to accumulate greater Raman signal without damaging the sample. To detect Raman scattering from the cyanide stretch, we acquired Raman spectra in 2000-2400 cm-1 region for 5 hours for pristine and TMSCN-treated MLs, and for 3 hours for the control pristine sapphire substrate exposed to

TMSCN. For the 100-400 cm-1 region, we measured spectra for 5 minutes. All spectra were intensity-calibrated using reference white-light source to remove any artifacts from the instrument.

7.10. Additional Details on the Raman Data

–1 The CN stretching mode for TMSCN is at 2190 cm . This peak is not observed in the

Raman spectrum of MLs after reaction with TMSCN.42, 43 When a control pristine sapphire substrate is exposed to TMSCN, a broad, asymmetrical peak at ~2230 cm–1 is observed in the

192 resulting Raman spectrum (Figure 7.9b, blue). The same peak is observed in the Raman spectrum of MLs after reaction with TMSCN (Figure 7.9b, red). We attribute this peak to the presence of adsorbed and/or decomposed TMSCN on the sapphire surface. While we could not find any published spectroscopic study for the adsorption/reaction of TMSCN on Al2O3 surface, closely related molecules such as HCN and MeCN are known to adsorb and react on Al2O3 and SiO2, causing the vibrational peaks in the spectral region 2000-2200 cm–1 to broaden and shift to higher wavenumbers.44-50

7.11. Additional Details on the Electronic Absorption Data

The macroscopic absorption spectra are fitted using a cubic baseline and a single Gaussian

(Figure 7.12a). The peak position of the Gaussian at the absorption edge red-shifts by 27 ± 2 meV upon TMSCN treatment, from 1.695 to 1.668 eV. Similarly, the microscopic absorption spectra are fitted using a linear baseline and a single Gaussian (Figure 7.12b). The peak position of the

Gaussian at the absorption edge red-shifts by 40 ± 2 meV upon TMSCN treatment, from 1.670 to

1.630 eV. The two independent sets of measurements clearly show that surface functionalization of Re6Se8Cl2 MLs with -CN shifts the absorption edge by ~30-40 meV.

7.12. Additional Details on EDS

Because the samples are deposited on carbon tape, there is background signal from carbon and oxygen. The mass percentages were normalized where %Re + %Se + %Cl = 100 %. At least

5 sample spots were averaged.

193

Table 7.1. Chemical compositions of Re6Se8Cl2 and alkali-doped Re6Se8Cl2 as determined by EDS Mass % Re Mass % Se Mass % Cl Theoretical 61.4 34.7 3.9 Pristine Re6Se8Cl2 62 ± 1 34 ± 1 4 ± 1 Li1.6Re6Se8Cl2 62 ± 1 34 ± 1 4 ± 1 Li1.6Re6Se8Cl2 treated with Br2 62 ± 1 34 ± 1 4 ± 1

7.13. Structural Determination

Structure solution and space group assignment were performed in ShelXT with no difficulty. Re6Se8Cl2 was solved and refined routinely in P-1.

Table 7.2. Selected crystallographic data for Re6Se8Cl2.

Compound Re6Se8Cl2

Formula Re6Se8Cl2

MW 1819.78 Space group P-1 a (Å ) 6.5785(7) b (Å ) 6.6199(7) c (Å ) 8.8013(9) α (°) 76.706(9) β (°) 70.201(9) γ (°) 86.371 (9) V (Å 3) 350.92(7) Z 1 -3 ρcalc (g cm ) 8.611

T (K) 100 λ (Å) 0.71073 2θmin, 2θmax 6.828, 52.742 Nref 1428 R(int), R(σ) 0.0376, 0.0311 μ(mm-1) 72.619 Size (mm) 0.042 x 0.029 x 0.008

Tmax, Tmin 0.108, 0.565 Data 1428

Restraints 0 Parameters 74 R1(obs) 0.0221

wR2(all) 0.0388 S 1.105 Peak, hole (e- Å -3) 1.21, -1.54

194

7.14. References

1. Novoselov, K. S.; Mishchenko, A.; Carvalho, A.; Neto, A. H. C., 2D Materials and van der Waals Heterostructures. Science 2016, 353, 9439.

2. Gong, C.; Li, L.; Li, Z. L.; Ji, H. W.; Stern, A.; Xia, Y.; Cao, T.; Bao, W.; Wang, C. Z.; Wang, Y. A.; Qiu, Z. Q.; Cava, R. J.; Louie, S. G.; Xia, J.; Zhang, X., Discovery of Intrinsic Ferromagnetism in Two-Dimensional van der Waals Crystals. Nature 2017, 546, 265-269.

3. Wang, Q. H.; Kalantar-Zadeh, K.; Kis, A.; Coleman, J. N.; Strano, M. S., Electronics and Optoelectronics of Two-Dimensional Transition Metal Dichalcogenides. Nat. Nanotechnol. 2012, 7, 699-712.

4. Fiori, G.; Bonaccorso, F.; Iannaccone, G.; Palacios, T.; Neumaier, D.; Seabaugh, A.; Banerjee, S. K.; Colombo, L., Electronics Based on Two-Dimensional Materials. Nat. Nanotechnol. 2014, 9, 768-779.

5. Jaramillo, T. F.; Jorgensen, K. P.; Bonde, J.; Nielsen, J. H.; Horch, S.; Chorkendorff, I., Identification of Active Edge Sites for Electrochemical H2 Evolution from MoS2 Nanocatalysts. Science 2007, 317, 100-102.

6. Deng, D. H.; Novoselov, K. S.; Fu, Q.; Zheng, N. F.; Tian, Z. Q.; Bao, X. H., Catalysis with Two-Dimensional Materials and their Heterostructures. Nat. Nanotechnol. 2016, 11, 218-230.

7. Voiry, D.; Yang, J.; Chhowalla, M., Recent Strategies for Improving the Catalytic Activity of 2D TMD Nanosheets Toward the Hydrogen Evolution Reaction. Adv. Mater. 2016, 28, 6197-6206.

8. Lu, Q. P.; Yu, Y. F.; Ma, Q. L.; Chen, B.; Zhang, H., 2D Transition-Metal Dichalcogenide Nanosheet-Based Composites for Photocatalytic and Electrocatalytic Hydrogen Evolution Reactions. Adv. Mater. 2016, 28, 1917-1933.

9. Perkins, F. K.; Friedman, A. L.; Cobas, E.; Campbell, P. M.; Jernigan, G. G.; Jonker, B. T., Chemical Vapor Sensing with Monolayer MoS2. Nano Lett. 2013, 13, 668-673.

10. Bonaccorso, F.; Colombo, L.; Yu, G. H.; Stoller, M.; Tozzini, V.; Ferrari, A. C.; Ruoff, R. S.; Pellegrini, V., Graphene, Related Two-Dimensional Crystals, and Hybrid Systems for Energy Conversion and Storage. Science 2015, 347, 1246501.

11. Pomerantseva, E.; Gogotsi, Y., Two-Dimensional Heterostructures for Energy Storage. Nat. Energy 2017, 2, 17089.

12. Georgakilas, V.; Otyepka, M.; Bourlinos, A. B.; Chandra, V.; Kim, N.; Kemp, K. C.; Hobza, P.; Zboril, R.; Kim, K. S., Functionalization of Graphene: Covalent and Non- Covalent Approaches, Derivatives and Applications. Chem. Rev. 2012, 112, 6156-6214.

195

13. Chhowalla, M.; Shin, H. S.; Eda, G.; Li, L. J.; Loh, K. P.; Zhang, H., The Chemistry of Two-Dimensional Layered Transition Metal Dichalcogenide Nanosheets. Nat. Chem. 2013, 5, 263-275.

14. Presolski, S.; Pumera, M., Covalent Functionalization of MoS2. Mater. Today 2016, 19, 140-145.

15. Voiry, D.; Goswami, A.; Kappera, R.; Silva, C. D. C. E.; Kaplan, D.; Fujita, T.; Chen, M. W.; Asefa, T.; Chhowalla, M., Covalent Functionalization of Monolayered Transition Metal Dichalcogenides by Phase Engineering. Nat. Chem. 2015, 7, 45-49.

16. Lei, S. D.; Wang, X. F.; Li, B.; Kang, J. H.; He, Y. M.; George, A.; Ge, L. H.; Gong, Y. J.; Dong, P.; Jin, Z. H.; Brunetto, G.; Chen, W. B.; Lin, Z. T.; Baines, R.; Galvao, D. S.; Lou, J.; Barrera, E.; Banerjee, K.; Vajtai, R.; Ajayan, P., Surface Functionalization of Two- Dimensional Metal Chalcogenides by Lewis Acid-Base Chemistry. Nat. Nanotechnol. 2016, 11, 465-471.

17. Nguyen, E. P.; Carey, B. J.; Ou, J. Z.; van Embden, J.; Della Gaspera, E.; Chrimes, A. F.; Spencer, M. J. S.; Zhuiykov, S.; Kalantar-Zadeh, K.; Daeneke, T., Electronic Tuning of 2D MoS2 through Surface Functionalization. Adv. Mater. 2015, 27, 6225-6229.

18. Fischer, O., Chevrel Phases - Superconducting and Normal State Properties. Appl. Phys. 1978, 16, 1-28.

19. Chevrel, R.; Hirrien, M.; Sergent, M., Superconducting Chevrel Phases - Prospects and Perspectives. Polyhedron 1986, 5, 87-94.

20. Leduc, L.; Perrin, A.; Sergent, M., The Structure of Hexarhenium Dichloride and Octaselenide, Re6Se8Cl2 - A Bidimensional Compound of Re6 Octahedral Clusters. Acta Crystallogr. C 1983, 39, 1503-1506.

21. Perrin, C.; Potel, M.; Sergent, M., Mo6Br6S3 - New Bidimensional Compounds with Mo6 Octahedric Clusters. Acta Crystallogr. C 1983, 39, 415-418.

22. Nicolosi, V.; Chhowalla, M.; Kanatzidis, M. G.; Strano, M. S.; Coleman, J. N., Liquid Exfoliation of Layered Materials. Science 2013, 340, 1226419.

23. Zheng, J.; Zhang, H.; Dong, S. H.; Liu, Y. P.; Nai, C. T.; Shin, H. S.; Jeong, H. Y.; Liu, B.; Loh, K. P., High Yield Exfoliation of Two-Dimensional Chalcogenides Using Sodium Naphthalenide. Nat. Commun. 2014, 5, 2995.

24. Zhong, X.; Lee, K.; Choi, B.; Meggiolaro, D.; Liu, F.; Nuckolls, C.; Pasupathy, A.; De Angelis, F.; Batail, P.; Roy, X.; Zhu, X., Superatomic Two-Dimensional Semiconductor. Nano Lett. 2018, 18, 1483-1488.

25. Eda, G.; Yamaguchi, H.; Voiry, D.; Fujita, T.; Chen, M. W.; Chhowalla, M., Photoluminescence from Chemically Exfoliated MoS2. Nano Lett. 2011, 11, 5111-5116. 196

26. Shearer, C. J.; Slattery, A. D.; Stapleton, A. J.; Shapter, J. G.; Gibson, C. T., Accurate Thickness Measurement of Graphene. Nanotechnology 2016, 27, 125704.

27. Backes, C.; Smith, R. J.; McEvoy, N.; Berner, N. C.; McCloskey, D.; Nerl, H. C.; O'Neill, A.; King, P. J.; Higgins, T.; Hanlon, D.; Scheuschner, N.; Maultzsch, J.; Houben, L.; Duesberg, G. S.; Donegan, J. F.; Nicolosi, V.; Coleman, J. N., Edge and Confinement Effects Allow in situ Measurement of Size and Thickness of Liquid-Exfoliated Nanosheets. Nat. Commun. 2014, 5, 4576.

4- 28. Kim, Y.; Fedorov, V. E.; Kim, S. J., Novel Compounds Based on [Re6Q8L6] (Q = S, Se, Te; L = CN, OH) and Their Applications. J. Mater. Chem. 2009, 19, 7178-7190.

29. Gabriel, J. C. P.; Boubekeur, K.; Uriel, S.; Batail, P., Chemistry of Hexanuclear Rhenium Chalcohalide Clusters. Chem. Rev. 2001, 101, 2037-2066.

30. Suginome, M.; Yamamoto, A.; Sasaki, T.; Murakami, M., Reactions of Cyanoboranes with a Palladium-PMe3 Complex: Mechanism for the Catalytic Cyanoboration of Alkynes. Organometallics 2006, 25, 2911-2913.

31. Hussain, M.; Kohser, S.; Janssen, K.; Wartchow, R.; Butenschon, H., The First η5- Indenylnickel Chelate Complexes with a Pendant Phosphane Tether. Organometallics 2009, 28, 5212-5221.

32. Bennett, M. V.; Beauvais, L. G.; Shores, M. P.; Long, J. R., Expanded Prussian Blue 3-/4- Analogues Incorporating [Re6Se8(CN)6] Clusters: Adjusting Porosity via Charge Balance. J. Am. Chem. Soc. 2001, 123, 8022-8032.

33. Gray, T. G.; Holm, R. H., Site-Differentiated Hexanuclear Rhenium(III) Cyanide Clusters n-4 [Re6Se8(PEt3)n(CN)6-n] (n = 4, 5) and Kinetics of Solvate Ligand Exchange on the Cubic 2+ [Re6Se8] Core. Inorg. Chem. 2002, 41, 4211-4216.

34. Koknat, F. W.; Adaway, T. J.; Erzerum, S. I.; Syed, S., Convenient Synthesis of the Hexanuclear Molybdenum(II) Halides Mo6Cl12 and Mo6Br12.2H2O. Inorg. Nucl. Chem. Lett. 1980, 16, 307-310.

35. Blanc, E.; Schwarzenbach, D.; Flack, H. D., The Evaluation of Transmission Factors and Their 1st Derivatives with Respect to Crystal Shape Parameters. J. Appl. Crystallogr. 1991, 24, 1035-1041.

36. Clark, R. C.; Reid, J. S., The Analytical Calculation of Absorption in Multifaceted Crystals. Acta Crystallogr. A 1995, 51, 887-897.

37. Version 1.171.37.35, Oxford Diffraction / Agilent Technologies UK Ltd, Yarnton, England. 2014.

38. Sheldrick, G. M., A Short history of SHELX. Acta Crystallogr. A 2008, 64, 112-122.

197

39. Sheldrick, G. M., Crystal Structure Refinement with SHELXL. Acta Crystallogr. 2015, 71, 3-8.

40. Palatinus, L.; Chapuis, G., SUPERFLIP - A Computer Program for the Solution of Crystal Structures by Charge Flipping in Arbitrary Dimensions. J. Appl. Crystallogr. 2007, 40, 786-790.

41. Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.; Puschmann, H., OLEX2: A Complete Structure Solution, Refinement and Analysis Program. J. Appl. Crystallogr. 2009, 42, 339-341.

42. Booth, M. R.; Frankiss, S. G., Constitution, Vibrational Spectra and Proton Resonance Spectra of Trimethylsilyl Cyanide and Isocyanide. Spectrochim. Acta A 1970, A26, 859- 869.

43. Szostak, R.; Hawranek, J. P., Ir Bandshapes of Liquid Trimethylsilyl Cyanide. Chem. Phys. Lett. 1981, 84, 331-334.

44. Low, M. J. D.; Ramasubr. N; Ramamurt. P; Deo, A. V., Infrared Spectrum Surface Reaction and Polymerization of Adsorbed Hydrogen Cyanide on Porous Glass. J. Phys. Chem. 1968, 72, 2371-2378.

45. Knoezinger, H.; Krietenbrink, H., Infrared Spectroscopic Study of Adsorption of Nitriles on Aluminum-Oxide - Fermi Resonance in Coordinated Acetonitrile. J. Chem. Soc. Farad. Trans. 1975, 71, 2421-2430.

46. Morrow, B. A.; Cody, I. A., Infrared Studies of Reactions on Oxide Surfaces: 3. HCN and C2N2 on Silica. J. Chem. Soc. Farad. Trans. 1975, 71, 1021-1032.

47. Kozirovski, Y.; Folman, M., Infra-Red Spectrum + Surface Polymerization of Adsorbed HCN. T. Faraday Soc. 1964, 60, 1532-1538.

48. Kim, S.; Sorescu, D. C.; Yates, J. T., Infrared Spectroscopic Study of ClCN Adsorption on Clean and Triethylenediamine-Precovered γ-Al2O3. J. Phys. Chem. C 2007, 111, 18226- 18235.

49. Ruttenberg, F. E.; Low, M. J. D., Reaction of Cyanogen with SiO2 Surfaces. J. Am. Ceram. Soc. 1973, 56, 241-244.

50. Kim, S.; Sorescu, D. C.; Yates, J. T., Infrared Spectroscopic Study of the Adsorption of HCN by γ-Al2O3: Competition with Triethylenediamine for Adsorption Sites. J. Phys. Chem. C 2007, 111, 5416-5425.

198