An Abstract of the Dissertation Of

AN ABSTRACT OF THE DISSERTATION OF

Danielle C. Hutchison for the degree of Doctor of Philosophy in Chemistry presented on December 5, 2019.

Title: Synthesis and Solution Characterization of Metal-Oxo Photoresist Precursors

Abstract approved: ______May Nyman

Metal-oxo clusters can be described as soluble pieces of metal-oxide frameworks. Currently, metal-oxo clusters are being considered as solution-based precursors for extreme ultraviolet (EUV, 13.5 nm) photoresist materials, which are a crucial component in the fabrication of microelectronic devices. Two different photoresist precursor systems have been investigated – alkyltin clusters and zirconium peroxide clusters. Four new butyltin structures have been crystallized, all having the β or γ Keggin topology:

β-[(BuSn)12(NaO4)(OCH3)12(O)5(OH)7] (β-NaSn12),

γ-[(BuSn)12(NaO4)(OCH3)12(O)5(OH)7] (γ-NaSn12),

γ-[(BuSn)12(NaO4)(OCH3)11(O)7(OH)6(BuSnOCH3)] (γ-NaSn13), 2+ β-[(BuSn)12(CaO4)(OCH3)12(O)4(OH)8] (β-CaSn12).

All four of these were synthesized by hydrolysis of BuSnCl3 with either NaOH or Ca(OH)2 in methanol. Two new zirconium peroxide structures have been characterized as well – an oxo-centered tetrahedron [Zr4(OH)4(μ-O2)2(μ4-O)(H2O)12](ClO4)6xH2O (ZrTd) and a 25-

membered wheel structure [Zr25O10(OH)50(O2)5(H2O)40](ClO4)10xH2O (Zr25). All of these new structures have been characterized extensively in solution by techniques including multinuclear NMR, small angle x-ray scattering (SAXS), and electrospray-ionization mass spectrometry (ESI-MS).

The alkyltin Keggin system is unique in that only the rarer β and γ isomeric forms have been crystallized. Furthermore, the Na-centered clusters have only been isolated as a mixture of β and γ isomers. We have therefore explored several factors which may influence the Keggin isomer such as the charge of the octahedral metal, the size and charge of the central metal, and the effects of aging, solvent, and heating. Changing the central metal from Na to Ca resulted in the stabilization of the β-isomer. Density functional theory

computations ranked the relative stabilities of these clusters γ-CaSn12 < γ-NaSn12 < β-

CaSn12 < β-NaSn12. Aging of Na- and Ca-centered clusters was studied in air by FT-IR and 1 in organic solvents (C6D6, CDCl3, 9:1 C6D6:MeOD, 9:1 CDCl3:MeOD) by H NMR. Results showed hydrolysis of the bridging methoxy ligands due to ambient humidity or residual water in the organic solvents. Addition of excess deuterated methanol was also

found to increase the rate of hydrolysis resulting in the following order: C6D6 < 119 C6D6/MeOD < CDCl3/MeOD ≈ CDCl3. Characterization by variable-temperature Sn NMR showed that the formation of additional isomers can be promoted by heating 23 solutions of NaSn12 and CaSn12 in C6D6. Na NMR characterization of heated solutions of

NaSn12 showed five chemical shifts, corresponding to all five Keggin isomers simultaneously in solution.

The aqueous chemistry of zirconium is typically dominated by the ubiquitous

square tetramer (Zr4) which spontaneously assembles upon dissolution of zirconium oxyhalide salts at low pH. By exchanging the halide with perchlorate and adding peroxide to the solution, we were able to crystallize two new topologies. Adding 1:1 peroxide/Zr

resulted in Zr25, and adding 10:1 peroxide/Zr resulted in ZrTd. To shed light on the role of peroxide in these systems, the reaction pathway was monitored by SAXS and pair distribution function (PDF). These studies revealed that when excess peroxide is present,

in the case of ZrTd, a large cluster species initially forms in solution before breaking down into the smaller tetrahedral cluster. Conversely, without excess peroxide, small trimer and

pentamer fragments are formed in solution which then assemble into Zr25 in the solid state at the interface between crystal and solution. The highly acidic nature of this solution prevents formation of the large cluster in the solution state. The trimer and pentamer

fragments have also been observed by ESI-MS and the intact Zr25 is never observed in water by any solution characterization methods.

Synthesis and Solution Characterization of Metal-Oxo Photoresist Precursors

by Danielle C. Hutchison

A DISSERTATION

submitted to

Oregon State University

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Presented December 5, 2019 Commencement June 2020

Doctor of Philosophy dissertation of Danielle C. Hutchison presented on December 5, 2019

APPROVED:

Major Professor, representing Chemistry

Head of the Department of Chemistry

Dean of the Graduate School

I understand that my dissertation will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request.

Danielle C. Hutchison, Author

ACKNOWLEDGEMENTS

To Dr. May Nyman, for your endless dedication to helping me become the best scientist that I can be and for providing me with so many opportunities to grow and learn.

To all former and present Nyman research group members, especially:

Morgan Olsen, for keeping me company during countless meetings and joining me into the exploration of tin chemistry.

Dr. Dylan Sures, for always being available to discuss science, current events, or life in general.

Ana Arteaga and Rachelle Smith, for always being down to get froyo with me and for any and all office conversations – scientific or otherwise.

To my wonderful husband, Adam – thank you for always loving me and encouraging me to reach my goals. You are a wonderful father to our son, and I wouldn’t have been able to do any of this without your support.

To Parker – I love being your mom more than anything, and watching you grow and learn brings me so much joy. I hope you never lose your sense of curiosity and excitement.

To all my friends and family, for supporting me through this crazy journey of graduate school, parenting, and living on the opposite side of the country.

CONTRIBUTION OF AUTHORS

M.N. provided significant guidance for all experiments, interpretation of results, and manuscript writing. In Chapter 3, M.R.O aided in the synthesis and crystallization of two

reported structures (β-NaSn12 and γ-NaSn13) In Chapters 3 and 4, R.D.S. performed all computational experiments and assisted in writing the computational portion of the manuscripts, and L.N.Z. performed single-crystal x-ray diffraction and solved all reported structures. K.A.P. provided computational insight. In Chapter 6, J.A.S. synthesized and crystallized all reported structures, and N.P.M performed single-crystal x-ray diffraction and solved crystal structures. K.K. collected x-ray total scattering data and performed PDF analysis; D.A.K. provided insight on manuscript writing.

TABLE OF CONTENTS

Page

1 Introduction to Metal-Oxo Clusters and the Keggin Structure...... 1 1.1 Metal-Oxo Clusters ...... 1 1.1.1 Characterization of Metal-Oxo Clusters ...... 1 1.2 The Keggin Structure ...... 3 1.3 Monoorganotin-Oxo Clusters ...... 7 1.4 Zirconium and Hafnium Metal-Oxo Clusters ...... 12

2 Nanolithography and Metal-Oxo Photoresists ...... 15 2.1 Nanolithography ...... 15 2.2 Metal-Oxo Photoresists ...... 18 2.2.1 Hf/Zr-Oxo Cluster Photoresists ...... 18 2.2.2 Organotin Photoresists ...... 19

3 Alkyltin Clusters: The Less Symmetric Keggin Isomers ...... 21 3.1 Abstract ...... 22 3.2 Introduction ...... 22 3.3 Experimental ...... 25 3.3.1 Synthetic Methods ...... 25 3.3.2 Characterization Techniques ...... 25 3.3.3 Computational Methods ...... 27 3.4 Results and Discussion ...... 28 3.4.1 Solution characterization ...... 32 3.4.2 Computational Studies ...... 37 3.5 Conclusions ...... 41 3.6 Acknowledgements ...... 42

4 Synthesis and Characterization of a Butyltin Keggin Ion with a Rare 4-Coordinate Ca Center ...... 43 4.1 Abstract ...... 44 4.2 Introduction ...... 44 4.3 Results And Discussion ...... 47

TABLE OF CONTENTS (Continued)

Page

4.3.1 Single crystal structure of β-CaSn12 ...... 47 4.3.2 Solution Characterization ...... 49 4.3.3 Aging Studies ...... 53 4.3.4 DFT Computational Analysis ...... 57 4.4 Conclusion ...... 60 4.5 Materials and Methods ...... 60

4.5.1 Synthesis of β-CaSn12 ...... 60 4.5.2 Computational Methods ...... 61 4.5.3 Characterization Techniques ...... 61 4.6 Acknowledgments ...... 64

5 Isomerization of Na-Centered Alkyltin Keggin Clusters ...... 65 5.1 Abstract ...... 66 5.2 Introduction ...... 66 5.3 Results and Discussion ...... 69 5.3.1 NMR Aging ...... 69 5.3.2 FT-IR Aging ...... 72 5.3.3 NMR HEATING ...... 73 5.4 Conclusion ...... 76 5.5 Experimental ...... 76 5.6 Acknowledgements ...... 77

6 Peroxide-Promoted Disassembly-Reassembly of Zr-Polyoxocations ...... 78 6.1 Abstract ...... 79 6.2 Introduction ...... 79 6.3 Results and Discussion ...... 82

6.3.1 Synthesis and structure descriptions of Zr4, ZrTd and Zr25 ...... 82 6.3.2 Solution characterization and disassembly-reassembly processes ...... 85 6.4 Conclusion ...... 93 6.5 Experimental Section ...... 94 6.5.1 Synthesis ...... 94

TABLE OF CONTENTS (Continued)

Page 6.5.2 Characterization Techniques ...... 95 6.6 Acknowledgments ...... 97

7 Conclusions ...... 98

References ...... 100

APPENDICES ...... 109

APPENDIX A Supporting Information for Chapter 3 ...... 110

APPENDIX B Supporting Information for Chapter 4 ...... 122

APPENDIX C Supporting Information for Chapter 5 ...... 132

APPENDIX D Supporting Information for Chapter 6 ...... 134

LIST OF FIGURES

Figure Page

1.1 Illustration of the process of collecting and interpreting SAXS data.[1] ...... 2

1.2 Example of small angle x-ray scattering curve highlighting regions of importance.3

1.3 Polyhedral representation of phosphotungstic acid (H3PW12O40). W is shown in light blue, P in pink, and O in red...... 4

1.4 Depiction of Keggin rotational isomers. Rotated trimers are shown in darker blue.[5] ...... 5

1.5 Structure of ethyltin dimer [EtSnCl2(OH)(H2O)]2. Sn is shown in gray, Cl in green, O in red and C in black...... 7

5+ 1.6 a) Structure of [(C7H7Sn)6(O)(OH)11(H2O)4] highlighting the trimeric unit also i observed in the Keggin structure. b) Structure of ( PrSn)9O8(OH)6Cl5. c) Structure 6+ of [Sn(OH)6(C7H7Sn)10(O)2(OH)16(H2O)2(OTf)2] . Gray polyhedra represent Sn, O is shown in red, Cl in green, S in yellow, F in orange, and C in black...... 8

2+ 1.7 Ball and stick (a) and polyhedral (b) representations of [RSn12(O)14(OH)6] . Sn is shown in gray, O in red, and C in black. Organic ligands have been shortened to the Sn-bound carbon for ease of viewing...... 9

i 5+ 1.8 Polyhedral representations of a) [( PrSn)12NaO4(OH)24] , b) [(BuSn)12(NaO4)(OCH3)12(O)9(OH)3(Sn(H2O)2)] (β-NaSn13) and c) [(BuSn)14(OCH3)10(OH)3O9(NaO4)(HBO3)2] (γ-NaSn14BO3). Blue and gray polyhedra represent Sn, O is shown in red, Na is shown in turquoise, B in pink, and C is shown in black. Additional capping Sn of β-NaSn13 and γ-NaSn14BO3 are shown in orange. Organic ligands have been shortened for ease of viewing...... 11

1.9 Ball and stick representation of a) tin carboxylate drum hexamer [RSn(O)O2CR’]6. b) tin carboxylate ladder [(R’Sn(O)O2CR)2R’Sn(O2CR)3]2. Sn is shown in gray, O in red, and C in black. Organic ligands have been shortened for ease of viewing. 12

8+ 1.10 Structure of [M4(OH)8(H2O)16] . M =Zr/Hf shown in green, O in red, and H in white...... 13

1.11 Ball-and-stick representation of [Zr6(O,OH)8] core. Zr is shown in green and O is shown in red. Terminal ligands have been omitted for ease of viewing...... 13

6- 1.12 Ball-and-stick representation of [Hf6(µ-O2)6(µ-OH)6(OH)12] . Hf is shown in blue and O in red...... 14

LIST OF FIGURES (Continued)

Figure Page

2.1 General schematic of the lithography process...... 16

2.2 8 nm lines patterned on Hf-based photoresist with EUV lithography.[93] ...... 19

3.1 Crystal structures of β-NaSn12 (left), γ-NaSn12 (center), and γ-NaSn13 (right). Tin atoms are represented by blue, gray, and orange polyhedra. Blue polyhedra represent trimers which have been rotated 60° with respect to the α-configuration. The orange polyhedron on γ-NaSn13 represents the additional capping tin. Na is shown in turquoise at the center of each cluster. Only the carbon that is bound directly to Sn is shown, for ease of viewing...... 30

3.2 (a) Experimental powder X-ray diffraction pattern for β,γ-NaSn12 (red) along with calculated powder patterns for β-NaSn12 (green) and γ-NaSn12 (blue). (b) SEM image of β,γ-NaSn12 crystals...... 32

3.3 (a) ESI-MS full spectrum for 0.1mM β,γ-NaSn12 in MeOH. Positive ionization mode, 100V fragmentation. (b) Overlay of experimental (blue) and simulated (red) ESI-MS peak envelopes for most intense peaks of spectrum. The simulated peak envelope is a combination of 9 overlapping peaks...... 33

3.4 Simulated (blue) and experimental (red) small angle X-ray scattering curves for β,γ- NaSn12 in benzene. Discrepancies between the simulated and experimental curves above q = 0.8 Å-1 are due to imperfect background subtraction and solvent scattering...... 35

1 3.5 H NMR spectrum of β,γ-NaSn12 in benzene-d6...... 36

13 3.6 C NMR spectrum of β,γ-NaSn12 in benzene-d6...... 36

119 3.7 Sn NMR spectrum of β,γ-NaSn12 in benzene-d6...... 37

3.8 Theoretical structures for α-NaSn12 (left) and α-NaSn13 (right). Tin atoms are represented by gray polyhedra. The orange polyhedron represents the additional capping tin on α-NaSn13. Na is shown as a turquoise sphere at the center of each cluster...... 38

2+ 4.1 Single crystal x-ray structure of β-CaSn12 [(BuSn)12(CaO4)(OCH3)12(O)4(OH)8] . Gray and blue polyhedra represent Sn; Ca is shown in teal, O in red, and C in black. Butyl chains have been eliminated with only the Sn-bound carbon shown, for ease of viewing...... 48

LIST OF FIGURES (Continued)

Figure Page

4.2 a) Full spectrum of β-CaSn12 in methanol. Positive ionization mode, 100V fragmentation. b) Experimental (red) and simulated (black) peak envelopes for main ion species. The simulation is a combination of 8 overlapping peaks...... 50

1 119 4.3 a) H NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue). b) Sn NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue). Inset: depiction of three Sn sites of β-CaSn12...... 52

4.4 Small angle x-ray scattering data for β-CaSn12 in THF (red) and simulated scattering curve (black)...... 53

1 4.5 H NMR spectra following aging of β-CaSn12 in C6D6 and 9:1 C6D6:MeOD...... 54

4.6 Average % methoxy ligands hydrolyzed vs number of days for β-CaSn12 in various solvents...... 54

4.7 FT-IR spectra showing the change in absorbance of the C-O stretch of solid β- CaSn12 over time...... 56

1 119 4.8 (a) H NMR spectra of β-CaSn12 in C6D6 heated in situ. (b) Sn NMR spectra of β- CaSn12 in C6D6 heated in situ. Gray arrows indicate growth of the γ isomer with heating...... 57

5.1 Structural representations of (a) β-NaSn12 and (b) γ-NaSn12. Blue and gray polyhedra represent Sn. Rotated trimers are shown in blue to differentiate isomers. Na is shown in turquoise, C is black, and O is red. Butyl chains have been shortened to the Sn- bound C for ease of viewing...... 68

1 5.2 H NMR spectra of β,γ-NaSn12 over time in (a) C6D6 and (b) CDCl3...... 70

5.3 Average percentage of methoxy ligands hydrolyzed vs number of days for β,γ- NaSn12 in various solvents...... 70

1 5.4 H NMR spectra of hydrolyzed β,γ-NaSn12 in C6D6 before (red) and after adding 12 eq MeOH (blue)...... 72

5.5 FT-IR spectra showing the change in intensity of the C-O stretch for solid β,γ- NaSn12 over time...... 73

1 119 5.6 (a) H NMR and (b) Sn NMR spectra of β,γ-NaSn12 C6D6 heated in situ...... 74

23 5.7 (a) Na NMR spectra of β,γ-NaSn12 in C6D6 heated in situ. (b) Expanded view of 23Na spectrum collected at 60°C after 3 hours...... 74

LIST OF FIGURES (Continued)

Figure Page

6.1 Schematic summarizing the reaction pathways of peroxide-promoted disassembly- reassembly of the ubiquitous Zr4 polycation to obtain new peroxide ligated cluster forms. The species in the yellow box are identified by SAXS and/or ESI-MS, but have not yet been isolated. Color scheme (used throughout the paper): Ligands-- turquoise = peroxide, green = hydroxide, gold = oxo, red = aqua. Black = Zr. For the ‘prototype ZrO2’ clusters that form prior to ZrTd, the nuclearity is approximately 22 Zr-centers (see text)...... 82

6.2 Representation of the Zr25 compound along the b axis, showing the layer- arrangement of the clusters. Color scheme is same as ...... 85

6.3 (A) X-ray scattering of Zr4 in water and organic solvents, along with simulated Zr4 (orange) and simulated scattering data (black) from a triple tetramer (B)...... 87

6.4 (A) Small angle x-ray scattering (SAXS) data for Zr25 in MeOH (red), DMF (blue), and water (green). Simulated scattering curve for Zr25 shown in black. (B) Two- phase spherical fit (red) to experimental SAXS curve of Zr25 in water (green). Population 1 radius: 7.9Å (13%). Population 2 radius: 3.1 Å (87%). Structure factor between 0.1-0.3 Å-1 was fit with a distance of 23 Å between clusters (see Table 6.1)...... 88

6.5 (A) SAXS of Zr25 reaction solution aged for three days along with Zr25 redissolved in water (green) and simulated Zr25 for comparison (black). (B) Experimental and simulated PDF of same reaction solution at day 1 and day 4 of aging. Crystal growth begins on day 3. Intensity of experimental data is normalized to the Zr-O pair at ~2 Å...... 90

6.6 (A) SAXS of ZrTd perchlorate crystals dissolved in methanol (red), DMF (blue) and water (green), and simulated from single-crystal structure (black). In these 50 mmolar solutions, there is linking between the clusters, via hydrolysis and condensation of water ligands. (B) Reaction solutions of ZrTd first indicating formation of larger clusters, followed by fragmentation by coordination with peroxide (see ...... 91

LIST OF TABLES

Table Page

3.1 Crystallographic information for reported structures ...... 31

3.2 Formulae and m/z of Simulated Peak Envelopes...... 34

3.3 Hydrolysis Gibbs free energy (kcal/mol) and HOMO-LUMO gaps of NaSn12 and NaSn13 isomers...... 40

4.1 Average bond lengths for O atoms of central tetrahedron ...... 49

4.2 ESI-MS peak assignments for β-CaSn12 in methanol ...... 50

4.3 Coupling constants, Sn-Sn distances, and peak assignments for 119Sn peaks of β- CaSn12 ...... 52

4.4 Calculated hydrolysis Gibbs free energy and HOMO-LUMO gap for β and γ isomers of NaSn12 and CaSn12 ...... 58

5.1 Peak Widths and Isomer Assignments of Chemical Shifts in 23Na NMR Spectrum of Heated NaSn12 Solution ...... 75

1 6.1 SAXS analysis of 2-phase fit of Zr25 reaction solutions with evaporation and aging ...... 88

6.2 SAXS analysis of ZrTd reaction solutions with evaporation and aging ...... 92

LIST OF APPENDIX FIGURES

Figure Page

A1 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)11(NaO4)(OH)17(OCH3)12] ...... 115

A2 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4)(O)10(OH)(OCH3)7] ...... 116

A3 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4)(O)9(OH)2(OCH3)8] ...... 116

A4 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4)(O)8(OH)3(OCH3)9] ...... 117

A5 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4)(O)7(OH)4(OCH3)10] ...... 117

A6 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4) (O)6(OH)5(OCH3)11] ...... 118

A7 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4)(O)4(OH)7(OCH3)13] ...... 118

A8 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4) (O)5(OH)6(OCH3)12] ...... 119

A9 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [(BuSn)12(NaO4)(O)5(OH)13(OCH3)9(CH3OH)9] ...... 119

A10 Size distribution analysis of SAXS data of β,γ-NaSn12 in benzene...... 120

A11 Modelling II results for β,γ-NaSn12 in benzene. The experimental scattering curve is in red and the calculated model in gray. The calculated radius is consistent with the radius determined from the experimental crystal structure...... 120

A12 IR spectrum of β,γ-NaSn12 showing the presence of methoxy ligands indicated by the strong C-O stretch at 1038 cm-1...... 121

B1 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)11(OH)6] ...... 124

LIST OF APPENDIX FIGURES (Continued)

Figure Page

B2 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)10(OH)8] ...... 124

B3 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)11(OH)4(OCH3)2] ...... 125

B4 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)11(OH)3(OCH3)3] ...... 125

B5 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)8(OH)11(OCH3)] ...... 126

B6 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)10(OH)4(OCH3)4] ...... 126

B7 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)8(OH)10(OCH3)2] ...... 127

B8 Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) for 2+ [(BuSn)12(CaO4)(O)10(OH)3(OCH3)5] ...... 127

1 B9 Full H NMR spectrum of β-CaSn12 in C6D6 (red) and β,γ-NaSn12 (blue) ...... 128

119 B10 Full Sn NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue) ...... 128

13 B11 C NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue) ...... 129

B12 Scattering curve of β-CaSn12 in THF (red) and spherical model of the data (gray). The model gives a cluster radius of 4.9Å, a center-to-center distance between clusters of 8.6Å, and 0.87 nearest neighbors...... 129

1 B13 (a) Aging of β-CaSn12 in CDCl3 monitored by H NMR. (b) Aging of β-CaSn12 1 in 90% CDCl3/10% MeOD monitored by H NMR...... 130

119 B14 Sn NMR spectrum of cooled β-CaSn12 solution after heating...... 130

B15 Full FT-IR spectra of β-CaSn12 from 1 to 28 days after isolation...... 131

1 C1 H NMR spectra of β,γ-NaSn12 over time in C6D6/CD3OD (a) and CDCl3/CD3OD (b)...... 132

C2 Full FT-IR spectra of β,γ-NaSn12 from 1 to 28 days after isolation...... 133

LIST OF APPENDIX FIGURES (Continued)

Figure Page

23 C3 Na NMR of β,γ-NaSn12 in C6D6 heated to 60°C (blue) and then cooled to room temperature (red)...... 133

D1 Representation of the Zr25 cluster with disordered atoms ...... 138

D2 Representation of the Zr25 cluster with 5% occupancy of the inner ring. View of the coordination of the delocalized Zr atom. Dashed bonds correspond to long bonds...... 138

D3 (right) Cylindrical fit (red) to experimental SAXS curve of Zr4 in methanol (black) having a radius of 3.3Å and a length of 22.8Å. (middle) Two-population spherical fit (red) to experimental SAXS curve of Zr25 in methanol (black). Population 1 radius:8.6Å. Population 2 radius: 3.2Å. (right) Cylindrical fit (gray) to experimental SAXS curve of ZrTd in methanol (red). Radius: 3.1Å, length: 21.6Å ...... 139

D4 Comparison of experimental scattering data for ZrTd reaction solution (day 4; blue) compared to simulated scattering curve for ‘Zr22’ (red). To create scattering data for ‘Zr22’, we used the structure of Pu22, approximately formulated [Pu22O28(OH)4Cl28(H2O)20]; changed the Pu to Zr and Cl ligands to oxygen to approximately simulate the size, shape and electron density of a hypothetical Zr22. Inset, Pu(Zr)22. Metals are green, oxygen is red...... 140

D5 Experimental and simulated PDF of the ZrTd reaction solution after 10 days of aging ...... 140

D6 (left) ESI-MS spectrum of Zr4 in methanol. Negative ionization mode, 30V fragmentation. (right) ESI-MS spectrum of Zr4 in water. Positive ionization mode, 100V fragmentation...... 141

D7 (left) ESI-MS spectrum of Zr25 in MeOH. Negative ionization mode, 100V fragmentation. (right) ESI-MS spectrum of Zr25 in water. Positive ionization mode, 100V fragmentation...... 141

D8 (left) ESI-MS spectrum of ZrTd in MeOH. Negative ionization mode, 30V fragmentation. (right) ESI-MS spectrum of ZrTd in water. Positive ionization mode, 30V fragmentation...... 142

D9 Experimental ESI MS (-, blue spectrum) and calculated peak positions (red) for - [H(ClO4)2] ...... 145

LIST OF APPENDIX FIGURES (Continued)

Figure Page

D10 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [(CH3OH)(ClO4)(H2O)5] ...... 145

D12 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr(OCH3)(ClO4)4] ...... 146

D11 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr(OCH3)2(ClO4)3] ...... 146

D13 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr(ClO4)5] ...... 147

D14 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr2(CH3OH)2O(OH)3(ClO4)4] ...... 147

D15 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr3O4(OCH3)(ClO4)4(H2O)2] ...... 148

D16 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr4(CH3OH)2(OH)11(ClO4)6] ...... 148

D17 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr4(OH)5(OCH3)6(ClO4)6] ...... 149

D18 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr4(CH3OH)(OH)4(OCH3)7(ClO4)6] ...... 149

D19 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr3(OCH3)4(ClO4)9(H2O)3] ...... 150

D20 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O2(OH)10(ClO4)] ...... 150

D21 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O5(OH)3(ClO4)2(H2O)2] ...... 151

D22 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O6(ClO4)3(H2O)3] ...... 151

D23 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr3O3(ClO4)5(H2O)2] ...... 152

LIST OF APPENDIX FIGURES (Continued)

Figure Page

D24 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O4(OH)3(ClO4)4(H2O)] ...... 152

D25 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O4(OH)3(ClO4)4(H2O)2] ...... 153

D26 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr6O3(OH)17(H2O)3] ...... 153

D27 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr6O5(OH)13(H2O)6] ...... 154

D28 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O5(ClO4)5(H2O)2] ...... 154

D29 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O5(ClO4)5(H2O)3] ...... 155

D30 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O(OH)9(ClO4)4(H2O)6] ...... 155

D31 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O4(OH)(ClO4)6(H2O)] ...... 156

D32 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr4O4(OH)(ClO4)6(H2O)2] ...... 156

D33 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5O7(ClO4)5(H2O)2] ...... 157

D34 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr6(OH)19(ClO4)4(H2O)2] ...... 157

D35 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr6(OH)7(O2)8(OCH3)2(H2O)2] ...... 158

D36 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(O)14(OH)(O2)(OCH3)2] ...... 158

D37 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(O)13(OH)(O2)(OCH3)4] ...... 159

LIST OF APPENDIX FIGURES (Continued)

Figure Page

D38 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(O)14(OH)5(CH3OH)3] ...... 159

D39 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(O)6(OH)19(O2)] ...... 160

D40 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(O)7(OH)12(O2)2(OCH3)3] ...... 160

D41 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr6(O2)11(OCH3)3(H2O)13] ...... 161

D42 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(O)4(OH)23(O2)(CH3OH)] ...... 161

D43 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(OH)22(O2)5(OCH3)] ...... 162

D44 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(OH)14(O2)8(OCH3)3] ...... 162

D45 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr8(O2)13(OCH3)7] ...... 163

D46 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr3(OH)8(O2)(H2O)4(ClO4)] ...... 163

D47 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr2(O)2(H2O)7(ClO4)3] ...... 164

D48 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5(OH)(O)4(O2)5] ...... 164

D49 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5(OH)3(O)7(O2)(H2O)4] ...... 165

D50 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5(OH)5(O)6(O2)(H2O)4] ...... 165

D51 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5(O)(OH)17] ...... 166

LIST OF APPENDIX FIGURES (Continued)

Figure Page

D52 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5(OH)19] ...... 166

D53 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5(OH)(O)3(O2)6(H2O)5] ...... 167

D54 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr5(OH)5(O)(O2)6(H2O)4] ...... 167

D55 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr6(OH)(O)8(O2)3(H2O)4] ...... 168

D56 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for + [Zr6(OH)3(O)3(O2)7(H2O)4] ...... 168

D57 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr3O(O2)2(ClO4)7(CH3OH)] ...... 169

D58 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr4O2(O2)2(OCH3)3(ClO4)6] ...... 169

D59 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr5O2(OH)8(OCH3)(ClO4)4(O2)2(CH3OH)2] ...... 170

D60 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr6O3(OH)14(ClO4)3(O2)(H2O)2] ...... 170

D61 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr3(ClO4)7(O2)3(H2O)9] ...... 171

D62 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr4O2(OCH3)4(ClO4)7(O2)] ...... 171

D63 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr4O(OCH3)6(ClO4)7(O2)] ...... 172

D64 Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) for - [Zr3O(OH)(ClO4)10(H2O)] ...... 172

D65 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr3(O)(OH)4(O2)(H2O)5(ClO4)3] ...... 173

LIST OF APPENDIX FIGURES (Continued)

Figure Page

D66 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr3(O)(OH)(O2)2(H2O)2(ClO4)4] ...... 173

D67 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr5(O)5(OH)4(O2)2(H2O)3(ClO4)] ...... 174

D68 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr4(O)(OH)4(O2)3(H2O)(ClO4)3] ...... 174

D69 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr3(O)(OH)2(O2)(H2O)5(ClO4)5] ...... 175

D70 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr4(O)2(OH)3(O2)2(H2O)3(ClO4)4] ...... 175

D71 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr6(O)6(OH)7(O2)(H2O)2(ClO4)2] ...... 176

D72 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr4(O)4(O2)(H2O)5(ClO4)5] ...... 176

D73 Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) for 1+ [Zr4(O)(OH)2(O2)(H2O)3(ClO4)9] ...... 177

D74 Experimental (black) and simulated (red) powder XRD diagrams of ZrTd...... 177

D75 Experimental (black) and simulated (red) powder XRD diagrams of Zr25...... 178

LIST OF APPENDIX TABLES

Table Page

A1 BVS for Oxo Ligands of β-NaSn12...... 110

A2 BVS for Oxo Ligands of γ-NaSn12...... 111

A3 BVS for Oxo Ligands of γ-NaSn13...... 113

B1 Bond Valence Sum for β-CaSn12 ...... 122

B2 Atomic percentages for selected elements in β-CaSn12 determined by SEM-EDX ...... 123

D1 Cl/Zr for Zr4 and Zr25 by wet and EDXS analyses ...... 134

D2 Atomic percentages of Zr and Hf in Zr4 determined by EDS ...... 134

D3 Crystal data and structure refinement details for Zr4, Zr25, and ZrTd compounds ...... 135

D4 Distances, Å, in CVT Motifs for several phases...... 136

D5 Bond valence sums and assignments of oxo and hydroxyl groups in Zr25...... 139

D6 Peak assignments for mass spectrum of Zr4 in MeOH ...... 142

D7 Peak assignments for mass spectrum of Zr4 in water ...... 143

D8 Peak assignments for mass spectrum of Zr25 in MeOH ...... 143

D9 Peak assignments for mass spectrum of Zr25 in water ...... 144

D10 Peak assignments for mass spectrum of ZrTd in MeOH ...... 144

D11 Peak assignments for mass spectrum of ZrTd in water ...... 144

1 Introduction to Metal-Oxo Clusters and the Keggin Structure 1.1 Metal-Oxo Clusters

Metal-oxo clusters are soluble, molecular species formed from both main group and transition metals, and often contain similar structural elements to bulk metal oxides. These clusters are larger than monomeric hydrated metal ions and smaller than nanoparticles. As the name suggests, metal-oxo clusters consist of metal centers bridged by oxygen in the form of oxo, hydroxyl, peroxo, or alkoxide ligands. Though formation mechanisms vary depending on the type of cluster, they are typically formed by a series of hydrolysis and condensation reactions in solution. Metal-oxo clusters must also have a stable terminal ligand to prevent the condensation of multiple clusters into a bulk metal oxide. The identity of the terminal ligand depends on many factors including the identity of the metal center, the solvent, and chemistry of the system. Some examples of common terminal ligands include oxo, aqua, halide, sulfate, and organic ligands (carboxylate, alkoxide, alkyl groups). The bridging and terminal ligands are not limited to those listed here. Metal-oxo clusters also have a wide variety of applications. Some examples of these include catalysis, thin film precursors, and nodes for metal-organic frameworks. Specific metal-oxo clusters and their applications will be discussed in more detail later.

1.1.1 Characterization of Metal-Oxo Clusters

Metal-oxo clusters are characterized in both the solution and solid states by a variety of methods. Typical solid state characterization methods include single crystal and powder x-ray diffraction and Fourier-transform infrared spectroscopy (FT-IR). Some examples of solution state characterization techniques are electrospray-ionization mass spectrometry (ESI-MS), nuclear magnetic resonance (NMR) spectroscopy, and small angle x-ray scattering (SAXS).

Small-angle x-ray scattering is a technique which can provide information about the size, shape, and distribution of particles in solution. In this technique, a solution containing a scattering species (i.e. metal-oxo clusters) is exposed to x-ray radiation. These x-rays scatter off of the sample and are collected on an image plate detector. The image is

2 then processed to create a scattering curve, from which information about the speciation of clusters in solution can be extracted. A schematic illustrating this process is shown in Figure 1.1. This technique relies on the electron density contrast between the scattering species and the solvent in order to produce a scattering image. Metal-oxo clusters are ideal subjects for study by SAXS as they are soluble and have high electron density.

Figure 1.1. Illustration of the process of collecting and interpreting SAXS data.[1]

An example of a scattering curve for a monodisperse solution of spherical particles is shown in Figure 1.2. The scattering intensity is plotted as a function of q, also called the scattering vector, which is related to the wavelength of x-ray radiation and the scattering angle θ by the equation 𝑞Å4𝜋/𝜆 ∙ sin𝜃. Both axes of the scattering plot are on a logarithmic scale. The scattering plot can be separated into three regions, as shown in Figure 1.2. The plateau region (typically low q values) can provide information about the interaction of the particles in solution. For example, a flat curve in the plateau region indicates a monodisperse solution. A curve which increases in intensity in the low-q region indicates aggregation of particles, and a curve which decreases in intensity indicates repulsion or ordering of particles in solution. The Guinier region is the portion of the scattering curve where the intensity rapidly decreases. This provides information about the size of the scattering species. The q values over which this Guinier region occurs are inversely related to the size of the particle. In other words, a Guinier region at higher q values indicates a smaller particle and vice versa. Finally, the Porod region contains the oscillations and solvent scattering. Well-defined oscillations indicate a high-purity solution.

Figure 1.2. Example of small angle x-ray scattering curve highlighting regions of importance. 1.2 The Keggin Structure

One structural topology which is commonly observed for metal-oxo clusters is the Keggin structure (Figure 1.3). The structure was first proposed in 1826 by Berzelius[2], but wasn’t experimentally determined by x-ray diffraction until Keggin structurally characterized crystals of phosphotungstic acid (H3[PW12O40]), more than 100 years later in [3] 1934. The Keggin structure generally consists of a central {AO4} tetrahedron surrounded by twelve {MO6} octahedra (where A is typically a main-group element and M is a transition metal), organized into four trimeric units which are linked by shared oxygens. In 3- the case of [PW12O40] A=P and M=W. Each trimeric unit is comprised of three {MO6} octahedra, in which each octahedron shares one edge with each of the other two trimers. The oxygen at the center of the trimer, where all three octahedra converge, is also shared with the central atom, forming one of the four corners of the {AO4} tetrahedron. All of the bridging oxygens within H3PW12O40 are oxo or hydroxyl ligands and each W has a double- bonded yl-oxo in the terminal position.

Figure 1.3. Polyhedral representation of phosphotungstic acid (H3PW12O40). W is shown in light blue, P in pink, and O in red. Based on the geometry of this structure, Baker and Figgis postulated five rotational isomers of the Keggin ion, denoted by the Greek letters α, β, γ, δ, and ε.[4] The

3- aforementioned [PW12O40] adopts the α-isomer, in which all four trimeric units are

connected to each other by corner-sharing and the molecule as a whole has Td symmetry. A 60° rotation of one trimeric unit results in formation of the β-isomer, thereby reducing the symmetry of the molecule to C3v, with the C3 axis passing through the center of the rotated trimer. All trimers are still connected to one another by corners. A second trimer rotation of 60° yields the γ-isomer with C2v symmetry and two trimers now connected by edge-sharing. The C2 axis passes through the shared edge connection. A third 60° trimer rotation results in the δ-isomer (C3v symmetry), with the C3 axis through the center of the unique non-rotated trimer. The three rotated trimers are connected by edges and the unique trimer is connected to the others by corners. A final 60° rotation forms the ε-isomer, with

Td symmetry and all trimers connected by edge-sharing. All five rotational isomers are shown in Figure 1.4 with rotated trimers highlighted. This convention will be used throughout this thesis to more easily distinguish between isomers.

Figure 1.4. Depiction of Keggin rotational isomers. Rotated trimers are shown in darker blue.[5] The Keggin ion in its various isomeric forms has been observed for many metals across the periodic table and as a building block of some naturally-occurring minerals. These minerals include the aptly-named Kegginite, containing an As-centered vanadium

ε-Keggin unit, zunyite which contains a Si-capped α-Al13 Keggin building block, and [6–9] ferrihydrite and magnetite which contain Fe13 ε and δ Keggin units, respectively. The most well-characterized Keggin ions are those belonging to the polyoxometalate family. Polyoxometalates are a subset of water-soluble anionic metal-oxo clusters which contain early d0 transition metals (typically V, Nb, Mo, Ta, or W) bridged by oxo or hydroxyl ligands. The defining feature of polyoxometalates is the stable terminal –yl oxo bond. 3- Polyoxometalate Keggin ions include the previously mentioned [PW12O40] and its 3- molybdenum analog, [PMo12O40] . Other members of this family include W or Mo in octahedral coordination with Si, As, B, Al, Ge, Fe, Co, Cu, etc. at the center.[10]

The most commonly observed isomer for this family of Keggin ions is the α-isomer, with the β-isomer being observed only occasionally. The structure of the β-isomer was not determined crystallographically until 1973[11], nearly four decades after the determination of the structure of the α-isomer. A number of experimental and computational studies have focused on the differences between the α and β isomers for tungstate and molybdate Keggin ions.[10,12–17] The main conclusion of these studies is that the β-isomer is generally less stable than the α-isomer, but that the β-isomer becomes more stable when the cluster is

6 reduced. However, factors controlling the stability of Keggin clusters are very complex and depend on a number of factors including the identity of the central heteroatom and the overall charge of the cluster.[10] It is worth noting that niobium also forms an α-Keggin with Si or Ge center, but the β-isomer has never been observed.[18,19]

Aluminum, gallium, and chromium also form cationic Keggin clusters and Keggin 7+ derivatives. The aluminum Keggin [Al13O4(OH)24(H2O)12] is typically found as the ε- isomer.[20] The δ and γ isomers have also been crystallized, and the other two isomers (β [21–25] and α) are presumed to exist in solution. Keggin derivatives include the Al30 and Ga30 polycations which contain δ-Keggin building units, and the Ga2Al18 cluster which contains ε-Keggin fragments.[25–30] The mixed Zn-Cr-Al Keggin 8+ [ZnO4Al5.1Cr6.9(OH)24(H2O)12Zn(H2O)3] crystallizes as the δ-isomer with the additional zinc cap stabilizing the rotated trimer.[31] The chromium Keggin also forms the δ-isomer both with and without caps.[32] Other Keggin ions and their preferred isomers include the [33,34] [35] [36–38] Fe13 Keggin (α) , organoanitmonate (ε) , and organotin Keggins (β,γ).

The most accurate way to determine the Keggin isomer is by single crystal x-ray diffraction, as most solution characterization methods (with the exception of NMR) are not sensitive to the arrangement of atoms within the molecule. NMR is a powerful characterization technique which provides information about the different chemical environments within a molecule, which is related to the symmetry of the overall molecule. In this way it can provide some information about the number and type of isomers present, but cannot definitively identify isomers. When analyzing the addenda (octahedral) metal of the Keggin ion, the number of chemical shifts depends on the symmetry of the isomer.

Both α and ε have Td symmetry which results in just one chemical shift in the NMR

spectrum. The β and δ isomers have C3v symmetry, resulting in three different chemical

shifts in a 1:2:1 ratio. Finally, the γ isomer with C2v symmetry has four chemical shifts in a 1:2:1:2 ratio. Additionally, the chemical shift value of the central cation of the Keggin may change depending on the isomer.[12,13,15,21,23,39]

1.3 Monoorganotin-Oxo Clusters

The tin-carbon bond is unique with respect to many other organometallic bonds in that it is hydrolytically stable. Since tin and carbon are both in group 14, there is good overlap of the molecular orbitals leading to a stable bond. The organic substituent makes an ideal terminal ligand for the formation of metal-oxo clusters and allows for solubility in organic solvents. Only monoorganotin (IV) compounds relevant to applications in nanolithography will be discussed here. A number of organotin-substituted polyoxometalates have been characterized as well, but these are outside the scope of this thesis. A variety of monoorganotin-oxo clusters have been known for some time, ranging in size from dimers to dodecamers. The dimer and dodecamer topologies appear to be robust – a variety of organotin starting materials will all afford the same types of structures. However, there are fewer examples of clusters with nuclearities between dimer and dodecamer. Here we provide a few examples of the structural diversity of monoorganotin- oxo clusters that have been reported.

Figure 1.5. Structure of ethyltin dimer [EtSnCl2(OH)(H2O)]2. Sn is shown in gray, Cl in green, O in red and C in black.

[40] [41] [42] The alkyltin dimer [RSnCl2(OH)(H2O)]2 (R= methyl , ethyl , isopropyl , n- butyl[43], or isobutyl[42]) shown in Figure 1.5, consists of two tin centers bridged by two hydroxyl ligands. There is one alkyl group, two chlorides, and one aqua ligand in the terminal positions on each tin. The synthesis for these dimeric compounds is typically very simple; in most cases they crystallize readily from hydrolysis of the RSnCl3 starting material by ambient moisture followed by slow evaporation.[41–43]

5+ Figure 1.6. a) Structure of [(C7H7Sn)6(O)(OH)11(H2O)4] highlighting the trimeric unit i also observed in the Keggin structure. b) Structure of ( PrSn)9O8(OH)6Cl5. c) Structure of 6+ [Sn(OH)6(C7H7Sn)10(O)2(OH)16(H2O)2(OTf)2] . Gray polyhedra represent Sn, O is shown in red, Cl in green, S in yellow, F in orange, and C in black. Relevant intermediate nuclearity clusters include a benzyltin hexamer[44], an isopropyltin nomer[45], and a benzyltin undecamer (Figure 1.6).[44] Both the hexamer 5+ [(C7H7Sn)6(O)(OH)11(H2O)4] and undecamer 6+ [Sn(OH)6(C7H7Sn)10(O)2(OH)16(H2O)2(OTf)2] (OTf = trifluoromethanesulfonate) were products of the same partial debenzylation reaction of a dibenzyltin decamer, [44] (PhCH2)2SnO)6[((PhCH2)2SnOH)2(CO3)]2, with trifluoromethanesulfonic acid. The hexamer structure (Figure 1.6a), can be described as a bi-capped tetrahedron. It also contains a trimeric subunit identical to the trimeric building block of the Keggin ion. The undecamer structure (Figure 1.6c) contains two pentamer units (one benzyltin removed from the previously mentioned hexamer) bridged by an inorganic Sn.[44] Characterization by ESI-MS and multinuclear NMR showed that these clusters are stable in solution. The i nononuclear isopropyltin cluster ( PrSn)9O8(OH)6Cl5 (Figure 1.6b) was recrystallized i [45] from a heated solution of PrSn(OH)2Cl ꞏ3/4 H2O in DMSO. No solution characterization was reported for this compound. Although these clusters have not been investigated for patterning applications, they have similar structural features to clusters which have been patterned. The variety of sizes and shapes make these intermediate nuclearity clusters good potential candidates for future patterning studies to investigate the effect of cluster size/shape and terminal ligand on the patterning performance.

The most well-known monoorganotin-oxo cluster is the dodecameric “football” 2+ cluster, [RSn12(O)14(OH)6] (Sn12) (Figure 1.7). This structure is a hollow cage composed of a hexameric ring of 5-coordinate tin, capped on each end by trimeric units of 6- coordinate tin. The cluster has an overall charge of 2+ and each tin has the organic substituent in the terminal position with oxo and hydroxyl bridging ligands. This structure has been synthesized with a multitude of organic substituents and counterions.[46–54] The most common synthetic method for the butyltin derivative uses the starting material BuSnOOH.[47] Derivatives of this structure have also been characterized with one[55] or three[56] vanadium atoms substituted for 5-coordinate tin.

2+ Figure 1.7. Ball and stick (a) and polyhedral (b) representations of [RSn12(O)14(OH)6] . Sn is shown in gray, O in red, and C in black. Organic ligands have been shortened to the Sn-bound carbon for ease of viewing.

Another dodecameric organotin cluster is the sodium-centered isopropyltin γ- Keggin, reported in 1991 (Figure 1.8a).[36] This cluster has the formula i 5+ [( PrSn)12NaO4(OH)24] and was crystallized from a solution of NaI, AgI, and i PrSn(OH)2Cl in DMSO. The sodium-centered Keggin cluster co-crystallizes with sheets of AgI and Cl- counterions. No solution characterization was provided for this compound; only the solid-state single crystal structure was reported.

A second sodium-centered organotin Keggin structure was reported in 2017

(Figure 1.8b). This new structure [(BuSn)12(NaO4)(OCH3)12(O)9(OH)3(Sn(H2O)2)], abbreviated β-NaSn13, forms the β-Keggin topology and was isolated from a recrystallization of BuSnOOH in methanol.[37] This cluster features terminal butyl ligands

10 and bridging oxo, hydroxyl, and methoxy ligands. The oxo and hydroxyl ligands are found between trimeric units of the Keggin and the methoxy bridges are found within each of the four trimeric units. β-NaSn13 also has an additional inorganic Sn cap with two terminal aqua ligands – most likely due to impurities in the BuSnOOH starting material.[37,57] Extensive solution characterization by SAXS, ESI-MS, and 119Sn NMR of this compound showed that the cluster remains intact upon redissolution, although there may be multiple isomers present in addition to uncapped clusters and impurities from the commercial [37] starting material. Though the reported crystal structure of β-NaSn13 contains 13 tin atoms, we consider it a dodecamer as bulk solution characterization did not show evidence of the inorganic Sn cap.[37]

Two additional butyltin γ-Keggin structures have also been reported with borate capping ligands. These structures are formulated

[(BuSn)14(OCH3)10(OH)3O9(NaO4)(HBO3)2] (γ-NaSn14BO3) (Figure 1.8c) and [38] [(BuSn)14(OCH3)10(OH)3O9(NaO4)(PhBO2)2] (γ-NaSn14PhBO2). Similar to the other Sn Keggin structures these have a Na center, bridging hydroxyl and methoxy ligands, and terminal butyl ligands. Additionally, these clusters have two butyltin caps and two borate caps, which are located on either side of the rotated edge-sharing trimers. Characterization by ESI-MS and multinuclear NMR, as well as DFT computations, show that the additional caps help to stabilize the γ isomer and prevent isomerization from occurring in solution.[38,58]

i 5+ Figure 1.8. Polyhedral representations of a) [( PrSn)12NaO4(OH)24] , b) [(BuSn)12(NaO4)(OCH3)12(O)9(OH)3(Sn(H2O)2)] (β-NaSn13) and c) [(BuSn)14(OCH3)10(OH)3O9(NaO4)(HBO3)2] (γ-NaSn14BO3). Blue and gray polyhedra represent Sn, O is shown in red, Na is shown in turquoise, B in pink, and C is shown in black. Additional capping Sn of β-NaSn13 and γ-NaSn14BO3 are shown in orange. Organic ligands have been shortened for ease of viewing.

Another class of organotin compounds is synthesized by reacting RSn(O)OH with carboxylic acids.[59–62]. These clusters typically adopt hexameric ‘drum’ or ‘ladder' topologies (Figure 1.9). The drum structure [RSn(O)O2CR’]6 has been investigated as a [63,64] possible photoresist precursor. This structure consists of two staggered Sn3O3 rings with the terminal organic ligands in the axial position (pointing outward from the top and bottom of the ‘drum’) and the bridging carboxylates in the equatorial positions (around the sides of the ‘drum’). The ladder topology [(R’Sn(O)O2CR)2R’Sn(O2CR)3]2 can be considered as an ‘unfolded drum’, and 119Sn NMR characterization showed interconversion between the ladder and drum structures.[59] In addition to the drum hexamer, the Sn12 and β-NaSn13 clusters have also shown promise as precursors for EUV photoresists. This application will be discussed in detail in Chapter 2.

Figure 1.9. Ball and stick representation of a) tin carboxylate drum hexamer [RSn(O)O2CR’]6. b) tin carboxylate ladder [(R’Sn(O)O2CR)2R’Sn(O2CR)3]2. Sn is shown in gray, O in red, and C in black. Organic ligands have been shortened for ease of viewing. 1.4 Zirconium and Hafnium Metal-Oxo Clusters

Zirconium and hafnium can also form metal-oxo clusters in solution, though the variation in structural topology is much more limited than for organotin clusters. These clusters differ from polyoxometalates in that they are cationic and lack a stable terminal - yl oxo. Additionally, they differ from organotin clusters in that the terminal ligands are typically inorganic. The most commonly observed Zr/Hf cluster in solution is a square 8+ - [65– tetramer (Figure 1.10) formulated [M4(OH)8(H2O)16] ꞏ8 X (M= Zr, Hf; X= halide). 67] Crystals of this tetramer are found in commercial ZrOX2 ꞏ 8 H2O and HfOX2 ꞏ 8 H2O salts. The four metal centers each have four terminal aqua ligands and are bridged by hydroxyl ligands. This cluster has been characterized extensively in solution by x-ray scattering, x-ray spectroscopy, and mass spectrometry.[68,69]

8+ Figure 1.10. Structure of [M4(OH)8(H2O)16] . M =Zr/Hf shown in green, O in red, and H in white. The above tetramer is the only reported structure for Zr and Hf which contains [70] terminal aqua ligands. A hexameric Zr6 structure has also been isolated with acetate or [71] glycine bridging terminal ligands (Figure 1.11). This structure contains a [Zr6(O,OH)8] core, in which the Zr centers form the corners of an octahedron. This hexameric structure is also found as the node of various metal-organic frameworks (MOFs)[72], such as the commercially-available UiO-66.

Figure 1.11. Ball-and-stick representation of [Zr6(O,OH)8] core. Zr is shown in green and O is shown in red. Terminal ligands have been omitted for ease of viewing. All of the above structures have been isolated from acidic solutions. One additional

relevant soluble cluster is an Hf6 wheel topology which was isolated from an alkaline tetramethylammonium hydroxide (TMAOH) solution. This cluster (Figure 1.12),

formulated [Hf6(μ-O2)6(μ-OH)6(OH)12][TMA]6 has bridging hydroxyl and peroxide linkages and terminal hydroxides.[73]

6- Figure 1.12. Ball-and-stick representation of [Hf6(µ-O2)6(µ-OH)6(OH)12] . Hf is shown in blue and O in red. The soluble Zr/Hf topologies are dominated by tetrameric and hexameric structures, and larger clusters have only been isolated with the addition of sulfate terminal ligands. The family of Zr/Hf-sulfate clusters contains structures ranging in nuclearity from 2 to 18, however these are typically insoluble.[69,74–77] Yet, these Zr/Hf-sulfate and Hf- peroxide clusters are relevant to applications in thin films and nanolithography. This application is discussed in detail in Chapter 2.

2 Nanolithography and Metal-Oxo Photoresists 2.1 Nanolithography

There is a continual demand for smaller electronic devices with higher memory density and faster performance. Moore’s Law states that the number of transistors per integrated circuit, which is directly correlated to computing speed, will double every two years.[78] This law has held true since Moore’s prediction in 1965, but the current push towards sub-10nm features means we are quickly approaching the lower size limit of patterned features.[79] Nanolithography, or nanopatterning, is the typical industrial process used for the fabrication of integrated circuits in microelectronic devices. This process uses radiation in the form of light to write features onto a substrate. Two of the most important components of this process are the radiation source and photoresist material. A photoresist, as the name implies, is a material which is both photo-sensitive and resistant to etching. The properties of the radiation source and photoresist are ultimately what control the size of the patterned features.

In general, the first step of the lithography process is the deposition of a thin film onto a silicon substrate. Next, the photoresist is deposited, typically by spin-coating, on top of the thin film. A mask is applied which covers portions of the photoresist while leaving other areas exposed. The masked film is exposed to radiation which induces a solubility change in the exposed regions. Following exposure, the film is washed in a development solution. If the exposed photoresist becomes more soluble in the development solution, it is known as a positive-tone resist. If the resist becomes less soluble in developer after exposure it is known as a negative-tone resist. This process is illustrated in

Figure 2.1. Finally, the pattern is transferred to the film by etching and the photoresist is stripped away. The film is often baked at low temperatures after deposition and after exposure to remove residual solvent and to increase the density of the film. The exact steps in the patterning process may change depending on a number of factors including the compositions of the film and photoresist and the desired application.[80]

Figure 2.1. General schematic of the lithography process. The three critical photoresist parameters for patterning sub-20nm features are sensitivity, linewidth roughness, and resolution.[81] Sensitivity refers to the amount of radiation required to induce a solubility change. Understandably, higher sensitivity is desired so that less radiation exposure is required, thereby enabling faster processing of chips. Linewidth roughness (LWR) describes the variability in the width of patterned lines and resolution refers to the dimensions of the smallest patternable features (high resolution = small features). Significant LWR can lead to unreliable transistor performance, and these effects become amplified as the size of patterned features decreases.[82] Additionally, high resolution allows for the patterning of smaller features and fabrication of higher density integrated circuits, facilitating the continuation of Moore’s Law. While high sensitivity, low LWR, and excellent resolution are all highly desired in a photoresist, typically there is a trade-off where only two of these properties can be achieved at the same time.[81]

There have been many changes and improvements to the lithography process and corresponding technology since the first integrated circuits were fabricated in 1957 using the Kodak thin-film resist (KTFR).[83] This early photoresist was composed of a photo- sensitive crosslinking organic polymer and was a negative-tone resist. The light source was a mercury vapor lamp with wavelengths of 446 nm (g-line), 405 nm (h-line) and 365 nm (i-line).[84,85] Traditionally, smaller features were patterned by decreasing the wavelength of light used. However, using a shorter wavelength results in a corresponding decrease in

17 intensity, so photoresist materials needed to become more sensitive in order to maintain the same patterning performance.

With the introduction of deep ultraviolet (DUV) light sources (248-193 nm), new and more sensitive resists were needed once again.[86] This prompted the industry to transition to the use of chemically amplified (positive-tone) resists (CARs) around the 1970s, which have remained the primary photoresist material since.[81,83] Chemically amplified resists (CARs) have increased sensitivity due to a catalytic chain reaction which is initiated by exposure to the appropriate wavelength of light. This exposure generates an acid catalyst (photoacid generator) within the polymer matrix of the photoresist which changes the solubility of the exposed resist.[86,87] This increases the sensitivity of the resist by requiring fewer photons to induce a solubility change. CARs often need to include a quencher in addition to the photoacid generator to control the reaction and keep it from diffusing into unexposed regions. The addition of a quencher can cause a decrease in sensitivity, thus leading to the trade-off between sensitivity, resolution, and LWR.[81,86]

Changing the radiation wavelength is not a simple task. It is very costly for the electronics industry as not only the radiation source needs to be changed, but all of the lithographic tools and photoresist materials must also be changed and optimized for the new wavelength. Rather than switching to 157 nm lithography, the industry opted to use immersion lithography to extend the lifetime of 193 nm lithography.[88] This process uses an immersion fluid with a higher refractive index than air to reduce the effective wavelength of the radiation source, but still requires multiple patterning steps to achieve 7 nm features.[79,88] During this process, the photoresist film is immersed in a liquid such as water and exposed to radiation through the immersion fluid.

The market is pushing for integrated circuits with patterned features beyond the resolution limit of 193 nm immersion lithography, and therefore extreme ultraviolet (EUV) lithography with a wavelength of 13.5 nm is being explored as a “next-generation” photolithographic technique to extend the lifetime of Moore’s law. The main advantage of EUV lithography is the short wavelength, which allows for the patterning of much smaller features than the current 193 nm wavelength. The major drawback, however, is the

18 corresponding decrease in photon flux with the decrease in wavelength and the need for increased sensitivity in the resist material. Organic polymer-based resists will no longer be suitable for EUV lithography, so alternate photoresist materials with high EUV absorption cross-sections, like metal-oxo clusters, are being explored.

2.2 Metal-Oxo Photoresists

Metal-oxo and organometallic compounds are currently being considered for EUV photoresist materials. There are several advantages of using metal-based resists rather than organic polymers. For one, metal-oxo clusters are monodisperse and molecular, with sizes on the order of 1 nm. The molecular size directly impacts the size of the features which can be patterned, so smaller molecules are preferred. A monodisperse precursor allows for greater precision of the patterned features, and the variety of structures and compositions of metal-oxo clusters allows for tuning of the film properties. Furthermore, metal-based resists are more sensitive to EUV radiation due to a higher EUV absorption cross section. Though several metal-oxo and organometallic systems are being investigated as photoresist materials, only those containing Hf/Zr and monoorganotin clusters will be discussed here.

2.2.1 Hf/Zr-Oxo Cluster Photoresists

One metal-oxo photoresist system which has demonstrated success with EUV patterning is based on the aqueous chemistry of hafnium and zirconium and is known as HafSOx and ZircSOx.[89] The precursor solution is a mixture of hafnium or zirconium oxychloride, hydrogen peroxide, and sulfuric acid.[89] These aqueous solutions produce high-quality thin films with dielectric properties and can be deposited by spin coating.[89] Hydrogen peroxide in the precursor solutions enables radiation sensitivity and allows for patterning of the resulting thin films.[90] The resulting thin film material has the formula [89] Hf/Zr(OH)4-2x-2y(O2)x(SO4)yꞏqH2O. When the film is exposed to radiation, the O-O peroxide bonds break, promoting condensation and the formation of an amorphous metal- oxide network.[91] This induces the solubility change required for patterning and results in a negative-tone resist. Furthermore, these inorganic films have demonstrated higher sensitivity to EUV radiation and higher resolution than traditional polymer films.[90,92,93]

As mentioned in Chapter 1, the square tetramer cluster topology spontaneously [65–67] assembles upon dissolution of Zr/HfOCl2 at low pH and a number of hafnium sulfate clusters also assemble in solution.[69,74–77] Although the exact nature of the interaction between the metal and hydrogen peroxide is unknown, characterization of these solutions by dynamic light scattering (DLS) and pair distribution function (PDF) analysis of x-ray total scattering show the formation of small cluster species.[91,94]

Hafnium-based resists have demonstrated excellent resolution with patterned features as small as 8 nm.[93] An SEM image of this resist with 8 nm features is shown in Figure 2.2. Despite these advantages, these aqueous solutions are prone to uncontrolled hydrolysis reactions which complicates the understanding of chemical events taking place throughout the lithography process and can cause inconsistent patterning results. The patterning mechanism needs to be well-understood, not only to create an optimized and cost-effective patterning process, but also to avoid contamination of expensive lithography equipment. For this reason, other non-aqueous metal-oxo photoresist precursors are being investigated.

Figure 2.2. 8 nm lines patterned on Hf-based photoresist with EUV lithography.[93] 2.2.2 Organotin Photoresists

Organotin compounds are another class of materials which have shown promise as EUV lithography precursors. As mentioned in Chapter 1, the Sn-C bond is hydrolytically stable and enables solubility in organic solvents, eliminating the background hydrolysis issues faced with aqueous systems. Additionally, the structural diversity of organotin compounds (also demonstrated in Chapter 1) means there are many possible photoresist

20 candidates from this family. Furthermore, tin oxide, the material left behind after patterning, is a well-known semiconductor with applications in transparent conducting oxides.

One organotin system which has demonstrated successful patterning is the alkyltin 2+ football cluster [(RSn)12(O)14(OH)6] , the structure of which is described in Chapter 1. The EUV radiation behavior of thin films of this compound has been studied to determine the effect of different organic ligands and counterions.[53,95] In this case, radiation exposure causes cleavage of the Sn-C bond, leading to condensation and formation of an insoluble tin oxide film (negative resist). The mechanism of the radiation-induced Sn-C bond cleavage is not fully understood, however the organic ligand and counterions are thought to play a role in determining the resist sensitivity and resolution. The best patterning [53] performance was achieved with [(PhSn)12O14(OH)6]Cl2 films and produced 18-nm lines.

The radiation chemistry of various organotin carboxylate drum clusters has been investigated as well.[63,64] These studies found that electron beam lithography did cause cleavage of the Sn-C bond and that the nature of the organic ligands did affect the sensitivity of the resist.[64] However, not all of the organic material was removed from the resist by patterning alone, and additional thermal treatment was required for complete transformation to tin oxide.[64]

The butyltin Keggin ion (β-NaSn13) and its precursor BuSnOOH have also shown promise as thin film nanolithography precursors.[37,96–98] The advantage of these materials is that they are charge neutral, which eliminates the effects of counterions on the radiation chemistry. On the other hand, these materials contain sodium which is not ideal for photoresist materials. However, these precursors are easily characterized and make dense, uniform films which have allowed them to serve as a model system for studying the radiation chemistry of these alkyltin clusters. Radiation exposure studies of BuSnOOH and

β-NaSn13 films showed negative-tone resist behavior and characterization by x-ray photoelectron spectroscopy (XPS) showed a loss of carbon in the films after exposure to x-rays or electrons.[96,98] Additionally, ambient-pressure XPS studies showed an increase in the amount of carbon lost from the films when oxygen is present during exposure.[97,98]

3 Alkyltin Clusters: The Less Symmetric Keggin Isomers

Danielle C. Hutchison, Rebecca D. Stern, Morgan R. Olsen, Lev N. Zakharov, Kristin A. Persson, May Nyman

Dalton Trans. 2018, 47 (29), Page 9804–9813

3.1 Abstract

The Keggin structure is prevalent in nature and synthesis, self-assembled from many metals across the periodic table as both isolated clusters and building blocks of condensed framework oxides. Here we present a one-step synthesis to obtain the sodium- centered butyltin Keggin ion in high yield and high purity, important for mechanistic nanolithography studies. Extensive solution characterization (small-angle X-ray scattering, 1H, 13C and 119Sn nuclear magnetic resonance, electrospray mass spectrometry) also confirms solutions contain only the Na-centered dodecamers. We report three butyltin

Keggin structures: the β-isomer (β-NaSn12), the γ-isomer (γ-NaSn12), and a γ-isomer capped with an additional butyltin (γ-NaSn13). All Keggin ions presented here have the

general formula [NaO4BuSn12(OCH3)12(O,OH)12] (Bu=butyl), and are of neutral charge. The lack of counterions (OH-) facilitates mechanistic lithographic studies without inference from hydrolysis chemistry. The methanol reaction media enables solubility and ligates the cluster, both important to obtain high purity materials. Despite the monospecific nature of the NaSn12 solutions, NMR reveals both isomer interconversion and ligand exchange. DFT computational comparisons of our three isolated structures, the capped β-isomer (β-

NaSn13), along with hypothetical α-isomers (α-NaSn12 and α-NaSn13), showed that the stability ranks β-NaSn12 > γ-NaSn12> α-NaSn12, consistent with experimental observation. The uncapped isomers were computationally determined to be more stable than the respective capped analogues. These clusters provide a unique opportunity to investigate the lower-symmetry Keggin isomers, and to determine structural factors that control isomer selectivity as well as isomer labilization.

3.2 Introduction

The Keggin cluster, first structurally characterized in 1934, is important in metal- oxyhydroxide chemistry in both nature and synthetic inorganic chemistry.[3,9,8,7,6] The Keggin structure consists of four trimers of edge-sharing octahedra, bridged by oxo, hydroxyl, or other coordinating ligands, surrounding a central tetrahedral oxoanion. The five rotational isomers of the Keggin structure are denoted by the Greek letters α, β, γ, δ, and ε, and differ in their symmetry and the nature of connectivity between the trimers

(edge-sharing vs corner-sharing).[99] The  and  isomers feature all corner-sharing

between the trimers and are respectively Td and C3v symmetry. The γ-isomer has one edge-

linkage and five corner-linkages between the trimers, with C2v symmetry. The δ-isomer is also of C3v symmetry, but with three corner linkages and three edge linkages. Finally, the

ε-isomer features all edge-sharing between the trimers, also with Td symmetry.

The group V/VI (Nb, Mo and W) polyoxometalates (POMs) have the most extensive structural variations of the Keggin cluster, with both lacunary fragments and larger aggregates of these fragments. The , and to a lesser extent, the  geometries dominate POM chemistry. Additionally, group 13 polycations (mainly Al and Ga) crystallize predominantly as  and  isomers, and derivatives thereof.[20,25–30] Meanwhile, Keggin structures have been isolated from across the periodic table featuring open shell transition metals, with and without organic ligands.[31,33,34] While the more symmetric isomers; particularly cubic α and ε, are favored by different metals, the tin Keggin ions uniquely favor the less symmetric β and γ isomers. This provides a new perspective for understanding this fundamental structure type, and the chemistries that adopt this structure type.

Two sodium-centered Sn Keggin structures have been isolated previously - we

reported the β-butyltin Keggin ion (β-NaSn13) in 2017, and Reuter reported the γ-Keggin in 1991.[36,37] Most chemically similar to the organotin clusters, organoantimony  and  Keggin ions were also reported.[35] Another dodecameric organotin cluster (termed the football cluster) has been known in the literature for nearly 30 years.[49,47,50] The football 2+ cluster {(RSn)12O14(OH)6} ; denoted Sn12, is composed of a hexameric ring of 5- coordinate tin, capped on each side of the ring by two trimers of octahedral tin. The overall cluster has a charge of 2+ and each tin has one organic terminal ligand (isopropyl, butyl, phenyl, etc.). Recently, the football cluster has shown promise as a precursor for nanolithography of films deposited from these clusters.[53] The tin Keggin ions are likewise of potential importance for this application; in particular for benchmarking understanding

of lithographic mechanisms. Our previously-isolated β-NaSn13 was recrystallized from commercial BuSnOOH, required two recrystallization steps to obtain a pure product, and

24 yielded inconsistent results, due to inhomogeneity in the BuSnOOH crude material.[37,57] The poor reproducibility and low yields hampers further applications studies. Therefore, moving forward, we have now developed a one-step, high yield, 24-hr synthesis that produces a crystalline mixture of β-NaSn12 and γ-NaSn12.

Here we report single-crystal X-ray structures of an uncapped β-Keggin ion (β-

NaSn12), an uncapped γ-Keggin ion (γ-NaSn12), and a capped γ-Keggin ion (γ-NaSn13), as

well as extensive solution characterization of the β,γ-NaSn12 mixture. The only way to determine the Keggin isomers is by single-crystal X-ray diffraction. Electrospray ionization mass spectrometry (ESI-MS) is blind to the different isomers, but reveals the presence of compositionally pure phase NaSn12, while small-angle X-ray scattering (SAXS) shows a monospecific solution, without inference of either larger or smaller aggregates. Nuclear magnetic resonance studies (NMR; 1H, 13C, 119Sn) show complex spectra with broadened, poorly resolved peaks. This is the only solution phase indication that the clusters are not only of mixed isomer forms, but also isomerize and undergo ligand exchange in solution.

We have also employed density functional theory (DFT) computations to help us better understand the relative stabilities of different Sn Keggin structures, particularly considering the apparent ease of β-γ isomerization. We have compared our four experimentally isolated structures (β-NaSn12, β-NaSn13, γ-NaSn12, and γ-NaSn13) as well as two theoretical α-Keggin ions (α-NaSn12 and α-NaSn13). The few studies that computationally model organotin compounds focused primarily on monomeric species.[100–103] Prior theoretical and experimental investigations on Keggin ion stability have focused mainly on polyoxometalates[17,13,10,15,104,5,39] or aluminum polycations,[23,31,105,106] with the exception of one computational study comparing the five rotational isomers of an organoantimonate Keggin.[107] Here we show a stability order of

β-NaSn12> -NaSn12> -NaSn12, and the capped clusters are all less stable, showing the same order. This is both consistent with experimental observation of these structures, and contrasts Keggin isomer stability of other chemistries. We offer some insight as to why these clusters of lower symmetry are more stable than the higher symmetry isomers.

3.3 Experimental

All reagents and solvents were purchased from commercial suppliers and used without further purification.

3.3.1 Synthetic Methods

Synthetic Procedure 1: Stock solutions of 0.1M n-BuSnCl3 (97%) in methanol and 0.1M NaOH in methanol were combined in a 1:4 Sn:OH ratio at room temperature, resulting in

a total solution volume of 10mL (2mL of BuSnCl3 and 8mL of NaOH). Colorless needle- like crystals began to form in the closed vial in 2-4 hours and crystallization was complete within 24 hours. Single crystal and powder X-ray diffraction revealed these crystals to be

a mixture of β-NaSn12 and γ-NaSn12, both having the same formula and charge:

[(BuSn)12(NaO4)(OCH3)12(O)5(OH)7]. Crystals of the two different isomers are indistinguishable by eye. The yield is ~30 mg (65% based on tin).

Synthetic Procedure 2: Crystals of both β-[(BuSn)12(NaO4)(OCH3)12(O)5(OH)7] (β-

NaSn12) and γ-[(BuSn)12(NaO4)(OCH3)11(O)7(OH)6(BuSnOCH3)] (γ-NaSn13) were isolated from the same closed vial of n-BuSnCl3 (96%, 0.11mmol) and NaOH (0.36mmol) in 9.2mL of methanol at room temperature. β-NaSn12 crystallized as colorless “leaf-

shaped” crystals and γ-NaSn13 crystallized as colorless needles. Unfortunately we were

unable to reproduce the crystals of γ-NaSn13. However this structure provides a model for computational studies and adds to the knowledge of isomers favored by alkyltin Keggin ions.

3.3.2 Characterization Techniques

Small Angle X-ray Scattering (SAXS): X-ray scattering data were collected on an Anton Paar SAXSess instrument using Cu-Kα radiation (1.54 Å) and line collimation. The instrument was equipped with a 2-D image plate for data collection in the q = 0.018-2.5 Å- 1 range. The lower q resolution is limited by the beam attenuator. Approximately 20 mmolar solutions were measured in 1.5 mm glass capillaries (Hampton Research). Scattering data of neat solvent was collected for background subtraction. Scattering was measured for 30 min for every experiment. We used SAXSQUANT software for data

26 collection and treatment (normalization, primary beam removal, background subtraction, desmearing, and smoothing to remove the extra noise created by the desmearing routine).

All analyses and curve-fitting to determine Rg, size, shape and size distribution were carried out utilizing IRENA macros with IgorPro 6.3 (Wavemetrics) software.[108] To simulate scattering data from the crystal structure, we used SolX software.[109]

Nuclear Magnetic Resonance Spectroscopy (NMR): 1H, 13C, and 119Sn NMR spectra were collected on a Bruker Ascend spectrometer (500 MHz for 1H, 125 MHz for 13C, 186 MHz for 119Sn) with a 5mm BBO probe at 25.0°C. Chemical shifts are reported in parts per

1 13 [110] 119 million () and H and C spectra are referenced to C6D6 solvent signals. Sn NMR is referenced to SnCl4 in C6D6.

Electrospray Ionization Mass Spectrometry (ESI-MS): ESI-MS was carried out using an Agilent 6230 ESI-MS system comprised of a Time-of-Flight (TOF) mass spectrometer

coupled to an electrospray ionizer. The crystallized Keggin product, β,γ-NaSn12, ([Sn]=1.2 mM), was dissolved in methanol and infused into the ESI-MS system at a flow rate of 0.4 -1 mL min using a syringe pump. The solutions were nebulized with the aid of heated N2 (325 °C) flowing at 8 L min-1 and a pressure of 35 psig (241 kPa). The voltages of the capillary, skimmer, and RT octopole were set at 3500, 65, and 750 V respectively, while the voltage of the fragmenter was set at 100 V. The data were collected in the positive ionization mode. No ion species were detected in the negative ionization mode

Single-Crystal X-ray Diffraction Data Collection Analysis: Diffraction intensities for β-

NaSn12 (CCDC-1838234), γ-NaSn13 (CCDC-1838235), and γ-NaSn12 (CCDC-1838236),

were collected at 153 K (β-NaSn12), 150 K (γ-NaSn13) and 173 K (γ-NaSn12) on a Bruker Apex2 DUO CCD diffractometer using CuK radiation, = 1.54178 Å. Absorption corrections were applied by SADABS.[111] Space groups were determined based on systematic absences. Structures were solved by direct methods and Fourier techniques and refined on F2 using full matrix least-squares procedures. All non-H atoms were refined with anisotropic thermal parameters except C atoms in the disordered terminal butyl groups which were refined with isotropic thermal parameters. All H atoms were refined in

27 calculated positions in a rigid group model. Diffraction for all three samples is very weak at high angles. Even using a strong Incoatec IµS Cu source for β-NaSn12 and γ-NaSn13 it

was possible to collect diffraction data only up 2θmax = 107.9 and 108.78°, respectively. Statistics of the reflections at the high angles are poor and it reduce structure resolution. However, it provides an appropriate number of reflections per number of refined

parameters; 5787/542 (β-NaSn12) and 12160/952 (γ-NaSn13). Terminal n-Bu groups in all structures are highly disordered; there are strong elongations for many C atoms and some groups were treated as being disordered over two positions. Terminal n-Bu groups in all structures were refined with restrictions; the standard bond distances have been used in the refinements as the targets for corresponding bonds. All structures were refined applying a rigid bond restrains; RIGU option in SHELX package. It should be mentioned that on the -3 residual density for γ-NaSn12 there are two relatively high peaks (4.77 and 4.67 eÅ )

located above three µ2-bridging O atoms, but not consistent with any atoms based on chemical information. We considered  isomer disorder in the lattice as an explanation. We overlay the two isomers (including the q-peaks for to determine if these peaks aligned with any atoms of the -isomer, and found this is not the case. All calculations were performed by the Bruker SHELXL-2014/7 package.[112]

Powder X-ray Diffraction Analysis: Powder X-ray diffraction data was performed on Rigaku Ultima-IV at 25°C with Cu Kα radiation. The diffraction patterns were obtained in the range from 3 to 60° (2θ) with a scan speed of 2°/min and step size of 0.03°.

Fourier Transform Infrared Spectroscopy: (FTIR) spectra were recorded on a Nicolet iS10 FTIR spectrometer with a secondary Nicolet iZ 10 module purchased from Thermo Fisher Scientific Inc. The instrument was equipped with a diamond plate for attenuated total reflectance (ATR) measurements. Spectra were collected in air from crushed crystals of

γ-NaSn12. This spectrum can be found in the SI.

3.3.3 Computational Methods

The computational results were obtained using Gaussian 09.[113] The geometry of each cluster was first optimized in the gas phase using the B3LYPfunctional.[114] The basis

28 set 6-31G(d) was used for elements Na, C, H, and O, while the basis set LANL2DZwas used for the element Sn.[115,116] Subsequently, with the same level of theory, a frequency calculation was done to verify the absence of imaginary vibration modes to confirm that the system was in a stable/metastable state. An effective core potential LANL2DZ was used for Sn. The geometry was further optimized in water using the continuum solvation model SMD.[117] The electronic energy was refined using the B3LYP single point with the basis set 6-311+G(d,p) for elements Na, C, H, and O, and basis set LANL2DZ for Sn.[118] The solvation energy was found using B3LYP/6-31G(d) single point with SMD for water.

3.4 Results and Discussion

Synthesis Discussion

The synthesis of Sn12 with hydroxide counterions reported by Eychenne-Baron and Sanchez is lengthy and complex, involving various purification and ion exchange steps.[47] Therefore it is difficult to reproduce, has low yields, and often contains impurities due to incomplete ion exchange. While Sn12 is important to provide a Na-free molecule for microelectronic applications, the NaSn12 clusters provide nearly identical chemistry for film deposition and pattering. Thereby, with its simple and reproducible chemical synthesis reported here, it can serve as a readily-obtained proxy for studying lithographic

mechanisms. The synthetic procedure for NaSn12 involves just one step, at room

temperature, and it can be performed in a 20 ml vial. Stock solutions of 0.1M BuSnCl3 in methanol and 0.1M NaOH in methanol are combined in a 1:4 ratio (usually 2mL of

BuSnCl3 to 8mL of NaOH) at room temperature and crystallization is complete within 24 hours. This synthetic procedure is highly reproducible and does not require any heating,

filtration, or recrystallization. Although our reaction produces a mixture of β-NaSn12 and

γ-NaSn12, both clusters have the same formula and 0-valency, and no counterions or other impurities appear to be present (discussed later). While hydroxides are ideal counterions for film deposition in that they are eliminated by mild heating, their presence can result in undesirable background hydrolysis that compromises sharp turn-on of solubility contrast by radiation-promoted linking of clusters.

Description of Structures

The pertinent data for the three reported structures is summarized in Table 3.1. The clusters β-NaSn12, γ-NaSn12, and γ-NaSn13 all possess the Keggin framework (Figure 3.1).

In these clusters, each trimer consists of three edge-sharing BuSnO5 octahedra with the butyl group serving as the terminal ligand on all tin atoms. The central atom in all three Keggin structures is Na. The twelve bridging oxygens within the trimers (three O2- in each of the four trimers) are methoxy ligands and the bridging oxygens between the four trimers

are oxo or hydroxyl ligands. The BuSn cap of γ-NaSn13 is 5-coordinate with 3 bonds to the cluster and two terminal ligands (butyl and methoxy). It is located in one of the pentagonal windows that are positioned on either side of the two edge-sharing trimers (Figure 3.1). One exception to the locations of the methoxy and oxo/hydroxyl ligands is found in the

structure of γ-NaSn13. One bridging oxygen within the trimer near the capping tin is a hydroxyl ligand rather than the expected methoxy. This could be a result of steric constraints due to the proximity of the capping tin, in addition to the fact that the methoxy ligands appear to be fairly labile (discussed later).

No counterions were located in any of the three lattices, indicating that β-NaSn12,

γ-NaSn12, and γ-NaSn13 all have a neutral charge. Bond valence sum (BVS) calculations

for the oxygens of β-NaSn12, γ-NaSn12, and γ-NaSn13 can be found in Appendix A (Tables A1, A2, and A3 respectively), which distinguished oxos from hydroxyls (see formulae above). Several BVS values for the bridging oxygens were ambiguous, with values falling between 1.2 and 1.4. A cutoff value to distinguish oxos from hydroxyls was determined for each structure, so that each formula contained the correct number of oxos and hydroxyls

to be consistent with a neutral charge. For γ-NaSn12, oxygens with a BVS of 1.35 or greater 2- - were assigned as O , and the remaining oxygens were assigned as OH . For γ-NaSn13, 2- oxygens with a BVS of 1.31 or greater were assigned as O . For β-NaSn12 however, some of the protons must be disordered over symmetrically identical sites in order to have the appropriate charge. Oxygens with a BVS of 1.38 or greater were assigned as O2-, accounting for four of the five required oxo ligands. The two oxos with a BVS of 1.35 are

30 mixed occupancy /O2-/OH-. Hydroxyl/oxo disorder and ambiguity is also common in polyoxometalate crystal structures.

Figure 3.1. Crystal structures of β-NaSn12 (left), γ-NaSn12 (center), and γ-NaSn13 (right). Tin atoms are represented by blue, gray, and orange polyhedra. Blue polyhedra represent trimers which have been rotated 60° with respect to the α-configuration. The orange polyhedron on γ-NaSn13 represents the additional capping tin. Na is shown in turquoise at the center of each cluster. Only the carbon that is bound directly to Sn is shown, for ease of viewing.

Table 3.1: Crystallographic information for reported structures

Compound β-NaSn12 γ-NaSn12 γ-NaSn13 moiety formula [(BuSn)12(NaO4)( [(BuSn)12(NaO4)(O [(BuSn)12(NaO4)(O OCH3)12 CH3)12 (O)5(OH)7] CH3)11(O)7(OH)6(B (O)5(OH)7] uSnOCH3)] empirical formula C60H151NaO28Sn12 C60H151NaO28Sn12 C64H159NaO29Sn13 molecular 2768.07 2768.07 2958.86 weight [g mol−1] T [K] 153±2 173±2 150±2 crystal system Orthorhombic Monoclinic Monoclinic space group Pbcm P21/c P21/n a [Å] 16.1584(6) 19.0758(6) 24.0683(8) b [Å] 24.8434(11) 24.1794(10) 18.8014(6) c [Å] 23.5910(10) 21.0889(7) 24.5975(7) α [°] 90 90 90 β [°] 90 94.047(2) 113.8910(17) γ [°] 90 90 90 volume [Å3] 9470.1(7) 9702.8(6) 10177.1(6) Z 4 4 4 3 ρcalcd [Mg/cm ] 1.941 1.895 1.931 μ [mm−1] 25.273 24.667 25.454 F(000) 5384 5384 5744 crystal size [mm] 0.150 x 0.080 x 0.120 x 0.090 x 0.140 x 0.120 x 0.040 0.080 0.090 radiation [Å] 1.54178 1.54178 1.54178 2 θ range for data 2.735 to 53.947 2.322 to 66.724 2.167 to 54.039 collection [°] index ranges -14≤h≤16 −22≤h≤19 −24≤h≤25 -25≤k≤23 −20≤k≤28 −18≤k≤19 -24≤l≤20 −25≤l≤21 −25≤l≤25 reflections collected 34416 53009 74384 independent 5787 [R(int) = 17123 [R(int) = 12160 [R(int) = reflections/Rint 0.0680] 0.0749] 0.0859] data/restraints/ 5787 / 803 / 542 17123 / 1520 / 909 12160 / 1548 / 952 parameters goodness of fit on F2 1.056 1.051 1.059 R1/wR2 [I>2 σ (I)] 0.0720 / 0.1765 0.0750 / 0.1934 0.0857 / 0.2133 R1/wR2 [all data] 0.1143 / 0.2131 0.1384 / 0.2397 0.1315 / 0.2601 largest diff. peak/ 1.359 and -0.961 4.770 and -1.057 2.033 and -1.744 hole [e Å−3]

Evidence for mixed  isomers in bulk solid

Powder X-ray diffraction of ground crystals shows a broad peak between 5-10 ° 2, and no other significant features (Figure3.2a), while the SEM image (Figure3.2b) shows good crystallinity. Simulated powder patterns for β-NaSn12 and γ-NaSn12 from the single crystal data are similar to each other and consistent with the experimental data. The apparent poor crystallinity of the ground powder is typical of cluster-containing crystals; in this case due to hydrolysis of methoxy ligands upon grinding in air. While analysis of multiple individual crystals revealed a mixture of β-NaSn12 and γ-NaSn12, only one isomer was present in each single-crystal that was assessed. This does not necessarily mean that

β-NaSn12 and γ-NaSn12 do not crystallize in the same lattice, rather it may infer that crystals containing mixed isomers form poor quality crystals, which we were not able to analyze.

Figure 3.2. (a) Experimental powder X-ray diffraction pattern for β,γ-NaSn12 (red) along with calculated powder patterns for β-NaSn12 (green) and γ-NaSn12 (blue). (b) SEM image of β,γ-NaSn12 crystals. 3.4.1 Solution characterization

Mass spectrometery. The NaSn12 clusters are soluble in organic media including alcohols, benzene, toluene, chloroform, THF, DMF, and 2-heptanone, so these solvents were employed for solution characterization. The ten natural isotopes of tin provide very distinctive peak envelopes of the multinuclear clusters, aiding accurate identification of the

33 soluble species detected by mass spectrometry. The TOF ESI-MS analysis of β,γ-NaSn12 crystals redissolved in methanol showed several peak envelopes, with the two most abundant peaks occurring at 2781.8808 and 2750.8555 m/z (Figure 3.3a, positive ionization mode, 100V fragmentation). All peak envelopes have peak separations of 1m/z, indicating a +1 charge. Major peak envelope assignments are shown in Figure 3.3b and

Table 3.2. The formula for β,γ-NaSn12 most similar to the formula determined from X-ray structure comprises ~17% of the observed species (underlined in Table 3.2). Prior mass

spectral analysis of Sn12 show that hydroxide ligands can be replaced by methoxy ligands in methanol solution.[51] The ESI-MS results likewise show extensive exchange between methoxy, oxo and hydroxyl, with multiple m/z formulae that differ only by the number of each ligand. The peak at 2558.6971 m/z contains 11 butyltin groups, likely a product of fragmentation. Additional peak simulations can be found in Appendix A.

Figure 3.3: (a) ESI-MS full spectrum for 0.1mM β,γ-NaSn12 in MeOH. Positive ionization mode, 100V fragmentation. (b) Overlay of experimental (blue) and simulated (red) ESI- MS peak envelopes for most intense peaks of spectrum. The simulated peak envelope is a combination of 9 overlapping peaks.

Table 3.2: Formulae and m/z of Simulated Peak Envelopes.

Formula m/z obs. m/z calc. % Contribution (for overlapping peaks)

1+ [(BuSn)11(NaO4)(OH)17(OCH3)8] 2558.6971 2558.8656 1+ [(BuSn)12(NaO4)(O)10(OH)(OCH3)7*] 2590.7214 2590.7252 1+ [(BuSn)12(NaO4)(O)9(OH)2(OCH3)8*] 2622.7508 2622.7515 1+ [(BuSn)12(NaO4)(O)8(OH)3(OCH3)9*] 2654.7756 2654.7777 1+ [(BuSn)12(NaO4)(O)7(OH)4(OCH3)10*] 2686.8016 2686.8040

1+ [(BuSn)12(NaO4)(O)6(OH)6(OCH3)10*] 2704.8146 3.94% 1+ [(BuSn)12(NaO4)(O)6(OH)5(OCH3)11*] 2718.8283 2718.8302 8.66%

1+ [(BuSn)12(NaO4)(O)4(OH)11(OCH3)9] 2726.8200 1.97%

1+ [(BuSn)12(NaO4)(O)4(OH)10(OCH3)10] 2740.8357 5.51% 1+ [(BuSn)12(NaO4)(O)5(OH)6(OCH3)12*] 2750.8555 2750.8565 17.32%

1+ [(BuSn)12(NaO4)(O)4(OH)9(OCH3)11] 2754.8514 7.87%

1+ [(BuSn)12(NaO4)(O)4(OH)8(OCH3)12] 2768.8671 16.93% 1+ [(BuSn)12(NaO4)(O)4(OH)7(OCH3)13] 2781.8808 2782.8828 34.65%

1+ [(BuSn)12(NaO4)(O)5(OH)3(OCH3)15*] 2792.9036 3.15% 1+ [(BuSn)12(NaO4)(O)5(OH)13(OCH3)9(CH3OH)9] 2940.8136 2940.9830

*Starred formulae have fewer than 40 required ligands (resulting from fragmentation). Underlined formula most similar to structural formula.

X-ray Scattering. Figure 3.4 shows the experimental (red) and simulated (blue) scattering curves for β,γ-NaSn12 in benzene. β-NaSn12 and γ-NaSn12 are indistinguishable by SAXS, so only the simulation for γ-NaSn12 is shown. The experimental scattering is consistent with the simulation, indicating a pure phase solution. Size distribution and modelling of the SAXS data give a diameter of 9.1Å and radius of 4.6Å, respectively (Figs. A10 and A11).

Figure 3.4. Simulated (blue) and experimental (red) small angle X-ray scattering curves for β,γ-NaSn12 in benzene. Discrepancies between the simulated and experimental curves above q = 0.8 Å-1 are due to imperfect background subtraction and solvent scattering.

1 13 119 NMR. Multinuclear ( H, C, Sn) NMR characterization of β,γ-NaSn12, like the ESI-MS data, suggests ligand exchange, in addition to the presence of two different isomers (Figure 3.5, Figure 3.6, Figure 3.7). Peaks were assigned based on the chemical shifts of β- [37] 1 13 NaSn13. The four chemical shifts in the H and C NMR spectra which correspond to the butyl chains on the cluster are broad and poorly defined, similar to the spectra for β- 119 NaSn13. The Sn spectrum shows an overlap of peaks from -450ppm to -500ppm which is consistent with a cluster (or mixture of clusters) that contains only 6-coordinate organotin

species. Ideally, β-NaSn12 (C3v symmetry) should have three different Sn chemical environments and γ-NaSn12 (C2v symmetry) should have a total of four different Sn chemical environments. In addition to the seven symmetrically inequivalent Sn environments, there is a distribution of hydroxide/methoxy ligands amongst the clusters, shown by ESI-MS. This increases the number of chemically inequivalent sites. It also appears that these effects are far-reaching. Even the protons on the butyl chain which are several bonds away from the tin are affected by ligand exchange. Similarly, prior studies suggest that changing the number or locations of protons on divacant lacunary tungstate γ-

Keggin ions significantly changes the 183W NMR spectrum – increasing the number of peaks and shifting the peaks.[104] Those protonation states having the lowest symmetry also exhibited the largest number of chemical shifts. Future studies will focus on monitoring and quantifying ligand exchange of these clusters, and the effect of different solvents.

1 Figure 3.5. H NMR spectrum of β,γ-NaSn12 in benzene-d6.

13 Figure 3.6. C NMR spectrum of β,γ-NaSn12 in benzene-d6.

119 Figure 3.7. Sn NMR spectrum of β,γ-NaSn12 in benzene-d6.

3.4.2 Computational Studies

Although there are five different rotational isomers of the Keggin ion (α, β, γ, δ, and ε), we and others have only been able to experimentally isolate the β and γ isomers, both with and without capping Sn. For those transition metals that adopt the Keggin structure (W, Mo, Nb), the α-isomer is typically the most stable, closely followed by the [5] β-isomer. In contrast, the Al13 Keggin ion favors the ε configuration which also has a

[20] high degree of symmetry; both the  and the  possess Td symmetry. Additionally, the Al-polycations favor the -isomer when capped on the unique corner-linked trimer,[106] and computational studies have shown the oxygens of this face are more basic, promoting binding of the capping metal cation.[31] Meanwhile, organoantimonate clusters favor the δ and ε isomers.[35] To our knowledge, no other systems exist which favor the formation of β and γ isomers. Furthermore, these two isomers have relatively low symmetry and any naturally occurring β or γ Keggin structures have yet to be identified. One might argue that higher symmetry clusters form ordered lattices more readily, which is why they are more frequently observed by this solid-state characterization method that provides absolute identification of isomers. However, this argument cannot be made for these uniquely lower-symmetry Keggin clusters. It is for these reasons that we have turned to computational modeling, to compare the relative stabilities of the α (hypothetical), β, and

γ tin Keggin clusters, as well as the Sn-capped analogues. For the computational comparisons, all clusters were assigned a charge of +1. Although the lack of counterions in the crystal structures indicate that all of the clusters have a neutral charge, adding one additional proton creates a more symmetrical structure (4 oxos and 8 hydroxyls for +1 charge vs 5 oxos and 7 hydroxyls for neutral charge). The +1 charge is also consistent with the species observed by mass spectrometry. Additionally, we use methyl instead of butyl as the terminal Sn-bound ligand for computation in order to simplify the models, since disorder in the butyl chains is prevalent in single-crystal X-ray structures. The theoretical structures for α-NaSn12 and α-NaSn13 are shown in Figure 3.8. The cap position and bonding for α-NaSn13 were chosen to resemble the capping structure of β-NaSn13.

Figure 3.8. Theoretical structures for α-NaSn12 (left) and α-NaSn13 (right). Tin atoms are represented by gray polyhedra. The orange polyhedron represents the additional capping tin on α-NaSn13. Na is shown as a turquoise sphere at the center of each cluster. The stabilities of the six Keggin structures were computationally compared using both the hydrolysis Gibbs free energy and the HOMO-LUMO gap. To reduce the number of degrees of freedom when modeling these systems and ease the comparison between these three structures, the uncapped α, β, and γ-isomers were all assigned the same formula 1+ and charge: [(MeSn)12(NaO4)(OCH3)12(O)4(OH)8] . The capped α, β, and γ-isomers were assigned slightly different formulae to reflect what was experimentally observed. The simulated β-NaSn13 was assigned a formula of 1+ [(MeSn)12(NaO4)(OCH3)12(O)8(OH)4(Sn(H2O)2)] which is exactly consistent with the experimental crystal structure. The experimentally characterized cap for γ-NaSn13 is

2+ [Sn(Me)(OCH3)] with 3 bonds to the cluster, and was likewise used for the computational 1+ study with a formula of [(MeSn)12(NaO4)(OCH3)12(O)6(OH)6(Sn(Me)(OCH3))] .

We used parameterized continuum solvation models to accurately determine solvation energies, using B3LYP.[119] We determined the hydrolysis Gibbs free energy in solution by using a thermodynamic cycle in which the hydrolysis energy is the sum of the corresponding gas-phase Gibbs free energy (ΔGgas) and the Gibbs free energies of solvation

(ΔGsolv; Equation 3.1). The gas-phase Gibbs free energy contains a correction term that takes into account the enthalpy, entropy, and temperature of the system when a frequency analysis is conducted. The term “n” is the coefficient of that species. An example of this thermodynamic cycle using β-NaSn12 is shown in Scheme 3.1. The dielectric constant for the solvent model was set to ~78.36, consistent with water, and we will exploit other solvents and/or explicit solvent molecules in our future research.[5,120,121] The computational results are summarized in Table 3.3.

Equation 3.1. Hydrolysis Gibbs free energy (∆Gaq) obtained from thermodynamic cycle in Scheme 3.1.

Scheme 3.1. Thermodynamic cycle for β-NaSn12.

Table 3.3: Hydrolysis Gibbs free energy (kcal/mol) and HOMO-LUMO gaps of NaSn12 and NaSn13 isomers.

Hydrolysis Gibbs ΔG Destabilization HOMO- Free Energy (kcal/mol) by capping LUMO Gap (kcal/mol) (kcal/mol) (eV) β-NaSn12 327.4 0 ─ 6.24 γ-NaSn12 337.4 10.0 ─ 5.91 α-NaSn12 342.7 15.3 ─ 6.20 β-NaSn13 364.9 37.5 37.5 5.28 γ-NaSn13 361.2 33.8 23.8 5.80 α-NaSn13 369.7 42.3 27.0 5.17

The most stable clusters are expected to have a relatively low hydrolysis Gibbs free energy and a large HOMO-LUMO gap. The hydrolysis Gibbs free energy differences between clusters are similar in magnitude to those reported prior.[5,122] The stability ranking of NaSn12 is β-NaSn12 > γ-NaSn12 > α-NaSn12; and the uncapped clusters are more stable than their capped counterparts, under the current assumption of no explicit solvent or counterions. These results are consistent with our experimental findings. β-NaSn12 appears to be the most favored, as it crystallizes under two different synthetic conditions, while γ-

NaSn12 and γ-NaSn13 are each crystallized under different conditions and always mixed with β-NaSn12. The calculated relative instability of the α-isomer is also consistent with our experimental results, as it has not been ever observed experimentally.

In the β isomer, all of trimers are joined by corners at six points, while the γ-isomer has one edge-linking and five corner-linkings. The Sn-Sn edge-sharing distance is 3.26Å, while the Sn-Sn corner-sharing distance is ~3.4-3.5 Å, causing relative instability of the - isomer from electrostatic repulsion. The lower stability of the capped clusters is contradictory to previous studies that showed capping stabilizes polyoxometalates and aluminum polycations by respectively decreasing the cluster’s overall charge or stabilizing a more negative cluster face.[31,123] Since all simulated clusters have the same charge in this study, the former explanation is irrelevant. Experimentally capping is charge neutral

because the attachment site is converted from hydroxide to oxide. The cap on the γ-NaSn13

41 yields a short Sn-Sn distance of 3.15Å between the cap and the nearest tin, and destabilization may be a result of repulsion. Capping of α-NaSn13 distorts the polyhedra surrounding the capping Sn (Figure 3.8), which may contribute to its destabilization. The

cap on β-NaSn13 does not lead to any distortion nor close Sn-Sn distances, so reasons for its destabilization are unclear.

When we consider Keggin structures from across the Periodic Table, the charge of the metal cations shows a consistent trend with preferred isomers. Highly charged transition metals (Nb5+, Mo6+, W6+) adopt the α and  isomers, which have the greatest number of corner-linked trimers and large distances between metal centers. Rotation from  to  is favored for more negative clusters; with either reduction of the W/Mo, or a lower- charged central cation.[10] Metals with a lower charge (Al3+, Cr3+, Ga3+) form the δ and ε- isomers (or derivatives thereof) which have the greatest degree of edge-sharing between trimers and smallest distance between metal centers.[20,29–31,124] Additionally, the -Keggin is represented as a building block in Fe3+-ferrihydrite.[8] Edge-sharing increases stability by adding rigidity to the molecule; so this is the energetic tradeoff for increased metal- metal repulsion. Presumably the lower-charged cations permit this edge-sharing, since the repulsion is less. All of the clusters in this investigation contain Sn having a 4+ oxidation state, which lies between metal charges for experimentally observed Keggin ions, consistent with progression of the favored isomers from  to .

3.5 Conclusions

To summarize, our one-step, reproducible synthesis of pure and high yield sodium- centered butyltin Keggin ions enables studying thin film deposition and lithographic

mechanisms, which is a topic of forthcoming papers. Unlike the related Sn12, β,γ-NaSn12 is crystallized readily without counterions, both increasing the simplicity of the synthesis, as well as improving the yield, purity, and reproducibility. The difficulties that arise in reproducibility of synthesis of related Sn12 was made evident in the study of these Keggin ions. Because the ligands (methoxy and hydroxyl) are extremely labile, crystallizing the clusters with ligands that can also serve as counterions exacerbates the challenge. Although the Sn-Butyl bond is hydrolytically inert, the bridging hydroxide ligands of alkyl tin

42 clusters readily exchange in response to solution conditions. Not only are the ligands of the BuSn Keggin ions labile, but also the connectivity between the trimers. Computational results provide essential insights toward helping understand the nature of this unique system that favors lower symmetry Keggin isomers. With this study, we have computationally and experimentally confirmed that the tin Keggin ions represent the only Keggin ion family (thus far) that favors the clusters of lower symmetry; providing a balance between corner-linking to minimize cation-cation repulsion, and edge-linking to maximize stability via bond formation.

3.6 Acknowledgements

This work was funded by the National Science Foundation, Center for Chemical Innovation, grant number CHE-1606982. We also acknowledge the support of the Oregon State University NMR Facility funded in part by the National Institutes of Health, HEI Grant 1S10OD018518, and by the M. J. Murdock Charitable Trust grant #2014162.

4 Synthesis and Characterization of a Butyltin Keggin Ion with a Rare 4- Coordinate Ca Center

Danielle C. Hutchison[a], Rebecca D. Stern[b], Lev N. Zakharov[a], Kristin A. Persson[b,c], May Nyman[a]*

Submitted to Inorganic Chemistry, 2019.

4.1 Abstract

Alkyltin clusters are exploited in nanolithography for fabrication of microelectronics. The alkyltin Keggin family is unique amongst Keggin clusters across the periodic table – they appear to favor the lower symmetry β and γ isomers, rather than the highly symmetrical α and ε isomers. Therefore, the alkyltin Keggin family may provide important fundamental information about formation and isomerization of Keggin clusters. We have synthesized and structurally characterized a new butyltin Keggin cluster with a 2+ 2+ tetrahedral Ca center, fully formulated [(BuSn)12(CaO4)(OCH3)12(O)4(OH)8] (β-

CaSn12). The synthesis is a simple one-step process. Extensive solution characterization including electrospray ionization mass spectrometry, small angle x-ray scattering, and 1 13 119 multinuclear ( H, C, and Sn) NMR shows β-CaSn12 is pure-phase and stable. This differs from the prior-reported Na-centered analogues that always form a mixture of β and γ isomers, with facile interconversion. Therefore, this study has clarified prior confusion over complex spectroscopic and crystallographic characterization of the Na-centered analogues. Density functional theory calculations showed a stability order of γ-CaSn12 < γ-

NaSn12 < β-CaSn12 < β-NaSn12; the  analogue is always more stable than , consistent with experiment. Notable outcomes of this study includes a rare tetrahedral Ca- coordination, a Na-free alkyltin cluster (important for microelectronics manufacturing), and a better understanding of Keggin families built of different metal cations.

4.2 Introduction

The Keggin ion is an important structure in inorganic chemistry, which consists of

a central tetrahedral {MO4} unit surrounded by 12 metal-oxo octahedra arranged into four trimers. This structure is formed by metals across the periodic table including W, Mo, Nb, Fe, Al, Cr, Sb, and Sn.[3,18,20,31–36] There are five rotational isomers of the Keggin ion[4], denoted by the Greek letters α, β, γ, δ, and ε, which differ in their symmetry and connectivity between the trimer units. The α isomer has all four trimers connected by corner sharing and has Td symmetry. Rotating one trimer by 60° reduces the symmetry to C3v in the β isomer, which also has all trimers connected by corners. A second trimer rotation of

60° yields the γ isomer (C2v symmetry), with two trimers corner-linked and two edge-

45 linked. Successive rotation of the third and fourth trimers by 60° results in the δ and ε isomers, respectively. The δ isomer has C3v symmetry with three trimers connected by edges and one linked by corners. All four trimers of the ε isomer are edge-linked with Td symmetry.

Recently, we added three new sodium-centered butyltin members to the Sn Keggin family – a β isomer (β-NaSn12), a γ isomer (γ-NaSn12), and a γ isomer capped with an

[125] additional Sn (γ-NaSn13). Other members of this family include our previously isolated

[37] Sn-capped butyltin β isomer (β-NaSn13) and the first reported alkyltin Keggin ion - an

isopropyltin γ isomer.[36] More recently, Zhu et al. stabilized the  isomer of Na-centered

[38] butyltin Keggin ion with borate-capping, including a bicapped NaSn14 structure. The alkyl-Sn Keggin family exhibits differentiating structural features. While the majority of metal cations favor the higher symmetry α (Nb, W, Mo, Fe) and ε (Al, Sb) structures, as well as the δ isomer (Cr, Al), we and others have only been able to experimentally isolate alkyltin clusters as the rarer β and γ isomers.[36–38,125] Isomerization between α and β tungstate Keggin ions has been investigated experimentally[13] and computationally.[10] Both conclude that 1) the α isomer is generally favored, and 2) α → β isomerization is n- favored with increasing XO4 charge on the central tetrahedron (X=central tetrahedral cation). Meanwhile, the growing alkytin Keggin cluster family is the only one thus far that favors the lowest symmetry γ isomer. In addition, our previously isolated butyltin Keggin clusters showed evidence of facile interconversion between isomers in solution,[125] complicating characterization by tin, carbon, and proton NMR.

In our previous study, we hypothesized that the charge of the 6-coordinate addenda metal may play a role in determining which Keggin isomer would be more favorable.[125] We also investigated the influence of capping for enhancing isomerization.[58] However, it is possible that the size and charge of the central atom of the Keggin ion may also be an important factor in promoting isomerization. Zhang et al. showed computationally that the central cavity for organoantimonate Keggin ions increases in size when going from the α to β isomer.[107] In addition, the central cavity increases in size as the charge of the central

[126] tetrahedron {MO4} becomes more positive. It is noteworthy that the Na central cation

46 for our butyltin Keggin clusters has a significantly larger ionic radius and lower charge (0.99Å, 1+) than the central cations of the most common Keggin structures; e.g. Si (0.26Å, 4+) and P (0.17Å, 5+).[127] In order to examine the effect of the central cation on the properties of the Keggin ion, we have crystallized a new butyltin β-Keggin ion with Ca at 2+ the center (formulated [(BuSn)12(CaO4)(OCH3)12(O)4(OH)8] , β-CaSn12 henceforth) via a one-step synthesis, and studied its solution behavior. Calcium is similar in size to sodium (ionic radius of 1.00Å)[127] which allows it to fit in the central cavity of the alkyltin Keggin

ion. To our knowledge, β-CaSn12 is the only Keggin ion which has been isolated having a Ca2+ ion in the tetrahedral center. The typical coordination number for Ca is 6-8 and it is rare to find examples of Ca with a coordination number as low as 4.[128] In fact, there is only one molecular compound in the literature containing the Td {CaO4} unit, and it required the use of very bulky alkoxide ligands to stabilize the Ca2+-centers.[129] Here we

provide complete solution characterization of β-CaSn12 by electrospray ionization – mass spectrometry (ESI-MS), multinuclear (1H, 13C, and 119Sn) NMR, and small angle x-ray scattering (SAXS).

Additionally, we report here room-temperature aging studies in the solution and solid-state for the Ca-centered butyltin Keggin cluster, and heating studies monitored by variable-temperature NMR. Understanding the long-term stability and solution behavior of these organotin materials is important for applications such as high resolution nanolithography.[53] These aging and heating studies provided insight to the prior suspected isomerization (or lack thereof) occurring in solution for these alkyl-Sn Keggin clusters. Over time, the methoxy ligands of the clusters (derived by synthesis in methanol) undergo hydrolysis and are replaced with hydroxyl ligands. This was monitored in solution by 1H NMR and in the solid state by FT-IR. We have also employed density functional theory (DFT) computational studies to compare the relative stabilities of the Ca- and Na- centered clusters.

4.3 Results And Discussion

Synthesis

The Na-centered Keggin clusters used for aging studies were synthesized by combining stock solutions of 0.1M BuSnCl3 in MeOH and 0.1M NaOH in MeOH at room

[125] temperature in a 1:4 ratio, as reported prior. β-CaSn12 was synthesized by combining a

0.1M solution of BuSnCl3 in MeOH with 2 molar equivalents of solid Ca(OH)2 in a Parr reactor which was heated to 100°C for 24 hours. Though the stoichiometry is the same between the two reactions (1 Sn: 4 OH), the reactant concentrations are much higher in the

β-CaSn12 synthesis as Ca(OH)2 is added as a solid rather than a solution. In addition, the

solvothermal conditions are necessary to dissolve the Ca(OH)2 and force the calcium into the low-coordination environment at the center of the cluster. The reaction does not proceed at room temperature.

4.3.1 Single crystal structure of β-CaSn12

The single crystal x-ray structure of β-CaSn12, fully formulated 2+ - [(BuSn)12(CaO4)(OCH3)12(O)4(OH)8] ꞏ 2[OH ], is shown in Figure 4.1. This compound has the same structure and ligand arrangement as our previously reported β-NaSn12, but has Ca at the center instead of Na.[125] Elemental analysis by SEM-EDX was used to

confirm the presence of Ca and lack of Na in the crystals of β-CaSn12 (Table B2). Each Sn has a terminal butyl chain and the bridging μ2-oxygens between trimers of the Keggin are

either oxo or hydroxyl ligands. The bridging μ3-oxygens found within each of the four trimers are methoxy ligands which come from the methanol reaction solution. The hydrolysis of these methoxy ligands will be discussed in detail later.

Figure 4.1. Single crystal x-ray structure of β-CaSn12 2+ [(BuSn)12(CaO4)(OCH3)12(O)4(OH)8] . Gray and blue polyhedra represent Sn; Ca is shown in teal, O in red, and C in black. Butyl chains have been eliminated with only the Sn-bound carbon shown, for ease of viewing.

β-CaSn12 has a charge of 2+ with two hydroxide counterions. The charge was determined based on ESI-MS data and bond valence sum (BVS) calculations (Table B1). Only one OH- was identified in the crystal lattice, but based on SQUEEZE[130] analysis there is room in the crystal packing for an additional hydroxide counterion. As is commonly observed for metal-oxo clusters, the BVS values for bridging oxo ligands are ambiguous, with values ranging from 1.16 to 1.41, due to disordered protons which cannot be located crystallographically. In order to be consistent with an overall charge of 2+, oxos with a BVS of 1.37 or higher were assigned as O2- and the remaining oxos assigned as OH-. Changing the central cation does have an effect on the bond lengths of the four central oxos. For β-NaSn12, the average bond distance from Na to the surrounding oxos is 2.32Å, 2+ and for β-CaSn12 the average Ca-O bond length is 2.28Å. The higher charge of Ca leads to shorter bonds to the surrounding oxos. Consequently, the Sn-O bonds of the central oxos of β-CaSn12 are lengthened. The average Sn-O bond length for the central oxos of β-CaSn12 is 2.11 Å and for β-NaSn12 it is 2.08 Å. These values are summarized in Table 4.1. The BVS of the central Ca is 1.7. Although this is a bit lower than the expected value of 2.0, this may be due to the fact that the parameters[131] used in the BVS calculations are empirically derived from metal oxide lattices, not molecules. As mentioned previously, the coordination environment of the central Ca is unique and there are no other structures in the literature which contain tetrahedral calcium within an inorganic framework.

Table 4.1. Average bond lengths for O atoms of central tetrahedron

Avg. Na/Ca-O bond length (Å) Avg. Sn-central O bond length (Å)

β-NaSn12 2.32 (1) 2.08 (1)

β-CaSn12 2.28 (2) 2.11 (1)

4.3.2 Solution Characterization

Electrospray Ionization – Mass Spectrometry (ESI-MS)

ESI-MS of β-CaSn12 crystals dissolved in methanol (Figure 4.2a) shows one main ion species at 1259.85 m/z, corresponding to a molecule with a 2+ charge (based on peak- to-peak separations in the peak envelope). Minor peaks are observed at 1282.83 m/z (2+ charged species) and 2557.63 m/z (1+ charged species). The peak envelopes corresponding to species with a 2+ charge were simulated as a combination of 8 overlapping peaks, shown in Figure 4.2b and listed in Table 4.2. The largest contributor to the overlapping peaks has 2+ the formula [(C4H9Sn)12(CaO4)(O)11(OH)4(OCH3)2] and comprises 59% of the total peak simulation. Though the intact ion with all ligands present was not detected, all of the assigned peaks have formulae which correspond to species containing 12 butyltin groups, a calcium center, and varying numbers of oxo, hydroxyl, and methoxy ligands. All of the formulae in Table 4.2 have fewer than the 40 ligands required for the complete cluster, but this is likely due to the fragmentation process. Unfortunately, the peak at 2557.63 m/z was too low in intensity to accurately assign. No peaks were detected in the negative ionization mode. Full peak assignments can be found in Figures B1-B8.

Figure 4.2. a) Full spectrum of β-CaSn12 in methanol. Positive ionization mode, 100V fragmentation. b) Experimental (red) and simulated (black) peak envelopes for main ion species. The simulation is a combination of 8 overlapping peaks.

Table 4.2: ESI-MS peak assignments for β-CaSn12 in methanol

Formula Observed Calculated % m/z m/z Contribution 2+ [(C4H9Sn)12(CaO4)(O)11(OH)6] 1245.83 1245.79 6.7% 2+ [(C4H9Sn)12(CaO4)(O)10(OH)8] 1254.85 1254.79 13.4% 2+ [(C4H9Sn)12(CaO4)(O)11(OH)4(OCH3)2] 1259.85 1259.80 59.0% 2+ [(C4H9Sn)12(CaO4)(O)11(OH)3(OCH3)3] 1266.85 1266.81 7.4% 2+ [(C4H9Sn)12(CaO4)(O)8(OH)11(OCH3)] 1279.84 1279.81 4.0% 2+ [(C4H9Sn)12(CaO4)(O)10(OH)4(OCH3)4] 1282.83 1282.83 3.4% 2+ [(C4H9Sn)12(CaO4)(O)8(OH)10(OCH3)2] 1286.83 1286.82 2.7% 2+ [(C4H9Sn)12(CaO4)(O)10(OH)3(OCH3)5] 1289.81 1289.83 3.4% Nuclear Magnetic Resonance (NMR) Spectroscopy

One of the most important features of CaSn12 is that it predominantly forms the β isomer. The Na-centered clusters always crystallize as a mixture of β and γ isomers, and solution state NMR characterization of these compounds suggests that there may be interconversion between isomers as well.[125] Having mainly one isomer present in solution significantly reduces the difficulty of solution characterization. The 1H and 119Sn NMR

spectra for β-CaSn12 are shown in Figure 4.3a and b, respectively and β,γ-NaSn12 is also included to highlight the difference in the complexity of NMR spectra when multiple isomers are present vs one. Expanded spectral views and 13C data can be found in Figures 1 B9-B11. The H NMR spectrum for β-CaSn12 shows much sharper and more well-defined 119 peak splitting than does β,γ-NaSn12, but the Sn NMR is most telling. Due to the mixture of isomers present in β,γ-NaSn12 as well as suspected isomerization in solution, there are

51 many different Sn environments present, resulting in several Sn peaks in the 6-coordinate 119 region. The Sn spectrum of β-CaSn12 has just three sharp peaks in the 6-coordinate region in a 2:1:1 ratio, consistent with the different Sn environments of the β isomer.

The three Sn sites are shown in Figure 4.3b, and each peak has been assigned to its respective site based on integration and 119Sn-117Sn coupling constants. Coupling 183 constants were also used to assign peaks in the W NMR of the β-AlW12O40 Keggin ion.[13] Site 1 (blue) contains the three Sn of the 60° rotated trimer at the top of the cluster

that defines the molecular C3 axis, site 2 (green) contains the 6 Sn in the middle “belt”, and site 3 (gray) contains the 3 Sn of the trimer opposite to those of site 1. The peak at - 469.5ppm with an integrated value of 2 (normalized to peak 1) corresponds to the 6 Sn in the middle belt of the cluster (site 2). The peak at -471.0ppm has a coupling constant of 147 Hz and corresponds to site 3, while the peak at -474.6ppm has a coupling constant of 175 Hz and corresponds to site 1. The larger coupling constant indicates a larger distance between neighboring Sn sites. The Sn-Sn distance between sites 1 and 2 in the crystal

structure of β-CaSn12 is 3.8Å and the Sn-Sn distance between sites 2 and 3 is 3.4Å. These values are also summarized in Table 4.3. The asymmetry of the satellites on peaks 2 and 119 3 of the Sn spectrum of β-CaSn12 will be discussed in further detail later. On the other

hand, we are unable to assign specific Sn sites for β,γ-NaSn12 due to the multitude of peaks, including satellite peaks from 119Sn-117Sn coupling.

1 119 Figure 4.3. a) H NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue). b) Sn NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue). Inset: depiction of three Sn sites of β- CaSn12.

Table 4.3: Coupling constants, Sn-Sn distances, and peak assignments for 119Sn peaks of β-CaSn12

δ(ppm) Distance between Sn-Sn distance to Assignment satellites (Hz) middle layer (Å) -469.5 165 - Site 2 -471.0 147 3.4 Site 3 -474.6 175 3.8 Site 1

Small Angle X-ray Scattering (SAXS)

Shown in Figure 4.4, small angle x-ray scattering data (SAXS) of β-CaSn12 in THF (red) shows a pure and monodisperse solution which is consistent with the simulated scattering curve (black) based on the single crystal structure. The deviation between the simulated and experimental curves around q=0.4Å-1 is due to a structure factor, or ordering of clusters in solution. This behavior has also been observed with our previously isolated β-NaSn13 Keggin cluster.[37] The experimental data were fit with a single-population spherical model which also accounted for the structure factor. The radius determined from this model was 4.92Å, which is consistent with the cluster radius determined from the single crystal structure (5.0Å). This model and its fit to the experimental data can be found in Figure B12.

Figure 4.4. Small angle x-ray scattering data for β-CaSn12 in THF (red) and simulated scattering curve (black). 4.3.3 Aging Studies

These butyl-Sn Keggin clusters are being considered for potential applications in solution-deposited thin films, so it is important to understand the aging behavior of these clusters both in solution and in the solid-state in air. We observed previously that the [125] methoxy ligands of β,γ-NaSn12 exhibited hydrolysis in solution. Here we describe the hydrolysis process for β-CaSn12 as a function of solvent composition (C6D6, CDCl3, 9:1 1 C6D6:CD3OD, and 9:1 CDCl3:CD3OD) via H NMR. FT-IR tracks the hydrolysis process in air via the intensity of the methoxy C-O stretch absorbance at ~1050 cm-1.

NMR

1 H NMR is a commonly used technique to monitor and quantify the extent of reactions in the solution state[21,132] and is ideal for measuring the hydrolysis of methoxy ligands to hydroxyl ligands. The peaks corresponding to bound methoxy and free methanol 1 protons are clearly visible in the H NMR spectrum of β-CaSn12. Since methanol is a product of the hydrolysis of methoxy ligands to hydroxyl, the relative integrated areas of the two peaks over time provides information about the reaction rate, and how it is affected 1 by different solvents. Figure 4.5 shows the H NMR spectra of β-CaSn12 in C6D6 and 9:1

C6D6:CD3OD for fresh solutions (0 days) through day 4 of aging. The analogous spectra recorded in CDCl3 and 9:1 CDCl3:CD3OD are shown in Figure B13. The broad peak at approximately 3.6 ppm corresponds to the bound –OCH3 protons and the sharp peak at 3.0-

3.2 ppm corresponds to methanol CH3- protons. These peaks were integrated a total of three times each, and their average relative areas compared for each spectrum (Figure 4.6).

1 Figure 4.5. H NMR spectra following aging of β-CaSn12 in C6D6 and 9:1 C6D6:MeOD.

β-CaSn12 exhibits hydrolysis in all of the solvents studied and shows that hydrolysis is the slowest in C6D6 and fastest in CDCl3. Adding excess deuterated methanol increases the rate of hydrolysis in benzene but does not increase the rate in chloroform. The difference in the rate of hydrolysis in different solvents is most likely due the amount of residual water present in solution. The deuterated solvents were not dried prior to the aging study and likely contain trace amounts of water, driving the hydrolysis process.

Figure 4.6. Average % methoxy ligands hydrolyzed vs number of days for β-CaSn12 in various solvents. Water is about twice as miscible with chloroform as it is with benzene, consistent [133] with the faster hydrolysis in CDCl3. Adding MeOD to C6D6 likely increases the amount of water in solution, thus increasing the rate of reaction. In the case of CDCl3/MeOD, it is

55 possible that the MeOD does not add a significant amount of water compared to what was already present in the CDCl3. Unfortunately it is difficult to quantify the amount of residual 1 water in solution, especially in CDCl3, as the H NMR peak corresponding to H2O (~1.56ppm) overlaps with the butyl chain protons on the cluster. There is a very small peak 1 at approximately 0.3ppm in the H spectra of β-CaSn12 in C6D6 which corresponds to water in solution (Figure B9).

Integrating the peaks of interest is straightforward in benzene where the peaks are well-separated, but is more complex in chloroform where the bound and free –OCH3 peaks overlap. As a further complication, the spectra with excess MeOD exhibit a second free methanol resonance slightly upfield of the main methanol peak, potentially due to the formation of CH3OD after exchange of CD3OD with –OCH3 ligands on the cluster. This is

not a major concern for the spectra in C6D6/MeOD, but does add another layer of

complexity to the integration of overlapping peaks in spectra taken in CDCl3/MeOD. Peaks also tend to broaden and overlap as aging progresses in all solvents. Despite these difficulties, the deviation between separate integration measurements is estimated to be between 1 and 8%. Perhaps with the exception of CDCl3, the curves appear to plateau; - - - suggesting an equilibrium is reached between OH and OCH3 /OCD3 ligation. The CDCl3 perhaps continues to absorb water with aging, driving the equilibrium forward towards more hydrolysis. However, since equilibrium appears to be reached, even in excess methanol, this suggests neither ligand is favored strongly, and both are labile. However, as noted prior, the methoxy-ligand seems necessary for crystallization; as this is the only form we and others have observed the various butyltin Keggin ions.[37,38,125]

FT-IR

FT-IR spectra for β-CaSn12 were taken at various time points after isolating crystals of the compound from the mother liquor. Spectra were collected at 1 day, 7 days, 14 days, 21 days and 28 days. Figure 4.7 shows an overlay of all the spectra focused on the C-O -1 stretch of OCH3 ligand (normalized to Sn-O absorbance at 530 cm ). Full spectra are

reported in Figure B15. The intensity of the C-O peak for β-CaSn12 fluctuates between measurements and does not follow a clear trend. This is consistent with the NMR studies;

56 methanol that is retained in the solid following hydrolysis or from the mother liquor can - - undergo exchange between OH and OCH3 ligation.

Figure 4.7. FT-IR spectra showing the change in absorbance of the C-O stretch of solid β- CaSn12 over time. In Situ Heating Monitored by NMR

1 119 A solution of β-CaSn12 in C6D6 was heated in situ and monitored by H and Sn NMR. Beginning at 30°C, 1H and 119Sn spectra were collected (approximately 1 hour data collection time) and then the temperature was increased by 10°C. 1H and 119Sn spectra were collected at each temperature. Once the temperature reached 60°C, the temperature was held constant for 3 hours. Three 119Sn spectra and four 1H spectra were collected during this time. Figure 4.8 shows the resulting spectra from this heating experiment.

The 1H spectra (Figure 4.8a) do not show much change other than the sharpening of peaks upon heating. Interestingly, the 119Sn spectra show the appearance of new peaks (denoted by arrows in Figure 4.8b) upon heating, indicating the formation of additional isomers in solution. The peak at -470.7ppm in the 119Sn spectrum was originally thought to be a satellite of the neighboring, more intense peak. However, this “satellite” increases in intensity with heating, meaning that it is due to the presence of an additional isomer. The

presence of this peak at room temperature indicates that the β-CaSn12 product is not purely β isomer, but additional isomers are only present in trace amounts. As there are four peaks which increase in intensity due to heating, we hypothesize that these are due to formation

57 of the γ isomer at higher temperatures. Upon cooling this heated solution to room temperature, the mixture of isomers is preserved (Figure B14). Initially this result was puzzling, since the synthesis is performed at elevated temperature, yet yields predominantly the β-form. However, there are multiple differences between the synthesis and the above-described NMR experiment including 1) solvent, 2) excess Ca(OH)2 in the synthesis, 3) reaction byproducts in the synthesis (i.e. Cl- from the butyltin trichloride). Despite this isomerization at elevated temperature, the mixture remains predominantly the β-form, illustrating the slower isomerization of the Ca-centered cluster compared to the Na-centered Keggin.

1 119 Figure 4.8. (a) H NMR spectra of β-CaSn12 in C6D6 heated in situ. (b) Sn NMR spectra of β-CaSn12 in C6D6 heated in situ. Gray arrows indicate growth of the γ isomer with heating. 4.3.4 DFT Computational Analysis

The β and γ isomers of the Na- and Ca-centered Sn Keggin ions were simulated and their hydrolysis Gibbs free energy and HOMO-LUMO gaps were determined using a thermodynamic cycle in which the hydrolysis energy is the sum of the corresponding gas- phase Gibbs free energy (ΔGgas) and the Gibbs free energies of solvation (ΔGsolv; Equation 4.1). The gas-phase term contains a correctional value that takes into account the enthalpy, entropy, and temperature of the system when a frequency analysis is performed. Parameterized continuum solvation models were used to determine the solvation energies, modeling the system in water with a dielectric constant of ~78.36. The thermodynamic

58 cycle for the calcium centered Keggin isomers is shown in Scheme 4.1, where “n” is the coefficient of that species.

Equation 4.1: Hydrolysis Gibbs free energy (ΔGaq) obtained from the thermodynamic cycle in Scheme 1.

Scheme 4.1: Thermodynamic cycle for all isomers of CaSn12.

These results are shown in Table 4.4 and are in order from most to least stable. We consider clusters to be more stable if they have a relatively low hydrolysis Gibbs free energy and a large HOMO-LUMO gap. To reduce the number of degrees of freedom, methyl was used as the terminal ligand on each tin rather than butyl. Both isomers of NaSn12 were assigned 1+ the same formula and charge: [(MeSn)12(NaO4)(OCH3)12(O)4(OH)8] . Similarly, both isomers of CaSn12 were assigned a formula and charge of 2+ [(MeSn)12(CaO4)(OCH3)12(O)4(OH)8] . These formulae are consistent with the single crystal structures and species detected by ESI-MS.

Table 4.4: Calculated hydrolysis Gibbs free energy and HOMO-LUMO gap for β and γ isomers of NaSn12 and CaSn12

Cluster Hydrolysis Gibbs free ΔG (kcal/mol) HOMO-LUMO 1+ energy (kcal/mol) Normalized to β-NaSn12 gap (eV) 1+ β-NaSn12 -367.0 0.0 6.21 2+ β-CaSn12 -354.7 12.3 6.23 1+ α-NaSn12 -350.5 16.5 6.18 1+ δ-NaSn12 -347.1 19.9 6.00 1+ γ-NaSn12 -344.8 22.2 5.67 2+ α-CaSn12 -342.8 24.2 6.20 2+ γ-CaSn12 -338.6 28.4 5.73 2+ δ-CaSn12 -338.3 28.7 6.01 1+ ε-NaSn12 -306.9 60.1 5.70 2+ ε-CaSn12 -297.2 69.8 5.81

Based on DFT calculations, the Na-centered clusters are more stable than their corresponding Ca-centered analogues, likely due to the fact that Na more easily adopts tetrahedral coordination than Ca. Additionally, the extra positive charge of the Ca-analogue generally decreases the stability. Importantly, the stability of the β-Keggin ion is highest for both the Na and Ca centered clusters, in agreement with experiment. In addition to having lower symmetry, the γ isomer has two edge-sharing trimers with a short metal-metal distance between the two Sn centers which decreases the overall stability of the cluster.

The difference in hydrolysis Gibbs free energy between the β and γ isomers for NaSn12 is

22.2 kcal/mol and for CaSn12 it is 16.1 kcal/mol, suggesting that CaSn12 may isomerize

more readily than NaSn12. Yet, we observe the opposite to be true in our experimental studies. We hypothesize that this difference may be attributed to kinetic factors related to the interconversion between isomers. A hydrolysis Gibbs free energy difference of more than 10 meV/atom is required to confidently state that one isomer is more stable than another.[106,107,117,122,134] Since these clusters each have 145 atoms, an energy difference of more than 34 kcal/mol signals definite favoritism of one isomer. The computational stability ordering differs slightly from what we observe experimentally, but the differences in hydrolysis Gibbs free energy are below the 34 kcal/mol threshhold. Therefore, we can only state with certainty that the β isomers for both systems are more stable than the ε analogues. Returning to the isomerization studies of the classic POM Keggin ions[10,13], n- isomerization is more favorable with increasing XO4 charge. Consistent with this, we 7- 6- observe greater isomerization behavior in the {NaO4} centered cluster than the {CaO4} centered cluster.

Although the structures of β-NaSn12 and β-CaSn12 differ only in the identity of the central cation, their behavior exhibits a few key differences. For one, crystallized β-CaSn12 contains only trace amounts of other isomers and only undergoes observable isomerization in solution when heat is applied. Conversely, NaSn12 crystallizes as a mixture of β and γ isomers and undergoes significant isomerization in solution at room temperature. Though the mechanism of isomerization is unknown, having a cation with a higher charge at the center of the Keggin ion appears to strengthen the metal-oxygen bonds of the central tetrahedron and decrease the likelihood of isomerization occurring.

4.4 Conclusion

In summary, we have successfully encapsulated Ca2+ into the central tetrahedral cavity of a butyltin β Keggin ion. This is the only member of the growing family of alkyltin + Keggin clusters with a central cation other than Na . In addition, β-CaSn12 provides the first example of 4-coordinate Ca2+ in an inorganic structure. While alkyltin clusters have been investigated for use in nanolithography, the central Na+ in the Keggin clusters has arisen concern about contamination of expensive lithography fabrication facilities. This study shows that other metal substitutions are possible, that may be later exploited in functionality of tin oxide materials. Solution characterization by ESI-MS, SAXS, and multinuclear NMR shows that β-CaSn12 is free of impurities and contains only trace

amounts of other isomers. In comparison to the analogous NaSn12 Keggin clusters that

always present as mixtures of γ and β isomers, CaSn12 is robustly dominated by the β isomer, and it must be heated in solution to drive conversion to a different isomer, presumably γ. This is consistent with the tungstate Keggin ions, in that the lower-charged central cation promotes α → β isomerization more readily. In the current study, the Na- centered cluster seems to establish an equilibrium between isomers at room temperature, while the Ca-centered cluster does not isomerize until heat is applied. DFT computations also show a difference in cluster stability with a change in the central cation. Overall, the size and charge of the central cation has a significant effect on the stability and isomerization of the Sn Keggin cluster as a whole. These experimental and computational studies provide additional fundamental insight into the factors which influence Keggin formation and isomer stability.

4.5 Materials and Methods

All reagents and solvents were purchased from commercial suppliers and used without further purification.

4.5.1 Synthesis of β-CaSn12

In a Parr reactor, 15mL of a 0.1M stock solution of n-BuSnCl3 (95%) in methanol

was combined with 0.22g solid Ca(OH)2 (4 eq. OH per Sn) to yield a thin, white suspension. The suspension was heated solvothermally at 100°C for 24 hours. Colorless

61 needle crystals were harvested from the walls of the reaction vessel. Yield approx. 200mg (45%).

4.5.2 Computational Methods

The computational results were obtained using Gaussian 09.[113] The geometry of each cluster was first optimized in the gas phase using the B3LYP functional.[114] The basis set 6-31G(d) was used for elements Na, Ca, C, H, and O, while the basis set LANL2DZ was used for Sn.[115,116] A frequency calculation was done to guarantee that there were no imaginary vibration modes confirming that the system was in a stable/metastable state. The effective core potential LANL2DZ was used for Sn. The geometry was further optimized in water using the continuum solvation model SMD and a solvent accessible surface (SAS).[117] The electronic energy was refined using the B3LYP single point with the basis set 6-311+G(d,p) for elements Na, Ca, C, H, and O, and basis set LANL2DZ for Sn.[118] The solvation energy was found using B3LYP/6-31G(d) single point with SMD in water and a solvent excluding surface (with minimum radius for added spheres as 0.20 and overlap index as 0.89).

4.5.3 Characterization Techniques

1 13 [110] million () and H and C spectra are referenced to C6D6 and CDCl3 solvent signals. 119 Sn NMR is referenced to SnCl4 in C6D6.

Electrospray Ionization Mass Spectrometry (ESI-MS): ESI-MS was carried out using an Agilent 6230 ESI-MS system comprised of a Time-of-Flight (TOF) mass spectrometer

coupled to an electrospray ionizer. The crystallized Keggin product, β-CaSn12, ([Sn]=1.2 mM), was dissolved in methanol and infused into the ESI-MS system at a flow rate of 0.4 -1 mL min using a syringe pump. The solutions were nebulized with the aid of heated N2 (325 °C) flowing at 8 L min-1 and a pressure of 35 psig (241 kPa). The voltages of the capillary, skimmer, and RT octopole were set at 3500, 65, and 750 V respectively, while the voltage of the fragmenter was set at 100 V. The data were collected in the positive ionization mode. No ion species were detected in the negative ionization mode.

Single-Crystal X-ray Diffraction Data Collection Analysis: X-ray diffraction intensities

for β-CaSn12 (CCDC-1949774) were collected at 173 K on a Bruker Apex2 CCD diffractometer using CuK radiation, 1.54178 Å. Space group were determined based on systematic absences. Absorption corrections were applied by SADABS.[111] Structure was solved by direct methods and Fourier techniques and refined on F2 using full matrix least- squares procedures. All non-H atoms were refined with anisotropic thermal parameters. All H atoms were refined in calculated positions in a rigid group model. H atoms in bridge –OH groups and in counter-ions were not found and not taken into consideration. The molecule is located in the crystal structure on a mirror plane. The found structure has a low resolution. The terminal t-Bu groups in the structure are significantly disordered and X-ray diffraction at high angles from crystals of this compound is very weak. Even using a strong

Incoatec IµS Cu source reflections with I/(I) more than 1 were only up to 2θmax = 100˚. Only these reflections were used in the final refinement. Anyway it provides appropriate number of reflections per refined parameters, 5031/509. The terminal t-Bu groups are highly disordered and its thermal ellipsoids are significantly elongated. One of these groups was refined as disordered over two positions related by a mirror plane. All terminal groups

63 were refined with restrictions on its geometry; the standard interatomic distances were used as targets for corresponding contacts. Some short H…H contacts in the structure don’t correspond to real contacts between the H atoms, but related to the fact that these groups were refined as a solid unit and its disorder was not resolved. One of two -OH counter-ions in the structure is disordered over two positions related by an inversion center. We could not find a position of the second –OH counter-ion needed to provide charge balance. Checking the structure by SQUEEZE[130] indicated that in the structure there is an additional empty space suitable for the second –OH counter-ion. Convergence of the final refinement was not very good due to a lot of highly disordered groups in the structure. RIGU option was used for the final refinement of the structure. Although the found structure has a low resolution it is given the paper because the framework of the structure seems to be clear and confirms the main chemical results discussed in the work. All calculations were performed by the Bruker SHELXL package.[112]

Crystallographic Data for β-CaSn12: C60H154CaO29Sn12, M = 2804.18, 0.11 x 0.06 x 0.03 mm, T = 173 K, Orthorombic, space group Pbcm, a = 16.1505(6) Å, b = 24.9896(11) Å,

c = 23.5117(11) Å, V = 9489.2(7) Å3, Z = 4, Dc = 1.963 Mg/m3, μ(Cu) = 25.663 mm-1,

F(000) = 5464, 2θmax = 100.0°, 20969 reflections, 5031 independent reflections [Rint = 0.1002], R1 = 0.0856, wR2 = 0.2263 and GOF = 1.042 for 5031 reflections (509 parameters) with I>2(I), R1 = 0.1455, wR2 = 0.2640 and GOF = 1.051 for all reflections, max/min residual electron density +0.885/-0.648 eÅ-3.

Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray Spectroscopy (EDS): Electron micrographs and atomic ratios (%) of the crystalline materials were obtained from a Quanta 600F instrument (FEI) combining a scanning electron microscope and an energy-dispersive x-ray spectrometer.

4.6 Acknowledgments

5 Isomerization of Na-Centered Alkyltin Keggin Clusters

Danielle C. Hutchison, May Nyman

In Preparation, 2019

5.1 Abstract

The alkyltin Keggin system is distinctive in that it is the only Keggin family that favors the rarer β and γ rotational isomers. This provides a unique opportunity to study the factors affecting Keggin isomerization from a fundamental standpoint. Aging studies of 1 β,γ-NaSn12 by H NMR and FT-IR showed that methoxy ligands undergo hydrolysis upon exposure to moisture and that this process is reversible with the addition of excess methanol. Furthermore, we were able to promote isomerization of these clusters by mild heating in solution. Characterization by 23Na NMR of a heated solution showed five chemical shifts which we believe correspond to all five of the Keggin rotational isomers simultaneously in solution.

5.2 Introduction

The Keggin ion is quickly becoming ubiquitous in inorganic chemistry – both in nature and synthesis. The Keggin structure consists of 12 metal-oxo octahedra, organized [3] into four trimers, surrounding a central cation in Td coordination. The Keggin ion has [4] five rotational isomers ; the α isomer has Td symmetry and all four trimers are connected

by corner sharing. A 60° rotation of one trimer results in the β isomer which has C3v symmetry with the C3 axis passing through the rotated trimer, and all trimers are still connected by corner sharing. A second trimer rotation of 60° yields the γ isomer (C2v symmetry) in which two trimers are connected by edge sharing and the remaining trimers are connected by corner sharing. Successive rotations of the remaining two trimers by 60° gives the δ and ε isomers, respectively. The δ isomer has C3v symmetry and 3 edge-sharing

attachments and one corner-sharing attachment. The ε-isomer has Td symmetry with all edge-sharing trimers. With each successive trimer rotation, the number of edge-sharing attachments increases.

The factors influencing isomerization of the Keggin ion have long been investigated both experimentally and computationally. As Keggin ions have been synthesized from metals across the periodic table including W, Mo, Nb, Fe, Al, Cr, and Sb,[3,18,20,31–36] this provides the opportunity to investigate the effects of the identity of the metals comprising

67 the Keggin as well as synthetic conditions and post-synthetic modification on the preferred isomer form. W and Mo Keggins typically favor the α isomer, with the β isomer becoming n- more stable upon reduction and with increasing charge of the {XO4} central tetrahedron (X=central cation).[10,12,13,15,126] On the other hand, Sb and Al have an inverted stability [21,23,106,107] order with the ε isomer being most stable. Yet, isomerization of the Al13 Keggin ion was promoted by addition of glycine and the isomerization process was monitored in 27 [21] solution by Al NMR. Chemical shifts for α, γ, δ, and ε isomers of Al13 have all been observed in the same solution by 27Al NMR.[21]

The family of alkyltin Keggin ions is unique in that only the β and γ isomer forms have been identified by single crystal x-ray diffraction. This family includes a sodium- centered isopropyltin γ isomer[36], two borate capped Na-centered butyltin γ-Keggins[38], as well as five new structures reported by our group – a capped sodium-centered butyltin β

isomer (β-NaSn13), a capped sodium-centered butyltin γ isomer (γ-NaSn13), uncapped

sodium-centered butyltin β and γ isomers (β-NaSn12, γ-NaSn12), and an uncapped calcium- [37,125,135] centered butyltin β isomer (β-CaSn12). β-NaSn12 and γ-NaSn12 always crystallize as a mixture of both isomers and cannot be separated. This product will be referred to as

β,γ-NaSn12 and the structures are shown in Figure 5.1. These clusters both have the same [125] formula and neutral charge [(BuSn)12(NaO4)(OCH3)12(O)5(OH)7]. The bridging µ2-

oxygens between trimers are oxo or hydroxyl ligands, and the bridging µ3-oxygens within the trimers are methoxy ligands. The methoxy ligands are derived from the methanol synthesis solvent and are important for crystallization. The terminal butyl chain on each tin atom enables solubility in organic solvents.

Figure 5.1. Structural representations of (a) β-NaSn12 and (b) γ-NaSn12. Blue and gray polyhedra represent Sn. Rotated trimers are shown in blue to differentiate isomers. Na is shown in turquoise, C is black, and O is red. Butyl chains have been shortened to the Sn- bound C for ease of viewing. Based on our studies, it appears that the charge and size of both the octahedral and tetrahedral metals play a role in determining the preferred Keggin isomer. When examining Keggin ions from across the periodic table, a trend begins to emerge. Those metals having a higher charge (W, Mo) tend to adopt the α isomer, as this allows for the greatest separation between highly charged metal centers. Metals with a lower charge (Cr, Al) tend to adopt the δ and ε isomers as the small size and low charge of these metals facilitates stabilization by edge-sharing. The Sn in our alkyltin clusters has an intermediate charge (4+) and therefore prefers the intermediate isomers, with a mix of corner and edge sharing to optimize the balance between the repulsion of metal centers and stabilization by edge- sharing.[125] Our studies also showed that the charge and size of the central cation plays a role in isomer determination. With Na+ as the central cation and without borate caps, a mixture of isomers is always observed.[37,125] We have also observed possible interconversion between isomers in solutions of Na-centered butyltin Keggins. However, changing the central cation to Ca2+ shortens the Sn-O bonds to the central tetrahedron and thereby stabilizes the β isomer.[135] We have also showed by 119Sn NMR that isomerization [135] of β-CaSn12 can be promoted by mild heating in organic solvent. Additionally, previous studies have shown that the methoxy ligands on the Keggin clusters can hydrolyze in solution, further complicating the characterization when multiple isomers are present.[135]

Here we investigate the effects of aging and heating on the hydrolysis of methoxy 1 23 119 ligands and isomer behavior of the β,γ-NaSn12 product. We employ H, Na, and Sn NMR for solution characterization and FT-IR for characterization in the solid state.

5.3 Results and Discussion 5.3.1 NMR Aging

1 The aging behavior of β,γ-NaSn12 was monitored by H NMR as a function of time

and solvent (C6D6, 9:1 C6D6:CD3OD, CDCl3, 9:1 CDCl3:CD3OD). Spectra were collected each day for five consecutive days, including the day the samples were prepared (Figure 5.2, Figure C1). Over time the bridging methoxy ligands of the cluster are hydrolyzed,

replacing the –OCH3 with –OH and forming CH3OH as a product. Thus, the peaks

corresponding to the bound –OCH3 and free CH3OH protons were integrated and the relative areas compared to determine the extent of the reaction in each solvent or solvent mixture. The integration of the peaks of interest is straightforward for the spectra taken in

C6D6 where the peaks are well-separated; however, it is more difficult for the spectra in

CDCl3 where the –OCH3 and CH3OH peaks overlap. Additionally, the spectra with excess MeOD exhibit a second free methanol resonance slightly upfield of the main methanol

peak, potentially due to the formation of CH3OD after exchange of CD3OD with –OCH3 ligands on the cluster. While this does not impact integration of the spectra taken in

C6D6/MeOD, it does add difficulty to the integration of overlapping peaks for spectra taken

in CDCl3/MeOD. Furthermore, peaks tend to broaden and overlap with the progression of time in all solvents. Despite these difficulties, each peak was integrated three times and the deviation between separate integration measurements is estimated to be between 1 and 5%. The percentage of methoxy ligands hydrolyzed was plotted as a function of time and solvent and is shown in Figure 5.3. An analogous experiment was performed for β-CaSn12 and was reported prior.[135]

1 Figure 5.2. H NMR spectra of β,γ-NaSn12 over time in (a) C6D6 and (b)CDCl3. As aging progresses, the relative area of the free methanol peak increases while the area of the -OCH3 peak decreases, indicating that methoxy ligands are hydrolyzed in all solvents. The peaks corresponding to the butyl chain protons are sharpest at 0 days and become broader over time. This can be attributed to a reduction in symmetry of the overall cluster as methoxy ligands are partially replaced with hydroxyl ligands.

Figure 5.3. Average percentage of methoxy ligands hydrolyzed vs number of days for β,γ- NaSn12 in various solvents.

Similar to β-CaSn12 reported prior, the rate of hydrolysis as a function of solvent is [135] as follows: C6D6 < C6D6/MeOD < CDCl3 = CDCl3/MeOD. The reaction proceeds most slowly in the least polar solvent, C6D, and the reaction rate is increased by adding

71 deuterated methanol. This trend can be attributed to residual water in the solvent, as solvents were not dried prior to use. Water is more soluble in chloroform than in benzene[133], and the addition of deuterated methanol likely adds water to the solution as well. The percentage of ligands hydrolyzed appears to level off after approximately 2 days, suggesting that the reaction may reach equilibrium. The reaction rates for CDCl3 and

CDCl3/MeOD appear to be similar, due to the fact that the reaction in these two solvents appears to reach the equilibrium condition rather quickly.

We also investigated the reversibility of this hydrolysis reaction by 1H NMR

(Figure 5.4). The hydrolyzed Keggin was a sample of crystalline β,γ-NaSn12 which had

been aged in air for several months and then dissolved in C6D6. As shown in Figure 5.4, the peaks corresponding to the methoxy and methanol protons (~3.5 and 3.2ppm, respectively) are very low in intensity compared to the butyl chain protons. Ideally, the

combined areas of the –OCH3 and CH3OH peaks should be equal to the area of peak a of the butyl chain (methyl group furthest from the Sn) as there are twelve methoxy ligands and twelve butyl chains per cluster in the crystal structure. For the hydrolyzed sample, the combined area of the methoxy and methanol peaks is about 10% of that of peak a. The butyl chain proton peaks show some defined splitting indicating that the hydrolyzed clusters may have a higher degree of symmetry once nearly all of the methoxy ligands have been exchanged with hydroxyls. A stoichiometric amount (12 eq., 2.8μL) of methanol was 1 added to the solution of hydrolyzed β,γ-NaSn12 in C6D6 and a second H NMR spectrum

was collected (blue spectrum in Figure 5.4). The intensities of both the bound –OCH3 and

free CH3OH proton peaks increased significantly after the addition of methanol to the solution, showing that some of the previously hydrolyzed ligands were replaced with methoxy groups. The butyl chain proton peaks broaden after the addition of methanol again due to a loss of symmetry as only some of the hydroxyl ligands have been exchanged with

methoxy ligands. We also noted that the hydrolyzed product had lower solubility in C6D6, with some material remaining undissolved. Within 5-10 minutes of adding the stoichiometric amount of methanol, all material dissolved and the solution became clear. Presumably, the replacement of methoxy ligands with hydroxyl makes the cluster slightly

72 less hydrophobic and therefore less soluble in deuterated benzene. Adding methanol reverses the hydrolysis process and increases the solubility in nonpolar solvents. It is interesting to note that the addition of methanol reverses the hydrolysis reaction for hydrolyzed clusters, but adding deuterated methanol appears to increase the rate of hydrolysis in the aging studies. We attribute this to the residual water in deuterated methanol having a greater effect on the forward hydrolysis reaction. Additionally, any – 1 OCD3 groups which are replaced on the cluster will not be observed by H NMR.

1 Figure 5.4. H NMR spectra of hydrolyzed β,γ-NaSn12 in C6D6 before (red) and after adding 12 eq MeOH (blue).

5.3.2 FT-IR Aging

The aging behavior of solid β,γ-NaSn12 in air was monitored by FT-IR (Figure 5.5, Fig C2). Spectra were collected on the crystalline product one day after isolation from the mother liquor and every 7 days for four weeks. The intensity of the C-O stretch (1040 cm- 1), representing the methoxy ligands, decreases over time. This is consistent with the

73 hydrolysis of methoxy ligands due to ambient humidity. Unlike β-CaSn12 reported prior, [135] the C-O stretch of β,γ-NaSn12 steadily decreases over time.

Figure 5.5: FT-IR spectra showing the change in intensity of the C-O stretch for solid β,γ- NaSn12 over time.

5.3.3 NMR HEATING

The solution behavior of β,γ-NaSn12 in C6D6 was monitored as a function of temperature by in situ 1H, 119Sn, and 23Na NMR (Figure 5.6 -Figure 5.7). Starting at 30°C, the temperature was held constant while collecting 1H, 23Na, and 119Sn spectra – approximately 1 hour of data collection. After this time, the temperature was increased by 10°C and all three spectra collected again at the new temperature. This process was repeated until the temperature reached 60°C and then the temperature held constant for 3 hours while spectra were collected.

1 119 Figure 5.6. (a) H NMR and (b) Sn NMR spectra of β,γ-NaSn12 C6D6 heated in situ. As shown in Figure 5.6a, the 1H NMR peaks broaden with heating, but do not change significantly. The 119Sn spectrum (Figure 5.6b) shows that the peaks present at 30°C decrease in intensity and new peaks become more intense as the temperature increases. This suggests that the β/γ isomers are being converted to different isomer forms. Unfortunately we are unable to assign any of the chemical shifts to specific isomers due to the complexity of the 119Sn spectrum.

23 Figure 5.7. (a) Na NMR spectra of β,γ-NaSn12 in C6D6 heated in situ. (b) Expanded view of 23Na spectrum collected at 60°C after 3 hours. The 23Na NMR spectra (Figure 5.7) are the most interesting in terms of highlighting structural differences in the Sn Keggin isomers at different temperatures. These 23Na peaks change significantly with heating, splitting and becoming sharper at higher temperatures. At 60°C, five peaks can be seen in the 23Na spectrum, potentially

75 corresponding to each of the 5 Keggin isomers. Since 23Na is quadrupolar (I=3/2), the peak width is correlated to the symmetry of the surrounding environment and therefore the identity and symmetry of the Keggin isomer. The two sharp peaks at 13.2 and 13.4 ppm

may correspond to the Td α and ε isomers, the two moderately broad peaks at 13.6 and 14.0

ppm could correspond to C3v β and δ, and the very broad peak from 15-20 ppm corresponds 119 to the C2v γ isomer. The increase in the number of Sn chemical shifts at high temperatures is also consistent with an increase in the number of Keggin isomers. The 23Na chemical shifts, peak widths and isomer assignments are summarized in Table 5.1. The peak widths for overlapping peaks were determined using the peak fitting tool in the ACD NMR processing software. Asymmetry of the peaks may be due to changes in the symmetry of the overall cluster as methoxy ligands are hydrolyzed, though these are several bonds away from the central Na. Unfortunately there is very little literature available regarding solution-state 23Na NMR, and the coordination environment of Na in these structures is unique, so we are unable to definitively assign peaks to specific isomers. Additionally, these heated solutions have not yielded any crystals, presumably due to the mixture of isomers present.

Table 5.1: Peak Widths and Isomer Assignments of Chemical Shifts in 23Na NMR Spectrum of Heated NaSn12 Solution

Chemical Shift (ppm) Peak Width (Hz) Isomer Assignment 13.17 13.4 α/ε 13.39 12.5 α/ε 13.61 21.8 β/δ 14.03 39.2 β/δ 17.07 350 γ

After these heated solutions are cooled to room temperature, the sharp peaks in the 23Na NMR spectrum are no longer present (Figure C3). The four overlapping peaks between 12-15 ppm (assigned to α, β, δ, and ε isomers) decreased in intensity and combined into one broad peak, and the broad peak from 15-20 ppm (assigned to the γ isomer) increased in intensity. Presumably, the more symmetric α and ε isomers are stabilized at high temperatures and convert to the more stable β or γ isomers upon cooling. This is

76 consistent with our previous computational studies which showed that the β and γ isomers of the Na-centered clusters are most stable.[125,135]

5.4 Conclusion

The Na-centered alkyltin clusters have provided a unique opportunity to study the effects of solvent, aging, and heating on the isomerization of the Keggin structure. Exposure to moisture in the form of ambient humidity or residual water in organic solvents promotes hydrolysis of methoxy ligands on the cluster, and this reaction can be reversed by adding additional methanol. We have observed through multinuclear NMR that interconversion between isomers occurs at room temperature and is further promoted by mild heating. Characterization by 23Na NMR shows five chemical shifts which we believe correspond to all five Keggin isomers simultaneously in solution.

5.5 Experimental

β,γ-NaSn12 was synthesized by combining stock solutions of 0.1M BuSnCl3 in MeOH and 0.1M NaOH in MeOH in a 1:4 ratio at room temperature, as reported prior.[125]

Reversibility Studies: A 15mg (5.8µmol) sample of crystalline β,γ-NaSn12 was aged in air

for approximately four months and then dissolved in 0.75mL of C6D6. A theoretical

formula of [(BuSn)12(NaO4)(OH)12(O)5(OH)7] with a molar mass of ~ 2600g/mol was used

to represent the fully hydrolyzed β,γ-NaSn12. A small amount of this material did not dissolve. After collecting a 1H NMR spectrum, 2.8µL (12 eq, 69 µmol) of MeOH was added to the solution. Approximately 5 – 10 minutes after mixing, the solution became completely clear and another 1H NMR spectrum was acquired.

Nuclear Magnetic Resonance Spectroscopy (NMR): 1H, 23Na, and 119Sn NMR spectra were collected on a Bruker Ascend spectrometer (500 MHz for 1H, 132 MHz for 23Na, 186 MHz for 119Sn) with a 5mm BBO probe at 30.0°C. Chemical shifts are reported in parts per

1 [110] 23 million () and H spectra are referenced to C6D6 and CDCl3 solvent signals. Na NMR 119 is referenced to NaCl in D2O and Sn NMR is referenced to SnCl4 in C6D6.

5.6 Acknowledgements

6 Peroxide-Promoted Disassembly-Reassembly of Zr-Polyoxocations

James A. Sommers1, Danielle C. Hutchison1, Nicolas P. Martin1, Karoly Kozma, Douglas A. Keszler, and May Nyman 1These authors contributed equally

J. Am. Chem. Soc. 2019, 141 (42), 16894–16902.

6.1 Abstract

Zr/Hf aqueous-acid clusters are relevant to inorganic nanolithography, metal- organic frameworks (MOFs), acid-catalysis and nuclear fuel reprocessing, but only two topologies have been identified. The (Zr4) polyoxocation is the ubiquitous square aqueous Zr/Hf-oxysalt of all halides (except fluoride), and prior-debated for perchlorate. Simply

adding peroxide to Zr oxyperchlorate solution leads to a striking modification of Zr4, yielding two structures identified by single-crystal X-ray diffraction. Zr25, isolated from a reaction solution of 1:1 peroxide:Zr, is fully formulated

[Zr25O10(OH)50(O2)5(H2O)40](ClO4)10xH2O. Zr25 is a pentagonal assembly of 25 Zr- oxy/peroxo/hydroxyl polyhedra and is the largest Zr/Hf cluster topology identified to date.

Yet is yet completely soluble in common organic solvents. ZrTd, an oxo-centered

tetrahedron fully formulated [Zr4(OH)4(μ-O2)2(μ4-O)(H2O)12](ClO4)6xH2O, is isolated

from a 10:1 peroxide:Zr reaction solution. The formation pathway of ZrTd and Zr25 in water were described by small-angle X-ray scattering (SAXS), pair distribution function

(PDF), and electrospray ionization mass spectrometry (ESI-MS). Zr4 undergoes

disassembly by one equivalent of peroxide (per Zr) to yield small oligomers of Zr25 which assemble predominantly in the crystalline lattice, an unusual crystal growth mechanism.

The self-buffering acidity of the Zr-center prevents Zr25 from remaining intact in water.

Identical species distribution and cluster fragments are observed in assembly of Zr25 and upon redissolution of Zr25. On the other hand, the 10:1 peroxide:Zr ratio of the ZrTd reaction solution yields larger prenucleation clusters before undergoing peroxide-promote disassembly into smaller fragments. Neither these larger cluster intermediates of ZrTd nor the smaller intermediates of Zr25 have yet been isolated and structurally characterized, and they represent opportunity to expand this new class of group IV polycations, obtained by peroxide reactivity and ligation.

6.2 Introduction

Metal-oxo clusters isolated from water are akin to molecular metal oxides in their role as active catalysts,[136–138] sorbents for sequestering aqueous contaminants,[139–141] and materials for energy production and storage.[142] In their molecular form, metal-oxo clusters

80 are ideal for elucidating reaction pathways of materials assembled from water, where cluster topology may predicate material form and function.[8,33,143,144] Metal-oxo clusters are long-known modular building blocks for zeolites,[145,146] and more recently for thin films,[37,147] porous frameworks,[148] and metal-organic frameworks (MOFs).[149,150]

Cluster-topologies that self-assemble in aqueous Zr/Hf solutions are limited in scope. The Zr/Hf clusters include tetrameric and hexameric forms; the former is ubiquitous in aqueous hydrohalic acid, and debated in perchloric acid solution.[151,152] The Clearfield- [65–68] Vaughan tetramer (Zr4), dubbed as such in reference to its initial isolation, is an acidic 8+ polycation formulated [Zr4(OH)8(H2O)16] (Figure 6.1), commonly described and sold

commercially as ‘ZrOX2ꞏ8H2O’ (X=halide). The hexamer topology, consisting of a

Zr6(O,OH)8 core with antiprismatic-Zr polyhedra arranged in a pseudo-octahedron, [70] presumably requires a bridging ligand to convert Zr4 to Zr6 in water, and has been widely exploited in MOFs,[72] now commercially available as UiO-66. Both Zr-MOFs and [153–155] Zr4 are robust and inexpensive acid catalysts. Despite the importance of these clusters, there has been minimal success in expanding the family of Zr/Hf- polyoxohydroxocations. In addition to growing our fundamental knowledge of aqueous group IV chemistry, new Zr/Hf cluster topologies could lead to new MOF reticulations from pre-formed clusters, and new homogenous and heterogeneous acid catalysts. Additionally, an improved understanding and control of aqueous speciation of Zr/Hf can innovate new separations chemistries[156,157], critical for nuclear technologies.

Combining sulfate with Hf/Zr4 in aqueous acid yields several cluster forms with nuclearities of 11, 17 and 18.[69,75,76,158] However, these clusters are largely insoluble, due to the network-forming and neutralizing tendency of divalent sulfate. High-resolution

lithography is achieved from acidic Zr/HfOCl2-sulfate-peroxide deposition solutions. The peroxide is converted to oxide upon radiation exposure and provides solubility contrast to create patterned coatings for fabrication of microelectronic devices.[89] From highly alkaline solution, we have isolated a Hf-hexamer ring linked by bridging peroxide,[73] which provided a model for understanding the development chemistry of HfO2 lithography. Aside from the radiation response of the peroxide in lithography, its coordination structure

81 and chemistry in the acidic deposition solutions and films has not been elucidated, due to lack of relevant structural information. However, recent X-ray scattering data suggest that [91] the sulfate and/or peroxide dissociates the Zr/Hf4 cluster via labile coordination; but the separate roles of sulfate and peroxide in speciation cannot be decoupled. Moreover, we do not understand why the maximum peroxide that can bind to Hf/Zr in these deposition solutions is 1:1,[91] where increased peroxide should correlate with improved radiation sensitivity. To elucidate the structure and reactivity roles of peroxide in these deposition solutions, we have targeted isolation of Zr/Hf clusters from acidic-peroxide solutions. In addition to achieving this goal, these studies have provided a means to open up the perpetually limited family of Zr/Hf aqueous polynuclear clusters. Here we describe a synthetic pathway to obtain group IV peroxide clusters that are completely inorganic and simple in composition (containing only Zr/O/H) and highly soluble in both water and organic solvents. We present two structurally characterized clusters – 20+ 6+ [Zr25O10(OH)50(O2)5(H2O)40] (Zr25) and [Zr4(OH)4(μ-O2)2(μ4-O)(H2O)12] ∙(ZrTd), as

well as sub-units of Zr25 described by small-angle X-ray scattering (SAXS) and electrospray ionization mass spectrometry (ESI-MS). Both are isolated from a parent

solution of Zr4 with peroxide, where an interplay of Zr concentration and the ratio of

peroxide:Zr4 dictates the obtained structure (Figure 6.1), and correlates with the proposed disassembly-reassembly processes. Zr25 is by far the largest Zr/Hf polycation cluster

reported, while ZrTd is an oxo-centered tetrahedron, a similar topology observed [159] previously for Th(IV). Intriguingly, subunits of Zr25 appear to assemble directly in the solid state lattice from water, and redissolve as these same subunits, a rare crystal growth process for polynuclear clusters. These clusters, their assembly pathways, and their solution behavior provide a path forward for diversifying aqueous polycation cluster chemistry, important in cluster-based materials such as MOFs. They also serve as models for elucidating peroxide-based lithography processes.

Figure 6.1. Schematic summarizing the reaction pathways of peroxide-promoted disassembly-reassembly of the ubiquitous Zr4 polycation to obtain new peroxide ligated cluster forms. The species in the yellow box are identified by SAXS and/or ESI-MS, but have not yet been isolated. Color scheme (used throughout the paper): Ligands--turquoise = peroxide, green = hydroxide, gold = oxo, red = aqua. Black = Zr. For the ‘prototype ZrO2’ clusters that form prior to ZrTd, the nuclearity is approximately 22 Zr-centers (see text). 6.3 Results and Discussion

6.3.1 Synthesis and structure descriptions of Zr4, ZrTd and Zr25

In order to isolate peroxide-containing clusters, it is important to eliminate all redox-active species in the reaction solution, otherwise side reactions complicate the peroxide-promoted cluster disassembly and reassembly processes, or the peroxide may be decomposed altogether. Therefore, we replace the commonly-used ‘ZrOCl2’ with a perchlorate salt of Zr4. This was achieved by dissolving basic zirconium carbonate in perchloric acid to prepare the reaction solution (see experimental section for details). Evaporation of this acidic zirconium perchlorate solution yields dendritic colorless crystals. Single-crystal X-ray analysis revealed the well-known Zr4 motif (Figure 6.1) in a tetragonal unit cell with completely disordered perchlorate counterions and a formulation of Zr4(OH)8(H2O)16(ClO4)8∙1.33H2O (Table D3). The 1.33 H2O molecules per {Zr4} unit

83 have been modeled with the “solvent mask” method in Olex2 software, similar to PLATON SQUEEZE.[160] More details about the solvent and counterion disorder are given in the SI.

Relevant atomic distances in Zr4 and in the ‘ZrOCl2’ commercial product (Zr4 with eight chloride counterions) are identical within ~0.04 Å, see Table D3. The exception is the O- O distances of the bridging hydroxide ligands. With the Cl- counterion, these ligands are ~0.12 Å further apart, promoted by H-bonding to the counterion (Cl---H-O). The oxygen- atoms of the perchlorate ligand do not associate with Zr4, leading to observed disorder in

the structure. Importantly, the structure of Zr4, along with prior-reported structures with

halide, asserts the ubiquity of this Zr4-polycation, regardless of the counterion, mineral acid, or electrolyte solution.

ZrTd is obtained from a 10:1 solution of peroxide:Zr and a 0.5 molar Zr solution. Crystals form after approximately 20 days and 80% evaporation of the reaction solution

(see experimental section for details). ZrTd is fully formulated Zr4(OH)4(μ-O2)2(μ4-

O)(H2O)12∙(ClO4)6∙4H2O (Figure 6.1). The μ4-oxo centered Zr-tetrahedron (Zr-Ooxo = 2.110(3)-2.138(3) Å) has four bridging hydroxyls (Zr-OH = 2.116(3)-2.146(3) Å) and two bridging peroxides (Zr-Op = 2.136(3)-2.173(3) Å) along the six edges of the tetrahedron,

and the molecule has C2 symmetry. Three terminal aqua ligands cap each of the four Zr

(Zr-Oaqua = 2.191(3)-2.265(3) Å), completing a distorted square antiprism geometry. All six perchlorate anions are located in the lattice along with four water molecules. The bridging rather than terminally-coordinated peroxide observed in both ZrTd and Zr25

(above) provides structural evidence that peroxide dissociates Zr4, suggesting a simple route to activate this ubiquitous specie and broaden the scope of Zr/Hf metal-oxo cluster chemistry.

Addition of peroxide to the zirconium carbonate-perchloric acid solution with a 1:1 peroxide:Zr ratio, 1 molar Zr solution (see experimental section) followed by evaporation

for four days (~50% loss of solvent volume) yields prism-shaped crystals of Zr25, which assembles in an orthorhombic space group Cmcm (Table D3). The cluster formulation 20+ [Zr25O10(OH)50(O2)5(H2O)40] was determined by assigning hydroxyl and oxo ligands based on bond valence sum calculations (BVS), since no H-atoms are observed in the

84 electron density map (Table D.5). The pentagonal topology of Zr25 (Figure 6.1) can be

described as an inner 5-membered ring connected to the outer ring via μ3 oxo and hydroxyl groups (Zr-O = 2.083(4)-2.119(3) Å and 2.369(3)-2.382(4) Å, respectively). The Zr atoms are linked within the ring via two bridging hydroxyl groups, similar to that observed in

Zr4. Disorder is observed for the entire inner ring (Figure D1).

The external ring consists of five tetrameric units (Figure D2) analogous to Zr4, but with slight distortions. The original {Zr4(OH)8} unit is a square (Zr-Zr distances = 3.55

Å) with Zr-OH bond distances of 2.13 Å. In Zr25, the Zr-μ2-OH distances are close to those

found in Zr4 (2.12-2.14 Å). However, there are two distinct Zr-Zr distances; 3.62 and 3.44 Å defining a rectangle. The short Zr-Zr edge has a single oxo linkage that bridges to a Zr

of the inner ring. Likewise, μ3-hydroxyl groups (bridging the inner ring) lead to the longer Zr-Zr distance on the adjacent side of the square. These five tetrameric units are linked together by disordered peroxo and a pair of hydroxyl groups with an occupancy of 0.5 for each. A short Zr-O distance (2.06-2.1 Å) is attributed to the presence of a peroxo group whereas a longer bond distance (2.16-2.21 Å) corresponds to OH groups. The distorted square antiprism geometry of the outer Zr is completed by two terminal water molecules. The Zr atoms of the inner ring exhibit the same coordination geometry, but bonded to only oxo and hydroxyl ligands. In its crystalline lattice, Zr25 clusters are oriented along the (ab) plane (Figure 6.2). PLATON indicates 48.8% of the lattice is considered to be voids. The distance between clusters in the ab plane is ~3.5 Å, measured from the terminal water ligands. The closest interlayer distance between clusters in the c-direction is ~6 Å. Between - these layers, many disordered H2O and ClO4 molecules reside without strong association to each other or the clusters. Details of modeling this disorder are summarized in Appendix D.

Figure 6.2. Representation of the Zr25 compound along the b axis, showing the layer- arrangement of the clusters. Color scheme is same as Figure 6.1. 6.3.2 Solution characterization and disassembly-reassembly processes

Crystals of Zr4, Zr25, and ZrTd perchlorate salts are highly soluble in water and in organic solvents, permitting detailed solution analyses. We analyzed 50 mM solutions of

Zr4, Zr25, and ZrTd in water, MeOH, and DMF by small angle X-ray scattering (SAXS), and diluted the aqueous and MeOH solutions for electrospray ionization mass spectral analyses (ESI-MS, see Appendix D).

The scattering curves for Zr4 in water methanol, and DMF compared to simulated scattering of Zr4 (Figure 6.3A) reveals larger species than a tetramer, evidenced by the shift of the Guinier region (0.1-0.4 Å-1) to lower-q. On the other hand, simulated scattering data from a model of three linked tetramers (Figure 6.3B) provided a good match, suggesting ladder-type polymerization in solution. In all three solutions, there is a broad ‘diffraction peak’ at q~1.8 Å-1 that matches the simulation. This is consistent with the 3.5 Å Zr-Zr distance of edge-sharing Zr-polyhedra (d=2/q). Polymerization is common for [69] Zr/Hf4 if acid is not added to prevent hydrolysis reactions, and highlights the consequence of highly acidic metal centers such as Zr/Hf on solution speciation. The pH of aqueous solutions of redissolved Zr4, (and Zr4Td and Zr25 discussed later) ranges from 1.5-3, depending on the concentration of metal centers indicating hydrolysis reactions. The 50 mmolar (Zr) solution has a pH of 1.39. We fit the experimental scattering curve in methanol with a 6.6 × 23 Å cylindrical model, consistent with the three-tetramer chain (Figure 6.3B). Finally, the coulombic peak between q=0.1-0.5 Å-1 (DMF and water)

86 indicates ordering of clusters in solution. Other than this coulombic peak, the scattering curves are similar to that of Zr4 dissolved in methanol. This is why we chose to model the methanol solution, but we can assume, based on the SAXS data, that the Zr-connectivity

and speciation is similar in all three solutions. The oligomeric forms of Zr4 observed by SAXS are clearly fragmented by ESI-MS analysis (Figures. D6, D9-34; Tables D5 & 6).

Interestingly, Zr4 dissolved in methanol converts to anionic methoxy-species, consistent + with the strong Lewis-Brønsted acid behavior of Zr4 ; monomer, dimer, trimer, and

tetramer species were observed. In water, Zr4 are predominantly 1+ tetramers. The higher

degree of fragmentation of Zr4 in methanol than water is clearly a result of the ionization

process. SAXS of Zr25 in MeOH and DMF are consistent with the scattering data simulated

from the Zr25 crystal structure, suggesting it dissolves intact (Figure 6.4A). The second plateau observed at q=0.5-1.0 Å-1 in the simulated data is commonly observed for non- spherical clusters with less electron density in its core. In the experimental data, this plateau is ‘smeared out’, perhaps indicating association of perchlorate ions in this small cavity.

ESI-MS peaks for Zr25 in methanol were again only observed in the negative ionization

mode with a 1- charge, and were assigned predominantly to Zr8 clusters (Table D8, Figures D7, D35-45), a result of ionization-promoted fragmentation. On the other hand,

the shift of the Guinier region (to higher-q) in the scattering data for Zr25 dissolved in water

indicates fragmentation (pH=2.36). Modeling the scattering data for Zr25 in water was satisfactorily fit using a two-phase spherical model (radii of 7.9Å (13%) and 3.2Å (86%)

respectively, Figure 6.4B). The ~8 Å radius fits precisely with intact Zr25 while the ~3 Å radius is consistent with any number of cluster fragments, i.e. trimers, tetramers and

pentamers. The ESI-MS of Zr25 in water is dominated by Zr trimers and pentamers, detected in the positive ionization mode (Table D9, Figure D7, D46-56).

Figure 6.3. (A) X-ray scattering of Zr4 in water and organic solvents, along with simulated Zr4 (orange) and simulated scattering data (black) from a triple tetramer (B).

SAXS analysis of the Zr25 reaction solution (1 molar Zr in the form of Zr4 perchlorate with a 1:1 Zr:peroxide ratio) by inspection is remarkably similar to Zr25 dissolved in water, consisting mostly of smaller cluster fragments (Figure 6.5A). However, a 2-phase fit of the freshly prepared solution (day 1), day 3, and day 4 (in contact with growing crystals)) while evaporating approximately half the solution volume showed a distinct trend (Table 6.1). As the Zr concentration increased with water evaporation and peroxide concentration decreased with evaporation/decomposition (Zr:peroxide=2:1), the smaller clusters increased in radius from ~1.5 to 3.5 Å, and the ratio of this population increased. The larger clusters increased in size from 4.5 to 8 Å, and the population decreased from 40 to 13%. The final sizes and distribution of sizes at day 4 (supersaturated solution) matches that of redissolved crystals of Zr25.

Pair distribution function (PDF) analysis of x-ray total scattering gave more detailed information of atom-pair correlations within the clusters in the reaction solutions (Figure 6.5B). At both day 1 and 4, the Cl-O correlation of perchlorate and the Zr-O (2.2 Å) correlation of all Zr-O bonds is observed. At day 1, the Zr-Zr correlation peak of (3.5

Å, i.e. the edge-sharing Zr in Zr4) is relatively small, suggesting the initial solution consists of smaller fragments including monomer forms. We surmise that in this high Zr (1 M) concentration solution, the polymerization observed prior in Zr4 solution (50 mmolar, no peroxide) is overcome by fragmentation via peroxide coordination and disassembly.

Figure 6.4. (A) Small angle x-ray scattering (SAXS) data for Zr25 in MeOH (red), DMF (blue), and water (green). Simulated scattering curve for Zr25 shown in black. (B) Two- phase spherical fit (red) to experimental SAXS curve of Zr25 in water (green). Population 1 radius: 7.9Å (13%). Population 2 radius: 3.1 Å (87%). Structure factor between 0.1-0.3 Å-1 was fit with a distance of 23 Å between clusters (see Table 6.1).

Table 6.1. SAXS analysis of 2-phase fit of Zr25 reaction solutions with evaporation and aging1

Age of Zr25 reaction Avg. radius of % smaller Avg. radius of % larger solution (days) smaller clusters clusters larger clusters clusters (Å) (Å)

1 1.4 (0.1) 59% 4.5 (2) 41%

3 2.7 (0.2) 63% 6.5 (2) 37%

4 3.4 (0.1) 88% 8.0 (4) 12%

Redissolved Zr25 3.1 (0.2) 87% 7.9 (2) 13% crystals

1radius uncertainties (Å) provided in parentheses

With solution aging, the Zr-Zr correlation peak at 3.5 Å becomes more prominent, indicating reassembly of the peroxide ligated cluster fragments with time, consistent with growth of small clusters (i.e. trimers to hexamers) observed by SAXS and ESI-MS.

However, we observe none of the strong higher distance peaks expected for the complete

Zr25 cluster in solution, consistent with the SAXS data. Although the studied reaction

solutions are in contact with growing crystals, Zr25 never dominates the solutions. On one

hand, this means Zr25 may assemble in the solid-state lattice via addition of small cluster fragments, rather than the expected crystal-growth mechanism of fully-formed clusters assembling with their counterions and lattice solvent. This has been observed prior for Cr- Al and Cr polycations,[31,32] where high acidity and/or excess nitrate prevents fully formed clusters from persisting in solution. Cluster assembly at the solid-liquid interface may become an emergent phenomenon as we continue to design synthetic strategies to isolate metastable aqueous-clusters, outside of standard pH control. The reason for the lack of

fully formed Zr25 in aqueous solution is not solubility (i.e. due to large size or low charge), since Zr25 can be dissolved intact in less polar organic solvents. Rather, the high acidity of

the Zr-centers likely leads to disassembly of Zr25 in water, so it can only reach the maximum observed fraction of ~13% of all dissolved species, in equilibrium with a larger population of small fragments. We are not able to accurately measure the acidity of the reaction solutions for either Zr25 or ZrTd (discussed below) because it is too low.

SAXS analysis of dissolved ZrTd perchlorate crystals (50 mmolar, pH=1.60)

shows that the species in solution, like Zr4, assemble into larger oligomers (Figure 6.6A). The scattering curve in methanol was fit with a 6.2Å × 21 Å cylindrical model (Figure D.3). The mass spectral analysis in methanol and water also supports the formation of larger species in solution. Peak envelopes were assigned to a variety of negatively (MeOH)

and positively (water) charged Zr3, Zr4, Zr5, and Zr6 species (Table D.10-11 Figure D.8, D.57-64 for methanol, D.65-73 for water). In all of the formulae determined from the ESI- MS data, the peroxide and hydroxyl (as well as oxo) ligands are still abundant. We envision the tetrahedral clusters can link via hydrolysis and condensation of Zr-centers, as is

proposed for Zr4. Again, this is due to the acid behavior of the metal centers that converts aqua ligands to reactive OH- ligands, observed by the low pH of the solution.

Figure 6.5. A: SAXS of Zr25 reaction solution aged for three days along with Zr25 redissolved in water (green) and simulated Zr25 for comparison (black). B: Experimental and simulated PDF of same reaction solution at day 1 and day 4 of aging. Crystal growth begins on day 3. Intensity of experimental data is normalized to the Zr-O pair at ~2 Å.

The ZrTd reaction solution initially contains 0.5 molar Zr as the Zr4 perchlorate salt, a 10:1 peroxide:Zr ratio (15 wt% peroxide). SAXS of this reaction solution revealed an entirely different disassembly-reassembly pathway than that of Zr25. Initially, monodisperse spherical particles form, approximately 12 Å in diameter (Figure 6.6B,

Table 6.2). This differs from the Zr4 solutions which exhibit elongated particles that are consistent with ‘ladder’ polymerization of the square clusters. The difference between the -1 Zr4 solution and ZrTd reaction solution is apparent between q=0.1-1.5 Å : the descent from the plateau region is shallower for Zr4. Additionally, the ZrTd solution scattering, particularly at day 4, exhibits a distinct oscillation, generally indicative of spheroidal species.

We simulated scattering data from numerous reported group IV, lanthanide and An(IV) clusters and cluster fragments, both ligated and unligated, since these have similar cluster and oxide forms. The best match we found is with Pu22, consisting of 22 Pu [161] polyhedra linked and terminated by oxo and chloride (Figure D4). The Pu22 cluster is structurally very similar to a fragment of cubic PuO2, isostructural with ZrO2, and distinctly contains the common molecular hexamer[70] building unit of the Zr-MOF, UiO-66.[72] [162– Moreover, the Pu22 cluster, the cubic oxide, and other related, larger nuclearity An(IV) 165] clusters contain the oxo-centered tetrahedral unit, OM4 (Figure 6.1) of ZrTd. This

91 result was initially unexpected: we anticipated the high peroxide content in this solution would immediately disassemble the tetramers, rather than promote cluster growth.

Figure 6.6. A: SAXS of ZrTd perchlorate crystals dissolved in methanol (red), DMF (blue) and water (green), and simulated from single-crystal structure (black). In these 50 mmolar solutions, there is linking between the clusters, via hydrolysis and condensation of water ligands. B: Reaction solutions of ZrTd first indicating formation of larger clusters, followed by fragmentation by coordination with peroxide (see Table 6.2 for fitting of these data).

We believe this is related to peroxide decomposition, because abundant O2 bubble formation was observed in these early stages of assembly. To determine if the rapid peroxide decomposition is promoted by acid (not expected) or Zr, we compare peroxide loss in the ZrTd. reaction solution to that of a solution containing just peroxide and perchloric acid, and no dissolved Zr. After 39 days, the ZrTd. reaction solution has 27% peroxide remaining (most of the decomposition occurs during the first few days) while the peroxide-HClO4 solution has 84% remaining (determined by permanganate assay). While redox-active metals are known to decompose peroxide, the activity of redox inert metals such as Zr(IV) is not well-understood and will be a topic of future study.

2- 2- The peroxide disproportionation reaction is decribed as: 2O2 → O2 (g) + 2O . In water, the resulting oxo (or hydroxyl) ligand will rapidly coordinate to oxophillic Zr, and hydrolysis and condensation reactions will promote assembly of polynuclear species. In the ZrTd reaction solution, the initial acid concentration is 1 molar when ‘prototype ZrO2’ is observed, and it is not possible to accurately measure the very low pH (i.e. below 0).

Additionally, peroxide decomposition does not increase the pH to a measurable value, because the resulting O/OH ligands immediately bind to Zr. Perhaps ‘prototype ZrO2’ could also be formed by adding base to an aqueous Zr4 solution, but it would likely be a polydisperse mixture that aggregates and precipitates. The role of the peroxide in addition to driving hydrolysis/condensation reactions is stabilizing the prenucleation clusters in solution by capping, as is observed for group V polyoxometalates.[166] Isolation of this initial form is a subject of future study, by replacing the capping peroxide with more stable terminal ligands.

After ten days when the solution is reduced to a volume of 10% the initial volume and the bubbling has subsided, the scattering data completely changes. There is extremely high concentration of species (5 molar based on volume reduction), indicated by the large structure factor (Table 6.2, Figure 6.6B). The scattering species are very small (~3 Å diameter), indicating the initial large cluster forms were disassembled by peroxide, but likely preserving the oxo-centered tetrahedron. We were not able to obtain sufficient quality PDF data on the early nucleation stages of the solutions due to the bubble formation from the peroxide, but PDF of the solution at day 10 helped interpret the SAXS data, and provided evidence for the OZr4 unit.

Table 6.2. SAXS analysis of ZrTd reaction solutions with evaporation and aging

1 2 Age of ZrTd Avg. radius shape Structure factor analysis reaction solution of clusters Distance Phi (days) (Å) (eta, Å)3 (dimensionless)4 1 6.6 (8) spherical 20 0.53 4 6.8 (8) spherical 25 0.71 10 1.5 (1.4) Spheroid, polydisperse5 15 2.1 Redissolved ZrTd 2.4 (2) Cylinder (length=16.3 N/A crystals (3)Å) 1radius uncertainties (Å) provided in parentheses 2parameters to fit coulombic peak caused by ordering of clusters in high concentration solutions 3distance between clusters ordered in solution 4’pack’, or term that relates to number of clusters each cluster associates with in solution at distance eta. 5shows high degree of polydispersity and large structure factor challenges accuracy of observed radius.

In addition to the Cl-O distance of perchlorate and the Zr-O bond distance at 2.2 Å, we observe the strong Zr-Zr correlation, present in ZrTd, and larger clusters containing the oxo-centered building unit (Figure D.5). All pair correlations present in the simulated

PDF for ZrTd, are qualitatively present, indicating ZrTd, is fully formed in solution.

Finally, Zr25 can be converted to ZrTd by dissolution and addition of more peroxide (10:1 peroxide:Zr), illustrating the robustness of this cluster form (isolated in 90% yield), and its reaction pathway.

6.4 Conclusion

Group IV Zr/Hf polycations are exploited in nanolithography for high resolution microcircuits, are the building block for MOFs and important catalytic metal centers. They have lacked the richness in structural diversity and nuclearity that has made its periodic table neighbors, polyoxometalates, an expansive class of molecular materials with tunable

properties. Namely, Zr/Hf square tetramers (Zr4) and cubic hexamers have dominated the solution phase chemistry, and materials formed from the solution phase. We have shown that the compositionally simple but complexly reactive peroxide ligand can disassemble

these stable building units, yielding unprecedented cluster topologies, including Zr25, the largest Zr/Hf polycation to date, yet is readily dissolved in organic solvent. Using this cluster and its smaller fragments to build cluster-based materials is one future pursuit. The reaction pathway of Zr4 to peroxide ligated oligomers (with 3 to 6 nuclearity) to Zr25 in water is partially reversible, controlled by the self-buffering pH of dissolved Zr. On the

other hand, the conversion of Zr4 to the oxo-centered tetrahedron ZrTd goes through formation of larger cluster assemblies that are yet to be isolated and identified. This study provides a new mechanism and dimension to expanding Zr/Hf cluster chemistry via peroxide-promoted disassembly-reassembly. In addition, the bridging motif of the

peroxide ligand observed in Zr25 and ZrTd provides structure and reactivity models for Zr/Hf-sulfate-peroxide lithographic processes (a topic of future study). The high solubility

of Zr25 and ZrTd, the yet-to-be isolated reaction intermediates, and analogous Hf chemistry all provide future direction and opportunity in group IV solution and solid-state chemistry.

6.5 Experimental Section 6.5.1 Synthesis

Zr4: Commercial zirconium basic carbonate (ZBC, Alfa Aesar, Lot X09A021, Hf unseparated, about 1 mol % Hf/Zr, assayed at 41.5 wt % (Zr+Hf)O2, 12.29 g (0.0377 mol

Zr+Hf), was added to a mixture of 25 mL H2O (deionized, 18 MΩ) + 12.20 g of 62.1 wt % perchloric acid, warmed to 65 °C. Dissolution was complete within 45 min. The volume of the final solution was 37 mL. Since neither the metal nor acid components are affected by subsequent evaporation steps, we regard this parent solution as being 1.02 M Zr+Hf,

and with perchlorate/Zr+Hf = 2.0. The pH of this reaction solution (and that of Zr25 and

Zr4Td, described below) is too low to measure accurately (i.e. below 0). An aliquot of this

was allowed to evaporate in air until dendritic crystals of Zr4 were formed. These were taken for x-ray crystallographic determination. The Zr and Hf composition is 99.3 and 0.7% (Table D2), a slight Hf decrease from the starting material (~2% Hf).

Zr25: To 0.50 mL of the parent solution was added 52 μL of 30 wt % H2O2, giving a starting mole ratio peroxide/Zr+Hf = 1.0. The solution was contained in a small polyethylene cup and was allowed to evaporate into the ambient atmosphere. After four days, with about 54 % loss of initial weight (water and peroxide), clear crystals in the form of prisms and lozenges, appeared, amounting to about 10% of the initial metal values. These were taken for X-ray crystallographic determination.

ZrTd: To 1.00 mL of the parent solution was added 1.00 mL of 30 wt % H2O2, giving a starting mole ratio peroxide/Zr +Hf = 9.6. This was allowed to evaporate into ambient air for 4 d after which it was placed into a dessicator over concentrated H2SO4. After 14 d and

loss of 80% of the initial mass, no solids had formed, so dessication over P2O5 was employed. This produced a mostly solid mass which re-hydrated rapidly in air with disappearance of solids. Re-placement over P2O5 resulted in a mixture of cylindrical crystals, ~0.2 mm by 1-5 mm, in contact with liquor. From this assemblage, the diffraction sample was taken. The final weight of the assemblage was 0.44 g compared to an estimated

(from eventual crystallographic formula weight) weight of 0.35 g of ZrTd solids plus 0.05 g “excess” perchloric acid = 0.40 g. In a parallel experiment, the evaporating mixture was

95 analyzed for peroxide and found to have peroxide/Zr = 0.52 (0.50 found by x-ray in final solid), indicating that the large initial excess of peroxide had evaporated or decomposed, leaving the expected final stoichiometry.

We note that despite the use of the same starting materials, the synthesis of ZrTd (with a much larger excess of peroxide) was prolonged and high-yielding, with dry-out-re- hydrate steps; in contrast to Zr25, which is comparatively rapid and low-yield. The crystal morphologies of the two are noticeably different: rhombohedra for Zr25, and elongated cylinders for the former. Attempts to increase the yield of Zr25 by forcing conditions of strong dessication or even mild warming (45°C), resulted in disruption of the crystals. This causes us to posit that the former phase is the more-stable one.

6.5.2 Characterization Techniques

Single crystal X-ray diffraction: Data of Zr4 and ZrTd were collected on a Bruker DUO- APEX II CCD area detector at 171 K using a Cu radiation (λ = 1.54178 Å). Data reduction was accomplished using SAINT V8.34a.[167] The substantial redundancy in data allowed a semiempirical absorption correction (SADABS V2.10)[168] to be applied, based on multiple

measurements of equivalent reflections. In the case of Zr25, data were collected at 100.0(1) K on a Rigaku Oxford Diffraction Synergy-S equipped with a PhotonJet-S Mo source (λ = 0.71073 Å) and HyPix-6000HE photon counting detector. All the images were collected and processed using CrysAlisPro Version 1.171.40.20a (Rigaku Oxford Diffraction, 2018). After integration, both numerical (Gaussian) absorption and empirical absorption (spherical harmonic, image scaling and detector scaling) corrections were applied.

All the structures were solved by the intrinsic phasing method from the SHELXT program, developed by successive difference Fourier syntheses, and refined by full-matrix least squares on all F2 data using SHELX[169] via OLEX2[170] interface. The crystal data of the

three compounds are given in Table D3. ICSD file numbers: 1891906 (Zr4) -1891907

(Zr25) -1898479 (ZrTd).

Small Angle X-ray Scattering (SAXS): X-ray scattering data were collected on an Anton Paar SAXSess instrument using Cu-Kα radiation (1.54 Å) and line collimation. The

96 instrument was equipped with a 2-D image plate for data collection in the q = 0.018-2.5 Å- 1 range. The lower q resolution is limited by the beam attenuator. Approximately 50 mmolar (Zr) solutions were measured in 1.5 mm glass capillaries (Hampton Research). Scattering data of neat solvent was collected for background subtraction. Scattering was measured for 30 min for every experiment. We used SAXSQUANT software for data collection and treatment (normalization, primary beam removal, background subtraction, desmearing, and smoothing to remove the extra noise created by the desmearing routine). All analyses and curve-fitting to determine Rg, size, shape and size distribution were carried out utilizing IRENA macros with IgorPro 6.3 (Wavemetrics) software.[108] To simulate scattering data from the crystal structure, we used SolX software.[109]

Powder X-ray diffraction: Powder XRD of compounds ZrTd and Zr25 were collected to ensure the single crystal is representative of the bulk. The data were collected on a Miniflex 600 in the range from 3 to 30° (2θ) with a scan speed of 5°/min and step size of 0.02°.

Zr4Td was mixed with paratone oil to limit the deliquescent behavior (Figure D74). For

similar reason, crystals of Zr25 were not ground, leading to a preferential orientation (Figure D75).

Electrospray Ionization Mass Spectrometry (ESI-MS): ESI-MS was carried out using an Agilent 6230 ESI-MS system comprised of a Time-of-Flight (TOF) mass spectrometer coupled to an electrospray ionizer. Samples of Zr25, Zr4Td and Zr4 ([Zr]=1.0 mM) were dissolved in neat water and neat methanol and infused into the ESI-MS system at a flow rate of 0.4 mL min-1 using a syringe pump. The solutions were nebulized with the aid of

heated N2 (325 °C) flowing at 8 L min-1 and a pressure of 35 psig (241 kPa). The voltages of the capillary, skimmer, and RT octopole were set at 3500, 65, and 750 V respectively, while the voltage of the fragmenter was set at 100V. The data were collected in both the positive and negative ionization modes.

Pair Distribution Function (PDF) analysis of X-ray total scattering: Solution measurements were performed with Rigaku Smartlab X-ray diffractometer with Mo-Kα irradiation (λ=0.71073 Å) in the 2θ range of 3.0-118.5° with transmission geometry. Solution samples were loaded in Kapton capillary with 1.46 mm inner diameter for

-1 irradiation. The theoretical q-range is up to a qmax of 15.5 Å . A 0.2 degree/minute data collection time and D/teX Ultra-HE high-speed one-dimensional X-ray detector were utilized to ensure high quality scattering data. PDFgetX3[171] was to process scattering data into the corresponding PDF. The corresponding simulations were created in PDFgui.[172]

6.6 Acknowledgments

The synthesis and solution characterization was supported by the National Science Foundation, Center for Chemical Innovation, grant number CHE-1606982. Structural characterization, PDF and manuscript drafting was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Material Sciences and Engineering, under award DE SC0010802. We acknowledge the Murdock Charitable Trust (grant # SR- 2017297) for acquisition of the single-crystal X-ray diffractometer.

7 Conclusions

The work presented here has focused on the synthesis and solution characterization of several new metal-oxo clusters which are being targeted as potential EUV lithography photoresist materials. We have investigated two different metal-oxo systems with radiation-sensitive ligands: the butyltin Keggin system and the zirconium peroxide system.

Four new butyltin Keggin structures have been reported (β-NaSn12, γ-NaSn12, γ-NaSn13, and β-CaSn12), highlighting the unique nature of the butyltin system in preferring the rarer

β and γ isomeric forms. Extensive solution characterization of the β,γ-NaSn12 mixture by electrospray-ionization mass spectrometry and small angle x-ray scattering showed that the clusters remain intact and are free from impurities. The presence of broad, overlapping peaks in the 1H NMR and additional signals in the 119Sn NMR point to interconversion between the β and γ isomers as well as the formation of new isomers upon dissolution. Aging studies of these Na-centered clusters by 1H NMR and FT-IR showed the hydrolysis

of methoxy ligands over time. Finally, mild heating of β,γ-NaSn12 in solution, monitored in situ by variable temperature 1H, 23Na, and 119Sn NMR, promoted the formation of multiple new Keggin isomers. All five of the Keggin rotational isomers were observed simultaneously in a heated solution by 23Na NMR.

Similar solution characterization studies were performed for β-CaSn12. In this case, however, the Ca2+ central cation stabilizes the β isomer and inhibits isomerization at room temperature. Aging studies also showed hydrolysis of methoxy ligands over time and isomerization to the γ isomer was promoted by heating in solution and observed in situ by variable temperature 119Sn NMR. While Na and Ca are not ideal for photoresist materials, we have shown that substitution of the central metal of the Keggin ion is possible. Future studies may focus on substituting the central cation with a transition metal or other main group metal.

Two new zirconium peroxide clusters (ZrTd and Zr25) were also reported and extensively characterized in solution. The isolation of these new structures helped to answer questions about the location of peroxide in the previously investigated HafSOx and ZircSOx photoresist deposition solutions. Monitoring the reaction pathways for both

99 clusters by small angle x-ray scattering revealed an interesting trend based on the ratio of peroxide to metal in the reaction solution. When the solution contains a 1:1 peroxide/Zr ratio, as in the synthesis of Zr25, small pentamer and trimer fragments form in solution and the intact cluster is never observed in aqueous solution. Therefore, we believe that these

prenucleation clusters assemble into Zr25 in the solid state, at the interface between solution and crystal. Conversely, with an excess of peroxide (10:1 peroxide/Zr), as in the synthesis

of ZrTd, a large spherical cluster first forms in solution before breaking down into the small tetrahedron just before crystallization.

As a final statement, not only have we added to the structural diversity of metal- oxo clusters with radiation-sensitive ligands, these well-characterized cluster solutions provide an ideal model system for the study of photoresist materials throughout the lithography process. Furthermore, solution characterization of these clusters and their reaction solutions has provided important fundamental insight into the factors influencing the structural formation and behavior of metal-oxo clusters.

100

References

[1] M. Nyman, Coord. Chem. Rev. 2017, 352, 461–472. [2] J. J. Berzelius, Ann. Phys. 1826, 82, 369–392. [3] J. F. Keggin, Proc R Soc Lond A 1934, 144, 75–100. [4] L. C. W. Baker, J. S. Figgis, J. Am. Chem. Soc. 1970, 92, 3794–3797. [5] X. López, J. J. Carbó, C. Bo, J. M. Poblet, Chem. Soc. Rev. 2012, 41, 7537–7571. [6] A. R. Kampf, J. M. Hughes, B. P. Nash, J. Marty, Am. Mineral. 2017, 102, 461–465. [7] L. Pauling, Z. Für Krist. - Cryst. Mater. 2015, 84, 442–452. [8] F. M. Michel, L. Ehm, S. M. Antao, P. L. Lee, P. J. Chupas, G. Liu, D. R. Strongin, M. A. A. Schoonen, B. L. Phillips, J. B. Parise, Science 2007, 316, 1726–1729. [9] M. E. Fleet, Acta Crystallogr. B 1981, 37, 917–920. [10] X. López, J. M. Maestre, C. Bo, J.-M. Poblet, J. Am. Chem. Soc. 2001, 123, 9571– 9576. [11] K. Yamamura, Y. Sasaki, J. Chem. Soc. Chem. Commun. 1973, 648–649. [12] J. J. Cowan, A. J. Bailey, R. A. Heintz, B. T. Do, K. I. Hardcastle, C. L. Hill, I. A. Weinstock, Inorg. Chem. 2001, 40, 6666–6675. [13] I. A. Weinstock, J. J. Cowan, E. M. G. Barbuzzi, H. Zeng, C. L. Hill, J. Am. Chem. Soc. 1999, 121, 4608–4617. [14] T. Ueda, K. Isai, Anal. Sci. 2013, 29, 447–453. [15] K. M. Sundaram, W. A. Neiwert, C. L. Hill, I. A. Weinstock, Inorg. Chem. 2006, 45, 958–960. [16] W. A. Neiwert, J. J. Cowan, K. I. Hardcastle, C. L. Hill, I. A. Weinstock, Inorg. Chem. 2002, 41, 6950–6952. [17] S.-H. Wang, S. A. Jansen, Chem. Mater. 1994, 6, 2130–2137. [18] M. Nyman, F. Bonhomme, T. M. Alam, M. A. Rodriguez, B. R. Cherry, J. L. Krumhansl, T. M. Nenoff, A. M. Sattler, Science 2002, 297, 996–998. [19] M. Nyman, Dalton Trans. 2011, 40, 8049–8058. [20] G. Johansson, L.-O. Gullman, A. Kjekshus, R. Söderquist, Acta Chem. Scand. 1960, 14, 771–773. [21] A. F. Oliveri, C. A. Colla, C. K. Perkins, N. Akhavantabib, J. R. Callahan, C. D. Pilgrim, S. E. Smart, P. H.-Y. Cheong, L. Pan, W. H. Casey, Chem. – Eur. J. 2016, 22, 18682–18685.

101

[22] O. Deschaume, E. Breynaert, S. Radhakrishnan, S. Kerkhofs, M. Haouas, S. Adam de Beaumais, V. Manzin, J.-B. Galey, L. Ramos-Stanbury, F. Taulelle, et al., Inorg. Chem. 2017, 56, 12401–12409. [23] S. E. Smart, J. Vaughn, I. Pappas, L. Pan, Chem. Commun. 2013, 49, 11352–11354. [24] S. Abeysinghe, D. K. Unruh, T. Z. Forbes, Inorg. Chem. 2013, 52, 5991–5999. [25] J. Rowsell, L. F. Nazar, J. Am. Chem. Soc. 2000, 122, 3777–3778. [26] L. Allouche, C. Gérardin, T. Loiseau, G. Férey, F. Taulelle, Angew. Chem. Int. Ed. 2000, 39, 511–514. [27] W. H. Casey, Chem. Rev. 2006, 106, 1–16. [28] W. H. Casey, M. M. Olmstead, C. R. Hazlett, C. Lamar, T. Z. Forbes, Inorganics 2015, 3, 21–26. [29] O. A. Gerasko, E. A. Mainicheva, D. Yu. Naumov, N. V. Kuratieva, M. N. Sokolov, V. P. Fedin, Inorg. Chem. 2005, 44, 4133–4135. [30] M. Fairley, K. W. Corum, A. Johns, D. K. Unruh, M. Basile, J. de Groot, S. E. Mason, T. Z. Forbes, Chem. Commun. 2015, 51, 12467–12469. [31] W. Wang, L. B. Fullmer, N. A. G. Bandeira, S. Goberna-Ferrón, L. N. Zakharov, C. Bo, D. A. Keszler, M. Nyman, Chem 2016, 1, 887–901. [32] W. Wang, M. Amiri, K. Kozma, J. Lu, L. N. Zakharov, M. Nyman, Eur. J. Inorg. Chem. 2018, 2018, 4638–4642. [33] O. Sadeghi, L. N. Zakharov, M. Nyman, Science 2015, 347, 1359–1362. [34] O. Sadeghi, C. Falaise, P. I. Molina, R. Hufschmid, C. F. Campana, B. C. Noll, N. D. Browning, M. Nyman, Inorg. Chem. 2016, 55, 11078–11088. [35] V. Baskar, M. Shanmugam, M. Helliwell, S. J. Teat, R. E. P. Winpenny, J. Am. Chem. Soc. 2007, 129, 3042–3043. [36] H. Reuter, Angew. Chem. Int. Ed. 1991, 30, 1482–1484. [37] S. Saha, D.-H. Park, D. C. Hutchison, M. R. Olsen, L. N. Zakharov, D. Marsh, S. Goberna-Ferrón, R. T. Frederick, J. T. Diulus, N. Kenane, et al., Angew. Chem. Int. Ed. 2017, 56, 10140–10144. [38] Y. Zhu, M. R. Olsen, M. Nyman, L. Zhang, J. Zhang, Inorg. Chem. 2019, 58, 4534– 4539. [39] K. Sugahara, S. Kuzuya, T. Hirano, K. Kamata, N. Mizuno, Inorg. Chem. 2012, 51, 7932–7939. [40] S. E. Johnson, C. B. Knobler, Organometallics 1994, 13, 4928–4938. [41] C. Lecomte, J. Protas, M. Devaud, Acta Crystallogr. B 1976, 32, 923–924. [42] H. Puff, H. Reuter, J. Organomet. Chem. 1989, 364, 57–65.

102

[43] R. R. Holmes, Soheila. Shafieezad, V. Chandrasekhar, J. M. Holmes, R. O. Day, J. Am. Chem. Soc. 1988, 110, 1174–1180. [44] L. Plasseraud, H. Cattey, P. Richard, Z. Für Naturforschung B 2010, 65, 1293– s1308. [45] H. Puff, H. Reuter, J. Organomet. Chem. 1989, 368, 173–183. [46] G. Kastner, H. Reuter, Main Group Met. Chem. 2011, 23, 331–336. [47] C. Eychenne-Baron, F. Ribot, C. Sanchez, J. Organomet. Chem. 1998, 567, 137– 142. [48] F. Ribot, D. Veautier, S. J. Guillaudeu, T. Lalot, J. Mater. Chem. 2005, 15, 3973– 3978. [49] H. Puff, H. Reuter, J. Organomet. Chem. 1989, 373, 173–184. [50] C. Eychenne-Baron, F. Ribot, N. Steunou, C. Sanchez, F. Fayon, M. Biesemans, J. C. Martins, R. Willem, Organometallics 2000, 19, 1940–1949. [51] D. Dakternieks, H. Zhu, E. R. T. Tiekink, R. Colton, J. Organomet. Chem. 1994, 476, 33–40. [52] F. Banse, F. Ribot, P. Toledano, J. Maquet, C. Sanchez, Inorg. Chem. 1995, 34, 6371–6379. [53] B. Cardineau, R. Del Re, M. Marnell, H. Al-Mashat, M. Vockenhuber, Y. Ekinci, C. Sarma, D. A. Freedman, R. L. Brainard, Microelectron. Eng. 2014, 127, 44–50. [54] L. Plasseraud, H. Cattey, P. Richard, Z. Für Naturforschung B 2014, 66, 262–268. [55] G. Kastner, H. Reuter, J. Organomet. Chem. 2000, 598, 381–386. [56] M. Izaaryene, G. Kastner, H. Reuter, Z. Für Krist. - Cryst. Mater. 2005, 220, 622– 625. [57] F. Ribot, C. Eychenne-Baron, F. Fayon, D. Massiot, B. Bresson, Main Group Met. Chem. 2002, 25, 115–119. [58] R. D. Stern, D. C. Hutchison, M. R. Olsen, L. N. Zakharov, M. Nyman, K. A. Persson, in Int. Conf. Extreme Ultrav. Lithogr. 2019, International Society For Optics And Photonics, 2019, p. 111471L. [59] R. R. Holmes, C. G. Schmid, V. Chandrasekhar, R. O. Day, J. M. Holmes, J. Am. Chem. Soc. 1987, 109, 1408–1414. [60] V. Chandrasekhar, R. O. Day, R. R. Holmes, Inorg. Chem. 1985, 24, 1970–1971. [61] V. Chandrasekhar, R. O. Day, R. R. Holmes, Phosphorus Sulfur Silicon Relat. Elem. 1995, 98, 417–421. [62] V. Chandrasekhar, S. Nagendran, V. Baskar, Coord. Chem. Rev. 2002, 235, 1–52.

103

[63] M. C. Sharps, D. A. Marsh, L. N. Zakharov, J. E. Hutchison, D. W. Johnson, Cryst. Res. Technol. 2017, 52, 1700081. [64] M. C. Sharps, R. T. Frederick, M. L. Javitz, G. S. Herman, D. W. Johnson, J. E. Hutchison, Chem. Mater. 2019, 31, 4840–4850. [65] A. Clearfield, P. A. Vaughan, Acta Crystallogr. 1956, 9, 555–558. [66] T. C. W. Mak, Can. J. Chem. 1968, 46, 3491–3497. [67] D. Sinram, B. Krebs, Z. Für Krist. - Cryst. Mater. 2009, 220, 166–168. [68] C. Hagfeldt, V. Kessler, I. Persson, Dalton Trans. 2004, 0, 2142–2151. [69] R. E. Ruther, B. M. Baker, J.-H. Son, W. H. Casey, M. Nyman, Inorg. Chem. 2014, 53, 4234–4242. [70] C. Hennig, S. Weiss, W. Kraus, J. Kretzschmar, A. C. Scheinost, Inorg. Chem. 2017, 56, 2473–2480. [71] L. Pan, R. Heddy, J. Li, C. Zheng, X.-Y. Huang, X. Tang, L. Kilpatrick, Inorg. Chem. 2008, 47, 5537–5539. [72] J. H. Cavka, S. Jakobsen, U. Olsbye, N. Guillou, C. Lamberti, S. Bordiga, K. P. Lillerud, J. Am. Chem. Soc. 2008, 130, 13850–13851. [73] S. Goberna-Ferrón, D.-H. Park, J. M. Amador, D. A. Keszler, M. Nyman, Angew. Chem. 2016, 128, 6329–6332. [74] D. B. McWhan, G. Lundgren, Inorg. Chem. 1966, 5, 284–289. [75] Q. Sun, C. Liu, G. Zhang, J. Zhang, C.-H. Tung, Y. Wang, Chem. – Eur. J. 2018, 24, 14701–14706. [76] A. Kalaji, L. Soderholm, Inorg. Chem. 2014, 53, 11252–11260. [77] A. Kalaji, L. Soderholm, Chem. Commun. 2013, 50, 997–999. [78] G. E. Moore, Proc. IEEE 1998, 86, 82–85. [79] L. Li, X. Liu, S. Pal, S. Wang, C. K. Ober, E. P. Giannelis, Chem. Soc. Rev. 2017, 46, 4855–4866. [80] C. Mack, in Fundam. Princ. Opt. Lithogr., John Wiley & Sons, Ltd, 2007, pp. 1–28. [81] S. Ghosh, C. P. Pradeep, S. K. Sharma, P. G. Reddy, S. P. Pal, K. E. Gonsalves, RSC Adv. 2016, 6, 74462–74481. [82] M. Brihoum, R. Ramos, K. Menguelti, G. Cunge, E. Pargon, O. Joubert, J. Appl. Phys. 2013, 113, 013302. [83] C. M. Garner, Philos. Trans. R. Soc. Math. Phys. Eng. Sci. 2012, 370, 4015–4041. [84] A. A. Lamola, C. R. Szmanda, J. W. Thackeray, Solid State Technol. 1991, 34, 53. [85] P. Bandelier, A.-L. Charley, A. Lagrange, in Lithography, John Wiley & Sons, Ltd, 2013, pp. 1–40.

104

[86] A. Jouve, M. May, I. Servin, J. Simon, in Lithography, John Wiley & Sons, Ltd, 2013, pp. 275–368. [87] S. A. MacDonald, C. G. Willson, J. M. J. Frechet, Acc. Chem. Res. 1994, 27, 151– 158. [88] D. P. Sanders, Chem. Rev. 2010, 110, 321–360. [89] J. T. Anderson, C. L. Munsee, C. M. Hung, T. M. Phung, G. S. Herman, D. C. Johnson, J. F. Wager, D. A. Keszler, Adv. Funct. Mater. 2007, 17, 2117–2124. [90] J. Stowers, D. A. Keszler, Microelectron. Eng. 2009, 86, 730–733. [91] R. T. Frederick, J. M. Amador, S. Goberna-Ferrón, M. Nyman, D. A. Keszler, G. S. Herman, J. Phys. Chem. C 2018, 122, 16100–16112. [92] R. P. Oleksak, R. E. Ruther, F. Luo, K. C. Fairley, S. R. Decker, W. F. Stickle, D. W. Johnson, E. L. Garfunkel, G. S. Herman, D. A. Keszler, ACS Appl. Mater. Interfaces 2014, 6, 2917–2921. [93] Y. Ekinci, M. Vockenhuber, M. Hojeij, L. Wang, N. M. Mojarad, 2013, pp. 867910- 867910–11. [94] R. P. Oleksak, R. E. Ruther, F. Luo, J. M. Amador, S. R. Decker, M. N. Jackson, J. R. Motley, J. K. Stowers, D. W. Johnson, E. L. Garfunkel, et al., ACS Appl. Nano Mater. 2018, 1, 4548–4556. [95] Y. Zhang, J. Haitjema, X. Liu, F. Johansson, A. Lindblad, S. Castellanos, N. Ottosson, A. M. Brouwer, J. MicroNanolithography MEMS MOEMS 2017, 16, 023510. [96] R. T. Frederick, S. Saha, J. Trey Diulus, F. Luo, J. M. Amador, M. Li, D.-H. Park, E. L. Garfunkel, D. A. Keszler, G. S. Herman, Microelectron. Eng. 2019, 205, 26– 31. [97] R. T. Frederick, J. T. Diulus, D. C. Hutchison, M. Nyman, G. S. Herman, ACS Appl. Mater. Interfaces 2019, 11, 4514–4522. [98] J. T. Diulus, R. T. Frederick, M. Li, D. C. Hutchison, M. R. Olsen, I. Lyubinetsky, L. Árnadóttir, E. L. Garfunkel, M. Nyman, H. Ogasawara, et al., ACS Appl. Mater. Interfaces 2019, 11, 2526–2534. [99] L. C. W. Baker, J. S. Figgis, J. Am. Chem. Soc. 1970, 92, 3794–3797. [100] A. Márquez, J. Anguiano, G. González, J. Fernández Sanz, J. Organomet. Chem. 1995, 486, 45–50. [101] M. R. Stem, M. L. Ellzey, Struct. Chem. 2010, 21, 43–58. [102] P. Matczak, Comput. Theor. Chem. 2013, 1013, 7–14. [103] M. R. Adams, E. A. C. Bushnell, T. Bruce Grindley, R. J. Boyd, Comput. Theor. Chem. 2014, 1050, 7–14.

105

[104] A. Sartorel, M. Carraro, A. Bagno, G. Scorrano, M. Bonchio, Angew. Chem. Int. Ed. 2007, 46, 3255–3258. [105] C. R. Armstrong, W. H. Casey, A. Navrotsky, Proc. Natl. Acad. Sci. 2011, 108, 14775–14779. [106] C. A. Ohlin, J. R. Rustad, W. H. Casey, Dalton Trans. 2014, 43, 14533–14536. [107] F.-Q. Zhang, W. Guan, Y.-T. Zhang, M.-T. Xu, J. Li, Z.-M. Su, Inorg. Chem. 2010, 49, 5472–5481. [108] J. Ilavsky, P. R. Jemian, J. Appl. Crystallogr. 2009, 42, 347–353. [109] X. Zuo, G. Cui, K. M. Merz, L. Zhang, F. D. Lewis, D. M. Tiede, Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 3534—3539. [110] H. E. Gottlieb, V. Kotlyar, A. Nudelman, J. Org. Chem. 1997, 62, 7512–7515. [111] G. M. Sheldrick, Siemens Area Detect. Absorpt. Correct. Program 1998. [112] G. M. Sheldrick, Acta Crystallogr. Sect. C 2015, 71, 3–8. [113] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson, H. Nakatsuji, et al., Gaussian 09, Gaussian, Inc., Wallingford, CT, 2013. [114] A. D. Becke, J. Chem. Phys. 1993, 98, 5648–5652. [115] P. C. Hariharan, J. A. Pople, Theor. Chim. Acta 1973, 28, 213–222. [116] W. R. Wadt, P. J. Hay, J. Chem. Phys. 1985, 82, 284–298. [117] A. V. Marenich, C. J. Cramer, D. G. Truhlar, J. Phys. Chem. B 2009, 113, 6378– 6396. [118] R. Krishnan, J. S. Binkley, R. Seeger, J. A. Pople, J. Chem. Phys. 1980, 72, 650– 654. [119] J. Ho, M. L. Coote, Theor. Chem. Acc. 2010, 125, 3. [120] J. A. Fernández, X. López, J. M. Poblet, J. Mol. Catal. Chem. 2007, 262, 236–242. [121] A. Kremleva, P. A. Aparicio, A. Genest, N. Rösch, Electrochimica Acta 2017, 231, 659–669. [122] X. López, J. M. Poblet, Inorg. Chem. 2004, 43, 6863–6865. [123] J.-H. Son, W. H. Casey, Chem. Commun. 2015, 51, 1436–1438. [124] E. Montargès, L. J. Michot, P. Ildefonse, Microporous Mesoporous Mater. 1999, 28, 83–101. [125] D. C. Hutchison, R. D. Stern, M. R. Olsen, L. N. Zakharov, K. A. Persson, M. Nyman, Dalton Trans. 2018, 47, 9804–9813. [126] F.-Q. Zhang, X.-M. Zhang, H.-S. Wu, H. Jiao, J. Phys. Chem. A 2007, 111, 159– 166.

106

[127] R. D. Shannon, Acta Crystallogr. A 1976, 32, 751–767. [128] A. K. Katz, J. P. Glusker, S. A. Beebe, C. W. Bock, J. Am. Chem. Soc. 1996, 118, 5752–5763. [129] K. F. Tesh, T. P. Hanusa, J. C. Huffman, C. J. Huffman, Inorg. Chem. 1992, 31, 5572–5579. [130] P. Van der Sluis, A. L. Spek, Acta Crystallogr. A 1990, 46, 194–201. [131] I. D. Brown, D. Altermatt, Acta Crystallogr. Sect. B 1985, 41, 244–247. [132] M.-C. Brochier Salon, P.-A. Bayle, M. Abdelmouleh, S. Boufi, M. N. Belgacem, Colloids Surf. Physicochem. Eng. Asp. 2008, 312, 83–91. [133] C. W. Clifford, J. Ind. Eng. Chem. 1921, 13, 631–632. [134] F. Clavería-Cádiz, R. Arratia-Pérez, D. MacLeod-Carey, Polyhedron 2016, 117, 478–486. [135] D. C. Hutchison, R. D. Stern, L. N. Zakharov, K. A. Persson, M. Nyman, Inorg. Chem. 2019, Submitted. [136] A. Sartorel, M. Carraro, F. M. Toma, M. Prato, M. Bonchio, Energy Environ. Sci. 2012, 5, 5592–5603. [137] S. J. Folkman, J. Soriano-Lopez, J. R. Galán-Mascarós, R. G. Finke, J. Am. Chem. Soc. 2018, 140, 12040–12055. [138] M. Blasco-Ahicart, J. Soriano-López, J. J. Carbó, J. M. Poblet, J. R. Galan- Mascaros, Nat. Chem. 2018, 10, 24–30. [139] H. Wang, S. Hamanaka, Y. Nishimoto, S. Irle, T. Yokoyama, H. Yoshikawa, K. Awaga, J. Am. Chem. Soc. 2012, 134, 4918–4924. [140] T. A. Stewart, D. E. Trudell, T. M. Alam, C. A. Ohlin, C. Lawler, W. H. Casey, S. Jett, M. Nyman, Environ. Sci. Technol. 2009, 43, 5416–5422. [141] S. Seino, R. Kawahara, Y. Ogasawara, N. Mizuno, S. Uchida, Angew. Chem. Int. Ed. 2016, 55, 3987–3991. [142] Y. Ji, J. Hu, L. Huang, W. Chen, C. Streb, Y.-F. Song, Chem. – Eur. J. 2015, 21, 6469–6474. [143] J. S. Weatherill, K. Morris, P. Bots, T. M. Stawski, A. Janssen, L. Abrahamsen, R. Blackham, S. Shaw, Environ. Sci. Technol. 2016, 50, 9333–9342. [144] X. Fan, J. Wang, K. Wu, L. Zhang, J. Zhang, Angew. Chem. Int. Ed. 2019, 58, 1320– 1323. [145] R. Chaves Lima, L. Bieseki, P. Vinaches Melguizo, S. B. Castellã Pergher, in Environ. Friendly Zeolites Synth. Source Mater. (Eds.: R. Chaves Lima, L. Bieseki, P. Vinaches Melguizo, S.B. Castellã Pergher), Springer International Publishing, Cham, 2019, pp. 21–63.

107

[146] J. D. Rimer, D. D. Roth, D. G. Vlachos, R. F. Lobo, Langmuir 2007, 23, 2784–2791. [147] D. A. Marsh, S. Goberna-Ferrón, M. K. Baumeister, L. N. Zakharov, M. Nyman, D. W. Johnson, Dalton Trans. 2017, 46, 947–955. [148] C. Falaise, K. Kozma, M. Nyman, Chem. – Eur. J. 2018, 24, 14226–14232. [149] D. Yang, S. O. Odoh, J. Borycz, T. C. Wang, O. K. Farha, J. T. Hupp, C. J. Cramer, L. Gagliardi, B. C. Gates, ACS Catal. 2016, 6, 235–247. [150] I. S. Kim, J. Borycz, A. E. Platero-Prats, S. Tussupbayev, T. C. Wang, O. K. Farha, J. T. Hupp, L. Gagliardi, K. W. Chapman, C. J. Cramer, et al., Chem. Mater. 2015, 27, 4772–4778. [151] M. T. Friend, N. A. Wall, Inorganica Chim. Acta 2019, 484, 357–367. [152] N. Wall, K. Nash, L. Martin, Managing Zirconium Chemistry and Phase Compatibility in Combined Process Separations for Minor Actinide Partitioning, Washington State Univ., Pullman, WA (United States), 2017. [153] P. García-García, M. Müller, A. Corma, Chem. Sci. 2014, 5, 2979–3007. [154] M. Rimoldi, A. J. Howarth, M. R. DeStefano, L. Lin, S. Goswami, P. Li, J. T. Hupp, O. K. Farha, ACS Catal. 2017, 7, 997–1014. [155] L.-P. Mo, Z.-H. Zhang, Curr. Org. Chem. 2011, 15, 3800–3823. [156] L. Y. Wang, M. S. Lee, J. Ind. Eng. Chem. 2016, 39, 1–9. [157] L. Xu, Y. Xiao, A. van Sandwijk, Q. Xu, Y. Yang, in Energy Mater. 2014, Springer International Publishing, 2016, pp. 451–457. [158] A. Kalaji, S. Skanthakumar, M. G. Kanatzidis, J. F. Mitchell, L. Soderholm, Inorg. Chem. 2014, 53, 6321–6328. [159] N. E. Travia, B. L. Scott, J. L. Kiplinger, Chem. – Eur. J. 2014, 20, 16846–16852. [160] A. L. Spek, Acta Crystallogr. Sect. C Struct. Chem. 2015, 71, 9–18. [161] G. E. Sigmon, A. E. Hixon, Chem. – Eur. J. 2019, 25, 2463–2466. [162] L. Soderholm, P. M. Almond, S. Skanthakumar, R. E. Wilson, P. C. Burns, Angew. Chem. Int. Ed. 2008, 47, 298–302. [163] C. Falaise, C. Volkringer, J.-F. Vigier, A. Beaurain, P. Roussel, P. Rabu, T. Loiseau, J. Am. Chem. Soc. 2013, 135, 15678–15681. [164] N. P. Martin, C. Volkringer, P. Roussel, J. März, C. Hennig, T. Loiseau, A. Ikeda- Ohno, Chem. Commun. 2018, 54, 10060–10063. [165] C. Falaise, C. Volkringer, C. Hennig, T. Loiseau, Chem. – Eur. J. 2015, 21, 16654– 16664. [166] L. B. Fullmer, C. E. Malmberg, D. B. Fast, L. A. Wills, P. H.-Y. Cheong, M. R. Dolgos, M. Nyman, Dalton Trans. 2017, 46, 8486–8493.

108

[167] SAINT Plus Version 8.34a, Bruker Analytical X-Ray Systems, Madison, WI, 2013. [168] SADABS, Bruker-Siemens Area Detector Absorption And Other Correction, 2008. [169] G. M. Sheldrick, Acta Crystallogr. A 2008, 64, 112–122. [170] O. V. Dolomanov, L. J. Bourhis, R. J. Gildea, J. a. K. Howard, H. Puschmann, J. Appl. Crystallogr. 2009, 42, 339–341. [171] P. Juhas, T. Davis, C. L. Farrow, S. J. L. Billinge, J. Appl. Crystallogr. 2013, 46, 560–566. [172] C. L. Farrow, P. Juhas, J. W. Liu, D. Bryndin, E. S. Božin, J. Bloch, T. Proffen, S. J. L. Billinge, J. Phys. Condens. Matter 2007, 19, 335219. [173] I. D. Brown, D. Altermatt, Acta Crystallogr. Sect. B 1985, 41, 244–247.

109

APPENDICES

110

APPENDIX A Supporting Information for Chapter 3

Table A1: BVS for Oxo Ligands of β-NaSn12.

Assignment Atom 1 Atom 2 d (A) BV BVS O2- O1 Sn1 2.090(16) 0.6 2.09 Central oxo O1 Sn2 2.085(10) 0.6 O1 Sn2 2.085(10) 0.6 O1 Na1 2.314(19) 0.3 O2- O2 Sn3 2.075(15) 0.6 2.16 Central oxo O2 Sn4 2.067(9) 0.6 O2 Sn4 2.067(9) 0.6 O2 Na1 2.330(16) 0.2 O2- O3 Sn5 2.094(12) 0.6 2.09 Central oxo O3 Sn6 2.075(12) 0.6 O3 Sn7 2.082(13) 0.6 O3 Na1 2.326(12) 0.2 OH- O5 Sn1 2.033(14) 0.7 1.34 O5 Sn5 2.074(15) 0.6 O2- O7 Sn2 2.031(12) 0.7 1.39 O7 Sn3 2.051(12) 0.7 OH- O8 Sn2 2.084(14) 0.6 1.24 O8 Sn6 2.079(14) 0.6 O2- O11 Sn4 2.056(13) 0.7 1.38 O11 Sn6 2.030(12) 0.7 OH-/O2- O12 Sn4 2.067(13) 0.6 1.35 O12 Sn7 2.033(13) 0.7 OH- O16 Sn5 2.065(7) 0.6 1.30 O16 Sn5 2.065(7) 0.6 OH- O17 Sn7 2.116(7) 0.6 1.13 O17 Sn7 2.116(7) 0.6 - OCH3 O4 Sn1 2.145(15) 0.5 1.87 O4 Sn2 2.161(14) 0.5 O4 C29 1.45(3) 0.8 - OCH3 O6 Sn2 2.133(13) 0.5 1.86 O6 Sn2 2.133(13) 0.5 O6 C30 1.48(4) 0.8 - OCH3 O9 Sn4 2.162(12) 0.5 1.77 O9 Sn4 2.162(12) 0.5 O9 C31 1.48(4) 0.8 - OCH3 O10 Sn3 2.119(13) 0.6 2.21 O10 Sn4 2.178(13) 0.5 O10 C32 1.33(3) 1.2

111

Table A1 (Continued)

- OCH3 O13 Sn5 2.15116 0.5 1.87 O13 Sn2 2.15316 0.5 O13 C33 1.453 0.8 - OCH3 O14 Sn5 2.132(17) 0.5 2.05 O14 Sn7 2.147(15) 0.5 O14 C34 1.39(3) 1.0 - OCH3 O15 Sn6 2.094(17) 0.6 1.86 O15 Sn7 2.148(17) 0.5 O15 C35 1.50(3) 0.7

Table A2: BVS for Oxo Ligands of γ-NaSn12.

Assignment Atom 1 Atom 2 d (A) BV BVS O2- O1 Sn1 2.114(11) 0.6 2.00 Central oxo O1 Sn2 2.102(11) 0.6 O1 Sn3 2.091(10) 0.6 O1 Na1 2.339(12) 0.2 O2- O2 Sn4 2.062(11) 0.7 2.07 Central oxo O2 Sn5 2.079(9) 0.6 O2 Sn6 2.125(10) 0.6 O2 Na1 2.329(12) 0.2 O2- O3 Sn7 2.078(9) 0.6 2.05 Central oxo O3 Sn8 2.074(11) 0.6 O3 Sn9 2.121(10) 0.6 O3 Na1 2.336(12) 0.2 O2- O4 Sn10 2.082(9) 0.6 2.03 Central oxo O4 Sn11 2.081(10) 0.6 O4 Sn12 2.118(10) 0.6 O4 Na1 2.356(11) 0.2 O2- O17 Sn2 2.035(11) 0.7 1.35 O17 Sn7 2.066(11) 0.6 OH- O18 Sn2 2.072(12) 0.6 1.21 O18 Sn8 2.1081(1) 0.6 OH- O19 Sn3 2.074(12) 0.6 1.23 O19 Sn10 2.091(12) 0.6 OH- O20 Sn3 2.110(11) 0.6 1.13 O20 Sn10 2.123(12) 0.6 OH- O21 Sn1 2.078(12) 0.6 1.22 O21 Sn4 2.097(11) 0.6

112

Table A2 (Continued)

OH- O22 Sn4 2.065(11) 0.6 1.31 O22 Sn12 2.059(11) 0.7 O2- O23 Sn1 2.042(10) 0.7 1.38 O23 Sn5 2.043(10) 0.7 OH- O24 Sn5 2.100(13) 0.6 1.33 O24 Sn7 2.017(13) 0.7 O2- O25 Sn6 2.040(12) 0.7 1.35 O25 Sn9 2.064(11) 0.7 O2- O26 Sn6 2.037(9) 0.7 1.38 O26 Sn12 2.045(10) 0.7 O2- O27 Sn9 2.039(10) 0.7 1.41 O27 Sn10 2.031(10) 0.7 OH- O28 Sn8 2.080(11) 0.6 1.22 O28 Sn11 2.094(11) 0.6 - OCH3 O5 Sn1 2.143(12) 0.5 1.99 O5 Sn2 2.147(12) 0.5 O5 C49 1.41(2) 0.9 - OCH3 O6 Sn1 2.128(11) 0.5 1.95 O6 Sn3 2.196(12) 0.5 O6 C50 1.41(2) 0.9 - OCH3 O7 Sn2 2.138(11) 0.5 1.92 O7 Sn3 2.149(12) 0.5 O7 C51 1.44(2) 0.9 - OCH3 O8 Sn4 2.144(11) 0.5 1.86 O8 Sn5 2.184(13) 0.5 O8 C52 1.44(2) 0.9 - OCH3 O9 Sn4 2.146(11) 0.5 1.88 O9 Sn6 2.170(13) 0.5 O9 C53 1.44(2) 0.9 - OCH3 O10 Sn5 2.173(11) 0.5 1.90 O10 Sn6 2.144(10) 0.5 O10 C54 1.43(2) 0.9 - OCH3 O11 Sn7 2.190(13) 0.5 1.88 O11 Sn8 2.114(12) 0.6 O11 C55 1.45(2) 0.8 - OCH3 O12 Sn7 2.152(11) 0.5 1.87 O12 Sn9 2.172(11) 0.5 O12 C56 1.44(2) 0.9 - OCH3 O13 Sn8 2.150(12) 0.5 1.89 O13 Sn9 2.159(13) 0.5 O13 C57 1.44(2) 0.9

113

Table A2 (Continued)

- OCH3 O14 Sn10 2.156(11) 0.5 2.10 O14 Sn11 2.153(10) 0.5 O14 C58 1.36(2) 1.0 - OCH3 O15 Sn10 2.171(12) 0.5 2.05 O15 Sn12 2.131(10) 0.5 O15 C59 1.38(2) 1.0 - OCH3 O16 Sn11 2.149(12) 0.5 2.00 O16 Sn12 2.192(12) 0.5 O16 C60 1.38(2) 1.0

Table A3: BVS for Oxo Ligands of γ-NaSn13.

Assignment Atom 1 Atom 2 d (Å) BV BVS O2- O1 Sn1 2.084(13) 0.6 2.00 Central oxo O1 Sn2 2.138(16) 0.5 O1 Sn3 2.093(17) 0.6 O1 Na1 2.324(16) 0.2 O2- O2 Sn4 2.05(2) 0.7 2.09 Central oxo O2 Sn5 2.111(16) 0.6 O2 Sn6 2.102(15) 0.6 O2 Na1 2.31(2) 0.2 O2- O3 Sn7 2.110(17) 0.6 2.02 Central oxo O3 Sn8 2.106(13) 0.6 O3 Sn9 2.086(18) 0.6 O3 Na1 2.320(15) 0.2 O2- O4 Sn10 2.100(16) 0.6 2.15 Central oxo O4 Sn11 1.99(3) 0.8 O4 Sn12 2.108(16) 0.6 O4 Na1 2.40(2) 0.2 O2- O17 Sn1 2.102(15) 0.6 2.04 O17 Sn8 2.143(15) 0.5 O17 Sn13 1.931(15) 0.9 OH- O18 Sn1 2.036(15) 0.7 1.30 O18 Sn8 2.098(16) 0.6 OH- O19 Sn2 2.095(18) 0.6 1.17 O19 Sn4 2.112(16) 0.6 O2- O20 Sn2 2.019(16) 0.7 1.48 O20 Sn6 2.015(16) 0.7

114

Table A3 (Continued)

O2- O21 Sn3 2.042(17) 0.7 1.79 O21 Sn11 2.087(15) 0.6 O21 Sn13 2.172(16) 0.5 O2- O22 Sn3 2.081(16) 0.6 1.31 O22 Sn12 2.045(16) 0.7 OH- O23 Sn4 2.105(16) 0.6 1.16 O23 Sn7 2.111(19) 0.6 O2- O24 Sn5 2.024(16) 0.7 1.45 O24 Sn7 2.024(16) 0.7 O2- O25 Sn5 2.18(3) 0.5 1.33 O25 Sn10 1.96(3) 0.9 OH- O26 Sn6 2.02(3) 0.7 1.30 O26 Sn12 2.11(3) 0.6 OH- O27 Sn9 2.102(17) 0.6 1.27 O27 Sn10 2.044(16) 0.7 O2- O28 Sn9 2.104(17) 0.6 1.85 O28 Sn11 2.111(15) 0.6 O28 Sn13 2.041(16) 0.7 OH- O12 Sn8 2.148(18) 0.5 1.07 O12 Sn9 2.124(14) 0.6 - OCH3 O5 Sn1 2.191(16) 0.5 1.84 O5 Sn3 2.140(14) 0.5 O5 C53 1.45(3) 0.8 - OCH3 O6 Sn1 2.113(16) 0.6 2.05 O6 Sn2 2.148(14) 0.5 O6 C54 1.40(3) 1.0 - OCH3 O7 Sn2 2.147(18) 0.5 1.80 O7 Sn3 2.163(18) 0.5 O7 C55 1.48(3) 0.8 - OCH3 O8 Sn4 2.09(2) 0.6 1.95 O8 Sn5 2.17(3) 0.5 O8 C56 1.44(5) 0.9 - OCH3 O9 Sn5 2.178(18) 0.5 1.89 O9 Sn6 2.149(18) 0.5 O9 C57 1.43(3) 0.9 - OCH3 O10 Sn4 2.094(18) 0.6 1.85 O10 Sn10 2.18(3) 0.5 O10 C58 1.48(4) 0.8

115

Table A3 (Continued)

- OCH3 O11 Sn8 2.154(14) 0.5 2.11 O11 Sn9 2.117(17) 0.6 O11 C59 1.37(6) 1.0 - OCH3 O13 Sn7 2.131(19) 0.5 1.87 O13 Sn9 2.190(18) 0.5 O13 C61 1.44(3) 0.9 - OCH3 O14 Sn10 2.25(3) 0.4 1.95 O14 Sn11 2.053(18) 0.7 O14 C62 1.43(4) 0.9 - OCH3 O15 Sn10 2.163(18) 0.5 1.84 O15 Sn12 2.163(18) 0.5 O15 C63 1.45(4) 0.8 - OCH3 O16 Sn11 2.090(18) 0.6 1.88 O16 Sn12 2.21(3) 0.4 O16 C64 1.45(5) 0.8 - OCH3 O29 Sn13 2.063(18) 0.7 1.73 On capping tin O29 C65 1.36(4) 1.1

Complete ESI-MS Peak Assignments

Figure A1. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)11(NaO4)(OH)17(OCH3)12] .

116

Figure A2. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4)(O)10(OH)(OCH3)7] .

Figure A3. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4)(O)9(OH)2(OCH3)8]

117

Figure A4. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4)(O)8(OH)3(OCH3)9] .

Figure A5. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4)(O)7(OH)4(OCH3)10] .

118

Figure A6. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4) (O)6(OH)5(OCH3)11] .

Figure A7. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4)(O)4(OH)7(OCH3)13] .

119

Figure A8. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4) (O)5(OH)6(OCH3)12] .

Figure A9. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [(BuSn)12(NaO4)(O)5(OH)13(OCH3)9(CH3OH)9] .

120

Figure A10. Size distribution analysis of SAXS data of β,γ-NaSn12 in benzene.

Figure A11. Modelling II results for β,γ-NaSn12 in benzene. The experimental scattering curve is in red and the calculated model in gray. The calculated radius is consistent with the radius determined from the experimental crystal structure.

121

Figure A12. IR spectrum of β,γ-NaSn12 showing the presence of methoxy ligands indicated by the strong C-O stretch at 1038 cm-1.

122

APPENDIX B Supporting Information for Chapter 4

Table B1: Bond Valence Sum for β-CaSn12

Assignment Atom 1 Atom 2 d (Å) BV BVS Ca1 O1 2.288(19) 0.4 Ca2+ Ca1 O3 2.263(16) 0.4 1.71 Ca1 O3 2.263(16) 0.4 Ca1 O2 2.31(2) 0.4 O1 Sn1 2.096(11) 0.6 O1 Sn1 2.096(11) 0.6 O2- 2.16 O1 Sn2 2.129(19) 0.5 O1 Ca1 2.288(19) 0.4 O2 Sn3 2.097(13) 0.6 O2 Sn3 2.097(13) 0.6 O2- 2.14 O2 Sn6 2.12(2) 0.6 O2 Ca1 2.31(2) 0.4 O3 Sn4 2.118(16) 0.6 O3 Sn5 2.100(16) 0.6 O2- 2.16 O3 Sn4 2.118(16) 0.6 O3 Ca1 2.263(16) 0.4 O4 Sn1 2.075(16) 0.6 OH- 1.34 O4 Sn5 2.031(17) 0.7 O5 Sn1 2.039(16) 0.7 OH- 1.35 O5 Sn1 2.061(17) 0.7 O8 Sn2 2.039(17) 0.7 O2- 1.37 O8 Sn3 2.050(16) 0.7 O11 Sn5 2.106(8) 0.6 OH- 1.16 O11 Sn5 2.106(8) 0.6 O14 Sn6 2.02(2) 0.7 O2- 1.41 O14 Sn7 2.047(19) 0.7 O15 Sn7 2.050(9) 0.7 OH- 1.35 O15 Sn7 2.050(9) 0.7 O17 Sn3 2.071(19) 0.6 OH- 1.26 O17 Sn4 2.08(2) 0.6 O6 Sn1 2.171(16) 0.5 OMe- O6 Sn2 2.139(18) 0.5 2.15 O6 C52 1.34(3) 1.1

123

Table B1 (Continued)

O7 Sn1 2.161(13) 0.5 OMe- O7 Sn1 2.161(13) 0.5 1.74 O7 C51 1.50(4) 0.7 O9 Sn3 2.110(18) 0.6 OMe- O9 Sn3 2.110(18) 0.6 1.99 O9 C54 1.45(5) 0.8 O10 Sn3 2.170(19) 0.5 OMe- O10 Sn6 2.16(2) 0.5 1.74 O10 C53 1.49(4) 0.8 O12 Sn4 2.11(2) 0.6 OMe- O12 Sn5 2.17(2) 0.5 1.90 O12 C50 1.45(4) 0.8 O13 Sn7 2.130(19) 0.5 OMe- O13 Sn5 2.162(18) 0.5 1.76 O13 C56 1.51(3) 0.7 O16 Sn4 2.158(16) 0.5 OMe- O16 Sn7 2.20(2) 0.4 1.92 O16 C55 1.40(3) 1.0

Table B2. Atomic percentages for selected elements in β-CaSn12 determined by SEM-EDX

Na At% Cl At% Sn At% Ca At% Area 1 2.40 1.35 78.81 17.44 Area 2 1.54 1.83 81.67 14.96 Area 3 2.44 3.02 77.89 16.65

124

Complete ESI-MS Peak Assignments

Figure B1. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)11(OH)6] .

Figure B2. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)10(OH)8] .

125

Figure B3. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)11(OH)4(OCH3)2] .

Figure B4. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)11(OH)3(OCH3)3] .

126

Figure B5. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)8(OH)11(OCH3)] .

Figure B6. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)10(OH)4(OCH3)4] .

127

Figure B7. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)8(OH)10(OCH3)2] .

Figure B8. Experimental ESI MS (+, blue spectrum) and calculated peak positions (red) 2+ for [(BuSn)12(CaO4)(O)10(OH)3(OCH3)5] .

128

1 Figure B9. Full H NMR spectrum of β-CaSn12 in C6D6 (red) and β,γ-NaSn12 (blue)

119 Figure B10. Full Sn NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue)

129

13 Figure B11. C NMR spectrum of β-CaSn12 (red) and β,γ-NaSn12 (blue)

Figure B12. Scattering curve of β-CaSn12 in THF (red) and spherical model of the data (gray). The model gives a cluster radius of 4.9Å, a center-to-center distance between clusters of 8.6Å, and 0.87 nearest neighbors.

(b 130

1 Figure B13. (a) Aging of β-CaSn12 in CDCl3 monitored by H NMR. (b) Aging of β-CaSn12 1 in 90% CDCl3/10% MeOD monitored by H NMR.

119 Figure B14. Sn NMR spectrum of cooled β-CaSn12 solution after heating.

131

Figure B15. Full FT-IR spectra of β-CaSn12 from 1 to 28 days after isolation.

132

APPENDIX C Supporting Information for Chapter 5

1 Figure C1. H NMR spectra of β,γ-NaSn12 over time in C6D6/CD3OD (a) and CDCl3/CD3OD (b).

133

Figure C2. Full FT-IR spectra of β,γ-NaSn12 from 1 to 28 days after isolation.

23 Figure C3. Na NMR of β,γ-NaSn12 in C6D6 heated to 60°C (blue) and then cooled to room temperature (red).

134

APPENDIX D Supporting Information for Chapter 6

Chemical Analysis

Chemical analysis by EDS and wet methods were undertaken for both Zr4 and Zr25. The wet method consisted of dissolution of usually damp crystals in water, neutralization with

excess NH4OH, filtration and ignition of the wet cake to ZrO2, evaporation of the filtrate

to recover dry NH4ClO4. This determined a ratio, r =mol perchlorate/mol Zr. For Zr4, this

value was 2.0 ±0.05. For Zr25, the combined factors of low yield and liquor viscosity meant that even when pressed between filter papers, the crystal were damp and cohesive, indicating unremoved liquor. The properties of the liquor were measured and it was found to have a density of 1.7 g/mL and to be 3M in Zr and 6 M in perchlorate. For a typical

sample of damp Zr25, r=0.90, and, based on recovered oxide, an “effective formula weight” (EFW) of 320/ mol Zr. Due to liquor, both r and EFW are overestimates, but this, and the

EDXS results tabulated below can assist the specification of the overall formula of Zr25, as noted below.

Table D1: Cl/Zr for Zr4 and Zr25 by wet and EDXS analyses

Phase Wet EDS

Zr4 2.0 ± 0.05 1.68 ± 0.47 Zr25 0.91 ± 0.03 0.76 ± 0.04 From the table, it appears that EDS underestimates perchlorate by a factor = 2.0/1.68 ≈1.2.

If this adjustment is made to the EDS result for Zr25, a perchlorate number of 0.90, or, 22.6 perchlorate per 25 Zr, which is close to the necessary 20 perchlorates for charge-balance of Zr25.

Table D2: Atomic percentages of Zr and Hf in Zr4 determined by EDS

Zr atomic % Hf atomic % Area 1 99.21 0.79 Area 2 99.48 0.52

135

Crystallographic data

Table D3: Crystal data and structure refinement details for Zr4, Zr25, and ZrTd compounds

Zr4 Zr25 ZrTd Empirical formula Cl24.02H96O168.1Zr12 H80O110Zr25 H36O49Cl6Zr4 Formula weight 4732.77 4121.14 1397.87 Temperature/K 171 100 173.15 Crystal system tetragonal orthorhombic monoclinic Space group I4/m Cmcm P21/n a/Å 16.7157(5) 31.6822(4) 10.4563(3) b/Å 16.7157(5) 23.4638(2) 21.6794(6) c/Å 27.2185(10) 27.1118(4) 17.1366(4) α/° 90 90 90 β/° 90 90 90.8320(10) γ/° 90 90 90 Volume/Å3 7605.2(5) 20154.5(4) 3884.22(18) Z 2 4 4 3 ρcalcg/cm 2.067 1.358 2.390 μ/mm-1 11.694 1.297 13.754 F(000) 4659.0 7840.0 2760.0 0.307 × 0.214 × 0.07 × 0.05 × Crystal size/mm3 0.05 × 0.04 × 0.03 0.193 0.03 MoKα (λ = CuKα (λ = Radiation CuKα (λ = 1.54178) 0.71073) 1.54178) 2Θ range for data 6.574 to 6.204 to 133.26 3.782 to 59.284 collection/° 133.288 -43 ≤ h ≤ 44, -32 -12 ≤ h ≤ 12, -25 -15 ≤ h ≤ 19, -19 ≤ k Index ranges ≤ k ≤ 30, -37 ≤ l ≤ ≤ k ≤ 25, -20 ≤ l ≤ 19, -23 ≤ l ≤ 32 37 ≤ 20 Reflections collected 13981 141850 31640 14706 [Rint = 6864 [Rint = 3457 [Rint = 0.0584, Independent reflections 0.0280, Rsigma = 0.0615, Rsigma = Rsigma = 0.0539] 0.0132] 0.0462] Data/restraints/parameters 3457/0/286 14706/3/359 6864/36/676 Goodness-of-fit on F2 1.041 1.049 1.047 Final R indexes [I>=2σ R1 = 0.0729, wR2 = R1 = 0.0484, wR2 R1 = 0.0333, (I)] 0.2106 = 0.1614 wR2 = 0.0788 R1 = 0.0963, wR2 = R1 = 0.0525, wR2 R1 = 0.0396, Final R indexes [all data] 0.2322 = 0.1645 wR2 = 0.0818 Largest diff. peak/hole / e 1.25/-1.33 0.94/-0.82 1.13/-1.31 Å-3

136

In Table D4, we show interatomic distances for CVTs in three phases to illustrate the invariance of the CVT to metal and anion identities

Table D4: Distances, Å, in CVT Motifs for several phases.

Phase ZOC HOC Zr4 M-aqua ligands 2.211 2.198 2.198 2.228 2.214 2.219 2.293 2.274 2.276 2.309 2.297 2.333 M-OH 2.114 2.101 2.100 2.116 2.106 2.107 2.150 2.138 2.110 2.153 2.151 2.113 OH-OH 2.339 2.325 2.240 M-M 3.551 3.537 3.557

Details of crystallographic disorder for Zr4 and Zr25.

- In the Zr4 compound, ClO4 ions located in the solvent region are highly disordered and delocalized. The occupancy of ions with Cl1, Cl3 and Cl5 was first refined then fixed to the found value (0.85, 0.87 and 0.33, respectively). The occupancies of oxygen atoms attached to these Cl have also been fixed to the same values. The other perchlorate ions possess a full occupancy but some oxygen atoms are also disordered on two positions and their occupancies were set to 0.5.

The electron densities are diffuse, and we cannot locate these exactly. The preliminary R1 value was reduced from 12% to 5% by applying a “solvent mask”. The solvent mask indicated the void space has a volume of 12229.6 Å3 and 2524.0 unassigned electrons. Considering the reaction solutions, we expect only perchlorate and water molecules in this interlayer space. These species contain 49 and 10 electrons respectively, indicating a

maximum of 12 ClO4 and 4 H2O, not sufficient to balance the 20+ charge of the cluster. Since the volume of the perchlorate ion is 100 - 150 Å3, there is room for up to 26 water

molecules per Zr25 cluster. This complicating factor challenged assigning a precise

composition for the Zr25 compound based on diffraction data alone. Here we address the

137

issues of perchlorate and solvent water stoichiometry, using data from chemical and energy dispersive spectroscopy (EDS) analyses.

As presented in Chapter 6, the central five-membered ring in the Zr25 cluster is slightly delocalized on two positions (opaque and transparent color in

Figure D1. Representation of the Zr25 cluster with disordered atoms

). The second one rotated by 36° and the occupancy of the Zr atoms and their μ2-hydroxyl groups was fixed to 0.05. First, we refined the occupancies of those Zr atoms and found values between 3.2 and 5.7% for the three independent cations and fixed it arbitrarily to 5%. The other Zr and O atoms in the central ring are then fixed to 95% occupancy to balance. Each delocalized Zr seems to be only linked to 4 μ2-hydroxyl groups (Zr-OOH =

2.03(5)-2.19(2) Å) and to two μ3-oxygen atoms (found as oxo groups in the non-disordered cluster) with bond distances in the range 2.05-2.25 Å. The potential linkage between these Zr and the oxygen atom from disordered hydroxyl/peroxo cannot exist due unrealistically long distances (2.96-3.11 Å). However, during the refinement process some very weak residual electronic densities were observed with reasonable bond distance (2.21-2.56 Å). We interpret this to mean the external ring might also rotate with the inner ring but the occupancy of this delocalized cluster is too weak to be conclusively observed.

138

Figure D1. Representation of the Zr25 cluster with disordered atoms

Figure D2. Representation of the Zr25 cluster with 5% occupancy of the inner ring. View of the coordination of the delocalized Zr atom. Dashed bonds correspond to long bonds The bond valence calculations[173] for some of the oxygen atoms have been done to confirm their nature: oxo (expected value: 2.0) or hydroxo group(expected value: 1.2). The results are given in the Table D5. These values support the assignment of oxo and hydroxyl groups and the cation formula.

139

Table D5: Bond valence sums and assignments of oxo and hydroxyl groups in Zr25.

Atom BVS Assignment O3 2.043 μ3-oxo O7 2.052 μ3-oxo O11 2.057 μ3-oxo O14 2.041 μ3-oxo O20 2.047 μ3-oxo O4 1.039 μ3-hydroxo O15 1.045 μ3-hydroxo O21 1.031 μ3-hydroxo O2b 1.001 μ2-hydroxo O5 1.101 μ2-hydroxo O6 1.151 μ2-hydroxo O8 1.152 μ2-hydroxo O10b 0.996 μ2-hydroxo O12 1.107 μ2-hydroxo O13 1.137 μ2-hydroxo O16 1.149 μ2-hydroxo O18b 1.028 μ2-hydroxo O19 1.119 μ2-hydroxo O22 1.17 μ2-hydroxo O23b 1.012 μ2-hydroxo O24b 1.189 μ2-hydroxo O25b 1.040 μ2-hydroxo Small angle X-ray scattering data

Figure D3. (right) Cylindrical fit (red) to experimental SAXS curve of Zr4 in methanol (black) having a radius of 3.3Å and a length of 22.8Å. (middle) Two-population spherical fit (red) to experimental SAXS curve of Zr25 in methanol (black). Population 1 radius:8.6Å. Population 2 radius: 3.2Å. (right) Cylindrical fit (gray) to experimental SAXS curve of ZrTd in methanol (red). Radius: 3.1Å, length: 21.6Å

140

Figure D4. Comparison of experimental scattering data for ZrTd reaction solution (day 4; blue) compared to simulated scattering curve for ‘Zr22’ (red). To create scattering data for ‘Zr22’, we used the structure of Pu22, approximately formulated [Pu22O28(OH)4Cl28(H2O)20]; changed the Pu to Zr and Cl ligands to oxygen to approximately simulate the size, shape and electron density of a hypothetical Zr22. Inset, Pu(Zr)22. Metals are green, oxygen is red.

Figure D5. Experimental and simulated PDF of the ZrTd reaction solution after 10 days of aging

141

ESI-MS Data

Figure D6. (left) ESI-MS spectrum of Zr4 in methanol. Negative ionization mode, 30V fragmentation. (right) ESI-MS spectrum of Zr4 in water. Positive ionization mode, 100V fragmentation.

Figure D7. (left) ESI-MS spectrum of Zr25 in MeOH. Negative ionization mode, 100V fragmentation. (right) ESI-MS spectrum of Zr25 in water. Positive ionization mode, 100V fragmentation.

142

Figure D8. (left) ESI-MS spectrum of ZrTd in MeOH. Negative ionization mode, 30V fragmentation. (right) ESI-MS spectrum of ZrTd in water. Positive ionization mode, 30V fragmentation.

Table D6: Peak assignments for mass spectrum of Zr4 in MeOH

Formula Observed m/z Calculated m/z - [H(ClO4)2] 198.9079 198.9054 - [(CH3OH)(ClO4)(H2O)5] 220.8904 221.0281 - [Zr(OCH3)2(ClO4)3] 450.7903 450.7856 - [Zr(OCH3)(ClO4)4] 518.7210 518.7156 - [Zr(ClO4)5] 588.7036 588.6444 - [Zr2(CH3OH)2O(OH)3(ClO4)4] 710.6443 710.6577 - [Zr3O4(OCH3)(ClO4)4(H2O)2] 802.6297 802.5266 - [Zr4(CH3OH)2(OH)11(ClO4)6] 1212.4284 1212.3909 - [Zr4(OH)5(OCH3)6(ClO4)6] 1232.3253 1232.4325 - [Zr4(CH3OH)(OH)4(OCH3)7(ClO4)6] 1278.3656 1278.4744 - [Zr3(OCH3)4(ClO4)9(H2O)3] 1346.2946 1346.3520

143

Table D7: Peak assignments for mass spectrum of Zr4 in water

Formula Observed m/z Calculated m/z + [Zr4O2(OH)10(ClO4)] 664.5571 664.5848 + [Zr4O5(OH)3(ClO4)2(H2O)2] 730.4878 730.5197 + [Zr4O6(ClO4)3(H2O)3] 812.4303 812.4647 + [Zr3O3(ClO4)5(H2O)2] 854.4353 854.4600 + [Zr4O4(OH)3(ClO4)4(H2O)] 894.3803 894.4098 + [Zr4O4(OH)3(ClO4)4(H2O)2] 912.3803 912.4204 + [Zr6O3(OH)17(H2O)3] 938.3805 938.4939 + [Zr6O5(OH)13(H2O)6] 956.3859 956.5045 + [Zr4O5(ClO4)5(H2O)2] 978.3266 978.3550 + [Zr4O5(ClO4)5(H2O)3] 996.3247 996.3656 + [Zr4O(OH)9(ClO4)4(H2O)6] 1038.3320 1038.4944 + [Zr4O4(OH)(ClO4)6(H2O)] 1060.2718 1060.3001 + [Zr4O4(OH)(ClO4)6(H2O)2] 1078.2701 1078.3107 + [Zr5O7(ClO4)5(H2O)2] 1100.2596 1100.2500 + [Zr6(OH)19(ClO4)4(H2O)2] 1304.1286 1304.2953

Table D8: Peak assignments for mass spectrum of Zr25 in MeOH

Formula Observed m/z Calculated m/z - [Zr6(OH)19(O2)3(H2O)3] 1020.3659 1020.4853 - [Zr8(O)14(OH)(O2)(OCH3)2] 1064.4062 1064.2004 - [Zr8(O)13(OH)(O2)(OCH3)4] 1110.4441 1110.2423 - [Zr8(O)14(OH)5(CH3OH)3] 1134.3380 1134.2635 - [Zr8(O)6(OH)19(O2)] 1180.3772 1180.2536 - [Zr8(O)7(OH)12(O2)2(OCH3)3] 1202.2671 1202.2745 - [Zr6(O2)11(OCH3)3(H2O)13] 1226.4188 1226.5128 - [Zr8(O)4(OH)23(O2)(CH3OH)] 1248.3080 1248.3010 - [Zr8(OH)22(O2)5(OCH3)] 1294.3486 1294.2702 - [Zr8(OH)14(O2)8(OCH3)3] 1316.2390 1316.2546 - [Zr8(O2)13(OCH3)7] 1362.2784 1362.2390

144

Table D9: Peak assignments for mass spectrum of Zr25 in water

Formula Observed m/z Calculated m/z + [Zr3(OH)8(O2)(H2O)4(ClO4)] 612.6141 612.7168 + [Zr2(O)2(H2O)7(ClO4)3] 638.6164 638.7170 + [Zr5(OH)(O)4(O2)5] 696.5712 696.4568 + [Zr5(OH)3(O)7(O2)(H2O)4] 722.5683 722.5300 + [Zr5(OH)5(O)6(O2)(H2O)4] 740.5727 740.5406 + [Zr5(O)(OH)17] 760.5265 760.5668 + [Zr5(OH)19] 778.5119 778.5774 + [Zr5(OH)(O)3(O2)6(H2O)5] 802.5127 802.5046 + [Zr5(OH)5(O)(O2)6(H2O)4] 820.5169 820.5152 + [Zr6(OH)(O)8(O2)3(H2O)4] 860.4608 860.4047 + [Zr6(OH)3(O)3(O2)7(H2O)4] 942.4048 942.3949

Table D10: Peak assignments for mass spectrum of ZrTd in MeOH

Formula Observed m/z Calculated m/z - [Zr3O(O2)2(ClO4)7(CH3OH)] 1080.4076 1080.3519 - [Zr4O2(O2)2(OCH3)3(ClO4)6] 1150.3291 1150.3330 - [Zr5O2(OH)8(OCH3)(ClO4)4(O2)2(CH3OH)2] 1180.3763 1180.3802 - [Zr6O3(OH)14(ClO4)3(O2)(H2O)2] 1198.3695 1198.3091 - [Zr3(ClO4)7(O2)3(H2O)9] 1226.4169 1226.4158 - [Zr4O2(OCH3)4(ClO4)7(O2)] 1248.3073 1248.3095 - [Zr4O(OCH3)6(ClO4)7(O2)] 1294.3473 1294.3514 - [Zr3O(OH)(ClO4)10(H2O)] 1318.2396 1318.2029

Table D11: Peak assignments for mass spectrum of ZrTd in water

Formula Observed m/z Calculated m/z 1+ [Zr3(O)(OH)4(O2)(H2O)5(ClO4)3] 776.4964 776.6069 1+ [Zr3(O)(OH)(O2)2(H2O)2(ClO4)4] 804.4955 804.5047 1+ [Zr5(O)5(OH)4(O2)2(H2O)3(ClO4)] 820.5021 820.4699 1+ [Zr4(O)(OH)4(O2)3(H2O)(ClO4)3] 860.4393 860.4495 1+ [Zr3(O)(OH)2(O2)(H2O)5(ClO4)5] 942.3866 942.4972 1+ [Zr4(O)2(OH)3(O2)2(H2O)3(ClO4)4] 962.3947 962.4208 1+ [Zr6(O)6(OH)7(O2)(H2O)2(ClO4)2] 1028.3368 1028.3258 1+ [Zr4(O)4(O2)(H2O)5(ClO4)5] 1048.3144 1048.3817 1+ [Zr4(O)(OH)2(O2)(H2O)3(ClO4)9] 1396.1119 1396.1726

145

Complete ESI-MS Peak Assignments for Zr4 in MeOH

Figure D9. Experimental ESI MS (-, blue spectrum) and calculated peak positions (red) - for [H(ClO4)2] .

Figure D10. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [(CH3OH)(ClO4)(H2O)5] .

146

Figure D11. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr(OCH3)2(ClO4)3] .

Figure D12. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr(OCH3)(ClO4)4] .

147

Figure D13. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr(ClO4)5] .

Figure D14. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr2(CH3OH)2O(OH)3(ClO4)4] .

148

Figure D15. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr3O4(OCH3)(ClO4)4(H2O)2] .

Figure D16. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr4(CH3OH)2(OH)11(ClO4)6] .

149

Figure D17. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr4(OH)5(OCH3)6(ClO4)6] ..

Figure D18. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr4(CH3OH)(OH)4(OCH3)7(ClO4)6] .

150

Figure D19. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr3(OCH3)4(ClO4)9(H2O)3] .

Complete ESI-MS Peak Assignments for Zr4 in Water

Figure D20. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O2(OH)10(ClO4)] .

151

Figure D21. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O5(OH)3(ClO4)2(H2O)2] .

Figure D22. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O6(ClO4)3(H2O)3] .

152

Figure D23. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr3O3(ClO4)5(H2O)2] .

Figure D24. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O4(OH)3(ClO4)4(H2O)] .

153

Figure D25. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O4(OH)3(ClO4)4(H2O)2] .

Figure D26. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr6O3(OH)17(H2O)3] .

154

Figure D27. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr6O5(OH)13(H2O)6] .

Figure D28. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O5(ClO4)5(H2O)2] .

155

Figure D29. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O5(ClO4)5(H2O)3] .

Figure D30. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O(OH)9(ClO4)4(H2O)6] .

156

Figure D31. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O4(OH)(ClO4)6(H2O)] .

Figure D32. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr4O4(OH)(ClO4)6(H2O)2] .

157

Figure D33. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5O7(ClO4)5(H2O)2] .

Figure D34. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr6(OH)19(ClO4)4(H2O)2] .

158

Complete ESI-MS Peak Assignments for Zr25 in MeOH

Figure D35. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr6(OH)7(O2)8(OCH3)2(H2O)2] .

Figure D36. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(O)14(OH)(O2)(OCH3)2] .

159

Figure D37. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(O)13(OH)(O2)(OCH3)4] .

Figure D38. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(O)14(OH)5(CH3OH)3] .

160

Figure D39. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(O)6(OH)19(O2)] .

Figure D40. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(O)7(OH)12(O2)2(OCH3)3] .

161

Figure D41. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr6(O2)11(OCH3)3(H2O)13] .

Figure D42. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(O)4(OH)23(O2)(CH3OH)] .

162

Figure D43. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(OH)22(O2)5(OCH3)] .

Figure D44. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(OH)14(O2)8(OCH3)3] .

163

Figure D45. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr8(O2)13(OCH3)7] .

Complete ESI-MS Peak Assignments for Zr25 in Water

Figure D46. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr3(OH)8(O2)(H2O)4(ClO4)] .

164

Figure D47. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr2(O)2(H2O)7(ClO4)3] .

Figure D48. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5(OH)(O)4(O2)5] .

165

Figure D49. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5(OH)3(O)7(O2)(H2O)4] .

Figure D50. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5(OH)5(O)6(O2)(H2O)4] .

166

Figure D51. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5(O)(OH)17] .

Figure D52. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5(OH)19] .

167

Figure D53. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5(OH)(O)3(O2)6(H2O)5] .

Figure D54. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr5(OH)5(O)(O2)6(H2O)4] .

168

Figure D55. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr6(OH)(O)8(O2)3(H2O)4] .

Figure D56. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) + for [Zr6(OH)3(O)3(O2)7(H2O)4] .

169

Complete ESI-MS Peak Assignments for ZrTd in MeOH

Figure D57. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr3O(O2)2(ClO4)7(CH3OH)] .

Figure D58. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr4O2(O2)2(OCH3)3(ClO4)6] .

170

Figure D59. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr5O2(OH)8(OCH3)(ClO4)4(O2)2(CH3OH)2] .

Figure D60. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr6O3(OH)14(ClO4)3(O2)(H2O)2] .

171

Figure D61. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr3(ClO4)7(O2)3(H2O)9] .

Figure D62. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr4O2(OCH3)4(ClO4)7(O2)] .

172

Figure D63. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr4O(OCH3)6(ClO4)7(O2)] .

Figure D64. Experimental ESI-MS (-, blue spectrum) and calculated peak positions (red) - for [Zr3O(OH)(ClO4)10(H2O)] .

173

Complete ESI-MS Peak Assignments for ZrTd in Water

Figure D65. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr3(O)(OH)4(O2)(H2O)5(ClO4)3] .

Figure D66. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr3(O)(OH)(O2)2(H2O)2(ClO4)4] .

174

Figure D67. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr5(O)5(OH)4(O2)2(H2O)3(ClO4)] .

Figure D68. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr4(O)(OH)4(O2)3(H2O)(ClO4)3] .

175

. Figure D69. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr3(O)(OH)2(O2)(H2O)5(ClO4)5] .

Figure D70. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr4(O)2(OH)3(O2)2(H2O)3(ClO4)4] .

176

Figure D71. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr6(O)6(OH)7(O2)(H2O)2(ClO4)2] .

Figure D72. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr4(O)4(O2)(H2O)5(ClO4)5] .

177

Figure D73. Experimental ESI-MS (+, blue spectrum) and calculated peak positions (red) 1+ for [Zr4(O)(OH)2(O2)(H2O)3(ClO4)9] .

Figure D74. Experimental (black) and simulated (red) powder XRD diagrams of ZrTd.

178

Figure D75. Experimental (black) and simulated (red) powder XRD diagrams of Zr25.