Guest Binding, Redox, and Molecular Transport Properties of Supramolecular

Coordination Assemblies

by

Bryan Erik Tiedemann

B.S. (California Institute of Technology) 2002

A dissertation submitted in partial satisfaction of the

requirements for the degree of

Doctor of Philosophy

in

Chemistry

in the

GRADUATE DIVISION

of the

UNIVERSITY OF CALIFORNIA, BERKELEY

Committee in Charge:

Professor Kenneth N. Raymond, Chair Professor Jeffrey R. Long Professor Alexander Katz

Spring 2007

Guest Binding, Redox, and Molecular Transport Properties of

Supramolecular Coordination Assemblies

Copyright © 2007

Bryan Erik Tiedemann

- 1 -

Abstract

Guest Binding, Redox, and Molecular Transport Properties of Supramolecular

Coordination Assemblies

by

Bryan Erik Tiedemann

Doctor of Philosophy in Chemistry

University of California, Berkeley

Professor Kenneth N. Raymond, Chair

The rich host-guest chemistry of the M4L6 tetrahedral cluster developed by

Raymond and coworkers has been studied for the last ten years, yet this dissertation will

argue that its full capabilities have yet to be realized. Several different physical

techniques have been used to explore the properties of the M4L6 supramolecular cluster,

providing a different perspective. A brief summary of the known properties of M4L6 is

given in Chapter 1, as well a description of how electrochemical techniques can apply to

supramolecular systems.

Chapter 2 is concerned with molecular transport, using diffusion NMR methods to

measure diffusion coefficients. Exterior binding of alkylammonium cations was clearly

observed from changes in the diffusion coefficient, confirming that strong ion pairing

interactions persist in aqueous solution. The effect of different alkali cations on the

diffusion rate is also discussed.

- 2 -

The thermodynamics of host-guest interactions are investigated in Chapter 3 using

isothermal titration calorimetry to directly measure reaction heats. Large heats are

observed when [Ga4L6]12- is titrated with R4N

+ in aqueous solution (R = Me, Et, Pr), and

the resulting thermodynamic parameters for the overall processes are reported.

A novel partial guest encapsulation mode is described in Chapter 4. Cationic

ruthenium sandwich complexes with pendant linear chains are encapsulated by [Ga4L6]12-

with the result that part of the long chains protruding out of the host to the exterior.

Chains terminated by an anionic group are permanently extruded through the three-fold

symmetric aperture in a triangular face of the host, confirming the non-dissociative guest

exchange mechanism described by Davis et al. The methyl terminus of neutral chains

can reside either inside or outside the cluster, and such a chain with six carbons rapidly

extends and retracts at room temperature, moving the methyl group in and out of the host

cavity. This fluxional structure exemplifies the dynamic second-order Jahn-Teller effect.

Chapters 5 and 6 describe electrochemical experiments with M4L6. Chapter 5

explores whether redox-active cations can be reduced while encapsulated within a redox-

silent host; the answer is no for the systems examined. Chapter 6 investigates redox-

active vanadium complexes, including the [V4L6]8- cluster with a redox-silent Et4N

+

guest. The results demonstrate that, while the vertices are redox-active in the M4L6

cluster, there is no interaction between them.

- i -

For my grandfather

Albert E. Morjig

- ii -

ACKNOWLEDGEMENTS

My family has always been extremely supportive of my interests, whatever they

turned out to be. My fiancée Anishya Mathai has provided critical support and

inspiration during the most stressful times of my thesis preparation. I thank my parents,

Al and Nancy Tiedemann, and my grandparents, Florence and Al Morjig, for their

continued love and support. My brother Sanders and my sisters Whitney and Lori have

always been dear friends to me, always in the mood for a good time together.

Ken Raymond has been a wonderful teacher and a great friend to me throughout

my graduate career. I’ve always enjoyed going up to the white board to muddle through

some theoretical analysis explaining exactly why my hypothesis was wrong. He

provided me opportunities to do world-class research, and showed me the joys of sailing.

I will truly miss working with him.

My colleagues in the Raymond Group have been wonderful coworkers, all very

talented, and all willing to help one another. Didier Pomeranc taught me synthetic

chemistry, particularly air-sensitive manipulations. Anna Davis, Rob Yeh, Dorothea

Fiedler, and Dennis Leung were my mentors in the supramolecular group. Dorothea

deserves a lot of credit for help with the partial guest encapsulation project. Georg

Seeber was a good friend and was great to work with while writing our review article.

Mike Pluth is without a doubt the most talented chemistry gradate student I’ve ever met,

and our regular discussions were always stimulating. Shannon Biros is not only an

extremely talented supramolecular chemist, but a wonderful person who made 505 a

- iii -

much warmer place. Jeff Mugridge will be my successor for the ITC experiments, and it

was a pleasure to work with him for the short period we overlapped.

Rudi Nunlist deserves full credit for making UC Berkeley’s NMR facility a world

class laboratory. He was always willing to teach me all sorts of crazy things about NMR

and Linux, and without his help setting up the diffusion NMR experiments on the AV-

500, Chapter 2 would have been quite a bit slimmer. Herman van Halbeek was

invaluable for helping Mike and I set up the initial diffusion NMR experiments on the

AVB-400, and without his help Chapter 2 might not even exist.

Our collaborators in Sicily, Prof. Giuseppe Arena, Dr. Carmelo Sgarlata, and

Valeria Zito at the Università di Catania, were wonderful colleagues to work with.

During the two weeks while I was there, we not only achieved some fantastic scientific

feats, but they also showed me the true meaning of hospitality. They are all highly

skilled analytical chemists.

The redox-active guest electrochemical studies was also a collaborative effort

between our group and Burak Ulgut and Prof. Héctor D. Abruña at Cornell University. I

worked in their lab for a week, where I picked up all sorts of great tips and tricks for

electrochemical experiments. Much of the vanadium electrochemistry work was done

with a lot of advice from Prof. Marcin Majda here at UC Berkeley. The propanethiol

treatment of the Hg electrode was his idea, and it worked great.

My undergraduate advisor, Harry Gray, allowed my passion for chemistry to

flourish, and always let me visit him in his office to ask stupid questions. Randy

Villahermosa was Harry’s graduate student who mentored me for three years, I

appreciate all his patience while we worked together.

- iv -

TABLE OF CONTENTS

CHAPTER 1 – Introduction 1

Overview 1

Supramolecular Hosts 2

Supramolecular Electrochemistry 5

Summary 10

References 10

CHAPTER 2 – Diffusion of a Supramolecular Cluster: Ion Pairing Effects in

Aqueous Solution

14

Introduction 14

Results and Discussion 19

Summary 27

Experimental 28

References 33

CHAPTER 3 – Thermodynamics of Guest Binding: Calorimetry Experiments 36

Introduction 36

Results and Discussion 41

Summary 51

Experimental 52

References 55

- v -

CHAPTER 4 – Partial Guest Encapsulation Modes 58

Introduction 58

Zwitterionic Guests 59

Monocations with Pendant Alkanes 69

Summary 78

Experimental 78

References 91

CHAPTER 5 – Electrochemical Properties of Monocations Encapsulated by

an M4L6 Host

93

Introduction 93

Results and Discussion 94

Summary 111

Experimental 112

References 119

CHAPTER 6 – Redox-Active Vanadium Complexes 122

Introduction 122

Synthesis and Characterization of Vanadium(IV) Complexes 124

Electrochemical Studies of Vanadium Complexes 128

Summary 150

Experimental 151

References 160

- vi -

APPENDIX 1 – Supplementary Spectra for Chapter 4 163

APPENDIX 2 – Born Solvation Energies and Guest Encapsulation 184

APPENDIX 3 – Crystallographic Data for K4[V2LH

3]⋅⋅⋅⋅6.7 DMF⋅⋅⋅⋅Et2O⋅⋅⋅⋅0.3 H2O 188

- 1 -

CHAPTER 1

Introduction

Overview

This dissertation focuses on various aspects of the M4L6 supramolecular clusters

developed by Raymond and coworkers, where M = GaIII or VIV and H4L = 1,5-bis(2,3-

dihydroxybenzamido)naphthalene (Figure 1.1).1 These chiral, self-assembled tetrahedral

complexes have proven to be extremely versatile hosts, encapsulating a wide variety of

guests within their hydrophobic cavities.3, 4 While encapsulated, guests can undergo

reactions – both stoichiometric and catalytic – with significant rate enhancement and

improved product selectivity in some cases.5-7 This may lead to practical applications,

warranting a detailed study of the guest binding, redox, and molecular transport

Figure 1.1. (Left) Schematic structure of the M4L6 tetrahedral cluster illustrating the structure of L4- and its coordination to the metal ion vertices. (Right) Diagram of the host-guest complex [Et4N ⊂ Ga4L6]

11-, based on the X-ray structure coordinates, with Et4N

+ guest shown in blue.

- 2 -

properties of the M4L6 supramolecular coordination assemblies, which are the subjects of

this thesis.

Supramolecular Hosts

Supramolecular chemistry takes advantage of self-assembly to prepare large,

discrete structures from relatively simple subunits. Using metal-ligand interactions,

π-stacking, and/or hydrogen bonds to link the various subunits together, many elegant

examples of supramolecular assemblies have been designed and synthesized which have

been the subject of some excellent review articles.8-10 Some of these assemblies feature a

binding site for guest encapsulation through noncovalent interactions. Examples of the

utilization of host-guest properties of such clusters to modify guest reactivity include

work done by Fujita,11, 12 Reek,13 and Rebek.14

Raymond and coworkers used symmetry and geometrical constraints to rationally

design the components of the M4L6 cluster shown in Figure 1.1.1, 9, 15 When the rigid

two-fold symmetric H4L ligands are combined with a hard octahedral metal ion

(M = AlIII, FeIII, GaIII, InIII, TiIV, VIV, GeIV, SnIV) in the presence of a suitable guest, the

only product which forms is the M4L6 tetrahedral cluster.3, 10 Four pseudo-octahedral

metal ions are located at the vertices of the tetrahedron, linked by six bis-bidentate

catecholamide ligands spanning the edges. Although the cluster is constructed from

achiral subunits, each tris-bidentate metal center exhibits helical chirality, and coupling

by the rigid ligands forces all vertices to be homochiral; thus, the assembly has pure

rotation point group T symmetry. When prepared with trivalent metal ions such as

gallium(III), the anionic cluster has an overall -12 charge, making K12[Ga4L6] very water

- 3 -

soluble. The interior of the tetrahedron is surrounded by six naphthalene rings from the

ligands, creating a hydrophobic pocket that sharply contrasts the polar exterior

environment surrounding the solvated host. This 250 – 500 Å3 expandable cavity binds

lipophilic guests, preferably monocations such as tetraethylammonium (Et4N+) and

cobaltocenium (CoCp2+).

The encapsulation of a solvated molecule G by the M4L6 host (abbreviated as H)

to form the host-guest complex [G ⊂ H] (abbreviated as HG), with the guest located

inside the host cavity, can be expressed by the following equilibrium:

H + G = HG Kb =[HG]

[H][G]Kb =

[HG]

[H][G]

When comparing guest binding affinities, saying that A is a “stronger” or “better” guest

than B is commonly used to mean Kb(A) > Kb(B). This jargon can be convenient, and

will be used periodically in this dissertation.

Two different types of binding interactions are possible for lipophilic cations with

the M4L6 cluster: encapsulation (discussed above), and exterior binding interactions to

form an ion pair. These two processes are interrelated, and guest exchange is thought to

proceed through an ion pair intermediate.7, 16 The X-ray crystal structure of

K5[Et4N]7[Fe4L6] shows one Et4N+ cation inside the host cavity, and the other six Et4N

+

cations closely associated with aromatic rings in the ligands.1, 3 This exterior binding

persists in aqueous solution, as evident from upfield shifts of exterior Et4N+ 1H NMR

resonances in the presence of [Ga4L6]12-. Exterior ion pairing is driven not only by

coulombic attraction with the anionic cluster, but also favorable cation-π interactions with

the aromatic rings, as well as van der Waals forces. Guests with π systems can also

participate in favorable internal π-π interactions with the naphthalene rings.11 Exterior

- 4 -

ion pairing is explored in Chapter 2 using diffusion NMR measurements, and guest

binding thermodynamics are discussed in Chapter 3, using data from isothermal titration

calorimetry (ITC) experiments.

Guest Exchange Mechanism

Interior and exterior guest species can dynamically exchange in solution, and the

kinetics of both self- and cross-exchange reactions have been measured for a variety of

guests.6, 16, 17 During a cross-exchange reaction, a strongly bound guest will displace a

weaker guest to form a more stable host-guest complex. Davis and Raymond observed

that [Ga4L6]12- and [Ti4L6]

8- hosts facilitate guest exchange at essentially the same rate,

thus demonstrating the guest exchange mechanism does not require a ligand dissociation

step.2 [Titanium(IV) catechol complexes are very inert to ligand exchange, while

gallium(III) catechol complexes are labile, so much slower guest exchange rates would

be observed for [Ti4L6]8- if ligand dissociation was required.2, 18 ]

Small openings exist in the triangular faces of the host, and concerted cluster

distortion acts to enlarge these gaps for guests to pass through (Figure 1.2). The

Figure 1.2. The egress of Et4N+ (green molecule) from the [Ga4L6]

12- cavity proceeds through the three-fold symmetric apertures of an intact cluster, without breaking any metal-ligand bonds. The reaction energy diagram shown at left reveals the transition state features the guest partially extruded through a dilated aperture (point C). The increase in energy at the far right of the reaction coordinate is an artifact of the calculation being performed in the gas phase. Reprinted with permission from Davis and Raymond.2

- 5 -

activation barrier height for the guest exchange process is determined by how wide the

aperture must be to accommodate the steric bulk of the guest in the transition state. Since

the distortion resembles a vibrational breathing mode, one can approximate the potential

energy profile with Hooke's law, proportional to the square of the radial distortion from

its equilibrium size. Small guests such as Me4N+ require small distortions, and fast self-

exchange rates are observed from line broadening analysis.19 Larger guests such as Pr4N+

require larger distortions to pass through the aperture, and slower self-exchange rates are

observed (for Pr4N+, k298 = 1.4 s-1).16 However, for asymmetric guests, where one

dimension is longer than the others, the aperture only has to open enough to allow the

smallest cross section to pass through. This is exemplified by Me2Pr2N+, whose self-

exchange rate of k298 = 4.4 s-1 is much faster than Pr4N+ because it can enter methyl-

group first, with the long propyl chains trailing behind.16 In contrast, ground state effects

of the smaller yet strong binding Et4N+ guest make its self-exchange rate very slow

(k298 = 0.009 s-1).16

Chapter 4 explores the consequences of the nondissociative guest extrusion

mechanism by synthesizing two sets of RuII sandwich complexes that have pendant alkyl

chains with different terminal groups. The mechanism is supported by the observation

of a stable cluster in D2O with part of the chain protruding through one of the three-fold

symmetric openings in the triangular faces of the tetrahedron.

Supramolecular Electrochemistry

The two primary topics of supramolecular electrochemistry are systems with

multiple redox sites and electrochemical switching. Chapter 5 explores the use of

- 6 -

[Ga4L6]12- as a redox-silent host for switchable binding of redox-active metallocene

guests, and Chapter 6 describes the electrochemical behavior of M4L6 with electroactive

vanadium ions at the vertices, as well as the analogous dinuclear vanadium helicate and a

mononuclear vanadium tris(catecholamide) model complex. The primary technique used

to study the electrochemical behavior of these systems is cyclic voltammetry. A brief

introduction to some important electrochemical concepts related to supramolecular

systems is given here.

Electrochemical Switching

Electrochemical switching is the most important application of electrochemistry

in the field of supramolecular chemistry.20 In a redox-switchable system, guest binding

events can be precisely controlled simply by applying voltage to a solution, changing the

oxidation state of a redox-active site on either the host or the guest. The two oxidation

states exhibit different binding equilibrium constants, preferably one much larger than the

other. For practical purposes, the redox-active site must exhibit reversible electron

transfer kinetics at the electrode; otherwise, kinetic limitations would render the switch

too slow to be useful.

The coupling of the electrochemical and guest binding equilibria can be described

by a simple square scheme. Following the description by Kaifer et al., such a scheme is

shown in Figure 1.3.20 For this example, the guest G is electroactive and is switched

from its low to high binding state by reduction, forming more stable complexes with

redox-silent host H when it is reduced (G) than where it is oxidized (G+). Therefore, the

binding equilibrium constant Kred is larger than Kox. The reduction potential of the host-

- 7 -

guest complex E±

b0E±

b0 is shifted from that for the free guest E±

f0E±

f0 according to the following

expression:

E±

b0 = E±

f0 +

RT

FlnKred

Kox

E±

b0 = E±

f0 +

RT

FlnKred

Kox (1.1)

where F is the Faraday constant, R is the molar gas constant, and T is the temperature in

Kelvin. A similar expression can be derived for a redox-active host binding a redox-

silent guest, with the reduction potentials E±

b0 and E±

f0 corresponding to the host with and

without bound guest, respectively.

With Kred > Kox, the redox potential of the host-guest complex will be

anodically shifted, i.e. oxidation will require a more positive potential due to the

stabilization imparted by guest binding. If both Kox and Kred are large (¸ 104M¡1),

diffusion of the empty host species is not a relevant factor, and if Kred=Kox ¸ 103, two

separate voltammetric waves can be observed.21 In this scenario, the magnitude of the

ratio Kred=Kox is defined as the binding enhancement, and can be estimated by the

difference in the half-wave potentials:

Kred

Kox

= e¡F (E±

f0¡E±

b0)=RT

(1.2)

Figure 1.3. Coupled electrochemical and binding equilibria for an electroactive guest G binding with a redox-silent host H.

- 8 -

In contrast, if Kox is small, the free guest is the species being reduced, not the

host-guest complex, and after reduction to the strongly bound G the complexation

process is limited by the diffusion of available empty host. In this case, a single wave

will be observed, with a shift in the observed half wave potential due to guest binding.20

Numerous electrochemical studies of redox-active guests encapsulated by a

redox-silent host have been reported, with ferrocene (Fc) derivatives being the most

popular choices for electroactive guest studies. Osella et al. studied the influence of Fc

encapsulation on its redox properties by β-cyclodextrin (β-CD), a well-characterized

neutral host.22 They observed a single FeIII/II wave in the cyclic voltammogram for the

host-guest complex, anodically shifted from the free Fc+/0 potential due to the high

binding constant, and hence stability, of the neutral host-guest complex. Diminished

currents were also observed upon encapsulation, due to the lower diffusion coefficient of

the encapsulated Fc relative to that for the free guest. The redox behavior of Fc in the

presence of β-CD served as the basis for models describing how host-guest interactions

affect electrochemical behavior, including an insightful analytical discussion by Mendoza

et al.23

In another study, Fujita and coworkers reported cyclic voltammetry measurements

with a PdII-linked supramolecular nanocage capable of encapsulating up to four

molecules of Fc.24 A single wave was observed with the host-guest complex pre-

synthesized with four encapsulated Fc molecules, and the observed 73 mV anodic shift

was attributed to a destabilization of the Fc+ oxidation product by the cationic

environment of the cage. When one equivalent of a water-soluble, neutral ferrocene

- 9 -

derivative was combined with the host, a single anodically shifted wave was observed,

with diminished currents due to the slower diffusion of the host-guest complex.

Multiple identical redox sites

If two or more redox-active sites are present in a molecule, the voltammetric

behavior can be strongly influenced by electronic coupling between the individual sites.

Extended π systems allow efficient long range communication, and π-d overlap for redox

active metal centers makes this interaction even stronger. The ultimate example of metal-

metal communication is the Creutz-Taube ion (Figure 1.4), where a single electron is

completely delocalized between both metals across the pyrazine bridge.25 The formal

oxidation states of both ruthenium atoms is 2.5, and this mixed valence state is denoted

[2,3] (for RuII and RuIII). Two reversible one-electron waves are observed: one for the

[2,2]/[2,3] couple at +0.37 V vs. NHE, and the other for the [2,3]/[3,3] couple at +0.76 V

vs. NHE.26 This separation of 0.39 V is a consequence of the strong communication,

since reduction of one site leads to an increase in electron density at the other site, and the

next oxidation becomes more difficult.

If the individual sites are not linked by a continuous extended π system, and are

separated far enough to make electrostatic interactions negligible, then little to no

communication between redox sites will exist. In such a case, the free energy of

successive reductions differ purely for entropic reasons. Statistical analysis dictates that

Figure 1.4. Creutz-Taube ion, where both ruthenium atoms share the same formal oxidation state of 2.5.

- 10 -

the kth successive formal reduction potential differs from the first according to27

E±

k ¡ E±

1 = ¡2RT

Fln kE±

k ¡ E±

1 = ¡2RT

Fln k

(1.3)

For two noninteracting sites, E±

2 ¡ E±

1 = ¡0:036E±

2 ¡ E±

1 = ¡0:036 V, which is extremely difficult to

measure experimentally. This is discussed in further detail in Chapter 6.

Summary

This dissertation uses several different approaches in order to understand the

guest binding, redox, and molecular transport properties of supramolecular coordination

assemblies. Both synthetic and analytical methods are used to investigate this common

theme, which is of direct interest for future industrial applications.

References

1. Caulder, D. L.; Powers, R. E.; Parac, T. N.; Raymond, K. N., “The Self-Assembly of a Predesigned Tetrahedral M4L6 Supramolecular Cluster.” Angew. Chem. Int. Ed.

1998, 37, 1840-1842.

2. Davis, A. V.; Raymond, K. N., “The Big Squeeze: Guest Exchange in an M4L6 Supramolecular Host.” J. Am. Chem. Soc. 2005, 127, 7912-7919.

3. Caulder, D. L.; Brückner, C.; Powers, R. E.; König, S.; Parac, T. N.; Leary, J. A.; Raymond, K. N., “Design, Formation, and Properties of Tetrahedral M4L4 and M4L6 Supramolecular Clusters.” J. Am. Chem. Soc. 2001, 123, 8923-8938.

4. a) Fiedler, D.; Leung, D. H.; Bergman, R. G.; Raymond, K. N., “Enantioselective Guest Binding and Dynamic Resolution of Cationic Ruthenium Complexes by a Chiral Metal-Ligand Assembly.” J. Am. Chem. Soc. 2004, 126, 3674-3675; b) Fiedler, D.; Pagliero, D.; Brumaghim, J. L.; Bergman, R. G.; Raymond, K. N., “Encapsulation of Cationic Ruthenium Complexes into a Chiral Self-Assembled Cage.” Inorg. Chem. 2004, 43, 846-848; c) Parac, T. N.; Caulder, D. L.; Raymond, K. N., “Selective Encapsulation of Aqueous Cationic Guests into a Supramolecular Tetrahedral M4L6 Anionic Host.” J. Am. Chem. Soc. 1998, 120, 8003-8004; d) Parac, T. N.; Scherer, M.; Raymond, K. N., “Host within a Host: Encapsulation of Crown

- 11 -

Ethers into Ga4L6 Supramolecular Host.” Angew. Chem. Int. Ed. 2000, 39, 1239-1242; e) Tiedemann, B. E. F.; Raymond, K. N., “Dangling Arms: A Tetrahedral Supramolecular Host with Partially Encapsulated Guests.” Angew. Chem. Int. Ed.

2006, 45, 83-86.

5. a) Fiedler, D.; Bergman, R. G.; Raymond, K. N., “Supramolecular catalysis of a unimolecular transformation: Aza-Cope rearrangement within a self-assembled host.” Angew. Chem. Int. Ed. 2004, 43, 6748-6751; b) Fiedler, D.; Leung, D. H.; Bergman, R. G.; Raymond, K. N., “Selective Molecular Recognition, C-H Bond Activation, and Catalysis in Nanoscale Reaction Vessels.” Acc. Chem. Res. 2005, 38, 351-360; c) Leung, D. H.; Fiedler, D.; Bergman, R. G.; Raymond, K. N., “Selective C-H Bond Activation by a Supramolecular Host-Guest Assembly.” Angew. Chem. Int. Ed. 2004, 43, 963-966.

6. Fiedler, D.; van Halbeek, H.; Bergman, R. G.; Raymond, K. N., “Supramolecular Catalysis of Unimolecular Rearrangements: Substrate Scope and Mechanistic Insights.” J. Am. Chem. Soc. 2006, 128, 10240-10252.

7. Leung, D. H.; Bergman, R. G.; Raymond, K. N., “Scope and Mechanism of the C-H Bond Activation Reactivity within a Supramolecular Host by an Iridium Guest: A Stepwise Ion Pair Guest Dissociation Mechanism.” J. Am. Chem. Soc. 2006, 126, 9781-9797.

8. a) Lawrence, D. S.; Jiang, T.; Levett, M., “Self-Assembling Supramolecular Complexes.” Chem. Rev. 1995, 95, 2229-2260; b) Leininger, S.; Olenyuk, B.; Stang, P. J., “Self-Assembly of Discrete Cyclic Nanostructures Mediated by Transition Metals.” Chem. Rev. 2000, 100, 853-908; c) Conn, M. M.; Rebek, J., Jr., “Self-assembling capsules.” Chem. Rev. 1997, 97, 1647-1668; d) Hamilton, T. D.; MacGillvray, L. R., “Enclosed Chiral Environments from Self-Assembled Metal-Organic Polyhedra.” Cryst. Growth Des. 2004, 4, 419-430; e) Manteos-Timoneda, M. A.; Crego-Calama, M.; Reinhoudt, D. N., “Supramolecular chirality of self-assembled systems in solution.” Chem. Soc. Rev. 2004, 33, 363-372; f) Johnson, D. W.; Raymond, K. N., “The Role of Guest Molecules in the Self-Assembly of Metal-Ligand Clusters.” Supramolecular Chem. 2001, 13, 639-659; g) Davis, A. V.; Yeh, R. M.; Raymond, K. N., “Supramolecular Assembly Dynamics.” Proc. Nat. Acad. Sci.

USA 2002, 99, 4793-4796.

9. a) Caulder, D. L.; Raymond, K. N., “Supermolecules by Design.” Acc. Chem. Res.

1999, 32, 975-982; b) Caulder, D. L.; Raymond, K. N., “The Rational Design of High Symmetry Coordination Clusters.” J. Chem. Soc., Dalton Trans. 1999, 8, 1185-2000.

10. a) Yeh, R. M.; Davis, A. V.; Raymond, K. N., “Supramolecular Systems: Self-Assembly.” In Comprehensive Coordination Chemistry II, Fujita, M.; Powell, A.; Creutz, A., Eds. Amsterdam, 2003; Vol. 7, pp 327-355; b) Seeber, G.; Tiedemann, B. E. F.; Raymond, K. N., “Supramolecular Chirality in Coordination Chemistry.” In Top. Curr. Chem., Reinhoudt, D. N.; Crego-Calama, M., Eds. Springer: Berlin, 2006; Vol. 265, pp 147-183.

- 12 -

11. Fujita, M.; Tominaga, M.; Hori, A.; Therrien, B., “Coordination Assemblies from a Pd(II)-Cornered Square Complex.” Acc. Chem. Res. 2005, 38, 369-378.

12. Yoshizawa, M.; Takeyama, Y.; Kusukawa, T.; Fujita, M., “Cavity-directed, highly stereoselective [2+2] photodimerization of olefins within self-assembled coordination cages.” Angew. Chem. Int. Ed. 2002, 41, 1347-1349.

13. a) Slagt, V. F.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.; Reek, J. N. H., “Encapsulation of Transition Metal Catalysts by Ligand-Template Directed Assembly.” J. Am. Chem. Soc. 2004, 126, 1526-1536; b) Slagt, V. F.; Reek, J. N. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M., “Assembly of Encapsulated Transition Metal Catalysts.” Angew. Chem. Int. Ed. 2001, 40, 4271-4274; c) Slagt, V. F.; van Leeuwen, P. W. N. M.; Reek, J. N. H., “Multicomponent Porphyrin Assemblies as Functional Bidentate Phosphite Ligands for Regioselective Rhodium-Catalyzed Hydroformylation.” Angew. Chem. Int. Ed. 2003, 42, 5619-5623.

14. a) Hof, F.; Craig, S. L.; Nuckolls, C.; Rebek, J., Jr., “Molecular Encapsulation.” Angew. Chem. Int. Ed. 2002, 41, 1488-1508; b) Kang, J. M.; Rebek, J., Jr., “Acceleration of a Diels−Alder reaction by a self-assembled molecular capsule.” Nature 1997, 385, 50-52; c) Kang, J. M.; Santamaria, J.; Hilmersson, G.; Rebek, J., Jr., “Diels-Alder Reactions through Reversible Encapsulation.” J. Am. Chem. Soc.

1998, 120, 3650-3656; d) Kang, J. M.; Santamaria, J.; Hilmersson, G.; Rebek, J., Jr., “Self-Assembled Molecular Capsule Catalyzes a Diels-Alder Reaction.” J. Am.

Chem. Soc. 1998, 120, 7389-7390.

15. Beissel, T.; Powers, R. E.; Raymond, K. N., “Symmetry-Based Metal Complex Cluster Formation.” Angew. Chem. Int. Ed. Engl 1996, 35, 1084-1086.

16. Davis, A. V.; Fiedler, D.; Seeber, G.; Zahl, A.; van Eldik, R.; Raymond, K. N., “Guest Exchange Dynamics in an M4L6 Tetrahedral Host.” J. Am. Chem. Soc. 2006, 128, 1324-1333.

17. Pluth, M. D.; Bergman, R. G.; Raymond, K. N., “Acid Catalysis in Basic Solution: A Supramolecular Host Promotes Orthoformate Hydrolysis.” Science 2007, 316, 85-88.

18. Lincoln, S. A.; Merbach, A. E., Adv. Inorg. Chem. 1995, 42, 1-88.

19. Caulder, D. L. The Rational Design of High Symmetry Coordination Clusters. University of California, Berkeley, CA, 1998.

20. Kaifer, A. E.; Gómez-Kaifer, M., Supramolecular Electrochemistry. Wiley-VCH: Weinheim, 1999.

21. Kaifer, A. E.; Mendoza, S., “Redox-switchable Receptors.” In Molecular

Recognition: Receptors for Cationic Guests, Gokel, G. W., Ed. Pergamon: Tarrytown, NY, 1996; Vol. 1, pp 701-732.

- 13 -

22. Osella, D.; Carretta, A.; Nervi, C.; Ravera, M.; Gobetto, R., “Inclusion Complexes of Ferrocenes and β-Cyclodextrins. Critical Appraisal of the Electrochemical Evaluation of Formation Constants.” Organometallics 2000, 19, 2791-2797.

23. Mendoza, S.; Castaño, E.; Meas, Y.; Godínez, L. A.; Kaifer, A. E., “Analysis of the Voltammetric Response of Electroactive Guests in the Presence of Non-Electroactive Hosts at Moderate Concentrations.” Electroanalysis 2004, 16, 1469-1477.

24. Sun, W.-Y.; Kusukawa, T.; Fujita, M., “Electrochemically Driven Clathration/Declathration of Ferrocene and Its Derivatives by a Nanometer-Sized Coordination Cage.” J. Am. Chem. Soc. 2002, 124, 11570-11571.

25. Creutz, C.; Taube, H., “A Direct Approach to Measuring the Franck-Condon Barrier to Electron Transfer between Metal Ions.” J. Am. Chem. Soc. 1969, 91, 3988-3989.

26. Ward, M. D., “Metal-Metal Interactions in Binuclear Complexes Exhibiting Mixed Valency, Molecular Wires and Switches.” Chem. Soc. Rev. 1995, 121-134.

27. Bard, A. J.; Faulkner, L. R., Electrochemical Methods: Fundamentals and

Applications. 2nd ed.; John Wiley & Sons: Hoboken, 2001.

- 14 -

CHAPTER 2

Diffusion of a supramolecular cluster: Ion pairing effects in aqueous solution

Introduction

In studies of guest exchange in the anionic [Ga4L6]12- host supramolecular cluster

it was found that ion pair formation between the cationic guest and the anionic [Ga4L6]12-

host is an important part of the guest exchange process. This was surmised from indirect

kinetic observations, sometimes involving a large number of experiments fit to

mechanistic models.1, 2 Measurements of complex diffusion offer means to explore ion

pairing more directly, since ion pair formation increases the size of the diffusing species,

which in turn decreases the rate of diffusion. Up to the late 1980’s, an entire day was

usually required to obtain a single diffusion coefficient measurement, with the sample

isolated from vibrations and under precise temperature control throughout the course of

the experiment. Thanks to major advancements in NMR methods and instrumentation

over the past two decades, particularly with the inclusion of z-gradient capabilities into

standard probes, diffusion coefficients can be measured in a single 40 minute experiment

with good accuracy (± 2%).

This chapter explores how diffusion NMR measurements can be used to

investigate exterior cation interactions with the [Ga4L6]12- anion by addressing two

important questions. First, will exterior binding of alkylammonium cations (Et4N+ =

tetraethylammonium, Pr4N = tetrapropylammonium) lead to observable changes in rates

of diffusion of the cluster or the exterior R4N+ ion? Second, how do different alkali

- 15 -

cations influence the diffusion of [Et4N⊂Ga4L6]11-? The first question assesses whether

or not diffusion NMR can actually detect ion pair formation, while the second question is

concerned with the basic mass transport characteristics for the cluster. Both questions are

addressed using quantitative measurements of diffusion coefficients in D2O solutions.

Background

Mass transport in solution occurs by diffusion, migration, and convection. In a

quiescent solution with no external electric field gradient, diffusion is the only means for

solutes to move from a region with high concentration to a region with low concentration.

Fick’s laws are differential equations describing the molar flux of a substance and its

concentration as a function of time and position. Following the description of Bard and

Faulkner,3 consider the case of linear (one-dimensional) diffusion. The flux of species i

at a position x and time t is Ji(x; t)Ji(x; t), with units of mol s-1 cm-2, and represents the number

of moles of i that pass point x per second per cm2 of area normal to the axis of diffusion

(Figure 2.1). According to Fick’s first law, the flux is proportional to the concentration

gradient:

Ji(x; t) = ¡Di

µ

@ci(x; t)

@x

¶

Ji(x; t) = ¡Di

µ

@ci(x; t)

@x

¶

(2.1)

x x0

Ji(x0 , t )

ci ci(x, t )

Figure 2.1. Schematic of one dimensional diffusion, where the decreasing concentration profile ci(x,t) results in a net flux of species i in the positive x direction.

- 16 -

where ci(x; t)ci(x; t) is the concentration of species i at position x and time t, and Di is the

diffusion coefficient of species i, with units of cm2 s-1. The negative sign arises because

the direction of flux is towards decreasing concentration. Fick’s second law describes the

change in concentration of i with time:

@ci(x; t)

@t= Di

µ

@2ci(x; t)

@x2

¶

@ci(x; t)

@t= Di

µ

@2ci(x; t)

@x2

¶

(2.2)

For a given initial concentration profile, the diffusion coefficient is the only factor

which can lead to different rates of diffusion for different chemical species. A higher

diffusion coefficient leads to larger flux at any given time, as well as a faster response to

changes in the concentration profile; this is experimentally observed as a faster rate of

diffusion.

The diffusion coefficient is an empirical quantity which depends on the solution

conditions. Due to the complex nature of solvent-solute interactions, there is no single

analytical model which reliably and accurately predicts diffusion coefficients, particularly

for ions in aqueous solutions. With that said, there are relatively simple models that are

very good at qualitatively predicting how the diffusion coefficient will change when a

specific parameter is varied, such as the Stokes-Einstein relation. By modeling diffusion

as a hydrodynamic process where a sphere moves through a viscous medium, Stokes and

Einstein obtained the following expression:4

D =kBT

6¼´rh

D =kBT

6¼´rh (2.3)

where kB is Boltzmann’s constant, T is the absolute temperature, η is the dynamic

viscosity of the solution, and rh is the “hydrodynamic radius” of the diffusing species.

- 17 -

The hydrodynamic radius is in fact defined by the Stokes-Einstein relationship,

and is an empirical quantity. For a large, neutral molecule, the hydrodynamic radius is

very similar to its crystallographic radius; but this is not so for ionic species. Solvent

molecules organize themselves around ions, and these solvation spheres tend to diffuse

with the ion, increasing the hydrodynamic radius. This is especially true for small ions in

aqueous solution: the crystallographic radius of Li+ in LiCl is 0.76 Å,5 but its

hydrodynamic radius in aqueous solution is 3.0 Å.6 The discrepancy between

crystallographic and hydrodynamic radii becomes much less apparent for larger ions,

allowing experimental trends in the diffusion coefficient to be correlated to trends in the

actual size of the diffusing species, measured from crystallographic data.

Diffusion NMR

A number of NMR techniques are available to monitor diffusion and have been

described in a recent review article.7 Pulsed gradient spin-echo (PGSE) NMR methods

have recently attracted increasing interest, since this technique allows diffusion

coefficients to be measured with high signal to noise ratios and have been successfully

used with 7Li and 31P as well as 1H NMR.8 In a PGSE experiment, gradient pulses are

used to spatially label nuclei along the z axis of the NMR tube, and after a delay ∆ to

allow for diffusion, a second gradient pulse is applied before data acquisition. This pulse

sequence is repeated while systematically changing the strength of the gradient. The

attenuation of resonance intensity I with applied gradient strength G is related to the

diffusion coefficient according to:

- 18 -

I = I0 exp

·

¡D°2 G2 ±2

µ

¢¡ ±

3

¶¸

I = I0 exp

·

¡D°2 G2 ±2

µ

¢¡ ±

3

¶¸

(2.4)

where I0 is the signal intensity measured with G = 0, D is the diffusion coefficient, γ is

the gyromagnetic ratio of the resonating nucleus, δ is the gradient pulse width in seconds,

and ∆ is the “diffusion time,” or time between the first and second gradient pulses. The

gradient strength G is typically expressed as G = f ¢GmaxG = f ¢Gmax, where Gmax is the maximum

gradient strength in gauss cm-1, and the fraction f serves as the dependent variable. In

practice, f is expressed as a percentage, and is limited to a range from 2% to 95% to

ensure linear behavior of the gradient amplifier in the probe.9 The observed data can be

fit to Equation 2.4 using nonlinear regression to obtain the value for D (Figure 2.2).

All values of D obtained from this fit are computed using the empirical value

Gmax, which should be accurately calibrated for the particular probe used for

measurements. This was done indirectly by measuring the diffusion decay curves for

dextrose and β-cyclodextrin, whose diffusion coefficients are known to a high degree of

Figure 2.2. Example of a diffusion decay curve with proper choices for δ and ∆. This particular curve corresponds to Li11[Et4N⊂Ga4L6] + 0.1 M LiCl in D2O at 300 K, with δ = 7 ms and ∆ = 90 ms.

0 20 40 60 80 1000

0.2

0.4

0.6

0.8

1

% Gradient Strength

No

rma

lize

d In

teg

ral

Observed Simulated Fit

- 19 -

accuracy. These reported values are Ddex = 6.728 x 10-6 cm2 s-1 and

Dβ-CD = 3.224 cm2 s-1, measured in aqueous solution by monitoring Rayleigh optical

interference fringes.10 The measured value of Gmax = 30.12 ± 0.1 gauss cm-1 was the

same for both probes used for this work, despite their use at different magnetic field

strengths (400 MHz vs. 500 MHz).

Results and Discussion

Ion pairing of alkylammonium cations

In collaboration with Michael D. Pluth, the ion pairing of Et4N+ and Pr4N

+ with

the [Ga4L6]12- host in D2O was investigated using PGSE 1H NMR to measure diffusion

coefficients. When a solution of K12[Ga4L6] is titrated with R4N+ (R = Et, Pr), the

observed diffusion coefficient of the host (DH) decreases with increasing [R4N+] (Figure

2.3) . This indicates that after the first equivalent is encapsulated, the excess R4N+ binds

to the exterior of the host, increasing its overall size and decreasing its diffusion

coefficient. Saturation occurs because the host has a limited number of exterior binding

0 5 10 152.0

2.1

2.2

2.3

DH ,

1

0-6 c

m2 s

-1

# equiv. Et4N

+

0 5 10 15

1.9

2.0

2.1

2.2

DH ,

10

-6 c

m2 s

-1

# equiv. Pr4N

+

a) Et4N+ b) Pr4N

+

Figure 2.3. Ion pairing interactions cause the diffusion coefficient of [Ga4L6]12- in D2O to decrease with

addition of a) Et4NCl with 0.1 M K2CO3 buffer (pD 11) in the host solution, and b) Pr4NBr.

- 20 -

sites available – as many as six sites are suggested from the crystal structure of

K5(Et4N)6[Et4N⊂Fe4L6], with each ligand providing a π-basic naphthalene ring surface

for binding lipophilic cations.11, 12 It is important to note that different solution

conditions were used for the Et4N+ and Pr4N

+ titration experiments in both Figure 2.3 and

Figure 2.4: samples with Et4N+ were prepared with 0.1 M K2CO3 buffer (pD 11.0), while

no buffer was used to prepare Pr4N+ samples.1 Therefore, it is difficult to assess the

relative binding affinities by comparing the data in Figure 2.3.

The weak exterior binding interactions occur in parallel with the much stronger

guest encapsulation equilibrium, causing most ion pair interactions to involve hosts with

an encapsulated guest. Identical diffusion coefficients are obtained from separate fits of

the diffusion decay curves for the signals from the host and encapsulated guest. This

confirms the host and guest diffuse together as a single host-guest complex.

Furthermore, the diffusion coefficient of [Ga4L6]12- measured in the absence of guest is

approximately the same as that measured with one equivalent of R4N+ present. This

implies that [Ga4L6]12- and [R4N⊂Ga4L6]

11- have similar hydrodynamic radii, and similar

solvation shell sizes. We see two possible explanations for this: either size differences

between solvation shells for z = -11 and z = -12 ions are fairly small due to the large size

of the clusters themselves, or the “empty” host contains a deuteron plus solvent (e.g.

D3O+, D5O2

+, etc.) leading to an actual charge of z = -11. These two situations cannot be

distinguished from diffusion data. Kinetic evidence suggests that neutral species are

encapsulated and later protonated inside of the [Ga4L6]12- assembly, and the effective

1 The experiments were carried out separately by two different people, explaining the different conditions.

- 21 -

shift in guest protonation constants has been seen to be no bigger than 4 log units, thus

favoring the first of the two hypotheses.13

Ion pair formation also decreases the observed exterior R4N+ diffusion coefficient,

since the bound alkylammonium cations must diffuse with the much larger [Ga4L6]12-

cluster. Recall that interior and exterior protons are readily distinguished by the large

upfield shifts observed upon encapsulation.11, 14 When a solution of K12[Ga4L6] with two

equivalents of R4N+ (R = Et, Pr) is titrated with KCl the observed diffusion coefficient of

the exterior alkylammonium cation rapidly increases with increasing [KCl] (Figure 2.4).

One R4N+ cation is encapsulated by the cluster, leaving the other R4N

+ cation on the

exterior to form an ion pair. Since the exterior R4N+ cation rapidly equilibrates between

the ion paired and free states on the NMR timescale, its observed diffusion coefficient is

the population average of the two states. In the absence of salt, the exterior R4N+

diffusion is only slightly faster than the host-guest complex, particularly for Et4N+,

consistent with tight ion pairing. The added salt disrupts the ion pairing of R4N+ to the

0.0 0.5 1.0 1.52.5

3.0

3.5

4.0

4.5

D,

10

-6 c

m2 s

-1

[KCl], M

0.0 0.5 1.0 1.5

3.0

3.5

4.0

4.5

D,

10

-6 c

m2 s

-1

[KCl], M

2 eq. Et4N+ 2 eq. Pr4N

+

Figure 2.4. Diffusion coefficient of exterior R4N+ as a function of KCl concentration. Addition of KCl to

K12[Ga4L6] in D2O with a) 2 equivalents of Et4NCl + 0.1 M K2CO3 buffer (pD 11) or b) 2 equivalents of Pr4NBr results in higher diffusion coefficients observed for the exterior alkylammonium cation, because the added salt disrupts ion pairing to the host exterior.

a) b)

- 22 -

host exterior, causing the diffusion coefficient of the exterior R4N+ cation to increase

dramatically.

Much higher diffusion coefficients are observed for the free alkylammonium

cations in the absence of host (Table 2.1). For Et4N+ with 1 M KCl, interaction with

[Et4N⊂Ga4L6]11- reduced the observed diffusion coefficient of the exterior cation to less

than half the value observed in the absence of host, despite the 100-fold excess of KCl;

similar effects were observed with Pr4N+, but to a somewhat lesser extent. Thus the

favorable exterior interactions between R4N+ and [Ga4L6]

12- cannot be solely attributed to

simple coulombic attractions, since K+ will exhibit similar, if not higher, coulombic

attractive forces to the anionic host. If R4N+ binding were caused by coulombic

attraction alone, a large excess of KCl would eliminate any interactions with the anionic

host, and the observed R4N+ diffusion rate would be equal to that observed in the absence

of host. This is clearly not the case, and additional attractive forces must be involved,

such as cation-π binding and/or van der Waals interactions.

Table 2.1. Diffusion coefficients of R4N+ in D2O with 1 M KCl measured in the absence and presence of

[Ga4L6]12- at 27 °C.

D – no [Ga4L6]12- (cm2 s-1) D – with [Ga4L6]

12- (cm2 s-1)

Et4NCl 9.9 x 10-6 4.3 x 10-6

Pr4NBr 7.6 x 10-6 4.1 x 10-6

Alkali cation interactions

In a study by Leung et al. involving reactive monocationic half-sandwich iridium

guests, a stepwise mechanism of guest dissociation through an ion paired intermediate

was described in detail based on kinetic data.2 In an elegant set of experiments, it was

- 23 -

demonstrated that K+ is much more effective than Na+ in disrupting the exterior binding

interaction, suggesting K+ interacts much more strongly with the ion pair adduct than

Na+. Whether or not K+ exhibits a particularly favorable interaction with the host itself,

however, remained unclear. Diffusion measurements may shed some light on whether

the anionic cluster shows different affinities for the various alkali cations.

Five different alkali salts of A11[Et4N⊂Ga4L6] (A+ = Li+, Na+, K+, Rb+, Cs+) were

prepared to investigate how different alkali cations interact with the cluster. The

crystallographic radii of the alkali cations increase with atomic number, i.e. Li+ < Na+ <

K+ < Rb+ < Cs+. However, their ionic mobilities observed in highly dilute aqueous

solutions indicate their hydrodynamic radii decrease with increasing atomic number, i.e.

for rh, Rb+ ≤ Cs+ < K+ < Na+ < Li+.6 The slower diffusion for smaller cations, particularly

for Li+, reflects their large solvation spheres. The solvation shells surrounding small ions

such as Li+ are tightly held to the central ion due to its high charge to surface area ratio.

Since the single positive charge for large ions is spread over a much larger area, Rb+ and

Cs+ exhibit relatively thin shells of loosely associated solvent molecules. Also, different

binding characteristics are expected for the different alkali cations based on hard-soft

acid-base theory, since Li+ is small and hard while Cs+ is large and soft due to

polarizability differences.15

The diffusion coefficients of [Et4N⊂Ga4L6]11- were measured for all five alkali

salts in D2O solutions, and the results are summarized in Figure 2.5. Changing the alkali

cation has a subtle but significant effect on the cluster’s diffusion coefficient. Two

different sets of samples were prepared: one set containing A11[Et4N⊂Ga4L6] (A+ = Li+,

Na+, K+, Rb+, Cs) in pure D2O (“non-salt” samples) and the other set containing the same

- 24 -

compounds but with 0.1 M ACl present (ACl is the chloride salt of the corresponding A+

cation). The data in Figure 2.5 were measured using two different spectrometers, with

much higher 1H sensitivity for the AV-500 compared to the AVB-400 due to different

probe configurations. All measurements with the “non-salt” samples were made with the

AVB-400 spectrometer, whereas the AV-500 instrument was primarily used for

measurements on samples containing 0.1 M ACl.2 The data for the “non-salt” samples –

measured on the AVB-400 spectrometer – suffered low poor signal to noise ratios at high

gradient strengths, suggesting their diffusion coefficients in Figure 2.5 may not be as

accurate as those shown for the samples with 0.1 M ACl.

2 Only one of the three values for the Li+ and Na+ systems was measured on the AV-400 out of all the samples with added salt, and the resulting diffusion coefficients were identical to one of the other two diffusion coefficients calculated from AV-500 to two decimal places (i.e. within 0.01 x 10-6 cm2 s-1).

Figure 2.5. Diffusion coefficients at 300 K measured for A11[Et4N⊂Ga4L6] (A+ = Li+, Na+, K+, Rb+, Cs+)

in pure D2O (blue bars) and in D2O with 0.1 M ACl present (purple bars). When average values of repeated experiments are reported, the number of experiments averaged together is listed above the corresponding bar, with the measured standard deviations used for the error bar. Otherwise, error bars for single measurements are calculated from the standard deviation of the measured gradient strength.

2.00

2.20

2.40

2.60

2.80

Li Na K Rb Cs

DH

G, 1

0-6

cm

2 s

-1

No added salt

With 0.1 M ACl

3

3

3

2

- 25 -

Comparing the measurements for the samples with 0.1 M ACl, the diffusion

coefficients for the Na+, K+ and Cs+ systems are all equal within experimental error, with

DK = 2.37(3) x 10-6 cm2 s-1. The Li+ system shows the slowest diffusion, with

DLi = 2.24(3) x 10-6 cm2 s-1, which is about 5% lower than the Na+, K+ and Cs+ values.

The Rb+ system exhibits the fastest diffusion, with DRb about 8% greater than the Na+, K+

and Cs+ diffusion coefficients, and about 14% higher than that for Li+. The non-salt data

also show the host-guest complex diffusion is fastest with Rb+ counterions, and the

slowest with Li+ counter ions.

Clearly, the alkali counterions must play an active role during the cluster diffusion

process. If A+ formed ion pairs with the [Et4N⊂Ga4L6]11- anion, and if different alkali

cations bind with different affinities to cause the diffusion coefficients to differ, then a

significant cation concentration dependence would be observed. However, the presence

of 0.1 M ACl – twenty times higher than the 5 mM cluster concentration – has very little

effect on the diffusion coefficient for Li+, Na+, and Cs+, implying their ion pairing can be

neglected (or the binding sites are already saturated). For Rb+, addition of salt leads to a

4% increase in the diffusion coefficient, and for K+ a 5% increase is observed with KCl.

These small changes may be artifacts from the poor signal to noise of the no-salt data

measured on the AVB-400.

The slow diffusion observed with Li+ counterions is consistent with its much

lower ionic mobility.6 To maintain charge neutrality, counterions must co-diffuse with

the solvated host-guest anion, and the observed diffusion coefficient of [Et4N⊂Ga4L6]11-

will depend on the alkali cation’s mobility. For infinitely dilute solutions, the diffusion

coefficient can be estimated from the individual ionic mobilities at infinite dilution:

- 26 -

D± =2u±+u

±

¡RT

zF (u±+ + u±¡)

D± =2u±+u

±

¡RT

zF (u±+ + u±¡)

(2.5)

where u±+u±

+ and u±¡u±¡

are the respective mobilities (cm2 V-1 s-1) of the cation and anion at

infinite dilution, z is the charge number (z = 11 for [Et4N⊂Ga4L6]11-), and F is the

Faraday constant.6, 16 At infinite dilution, the cluster’s mobility u±HGu±HG is unaffected by the

nature of the cation, and we can calculate the ratio of diffusion coefficients for two

different cations with mobilites u±1u±

1 and u±2u±

2:

D±

2

D±

1

=

µ

u±2u±1

¶

u±1 + u±HG

u±2 + u±HG

D±

2

D±

1

=

µ

u±2u±1

¶

u±1 + u±HG

u±2 + u±HG (2.6)

Due to its large size and charge, the cluster’s mobility u±HGu±HG should be less than

that for any alkali cation. If u±2 > u±1 > u±HGu±2 > u±1 > u±HG, then D±

2 > D±

1D±

2 > D±

1 according to Equation 2.6.

Since Li+ has the lowest mobility of any alkali cation, Equation 2.6 predicts the diffusion

of [Et4N⊂Ga4L6]11- will be slowest for the lithium salt, consistent with the experimental

observations in Figure 2.5.

Rb+ has the highest ionic mobility in aqueous solution, about 6% faster than K+,

and much higher than Na+ or Li+, so this may explain why the host diffuses fastest with

Rb+ cations. It remains unclear why the cluster’s diffusion coefficient observed with Rb+

is relatively high compared to Cs. The mobilities of Rb+ and Cs+ are very similar, so

Equation 3.6 cannot explain why the diffusion coefficient with Rb+ is about 8% larger.

Ion pairing effects are unlikely – addition of KCl or RbCl to leads to a 5% and 4%

increase in the observed diffusion coefficients, respectively, with essentially no change

for Cs+. For Et4N+ and Pr4N

+, exterior binding led to slower diffusion, since the

relatively large cations increased the hydrodynamic radius of the ion pair. Although the

radii of K+ and Rb+ are much smaller than the alkylammonium ions, ion pairing still

- 27 -

cannot account for the faster diffusion observed in the presence of salt. The

crystallographic radii for the two cations are r(K+) = 1.38 Å and r(Rb+) = 1.52 Å,5 so if

the two cations did bind to the anionic cluster, the size differences predict faster diffusion

for K+, but the larger Rb+ shows faster diffusion, inconsistent with ion pair formation.

The solvated radii of the two ions are approximately equal, so binding of the solvated

cations cannot account for the difference either.3

The diffusion coefficients observed with Na+ and K+ are about the same,

suggesting the anionic cluster does not show any particular preference for binding one

cation over the other. In contrast, Leung et al. observed K+ was much more effective

than Na+ at disrupting the binding of an iridium half-sandwich cation to the exterior of

[Ga4L6]12-. While the reason for this cation dependence remains unclear, the diffusion

data suggest that ion pairing to the host itself is not significant for alkali cations. Perhaps

the stronger solvation of Na+ compared to K+ due to its smaller radius is an important

factor. The crystallographic radii for the chloride salts are rNa = 1.02 Å for Na+ and

rK = 1.38 Å for K+, and from the Born model K+ desolvation requires about 35% less

energy than Na+ in the same solution. Future studies are needed to test this hypothesis.

Summary

Diffusion NMR can be a very powerful tool to observe relatively weak

interactions with ease. Ion pairing interactions can be observed by monitoring changes in

the diffusion coefficients for the host and the guest, rather than relying on detailed kinetic

studies involving complex reaction mechanisms to indirectly measure exterior binding

3 Ternary species with bridging ion-paired cations such as HG11-⋅⋅⋅A+⋅⋅⋅HG11- will be strongly disfavored by the large coulombic repulsive forces between the two -11 anions.

- 28 -

equilibria. The exterior binding of Et4N+ and Pr4N

+, originally deduced from the Et4N+

concentration dependence observed for the NMR chemical shifts, has been firmly

established by diffusion NMR methods as an important secondary binding interaction.

Furthermore, diffusion measurements provided valuable insight on how different alkali

cations interact with the anionic host. Since diffusion coefficients are related to

molecular size, diffusion NMR can be a valuable analytical method to study

supramolecular systems.

Experimental

General considerations

Unless noted otherwise, reagents were obtained from commercial suppliers and

used without further purification. Standard Schlenk techniques were used for reactions

carried out under argon, and a glove box continuously purged with nitrogen was used to

manipulate and store air-sensitive solids. When necessary, solvents were degassed by at

least six pump/fill cycles while vigorously stirring, using argon for the fill step.

Methanolic alkali hydroxide stock solutions were standardized by Michael D. Pluth by

performing a titration with aqueous HCl using a colored pH indicator, taking the average

value from three titrations for each solution. Tetraethylammonium chloride (Et4NCl) was

recrystallized from absolute ethanol/ether and dried in vacuo over P2O5 at room

temperature for 12 hours, then dried in vacuo over molecular sieves at 60 °C for 18 hours

and stored under nitrogen. H4L (H4L = 1,5-bis(2,3-dihydroxybenzamido)naphthalene)

and K12[Ga4L6] were synthesized according to literature procedures.11, 17 Routine mass

spectrometry and elemental analysis was performed by the Mass Spectrometry

- 29 -

Laboratory and Microanalysis Facility in the College of Chemistry at the University of

California, Berkeley.

Synthetic procedures

Cs11[Et4N ⊂⊂⊂⊂ Ga4L6]·(Me2CO).12 In a 250 mL round-bottom Schlenk flask, 200 mg

(0.465 mmol) of H4L and 113.8 mg (0.310 mmol) of Ga(acac)3 (Aldrich) were combined,

to which 100 mL of methanol was added. To this opaque white mixture was added

0.79 mL (0.077 mmol) of methanolic Et4NCl (97 mM) via syringe, the reaction mixture

was degassed via several pump/fill cycles, and 4.2 mL (0.95 mmol) of methanolic CsOH

(0.226 M) was added via syringe, causing the color to change from white to yellow. The

reaction mixture was degassed again, and stirred under argon at room temperature for

18 hours. The pale yellow solid suspended in the methanolic reaction mixture was

filtered under a stream of nitrogen, washed with acetone (3 x 10 mL) and petroleum ether

(2 x 15 mL), and dried in vacuo overnight for 12 hours to yield 300 mg (86%) of yellow

powder. The 1H NMR spectrum was similar to that published for K11[Et4N⊂Ga4L6]. The

number of co-precipitated acetone molecules was determined from 1H NMR integration.

Rb11[Et4N⊂⊂⊂⊂Ga4L6]·(Me2CO)(Et2O).12 A procedure similar to the Cs+ salt synthesis was

used, with 200 mg (0.465 mmol) of H4L, 113.8 mg (0.310 mmol) of Ga(acac)3, 0.75 mL

(0.073 mmol) of methanolic Et4NCl (97 mM), and 2.4 mL (0.95 mmol) of methanolic

RbOH (0.397 M). After stirring at room temperature for 4 days, the reaction mixture was

cloudy, indicating some product was suspended. After reducing the volume to

approximately 50 mL using a vacuum pump to remove solvent, the reaction mixture was

- 30 -

filtered, leaving a small amount of yellow solid on the frit. Additional solvent was

removed from the clear yellow filtrate to reduce its volume to about 30 mL, generating a

small amount of yellow precipitate. Addition of 30 mL of degassed acetone caused this

solid to re-dissolve, affording a transparent yellow solution. Addition of 30 mL of Et2O

to this methanol/acetone solution led to precipitate formation after 10 min., and 30 L of

additional Et2O was added and allowed to equilibrate for an additional 15 min. The solid

was collected on the same frit used for the first filtration, washed with acetone/ether

(1 x 10 mL; some product dissolved in this washing), then pure Et2O (2 x 15 mL). The

solid was dried on the frit under a stream of nitrogen, yielding 220 mg (70%) of pale

yellow powder. The 1H NMR spectrum was similar to that published for

K11[Et4N⊂Ga4L6]. The number of co-precipitated solvent molecules was determined

from 1H NMR integration.

Li11[Et4N⊂⊂⊂⊂Ga4L6]·(Me2CO)5.12 A procedure similar to the Cs+ salt synthesis was used,

with 200 mg (0.465 mmol) of H4L, 113.8 mg (0.310 mmol) of Ga(acac)3, 0.75 mL

(0.073 mmol) of methanolic Et4NCl (97 mM), and 2.8 mL (0.95 mmol) of methanolic

LiOH (0.343 M). After stirring at room temperature for 7 days, solvent was removed

with a vacuum pump to reduce the reaction mixture volume to 60 mL. The slightly

turbid mixture was passed through 0.2 µm nylon syringe filter disks, and the volume of

the clear yellow filtrate was reduced to approximately 10 mL using a vacuum pump to

remove solvent. Degassed acetone (150 mL total) was gradually added via cannula while

slowly stirring to form a precipitate, which was collected on a frit under a stream of

nitrogen, washed with acetone (2 x 10 mL) and petroleum ether (3 x 20 mL), yielding

- 31 -

160 mg (62%) of yellow-green powder. This solid is hygroscopic, steadily gaining mass

while exposed to air. The 1H NMR spectrum was similar to that published for

K11[Et4N⊂Ga4L6]. The number of co-precipitated acetone molecules was determined

from 1H NMR integration.

Na11[Et4N⊂⊂⊂⊂Ga4L6]·(Me2CO)3.12 A procedure similar to the Li+ salt synthesis was used,

with 1.45 mL (0.93 mmol) of methanolic NaOH (0.640 M) as the base, stirring at room

temperature overnight. Yield: 180 mg (68%) of yellow powder. The 1H NMR spectrum

was similar to that published for K11[Et4N⊂Ga4L6]. The number of co-precipitated

acetones was determined from 1H NMR integration.

K11[Et4N⊂⊂⊂⊂Ga4L6]·(Me2CO)2.12 A procedure similar to the Li+ salt synthesis was used,

with 1.9 mL (0.95 mmol) of methanolic KOH (Aldrich, 0.5 M) as the base, stirring at

room temperature overnight. Yield: 100 mg (37%) of yellow powder. The 1H NMR

spectrum was similar to that published for K11[Et4N⊂Ga4L6]. The number of co-

precipitated acetone molecules was determined from 1H NMR integration.

Diffusion NMR experiments

PGSE diffusion 1H NMR measurements were performed on either a Bruker

AVB-400 spectrometer with a z-gradient broadband coil or a Bruker AV-500

spectrometer with a TBI-P probe with a z-gradient coil, using the ledbpgp2s pulse

program with diffusion time ∆ = 90-100 ms, bipolar gradient pulse duration δ = 7 ms

(2 x 3.5 ms), 8 scans per experiment, pre-pulse delay of 6 sec., and a linear gradient

- 32 -

strength ramp of 32 increments from 2% to 95%.9 The 90° RF pulse width was

calibrated for each sample, particularly with different ionic strengths. A constant

temperature of 300 K was maintained using an automated temperature controller,

allowing the sample temperature to equilibrate in the probe for at least 10 minutes before

starting each measurement. The integrated areas (normalized by the low-G integral

value) were averaged for resonances on the same molecule, and fit to the expected

exponential decay equation18 using nonlinear regression to evaluate the diffusion

coefficient. The regression weighting scheme was based on the observed standard

deviation for the averaged resonances. The probe gradient power was calibrated from a

fit of the diffusion decay curve of dextrose and β-cyclodextrin in D2O using literature

values for the diffusion coefficients.10

Sample preparation: Et4N+ titration. In a thin-walled NMR tube, 25 mg (7 µmol) of

K12[Ga4L6] was dissolved in 500 µL of D2O buffered to pD = 11.0 with 0.1 M K2CO3,

and a sealed capillary with ferrocene in CDCl3 was inserted in the tube for an internal

diffusion standard. For each titration point, a stock solution of 1.0 M Et4NCl in D2O was

added to the NMR tube in 7 µL increments, and mixed with the host solution by inverting

the tube several times before replacing the tube in the probe for the next measurement.

Actual number of equivalents of Et4N+ was determined from 1H NMR integrals.

Sample preparation: Pr4N+ titration. In a glass vial, 14 mg (4 µmol) of K12[Ga4L6]

was dissolved in 0.7 mL of D2O, filtered through a glass wool plug, and the filtrate was

collected in a thin-walled NMR tube. For each titration point, a stock solution of 0.2 M

- 33 -

Pr4NBr in D2O was added to the NMR tube in 20 µL increments in a similar manner

described for the Et4N+ titration.

Sample preparation: KCl titration with Et4N+. Using six thin-wall NMR tubes, 10 mg

(2.8 µmol) of K12[Ga4L6] was combined with different amounts of KCl in each tube

(0 mg, 2.16 mg, 3.44 mg, 9.15 mg, 18.06 mg, 33.75 mg) and each dissolved in 400 µL of

D2O buffered to pD = 11.0 with 0.1 M K2CO3. To each sample was added 100 µL of a

stock solution with 60 mM Et4NCl in the same D2O buffer for a final volume of 500 µL

in each tube.

Sample preparation: KCl titration with Pr4N+. Two separate stock solutions were

prepared in D2O: one with 62.1 mg (16.9 µmol) of K12[Ga4L6] and 9.0 mg (34 µmol) of

Pr4NBr dissolved in D2O to 2.00 mL for a stock solution with 8.5 mM host and 19 mM

guest; the other with 3.0 M KCl. Using seven medium-walled NMR tubes, 200 µL of

host/guest stock solution was added to each tube, and different amounts of KCl stock

solution was added (0 µL, 13.3 µL, 33.3 µL, 66.7 µL, 100 µL, 133 µL, 200 µL) and the

appropriate amount of pure D2O was added for a final volume of 400 µL per tube. Each

sample was degassed in its tube via three freeze-pump-thaw cycles, flame sealed under

vacuum, and allowed to equilibrate overnight before measurements.

References

1. a) Davis, A. V.; Fiedler, D.; Seeber, G.; Zahl, A.; van Eldik, R.; Raymond, K. N., “Guest Exchange Dynamics in an M4L6 Tetrahedral Host.” J. Am. Chem. Soc. 2006, 128, 1324-1333; b) Fiedler, D.; Bergman, R. G.; Raymond, K. N., “Supramolecular catalysis of a unimolecular transformation: Aza-Cope rearrangement within a self-

- 34 -

assembled host.” Angew. Chem. Int. Ed. 2004, 43, 6748-6751; c) Fiedler, D.; van Halbeek, H.; Bergman, R. G.; Raymond, K. N., “Supramolecular Catalysis of Unimolecular Rearrangements: Substrate Scope and Mechanistic Insights.” J. Am.

Chem. Soc. 2006, 128, 10240-10252.

2. Leung, D. H.; Bergman, R. G.; Raymond, K. N., “Scope and Mechanism of the C-H Bond Activation Reactivity within a Supramolecular Host by an Iridium Guest: A Stepwise Ion Pair Guest Dissociation Mechanism.” J. Am. Chem. Soc. 2006, 126, 9781-9797.

3. Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and

Applications, 2nd ed.; John Wiley & Sons: Hoboken, 2001; p 148-150.

4. Welty, J. R.; Wicks, C. E.; Wilson, R. E.; Rorrer, G. Fundamentals of Momentum,

Heat, and Mass Transfer, 4th ed.; John Wiley & Sons: New York, 2001.

5. Lide, D. R., Handbook of Chemistry and Physics. 81st ed.; CRC Press: Boca Raton, 2000; p 12.14-15.

6. Stern, K. H.; Amis, E. S., “Ionic Size.” Chem. Rev. 1959, 59, 1-64.

7. Johnson, C. S., Jr., “Diffusion ordered nuclear magnetic resonance spectroscopy: principles and applications.” Prog. NMR. Spectrosc. 1999, 34, 203-256.

8. a) Fernández, I.; Martínez-Viviente, E.; Breher, F.; Pregosin, P. S., “7Li, 31P, and 1H Pulsed Gradient Spin-Echo (PGSE) Diffusion NMR Spectroscopy and Ion Pairing: On the Temperature Dependence of the Ion Pairing in Li(CPh3), Fluorenyllithium, and Li[N(SiMe3)2] amongst Other Salts.” Chem. Eur. J. 2005, 11, 1495-1506; b) Pregosin, P. S.; Martínez-Viviente, E.; Anil Kumar, P. G., “Diffusion and NOE spectroscopy. Applications to problems related to coordination chemistry and homogeneous catalysis.” Dalton Trans. 2003, 4007-4014.

9. Kerssebaum, R., DOSY and Diffusion by NMR. In User Guide for XWinNMR 3.1/3.5

Version 1.03, Bruker BioSpin GmbH: Rheinstetten, Germany, 2002.

10. Longsworth, L. G., “Temperature Dependence of Diffusion in Aqueous Solutions.” J.

Phys. Chem 1954, 58, 770-773.

11. Caulder, D. L.; Powers, R. E.; Parac, T. N.; Raymond, K. N., “The Self-Assembly of a Predesigned Tetrahedral M4L6 Supramolecular Cluster.” Angew. Chem. Int. Ed.

1998, 37, 1840-1842.

12. Caulder, D. L.; Brückner, C.; Powers, R. E.; König, S.; Parac, T. N.; Leary, J. A.; Raymond, K. N., “Design, Formation, and Properties of Tetrahedral M4L4 and M4L6 Supramolecular Clusters.” J. Am. Chem. Soc. 2001, 123, 8923-8938.

- 35 -

13. Pluth, M. D.; Bergman, R. G.; Raymond, K. N., “Acid Catalysis in Basic Solution: A Supramolecular Host Promotes Orthoformate Hydrolysis.” Science 2007, 316, 85-88.

14. Chatterjee, A. K.; Choi, T.-L.; Sanders, D. P.; Grubbs, R. H., “Catalyzed Alkene Metathesis.” J. Am. Chem. Soc. 2003, 125, 11360-11370.

15. Ahrland, S., Factors Contributing to (b)-Behaviour in Acceptors. In Structure and

Bonding, Jørgensen, C. K.; Neilands, J. B.; Nyhoum, R. S.; Reinen, D.; Williams, R. J. P., Eds. Springer-Verlag: Berlin, 1966; Vol. 1, pp 207-220.

16. Euken, A. Lehrbuch der chemischen Physik, Akademische Verlagsgesellschaft: Leipzig, 1949; Vol. 2, p 798.

17. Parac, T. N.; Caulder, D. L.; Raymond, K. N., “Selective Encapsulation of Aqueous Cationic Guests into a Supramolecular Tetrahedral M4L6 Anionic Host.” J. Am.

Chem. Soc. 1998, 120, 8003-8004.

18. Stejskal, E. O.; Tanner, J. E., “Spin Diffusion Measurements: Spin Echoes in the Presence of a Time-Dependent Field Gradient.” J. Chem. Phys. 1965, 42, 288-292.

- 36 -

CHAPTER 3

Thermodynamics of Guest Binding: Calorimetry Experiments

Introduction

The fact that guest encapsulation by the [Ga4L6]12- host can be highly favorable is

readily apparent from 1H NMR spectroscopy. When K12[Ga4L6] is combined with one

equivalent of Et4N+ in D2O, almost quantitative encapsulation is observed from the

integrated areas of the upfield-shifted CH2 and CH3 interior resonances, with very little

signal remaining for the exterior Et4N+ protons. The binding equilibrium constant for this

reaction, measured by 1H NMR integral ratios, is 1.96 x 104 M-1 in D2O at 25 °C.1 What

is the driving force behind this strong binding affinity? Previous van't Hoff studies using

NMR data indicated that guest encapsulation was an endothermic process, driven by a

large increase in entropy due to solvent release.2 However, the van't Hoff plot used for

that study assumed the enthalpy and entropy of binding were both temperature

independent, which in general is not the case. By neglecting the ∆Cp contribution –

which can be quite relevant for host-guest binding interactions – the values of ∆H° from

a van’t Hoff plot can be incorrect.3 Isothermal titration calorimetry (ITC) offers a direct

method to measure the enthalpy of a reaction, ∆H°, without relying on any assumptions

about temperature dependence.4, 5 ITC experiments were carried out in collaboration

with Prof. Giuseppe Arena and Dr. Carmelo Sgarlata at the Università di Catania in

Catania, Italy. This chapter represents the initiation of a long-term collaboration between

the Raymond and Arena groups for thermodynamic studies.

- 37 -

Background: Guest binding interactions

The host-guest chemistry of the M4L6 tetrahedral cluster has been studied in great

detail since its preparation was first reported in 1998 by Caulder et al.1 When prepared

with trivalent metal ions such as gallium(III), the anionic M4L6 cluster has an overall -12

charge, making K12[Ga4L6] very water soluble. The interior of the tetrahedron is

surrounded by six naphthalene rings from the ligands, forming a 300–500 Å3

hydrophobic cavity that preferentially binds lipophilic monocations as guests.2 The X-

ray crystal structure of K5(Et4N)7[Fe4L6] shows one Et4N+ cation occupying the interior

cavity, and six Et4N+ cations in close contact with the aromatic rings of the ligands.6 As

discussed in Chapter 2, these exterior Et4N+ cations remain closely associated with the

cluster in aqueous solution, demonstrating that the M4L6 cluster has a second set of guest

binding sites on the exterior.

Cross peaks in the 1H NOESY spectrum have been observed between the exterior

guest resonances and the catechol and naphthalene signals from the host, in contrast to

the interior guest signals which only show cross peaks with naphthalene proton

resonances.6-8 For Et4N+, the intensities of the catechol and naphthalene cross peaks are

similar, consistent with the X-ray crystal structure showing Et4N+ associated with both

the catechol and naphthalene rings.6 For the half-sandwich iridium complex 1, cross

peaks between the Cp* proton signals and the

host catechol protons are much more intense

than those with naphthalene.7 The anionic

charge of the [Ga4L6]12- cluster is

concentrated at the metal catecholate vertices

- 38 -

of the assembly, suggesting electrostatic attraction to monocationic 1 is the predominant

interaction, which causes strong ion pairing to the anionic host. In contrast, ammonium

cations such as Me4N+ and Et4N

+ can also bind to the aromatic naphthalene faces of the

ligand scaffold via cation-π interactions. Encapsulated guests can only interact with the

naphthalene ring walls surrounding the cavity, so cation-π and π-π interactions are very

favorable for interior binding. Clearly, interactions with the ligand are extremely

important for favorable guest binding interactions. To illustrate the charge distribution of

a coordinated ligand in the [Ga4L6]12- host, the potential density surface of the

hypothetical model complex [K2L]2- was generated from a quantum mechanical

computational model (Figure 3.1).

Exterior binding and guest encapsulation are not independent processes. In fact,

recent kinetic studies have demonstrated that guest encapsulation and dissociation

Figure 3.1. Potential density surface of the [K2L]2- complex generated from the Hartree-Fock quantum mechanical calculation, as a hypothetical model to approximate the electronics of the coordinated ligands in the [Ga4L6]

12- cluster. Negatively charged regions are shown in red, and positive charges are colored blue.

- 39 -

proceed in a stepwise process via an exterior ion-pair intermediate.7, 9, 10 What are the

factors which determine how favorable interior and exterior guest binding will be?

Noncovalent interactions such as coulombic attraction, cation-π interactions, π-π

interactions, and van der Waals forces are all enthalpically favorable in the gas phase, but

entropy is lost when the guest binds to the host. However, due to the large charge of the

[Ga4L6]12- host, solvation effects can be extremely powerful, particularly in aqueous

solution. When a monocationic guest is encapsulated, the overall charge of the host

decreases from -12 to -11, and the solvated cation must be completely stripped of solvent

to enter the interior cavity. Both desolvation processes involve an enormous enthalpic

cost, but the liberation of water molecules from the highly organized solvation shells

leads to a large increase in entropy. Thus, guest encapsulation may be entropically

driven, even if it is an endothermic process.

Solvation of soft, polarizable donors and acceptors is usually less powerful, so

binding is typically enthalpically favorable but entropically disfavored.11 Cation-π

interactions involve soft, polarizable acceptors, and exothermic binding interactions are

often observed even in aqueous solution.12 Thus, exterior binding of ammonium cations

such as Me4N+ and Et4N

+ to the naphthalene ring faces of the host may be enthalpically

driven, but entropically disfavored.

Born solvation theory

According to the Born solvation model, for a spherical ion of charge zi with a

radius ri the Gibbs free energy of solvation relative to the gas phase is given by:13

- 40 -

¢Gsolv = ¡µ

Nae2

8¼²0

¶

z2i

ri

µ

1¡ 1

²

¶

= ¡¯µ

z2i

ri

¶ µ

1¡ 1

²

¶

where ¯ ´ Nae2

8¼²0

¢Gsolv = ¡µ

Nae2

8¼²0

¶

z2i

ri

µ

1¡ 1

²

¶

= ¡¯µ

z2i

ri

¶ µ

1¡ 1

²

¶

where ¯ ´ Nae2

8¼²0

(3.1)

where ² is the dielectric constant of the solvent (modeled as a continuum), Na is

Avogadro’s number, e is the charge of an electron, and ²0 is the permittivity of vacuum.

Since dG = V dP ¡ SdTdG = V dP ¡ SdT , S = ¡µ

@G

@T

¶

P

S = ¡µ

@G

@T

¶

P

, and thus the entropy of solvation is:

¢Ssolv = ¯z2

i

ri

µ

1

²2d²

dT

¶

¢Ssolv = ¯z2

i

ri

µ

1

²2d²

dT

¶

(3.2)

Furthermore, since ¢G = ¢H ¡ T¢S¢G = ¢H ¡ T¢S, the enthalpy of solvation is:

¢Hsolv = ¡¯z2

i

ri

µ

1¡ ²¡1 ¡ T²¡2 d²

dT

¶

¢Hsolv = ¡¯z2

i

ri

µ

1¡ ²¡1 ¡ T²¡2 d²

dT

¶

(3.3)

In H2O, = 78.36 and at 25 °C,14 and the constant

β = 6.95 x 10-5 J·m mol-1, so Equations 3.1 – 3.3 can be expressed numerically for

convenience:

where 1 e.u. ≡ 1 cal mol-1 K-1, and ri has units of Å in the above formulae.

The Born solvation free energies, entropies, and enthalpies for the ions considered

in this chapter are listed in Table 3.1 for comparison purposes. Note that the host

solvation enthalpy cost for guest encapsulation is about 400 kcal mol-1 due to the

reduction in overall charge from -12 to -11. Furthermore, note the enthalpy cost for

desolvation of a dicationic species is much larger than for the monocation due to the z2

- 41 -

dependence of Equation 3.3, explaining why a dication has never been observed inside

the [Ga4L6]12- host cavity.2

Table 3.1. Born thermodynamic parameters for solvation of various ions in H2O at 25 °C calculated using Equations 3.1 – 3.3. In addition to the alkylammonium monocations studied in this chapter, a hypothetical dication with the same radius as Et4N

+ is listed for comparison.

Ion ri (Å) zi ¢Gsolv¢Gsolv

(kcal mol-1) ¢Hsolv¢Hsolv

(kcal mol-1) ¢Ssolv¢Ssolv (e.u.)c

Me4N+ 2.93a 1 -55.9 -56.9 -3.3

Et4N+ 3.48a 1 -47.1 -47.9 -2.8

Pr4N+ 3.90a 1 -42.0 -42.7 -2.5

Bu4N+ 4.25a 1 -38.6 -39.2 -2.3

Hypothetical dication 3.48 2 -188 -192 -11.1

[Ga4L6]12- 9.5b -12 -2483 -2527 -146

[R4N⊂Ga4L6]11- 9.5b -11 -2087 -2123 -122

a) Radius calculated from volume of molecular model (semi-empirical minimization). b) Radius estimated from X-ray crystal structure of K5(Et4N)7[Fe4L6]. c) 1 e.u. ≡ 1 cal mol-1 K-1.

Results and Discussion

When K12[Ga4L6] is dissolved in H2O with 0.1 M KCl and titrated with R4NCl

(R = Me, Et, Pr, Bu), the observed heats are very different depending on the size of the

guest being added (Figure 3.2). Exothermic reactions are observed for R = Me, Et, and

Pr, but Bu4N+ switches from exothermic to endothermic as the titrant concentration is

increased. (The addition range for Bu4N+ was limited to 7 equivalents due to

precipitation problems at higher ratios). Values for ∆H° and log β were obtained for

Me4N+, Et4N

+, and Pr4N+ by fitting the data with Hyp∆H,15 and ∆S° was calculated from

the fitted quantities (Table 3.2). Simple 1:1 (guest:host) binding models were used for

Me4N+ and Et4N

+, but an additional 2:1 interaction could be elucidated with the Pr4N+

- 42 -

system. Unfortunately, the Bu4N+ titration curve could not be fit at all to any binding

model, probably because the host does not encapsulate the large cation. It is thought that

Bu4N+ may bind to the exterior of the host,10 but the observed ITC data cannot confirm

nor deny this hypothesis. Since this remains unresolved, Bu4N+ will not be discussed

further in this chapter.

Table 3.2. Thermodynamic parameters of complex formation of Me4N+, Et4N

+, and Pr4N+ with the

[Ga4L6]12- host calculated from isothermal titration calorimetry data measured at 25 °C in 0.1 M KCl.

Titrant Product Species log β ∆H° (kcal mol-1) ∆S° (e.u.)a

Me4NCl (Me4N)1Host 1.1(3) -6.02(5) -15 ± 1

Et4NCl (Et4N)1Host 1.8(1) -8.82(1) -21.3(5)

(Pr4N)1Host 2.0(3) -5.97(5) -11 ± 1 Pr4NCl

(Pr4N)2Host 4.5(4) -0.7(1) 18 ± 2

a) 1 e.u. = 1 cal mol-1 K-1.

The data in Table 3.2 clearly demonstrate that the interactions of R4N+ cations

with the [Ga4L6]12- cluster are enthalpically favored (R = Me, Et, Pr). The values of log β

for Me4N+ and Pr4N

+ are similar to the equilibrium constants for guest encapsulation by

0 5 10 15 20 25

-5

0

5

10

15

20

To

tal H

ea

t, m

J

[G]/[H]

Me4N

+

Et4N

+

Pr4N

+

Bu4N

+

0 5 10 15 20 25

-2

-1

0

1

2

Incre

me

nta

l H

ea

t, k

J m

ol-1

titra

nt

[G]/[H]

Me4N

+

Et4N

+

Pr4N

+

Bu4N

+

Figure 3.2. ITC reaction heats observed when 1 mM K12[Ga4L6] is titrated with R4NCl (R = Me, Et, Pr, Bu) in aqueous 0.1 M KCl at 25 °C. a) Incremental heat released per mole of titrant injected. b) Cumulative heat curves observed for the four different cations interacting with the host. All data were corrected for dilution effects. Note that the vertical axes correspond to the heat released by the system, and therefore an exothermic reaction is observed as a positive heat.

- 43 -

[Ga4L6]12- observed via 1H NMR in D2O

4 (for Pr4N+, Kb = 1.1 x 102 M-1, and for Me4N

+

Kb << 102 M-1),1 but the value of log β for Et4N+ listed in Table 3.2 is two orders of

magnitude lower than its value of log Kb = 4.3 measured via NMR!1 The high affinity of

Et4N+ for the [Ga4L6]

12- interior cavity has been verified by literally hundreds of NMR

experiments independently performed by several different researchers over the last ten

years, so its validity is not in question. However, ten separate ITC experiments have

been performed that all indicate log β = 1.8 for the interaction of Et4N+ with the host in

0.1 M KCl, ruling out artifacts from experimental errors. If additional equilibria with

different stoichiometries are included in the model, such as the 2:1 (or even 7:1) species

expected for ion pairing to the host-guest complex, the fit does not converge.5 Thus, the

equilibrium constant for Et4N+ listed in Table 3.2 may be primarily describing exterior

binding interactions, or perhaps something much more complicated; one hypothesis

proposed involves the guest encapsulation and host assembly equilibria coupled in a

square scheme. Work towards understanding this system is currently in progress, and we

hope to resolve this issue as the collaborative project continues.

The low but nonzero value of log β for Me4N+ is driven entirely by enthalpy, and

is strongly entropically disfavored. Whether Me4N+ interacts primarily with the exterior

or interior of the host is unclear. The favorable exothermic interaction is almost

4 For the remainder of this chapter, Kb refers to the equilibrium constant for encapsulation of a guest into the interior cavity of [Ga4L6]

12- in aqueous solution. 5 ITC experiments with [Ga4L6]

12- had been performed previously by Dr. Martin Michels in collaboration with Dr. Linfeng Rao at the Lawrence Berkeley National Laboratory. Titrations of the host with many different guests were performed, including Et4N

+, and the solution conditions he used were the same as those used in this chapter. Dr. Michels tabulated both exterior and interior binding equilibrium constants, stoichiometries, and reaction enthalpies for Et4N

+ from his ITC measurements in an unpublished report (which has been cited several times in various publications). The data were fit using the BindWorks software, with a multiple binding sites model. The fits obtained with this software are questionable, since its computational algorithms are prone to false minima. Furthermore, five variables were refined simultaneously in a nonlinear fit, with a large degree of correlation between the parameters. These data have not yet been analyzed using the methods described in this chapter.

- 44 -

completely neutralized by the free energy cost associated with the large decrease in

entropy. Cation-π interactions with the aromatic naphthalene rings of the ligands are

likely responsible for the favorable enthalpy, and numerous examples of exothermic

cation-π interactions with Me4N+ in aqueous solution have been described.12, 16 The

highly negative ∆S° is contrary to what one would expect for interior binding based on

solvation effects – the entropy gained due to desolvation of host (z = -12 to -11) and guest

from the Born equation is about 25 e.u. It is possible that the observed heat is due to

exterior binding interactions, but encapsulation may be involved as well. Similar

unexpected entropic costs have been observed by Arena et al., where encapsulation of a

pendant Me3N+ group into the pocket of a calixarene host is entropically disfavored in

aqueous solution as well.16 The authors attributed this entropy loss to both confinement

of the guest to the cavity and a stiffening of the surrounding host system.

A “stiffening” of the [Ga4L6]12- cluster upon encapsulation of Me4N

+ may also be

involved here. In the absence of guest, the cluster is ill-formed in aqueous solution, with

ligand dissociation readily observed via 1H NMR as extra peaks gradually appear in the

aromatic region of the spectrum of K12[Ga4L6] in degassed unbuffered D2O (pD ~ 8).

Addition of one equivalent of guest causes the cluster to snap together, and six sharp

aromatic signals are observed (characteristic of T-symmetric host), with no extra peaks.

This guest-templated self-assembly results in a large negative ∆S due to many partially

assembled microstates collapsing into one single species. Furthermore, ligand

coordination may contribute to the relatively high value of -∆H°, since the GaIII-

catecholamide coordination is most likely exothermic. While no experimental enthalpy

information with GaIII catecholates was found in the literature, this prediction is justified

- 45 -

since the coordination of enterobactin to FeIII at pH 9 is exothermic for the following

equilibrium reaction17

H3ent3- + Fe3+ = [Fe(ent)]3- + 3H+ ∆H = -6.5(3) kcal/mol

Since both enterobactin and H4L feature catecholamide chelating moieties, and the

chemistry of GaIII and FeIII are similar,18 the coordination of catecholamides to GaIII are

likely exothermic reactions as well.

The Pr4N+ system is very interesting, since the thermodynamic parameters for

both the 1:1 and 2:1 complexes could be quantified. For the 1:1 complex, does the Pr4N+

cation bind to the interior cavity to form [Pr4N+⊂Ga4L6]

11-, or does it bind to one of the

naphthalene rings on the exterior to form the [(Pr4N)···[Ga4L6]]11- ion pair? This may be

explored by comparing the stepwise binding equilibria (Table 3.3).

Table 3.3. Thermodynamic parameters for the stepwise binding equilibria for Pr4N+ with the [Ga4L6]

12- host (H) in 0.1 M aqueous KCl at 25 °C, computed from the calorimetry data listed in Table 3.2.

Reaction logKilogKi ¢H±

i¢H±

i , kcal/mol ¢S±

i¢S±

i , e.u.

H + Pr4N+ ÐÐ [(Pr4N

+)1·H] 2.0(3) -5.97(5) -11 ± 1

[(Pr4N+)1·H] + Pr4N

+ ÐÐ [(Pr4N+)2·H] 2.5(7) 5.2(1) 29 ± 4

The first Pr4N+ cation binding step with the free [Ga4L6]

12- cluster is an

exothermic process, but the second Pr4N+ binding reaction with the [(Pr4N

+)1H] adduct is

endothermic. Conversely, binding the first Pr4N+ cation is entropically disfavored, but

the second cation binding event is driven by a large increase in entropy. The two

stepwise binding equilibrium constants are similar, with their difference less than the

standard error of the values. However, K1 is driven entirely by enthalpy, and K2 is driven

entirely by entropy.

- 46 -

The value of ¢H±

1¢H±

1 is equal to ¢H±¢H± observed for Me4N+ within error, but less

entropy is lost during the first Pr4N+ binding step than the Me4N

+ reaction, causing the

1:1 binding constant for Pr4N+ to be an order of magnitude higher than that for Me4N

+.

The Born solvation entropy for Pr4N+ is only about 1 e.u. less negative than that for

Me4N+, so solvation cannot account for difference in reaction entropy observed for

forming 1:1 complexes with Me4N+ and Pr4N

+. However, if the first Pr4N+ cation binds

to the exterior and the second is encapsulated inside the host cavity, solvation effects can

account for the large difference between ¢S±

1¢S±

1 and ¢S±

2¢S±

2. This endothermic, entropically

driven encapsulation equilibrium for Pr4N+ was observed from a van’t Hoff plot using

variable temperature NMR to measure the encapsulation equilibrium constant.2

As discussed by Parac et al.,2 the host charge changes from z = -12 to z = -11

upon guest encapsulation, and the Born model predicts an entropy gain of about 23 e.u.

due to the change in solvation. In addition, the aqueous Pr4N+ ion must be completely

stripped of solvent to fit inside the host cavity, leading to an additional entropy gain of

about 2.5 e.u. Finally, about 8-10 water molecules are expected to occupy the interior

cavity of aqueous [Ga4L6]12-, which are liberated upon Pr4N

+ encapsulation for an

additional entropy gain. (The transfer of water molecules “frozen” in hydrated salts into

bulk water typically show an entropy change of about 6.7 e.u.).19 The positive value

observed for ¢H±

2¢H±

2 can also be attributed to solvation changes, since a large enthalpic cost

must be paid for the desolvation of the host and guest. Furthermore, the larger

Pr4N+ cation does not fit inside the host cavity as well as Et4N

+, and expansion of the host

cavity to accommodate the larger guest may also contribute to the positive ∆H2 value.

- 47 -

When weaker guests such as Me4N+ or Pr4N

+ are present in the working cell with

the host, the heat curves observed during titration of Et4N+ are significantly altered

(Figure 3.3). When two equivalents of the weakly encapsulated Me4N+ guest are

combined with the host solution in 0.1 M KCl and allowed to equilibrate for over an

hour, the heat released during the first addition of Et4N+ (~0.8 equiv) is five times larger

than that observed with the host alone. However, the two cumulative heat curves are

almost perfectly parallel, demonstrating the equilibrium constants characterizing these

two systems are essentially the same. Thus, the presence of two equivalents of Me4N+ in

solution with [Ga4L6]12- causes the reaction of the first equivalent of Et4N

+ with the host

to be much more exothermic. The difference in the Born solvation enthalpies for the two

cations is consistent with the observed increase in heat released. Consider the guest

exchange reaction where Me4N+ is displaced by Et4N

+:

[Me4N⊂Ga4L6]11- + Et4N

+(aq) ÐÐ [Et4N⊂Ga4L6]

11- + Me4N+

(aq)

0 5 10 15 200

5

10

15

20

25

30

To

tal H

ea

t, m

J

[Et4N

+]/[Host]

0.1 M KCl

KCl + 2 eq. Me4NCl

0.1 M Me4NCl

0 5 10 15 200

5

10

15

20

25

30

To

tal H

ea

t, m

J

[Et4N

+]/[Host]

0.1 M KCl

KCl + 2 eq. Pr4NCl

0.1 M Pr4NCl

a) b)

Figure 3.3. The thermograms observed during the titration of K12[Ga4L6] with Et4NCl are strongly affected by the presence of a) Me4N

+ and b) Pr4N+ as secondary guests. The black curves in both plots are identical,

measured in the absence of secondary guest with 0.1 M KCl as the ionic medium. The red curves were measured with 2 equivalents of secondary guest added to the host solution, also with 0.1 M KCl as the ionic medium. The green curves were measured when the KCl ionic buffer was replaced with the specified alkylammonium salt in both the analyte and titrant solutions. Heats are corrected for dilution effects.

- 48 -

The enthalpic cost of Et4N+ desolvation is offset by the much more negative resolvation

enthalpy of the smaller Me4N+ cation upon ejection from the host cavity, with the Born

model predicting a net ∆Hsolv ≈ -9 kcal mol-1.

For the larger Pr4N+ cation, the solvation enthalpy is less favorable than that for

Et4N+ due to the 1/r dependence in Equation 3.2. The simple solvation model described

above predicts that when two equivalents of Pr4NCl are combined with the host in a

similar fashion, the heat released after the first injection of Et4N+ should be lower than

that observed in the absence of secondary guest. This is not the case – the heat is higher

in the presence of two equivalents of Pr4N+, although to a lesser extent than with Me4N

+.

The reason for this is not known at this time, and is currently under investigation.

However, one possible explanation is that the interior binding affinity of Et4N+ is truly

enthalpically favored.

With a large excess of Me4N+ present, very different results are observed. The

green curve in Figure 3.3a was obtained by replacing the 0.1 M KCl ionic strength buffer

with 0.1 M Me4NCl in both the titrant (Et4NCl) and analyte (K12[Ga4L6]) solutions.

Thus, approximately 100 equivalents of Me4N+ per [Ga4L6]

12- host was present

throughout the experiment. A sharp break in the total heat curve, followed by a relatively

flat plateau, indicates that the primary exothermic reaction is essentially complete after

addition of two equivalents of Et4N+ per unit host. Sharp break points are characteristic

of systems with high binding constants (Kb ≥ 104 M-1).4 It is possible that the exterior

binding sites of the host are saturated by the large excess of Me4N+, allowing the Et4N

+

cation to bypass the exterior binding intermediate and directly bind to the host interior.

Despite its much weaker binding affinity, the very large excess of Me4N+ competes to a

- 49 -

small extent with the much stronger Et4N+ guest for binding to the host cavity. For this

reason, a satisfactory fit of the data can only be obtained if the model includes the 1:1

binding equilibria of both Et4N+ and Me4N

+. Furthermore, such a model accurately

accounts for the displacement reaction of encapsulated Me4N+ by the superior Et4N

+

guest. This analysis revealed that for Et4N+ encapsulation, ∆H° = -4.8 kcal mol-1,

log Kb = 4.6, and ∆S° = 5 e.u. Note that by including the Me4N+ binding equilibrium in

the model, these values are for the true guest binding equilibrium between Et4N+ and

[Ga4L6]12-, as opposed to the displacement reaction. These preliminary results must be

confirmed by titrations over a narrower concentration range by injecting a more dilute

Et4N+ stock solution; such experiments were being performed by our collaborators in

Catania at the same time this dissertation was written.

A similar effect was observed with 0.1 M Pr4NCl, although the break point in

Figure 3.3b is much less pronounced compared to that observed with Me4N+. The

shallower curve is consistent with the higher binding constant for Pr4N+, requiring higher

amounts of Et4N+ to fully displace the encapsulated Pr4N

+. After the break point is

attained early in the titration, corresponding to quantitative Et4N+ encapsulation, heat

continues to be released as more Et4N+ is injected. This may be due to displacement of

exterior bound Pr4N+ by the smaller Et4N

+, since cation-π interactions are typically more

favorable for smaller cations.12

The binding of Et4N+ with [Ga4L6]

12- in 0.1 M KCl is clearly an exothermic

process in all three cases, as shown in Table 3.2 and Figure 3.3a. However, the

enthalpies inferred from van’t Hoff plots of alkylammonium binding to the [Ga4L6]12-

host in D2O indicated guest encapsulation was endothermic.2 Dr. Martin Michels, a

- 50 -

previous postdoctoral scholar working with Professor Raymond through May of 2001,

also noted this discrepancy during previous calorimetry experiments with the [Ga4L6]12-

host. Using variable temperature NMR spectroscopy to generate van’t Hoff plots, Dr.

Michels found that in the absence of ionic strength buffer, the observed slopes indicate

encapsulation of both Et4N+ and Pr4N

+ are endothermic. In the presence of KCl,

however, the slopes change sign, indicating that encapsulation becomes exothermic for

both guests at higher ionic strength.20

To test whether added salt has an appreciable impact on the guest binding

enthalpy, two similar ITC experiments were performed: one with 0.1 M aqueous KCl,

and the other using only water (“non-salt” system). Both analyte solutions contained

1 mM K12[Ga4L6], and both titrant solutions contained 10 mM Et4NCl. Although

numerical values could not be fit to the non-salt data due to spurious dilution heats

caused by large ionic strength changes, the data in Figure 3.4 clearly show that the

reaction is exothermic whether or not KCl is present. The difference in slope may be due

to uncompensated endothermic host dilution that becomes significant in the absence of

ionic strength buffer.

Since metal-catechol binding equilibria are strongly pH dependent, with

formation of tris-bidentate metal catechol complexes favored in basic solution, should a

buffer system be used to regulate pH during the encapsulation process? For equilibrium

studies, buffers can be very useful if a chemical reaction leads to a pH change. A 1 mM

solution of [Ga4L6]12- in 0.1 M KCl was prepared, and the pH was measured to be 7.56 in

the absence of guest. To this was added Et4NCl titrant solution in several aliquots,

allowing the system to equilibrate for 20 minutes between each addition. No significant

- 51 -

pH change was observed (i.e. ±0.03 pH units) even up to 16 equivalents of Et4N+ per

host. This clearly demonstrates that protons are neither consumed nor released during the

guest binding interaction.

Summary

Isothermal titration calorimetry is the most accurate method for measuring

reaction enthalpies, since heat is measured directly. The calorimetry data described in

this chapter unambiguously demonstrate that the encapsulation of Et4N+ by the [Ga4L6]

12-

host in aqueous solution is exothermic by nearly 5 kcal mol-1, in contrast to the

endothermic binding enthalpy inferred from the van’t Hoff plot. However, the complex

binding interactions are difficult to identify from this method, since the only observable

parameter involved is heat. This chapter represents the beginning of a long term research

Ethyl With or Without KCl

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

0.00 0.20 0.40 0.60 0.80 1.00

# eq. ethyl

Cu

mu

lati

ve h

eat,

mJ

no KCl

with KCl

Figure 3.4. Comparison of the cumulative heats released during titration of 1 mM [Ga4L6]12- with

9.23 mM Et4NCl at 25 °C for solutions prepared with and without KCl ionic strength buffer.

- 52 -

collaboration, and the unresolved discrepancies between NMR and ITC observations will

be addressed as this project continues.

Experimental

General considerations

Unless noted otherwise, reagents were obtained from commercial suppliers and

used without further purification. Standard Schlenk techniques were used for reactions

carried out under argon, and a glove box continuously purged with nitrogen was used to

manipulate and store air-sensitive solids. (Due to less rigorous glovebox conditions in

Catania, vials containing host samples were flushed with argon, capped, wrapped in

Parafilm, and placed in a -20 °C freezer for later use). When necessary, solvents were

degassed by at least six pump/fill cycles while vigorously stirring, using argon for the fill

step. H4L (H4L = 1,5-bis(2,3-dihydroxybenzamido)naphthalene) and K12[Ga4L6] were

synthesized according to literature procedures.1, 2 Tetraethylammonium chloride (Et4NCl)

was recrystallized from absolute ethanol/ether and dried in vacuo over P2O5 at room

temperature for 12 hours, then dried in vacuo over molecular sieves at 60 °C for 18 hours

and stored under nitrogen. Tetramethylammonium chloride (Me4NCl),

tetrapropylammonium chloride (Pr4NCl), and tetrabutylammonium chloride (Bu4NCl)

were not recrystallized, but were dried over P2O5 in a manner similar to that for Et4NCl.

Aqueous silver nitrate, used as a titrant for chloride determination, was standardized by

Valeria Zito (Dipartimento di Scienze Chimiche at the Università di Catania, Italy) via

seven titrations into aqueous NaCl, using K2CrO4 as an indicator

([AgNO3] = 0.0488(2) M). This AgNO3 solution was used within one week after

- 53 -

standardization. Stock solutions of the R4NCl titrants (R = Me, Et, Pr, Bu) were

standardized using this AgNO3 titrant (with K2CrO4 indicator) to determine the chloride

concentration to three significant figures. Routine mass spectrometry and elemental

analysis was performed by the Mass Spectrometry Laboratory and Microanalysis Facility

in the College of Chemistry at the University of California, Berkeley.

Thermogravimetric analyses (TG) were performed by the analytical facilities in the

Dipartimento di Scienze Chimiche at the Università di Catania, Italy.

K12[Ga4L6]·(Me2CO)2(H2O)8. The following reaction was carried out under argon. A

suspension of 1.51 g (3.50 mmol) of H4L and 855 mg (2.33 mmol) of Ga(acac)3 in

150 mL of degassed methanol was prepared, and the slurry was degassed via seven

pump/fill cycles. Via syringe, 16 mL of 0.5 M KOH in methanol was added to the

reaction mixture while stirring, immediately followed by via seven pump/fill cycles to

remove any oxygen introduced with the base solution. Addition of base caused most

solid to dissolve, and the color changed from white to yellow. Nearly all solid dissolved

after stirring for 15 min. The dark yellow solution was stirred overnight at room

temperature, filtered through a fine frit to remove insoluble residual impurities, filtered

again through several 0.2 µm nylon syringe filter disks, and the volume was reduced to

30 mL with a vacuum pump. Any solid that formed during this evaporation step was re-

dissolved with degassed methanol. The product was precipitated by slowly adding

600 mL of degassed acetone via cannula, stirred slowly for 5 min to aid mixing of

acetone and methanol, and left alone for two hours. The yellow precipitate that formed

was collected on a frit under a stream of nitrogen, washed with acetone (3 x 50 mL), and

- 54 -

dried in vacuo 12 hours to yield 1.3 g (65%) of light yellow powder. Duplicate Anal.

Calc. (found) for C144H84Ga4K12N12O36·(Me2CO)2·(H2O)8: %C, 50.51 (50.86); H, 3.17

(3.12); N, 4.71 (4.75). Thermogravimetric Analysis: ~8% volatile components. The solid

is somewhat hygroscopic, gaining mass on an analytical balance after removing from

glove box. Samples for thermogravimetric analysis were stored in an air-filled vial for

several hours before the TG experiment, so the sample had become fully hydrated by the

time the experiment was performed.

Isothermal titration calorimetry (ITC) experiments

Measurements were performed using a CSC 5300 Nano-Isothermal Titration

Calorimeter III at the Università di Catania, Italy. The ambient temperature of the room

was controlled to 25 ± 1 ºC, with double doors for the entrance minimizing temperature

fluctuations. All measurements were performed with a cell temperature of 25.000(5) ºC,

with 0.1 M KCl in the dummy cell. Samples were degassed by stirring under vacuum for

20 minutes to ensure no air bubbles formed in the cell or syringe. Prior to beginning each

experiment, the instrument response was calibrated with 10 identical power pulses

(50-300 uJ) using the sample for the specific experiment in the working cell. The cell

volume was 1036 µL, and was operated on overfill mode.21 Titrant was automatically

injected in 8 µL increments using a special syringe with a stirrer tip, rotating at 200 rpm

to mix the solution. For experiments in the presence of host, a 600 s delay between

injections was used up to 4:1 guest/host ratio due to sluggish kinetics in this region. The

time intervals between injections were incrementally shortened as higher guest/host ratios

were reached, eventually reaching 300 s as the minimum delay for greater than 10 equiv.

- 55 -

guest. A dilution experiment in the absence of host was performed with the same titrant

solution and ionic medium for each set of conditions.

The raw data were corrected for baseline heat and integrated using BindWorks,22

and the integrated dilution heat was subtracted from the integrated gross heat for each

interval to obtain the net incremental heat for each injection. Values for K and ∆H° were

obtained with the computer program Hyp∆H,15 which was derived from the general

analysis suite HYPERQUAD.23

Molecular modeling calculations

Quantum mechanical calculations of the hypothetical [K2L]2- complex were

performed with Spartan ’02 for Linux.24 The equilibrium geometry of the complex was

first determined using a molecular mechanics calculation (MMFF method). The

electronic structure of this minimized structure was then computed with a Hartree-Fock

molecular orbital calculation, using the 6-31G* basis set. From the results of this single-

point energy calculation, the potential density surface map was calculated to produce

Figure 3.1.

References

1. Caulder, D. L.; Powers, R. E.; Parac, T. N.; Raymond, K. N., “The Self-Assembly of a Predesigned Tetrahedral M4L6 Supramolecular Cluster.” Angew. Chem. Int. Ed.

1998, 37, 1840-1842.

2. Parac, T. N.; Caulder, D. L.; Raymond, K. N., “Selective Encapsulation of Aqueous Cationic Guests into a Supramolecular Tetrahedral M4L6 Anionic Host.” J. Am.

Chem. Soc. 1998, 120, 8003-8004.

3. a) Diederich, F.; Dick, K.; Griebel, D., “Complexation of arenes by macrocyclic hosts in aqueous and organic solutions.” J. Am. Chem. Soc. 1986, 108, 2273-2286; b) Meric, R.; Vigneron, J.-P.; Lehn, J.-M., “Efficient complexation of quaternary

- 56 -

ammonium compounds by a new water-soluble macrobicyclic receptor molecule.” J.

Chem. Soc. Chem. Comm. 1993, 129-131; c) Smithrud, D. B.; Wyman, T. B.; Diederich, F., “Enthalpically driven cyclophane-arene inclusion complexation: Solvent-dependent calorimetric studies.” J. Am. Chem. Soc. 1991, 113, 5420-5426; d) Ferguson, S. B.; Sanford, E. M.; Seward, E. M.; Diederich, F., “Cyclophane-arene inclusion complexation in protic solvents: solvent effects versus electron donor-acceptor interactions.” J. Am. Chem. Soc. 1991, 113, 5410-5419; e) Stauffer, D. A.; Barrans, R. E., Jr.; Dougherty, D. A., “Concerning the thermodynamics of molecular recognition in aqueous and organic media. Evidence for significant heat capacity effects.” J. Org. Chem. 1990, 55, 2762-2767; f) Diederich, F., “Complexation of Neutral Molecules by Cyclophane Hosts.” Angew. Chem. Int. Ed. Engl. 1988, 27, 362-386; g) Ferguson, S. B.; Seward, E. M.; Diederich, F.; Sanford, E. M.; Chou, A.; Inocencio-Szweda, P.; Knobler, C. B., “Strong enthalpically driven complexation of neutral benzene guests in aqueous solution.” J. Org. Chem. 1988, 53, 5593-5595; h) Petti, M. A.; Shepodd, T. J.; Barrans, R. E., Jr.; Dougherty, D. A., “"Hydrophobic" binding of water-soluble guests by high-symmetry, chiral hosts. An electron-rich receptor site with a general affinity for quaternary ammonium compounds and electron-deficient π systems.” J. Am. Chem. Soc. 1988, 110, 6825-6840.

4. Fisher, H. F.; Singh, N., “Calorimetric Methods for Interpreting Protein-Ligand Interactions.” Methods in Enzymology 1995, 259, 194-221.

5. a) Horn, J. R.; Brandts, J. F.; Murphy, K. P., “van't Hoff and Calorimetric Enthalpies II: Effects of Linked Equilibria.” Biochemistry 2002, 41, 7501-7507; b) Arena, G.; Calì, R.; Maccarrone, G.; Purrello, R., “Critical review of the calorimetric method for equilibrium constant determination.” Thermochim. Acta 1989, 155, 353-376.

6. Caulder, D. L.; Brückner, C.; Powers, R. E.; König, S.; Parac, T. N.; Leary, J. A.; Raymond, K. N., “Design, Formation, and Properties of Tetrahedral M4L4 and M4L6 Supramolecular Clusters.” J. Am. Chem. Soc. 2001, 123, 8923-8938.

7. Leung, D. H.; Bergman, R. G.; Raymond, K. N., “Scope and Mechanism of the C-H Bond Activation Reactivity within a Supramolecular Host by an Iridium Guest: A Stepwise Ion Pair Guest Dissociation Mechanism.” J. Am. Chem. Soc. 2006, 126, 9781-9797.

8. Tiedemann, B. E. F.; Raymond, K. N., “Dangling Arms: A Tetrahedral Supramolecular Host with Partially Encapsulated Guests.” Angew. Chem. Int. Ed.

2006, 45, 83-86.

9. Davis, A. V.; Fiedler, D.; Seeber, G.; Zahl, A.; van Eldik, R.; Raymond, K. N., “Guest Exchange Dynamics in an M4L6 Tetrahedral Host.” J. Am. Chem. Soc. 2006, 128, 1324-1333.

10. Fiedler, D.; van Halbeek, H.; Bergman, R. G.; Raymond, K. N., “Supramolecular Catalysis of Unimolecular Rearrangements: Substrate Scope and Mechanistic Insights.” J. Am. Chem. Soc. 2006, 128, 10240-10252.

- 57 -

11. Ahrland, S., “Thermodynamics of Complex Formation between Hard and Soft Acceptors and Donors.” In Structure and Bonding, Jørgensen, C. K.; Neilands, J. B.; Nyhoum, R. S.; Reinen, D.; Williams, R. J. P., Eds. Springer-Verlag: Berlin, 1968; Vol. 5, pp 118-149.

12. Dougherty, D. A.; Ma, J. C., “The Cation-π Interaction.” Chem. Rev. 1997, 97, 1303-1324.

13. Born, M., “Volumen und Hydratationswärme der Ionen.” Z. Physik 1920, 1, 45-48.

14. Lide, D. R., Handbook of Chemistry and Physics. 81st ed.; CRC Press: Boca Raton, 2000; p 6.149-6.150.

15. Gans, P.; Sabatini, A.; Vacca, A. Hyp∆H, 1.0.62; Protonic Software: Leeds, UK, 2007.

16. Arena, G.; Casnati, A.; Contino, A.; Lombardo, G. G.; Sciotto, D.; Ungaro, R., “Water-Soluble Calixarene Hosts that Specifically Recognize the Trimethylammonium Group or the Benzene Ring of Aromatic Ammonium Cations: A Combined 1H NMR, Calorimetric, and Molecular Mechanics Investigation.” Chem.

Eur. J. 1999, 5, 738-744.

17. Scarrow, R. C.; Ecker, D. J.; Ng, C.; Liu, S.; Raymond, K. N., “Iron(III) Coordination Chemistry of Linear Dihydroxyserine Compounds Derived from Enterobactin.” Inorg. Chem. 1991, 30, 900-906.

18. Loomis, L. D.; Raymond, K. N., “Kinetics of gallium removal from transferrin and thermodynamics of gallium-binding by sulfonated tricatechol ligands.” J. Coord.

Chem. 1991, 23, 361-387.

19. Phillips, C. S. G.; Williams, R. J. P., Inorganic Chemistry. 1st ed.; Oxford University Press: New York, 1965; Vol. 1, p 260.

20. Michels, M.; Raymond, K. N., unpublished results.

21. Spokane, R. B.; Gill, S. J., “Titration microcalorimeter using nanomolar quantities of reactants.” Rev. Sci. Instrum. 1981, 52, 1728-1733.

22. BindWorks, Version 3.1.8; Calorimetry Sciences Corp.: Lindon, Utah, 2007.

23. Gans, P.; Sabatini, A.; Vacca, A., “Investigation of equilibria in solution. Determination of equilibrium constants with the HYPERQUAD suite of programs.” Talanta 1996, 43, 1739-1753.

24. Spartan '02 for Linux, Build 119; Wavefunction, Inc.: Irvine, CA, 2001.

- 58 -

CHAPTER 4

Partial Guest Encapsulation Modes

Introduction

Guest exchange for the M4L6 tetrahedral host was demonstrated to proceed

through a nondissociative mechanism, where the cluster remains fully intact throughout

the entire process.1, 2 Guest ingress and egress into and out of the host cavity is thought

to occur through expandable apertures in each of the host’s four triangular faces, centered

along the three-fold rotation axis passing through the opposite vertex. This conjecture

raised the following question: could this aperture be used to extend part of a guest outside

the cavity while another part remains bound to the host interior? The answer is yes. This

unique encapsulation mode will be called “partial guest encapsulation.” Two different

categories of guests are discussed in this chapter where partial guest encapsulation has

been observed: zwitterions (RuCn = [CpRu(η6-C6H5(CH2)nSO3)], n = 4, 6, 8, 10) and

monocations with pendant linear alkanes (RuAn+ = [CpRu(η6-C6H5(CH2)nH)]+, n = 4, 6,

8, 10, 12) whose structures are shown in Scheme 4.1. Both guest systems feature linear

Scheme 4.1. Structural diagrams of the RuCn zwitterion and the RuAn+ cation, where n corresponds to the

total number of carbon atoms contained in the linear hydrocarbon chains for both structures.

- 59 -

hydrocarbon chains to pass through the small aperture in the host’s triangular faces

without much distortion, minimizing the energy cost. The work described in this chapter

has been previously described in two separate communications.3, 4

Zwitterionic Guests

With the [Ga4L6]12- ion as the host, suitable guest compounds had to be designed

with appendages capable of protruding through the opening of the host into the bulk

solvent. Such a guest should have three regions: a lipophilic monocationic head group

with a high affinity for the host interior, a linear chain of variable length to protrude

through the aperture, and a hydrophilic anionic end group that cannot be encapsulated

(Figure 4.1). The sandwich complex [CpRu(η6-C6H6)]+ is known to be encapsulated by

the [Ga4L6]12- ion,5 and was chosen as the cationic head group for RuCn. This series of

compounds features linear alkyl chains, with one end bound to the cationic sandwich

complex at the phenyl ligand, and the other bound to a sulfonate anion.

Although several other choices are feasible for the cationic head group, such as

Et4N+ derivatives, the RuII sandwich complex was chosen because its synthesis was

Figure 4.1. a) Cartoon of the M4L6 host encapsulating the cationic head group (red ball) of a zwitterion while the anionic end (blue ball) remains outside the cavity. For this to occur, the linear alkyl chain (wavy line) must pass through an aperture in one of the host’s triangular faces. b) Relationship between the cartoon guest and the actual structure of the RuCn zwitterion.

- 60 -

relatively straightforward. (Attempts to prepare Et3N+-(CH2)n-SO3

- were hindered by

purification difficulties). Commercially available n-phenylalcohols were treated with

PBr3, and the reaction of the resulting bromide with sodium sulfite afforded the

phenylalkanesulfonate product as the sodium salt. These soap-like compounds were then

combined with [CpRu(MeCN)3]PF6 in degassed, anhydrous chloroform to yield the

desired products as the NaPF6 adduct (Scheme 4.2).

Addition of a stoichiometric amount of RuCn to K12[Ga4L6] in D2O led to the

encapsulation of the ruthenium head group for all chain lengths (n = 4 – 10). The 1H

NMR spectra of the resulting host-guest complexes display signals for the Cp and phenyl

rings of the guest shifted to significantly higher fields relative to the values observed in

the absence of host (Figure 4.2). Such an upfield shift is a diagnostic feature observed for

encapsulated guest protons and is caused by the magnetic shielding from the naphthalene

groups surrounding the cavity of the [Ga4L6]12- host.6-8 Furthermore, the two sets of

mirror-related phenyl protons become diastereotopic upon encapsulation of the sandwich

complex by the host because of the chiral environment of the cavity.

Diastereotopic splitting is also observed for most, but not all, geminal methylene

resonances, accompanied by varying upfield shifts. With the aid of 2D COSY and/or

Scheme 4.2. Generalized synthesis for RuCn·NaPF6

- 61 -

TOSCY NMR spectroscopic analysis (Appendix), all chain protons were fully assigned

(Figure 4.2). Methylene carbon toms are numbered sequentially along the chain and

begin with the carbon atom bound to the sandwich complex. For RuC4, the geminal

methylene protons on C1–C3 become diastereotopic upon encapsulation by [Ga4L6]12-,

and diastereotopic geminal methylene protons are observed for C1–C5 for encapsulated

RuC6, RuC8, and RuC10. Upfield shifts tend to increase for protons closer to the head

Figure 4.2. Guest region of the 1H NMR spectra (D2O, 500 MHz) of the host-guest complexes [RuCn⊂Ga4L6]

12-. Guest resonance assignments follow the labeling scheme illustrated for RuC4 (top), with the same numbering pattern used for all chain lengths. Cp denotes the 5H singlet from the Ru-bound cyclopentadiene ring. Interior protons are highlighted with red labels, and all signals that integrate to two protons are identified by subscripted labels.

- 62 -

group. In comparison, several methylene protons remain enantiotopic and show little, if

any, upfield shift. Furthermore, the signal from the methylene group adjacent to the

sulfonate moiety is essentially unaffected by encapsulation of the cationic head group for

all chain lengths. These 1H NMR observations indicate that part of the alkyl chain bound

to the cationic sandwich complex resides within the chiral host, and the rest of that alkyl

chain lies outside the cavity with the terminal sulfonate group. At least one methylene

group is found outside the host in all four systems. Thus, the [Ga4L6]12- cluster

encapsulates only part of the RuCn zwitterion. This represents the first reported example

of partial guest encapsulation with a “closed” host.3

Symmetry of host-guest complexes

The M4L6 cluster has pure rotation point group T symmetry. Three-fold rotation

axes pass through each metal center at the vertices of the tetrahedron, and two-fold

Figure 4.3. Host resonances from the 1H NMR spectra observed for a) point group T typically observed for host-guest complexes, and b) [RuC10⊂Ga4L6]

12- characteristic of C3 symmetry.

- 63 -

rotation axes pass through opposite pairs of naphthalene rings surrounding the cavity.

Most fully encapsulated guests do not change the observed symmetry of the host, because

the guest can rapidly tumble inside the cavity. For these T-symmetric host-guest

complexes, the signals for the 72 host protons are observed as six sets of 12 protons

(Figure 4.3a) because the catechol and naphthalene ring edges are interrelated, each with

three adjacent nonequivalent protons.

In the presence of the protruding arm, however, the host aromatic resonances are

split into 24 sets of three protons (Figure 4.3b). This observation indicates that the

overall symmetry of the cluster is reduced from point group T for the host alone to C3

upon encapsulation of RuCn. Such a symmetry reduction is expected to occur when the

alkyl sulfonate arm of the guest is extruded through the opening in a triangular face of the

tetrahedron (Figure 4.4). This perturbation breaks the ligand two-fold symmetry, but

C3 (observed) D2 (not observed)

Figure 4.4. Breaking the T symmetry of the M4L6 cluster can result in two different lower symmetry subgroups. The protruding tail for RuCn is collinear with one three-fold axis, but breaks the two-fold ligand symmetry, resulting in 24 resonances from the two sets of unsymmetric ligands. The above C3 structure shows one set in red, and the other in blue, with different shades to highlight the lack of ligand symmetry. Alternatively, distortion along a two-fold axis results in the D2 subgroup, which exhibits 18 resonances from the three sets of chemically inequivalent C2 symmetric ligands.

- 64 -

retains the C3 axis running from a gallium(III) vertex through the aperture of interest.

The six ligands that span the edges of the tetrahedron separate into two chemically

nonequivalent groups: three “base” ligands that surround the protruding arm and three

“side” ligands connected to the gallium(III) vertex opposite the protrusion, with 12

different protons for each. According to 2D COSY NMR spectroscopy, the 24 host

signals originate from four sets of catechol protons and four sets of naphthalene protons

(Appendix).

Structural details from NOE interactions

The 2D NOESY NMR spectra of the [RuCn⊂Ga4L6]12- system provide additional

information about the structures of the host-guest complexes. Figure 4.4 shows the

spectrum of the [RuC6⊂Ga4L6]12- system with focus on the cross section between the host

and guest resonances. The phenyl and Cp signals of the guest show strong correlations

with signals from three out of the four sets of host naphthalene protons. No cross peaks

are observed between the phenyl or Cp resonances of the guest and catechol proton

signals of the host. This confirms that the cationic head group is buried deep within the

host cavity, near the naphthalene ring walls. Similar 2D NOESY NMR spectroscopic

observations have been reported for other guests within the [Ga4L6]12- cluster, including

Et4N+ and [CpRu(η6-C6H6)]

+ ions.5, 6

- 65 -

Cross peaks are also observed between several host resonances and most, if not

all, methylene groups on the alkyl chain. The first six to eight methylene groups all show

cross peaks with the same two host resonances – one doublet and one triplet, highlighted

in Figure 4.5. According to molecular modeling studies (Figure 4.5) and 2D COSY

NMR spectra (Appendix), these signals originate from two adjacent naphthalene protons,

located ortho and meta to the amide nitrogen atom. Because of the orientation of the

naphthalene ring, the para proton is directed away from the alkyl chain and strong cross

peaks are not observed with the guest.

Figure 4.5 The 2D NOESY NMR spectrum (D2O, 500 MHz) of the [RuC6⊂Ga4L6]12- system shows cross

peaks between host resonances (horizontal axis) and guest resonances (vertical axis). Resonances from naphthyl protons that border the aperture, ortho and meta to the amide nitrogen atom, are highlighted in red. The catechol resonance highlighted in blue shows cross peaks with signals from exterior methylene protons (nap = naphthalene, cat = catechol).

- 66 -

Relative distances between the host and guest protons in the [RuC6⊂Ga4L6]12-

system were determined from NOE interaction growth rates, showing these ortho and

meta protons are the closest to the protruding alkyl chain, so they must be located on the

C3-related edges of the three naphthalene rings that surround the aperture. These

hydrogen atoms are directed away from the cluster center and can be used as boundary

markers to distinguish the host interior from the exterior. Considering the guest’s alkyl

chain, methylene groups 4 and 5 are closest to the boundary, whereas methylene group 1

is the farthest. This difference suggests that the two naphthyl protons are situated

between the C4 and C5 carbons of the alkyl chain, with C4 on the interior side and C5 on

the exterior side of the cavity boundary. C1, immediately adjacent to the encapsulated

cationic head group, lies deep inside the cluster.

Cross peaks between protons at different positions along the alkyl chain provide

information about the specific conformation of the guest’s alkyl chain. The actual

conformation adopted by the alkyl chain can be described by two extremes: fully

extended for the longest chain, or helical coiling for the shortest chain. In a fully

Figure 4.6. MM3 minimized structural model of the [RuC10⊂Ga4L6]12- host-guest complex.

- 67 -

extended chain, protons on carbon C(i) will only show NOE interactions with C(i+2)

and/or C(i-2), since all dihedral angles are 180°. Helical coiling leads to additional NOE

interactions such as C(i+3) and C(i+4) due to the gauche orientations of the chain. Such

gauche interactions were observed by Rebek and co-workers upon encapsulation of n-

alkanes by a cylindrical supramolecular host, particularly when the length of the extended

alkane exceeded the dimensions of the dimer’s interior cavity.9 For the [RuCn⊂Ga4L6]12-

system, the only intra-chain cross peaks observed are from C(i ± 1) methylenes and

diastereotopic geminal protons bound to the same carbon atom, indicating the alkyl chain

is in the extended conformation. The chain is probably drawn into solution by the

sulfonate anion, extended away from the anionic host.

Guest binding affinities

Competitive binding experiments with Et4N+ were carried out to evaluate the

binding affinities of RuCn in D2O. Equimolar amounts of RuCn, Et4N+, and [Ga4L6]

12-

were combined in D2O and the system was allowed to equilibrate overnight. With Kref

known (for Et4N+ in D2O with no added KCl, Kref = 1.96 x 104 M-1),8 the unknown

binding constant Kn can be obtained from the relative concentrations of the two host-

guest complexes from the following equation:

Kn = Kref

µ

[RuCn ½ Ga4L6]

[Et4N ½ Ga4L6]

¶2

Kn = Kref

µ

[RuCn ½ Ga4L6]

[Et4N ½ Ga4L6]

¶2

(4.1)

where n refers to the length of the guest’s alkyl chain. The results are summarized in

Table 4.1.

- 68 -

Table 4.1. Host-guest binding equilibrium constants for [RuCn⊂Ga4L6]12- in D2O, from Et4N

+ competitive binding experiments.

Guest Kn, M-1 log10 Kn

RuC4 (≤ 103) ---

RuC6 1.7 x 103 3.2

RuC8 8.7 x 103 3.9

RuC10 6.9 x 103 3.8

No signal for [RuC4⊂Ga4L6]12- was observed by NMR when 1 equivalent of

Et4N+ was added, so a value for K4 could not be obtained from these experiments.

However, these data clearly establish the relative ordering for binding affinities:

K4 < K6 < K10 ~ K8

For formation of a host-guest complex with the M4L6 cluster, the optimal chain length for

[RuCn⊂Ga4L6]12- is eight carbons. As the chain length increases from n = 4 to n = 8, the

sulfonate anion can move farther away from the anionic host to minimize coulombic

repulsion, lowering the energy of the host-guest complex.

Mass spectrometry

Host-guest complex formation was also confirmed by high resolution negative ion

electrospray mass spectrometry (ESI-MS) (Figure 4.7). Spectra were obtained for

solutions of all four [RuCn⊂Ga4L6]12- systems, showing peaks for the z = -3 and -4 charge

states of the host-guest complexes with K+, Na+, and/or H+ counterions. The mass

spectra of the [RuC4⊂Ga4L6]12- and [RuC10⊂Ga4L6]

12- systems show additional peaks that

correspond tot the z = -5 charge state of the host-guest complex.

- 69 -

The above study shows that the [Ga4L6]12- tetrahedron remains intact upon

incorporation of the cationic head of a zwitterion, while the alkyl chain attached to the

anionic tail passes through one of the small openings at the centers of the triangular faces.

This process is fully consistent with the nondissociative guest egress mechanism

described by Davis and Raymond in a previous publication.1

Monocations with Pendant Alkanes

When the cationic head group of RuCn is encapsulated, coulombic repulsion with

the anionic host requires the sulfonate anion at the opposite end of the hydrocarbon chain

to remain in solution. This design feature forced the hydrocarbon to be permanently

extruded through one aperture in the cluster as long as the cation remained bound to the

Figure 4.7. Portion of the ESI MS of the [RuC10⊂Ga4L6]12- system that shows two adjacent peaks for the

z = -4 charge state, with predicted isotopic distribution patterns for two particular fragment ion formulae.

- 70 -

host interior.3 What happens if the sulfonate group were replaced by a simple hydrogen

atom, so the pendant chain terminates with a neutral methyl group? To answer this

question, the series of cations RuAn+ was prepared (n = 4, 6, 8, 10, 12) by combining the

commercially available n-phenylalkanes with [CpRu(CH3CN)3]PF6 in a manner similar

to the RuCn synthesis. By eliminating repulsion from the sulfonate anion, the neutral

chain is free to completely retract inside the host, as long as there is enough space to

accommodate it.

When equimolar amounts of solid [RuAn]PF6 and K12[Ga4L6] were combined in

D2O (n = 4, 6, 8, 10), homogeneous solutions were obtained despite the very low water

solubilities of [RuAn]PF6 , particularly for longer chain lengths. This enhanced solubility

is due to encapsulation of the cationic ruthenium center by the water soluble [Ga4L6]12-

host, with formation of [RuAn⊂Ga4L6]11- readily apparent from the large upfield shifts of

the guests’ Cp and phenyl resonances in the 1H NMR spectra (Figure 4.8).3, 6 The

diastereotopic splitting patterns for the five phenyl proton resonances observed upon

encapsulation of RuAn+ are very similar to those for [RuCn⊂Ga4L6]

12-. For RuA12+,

MeOD was used instead of D2O to combine the two solids, but similar upfield shifts

observed upon encapsulation of the head group confirmed quantitative formation of

[RuA12⊂Ga4L6]11- in MeOD. Host-guest complex formation was confirmed via ESI MS

for all chain lengths (Appendix).

In contrast to the RuCn system, the point symmetry of the [RuAn⊂Ga4L6]11- host-

guest complexes is chain length dependent. For long side chains (n ≥ 8), the entire alkane

cannot fit inside the cavity and remains extruded through one of the host’s faces, with C3

symmetry observed for the host-guest complexes. With a short side chain (n = 4, RuA4+),

- 71 -

the entire guest is completely encapsulated by the host, and rapid tumbling of the guest

within the cavity leads to overall T symmetry for the [RuA4⊂Ga4L6]11- host-guest

complex. With an intermediate chain length (n = 6, RuA6+), the side chain rapidly

Figure 4.8. Guest region of the 1H NMR spectra (500 MHz) of [RuAn⊂Ga4L6]11- for the different chain

lengths n. Samples were dissolved in D2O for all spectra except for n = 12 (CD3OD). Selected assignments for the guest protons are shown in blue, with the regions where different groups of protons resonate labeled as appropriate. A label printed in one spectrum applies to all those below it as well.

- 72 -

extends and retracts at room temperature such that it dynamically protrudes through all

four triangular faces of [Ga4L6]12- on the NMR timescale (Figure 4.9). During this

dynamic process, the side chain retracts such that RuA6+ is fully encapsulated long

enough so that its tumbling averages out to T symmetry for the host-guest complex.

Lowering the temperature reveals that the ground state configuration has C3 symmetry,

with the alkyl chain fully extended and protruding through one of the host’s four

triangular faces. Since all four apertures are identical in the retracted, T-symmetric state,

there are four degenerate C3 states. Rapid interconversion between C3 states on the NMR

timescale leads to time-averaged T symmetry at higher temperatures.

The dynamic nature of the [RuA6⊂Ga4L6]11- system is apparent by monitoring the

host region during variable temperature 1H NMR measurements. At room temperature,

Figure 4.9. While the cationic RuII sandwich complex (red ball at top) of RuA6+ remains within the

[Ga4L6]12- cluster, the pendant alkyl chain rapidly retracts and extends at room temperature, moving the

terminal methyl group (black ball) inside and outside the host cavity. The higher energy retracted state exhibits T symmetry, and the four degenerate C3-symmetric states are lower in energy.

- 73 -

six signals – two broad and four sharp – integrating to 12H each are observed for the 72

aromatic host protons, consistent with point group T (Figure 4.10). At low temperature

(T < -40 °C), the host resonances split into 24 overlapping signals integrating to 3H each,

with the 2D COSY spectrum at -60 °C (Appendix) consistent with a C3-symmetric host-

guest complex whose guest side chain protrudes from one of the face apertures. At high

temperature (T ≥ 50 °C), the host signals coalesce into six sharp peaks.

Coalescence of the host resonances for [RuA8⊂Ga4L6]11- was observed at a much

higher temperature, with six broad host resonances observed at 75 °C (Figure 4.11). This

demonstrates that the longer eight carbon chain can fully retract into the cavity, but the

Figure 4.10. Aromatic region of the 1H NMR spectra for [RuA6⊂Ga4L6]11- showing the signals for the

host’s 72 aromatic protons observed at a) -60 °C in MeOD, b) 20 °C in D2O, and c) 50 °C in D2O.

- 74 -

high coalescence temperature indicates that the T state is much higher in energy than the

C3 ground state with the eight carbon chain, most likely due to the limited space of the

host cavity.

The modest upfield shift observed for the CH3 triplet (Figure 4.8) depends on the

percentage of time the alkyl chain spends retracted (T state) vs. extended (C3 state), since

the magnetically shielded environment of the host cavity leads to an upfield shift for

interior protons. For [RuA6⊂Ga4L6]11-, conversion between C3 and T states at room

temperature is fast on the NMR timescale, and the observed chemical shift for the CH3

Figure 4.11. Aromatic region of the 1H NMR spectra for [RuA8⊂Ga4L6]11- (D2O, 500 MHz) showing the

signals for the host’s 72 aromatic protons observed at a) room temperature, b) 50 °C, and c) 75 °C.

- 75 -

signal δobs is a population average of the chemical shifts for the C3 and T states, denoted

δC3 and δT, respectively:10

±obs = ±C3xC3

+ ±TxT

= ±C3+ (±T ¡ ±C3

)xT

±obs = ±C3xC3

+ ±TxT

= ±C3+ (±T ¡ ±C3

)xT (4.2)

where xC3xC3

and xTxT denote the fractional populations for the two states, with xC3+ xT ´ 1xC3+ xT ´ 1.

Substitution and rearrangement leads to the following expression for xTxT :

xT =±obs ¡ ±C3

±T ¡ ±C3

xT =±obs ¡ ±C3

±T ¡ ±C3 (4.3)

With [RuA4⊂Ga4L6]11- and [RuA10⊂Ga4L6]

11- as systems in the T and C3 states, values

for δT = -0.116 ppm and δC3 = 1.01 ppm can be estimated from the methyl triplet in their

1H NMR spectra. From Equation 4.3, with δobs = 0.741 ppm, the [RuA6⊂Ga4L6]11- system

exists in the T state (retracted chain) for 24% of the time, and exists in the C3 state

(extended chain) the remaining 76% of the time. This corresponds to an equilibrium

constant K = [T]/[C3] = 1/3, which means the energy difference between the T and C3

states is about 0.6 kcal mol-1 (Figure 4.9).

The 2D NOESY spectrum of [RuA6⊂Ga4L6]11- at 27 °C shows a cross peak

between the guest’s terminal methyl group and one of the host’s catechol protons

(Appendix). Because the catechol protons point away from the interior of [Ga4L6]12-, this

confirms that, on average, the methyl group is located outside the host cavity. Similar

NOE interactions were observed with the RuCn zwitterions, where a host catechol

resonance showed cross peaks with signals from the unencapsulated portion of the

guest’s alkylsulfonate chain.3 In contrast, fully encapsulated guests such as RuA4+ and

Et4N+ only show NOE cross peaks with the three naphthalene signals.6

- 76 -

The Second-Order Jahn Teller Effect in [RuA6⊂Ga4L6]11-

The preceding observations demonstrated that the [RuA6⊂Ga4L6]11- system

rapidly oscillates between four degenerate C3 ground states through a T-symmetric

intermediate at room temperature (Figure 4.9). The reaction mechanism consequences of

such a dynamical system have been described by Pearson as based on the second-order

Jahn-Teller effect.11 The original Jahn-Teller theorem applies to specific types of

degenerate electronic systems.12 Chemists are generally familiar with the first-order

Jahn-Teller effect explaining, for example, the elongated axial bond distances in pseudo-

octahedral copper(II) complexes.13 An often cited example of a dynamic second-order

Jahn-Teller distortion is XeF6 in the gas phase.14 Its sterically active lone pair distorts

one face of the octahedron to yield a C3v ground state, yet is in dynamic exchange

through an essentially octahedral intermediate on its way to the other seven degenerate

C3v structures, with the lone pair occupying the other seven faces in the octahedron.15

Most studies concerning second-order Jahn-Teller effects involve bond distortions or

ruptures within a single covalent network.16 However, the fluxional structure for

[RuA6⊂Ga4L6]11- described above represents the first reported example of a second-order

Jahn-Teller distortion in a supramolecular host-guest system.4

Following the description of Pearson, after a displacement q = Q – Q0, the

Hamiltonian may be expanded as a Taylor-Maclaurin series as a function of the reaction

coordinate Q:

H = H0 +

µ

@U

@Q

¶

q +1

2

µ

@2U

@Q2

¶

q2 + ¢ ¢ ¢H = H0 +

µ

@U

@Q

¶

q +1

2

µ

@2U

@Q2

¶

q2 + ¢ ¢ ¢

(4.4)

with U equal to the potential energy of the system. From second-order perturbation

theory, the energy E becomes:

- 77 -

E = E0 +

¿

Ã0

¯

¯

¯

¯

@U

@Q

¯

¯

¯

¯

Ã0

À

q +

¿

Ã0

¯

¯

¯

¯

@2U

@Q2

¯

¯

¯

¯

Ã0

À

q2

2+

X

k>0

D

Ã0

¯

¯

¯

@U@Q

¯

¯

¯Ãk

E2

q2

E0 ¡ Ek

E = E0 +

¿

Ã0

¯

¯

¯

¯

@U

@Q

¯

¯

¯

¯

Ã0

À

q +

¿

Ã0

¯

¯

¯

¯

@2U

@Q2

¯

¯

¯

¯

Ã0

À

q2

2+

X

k>0

D

Ã0

¯

¯

¯

@U@Q

¯

¯

¯Ãk

E2

q2

E0 ¡ Ek

(4.5)

where E0 is the original energy at point Q0 (q = 0), Ã0Ã0 is the electronic wave function for

the ground state, and ÃkÃk is the wave function for the kth excited state (k = 0, 1, 2, …).11

Although Figure 4.9 is a cartoon diagram of the energy of the host-guest complex,

it serves to illustrate Equation 4.5. At maxima or minima, the slope (@U=@Q) = 0(@U=@Q) = 0, and

the term linear in q vanishes. For Q0 defined as the C3 ground state with a fully extended

chain, E = E0E = E0, and the force constant (@2U=@Q2) > 0(@2U=@Q2) > 0. When Q is at the transition state,

with the chain partially retracted, E > E0E > E0, and the force constant (@2U=@Q2) < 0(@2U=@Q2) < 0. When

Q is at T, with the chain fully retracted inside the cavity, ET > E0ET > E0 by 0.6 kcal mol-1, but

now (@2U=@Q2) > 0(@2U=@Q2) > 0 because the system is at a local minimum. The last term in

Equation 4.5 is negative, and mixing excited states with Ã0Ã0 lowers the energy when Q

approaches T from the C3 transition state. These are the descriptors of a second-order

Jahn-Teller distortion.

At low temperatures, the system does not have enough thermal energy to

sufficiently populate the appropriate excited states to overcome the larger activation

barrier ¢E1¢E1 (defined in Figure 4.9), and the guest’s alkyl chain remains extruded through

a single face. At higher temperatures, there is a greater statistical population of excited

electronic states ÃkÃk, permitting distortion to the transition state towards the T-symmetric

intermediate. The system can easily overcome the smaller activation barrier ¢E2¢E2 to

return to one of the C3 ground states. At 50 °C, this distortion is very fast, allowing the

system to rapidly interconvert between the four degenerate C3 states through the T

intermediate state, leading to the T time-averaged symmetry on the NMR timescale.

- 78 -

Summary

The development of the novel partial guest encapsulation mode illustrates the

power of rational design. Once the nondissociative guest egress mechanism for the M4L6

host was established, a set of simple design principles could be developed to take

advantage of its predictions. By deliberately creating the partial guest encapsulation

mode, complex interactions could be studied and carefully controlled simply by changing

the length and terminus of a linear hydrocarbon. The flexible apertures of the M4L6

tetrahedron might be used for future applications such as controlling catalytic polymer

growth to completely prevent branching, or linking two M4L6 clusters to form a

supramolecular “dumbbell.”

Experimental

General considerations

All reagents were obtained from commercial suppliers and used without further

purification unless noted otherwise. Standard Schlenk techniques were used for reactions

carried out under argon, and a glove box continuously purged with nitrogen was used to

store the moderately air sensitive cluster. When necessary, solvents for reactions were

degassed by at least six pump/fill cycles while vigorously stirring, backfilling with argon.

H4L (H4L = 1,5-bis(2,3-dihydroxybenzamido)naphthalene) and K12[Ga4L6] were

synthesized according to literature procedure.7, 8 NMR spectra of host-guest complexes

were measured on a Bruker AV-500 spectrometer with a TBI-P probe featuring a high-

sensitivity inner 1H coil. Chemical shifts δ vs. SiMe4 are listed for 1H NMR spectra, and

vs. CFCl3 for 19F NMR spectra. Routine mass spectrometry and elemental analysis was

- 79 -

performed by the Mass Spectrometry Laboratory and Microanalysis Facility in the

College of Chemistry at the University of California, Berkeley. High resolution TOF

electrospray mass spectra (ESI-MS) were recorded on a Waters QTOF API mass

spectrometer equipped with a Z-spray source, located either at the UC Berkeley Mass

Spectrometry facility (RuAn+ complexes) or the Waters Corporation, Dublin, CA (RuCn

complexes).

Guest synthetic procedures

1-bromo-4-phenylbutane (C4Br). A stirred solution of 1.0 g (6.7 mmol) of 4-phenyl-1-

butanol and 3 mL of pyridine in 75 mL of CH2Cl2 was cooled to -10 °C in an ice/salt

bath. An addition funnel containing 6 g (20 mmol) of phosphorous tribromide dissolved

in 25 mL of CH2Cl2 was affixed to the reaction vessel, and this solution was added

dropwise to the stirred solution over a 30 min period. The reaction mixture was stirred

for 15 h at room temperature, filtered to remove an orange solid, then washed with dilute

brine (2 x 300 mL), dilute sulfuric acid (250 mL), 1 M hydrochloric acid (2 x 250 mL),

and concentrated brine (250 mL). The organic fraction was collected, dried over MgSO4,

and the solvent removed via rotary evaporation, yielding a yellowish oil. Purification via

chromatography (basic alumina, CH2Cl2) and removal of solvent yielded 0.50 g (36%) of

a clear oil. 1H NMR (CDCl3, 400 MHz): δ 1.77 (m, 2H), 1.9 (m, 2), 2.65 (t, 2, J=7.6 Hz),

3.42 (t, 2, J=6.8 Hz), 7.2 (m, 5).

Sodium(4-phenylbutane-1-sulfonate) (C4SO3Na).17 A solution of 1.0 g (8.0 mmol) of

sodium sulfite in 20 mL of H2O was added to 0.50 g (2.3 mmol) of C4Br. The reaction

- 80 -

mixture was heated at reflux for 24 h, cooled to room temperature, and filtered to collect

the white crystals, washing once with cold H2O. The volume of the filtrate was reduced,

cooled to 4 °C, and the white crystalline solid was collected on a frit and washed with

cold H2O, then dried for 24 h in vacuo at 50 °C to yield 570 mg (96%) of white solid.

Anal. Calc. (found) for C10H13NaO3S: %C, 50.84 (50.45); H, 5.55 (5.31). MS(FAB+):

m/z 237 ([MNaH]+), 259 ([MNa2]+). 1H NMR (D2O, 300 MHz): δ 1.73 (m, 4H), 2.67 (t,

J=7.1 Hz, 2H), 2.93 (t, J=7.5 Hz, 2H), 7.3 (m, 5H).

1-bromo-6-phenylhexane (C6Br). A stirred solution of 2.52 g (14.1 mmol) of 6-phenyl-

1-hexanol in 80 mL of Et2O was cooled to 0 °C in an ice bath. An addition funnel

containing 4 g (20 mmol) of phosphorous tribromide in 25 mL of Et2O was affixed to the

reaction vessel, and this solution was added dropwise to the stirred solution over a 30 min

period. The reaction mixture was stirred for 15 h, allowing the ice bath to gradually

return to room temperature. The ether solution was treated with aqueous sodium

bicarbonate (2 x 250 mL) followed by brine (100 mL). The organic layer was collected,

dried over MgSO4, and the solvent removed via rotary evaporation to yield a cloudy,

colorless oil. Purification via chromatography (basic alumina, CH2Cl2) and removal of

solvent yielded 1.16 g (34%) of a clear oil. 1H NMR (CDCl3, 400 MHz): δ 1.39 (m, 2H),

1.49 (m, 2H), 1.66 (m, 2H), 1.88 (m, 2H), 2.64 (t, J = 7.6 Hz, 2H), 3.42 (t, J = 6.8 Hz,

2H), 7.21 (m, 3H), 7.31 (m, 2H).

Sodium(6-phenylhexane-1-sulfonate) (C6SO3Na).17 A solution of 2.1 g (15 mmol) of

sodium sulfite in 50 mL of H2O/EtOH (4:1) was added to 1.16 g (4.81 mmol) of C6Br.

- 81 -

The reaction mixture was heated at reflux for 24 h, cooled to 4 °C, and filtered to collect

the white crystals. The solid was dried for 24 h in vacuo at 50 °C to yield 1.2 g (96%) of

white solid. Anal. Calc. (found) for C12H17NaO3S·(H2O)0.5: %C, 52.73 (52.21); H, 6.64

(6.28). MS(ES-): m/z 241.1 (M-), 483.3 ([HM2]-), 505.2 ([NaM2]

-). 1H NMR (MeOD,

300 MHz): δ 1.39 (m, 4H), 1.62 (m, 2), 1.77 (m, 2), 2.60 (t, 2, J=7.8 Hz), 2.76 (t, 2,

J=8.1 Hz), 7.17 (m, 5).

1-bromo-8-phenyloctane (C8Br). A stirred solution of 2.87 g (13.9 mmol) of 8-phenyl-

1-octanol in 75 mL of anhydrous Et2O was cooled to 0 °C under air in an ice/salt bath. A

solution of 3 g (10 mmol) of phosphorous tribromide in 15 mL of Et2O was added

dropwise over 30 min, and the reaction mixture was stirred for 90 additional min at 0 °C.

The reaction mixture was treated with aqueous NaHCO3, the organic layer collected and

dried with MgSO4, and the solvent removed to yield a cloudy colorless oil. Purification

via chromatography (basic alumina, CH2Cl2) and removal of solvent yielded 1.0 g (27%)

of a clear, colorless oil. 1H NMR (CDCl3, 400 MHz): δ 1.38 (br s, 8H), 1.65 (m, 4H),

2.66 (t, 2H, J=7.6 Hz), 3.46 (t, 2H, J=6.8 Hz), 7.3 (m, 5H).

Sodium(8-phenyloctane-1-sulfonate) (C8SO3Na).17 To 1 g of C8Br was added a

solution of 1.5 g (12 mmol) of sodium sulfite dissolved in 40 mL of H2O with 10 mL of

ethanol. The reaction mixture was heated at reflux under air for 24 hours, then allowed to

cool to room temperature. A white polycrystalline solid formed, which was collected on

a frit and washed with cold H2O (2 x 5 mL and isopropyl alcohol (2 x 20 mL), then dried

overnight to yield 0.50 g (46%) of shiny white flakes. Anal. Calc. (found) for

- 82 -

C14H21NaO3S: %C, 57.51 (57.15); H, 7.24 (7.41); S, 10.97 (10.90). MS(ES-): m/z 269.1

(M-). 1H NMR (D2O, 300 MHz): δ 1.3 (m, 8H), 1.57 (m, 2H), 1.67 (m, 2H), 2.59 (t,

J=7.8 Hz, 2H), 2.84 (t, J=8.0 Hz, 2H), 7.3 (m, 5H).

1-bromo-10-phenyldecane (C10Br). A stirred solution of 2.56 g (10.9 mmol) of 10-

phenyl-1-decanol in 75 mL of anhydrous Et2O was cooled to 0 °C under air in an ice

bath. An addition funnel containing 3 g (10 mmol) of phosphorous tribromide in 25 mL

of Et2O was affixed to the reaction vessel, and this solution was added dropwise to the

stirred solution over a 30 min period. The reaction mixture was stirred for 90 min, then

treated with aqueous sodium bicarbonate (250 mL). The organic layer was collected,

dried over MgSO4, and its solvent removed to yield a cloudy, colorless oil. Purification

via chromatography (silica, CH2Cl2) yielded 0.85 g (25%) of a colorless oil. 1H NMR

(CDCl3, 400 MHz): δ 1.30 (br s, 10H), 1.43 (m, 2H), 1.62 (m, 2H), 1.86 (m, 2H), 2.615

(t, 2H, J=7.6 Hz), 3.421 (t, 2H, J=6.8 Hz), 7.24 (m, 5H).

Sodium(10-phenyldecane-1-sulfonate) (C10SO3Na).17 To 0.84 g (2.8 mmol) of C10Br

was added a solution of 1.2 g (8.4 mmol) of sodium sulfite dissolved in 35 mL of H2O

with 10 mL of ethanol. The reaction mixture was heated at reflux under air for 24 h, then

allowed to cool to room temperature. A white polycrystalline solid formed, which was

collected on a frit and washed with 10 mL of cold H2O, then for two days at 60 °C in a

vacuum oven to yield 0.76 g (84%) of shiny white flakes. Anal. Calc. (found) for

C16H25NaO3S: %C, 59.97 (60.16); H, 7.86 (7.73). MS(ES-): m/z 297.2 (M-). 1H NMR

- 83 -

(MeOD, 400 MHz): 1.35 (m, 12H), 1.60 (m, 2H), 1.78 (m, 2H), 2.59 (t, J=7.6 Hz, 2H),

2.78 (t, J=8.0 Hz, 2H), 7.2 (m, 5H).

[Ru(η5-C5H5)(η6-C6H5(CH2)10SO3)]·NaPF6 (RuC10·NaPF6). This reaction was

performed under argon using standard Schlenk techniques. To 160 mg (0.49 mmol) of

C10SO3Na and 200 mg (0.46 mmol) of [Ru(η5-C5H5)(CH3CN)3]PF6 was added 100 mL of

anhydrous CHCl3. The reaction mixture was heated at reflux for 24 h, during which time

a brown solid appeared on the walls of the flask. The solvent was removed, and the light

brown residue was dissolved in methanol, filtered, and precipitated with ether to yield

150 mg (52%) of beige solid. Anal. Calc. (found) for C21H30F6NaO3PRuS: %C, 39.94

(39.05); H, 4.79 (4.73); S, 5.08 (4.82). MS(ES+): m/z 465.1 ([MH]+), 487.1 ([MNa]+).

1H NMR (MeOD, 500 MHz): δ 1.34 (s, 6H), 1.40 (m, 6H), 1.63 (m, 2H), 1.78 (m, 2H),

2.53 (t, J=8.0 Hz, 2H), 2.78 (t, J=8.0 Hz, 2H), 5.41 (s, 5H), 6.13 (m, 1H), 6.17 (m, 2H),

6.23 (d, J=6.0 Hz, 2H). 1H NMR (D2O, 300 MHz): δ 1.32 (m, 12H), 1.58 (m, 2H), 1.71

(m, 2H), 2.48 (t, J=7.8 Hz, 2H), 2.87 (t, J=8.0 Hz, 2H), 5.33 (s, 5H), 6.07 (m, 3H), 6.14

(m, 2H). 19F NMR (MeOD, 376 MHz): δ -74.0 ppm vs. CFCl3 (d, J=709 Hz).

[Ru(η5-C5H5)(η6-C6H5(CH2)4SO3)]·NaPF6 (RuC4·NaPF6). A procedure similar to the

synthesis of RuC10·NaPF6 was used, with 127 mg (0.278 mmol) of [Ru(η5-

C5H5)(CH3CN)3]PF6 and 71 mg (0.28 mmol) of C4SO3Na in place of C10SO3Na.

Recrystallization from methanol/ether yielded 60 mg (40%) of beige powder. Anal.

Calc. (found) for C15H18F6NaO3PRuS: %C, 32.91 (32.35); H, 3.31 (3.55); S, 5.86 (5.50).

MS(ES+): m/z 381.1 ([MH]+), 403.1 ([MNa]+). 1H NMR (D2O, 500 MHz): δ 1.74 (m,

- 84 -

2H), 1.81 (m, 2H), 2.55 (t, J=7.7 Hz, 2H), 2.94 (t, J=7.5 Hz, 2H), 5.35 (s, 5H), 6.08 (m,

3H), 6.18 (d,J=5.5 Hz, 2H).

[Ru(η5-C5H5)(η6-C6H5(CH2)6SO3)]·NaPF6 (RuC6·NaPF6). A procedure similar to the

synthesis of RuC10·NaPF6 was used, with 215 mg (0.495 mmol) of [Ru(η5-

C5H5)(CH3CN)3]PF6 and 140 mg (0.50 mmol) of C6SO3Na in place of C10SO3Na.

Recrystallization from methanol/ether yielded 150 mg (53%) of beige powder. Anal.

Calc. (found) for C17H22F6NaO3PRuS: %C, 35.48 (35.30); H, 3.85 (4.15); S, 5.57 (5.29).

MS(ES+): m/z 409.1 ([MH]+), 431.1 ([MNa]+). 1H NMR (D2O, 400 MHz): δ 1.43 (m,

4H), 1.61 (m, 2H), 1.72 (m, 2H), 2.50 (t, J=7.8 Hz, 2H), 2.89 (t, J=8.0 Hz, 2H), 5.34 (s,

5H), 6.06 (m, 3H), 6.16 (d, J= 5.6 Hz, 2H).

[Ru(η5-C5H5)(η6-C6H5(CH2)8SO3)]·NaPF6 (RuC8·NaPF6). A procedure similar to the

synthesis of RuC10·NaPF6 was used, with 185 mg (0.425 mmol) of [Ru(η5-

C5H5)(CH3CN)3]PF6 and 125 mg (0.425 mmol) of C8SO3Na in place of C10SO3Na.

Recrystallization from methanol/ether yielded 170 mg (66%) of beige powder. Anal.

Calc. (found) for C19H26F6NaO3PRuS: %C, 37.81 (37.13); H, 4.34 (4.56); S, 5.31 (5.28).

MS(ES+): m/z 437.1 ([MH]+), 459.1 ([MNa]+). 1H NMR (D2O, 400 MHz): δ 1.38 (m,

8H), 1.60 (m, 2H), 1.71 (m, 2H), 2.49 (t, J=7.8 Hz, 2H), 2.88 (t, J=8.0 Hz, 2H), 5.34 (s,

5H), 6.07 (m, 3H), 6.16 (d, J=5.6 Hz, 2H). 19F NMR (MeOD, 376 MHz): δ -71.3 ppm

vs. CFCl3 (d, J=709 Hz).

- 85 -

[(cyclopentadienyl)(n-phenylbutane)ruthenium(II)]hexafluorophosphate

([RuA4]PF6). The following reaction was carried out under argon. To 95 mg

(0.22 mmol) of [CpRu(CH3CN)3]PF6 (Strem, 98%) was added 50 mL anhydrous

chloroform (Aldrich, 99%, stabilized with amylenes) via cannula, then 70 µL (60 mg,

0.5 mmol) of n-butylbenzene (TCI) was added to the orange solution via syringe. An

additional 5 mL of chloroform was added to wash any stray reactant from the walls of the

flask. After heating the reaction mixture at 75 °C for 24 h, the solvent was removed via

rotary evaporation. The residue was re-dissolved under air in 15 mL of dichloromethane,

filtered through Celite, and after washing the Celite plug with ~100 mL of

dichloromethane, the solvent was removed again via rotary evaporation. The residue was

re-dissolved in a minimum volume (~5 mL) of fresh dichloromethane, and 50 mL of

diethyl ether was added to this yellow-brown solution to form a precipitate. The solid

was collected on a frit, washed with diethyl ether (3 x 15 mL) and petroleum ether

(2 x 10 mL), and dried in vacuo for 6 h to yield 70 mg (72%) of off-white powder. Anal.

Calc. (found) for C15H19F6PRu: %C, 40.45 (40.35); H, 4.30 (4.44). MS (ESI+, MeOH):

m/z 301.1 (RuA4+). 1H NMR (CDCl3, 400 MHz): δ 0.98 (t, 3H, J=7.2 Hz), 1.44 (m, 2H),

1.6 (m, 2H), 2.54 (t, 2H, J=7.6 Hz), 5.40 (s, 5H), 6.15 (m, 5H).

[(cyclopentadienyl)(n-phenylhexane)ruthenium(II)]hexafluorophosphate

([RuA6]PF6). The following reaction was carried out under argon. To 98 mg

(0.22 mmol) of [CpRu(CH3CN)3]PF6 was added 40 mL of anhydrous chloroform and

150 µL (130 mg, 0.80 mmol) of n-hexylbenzene. After stirring at room temperature for

2 days, a mustard yellow solution with a ring of dark brown solid was present in the

- 86 -

reaction vessel. The reaction mixture was filtered through Celite under air, and after

thoroughly washing the Celite plug with 150 mL of dichloromethane, the solvent was

removed via rotary evaporation, leaving a dark, viscous, mustard colored oil in the flask.

This oil was dissolved in 3 mL of dichloromethane, and addition of 50 mL of diethyl

ether led to phase separation after 20 min, with droplets of mustard colored oil at the

bottom of the flask. After adding an additional 50 mL of diethyl ether, the biphasic

mixture was vigorously stirred for 45 min to afford a light-colored solid precipitate. The

solid was collected on a medium frit, washed with diethyl ether (3 x 15 mL), and dried in

vacuo for 4 h to yield 55 mg (52%) of pale tan solid. Anal. Calc. (found) for

C17H23F3PRu: %C, 43.13 (42.83); H, 4.90 (4.79). MS (ESI+, MeOH): m/z 329.2

(RuA6+). 1H NMR (CDCl3, 300 MHz): δ 0.87 (t, 3H, J=6.6 Hz), 1.3 (m, 6H), 1.6 (m,

2H), 2.47 (t, 2H, J=7.8 Hz), 5.35 (s, 5H), 6.1 (m, 5H).

[(cyclopentadienyl)(n-phenyloctane)ruthenium(II)]hexafluorophosphate

([RuA8]PF6). The following reaction was carried out under argon. To 98 mg

(0.23 mmol) of [CpRu(CH3CN)3]PF6 was added 50 mL of anhydrous chloroform via

cannula, then 1.0 mL (0.86 g, 4.5 mmol) of n-octylbenzene was added to the orange

solution via syringe. After heating the reaction mixture at reflux for 18 h, the reaction

mixture was allowed to cool, filtered through Celite, and the plug was washed with

CH2Cl2 to obtain a yellow-brown solution. Solvent was removed via rotary evaporation,

leaving a biphasic oily residue: small yellow-brown drops within a clear oily liquid. The

residue was dissolved in 5 mL of CH2Cl2, and 50 mL of ether was added and stirred for

one hour, forming a silky precipitate. The powder was collected on a frit, washed with

- 87 -

ether (2 x 20 mL), and dried in vacuo for three days to yield 65 mg (58%) of off-white

powder. Anal. Calc. (found) for C19H27F6PRu: %C, 45.51 (45.39); H, 5.43 (5.36). MS

(ESI+, MeOH): m/z 357.1 (RuA8+). 1H NMR (CDCl3, 300 MHz): δ 0.87 (t, 3H,

J=6.6 Hz), 1.3 (m, 10H), 1.57 (m, 2H), 2.48 (t, 2H, J=7.8 Hz), 5.36 (s, 5H), 6.1 (m, 5H).

[(cyclopentadienyl)(n-phenyldecane)ruthenium(II)]hexafluorophosphate

([RuA10]PF6). The title compound was prepared using the same procedure described for

[RuA8]PF6, with 100 mg (0.23 mmol) of [CpRu(CH3CN)3]PF6, 0.40 mL (0.34 g,

1.6 mmol) of n-phenyldecane, and 40 mL of anhydrous CHCl3. Recrystallization of the

product from CH2Cl2/ether yielded 50 mg (41%) of off-white powder. Anal. Calc.

(found) for C21H31RuPF6: %C, 47.63 (47.32); H, 5.90 (5.88). MS (ESI+, MeOH):

m/z 385.1 (RuA10+). 1H NMR (CDCl3, 400 MHz): δ 0.86 (t, 3H, J=6.2 Hz), 1.24 (m,

12H), 1.32 (m, 2H), 1.56 (m, 2H), 2.45 (t, 2H, J=7.8 Hz), 5.34 (s, 5H), 6.05 (m, 2H), 6.11

(m, 3H).

[(cyclopentadienyl)(n-phenyldodecane)ruthenium(II)]hexafluorophosphate

([RuA12]PF6). The following reaction was carried out under argon. To 99 mg

(0.23 mmol) of [CpRu(CH3CN)3]PF6 was added 40 mL of anhydrous 1,2-dichloroethane

via cannula, then 0.3 mL (0.3 g, 1 mmol) of n-dodecylbenzene was added to the orange

solution via syringe. The reaction mixture was heated at reflux for 5 h, then removed

from heat and stirred for an additional 12 h at room temperature. The brown reaction

mixture was filtered through Celite under air, and after thoroughly washing the Celite

plug with 100 mL of CH2Cl2, the solvent was removed via rotary evaporation. The

- 88 -

yellow-brown residue was dissolved in 3 mL of CH2Cl2, and addition of approximately

40 mL of diethyl ether afforded a light-colored precipitate. The solid was collected on a

fine frit, washed with diethyl ether (3 x 5 mL) and petroleum ether (2 x 15 mL), and dried

in vacuo overnight to yield 65 mg (50%) of off white powder. Anal. Calc. (found) for

C23H35RuPF6: %C, 49.55 (49.38); H, 6.33 (6.48). MS (FAB+): m/z 413 (RuA12+). 1H

NMR (CDCl3, 400 MHz): δ 0.86 (t, 3H, J=7.2 Hz), 1.23 (m, 16H), 1.32 (m, 2H), 1.56 (m,

2H), 2.46 (t, 2H, J=8.0 Hz), 5.34 (s, 5H), 6.06 (m, 2H), 6.12 (m, 3H).

Host-guest complex synthetic procedures

K12[RuC10⊂⊂⊂⊂Ga4L6]·(Me2CO)·7H2O. This reaction was performed under argon using

standard Schlenk techniques. A suspension of 102 mg (0.237 mmol) of H4L, 58 mg

(0.16 mmol) of Ga(acac)3, and 25 mg (0.040 mmol) of RuC10·NaPF6 in 75 mL of MeOH

was heated at reflux for 12 h. The solvent was removed to leave a light tan residue.

After drying this residue in vacuo for three hours, 50 mL of MeOH was added. Addition

of 0.95 mL (0.47 mmol) of methanolic KOH (0.5 M) caused the off-white suspension to

become a yellow solution. The reaction mixture was re-degassed with three pump/fill

cycles immediately after the addition of base, then stirred at room temperature for 2 h.

Undissolved solid was removed by cannula filtration. The volume was reduced to 5 mL,

and 150 mL of acetone was added to precipitate a pale yellow/brown solid. This solid

was collected on a frit under a stream of nitrogen, washed with acetone (4 x 10 mL), and

dried in vacuo overnight to yield 120 mg (80%) of yellow-beige powder. Anal. Calc.

(found) for C165H114Ga4K12N12O39RuS·Me2CO·7H2O: %C, 51.03 (51.01); H, 3.42 (3.29);

N, 4.25 (4.13). MS(ES-): (see text). 1H NMR (D2O, 500 MHz): δ -1.20 (m, 1H), -1.13

- 89 -

(br m, 1H), -1.02 (br m, 1H), -0.91 (br m, 1H), -0.25 (m, 2H), 0.29 (m, 1H), 0.52 (m, 1H),

0.66 (m, 1H), 0.80 (m, 1H), 1.18 (m, 2H), 1.36 (m, 2H), 1.49 (m, 2H), 1.87 (m, 2H), 2.29

(s, 5H), 3.05 (m, 3H), 3.14 (d, J=6.0 Hz, 1H), 3.20 (d, J=5.9 Hz, 1H), 3.28 (t, J=5.8 Hz,

1H), 4.35 (t, J=5.6 Hz, 1H), 6.49 (t, J=7.8 Hz, 3H), 6.58 (t, J=7.8 Hz, 3H), 6.64 (m, 9H),

6.74 (d, J=7.3 Hz, 3H), 6.79 (m, 6H), 6.82 (t, J=8.2 Hz, 3H), 6.90 (t, J=8.0 Hz, 3H), 7.17

(m, 6H), 7.23 (d, J=8.3 Hz, 3H), 7.27 (t, J=8.3 Hz, 3H), 7.36 (d, J=8.3 Hz, 3H), 7.39 (d,

J=8.3 Hz, 3H), 7.47 (d, J=7.6 Hz, 3H), 7.76 (d, J=8.6 Hz, 3H), 7.79 (d, J=8.6 Hz, 3H),

7.80 (d, J=8.6 Hz, 3H), 7.87 (d, J=7.8 Hz, 3H), 8.08 (d, J=8.7 Hz, 3H), 8.11 (d, J=7.8 Hz,

3H), 8.37 (d, J=7.8 Hz, 3H). 19F NMR (D2O, 376 MHz): No signal observed.

NMR Experiments of Host-Guest Complexes

General considerations. NMR spectra of host-guest complexes were acquired using a

Bruker AV-500 spectrometer with a TBI-P probe featuring a high-sensitivity inner 1H

coil. All 2D NMR spectra were measured at a constant, controlled temperature. For the

2D gCOSY spectrum of [RuA6⊂Ga4L6]11- measured at -60 °C in MeOD, a 12 second

delay was used between FID acquisition and the next pulse due to the very long 1H

relaxation times at low temperature.

Sample preparation: RuCn guests. K12[Ga4L6] (10.0 mg, 2.97 µmol) and RuCn·NaPF6

(3 µmol) were combined in a vial and dissolved in 0.6 mL of D2O at room temperature.

The solution was filtered through a glass fiber plug and transferred to an NMR tube, and

the spectrum recorded 10 min after dissolution. Solutions were discarded within 24 hours

due to slow oxidation of the ligands in the cluster.

- 90 -

Sample preparation: RuAn+ guests. Aqueous samples of host-guest complexes for

NMR spectroscopy were prepared using the same method, so the procedure used for

[RuA6⊂Ga4L6]11- is given as a representative example. In a threaded glass vial, 30.0 mg

(8.62 µmol) of K12[Ga4L6]·2(Me2CO) and 4.08 mg (8.62 µmol) of [RuA6]PF6 were

combined, corresponding to equimolar proportions. To this was added 2.0 mL of D2O,

and the mixture was sonicated for approximately 2 min to aid solvation. The clear yellow

solution was passed through a 0.2 µm nylon disk filter, and 0.5 mL of the filtrate was

transferred to a medium wall glass NMR tube. The solution was degassed in the tube via

three freeze-pump-thaw cycles, and flame sealed under vacuum. The sample was

allowed to equilibrate at room temperature for at least 12 h before measurements.

The NMR sample of [RuA6⊂Ga4L6]11- in MeOD used for low temperature studies

was prepared in a similar fashion, with 15.0 mg (4.31 µmol) of K12[Ga4L6]·2(Me2CO)

and 2.04 mg (4.31 µmol) of [RuA6]PF6 dissolved in 0.7 mL of MeOD after brief

sonication (< 1 min). The clear yellow solution was filtered through a 0.2 µm nylon disk

filter into a thin wall glass NMR tube. The solution was degassed and sealed under

vacuum as above, and allowed to equilibrate overnight at room temperature before use.

Spectra of [RuA6⊂Ga4L6]11- in D2O and MeOD were nearly identical, showing the same

host signal broadening behavior between 1 °C and 40 °C.

- 91 -

Mass spectrometry: Sample preparation

Samples for ESI-MS were prepared by combining equimolar amounts of RuCn·NaPF6 or

[RuAn]PF6 and K12[Ga4L6] in methanol or H2O/methanol, except for [RuC10⊂Ga4L6]12-

where the pre-synthesized host-guest complex was used.

References

1. Davis, A. V.; Raymond, K. N., “The Big Squeeze: Guest Exchange in an M4L6 Supramolecular Host.” J. Am. Chem. Soc. 2005, 127, 7912-7919.

2. Davis, A. V.; Fiedler, D.; Seeber, G.; Zahl, A.; van Eldik, R.; Raymond, K. N., “Guest Exchange Dynamics in an M4L6 Tetrahedral Host.” J. Am. Chem. Soc. 2006, 128, 1324-1333.

3. Tiedemann, B. E. F.; Raymond, K. N., “Dangling Arms: A Tetrahedral Supramolecular Host with Partially Encapsulated Guests.” Angew. Chem. Int. Ed.

2006, 45, 83-86.

4. Tiedemann, B. E. F.; Raymond, K. N., “Second-Order Jahn-Teller Effect in a Host-Guest Complex.” Angew. Chem. Int. Ed. 2007, in press.

5. Fiedler, D.; Pagliero, D.; Brumaghim, J. L.; Bergman, R. G.; Raymond, K. N., “Encapsulation of Cationic Ruthenium Complexes into a Chiral Self-Assembled Cage.” Inorg. Chem. 2004, 43, 846-848.

6. Caulder, D. L.; Brückner, C.; Powers, R. E.; König, S.; Parac, T. N.; Leary, J. A.; Raymond, K. N., “Design, Formation, and Properties of Tetrahedral M4L4 and M4L6 Supramolecular Clusters.” J. Am. Chem. Soc. 2001, 123, 8923-8938.

7. Caulder, D. L.; Powers, R. E.; Parac, T. N.; Raymond, K. N., “The Self-Assembly of a Predesigned Tetrahedral M4L6 Supramolecular Cluster.” Angew. Chem. Int. Ed.

1998, 37, 1840-1842.

8. Parac, T. N.; Caulder, D. L.; Raymond, K. N., “Selective Encapsulation of Aqueous Cationic Guests into a Supramolecular Tetrahedral M4L6 Anionic Host.” J. Am.

Chem. Soc. 1998, 120, 8003-8004.

9. Scarso, A.; Trembleau, L.; Rebek, J., Jr., “Helical Folding of Alkanes in a Self-Assembled, Cylindrical Capsule.” J. Am. Chem. Soc. 2004, 126, 13512-13518.

10. Macomber, R. S., A Complete Introduction to Modern NMR Spectroscopy. John Wiley & Sons: New York, 1998; pp 164-165.

- 92 -

11. a) Pearson, R. G., Symmetry Rules for Chemical Reactions. John Wiley and Sons: New York, 1976; pp 12-25; b) Pearson, R. G., “The Second-Order Jahn-Teller Effect.” Journal of Molecular Structure - Theochem 1983, 103, 25-34.

12. Jahn, H. A.; Teller, E., “Stability of Polyatomic Molecules in Degenerate Electronic States. I. Orbital Degeneracy.” Proc. R. Soc. London, A 1937, 161, 220-235.

13. Cotton, F. A.; Wilkinson, G.; Murillo, C. A.; Bochmann, M., Advanced Inorganic

Chemistry. Sixth ed.; John Wiley and Sons: New York, 1999; p 865.

14. Bartell, L. S., “Evidence for Pseudo-Jahn-Teller Effect in XeF6.” J. Chem. Phys.

1967, 46, 4530-4531.

15. Pitzer, K. N.; Bernstein, L. N., “Molecular Structure of XeF6.” J. Chem. Phys. 1975, 63, 3849-3856.

16. Pearson, R. G., “Concerning Jahn-Teller Effects.” Proc. Nat. Acad. Sci. USA 1975, 72, 2104-2106.

17. Truce, W. E.; Hoerger, F. D., “The Chemistry of Sultones. II. Alkylation of Organometallic and Related Compounds by Sultones.” J. Am. Chem. Soc. 1955, 77, 2496-2500.

- 93 -

CHAPTER 5

Electrochemical Properties of Cations Encapsulated by an M4L6 Host

Introduction

Supramolecular chemistry relies on labile interactions to assemble intricate

structures from relatively simple components.1 Large, discrete molecular systems may be

designed and synthesized, sometimes exhibiting rich functionality such as guest binding

equilibria.2 While X-ray crystallography and NMR are often the preferred analytical

tools used to characterize supramolecular assemblies, electrochemical methods offer

unique advantages for investigating supramolecular systems with redox-active

components.3 Supramolecular electrochemistry goes beyond the typical redox activity

encountered with most unimolecular covalent species, involving complex behavior such

as electrochemical switching.4, 5 Direct modulation of host-guest binding properties

through electron transfer events can be an elegant means for applications ranging from

chemical sensing to electrochemically driven self-assembly reactions.3, 6, 7 The work

described in this chapter explores the electrochemical behavior of redox-active guests

encapsulated by an M4L6 cluster assembled with redox-silent M.

Ferrocene (Fc) and cobaltocenium (CoCp2+) are popular redox-active guests for

electrochemical binding studies with redox-silent hosts, since they typically show

reversible redox behavior. Cationic sandwich complexes such as CoCp2+ can be

encapsulated by the [Ga4L6]12- host, with binding constants exceeding 104 M-1 in some

cases.8, 9 Reduction of GaIII and naphthalene require extremely negative potentials,

- 94 -

making [Ga4L6]12- electrochemically inert at potentials below the onset of ligand

oxidation. If a guest is electrochemically active within this window, is it possible to

reduce a monocationic guest while encapsulated within the host? To address this

question, cyclic voltammetry experiments were carried out in collaboration with Burak

Ulgut and Prof. Héctor D. Abruña at Cornell University, with CoCp2+,

decamethylcobaltocenium (CoCp*2+), and decamethylferrocenium (FeCp*2

+) chosen as

electroactive species encapsulated by [Ga4L6]12-. To support the electrochemical data, the

binding constant of CoCp2+ with [Ga4L6]

12-, the diffusion coefficient of its host-guest

complex and the rate of encapsulation of CoCp*2+ by [Ga4L6]

12- were measured in DMF-

d7 solutions using 1H NMR methods.

At negative potentials associated with cobalt(III) reduction, the cationic guest

could not be reduced while inside the cavity, suggesting encapsulation by [Ga4L6]12-

rendered the guest electrochemically inert. Adsorption of the host-guest complex onto

the platinum working electrode occurs at more positive potentials associated with

iron(III) reduction, causing the observed response to be much different than the

analogous cobalt systems. Furthermore, exterior ion pairing of the unencapsulated

metallocenium cations with the [Ga4L6]12- anionic host affects both the current and

potential of the observed waves.

Results and Discussion

Synthesis and characterization of host-guest complexes

Solid samples of the 1:1 host-guest complexes were prepared by combining

equimolar quantities of K12[Ga4L6]·(Me2CO)3 and the hexafluorophosphate salt of the

- 95 -

appropriate cationic sandwich complex in methanol under argon, stirring overnight, and

evaporating the solvent in vacuo. The resulting 1H NMR spectra showed the

encapsulated CoCp2+, CoCp*2

+, and FeCp*2+ resonances shifted to significantly higher

fields relative to their unencapsulated values. This diagnostic upfield shift is a

consequence of the host cavity’s magnetically shielded environment, due to ring current

effects in the six surrounding naphthyl moieties.10 In contrast, the neutral FeII

counterparts do not show an appreciable affinity for the host interior. When K12[Ga4L6]

was combined with excess neutral Fc or FeCp*2 in DMF-d7, no peaks were observed for

an encapsulated guest, even after one week.

Figure 5.1. (Top) 1H NMR spectrum (300 MHz) of [FeCp*2]PF6 in acetone-d6 at 20 °C. (Bottom) 1H NMR spectrum (500 MHz) of [FeCp*2⊂Ga4L6]

11- in CD3OD at -20 °C, with an expansion of the central region shown as an inset. The broad resonance for the Cp* methyl protons is marked with a red asterisk in each spectrum.

- 96 -

The 1H NMR spectra for [CoCp2⊂Ga4L6]11- and [CoCp*2⊂Ga4L6]

11- match the

spectra from previous publications, and the signals from the host’s 72 aromatic protons

convey information about the overall symmetry of the host-guest complex.8, 11 Six

signals integrating to 12H each are observed for complexes with T symmetry, including

[CoCp2⊂Ga4L6]11-, but the very large size of CoCp*2

+ causes the host to bulge along one

of its 2-fold axes to accommodate the bulky guest, reducing the overall symmetry to D2

due to loss of the C3 axes. The resulting 1H NMR spectrum of a complex with D2

symmetry will display 18 aromatic resonances integrating to 4H each, observed as

overlapping signals for [CoCp*2⊂Ga4L6]12-. Despite its paramagnetism, the

encapsulation of FeCp*2+ by [Ga4L6]

12- is confirmed by the resulting 1H NMR spectrum

(Figure 5.1). The broad 30H guest resonance, typically observed at very high fields in

the absence of host due to paramagnetism, is shifted nearly 10 ppm further upfield upon

encapsulation. The aromatic host signals, which normally appear between 6 – 9 ppm, are

observed as 18 resonances spread from -3 to +28ppm, each integrating to 4H. (Two of

these resonances overlap to yield a single 8H peak). Interactions with the paramagnetic

FeIII complex are responsible for the wide chemical shift distribution of the aromatic

signals, but the observed splitting from 6 to 18 resonances is caused by a reduction of

symmetry from T to D2. This is exactly the same behavior observed for the isostructural

CoIII analogue, but the paramagnetic FeIII ion makes the splitting due to loss of host

symmetry much more pronounced.

- 97 -

UV-visible spectroscopy has a limited role for characterization of host-guest

complexes. The electronic spectrum of the [Ga4L6]12- host consists of very strong ligand

π-π transitions in the near UV region, with a tail extending through 420 nm responsible

for its yellow color. The much weaker CoIII d-d transitions occur in the same region as

the tail, and therefore cannot be readily observed. For the 17-electron ferrocenium

derivatives, there is a ligand to metal charge transfer (LMCT) band in the visible region

overlapping with two weaker d-d transitions at shorter wavelengths.12 UV-visible

spectroscopy was used to show whether or not the FeIII sandwich complex remained

intact when combined with [Ga4L6]12-. For [FeCp*2]PF6, λmax = 779 nm for the LMCT

band in methanol with no host. Upon encapsulation, this LMCT peak is slightly red

shifted, with λmax = 786 in the presence of host (Figure 5.2). The persistence of that

transition after encapsulation indicates the FeCp*2+ sandwich complex remains intact,

while the small red shift of the LMCT peak is due to a change in solvation environment

(from methanol to the hydrophobic host interior).

Figure 5.2. Visible absorption spectra (methanol, room temperature) showing d-d bands and a ligand to metal charge transfer transition for FeCp*2

+ (blue) and [FeCp*2⊂Ga4L6]11- (red).

500 600 700 800 900

0

100

200

300

400

500

600

786 nm

779 nm

ε ,

M-1cm

-1

Wavelength, nm

Free

Encapsulated

- 98 -

In contrast to the fully methylated derivative, when either Fc+ or Me2Fc+

(Me2Fc+ = 1,1’-dimethylferrocenium) is combined with [Ga4L6]12-, decomposition

occurs rather than encapsulation of the cationic sandwich complex, associated with a

color change and formation of black insoluble solid. This decomposition was monitored

for Me2Fc+ via UV-visible spectroscopy in degassed methanol, with the intensity of the

absorption at λmax = 649 nm diminishing in the presence of [Ga4L6]12-. The stronger

oxidizing abilities of Fc+ and Me2Fc+ relative to FeCp*2+ account for their different

reactivities observed with the host; oxidation of the catechol chelating groups in the host

ligands, accompanied by reduction of FeIII to FeII is the most likely decomposition

pathway.

Guest binding and diffusion measurements

The binding constant of CoCp2+ with [Ga4L6]

12- was measured in DMF-d7 from

1H NMR integration ratios. Four different samples containing 1:1 host-guest

stoichiometries (0.4 mM) were used for redundant data points: three prepared using stock

solutions of [CoCp2]PF6 and K12[Ga4L6]·(Me2CO)4 and one from the solid host-guest

sample used for electrochemical measurements. The binding constant

Kb = 3.3 ± 0.2 x 105 M-1 in DMF-d7 is over ten times higher than that measured in D2O

(1.6 x 104 M-1)13 for the same host-guest complex. Although binding constants have not

been quantified for other guests in DMF, competitive binding between CoCp2+ and Et4N

+

has been observed via NMR. When 1 equivalent of Et4NCl is added to a solution of

[CoCp2⊂Ga4L6]11- in DMF-d7 and allowed to equilibrate at room temperature, over 60%

- 99 -

of the CoCp2+ guest is displaced by Et4N

+.6 This indicates Et4N+ has a slightly higher

binding constant than CoCp2+ in DMF, which is also the case for aqueous solutions (for

Et4N+ binding in D2O, Kb = 1.96 x 104 M-1).14 Thus, changing the solvent from H2O to

DMF leads to very similar increases in the binding affinities for Et4N+ and CoCp2

+,

despite their very different structures, suggesting a general solvent dependence for Kb.

The observed solvent dependence of the binding constant can be partially

explained by the differences in the Born solvation energies. The dielectric constant of

DMF is 38, compared to 78 for H2O, so there is a lower guest desolvation energy cost in

DMF compared to H2O. By invoking a simple thermodynamic cycle to describe the

encapsulation process in the gas phase, solvation effects on the free energy of the

encapsulation reaction ∆Genc can be isolated. Using the Born theory of solvation, the free

energy difference ∆Genc(DMF) – ∆Genc(H2O) ≈ -5.6 kcal mol-1 (Appendix 2). This

estimate qualitatively agrees with experimental observations, predicting a higher binding

constant in DMF compared to H2O, although the energy difference calculated from the

Born model is three times larger than the energy difference calculated from the observed

binding constants.

In cyclic voltammetry measurements with solvated redox couples, the magnitude

of the observed current is proportional to the square root of the diffusion coefficient for

the electroactive species. To properly interpret changes in current, accurate values for

diffusion coefficients are needed. The diffusion coefficient DHG of [CoCp2 ⊂ Ga4L6]11- in

DMF-d7 with 11 mM Bu4NPF6 was measured to be DHG = 3.1 ± 0.1 x 10-6 cm2 s-1 at

6 The exterior Et4N

+ resonances were near the large solvent residual peaks in DMF-d7, preventing accurate binding constant measurements.

- 100 -

25 °C by pulsed gradient spin-echo (PGSE) diffusion NMR (Figure 5.3).7 This value was

measured with a host-guest concentration of 0.40 mM, with potassium counterions.

The binding affinity of the sterically demanding CoCp*2+ guest in DMF is similar

to that for the smaller CoCp2+, as indicated from the small amount of dissociation

observed in the 1H NMR spectrum of 0.4 mM [CoCp*2⊂Ga4L6]11- two weeks after

preparation. Similar diffusion coefficients are also expected for the two host-guest

complexes, since guest encapsulation has a very small effect on the effective radius.

However, the two cations will be encapsulated by the host at very different rates due to

their size differences, with slower kinetics expected for the bulky CoCp*2+. To assess the

upper limits of the time scale where significant guest encapsulation occurs, the rate of

encapsulation of CoCp*2+ by [Ga4L6]

12- in DMF-d7 was measured via 1H NMR

(Figure 5.4). The linear relationship of xf-1 vs. time, where xf is the mole fraction of free

7 For a detailed discussion of diffusion NMR techniques, see Chapter 2 in this dissertation.

Figure 5.3. PGSE 1H NMR diffusion decay curve for [CoCp2⊂Ga4L6]11- in DMF-d7 with 11 mM Bu4NPF6,

using data obtained from host and guest signals. The solid line was fit to the mean of the three decay curves shown.

0 20 40 60 80 100

0

0.2

0.4

0.6

0.8

1

% Gradient Strength

Norm

aliz

ed Inte

gra

l

Host amide Host aromaticsGuest Cp Simulated Fit

- 101 -

(unencapsulated) CoCp*2+, indicates the overall rate law is second order, with the

reaction proceeding via a bimolecular process:

−−+ ⊂→+ 11642

12642 ]LGa*[CoCp]L[Ga*CoCp 2k

The second order reaction rate constant k2 characterizing this reaction is 2.6 M-1 s-1. At

0.4 mM, this corresponds to a reaction half life of approximately 7 min.

Cyclic voltammetry

Cyclic voltammetry of solutions of the free metallocenes in DMF led to reversible

or quasireversible waves for the MIII/II redox couple (Table 5.1). Both cobalt complexes

showed reversible waves for the CoIII/II couples, with E1/2(CoCp*2) occurring at 0.55 V

more negative potentials than the unsubstituted CoCp2+/0 wave. A second quasireversible

wave appeared at very negative potentials with CoCp2+, associated with the CoII/I redox

couple. The larger ∆Ep values associated with the ferrocene derivatives seem to suggest

the FeIII/II redox processes are quasireversible, particularly for FeCp*2. Values reported

0 10 20 30 400.1

0.2

0.3

0.4

0.5

0.6

0.7

Mo

le f

ractio

n F

ree

Co

Cp

* 2

+

Time, minutes

Figure 5.4. Kinetic trace obtained by monitoring the integrated 1H NMR signal from free CoCp*2+

diminishing due to encapsulation by [Ga4L6]12- in DMF-d7 at room temperature.

- 102 -

for the heterogeneous rate constant k0 in the literature reveal ferrocene and its derivatives

demonstrate mediocre electron transfer kinetics with a Pt electrode in DMF solutions,

leading to the observed quasireversible behavior.15

Table 5.1. Half wave potentials (E1/2) and peak separations (∆Ep) measured from the cyclic voltammograms observed for various metallocenes in DMF with 0.1 M Bu4NPF6 (v = 100 mV/s).

Couple E1/2, V vs. Ag/AgCl ∆Ep, mV

CoIII/II –0.76 60 [CoCp2]PF6

CoII/I –1.73 107

[CoCp*2]PF6 CoIII/II –1.31 59

FeCp*2 FeIII/II 0.09 88

Me2Fc FeIII/II 0.47 66

Cyclic voltammograms of [CoCp2⊂Ga4L6]11-, [CoCp*2⊂Ga4L6]

11- and

[FeCp*2⊂Ga4L6]11- in DMF measured with a platinum working electrode are shown in

Figure 5.5. A single wave is observed for the MIII/II couple in all three cases. Comparing

the three systems, it becomes clear that the currents measured for the two cobalt systems

are about an order of magnitude less than that for the iron system at the same scan rate,

despite the fact that the concentrations are approximately the same for all three systems.

Furthermore, the iron system features large anodic currents and small cathodic currents,

but the cathodic and anodic peak currents are approximately equal for the two cobalt

systems. Clearly, there is a different process occurring in the iron system that is not

present in the cobalt systems.

Consider first the cobalt systems. The diffusion coefficient of the host-guest

complex is about a third of that for the guest alone, which would reduce the peak current

by only 32% in a reversible system – not nearly enough to account for the observed order

- 103 -

of magnitude change. Although interference with background current makes

measurement of peak parameters somewhat ambiguous, the peak potential differences

∆Ep measured for the encapsulated CoCp2+/0 couple do not increase with scan rate,

suggesting electrode kinetics do not limit the observed current. (Peak parameters could

not be accurately measured for the encapsulated CoCp*2+/0 couple due to the wave’s

proximity to the residual H2O reduction barrier). The half wave potentials are shifted to

slightly more negative potentials than their values measured in the absence of host. The

observed cathodic shift is somewhat larger for CoCp2+/0 (40-80 mV) than for CoCp*2

+/0

Figure 5.5. Cyclic voltammograms for a) [CoCp2⊂Ga4L6]

11-, b) [CoCp*2⊂Ga4L6]11-, and

c) [FeCp*2⊂Ga4L6]11- measured at 50 mV/s in DMF

using a polished Pt electrode. Data for the host-guest complexes are shown in blue, whereas data measured for the corresponding guests in the absence of host are shown in red. For the cobalt systems shown in a) and b), the currents displayed for the free guests are reduced by 20% from their actual values to aid visual comparison with the much weaker signal for the host-guest complex.

-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6[CoCp

2 < Ga

4L

6]11-

: 50 mV/s

x 0.2

Curr

ent,

µA

Potential, V vs. Ag/AgCl

Host-Guest

Free Guest

a)

-1.8 -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

[CoCp*2 < Ga

4L

6]11-

: 50 mV/s

x 0.2

Cu

rre

nt, µ

A

Potential, V vs. Ag/AgCl

Host-Guest

Free Guest

b)

-0.2 0.0 0.2 0.4

-2

-1

0

1

2

3

[FeCp*2 < Ga

4L

6]11-

: 50 mV/s

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

Host-Guest

Free Guestc)

- 104 -

(10-20 mV). A cathodic shift indicates the cationic state is stabilized in the presence of

host, as discussed later in this chapter.

The reason the currents are so low for the cobalt host-guest systems is that the

guest is rendered electroinactive while encapsulated within the [Ga4L6]12- host. The

observed signal is from the small amount of unencapsulated CoCp2+ present at

equilibrium. This was confirmed by adding fractional equivalents of CoCp2+ to a

solution of K6(Me4N)5[Me4N⊂Ga4L6] in

DMF (Figure 5.6). The very weak Me4N+

guest is readily displaced by the much

more strongly bound CoCp2+, and served

only as a filler to ensure the cluster

remained intact. As expected, no signal

was observed for [Me4N⊂Ga4L6]11- at

potentials below 0 V, with the solvent

reduction barrier appearing below -1.5 V

due to water introduced with the solid

host-guest sample (a dihydrate). When

0.5 equivalents of CoCp2+ was added and

the solution was allowed to equilibrate for

35 minutes, essentially no signal was

observed in the cyclic voltammogram

other than a small feature in the anodic

segment appearing at -0.76 V, with a

-1.4 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2

1 eq.

0.5 eq.

0 eq.

Potential, V vs. Ag/AgCl

1 uA

Figure 5.6. Following the addition of 0.5 equivalents of CoCp2

+ to a solution of [Me4N⊂Ga4L6]11- in DMF,

no signal for the CoIII/II redox process is observed in the cyclic voltammogram. When 1 equivalent of CoCp2

+ is present, a small wave appears. Scan rate: 100 mV/s

- 105 -

current of approximately 30 nA above the baseline at 100 mV/s. A second addition of

CoCp2+ increased the total CoCp2

+:[Ga4L6]12- ratio to 1:1, and the solution was allowed to

equilibrate. The resulting cyclic voltammogram was very similar to that for the

[CoCp2⊂Ga4L6]11- host-guest system, featuring a small reversible wave centered at

-0.79 V vs. Ag/AgCl, with peak currents of about 0.15 µA at 100 mV/s. Unfortunately,

addition of CoCp2+ beyond 1 equivalent led to precipitation of the host.

The absence of a signal with 0.5 equivalents of CoCp2+ but the presence of a

signal with 1 equivalent of CoCp2+ per host is due to the much lower amount of free

CoCp2+ available at equilibrium with 0.5 equivalents added compared to 1 equivalent.

The observed binding constant of CoCp2+ encapsulation by [Ga4L6]

12- is 3.3 x 105 M-1 in

DMF, and assuming the competitive binding of Me4N+ is negligible, the concentration of

free CoCp2+ is about 30 µM for a 1:1 host-guest stoichiometry at 0.4 mM. In contrast,

with only 0.5 equivalents of CoCp2+ present under the same conditions, the free CoCp2

+

concentration is only about 3 µM. This large concentration difference explains the

observed CoCp2+/[Me4N⊂Ga4L6]

11- titration results if [CoCp2⊂Ga4L6]11- is redox silent.

Furthermore, the magnitudes of the peak currents observed in the [CoCp2⊂Ga4L6]11- and

[CoCp*2⊂Ga4L6]11- cyclic voltammograms are consistent with the predicted current for a

reversible system with a bulk concentration of about 30 µM.

The reason why the observed current is much larger for [FeCp*2⊂Ga4L6]11- is that

the anionic host has a tendency to adsorb onto the Pt electrode at more positive potentials.

The current response for this system is extremely sensitive to the condition and history of

the electrode: after polishing the electrode, the response will vary from one experiment to

the next when the same experimental parameters are used (Figure 5.7). When the

- 106 -

electrode is polished prior to each run, the anodic current is much larger than that

observed for the initial cathodic sweep, presumably due to some irreversible host

oxidation process. After many experiments without polishing, the current response

stabilized, exhibiting quasireversible behavior with the half wave potential cathodically

shifted approximately 20 mV versus free FeCp*2+/0. When the Pt working electrode was

removed from the cell after the response stabilized, the metal disk was yellow!

Obviously, some bulk layer formed on the electrode over the course of the experiment,

probably containing the yellow [Ga4L6]12- cluster.

To test whether or not this yellow adsorbate was electroactive, the working

electrode was rinsed with clean electrolyte and immersed in fresh degassed DMF with

0.1 M Bu4NPF6, and then the cyclic voltammetric response was observed (Figure 5.8).

Two different redox processes are observed for this surface-confined material: a

quasireversible wave centered at about 0.06 V, and an irreversible oxidation with a peak

at about 0.7 V vs. Ag/AgCl. The quasireversible process persisted upon further cycling,

but the irreversible oxidation did not. The quasireversible wave is probably from

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

[FeCp*2 < Ga

4L

6]11-

: Polished Pt

Cu

rre

nt, µ

A

Potential, V vs. Ag/AgCl

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

[FeCp*2 < Ga

4L

6]11-

: Nonpolished Pt

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

Several experiments

No polishing

Figure 5.7. Cyclic voltammograms for [FeCp*2⊂Ga4L6]11- measured at 50 mV/s with a freshly polished Pt

electrode (left) and with the same electrode after 20 consecutive experiments (right).

- 107 -

FeCp*2+ trapped in the film, and oxidation of the catecholamide groups of L4- to the

quinone form leads to the irreversible anodic wave.

If a solution of K12[Ga4L6] is titrated with neutral FeCp*2, another type of

behavior is observed. Cyclic voltammetry experiments started with an anodic sweep

from -0.3 V to 0.3 V to generate the cation, which could then go on to interact with

[Ga4L6]12- before the cathodic sweep regenerated the neutral species. By adding aliquots

of FeCp*2 stock solution to K12[Ga4L6] in DMF electrolyte, measurements were made at

several guest:host concentration ratios ranging from 0.25 to 4 molar equivalents of

FeCp*2 per host (Figure 5.9). The resulting cyclic voltammograms exhibit peak current

ratios which depend not only on scan rate but on the guest:host ratio as well. The FeCp*2

concentration dependency is most pronounced at slow scan rates, with the ipc/ipa values

measured at 10 mV/s increasing linearly with [FeII] up to 1 equivalent, then increasing

much more gradually as excess FeCp*2 is added. The half wave potentials observed for

-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2

-2

0

2

4

6

8

Adsorbate Film in Clean Electrolyte

Curr

en

t, µ

A

Potential, V vs. Ag/AgCl

Figure 5.8. After many cyclic voltammetry experiments with [FeCp*2⊂Ga4L6]11- in DMF, the yellow film

deposited on the Pt electrode surface remained adsorbed when transferred to clean electrolyte, and its redox activity was demonstrated by measuring the above cyclic voltammogram for the film itself.

- 108 -

the FeIII/II redox couple are shifted 10-20 mV cathodically in the presence of [Ga4L6]12-,

similar to the shifts observed for the cobalt host-guest systems.

The reason for the lower cathodic currents at slow scan rates is likely due to a

reduction in the effective FeCp*2+ diffusion coefficient due to association with [Ga4L6]

12-.

However, the rate of encapsulation is too slow to account for this association: at 1:1 host-

guest stoichiometry, the second order reaction half life is about 7 minutes for

encapsulation of CoCp*2+, but at v = 10 mV/s FeCp*2 is oxidized and then re-reduced in

about 30 seconds for the potential range used. The host-guest interaction involved in this

situation must then be exterior ion-pair formation. Ion pairing with [Ga4L6]12- has been

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3-0.2

0.0

0.2

0.4

0.6

Potential, V vs. Ag/AgCl

Host OnlyC

urr

ent,

µA

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0.25 eq.

Cu

rre

nt, µ

A

Potential, V vs. Ag/AgCl

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

-0.5

0.0

0.5

1.0

1.5

2.0

0.5 eq.

Cu

rre

nt, µ

A

Potential, V vs. Ag/AgCl

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

0.75 eq.

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

1 eq.

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

-2

-1

0

1

2

2 eq.

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

Figure 5.9. Cyclic voltammograms of FeCp*2 with K12[Ga4L6] in DMF at 10 mV/s. The host solution was titrated with a FeCp*2 stock solution for 0.25 – 2 equivalents of Fe per host.

- 109 -

shown to play an important role in other systems, proposed as an intermediate step

between guest ejection and complete guest dissociation.8, 16

Effect of ion pairing on redox potentials

The small but real cathodic shifts observed for MIII/II redox waves in the presence

of [Ga4L6]12- indicate the cluster stabilizes the metallocenium cation, making its reduction

slightly more difficult. Metallocenium complexes can interact with the cluster in two

different ways: encapsulation or exterior binding. As discussed in Chapter 2, cations are

known to bind to the exterior of the cluster to generate an ion pair, even in aqueous

media, and ion pair formation has been shown to be an important intermediate step

during the guest exchange process.8, 16, 17

Since encapsulated cobalt guests are redox-silent, but the residual signal from

exterior CoIII exhibits a cathodic shift, the stabilizing interaction must be exterior binding.

The electrochemical and exterior binding equilibria are coupled to one another in the

square scheme shown in Figure 5.10.7 Let E±

f0E±

f0 and E±

b0E±

b0 denote the reduction potentials for

the unencapsulated guest when free (solvated) and bound to the exterior of the cluster,

respectively, and let Kox and Kred denote the equilibrium constants for exterior binding of

Figure 5.10. Coupled electrochemical and binding equilibria for a redox-active guest G+/0 forming an exterior adduct with a redox-silent host H.

- 110 -

the cationic and neutral sandwich complexes, respectively. The reduction potential of the

host-guest adduct (guest bound to exterior, not encapsulated) is determined by the other

three equilibria:

E±

b0 = E±

f0 +

RT

FlnKred

Kox

E±

b0 = E±

f0 +

RT

FlnKred

Kox (5.1)

Separate waves for the solvated and bound species will not be observed unless the

exterior binding constant is very high (≥ 105 M-1); rather, a single wave for the free guest

will appear, shifted cathodically in the presence of host.4 A kinetic study by Leung et al.

exploring exterior ion pairing interactions with [Ga4L6]12- indicates Kox ~ 102 M-1 – far

too small to expect two-wave behavior to be observed.16 Both coulombic attraction and

cation-π interactions operate in favor of exterior binding interactions for the cationic

species, but not the neutral species, suggesting Kred < Kox, and a cathodic shift is predicted

from Equation 5.1.

For CoCp*2+ and FeCp*2, the redox waves shift cathodically 10-20 mV when

[Ga4L6]12- is present, while a larger shift of 40-80 mV is observed with CoCp2

+. The

magnitudes of these shifts are consistent with an exterior binding constant Kox ~ 101 -

102 M-1 for the cationic species. The unsubstituted CoCp2+ has a higher affinity for the

exterior of [Ga4L6]12- compared to its decamethyl analogues. One possible explanation

for this difference is that the π-acidic Cp rings of the sandwich complex favorably

interact with the π-basic naphthalene rings in the ligands of the cluster, but the steric bulk

from the methyl groups prevents the Cp* rings from forming as close a contact with the

naphthalene π system, reducing the strength of the favorable π-π interaction.

Although structurally FeCp*2+ and CoCp*2

+ are nearly identical, they display very

different electrochemical responses when combined with [Ga4L6]12-. The redox potential

- 111 -

of the cobalt complex is much more negative than the iron complex, indicating different

electrode interactions for the host-guest complexes are involved at negative and positive

potentials. Even with these problems, platinum working electrodes led to much simpler

voltammograms than glassy carbon or mercury electrodes. With mercury, multiple

adsorption peaks dominated the observed response for [CoCp2⊂Ga4L6]11-, while data

from glassy carbon suffered from a large background signal and poor reproducibility.

Summary

Encapsulation of an electroactive metallocenium cation within [Ga4L6]12- renders

it electrochemically inert. The host prevents electron transfer from the electrode to the

encapsulated guest, demonstrating a neutral guest cannot be electrochemically generated

from the monocation while it resides in the [Ga4L6]12- host cavity. This is likely due to

coulombic repulsion between the negatively charged cathode and the highly anionic host-

guest complex. The residual unencapsulated guest can bind to the host exterior, and this

ion-paired complex is redox-active. Unsubstituted CoCp2+ binds more strongly to the

exterior of [Ga4L6]12- than CoCp*2

+ and FeCp*2+, as inferred from the relative

magnitudes of the cathodic shifts of the half wave potentials. At more positive potentials,

the cluster strongly adsorbs to the Pt electrode surface, forming an electroactive bulk film

with the guest trapped inside of it. These results demonstrate that [Ga4L6]12- does not

exhibit redox-switchable guest binding – encapsulation simply shuts down a guest’s

redox activity.

- 112 -

Experimental

General Considerations

All reagents were obtained from commercial suppliers and used without further

purification unless noted otherwise. A glove box continuously purged with nitrogen was

used to manipulate and store all air-sensitive solids. Standard Schlenk techniques were

used for reactions carried out under argon. When necessary, solvents were degassed by

at least six pump/fill cycles, using argon for the fill step. DMF was dried and stored over

molecular sieves. Tetraethylammonium chloride (Et4NCl) was recrystallized from

absolute ethanol/ether and dried in vacuo over P2O5 at room temperature for 12 hours,

then dried in vacuo over molecular sieves at 60 °C for 18 hours and stored under

nitrogen. Ferrocene was resublimed in air prior to use, and 1,1’-dimethylferrocene

(Me2Fc) was recrystallized under argon from ethanol/water. H4L (H4L = 1,5-bis(2,3-

dihydroxybenzamido)naphthalene) was synthesized according to literature procedure,10

and purified by washing the solid with methanol. Routine mass spectrometry and

elemental analysis was performed by the Mass Spectrometry Laboratory and

Microanalysis Facility in the College of Chemistry at the University of California,

Berkeley. Elemental analysis of host-guest complexes was performed by Desert

Analytics, Tucson, AZ.

Synthetic procedures

Decamethylferrocenium hexafluorophosphate ([FeCp*2]PF6).18 In air, 3 mL of

concentrated H2SO4 was added to 0.40 g (1.2 mmol) of decamethylferrocene (FeCp*2).

The orange-red color of FeCp*2 immediately changed to dark green upon addition of

- 113 -

acid. After stirring overnight (~12 h) at room temperature, the very dark green viscous

reaction mixture was diluted with water to 50 mL, and the resulting green mixture was

filtered through paper to remove the yellow solid byproduct. A solution of 1.03 g

(6.13 mmol) of NaPF6 dissolved in 5 mL of H2O was gradually added to the filtrate while

stirring, immediately forming a green precipitate. The suspension was stirred for 30 min,

and the green solid was collected on a medium frit, washed with cold H2O (4 x 10 mL),

transferred to a vial, and dried overnight in vacuo over Drierite to yield 0.49 g (87%) of

green powder. Anal. Calc. (found) for C20H30FePF6: %C, 50.97 (51.08); H, 6.42 (6.16).

MS(ESI+, CH3CN): m/z 326.2 ([FeCp*2]+). 1H NMR (acetone-d6, 300 MHz):

δ -37.3 ppm (br s, 30H, -CH3). Visible spectrum (MeOH, λ > 400 nm): λmax(εmax) 779 nm

(543 M-1 cm-1); Gaussian Fit: Peak 1: 646 nm (168 M-1 cm-1); Peak 2: 740 nm

(299 M-1 cm-1); Peak 3: 786 nm (363 M-1 cm-1).

1, 1’-dimethylferrocenium hexafluorophosphate ([Me2Fc]PF6).19 The title compound

was prepared using a modified literature procedure reported for the synthesis of the

unsubstituted FcPF6. In air, 10 mL of concentrated H2SO4 was added to 1.0 g (4.7 mmol)

of recrystallized Me2Fc. The orange color of Me2Fc immediately changed to dark blue

upon addition of acid. After stirring for 4 h at room temperature, the very dark blue

viscous reaction mixture was diluted to 150 mL with water, and the resulting blue

mixture was filtered through paper to remove a dark solid by-product. A solution of 3.6 g

(21 mmol) of NaPF6 dissolved in 10 mL of H2O was gradually added to the filtrate while

stirring, forming a dark precipitate suspended in a dark mother liquor. The reaction

vessel was sealed with a glass stopper and placed in a refrigerator to cool overnight. A

- 114 -

dark blue amorphous solid was collected on a course frit, washed with cold H2O

(3 x 15 mL), and partially dried on frit by pulling vacuum through solid cake under a

stream of nitrogen. Note that the filtrate was dark blue, due to the moderate solubility of

the product in H2O. The gooey solid was transferred to a vial and dried overnight in

vacuo over Drierite (periodically pausing to break up clumps of solid with spatula) to

yield 0.94 g (56%) of dark cerulean amorphous powder. While the product does have a

faint sulfur odor, elemental analysis indicates its purity is sufficient for use without

further purification. Anal. Calc. (found) for C12H14FePF6: %C, 40.14 (40.34); H, 3.93

(3.84). MS(ESI+, CH3CN): m/z 214.0 (Me2Fc+). 1H NMR (D2O, 500 MHz): δ 34.5 (br s,

4H, Cp-Ha), 31.5 (br s, 4H, Cp-Hb), 10.39 ppm (br s, 6H, -CH3). Visible spectrum

(MeOH, λ > 400 nm): λmax(εmax): 649 nm (304 M-1 cm-1), 569 nm (sh, 202 M-1 cm-1),

466 nm (184 M-1 cm-1).

K12[Ga4L6]·4Me2CO.14 The following reaction was carried out under argon using a

procedure adapted from the literature. A suspension of 2.00 g (4.65 mmol) of H4L in

125 mL of degassed methanol was prepared, to which 19.5 mL of 0.5 M KOH in

methanol was added via syringe while stirring, followed by via three pump/fill cycles to

remove any oxygen introduced with the base solution. Addition of base caused most

solid to dissolve, and the color changed from white to yellow. Addition of 25 mL

degassed methanol to the turbid reaction mixture caused the remaining solid to dissolve

after stirring for 10 min at room temperature. To this yellow solution, 1.14 g

(3.11 mmol) of Ga(acac)3 was added, and the reaction mixture was immediately degassed

via six pump/fill cycles. The dark yellow solution was stirred overnight at room

- 115 -

temperature, filtered through a fine frit to remove insoluble residual impurities, and

solvent was removed using a vacuum pump. The dark yellow residue was dried in vacuo

for 3 h to remove 2,4-pentanedione, re-dissolved in a minimal amount of methanol

(100 mL), and the product was precipitated with 600 mL of degassed acetone, then stirred

for 2 h to allow the heterogeneous system to equilibrate. The yellow precipitate was

collected on a frit under a stream of nitrogen, washed with acetone (3 x 50 mL), and dried

in vacuo over molecular sieves and Drierite for 12 h to yield 2.5 g (93%) of light yellow

powder. 1H NMR integrals confirmed the number of acetone molecules present in the

sample.

K11[CoCp2⊂⊂⊂⊂Ga4L6]·KPF6·2H2O. The following reaction was carried out under argon.

To 300 mg (0.0848 mmol) of K12[Ga4L6]·4Me2CO and 28.3 mg (0.0848 mmol) of

[CoCp2]PF6 (Aldrich) was added 50 mL of degassed methanol, leading to a homogeneous

solution. After stirring at room temperature overnight, solvent was removed with a

vacuum pump, and the residue was dried in vacuo for 3 h. The yellow solid was scraped

from flask walls, transferred to an Erlenmeyer flask, swirled in pentane, collected on a

frit under a stream of nitrogen, washed with pentane (3 x 20 mL), and dried overnight in

vacuo over molecular sieves to yield 285 mg (92%) of orange-yellow powder. Anal.

Calc. (found) for C154H98CoF6Ga4K12N12O38P: %C, 50.31 (50.32); H, 2.68 (2.76); N, 4.57

(4.48); Co, 1.60 (1.46); Ga 7.59 (8.15). The number of co-crystallized H2O molecules

per host was confirmed by 1H NMR in DMSO-d6, using the DMSO/H2O peak integrals

from pure DMSO-d6 for comparison. 1H NMR (D2O, 300 MHz): δ 7.98 (br, 12H, ArH),

- 116 -

7.84 (br, 12H, ArH), 7.30 (d, J=8.1 Hz, 12H, ArH), 7.11 (t, J=8.2 Hz, 12H, ArH), 6.75

(d, J=7.0 Hz, 12H, ArH), 6.59 (t, J=7.8 Hz, 12H, ArH), 2.21 (s, 10H, Cp-H, encaps.).

K11[CoCp*2⊂⊂⊂⊂Ga4L6]·KPF6·2H2O. The title compound was synthesized from 301 mg

(0.0850 mmol) of K12[Ga4L6]·4Me2CO and 40.2 mg (0.0848 mmol) of [CoCp*2]PF6

(Aldrich) in a manner analogous to that described for K11[CoCp2⊂Ga4L6]·KPF6·2H2O.

Yield: 290 mg of dark yellow grainy solid. 1H NMR (D2O, 400 MHz): δ 8.26 (d,

J=7.9 Hz, 4H, ArH), 8.22 (d, J=7.8 Hz, 4H, ArH), 7.87 (d, J=7.7 Hz, 4H, ArH), 7.68 (d,

J=8.5 Hz, 4H, ArH), 7.62 (d, J=8.5 Hz, 4H, ArH), 7.47 (m, 8H, ArH), 7.37 (m, 8H, ArH),

6.85 (t, J=8.2 Hz, 4H, ArH), 6.81 – 6.62 (m, 28H, ArH), 6.57 (t, J=7.8 Hz, 4H, ArH),

-0.58 (s, 30H, CH3, encaps.).

K11[FeCp*2⊂⊂⊂⊂Ga4L6]·KPF6·2H2O. The title compound was synthesized from 150 mg

(0.0425 mmol) of K12[Ga4L6]·4Me2CO and 20.1 mg (0.0426 mmol) of [FeCp*2]PF6 in a

manner analogous to that described for K11[CoCp2⊂Ga4L6]·KPF6·2H2O. Yield: 140 mg

(88%) of light green powder. Anal. Calc. (found) for C164H118FeF6Ga4K12N12O38P:

%C, 51.65 (51.80); H, 3.12 (3.45); N, 4.41 (4.28); Fe, 1.46 (1.38). 1H NMR (MeOD,

500 MHz, -20 °C): δ 27.86 (s, 4H, ArH), 21.66 (s, 4H, ArH), 20.32 (s, 4H, ArH), 9.37 (d,

J=8.9 Hz, 4H, ArH), 9.01 (s, 4H, ArH), 8.91 (d, J=8.1 Hz, 4H, ArH), 8.43 (s, 4H, ArH),

6.77 (s, 4H, ArH), 6.42 (d, J=7.9 Hz, 4H, ArH), 4.70 (s, 8H, ArH), 4.06 (m, 4H, ArH),

3.85 (d, J=7.6 Hz, 4H, ArH), 2.79 (br s, 4H, ArH), 2.09 (br s, 4H, ArH), 0.20 (br s, 4H,

ArH), -2.01 (br s, 4H, ArH), -3.08 (br s, 4H, ArH), -46.78 (br s, 30H, -CH3, encaps.).

- 117 -

Binding constant measurements

The binding equilibrium constant for encapsulation of CoCp2+ by [Ga4L6]

12- in

DMF-d7 was calculated from the 1H NMR spectra measured for four different samples

using a Bruker AV-500 spectrometer with a TBI-P probe at ambient temperature. All

four samples were prepared with [host]=[guest]=0.4 mM. One sample was prepared

using pre-synthesized K11[CoCp2⊂Ga4L6]·KPF6, and the other three samples were

prepared by mixing stock solutions of K12[Ga4L6] and [CoCp2]PF6 in a nitrogen-filled

glovebox. The binding constants were computed from the integrated areas of the exterior

and interior guest resonances for each spectrum, and the resulting values were averaged

to obtain the reported binding constant.

Encapsulation kinetic measurements

In a screw-cap vial, 0.83 mg (0.24 µmol) of K12[Ga4L6] was dissolved in 0.6 mL

of DMF-d7 and transferred to an NMR tube. A Bruker AV-500 spectrometer with a

TBI-P probe was used to monitor the 1H NMR signal. After shimming and tuning, the

sample was ejected from the probe, and 10 µL (0.2 µmol) of [CoCp*2]PF6 stock solution

(20 mM in DMF-d7) was added to the NMR tube at t = 0. The tube was repeatedly

inverted to mix the solution, and spectra were acquired at approximately 1 min intervals

using one scan each with a 90° pulse. The first point corresponded to t ≈ 2 min, and the

last to t ≈ 40 min.

- 118 -

Diffusion coefficient measurements

In a nitrogen-filled glovebox, a solution containing 0.4 mM K12[Ga4L6], 0.4 mM

[CoCp2]PF6, and 11 mM Bu4NPF6 in DMF-d7 was prepared, transferred to a standard

NMR tube, and capped with a rubber septum. PGSE diffusion 1H NMR measurements

were performed on a Bruker AVB-400 spectrometer with a z-gradient coil, using the

ledbpgp2s pulse program with diffusion time ∆ = 100 ms, bipolar gradient pulse duration

δ = 7 ms (2 x 3.5 ms), 108 scans per experiment, and a linear gradient strength ramp of

32 increments from 2% to 95%.20 A constant temperature of 298 K was maintained

during the eight hour experiment. The integrated areas were fit to the exponential decay

equation21 using nonlinear regression to evaluate the diffusion coefficient. The probe

gradient power was calibrated from a fit of the diffusion decay curve of dextrose and β-

cyclodextrin in D2O using literature values for the diffusion coefficients.22

Electrochemical measurements

All electrochemical experiments were carried out at ambient temperature (~22 °C)

using either a Solartron 1280B (at Cornell University) or a BAS 100A potentiostat (at UC

Berkeley). Data from the Solartron 1280B were collected without noise filtration in the

post processing. Dry DMF was used for the solvent, and 0.1 M Bu4NPF6, recrystallized

from hot ethyl acetate, served as the supporting electrolyte. Free guest solutions (no

[Ga4L6]12- present) were prepared with analyte concentrations of approximately 1 mM,

and host-guest solutions were prepared with 0.4 mM [Ga4L6]12- due to the host’s limited

solubility in the electrolyte solution. All samples were sparged with dry nitrogen or

- 119 -

argon for at least 10 minutes before the first measurement, and then held under a

continuous stream of inert gas until measurements were finished.

A scintillation vial served as the vessel for the standard three-electrode

electrochemical cell. A platinum disc working electrode and a coiled platinum wire

counter electrode were used, and a homemade Ag/AgCl reference electrode was used

with saturated NaCl filling solution. All potentials are reported relative to this electrode

potential. The working electrode was mechanically polished with 1 µm diamond paste

and electrochemically polished in 0.1 M aqueous H2SO4 at least once before starting

experiments with a different sample. The electrochemical polishing procedure consisted

of four steps (potentials vs. Ag/AgCl electrode described above): 1) hold potential at

1.6 V for 30 sec, 2) hold potential at -0.3 V for 30 sec, 3) rapid cyclic voltammetry

(v = 3 – 5 V/s) from -0.3 V to 1.6 V for 10 cycles, returning to -0.3 V and 4) slow cyclic

voltammogram (v = 0.1 V/s) to test surface characteristics (usual range of -0.2 V to 1.5 V,

depending on H2 reduction barrier). For data collection, potentials were held at the

starting potential for several seconds before beginning the cyclic voltammetry runs.

References

1. a) Atwood, J. L.; Davies, J. E. D.; MacNicol, D. D.; Vogtle, F.; Lehn, J.-M., Comprehensive Supramolecular Chemistry. Pergamon: Oxford, 1996; b) Lehn, J.-M., Supramolecular Chemistry: Concepts and Perspectives. VCH: Weinheim, 1995; c) Conn, M. M.; Rebek, J., Jr., “Self-assembling capsules.” Chem. Rev. 1997, 97, 1647-1668; d) Lawrence, D. S.; Jiang, T.; Levett, M., “Self-Assembling Supramolecular Complexes.” Chem. Rev. 1995, 95, 2229-2260; e) Leininger, S.; Olenyuk, B.; Stang, P. J., “Self-Assembly of Discrete Cyclic Nanostructures Mediated by Transition Metals.” Chem. Rev. 2000, 100, 853-908.

2. a) Davis, A. V.; Yeh, R. M.; Raymond, K. N., “Supramolecular Assembly Dynamics.” Proc. Nat. Acad. Sci. USA 2002, 99, 4793-4796; b) Fiedler, D.; Leung, D. H.; Bergman, R. G.; Raymond, K. N., “Selective Molecular Recognition, C-H Bond Activation, and Catalysis in Nanoscale Reaction Vessels.” Acc. Chem. Res.

- 120 -

2005, 38, 351-360; c) Johnson, D. W.; Raymond, K. N., “The Role of Guest Molecules in the Self-Assembly of Metal-Ligand Clusters.” Supramolecular Chem.

2001, 13, 639-659; d) Pluth, M. D.; Raymond, K. N., “Reversible Guest Exchange Mechanisms in Supramolecular Host-Guest Assemblies.” Chem. Soc. Rev. 2007, 36, 161-171; e) Rebek, J., Jr., “Reversible Encapsulation and Its Consequences in Solution.” Acc. Chem. Res. 1999, 32, 278-286.

3. Kaifer, A. E.; Gómez-Kaifer, M., Supramolecular Electrochemistry. Wiley-VCH: Weinheim, 1999.

4. Kaifer, A. E.; Mendoza, S., “Redox-switchable Receptors.” In Molecular

Recognition: Receptors for Cationic Guests, Gokel, G. W., Ed. Pergamon: Tarrytown, NY, 1996; Vol. 1, pp 701-732.

5. Boulas, P. L.; Gómez-Kaifer, M.; Echegoyen, L., “Electrochemistry of Supramolecular Systems.” Angew. Chem. Int. Ed. 1998, 37, 216-247.

6. a) Venturi, M.; Credi, A.; Balzani, V., “Electrochemistry of coordination compounds: an extended view.” Coord. Chem. Rev. 1999, 185-186, 233-256; b) Gisselbrecht, J.-P.; Gross, M.; Lehn, J.-M.; Sauvage, J.-P.; Ziessel, R.; Piccinni-Leopardi, C.; Arrieta, J. M.; Germain, G.; Van Meerssche, M., “p-Quaterpyridine complexes: crystal structure of the mononuclear copper(II) complex. Electrochemical studies of the monomeric copper(II) and dimeric copper(I) complexes, of their interconversion, and of the bis[RuII(bipy)2] complex.” Nouv. J. Chimie 1984, 8, 661-667; c) Moon, K.; Kaifer, A. E., “Dimeric Molecular Capsules under Redox Control.” J. Am. Chem.

Soc. 2004, 126, 15016-15017; d) Fabbrizzi, L.; Poggi, A., “Sensors and Switches from Supramolecular Chemistry.” Chem. Soc. Rev. 1995, 24, 197-202.

7. Miller, S. R.; Gustowski, D. A.; Chen, Z. H.; Gokel, G. W.; Echegoyen, L.; Kaifer, A. E., “Rationalization of the Unusual Electrochemical Behavior Observed in Lariat Ethers and Other Reducible Macrocyclic Systems.” Anal. Chem. 1988, 60, 2021-2024.

8. Davis, A. V.; Fiedler, D.; Seeber, G.; Zahl, A.; van Eldik, R.; Raymond, K. N., “Guest Exchange Dynamics in an M4L6 Tetrahedral Host.” J. Am. Chem. Soc. 2006, 128, 1324-1333.

9. Fiedler, D.; Pagliero, D.; Brumaghim, J. L.; Bergman, R. G.; Raymond, K. N., “Encapsulation of Cationic Ruthenium Complexes into a Chiral Self-Assembled Cage.” Inorg. Chem. 2004, 43, 846-848.

10. Caulder, D. L.; Powers, R. E.; Parac, T. N.; Raymond, K. N., “The Self-Assembly of a Predesigned Tetrahedral M4L6 Supramolecular Cluster.” Angew. Chem. Int. Ed.

1998, 37, 1840-1842.

11. a) Davis, A. V.; Raymond, K. N., “The Big Squeeze: Guest Exchange in an M4L6 Supramolecular Host.” J. Am. Chem. Soc. 2005, 127, 7912-7919; b) Tiedemann, B. E.

- 121 -

F.; Raymond, K. N., “Dangling Arms: A Tetrahedral Supramolecular Host with Partially Encapsulated Guests.” Angew. Chem. Int. Ed. 2006, 45, 83-86.

12. Sohn, Y. S.; Hendrickson, D. N.; Gray, H. B., “Electronic Structure of Metallocenes.” J. Am. Chem. Soc. 1971, 93, 3603-3612.

13. Parac, T. N.; Raymond, K. N. Unpublished results.

14. Parac, T. N.; Caulder, D. L.; Raymond, K. N., “Selective Encapsulation of Aqueous Cationic Guests into a Supramolecular Tetrahedral M4L6 Anionic Host.” J. Am.

Chem. Soc. 1998, 120, 8003-8004.

15. Matsumoto, M.; Swaddle, T. W., “The Decamethylferrocene(+/0) Electrode Reaction in Organic Solvents at Variable Pressure and Temperature.” Inorg. Chem. 2004, 43, 2724-2735.

16. Leung, D. H.; Bergman, R. G.; Raymond, K. N., “Scope and Mechanism of the C-H Bond Activation Reactivity within a Supramolecular Host by an Iridium Guest: A Stepwise Ion Pair Guest Dissociation Mechanism.” J. Am. Chem. Soc. 2006, 126, 9781-9797.

17. Fiedler, D.; Bergman, R. G.; Raymond, K. N., “Supramolecular catalysis of a unimolecular transformation: Aza-Cope rearrangement within a self-assembled host.” Angew. Chem. Int. Ed. 2004, 43, 6748-6751.

18. Nesmeyanov, A. N.; Materikova, R. B.; Lyatifov, I. R.; Kurbanov, T. K.; Kochetkova, N. S., “sym-Polymethylferricinium Hexafluorophosphates.” J.

Organomet. Chem. 1978, 145, 241-243.

19. Yang, E. S.; Chan, M.-S.; Wahl, A. C., “Rate of Electron Exchange between Ferrocene and Ferricenium Ion from Nuclear Magnetic Resonance Studies.” J. Phys.

Chem 1975, 79, 2049-2052.

20. Kerssebaum, R., “DOSY and Diffusion by NMR.” In User Guide for XWinNMR

3.1/3.5 Version 1.03, Bruker BioSpin GmbH: Rheinstetten, Germany, 2002.

21. Stejskal, E. O.; Tanner, J. E., “Spin Diffusion Measurements: Spin Echoes in the Presence of a Time-Dependent Field Gradient.” J. Chem. Phys. 1965, 42, 288-292.

22. Longsworth, L. G., “Temperature Dependence of Diffusion in Aqueous Solutions.” J.

Phys. Chem 1954, 58, 770-773.

- 122 -

CHAPTER 6

Redox-Active Vanadium Complexes

Introduction

The work in Chapters 2 through 5 used only M4L6 clusters assembled with GaIII

metal ions, since gallium(III) has several properties which are very convenient for

different applications. For example, the d10 GaIII ion is diamagnetic and thus NMR silent,

and the very high reduction potential of GaIII makes it electrochemically inert. However,

other metal ions can be used to prepare M4L6 tetrahedra, since the catechol binding sites

on L4- are very powerful chelators for hard metal ions with pseudo-octahedral

coordination geometries. Previous studies have shown M4L6 clusters can be assembled

with trivalent MIII = AlIII, FeIII, GaIII, InIII, as well as tetravalent MIV = TiIV, GeIV, and

SnIV.1 In this chapter, redox-active vanadium(IV) is used to prepare the M4L6 tetrahedral

complex, as well as the related M2L3 dinuclear helicate and a mononuclear ML3 model

complex. Where the preceding chapter investigated redox-active guests with redox-silent

hosts, this chapter concerns redox-active hosts with redox-silent guests.

Vanadium tris(catechol) complexes have been characterized with trivalent,

tetravalent, and pentavalent vanadium ions.2 The three oxidation states for [V(cat)3]z-

(z = 1, 2, 3) can be electrochemically interconverted, with E±0(VV=IV) = ¡0:035 VE±0(VV=IV) = ¡0:035 V and

E±0(VIV=III) = ¡0:86 VE±0(VIV=III) = ¡0:86 V vs. SCE for [V(cat)3]z- in acetonitrile (cat2- = ortho-catechol

dianion). This means that for V4L6, eight electrons are released when oxidizing all four

vanadium(III) ions in [V4L6]12- to vanadium(V) in [V4L6]

4-. Due to the large difference

- 123 -

between the two reduction potentials, the waves for [V4L6]4-/8- and [V4L6]

8-/12- are well

separated, but how the reduction of individual vanadium ions will proceed must be

determined. Will reduction of one vanadium in a complex influence the reduction of

adjacent vanadium ions?

When multiple redox sites are involved, three scenarios can occur, depending on

how the individual redox sites interact. Consider the reduction of four identical sites. At

one extreme, if the reduction of one site leads to cooperative interactions that make the

reduction of the remaining three sites much more favorable than the first, all reductions

occur simultaneously and a single four electron wave is observed in the cyclic

voltammogram (Figure 6.1a). At the other extreme, if reduction of one site makes the

next more difficult, each consecutive reduction requires more energy than the last,

requiring more negative electrode potentials, and four separate one electron waves are

observed (Figure 6.1b). Finally, if individual redox sites do not interact at all with one

Figure 6.1. With four identical redox sites, the observed redox wave depends on how the sites interact. a) If reduction of one site makes the next reduction more favorable, all four electrons are transferred at once. b) If consecutive reductions become more difficult, four separate one electron waves are observed. c) If there is no interaction between sites, the four one electron waves overlap to generate one broad wave.

60

40

20

0

-20

-40

-60

-0.25 -0.5 -0.75 -1 -1.25E(V):

I/µA

a)

15

10

5

0

-5

-10

-0.25 -0.5 -0.75 -1 -1.25E(V):

I/µA

b) 37.5

25

12.5

0

-12.5

-25

-0.25 -0.5 -0.75 -1 -1.25E(V):

I/µA

c)

- 124 -

another, the free energy of successive reductions are affected only by statistical (entropic)

factors, and the small differences in potentials result in four overlapping one-electron

waves, observed as a single broad wave in the cyclic voltammogram (Figure 6.1c).3

Synthesis and characterization of vanadium(IV) complexes

In addition to the tetranuclear [V4L6]8- cluster, the mononuclear complex

[V(cam)3]2- and the dinuclear helicate [V2L

H3]

4- were synthesized as the potassium salts,

with the structures of H2cam and H4LH shown in Figure 6.2 (H2cam = N-methyl-2,3-

dihydroxybenzamide; H4LH = 1,4-bis(2,3-dihydroxybenzamido)benzene). A similar

400 500 600 700 800 9000

5

10

15

20

25

ε ,

10

3 M

-1 c

m-1

Wavelength, nm

[V2LH

3]4-

300 400 500 600 700 8000.0

0.2

0.4

0.6

0.8

1.0

Absorb

an

ce

Wavelength, nm

[V(cam)3]2-

400 500 600 700 800 9000

10

20

30

40

50

60

70

ε,

10

3 M

-1 c

m-1

Wavelength, nm

[V4L6]8-

H2cam =

H4LH

a)

b) c)

Figure 6.2. Structural formulae of ligands H2cam and H4LH (upper right), and the UV-visible spectra for

aqueous solutions of a) K2[V(cam)3], b) K4[V2LH

3], and c) K7[Et4N⊂V4L6].

- 125 -

procedure was used to prepare the three vanadium(IV) complexes, where a suspension of

ligand in degassed methanol was combined with the appropriate stoichiometric amount of

VO(acac)2 under argon (H-acac = 2,4-pentanedione) and 2 equivalents of KOH per

VO(acac)2, and stirred at room temperature for several hours. The amount of KOH must

be carefully measured, since the VO2+ ion is favored in basic conditions, and the oxo

ligand must be displaced by the bidentate catechol unit to form the pseudo-octahedral

VL3 coordination geometry in each complex. The resulting products are extremely dark

blue, appearing black in the solid state. This intense color is from broad overlapping

ligand to metal charge transfer bands spanning the entire visible region (Figure 6.2), and

is characteristic of vanadium(IV) tris(catechol) derivatives.

Negative ion electrospray mass spectrometry confirmed the formation of

[V(cam)3]2-, [V2L

H3]

4-, and [V4L6]8-, and elemental analysis confirmed the purity of the

powders. The infrared spectra showed no evidence for a V=O stretch expected near

2500 3000 3500 4000

H (gauss)

[V2LH

3]4-

[V(cam)3]4-

Figure 6.3. EPR spectra of glassy (Et3NH)4[V2LH

3] and K2[V(cam)3] in MeOH/EtOH (9:1) at 8 K.

- 126 -

1000 cm-1 for the vanadyl ion, consistent with pseudo-octahedral vanadium coordination

spheres in all three complexes. EPR spectra of K2[V(cam)3] and (Et3NH)4[V2LH

3] were

measured at 8 K as MeOH/EtOH glasses (Figure 6.3). Both displayed the eight-line

splitting pattern characteristic of the 51V nucleus (I = 7/2), and are similar to the EPR

spectrum of glassy (Et3NH)2[V(cat)3] in aqueous catechol.4 More informative is that the

two spectra are very different from that observed for VO(cat)2, since the lower symmetry

of the square pyramidal vanadyl complex leads to a high degree of anisotropy observed in

the EPR spectra. This further supports the tris(bidentate) coordination mode for

vanadium(IV) expected with formation of [V(cam)3]2- and [V2L

H3]

4-. Due to the

paramagnetic nature of the d1 vanadium(IV) ion, NMR methods could not be used to

characterize the complexes described here.

The X-ray crystal structure of K4[V2LH

3] was determined, confirming that

[V2LH

3]4- is a discrete, dinuclear triple helicate (Figure 6.4). This represents the first

structural characterization of a non-oxo vanadium(IV) supramolecular assembly. The

general structural features of the [V2LH

3]4- helicate are very similar to other dinuclear

Figure 6.4. Illustrations of the anionic fragment [V2LH

3]4- based on the X-ray crystal structure coordinates,

as viewed from the side of the helicate (left), and along the metal-metal axis (right).

- 127 -

helicate structures reported by Raymond and coworkers: two pseudo-octahedral metal

ions, each coordinated by three catecholamide groups joined by a rigid para-phenylene

linker. The compound crystallizes in space group P21/n with Z = 4, with a large amount

of disordered solvent (one Et2O, about seven DMF molecules, and about 0.3 H2O

molecules per helicate). Each helicate is homochiral, with both ∆,∆ and Λ,Λ enantiomers

present in the centrosymmetric unit cell. The V1 – V2 distance is 12.04 Å, with the

metal-metal axis distorted from three-fold symmetry due to crystal packing effects. The

average trigonal twist angle is 33°, but this varies up to three degrees between the six

chelating groups due to packing effects. A summary of bond lengths and angles for the

vanadium coordination spheres is given in Appendix 3.

To gauge the influence of the VIV coordination geometry on the helicate structure,

the structure of K4[V2LH

3] will be compared with two previously published dinuclear

helicates: (s-nic)6[Ga2LH

3] and K4[Ti2LA

3] (H4LA = 2,6-bis(2,3-

dihydroxybenzamido)anthracene).5 The average trigonal twist angle for the GaIII

structure is 46.5°, and only varies up to one degree from this mean value. For the TiIV

structure, the average twist angle is 34.5°, just slightly greater than the average for

vanadium. However, due to the variation between individual dihedral angles, the TiIV

and VIV twist angles are equal within the standard deviation. Thus, the additional d

electron in VIV does not appreciably influence the twist angle. Furthermore, the Ga1 –

Ga2 distance in [Ga2LH

3]6- is 11.71 Å, 0.33 Å shorter than the V1 – V2 distance in

[V2LH

3]4-. The shorter distance in the GaIII structure is due to the larger twist, coiling the

ligands inward. Differences between V-O bond distances observed here and M-O bond

- 128 -

distances (M = TiIV, GaIII) in analogous structures are consistent with differences in ionic

radii.

Electrochemical Studies of Vanadium Complexes

The electrochemical behavior of tris(bidentate) catecholamide vanadium(IV)

complexes is dependent on experimental conditions. In particular, the choice of solvent,

pH of aqueous solutions, and type and conditioning of the working electrode surface are

all factors which strongly influence the observed response during a cyclic voltammetry

experiment. For this reason, this section will be divided into two parts: DMF solutions

will be described first, followed by aqueous solutions. Note that the potential of the

Ag/AgCl (sat’d NaCl) reference electrode used here was -0.080 V vs. SCE.

DMF Solutions – [V2L3]4-

and [V4L6]8-

Cyclic voltammograms of [V2LH

3]4- and [V4L6]

8- in DMF measured using a

hanging mercury drop electrode (HMDE) are shown in Figure 6.5, with potentials

Figure 6.5. Cyclic voltammograms for a) [V2LH

3]4- and b) [V4L6]

8- in DMF. The scan rates range from 0.4 V/s (red) to 2.0 V/s. Inset: peak current for the cathodic (blue) and anodic (red) waves plotted against the square root of the scan rate.

-0.6 -0.8 -1.0 -1.2 -1.4

-5

0

5

10

0.0 0.5 1.0 1.50

2

4

6

8

10

Pe

ak C

urr

en

t, µ

A

sqrt(v), (V/s)1/2

Cathodic

Anodic

Curr

en

t, µ

A

Potential, V vs. Ag/AgCl

-0.6 -0.8 -1.0 -1.2 -1.4

-5

0

5

10

0.0 0.5 1.0 1.50

2

4

6

Pe

ak C

urr

en

t, µ

A

sqrt(v), (V/s)1/2

Cathodic

Anodic

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

a) V2LH

3 b) V4L6

- 129 -

reported vs. Ag/AgCl (saturated NaCl filling solution). Note that all DMF solutions

prepared for electrochemical experiments contained 0.1 M Bu4NPF6 as the supporting

electrolyte. Quasireversible waves for the VIV/III redox couple are observed for both

supramolecular systems, with E1/2 = -0.87 V vs. Ag/AgCl for [V2LH

3]4-/6- and

E1/2 = -0.88 V vs. Ag/AgCl for [V4L6]8-/12-. Plots for the peak cathodic and anodic

currents vs. pv (where vv is the scan rate in V/s) yield straight lines for both systems,

characteristic of a diffusion-controlled process (insets in Figure 6.5).

Assuming reversible behavior, the slopes of the regression lines fit to the peak

current plots in Figure 6.5 can give information about the number of electrons involved.

For the reversible reaction O + ne = R, the peak cathodic current (in µA) for a linear

potential sweep is3

ip;c = (2:69£ 105)n3=2AD1=2O C¤

O

pvip;c = (2:69£ 105)n3=2AD

1=2O C¤

O

pv (6.1)

where n is the number of electrons transferred in a single step, A is the electrode area

in cm2, DO is the diffusion coefficient of species O in cm2 s-1, CO* is the concentration of

O in the bulk solution in mM, and vv is the scan rate in V/s. In theory, the anodic peak

current ip,a measured during the reverse sweep can also be described using the same

equation.

From the slope of ip,c vs. pv , n can be determined if A, DO, and CO* are known.

The value of CO* corresponds to the analyte concentration determined from the sample

preparation. The electrode surface area A = 0.01 cm2 (“SMALL” drop size setting) was

determined from the average mass of the Hg drop dispensed by the apparatus while the

capillary tip is immersed in electrolyte solution. The diffusion coefficients of [V2LH

3]4-,

- 130 -

[V2LH

3]6-, [V4L6]

8-, and [V4L6]12- have not been measured, but they can be estimated from

measured values for similar complexes by using the Stokes-Einstein equation6

D =kBT

6¼¹rh

D =kBT

6¼¹rh (6.2)

where kB is Boltzmann’s constant, T is the temperature in Kelvin, µ is the dynamic

viscosity of the solvent, and rh is the hydrodynamic radius of the diffusing species.

Although rh is unknown, for large molecules we can assume it varies proportionally with

changes in the ion’s size as measured from crystallographic data.8

The diffusion coefficient of [CoCp2 ⊂ Ga4L6]11- with 11 mM Bu4NPF6 in DMF-d7

is DGa = 3.1(1) x 10-6 cm2 s-1 at 25 °C (Chapter 5), measured by diffusion NMR. To use

this observed value for the vanadium system, we can make the following assumptions:

1) the identity of the encapsulated guest does not affect the exterior dimensions of the

cluster, and 2) size changes due to substitution of VIII for GaIII can be neglected.

Therefore, the hydrodynamic radii of [CoCp2⊂Ga4L6]11- and [Et4N⊂V4L6]

11- are equal,

and D = 3.1 x 10-6 cm2 s-1 for [Et4N⊂V4L6]11-. For large ions in nonaqueous solvents,

DO ≈ DR (where R denotes the reduced species), so the diffusion coefficients of the VIII

and VIV clusters are approximately equal.

Although the dinuclear helicate is certainly not a sphere, its effective radius can

be estimated to be about half the center-to-center distance between adjacent complexes in

the crystal lattice. This approximation seems reasonable because a large amount of

solvent co-crystallized with each [V2LH

3]4- ion. From the crystal structure described

earlier in this chapter, the above estimate gives reff ≈ 6.8 Å. This means the effective

radius of the M2L3 helicate is about 2/3 of that for the M4L6 cluster, estimated to be 10 Å

8 For further information about diffusion, see Chapter 2 in this dissertation.

- 131 -

from the center-to-center distances between clusters in the K5(Et4N)6[Et4N⊂Fe4L6]

crystal lattice.7 According to the Stokes-Einstein equation, the diffusion coefficient

varies with 1/r, and thus DM2L3 ≈ (3/2)·DM4L6 = 4.5 x 10-6 cm2 s-1 in DMF electrolyte.

Using the above values for the diffusion coefficients, the slopes of ip,c vs. pv for

the data in Figure 6.5 correspond to n ≈ 1.0 for [V2L3]4-/6- and n ≈ 2.3 for [V4L6]

8-/12-. It

should be emphasized that in general the value of n computed using Equation 6.1 does

not necessarily correspond to the actual number of electrons transferred for multielectron

processes. Only if all electrons are transferred simultaneously in a single, perfectly

reversible step will this value of n actually equal the number of electrons involved. One

familiar example of a true multielectron process is the reduction of aqueous CuII to Cu0.

The peak to peak separation ∆Ep in the cyclic voltammogram would be 59/n mV for a

true multielectron process. For the cyclic voltammograms in Figure 6.5, ∆Ep ≈ 100 mV,

independent of scan rate.

The observed behavior does correspond to that expected if a species contains

multiple identical non-interacting sites, each capable of a one electron reversible redox

reaction. For [V2LH

3]4- there are two sites, and for [V4L6]

8- there are four. In other

words, the reduction of one vanadium(IV) vertex to vanadium(III) has little to no

influence on the reduction potential for the remaining vanadium(IV) ions in the same

molecule. With no communication between metal centers, the free energy of successive

reductions differ purely for entropic reasons. Statistical analysis dictates that the kth

successive formal reduction potential differs from the first according to:3

E±

k ¡ E±

1 = ¡2RT

Fln k

(6.3)

- 132 -

where R is the universal gas constant and F is the Faraday constant. For this sort of

system the wave observed in the cyclic voltammogram will resemble a broadened one-

electron reversible wave centered about the average of the individual reduction potentials.

The value of n calculated from the slope of ip,c vs. pv

pv using Equation 6.1 will be less

than the number of individual one-electron redox steps contributing to the observed

wave.

The oxidation wave for the VV/IV couple can be observed using a glassy carbon

electrode instead of a mercury electrode. A wide potential window allows both waves for

the VV/IV and the VIV/III redox couples to be observed for [V4L6]8- in a single cyclic

voltammogram (Figure 6.6). The VV/IV quasireversible wave appears at 0.47 V vs.

Ag/AgCl, and the two waves are separated by a very large potential difference.

0.5 0.0 -0.5 -1.0 -1.5

-20

-10

0

10

V(4+/3+)

V(5+/4+)

Cu

rren

t, µ

A

Potential, V vs. Ag/AgCl

100 mV/s

Figure 6.6. Cyclic voltammogram of K7[Et4N⊂V4L6] in DMF at a glassy carbon electrode showing the two separate waves for the VV/IV oxidation and the VIV/III reduction at 100 mV/s.

- 133 -

Aqueous Solutions – [V(cam)3]2-

The VIV/III redox couple of this mononuclear model compound was studied with

cyclic voltammetry at a mercury drop electrode. The observed electrochemical behavior,

while highly reproducible, depended strongly on the composition and pH of the

electrolyte solution. The common feature between all systems was a diffusion-controlled

redox wave, corresponding to the reaction

[V(cam)3]2-

(aq) + e- [V(cam)3]3-

(aq).

In addition to the diffusion-controlled wave, various adsorption phenomena were

observed. Treatment of the mercury electrode surface with propanethiol (PrSH) in situ

had a noticeable effect on adsorption. To facilitate discussion, the conditions used are

divided into five categories:

(1) “Naked Electrode” – Low pH: Buffered with phosphate to pH < 7.

(2) “Naked Electrode” – High pH: Buffered with carbonate to pH > 8.

(3) TRIS Solutions – Buffered with TRIS to pH ≥ 8.

(4) Propanethiol Treatment

(5) Excess Ligand –Saturated H2cam in pH 9.5 carbonate buffer solution

A discussion of the behavior observed under these five different conditions will follow

the description of the results, including the influence of complex formation equilibria.

Naked Electrode – Low pH

Cyclic voltammograms in pH 6.7 KCl and pH 6.3 NaClO4 appear very similar.

Representative voltammograms for each solution at low (25 mV/s) and high (500 mV/s)

scan rates are shown in Figure 6.7. The dominant feature at all scan rates is a large

- 134 -

cathodic postpeak appearing as the potential is swept in the negative direction (cathodic

sweep); no corresponding anodic postpeak is observed in the reverse scan. This postpeak

shifts cathodically (to more negative potentials) as the scan rate is increased. At pH 6.7,

an adsorption prepeak is also observed during the initial cathodic sweep, overlapping

with the diffusion wave. The prepeak is not observed after the first cycle. Although an

overlapping prepeak is not observed at pH 6.3, a small peak is present during the first

cycle around -0.2 V that is absent from additional cycles. Additionally, at pH 6.3 there is

an additional feature in the anodic wave near -0.25 V that becomes more prominent at

Figure 6.7. Cyclic voltammograms for aqueous [V(cam)3]2- (pH < 7) measured with a mercury electrode.

0.0 -0.2 -0.4 -0.6 -0.8

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

pH 6.34

(1 M NaClO4)

v = 25 mV/s

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

Cycle 1

Cycle 2

0.0 -0.2 -0.4 -0.6 -0.8

-1

0

1

2

3

4

5

6

pH 6.34

(1 M NaClO4)

v = 500 mV/s

Curr

ent,

µA

Potential, V vs. Ag/AgCl

Cycle 1

Cycle 2

Cycle 3

Cycle 4

Cycle 5

-0.2 -0.4 -0.6 -0.8 -1.0

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

pH 6.73

(0.5 M KCl)

v = 25 mV/s

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

Cycle 1

Cycle 2

Cycle 3

Cycle 4

-0.2 -0.4 -0.6 -0.8 -1.0

-1

0

1

2

3

4

5

6

pH 6.73

(0.5 M KCl)

v = 500 mV/s

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

Cycle 1

Cycle 2

Cycle 3

Cycle 4

Cycle 5

- 135 -

higher scan rates (vv ≥ 200 mV/s). A similar oxidation process may account for the

“flattened tail” of the anodic wave at pH 6.7.

Repeated cycling of the electrode potential affects both the diffusion wave and the

adsorption postpeak (Figure 6.8). For the diffusion wave, the peak current ratio ip,c/ip,a

increased with additional cycles until it reaches a plateau. In particular, at 500 mV/s in

pH 6.7 KCl, the ratio holds constant at 0.95 by the eighth scan. The variation of the

cathodic to anodic peak current ratio ip,c/ip,a with repeated cycling primarily reflects the

increasing cathodic peak current for the diffusion wave, since the peak anodic current

remained relatively constant. For the cathodic adsorption postpeak, the peak current

decreased with additional cycles. It seems likely that the two processes are

interdependent. This behavior is consistent with reduction of one species that is

electroactive toward reduction both while solvated and while adsorbed on the mercury

surface, with rapid desorption of the reduced species once formed.

Figure 6.8. Effect of repeated cycling on: (a) the diffusion wave peak current ratio and (b) the cathodic postpeak current observed in the cyclic voltammogram of [V(cam)3]

2- at pH 6.7 (scan rate: 500 mV/s).

0 2 4 6 8 10 12 14 16

0.6

0.7

0.8

0.9

1.0

i p,c /

ip

,a

Cycle Number

0 2 4 6 8 10 12 14 160.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Pe

ak C

urr

en

t, µ

A

Cycle Number

a)

b)

- 136 -

Naked Electrode – High pH

Solutions buffered with carbonate to pH 8.7 (KCl) and pH 9.6 (NaClO4) exhibited

sharp adsorption postpeaks in both the cathodic and anodic sweep of their cyclic

voltammograms (Figure 6.9). These peaks are sharper and much less intense than the

enormous postpeak observed for solutions at lower pH. The postpeak intensity remained

essentially unchanged upon repeated cycling, but additional cycles resulted in peaks

slightly shifted (~8 mV) to more negative potentials from the peak observed in the first

cycle. For the pH 9.6 solution, the cathodic postpeak current increased linearly with the

scan rate. The small anodic postpeak is fairly well separated from the main anodic

diffusion wave at pH 9.6, but at pH 8.7 the anodic adsorption peak and diffusion wave

overlap. Two prepeaks were also observed for both solutions during the initial cathodic

sweep: one located around -0.3 V, and another overlapping with the main diffusion-

controlled wave at about -0.5 V. These prepeaks were not observed after the first cycle.

Figure 6.9. Cyclic voltammograms for K2[V(cam)3] measured with a mercury electrode in pH > 8 aqueous solutions buffered with carbonate.

0.0 -0.2 -0.4 -0.6 -0.8 -1.0

pH 9.6 (NaClO4)

Potential, V vs. Ag/AgCl

200 mV/s

500 mV/s

1003 mV/s

2 uA

0.0 -0.2 -0.4 -0.6 -0.8

pH 8.7 (KCl)

Potential, V vs. Ag/AgCl

1 uA

- 137 -

The [V(cam)3]2-/3- diffusion wave is quasireversible in both cases. In the first

cycle, the cathodic wave is shifted anodically and has a larger peak current compared to

subsequent cycles. In the first cycle, ip,c/ip,a = 1.1 – 1.2, whereas for every subsequent

cycle ip,c/ip,a = 0.9 – 1.0. Since additional cycles produce nearly identical waves, the

second cycle will be used for analysis. At pH 9.6, E1/2 = -0.48 V vs. Ag/AgCl (sat’d

NaCl). The peak separation was scan rate dependent, ranging from ∆Ep = 168 mV at

50 mV/s to ∆Ep = 336 mV at 1000 mV/s.

TRIS solutions

The cathodic postpeak decreased at pH 8.1, but a shoulder in the positive sweep

suggests an overlapping anodic postpeak is present (Figure 6.10). At pH 9.2, with

[TRIS] = 0.76 M, the postpeak disappears, but its shape suggests weak adsorption of

[V(cam)3]2- may occur.8 This buffer system should not be used with mercury for these

compounds, since its specific adsorption complicates the system unnecessarily.

Figure 6.10. Cyclic voltammograms for K2[V(cam)3] in aqueous TRIS/TRIS·HCl buffered at a) pH 8.1 and b) pH 9.2, with 0.5 M KCl supporting electrolyte (Hg electrode, scan rate: 200 mV/s).

-0.2 -0.4 -0.6 -0.8 -1.0

-2

-1

0

1

2

3

pH 8.1

[TRIS] = 0.17 M

Cu

rre

nt, µ

A

Potential, V vs. Ag/AgCl

-0.2 -0.4 -0.6 -0.8 -1.0

-1

0

1

2

pH 9.2

[TRIS] = 0.76 M

Cu

rrent,

µA

Potential, V vs. Ag/AgCl

a) b)

- 138 -

Propanethiol treatment

Alkanethiols (CnH2n+1SH) are known to undergo oxidative adsorption onto the

surface of mercury.9 Short-chain thiols (n ≤ 5) form bulk low-density films at open

circuit potential or at potentials more positive than -0.7 V vs. Ag/AgCl.9, 10 In contrast to

their longer chain counterparts that form self-assembled monolayers,11 these low-density

films are permeable to solvated species, and do not hinder electron transfer appreciably.

However, adsorption of the short-chain thiols will decrease the mercury surface tension,

reducing the free energy released upon adsorption of other species (and thus their binding

affinities), including [V(cam)3]2-.

Addition of PrSH directly to the analyte solution allowed modification of the

mercury electrode surface at thiol concentrations of 10 µM. At these PrSH

concentrations, the prepeaks were eliminated, even with a mercury drop less than a few

seconds old (Figure 6.11a). The adsorption postpeak intensity may be reduced by

allowing the mercury drop to “age” in the presence of PrSH at open circuit potential for

several minutes. At high pH, the sharp adsorption postpeaks broaden as the drop is

Figure 6.11. Cyclic voltammograms of aqueous K2[V(cam)3] (pH 9.8, 1 M NaClO4) with 10 µM PrSH measured at a mercury electrode (a) immediately after dispensing a fresh drop, and (b) after aging another mercury drop for 10 min. at open circuit (scan rate: 200 mV/s).

0.0 -0.2 -0.4 -0.6 -0.8 -1.0

-1.0

-0.5

0.0

0.5

1.0

1.5

Cu

rre

nt,

µA

Potential, V vs. Ag/AgCl

Cycle 1

Cycle 2

0.0 -0.2 -0.4 -0.6 -0.8 -1.0

-1.0

-0.5

0.0

0.5

1.0

1.5

Cu

rre

nt, µ

A

Potential, V vs. Ag/AgCl

Cycle 1

Cycle 2

a) b)

Fresh drop Aged 10 min.

- 139 -

“aged,” with a corresponding decrease in the peak current relative to baseline. After

10 min, the postpeaks were gone, leaving only the reversible VIV/III diffusion wave

(Figure 6.11b). The center of this wave is located at E1/2 = -0.48 V vs. Ag/AgCl at

pH 9.8, which corresponds to the VIV/III reduction potential for [V(cam)3]2-.

The surface coverage of PrSH on the mercury surface was estimated by

chronocoulometry to be about 6 x 10-11 mol cm-2. This measurement was made with

50 µM PrSH in aqueous carbonate buffer (pH 9.4) with 1 M NaClO4 by aging the

mercury drop for 10 min. at open circuit, followed by a double potential step from 0 V to

-1.0 V for 0.5 sec. and back to 0 V vs. Ag/AgCl, taking the difference between the two to

correct for double layer charging effects. The charge per unit area is 11 µC/cm2 from the

average of three measurements, and knowing n = 2 electrons for the reduction of PrSH on

Hg leads to the surface concentration given above. This number is about ten times lower

than the alkanethiol surface coverages reported by Majda and coworkers,12 possibly due

to the lower PrSH concentrations used here or the fact that PrSH accumulation was

performed at open circuit rather than at a fixed potential with a completed circuit.

In contrast, at pH 6.3 the large adsorption postpeak is not completely quenched by

a long soak in PrSH, although its intensity is reduced considerably. Even after aging the

Hg drop for two hours, a significant adsorption postpeak remained for the acidic systems.

This is consistent with the pH dependence of mercuric thiolate formation, with adsorption

strongly preferred in basic solution.13 At low pH, a longer drop age resulted in lower

peak currents in the second cycle compared to those observed with a fresh drop; the

original peak currents could be restored by forming a fresh mercury drop.

- 140 -

Eliminating the adsorption postpeaks at high pH leads to approximately reversible

waves, and the peak current is described by Equation 6.1. This equation predicts ip;c=pvip;c=

pv

should remain constant as long as the bulk analyte concentration CO* does not change.

With PrSH-treated electrodes at pH 9.8, a plot of ip;c=pvip;c=

pv vs. time shows this ratio is

certainly not constant (Figure 6.12). Instead, it decays over time as the solution

equilibrates. (Time is measured using the time stamp written in the data file for the

experiment; t = 0 is defined as the time of the first data file). The decay is significant –

after about 80 minutes, ip;c=pvip;c=

pv is half its original value. Since the electrode area remains

essentially constant over time, the observed decay must be due to a steady decrease in

analyte concentration. In the low pH PrSH system, however, the peak current of the

diffusion wave measured at a newly formed drop remained essentially constant over time.

Note that the solution remained under argon at all times. Thus, some homogeneous

decomposition process occurs in basic solution.

Figure 6.12. Time dependence of ip,c/v1/2 ratio for K2[V(cam)3] in aqueous solution (pH 9.8) with 10 µM

PrSH. For each measurement, the mercury drop electrode was aged for at least five minutes.

10 20 30 40 50 60 70 801.5

2.0

2.5

3.0

3.5

i p,c /

v1

/2,

µA

*(V

/s)-1

/2

Elapsed Time, minutes

- 141 -

Excess ligand

Solid H2cam was added to a 1 mM solution of K2[V(cam)3] in 1 M NaClO4,

buffered with carbonate to pH 9.5 (pH measured before adding H2cam and K2[V(cam)3]).

Some undissolved H2cam remained in the cell, indicating the solution was saturated with

camH2. With judicious choice of a potential range, a reversible [V(cam)3]2-/3- wave was

observed (Figure 6.13). No adsorption peaks were observed with a potential window

[Emax, Emin] = [-0.25, -0.725] V vs. Ag/AgCl. When the potential window was expanded

to [0, -1.0] V, additional broad features appeared. However, these are well separated

from the main diffusion wave, and could be avoided with a narrower potential range.

The wave is centered at E1/2 = -0.48 V vs. Ag/AgCl The observed peak currents

exhibited a linear dependence on pv

pv . After sitting under argon for about 3 hours, a 20%

decrease was observed in the peak cathodic current. This amount of decay was consistent

between three different scan rates.

Figure 6.13. Cyclic voltammograms of aqueous K2[V(cam)3] with excess H2cam measured with a mercury electrode using a (a) wide and (b) narrow potential window (scan rate: 200 mV/s). Prior to adding H2cam, the pH was 9.5 (carbonate buffer with 1 M NaClO4 supporting electrolyte).

0.0 -0.2 -0.4 -0.6 -0.8 -1.0-2

-1

0

1

2

Curr

ent, µ

A

Potential, V vs. Ag/AgCl

-0.2 -0.3 -0.4 -0.5 -0.6 -0.7-2

-1

0

1

2

Cu

rre

nt, µ

A

Potential, V vs. Ag/AgCl

a) b)

- 142 -

Discussion – Mononuclear complex

The behavior observed for aqueous [V(cam)3]2- at pH < 7 is due to

the large difference in affinities for the mercury electrode surface between

the VIV and VIII complexes, and is very similar to that for aqueous [V(cat)3]2-

with mercury electrodes reported in the literature.14 In that study, the authors reported

that the amount of adsorbed [V(cat)3]2- is limited only by the rate of diffusion to the

mercury surface, and not on accumulation potential. The favorable binding interaction

stabilizes the adsorbed VIV species, and thus more negative potentials are required to

reduce the adsorbed species. Following reduction of the adsorbed VIV complex, which

leads to the observed postpeak current, the VIII complex immediately desorbs from the

Hg surface, diffusing into solution.

With both oxidation states electroactive in the solvated state, this difference in

surface binding accounts for the behavior of both the diffusion wave and the adsorption

postpeak observed upon repeated cycling in Figure 6.8. To facilitate discussion, VIV

refers to [V(cam)3]2-, and VIII refers to [V(cam)3]

3-. The VIII formed at the end of the first

cathodic sweep rapidly desorbs, and only solvated VIII is present during the anodic return

sweep, which is oxidized to generate solvated VIV. Before the surface concentration of

VIV is able to return to its original value, the next cycle begins. With more solvated VIV

present at the start of the next cycle, a higher cathodic current is observed for the

diffusion-controlled reduction process, and a lower current is observed for the postpeak.

Each additional cycle begins with more VIV in solution and less VIV adsorbed on the

electrode compared to the previous cycle, eventually reaching a steady state after a large

number of cycles.

- 143 -

In basic solutions, changes in the electrode-solution interface lead to very

different features in the observed cyclic voltammograms. When a non-adsorbing

carbonate buffer is used, adsorption features no longer dominate the observed current

response, but large peak separations in the diffusion wave suggest sluggish electrode

kinetics are involved. Changing the buffer system to TRIS while maintaining a similar

pH dramatically changes the behavior, since TRIS specifically adsorbs onto mercury.

This highlights just how sensitive the electrochemical response can be for complexes

such as [V(cam)3]2- due to changes in solution conditions that affect the composition of

the electrochemical double layer.

The fact that PrSH treatment is able to suppress the stray peaks confirms they

originate from adsorbed electroactive species. Adsorption is suppressed not only by

PrSH treatment, but by excess H2cam as well. Catechol (H2cat) is known to adsorb onto

mercury,15 so its derivative H2cam should exhibit similar adsorption properties. Previous

aqueous electrochemical studies of [V(cat)3]2- derivatives published by Raymond and

coworkers were performed in the presence of excess ligand. In particular, Cooper et al.

merely note that for aqueous [V(cat)3]2-, reduction at a mercury drop “is not reversible in

solutions that lack excess catechol.”16 They attribute this “irreversible cyclic

voltammogram” to dissociation of catechol from the VIII species, which they expected to

exhibit a smaller stability constant than the higher oxidation states. No mention of

adsorption is made. However, the presence of excess ligand presumably prevented

electrochemical observation of the decomposition reaction in that study.

With interference from adsorption peaks successfully suppressed by PrSH, the

decomposition of solvated V(cam)32- may be studied without adding additional ligand. In

- 144 -

aqueous solution, several homogeneous chemical equilibria occur in parallel to

heterogeneous electron transfer and adsorption reactions at the electrode. These

homogeneous reactions are:

[V(cam)3]2¡ + H2O Ð [VO(cam)2]

2¡ + H2cam KVO =[VO(cam)2][H2cam]

[V(cam)2¡3 ]

[V(cam)3]3¡ + 2H+

Ð [V(cam)2]¡ + H2cam K3 =

[V(cam)¡2 ][H2cam]

[V(cam)3¡3 ][H+]2

H2cam Ð Hcam¡ + H+ Ka1 =[Hcam¡][H+]

[H2cam]

Hcam¡Ð cam2¡ + H+ Ka2 =

[cam2¡][H+]

[Hcam¡]

[V(cam)3]2¡ + H2O Ð [VO(cam)2]

2¡ + H2cam KVO =[VO(cam)2][H2cam]

[V(cam)2¡3 ]

[V(cam)3]3¡ + 2H+

Ð [V(cam)2]¡ + H2cam K3 =

[V(cam)¡2 ][H2cam]

[V(cam)3¡3 ][H+]2

H2cam Ð Hcam¡ + H+ Ka1 =[Hcam¡][H+]

[H2cam]

Hcam¡Ð cam2¡ + H+ Ka2 =

[cam2¡][H+]

[Hcam¡]

To assess the significance of the four equilibria under various conditions, one may

approximate the four homogeneous equilibrium constants by using literature data

reported for similar compounds. The vanadium(IV)-Tiron system reported by Buglyó

and Kiss17 was chosen to approximate KVO, while the iron(III)-DMB system reported by

Harris et al.18 was chosen to approximate K3 (DMB = N,N-dimethyl-2,3-

dihydroxybenzamide). For vanadium(IV)-Tiron at 0.2 M ionic strength, log KVO = -2.0,

pKa1 = 7.47, and pKa2 = 12.2. For iron(III)-DMB at 0.1 M ionic strength, log K3 = 12.0,

pKa1 = 8.42, and pKa2 = 12.1. Using these values, the following equilibrium relations are

obtained for the VIV and VIII concentration ratios:

log

µ

[V(cam)2¡3 ]

[VO(cam)2¡2 ]

¶

= 2 + log[H2cam]log

µ

[V(cam)2¡3 ]

[VO(cam)2¡2 ]

¶

= 2 + log[H2cam]

(6.4)

log

µ

[V(cam)3¡3 ]

[V(cam)¡2 ]

¶

= 2pH + log[H2cam]¡ 12log

µ

[V(cam)3¡3 ]

[V(cam)¡2 ]

¶

= 2pH + log[H2cam]¡ 12

(6.5)

Consider first the VIV equilibrium. If the concentration of ligand in solution is

small, most VIV exists as the vanadyl complex at equilibrium. In most experiments

- 145 -

described above, no H2cam was added when preparing the solution for analysis. In these

systems, ligand dissociation from the tris complex accounts for all free H2cam present in

solution. Most free ligand exists as H2cam at pH 7 and below. By setting [H2cam] =

[VO(cam)22-] and CM° = [V(cam)3

2-] + [VO(cam)22-], one can simplify Equation 6.4 and

solve the resulting quadratic equation for the equilibrium concentration of the vanadyl

product if the initial V(cam)32- concentration CM° is known. Using the above

assumptions, and with CM° = 10-3 M, 90% of the vanadium will be found as the vanadyl

complex at equilibrium, with only 10% of the original tris-bidentate complex remaining.

In a more basic solution, such as the pH ~9-10 solutions described above, most of

the free ligand will exist as Hcam-. Under these conditions, the VIV equilibrium can be

re-expressed as:16

[V(cam)3]2¡ +OH¡

Ð [VO(cam)2]2¡ + Hcam¡ K¤

VO = KVOKa1

Iw

[V(cam)3]2¡ +OH¡

Ð [VO(cam)2]2¡ + Hcam¡ K¤

VO = KVOKa1

Iw

where the ionization constant of water Iw = 10-14 at 25 °C. For DMB, pKa1 = 8.4, and

using logKVO = ¡2logKVO = ¡2 gives logK¤

VO = 3:6logK¤

VO = 3:6. Performing a similar calculation with

[V(cam)32-]0 = 10-3 M reveals that at pH 10.4, VO(cam)2

2- is the only vanadium species

present at equilibrium. (This estimate considers only one equilibrium, with Hcam- as the

only free ligand species present in solution). The decomposition reaction is slow

compared to the timescales used in cyclic voltammetry.17

The decomposition of [V(cam)3]2- into [VO(cam)2]

2- and Hcam- in basic solution

accounts for the observed decay of ip;c=pvip;c=

pv over time shown in Figure 6.12. At

equilibrium, nearly all VIV will be found as the vanadyl species according to the above

- 146 -

calculations. Nearly two hours after the solution was first prepared,9 over half of the

vanadium in solution was converted to the vanadyl complex. Despite this, the slope in

Figure 6.12 remained essentially constant, indicating the system was far from

equilibrium. Addition of H2cam results in a higher concentration of [V(cam)3]2- at

equilibrium. This is consistent with the slower decay of ip;c=pvip;c=

pv observed for solutions

saturated with H2cam.

Next consider the VIII equilibrium, using data reported for [Fe(DMB)3]3- to

estimate its formation constants. At pH ≤ 7, the ligand exists primarily as the fully

protonated species H2cam. If no free H2cam was added to the system, [H2cam] =

[V(cam)2-]. If the initial V(cam)3

3- concentration is 1 mM, then at equilibrium 90% of the

complex dissociates into V(cam)2- at pH 7. Therefore, following the reduction of

[V(cam)3]2- to [V(cam)3]

3- at pH ≤ 7, the ligand dissociation process should also be

considered when interpreting the results. Furthermore, if [V(cam)2]- is electroactive

towards oxidation, its formation at pH < 7 could account for the anodic feature observed

at higher scan rates in Figure 6.7.

At pH ≥ 9.5, the ligand primarily exists as the monobasic species Hcam-, and

dissociation primarily proceeds according to:

[V(cam)3]3¡ + H+

Ð [V(cam)2]¡ +Hcam¡ K¤

3 = K3Ka1[V(cam)3]3¡ + H+

Ð [V(cam)2]¡ +Hcam¡ K¤

3 = K3Ka1

with logK¤

3 = logK3 ¡ pKa1 = 3:6logK¤

3 = logK3 ¡ pKa1 = 3:6. Using this equilibrium, the vanadium(III) complex

ratio is given by the following expression:

log

µ

[V(cam)3¡3 ]

[V(cam)¡2 ]

¶

= pH+ log [Hcam¡]¡ 3:6log

µ

[V(cam)3¡3 ]

[V(cam)¡2 ]

¶

= pH+ log [Hcam¡]¡ 3:6

(6.6)

9 While the last data point in Figure 6.12 is at t = 80 min, the solution was actually prepared approximately 30 min prior to t = 0, since the degassing process had to be completed before the first measurement.

- 147 -

If no free H2cam was added to the system, [H2cam] = [V(cam)2-]. If the initial

[V(cam)3]3- concentration is 1 mM, then at equilibrium only 3% of the complex

dissociates into [V(cam)2]- at pH 9.5, and only 2% at pH 10. Therefore, during cyclic

voltammetry experiments with basic solutions the homogeneous dissociation process of

the VIII product can be neglected.

The reversible waves observed with PrSH treatment and with excess ligand in

basic aqueous buffer exhibit the same half-wave potentials, indicating their half-wave

potentials correspond to the standard reduction potential for the VIV/III

couple. This

reduction potential is -0.48 V vs. Ag/AgCl (sat’d NaCl), and calibration of the reference

electrode potential gives E°’ = -0.56 V vs. SCE. This is 0.16 V more positive than the

reduction potential for the aqueous tris(catechol) complex reported in the literature,2

which means reduction to the vanadium(III) complex is easier for the catecholamide

complex than with unsubstituted catechol ligands. The electron-withdrawing amide

group reduces the electron density surrounding the metal, leading to the observed anodic

shift. In contrast, the unsubstituted catechol dianion is highly electron rich, making its

VIII complex a very strong reductant.

Aqueous Solutions – [V2LH3]4-

and [V4L6]8-

.

Complex redox behavior is exhibited by both [V2LH

3]4- and [V4L6]

8- in aqueous

solution. In addition to adsorption processes similar to those observed for [V(cam)3]2-,

additional reduction waves appear at more negative potentials which do not exhibit

typical quasireversible behavior.

- 148 -

The dinuclear complex [V2LH

3]4- exhibits two well separated sets of redox waves

during cyclic voltammetry (Figure 6.14a). The first is a quasireversible diffusion wave

centered at E1/2 = -0.57 V vs. SCE, with an overlapping adsorption-controlled wave

present in the first cycle only. The adsorption-controlled wave may be isolated by

subtracting the current-voltage data for cycle 2 from cycle 1, generating a peak such as

that illustrated in Figure 6.14b. The difference peak current is linearly dependent on the

scan rate, consistent with an adsorption-controlled process present in cycle 1 but absent in

cycle 2 (and additional cycles). Treatment with PrSH reduces the interference from this

adsorption process to a certain extent, although a small peak remains in the difference

wave between the first and second cycles, indicating reactant adsorption still occurs.

Plotting the peak cathodic current of the first redox wave in cycle 2 versus the

square root of the scan rate yields a straight line, characteristic of a diffusion-controlled

process. Assuming D2 = 3 x 10-6, the slope gives a value of n = 1, similar to the DMF

results. A second, smaller redox wave appears at higher scan rates (vv > 0.2 V/s), with a

peak current around -1.23 V vs. SCE. The nature of this wave is not clear – plotting the

Figure 6.14. (a) Cyclic voltammograms for [V2LH

3]4- in aqueous solution (pH 9.8). The first and second

cycles are shown in different colors. (b) Plot of (cycle 1) – (cycle 2) difference, illustrating the absorption peak. The red line is the absorption peak baseline used for analysis.

0.0 -0.5 -1.0 -1.5

1.0 V/s

0.5 V/s

0.2 V/s

Potential, V vs. SCE

Cycle 1

Cycle 2

1 uAa)

-0.2 -0.4 -0.6 -0.8 -1.0

0.0

0.5

Curr

ent

Diffe

rence

, µ

A

Potential, V vs. SCE

b)

- 149 -

peak current against either pv

pv or vv yields a curve which may be considered a straight line

in either case.

For the tetranuclear complex [V4L6]8-, at least three separate reduction waves

appear in the cathodic sweep, with a possible fourth wave appearing in the solvent

reduction tail (Figure 6.15). The first three waves seem to be diffusion controlled, as

suggested by linear plots of peak current versus pv

pv . The first wave is quasireversible,

centered at E1/2 = -0.57 V vs. SCE. The next three waves are irreversible, with the last

exhibiting erratic behavior. Since no additional waves were observed for [V4L6]8- in

DMF, these additional waves are most likely due to various side reactions specific to

aqueous solutions, as opposed to communication between redox sites. The fact that four

peaks can be observed for the cluster with four vanadium redox sites may simply be a

coincidence. Since simple cyclic voltammograms can be observed in DMF, the

complicated aqueous behavior will not be explored further.

-0.5 -1.0 -1.5

-1

0

1

2

Cu

rre

nt, µ

A

Potential, V vs. SCE

Cycle 1

Cycle 2

Figure 6.15. Cyclic voltammogram of [V4L6]8- in H2O (pH 9). Scan rate: 0.2 V/s.

- 150 -

Discussion – Supramolecular complexes

The half-wave potentials of the quasireversible diffusion waves for [V2LH

3]4- and

[V4L6]8- are both equal to E1/2 = -0.57 V vs. SCE = -0.49 V vs. Ag/AgCl. This is very

close to the VIV/III reduction potential of -0.48 V vs. Ag/AgCl for the mononuclear

[V(cam)3]2- complex in basic solution, confirming that the additional vanadium metal

centers have no influence on the individual VIV/III reduction potentials in aqueous

solution. This is also true for DMF solutions, where the reduction potentials for the

dinuclear and tetranuclear complexes differ by only 0.01 V. However, the potentials of

the supramolecular complexes in DMF are 0.38-0.39 V more negative than in aqueous

solutions. This is due to stabilization of the VIII complex in aqueous solution from

hydrogen bonding, which acts to delocalize the excess negative charge into the

surrounding solvent. DMF is aprotic, and no hydrogen bonding interactions exist to

stabilize the anion. Similar solvent effects were observed for reduction of [V(cat)3]2-,

where the VIV/III redox wave was 0.14 V more negative in aprotic CH3CN compared to its

aqueous reduction potential; the smaller cathodic shift in that particular case was due to

stabilization from the Et3NH+ counterions that form strong hydrogen bonds with the

anionic complex in acetonitrile.2

Summary

When redox-active metal ions are used to assemble large supramolecular

complexes such as the M2LH

3 helicate or the M4L6 tetrahedron, there is no interaction

between the individual redox sites. The individual reduction potentials differ only due to

statistical (entropic) factors, with the observed half-wave potentials nearly identical for

- 151 -

the mononuclear, dinuclear, and tetranuclear vanadium catecholamide complexes.

Hydrogen bonding stabilizes the anionic vanadium(III) complexes in aqueous solution,

but the observed electrochemical response in water is complicated by adsorption to the

mercury electrode. Adsorption can be minimized by treating the electrode with

propanethiol, or by adding excess free ligand if sufficiently soluble; basic conditions

should be maintained by buffering with non-adsorbing carbonate. Both VV/IV and VIV/III

redox couples lead to quasireversible waves in DMF for the multiple non-interacting

redox sites, with the two waves separated by nearly 1.5 V. For [V4L6]12-, oxidation of the

VIII ions to VV releases eight electrons in two sets of four one electron reactions. This

suggests potential applications as an electron transport mediator.

Experimental

General considerations

Reagents were obtained from commercial suppliers and used without further

purification unless noted otherwise. Reactions were carried out under argon using

standard Schlenk techniques, and a glove box continuously purged with nitrogen was

used to manipulate and store all air-sensitive solids. Water was purified by passing

house-distilled water through a Millipore Milli-Q system until the effluent stream

conductivity reached 18 MΩ·cm. Dry CH2Cl2 and THF were obtained by flowing

through a bed of activated alumina under nitrogen. When noted, methanol (MeOH),

triethylamine (Et3N), and CH3CN were distilled from CaH2 under N2. H4L and H4LH

were synthesized according to literature procedure.19 Solvents were removed from air-

sensitive systems with a vacuum pump connected to a Schlenk line, and from non-air-

- 152 -

sensitive systems with a rotary evaporator. Electrospray ionization (ESI) mass

spectrometry was performed on K4[V2LH

3] by Robert M. Yeh on a Finnigan LCQ

quadrupole ion trap mass spectrometer equipped with a microspray ionization source, and

on K7[Et4N⊂V4L6] by Michael D. Pluth on a Waters Q-TOF API mass spectrometer

equipped with a Z-spray source. All other mass spectrometry and elemental analyses

were performed by the Mass Spectrometry Laboratory and Microanalysis Facility,

respectively, in the College of Chemistry at the University of California, Berkeley.

Synthetic procedures

N-methyl-2,3-dimethoxybenzamide (Me2cam). To 6.1 g (33 mmol) of 2,3-

dimethoxybenzoic acid was added 40 mL (55 mmol) of thionyl chloride (SOCl2) and

1 mL of DMF. The reaction mixture was allowed to stir for 12 hours under a moderate

flow of N2, and the exhaust stream containing the gaseous HCl and SO2 products passed

through a bubbler filled with aqueous NaOH and vented into a fume hood. The resulting

solid was dried under vacuum for one hour, yielding the acid chloride as a yellow-white

residue. The flask was filled with argon, and 150 mL of dry THF was added to dissolve

the residue. The resulting solution was transferred via cannula to a dry addition funnel

temporarily affixed to an empty flask to minimize exposure to air. To a separate 500 mL

Schlenk flask was added 8.4 mL (97 mmol) of methylamine (40% solution in H2O),

50 mL of THF, and a magnetic stirbar. This flask was affixed to the addition funnel, and

its contents were allowed to cool to -10 °C in an ice/salt bath while stirring vigorously.

The acid chloride solution was added dropwise over the course of two hours, and the

reaction mixture was vigorously stirred for two additional hours at -10 °C. Solvent was

- 153 -

removed, and excess water was removed by azeotropic evaporation with ethanol to form

an oily yellowish residue. The residue was dissolved in 100 mL of CH2Cl2, washed with

1 M HCl (1 x 350 mL), 1 M NaOH (1 x 350 mL), and brine (1 x 200 mL). The organic

layer was collected and its solvent removed to form an oil. Column chromatography on

silica gel with 95%/5% CHCl3/MeOH (500 mL) was performed, with the desired product

collected in the first of two bands. The volume was reduced to 100 mL before drying

with MgSO4 and filtering through a fine frit. The remaining solvent was removed with a

rotary evaporator to form an oil, which slowly solidified overnight at 4 °C.

Recrystallization from CH2Cl2/hexane yielded 3.6 g (55%) of large colorless

orthorhombic crystals, mp 79-80 °C. IR (Nujol): 3329, 1638 cm-1. 1H NMR (400 MHz,

CDCl3): δ 3.02 (d, 3H, J=4.8 Hz), 3.89 (br s, 6H), 7.03 (dd, 1H, J1=1.6 Hz, J2=8.0 Hz),

7.14 (t, 1H, J=8.0 Hz), 7.70 (dd, 1H, J1=1.6 Hz, J2=8.0 Hz), 7.93 (br s, 1H). 13C NMR

(100 MHz, CDCl3): δ 26.59, 56.04, 61.30, 115.19, 122.75, 124.37, 126.76, 147.41,

152.53, 165.87. MS (FAB+): m/z 196.1 (M+), 165.1 ([M–CH3NH·]+). Anal. Calcd.

(found) for C10H13N1O3: %C, 61.53 (61.36); H, 6.71 (6.83); N, 7.18 (7.16). The structure

was confirmed by X-ray crystallography.

N-methyl-2,3-dihydroxybenzamide monohydrate (H2cam·H2O). To a solution of

1.70 g (8.71 mmol) of Me2cam in 150 mL of CH2Cl2 cooled to -78 °C was added 8.7 g

(35 mmol) of BBr3 via syringe. (CAUTION: BBr3 is very corrosive, and violently reacts

with water and alcohols). The reaction mixture was allowed to reach room temperature

and stirred overnight. Solvent was removed under vacuum, leaving a pale yellow

residue. The flask was uncapped, immersed in liquid nitrogen, and 200 mL of methanol

- 154 -

was added in small portions. The open flask was heated at reflux for 8 hours,

maintaining a minimum volume of 75 mL with periodic additions of fresh methanol. The

solvent was then allowed to evaporate, and 150 mL of H2O was added to the grey solid.

The reaction mixture was boiled on a hot plate for three hours, the volume was reduced to

25 mL, and the mixture was cooled in an ice bath to 0 °C. The white precipitate was

collected on a frit, washed with water (4 x 15 mL), and dried overnight under vacuum to

yield 1.36 g (84%) of off-white needles. 1H NMR (d6-acetone, 400 MHz): δ 2.93 (d, 1H,

J=4.4 Hz), 6.72 (t, 1H, J=8.0 Hz), 6.97 (dd, 1H, J1=1.2 Hz, J2=8.0 Hz), 7.23 (dd, 1H,

J1=1.2 Hz, J2=8.0 Hz), 7.7 (br s, 1H), 8.1 (br s, 1H), 13.2 (br s, 1H). 13C NMR (d6-

acetone, 100 MHz): δ 25.46, 114.47, 116.46, 118.08, 118.23, 146.45, 150.00, 170.91.

MS (FAB+): m/z 168.1 (MH+). Anal. Calcd. (found) for C8H9N1O3·H2O: %C, 51.89

(51.95); H, 5.99 (6.03); N, 7.56 (7.51). Note: If further purification is desired, camH2

may be slowly recrystallized from hot water to yield off-white flakes (needles merged

together). The structure was confirmed by X-ray crystallography.20

K2[V(cam)3]·(CH3OH). A slurry of 285 mg (1.54 mmol) of H2cam in 45 mL of MeOH

was prepared, and addition of 1.8 mL (0.90 mmol) of 0.5 M KOH in methanol caused the

solid to dissolve, forming a pale pinkish-brown solution. A solution of 120 mg

(0.45 mmol) of vanadyl bis-acetylacetonate (VO(acac)2) in 10 mL MeOH was added via

cannula, immediately causing the color to change to intense blue-violet. The reaction

mixture was stirred for 90 min at room temperature. Solvent was removed, and the black

residue was dried under vacuum for one hour. The solid was redissolved in 5 mL of

methanol, and addition of 60 mL of EtOAc/Et2O (1:1) formed a powdery precipitate.

- 155 -

The solid was collected on a fine frit, washed with cold EtOAc/Et2O (3 x 20 mL) and

Et2O (2 x 20 mL) and dried overnight under vacuum to yield 225 mg (76%) of fine black

powder. MS(ESI-): 585.0 (KM-), 569.1 (NaM-), 547.1 (HM-), 273.1 (M2-). UV-visible

(H2O): λmax, nm (εmax, M-1s-1) 320 (11700), 425 (sh, 3200), 590 (5020), 680 (sh, 4600).

Anal. Calcd. (found) for C24H21K2N3O9V·(CH3OH): %C, 45.73 (45.62); H, 3.84 (3.91);

N, 6.40 (6.28). IR (Nujol) showed no V=O stretch near 1000 cm-1. EPR (9:1

MeOH/EtOH, 8 K): see Figure 6.3.

K4[V2LH

3]·3(CH3OH)·2H2O. The title compound was synthesized from 350 mg

(0.92 mmol) of H4LH, 160 mg (0.61 mmol) of VO(acac)2, and 2.5 mL (1.2 mmol) of

0.5 M KOH in methanol following a procedure similar to that used to prepare

K2[V(cam)3]·CH3OH. After stirring for three hours at room temperature, the volume was

reduced to 5 mL. Ethyl acetate was steadily added via cannula until solid began to form

(100 mL total), and 25 mL of Et2O was added to form a precipitate suspended in solution.

The flask was immersed in an ice bath for 30 min, and the black solid was collected on a

frit, washed with ethyl acetate (3 x 20 mL) and Et2O (2 x 15 mL), and dried overnight

under vacuum to yield 420 mg (91%) of coarse black solid. Anal. Calcd. (found) for

C60H36K4N6O18V2·3CH3OH·2H2O: %C, 49.80 (49.84); H, 3.45 (3.45); N, 5.53 (5.36).

UV-visible (H2O): λmax, nm (εmax, M-1s-1) 432 (sh, 9420), 583 (11900), 666 (sh, 11100).

IR (Nujol) showed no V=O stretch near 1000 cm-1. The structure was confirmed by

X-ray crystallography (see Figure 6.4, experimental details listed separately below).

- 156 -

(Et3NH)4[V213]·3H2O. To a suspension of 100 mg (0.26 mmol) of H4LH in 40 mL of

distilled methanol was added 0.3 mL (2 mmol) of distilled Et3N. A solution of 46 mg

(0.18 mmol) of VO(acac)2 in 5 mL of distilled methanol was added to the reaction

mixture via cannula, and the color immediately changed to an intense blue. The solution

was stirred for 5 hours at 50 °C, followed by stirring at room temperature for an

additional 12 hours. Removal of solvent afforded a black residue, which was dried for

two hours under vacuum. Recrystallization from CH2Cl2/hexane overnight afforded

extremely thin black plates. The solid was collected on a frit, washed with hexane

(2 x 30 mL) and dried overnight under vacuum to yield 120 mg (81%) of black solid.

Anal. Calcd. (found) for C84H100N10O18V2·3H2O: C, 59.57 (59.74); H, 6.31 (6.09);

N, 8.27 (8.19). UV-visible: λmax, nm (εmax, M-1s-1) 438 (12,900), 578 (17,100), 660 (sh,

16,000). IR (Nujol) showed no V=O stretch near 1000 cm-1. EPR (9:1 MeOH/EtOH,

8 K): see Figure 6.3.

K7[Et4N⊂⊂⊂⊂V4L6]. In a 250 mL Schlenk flask, 1.00 g (2.32 mmol) of H4L and 411 mg

(1.55 mmol) of VO(acac)2 were combined as solids and placed under argon. To this was

added 125 mL of degassed methanol via cannula while stirring, and the resulting mixture

consisted of light colored solid suspended in a very dark blue solution. A stock solution

with 100 mM of Et4NCl in methanol was separately prepared, and 3.8 mL (0.38 mmol) of

this solution was added to the stirred suspension, and the reaction mixture was degassed

via four pump/fill cycles. After stirring at room temperature for 25 min, 6.2 mL

(3.1 mmol) of 0.5 M KOH in methanol was added via syringe, and the reaction mixture

was degassed again immediately after addition of base via several pump/fill cycles. The

- 157 -

nearly black reaction mixture was heated at reflux overnight under argon. After allowing

the mixture to cool to room temperature, the black reaction mixture was filtered using a

Buchner funnel to remove undissolved solids, degassed again, and 150 mL of 1:1

acetone/Et2O was added via cannula to precipitate a black solid. This solid was collected

on a frit, washed with 1:1 acetone/Et2O (3 x 50 mL) and petroleum ether (2 x 50 mL),

and dried overnight in vacuo to yield 1.05 g (83%) of grainy black solid. Anal. Calc.

(found) for C152H104N13O36K7V4·(H2O)6: %C, 55.76 (55.89); H, 3.57 (3.61); N, 5.56

(5.54). UV-visible (H2O): λmax, nm (εmax, M-1s-1) 435 (18150), 593 (24680), 683 (sh,

22420). IR (Nujol) showed no V=O stretch near 1000 cm-1.

X-Ray crystallography

Black crystals of K4[V2LH

3]·6.7DMF·Et2O·0.3H2O were grown by vapor diffusion

of ether into a slightly wet DMF solution of K4[V2LH]3 at 4 °C. Crystal data were

collected using a Bruker SMART diffractometer21 equipped with a CCD area detector

with graphite monochromated Mo Kα radiation (λ = 0.71073 Å). Frames corresponding

to an arbitrary hemisphere of data were collected using ω scans of 0.3° counted for a total

of 30 seconds per frame at T = -151 °C. Peak integrations, cell refinement, and data

reduction were performed by using the Bruker SAINT software package.22 Data were

corrected for Lorentz and polarization effects. The compound crystallizes in monoclinic

space group P21/n, with a = 26.129(5), b = 12.842(2), c = 29.694(5) Å, β = 99.738(2)°,

V = 9820(3) Å3, Z = 4. An empirical absorption correction was applied using

SADABS.23 The structure was solved by direct methods (SIR92) and expanded using

Fourier techniques using the teXsan crystallographic software package.24 Further least

- 158 -

squares refinement to model solvent disorder was done in SHELXL-97.25 All non-

hydrogen atoms, excluding solvent molecules, were refined anisotropically; solvent

atoms were refined with isotropic thermal parameters. The positions of the hydrogen

atoms were included in the structure factor calculation but were not refined. Additional

experimental details for this structure are listed in Appendix 3.

Five of the seven DMF molecules in each asymmetric unit were disordered. Four

of these five were modeled by allowing each DMF to partially occupy two positions. For

three of the four pairs of DMF molecules, the thermal parameters were refined with the

occupancy fixed. The atoms in the two parts of the fourth DMF pair were constrained to

lie in a rigid group corresponding to an idealized DMF, and the occupancy was refined

with fixed isotropic thermal parameters. The fifth disordered DMF showed substitutional

disorder, and was modeled as a DMF with 70% occupancy and a water with 30%

occupancy. Full matrix least squares with 925 parameters yielded R1 = 0.088 for 5711

reflections with I > 2σ(I).

Electrochemical measurements

All electrochemical experiments were carried out at ambient temperature (~22 °C)

using a Bioanalytical Sciences BAS-100A Electrochemical Analyzer. No corrections

were made for uncompensated resistance or junction potential, but uncompensated

resistance was estimated using the “IR Test” function in the control software. Analyte

concentrations of 0.3-1.1 mM were used. Unless noted otherwise, a Princeton Applied

Research EG&G Model 303A Static Mercury Drop Electrode (SMDE) apparatus was

used as the working electrode. A platinum wire served as the counter electrode. A

- 159 -

silver/silver chloride reference electrode filled with saturated NaCl/AgCl was used. The

potential of the reference electrode was -0.080 V vs. SCE. This was measured in aqueous

electrolyte using a voltmeter and a commercial electrode of known potential, and

confirmed using [CoCp2]+/0 as a reference compound in DMF with 0.1 M

tetrabutylammonium hexafluorophosphate (Bu4NPF6). For measurements with glassy

carbon, the working electrode was a BAS glassy carbon disk electrode (3 mm diameter)

polished with 0.1 µm alumina before use, and a coiled platinum wire counter electrode

was used. A similar Ag/AgCl reference electrode filled with saturated NaCl was used

with glassy carbon, and exhibited the same potential as the electrode used with mercury.

Unless noted, a fresh drop of mercury was dispensed immediately before each

experiment. Samples were bubbled with inert gas while stirring for at least twenty

minutes before any measurements, and for at least 30 s between each individual

experiment. A steady stream of argon flowed over the solution surface when not

bubbling. DMF solutions contained 0.1 M Bu4NPF6 as the supporting electrolyte. For

aqueous solutions, electroanalytical grade KCl (0.5 M) or reagent grade NaClO4 (1 M)

was used as the supporting electrolyte. Solution pH was buffered using potassium or

sodium phosphate (pH 6.34, 6.73), Tris/HCl (pH 8.13, 9.16), or NaHCO3/Na2CO3

(pH 8.73, 9.53, 9.64, 9.79) at 0.1 M buffer ionic strength. The argon used for aqueous

measurements was first passed through a gas scrubbing tower filled with aqueous

VCl2/HCl and then a bubbler filled with 1 M aqueous NaCl.

Experiments with propanethiol-modified mercury electrodes were performed with

the following procedure. Propanethiol was added to a prepared NaClO4/carbonate buffer

solution, and this solution was either used to prepare the analyte solution, or was added to

- 160 -

an existing analyte solution via syringe, followed by bubbling with argon for 4 min while

stirring. In either case, [PrSH] ≈ 10 µM in the actual analyte solution. Mercury drops

were “aged” as follows: after the normal 30 s bubble/stir cycle between experiments, the

solution was allowed to settle for 1 min, and at least five drops were dispensed and

dislodged to remove any contaminants from the capillary tip. A new drop was dispensed,

and the system was left undisturbed without potential control. Time was measured using

a stopwatch. After the desired time elapsed, the experiment was started.

Digital cyclic voltammetry simulations were performed using BAS DigiSim.26

References

1. Seeber, G.; Tiedemann, B. E. F.; Raymond, K. N., “Supramolecular Chirality in Coordination Chemistry.” In Top. Curr. Chem., Reinhoudt, D. N.; Crego-Calama, M., Eds. Springer: Berlin, 2006; Vol. 265, pp 147-183.

2. Cooper, S. R.; Koh, Y. B.; Raymond, K. N., “Synthetic, Structural, and Physical Studies of Bis(triethylammonium) Tris(catecholato)vanadate(IV), Potassium Bis(catecholato)oxovanadate(IV), and Potassium Tris(catecholato)vanadate(III).” J.

Am. Chem. Soc. 1982, 104, 5092-5102.

3. Bard, A. J.; Faulkner, L. R., Electrochemical Methods: Fundamentals and

Applications. 2nd ed.; John Wiley & Sons: Hoboken, 2001.

4. Branca, M.; Micera, G.; Dessi, A.; Sanna, D.; Raymond, K. N., “Formation and Structure of the Tris(catecholato)vanadate(IV) Complex in Aqueous Solution.” Inorg.

Chem. 1990, 29, 1586-1589.

5. a) Yeh, R. M.; Ziegler, M.; Johnson, D. W.; Terpin, A. J.; Raymond, K. N., “Imposition of Chirality in a Dinuclear Triple-Stranded Helicate by Ion Pair Formation.” Inorg. Chem. 2001, 40, 3922-3935; b) Scherer, M.; Caulder, D. L.; Johnson, D. W.; Raymond, K. N., Angew. Chem. Int. Ed. 1999, 38, 1588-1592.

6. Welty, J. R.; Wicks, C. E.; Wilson, R. E.; Rorrer, G., Fundamentals of Momentum,

Heat, and Mass Transfer. 4th ed.; John Wiley & Sons: New York, 2001.

7. Caulder, D. L.; Powers, R. E.; Parac, T. N.; Raymond, K. N., “The Self-Assembly of a Predesigned Tetrahedral M4L6 Supramolecular Cluster.” Angew. Chem. Int. Ed.

1998, 37, 1840-1842.

- 161 -

8. Wopschall, R. H.; Shain, I., “Effects of Adsorption of Electroactive Species in Stationary Electrode Polarography.” Anal. Chem. 1967, 39, 1514-26.

9. Stevenson, K. J.; Mitchell, M.; White, H. S., J. Phys. Chem. B 1998, 102, 1235-1240.

10. Muskal, N.; Mandler, D., “Thiol self-assembled monolayers on mercury surfaces: the adsorption and electrochemistry of omega-mercaptoalkanoic acids.” Electrochim.

Acta 1999, 45, 537-548.

11. Slowinski, K.; Chamberlain, R. V.; Miller, C. J.; Majda, M., J. Am. Chem. Soc. 1997, 119, 11910-11919.

12. Slowinski, K.; Chamberlain, R. V.; Miller, C. J.; Majda, M., “Through-Bond and Chain-to-Chain Coupling. Two Pathways in Electron Tunneling through Liquid Alkanethiol Monolayers on Mercury Electrodes.” J. Am. Chem. Soc. 1997, 119, 11910-11919.

13. Birke, R. L.; Mazorra, M., “A Study of the Electrochemical Characteristics of some Thiols by Differential Pulse Polarography and Other Electrochemical Techniques.” Anal. Chim. Acta 1980, 118, 257-269.

14. Ivanov, V. D.; Kaplun, M. M., “Studying the adsorption accumulation of vanadium-catechol complexes at a mercury electrode.” Zhurnal Anal. Khim. 1997, 52, 362-368.

15. Sarangapani, S.; Venkatesan, V. K., “Adsorption of Phenols at Mercury-Solution Interface.” Proc. Indian Natn. Sci. Acad. A 1983, 49, 124-142.

16. Cooper, S. R.; Koh, Y. B.; Raymond, K. N., J. Am. Chem. Soc. 1982, 104, 5092-5102.

17. Buglyo, P.; Kiss, T., “Formation of a Tris Complex in the Vanadium(IV)-Tiron System.” J. Coord. Chem 1991, 22, 259-268.

18. Harris, W. R.; Carrano, C. J.; Cooper, S. R.; Sofen, S. R.; Avdeef, A. E.; McArdle, J. V.; Raymond, K. N., J. Am. Chem. Soc. 1979, 101, 6097-6104.

19. a) Kersting, B.; Meyer, M.; Powers, R. E.; Raymond, K. N., J. Am. Chem. Soc. 1996, 118, 7221-7222; b) Caulder, D. L.; Powers, R. E.; Parak, T. N.; Raymond, K. N., Angew. Chem. Int. Ed. 1998, 37, 1840-1842.

20. Escalada, J.; Freedman, D.; Werner, E. J., “2,3-Dihydroxy-N-methylbenzamide monohydrate.” Acta Crystallogr. Sect. E 2004, E60, o1296-o1298.

21. SMART Area Detector Software Package, v5.052d; Bruker Analytical X-ray Systems, Inc.: Madison, WI, 1999.

22. SAINT: SAX Area-Detector Integration Program, v7.01A; Bruker Analytical X-ray Systems, Inc.: Madison, WI, 2002.

- 162 -

23. Blessing, R. H., “An Empirical Correction for Absorption Anisotropy.” Acta

Crystallogr. Sect. A 1995, 51, 33-38.

24. teXsan: Crystal Structure Analysis Package, Molecular Structure Corp.: The Woodlands, TX, 1992.

25. Sheldrick, G. M. SHELXL97, Universität Göttingen: 1997.

26. Rudolph, M.; Feldberg, S. W. DigiSim, v3.03b; Bioanalytical Systems, Inc.: West Lafayette, IN, 2004.

- 163 -

APPENDIX 1

Supplementary Spectra for Chapter 4

2D NMR spectra for [RuCn⊂⊂⊂⊂Ga4L6]12-

Figure A1.1. Portion of the 2D TOCSY 1H NMR spectrum (D2O, 400 MHz) of

[RuC4⊂Ga4L6]12- showing the cross peaks between encapsulated RuC4 resonances.

- 164 -

Figure A1.2. Portion of the 2D COSY 1H NMR spectrum (500 MHz, D2O) of

[RuC6⊂Ga4L6]12- showing the cross peaks between encapsulated RuC6 resonances.

- 165 -

Figure A1.3. Portion of the 2D COSY 1H NMR spectrum (500 MHz) of

[RuC10⊂Ga4L6]12- showing the cross peaks between encapsulated RuC10 resonances.

- 166 -

Figure A1.4. Portion of the 2D COSY 1H NMR spectrum (D2O, 500 MHz) for

[RuC10⊂Ga4L6]12-, showing the coupling between host protons.

- 167 -

Figure A1.5. Portion of the 2D NOESY 1H NMR spectrum (D2O, 400 MHz) of

[RuC4⊂Ga4L6]12-, showing the cross peaks between host and guest signals.

- 168 -

Figure A1.6. Portion of the 2D NOESY 1H NMR spectrum (D2O, 400 MHz) of

[RuC10⊂Ga4L6]12-, showing the cross peaks between host and guest signals.

- 169 -

ESI MS for [RuCn⊂⊂⊂⊂Ga4L6]12-

Figure A1.7. Portion of the ESI- mass spectrum of [RuC4⊂Ga4L6]12- in 75% H2O with

25% methanol. Peaks from fragments containing the host-guest complex in -3,-4, and -5

charge states are shown here.

- 170 -

Figure A1.8. Portion of the ESI- mass spectrum of [RuC6⊂Ga4L6]12- in methanol. Peaks

from fragments containing the host-guest complex in -3 and -4 charge states are shown

here.

- 171 -

Figure A1.9. Portion of the ESI- mass spectrum of [RuC8⊂Ga4L6]12- in methanol. Peaks

from fragments containing the host-guest complex in -3 and -4 charge states are shown

here.

- 172 -

Figure A1.10. Portion of the ESI- mass spectrum of [RuC10⊂Ga4L6]12- in methanol.

Peaks from fragments containing the host-guest complex in -3, -4, and -5 charge states

are shown here.

- 173 -

Full 1D NMR spectra for [RuAn⊂⊂⊂⊂Ga4L6]11- (n = 4, 6, 8)

Figure A1.11. 1H NMR spectrum of [RuA4⊂Ga4L6]11- in D2O at room temperature.

The guest’s four carbon side chain is entirely contained within the host cavity, resulting

in a T-symmetric host-guest complex.

- 174 -

Figure A1.12. 1H NMR spectrum of [RuA6⊂Ga4L6]11- in D2O at room temperature. The

guest’s six carbon side chain rapidly extends and retracts in this system.

- 175 -

Figure A1.13. 1H NMR spectrum of [RuA8⊂Ga4L6]11- in D2O at room temperature. The

guest’s eight carbon side chain is extruded through one of the host’s facial apertures,

resulting in a C3-symmetric host-guest complex.

- 176 -

2D NOESY NMR Spectra for [RuAn⊂⊂⊂⊂Ga4L6]11- (n = 4, 6, 8)

Figure A1.14. 2D NOESY 1H NMR spectrum of [RuA6⊂Ga4L6]11- at 27 °C in D2O, with

a mixing time τ = 0.4 sec.

- 177 -

Figure A1.15. Expanded region of the 2D NOESY spectrum of [RuA6⊂Ga4L6]11- at

27 °C in D2O shown in Figure S4 showing the cross peaks between host and guest

signals.

- 178 -

Figure A1.16. 2D NOESY 1H NMR spectrum of [RuA4⊂Ga4L6]11- (n = 4) at 27 °C in

D2O, with a mixing time τ = 0.4 sec.

- 179 -

2D COSY NMR Spectra: [RuAn⊂⊂⊂⊂Ga4L6]11- (n = 4, 6, 8)

Figure A1.17. 2D COSY 1H NMR spectrum of [RuA6⊂Ga4L6]11- at 27 °C in D2O

showing the aromatic host resonances for T symmetry (time-averaged). Assignments are

also included.

- 180 -

Figure A1.18. 2D COSY 1H NMR spectrum of [RuA6⊂Ga4L6]11- at -60 °C in MeOD

showing the aromatic host resonances for the C3 state. Although the peaks are broad and

overlap with each other, especially below 7 ppm, the cross peaks are consistent with eight

sets of three adjacent protons, expected for point group C3. This COSY spectrum is

qualitatively similar to that observed upon encapsulation of RuCn.

- 181 -

Electrospray Ionization Mass Spectra (ESI-MS) of [RuAn⊂⊂⊂⊂Ga4L6]11-

Pluth /R aym ond, B ryan4 in M eO H

815 820 825 830 835 840 845 850 855 860 865 870m /z0

100

%

Q T 0155 50 (0.855) Cm (1 :153) TO F M S E S - 7 .19e3833.542

833.287

833.043

832.788

824.046

823.803

823.549

823.306

823.053

822 .810

815.056814.068

822.546

815.560

832 .543

824 .796

825.050832.288

825.304

832 .044825.547

831 .789825.812

834.042

843.029

834 .286

842.783

834.542

842.526

834 .786

842.281

835.041842.024

835.286841.779

841 .533835.542

835.786

843 .274

843 .777

844.034

844 .280

844.537

852 .762844.783

852.526845.039

852.279

845.275

853.021

855.032

856.012

856.346

a)

Pluth /R aym ond, B ryan4 in M eO H

815 820 825 830 835 840 845 850 855 860 865 870m /z0

100

%

Q T 0155 50 (0.855) Cm (1 :153) TO F M S E S - 7 .19e3833.542

833.287

833.043

832.788

824.046

823.803

823.549

823.306

823.053

822 .810

815.056814.068

822.546

815.560

832 .543

824 .796

825.050832.288

825.304

832 .044825.547

831 .789825.812

834.042

843.029

834 .286

842.783

834.542

842.526

834 .786

842.281

835.041842.024

835.286841.779

841 .533835.542

835.786

843 .274

843 .777

844.034

844 .280

844.537

852 .762844.783

852.526845.039

852.279

845.275

853.021

855.032

856.012

856.346

a)

Pluth/Raymond, Bryan4 in MeOH

830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848m/z0

100

%

0

100

%

0

100

%

QT0155 (0.017) Is (1.00,1.00) C159H102N12O36Ga4Ru1K4Na2 TOF MS ES- 1.19e12833.767

833.267

833.017

832.767

832.517

832.267

832.017

834.267

834.517

834.768

835.018

835.268

QT0155 (0.017) Is (1.00,1.00) C159H101N12O36Ga4Ru1K5Na2 TOF MS ES- 1.18e12843.256

842.756

842.506

842.256

842.006

841.756

841.506

843.756

844.006

844.256

844.507

844.757845.007

QT0155 50 (0.855) Cm (1:153) TOF MS ES- 7.19e3833.542

833.287

833.043

832.788

832.543

832.288

832.044829.540830.303

834.042

843.029834.286

842.783834.542

842.526834.786

842.281

835.041842.024

835.286 841.779835.542

843.531

843.777

844.034

844.280

844.537

845.039

845.789

K4Na2H1[1 ⊂ Ga4L6]4-

(simulated)

K5Na2[1 ⊂ Ga4L6]4-

(simulated)

b)

Observed Data

Pluth/Raymond, Bryan4 in MeOH

830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848m/z0

100

%

0

100

%

0

100

%

QT0155 (0.017) Is (1.00,1.00) C159H102N12O36Ga4Ru1K4Na2 TOF MS ES- 1.19e12833.767

833.267

833.017

832.767

832.517

832.267

832.017

834.267

834.517

834.768

835.018

835.268

QT0155 (0.017) Is (1.00,1.00) C159H101N12O36Ga4Ru1K5Na2 TOF MS ES- 1.18e12843.256

842.756

842.506

842.256

842.006

841.756

841.506

843.756

844.006

844.256

844.507

844.757845.007

QT0155 50 (0.855) Cm (1:153) TOF MS ES- 7.19e3833.542

833.287

833.043

832.788

832.543

832.288

832.044829.540830.303

834.042

843.029834.286

842.783834.542

842.526834.786

842.281

835.041842.024

835.286 841.779835.542

843.531

843.777

844.034

844.280

844.537

845.039

845.789

K4Na2H1[1 ⊂ Ga4L6]4-

(simulated)

K5Na2[1 ⊂ Ga4L6]4-

(simulated)

b)

Observed Data

Figure A1.19. (a) Electrospray ionization mass spectrum (ESI-MS) for [RuA4⊂Ga4L6]11-

(n = 4) in MeOH showing peaks for the z = -4 charge state. (b) Simulated isotopic

distributions for two particular fragment ion formulae, with the observed data shown

below. Here, 1+ = RuA4+.

- 182 -

Pluth/Raym ond, Bryan6 in M eOH

828 830 832 834 836 838 840 842 844 846 848 850 852 854 856 858 860 862 864m /z0

100

%

Q T0156 41 (0.701) Cm (1:133) TO F M S ES- 1.33e4850.292

849.798

849.540

840.551

840.305

840.049

839.804

839.558

839.292

839.057831.067

830.558

830.060

838.801831.810 836.553

840.796

841.052 849.293

841.298

849.047841.554

841.800848.789

842.056848.543

842.302848.285

850.538

850.796

851.043

860.035

851.290

859.279

859.031851.548

858.783851.795

858.534852.042

852.289 858.286

860.283

860.531

860.780

861.039

861.287

861.547

861.795

862.303

862.801

a)

Pluth/Raym ond, Bryan6 in M eOH

828 830 832 834 836 838 840 842 844 846 848 850 852 854 856 858 860 862 864m /z0

100

%

Q T0156 41 (0.701) Cm (1:133) TO F M S ES- 1.33e4850.292

849.798

849.540

840.551

840.305

840.049

839.804

839.558

839.292

839.057831.067

830.558

830.060

838.801831.810 836.553

840.796

841.052 849.293

841.298

849.047841.554

841.800848.789

842.056848.543

842.302848.285

850.538

850.796

851.043

860.035

851.290

859.279

859.031851.548

858.783851.795

858.534852.042

852.289 858.286

860.283

860.531

860.780

861.039

861.287

861.547

861.795

862.303

862.801

a)

Pluth/Raymond, Bryan6 in MeOH

838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853m/z0

100

%

0

100

%

0

100

%

QT0156 (0.017) Is (1.00,1.00) C161H106N12O36Ga4Ru1K4Na2 TOF MS ES- 1.19e12840.775840.525

840.275

840.025

839.775

839.525

839.275

839.025

841.025

841.275

841.525

841.775

842.026

842.276

842.526

QT0156 (0.017) Is (1.00,1.00) C161H105N12O36Ga4Ru1K5Na2 TOF MS ES- 1.18e12850.264850.014

849.764

849.514

849.264

849.014

848.764

848.514

850.514

850.764

851.014

851.264

851.514

851.765

852.015

QT0156 41 (0.701) Cm (1:133) TOF MS ES- 1.33e4850.292850.045

849.798

849.540

840.551840.305

840.049

839.804

839.558

839.292

838.801

840.796 849.293

841.298849.047

841.554

841.800 848.789

842.056 848.543

850.538

850.796

851.043

851.290

851.548

851.795

852.042

852.547853.300

K5Na2[2 ⊂ Ga4L6]4-

(simulated)

K4Na2H1[2 ⊂ Ga4L6]4-

(simulated)

b)

Observed Data

Pluth/Raymond, Bryan6 in MeOH

838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853m/z0

100

%

0

100

%

0

100

%

QT0156 (0.017) Is (1.00,1.00) C161H106N12O36Ga4Ru1K4Na2 TOF MS ES- 1.19e12840.775840.525

840.275

840.025

839.775

839.525

839.275

839.025

841.025

841.275

841.525

841.775

842.026

842.276

842.526

QT0156 (0.017) Is (1.00,1.00) C161H105N12O36Ga4Ru1K5Na2 TOF MS ES- 1.18e12850.264850.014

849.764

849.514

849.264

849.014

848.764

848.514

850.514

850.764

851.014

851.264

851.514

851.765

852.015

QT0156 41 (0.701) Cm (1:133) TOF MS ES- 1.33e4850.292850.045

849.798

849.540

840.551840.305

840.049

839.804

839.558

839.292

838.801

840.796 849.293

841.298849.047

841.554

841.800 848.789

842.056 848.543

850.538

850.796

851.043

851.290

851.548

851.795

852.042

852.547853.300

K5Na2[2 ⊂ Ga4L6]4-

(simulated)

K4Na2H1[2 ⊂ Ga4L6]4-

(simulated)

b)

Observed Data

Figure A1.20. (a) ESI-MS for [RuA6⊂Ga4L6]11- (n = 6) in MeOH showing peaks for the

z = -4 charge state. (b) Simulated isotopic distributions for two particular fragment ion

formulae, with the observed data shown below. Here, 2+ = RuA6+.

- 183 -

Pluth /R aym ond, Bryan8 in M eO H

836 838 840 842 844 846 848 850 852 854 856 858 860 862 864 866 868 870 872m /z0

100

%

Q T0157 14 (0.239) Cm (1:153) TO F MS ES- 1.75e4857.305

856.810

856.562

847.567

847.320

847.063

846.817

846.560

846.313

846.056838.076

837.330

836.830

838.822 845.810

856.303848.060

848.317

856.055

848.564

848.811 855.808

849.057855.560

849.315855.312

857.553

857.801

858.060

867.045866.546

858.308

866.297

866.048858.556

865.799858.804

865.550859.063

865.301859.311

867.294

867.544

867.804

868.043

868.303

868.552

868.802

869.312

869.562

869.801

a)

Pluth /R aym ond, Bryan8 in M eO H

836 838 840 842 844 846 848 850 852 854 856 858 860 862 864 866 868 870 872m /z0

100

%

Q T0157 14 (0.239) Cm (1:153) TO F MS ES- 1.75e4857.305

856.810

856.562

847.567

847.320

847.063

846.817

846.560

846.313

846.056838.076

837.330

836.830

838.822 845.810

856.303848.060

848.317

856.055

848.564

848.811 855.808

849.057855.560

849.315855.312

857.553

857.801

858.060

867.045866.546

858.308

866.297

866.048858.556

865.799858.804

865.550859.063

865.301859.311

867.294

867.544

867.804

868.043

868.303

868.552

868.802

869.312

869.562

869.801

a)

Pluth/Raymond, Bryan8 in MeOH

845 846 847 848 849 850 851 852 853 854 855 856 857 858 859m/z0

100

%

0

100

%

0

100

%

QT0157 (0.017) Is (1.00,1.00) C163H110N12O36Ga4Ru1K4Na2 TOF MS ES- 1.19e12847.783847.533

847.283

847.033

846.783

846.533

846.283

846.033

848.033

848.283

848.533

848.783

849.033

849.284

849.534

QT0157 (0.017) Is (1.00,1.00) C163H109N12O36Ga4Ru1K5Na2 TOF MS ES- 1.18e12857.272857.022

856.772

856.522

856.272

856.022

855.772

855.522

857.522

857.772

858.022

858.272

858.522

858.772

859.023

QT0157 14 (0.239) Cm (1:153) TOF MS ES- 1.75e4857.305857.057

856.810

856.562

847.567847.320

846.817

846.560

846.313846.056

845.575

847.814 856.303

848.317856.055

848.564

848.811855.808

855.560849.315

857.553857.801

858.060

858.308

858.556

858.804859.063

859.560

b)

K5Na2[3 ⊂ Ga4L6]4-

(simulated)

K4Na2H1[3 ⊂ Ga4L6]4-

(simulated)

Observed Data

Pluth/Raymond, Bryan8 in MeOH

845 846 847 848 849 850 851 852 853 854 855 856 857 858 859m/z0

100

%

0

100

%

0

100

%

QT0157 (0.017) Is (1.00,1.00) C163H110N12O36Ga4Ru1K4Na2 TOF MS ES- 1.19e12847.783847.533

847.283

847.033

846.783

846.533

846.283

846.033

848.033

848.283

848.533

848.783

849.033

849.284

849.534

QT0157 (0.017) Is (1.00,1.00) C163H109N12O36Ga4Ru1K5Na2 TOF MS ES- 1.18e12857.272857.022

856.772

856.522

856.272

856.022

855.772

855.522

857.522

857.772

858.022

858.272

858.522

858.772

859.023

QT0157 14 (0.239) Cm (1:153) TOF MS ES- 1.75e4857.305857.057

856.810

856.562

847.567847.320

846.817

846.560

846.313846.056

845.575

847.814 856.303

848.317856.055

848.564

848.811855.808

855.560849.315

857.553857.801

858.060

858.308

858.556

858.804859.063

859.560

b)

K5Na2[3 ⊂ Ga4L6]4-

(simulated)

K4Na2H1[3 ⊂ Ga4L6]4-

(simulated)

Observed Data

Figure A1.21. (a) ESI-MS for [RuA8⊂Ga4L6]11- (n = 8) in MeOH showing peaks for the

z = -4 charge state. (b) Simulated isotopic distributions for two particular fragment ion

formulae, with the observed data shown below. Here, 3+ = RuA8+.

- 184 -

APPENDIX 2

Born Solvation Energies and Guest Encapsulation

According to the Born theory of solvation, a spherical ion of charge zi with a

radius ri, the Gibbs free energy of solvation relative to the gas phase is given by:1

¢Gsolv = ¡µ

Nae2

8¼²0

¶

z2i

ri

µ

1¡ 1

²

¶

= ¯

µ

z2i

ri

¶ µ

1¡ 1

²

¶

where ¯ ´ Nae2

8¼²0

¢Gsolv = ¡µ

Nae2

8¼²0

¶

z2i

ri

µ

1¡ 1

²

¶

= ¯

µ

z2i

ri

¶ µ

1¡ 1

²

¶

where ¯ ´ Nae2

8¼²0

(A2.1)

where ²² is the dielectric constant of the solvent, Na is Avogadro’s number, e is the charge

of an electron, and ²0²0 is the permittivity of vacuum. Suppose an anionic host H12-

encapsulates a monocationic guest G+ to form the host guest complex [G⊂H]11-,

abbreviated as HG11-. The solution-state binding equilibrium reaction is

H12¡(s) +G+

(s) Ð HG11¡(s) ¢G(s)

enc = ¡RT lnKbH12¡(s) +G+

(s) Ð HG11¡(s) ¢G(s)

enc = ¡RT lnKb

where subscript (s) denotes solvated species, ¢G(s)enc¢G(s)enc is the Gibbs free energy of the

binding reaction in the solution state, Kb is the binding equilibrium constant, R is the

universal gas constant, and T is the absolute temperature.

Solvation effects can be analyzed by considering a hypothetical pathway where

guest encapsulation occurs in the gas phase, denoted by subscript (g):

G+(s) ¡! G+

(g) ¡¢Gsolv(G+)

H12¡(s) ¡! H12¡

(g) ¡¢Gsolv(H12¡)

H12¡(g) +G

+(g) ¡! HG11¡

(g) ¢G(g)enc

HG11¡(g) ¡! HG11¡

(s) ¢Gsolv(HG11¡)

G+(s) ¡! G+

(g) ¡¢Gsolv(G+)

H12¡(s) ¡! H12¡

(g) ¡¢Gsolv(H12¡)

H12¡(g) +G

+(g) ¡! HG11¡

(g) ¢G(g)enc

HG11¡(g) ¡! HG11¡

(s) ¢Gsolv(HG11¡)

- 185 -

¢G(s)enc = ¢G(g)

enc +¢Gsolv(HG11¡)¡¢Gsolv(H

12¡)¡¢Gsolv(G+)¢G(s)

enc = ¢G(g)enc +¢Gsolv(HG

11¡)¡¢Gsolv(H12¡)¡¢Gsolv(G

+) (A2.2)

Assume that G+, H12-, and HG11- are spheres with effective radii rG for G+ and rH

for H12- and HG11-. Since guest encapsulation does not change the overall size of the

cluster very much, one can assume that the effective radii of H12- and HG11- are equal. Of

course, neither [Ga4L6]12- nor guests such as CoCp2

+ are actually spheres, but this

assumption is sufficient for a rough estimate of solvation effects. Using the Born

equation (Equation A2.1), the free energies of solvation are:

¢Gsolv(G+) = ¡ ¯

rG

µ

1¡ 1

²

¶

¢Gsolv(G+) = ¡ ¯

rG

µ

1¡ 1

²

¶

(A2.3)

¢Gsolv(HG11¡)¡¢Gsolv(H

12¡) = ¯

µ

1¡ 1

²

¶

(112 ¡ 122)

rH

= 23¯

rH

µ

1¡ 1

²

¶

¢Gsolv(HG11¡)¡¢Gsolv(H

12¡) = ¯

µ

1¡ 1

²

¶

(112 ¡ 122)

rH

= 23¯

rH

µ

1¡ 1

²

¶

(A2.4)

Upon substitution, the free energy of encapsulation in the solution state is

¢G(s)enc = ¢G(g)

enc + ¯

µ

1¡ 1

²

¶ µ

23

rH

+1

rG

¶

¢G(s)enc = ¢G(g)

enc + ¯

µ

1¡ 1

²

¶ µ

23

rH

+1

rG

¶

(A2.5)

Assuming ¢G(g)enc¢G(g)enc is independent of solvent properties because it corresponds to a gas

phase reaction, the above expression can be used to estimate the difference in guest

binding affinities between two different solvents.

Suppose there are two different solvents, solvent A and solvent B, with dielectric

constants ²A²A and ²B²B, respectively. If guest binding occurs in both solvent systems, and if

²A 6= ²B²A 6= ²B, the difference in the free energy of guest binding in the two solvents is

- 186 -

¢G(B)enc ¡¢G(A)

enc = ¯

µ

23

rH+

1

rG

¶ µ

1¡ 1

²B

¶

¡ ¯

µ

23

rH+

1

rG

¶ µ

1¡ 1

²A

¶

= ¯

µ

23

rH+

1

rG

¶ µ

1

²A¡ 1

²B

¶

¢G(B)enc ¡¢G(A)

enc = ¯

µ

23

rH+

1

rG

¶ µ

1¡ 1

²B

¶

¡ ¯

µ

23

rH+

1

rG

¶ µ

1¡ 1

²A

¶

= ¯

µ

23

rH+

1

rG

¶ µ

1

²A¡ 1

²B

¶

(A2.6)

Equation A2.6 allows the relative binding affinities for two different solvent

systems to be estimated from their dielectric constants and the effective radii of host and

guest. If ²B < ²A²B < ²A, then ¢GBenc ¡¢GA

enc < 0¢GBenc ¡¢GA

enc < 0, and therefore the binding constants

KBb > KA

bKBb > KA

b . In other words, the binding constant is higher in less polar solvents because

the desolvation energy cost is smaller, since the reactants are less strongly solvated. Note

that most of this energy cost is due to reducing the host charge from -12 to -11, whereas

much less energy is required to fully desolvate the monocationic guest; this observation

was discussed in a previous publication.2

We would like to compare the observed binding constant differences for CoCp2+

encapsulation by [Ga4L6]12- in water and DMF solutions with the predicted difference

from Equation A2.6. For ferrocenium (Fc+), rG ≈ 4 Å,3 and for [Et4N⊂Fe4L6]11-, the

center-to-center distance between adjacent clusters in the crystal lattice gives rH ≈ 9.5 Å.4

At 25 °C, the dielectric constant for water ²A = 78²A = 78 and for DMF ²B = 38²B = 38. The energy

difference ¢GBenc ¡¢GA

enc¢GBenc ¡¢GA

enc from Equation A2.6 is -5.6 kcal mol-1. This predicts a higher

binding constant for DMF than H2O based purely on solvation differences, in qualitative

agreement with the observed differences. However, the experimentally observed free

energy difference is -1.8 kcal mol-1, much smaller than the value from Equation A2.6.

References

1. Born, M., “Volumen und Hydratationswärme der Ionen.” Z. Physik 1920, 1, 45-48.

- 187 -

2. Parac, T. N.; Caulder, D. L.; Raymond, K. N., “Selective Encapsulation of Aqueous Cationic Guests into a Supramolecular Tetrahedral M4L6 Anionic Host.” J. Am.

Chem. Soc. 1998, 120, 8003-8004.

3. Matsumoto, M.; Swaddle, T. W., “The Decamethylferrocene(+/0) Electrode Reaction in Organic Solvents at Variable Pressure and Temperature.” Inorg. Chem. 2004, 43, 2724-2735.

4. Caulder, D. L.; Brückner, C.; Powers, R. E.; König, S.; Parac, T. N.; Leary, J. A.; Raymond, K. N., “Design, Formation, and Properties of Tetrahedral M4L4 and M4L6 Supramolecular Clusters.” J. Am. Chem. Soc. 2001, 123, 8923-8938.

- 188 -

APPENDIX 3

Crystallographic Data for K4[V2LH

3]·6.7DMF·Et2O·0.3 H2O

formula C87.1H92.2K4N13.7O27V2 formula weight 2021.22 temperature (K) 122 crystal system monoclinic space group P21/n (no. 14) a (Å) 26.129(5) b (Å) 12.842(2) c (Å) 29.694(5) β (degrees) 99.738(2) Z 4 V (Å3) 9820.15 µcalc (mm-1) 0.44 Tmin, Tmax (transmission) 0.8979, 0.9783 F(000) 4194.8 crystal size (mm) 0.25 x 0.15 x 0.05 radiation Mo Kα (λ = 0.71073 Å) h,k,l range collected -26 ≤ h ≤ 20, -12 ≤ k ≤ 10, -29 ≤ l ≤ 29 2θ range 3.54° – 41.75° scan type ω scans scan speed (°/min) 0.6 reflections collected 31142 unique reflections 10302 data: Fo

2 > 2σ(Fo2) 5711

number of parameters 925 data/parameter ratio 6.2 for Fo

2 > 2σ(Fo2)

number of restraints 44 R 0.088 with Fo

2 > 2σ(Fo2)

wR2 0.2130 with Fo2 > 2σ(Fo

2) goodness of fit 1.25

- 189 -

Vanadium Coordination Sphere Information

Twist Angle O – V – O Angle O1, O2 35.49° 80.38° O5, O6 36.48 79.66 O9, O10 32.16 80.37 Average for V1 34.71° 80.14° Std Deviation 2.26° 0.41°

Twist Angle O – V – O Angle O4 , O3 33.12° 79.89° O8, O7 30.56 79.94 O12, O11 35.95 79.65 Average for V2 31.41° 79.83° Std Deviation 1.48° 0.15°

Distance (Å) Distance (Å) V1-O1 1.9126 V1-O2 1.9596 V1-O5 1.9311 V1-O6 1.9618 V1-O9 1.9248 V1-O10 1.9370 V2-O4 1.9377 V2-O3 1.9572 V2-O8 1.9142 V2-O7 1.9562 V2-O12 1.9367 V2-O11 1.9374 Avg dout 1.926(11) Avg din 1.952(11)

V1 – V2 Distance = 12.041 Å

V1

O5

O9

O2

O6

O1

O10

V2

O4

O8

O11

O3

O12

O7