Progress in inorganic chemistry volume 57

Careful research on the[Mbpy3]3þalkaline decomposition reactions ultimately led to the realization thatthe major, if not sole, pathways for metal ion reduction involved irreversible liga

Inorganic Chemistry Volume 57

PROGRESS IN

INORGANIC CHEMISTRY

Edited byKENNETH D KARLIN

DEPARTMENT OFCHEMISTRY

JOHNSHOPKINSUNIVERSITY

BALTIMORE, MARYLAND

VOLUME 57

Published by John Wiley & Sons, Inc., Hoboken, New Jersey

Published simultaneously in Canada

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers,

MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission.

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose No warranty may be created or extended by sales representatives or written sales materials The advice and strategies contained herein may not be suitable for your situation You should consult with a professional where appropriate Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.

Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not be available in electronic formats For more information about Wiley products, visit our web site at www.wiley.com.

Library of Congress Catalog Card Number: 59-13035

Chapter 1 Mechanisms of Water Oxidation Catalyzed

by Ruthenium Coordination Complexes 1

AURORAE CLARKand JAMESK HURSTChapter 2 Biomimetic and Nonbiological Dinuclear Mxþ

Complex-Catalyzed Alcoholysis Reactions of

Phosphoryl Transfer Reactions 55

R STANBROWNChapter 3 Photoactivated DNA Cleavage and Anticancer

Activity of 3d Metal Complexes 119

AKHILR CHAKRAVARTYand MITHUNROYChapter 4 Design and Evolution of Artificial Metalloenzymes:

Biomimetic Aspects 203

MARCCREUSand THOMASR WARDChapter 5 Functionalization of Fluorinated Aromatics by

Nickel-Mediated C–H and C–F Bond Oxidative

Addition: Prospects for the Synthesis of

Fluorine-Containing Pharmaceuticals 255

SAMUELA JOHNSON, JILLIANA HATNEAN, and

MEGHANE DOSTERChapter 6 DNA Based Metal Catalysis 353

JENSOELERICHand GERARDROELFESChapter 7 Metallo-b-lactamases and Their Synthetic Mimics:

Structure, Function, and Catalytic Mechanism 395

MUTHAIAHUMAYAL, A TAMILSELVI, and

v

Chapter 8 A New Class of Nanostructured Inorganic–Organic

Hybrid Semiconductors Based on II–VI

Bu 2 Q)(btpyan)] ion [3,6-Bu 2 Q ¼ 3,6-di-tert-butyl-1,2-benzoquinone and btpyan ¼ 1,8-bis (2,20:60,200-terpyrid-40-yl)anthracene] ( See text for full caption.)

360 350 340 330

Magnetic field (mT)

80 60 40 20 0

Time (s)

(c)

400 300 200 100 0.06 0.07 0.08 0.09 0.10

an ORTEP drawing at the 50% probability level (Hydrogens and counterions are omitted for clarity.)

Chapter 4, Figure 12 Enantioselectivity of artificial-transfer hydrogenases for acetophenone tion In the achiral (planar trigonal) intermediate during catalytic turnover, incorporation of a hydride from one of the two possible prochiral faces will lead to enantiomers of the three-legged d6piano stool complex (See text for full caption.)

reduc-X HN

X

Cu 2+

HN O N N

O N N

Chapter 6, Scheme 17 Covalent approach to asymmetric DNA based catalysis.

Chapter 7, Figure 4 Mono- and binuclear structures of m bls from Bacillus cereus (BcII) of subclass B1 (a) Panel a represents the overall protein structures of mbls, A, B and C from BcII with 2.5-, 1.85-, and 1.9-

A ˚ resolution, respectively (b) panel b (D–F) represents the active sites of corresponding protein structures Water molecule and hydroxide ions are shown as red spheres, whereas Zn(II) ions are shown as gray spheres [PDB codes for structures A, B and C are 1BMC, 1BVT, and 1BC2, respectively (60–62).

H and J forms Panel a represents the overall protein structures, whereas panel b represents the active sites of these proteins [PDB codes for structures G and H are 1KO3 and 1KO2, respectively Here Ocs221 represents the cysteine sulfonic acid (66).]

Chapter 8, Figure 32 ( a) A reference UV LED (360 nm) illuminating blue light (commercially available) ( b) Image of the same LED coated with a thin layer of 2D-[Cd 2 S 2 (ba)] before illumination ( c) The illuminating image of the coated LED (d) The illuminating image of the coated LED after Mn2þdoping (0.1 mol%).

by Ruthenium Coordination Complexes

AURORA E CLARK AND JAMES K HURST

Department of Chemistry, Washington State University Pullman, WA

CONTENTS

I INTRODUCTION

II OXYGEN–OXYGEN COUPLING OF COORDINATED WATER

A The [RuII(tpy)(H 2 O)] 2 ( m-bpp) 3þ Ion

B The “Tanaka Catalyst”

III HOMOLYTIC CLEAVAGE OF O–H BONDS: THE “BLUE DIMER”

A Structure

B Redox States

C Isotopically Defined Reaction Pathways

D Theoretical Analyses

E “Noninnocent” Involvement of Bipyridine Ligands

IV NUCLEOPHILIC ADDITION OF WATER TO ELECTROPHILIC RUTHENYL OXO LIGANDS

A General Reaction Characteristics

B [Ru(bpm)(tpy)(H 2 O)]2þand Related Ions

1 Reaction Pathways

2 Alternative Theoretical Analyses

V EXPANSION OF THE COORDINATION SPHERE

VI MEDIUM EFFECTS

A Ion Pairing

B Anation

Progress in Inorganic Chemistry Volume 57, First Edition Edited by Kenneth D Karlin.

 2012 John Wiley & Sons, Inc Published 2012 by John Wiley & Sons, Inc.

1

C Influence on Catalytic Rates

VII FUTURE DIRECTIONS

A Tuning Reactivities through Modification of Organic Ligands

8 (VIIIB) ions formed ozone (O3) and hydrogen peroxide (H2O2) during theiralkaline decomposition to the corresponding M(II) ions and the subsequentrecognition by Creutz and Sutin (2) that this instability could form the basis forwater photolysis by visible light using [Ru(bpy)3]2þas photosensitizer Since directone-electron (1e) reduction of H2O to HO.is thermodynamically disallowed,considerable attention was given to characterizing the reaction dynamics with theintention of identifying reactive intermediates A brief review of this early literaturecan be found in (3) Speculations concerning the nature of these intermediatesranged from species with chemically altered bpy ligands to ion aggregates contain-ing stabilized HO.radical [e.g., HO.(HO)n], and evenm-oxo dinuclear bridgedions generated in a complex sequence of reactions initiated by HO.substitution onthe metal to form seven-coordinate intermediates This last suggestion was appar-ently inspired by contemporaneous research from Meyer and co-workers (4, 5)demonstrating that [Ru(bpy)2(H2O)]2O4þ was an effective catalyst for wateroxidation in acidic solutions containing strong oxidants Careful research on the[M(bpy)3]3þalkaline decomposition reactions ultimately led to the realization thatthe major, if not sole, pathways for metal ion reduction involved irreversible ligandoxidation accompanied by negligible formation of O2(6, 7), and interest in theseions as potential water oxidation catalysts waned A decade later, however, in apublication that did not receive much attention, Ledney and Dutta (8) reported that

[Ru(bpy)3]3þencapsulated within Y-zeolite supercages decomposed in alkalinesolution with near-stoichiometric formation of O2 Transient species suggestive

of bpy ligand modification were detected by resonance Raman (RR), cryogenicelectron paramagnetic resonance (EPR), and diffuse reflectance spectroscopy,prompting the researchers to propose a mechanism based upon HO.addition tothe ligand This general type of mechanism involving “noninnocent” participa-tion of coordinated nitrogen heterocyclic ligands had been previously exploredwithin a wider context of metal ion reactivity without any definitive supportingevidence having been found (9–11), and had also been considered by theBrookhaven group (2) as a potential mechanism for [Ru(bpy)3]3þ catalyzedwater oxidation The dramatic change of reaction course attending zeoliteencapsulation was attributed to elimination of bimolecular reactions, amongwhich were presumably the ligand degradation pathways observed in homoge-neous solution Indeed, other research indicated that when [Ru(bpy)3]3þ wasreacted with HO. at high cage occupancies, dioxygen (O2) was not formed.Rather, carbon dioxide (CO2) evolved in a manner that evoked the solutionreactions, indicating that extensive ligand degradation had occurred (12) None-theless, the study made on [Ru(bpy)3]3þat low zeolite loadings provided the firstindication that, under suitably restrictive conditions, a coordinately saturatedsingle ruthenium center is capable of catalyzing water oxidation

A second instructive point arising from the early studies was that in the presence

of certain redox metal ions [e.g., Co(II)] (6, 13, 14) and metal oxides (15–18), whichfunctioned as cocatalysts, O2formation by [M(bpy)3]3þreduction could becomenearly quantitative Indeed, these observations formed the basis for several fairlyefficient photocycles for water oxidation by electron donors using [Ru(bpy)3]2þas aphotosensitizer (Fig 1) In these cases, in addition to functioning as the true catalyst,the second metal ion most likely protected the [M(bpy)3]3þ by introducing acompetitive reduction pathway that did not involve ligand degradation

During the 1980–1990s, the perception developed in the field that efficienthomogeneous catalysis of water oxidation required the presence of at least twometal centers within the complex Factors contributing to this viewpoint included theintense focus on understanding biological water oxidation (24–26), then alreadyknown to involve a tetranuclear Mn cluster (27–30), and the repeated demonstrationsthat the ruthenium “blue dimer” (cis,cis-[Ru(bpy)2(H2O)]2O4þ) and analogousm-oxobridged diruthenium ions were efficient catalysts (31–35) but, in addition to[Ru(bpy)3]3þ, monomeric complexes containing water ligands, including speciesthat might be considered dimer fragments (e.g., cis-[Ru(bpy)2(H2O)2]3þ) wereapparently devoid of activity ((4, 31, 36); see, however, 37) Indeed, the discoverythat only two of the four Mn centers in the oxygen-evolving complex undergo redoxcycling further heightened suspicions that dinuclear centers were somehow uniquelyassociated with catalytic activity (38, 39) However, very recent discoveries have now

made it abundantly clear that this general assumption is invalid Examples of efficientcatalysis by mononuclear, dinuclear, and tetranuclear Ru complexes, as well assimilar complexes containing other metal centers, have now surfaced; moreover, thisbody of emerging work is transformative in that one no longer seeks to unlock themystery of how the O–O bond could possibly form, but rather how to distinguishamong the many demonstrated and proposed pathways that are revealed in thesereactions and to understand how structural factors dictate the expression of onepathway over another.

This chapter reviews the current state of knowledge concerning water oxidation

as revealed by reactions involving heterocyclic Ru coordination complexes Theseions possess spectroscopic signatures that make them particularly suited tomechanistic studies and often accumulate intermediary species during turnoverthat can provide important clues to reaction mechanisms Moreover, advancedcomputational analyses based upon density functional theory (DFT), as well asmulticonfigurational self-consistent field (MCSCF) and perturbation theories havebeen utilized, which are extremely helpful in evaluating the plausibility ofproposed mechanisms Although application of DFT and wave function basedmethods is now widespread within this field (40–52), it is perhaps worthwhile toemphasize that, although important as validatory tools, their full predictive powerhas not yet been realized As a recent report suggests (40), difficulties in reliably

O2+4H+WOCn

2 (bpy)]2þ, where dcb ¼ 4,4 0

-dicar-bethoxy-2,2-bipyridine (20) The strongly oxidizing sulfate radical anion [Eo(SO 4 /2 ) ¼ 2.4 V] formed upon 1ereduction of S 2 O 8  reacts with ruthenium bipyridine complexes at near-diffusion controlled rates (23) and participates in water oxidation by oxidizing both [RuL 3 ]2þand intermediary oxidation states of the WOC.

predicting mechanisms may be due to limitations in the chemical model that isstudied rather than the computational method that is employed Indeed, mosttheoretical studies do not consider the role of extended explicit solvation of thecomplex during the myriad transformations that occur along the reactivepotential energy surface, thus ignoring a key facet of the experimental reactionconditions A pertinent case in point is the study of water oxidation catalyzed

by [RuII(tpy)(H2O)]2(m-bpp)3þ (bpp¼ 2,6-bis(pyridyl)pyrazolate anion andtpy¼ 2,20:60.200 terpyridine) ion, which is discussed in detail in Section II.This complex contains two hexacoordinate Ru ions templated within a hetero-cyclic bridging bpp (Fig 2) The coordination environment enforces a geometry

in which the water ligands are facially oriented with an O  O separation distance

of only2.09 A˚ Four-electron (4e) oxidation to the corresponding [RuIV

(tpy)(O)]2(m-bpp)3 þ ion leads to O

2 evolution by a unimolecular pathway (54);

18O-isotopic labeling studies indicate that both O atoms are obtained from thecoordination sphere of the complex ion (53) These data strongly implicate amechanism involving coupling between two adjacent RuIV¼O atoms, followed

by reductive elimination of O2and regeneration of [RuII(tpy)(H2O)]2(m-bpp)3þ,

as illustrated in Fig 2 However, a DFT computational analysis made prior to thedefinitive isotope-labeling study predicted the existence of a prohibitively highactivation energy barrier for this reaction pathway (44) In this study, it was foundthat a 1,2-peroxo-bridged intermediate readily formed from [RuIV(tpy)(O)]2(m-bpp)3þ, but that decomposition of this intermediate was energetically verydemanding Thus, by this analysis, the peroxo-bridged complex was identified

as a dead-end species An alternative low-energy pathway was found thatinvolved protonation of one of the ruthenyl oxo atoms, causing electron density

to be withdrawn from the adjacent ruthenyl group This electronic polarizationrendered the ruthenyl oxygen atom sufficiently electrophilic to undergo nucle-ophilic attack by a solvent molecule with formation of a hydroperoxo–hydroxointermediate Internal electronic rearrangement then led to release of O2withregeneration of the catalyst in its original form (Fig 2) However plausible thismechanism may be, the subsequently published 18O labeling studies clearlyshow it is not operative under the reaction conditions investigated Specifically,this mechanism requires that one O atom be obtained from solvent and the otherfrom the coordination sphere of the catalyst, which is clearly not the case (53).This set of studies constitutes an example of the subtlety of forces at play that candetermine which of several potential pathways for water oxidation are expressed,

as well as the extreme challenge this presents to theorists in accurately predictingactivation barriers Correspondingly, this chapter first focuses attention uponcatalysts for which experimental evidence has given some indication of the actualreaction pathways and then enumerates other catalytic systems where experi-mental evidence on proposed reaction pathways is less definitive

II OXYGEN–OXYGEN COUPLING OF COORDINATED WATER

A The [RuII(tpy)(H2O)]2(m-bpp)3þIonThis bis-(pyridyl)pyrazolate-bridged dimer is particularly amenable to analysis

of water oxidation because each of the oxidation steps is thermodynamicallyand kinetically resolved and each of the oxidation states has a distinct opticalspectroscopic signature (45, 53) Moreover, following oxidation to the highestaccessible state ([RuIV(tpy)(O)]2(m-bpp)3þ), a transient species accumulateswhose first-order decay parallels O2release Consequently, this species could be

a bona fide reaction intermediate in the O forming cycle; its accumulation presents

(tpy)Ru II -L-Ru II (tpy)

pathway b

pathway a

N N

N N

O4 1.854A O3

Figure 2 Optimized calculated structure of [RuII(tpy) 2 (H 2 O)] 2 ( m-bpp) 3þ and alternative proposed pathways for catalyzed water oxidation For pathway a, both O atoms are derived from the coordination sphere, whereas for pathway b, one atom is from the coordination sphere and the other is from the solvent (as identified by the solid circle) [Adapted from (53).]

a unique opportunity for structural characterization that is lacking in other catalyticsystems The cyclic voltammogram (CV) of [RuII(tpy)(H2O)]2(m-bpp)3þdisplaysthree quasireversible (1e) waves in acidic aqueous solutions; a fourth irreversibleoxidation can be detected at potentials approaching catalytic water oxidation.These data indicate a regular progression in thermodynamic stabilities that followthe order: {2,2}! {2,3} ! {3,3} ! {3,4} ! {4,4} (where the notation given ismeant to indicate only the overall oxidation state of the complex based uponassignment of formal charges, i.e., not the actual electronic distribution) Oxida-tion is accompanied by release of protons, as dictated by the increasing acidities ofthe higher oxidation states so that, upon complete oxidation to {4,4}, the coor-dinated aquo ligands are completely deprotonated to give ruthenyl oxo atoms Rateconstants for stepwise oxidation by Ce4þprogressively decrease with increasingoxidation state, so that each of the intermediary oxidation states can be isolated andphysically characterized Upon oxidation to {4,4}, however, spontaneous O2evolution occurs in a reaction that is associated with first-order formationand decay of a spectroscopically distinct reaction transient The visible spectra

of both {4,4} and the transient species (I) have been obtained by global kineticanalysis

Species I is suggested to be a 1,2-m-peroxo-bridging intermediate formed bycoupling of the two juxtaposed oxo radicaloid atoms on the adjacent Ru atoms of{4,4} Due to the close energetic spacing of the various electronic states of I, thetheoretically predicted ground state is dependent on the exact density functionalused within DFT (43, 44) However, complete active space self-consistent fieldcalculations with second- order M€oller–Plesset perturbation theory (CASPT2)generally agrees quite well with the M06-L DFT implementation, predicting thateach low-spin Ru(III) couples as a triplet with its respective O., with the twotriplet RuIII–O.units coupling as a net S¼ 2 configuration; these calculations alsoindicate that the low-lying S¼ 0 state lies within 4 kJ mol1 From a computationalperspective, the reaction energetics of I are somewhat sensitive to the specificdensity functional used Yet the chemical model employed to mimic both I and itssolvation environment is significant and may be more important The direct O–Ocoupling pathway (Fig 2) is predicted by both B3LYP and M06-L functionals tohave a reasonable activation barrier for formation of the first intermediate, a cyclic1,2-peroxo bridged Ru–O–O–Ru3þ{3,3}ion However, discrepancies exist overthe appropriate treatment of the second transition state to form the {2,2}3þprotocatalyst Irrespective of whether the calculation is performed in the gasphase or utilizing a solvent continuum model to mimic the effects of the bulkdielectric, it is apparent that the activation barrier is much too high unless thechemical model is expanded to include more of the explicit solvation environmentsurrounding the Ru–O–O–Ru3þ{3,3} intermediate The approach of Yang andBaik (44) was to take into account the effects of acidity present in the experimentalsolution by examining formation of {2,2}3þfrom protonated Ru–O–O–Ru3þ{3,3}

This approach did not yield significantly improved energetics and, as such, thisreaction pathway was dismissed as a viable mechanism for I Instead, Yang andBaik (44) proposed that an alternate pathway consisting of coupling of theterminal oxo and water oxygen atoms (Fig 2) would be energetically morefavorable However, improvement in the microsolvation environment aroundRu–O–O–Ru3þ{3,3} through addition of two waters of hydration yielded acalculated activation barrier for formation of {2,2}3þ(45) that agreed within

9 kJ mol1with the experimental value

Although the experimental and theoretical results present a self-consistentand intuitively reasonable model for catalyzed water oxidation, the reactionitself presents some unexplained anomalies The rate laws for oxidation of {2,2}

to {3,3} are first order in both Ce4þ and the dimer However, the rate lawfor oxidation of {3,4} shows apparent saturation of the dependence upon Ce4þconcentration Potential causes are discussed below in Section VI on mediumeffects More strikingly, the global kinetic analyses for reactions made at ambienttemperatures indicate that, following a single turnover, the {2,2} product under-goes apparent sequential conversion to two new species that have markedlyaltered optical absorption spectra (45) These are suggested to be anated speciesthat may be similar to Ru2–bpp complexes that have been isolated containingbridging Cl, MeCOO, and CF3SO3anions in place of the coordinated watermolecules (53, 54) However, the optical changes are considerably greater thanhave been reported for m-oxo bridged Ru dimers, where SO4  substitutionoccurs (32, 33) and where ClO4and CF3SO3anation has been proposed basedupon kinetic effects (55) (Section VI.A) In those cases, the modified catalystsexhibit optical spectra that are almost indistinguishable from the correspondingcatalytically active diaquo forms Under conditions where Ce4þis in large excess,[RuII(tpy)(H2O)]2(m-bpp)3 þ is reported to catalyze water oxidation through asmany as500 cycles prior to deactivation, so it appears that either the structuralchanges implied by the optical spectra occurring after a single cycle are reversible orthe chemically modified complexes are also capable of catalyzing water oxidation

It was also reported that “exhaustive” electrochemical oxidation led to formation of

a small amount of dinuclear complex containing an oxidized bpp ligand

B The “Tanaka Catalyst”

A long-lived diruthenium catalyst for water oxidation containing a binucleatinganthracene-linked pair of terpyridyl groups with redox-active benzoquinone andhydroxide ions as additional ligands (Fig 3) was first reported in 2000 (57).Athough this complex, isolated as [Ru2(OH)2(3,6-Bu2Q)(btpyan)](SbF6)2; struc-ture given in (Fig 3), is water insoluble, Tanaka and co-workers (68) were able todemonstrate limited electrocatalytic activity by constant potential electrolysis

(CPE) in trifluoroethanol containing 10% water When the complex was deposited

as a solid on an indium–tin oxide (ITO) electrode, remarkably efficient catalyzed water oxidation could be achieved in aqueous media, with O2evolutionturnover numbers per catalyst molecule exceeding 33,000 being measured.However, the catalytic rate constant was very low Several structurally similarcomplexes containing modifications within the bridging group (xanthene foranthracene) of the templating macrocyclic ligand (59) or different substitutedquinones (46) have been prepared in efforts to improve catalytic rates within thisclass of compounds However, to date, none of these complexes have been found toexhibit detectable electrocatalytic activity

electro-Figure 3 The DFT predicted mechanism for water oxidation catalyzed by [Ru 2 (OH) 2 (3,6-Bu 2 Q) (btpyan)]2þion [3,6-Bu 2 Q¼ 3,6-di-tert-butyl-1,2-benzoquinone and btpyan ¼ 1,8-bis(2,2 0:60,200-ter-

pyrid-40-yl)anthracene] Two proton-coupled electron transfer (PCET) steps on the resting form of the catalyst (top) lead to oxidation of juxtaposed hydroxo ligands, which couple to form a bridging superoxo ion (bottom), with the additional electron being distributed over the quinone ligands Further PCET reoxidizes the quinones, leading to incorporation of solvent into the coordination sphere (left); at this point, the superoxo ligand is terminally coordinated The final PCEToxidizes the superoxide and returns the catalyst to its original form The RIMP2 calculated geometric structure of the complex ion containing 3,5-dimethyl-substituted quinone ligands (in place of tert-butyl substituents) is shown within the catalytic cycle [Adapted from (56).] (See the color version of this figure in Color Plates section.)

The aqueous insolubility and the “noninnocent” nature of the quinone ligandspresent formidable challenges to characterization of the “Tanaka complex”, as it isnow known, in its various accessible oxidation states In particular, the complex

is representative of a large class of Ru–NIL (NIL¼ noninnocent ligand) plexes whose ligand and metal orbitals are extensively mixed, giving rise toapparent noninteger oxidation states and nearly isoenergetic electronic states withdiffering spin multiplicities (56, 60), so that even ground-state configurations aredifficult to assign Despite the challenges, mechanistic analyses of this reactionhave been carried forward with considerable success by the Tanaka and Fujita/Muckerman groups using a combination of experimental and theoreticalapproaches These efforts have been aided by the availability of a model of the

com-“half-molecule”, (i.e., [Ru(H2O)(3,5-Bu2Q)(tpy)]2þ) (61) Although apparentlynot capable of oxidizing water itself (46), this ion is more amenable to compu-tational and physical analyses than the dimer Controversies concerningthe ground-state representation of this ion, prevalent in the earlier literature(46, 61), appear to have been recently resolved through in-depth electrochemical,spectroscopic, and computational analyses (47, 56)

The computational studies utilized a combination of DFT, time-dependent DFT(TD-DFT) (using the B3LYP functional) and CASSCF (complete active spaceself-consistent field) methodologies to probe the relative energies of the variousavailable spin states of the reaction intermediates Despite the relative simplicity ofthe monomer relative to the dimer, significant computational difficulty wasencountered Although the authors utilized the broken-spin broken-symmetry(BS/BS) method (62–64) to obtain open-shell singlet states, a wide variety ofhS2ivalues were observed, indicating spin contamination from alternative S states withthe same Msvalues Indeed, spin contamination was even observed for the open-shell triplet states using DFT Interestingly, the authors avoided using theNoodleman’s spin projection correction to the BS/BS singlet-state energy withintheir calculations, perhaps due to the large amount of spin contamination observed

in the open-shell singlet states To further test the relative energies of the variousspin states, the authors utilized TD-DFT to examine which spin states were higherthan the predicted ground state Unfortunately, many of the excited statesencountered were charge transfer (CT) in nature, bringing into question thereliability of the calculations, as DFT (specifically density functionals withoutlong-range corrections) is known to perform very poorly for CT excitations (65).The results for the “half-molecule” most relevant to the catalytic activity ofthe binuclear ion are that the best description of the formal oxidation state ofthe aquo complex is [RuII(H2O)(Q)(tpy)]2þ, rather than the initially proposed[RuIII(H2O)(SQ.)(tpy)]2þ (SQ.¼ 3,5-di-tert-butylbenzosemiquinone) (61),and that sequential deprotonation leads to [RuII(OH)(Q)(tpy)]þand [RuII(O.)(SQ.)(tpy)]0 The doubly deprotonated molecule is unique in possessing an oxylradical ligand, formed by internal transfer of an electron to the quinone

This radical is expected to be highly reactive and, in experimental systems, appears

to abstract a hydrogen atom (from unspecified sources) to give [RuII(OH)(SQ.)(tpy)]0as the final product The calculated electronic spin states for these threeprotonation states are difficult to assign using DFT, as spin contamination isobserved for the varying states As such, thehS2i values were interpreted in terms

of a simple generalized valence bond configuration interaction (GVB-CI) within a(2,2)CAS type model as in stretched H2 This interpretation suggests that low-lyingsinglet, open-shell singlet, and triplet spin multiplicities can exist that contain Ru

in formal oxidation states ranging from Ru(II) to Ru(IV) (47)

The water-oxidizing capacity of the dinuclear catalyst is attributed to formation

of intermediates similar to [RuII(O.)(SQ.)(tpy)]0, in which the templatingbtpyan ligand juxtaposes the coordinated oxyl groups to direct O–O bondformation via radical coupling (Fig 3) These researchers originally proposed amechanism based upon DFT computational results in which sequential deproto-nation of the resting form of the catalyst ([(RuII)2(OH)2(Q)2(btpyan)]2þ) led to anintermediate containing a bridging superoxide anion with electron density shifting

to the quinone ligands (i.e., best described as [(RuII)2(O2)(Q1.5)2(btpyan)]0),following which net 4eoxidation led to release of O2with regeneration of theresting form of the catalyst (46) More recently, this mechanism has been modified

so that the overall cycle contains a series of four PCET steps (Fig 3) (56) Here, theresting form of the catalyst is indicated as an asymmetrically hydrogen-bondedpair of coordinated hydroxo ligands The intermediate formed following the firstPCET step contains an oxyl anion that is stabilized by hydrogen-bonding to theadjacent hydroxyl ligand Loss of this proton in the second oxidation step thenallows O–O bond formation, in which the 1,2-bridging O2group is formulated assuperoxo anion with the additional electron density shifting to the quinone ligands.Subsequent PCET then leads to formation of a terminally coordinated superoxoanion via addition of solvent and, in the final step, oxidation of the coordinated

O2.releases O2, closing the catalytic cycle

One remarkable feature of this reaction as written is that the Ru ions do notchange their formal oxidation states throughout the cycle Instead, redox changesoccur primarily through complementary changes in electron density in orbitals thatare centered in the oxo and quinone ligands and reflect the highly delocalizedcharacter of the frontier orbitals in this coordination complex Nonetheless, thecomplex nature of the wave functions observed here and elsewhere, as well as thesmall energy differences between spin states, call for more thorough computa-tional studies In particular, note that few benchmarking calculations have beenperformed on Ru catalysts so as to understand more broadly the performance ofvarious density functionals and how that performance changes with varyingsystems While it is becoming more commonplace for CASSCF and CASPT2methods to be used in conjunction with DFT, this needs to become standardpractice and researchers must ensure that the size of the active space in which the

electronic excitations are allowed to occur is sufficiently large to capture theessential aspects of the wave function Moreover, in both DFT and wave functionbased methods, benchmarking of the basis sets used to describe the metal andligands must be performed To our knowledge, no studies have examined the basisset dependence of the reaction energetics and spin state distributions, nor have anyattempts been made to extrapolate any type of basis set or Kohn–Sham limit for anymethodology employed Similarly, no examples exist that have benchmarkedthe performance of varying continuum approximations and their effects upon thereaction energetics.

THE “BLUE DIMER”

A StructureThe water oxidizing capacity of the m-oxo bridged cis,cis-[RuIII(bpy)2(H2O)]2O4þ“blue dimer” (hereafter identified as {3,3}) was originally reported

by Meyer’s group (4) in 1982 For the ensuing20 years, this ion and structuralanalogues bearing substituted bipyridine ligands were the only known homoge-neous catalysts for water oxidation whose reactivity could be reproduciblydemonstrated (5, 31–35) Correspondingly, they are the ions whose physicalproperties and reactivities have been most extensively investigated X-raycrystallographic analyses of {3,3} (5) and {3,4} (as the dihydroxy-ligated[Ru(bpy)2(OH)]2O3þion) (66) reveal a nearly linear oxo-bridge and torsionaldislocation about the Ru–O–Ru bond that places the O atoms of the adjacentlycoordinated H2O or OH ligands at a distance of4.5 A˚ The DFT calculationsindicate that this same general orientation is maintained in the chemically unstable,catalytically relevant higher oxidation states of the complex (Fig 4) (67), and thenear-linear bridging character of the Ru–O–Ru bond over the entire range ofaccessible oxidation states ({3,3} to {5,5}) has been experimentally confirmed byresonance Raman (RR) measurements of the18O isotope-dependent frequencyshifts occurring in the ns(Ru–O–Ru) symmetric stretching vibrational modes (68).CASSCF methods have characterized the electronic ground state of {3,3} as

a weakly antiferromagnetically coupled singlet (43) In the computed structures,progressive oxidation of the metal centers leads primarily to modest shortening

of the metal–ligand bonds throughout the complex accompanied by an increase

in the torsional angle between the adjacently coordinated terminal oxo ligands,the net effect being that their critical O  O distances do not change appreciablyupon oxidation (67) Consequently, although compositionally similar tothe bis(pyridyl)pyrazolate-bridged diruthenium complex recently described by

Llobet and associates (53, 54), the conformational differences suggest that asignificant activation barrier to intramolecular coupling of oxo atoms may exist

in the “blue dimer” arising from the molecular distortions required to bring thesegroups into close contact Indeed, the18O isotope labeling studies described belowreveal that these two dimers catalyze water oxidation by distinct mechanisms

B Redox StatesExtensive mechanistic investigations have been undertaken by two groups, whohave generally used alternative approaches of analysis (67, 69) Although this hasled to somewhat different viewpoints, particularly concerning the nature of reactionintermediates, the groups concur that the oxygen-evolving form of the catalyst is{5,5}, an oxidation state in which the coordinated water molecules have been fullydeprotonated to generate ruthenyl oxo atoms, (i.e., [RuV(bpy)2(O)]2O4þ) Theidentity of this species was first inferred by Meyer and co-workers (5) usingelectrochemical analyses and later confirmed by redox titrations in our laboratory,which made use of a columnar flow-through carbon fiber electrode for fast CPE (70).Resonance Raman spectroscopy clearly identified Ru¼O stretching vibrationalmodes in the {5,5} ion at800 cm1(Fig 5) (70, 71); furthermore, {5,5} underwentfirst-order decay with a rate constant that was equal to the rate constant for O2evolution measured under steady-state catalytic conditions (70, 72, 73)

Under most experimental conditions, CVs of the “blue dimer” in water exhibit twowell-defined oxidation waves above {3,3} whose relative amplitudes indicate thatthey are {3,4} and {4,5}, as well as an additional wave that just precedes the onset ofFigure 4 The B3LYP/6-31G/LANL2DZ high-spin ferromagnetically coupled optimized confor- mation of the “blue dimer” in its catalytically active {5,5} ([Ru(bpy) 2 (O)] 2 O4þ) oxidation state.

solvent breakdown (5, 23) Based upon this behavior, one can assign the followingsequence of accumulating redox states: {3,3}! {3,4} ! {4,5} ! {5,5} Thesepotentials are pH dependent, reflecting the different states of protonation of thecoordinated aquo ligands under varying medium conditions Below pH 2, the twomore anodic waves coalesce, so that the voltammograms appear as two waves withrelative amplitudes of 1:3, indicating that the higher oxidation step appears as thethree-electron (3e) process: {3,3}! {3,4} ! {5,5} (5) However, an intermediatespecies can still be detected when more sensitive methods are used For example,redox spectrometric titrations utilizing the flow CPE cell described above with RRdetection clearly demonstrate the accumulation of an intermediary oxidation state atpotentials slightly lower than those required to oxidize the complex to {5,5} (Fig 5);furthermore, decay of flow CPE prepared {5,5} is biphasic, with the first stepproceeding to an intermediary species that only slowly converts to {3,4}, the higheststable oxidation state (70) The identity of this intermediate has been controversial.Based primarily upon titrimetric and transient kinetic studies using Ce4þas oxidant

Figure 5 Resonance Raman spectroelectrochemical titration of the “blue dimer” {3,4} ion in 0.5 M

CF 3 SO 3 H The inset shows the low-frequency spectra of the various detectable oxidation states Bands highlighted in light gray are the Ru–O–Ru symmetric stretching frequency and its first overtone; the band highlighted in dark gray (lowest trace) is the stretching frequency of the terminal Ru¼O bond [Adapted from (70).]

and employing global kinetic analysis for spectral deconvolution, Meyer and workers assigned this oxidation state as {4,5} (55, 74); their kinetic analyses identified{4,4}as an unstable transient species whose concentration levels were vanishinglysmall However, several different titrimetric measurements made in our laboratoryusing flow CPE prepared solutions in various oxidation states (70), as well as directtitration with Ce4þ(71) indicate that the accumulating intermediary oxidation state isactually {4,4} Recent RR and optical spectroscopic measurements have confirmedthis assignment Specifically, as anticipated from the CV analyses (5), {4,5} containsruthenyl bonds, which are readily detected in the RR spectrum by their isotope-sensitive Ru¼O stretching modes (23) These bands are not observed in theintermediate that accumulates in acidic solutions, however (Fig 5) (70) Furthermore,the optical spectrum of {4,4} determined in neutral solutions by pulse radiolysis isunlike that of {4,5}, but identical to the spectrum of the accumulating intermediate inacidic solutions (23) Assignment as {4,4} is also supported by pH jump experiments

co-in which solutions of {4,5} are rapidly acidified One observes by X-band EPRspectroscopy the immediate formation of {5,5}, but no {3,4}, the inference being thatthe other accumulating oxidation state is {4,4}, which is EPR silent Upon standing,the EPR signal of {3,4} slowly appears as the signal associated with {5,5} disappears at

a rate characteristic of water oxidation; that is, the following reaction sequence:2{4,5}! {4,4} þ {5,5} ! ! (redox decay to {3,4} and O2) (Fig 6) (23).Collectively, this body of evidence forms overwhelming support for the reactionsequence (Scheme 1), in which {4,4} is the accumulating intermediary state in acidicsolutions, but {4,5} is the accumulating state under more alkaline conditions:

4000 3600 3200 2800

Figure 6 The X-band cryogenic EPR spectra of paramagnetic “blue dimer” oxidation states in 0.5 M

CF 3 SO 3 H Panel a: spectra of {3,4} and {5,5} formed by flow CPE at the indicated potentials (vs NHE); panel b: spectral changes accompanying a pH jump of {4,5} from pH 7 to 0.3 Formation of {5,5} is immediate and its subsequent decay is accompanied by slow accumulation of {3,4}, consistent with the reaction sequence: 2{4,5} ! {4,4} þ {5,5} ! ! ! {3,4} [Adapted from (70 and 23).]

A similar pH dependent cross over of relative stabilities of {4,4} and {4,5} hasbeen reported for the analogous [(bpy)2OsIII(H2O)]2O4þ ion; as discussed byMeyer and co-workers (75), this unusual behavior reflects the influence of proticequilibria involving the aquo/hydroxo/oxo ligands upon the reduction potentials ofthe various redox states The [RuIII(tpy)2(H2O)2]O4þ ion exhibits a somewhatdifferent pattern, in which {3,3}, {3,4}, and {4,4} are stable oxidation statesthroughout the pH range, but {4,5} becomes unstable to disproportionation to{4,4}and {5,5} below pH2 (76).

In addition to {5,5}, the complex in its {4,5} oxidation state has thethermodynamic potential to oxidize water, in this case according to the reaction:2{4,5}þ 2H2O! 2{3,4} þ O2 At pH 7, the thermodynamic driving force forthis reaction (DG) is 1.0 V (5) Involvement of two dimer molecules in theoverall reaction necessitates a multistep reaction mechanism, however At pH 7,{4,5}decays by a complex rate law without generating any O2(23, 77) At pH5–6, Meyer’s group has reported O2 formation, but we have been unable toconfirm this, and detect no O2 accumulation during decay under these condi-tions (23) The rates of {4,5} decay increase with increasing solution alkalinity,which is a feature commonly shared with the group 8 (VIII B) [M(bpy)3]3þionsnoted above, as well as other highly oxidizing monomeric Ru species, such as[Ru(tpy)(bpy)O]2þ and several Os analogues, including [(bpy)2OsV(O)]2O4þ(75) These reactions have been attributed to oxidative degradation of the polyimineligands In acidic solutions, {4,4} is also thermodynamically capable of oxidizingwater in a reaction requiring two dimers In this case,DG¼ 0.4 V for thereaction: 2{4,4}þ 2H2O! 2{3,3} þ O2 The driving force for this reaction is pH independent

in the acid range because the protons released upon oxidation of water are consumed

in conversion of the dihydroxy-ligated {4,4} to the diaquo-ligated {3,3} tally, one observes that electrochemically or chemically prepared {5,5} decays byfirst-order kinetics to a redox-equilibrated solution containing primarily {4,4}, whichthen undergoes considerably slower reduction by a complex reaction mechanism tothe stable {3,4} ion (70) As noted above, the rate constant for the first step, (i.e., {5,5}

–e-O2

Scheme 1 Thermodynamically accessible redox states of the “blue dimer”.

reduction) parallels turnover rate constants for O2evolution measured under state conditions (70, 73), identifying it as the catalytically active species The capacityfor {4,4} to oxidize water has not been determined, although it is apparent that if thisreaction occurs at all, it is considerably slower than the reaction catalyzed by {5,5}.Redox equilibrated solutions at this level of oxidation necessarily contain smallamounts of {5,5} and {3,4} (70), as governed by the equilibrium: 3{4,4} $2{3,4}þ {5,5} Consequently, it might prove difficult to distinguish between O2generated by residual {5,5} and {4,4} under these conditions In any event, {4,4} is notthe O2evolving species under normally measured catalytic conditions where strongoxidants are in considerable excess.

steady-C Isotopically Defined Reaction Pathways

Oxygen isotopic labeling studies have been particularly informative in mining the reaction dynamics of the “blue dimer” Early studies from theMeyer (78) and Hurst (68) laboratories using [18O]–H

deter-2O labeled complexsuggested that several pathways may exist for O2formation These studies weremade using different oxidants (Ce4þand Co3þ, respectively) and different reactantstoichiometries (slight excess of {3,3} and Co3þ, respectively); both studies gave

O2isotopomer distributions that identified two major pathways, one in whichone O atom was derived from the coordination sphere and the other from thesolvent (pathway a), and a second in which both O atoms were derived fromsolvent (pathway b) A minor pathway comprising10% of the total reaction inwhich both O atoms were obtained from the complex coordination sphere(pathway c) was suggested from the Ce4þoxidation study, but this pathway wasnot detected in the study using Co3þas oxidant The basis for the quantitativedifferences in these two studies is uncertain One possible explanation is basedupon differing reactant compositions; in the study using Ce4þ as oxidant, thepredominant oxidation state was {4,4}, whereas in the study with Co3þit was {5,5}

If {4,4} were contributing to O2 formation in the Ce4þ experiments, a likelypathway would involve bimolecular reaction between these ions, which couldaccount for the product derived from two coordinated oxo atoms observed in thisstudy (78)

More recently, we developed methods that allow real-time mass spectrometricdetermination of evolved O2during catalytic turnover (72) This procedure has theadvantages that it provides a temporal record that can be used to test reaction kineticschemes and allows measured isotopomer ratios to be extrapolated to zero time tocorrect for isotopic dilution of the18O label in the coordination sphere as the reactionproceeds This approach has been used to probe reaction pathways for catalysis ofwater oxidation by the “blue dimer” and several congeners whose bpy ligandscontain electron-donating or -withdrawing substituents Typical results are illus-trated in panel a of Fig 7; under the experimental conditions, the oxidant (Ce4þ) is in

N N

25–100-fold excess, so that {5,5} remains the O2evolving species over the course ofthe measurements Only traces of CO2are detected, indicating that ligand decom-position is negligible The results displayed in Fig 8 confirm the absence ofpathways involving intramolecular or bimolecular reductive elimination of coor-dinated oxo atoms, as determined in the earlier work using Co3þas oxidant (68) Thetwo pathways expressed correspond to the major ones identified in the earlierwork (68, 78); their distribution depends modestly on the electronic character of thebpy ligand substituent groups, for which electron withdrawal increases the relativecontribution of the pathway involving O2formation from two water molecules.

In principle, pathway b, involving O2 formation from two solvent watermolecules, could be artifactual if isotopic scrambling occurred on the time scale

of the O2measurements For example, a mechanism is illustrated in Scheme 4 thatinvolves competitive water exchange on one of the intermediary oxidation statesleading to the appearance of a second pathway involving two water moleculesderived from solvent Direct measurement of water exchange on the stable {3,3}and {3,4} ions was made using RR spectroscopy by a technique that involvedincubation of labeled mixtures of these oxidation states, followed by oxidation to{5,5}and determination of the relative intensities of the well-separated Ru¼16Oand Ru¼18O stretching vibrations (71) These experiments indicated that, whereaswater exchange at the cis-aquo positions was relatively rapid on {3,3} (t1/2 100 s

at 23C), no exchange occurred within 24 h when the complex was oxidized to{3,4} Kinetic modeling of O2 evolution profiles for the various isotopomersprovided indirect evidence that water exchange also did not occur from interme-diary oxidation states, such as {4,4}; specifically, introducing water-exchangesteps into the model introduced severe distortion within the elution profiles for thevarious isotopomers [cf., Fig 7c], whereas a model based upon two independent

150 100

Time (s)

Time (s)

150 100 50 0 0 20 40 60 80 100

Time (s)

150 100 50 0 0 40 80 120 160

Figure 7 Kinetic traces for evolution of O 2 isotopes from 90% H 2 18O enriched [Ru

(bpy) 2 (H 2 O)] 2 O4þin 8% enriched solvent during water oxidation by Ce 4þ ion Panel a: experimental data (72); panel b: kinetic simulation based upon concurrent water addition pathways described in Schemes 2 and 3; panel c: kinetic simulation based upon water exchange (Scheme 4) Solid, dash–dotted, and dotted lines show the time course evolution of32O2,34O2, and36O2isotopomers, respectively [Adapted from (79).]

“blue dimer” under these conditions to that obtained in 50 mM phosphate, pH 7.0, using a [RuII(dcb) 2 (bpy)]2þ–S 2 O 8  photocatalytic system (Fig 1) [Adapted from (79).]

reaction pathways (a and b) running concurrently easily fits the experimentalkinetic profiles [Fig 7(b) 79] Since {3,3} is in vanishingly small concentrationsunder turnover conditions and the higher oxidation states are substitution inert,water exchange on the complex ion (e.g., Scheme 4) cannot account for the results.The alternative possibility that accumulating O2might undergo exchange withsolvent by an unrecognized catalyzed mechanism was explored by mass spec-trometry (MS) by running the reaction using components with natural isotopicabundances of oxygen in an atmosphere of36O

2 A scrambling mechanism of thistype could then be detected by the appearance of34O

2in the product gases, (i.e., thereaction):32O

2 þ36O

2$ 2 34O

2 No mixed-isotope 34O2 was observed in theproduct–gas mixture, however, indicating the absence of any scrambling mech-anism of this type as well (79)

Another possibility is that the oxidant participates directly in the forming reaction In this case, the reaction could occur through a single reactiveintermediate, but with two decay channels that lead to isotopically distinguish-able products A hypothetical example is given in Scheme 5, wherein onepathway involves dissociation of O2from the coordination sphere of a hydro-peroxo-hydroxo intermediate formed by reaction of a Ru¼O center with H2O andthe other from oxidative reaction of the hydroperoxide with Ce4þ, which carriesout O atom transfer to the hydroperoxide, forming an O2product with both

dioxygen-O atoms originating in solvent Experimental data suggest that this secondpathway is unimportant in reactions catalyzed by {5,5}, however Specifically,the isotopomer ratio is insensitive to the Ce4þ concentration over a widerange (72, 78), whereas the suggested mechanism corresponding to Scheme 5would predict a strong dependence; moreover, the same ratio is obtained withvarious oxidants, including Ce4þand Co3þions (68, 72) and a photochemicalsystem (Fig 8) that uses [Ru(bpy)3]3þ and SO4. under entirely differentmedium conditions (79) Similar mechanisms involving direct addition of water

to the hydroperoxide to form a terminally coordinated HO3intermediate or

Scheme 5 Competing pathways for decay of a common hydroperoxo/hydroxo intermediate leading

to different O 2 isotopomers Solid circles indicate solvent-derived O atoms.

collapse of the hydroperoxo-hydroxo intermediate to give a binuclear ioncontaining a 1,3-m-bridging ozonide (O2 

3 ) ligand, for example,

½RuðbpyÞ2ðOOHÞORuðOHÞðbpyÞ24þ!½ðRuðbpyÞ2Þ2ðmOÞðm1;3O3Þ2þþ2Hþ

ð1Þ

followed by H2O attack at the central O atom (80) could also give an O2moleculewhose atoms are both obtained from solvent These suggestions are plausible inthe sense that dihydrogen trioxide (H2O3) and higher polyoxo analogues of waterare known chemical entities that decompose spontaneously to give O2and otheroxidized water species (81–84) However, rough thermodynamic estimatesbased upon a recent determination of the enthalpy of formation of H2O3(85),suggest that formation of coordinated O2 

3 entities in reaction (1) or similarreactions is energetically highly unfavorable, (i.e., withDH  1.0 V) (72) Thisconclusion is reinforced by DFT calculations made by Yang and Baik (42), whoexamined various possible scenarios for water addition to form terminallycoordinated HO3or bridging O 2

3 complexes from other reaction intermediatesand concluded that these reactions would be energetically uphill by at least

36 kcal mol1 Overall, it appears that conceptually reasonable mechanismsinvolving formation of a hydroperoxo-hydroxo intermediate as a commonprecursor to two different isotopically distinguishable products are improbable

on energetic grounds

Very large H/D kinetic isotope effects are often found for oxidation of O–H andC–H bonds in peroxides, hydroquinones, alcohols, and arenes by monomericruthenyl polypyridyl compounds (e.g., [RuIV(bpy)2(py)O]2þ (py¼pyridine))suggesting that these reactions occur by hydrogen-atom abstraction mechanisms(86–89) Assuming similar processes occur in water oxidation by the diruthenyl{5,5}ion, one can write self-consistent mechanisms that rationalize the existence

of two distinct pathways (Schemes 2 and 3) Formation of free HO.by H atomabstraction from H2O is energetically prohibitive, but can be avoided byconcerted addition of the nascent HO fragment to an adjacent atom In Scheme 2,reaction at the adjacent ruthenyl oxo atom leads directly to the hydroperoxo-hydroxo intermediate, which then undergoes internal electronic rearrangement,leading to release of O2 In this pathway, one O atom is obtained from thecoordination sphere and the other from solvent Hydrogen bonding of the reactivewater molecule to the bridgingm-oxo atom is inferred from RR measurements ofRu–O–Ru stretching frequencies that, at least in their stable oxidation states,undergo small shifts to lower energies in D2O (72) Resonance Ramanspectroscopy also has been used to probe whether or not the bridging atomundergoes exchange during catalysis The Ru–O–Ru bond is slightly nonlin-ear, so that the stretching frequency is weakly dependent on the mass of the

bridging atom, allowing 18O to be distinguished from 16O No loss of theisotopic label was measured in the RR spectrum following as many as 10turnovers of the catalyst, indicating that the bridging O atom is not a netparticipant in the dioxygen-forming reactions (72).

D Theoretical AnalysesThis reaction has been treated theoretically using DFT and CASSCF meth-ods (41–43) The only low-energy pathway for water oxidation by {5,5} found inthese studies was the one described by Scheme 2, (i.e., homolytic cleavage of awater O–H bond with concerted addition of the fragments across the two ruthenyl

O atoms) These studies provide important mechanistic insight into how thesereactions might occur The electronic structure of the {5,5} core is best described

as an antiferromagnetically coupled (RuIV–O.)2O4þ ion, whose diradicaloidnature is well suited for promoting the water homolysis reaction Interestingly,B3LYP does not predict this electronic state to be the lowest in energy, insteadpreferring the ferromagnetically coupled state Subsequent CASSCF studies haverevealed the correct open-shell antiferromagnetic ground state; nonetheless,the discrepancy between the B3LYP and CASSCF calculations is limited only

to the relative energetics of the two states, as both share very similar descriptions

of the electronic structure [molecular orbitals (MOs), charges, etc.] Within theB3LYP study, rate-limiting addition of water ensues, with a calculated activationbarrier of 25.9 kcal mol1, which is very similar to experimentally determinedactivation free energies [DG‡¼ 20.3 and 20.9 kcal mol1at 23C for O

2evolutionand {5,5} decay, respectively (72)] The {4,4}-hydroperoxo-hydroxo intermediateformed undergoes rapid proton-coupled internal electron transfer to give an aquo-superoxo bound {3,4} complex that, upon undergoing spin exchange and anadditional electron transfer step, releases3SO2, regenerating {3,3} This analysiscorrectly identifies {5,5} decomposition as the rate-determining step Additionally,the possibility that water addition might produce bridging or terminally coordi-nated ozonides as reaction intermediates, as discussed above, was explicitlyconsidered and rejected as a potential mechanism on the basis that formation ofbound trioxides from the {3,4}-aquo–superoxo intermediate was energeticallyhighly unfavorable, (i.e.,DG  36 kcal mol1) (42) Like most studies performed

on this system, only a single density functional (B3LYP) and modest basis set havebeen used, and it is unclear how different density functionals or a more thoroughinvestigation of basis set dependence would alter the proposed energetics Withinthe field, much use is made of Mulliken spin population analysis to determine theextent of radical character on a given center However, care must be taken with thismethodology, as recent work has highlighted the sensitivity of spin populations tovarying basis sets, which may have a significant effect upon the interpretation ofmechanisms (90)

E “Noninnocent” Involvement of Bipyridine Ligands

The other proposed pathway (Scheme 3) is based upon the propensity of HO.toadd to aromatic and heterocyclic rings (2, 15, 91–94), and is suggested to arise duringhydrogen atom abstraction from water molecules that are not oriented to couple thedeveloping HO fragment to the adjacent ruthenyl group (72) As envisioned, theligand radical intermediate formed then adds a second water molecule to generate a{4,4}intermediate containing a diol-derivatized bpy This intermediate then under-goes internal 2e transfer with proton rearrangement to form a {3,3} complexcontaining a dioxetane, which spontaneously releases O2, regenerating the catalyst inits original form As written, one key feature is that no change occurs in theelectrostatic charge of the complex throughout the cycle, so that there should be

no large solvent reorganization contributions to activation barriers at any of the steps.The first step in this pathway is similar to previously proposed steps in thealkaline decomposition reactions of [M(bpy)3]3þions (2, 3, 6) Optical spectro-scopic features consistent with the proposed HO.addition to the bpy ligands havebeen observed during catalytic turnover of the “blue dimer” under certain con-ditions Specifically, a diagnostic feature of bpy–OH adduct formation is theappearance of broad, relatively weak absorption bands in the near-infrared (NIR)region of the optical spectrum (2, 15, 93, 94) In photocatalyzed reactions of thetype described in Fig 1, deconvolution of the steady-state absorption envelope intoits component dyes requires inclusion of a broad near-IR band centered at780 nm(Fig 9) that can be attributed to similar ligand modifications within the dimer Thespecies giving rise to this band passes kinetic competency tests as a bona fidereaction intermediate, (i.e., the NIR band appears and disappears during repetitivelight–dark cycling of the photochemical system with a decay rate constant that isidentical to that for O2evolution) (20) This demonstration is important because

HO.addition is also a plausible first step in irreversible oxidation of the bpy ring (6),although in this case decay of the intermediate would not be linked to O2formation,and therefore would not necessarily occur at the same rate Finally, similar NIRbands are also observed as transient species in reactions between {3,3} andradiolytically generated HO., that is, the reaction: {3,3}þ HO. ! {3,3}-OH.

! {3,4} þ OH(95), confirming that the predominant reaction of HO.with thedimer is ring addition, rather than hydrogen-atom abstraction or electron transfer.Solutions of {5,5} exhibit an anomalous X-band EPR signal that is unexpectedsince the ions should either be diamagnetic or contain an even number of unpairedspins, hence be EPR silent This signal is unusual in exhibiting a relatively narrow andnearly axial band shape, whereas the paramagnetic states of Ru m-oxo dimersgenerally give broad rhombic bands (e.g., Fig 6); other characteristics include asix-line hyperfine splitting evident on the g?component that comprises20% of thesignal, consistent with IN¼ 5/2 nuclearhyperfine coupling to99Ru and101Ru, whichcollectively are present in30% natural abundance, and a power saturation profile

that indicates rapid spin relaxation The integrated intensity of this signal is<10% ofthe total amount of complex present, based upon comparison to the intensity of the{3,4}signal Moreover, in photochemical experiments this signal also decays at thesame rate as O2evolution, as determined by varying the time interval of incubation inthe dark prior to cryogenic quenching (Fig 9) (20) Based upon these properties, wehad earlier suggested that this signal might be associated with a ligand-centeredradical, such as the first intermediate in Scheme 3 (3, 67); EPR spectra with similarband shapes have been reported for dimeric Ru complexes containing coordinatedphenoxyl radicals, [e.g., (96)] However, ongoing one-dimensional electron spinecho envelope modulation (1D-ESEEM) and electron-nuclear double resonance(ENDOR) experiments (97) have detected only weak proton coupling to this signal,and the relatively large axial Ru nuclear hyperfine coupling constants (130 MHz)that are observed suggest that the unpaired spin density resides primarily on a single

Ru center (96, 98) Consequently, any assignment given to the near-axial signalobserved under catalytic conditions must be considered provisional pending theoutcome of additional structural studies

Chemical precedents may exist for the addition of a second water molecule

to the bpy ring to form a diol, as shown in the second reaction step of Scheme 3

360 350 340 330

Magnetic field (mT)

80 60 40 20 0 Time (s)

(c)

400 300 200 100 0.06 0.07 0.08 0.09 0.10

Both 2,20-bpyOH. and Ru(bpy)2(bpyOH.) adducts have been shown to bemoderately strong reducing agents (94) Their formation in the presence of[Fe(CN)6]3led to net reduction of ferricyanide with net loss of OHfrom themedium, as measured by conductivity changes following pulse radiolytic gener-ation of HO. The reaction sequence proposed to account for these reactions,illustrated for bpy, is [Fe(CN)6]3þ bpyOH. ! [Fe(CN)6]4þ bpyOHþ;bpyOHþþ OH! bpy(OH)2 A similar set of reactions appears to occur when{3,4}is reacted with radiolytically generated HO. Unlike {3,3}, rapidly decayingtransients are not detected in this reaction; however, rather than being oxidized,{3,4}undergoes net reduction to {3,3} (Fig 10) This behavior is unique to HO.;reaction of {3,4} with other radiolytically generated 1eoxidants (SO4., CO3.gives quantitative oxidation to {4,4} (23) Following addition of HO.to {3,4}, thesecond Ru center may function analogously to [Fe(CN)6]3in the presence ofRu(bpy)2(bpyOH.) by acting as an electron sink for the unpaired electron on theligand radical, allowing net oxidation and addition of a hydroxide ion (Scheme 3).The products of this reaction have not yet been identified, however; the diol, ifformed, may be unstable (94), and alternative reactions not involving diolformation are possible (e.g., Scheme 6) Accumulation of H2O2did not occurduring limited g-irradiation of {3,4} under conditions that generate HO., implyingthat the reactions involve net oxidation of the bipyridine ligand rather than water.Unlike the first reaction step, addition of water in the second step of theproposed mechanism need not be concerted, and is more likely to involve pseudo-base addition of OHto any of several positions on the ring accompanied byprotonation of the terminal oxo ligand of the Ru undergoing 1ereduction Theputative product is an analogue of the {4,4} intermediate that accumulates during

40 30 20 10 0 –40

{5,5}decay This species is unique, however, in possessing an unstable derivatized bpy coordinated to a dinuclear 2e sink that is almost as stronglyoxidizing as the original {5,5} ion [at pH 1, {5,5}þ 2Hþþ 2e! {4,4}, Eo0

dihydroxy-¼ 1.58 V (NHE); {4,4} þ 2Hþþ 2e! {3,3}, Eo0¼ 1.28 V (NHE)] Furtheroxidation of the modified bpy might then occur by pathways that allow formation

of O2via internal redox cycling within the ring, such as suggested in Scheme 3

We are investigating by DFT the thermodynamic aspects of water addition to bpywithin model mer- and fac-[RuIV/V(NH3)3(bpy)O]3þ/2þand [RuIII/IV(NH3)3(bpy)(OH)]3þ/2þ complexes; reactions in both gas and solution phases are beingconsidered In these reactions, the products contain ligand-centered cations whoseelectron and spin densities are reflected not only in the relative spin-state distribu-tions, but also in calculated EPR spectra and its relevant terms [zero-field splitting(ZFS), g-tensor, and nuclear quadrupole coupling elements] Our preliminaryresults indicate that the thermodynamics of these reactions are tuned by a variety

of factors, including the metal oxidation state, site of addition to the ring, and spinstate of the ligand radical products Water addition to both fac- and mer-isomers of[RuV(NH3)3(bpy)O]3þis energetically favorable, with the fac-isomeric productsgenerally being more stable Most of the stable [RuIV(NH3)3(OH)(bpyHO.)]3þproducts are in their closed shell S¼1

2states, with the most exergonic reactionsoccurring at the meta position on the bpy ring A notable exception is the fac-[RuIV(NH3)3(OH)(p-bpyOH.)]3þ, whose formation in the S¼3

2ground state isthe most exothermic of all of the reactions Water addition to the bpy ligand of mer-[RuIV(NH3)3(bpy)(O)]2þ is highly unfavorable in all spin states, whereas thecorresponding reaction of mer-[RuIV(NH3)3(bpy)(OH)]3þto generate S¼ 0 pro-ducts is slightly favored at each of the ortho-, meta-, and para-positions The EPRsimulations indicate that the ligand radical signatures of the various paramagneticproducts of [RuV(NH3)3(bpy)O]3þreactions are unique; comparison to authenticEPR signals from the “blue dimer” in its oxidized {5,5} state should aid in itsScheme 6 Alternative bpy ligand oxidation products in reactions of HO.with the “blue dimer” {3,4} ion.

structural characterization In any event, the facile addition of water to the modelcomplex in its Ru(V) state supports the plausibility of similar “covalent hydration”reactions of the {5,5} ions of the “blue dimer” and related congeners.

RUTHENYL OXO LIGANDS

A General Reaction Characteristics

In 2005, during the course of investigating the catalytic capabilities of a set ofdiruthenium complexes contained within a nitrogen macroheterocyclic ligand (99),Thummel and Zong (100) found that monomeric complexes that represented half-dimer analogues were themselves effective catalysts for water oxidation by Ce4þ

in acidic media Subsequent research from several laboratories has shown thatwater oxidation can be catalyzed by a wide range of monomeric aquapolyimineand structurally related Ru complexes Several of these are shown in Fig 11;additional compounds have been reported (37, 49, 101, 103, 105, 106), includingexamples containing carbene derivatives and 2,6-bis((dimethylamino)methyl)pyri-dine and acetylacetone ligands These complexes typically exhibit CV that displayquasireversible proton-coupled electron transfer in two 1esteps or one 2estep,indicating oxidation to the L5RuIV¼O state, with an irreversible additionaloxidation step appearing immediately before the onset of a catalytic wave foroxidation of solvent In titrimetric studies with Ce4þ, oxidation of water was notobserved until 3 equiv of the oxidant were added, indicating that L5RuV¼O is thecatalytically active species In their studies, Meyer and co-workers (101) noted aninverse relationship between the half-time of the catalyzed reactions and thecatalyst Ru(V/III) half-cell potentials, presumably reflecting the importance of theelectrophilic nature of the ruthenyl oxo atom upon the reaction dynamics Instudying the influence of electron-donating and -withdrawing substituent groups

on the polypyridyl ligands of [RuII)(tpy)(bpy)(H2O)]2þ(compounds D, Fig 11),Berlinguette and co-workers (103) found an inverse correlation between theturnover number (i.e., the maximal amount of O2 formed) and the apparentcatalytic rate constant, as measured by fitting O2evolution curves to a first-orderdecay profile Under the experimental conditions, where Ce4þwas in very large(5000-fold) excess, decomposition of the ligands was extensive Since thedecline in O2 evolution under these conditions must be caused primarily byinactivation of the catalyst, it is not surprising that the measured rate constantscorrelate inversely with the amount of O2 produced, (i.e., the more rapidlydegraded catalysts produce less O2) A more useful comparison would be that ofthe influence of structural modifications upon initial rates of O2formation underconditions where decomposition was minimized

B [Ru(bpm)(tpy)(H2O)]2þand Related Ions

1 Reaction PathwaysMeyer and co-workers (48) have recently concluded an elegant and thoroughkinetic study of some of the more reactive monomeric catalysts in acidic solutions.Their studies have identified at least two reaction pathways, one in which net 4eoxidation by Ce4þ generates a spectroscopically identifiable transient specieswhose decay occurs spontaneously with O2evolution in a first-order process andanother in which this intermediate undergoes rate-limiting 1eoxidation by anadditional Ce4þion, which is accompanied by O evolution The latter pathway is

N Me

+

R R

N N

N Ru N N Me

N

Me 2+

N N Ru

H2O N N N

Me

Me 2+

N N Ru

H2O N N N

Me

Me 2+

G F

Figure 11 Representative monomeric Ru water oxidation catalysts Compound A, [Ru(bpm)(tpy) (H 2 O)]2þ(48); compound B, [Ru(Mebimpy)(bpy)(H 2 O)]2þ(101); compound C, [Ru(Mebimpy)(4,40- ((HO) 2 OPCH 2 ) 2 bpy)(H 2 O)]2þ(102); compounds D, [Ru(40-R 1 -tpy)(4,40-(R 2 ) 2 -bpy)(H 2 O)]nþions (R 1 ,

R 2 ¼ H, OMe, Cl, COOH) (103); compounds E, [Ru(tpy)(4,4-R 2 bpy)Cl]þ, R ¼ H, Me, OMe, NO 2 , C(O) OEt (Et ¼ CH 3 CH 2 -)(49); compound F, [Ru(dpp)(pic) 2 ]2þ(49); compound G, [Ru(4,40-CO 2 -bpy) (pic) 2 ] (104) Ligands: bpm ¼ 2,2 0-bipyrimidine; 4,40-CO

2 -bpy ¼ 2,2 0-bipyridine-4,40-dicarboxylate;

4,40-((HO) 2 OPCH 2 ) 2 bpy ¼ 2,2 0-bipyridine-4,40-bis-phosphonic acid; dpp¼ 2,9-dipyridyl-2 0

-yl-1,10-phenanthroline, Mebimpy ¼ 2,6-bis-(1-methylbenzimidazole-2-yl)pyridine; pic ¼ 4-picoline.

dominant in strongly acidic solution (1.0 M HNO3), but is not observed in 0.1 MHNO3 This behavior is thought to relate to the differing oxidation potentials of

Ce4þin the two media (69), with thermodynamic constraints limiting oxidation ofthe first intermediate by Ce4þto the more acidic environment Addition of 3 equiv

of Ce4þto [RuII(bpm)(tpy)(H2O)]2þ(compound A, Fig 11) generated a complexkinetic waveform, ultimately resulting in accumulation of a spectroscopically andelectrochemically distinct species formed by first-order conversion from[RuV(bpm)(tpy)(O)]3þ This species was shown by redox titration with Fe2þto

be a 3eoxidant The proposed structures of the intermediates include a terminallycoordinated peroxo ligand ([RuIII(bpm)(tpy)(OOH)]2þ), formed by nucleophilicaddition of H2O to the ruthenyl oxo atom of [RuV(bpm)(tpy)(O)]3þ, and 1e([RuIV(bpm)(tpy)(OO)]2þ) and 2e ([RuV(bpm)(tpy)(OO)]3þ) oxidized analo-gues Their relationships in the proposed catalytic cycle are shown in Scheme 7 Inthis cycle, [RuIV(bpm)(tpy)(OO)]2þpresents a branch point for the reaction, eitherundergoing spontaneous conversion with redox cycling between the Ru(IV)-peroxo and Ru(II) states or oxidation to (formally) Ru(V) with cycling betweenthe Ru(V)-peroxo and Ru(III) states Consistent with this mechanism, RR spectraexhibiting a unique band at 1015 cm1has been reported for the [RuIV(Mebimpy)(bpy)(OO)]3þ analogues of [RuIV(bpm)(tpy)(OO)]2þ (compound B, Fig 11),which is consistent with DFT calculations for the O–O stretching frequency of

an h2-bound peroxo ligand; confirmation of this assignment by analysis of theoxygen-isotope dependence of the frequency is desirable Oxygen isotope-label-

[RuII(bpm)(tpy)(H2O)]2+