Electrochemical investigation of Mn4O4-cubane water-oxidizing clusters

Electrochemical investigation of Mn4O4-cubane water-oxidizing clustersw

Robin Brimblecombe,ab

Alan M. Bond,*aG. Charles Dismukes,*

b

Gerhard F. Swiegerscand Leone Spiccia*

a

Received 22nd January 2009, Accepted 16th April 2009

First published as an Advance Article on the web 26th May 2009

DOI: 10.1039/b901419e

High valence states in manganese clusters are a key feature of the function of one of the most

important catalysts found in nature, the water-oxidizing complex of photosystem II. We describe

a detailed electrochemical investigation of two bio-inspired manganese–oxo complexes, [Mn4O4L6]

(L = diphenylphosphinate (1) and bis(p-methoxyphenyl)phosphinate (2)), in solution, attached to

an electrode surface and suspended within a Nafion film. These complexes contain a cubic

[Mn4O4]6+ core stabilized by phosphinate ligands. They have previously been shown to be active

and durable photocatalysts for the oxidation of water to dioxygen. A comparison of catalytic

photocurrent generated by films deposited by two methods of electrode immobilization reveals

that doping of the catalyst in Nafion results in higher photocurrent than was observed for a solid

layer of cubane on an electrode surface. In dichloromethane solution, and under conditions of

cyclic voltammetry, the one-electron oxidation processes 1/1+ and 2/2+ were found to be

reversible and quasi-reversible, respectively. Some decomposition of 1+ and 2+ was detected on

the longer timescale of bulk electrolysis. Both compounds also undergo a two-electron, chemically

irreversible reduction in dichloromethane, with a mechanism that is dependent on scan rate and

influenced by the presence of a proton donor. When immersed in aqueous electrolyte, the reduction

process exhibits a limited level of chemical reversibility. These data provide insights into the

catalytic operation of these molecules during photo-assisted electrolysis of water and highlight

the importance of the strongly electron-donating ligand environment about the manganese ions in

the ability of the cubanes to photocatalyze water oxidation at low overpotentials.

Introduction

The catalytic site of the O2-evolving water-oxidizing complex

(WOC) in photosystem II (PSII) features an inorganic cluster

composed of an Mn4Ca1 core. This core is conserved in

all known oxygen photosynthetic organisms.1 A range of

manganese complexes inspired by the PSII-WOC have been

reported. A small number of these are capable of catalytically

oxidizing water, although usually only at slow rates and with

limited turnovers in the presence of sacrificial chemical

oxidants like CeIV (1.72 V vs. NHE).2,3 Chemical oxidants

are commonly used in lieu of more complex, light-driven

photo-oxidants in such homogeneous ‘‘artificial photo-

synthesis’’ systems. Such chemical oxidants have the dis-

advantage that they complicate both the system and the

interpretation of the results. Two approaches are available

to eliminate the need for sacrificial chemical oxidants. These

involve the attachment of the catalysts to conductive surfaces

for direct electrochemical oxidation or to photoanodes for

light-induced oxidation, as in a solar cell. In this work we

consider only the former approach.

In a recent study, we described the sustained, catalytic, light-

driven electro-oxidation of water by a bio-inspired Mn

complex containing a [Mn4O4]6+/7+ cubane core,4 supported

by six facially capping diarylphosphinate ligands, L, giving a

[Mn4O4L6] molecule (Fig. 1).5 The formal oxidation states of

the manganese ions in this cluster are [MnIII2MnIV2]. Two

ligands have been employed: L = diphenylphosphinate,

[Mn4O4(Ph2PO2)6] (1), and bis(p-methoxyphenyl)phosphinate,

[Mn4O4((CH3OPh)2PO2)6] (2).5,6 Both complexes have been

investigated in detail from both a structural and a chemical

reactivity perspective.5–12 The oxidized cubanes,

[MnIIIMn3IVO4L6]

+(ClO4�), 1+ClO4

� and 2+ClO4

�, hereafter

denoted the cubium oxidation state, have been previously

isolated and characterized.8,11

When trapped in a thin layer of Nafion cast onto an

electrode surface, 2+ is an active and durable catalyst capable

of photo-oxidizing water at an applied potential of 1.00 V vs.

Ag/AgCl.4 Chemical reactivity studies have revealed that the

[Mn4O4L6]0/1+ cubane complexes are exceptionally strong

oxidizing agents, with the core oxygen atoms capable of

oxygenation and H-atom abstraction reactions.7,8 The neutral

cubane is capable of H-atom abstraction by cleaving RO–H

a School of Chemistry, Monash University, Victoria 3800, Australia.E-mail: [email protected],[email protected]

bDepartment of Chemistry & Princeton Environmental Institute,Princeton University, NJ 08544, USA.E-mail: [email protected]

c Intelligent Polymer Research Institute, University of Wollongong,Wollongong, NSW 2522, Australiaw Electronic supplementary information (ESI) available: Tables S1, S2and S3 provide summaries of cyclic voltammetry data for solutionoxidation and reduction processes. Cyclic voltammograms obtained ata glassy carbon electrode for the oxidation and reduction of 1 and 2 indichloromethane (Fig. S1) and glassy carbon rotating disc voltammo-gram for the reduction of 1 in dichloromethane (Fig. S2). SeeDOI: 10.1039/b901419e

This journal is �c the Owner Societies 2009 Phys. Chem. Chem. Phys., 2009, 11, 6441–6449 | 6441

PAPER www.rsc.org/pccp | Physical Chemistry Chemical Physics

and RN–H bonds that are greater than 390 kJ mol�1 in

strength while the oxidized cubane (1+) can abstract hydride

anions, cleaving bonds of 530 kJ mol�1 (top line in Fig. 1).13

One striking comparison is that the O–H bond in the reduced

complex with the [Mn4O3(OH)]6+ core requires in excess of

390 kJ mol�1 to dissociate H and regenerate the [Mn4O4]6+

core vs. only 320 kJ mol�1 for the O–H bond in analogous

dimanganese complexes containing a rhombohedral core,

[Mn2O(OH)]3+ - [Mn2O2]3+.7 Thus, the oxo bridges of the

expanded [Mn4O4]6+ cubane cores form stronger O–H bonds

than those in the [Mn2O2]3+ rhombohedral core.13

The redox potentials of manganese complexes are strongly

influenced by the donor atoms and in particular the electron-

donating/withdrawing potential of the ligands.14 The presence

of a high charge of 7+ in the [Mn4O4] core elevates

the potential relative to monomeric metal ions. The higher

oxidation states of the cubane core can be stabilized by the

addition of electron-donating methoxy groups at the para-

position of the phosphate phenyl rings.6,11

Previous attempts to covalently attach the cubane to

conductive surfaces and to thereby form monolayers on gold

surfaces were successful using the iodine atom of p-iodo-

phenylphosphinate to bridge between the cubane core and a

gold surface.15 These monolayers were shown to be electro-

chemically active. However, the bridge was unstable and

shown to oxidatively cleave upon reaching the I�/IO3�

potential. Attempts to functionalize the phenyl rings at the

para-position with anchoring groups, such as carboxylic acid

and hydroxyl groups, have proven unsuccessful due to the

interference of these groups in cube formation.

In order to better understand the basis for the strong oxidative

capacity of the cubium state and to characterize the lower redox

states of these complexes, we have examined in detail, and describe

herein, their solution phase electrochemistry. We also compare

catalytic activity in aqueous electrolyte environments, when

the complexes are immobilized on electrode surfaces either by

deposition as a solid film or when ion-exchanged within thinNafion

membranes cast on the electrode surface.

Experimental section

Materials and methods

Diphenylphosphinic acid and bis(methoxyphenyl)phosphinic

acid were purchased from Lancaster and Aldrich, respectively,

and used without further purification. Tetrabutylammonium

hexafluorophosphate (Bu4NPF6) was obtained from GFS

Chemicals and was used as the electrolyte in organic solvents.

It was purified as follows prior to use: a sample of Bu4NPF6

was dissolved in acetone and KPF6 was added to precipitate

iodide impurities (as potassium iodide). The solution was

then filtered, evaporated to dryness and recrystallized from

ethanol. The resulting crystalline solid was dissolved in

dichloromethane. An insoluble white powder was removed

by filtration and the solvent evaporated to dryness to produce

electrochemically pure Bu4NPF6. Nafions was purchased

from Dupont as a Nafion-PFSA polymer dispersion DE

1020, which is an aqueous dispersion of 10–12% Nafion. All

other reagents were purchased from BDH or Aldrich and

used as received. [Mn4O4L16], compound 1 (L1 = diphenyl-

phosphinate),5 [Mn4O4L26], compound 2 (L2 = bis(p-methoxy-

phenyl)phosphinate),6 and 2+ClO4� were prepared11

as described in the literature. The pH of the aqueous solutions

and buffers used in the electrochemical studies were adjusted

by addition of either 0.1 M H2SO4 or 0.1 M NaOH to

aqueous 0.1 M Na2SO4 supporting electrolyte. A 10 mM

MOPS buffer was prepared by dissolving 3-(N-morpholino)-

propanesulfonic acid in water and adjusting the pH to 7 using

NaOH. The ionic strength of this buffer solution was adjusted

to 0.1 M using Na2SO4. pH values were measured with a

Metrohm 6.0234.100 pH electrode and 744 pH meter.

Electrochemistry

Electrochemical experiments were conducted at 22 (�2) 1C

with BAS (Bio Analytical Systems) Epsilon CS3 or BAS 100B

workstations. Cyclic voltammograms were obtained at scan

rates of 5 to 500 mV s�1 in a conventional three electrode

electrochemical cell containing inlet and outlet ports for

degassing solutions with nitrogen. Experiments in CH2Cl2 or

CH3CN (0.1 M Bu4NPF6) used an Ag/Ag+ (0.01 M AgNO3

in CH3CN) reference electrode, with a double glass frit

separating the electrode from the test solution. The ferrocene/

ferrocenium (Fc/Fc+) oxidation process was used to provide

an internal reference potential calibration system for

electrochemical studies in organic solvents, with potentials

adjusted to the Fc/Fc+ scale (0.400 V vs. NHE at 25 1C).

Aqueous experiments were conducted in distilled water

containing 0.1 M Na2SO4 as the electrolyte. In this case,

potentials were referenced against a BAS Ag/AgCl (3 MNaCl)

glass bodied reference electrode separated from the test solu-

tion by a Vycor frit, with a potential of 0.200 V vs. NHE at

25 1C (BAS reference electrode users manual). Potentials

(measured against Ag/AgCl (3 M NaCl)) are presented vs.

Fig. 1 Schematic representation of [Mn4O4L6], 1: L = diphenly-

phosphinate (PO2(Ph)2), and 2: L = bis(p-methoxyphenyl)phosphinate

(PO2(PhOMe)2), and previously observed oxidation states and

proposed reduced intermediates.

6442 | Phys. Chem. Chem. Phys., 2009, 11, 6441–6449 This journal is �c the Owner Societies 2009

Fc/Fc+ to allow comparison to organic solution measurements,

converted by adjusting the potential by�0.200 V. All voltammetric

experiments used Pt auxiliary electrodes. The working

electrodes used were 3 mm diameter glassy carbon and Pt disc

electrodes. The glassy carbon working electrode had an

electroactive area of 5.9 mm2 as determined by oxidation of

1.0 mM ferrocene in CH3CN (0.5 M Bu4NPF6) and use of the

Randles–Sevcik equation2 with a diffusion coefficient (D) of

1.7 � 10�5 cm�2 s�1 (CH3CN, 0.5 M Bu4NBF4).2

Bulk electrolysis experiments in CH2Cl2 (0.1 M Bu4NPF6)

were conducted under a nitrogen atmosphere with the same

Ag/Ag+ (0.01 M AgNO3 in CH3CN) reference electrode as

used in voltammetry, a large area glassy carbon working

electrode, and a platinum basket counter electrode. The bulk

electrolysis cell consisted of two compartments. The outer one

contained the counter electrode and a stirring rod immersed in

CH2Cl2 (0.1 M Bu4NPF6). This compartment was separated

by a glass frit from the inner compartment which contained 1

or 2 dissolved in CH2Cl2 (0.1 M Bu4NPF6) as well as the glassy

carbon working and the Ag/Ag+ reference electrode. Bulk

electrolysis experiments were stopped when the current

decayed to 1% of the initial value.

Solid film preparation

Electrodes were coated with a thin film of cubane by casting a

small volume of a CH2Cl2 solution of the complex onto the

electrode surface and allowing it to dry. Alternatively, films

were prepared by rubbing the face of the electrode over a small

amount of dry sample.

Nafion film deposition and Nafion doping

Nafion-modified electrodes were prepared by drop-casting

0.5 mL of 10% Nafion in ethanol onto a 3 mm diameter

electrode and allowing it to air dry before heating in a

laboratory oven at 120 1C for 20 min. Doping of the cast

Nafion membrane was achieved by immersion in a 2 mM

CH3CN solution of 2+ClO4�. 2+ in Nafion is reduced to 2, at

the initial potential used in cyclic voltammetry.

UV-visible spectrophotometry

The UV-visible spectra of 2+ in CH2Cl2 and 2+ supported in

Nafion in water were measured in 1 cm quartz cuvettes using a

Varian Cary 300 BIO spectrometer. Nafion membranes were

cast on the inside of quartz cuvettes, dried at 120 1C for

20 min, doped in situ with 2 mM solutions of 2+, rinsed with

acetonitrile, and then air-dried.

Light source

The xenon light source used for these experiments generated

white light with a stable output over the range 250–750 nm, from

a Rofin Australia-Polilight PL6, passed through a 1 m long

liquid light guide. Illumination experiments were conducted at a

light intensity of approximately 500 mW cm�2, measured at the

electrode surface.

Results and discussion

The cyclic voltammetry of 1 and 2 was investigated in three

different environments: (i) dissolved in CH2Cl2, (ii) as a solid

layer on the electrode surface in contact with water, and

(iii) when suspended in a Nafion membrane cast on the

electrode surface, again in contact with water. Using these

procedures, oxidation and reduction processes of the cubanes

are now described in detail.

Compound 1 has a very low solubility in many solvents

arising from the hydrophobic outer shell formed by the 12

phenyl rings of the phosphinate ligands. However, moderate

solubility is achieved in dichloromethane. As a consequence,

the solution electrochemistry for both compounds could be

studied over a concentration range of 0.4–1.0 mM in this

solvent. The nature of both oxidation and reduction processes

in dichloromethane was investigated by cyclic voltammetry,

rotating disc and bulk electrolysis; the reduction process was

studied in the presence of proton donors.

The electrochemistry of 1 was examined as a solid layer on

the surface of a glassy carbon electrode in contact with

aqueous electrolyte. Layers of this type were prepared by both

rubbing the electrode in solid cubane or casting a solution of

cubane dissolved in dichloromethane onto the electrode

surface and allowing the solvent to evaporate. Solid layers

were immersed in aqueous solutions of 0.1 M Na2SO4 and

were physically stable for at least 24 h.

The incorporation of 2+ into Nafion films was achieved by

immersing Nafion-modified glassy carbon electrodes into a

solution of 2+ClO4� (2 mM in CH3CN).4 Cubane in-

corporation was confirmed by the UV-visible and NMR

spectra of doped membranes on quartz,4 and by cyclic

voltammetry of doped films deposited on electrode surfaces

and immersed in an aqueous (0.1 M Na2SO4) electrolyte. The

voltammetry was also investigated when the Nafion-modified

electrode was in contact with buffered aqueous electrolyte.

Representative cyclic voltammograms obtained at slow scan

rates for each environment studied are displayed in Fig. 2.

Compounds 1 and 2 each exhibit one chemically reversible

oxidation and one chemically irreversible process in CH2Cl2(0.1 M Bu4NPF6). Voltammetry of 1 as a solid film on the

electrode surface in contact with aqueous 0.1 M Na2SO4

exhibits both an oxidation and a reduction process as well as

producing additional oxidation peaks at less positive potentials

after sweeping through the reduction step. Suspension of 2+ in

Nafion and immersion in aqueous electrolyte (0.1 M Na2SO4)

also reveal a well-defined process at positive potentials that is

known to generate a highly reactive 2+ species that acts as a

water-oxidation catalyst under conditions of bulk electrolysis.

For the Nafion system, the peak current magnitudes for the

reduction process are relatively weak compared to the waves

for the oxidation process (Fig. 2).

Oxidation in dichloromethane solution

Cyclic voltammograms for oxidation of 1 and 2 (scan rate of

10 mV s�1) are depicted in Fig. 3(A). A summary of

the electrochemical data derived from these experiments is

presented as a function of scan rate in Tables S1 and S2, ESIw.For compound 1, the reversible formal potential (E1f) derived


from the average of the oxidation (Eoxp ) and reduction (Ered

p )

peak potentials [(Eoxp � Ered

p )/2] gives a value of 690 � 3 mV,

which, as expected, is almost unaffected by scan rate, whilst

the peak-to-peak separation (DEp) values observed for 1 are

slightly larger than the scan rate independent value of about

56 mV at 22 1C that is theoretically predicted for a reversible

one-electron process. They are equivalent to those obtained

for the known, reversible, one-electron Fc/Fc+ oxidation (for

the same Fc concentration and at the same scan rate in CH2Cl2(0.1 M Bu4NPF6)). Thus, oxidation of 1 is concluded to

represent an electrochemically and chemically reversible

one-electron process, with the higher than theoretically

expected DEp value predominantly attributed to un-

compensated ohmic IR drop in the high resistance CH2Cl2medium. The even larger DEp values for the oxidation of 2

imply that the rate of electron transfer for the 20/2+ process is

slower than for either 10/+ or Fc0/+. The oxidation of 2 is

therefore considered to be quasi-reversible at a glassy carbon

electrode. The E1f values determined by cyclic voltammetry for

the Mn4O4L6 " [Mn4O4L6]+ + e� processes are 690 � 3 mV

and 542 � 8 mV for 1 and 2, respectively, representing a

difference in E1f values of 150 mV.

In the initial brief report by Wu et al.,6 a difference in the E1fvalues of 109 mV was reported in CH2Cl2 (0.1 M Bu4NPF6)

(E1f values for 1 and 2 (0.4 mM) were 813 mV and 704 mV,

respectively (vs. Ag/Ag+ (0.1 M AgNO3 in CH3CN))). Also,

DEp values were reported to be much larger than those found

here (297 mV vs. 95 mV for 2 at a scan rate of 100 mV s�1),

suggesting that a higher level of uncompensated resistance was

present. This would have introduced considerable uncertainty

into the E1f values.6 The difference in the E1f values, for

1+ClO4�/1 and 2+ClO4

�/2, of 150 mV found in a more recent

study11 compares favorably with the result from the

current study.

The half-wave (E1/2) potentials obtained from rotating

disc voltammetry (Fig. 3(B)) were slightly more positive than

the E1f values, again consistent with the presence of un-

compensated IR drop and quasi-reversibility. For 1, E1/2

becomes slightly more positive with increasing rotation speed.

In contrast, E1/2 increased significantly with rotation rate for

2, as expected for a quasi-reversible system. E1/2 values range

from 696 to 709 mV for 1 and 542 to 607 mV for 2 (vs. Fc/Fc+)

when the scan rate changed from 500 to 5000 rpm. Diffusion

coefficients of 4.6 (�0.1) � 10�6 cm2 s�1 for 1 and 3.5 (�0.2) �10�6 cm2 s�1 for 2 were calculated from the rotation rate data

Fig. 2 Cyclic voltammograms obtained at 22 1C and with a scan rate

of 10 mV s�1 at a glassy carbon electrode for oxidation and reduction

of: (A) 0.4 mM CH2Cl2 (0.1 M Bu4NPF6) solutions of 1 (----) and 2

(—), 10 mV s�1; (B) solid layer of 1 adhered to the glassy carbon

electrode surface in contact with aqueous (0.1 M Na2SO4);

(C) 2+-doped in Nafion adhered to a glassy carbon electrode and

placed in contact with aqueous (0.1 M Na2SO4) electrolyte.

Fig. 3 Voltammograms obtained for oxidation of 1 and 2 at 22 1C at

a glassy carbon electrode: (A) Top: cyclic voltammetry with a scan rate

of 10 mV s�1 for 0.4 mM 1 (----) and 2 (—) in CH2Cl2 (0.1 M

Bu4NPF6); (B) Bottom: rotating disc voltammogram obtained for

1 mM 2 with a rotation rate of 500 rpm and a scan rate of 5 mV s�1.


using the Levich equation. The slightly lower value for 2 may

be attributed to its higher molecular weight (1952 amu for 2 vs.

1587 amu for 1).

Bulk oxidation at a glassy carbon electrode of solutions of 1

(0.4 mM) and 2 and (1 mM) in CH2Cl2 was conducted at

a potential 100 mV more positive than the rotating disc

voltammetric E1/2 value for each complex. Coulometric

analysis of these experiments gave an n-value of 0.97 � 0.05

electrons, providing confirmation that the oxidation of 1 and 2

is a one-electron process (eqn (1)).

[MnIII2MnIV2O4L6] " [MnIIIMnIV3O4L6]+ + e� (1)

Reductive back electrolysis of the oxidized compounds yielded

an average of 67% of the charge collected in the oxidation

process, suggesting [Mn4O4L6]+ is not entirely stable in

CH2Cl2 over the time period of the experiment (ca. 1 hour).

The limiting current recorded by rotating disc voltammetry,

after bulk oxidation followed by back electrolysis, gave an

average limiting current of 68% of the magnitude of the

pre-electrolysis limiting current value, consistent with the

coulometry. Analysis of the rotating disc voltammogram

obtained after bulk oxidation indicated that the E1/2 values

of 1 and 2, after allowing for the uncompensated IR drop,

were consistent with those obtained before electrolysis.

The UV-visible spectrum before and after bulk electrolysis

of 1 is shown in Fig. 4. The sharp absorption maxima at ca.

230 nm is attributed to a p - p* transition within the phenyl

rings of the ligands and, as expected, remains present when the

cubane is oxidized. There is a significant increase in absorption

in the range of 300–400 nm after oxidation. The intensity of

the bands in this region suggests that they are due to charge

transfer transitions. Thus, oxidation appears to reduce the

energy required for charge transfer within the complex which

agrees with the shortened metal–ligand bond lengths observed

in the crystal structure of 1+ClO4�.11

No other oxidation processes were observed for either

cubane between 1.0 and 2.0 V (vs. Fc/Fc+) (the latter potential

being the solvent/electrolyte limit). Thus, in CH2Cl2, even with

each manganese center coordinated to three oxo and three

phosphinate oxygens, it is not possible to reach a 4 MnIV state

at potentials less positive than 2.00 V.

Oxidation in Nafion

The oxidation process of the 2/2+ couple in Nafion at a scan

rate of 50 mV s�1 gave Eoxp = 738 mV, Ered

p = 517 mV,

and a mid-point potential of Em calculated as the value

[(Eoxp + Ered

p )/2] = 628 mV vs. Fc/Fc+ (Fig. 5(A)). Ep varied

with scan rate, suggesting significant resistance within the

system. The substantial level of chemical reversibility of the

2/2+ process suggests that the majority of the neutral cubane,

2 (generated electrochemically within the Nafion), is retained

within the membrane. This is likely due to an association

between the hydrophobic regions within the substructure of

the Nafion membrane and the hydrophobic aryl groups of 2,

which are insoluble in aqueous solution. Evidence in support

of this interpretation comes from the substantial red shift

(30 nm) in the ligand p - p* transition at 230 nm.4

In unbuffered aqueous electrolyte, the 2/2+ Nafion/GCE

oxidation process is initially well defined. However, with

potential cycling, the peak current deteriorates and the peak

potentials shift to less positive values (Fig. 5(A)). This pheno-

menon was also observed for oxidation of Mn(II) in Nafion/

GCE (Fig. 5(B)), revealing it to be a feature of the Nafion

support. Equivalent experiments in buffered solution at pH 3.3

lead to stable peak currents and potentials with smaller peak-

to-peak separations (Fig. 6(A)) than those observed in

unbuffered solution (Fig. 5(A)). In buffered solution at

pH 7, the peak current rapidly diminishes upon repetitive

cycling of the potential with the process becoming poorly

defined even after one sweep (Fig. 6(B)). After transferring

this electrode to a pH 3.3 solution environment, the oxidation

Fig. 4 UV-visible absorption spectra in CH2Cl2 (0.1 M Bu4NPF6) at

22 1C of 1 before (TT) and after (---) one-electron oxidative bulk

electrolysis with the potential of the glassy carbon working electrode

held at 950 mV (vs. Fc/Fc+) and after (—) two-electron bulk reduction

of 1, with the electrode potential held at �700 mV vs. Fc/Fc+, stirring

under N2.

Fig. 5 Cyclic voltammetry at 22 1C of 2+ (A) andMn2+ (B) doped in

Nafion on a glassy carbon electrode in H2O (0.1 MNa2SO4), scan rate:

50 mV s�1.


process again became well defined. Thus, the voltammetric

response and conductivity of the system appear to be depen-

dent on the proton concentration within the Nafion.

Our findings imply that the Nafion support provides good

electrical contact between a fraction of the absorbed cubane

and the electrode, with some variability in peak current

magnitudes observed between different preparations. This

is presumably due to the variability of hand casting the

membranes onto the electrode surface. The success of this

system facilitated the investigation of the catalytic potential of

these complexes.4

Electro-photocatalytic oxidation of water

In a previous communication, we reported the photo-assisted

electrocatalytic oxidation of water using 2+–Nafion-coated

electrodes.4 In those experiments 2+–Nafion-modified electro-

des were poised at 0.8 V (vs. Fc/Fc+), above the oxidation

potential of the manganese cubanes, and illuminated with

light. The resulting photocurrent was shown to correlate with

the oxidation of water to oxygen and protons. In the present

study, we compare the photocurrent from the two methods of

electrode immobilization described above. Attempts to dope

1+ into Nafion were unsuccessful due to its insolubility

in polar solvents, which appears to be a pre-requisite for

achieving successful penetration into the Nafion membranes.

Photocurrent comparisons are thus made only for 2+, which is

readily soluble in acetonitrile.

As shown in Fig. 7, illumination of electrodes coated with

a solid layer of 2+ yielded a measurable and higher photo-

current under illumination than the unmodified glassy carbon

electrode alone. The solid layer was found to be unstable

under extended illumination, resulting in detachment of the

solid from the electrode surface on the hour time scale. Wada

and Tanaka reported a similar detachment in their study of the

catalytic properties of [RuII2(OH)2(3,6-t-Bu2quinone)2-

(btpyan)]2+ (btpyan = 1,8-bis(2,20:60200-terpyridyl anthracene))

adsorbed onto electrode surfaces.16 The thickness of the solid

layer could be adjusted by varying the amount of 2+ on the

surface. Increasing the thickness typically reduced the stability

of the solid layer at the electrode surface, leading to easier

detachment during cycling. A decrease in photocurrent from

that shown in Fig. 7 was also observed, as the solid layer

became too thick to allow light to penetrate to the adsorbed

cubane molecules in electrical contact with the electrode

surface.

Nafion-modified electrodes doped with 2+ displayed a large

photocurrent under illumination compared to undoped

Nafion-modified electrodes. Notably, this photocurrent was

almost five times larger than the photocurrent observed from

the electrode coated with a solid layer, under equivalent

conditions. We previously reported that the 2+–Nafion system

produced photocurrent for a period of 65 hours.4 These results

reveal that immobilizing 2+ in a Nafion membrane at the

electrode surface is a more effective method of utilizing

the cluster for photocatalysis than the deposition of a solid

layer on an electrode surface.

Reduction in dichloromethane solution

The reduction of 1 and 2 is chemically irreversible in solution

at the low scan rates used in Fig. 2(A) and 7. At higher scan

rates (Fig. S1, ESIw), an oxidation peak is just detectable in the

positive potential scan direction. The scan rate dependence

implies that the reduced cubane (0.4 mM) undergoes a rapid

chemical reaction that inhibits re-oxidation. Reduction peak

potentials for 1 and 2 are highly dependent on scan rate

with average values of around �620 mV and �650 mV

(vs. Fc/Fc+), respectively, over the scan rate range of 10 to

500 mV s�1 (Table S3, ESIw). The mechanism is clearly

complex, but appears to involve a fast chemical reaction after

an initial quasi-reversible (moderately slow) one-electron step

Fig. 6 Cyclic voltammetry at 22 1C of 2/2+ in Nafion/Pt, immersed

in NaHSO4 buffer, pH 3.3 (A) and MOPS buffer (see Experimental

section), pH 7 in Na2SO4 (B), at 50 mV s�1.

Fig. 7 Controlled potential electrolysis obtained at 22 1C, 0.8 V vs.

Fc/Fc+, at a glassy carbon electrode for: (left) Solid layer of 2+ClO4

(—) adhered to the glassy carbon electrode surface, and unmodified

glassy carbon electrode (---) in contact with aqueous (0.1 M Na2SO4);

(right) 2+-doped in Nafion (—) and undoped Nafion (---) adhered to a

glassy carbon electrode, in contact with aqueous (0.1 M Na2SO4)

electrolyte.


(EC mechanism) perhaps followed on longer time scales (see

Fig. 1 (without acid) and equations below) by a further one-

electron reduction to give an ECEC type process of the kind:

[MnIII2MnIV2O4L6] + e� " [MnIII3MnIVO4L6]� (2a)

[MnIII3MnIVO4L6]� - product 1 (2b)

product 1 + e� " reduced product(s) 2 (2c)

The higher scan rate data imply that the reversible potentials

for the initial one-electron reduction processes are ca. 505 mV

and 530 mV for 1 and 2, respectively. This small difference of

only 25 mV is significantly less than that of about 150 mV

found for the oxidation process. Wu et al.6 previously reported

reduction peak potentials for 1 and 2 of �571 mV and

�884 mV, respectively (vs. Ag/AgNO3), with a difference of

over 300 mV. The reasons for this discrepancy are unclear but

as the reduction process is irreversible, significant uncertainty

is associated with values obtained at a single scan rate in a high

resistance medium.

Peak current magnitudes per unit concentration for

reduction and oxidation of 1 and 2 are similar, which may

imply that they are both one-electron processes (n = 1).

However, the peak heights for the reduction process

(eqn 2(a–c)) are a complex function of many variables in

addition to the n-value. In contrast, the limiting current (iL)

value at a rotating disc electrode directly reflects the n-value, if

the processes are mass transport limited in this potential

region.

Rotating disc experiments conducted on the reduction

processes (Fig. S2, ESIw) revealed limiting current magnitudes

(opposite sign) per unit concentration that were approximately

double that of the known one-electron oxidation process. For

example, in the case of 1, 0.4 mM in CH2Cl2 (0.1 M Bu4NPF6,

at a rotation rate of 1000 rpm and scan rate of 5 mV s�1) ioxL =

9.9 mA, iredL = �20 mA, suggesting the overall reduction step is

a two-electron process. Bulk reductive electrolysis of 0.4 mM 1

and 1 mM 2 in CH2Cl2 (0.1 M BuNPF6), at potentials 100 mV

more negative than Eredp , gave an n-value of 1.96 � 0.05

electrons, confirming that this is an overall two-electron

reduction process on long time scales. Oxidative bulk electro-

lysis of the reduction product yielded an insignificant amount

of charge, confirming that the process is chemically irreversible

on long time scales.

The intensity of the UV absorption at 250–400 nm became

significantly lower after a two-electron reduction (Fig. 4). This

may be due to a decrease in ligand to metal charge transfer

transitions, caused by an elongation of the core metal–ligand

bonds arising from the decrease in the manganese oxidation

state. However, it is likely that the two-electron reduced

species has undergone a change in structure, so the decrease

in intensity may also be due to the reduced cubane having

undergone a chemical change, such as the release of phosphinate

ligands.

Another fully irreversible reduction process is observed

at approximately �1500 mV in cyclic voltammograms.

Rotating disc experiments for this reaction give an iredL value

that is at least 10 times greater than the observed ioxL for

the reversible oxidation process, implying that this is a

multi-electron reduction process. A brown-black precipitate

is formed on the electrode surface which blocks the surface on

repeated cycling in protic media.

A 0.4 mM solution of 1 in CH2Cl2 (0.1 M BuNPF6)

was analyzed by cyclic voltammetry in the presence of

2.4 mM diphenylphosphinic acid. In the presence of the

proton source, the oxidation wave was slightly broadened

and two reduction processes were observed in the potential

range between 0 and �1.00 V (Fig. 8(A)). These processes are

also seen after addition of p-toluenesulfonic acid. A 2.4 mM

diphenylphosphinic acid solution in CH2Cl2 (0.1 M BuNPF6)

displays no faradaic current at potentials between�1.00 V and

1.00 V. Furthermore, splitting of the processes was not

observed when tetrabutylammonium diphenylphosphinate

was added to dichloromethane solutions of the cubane,

indicating that the presence of a proton donor is the origin

of the changes in the voltammetry. The first of the two

processes occurs at a less negative potential than observed

for the previously described cubane reduction (Fig. 5) and

the second at more negative potential than for the cubane

reduction. Analogous splitting into two processes was

observed in the solid film and Nafion supported systems when

these electrodes were immersed in pH 3.3 aqueous electrolyte

solutions (Fig. 8(B)).

Bulk reductive electrolysis was conducted at potentials

slightly more negative than those for the first reduction

Fig. 8 Cyclic voltammograms obtained at 22 1C at a glassy carbon

electrode with a scan rate of 10 mV s�1 for reduction of: (A) 0.4 mM 1

in CH2Cl2 (0.1 M Bu4NPF6), in the absence (---) and presence (—) of

the proton donor (2.4 mM diphenylphosphinic acid); (B) 2

(2+ converted to 2) doped in Nafion adhered to electrode in contact

with (---) aqueous 0.1 M Na2SO4 and (—) 0.1 M NaHSO4, pH 3.3

electrolyte.


process (�450 mV vs. Fc, Fig. 8), on a 0.4 mM solution of 1

with 2.4 mM diphenylphosphinic acid. Coulombic analysis

again implied that an overall two-electron reduction occurred,

as in the absence of acid. After bulk electrolysis, the second

more negative process was absent. Proton assisted reduction

would explain the positive shift in potential for the first process

and the splitting into two processes. Thus, the initial step is

modified from that in eqn (2) to:

[Mn4O4]6+ + aH+ + e� - [Mn4O4Ha]

(5+a)+

- electroactive product(s) (3)

Multiple-process reduction is likely to be occurring on the

longer time scale of bulk electrolysis via an ECE type scheme.

This proton-assisted one-electron process is consistent with

our previous report describing the isolation and purification

of stable 1H via H atom transfer from phenothiazine (top line

in Fig. 1).7 This species was identified as having the core

composition of [Mn4O3(OH)]6+.

The addition of two electrons in this bulk electrolysis

experiment led to precipitation of a red-brown solid, leaving

a cloudy white solution. The electronic absorption spectra

of this solution reveal only the ligand absorption maxima

(B230 nm), implying that substantial decomposition of the

manganese core has occurred on long time scale experiments.

In addition to the solid product, water droplets formed on the

side of the glass cell, which was not observed in any of

the previously described bulk electrolysis experiments. The

red-brown solid obtained from the acidic bulk electrolysis

reduction experiment was insoluble in organic and aqueous

solvents and is proposed to be a reduced manganese oxide.

The cloudy white solution has been seen before in the course of

cubane reduction.10 Infrared data collected on this previous

solid indicated that it corresponded to a polymeric

Mn(II)(phosphinate)2.

Previous chemical reduction studies have revealed that the

cubane can undergo four-electron/four-proton reduction

process via proton-coupled electron transfer (PCET), yielding

two water molecules and the [Mn4O2L6] complex.7,10,17 PCET

was shown to occur via a sequential process with core

oxo-ligands being protonated to hydroxides and subsequently

water. This process goes via a reduced intermediate, Mn4O3L6,

that was detected by mass spectrometry after 2 reducing

equivalents were added. This intermediate forms when a

corner oxo is converted to a water molecule in the reduction

process. The presence of water after the reduction of 1 and 2 in

degassed CH2Cl2 solution suggests that the reduction of the

manganese atoms causes a weakening of the Mn–O bonds,

facilitating protonation of the core oxygens and their sub-

sequent release as water molecules (see Fig. 1 (with acid)).

Reduction of quasi-solid phases in contact with aqueous

environments

The electrochemistry of 1 as a solid (Fig. 2(B)) or 2 supported

in Nafion in aqueous 0.1 M Na2SO4 electrolyte yields new

oxidation processes in potential cycling experiments but only

after reduction of the cubane. Furthermore, these processes

are sensitive to the pH of the aqueous electrolyte solution.

Thus, a change of electrolyte to more acidic 0.1 M NaHSO4

causes the reduction component to split into two processes

(Fig. 2(B)), as also occurs when a weaker acid is added to

dichloromethane solutions subjected to solution phase

voltammetry. Clearly, there is a close relationship between

the voltammetry in the organic solvent and the solid phase

studies at chemically modified electrodes, even through

products of the reduction differ.

Conclusions

The high charge of the [Mn4O4]7+ core present in 1+ and 2+

means that the application of very positive electrode potentials

is needed for their generation. This provides a highly reactive

species, previously shown to be able to catalyze the oxidation

of water under illumination.4 We have confirmed that the

one-electron oxidation of the cubane to the cubium form is a

chemically reversible process on the voltammetric time scale

and displays significant stability under synthetic (bulk electro-

lysis) conditions (1 2 1+ transition in Fig. 1). The reversible

potentials for these 10/+ and 20/+ couples are ligand-dependent,

implying that the generation of a high valence state on a tetra-

manganese cluster is facilitated by strong electron donation

from the supporting ligands.6 In the [Mn2O2(bipyridine)4]3+

complex, charge transfer from two oxo-ligands and four

nitrogen lone pairs on the bipyridines gives rise to an electro-

chemical oxidation potential of the MnIIIMnIV core to 2MnIV

at 1.25 V vs. SCE.18 This potential is of the magnitude required

to oxidize aqueous MnII to MnIII (1.27 V vs. SCE). Increasing

electron donation from the ligands to the [Mn2O2] cluster, by

replacing two bipyridine ligands with two terminal dihydrogen

phosphates and one bridging hydrogen phosphate, means that

the 2 MnIV state in [Mn2O2(bpy)2(m-HPO4)(H2PO4)2] can be

achieved at 0.7 V vs. SCE.14

In the cubane cluster, each metal center is supported by

three oxo-ligands and three phosphinate oxygen donors to

achieve a large electron-donating field around the metals,

which causes oxidation of 2MnIII2MnIV to MnIII3MnIV to

occur at potentials as low as 0.73 V vs. SCE, yielding the

[Mn4O4]7+ species described above. The cubium state could be

achieved at even less positive potentials by developing cubanes

with still more strongly electron-donating phosphinate groups.

In principle, this would lower the required energy in catalytic

applications of these clusters.

Catalytic utilization of the cubium state has been achieved

by immobilization of the cluster at electrode surfaces.

A comparison of two methods of catalyst immobilization,

solid layer adsorption16 and doping into electrode bound

Nafion membranes,19–23 revealed that Nafion membranes offer

superior catalyst support, higher photocurrents and greater

stability. The Nafion appears to create a more efficient

distribution of catalytic molecules over the electrode surface

and protects the catalysts from the bulk water electrolyte.

Further investigation of this system is currently underway.

We note that to achieve effective doping of the Nafion

membrane, the catalyst must be positively charged and soluble

in a polar solvent. The casting of solid layers onto an electrode

surface is, therefore, a useful technique for screening

the catalytic potential of molecules that do not meet these

requirements.


Electrochemical investigation of the reduction has revealed it to

involve an overall two-electron process, with products and inter-

mediates being sensitive to the presence of protons (see Fig. 1,

proposed electrochemical reduction with acid and without acid).

Previous chemical reduction studies have revealed that the cubane

can undergo proton-coupled reduction, yielding two water

molecules and a [Mn4O2L6] complex.10 PCET was shown to be

a sequential process, with core oxo-ligands being protonated to

hydroxides and subsequently water. The observed resolution into

two reduction processes in this study, upon addition of protons,

demonstrates that reduction is more readily achieved when

coupled with protonation. The ability of the cubane to undergo

PCET may be an essential property in its ability to catalytically

oxidize water.7

The chemical irreversibility of the two-electron process

2MnIII2MnIV - 4MnIII in dichloromethane solution phase

suggests that a structural or chemical change prevents

re-oxidation of the cluster. A transient one-electron reduced

cluster is probably an intermediate. In the solid films, inter-

mediates are detected although it is unclear whether the original

complex may be reformed under conditions of cyclic voltammetry

or a new water-ligated product is generated. For the Nafion

supported cubane, the peak currents and potentials have been

shown to be sensitive to the pH of the supporting electrolyte, with

the reduction process appearing to undergo a PCET process

similar to that observed in dichloromethane solution. As reported

by Brimblecombe et al.,4 the development of the Nafion system

has provided a significant step forward in the use of these

complexes as photo-assisted oxidation catalysts.

Acknowledgements

This work was supported by the Australian Research Council

through the Discovery Program (LS/GCD/GFS/AMB) and

Federation Fellowship Scheme (AMB), the US National

Institutes of Health (GCD), a Lemberg Fellowship (GCD),

an Australian Academy of Sciences Travel Fellowship (GFS),

an Australian Postgraduate Award (RB), a Fullbright Post-

graduate Award (RB) and a Monash University Postgraduate

Publication Award (RB).

Notes and references

1 J. Barber and J. W. Murraya, Coord. Chem. Rev., 2008, 252,233–243.

2 A. J. Bard and L. R. Faulkner, Electrochemical Methods:Fundamentals and Applications, John Wiley & Sons Inc., New York,1990.

3 C. W. Cady, R. H. Crabtree and G. W. Brudvig, Coord. Chem.Rev., 2008, 252, 444–455.

4 R. Brimblecombe, G. F. Swiegers, G. C. Dismukes and L. Spiccia,Angew. Chem., Int. Ed., 2008, 47, 7335–7338.

5 W. Ruettinger and G. C. Dismukes, Chem. Rev., 1997, 97

1–24.6 J.-Z. Wu, E. Sellitto, G. P. A. Yap, J. Sheats and G. C. Dismukes,Inorg. Chem., 2004, 43, 5795–5797.

7 T. G. Carrell, E. Bourles, M. Lin and G. C. Dismukes, Inorg.Chem., 2003, 42, 2849–2858.

8 T. G. Carrell, S. Cohen and G. C. Dismukes, J. Mol. Catal. A:Chem., 2002, 187, 3–15.

9 W. Ruettinger, M. Yagi, K. Wolf, S. Bernasek andG. C. Dismukes, J. Am. Chem. Soc., 2000, 122, 10353–10357.

10 W. F. Ruettinger and G. C. Dismukes, Inorg. Chem., 2000, 39,1021–1027 (Inorg. Chem., 2000, 39, 4186).

11 J.-Z. Wu, F. D. Angelis, T. G. Carrell, G. P. A. Yap, J. Sheats,R. Car and G. C. Dismukes, Inorg. Chem., 2006, 45, 189–195.

12 M. Yagi, K. V. Wolf, P. J. Baesjou, S. L. Bernasek andG. C. Dismukes, Angew. Chem., Int. Ed., 2001, 40, 2925–2928.

13 W. F. Ruettinger, D. M. Ho and G. C. Dismukes, Inorg. Chem.,1999, 38, 1036–1037.

14 R. Manchanda, G. W. Brudvig and R. H. Crabtree, Coord. Chem.Rev., 1995, 144, 1–38.

15 Y. Yu, M. Dubey, S. L. Bernasek and G. C. Dismukes, Langmuir,2007, 23, 8257–8263.

16 T. Wada and K. Tanaka, Eur. J. Inorg. Chem.,2005, 3832–3839.

17 M. Maneiro, W. F. Ruettinger, E. Bourles, G. L. McLendon andG. C. Dismukes, Proc. Natl. Acad. Sci. U. S. A., 2003, 100,3707–3712.

18 S. R. Cooper and M. Calvin, J. Am. Chem. Soc., 1977, 99,6623–6630.

19 H. Hagiwara, N. Ono, T. Inoue, H. Matsumoto and T. Ishihara,Angew. Chem., Int. Ed., 2006, 45, 1420–1422.

20 M. A. Gondal, A. Hameed and Z. H. Yamani, Energy Sources,2005, 27, 1151–1165.

21 R. Bandyopadhyay, S. K. Maiti and R. Bhattacharyya, Inorg.Chem. Commun., 2002, 5, 452–452.

22 K. Sayama, K. Yase, H. Arakawa, K. Asakura, A. Tanaka,K. Domen and T. Onishi, J. Photochem. Photobiol., A, 1998,114, 125–135.

23 K. Hashimoto, H. Irie and A. Fujishima, Jpn. J. Appl. Phys., 2005,44, 8269–8285.


Electrochemical investigation of Mn4O4-cubane water-oxidizing clusters

Documents

Transcript of Electrochemical investigation of Mn4O4-cubane water-oxidizing clusters