Fundamental Studies on the Electrocatalytic Properties of Metal Macrocyclics and Other Complexes for the Electroreduction of O₂

Abstract

The high prospects of exploiting the oxygen reduction reaction (ORR) for lucrative technologies, for example, in the fuel cells industry, chlor-alkali electrolysis, and metal-air batteries, to name but a few, have prompted enormous research interest in the search for cost-effective and abundant catalysts for the electrocatalytic reduction of oxygen. This chapter describes and discusses the electrocatalysis of oxygen reduction by metallomacrocyclic complexes and the prospect of their potential to be used in fuel cells. Since the main interest of most researchers in this field is to design catalysts which can achieve facile reduction of O₂ at a high thermodynamic efficiency, this chapter aims to bring to light the research frontiers uncovering important milestones towards the synthesis and design of promising metallomacrocyclic catalysts which can accomplish the four-electron reduction of O₂ at low overpotential and to draw attention to the fundamental requirements for synthesis of improved catalysts. Particular attention has been paid to discussion of the common properties which cut across these complexes and how they may be aptly manipulated for tailored catalyst synthesis. Therefore, besides discussion of the progress attained with regard to synthesis and design of catalysts with high selectivity towards the four-electron reduction of O₂, a major part of this chapter highlights quantitative structure–activity relationships (QSAR) which govern the activity and stability of these complexes, which when well understood, refined, and carefully implemented should lead to rational design of better catalysts. A brief discussion about nonmacrocyclic copper (I) complexes, particularly Cu(I) phenanthrolines, and those with a laccase-like structure which exhibit promising activity for ORR has been included in a separate section at the end.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

7.1 Introduction

Providing adequate, clean, and sustainable energy, from the source of exploitation through eventual disposal, is one of the grand challenges of the twenty-first century. Electrochemical energy storage and conversion systems offer some of the most appealing possibilities for providing clean energy. Of the electrochemical energy systems, fuel cells [1] and metal-air batteries [2] have particularly attracted great research interest. The oxygen reduction reaction (ORR) is one of the key reactions in both of these systems, and Pt is the premium cathode catalyst for the reaction [3]. There is also a growing interest of the importance of ORR in chlor-alkali electrolysis due to its energy saving benefits [4]. Owing to the scarcity of Pt, intensive research has been undertaken aimed at improving both its mass- and area-specific activities through its alloying with less expensive metals and through developments in the synthesis of nanoparticles [5–7]. However, with the awareness that even the most ingenious improvements in catalyst synthesis cannot dispel the issue of Pt scarcity and the potential escalation of its cost upon increased demand, it is a prudent endeavor to explore alternative inexpensive catalysts. Metallomacrocyclic complexes, particularly metalloporphyrins (MPs) and metallophthalocyanines (MPcs), have been widely investigated for ORR since the 1960s [8]. However, due to lack of systematic methodologies for prediction of metallomacrocyclic complexes with satisfactory activity and stability, their performances with regard to both activity and stability generally trail those of catalysts derived from Pt. Of the metallomacrocyclic complexes that have been investigated for ORR, the N₄ metallomacrocyclic complexes, specifically MPs and MPcs, generally exhibit better activity compared to N₂O₂ (Pfeiffer complexes), O₄, N₂S₂, and S₄ macrocyclic chelate complexes [9, 10]. These materials are particularly interesting because of their lower costs compared to noble metals, and their high tolerance to methanol crossover, for the case of methanol fuel cell applications. They are also interesting because they provide models where active centers can be identified, and their catalytic activity can be modulated by changing the structure of the macrocyclic ligand [11]. Several factors influence the activity and stability of metallomacrocyclic complexes for ORR. For a given macrocyclic ligand, ORR will vary with the type of central metal ion [12–15]. Conversely, for a given metal ion, ORR will vary with the nature of substituents on the macrocyclic ligand [16–19], due to the electronic density changes they induce on the metal ion. In addition to these inherent limitations, the ORR activity and stability of a given metallomacrocyclic complex are highly dependent on the pH of the electrolyte. This effect is more pronounced in acidic electrolytes than in alkaline ones [20]. The other factors which influence the activity and stability of a given metallomacrocyclic complex include its solubility in a given electrolyte, the method of its immobilization on the electrode, the operating conditions, and whether the ORR is measured with the metallomacrocyclic complex in solution or adsorbed on an electrode [21]. The most commonly used methods for immobilization of metallomacrocyclic complexes on an electrode include dip coating, drop dry, spin coating, electropolymerization, grafting, self-assembled layer(s), sublimation, and spraying. As such, the ORR performance of a given metallomacrocyclic complex may also vary depending on the method of immobilization used. For example, it has been reported in some literature that the potential of the M(III)/M(II) redox couple and oxygen reduction are shifted to more positive values for films formed by electropolymerization compared to films formed by dip coating or drop coating. Therefore, because of the inevitable variations in experimental procedures and conditions from one laboratory to another, it is difficult to make cross-laboratory comparisons of results reported in literature. To avoid delving into this complexity, qualitative and quantitative structure–activity relationships (QSAR) which govern the activity and stability of metallomacrocyclic complexes have been highlighted, which when well understood, refined, and carefully implemented should form a basis for rational design of improved metallomacrocyclic catalysts for oxygen reduction.

MPs and MPcs show very similar physical and chemical properties and they are structurally related to biological catalysts like cytochrome c and hemoglobin. The basic difference between their structures is shown (Fig. 7.1). As it will be discussed in details later, the properties of these complexes are very dependent on the type of central metal (M) and on the nature of substituents on the ligand. It is important to remark that the choice of substituents for the macrocyclic ligands is inexhaustible which leaves plenty of room for tailoring their properties. For example, the properties of metalloporphyrins may be varied widely by means of substitution groups at the β and meso-positions of the ring (Fig. 7.1a). Furthermore, the concepts of supramolecular chemistry and molecular self-assembly offer additional possibilities to vary the properties of metallomacrocyclic [22–24].

Fig. 7.1

Basic structure of a metalloporphyrin (a) and a metallophthalocyanine (b)

Figure 7.2 presents an overview of the intensity of research in the use of MPs and MPcs as electrocatalysts for ORR during the last 4 decades (1982–2011). The graph is interesting as it clearly depicts that intensity of research in ORR follows the trends in global oil prices and world events: increasing as oil prices increase and decreasing as oil prices reduce (see “Oil Price History and Analysis” at http://www.wtrg.com/prices.htm). The 1981/1982 period recorded one of the peak oil prices in history. The increased research interest in ORR using MPs and MPcs in the 1982–1986 period may be related to two major events that led to high oil prices: the Iran/Iraq war and the imposition of price controls by the United States on her domestically produced oil, resulting in the US consumers paying more for imports than domestic production.

Fig. 7.2

Graphical representation of the publication growth (% growth = number of publications per 5-year period divided by the total number of publications in the last 30 years, multiplied by 100) from 1982 until 2011. Raw data obtained from SciFinder Scholar(R) search engine

The increased research activity in the 1987–1991 period is associated with the spike in oil prices in 1990 due to low production and the uncertainty relating to invasion of Kuwait by Iraqi and the ensuing Gulf war. Crude oil prices were low over the 1992–1996 period (about US$ 20 per barrel). The 1997–2001 period was the period of the Asian financial crisis that led OPEC to increase oil production quota by 10 %. In fact, this period recorded one of the lowest oil prices in history (about US$ 18 per barrel in 1998). It is not surprising that the intensity of research in ORR for fuel cells and metal-air batteries was at its lowest during this period. Since 2002 to date, research in ORR has continued to increase. This is mostly due to the high oil prices as a result of the Asian economic growth, weak US dollar, Iraq war, world economic recession, and the Arab uprising termed the “Arab spring,” and the increased awareness that the current energy sources cannot fulfill future energy demands. The increased global awareness about the detrimental effects of using fossil energies to the environment and the fact that fossil energy reserves are finite have also in part contributed to the increased intensity of research in ORR for fuel cells.

Apart from high oil prices, the world has increasingly become concerned about the need for environmental protection by minimizing the emission of greenhouse gases emanating from the burning of coal and oil. To address these environmental concerns, most countries of the world are now investing in Research and Development (R&D) for advancement of renewable energy technologies, which may further explain the recent increased research activities in ORR (2002–2011). For example, the USA introduced the Energy Policy Act of 2005 to encourage investment in renewable technologies (http://www.gpo.gov/fdsys/pkg/PLAW-109publ58/pdf/PLAW-109publ58.pdf). In a similar vein, an Advanced Energy Initiative (AEI) related to climate change was introduced in 2006 by the USA to accelerate renewable energy and technologies and reduce greenhouse gas emissions. The AEI provided for a 22 % increase in funding for clean-energy technology research in clean coal technology, nuclear power, and renewable solar and wind energies [25].

7.1.1 Reaction Pathways for the Reduction of Molecular Oxygen

The electrochemical reduction of oxygen in aqueous solutions is a complex multielectron reaction that occurs via two main pathways: one involving the transfer of two electrons to give peroxide and the so-called direct four-electron pathway to give water. The latter involves the rupture of the O–O bond. The nature of the electrode strongly influences the preferred pathway. Most electrode materials catalyze the reaction via two electrons to give peroxide. The several possible pathways are summarized in Table 7.1.

Table 7.1 Possible pathways for the oxygen reduction in aqueous media

In strongly alkaline solutions or in organic solvents, O₂ is reduced via the transfer of a single electron to give a superoxide ion: ${{\mathrm{ O}}_2}+{e^{-}}\rightleftarrows {{\mathrm{ O}}_2}^{-}$ (E° = −0.33 V vs. NHE). The maximum free energy or the highest oxidant capacity of O₂ is obtained when this molecule reacts on the cathode of a fuel cell via four-electrons. So there is a need for catalysts that promote the four-electron reduction pathway. Most common electrode materials only promote the two-electron pathway, which releases almost one-half the free energy compared to that of the four-electron pathway. This is due in part to the relatively high dissociation energy of the O–O bond (118 kcal/mol). The four-electron reduction of O₂ to give water involves the rupture of the O–O bond and can involve the interaction of O₂ with one site (single site) or with two active sites simultaneously (dual site) on the electrode surface (Fig. 7.3).

Fig. 7.3

Different spatial configurations for molecular oxygen when it interacts with metal sites

Upon these possible interactions, the energy of the O–O bond decreases, favoring its rupture since electrons accepted by the O₂ molecule will occupy antibonding π* orbitals. On platinum, O₂ reduction occurs almost entirely via four-electrons [26]. It is likely that on this metal O₂ interacts via the “bridge cis” conformation, involving two metal active sites (see Fig. 7.3) since the Pt–Pt separation in certain crystallographic orientations is optimal for this type of interaction. It is then crucial to develop low-cost catalysts that decrease the overpotential of the reduction of O₂ and that can also promote the four-electron reduction.

7.1.2 Evaluation of Catalysts for ORR

It is desirable to evaluate potential catalysts for fuel cell reactions in fuel cell prototypes under the real conditions of application. The standard approach involves profiling the voltage and power output as a function of the current drawn by a load during operation of a fuel cell. A schematic representation of the typical features of the voltage/power–current graphs is shown in Fig. 7.4. Fuel cells generally exhibit a substantial drop in voltage from the theoretically expected maximum voltage, even at their open circuit potential. When a load is applied to the cell, the cell voltage drops further as more and more current is drawn from it. For intermediate current densities, the voltage drops linearly as the current drawn from it is increased as shown in Fig. 7.4. As the current drawn from the cell is increased further up to some point, mass-transport limitation of either reactants, products, or both ensues and a drastic decline in the cell voltage is observed.

Fig. 7.4

Schematic representation of the typical voltage/power–current curves during operation of a fuel cell

The effective voltage E of a fuel cell at a given current density, taking into account the various voltage loses, is given by $E={E_{\mathrm{ eq}}}-{\eta_{\mathrm{ ORR}}}-{\eta_{\mathrm{ IR}}}-{\eta_{\mathrm{ diff}}}$, where E _eq is the theoretical thermodynamic voltage, η _ORR is the activation overpotential due to slow electrode reactions, η _IR is the overpotential due to ohmic resistances in the cell, and η _diff is the overpotential due to slow diffusion of reactants, products, or both. For hydrogen fuel cells, the voltage losses due to electrooxidation of H₂ at the anode are insignificant compared to the voltage losses due to ORR. Therefore, when comparing different catalysts for a given fuel cell configuration, the best catalyst or pair of catalysts for the anodic and cathodic reactions should be the one(s) which exhibit(s) minimum loss in cell voltage at a given current density. Said otherwise, the best catalyst should be the one which operates at the highest efficiency (η), η being defined as the ratio of the operating voltage (E) to the theoretical thermodynamic voltage (E _eq): η(%) = (E/E _eq) × 100. In addition to the efficiency criterion, long-term operation of a fuel cell is an equally important requirement. A desirable catalyst should therefore be characterized with a high efficiency and also exhibit minimal loss in activity over long-term use.

However, the costs for installation of a fuel cell prototype are fairly high, and the devise is fairly expensive to operate. Additionally, long-term stability tests are particularly time consuming, which would make preliminary evaluation of catalysts using a fuel cell prototype unnecessarily costly, both resource-wise and time-wise. Preliminary evaluation of catalysts for fuel cells therefore often involves independent investigation of their half-cell reactions. A brief description of the methods that are most commonly used to investigate catalysts for the oxygen half-cell reaction follows in the next section.

7.1.3 Determination of the Selectivity of O₂ Electroreduction

Ordinarily, the selectivity of ORR is determined by means of rotating disk electrode (RDE) or rotating ring-disk electrode (RRDE) voltammetry. The RDE is designed to boost the diffusion of an electroanalyte in conditions where an electrochemical reaction is limited by diffusion of the analyte to the electrode. In RRDE, a ring (often platinum) surrounds the disk electrode with an insulating material (usually Teflon) between them (Fig. 7.5).

Fig. 7.5

Schematic representation of the RDE and RRDE (left panel) and of the possible reactions that take at the RRDE during ORR (right panel)

During rotation of the electrode, the electrolyte is tangentially drawn to the disk electrode and radially swept away from it under controlled hydrodynamic conditions. Levich showed that the diffusion limited current (i _d) measured at a smooth disk electrode under controlled rotation is related to the angular velocity speed of rotation of the electrode according to Eq. (7.1) [27]:

$$ {i_{\mathrm{ d}}}=0.62nFAD_0^{2/3}{\omega^{1/2 }}{\upsilon^{-1/6 }}{C_0} $$

(7.1)

where n is the number of electrons exchanged per molecule, F is the Faraday constant, A is the surface area of the electrode, D ₀ is the diffusion coefficient of the electroanalyte, ω is the angular velocity of the electrode, υ is the kinematic viscosity of the electrolyte, and C ₀ is the bulk concentration of the electroanalyte. For a reaction which is essentially under diffusion limitation, a graph of i _d against ω ^1/2 yields a straight with a slope = 0.62nFAD ₀ ^2/3 C ₀υ^–1/6 from which n may be determined. Note that the coefficient 0.62 is used when ω is expressed in rad s^–1, while 0.21 is used when ω is expressed in revolution per minute (i.e., 0.62 × (2π/60)^1/2 = 0.21) [28]. A more commonly used approach employs a modification form of Eq. (7.1), called the Levich-Koutecky analysis [Eq. (7.2)], for reactions which are under mixed kinetic and diffusion control, where i is the measured current, i _k is the kinetic current defined by Eq. (7.3), and i _d is the term in Eq. (7.1):

$$ \frac{1}{i}=\frac{1}{{{i_{\mathrm{ k}}}}}+\frac{1}{{{i_{\mathrm{ d}}}}}\;\mathrm{ or}\;\frac{1}{i}=\frac{1}{{{i_{\mathrm{ k}}}}}+\frac{1}{{B\sqrt{\omega }}} $$

(7.2)

$$ {i_{\mathrm{ k}}}=knAF{C_0} $$

(7.3)

From Eq. (7.2), a graph of the inverse of the measured current at a given potential i ^–1 against ω ^–1/2 gives a straight line with a slope of B ^–1 from which n can be determined and i _k ^–1 as the intercept on the i ^–1 axis [27].

The RRDE is designed in such a way that under the hydrodynamic conditions, an electroactive species generated at the disk may be detected at the ring electrode. The species generated at the disk electrode must therefore be sufficiently long-lived to be able to traverse the radius of the disk electrode and be detected at the ring. For the case of the ORR, the ring electrode is always poised at a potential where any H₂O₂ generated at the disk is oxidized at the ring. The fraction of H₂O₂ generated during ORR is calculated from Eq. (7.4), while the number of electrons transferred (n) is obtained from Eq. (7.5):

$$ X{{\mathrm{ H}}_2}{{\mathrm{ O}}_2}=\frac{{2{I_{\mathrm{ R}}}/N}}{{{I_{\mathrm{ D}}}+{I_{\mathrm{ R}}}/N}} $$

(7.4)

$$ n=\frac{{4{i_{\mathrm{ O}2}}}}{{{i_{\mathrm{ H}2\mathrm{ O}2}}/N+{i_{\mathrm{ O}2}}}} $$

(7.5)

where N is the collection efficiency of the ring electrode, I _R is the ring current, I _D is the disk current, while i _O2 is the electrocatalytic oxygen reduction current at the disk, and i _H2O2 is the corresponding ring current due to H₂O₂ oxidation at a specific potential. The value of N is normally supplied by the manufacturer but it is advisable to verify it as often as possible using a suitable redox pair [29].

RDE and RRDE are very convenient voltammetric methods for studying the mechanism and kinetics of ORR and are by far the most widely used methods. However, it is important to bear in mind that the underlying mathematical formulations of these methods are theorized for smooth electrode surfaces under laminar flow hydrodynamics. There are many examples in recent literature where RDE and RRDE have been used to study catalyst films for which turbulent flow hydrodynamics is quite obvious. The collection efficiency of RRDE for microscopically disordered films, for example, very porous materials and irregularly built-up films (as may be the case for catalysts modified with nanocarbons such as carbon nanotubes and graphenes), is likely to be determined erroneously due to sporadic hydrodynamics. Therefore, the quality of a given catalyst film has a great influence on the correctness of results obtained from RDE and RRDE. It is generally recommended that catalyst films for RDE and RRDE studies should be as thin as possible [29]. Thick films may lead to increased mass-transport resistance through the catalyst layer and incomplete utilization of the catalyst which certainly lead to incorrect interpretation of results. These factors have to be considered critically when performing and interpreting RDE and RRDE measurements. A depiction of the possible processes that occur during the electroreduction of oxygen at RRDE electrodes is schematically shown in Fig. 7.4 to draw attention to some possible sources of error in treatment and interpretation of RRDE results. If O₂ is reduced by the transfer of two electrons to form H₂O₂, at least four competing reactions may follow and these include a competition between further electroreduction of H₂O₂ and its disproportionation on the disk, and if the H₂O₂ makes it to the ring, a competition between its electrochemical detection and disproportionation takes place since H₂O₂ is known to disproportionate on Pt surfaces. Therefore, the amount H₂O₂ detected at the ring is likely to be much lower than the amount actually produced because of the competing reactions. The magnitude of the error encountered in the determination of n and H₂O₂ using RRDE is likely to be even larger the less smooth the catalyst film is. However, in the absence of a more reliable method for studying the mechanism and kinetics of ORR, RDE, and RRDE voltammetries continue to be useful and handy. In the face of these drawbacks, several groups have proposed the use of scanning electrochemical microscopy (SECM) to study the selectivity of electrocatalysis of ORR [30–34].

7.2 From Model Structures to Active N₄-Metallomacrocyclic Catalysts

There is a vast amount of scientific literature about ORR catalysis by metallomacrocyclic. As mentioned earlier in the introduction, a variety of factors influence the ORR activity of these complexes, including the method used for immobilization of the complex, the pH of the electrolyte, and the conditions and quality of the experiments among others. As such, there is some ambiguity concerning the ORR selectivity of some complexes. In the next Sects. 2.1–2.4, a discussion of ORR catalysis by N₄-metallomacroyclic complexes is given with special emphasis placed upon those complexes and their design aspects which facilitate the four-electron reduction of oxygen. Nature achieves facile reduction of O₂ to water at very high turnover frequencies in the terminal respiration chain by cytochrome c oxidases (CcO) at their heme (Fe a₃)/Cu (Cu_B) bimetallic active site (Fig. 7.6a). This makes CcO interesting model systems to emulate, in what should be appropriately termed as bioinspired catalyst design. The next section gives a brief insight of the electrocatalysis of oxygen reduction by CcO and the progress attained in the biomimetic design of artificial heme/Cu-like catalysts for oxygen reduction.

Fig. 7.6

Active structure of cytochrome c oxidase (a) and of a synthetic analog of cytochrome c oxidase (b) [35]

7.2.1 Oxygen Reduction by Cytochrome c Oxidases

Natural cytochrome c oxidases (CcO) catalyze the reduction of oxygen at their heme a₃/Cu_B (Fig. 7.6a) bimetallic site directly to water without the release of superoxide or peroxide. The Fe–Cu distance in CcO varies in the range of ~4.9–5.3 Å [36], depending on the redox states of the metal ions and ligation thereof, depending on the protein environment.

Recent work by Collman and Ghosh [35] has provided additional evidence that the active site in CcO is comprised of a heme a₃ of an iron porphyrin and Cu (Cu_B) coordinating three histidine groups with one of the histidine groups bound to a posttranslationally modified tyrosine residue (Fig. 7.6b). It is reported that the primary role of the redox centers is to rapidly provide the four electrons needed to reduce oxygen directly to water without the release of toxic superoxide or peroxide species.

The first step of the reduction process involves adsorption of oxygen at the reduced Fe^II/Cu^I center to form an Fe^III–O₂ ^– superoxide adduct with subsequent formation of an intermediate comprised of oxidized Cu^II, an Fe^IV=O ferryl radical, and a peripheral phenoxide radical (Fig. 7.7). The oxidized intermediate is then reduced directly to water by simultaneous transfer of four electrons [35].

Fig. 7.7

A simplified mechanism leading to formation of the oxidized intermediate in the course of oxygen reduction at the active site of a functional heme/Cu analog of CcO [35]

Tremendous effort has been devoted to synthesize artificial catalysts which can achieve reduction of oxygen directly to water by mimicking the active site of CcO [35]. A key motivation of this endeavor has also been to use the CcO synthetic analogs as biomimetic models to probe the structure and function of CcO in the respiration chain. After an enduring effort spanning about three decades, successful synthesis of a functional heme/Cu analog with the ability to reduce oxygen directly to water at physiological conditions without generation of toxic peroxide and superoxide species was reported [35–39]. It is generally believed that natural enzymes exhibit unique flexibility, with the ability to sustain long range open-to-closed conformational changes, which is necessary for binding and catalyzing the reaction of small molecules [40]. Synthetic mimics of the heme/Cu active site in CcO must therefore be faultlessly designed with such conformational flexibility as to facilitate adsorption of oxygen, retention of partially reduced oxide species (PROS) intermediates until final products are formed, and release of the products. Artificial analogs of the heme/Cu sites in CcO have tended to be rather rigid, devoid of this conformational flexibility. A number of factors which complicate replication of the heme/Cu site of the CcO system in biomimetic design of functional heme/Cu analogs for oxygen reduction have been discussed [39]. Functional analogs of CcO incorporating dissimilar metal centers other than a heme/Cu active site have also been reported [41]. The success achieved in synthesizing functional CcO analogs and in the elucidation of their mechanism for oxygen reduction not only is useful for understanding the role of CcO in the terminal respiration chain but also gives new impetus to the design of effective molecular catalysts for four-electron reduction of oxygen.

7.2.2 Oxygen Reduction by Bimetallic Cofacial Porphyrins

Cofacial metalloporphyrins refers to a molecular arrangement whereby the metal ions of two independent planar metalloporphyrins exist face-to-face with each other. This molecular configuration may arise by spontaneous molecular arrangement, for example, due to π–π stacking, or by employing special synthetic schemes [42]. Dicobalt cofacial diporphyrins hinged on amide bridges were the earliest bimetallic diporphyrins with close semblance to the heme/Cu system that achieved four-electron reduction of oxygen to water in acidic electrolytes at remarkably low overpotentials [43]. Minor alterations in the dicobalt diporphyrin structures are reported to drastically poison the potency of the catalysts or cause them to revert to two-electron oxygen reduction catalysts [44]. Selection of the right anchoring system for the two individual cobalt porphyrin units to achieve just the right interplanar conformational separation between them is one critical factor in designing dicobalt cofacial diporphyrins with the ability to reduce oxygen to water. The first successful bridging system for two dicobalt diporphyrins was reported by Collman et al. [42, 43] and comprised of two diametrically positioned four-atom amide chains. The potential at which oxygen reduction commences in these complexes is substantially more positive than the standard potential of the O₂/H₂O₂ couple (E = 0.68 V vs. SHE) which thermodynamically precludes production of hydrogen peroxide by any mechanism. Table 7.2 shows examples of dimetal cofacial porphyrins which achieve the four-electron reduction of O₂ to water at substantially low overpotentials.

Table 7.2 Dicobalt cofacial porphyrins with a high selectivity for direct four-electron reduction of oxygen

If one or both cobalt atoms were replaced by other metal atoms, hydrogen peroxide was formed either as the main product or as an intermediate. Chang et al. [40] found that dicobalt diporphyrins anchored on dibenzofuran (DBD) and xanthene (DPX), Fig. 7.8A, exhibited remarkable conformational flexibility, and in both cases, the complexes were able to reduce oxygen directly to water despite having large differences (~4 Å) in the interplanar separation of their metal centers. Two other anchoring systems, anthracene and bisphenylene, have also been successfully used to design four-electron oxygen reduction “pillared” cobalt (II) cofacial diporphyrins, Fig. 7.8A [45, 49]. The flexibility of the dicobalt cofacial porphyrins anchored on DPD and DPX was attributed to the ability of these hinge-like frameworks (or pillared platforms) to considerably open their bite “the Pac-Man effect” as illustrated in Fig. 7.9 in accommodating exogenous ligands.

Fig. 7.8

(A) Examples of dicobalt cofacial bisporphyrins: DPX diporphyrin xanthene (a); DPD diporphyrin dibenzofuran (b); DPXM diporphyrin xanthene methoxyaryl (c); DPDM diporphyrin dibenzofuran methoxyaryl (d). (B) Rotating ring-disk voltammograms of O₂ reduction at pyrolytic graphite disks modified with (a), (b), (c), and (d) [48]

Fig. 7.9

Illustration of the flexibility of dicobalt cofacial porphyrins with a dibenzofuran bridging system showing the “Pac-Man effect.” White = hydrogen, gray = carbon, red = oxygen, blue = nitrogen, and crimson = Co

The effect of the type of anchoring system and porphyrin structure on the ORR activity of dicobalt cofacial diporphyrins can be observed in Fig. 7.8A. The ORR activity is drastically affected by the introduction of 2,6 dimethoxyphenyl groups to the meso-position of the porphyrin ring as can be seen by the increased anodic current registered at the ring. It is however not definitively clear whether it is the electronic changes or steric effects which lead to this sharp drop in activity. Some quantitative results showing the effect of the type of anchoring system on the ORR activity of dicobalt cofacial porphyrins have been reported in the review by Collman et al. [50].

The two cobalt centers in dicobalt cofacial diporphyrins have been reported to act in concert in order to achieve four-electron reduction of oxygen [51, 52]. One of the sites reportedly functions as a Lewis acid to stabilize the intermediate(s) in the cavity, ensuring that it does not dissociate before completion of the reaction [53]. To corroborate the dual-site postulate and specificity of the reaction site, a parallel type of mechanism involving both the two- and four-electron reduction was observed when one of the Co(III) centers in Co₂FTF4 was replaced by Al(III) [53]. A simplified scheme of the proposed mechanism of oxygen by dicobalt cofacial porphyrins is shown in Fig. 7.10 [48].

Fig. 7.10

Proposed mechanism for oxygen reduction at an active site of a dicobalt cofacial diporphyrin (adopted from [48])

Electrocatalysis of O₂ reduction using bimetallic cofacial diporphyrins bearing dissimilar metal ions has been investigated by some groups. Using anthracene as the bridging system, Ni et al. [54] investigated the influence of the nature of the metal centers including Co–Co, Co–Cu, Co–Fe, Fe–Fe, and Fe–H₂ on the electrocatalysis of ORR by cofacial dimeric porphyrins. The cofacial dimeric porphyrin was found to reduce oxygen to hydrogen product as the final product. The other complexes, namely, Co–Fe, Fe–Fe, and Fe–H₂, were reported to catalyze the reduction of O₂ by a parallel mechanism involving both the two-electron and the four-electron reduction pathways. The researchers reported that for those catalysts which reduced O₂ to water, H₂O₂ was not formed as an intermediate.

A major drawback associated with dicobalt cofacial diporphyrins is their drastic decline in activity when adsorbed on other electrode surfaces other than edge-plane graphite (EPG). The catalysts have been reported to revert to two-electron reduction catalysts when adsorbed on other electrode surfaces other than EPG [50]. This has led to the conclusion that axial ligation of surface oxygen groups on EPG to the cobalt ion bears a significant contribution to the ORR activity of these complexes [50]. A second disadvantage is that these complexes are active in limited potential and pH windows outside of which they become readily deactivated. In a particular study which investigated the influence of site availability of these complexes for oxygen adsorption and for axial ligation on their activity for oxygen reduction, that is, whether the sites were located inside or on the outside of the cavity, it was found that four-electron reduction of oxygen was only possible when the sites inside the cavity were both available for interaction with oxygen. As with monomeric mononuclear porphyrins, dicobalt cofacial diporphyrins also gradually lose their activity upon repetitive potential cycling for a few cycles. Pretreatment of the active catalysts with hydrogen peroxide was found to rapidly deactivate them indicating that the mechanism of their deactivation involves chemical attack by peroxide and superoxide species. Despite all these existing drawbacks, the success achieved thus far should stimulate enthusiasm for synthesis of molecular heme/Cu or heme/Cu-like analogs of CcO and dinuclear cofacial porphyrin which can function under the conditions desirable for technological applications. As proposed way back in the original work by Collman et al. [43] it should be possible to modulate the ORR activity of the dicobalt cofacial porphyrins by judicious modification of the porphyrin ring and possibly by exploration of other bridging groups.

7.2.3 ORR by Simple N₄-Metallomacrocyclic Complexes

There is a large amount of literature available about the pathways for oxygen reduction by N₄-metallomacrocyclic complexes, with the reported results conflicting in some cases, which complicates a balanced review of the work. Additionally, the activity and stability of N₄-metallomacrocyclic complexes are affected by a variety of factors as spelt out in the introduction. This makes it difficult to make reliable cross-laboratory comparisons. However, by and large, the number of simple N₄-metallomacrocyclic complexes that can achieve the reduction of oxygen exclusively to water without generation of substantial amounts of hydrogen peroxide is generally very limited. Much of the attention in this section is devoted to research frontiers in view of promising simple N₄-macrocyclic catalysts that can achieve four-electron reduction of oxygen as this is the primary desire of many researchers in the field. To simplify our task, bearing in mind that thorough coverage of all the important works reported over the last four decades by various authors cannot be achieved faultlessly, emphasis of the discussion was placed more on specific structural properties or special catalyst preparation procedures that achieve four-electron reduction of oxygen either (a) in a direct four-electron reduction process, or (b) in a series process via hydrogen peroxide as an intermediate with its further electrochemical reduction to water or dismutation to water and oxygen, or (c) in a parallel type mechanism involving both (a) and (b). Unique cases involving sophisticated modifications, or otherwise, by means of which oxygen is reduced by the transfer of four electrons have also been included in as far as we could access the concerned literature. The rationale for discussing the electrocatalysis of oxygen reduction by these complexes in slight detail was to draw attention to their important properties which furnish them with the unique ability to facilitate the four-electron reduction of oxygen.

7.2.4 Direct Four-Electron Reduction of O₂ by Simple Monomeric N₄-Macrocycles

It has been shown in several studies that the ORR activity of metallomacrocyclic complexes is very dependent on the pH of the electrolyte. Therefore, a given metallomacrocyclic complex may catalyze the reduction of oxygen via the four-electron reduction in a specific pH window, outside of which, it may only afford to it via the two-electron pathway. This is particularly true for Fe and Co porphyrins and phthalocyanines. Fe phthalocyanines generally reduce oxygen via the four-electron pathway under alkaline conditions, whereas the two-electron transfer pathway is predominant under acidic conditions. This is quite the oppositee for Co phthalocyanines which show a higher selectivity towards the four electron trasfer pathway in acidic electrolytes wheras they are essentially two-electron reduction catalysts under alkaline conditions. This subject will be discussed in details in Chap. 3. Most simple monomeric, monometallic N₄-metallomacrocyclic complexes can only achieve the reduction of oxygen to hydrogen peroxide by the transfer of two electrons. A few metallo N₄-macrocyclic complexes can however catalyze the reduction of oxygen in a direct four-electron transfer process and they constitute the subject of discussion in this section.

Many iridium porphyrins, specifically Ir(OEP)R, R = H, alkyl or aryl derivatives, and Ir(OEP)I and Ir(OEP)OOH, catalyze the direct four-electron reduction of oxygen at substantially low overpotentials [55]. With the exception of Ir(OEP)H, the rest of the complexes require preconditioning at specific potentials to be activated, the required conditioning potential varying depending on the nature of substituent groups.

According to studies by Collman et al. [47] Ir(OEP)R complexes specifically required conditioning at high positive potentials (>0.8 V vs. NHE), while Ir(OEP)I and Ir(OEP)OOH required conditioning at negative potentials (<−0.2 V vs. NHE) to be activated. As can be seen in Fig. 7.11a, Ir(OEP)H has been reported to exhibit the best activity similar to or better than that observed for dicobalt cofacial porphyrins [47, 55]. It was reported in a study by Shi et al. [46] that the presence of a cationic surfactant (DDAB) improved both the activity and stability of Ir(OEP)H and also widened its active potential window, Fig. 7.11b.

Fig. 7.11

(a) Rotating ring-disk voltammograms of iridium chelates irreversibly adsorbed on pyrolytic graphite (Cp) in 0.5 M H₂SO₄ at a rotation of 16 rps and scan rate of 10 mV s^–1 [56] (reproduced with permission of Elsevier); (b) rotating ring-disk voltammograms of a Pt (ring) and graphite (disk) coated with DDAB and [Ir(OEP)]₂ (1.3 × 10^–9 mol cm^–2) upper panel (A) and lower panel (B) Pt ring (uncoated)-Pt disk in 0.5 M H₂SO₄ saturated with air. The Pt rings were maintained at 1.0 V in both cases [46]

The dimeric form of Ir(OEP), [Ir(OEP)]₂, also catalyzes the direct four-electron reduction of oxygen, but unlike its monomer, it does not require any conditioning. As shown in Fig. 7.11a, the iridium chelate complexes become severely but reversibly deactivated at low potentials [56]. Their activity is not however affected by pH as dramatically as the dicobalt cofacial diporphyrin complexes. As with the case for Co₂FTF catalysts, it has been reported that when the Ir(OEP)R catalysts are adsorbed on other electrode surfaces other than EPG, their activity declines to two-electron reduction catalysts. Quite surprisingly, the main product of oxygen reduction was reported to be hydrogen peroxide after pyrolysis of Vulcan supported IrOEP and IrTPP. This is a rather peculiar case since the activity of most N₄-metalloporphyrins increases upon their pyrolysis. It has also been reported that the Ir complexes are unstable in air, losing all their activity after a few months. The four-electron reduction of oxygen by iridium porphyrins has been attributed to a single site which facilitates the adsorption of oxygen on Ir(II)in a side-on configuration, with scission of the O–O bond being a likely step in the mechanism.

Unfortunately, Ir is one of the rarest and most expensive metals which would render its use very costly. Nonetheless, the complex should serve as a suitable model, such that theoretical and experimental knowledge gained from its study may serve to tailor the synthesis of improved catalysts. Table 7.3 lists some metallomacrocyclic complexes which accomplish the reduction of oxygen via the four-electron transfer pathway in acidic electrolytes.

Table 7.3 Metallomacrocyclic complexes with high activity and significantly low overpotentials for oxygen in acidic electrolytes

The other monomeric mononuclear porphyrins which mediate the four-electron reduction of oxygen are cobalt porphyrin (CoP) and cobalt meso-tetramethyl porphyrin (CoTMeP), with CoP exhibiting a higher activity than CoTMeP [57]. The unexpectedly high activity of CoP has been attributed to its likely spontaneous dimerization to produce a more catalytically active species. The synthesis of CoP or its free base (porphine) is quite challenging specifically regarding the attainable yield and purity, which makes it overly expensive for a nonprecious metal catalyst. Another drawback is that the catalyst is highly susceptible to oxidative degradation which severely affects its merit. In a later study which compared the ORR activity of several meso-tetraalkyl cobalt porphyrins by the same group, activity was found to follow the order CoP > CoTmeP > CoTBuP > CoTPrP > CoTEtP > CoPeP [16]. This led to the conclusion that the rate of adsorption of oxygen decreased as the bulkiness of the alkyl substituents increased. At the moment, there is no foreseeable strategy for improving the stability of CoP. Nevertheless, understanding its properties which furnish it with this exceptional activity deserves to be investigated in detail.

7.2.5 Specially Modified N₄-Metallomacrocyclic Complexes

Some studies have shown that certain modification procedures can be used to transform two-electron reduction metalloN₄-macrocyclic complexes into hybrid materials with the capability to reduce oxygen to water, either via the direct four-electron transfer pathway or in the series two-electron transfer pathway. Carbon nanomaterials, carbon nanotubes in particular [58–65], have been reported to significantly increase the catalytic oxygen reduction current, with a substantial reduction of the overpotential for ORR reported in some cases, as shown by the examples in Table 7.4.

Table 7.4 Comparative onset potentials and current densities for MPc and MPc-CNT hybrid catalysts

Ozoemena’s group has recently reported that MOCPcPt (where M = Fe, Ru) (Fig. 7.12a) supported on multiwalled carbon nanotubes afford the reduction of O₂ in a direct four-electron transfer process in 0.1 M NaOH [60, 61, 67]. There was no significant difference between the ORR activity at the FeOCPcPt and RuOCPcPt except that the latter gave a slightly higher kinetic rate constant (~3.6 × 10^–2 cm s^–1) than the former (~2.8 × 10^–2 cm s^–1).

Fig. 7.12

(a) Structure of metallo-tetrakis-(diaquaplatinum)octacarboxy phthalocyanine (MOCPcPt, where M = Fe, Ru) and (b) structure of Co(TPPS) and Co(TMPyP)

The other material design strategies that have been reported to yield four-electron reduction metalloN₄-macrocyclic complexes include the use of supramolecular assembly [68, 69], by exploiting the electrostatic interaction between oppositely charged ions as in the work of D’Souza et al. [70] and Liu et al. [71]. D’Souza et al. reported that a 98 % selectivity of O₂ reduction to water was achieved when a dimeric porphyrin formed by electrostatic coupling of [meso-tetrakis(N-methylpyridyl)porphyrinato]cobalt tetrachloride ([Co(TMPyP)]⁴⁺ Cl₄ ^–) with tetrasodium [meso-tetrakis(4-sulfonatophenyl)porphyrinato]cobalt ([(Na⁺)₄Co(TPPS)]^4–) was used as a catalyst for ORR (Fig. 7.12b) [70]. In a related study, Liu et al. reported the formation of a supramolecular complex [CoTBPyP][SiW₁₂O₄₀] by the electrostatic interaction of meso-tetrakis(4-N-benzylpyridyl)porphyrinatocobalt (CoTBPyP) and a polyoxometalate anion, silicotungstate (SiW₁₂O₄₀)^4–, which was capable of reducing O₂ to water. The application of supramolecular porphyrins for oxygen reduction has been reviewed by Araki and Toma [72].

The multinulear complexes reported by Shi and Anson are another interesting example of specially designed metalloN₄-macrocyclic complexes with exceptional activity for ORR. Cobalt (tetrakis(4-pyridyl)porphyrin), with four [Ru(NH₃)₅]²⁺ and ([(NH₃)₅Os]ⁿ⁺ (n) 2, 3) groups around the porphyrin periphery, generally termed as “multinuclear catalysts” reduce oxygen nearly exclusively to water [52, 73–76]. As shown in Fig. 7.13b, the evident increase in the diffusion-limited current for oxygen reduction at the disk and the large decrease in the anodic current at the platinum ring clearly confirm that modification of CoP(py)₄ with [Ru(NH₃)₅]²⁺ converts it from a predominantly two-electron O₂ reduction catalyst to a nearly exclusively four-electron reduction catalyst with minimal generation of hydrogen peroxide.

Fig. 7.13

(a) Reaction for conversion of CoP(py)₄ to [Co^IIP(pyRu^II(NH₃)5)⁴]⁸⁺ (adopted from [52]). (b) RRDE voltammograms of (A) CoP(py)₄, (B) [Co^IIP(pyRu^II(NH₃)₅)₄]⁸⁺ (reproduced from [74])

The proposed mechanism for oxygen reduction by the multinuclear ORR catalysts is outlined in Fig. 7.14 for tetraruthenated [CoP(pyRu(NH₃)₅)₄]⁸⁺ but is also applicable to osmiumated porphyrins.

Fig. 7.14

Proposed mechanism for O₂ reduction by [CoP(pyRu(NH₃)₅)₄]⁸⁺ and other related multinuclear complexes (reproduced from [52])

The mechanism has been proposed to involve π-backdonation from Ru(II) or Os(II) to the oxygen adsorbed porphyrin adduct Co(II)–O₂, where the cobalt porphyrin is the site for oxygen reduction while Ru and Os serve as cocatalysts which affect the relative rates of two competing pathways for oxygen. The first step in the mechanism (1) involves generation of Co(II) porphyrin with subsequent adsorption of oxygen on the active-site Co(II) in the next reaction (2). The next reaction (3) generates the backbonding Ru(II) which drives the reduction of oxygen by four electrons as shown in reaction (4) and finally, regeneration of the Co(II)–O₂ in reaction (5). It has been proposed that the electronic effects produced by the backdonation of the Ru(II) and Os(II) complexes on Co(II)–O₂ might be achieved by other suitable electron-donating nonmetallic functional groups attached to the porphyrin ring. It was however observed that differences in catalytic activity of multinuclear complexes were correlated to the relative backbonding strengths of the coordinated metal complexes but not on their relative reducing strengths.

Anson et al. postulated that their multinuclear catalysts described above may actually be supramolecular porphyrins with a structure similar to that shown in Fig. 7.15.

Fig. 7.15

Proposed structure of the supramolecular structure formed by complexation of meso-tetrakis((4-pyridyl)porphyrinato)cobalt (CoP(py)₄) with [Ru(NH₃)₃(H₂O)₃]²⁺ on a graphite electrode

Most cobalt porphyrins can only reduce oxygen to hydrogen peroxide with no further reduction or dismutation of the peroxide [77]. Some recent studies have proposed a hybrid multifunctional catalyst incorporating a metallomacrocyclic complex and Prussian blue (PB) or horseradish peroxidase (HRP) so that the hydrogen peroxide generated by the metallomacrocycle can be reduced further by the PB/HRP to water. For example, reduction of O₂ to H₂O was achieved by designing a multicomponent-multifunctional catalyst incorporating carbon nanotubes (CNTs), CoPIX, and PB. The CNTs provide a high surface area matrix for dispersion of the catalysts [78].

According to a study by Forshey et al. [ 79 ] electrodeposited iron meso-tetrakis(N-methyl-4-pyridyl)porphyrin and electropolymerized meso-tetrakis(2-thienyl)porphyrinato]cobalt(II) (pCoTTP) accomplsih the reduction of O₂ to water. The ability of pCoTTP to reduce O₂ directly to water was attributed to a conducting network of CoTTP nodal points where multiple layers are arranged in such a way that they form suitable Co–Co bifacial binding clefts for O₂, thus allowing four-electron reduction of oxygen to water. A study by Elbaz et al. [80] has reported that Co(III) meso-tetra(o-aminophenyl)porphyrin (Co(III)TAPP) and Co(III)(p-sulfonatedphenyl)porphyrin (Co(III)TPPS) incorporated into aerogel carbon (AEG) electrodes by adsorption or electropolymerization achieved four electrons of oxygen.

7.3 Fundamental Studies of O₂ Electroreduction by N₄-Metallomacrocyclic Complexes

The preceding section focused mostly on qualitative description of the dependence of ORR activity on the structure of N₄-metallomacrocyclic complexes and some special modification procedures which can be used to improve activity. In this section, much of the discussion will focus on quantitative decription of the parameters which influence the ORR activity of N₄-metallomacrocyclic complexes. Specifically, the dependence of activity on the properties of the central metal ion will be dicussed in relation to the driving force of the reaction. In addition to this, the molecular orbital theory and the concept of intermolecular are used to describe the interaction between oxygen the central metal ion in N₄-macrocyclic complexes and how this interaction influences the ORR activity of the complex.

7.3.1 Effect of the Central Metal on the ORR Activity of N₄-Macrocylic Complexes

The electroreduction of oxygen by N₄-macrocylic complexes reaction is very sensitive to the nature of the metal center in the complex. For Fe and Mn phthalocyanines, at low overpotentials a four-electron reduction is observed with rupture of the O–O bond [14, 81–83], without the formation of peroxide. In contrast Co, Ni, and Cu phthalocyanines promote the reduction of O₂ only via two electrons to give peroxide [14] as the main product of the reaction. Polymerized Co tetraaminophthalocyanines promote the four-electron [84, 85] reduction, whereas polymerized Fe tetraaminophthalocyanines only promote the two-electron reduction [86] in contrast to their monomeric counterparts. The net catalytic activity of metal macrocyclics is linked to the redox potential of M(III)/(II) of the complexes, the more positive the redox potential, the higher the activity. This trend is the opposite to what is expected from a simple redox catalysis mechanism, which is generally observed for the reduction of O₂ catalyzed by immobilized enzymes. The metal-N₄ chelates need to be supported on a conducting support, like carbon or graphitic materials. Long-term stability is a problem with N₄ chelates. Heat treatment in an inert atmosphere increases both the stability and catalytic activity [87–92].

Even though a few studies have been carried out using the complexes in solution [93], most studies have been performed with the metal chelates confined on an electrode surface, generally graphite or carbon supports, since this is closer to the situation in a fuel cell, where catalysts are absent in the electrolyte. Since the support can act as an axial ligand, the properties of the complexes in solution or on the adsorbed state could be different. So most studies discussed here have been carried out with the complexes immobilized on graphite or carbon supports. Smooth electrodes have been used to study mechanistic aspects of the reaction.

7.3.1.1 Interaction of O₂ with Active Sites and the Redox Mechanism

The one-electron reduction of O₂ to give superoxide is an outer-sphere reaction and does not involve the interaction of the molecule with an active site on the electrode surface. The electron transfer process probably occurs at the outer Helmholtz plane. This process is observed in nonaqueous media or in strongly alkaline aqueous solutions and is not relevant to fuel cell development since little free energy is liberated in the process. In contrast, ORR occurring via the transfer of more than one electron (two or four) is an inner-sphere reaction, involving the interaction of the molecule and or intermediates with active sites present on the electrode surface. Since we are interested in discussing the electrocatalytic reduction of O₂, we will focus our attention on the inner-sphere reduction processes.

O₂ interacts with the N₄ catalysts usually binding to the d orbitals of the central metal in the macrocyclic structure. The energy of the interaction will depend on the energy and the electronic density located on those orbitals. Figure 7.16 illustrates some different possible interactions (end-on and side-on) of the orbitals of the oxygen molecule with the orbitals of the metal in the M–N₄ molecule for end-on and side-on interactions, respectively.

Fig. 7.16

End-on and side-on interactions of frontier orbitals of O₂ with the frontier orbitals of a metal site (reproduced from [94])

For end-on M–O₂ complexes (see Figs. 7.16 and 7.17), the most important interaction for both σ and π bondings occurs with the π* antibonding orbitals of the O₂. The σ interaction is primarily between the metal 3dz² and the in-plane antibonding π_g ^s orbital on the O₂, where the superscript “s” refers to whether or not the orbital is symmetric (or antisymmetric) in relation to the MO₂ plane. This interaction involves a transfer electron density from the metal to the O₂ molecule. The π interaction is primarily between the metal 3dyz and the 1π_g ^a (π antibonding antisymmetric orbital) on the O₂ and can be viewed as a backbonding interaction. In both of these interactions, the π_u ^s and π_u ^a (bonding π orbitals symmetric and antisymmetric, respectively) play a lesser role in the composition of the bonding orbitals [95]. In the case of the side-on interaction, a 3dz² orbital of the metal interacts with the 1π_s ^u bonding orbital (backbonding interaction) and a 3 dxz with a 1π_g ^s antibonding orbital of oxygen as shown in Figs. 7.16 and 7.18. Figure 7.19 illustrates the optimized structural configurations for the O₂ adsorbed on FePc and CoPc molecules according to calculations by Wang et al. [96].

Fig. 7.17

Qualitative molecular orbital diagram for the end-on M(O₂)PL dioxygen adduct. M = Fe, Co; P = porphyrin; L = NH₃ or imidazole (reprinted with permission from American Chemical Society [95])

Fig. 7.18

Qualitative molecular orbital diagram for a side-on Mn(O₂)PL dioxygen adduct (reprinted with permission from [95])

Fig. 7.19

Optimized structural configurations for the O₂ adsorbed on FePc and CoPc molecules. The upper portion shows two end-on configurations. The lower portion of the figure shows two side-on configurations. The central yellow sphere depicts the metal Fe or Co atom, the central two red spheres represent the adsorbed O₂ molecule, blue spheres represent N atoms, gray spheres represent C atoms, and light white spheres represent H atoms [96]

The predominating interactions will weaken the O–O bond and increase the O–O distance. The metal in the complex should be in the M(II) state, so, for example, in alkaline solution a step will require the reduction of M(III):

$$ \mathrm{ M}(\mathrm{ II}\mathrm{ I})-\mathrm{ OH}+\mathrm{ e}\rightleftarrows \mathrm{ M}(\mathrm{ II})+\mathrm{ O}{{\mathrm{ H}}^{-}} $$

(7.6)

An adduct will be formed according to

$$ \mathrm{ M}(\mathrm{ II})+{{\mathrm{ O}}_2}\rightleftarrows \mathrm{ M}(\mathrm{ II}\mathrm{ I})-{{\mathrm{ O}}_2}^{-}\;\mathrm{ or}\;\mathrm{ M}(\mathrm{ II})-{{\mathrm{ O}}_2} $$

(7.7)

This adduct must be short-lived. Otherwise it will hinder further O₂ molecules from interacting with the active site. The adduct will undergo reduction as follows:

$$ \mathrm{ M}(\mathrm{ II}\mathrm{ I})-{{\mathrm{ O}}_2}^{-}+{e^{-}}\rightleftarrows \mathrm{ M}(\mathrm{ II})+\mathrm{ intermediates} $$

(7.8)

where M(II) is the active site. The last reaction shows the process in alkaline media and could involve M(II)–O₂ instead of M(III)–O₂ ^– especially when Co is the metal center. In acid media the process will involve a proton. The scheme above is applicable to Mn and Fe complexes. In the case of Co complexes, Co(III) is probably not formed upon its interaction with the oxygen molecule since the Co(III)/Co(II) formal potential is much more positive than the M(III)/(II) formal potentials (M = Mn, Fe). However, step 2 is still important since the catalytically active site is Co(II) [82, 97]. Density functional theory calculations [96] have shown that that with CoPcs only end-on interaction is possible, whereas for FePc both end-on and side-on interactions are plausible. As it will be discussed further, only Mn and Fe phthalocyanines promote the four-electron reduction of O₂ and this can be attributed to a side-on interaction of O₂, particularly for Fe complexes.

Since the interaction of the oxygen with the active site involves a partial oxidation of the metal in the complex or at least a decrease of electron density in the metal upon interacting with O₂, it is interesting to compare the catalytic activity of metallomacrocyclic with their M(III)/(II) formal potential. Since the formal potentials are sensitive to the pH of the electrolyte, it should be measured in the same media in which the catalytic activity is examined [14, 20, 98]. Further, they should also be measured with the complex adsorbed on the electrode and not in solution phase. When comparing phthalocyanines, the Co and Fe derivatives show the highest activity for the reduction of O₂ but they behave differently. As pointed out above, Co complexes exhibit Co(III)/(II) transition that is far more positive than the onset potential for the reduction of O₂, whereas for Fe complexes the onset potential for the catalytic reduction of O₂ is very close to the Fe(III)/(II) transition [14, 82, 97, 99]. For both types of complexes, there is in situ spectroscopic evidence for the reversible transition involving the M(III)/(II) couples [97, 100, 101]. For example, for Fe phthalocyanine adsorbed on ordinary pyrolytic graphite, Scherson et al. used Fe K-edge XANES (X-ray Absorption Near Edge Structure) recorded in situ in 0.5 M H₂SO₄ to prove the evidence for the redox transition of Fe(III)/(II) involving a metal-based orbital.

Many authors have discussed that a correlation should exist between the formal potential of the catalysts and its activity for ORR and it is yet not clear what sort of correlation should be expected. Reduction of O₂ should be observed at the potential of reduction of the M(III)O₂ ^– adduct and not at the potential of the M(III)/(II) couple if adduct formation takes place before the transfer of an electron from the electrode. The latter should only be observed if the reaction were outer sphere, where O₂ would only collide with the redox center without the formation of a bond. In the special case of iron phthalocyanines and other macrocyclics, O₂ reduction usually starts at potentials very close to the Fe(III)/(II) couple [82, 102, 103]. In contrast, for cobalt macrocyclics reduction of O₂ begins at potentials much more negative than those corresponding to the Co(III)/(II) couple [14]. Several authors have reported correlations between activity (measured as potential at constant current) and the M(III)/(II) formal potential and volcano-shaped curves have been obtained [14, 15, 104, 105].

This could indicate that the redox potential needs to be located in an appropriate window to achieve maximum activity. In other words, a M(III)/(II) formal potential that is too negative (easily oxidable metal center) or a M(III)/(II) formal potential that is too positive (metal center that is more difficult to oxidize) does not favor the catalysis. However, more recent studies [94, 106–109] have shown that when comparing families of metallophthalocyanines, linear correlations are obtained when plotting log k or log I (rate constant or current at fixed potential) vs. the M(III)/(II) formal potential, as illustrated in Fig. 7.20. First-order rate constants were calculated as k = I/nAFc, where I is the current at a given potential, n is the total number of electrons transferred which is 2 for the peroxide pathway and 4 for the reduction to H₂O, A is the area in cm², F is the Faraday constant, and c is the oxygen concentration in moles per cm³. One linear correlation is obtained for Cr, Mn, and Fe complexes and these metals have configurations d⁴(Cr), d⁵(Mn), and d⁶(Fe). Another linear correlation is obtained for Co complexes, which have a configuration d⁷. An interesting feature in the data of Fig. 7.20 is that the lines are parallel with a slope close to +0.15 V/decade. It is possible that the straight lines in Fig. 7.20 are part of an incomplete volcano correlation. If so, the slope 0.15  V/decade might have a physical meaning as discussed for the oxidation of thiols [110]. The data in Fig. 7.20 strongly suggest that more positive redox potentials will increase the catalytic activity, and it is good that no volcano correlation is obtained because there seems to be room for improving the catalytic activity of phthalocyanines or other macrocyclics. The M(III)/(II) redox potential of some macrocyclics can be shifted in the positive direction with heat treatment, and this could increase the catalytic activity. For example, when iron tetraphenylporphyrin [92], FeTPP, is heat treated, the Fe(III)/(II) redox transition is shifted from 0.2 V vs. RHE for fresh FeTPP to 0.4 V for FeTPP heat treated at 700 °C. Intermediate redox potentials are obtained for heat treatments at intermediate temperatures [92], and the catalytic activity increases with heat treatment, showing that a more positive redox potential of the catalyst favors the O₂ reduction reaction rate.

Fig. 7.20

Plot of log k (at constant potential vs. the M(III)/(II) formal potential of the MN₄ macrocyclic for the reduction of oxygen in 0.2 M NaOH) (from [106])

In a recent study [111], it has been discussed that the changes in the formal potential of the catalyst could explain the high catalytic activity of ORR that has been obtained by heat treatment of metallomacrocyclics and other starting materials. For example, when comparing data obtained with heat-treated catalysts prepared by very different techniques and starting materials, not necessarily involving metallomacrocyclic complexes [112], a correlation of log i (as currents normalized per mass of catalysts) vs. the formal potential of the catalyst gives what could be considered a linear correlation (see Fig. 7.21).

Fig. 7.21

Plot of log i _M (in Ag^–1) at E = 0.8 V vs. the formal potential of the catalyst for ORR [111, 112]

The scatter of the data could be attributed to differences in the porosity of the electrode prepared and also to differences in the number of active sites per unit of surface area. But in spite of this, a clear trend in the figure shows that more positive formal potentials seem to favor the catalytic activity of the different materials. Since the structures of the different catalysts used in the correlation of Fig. 7.21 are not known, no chemical formulas can be given and the numbers correspond exactly to those given in the original paper. More discussion on heat-treated materials will be given on other chapters of this book.

The increase in activity as the M(III)/(II) redox potential of the catalysts is more positive is in contrast to what was previously found in volcano correlations where a maximum activity is observed for intermediate redox potentials not only for the reduction of O₂ but for other reactions such as the oxidation of thiols or of hydrazine [99]. When studying a series of unsubstituted and substituted Mn phthalocyanines, Nyokong and Sehlotho [109] have also found that the catalytic activity of these complexes increases as the Mn(III)/(II) becomes more positive (see Fig. 7.22) with a slope of 0.24 V/decade. It is very likely that the data in Figs. 7.20, 7.21, and 7.22 are part of an incomplete volcano and this is very important because it means that hypothetically that catalysts with higher activities could be prepared by shifting the formal potential to more positive values.

Fig. 7.22

Plot of log i vs. MnIII/MnII redox potential for oxygen reduction in pH 5 buffer. Manganese phthalocyanine (MnPc, 1), manganese tetraaminophthalocyanine (MnTAPc, 2), manganese tetrapentoxy pyrrole phthalocyanine (MnTPePyrPc, 3), manganese tetra phenoxy pyrrole phthalocyanine (MnTPPyrPc, 4), manganese tetra mercaptopyrimidine phthalocyanine (MnTMPyPc, 5), and manganese tetra ethoxy thiophene phthalocyanine (MnTETPc, 6) [109]

A possible explanation for the results in Figs. 7.20 and 7.22 (activity decreases as the driving force of the catalyst increases) is that the electronic coupling between the donor (MPc) and the acceptor (O₂) decreases as the electron-donating capacity of the substituents increases, due to a shift in the energy of the frontier orbitals of the metallophthalocyanine [113–115]. The shift in the energy of the frontier orbital with substituents on cobalt phthalocyanines has been calculated by Schlettwein [113] and Cárdenas-Jirón [114–116] using PM3 and ZINDO/S semiempirical theoretical calculations. There are several approaches to estimate the electronic coupling matrix elements between the donor and the acceptor in electron transfer reactions. One of them considers the energy difference between the LUMO (lowest unoccupied molecular orbital) of the electron acceptor and the HOMO (highest occupied molecular orbital) of the electron donor [117] but this requires to know the distance that separates the donor from the acceptor. This is not simple for an inner-sphere reaction where the M⋯O₂ distance could vary from complex to complex. To avoid this difficulty, another reactivity index can be used to explain the data in Fig. 7.20 and this is the concept of molecular hardness which is a commonly used criterion of reactivity in organic reactions as proposed by Pearson [118, 119]. The hardness η of a single molecule is approximately one-half the energy gap of the HOMO–LUMO, so the larger the gap, the greater the hardness, the more stable the molecule (the harder the molecule, the less its reactivity). The opposite corresponds to a molecule with a narrow HOMO–LUMO gap (soft molecule) situation that will correspond to a very reactive molecule. Now for a donor–acceptor pair, it is more convenient to use the concept of donor–acceptor hardness η _DA which is one-half the difference between the energy of the LUMO of the acceptor (O₂ molecule) and the energy of the HOMO of the donor (metal complex):

$$ {\eta_{\mathrm{ DA}}}=\frac{1}{2}({\varepsilon_{\mathrm{ LUMOacceptor}}}-{\varepsilon_{\mathrm{ HOMOdonor}}}) $$

(7.9)

The donor–acceptor intermolecular hardness can also be described as one-half the difference between the ionization potential of the donor and the electron affinity of the acceptor:

$$ {\eta_{\mathrm{ DA}}}=-\frac{1}{2}({\varepsilon_{\mathrm{ acceptor}}}-{\varepsilon_{\mathrm{ donor}}}) $$

(7.10)

For a gas phase reaction involving the transfer of a single electron, this will be equivalent to the Gibbs free energy of the process ΔG ^∘. For the special case of Co phthalocyanine in its ground state, the HOMO is occupied with a single electron (doublet state) so it corresponds to a single occupied molecular orbital (SOMO). The same is valid for molecular oxygen, which in its ground state has two unpaired electrons in two degenerate π* antibonding orbitals. In this case, the formation of an adduct CoPc-O₂ involves the interaction of two SOMOs and η _DA is given by

$$ {\eta_{\mathrm{ DA}}}=\frac{1}{2}({\varepsilon_{\mathrm{ SOMOacceptor}}}-{\varepsilon_{\mathrm{ SOMOdonor}}}) $$

(7.11)

Figure 7.23 illustrates the calculated energy levels of the SOMOs of the different cobalt phthalocyanines with respect to the SOMO of oxygen using PM3.

Fig. 7.23

(a) Relative energies of frontier orbitals of dioxygen and substituted Co phthalocyanines. For simplicity, only one electron is shown on the SOMO of the CoPcs [115]. (b) Plot of log k (at constant potential) vs. the donor–acceptor intermolecular hardness for the different O₂–CoPc pairs (reproduced from [114])

Electron-withdrawing substituents (sulfonate, fluoro) on the phthalocyanine ring stabilize the SOMO and the opposite is true for electron-donating groups (methoxy and neopentoxy). So electron-withdrawing groups, even though they decrease the electron density on the cobalt (more positive redox potential), also decrease the gap between the energy of the SOMO of the phthalocyanine and the energy of the SOMO of oxygen. The bottom of Fig. 7.23 shows that log k for O₂ reduction increases as the chemical hardness of the system decreases or as the softness of the system increases (more reactivity). The trend in reactivity is exactly the same as that illustrated in Fig. 7.20. It can be concluded then that hardness could be used as a criterion for reactivity of these systems when comparing complexes that bear the same structure and could explain why, for example, perfluorinated phthalocyanine, which has the most positive redox potential (the most oxidant), is the best catalyst for O₂ reduction in alkaline media in the series of cobalt phthalocyanines examined.

Quantum theories of elementary heterogeneous electron transfer (ET) reactions in polar media have been extended to reactions which proceed through active intermediate electronic surface band states or bands. On the basis of this theoretical framework, Ulstrup [120] has interpreted experimental data obtained for O₂ reduction catalyzed by metal phthalocyanines.

When comparing activities of complexes by plotting constant potential vs. redox potential of the catalyst, linear correlations are also obtained (see Fig. 7.24) and this was predicted theoretically by the work of Ulstrup [120]. The slope of the lines in Fig. 7.24 is less than one, which was also predicted by Ulstrup and is attributed to the excitation of intramolecular modes of relatively low frequencies in the cathodic range. The data shown in Fig. 7.24 is essentially similar to that shown in Fig. 7.20 but the comparison was made at constant current. So essentially, the graph of Fig. 7.20 is a plot of driving force vs. driving force. This carries the assumption that the M(III)/(II) redox potential provides a measure of the driving force of the catalysts. Since catalysts produced by heat treatment, using MN₄ metal macrocyclics as the starting materials, or other ingredients like metal salts and nitrogen-containing compounds show very high activities and stabilities, especially in acidic media, it is interesting to compare activity parameters such as those in Fig. 7.20, that is, potential at constant current vs. the formal potential of the catalyst for this type of catalysts.

Fig. 7.24

Plot of log E (at constant current) vs. the M(III)/(II) formal potential of the MN₄ macrocyclic for the reduction of oxygen in 0.2  M NaOH (from [106])

A correlation of this sort is illustrated in Fig. 7.25 [111] and includes catalysts reported by several groups in a joint article [112]. It is interesting to note that in spite of the scatter of the data, it seems that a linear correlation does exist, similar to that illustrated in Fig. 7.20. The slope is 1.2, which is not too far from unity as predicted by Ulstrup. If one compares the data in Fig. 7.21 with that in Fig. 7.25, the scatter of the data in Fig. 7.20 is much lower than that in Fig. 7.25 because it was obtained with graphite electrodes modified with monolayers of MN₄ macrocyclics. So the amount of active sites per unit of real area is very similar for all cases, which might not be true for the data in Fig. 7.25, in spite of the normalization of the current for the mass of catalysts per unit of geometric area. However, a trend is observed and again suggests that more positive formal potentials of the catalyst seem to favor the ORR catalytic process. Finally, it is important to point out that the formal potential plotted in Fig. 7.25 does not necessarily involve the M(III)/(II) but might involve metal-free redox functionalities generated after heat treatment as discussed in [112].

Fig. 7.25

Plot of log E (at constant current) vs. the formal potential of catalysts obtained by heat treatment for the reduction of oxygen in 0.05 M H₂SO₄ at 25 °C (reproduced by permission of the Electrochemical Soc [111], data from [112]). The labeling of catalysts is the same as that used in the original reference [112]

Not all metals of the first transition series exhibit the M(III)/(II) processes, so if one compares macrocyclics of different metals, it is convenient to use another parameter, for example, the number of d electrons in the metal as shown in Fig. 7.26. In this figure, since different Tafel slopes are obtained for the different catalysts, it is not simple to compare activities as current at constant potential. So instead, as a criterion of activity, potential at constant currents is used.

Fig. 7.26

Volcano plot for the electrocatalytic activity of different M-tetrasulfonated phthalocyanines adsorbed on graphite for O₂ reduction in 0.1 m NaOH, as a function of the number of d electrons in the metal (from [15])

Figure 7.26 clearly shows that Fe derivatives exhibit the highest activity, followed by Mn and Co and also illustrates a common observation in catalysis that metals with nearly half-filled d-energy levels exhibit the highest activity. So a redox type of mechanism does not operate for metals that do not exhibit the M(III)/(II) transition in the potential window examined for O₂ reduction. This is the case for Ni, Cu, and Zn phthalocyanines. It is important to point out that in order to have catalytic activity, the frontier orbital of the MN₄ needs to have some d character as illustrated in Fig. 7.16. The catalysts with higher activity included in Fig. 7.24 (Cr, Mn, Fe, Co) have frontier orbitals with d character, whereas in Ni and Cu phthalocyanines, the frontier orbitals have more ligand character [121]. This is illustrated in Fig. 7.27 [108] that compares the frontier orbitals of CoPc and CuPc. CoPc shows a well-defined dz² orbital sticking out of the plane of the phthalocyanine, whereas CuPc does not and shows low activity for O₂ reduction.

Fig. 7.27

Illustration of the frontier orbitals involved in the interaction of cobalt phthalocyanine with O₂ (reprinted from [108])

There is experimental evidence to support this using tunneling electron microscopy. It has been shown a strong d orbital dependence on the images of metal phthalocyanines (see Fig. 7.28).

Fig. 7.28

STM surface plot image of CoPc and CuPc coadsorbed on the (111) plane of Au [122]

Unlike copper phthalocyanine where the metal appears as a hole in the molecular image, cobalt phthalocyanine shows the highest point in the molecular image [122]. The benzene regions of CoPc and CuPc show the same height. So, essentially the data using images generated using electron tunneling microscopy is in agreement with the images generated by theoretical calculations.

Even though most authors agree that the M(II) state is the active site for O₂ reduction [15, 20, 65–67, 70, 72, 88, 89, 97–99, 102, 105, 123, 124 ] for FePc (iron phthalocyanine) and FeNPc (iron naphthalocyanine) [125–127], it has been proposed that Fe(I) could also play a role in the electrocatalytic process. This was based on electroreflectance experiments that indicated that Fe(I) interacts with O₂ whereas Fe(II) does not. However, many authors have shown experimental evidence that O₂ reduction commences at potentials much more positive than those corresponding to the Fe(II)/(I) couple [14, 82, 97, 99]. On the contrary, the reduction currents are observed at potentials close to the potential of the Fe(III)/(II) couple, so it seems unlikely that Fe(I) could be the active site. Worse, as shown from rotating ring-disk experiments, Fe(I) only favors the two-electron reduction in contrast to Fe(II) [82, 88].

The catalytic activity can also depend on the amount of metal complex present on the electrode surface. In general, the amount of catalyst present on the surface is evaluated from cyclic voltammograms, measuring the electrical charge under reversible peaks. This carries the assumption that all adsorbed catalyst gives an electrochemical signal. This might not necessarily be true and there could be a fraction of complexes present on the surface that are electrochemically silent. It is assumed that the “electroactive” adsorbed species are also active for the reduction of O₂.

It has been found that the O₂ reduction currents are directly proportional to the amount of catalyst present [21, 128], when the catalysts is adsorbed on the electrode surface indicating that the reaction is first order in the surface concentration of catalyst. This is not true for cases where the catalyst is incorporated to the surface by vapor deposition or when the catalyst is deposited from solutions and the solvent is completely evaporated [21]. An explanation for these different observations is that when the catalyst is deposited by vapor deposition or from complete evaporation of solutions, multilayers are formed, and the metal active centers are not all completely accessible to the O₂ molecule. Another example of this type of behavior is the case for polymerized multilayers of cobalt tetraaminophthalocyanines, where it has been demonstrated that only the outermost layer is active for the reduction of O₂ [84, 85]. Scherson et al. [129] have reported that when (FeTMPP)₂O is deposited on a porous support, only 30 % of the amount deposited is found to be electrochemically active. Anson et al. [130] have found that for the case of CoPc(CN)₁₆ and CoPcF₁₆ that were deposited from solutions where the solvent was completely evaporated, again it was found that only 30 % of the amount deposited was electrochemically active. It was concluded that only those molecules that directly exposed to the electrolyte and to the incoming O₂ molecules and at the same time are in electric contact to the electrode are active for the reduction of O₂. These results are not surprising since when multilayers are deposited, not all are necessarily in electrical contact with the electrode, which is not the case for adsorbed layers, where molecules are probably lying flat on the electrode surface, interacting with the π system of the graphitic planes. Van der Putten et al. [131] have observed catalytic activity with vacuum-deposited layers in spite of the fact that these layers are electrochemically silent. As pointed out before when multilayers of metal phthalocyanines are deposited on an electrode surface, only the outermost layer is active for the reduction of O₂ [131] and this is also true for other electrochemical reactions. This shows that multilayers of phthalocyanines or polymerized multilayers of phthalocyanines are rather compact and the inner layers are not accessible to O₂ molecules [85].

7.3.1.2 N₄-Metallomacrocyclic Catalysts for One-Electron Reduction of O₂

An electrode surface that has no active sites should only promote the outer-sphere one-electron reduction of dioxygen. An example of such surface is a defect-free electrode surface that exposes the basal plane of graphite [124]. On the basal plane, all carbon atoms are fully coordinated so they cannot bind an oncoming molecule like O₂. On electrodes modified with adsorbed catalysts, to the best of our knowledge, there is only one report that shows evidence for the one-electron reduction of O₂ in alkaline media. A reversible one-electron reduction of O₂ to produce the stable superoxide ion was observed in an aqueous solution of 1 M NaOH. This electroreduction of O₂ was catalyzed by Cobalt(II) 1,2,3,4, 8,9,10,11, 15,16,17,18,22,23,24,25-hexadecafluoro-29H,31H-phthalocyanine (abbreviated as Co^IIPcF₁₆) adsorbed on a graphite electrode [132]. What is curious is that Co^IIPcF₁₆ is the most active catalyst for the two-electron reduction of O₂ in the correlations shown in Figs. 7.17, 7.19, and 7.20, so it is very surprising that it can promote the one-electron reduction of O₂. The OH⁻ concentration has a very strong effect on the reduction process. Higher OH⁻ concentrations could stabilize the superoxide ion [124] but 1 M NaOH might not be concentrated enough to achieve this purpose so it could be interetsing to check this experiments by using electron paramagnetic resonance (EPR) techniques to detect any superoxide formation.

7.3.1.3 N₄-Metallomacrocyclic Catalysts for Two- and Four-Electron Reduction of O₂

Most mononuclear Co macrocyclics catalyze the reduction of dioxygen via two electrons to give peroxide [28, 82, 97, 99, 123]. The activity of Fe phthalocyanines in general is higher than those of Co phthalocyanines and the opposite is true for porphyrins, which reveals the importance of the nature of the ligand in determining the catalytic activity [96]. The opposite is true for heat-treated materials [133]. Cobalt complexes are more stable than iron complexes and this trend is maintained after heat treatment [72, 105]. However, iron complexes tend to promote the four-electron reduction of dioxygen and this will be discussed further on.

Lamy et al. [134] and van der Putten et al. [135] conducted spectroscopic investigations of polymer-modified electrodes containing CoTSPc using UV–visible differential reflectance spectroscopy and were able to identify Co(III)/Co(II) transition when varying the electrode potential. They used electron spin resonance on the Ppy (polypyrrole) and Ppy-CoTSPc electrodes in deoxygenated and oxygen-saturated solutions. It was shown that the Co(III)TSPc species is effective in the electroreduction of oxygen and that this species is more stable in oxygen-saturated medium than in deoxygenated medium because of its stabilization under the following form: Co(III)–O₂. In the case of Ppy-CoTSPc film, the polypyrrole matrix undergoes strong interactions with oxygen species and, most likely, with hydrogen peroxide.

Phosphoric acid is one of the electrolytes used in fuel cells although very few studies have focused attention on the activity of metallomacrocyclic complexes in this electrolyte. Vasudevan and Phougat [136] investigated the electrocatalytic activity of cobalt phthalocyanine monomers and polymers with imido and carboxylic group ends. The complexes were mixed with carbon powder and polyethylene powder. The activity of the monomeric compounds was found to be higher than that of polymeric compounds.

However, in recent work, Ponce et al. [137] have shown that when Fe phthalocyanine is anchored on gold via self-assembled axial ligands, the electrocatalytic activity for ORR increases (see Fig. 7.29). The increase in activity seems to be associated to the electron-withdrawing effect of the axial ligand. This will shift the Fe(III)/(II) formal potential in the positive direction which favors the catalysis as shown in Fig. 7.24.

Fig. 7.29

Illustration of the catalytic action of FePc bound to gold via a self-assembled monolayer of axial ligand bound to Au [137]

When the catalytic activity of graphite modified with mixtures of different proportions of FeTSPc and CoTSPc was tested for ORR, it was found that the catalysts acted independently, that is, the amount of peroxide generated was directly proportional to the fraction of CoTSPc present on the electrode surface. FeTSPc did not promote the decomposition or reduction of peroxide generated on sites occupied by CoTSPc. However, the possibility for the Fe centers to form hydrogen peroxide and promote its decomposition cannot be ruled out completely since Fe(II) sites are known for their catalase activity [138]. Indeed for some metal complexes van Veen et al. have found that their catalytic activity for peroxide decomposition is directly proportional to its activity for O₂ reduction [139].

Again, as observed with FeTSPc and FePc [77, 87, 140, 141], a prewave is observed for the reduction currents and corresponding to the direct four-electron reduction. In contrast to FeTSPc or FePc, production of peroxide was attributed to reduction of the ligand and not reduction of the metal (Fe(I) formation). For this particular catalyst, it was suggested that dioxygen can bind to the Fe center and to a highly electronegative nitrogen in the ring, which will avoid the desorption of peroxide before it is reduced. This dual-site mechanism would aid charge injection via backbonding from the macrocycle into antibonding orbitals of O₂ or bound peroxide causing the destabilization and further rupture of the O–O bond [142]. At more negative potentials at which reduction of the ligand takes place, this mechanism becomes inoperative, the O₂ molecule only binds to the Fe center, and peroxide can desorb into the solution. In a study involving heat-treated FeTPP [143] and deposited by thick layers on glassy carbon, it was found that the amount of hydrogen peroxide decomposed, compared to the amount of oxygen and hydrogen peroxide reduced, was so small that chemical decompositions were ruled out.

Van den Brink et al. [141] when using vacuum-deposited layers of FePc have found that when examining the catalytic activity of these layers, the first reduction wave scan was very different from subsequent scans, indicating that some reorganization of the deposited layers took place since this phenomenon is not observed on adsorbed layers of FePc. The effect of irreversible changes of FePc when treated under potential load with oxygen is only observed using vacuum-deposited multilayers [125]. Léger et al. pointed out that the structure of FePc films influences their electrocatalytic activity.

XRD studies have shown that non-heat-treated FePc is under the α-phase whereas heat-treated FePc is under the β-phase. Surprisingly, the α-phase shows higher activity than the β-phase (see Fig. 7.30). These authors have also shown using electrochemical quartz crystal microbalance (EQCM) that α-phase FePc probably forms μ-oxo dimers at potentials higher than 700 mV vs. RHE. These μ-oxo dimmers are reduced at the same potential than the monomer of α-phase FePc [142].

Fig. 7.30

Polarization curves for the oxygen reduction on a α-FePc/C (m) and a β-FePc (d) disk electrode recorded at 2,500 rpm in O₂-saturated 0.5 M H₂SO₄ electrolyte (T = 20 °C, v = 5 mV s^–1) [142]

Theoretical studies performed by Anderson and Sidik [144] using spin-unrestricted hybrid gradient-corrected density functional calculations have predicted that Fe(II) is the active site for four-electron reduction of oxygen by iron in the N₄Fe systems employed in the calculation, and it may be suggested that the same should be expected for heat-treated iron macrocyclics. The calculations have shown that Fe(II) is favored over Fe(III) because H₂O bonds strongly to the Fe(III) site, preventing O₂ adsorption and water does not bond strongly to Fe(II). On a first step, –OOH bonds more strongly to Fe(II) than to Fe(III), which results in a calculated more reversible potential for its formation over Fe(II). Calculations show that subsequent reduction steps have very reversible potentials over both centers (Fe(II) or Fe(III)). Calculations also show a hydrogen bonding interaction between –(OHOH) bonded to Fe(II) and to a nitrogen lone-pair orbital in the N₄ chelate. This interaction prevents peroxide from desorbing as a two-electron reduction product. So essentially these studies show that adsorbed hydrogen peroxide in an intermediate formed from N₄Fe–OOH on Fe(II) sites and can be released into the solution at more negative potentials as found experimentally [14, 28].

Most catalysts investigated usually have Fe and Co as metal centers. However, complexes of other metals have also been studied. For example, CrTSPc and MnTSPc exhibit catalytic activity for ORR [14] and they somehow resemble the behavior of Fe complexes, especially MnTSPc, in the sense that it shows a prewave where O₂ reduction proceeds entirely via four electrons to give water. Peroxide is produced at higher polarizations. The lower activity of Cr and Mn phthalocyanines compared to Fe phthalocyanines can be attributed to their low redox potential, that is, they are easily oxidized [82, 105]. The activity of most macrocyclic metal complexes increases after heat treatment [145]. However, the opposite is observed for manganese complexes probably because the metal is lost from the N₄ structure. So, most work dealing with heat-treated materials has focused on Fe and Co macrocycles. Complexes of Mo can only be used in alkaline solution since they are not stable in acid media. MoNPc is less activity than FeNPc as reported by Magner [146].

Very few authors have studied the effect of temperature [147] on the catalytic activity of phthalocyanines. Baker et al. [147] carried out experiments at different temeperatures in an acidic electrolyte to simulate the environment of an operating PEM fuel cell. They conducted experiments in the temperature range of 20–80 °C and using unsubstituted and substituted Fe phthalocyanines. The surface electrochemical responses of the FePc species were characterized with respect to their surface concentrations and adsorbed surface orientations. Depending on the type of substituent, the adsorption mode could be flat, edge-on, as a dimer, or as an agglomerate, suggesting that the substituent has a strong effect on the FePc species’ adsorption mode. Substitution also has a significant effect on stability. With respect to their electrocatalytic activity, both temperature, substitution, and possibly mode of adsorption can significantly affect the ORR mechanism. For example, the overall electron transfer number observed can change from 1 to 3 depending on the type of substituent and the reaction temperature. Further research is required to determine if this change in n reflects a change in the ORR pathway and/or a decrease in the stability of the adsorbed ORR intermediates. Based on the various approaches found in the literature and the current understanding, a mechanism for the FePc species catalyzed ORR was suggested as follows:

$$ \mathrm{ FeIIIPc}+{e^{-}}\rightleftarrows \mathrm{ FeIIPc} $$

(i)

$$ \mathrm{ FeIIPc} + {{\mathrm{ O}}_2}\rightleftarrows {{\mathrm{ O}}_2}-\mathrm{ FeIIPc} $$

(ii)

$$ {{\mathrm{ O}}_2}-\mathrm{ FeIIPc}+{e^{-}}+\mathrm{ H}+\rightleftarrows \left[ {\mathrm{ H}{{\mathrm{ O}}_2}-\mathrm{ FeIIIPc}} \right]\ (\mathrm{ r}.\mathrm{ d}.\mathrm{ s}) $$

(iii)

$$ \left[ {\mathrm{ H}{{\mathrm{ O}}_2}-\mathrm{ FeIIIPc}} \right]\to \mathrm{ H}{{\mathrm{ O}}_2}{\bullet^{-}}+\mathrm{ FeIIIPc}\ (\mathrm{ hydroperoxyl}\mathrm{ radical}\mathrm{ formation}) $$

(iv)

$$ \left[ {\mathrm{ H}{{\mathrm{ O}}_2}-\mathrm{ FeIIIPc}} \right]+{e^{-}}+1{{\mathrm{ H}}^{+}}\to {{\mathrm{ H}}_2}{{\mathrm{ O}}_2}+\mathrm{ FeIIIPc}\;({{\mathrm{ H}}_2}{{\mathrm{ O}}_2}\;\mathrm{ formation}) $$

(v)

$$ \left[ {\mathrm{ H}{{\mathrm{ O}}_2}-\mathrm{ FeIIIPc}} \right]+3{e^{-}}+3{{\mathrm{ H}}^{+}}\to 2{{\mathrm{ H}}_2}\mathrm{ O}+\mathrm{ FeIIIPc}\ ({{\mathrm{ H}}_2}\mathrm{ O}\ \mathrm{ formation}) $$

(vi)

It is worth mentioning that Wilkinson et al. [148] have also conducted studies at different temperatures using Fe fluoroporphyrin and reported some kinetic parameters such as activation energies. They found that the results were essentially similar to those found with FePc.

7.4 Oxygen Reduction Catalyzed by Nonmacrocyclic Cu Complexes

Copper (I) complexes exhibit catalytic activity for the four-electron (4-e) reduction of O₂ to water. Natural occurring enzymes like Cu-containing fungal laccase reduce O₂ directly to water very efficiently at very positive potentials, not far from the thermodynamic standard potential of the O₂/H₂O couple. These enzymes involve a trinuclear Cu active site [149–153]. For this reason some authors have investigated the catalytic activity of Cu(I) complexes for ORR, in particular Cu phenanthrolines confined on graphite or glassy carbon surfaces [154–169], with the aim of achieving the total reduction of O₂ via the transfer of four-electrons.

For example, copper(I) 1,10-phenanthroline, Cu(phenP), reduces O₂ almost entirely via the transfer of four electrons and four protons to give water [156, 157, 170]. This is quite interesting since, similarly to what is observed with Fe phthalocyanines [99, 171], the four-electron reduction of O₂ is promoted by single-site catalysts. The O₂ molecule cannot interact simultaneously with two Cu active sites. If this were the case, the order of the reaction in surface concentration would equal to two. Anson et al. [156, 157, 170] checked this and found that the reaction is first order in Cu coverage, suggesting of a mononuclear Cu site as the active catalyst. As mentioned above, Cu-containing fungal laccase enzymes promote the reduction of molecular oxygen directly to water at a trinuclear Cu active site at very low overpotentials [149, 150]. Using specially designed laccase-modified electrodes, Heller et al. [172–174] have measured current densities of 5 mA/cm² at overpotentials (vs. the O₂/H₂O couple) as low as −0.070 V at pH 5. As discussed by Chidsey et al. [168], this current density corresponds to a turnover rate of 2.1 O₂ reduced per laccase s^–1, or 0.7 O₂ per Cu s^–1. In comparison, as pointed out by Chidsey et al. [168], Pt nanoparticle catalysts reduce O₂ with rates of 2.5 O₂ per active surface Pt s^–1 or 0.25 O₂ per total Pt s^–1, at an overpotential of −0.350 V [168, 175]. This normalization allows the comparison of atom efficiency of the catalysts regardless of catalyst loading [166]. With this evidence it is possible to design and prepare Cu complexes that can mimic enzymatic systems using mononuclear Cu complexes.

Anson et al. found that 1,10-phenanthroline (phen) complexes of Cu, adsorbed on the edge-plane orientation of graphite, catalyze the four-electron reduction of O₂ to H₂O [156–159, 170]. They proposed a mechanism for the electrocatalytic four-electron reduction of O₂. Using complexes having three different phen ligands (phen, 2,9-Me2-phen, 5-Cl-phen), they found that reduction of O₂ occurs at potentials negative to the formal Cu (II/I) redox potential of the respective complex strongly suggesting that the active species is the Cu(I) complex [158]. They conducted their studies in the presence of buffers containing acetate, phosphate, and borate. However, the mechanism proposed did not involve the coordination of these anions to the Cu center, even though these anions can affect the potentials and electrocatalytic activity of similar complexes [174]. Chidsey et al. [168] conducted a systematic study of the catalytic activity of several Cuphen complexes for ORR in the presence of acetate. They found that the reduction of Cu(II) on the adsorbed complexes occurs with the concerted dissociation of an acetate ion as [phenCu(II)AcO^–]_ad + e ^– > [phenCu(I)]_ad + AcO^– and that this can happen with any coordinating anion. They found that catalytic currents are observed at potentials very close to the Cu(II)/(I) formal potential of the adsorbed Cu phen complex. This is illustrated in Fig. 7.31. They found that the presence of the anion is crucial in the stability of the catalyst since irreversible degradation of the catalyst is observed if an O₂ reduction experiment is conducted in the absence of acetic acid. The effect of the E ^∘′, the Cu(II)/(I) formal of the catalyst on the ORR activity was studied by investigating some complexes with electron-withdrawing groups located on the phenanthroline ligand as illustrated in Fig. 7.31. As expected, these groups, according to their electron-withdrawing power, shift the Cu(II)/(I) formal potential to more positive values, as seen by the shift of the reversible waves (dotted lines) in Fig. 7.31 assigned to the Cu(II)/(I) transition.

Fig. 7.31

Cyclic voltammograms of edge-plane graphite modified with adsorbed Cu complexes of (a) phen, (b) 3,8-(CO₂Et)₂-4,7-Cl₂-phen, (c) 2,9-Me₂-phen, and (d) 2,9-Me₂-5-NO₂-phen, in N₂-purged (dotted lines) and air-saturated (continuous lines) solutions of 0.1 M NaClO₄, 0.020 M NaAcO, 0.02 M AcOH, pH 4.8 at 0.1 V s^–1 as reported by Chidsey et al. [168]

The shift to more positive potential of the Cu(II)/(I) formal potential has a positive effect in the catalytic activity of the complexes since the ORR waves (solid lines) also shift in the same direction, that is, to lower ORR overpotentials. Figure 7.32 illustrates the effect of the formal potential on the catalytic currents. The currents were measured at the formal potential of each catalyst, not at a constant potential, and they are normalized for the amount of catalyst present on the surface (the amount of catalyst is proportional to the electrical charge under the reversible peaks, dashed lines, in Fig. 7.31). It could be interesting to compare the currents at constant electrode potential vs. the formal potentials since it would be likely that a similar correlation to that shown in Fig. 7.20 illustrated for ORR on metallophthalocyanines vs. the M(III)/(II) formal potential, judging from the data in Fig. 7.31.

Fig. 7.32

Normalized kinetic currents for Cu complexes adsorbed on the edge-plane of graphite, measured at E ^∘′ (circles) and at −0.150 V vs. NHE (squares) plotted vs. E ^∘′ for each Cu complex. The dotted line is the expected behavior for the kinetic current measured at E _1/2. All data measured in 0.020 M NaAcO and 0.020 M AcOH pH 4.8, 0.1 M NaClO₄ except 2,9-Me₂-phen measured in 0.020 M NaAcO and 0.072 M AcOH pH 4.2, 0.1 M NaClO₄

Chidsey et al. [169] have also studied the catalytic activity of a Cu(I) complex of 3-ethynyl-phenanthroline covalently attached to an azide-modified glassy carbon electrode. This catalyst promotes the four-electron reduction of O₂ at pH 4.8 (acetate buffer). In contrast to what is observed by Anson et al. [156–158, 170] using Cu phenanthroline, Chidsey et al. found a second-order dependence of the rates on Cu coverage at moderate overpotentials. This suggests that O₂ interacts with two Cu centers simultaneously, forming a bridge structure, as illustrated in Fig. 7.33.

Fig. 7.33

Binding of dioxygen to two proximal Cu centers of 3-ethynyl-phenanthroline covalently attached to an azide-modified glassy carbon electrode according to Chidsey et al. [169]

Copper(I) complexes confined on electrode surfaces are promising catalyst for the four-electron ORR and seem to require the presence of coordinating anions like acetate to stabilize the adsorbed. This is a much less explored field of nonprecious metal catalysts compared to MN₄ macrocyclics for ORR and might be interesting to mimic laccase-like structures as this particular enzyme catalyzes the reduction of O₂ at very low overpotentials [151].

7.5 Conclusions

It can be concluded that in spite of the rather large amount of work published in the literature, there are still many questions that remain unsolved about the electrocatalytic reduction of O₂ mediated by N₄-metallomacrocyclics confined on electrode surfaces. Improving the activity of these complexes beyond the present state-of-the-art catalysts will require development of rigorous qualitative and quantitative structure–activity relationships (QSAR) [176] for judicious tailoring of activity. Some useful trends do exist. For example, it is now well established that Fe and Co macrocyclic complexes are by far the best catalysts for oxygen reduction even though some studies have demonstrated that cofacial Ir complexes are also very active. These complexes are characterized by exhibiting a reversible redox transition involving the M(III)/(II) couple. Some authors have found volcano-shaped correlations between activity (measured as current at constant potential or potential at constant current) vs. M(III)/(II) formal potential of the catalyst suggesting that an optimal M(III)/(II) redox potential does exist for maximum activity. However, other authors have found only linear correlations between activity and M(III)/(II) redox potential, which indicates that the more positive the redox potential, the higher the activity. In the latter correlations, the activity decreases with increasing the driving force of the catalyst. This finding is important since a priori one would expect that the more negative the M(III)/(II) formal potential, the higher the activity, since this could favor the partial reduction of O₂ upon interacting with the metal center, that is, M(III)-O₂-. It is also possible that the linear correlations found are part of an incomplete volcano. However, in these correlations it was found that Cr, Mn, Fe, and Co complexes, which exhibit a M(III)/(II) transition, give rise to two separated correlations or families of compounds. This is a reflection of the fact that the reduction wave for O₂ reduction on, for example, Mn and Fe complexes starts at a potential very close to the M(III)/(II) formal potential of the catalyst. In contrast, for Co macrocyclics, the reduction wave starts at potentials far more negative than the Co(III)/(II) formal potentials. The proximity of the O₂ reduction wave to the M(III)/(II) for some complexes is also reflected in the observation that a direct four-electron reduction process operates, as observed for Fe and Mn complexes. For most monomeric or monolayers of Co complexes, the onset for O₂ reduction is far removed from the Co(III)/(II) transition and only the peroxide pathway is observed.

Biomimetic catalyst design schemes have been very successful in tailoring the properties of catalysts, for example, as demonstrated for functional heme/Cu analogs and bimetallic cofacial diporphyrins. The knowledge gained thus far provides a promising leeway for the design of efficient N₄-metallomacrocyclic based catalysts for four-electron reduction of oxygen.

Despite Ir- and Ru-based N₄-metallomacrocyclic complexes reporting some of the best activities in acidic and alkaline media, respectively, being rare metals, their advantageous superiority would be offset by their high cost. Nonetheless, these complexes should serve as appropriate models, such that theoretical and experimental knowledge gained from studying them may serve to tailor the synthesis of improved catalysts. Particularly, the energetic aspects that furnish some monomeric complexes the unique ability to reduce oxygen directly to water as opposed to other monomeric complexes warrant detailed investigation.

The relative backdonating power of metal ions attached to the pendant groups of the multinuclear oxygen reduction catalysts discussed in this work is the fundamental parameter for modulating the ORR activity of these complexes. It is therefore possible that the electronic effects introduced due to π backdonation by Ru(II) and Os(II) may also be fulfilled by other groups. This rationale constitutes the prospect for modulation of the ORR activity of these complexes.

The stability of metalloporphyrins and metallophthalocyanines alike is far from satisfactory for any practical application. Whereas there is irrefutable evidence for improved activity upon heat treatment of the N₄-metallomacrocyclic complexes, the loss of their structural merit is obvious. The fact that the performance of nonprecious metal catalysts synthesized by heat treatment of inexpensive nitrogen, carbon, and metal precursors rivals that of heat-treated N₄-metallomacrocyclic complexes, which are relatively more expensive, renders the latter to be noncompetitive precursors [177]. On the other hand, improving the stability and activity of N₄-metallomacrocyclic complexes while conserving their molecular integrity is a difficult task, and indeed very little progress has been reported in line with this.

New material design approaches including exploitation of the synergetic benefits reported for some catalyst supports, for example, carbon nanotubes, graphene, nitrogen-doped carbons, and titanium oxide [178, 179] and other conductive mesoporous materials with high surface areas, among others, might help to solve some of the current challenges. The multicomponent and multifunctional approach [78, 180, 181], where the individual components of a composite catalyst perform specific reactions to ensure complete reduction of oxygen, is worthwhile engendering.

Finally, it is worth mentioning that one of the advantages of MN₄ macrocyclics over Pt catalysts is their tolerance to methanol crossover [14], which is a serious problem in methanol-air fuel cells [83]. For example, Léger et al. have shown that FePc is highly tolerant to methanol [182]. The same is true for CoPc [183]. When methanol crossover from the anode through the electrolytic membrane to the cathode occurs, electroreduction of dioxygen and electrooxidation of methanol occur simultaneously, which is detrimental to the overall performance of the fuel cell since the fuel efficiency decreases and so does the power output. In general MN₄ macrocyclics are poor catalysts for the oxidation of methanol so this reaction should not occur and this would avoid the problems mentioned above.

Cu(I) complexes that somehow mimic natural-occuring laccase represent an interesting class of catalysts since they catalyze the four-electron reduction of O₂ directly to water at rather low overpotentials at pH 4.8 but they lack long-term stability. However, it might be possible to solve this problem in the future.