In this chapter, first, we hope to bring to readers some basic electrochemical fundamentals of solution-phase voltammetry and then show how these simple equations may be applied to characterize specifically porphyrinoids electrochemically. Each example was chosen to demonstrate a different electrochemical aspect. The focus then develops into changes brought about when porphyrinoids are adsorbed on the surface of an electrode first by nonspecific absorption and then by specific adsorption. When in solution, the analyte can diffuse to and from the electrode, and electron transfer between porphyrinoid and electrode is a relatively slow outer-sphere process. Provided the rate of electron transfer is still much faster than the rate of diffusion, the porphyrinoid continues to exhibit electrochemically reversible or quasi-reversible behavior. However, when specifically adsorbed, either by multiple secondary forces such as π–π stacking or by formal bond formation, diffusion is not anymore possible. Electron transfer takes place by an inner-sphere mechanism that is much faster than outer-sphere electron transfer. Measured electrochemical responses also differ. For example, for specifically adsorbed porphyrinoids, peak currents are linearly related to scan rates, while for dissolved porphyrinoids, peak currents are linearly related to the square root of scan rate.

Specific adsorption is a key aspect in electrocatalysis, and we then spontaneously digress into some simplified fundamental aspects of heterogeneous electrocatalysis. Electrocatalytic materials are researched in a variety of fields including fuel cells,¹  chemical synthesis,²  energy storage,³  and sensors.^4,5  The chapter concludes with examples in which porphyrinoids have been used for electrocatalysis.

The examples discussed here have never been intended as an exhaustive review of electrocatalysis by porphyrinoids. Rather, they were chosen to demonstrate how electrocatalytic activity changes due to porphyrinoid structural changes. That said, examples other than those discussed, have also been cited for the benefit of readers. We hope this work will assist many beginning and experienced researchers in the field of solution-phase electrochemistry and electrocatalysis.

Electrochemical properties of a compound are conveniently probed in solution by techniques such as cyclic voltammetry (CV), square wave voltammetry (SWV), and linear sweep voltammetry (LSV).⁶  When two closely overlapping redox processes of a compound follow each other, SWV can make it a little easier to resolve the potentials at which they occur. Following the procedure developed by Taube,⁷  for electrochemical reversible redox processes, the limit of potential resolution of two separate redox processes (differences in formal reduction potential, ) by both CV and SWV is in the vicinity of 50 mV. However, to accurately obtain them by CV using the Taube method is much more cumbersome when E_p values cannot be measured directly. The key factor is identifying peak potentials, E_p, of both peaks accurately. Experimental measurements and theoretical simulations show E_p measurements can still be done easily with SWV if ΔE°′ = 80 mV, but in CV, E_p identification struggles even at ΔE°′ = 100 mV. Only when both peaks can be measured directly with CV (at ΔE°′ = ca. 250 mV) will direct CV-measured E°′ values be accurate (within 1 mV of the correct potential).⁷  An LSV is a CV performed at a very slow scan rate, 1 or 2 mV s⁻¹. It is the electrochemical equivalent of NMR integration giving the relative number of electrons that flow in each redox process in a multi-redox system.

Electrochemical reversible redox processes are characterized in CV by three equations.^8,9 

In eqn (1.1)–(1.3), n is the number of electrons transferred in the balanced equation of the redox process, i_pa is the peak anodic current at the peak anodic potential (E_pa), i_pc is the peak cathodic current at the peak cathodic potential (E_pc), and E°′ is the formal reduction potential of the reversible redox process. For eqn (1.2), i_pa/i_pc is meant to be the “reverse scan current” divided by the “forward scan current.” If this is not so, the ratio should be inverted to i_pc/i_pa = 1. Care must be taken when choosing a solvent and supporting electrolyte for a particular system because popular solvents like acetonitrile may complex, or at the very least associate, with a metal, and often-used supporting electrolytes like [ⁿBu₄N][PF₆] may form ion pairs of the type (cation)⁺⋯⁻(PF₆). For this reason, many CV/SWV/LSV measurements are now conducted in CH₂Cl₂ utilizing a super non-ion pair forming supporting electrolyte like [ⁿBu₄N][B(C₆F₅)₄]. Working electrodes are chosen in such a way as to minimize any layer forming (coating) interactions with analytes.

Figure 1.1 shows the general structure of porphyrin (por), phthalocyanines (Pcs), and subphthalocyanines (subPcs). Electrochemically, most porphyrinoids exhibit two ring-based oxidations and four ring-based reductions, but all six redox processes are seldom observed in a solvent potential window.

Scheme 1.1 shows the synthetic route toward metallated tetrabenzoporphyrins, TBPs, tetrabenzomonoazaporphyrins, TBMAPs, tetrabenzodiazaporphyrins, TBDAPs, and tetrabenzotriazaporphyrins, TBTAPs, 1–4.¹⁰  These porphyrinoid types have not been extensively studied.^11,12  The metal-free derivatives showed in CH₂Cl₂/0.1 mol dm⁻³ [ⁿBu₄N][B(C₆H₅)₄] two ring-based oxidations and two ring-based reductions. Some copper complexes also exhibit a Cu^II reduction wave in this potential region (see Figure 1.2). The relationship between E°′ and Σχ_R where Σχ_R is the sum of the Pauling scale group electronegativities of the apex atoms N and CH in 1–4 may also be found in Figure 1.2.¹⁰  Σχ_R is calculated, for example for 2, as follows: Σχ_R(2) = χ_N + 3 · {χ_C + χ_H} = 3.04 + 3(2.55 + 2.20) = 17.29. The larger this number, the more electron-withdrawing the macrocycle becomes, meaning that 4 with three aza groups and Σχ_R = 13.87 is more electron-rich than 2 having only one aza group and Σχ_R = 17.29.

For wave 3, linearity in the relationship between copper complex E°′ and Σχ_R value breaks down. This is consistent with the source of this reduction in 1b (N = 0) and 2b (N = 1) not being the same as for 3b (N = 2) and 4b (N = 3). DFT (density functional theory) calculations showed that for 1b and 2b, the first reduction wave labeled 3 in Figure 1.2 is not a ring-based reduction as it is for 3b and 4b, but a Cu^II reduction.¹⁰ 

The linear relationship between E°′ and Σχ_R (see Figure 1.2) is indicative of good communication within the porphyrinoid macrocycle. Good electronic communication also extends to substituents on the macrocycle and axial ligands of coordinated metals. Figure 1.3 shows the structures of 5,15-diphenylporphyrin, 5; 5-ferrocenyl-10,20-diphenylporphyrin, 6; 5-ferrocenyl-15-phenylporphyrin, 7; and 5,15-diferrocenyl-15-phenylporphyrin, 8 obtained after scrambling of intermediate products of the MacDonald-type 2 + 2 condensation of 5-ferrocenyldipyrromethane, dipyrromethane and benzaldehyde in CH₂Cl₂ at room temperature in the presence of trifluoroacetic acid.¹³  Oxidation was with 2,3-dichloro-5,6-dicyanobenzoquinone, and the trifluoroacetic acid was quenched at the end of the reaction with triethylamine.

The CV, LSVs, and SWVs of 8 as representative examples are also shown in Figure 1.3. Key observations relate first to the LSVs that demonstrate each redox process involving the same number of electrons; here it is representative of a one-electron transfer process. Next, it is clearly easier to identify accurately peak potentials of wave 3 in the SWV than in the CV. Finally, the two ferrocenyl (= Fc) groups of 8 do not exhibit equivalent electrochemical behavior. When the first is oxidized to ferrocenium, the Fc group's Gordy scale group electronegativity¹⁴  is changed from χ_Fc = 1.87 to χ_{Fc +} = 2.82 (i.e., the strongly electron-donating Fc group became strongly electron-withdrawing). This caused the second Fc group to be oxidized at a potential 132 mV larger than the first Fc group. Figure 1.3, bottom right, shows that communication between different meso substituents in 5–8 is good enough that a linear dependency exists for all four redox waves between E°′ of waves 1, 2, 3 (the ferrocenyl redox processes) and 4 and Σχ_R = χ_R1 + χ_R2 + χ_R3 + χ_R4. χ_R1 through χ_R4 can be deduced from the structures of 5–8, and χ_C6H5 = 2.21, χ_C6F5 = 3.01 (9 possess the C₆F₅ group), and χ_H = 2.20.¹⁴  Electronic communication between Fc substituents was also confirmed for 3,3′-diferrocenyl azaBODIPYs (difluoroboryl complex of azodipyrromethene),¹⁵  tetraferrocenylporphyrin (H₂TFcP),¹⁶  corroles,¹⁷  and many others.¹⁸  The H₂TFcP study also highlighted the advantage of using CH₂Cl₂ over acetonitrile, DMF, and THF as solvent (but see advantages under certain conditions of THF as solvent in Section 1.3.4).¹⁶  Nemykin wrote a good review on the magic of compounds obtained as a result of a marriage between metallocenes and porphyrinoids.¹⁹ 

Communication to axial ligands is just as good as to ring substituents. Figure 1.4, top left, shows the structures and voltammograms of subphthalocyanines (subPcs) 10–13.²⁰  SubPcs 10 and 11 have hydrogen atoms on the annulated benzene rings, while 12 and 13 have fluorine atoms. Only one ring-based oxidation, wave A, was observed in the CH₂Cl₂ potential window, but E°′ for this wave for the fluoro complexes 12 and 13 was 395 mV more positive than E°′ of 10 and 11. E°′ of the first reduction wave, wave I, of 13 was offset with 544 mV from that of 11, and the second ring-based reduction of 12 occurred at a potential almost 600 mV more positive than that of 10.

Most striking, the fluorinated complexes 12 and 13 exhibited a third ring base reduction, while compounds 10 and 11 did not show this redox process within the potential window of the solvent.

The axial ligand in 10–13 is either OOC–CHCH–Fc (in 10 and 12) or OOC–CH₂–CH₂–Fc (in 11 or 13). The CHCH moiety allows communication between the subPc core of 10 and 11 and the Fc group, while the CH₂CH₂ atom chain in the axial ligand of 12 and 13 isolates the ferrocenyl group from the subPc core. This isolating effect of the CH₂CH₂ group manifested in the Fc group of 11 and 13 being oxidized at almost the same potential, −0.058 and 0.050 mV versus FcH/FcH⁺, respectively, despite 11 having 12 hydrogen atoms and 13 having 12 fluorine atoms bound to the subPc macrocycle.²⁰ 

The situation is rather different for 10 and 12 where the CHCH group separates the Fc group from the subPc cores. The formal reduction potentials of the Fc group of 10 and 12 (the compound with 12 F atoms) were 0.119 and 0.243 V versus FcH/FcH⁺, respectively. These four Fc E°′ values indicate that axial ligands and substituents on the macrocycle core, when conjugation is possible, influence the redox properties of each other. Communication between the Fc group of ferrocene-containing β-diketonato ligands and the phthalocyaninato macrocycle when coordinated with Zr and Hf Pcs was also observed and discussed.²¹ 

It follows that a careful choice of ring substituents and axial ligands provides a route to fine adjustments of formal reduction potentials, E°′, of redox-active components in porphyrinoid complexes.

Sandwiched (double-decker) and triple-decker porphyrinoids are especially known for lanthanide and actinide complexes.^22–25  Utilizing the example of a non-rare earth cadmium triple-decker phthalocyanine,²⁶  it is shown how the careful choice of phthalocyanine (Pc) ring substituents can limit Pc aggregation (i.e., enhance solubility) and how solvent choice with a large working potential window may allow observation of more than just three or four ring-based porphyrinoid redox processes. The CVs of subPcs 10 and 11 exhibited three ring-based redox processes, and 12 and 13 exhibited four of the expected six redox processes in the solvent (CH₂Cl₂) potential window (see Figure 1.4). To observe all six expected porphyrinoid ring-based redox processes in a particular solvent is unusual. One such example in which it was observed is the cadmium phthalocyaninato triple-decker, 14, for which 17 of the possible 18 redox processes could be observed in THF as solvent (see Figure 1.5).²⁶  The relative number of electrons flowing for each redox process is shown by the LSVs. Especially the SWV in CH₂Cl₂ of waves IV and C and the SWV of waves V1 and V2 in THF demonstrate better potential resolution than the CVs. Notable in Figure 1.5 are the differences in measured anodic potentials in THF and CH₂Cl₂. The free electron pairs on THF make it nucleophilic and capable of interacting or associating with oxidized species having a positive charge. This caused the six anodic waves of 14 to coalesce into two major waves, waves A and B, while in CH₂Cl₂ (which do not interact with positively charged species), the same redox processes were observable in three redox waves, waves A, B, and C. However, the interaction of THF with negatively charged reduced species is very weak or non-existent; hence, the cathodic CV fingerprint of 14 in THF and CH₂Cl₂ is similar. Because THF has a wider cathodic potential window than CH₂Cl₂, the redox process labeled VI of 14 could be detected in THF but not in CH₂Cl₂.²⁶ 

One of the reasons 14 shows such well-defined CV shapes is the good solubility the eight non-peripheral C₆H₁₃ substituents impose onto the porphyrinoid. It was shown that solubility increases substantially (i.e., the tendency to aggregate decreases) in organic solvents as the alkyl substituent chain length (n in C_nH_{2n + 1}) increases. Listed below is how the onset aggregation concentration increases with increasing n where eight C_nH_{2n + 1} groups are bound to the non-peripheral positions of ZnPc.²⁷  {(n; [octa-subst. ZnPc]/μM) = (5; 9), (6; 11), (7; 32), (8; 27), (9; 42), (10; 152), (12; 400)}.

Lowering the tendency to aggregate is much enhanced from n = 10. This effect is less pronounced if the substituents are on the non-peripheral positions.

The above-described CVs have near-ideal shapes and obeyed eqn (1.1)–(1.3) because they were involved in near-ideal, outer-sphere, heterogeneous electron transfer between electrode and porphyrinoid substrate.²⁸  This means that porphyrinoid reactants, intermediates, and products do not adsorb (e.g., via aggregation) on the electrode surface. Heterogeneous electron transfer between electrode and porphyrinoid must occur by electron tunneling across at least a mono layer of solvent separating electrode from porphyrinoid.^28,29  This process is demonstrated in Figure 1.6.

Redox potentials in organic media according to IUPAC should be referenced versus the FcH/FcH⁺ redox couple at 0.00 V.³⁰  However, the potential of the FcH/FcH⁺ couple is dependent on the solvent used. If this redox process overlaps with redox processes of the analyte, decamethylferrocene, Fc*, is often used in its place. The Fc* redox potential relative to ferrocene in different solvents is well described elsewhere,^31–34  but for convenience, some conversion factors are given in Table 1.1.

If, however, the porphyrinoid interacts weakly with the electrode, it may nonspecifically adsorb on the surface of the electrode.^35,36  When a porphyrinoid in the oxidized (or reduced) form is nonspecifically (= weakly) absorbed on an electrode surface, the height of the cathodic peak, i_pc, (or the anodic peak current, i_pc) will increase faster than the height of peaks when the porphyrinoid is involved in normal ideal outer-sphere electron transfer processes because both absorbed and diffusing oxidized porphyrinoid contribute to the current.³⁵  The effects of nonspecific adsorption of a compound on an electrode are demonstrated in Figure 1.7, left, for phthalocyanine 15,³⁷  where the cathodic peak height of reduction half wave III_ads is larger than the anodic peak height (i_pa of the anodic CV half wave III), as well as the cathodic peak heights of waves I and II. This implies that i_pa/i_pc is substantially less than 1. Since i_pc would become larger more quickly with faster scan rates than i_pa³⁵  would, the current ratio i_pa/i_pc would deviate further from unity at faster scan rates. Aggregation probably plays a large role in observation of this effect and could be minimized or completely avoided if the substituents on the peripheral positions had C₁₀ or longer chain lengths than the 2-ethylhexyl chains that were used here (see also Section 1.3.4). Other examples of this effect related to ferrocene-containing porphyrinoids may be found in the literature.^38,39 

However, if the porphyrinoid is strongly absorbed onto the electrode surface, it cannot continue to diffuse away from the electrode, eqn (1.1)–(1.3) are not valid anymore, and ΔE_p becomes zero (see Figure 1.7 right, wave 1). Strong adsorption is normally associated with specific adsorption (or in some cases, even bond formation) of a substrate onto an electrode forming a layer of adsorbate on it. Specific adsorption may be described by the Langmuir, Temkin, or other isotherms.⁴⁰  Electron transfer in such a system is very fast as it proceeds via an inner-sphere mechanism.²⁸  Different potentials are observed for specific adsorbed substrates and dissolved substrates. If some dissolved substrate is still left in the solvent after specific substrate adsorption on the electrode, this will be electrochemically observable as a near-ideal, outer-sphere, heterogeneous electron transfer process between electrode and porphyrinoid substrate [i.e., almost a “normal” CV portion as shown in wave 2 (see Figure 1.7)].^35,36  This figure describes the following scenario.³⁵  First, the porphyrinoid substrate is dissolved in the oxidized form, O_dissolved. A portion of O_dissolved is then reduced (wave 1), and the reduced species R simultaneously adsorbs strongly on the electrode forming a layer of this reduced product (R_adsorbed). This wave 1 reduction occurs at potentials more positive than the diffusion-controlled reduction process at wave 2 because the free energy of adsorption of the reduced form, R_adsorbed, makes simultaneous reduction and adsorption of the dissolved oxidized form, O_dissolved, easier than to generate the reduced form in solution. Wave 1 associated with the half reaction O_dissolved + e⁻ → R_adsorbed is then followed by wave 2, which is associated with the half reaction O_dissolved + e⁻ → R_dissolved. The peak current of wave 1 (i.e., of the specific adsorbed species) increases linearly with scan rate υ, while the peak current of wave 2 increases linearly with υ^1/2. Finally, if the electrode covered with R_adsorbed is then removed from the electrochemical cell, rinsed to remove all solvent containing substrate and electrolyte, and then immersed into a fresh solution of electrolyte only, and if the adsorbed layer is still intact (i.e., it was adsorbed strong enough not to be washed away), a new CV experiment would give only wave 1 in the voltammogram.

A key factor to observe is that electron transfer between electrode and specific adsorbed species follows an inner-sphere pathway that allows for much faster rates of electron transfer than outer-sphere processes. In homogeneous terms, this is called the kinetic advantage, and the ratio of inner-sphere to outer-sphere rate constants, k_inner/k_outer, has recently been shown⁴¹  to be as high as 10^4.81 for electron transfer reactions between a nickel(iii) peptide and Ir^IIICl₆^{3 −} or 10^4.12 between a nickel(iii) peptide and Mo(CN)₈^{4 −}. As shown in Figure 1.8, inner-sphere heterogeneous electron transfer between electrode and substrate occurs while the substrate is in firm, stable contact with the electrode (i.e., specifically adsorbed) either through a formal bond or via very strong secondary forces. This adsorption mode provides the shortest possible distance between electrode and substrate. The overlapping orbitals of substrate and electrode provide a “highway” for electron flow between electrode and substrate. In this scenario, no diffusion of the substrate is possible and ΔE_p = 0.

Closer proximity of the redox center to the electrode surface has a profound influence on the rate of electron transfer. It was shown that the rate of electron transfer between ferrocenyl derivatives, Fc(CH₂)_nS–, bound to the surface of a gold electrode differs with n as follows:^42,43 

The dissolved decamethylferrocene in the CV shown in Figure 1.8 highlights the linear dependence of peak currents against the square root of scan rate for solution-based CVs according to the Randles–Sevcik equation,^8,9  eqn (1.4) and (1.5), where A is the electrode surface area, C is the bulk analyte concentration, F is the Faraday constant, D is the diffusion constant, υ is the scan rate, and n is the number of electrons transferred in the balanced redox reaction.

In contrast, for electrochemical reversible systems, anodic and cathodic waves of surface-bound analytes, in Figure 1.8, the self-assembled monolayer of Fc(CH₂)₄S– on gold, should be symmetrical, and ΔE = E_pa − E_pc should be 0 mV.⁴²  The relationship between peak currents and scan rate is given by eqn (1.6) where A_sur is the electrode surface area, and Γ is the surface coverage.^42,43 

For dissolved species, the slope of an i_p versus υ^1/2 plot is used to obtain the diffusion constant of the species, while the slope of i_p versus υ plots leads to the surface coverage Γ.

A final thought is that some species dissolved in a solvent may adsorb and pollute the electrode to the point where it becomes pacified or useless for electrochemical measurements. But this is not always the case. The above-described specific adsorption effects that lead to inner-sphere electron transfer are, in fact, the bases of electrocatalysis. The remainder of this chapter will focus on simplified fundamental thoughts required for treatment of data of electrocatalytic systems followed by some selected examples of this important heterogeneous process.

Electrocatalysis is just another form of heterogeneous catalysis. It catalyzes a reaction involving electron flow to and from an electrode. Electron flow takes place at the surface of the electrode, but as shown below, the catalyst does not need to be adsorbed on the electrode. It can also be in solution as long as it has access to the electrode via, for example, diffusion. Being involved in “catalytic” processes, the catalyst must also fit in with general principles of heterogeneous/homogeneous catalysis. The most important catalyst characteristics are as follows:

• a. It must be stable.

• b. It must be active.

• c. It must be selective for the process for which it was chosen.

One would like to have the catalytic process operating as fast as possible at potentials as close as possible to the thermodynamic reversible electrode potential. A key equation to evaluate this, is the Butler–Volmer equation, eqn (1.7).⁴⁴ 

In eqn (1.7), i is the total electrode current density of the electrode reaction in units of A m⁻², the first term in brackets is the cathodic component, the second term is the anodic component, i₀ is the exchange current density (A m⁻²), n = number of electrons transferred in the rate-determining step, α_a is the dimensionless anodic transfer coefficient, η is the overpotential of the reaction in volt, F = Faraday constant, R = gas constant, and T = temperature in Kelvin. Note 1 − α_c = α_a where subscripts a and c denote anodic and cathodic, respectively. Also, η = E_measured − E°_eq and E°_eq = E°_red hr − E_ox hr, hr = half reaction, red = reduction, and ox = oxidation. Provided there is no mass transport in the rate-determining step, the overpotential associated with any given current serves only as activation energy. In Figure 1.9(a), eqn (1.7) is used to obtain the green and black curves (when negative currents are found, log i is obtained by using the absolute value) as well as the purple line in Figure 1.9(c). Figure 1.9(a) shows the relationship between log i and overpotential. This particular plot is not symmetrical since α_c was chosen to be 0.4, implying that α_a = 1 − 0.4 = 0.6. If α_a = α_c = 0.5, then the plot is symmetrical. Figure 1.9(c) shows the relationship between i (rather than log i) and overpotential.

If the overpotential, η, is large (η > 120/n mV), eqn (1.7) simplifies to contain the anodic term only. (Remember that 1 − α_c = α_a.)

or in logarithmic form

Eqn (1.8a) was used to obtain the green curve in Figure 1.9(c); eqn (1.8b) generated the linear green part in Figure 1.9(a). In accordance with eqn (1.8b), at η = 0, i₀ may be found from the Y-intercept of a log i versus η plot [Figure 1.9(a)].

Rearrangement of eqn (1.8b) to make η the subject of the formula gives

Eqn (1.11) is the empirical Tafel equation⁴⁴  that dates back to 1905 for anodic (oxidation) reactions. Comparison of eqn (1.10) and (1.11) gives

An empirical Tafel plot utilizing experimental data would give something like Figure 1.9(b). Note that the X and Y axes in Figure 1.9(a) must just be interchanged to fit Figure 1.9(b), and one should use the absolute values of the overpotentials. One can then calculate either n or α from eqn (1.12), and i₀ may be found either from the X-axis intercept or (but less accurately) from the Y-intercept using eqn (1.13).

Similar equations may be derived for cathodic (reduction) processes.

From the Tafel diagram of eqn (1.10) (see Figure 1.9, right) and eqn (1.12), the Tafel slope, b, is only dependent on the variables α and n. The larger b is, the faster the overpotential η increases. Thus, for an electrochemical reaction to obtain a large current at low overpotentials, it should exhibit a small Tafel slope b that translates to a large αn. Figure 1.9(b) demonstrates this: α for the green line is 0.6, and this line has a smaller slope than the black line that has α = 0.4.

Volcano plots [see Figure 1.9(d)] are generated when electrocatalyst activity expressed as log[(i₀ (A′ cm⁻²)] is graphed (normally) against heats of adsorption (ΔG_H in eV or other suitable units) of a reacting molecule that must react with a series of different catalysts following the same electron transfer mechanism.⁴⁵  The reaction conditions under which the reactions were performed and the rate-performing step for each reaction must be the same for all catalysts used.⁴⁶  If ΔG_H is not available from literature sources, other parameters may be used, such as redox potentials [used in Figure 1.9(d)]⁴⁶  or number of d electrons if the catalyst metal differs.⁴⁷  Bond strengths between the active site and adsorbed molecule are also used often.⁴⁸  Any parameter related to the ability of the catalyst to form chemical bonds with reactants, reaction intermediates or products would suffice. Volcano plots are important in catalysis and electrocatalysis since they offer a guide to identify the most active (electro)catalyst for a particular reaction, and they give clues on what structural modifications could be considered to further enhance catalytic activity. The name “volcano” plots are self-explaining. The shape of the graph looks like a volcano mountain.

When a rotating disk electrode setup is available, the Koutecky–Levich equation [eqn (1.14a, b)] may be used to obtain insight into the mechanism of electron transfer.⁴⁹  One may also evaluate half reaction rates occurring at an electrode that is hampered by a combination of sluggish kinetics and mass transport.

In eqn (1.14a), i is the measured current density (A cm⁻²), and i_K is the kinetic current. The first term in eqn (1.14a, b) is the Levich current density, i_lev, which is used to measure the rate of current-limiting chemical reactions. This means that one can also rewrite eqn (1.14a) as in eqn (1.14b). The value of n in eqn (1.14a) is the number of electrons transferred in the overall electrode reaction, F is the Faraday constant (96 485.3 C mol⁻¹), D and C_o* are the diffusion coefficient (cm² s⁻¹) and bulk concentration of the species that undergoes electrocatalysis, ν is the kinematic viscosity (cm² s⁻¹) of the solvent, and ω is the angular rotation speed (rad s⁻¹) of the electrode.

The first step in an experiment is to collect CVs at different rotation rates at the rotating disk electrode while scanning from low to high potential (or from high to low). The measured current should start at zero, rise quickly, and then flatten off as overpotentials get large. The shape of the i–E curve will be sigmoidal.⁵⁰  This is demonstrated in Section 1.6.3 (see Figure 1.16).

A plot of 1/i versus 1/(ω)^0.5 for a number of data points at the same potential should be linear, and the number of electrons flowing at the rate determining step, n, can be calculated from the slope.⁴⁹  Data collected at a large enough overpotential (i.e., from the limiting region in a current potential plot) will pass through the origin in a Koutecky–Levich plot, which implies that there are no kinetic limitations in electron transfer. Only mass transport limits the current. When currents are measured at an overpotential so small that i is not constant with an increase in potential (i.e., i still rises with increasing E in a CV), a Koutecky–Levich plot gives a non-zero intercept of 1/i_K, which indicates a kinetic limitation. This non-zero intercept implies that even at infinite rotation rate (i.e., mass transport is infinite), the rate of the electrode reaction is limited by slow kinetics at the electrode surface. The intercept term 1/i_K is the reciprocal kinetic current in the absence of any mass transport limitations. By measuring the kinetic current at a variety of different overpotentials along the voltammogram, it is possible^49,50  to determine the standard rate constant k_f for the electrochemical half reaction.

With this background the four quantities important for electrocatalyst activity may be summarized as follows:

• a. The exchange current density i₀.

• b. The Tafel slope b (mV dec⁻¹) obtained from a plot of overpotential η against log i. The shape of this graph is also shown in Figure 1.9, and the linear portion satisfies the equation η = a + b log i. b may also be obtained from eqn (1.12) if α is known. Or once b has been determined experimentally, eqn (1.12) may be used to obtain α.

• c. The current density i (in units of mA cm⁻²) at a given overpotential. Current density may be reported in any of three ways:

  i. Amp per geometric area, A cm⁻²_geo,

  ii. Amp per surface area, A cm⁻²_real, and

  iii. Amp per electrochemically active surface area, A cm⁻²_act. Here, a roughness correction must be performed (see literature⁴² ).

  The last option of these three is the best to use because it gives the correct turnover frequency (TOF), but the first option is often used because it is easier to obtain.

• d. The overpotential η_i required to reach a target current density

To illustrate some of the preceding concepts, this section starts with a discussion of nickel hydroxide nanoparticles (= Ni(OH)² NPs) supported on polycrystalline boron-doped diamond electrodes (pBDD) for electrocatalysis of oxidation reactions.⁵¹  In situ generated Ni^{3 +} in Ni^IIIOOH are the active catalytic species in alcohol oxidations because empty d-orbitals or unpaired d-electrons are available for bond formation with redox intermediates, adsorbed species, or electrodes. To deposit Ni(OH)₂ on pBDD electrodes, nitrate ions from the nickel source, Ni(NO₃)₂, were electrochemically reduced at −1.1 V versus Ag/AgCl according to the reaction

The electrochemically generated OH⁻ ions reacted with Ni^{2 +} to form poorly soluble Ni(OH)₂ (K_sp = 5.48 × 10⁻¹⁶, 25 °C), which immediately adsorbed on the electrode. This method is easy to use, is cheap, and provides the capability of controlling Ni(OH)₂ NPs size by choosing appropriate times of electrolysis. Ni(OH)₂ NPs were characterized by XPS, AFM, and field-emission scanning electron microscopy. XPS measurements confirmed that the NPs were void of Ni metal, NiOOH, or NiO species; it consisted only of Ni(OH)₂. AFM measurements showed Ni(OH)₂ NPs aggregates had conelike structures, heights of 250 nm, and base diameters of ca. 200 μm; smaller NPs had a controlled height of 10–20 nm and base size of ca. 33 nm. The amount of deposited Ni(OH)₂ could be controlled and was calculated⁵¹  as an effective surface concentration Γ from eqn (1.16) as ca. 20 ± 4, ca. 140 ± 20, or ca. 420 ± 70 nmol cm⁻² (Q_ox = charge associated with oxidation), depending on electrolysis time. FE-SEM showed that adsorbed particles are distributed uniformly over the pBDD.

Figure 1.10 shows the CV of a pBDD electrode modified with ca. 20 nmol cm⁻² Ni(OH)₂ in water containing 0.1 M KOH. Although the ratio i_pc/i_pa ≠ 1, the charge ratio Q_red/Q_ox ≈ 1 (Q_red = reduction charge and Q_ox = oxidation charge that flowed). The redox process that takes place is

This Ni(OH)₂-modified pBDD electrode was then tested as electrocatalyst in oxidation of glucose, methanol, and ethanol. In the absence of glucose or alcohol, the parasite (background) current was found to be negligible in 0.1 M KOH (inset in Figure 1.10). When 0.5 M ethanol was added to the solution, the Ni(OH)₂ was electrochemically oxidized at ca. 0.22 V versus Ag/AgCl [eqn (1.17)]. The oxidized product, Ni^IIIOOH, then oxidizes alcohols and regenerates Ni(OH)₂. The next catalytic cycle continues when the electrocatalyst Ni^IIIOOH is electrochemically regenerated. At a potential of 0.3 V versus Ag/AgCl, Ni^IIIOOH generation is fast with a large current flow, which is the result of alcohol oxidation driving the catalytic cycle. This potential (0.3 V) is so low that OH⁻ is not oxidized simultaneously with alcohol oxidation. The activity of the Ni(OH)₂ NPs is high. Keeping in mind that the most efficient electrodes were covered with only ca. 2 μg Ni(OH)₂ cm⁻², the normalized catalyst current for ethanol oxidation of 1010 A g⁻¹ is impressive.⁵¹ 

The authors⁵¹  found that the most effective catalyst particles had cone heights of ca. 25 nm. NPs with cone heights of 12 nm are less active than NPs with cone heights of 25 nm. Increasing the NPs cone heights beyond 25 nm again caused a lowering of catalyst efficiency.

This study⁵¹  also demonstrates the properties of good electrode material. The properties must be as follows:

• a. The electrode surface must be such that strong Ni(OH)₂–electrode surface material interaction exists. Here, no covalent bond between pBDD electrode and Ni(OH)₂ is formed, yet the specific adsorption of Ni(OH)₂ onto pBDD led to secondary forces that are strong enough to prevent loss of the NPs under the conditions used.

• b. The NPs itself must exhibit chemical reversibility in the redox process with which it is involved over many cycles while staying adsorbed on the electrode.

• c. The electrode itself must be robust and chemically inert under the conditions used (temperature, solvent type, electrolytes, and other species in the solution to avoid catalyst poisoning by any products).

• d. The potential window in which the electrode operates must also be wide enough to prevent electrochemically induced processes to shorten the active life of the catalyst/support system.

• e. The contribution of the parasitic (background) current from the electrode in the operating potential window should be negligible.

As described,⁵¹  pBDD electrodes are well suited as inert support electrode for electrocatalysis.

The above study made use of a catalyst, Ni(OH)₂, that was not covalently anchored to the pBDD electrode. In general, this is dangerous in that the secondary forces that keep the specific adsorbed particles in place on the electrode surface may be too weak to prevent catalyst particles to fall off from the electrode partially or completely under the conditions of an electrocatalysis experiment. To make sure this does not happen, one would measure the current during electrocatalysis for an extended time at a fixed potential. In the ideal case, no current loss will be observed.

The importance of doing blank studies (i.e., checking whether the electrode itself does not have catalytic activity) cannot be overstressed. Kozub and Compton⁵²  demonstrated how to prevent interpretation of results as an electrocatalytic effect by a potential catalyst adsorbed on the electrode surface when, in fact, the electrode itself behaved as the electrocatalyst. To do this, they observed that cobalt(II) phthalocyanine, CoPc, specifically adsorbed on a suitable electrode has been used as an electrochemical sensor to detect molecules or ions such as nitrite, glutathione, hydrazine, glucose, hydrogen peroxide, and many others (43 different species have been listed where CoPc was used for detection).⁵²  They focused on nitrite detection. The popularity of CoPc as detecting substrate specifically adsorbed onto a suitable electrode stems from a volcano plot where different MPcs and MTSPcs (M = metal, TSPcs = tetrasulfonic acid phthalocyanines) were used to detect a specific species X.⁴⁷  It was found that the effectivity of electro-oxidation of cysteine by these compounds was in the M order of Co > Fe > Mn > Ni > Cu. The detection principle to detect such species (e.g., X) operates as follows:⁵² 

Because the potential required for the mediated oxidation of compounds or ions X to be detected is less than the potential required for the direct electrochemical oxidation of these species according to reaction (1.20), X may be detected in a cheaper and more effective electrocatalytic way.⁵² 

It was shown that the electrochemical response of an edge plane pyrolytic graphite (EPPG) electrode exhibited within 98% of the same current responses during the oxidation of nitrite than a CoPc-modified EPPG electrode in water at pH = 7.2.⁵²  This means that CoPc is not the active catalyst during nitrite detection. That the CoPc was (initially) adsorbed on the EPPG electrode stems from the fact that i_p–υ plots (υ = scan rate) were linear in DMSO and water for Co^II/I processes [i.e., eqn (1.6) was obeyed]. In THF, the plot i_p–υ^1/2 was linear, which, according to eqn (1.5), is indicative of a diffusion-controlled dissolved species undergoing redox processes. However, nitrite oxidation does not benefit from the Co^II/I redox reaction. It is the Co^III/II couple that is electroactive in the oxidation of nitrite to nitrate. Further research showed that Co^IIPc is adsorbed on the electrode surface in microcrystallite form.⁵²  These crystals are insoluble in water. However, when it is oxidized to Co^IIIPc⁺, as in eqn (1.18), it undergoes solubilization with loss of electrode surface modification. Electrode surfaces were characterized with SEM and EDS. Thus, it was clearly shown in this project that if one does not know that the electrode itself can catalyze the desired oxidation of the compound to be detected, one can easily fall into the trap of making wrong conclusions by attributing catalytic activity to the envisaged catalyst while the electrode is actually the active catalyst. As the authors so elegantly demonstrated, the modified electrode should be checked for stability under the conditions of use (here utilizing the Co^III/II couple), not conditions that activate another redox process unimportant for the task (here utilizing the Co^II/I couple).

To overcome the solubility issues of adsorbed porphyrinoid electrocatalysts as described in Section 1.5.2, one can either covalently bind the catalyst to the surface of the electrode or it can be structurally modified to be less soluble, for example, by increasing porphyrinoid aggregation properties. Covalent binding of the catalyst to the electrode utilizing electrochemical methods is demonstrated in Figure 1.11.⁵³  An amino porphyrin was converted to a diazonium salt and then reduced electrochemically either at a platinum or glassy carbon (GC) or indium tin oxide (ITO) electrode utilizing five CV scan cycles between 0.5 and 0.05 V versus SCE. Cycles between 0.5 and 0.35 V were less efficient in generating the electrode–porphyrin bond. The electrode-grafted porphyrin could be modified chemically by inserting zinc into the cavity and then again removing it with 0.2 M HCl. Surprisingly though, the grafted porphyrin was not completely stable. Scanning between 0.00 and −1.80 V versus SCE for 100 cycles at 1 V s⁻¹ of the metal-free porphyrin-modified glassy carbon electrode led to 37% signal intensity current loss. The Pt- and ITO-modified electrodes lost 19% signal strength. Zn^{2 +}-porphyrin-modified GC electrodes were more stable than GC electrodes modified with the metal-free porphyrin because only 27% of the anodic and cathodic peak current values were lost after 100 CV cycles. However, when modified GC electrodes had Ni^{2 +} coordinated to the center porphyrin cavity, 55% of the signal strength was lost. Slow scan rates of 0.1 V s⁻¹ resulted in 100% loss of porphyrin redox features after 13 cycles.

In an alternative approach, the porphyrin structure may be altered to be less soluble than simple CoPc. Shown in Figure 1.12 is an electrochemically induced polymerization reaction of magnesium porphyrin. By applying 0.65 V versus SCE to the electrode, porphyrin units were linked at the meso position in a potentiostatic polymerization.⁵⁴  To prevent unwanted side reactions, polymerization was stopped when the deposition charge reached 10 mC cm⁻². The species pMgP-I was deposited as a thin film on the surface of the electrode. Acid treatment of this film leaves the metal-free complex p2HP-I still intact on the electrode.⁵⁵  Co^II can then be inserted into this film on the electrode surface to give a pCoP-I modified electrode by treatment with a saturated CH₃CN solution of cobalt acetate.

This procedure was demonstrated on GC, Pt, and ITO electrodes. Oxidation of the thin pCoP-I film to a pCoP-II film having a ladder structure was achieved by cycling the electrode potential 20 times between −1.80 and 1.40 V versus SCE.⁵⁴  Films of pCoP-II were characterized by XPS and SEM and, of course, electrochemically.

It was found that the (pCoP-II)-modified electrodes from the previous section could be used for oxidation and sensing of sulfite ions in aqueous solution (see Figure 1.13). The limit of quantification was 0.649 mmol dm⁻³, and the limit of detection was 0.195 mmol dm⁻³. This is about twice the sensitivity of glassy carbon electrodes modified with a copolymer of Co-tetrakis(para-aminophenyl)porphyrin and ortho-phenylene-diamine developed earlier by Arce and coworkers.⁵⁶  Detection of sulfites is an important analytical experiment since sulfites are used to prolong the shelf life of many food stuffs, including red wine.

Tests showed that a bare glassy carbon electrode in 0.1 M NaNO₃/water was inactive in the absence or presence of 10 mM sulfite (see Figure 1.13). The (pCoP-II)-modified electrode showed in the absence of any sulfite only a redox wave associated with adsorbed Co^II at ca. 0.240 V versus SCE, but in the presence of 10 mM SO₃^{2 −}, sulfite oxidation was observed at 0.46 [wave O1(S)] and 0.79 V [wave O2(s)] at a pH of ca. 9.7.⁵⁴ 

Wave O1(S) is only present in the first of multi-sweep CVs (see Figure 1.13, bottom right). However, when the cell is allowed enough time between scans to homogenize, wave O1(S) is observed on every scan. Furthermore, wave O1(S) varies linearly with scan rate, while wave O2(S) varies linearly with (scan rate)^1/2. Results were consistent with wave OI(S) representing oxidation of sulfite, which, prior to commencing the scan, diffused into the polymer film pores and was trapped there. Thus, oxidation of these sulfite ions is not dependent on diffusion; hence, a linear I/υ relationship was observed. It can only be observed in the first scan because sulfite diffusion into the polymer film is not fast enough to allow observation of wave O1(S) in successive uninterrupted CV scans. In contrast, wave O2(S) is considered the wave associated with electrocatalytic oxidation of dissolved sulfite that can diffuse freely to the surface of the (pCoP-II)-modified glassy carbon electrode; hence, its oxidation shows a linear dependence for the I/υ^1/2 relationship.⁵⁴  (pCoP-I)- and (pCoP-II)-modified Pt and glassy carbon electrodes were also found to be effective oxygen reduction electrocatalysts but will not be discussed here.⁵⁷ 

A discussion of electrocatalysis that does not include reduction of CO₂(g), O₂(g), and production of O₂(g) and H₂(g) from water (i.e., water splitting) would be incomplete because these processes are very important to fuel production and energy storage and a huge research effort has already gone into it. Important reactions in fuel production and consumption industries that would benefit from improved electrocatalysis include:

On the right-hand side of eqn (1.21)–(1.24) are fuels, and on the left-hand side are the products of fuel consumption. Eqn (1.24) is included in the array of equations above since the Fischer–Tropsch reaction [eqn (1.25)] involves formation of paraffins (alkanes, including petrol and waxes), α-olefins (alkenes), and water.

CO may be generated from CO₂ electrocatalytically and by reacting it with H₂. Utilizing the sun as an electrical energy source (solar power) would further assist the energy requirements of an energy-hungry future world.

We begin the story with the reduction of CO₂(g) to CO(g), but it can also be reduced to CH₄(g), C₂H₄(g), CH₃OH(g), or HCOOH(ℓ), to name only a few. CO is a feedstock material in the Fischer–Tropsch reaction and hydroformylation to produce liquid hydrocarbon fuels.

Often a misconception in electrocatalysis is that the electrocatalyst must be deposited on the surface of an electrode. This conception is incorrect. Electron transfer takes place at the electrode, and as long as a catalyst has access to the surface of the electrode, even from a homogeneous solution through diffusion, electrocatalysis can still take place. Officer⁵⁸  demonstrated this recently with the elegant electroreduction of CO₂ to CO catalyzed by a dissolved Fe(0) tetraphenylporphyrin/ionic liquid co-catalytic system in the presence of trifluoroethanol (TFE). [Fe⁰TPP]²⁻ is generated from Fe^IIITPPCl by an in situ electrochemical reduction. To place the energy saving this brought about in perspective, one should recognize that simple CO₂ reduction generating the first intermediate CO₂^•− according to the reaction CO₂ + e⁻⇌CO₂˙⁻ takes place at −1.97 V in DMF versus NHE. Addition of protons to the reaction leads to less energy-consuming reductions.^58,59  The electrode potential under acidic conditions falls with ca. 0.4 V for the reaction CO₂ + 2 e⁻ + 2 H⁺⇌CO + H₂O. By performing the reaction in the ionic liquid 1-butyl-3-methylimidazolium tetrafluoroborate, [BMIM][BF₄], a further energy saving was achieved. [BMIM][BF₄] does not ease the reduction potential of CO₂. That is the job of the weak acid trifluoroethanol (TFE). As shown in Figure 1.14, top right, addition of [BMIM][BF₄] to the reaction medium led to a decrease of 0.23 V (from −1.63 to −1.40 V versus NHE) in the reduction potential of the [Fe^ITPP]⁻/[Fe⁰TPP]²⁻ couple, probably through an electrostatic interaction between [BMIM]⁺ cations, the anions [Fe^ITPP]⁻ and [Fe⁰TPP]²⁻. The CV wave at ca. −0.98 to −1.06 V versus NHE corresponds to the [Fe^IITPP]⁻/[Fe^ITPP]²⁻ couple (best seen in the inset), and the wave at −0.14 to −0.23 V is associated with the [Fe^IIITPP]⁻/[Fe^IITPP]²⁻ couple. Use of 0.3 M [BMIM][BF₄] in DMF containing 1 M TFE led to a lowering of the onset potential of CO₂ reduction where CO could first be detected from −1.51 V to −1.36 V versus NHE. Gas evolution was measured by gas chromatography. This corresponds to a decrease of the overpotential η from 0.82 to 0.67 V relative to the equilibrium CO₂/CO potential of E° = 0.69 V in DMF.⁵⁸ 

The CVs on the right were recorded under argon (top) and CO₂ (bottom). Current densities were increased from 1.14 mA cm⁻² when a potential of −1.75 V versus NHE was applied to systems in the absence of [BMIM][BF₄] but increased to 3.26 mA cm⁻² at an applied potential of −1.56 V for the system containing [BMIM][BF₄]; both contained TFE. The electrode was stable over 4 h, and turnover frequencies (TOFs) were calculated as 47 (at −1.75 V, no ionic liquid) or 190 s⁻¹ (at −1.56 V in the presence of ionic liquid). Turnover numbers were found to be 6.77 × 10⁵ or 27.4 × 10⁵, respectively, after 4 h.⁵⁸ 

Tafel plots (Figure 1.14) had slopes of 197 mV dec⁻¹ for the system containing 0.3 M [BMIM][BF₄] and 286 mV dec⁻¹ for the system without any [BMIM][BF₄]. The smaller slope indicated that the rate of electron transfer in the presence of [BMIM][BF₄] is much faster (Section 1.4) and highlighted the important contribution of ionic liquids to this reaction.⁵⁸ 

Regarding hydrogen evolution reactions (= HERs), Wu and coworkers⁶⁰  demonstrated that β-hydrogenation of porphyrin followed by Ni^II insertion to give nickel(II) porphyrin, N-1, chlorin, N-2, porpholactone, N-3, and β-hydroporpholactone, N-4 (see Figure 1.15) increased TOF values for H₂(g) evolution from CH₃CN/CF₃CO₂H from 265 for N-1 to 6287 s⁻¹ for N-2 (i.e., a 24-fold increase in activity) and from 342 for Ni-3 to 1737 s⁻¹ for N-4 (a 5-fold increase in activity). DFT calculations showed that after initial 2 e⁻ reduction to generate the pre-catalytic species [Ni-1]²⁻ and [Ni-2]²⁻, the β-hydrogenated species [Ni-2]²⁻ has a more electron-rich Ni center than [Ni-1]²⁻. Interaction of CF₃CO₂H with [Ni-2]²⁻ then forms the active catalyst, which is a nickel hydride species, while in [Ni-1]²⁻, a protonated pyrrole species is generated. The nickel hydride catalyst generated from [Ni-2]²⁻ then reacts much faster (k₂ ≈ 3.03 M⁻¹ s⁻¹) with an additional CF₃CO₂H species than the pyrrole-protonated catalyst from [Ni-1]²⁻ (k₂ ≈ 0.0917 M⁻¹ s⁻¹) to generate H₂(g) with TOF values that are 24 times larger for Ni-2 than Ni-1.

The potential at which H₂(g) forms for the Ni-2 catalyst range from ca. −1.75 V at [TFA] = 100 mM to ca. −2.2 V versus Fc/Fc⁺ (see Figure 1.15, bottom left), which is slightly better than what was found⁶¹  for Pd and Cu meso-substituted tetraferrocenylporphyrin (− 2.0 until −2.4 V versus Fc/Fc⁺), but TOF_max values were still very good (ca. 6000). Despite the high TOF values, by utilizing the standard potential estimates of the H⁺/H₂ couple in organic media,^62,63  overpotentials are in the region of 0.8–1.2 V. This is very high. Good catalysts should have high turnover at low (< 0.5 V) overpotentials. It follows that the above catalysts will have a low power:chemical conversion ratio and need further refinement for industrial applications.

As far as reduction of O₂(g) is concerned, the work of Nemykin and Kadish comes to mind.⁶⁴  These authors used a series of three different meso-ferrocenylated cobalt porphyrins, (Fc)_n(p-CH₃Ph)_4−nPorCo with n = 0 − 4, adsorbed on an EPPG rotating disk as well as a rotating ring-disk electrode in DMF to reduce oxygen electrocatalytically. The Co^II/III couple provides the electrons to reduce O₂ to H₂O₂ selectively under acidic conditions in a two-electron process (see Figure 1.16). Hydrogen peroxide is a promising candidate as a sustainable energy carrier because it has a high energy density and emits no CO₂. It is, therefore, an environmentally friendly oxidant. Fuel cells running on hydrogen peroxide produce only water and oxygen as waste products after power generation, and over the years, cell performance has been improved significantly.⁶⁵  Interestingly, in the Nemykin study,⁶⁴  the electron-donating properties of the Fc groups lowered the potential at which O₂ reduction took place but did not participate in the electron transfer steps to O₂. Because of its bulky size, the Fc group did prevent dimerization of porphyrin intermediates; thus, selective oxidation of O₂ to only H₂O₂ was possible. In the absence of reactants that prevent dimerization of intermediates, four-electron oxidation to H₂O often occurs selectively⁶⁶  or a mixture of two- and four-electron oxidation takes place.^64,67 

Use of a rotating disk electrode gave sigmoidal CVs as shown in Figure 1.16(b). By plotting reciprocal i (written as j in this publication) against 1/(ω)^0.5 [see Figure 1.16(c)], the number of electrons flowing to O₂ at the electrode surface could be calculated using eqn (1.14a) as 2.1 (i.e., ≈ 2) even though the Co^III/II catalyst is involved in a one-electron redox process. This is consistent with H₂O₂ forming 95% selectively. The rotating ring-disk electrode gave slightly lower selectivity: 88% of products were H₂O₂.

Identification of the Co^III/II couple in the CVs came from spectro-electrochemical measurements. In a nutshell, the wavelength of maximum absorption of the Soret band of the studied complexes, which lies in the UV/vis wavelength region 400 < λ_max < 450 nm, shifted by as much as 21 nm to longer wavelengths when Co^II was oxidized to Co^III. In contrast, it stayed at the same wavelength when the Fc group(s) were oxidized. When the ferrocenyl groups were oxidized, the Soret band absorbance maximum fell by ca. 13%, but when ring-centered oxidation took place, the absorbance maximum of the Soret band fell more than 65%. This is an example of the use of spectro-electrochemistry to identify which electrochemical oxidation or reduction wave belongs to which part of a molecule. One could also have used DFT calculations to obtain the same information.¹⁰ 

Although the Fc group did not participate in the electron transfer process of Figure 1.16, free small ferrocene derivatives can be involved with two-electron O₂ reduction to H₂O₂ or four-electron reduction to H₂O. Thus, as reactant partners, a cobalt chlorin porphyrinoid, Co^II(Ch), catalyst with either dimethylferrocene (Me₂Fc) or Br₂Fc as reductants reduced O₂ in acidic media to H₂O₂ with TON numbers exceeding 30 000 at low overpotentials.⁶⁵  Cobalt corrole porphyrinoids also generate H₂O₂ in a two-electron acidic reduction of O₂,⁶⁸  while both catalytic two-electron⁶⁹  and catalytic four-electron⁷⁰  reduction of O₂ by free small ferrocene derivatives in the presence of non-porphyrinoid copper catalysts generate H₂O₂ or H₂O selectively.

In terms of water splitting reactions, energy requirements get a bit more demanding. Water is the biggest natural source available to generate H₂, H⁺, and O₂. O₂ forms during a four-electron oxidation of water [eqn (1.26)].

However, the oxygen evolution reaction (OER) above requires much energy since it occurs at potentials close to 2 V versus RHE. Not much success with porphyrinoids as electrocatalysts have been achieved in the OER reaction; hence, only two of them are discussed briefly. Cobalt phthalocyanine (CoPc, see Figure 1.16) and the perfluorinated derivative CoPcF₁₆ have been shown⁷¹  to electrocatalyze reaction 1.26 when adsorbed onto a hematide (a photoanode material) film previously evaporated onto fluorine-doped tin oxide glass (FTO). The photoanode material allows photocatalysis to work in tandem with electrocatalysis during the OER. CVs showed that the onset potential of OERs for CoPcF₁₆- and CoPc-modified FTO electrodes was ca. −1.67 and −1.78 V versus RHE, respectively. Irradiation of a CoPcF₁₆/hematite/FTO-modified electrode with 5 s interval chopped light from an AM1.5G solar illumination source lowered the OER onset potential to ca. 0.8 V versus RHE, and the current density increased to ca. 1 mA cm⁻² at 1.23 V versus RHE. This current density is more than twice as high as that of a known cobalt–phosphate/hematite water oxidation photocatalyst system under the same conditions.⁷¹  The CoPcF₁₆/hematite/FTO-modified electrode was stable for 60 000 s at 1.7 V versus RHE under illumination; Faradaic efficiency was close to 100%. Oxygen was measured with an in-line micro gas chromatograph. In another report,⁷²  a CoPcF₁₆/FTO-modified electrode operating at 2 V versus RHE (OER onset potential = 1.7 V) for 8 h (TON ≈ 1 × 10⁵) at pH = 7 produced about 4 μmol O₂ h⁻¹ cm⁻² at a turnover frequency (TOF) of 3.6 s⁻¹. This is an efficient electrocatalytic process with current knowledge, but it is not spectacular. It means that a 1 cm² electrode produces over 1 h less than 0.1 mL O₂ gas.

One of the problems with heterogeneous catalysis, which includes electrocatalysis by material adsorbed on the surface of an electrode, is particle size. If the particles are large, there is a large portion of the material in the interior of these particles to which reacting molecules never gain access. It means that the measured activity of an electrocatalyst is much less than the potential (theoretical) activity simply because large portions of a catalyst are hidden (or “dead”) in the interior of large particles. Compounds that easily aggregate, as many phthalocyanine porphyrinoids do, are especially prone to this limitation. To overcome it, CoPc and octa(octyloxy)CoPc (the latter is labeled CoPc-A in Figure 1.16) were prepared and adsorbed onto CCG (chemically converted graphene) electrodes. Then electrocatalyst activity for these two materials was compared in the reduction of CO₂ to CO.⁷³  Characterization of the surfaces was with UV–vis, Raman, XPS, and TEM coupled with energy-dispersive X-ray spectroscopy (TEM–EDS). Bright field TEM images showed the morphologies of a bare CCG electrode, and both electrodes modified with layers of phthalocyanine adsorbed on it were the same [see Figure 1.16 bottom, inset (e), left]. EDS mapping of the CoPc-A-modified CCG electrode showed a homogeneous distribution of Co, C [see Figure 1.16, inset (e), middle and right], and N over the composite material.

Integration of the CV Co^II/I oxidation allowed estimation of the amount of electrochemical active CoPc or CoPc-A molecules on the CCG electrode surface as 1.4 × 10⁻⁹ and 9.0 × 10⁻¹⁰ mol cm⁻², respectively. This led to a larger observed total current density (e.g., ca. 9 mA cm⁻² if the applied potential = −0.8 V versus RHE) for the CoPc-A-modified CCG electrode compared with the CoPc-CCG-modified electrode (ca. 6 mA cm⁻² under identical conditions) during electrocatalytic CO₂ reduction. In addition, the faradaic efficiency for CO formation for the CCG/CoPc-A-modified electrode is ca. 77% at −0.59 V, while that of the CCG/CoPc-modified electrode is only 63% at the same potential.⁷³  These results are consistent with an interpretation that the long-chain octyloxy substituents of CoPc-A prevented Pc aggregation much better than aggregation is prevented by simple CoPc.

The authors⁷³  developed the concept of effective turnover frequency (eTOF) as a measure of the minimum activity of an aggregated catalyst. The lower aggregation of CoPc-A on the graphene sheets resulted in an effective catalytic activity per single CoPc-A molecule of ∼5 s⁻¹ at −0.59 V compared to an eTOF of ca. 2 s⁻¹ at −0.59 V for the CCG/CoPc-modified electrode. In addition, the CoPc-A-modified CCG electrode exhibited long-term stable CO production with eTOF increasing to ∼6 s⁻¹ and eTON ∼6.7 × 10⁵ for CO evolution over 30 h of electrolysis. The compound minimizing aggregation, octa(octyloxy)CoPc = CoPc-A, was thus [(5 to 6)/2] = more than 2.5 times more active as the compound that aggregated strongly, CoPc. The authors also showed that CoPc-A was more active than almost all other state-of-the-art CO₂ reduction electrocatalyst systems.⁷³  The idea of using long-chain substituents to prevent or minimize aggregation is, however, not new. Cook²⁷  determined the onset aggregation concentration of non-peripherally substituted phthalocyanines as a function of substituent chain length, and others used this information to overcome aggregation in spectroscopic and electrochemical studies.^10,26,37 

The last support system discussed is the covalent organic framework (COF) system where CoPc was covalently bound into a cavity-forming structure (see Figure 1.17).^74,75  Films of these COFs were adsorbed onto an HOPG electrode (HOPG = highly oriented pyrolytic graphite⁷⁶ ), characterized thoroughly and then tested in the reduction reaction of CO₂ to CO. The most active CO₂ reduction film was found to have R₁ = R₂ = F. For this complex, CO₂ was reduced at a low (550 mV) overpotential with 87% faradaic efficiency. When a potential of −0.67 V versus RHE was applied, the current density achieved was 65 mA mg⁻¹ COF.⁷⁴  Derivatives with four rather than three phenyl rings in the linking unit, linking CoPc functionalities together and being void of F substituents, were slightly less active.⁷⁵ 

In contrast to COFs, MOFs (metal–organic frameworks) are not held together by covalent bonds. Building blocks are held together by coordination bonds that are not as strong as covalent bonds. However, this does not mean MOFs are excluded from electrocatalysis. For example, a redox-active iron porphyrin was built into the robust Zr-based PCN-223-Fe MOF, [Zr₆O₄(OH)₄(Fe(III)-(TCPP)₃] with TCPP = meso(tetracarboxyphenyl)porphyrin)] and films of it adsorbed onto an FTO substrate.⁷⁷  This working electrode was then used for the O₂ reduction reaction (the ORR), and it was found that the MOF stayed intact and was capable of reducing O₂ 94% selectively to water in a four-electron process. Only 6% of products were found to be H₂O₂, which was generated in a two-electron reduction of O₂ (compare Section 1.6.3). The current density was ca. 7.5 mA at 0.8 V versus NHE. A review concerning design strategies for MOF electrocatalysts is available.⁷⁸ 

This work started with some fundamentals of electrochemistry and showed how it is applied in simple solution-phase electrochemical experiments. These were all outer-sphere processes that have little or no interaction between the electrode surface and the redox species. Such redox processes are fully described by formal reduction potentials, E°′, standard rate constants, k°, charge transfer coefficients, α, and double layers next to an electrode surface as defined by standard Butler–Volmer treatments.⁷⁹  Questions that these studies answer tend to focus on the influence structural and electronic changes the redox species has on its electrochemical properties, as well as the influence of solvent and electrolyte interactions with redox species. The chapter then progressed into the realm of inner-sphere redox processes. The chapter discussed some theoretical background and presented case studies. Here, the bonding or adsorption of reactants, intermediates, and products to the electrode surface enhances the rate of electrode reactions by several orders of magnitude and is the gateway to electrocatalysis. Inner-sphere redox processes are characterized by overpotentials, η, Tafel slopes (which is an indicator of rates of electron transfer in the rate-determining mechanistic step) and exchange current densities at equilibrium (also an indicator of charge transfer kinetics) as defined in the Tafel equation.⁷⁹  Several examples of the application of these concepts were provided.

While all of these aspects remain important in a modern era, leading future studies need to strive to bring additional information to the party. These involve more information of how electrode interactions with adsorbed species enhance electrode reactions and what structural properties of the electrode as well as the adsorbed particle will enhance catalytic reactions. One needs to understand why certain changes will have no influence on electrocatalysis, what types of changes will actually be detrimental, and why other changes will be very beneficial to electrocatalysis. The emphasis is on the “why”. It means graduating from a black box approach, which simply reports “if you do ‘this,’ ‘this’ happens” to understanding “why” it happens. The new leading modern publication should report “why” it happens.

Toward realization of this goal, more use of heterogeneous surface characterization techniques applied to electrode surfaces and adsorbed particles becomes increasingly important. Spectroscopic techniques in a multidisciplinary approach to enhance insight and understanding already play a big role. The chapter tried to show how XPS, TEM and TEM-EDS, AFM, SEM, FE-SEM, and DFT calculations brought new insights and enhanced the quality of results. This is not an exhaustive list of possible techniques to utilize. Additionally, electrochemical scanning tunneling microscopy, SEM, scanning ion conductance microscopy, and PXRD come to mind; the list is almost endless. Understanding how adsorbed nanoparticles may diffuse, aggregate, or sinter during electrocatalysis and how it influences the electrocatalytic reaction is another important aspect of research. The properties of electrodes are another topic that needs to be researched thoroughly. It is not any more cutting-edge science to report that on platinum you get “this,” on glassy carbon you get “that,” and on single-walled carbon nanotubes or graphene, you get “another” result. One needs to develop knowledge on why these results are expected. One also needs to understand the effect of composite electrode material, grain boundaries in such electrodes, and the effect defects on the surface of single crystals have on electrode performance.⁷⁹  This is only possible if several multidisciplinary labs collaborate on projects. This is also where the bigger funding opportunities are.

The days of a single researcher doing it all as described above are gone. We hope that this chapter has guided, challenged, and inspired new research avenues and collaborations for the benefit of all interested in electrochemistry and electrocatalysis.

The authors acknowledge the Central Research Fund of the University of the Free State, Bloemfontein, South Africa for financial support.