1 Introduction

Since the discovery of its sandwich structure [1], ferrocene, an orange d6 Fe(II) 18-electron pseudo-octahedral complex, has been associated with its oxidized green d5 Fe(III) 17-electron ferricenium form and with redox properties of the Fe(III)/Fe(II) redox couple [13]. Upon oxidation, the ferrocene sandwich structure is essentially retained in ferricenium salts, with a lengthening of only approximately 0.1 Å of the 10 Fe–C bonds from 2.02 Å in ferroceneto 2.10 Å in ferricenium (Eq. 1).

(1)

The very rich chemistry of ferrocene related to its aromaticity has brought the ferrocene area to the forefront of organometallic chemistry [4, 5]. Applications of ferrocenes are multiple in polymer and materials science [611], medicine, [1217] sensors, [1824] and asymmetric catalysis. [2527]. More than 1,000 publications have been dealing with the electrochemistry of ferrocenes with emphasis particularly on polymers [2, 6, 11, 2830], thin-layer electrochemical cells [31], and ionic liquids [32]. Ferrocene is very often found in macromolecular structures in which its redox chemistry is of interest to modify the properties of the materials. In this micro-review that is in large part devoted to research in the author’s laboratory, various fundamental and applied aspects of the redox chemistry of ferrocenes and other iron–sandwich complexes are summarized.

2 Recent Syntheses of Ferrocenes

The synthesis of ferrocenes has been reviewed several times [3337] and commonly starts from ferrocene that is an inexpensive commercial product [37]. Recently, however, new synthetic methods of functional complexes that are not otherwise accessible have been synthesized by visible-light photolysis (using a simple desk lamp) of functional mixed sandwich complexes of the type [Fe(η5-C5H4R)(η6-toluene][PF6] (R = functional group) producing the complexes [Fe(η5-C5H4R)(NCMe)3][PF6] [38, 39] that react with functional cyclopentadienyl lithium or sodium salts (Eq. 2) [40]. Likewise, bifunctional biferrocenes are accessible from the binuclear mixed-sandwich complexes (Scheme 1) [40, 41].

(2)
Scheme 1
scheme 1

Selective synthesis of functional biferrocenes by visible-light photolysis of mixed-sandwich complexes under ambient conditions (desk lamp, r.t.) [40]

Whereas ferricenium salts are relatively stable in the solid state, these salts are fragile in organic solutions and react with dioxygen from air [42]. More robust ferricenium complexes are obtained from ferrocene derivatives containing electron-releasing substituents that strengthen the Fe-cyclopentadienyl bonds [43, 44]. This is particularly the case in biferrocene complexes [4547], because ferrocene is an electron-releasing substituent for the nearby ferrocenyl moiety. These biferrocenyl complexes present a localized mixed valency (class 2 mixed-valent complexes [48]), i.e. they are Fe(II)–Fe(III) complexes with some electronic coupling [47]. Thus the biferroceniummonocation is stabilized by its ferrocenyl substituent compared to the single ferricenium cation, and biferrocenium salts are robust [49, 50] and can be introduced into macromolecular structures [51, 52], unlike biferricenium dication.

3 Ferrocenes as Stoichiometric Redox Reagents

Although they are weak reductants because the oxidation potential of ferrocene is between 0.4 and 0.5 V versus SCE depending on the solvent (vide infra), ferrocenes are useful reducing agents towards strong oxidants because of the relative stability, a least for a short time, of the ferricenium salts. Thus ferrocene has been used to reduce Au(III) that is coordinated to nitrogen ligands such as triazoles, leading to the formation of gold nanoparticles (AuNPs). These nanoparticles are much larger than those obtained by NaBH4 reduction, because the later reductant is much stronger than ferrocene, and the NP size is linked to the reduction rate that depends on the driving force (Fig. 1) [53, 54]. Along this line, biferrocene is a stronger reductant than ferrocene both because of the presence of the electron-releasing ferrocenyl group as substituent and the mixed valency; thus smaller AuNPs are formed than with ferrocene under comparable conditions, and the biferrocenium AuNP stabilizer is fully stable [51, 52].

Fig. 1
figure 1

Biferrocene-containing polymers as Au(III) reductants to form nanosnake-shaped polymers that encapsulate AuNPs

Decamethylferrocene is an even stronger reductant than biferrocene, with an oxidation potential approximately 0.5 V less positive than that of ferrocene (depending on the nature of the solvent, vide infra) and has also been used to reduce functional ferricenium complexes [50] and triazole-coordinated Au(III) to AuNPs [53]. As mentioned above, ferricenium salts are relatively stable, which is not the case for boro- or alumino-hydrides that form ill-defined clusters. Other neutral iron–sandwich complexes are much better reductants than ferrocenes and also form well-defined products (vide infra) [37, 43, 44].

Ferricenium salts are common mild oxidants that are convenient to prepare and handle (Scheme 2). They are popular in the organometallic community and conveniently form ferrocenes upon redox reactions. They have been used for a long time, in particular to oxidize anionic and neutral organometallic derivatives [55, 56]. They oxidize pentamethylferrocene-containing dendrimers to pentamethyferricenium derivatives (Scheme 3) and biferrocenes to biferrocenium cations (Scheme 4). Their use has been reviewed, in particular by Geiger et al. [43, 44].

Scheme 2
scheme 2

Use of decamethylferrocene as an exergonic reductant of monosubstituted ferricenium dendrimers (den = CH2-dendrimer)

Scheme 3
scheme 3

Use of a ferricenium salt to exergonically oxidize pentamethylferrocene dendrimers (den = CH2-dendrimer)

Scheme 4
scheme 4

Use of a ferricenium salt to exergonically oxidize a biferrocenyl dendrimer to biferrocenium dendrimer (R = dendritic core). These reactions were also carried out with [FeCp2][BAr F4 ] {Cp = η5-C5H5; ArF = 3,5-bis(trifluoromethyl)phenyl)}

Decamethylferricenium salts are less useful oxidants, because they are weaker oxidants than ferricenium salts. On the other hand, acetylferricenium salts [43, 44, 57, 58] are very useful, because they are stronger oxidants than ferricenium salts and thus can stoichiometrically oxidize a large variety of ferrocene derivatives that do not contain electron-withdrawing substituents. For instance, they can oxidize ferrocene dendrimers to ferricenium dendrimers [50].

4 Other Iron–Sandwich Complexes as Stoichiometric Redox Reagents

The other stable iron–sandwich complexes that are very useful redox reagents are the mixed sandwich complexes of the [FeCp(η6-arene)]n+ family (n = 0–2) that are robust (even in concentrated sulfuric acid) for n = 1 (yellow, d6, Fe(II) [59]), are stable in their green Fe(I) form (n = 0) for peralkylated arenes [6062] and for the water-soluble complex [Fe(η5-C5H4CO2 )(C6Me6)] [63] and their Fe(III) form for the purple 17-electron d5 complexes [Fe(η5-C5Me5)(η6-C6Me6)][SbX6]2 (X = F or Cl). This latter fully methylated complexes have a redox potential of 1.85 V vs. SCE for the Fe(III)/Fe(II) redox couple in SO2 and are obtained upon oxidation of the Fe(II) precursor by reaction with SbCl5 in CH2Cl2(X = Cl) or SbF5 in SO2 (X = F). They are very strong oxidants that are thermally stable and can oxidize [Ru(bipy)3][PF6]2 to the 17-electron RuIII complex and the neutral cluster [FeCp(CO)]4 to its mono- and dications [64]. These fully methylated complexes are stable in three oxidation states, the Fe(I) complex being the most electron-rich neutral molecule known with an ionization potential determined by He(I) photoelectron spectroscopy as low as 4.2 eV [65]. The most common and currently used Fe(I) complex of this family is [FeCp(η6-C6Me6)] [60] that reduces C60 up to the tris-anion C60 3− (Scheme 5) [66].

Scheme 5
scheme 5

The three stable oxidation states of the permethylated complexes [Fe(η5-C5R5)(η6-arene]+n (n = 0–2) and representative redox reactions. The 18e d6, Fe(II) monocationic precursors [Fe(η5-C5R5)(η6-C6Me6][PF6] are accessible upon reactions of C6Me6 with AlCl3 and ferrocene (R = H) or [FeCp*(CO)2Br] (R = Me; Cp* = η5-C5Me5) [60, 64]

The [FeCp(η6-arene)]n+ family is also known with binuclear [67] and dendritic complexes with up to 27 [FeCp(η6-C6H2Me3−)][PF6] [68] or 64 [Fe+5-C5H4CO2−)(η6-C6Me6)][PF6] termini, the later dendritic complex being able, in it 19-electron Fe(I) form, to reduce C60 [69].

Finally, another useful electron-transfer reagent is the 20-electron complex [Fe(η6-C6Me6)2] first reported by Fischer and Röhrscheid, by reduction of the 18-electron dicationic Fe(II) precursor [70], and for which organometallic chemistry was later examined [71, 72]. The Fe(I)/Fe(0) redox couple has a redox potential of −1.4 V versus SCE and is thus a rather strong reductant. It reduces dioxygen, yielding a thermally stable 18-electron o-xylylene complex, and it also reduces various organic halides RX giving 18-electron Fe(II) complexes [Fe(η6-C6Me6)(η5-exo-C6Me5R)] as a result of in situ coupling between the radical R., formed by reduction of RX, and the arene ligand of the 19-electron intermediate [73].

5 Electrochemical References

The ferricenium/ferrocene system has long been considered by the IUPAC as the official reference for the determination of electrochemical potentials determined by cyclic voltammetry [74]. It is indeed the prototype of electrochemically reversible systems, because there is no significant structural change between the oxidized and reduced forms, the bond lengthening upon oxidation being minute (cf. introduction). In such a case no kinetic barrier is involved along the electron transfer that is much faster than the electrochemical time scales. Ferrocene is also readily available, and it mild oxidation potential is located in a readily accessible region of the electrochemical scale. It had been found, however, that ferricenium was the subject of specific solvation that influenced the redox ferricenium/ferrocene potential differently in different solvents, so that the so-called “ferrocene assumption” has been questioned [75, 76]. Indeed, the thermodynamics of ion pairing between ferricenium cation and electrolyte anion varies with the nature of the solvents, because the central metal is not protected against nucleophilic attack between the two rings of ferrocene. On the other hand, such effect has been found to be negligible in decamethylferricenium due to the shell protection of the metal by the permethylated ligands, subsequent to the finding of the electrochemical reversibility of the oxidation of decamethylferrocene to its cation [7780].

Consequently, the decamethylferrocene (Fe(II/Fe(III) couple has been proposed as a more reliable reference than the ferrocene Fe(II)/Fe(III) system [8184]. Along with decamethylferrocene, it has been proposed that other permethylated late transition-metal sandwich complexes such as decamethylcobaltocene [85] and [FeCp*(η6-C6Me6)] (vide supra) [60] with their monocations may also serve as complementary references when the redox potential to be measured interferes in the same potential region as decamethylferrocene. Table 1 gives the comparison between the values of the redox potentials of the permethylated iron–sandwich complexes and comparisons with decamethylcobalticenium (Table 1). Consequently values measured in the literature versus ferrocene as the reference can be converted using Table 1 to values versus decamethylferrocene or other references of this table in the given solvent [83, 84].

Table 1 E1/2 values (V) of iron–sandwich complexes determined (a) versus SCE and (b) versus FeCp +/02

6 Electrocatalysts [86, 87]

The 17-electron Fe(III) salts ferricenium, [FeCp2][PF6], and [Fe(η5-C5Me5)(η6-C6Me6)][SbX6]2 (X = F or Cl) and the 19-electron Fe(I) complexes [Fe(η5-C5R5)(η6-C6Me6)] (R = H or Me) are excellent electron-transfer reagents for stoichiometric oxidation and reduction respectively, but they can also be used catalytically as sources of electron hole or electron respectively in electron-transfer-chain catalyzed (electrocatalytic) reactions [87, 88]. For instance ferricenium as been used as electrocatalysts in the [W(CO)3(MeCN)3]-catalyzed terminal alkyne polymerization [88] according to a mechanism that combines the electrocatalytic acetonitrile exchange by alkyne ligands at the tungsten center with the metathesis polymerization of the terminal alkynes catalyzed by W-vinylidene species according to the Chauvin–Katz mechanism [8991]. In this mechanism, it is proposed that coordination of the terminal alkyne onto the W-vinylidene species resulting from rearrangement of a W(η2-alkyne) species produces a tungstacyclobutene. The role of the electrocatalyst is to considerably accelerate the acetonitrile/alkyne ligand substitution by processing via odd-electron W species, whereas ligand substitution from an 18-electron W catalyst requires overtaking a high-energy barrier for ligand dissociation to form a 16-electron species (see a proposed mechanism in Scheme 6). Indeed the polymerization proceeds immediately at room temperature in the presence of the ferricenium electrocatalyst instead of requiring heating at 100 °C for several hours in its absence [90].

Scheme 6
scheme 6

Proposed mechanism for the W-catalyzed polymerization of terminal alkynes (R = Ph, n-Pr, n-Bu; Ln are undetermined ancillary ligands in the intermediate species) accelerated by catalytic amounts of [FeCp2][PF6] (optimized amount: 20 %vs. the W catalyst). W catalyst: 10 % versus alkyne; r.t., 10 min.; solvent: THF or CH2Cl2 [88, 90]. Molecular weights up to 32,000 were obtained (yields: 40 ± 5 %; polydispersity indices: 2.2-2.4)

In electrocatalytic reactions, the driving force of the electrocatalyst must be sufficient to initiate the reaction. For instance a catalytic amount (0.1 equiv) of [FeCp(η6-C6Me6)] allows facilitating the addition of PMe3to the Ru center of the fulvalene complex [RuW(η102-C10H8)(CO)5] to produce the zwitterionic complex [(CO)2(PMe3)Ru+102-C10H8)W(CO)3], but it cannot further catalyze CO substitution by PMe3. On the other hand, the more powerful reductant [Fe(η5-C5R5)(η6-C6Me6)] in 0.15 equiv. can do it to form the other zwitterion [(CO)(PMe3)2Ru+102-C10H8)W(CO)3] [92]. This example illustrates the advantage of the use of several electrocatalysts of various driving forces to selectively conduct precise reactions, and iron–sandwich complexes that can bear a variable number of methyl substituents perfectly fulfill this role.

These odd-electron iron–sandwich electrocatalysts can also play the role of redox catalysts, i.e. catalysts of redox reactions; for instance the complexes [Fe(η5-C5H4R)(η6-C6Me6)] (R = H or CO2 ) catalyze the cathodic reduction of nitrates and nitrites to ammonia, and this function has been reviewed earlier [23, 93].

7 Endo Receptors and Exoreceptors

The use of ferrocenes attached to endo-receptors has been prolific for the recognition of anions, and in particular those that play important roles in biology such as oxoanions including adenosyl triphosphate (ATP and other DNA fragments) or in the pollution of the environment (nitrate, phosphate, radioactive pertechnetate). These endoreceptors (chelates, tripods, crowns, porphyrins, calixarenes) incorporating ferrocenyl groups have been extensively developed by the group of Beer and involve the synergy between electrostatic bonding with ferricenium upon anodic oxidation, supramolecular interaction between the anion and a functional substituent of the ferrocenyl group such as amido and the topology effect of the encapsulation [19, 22, 94, 95]. The resulting cyclic voltammogram (CV) undergoes either a cathodic wave shift of the ferrocene oxidation CV wave or the appearance of a new CV wave depending on the strength of interaction [96], and the subject has been reviewed [94]. On the other hand, ferrocenyl-terminated dendrimers behave as exo-receptors, with anion recognition near the ferrocenyl periphery [97]. A remarkable positive dendritic effect was observed, i.e. the CV wave shift was all the larger as the metallodendrimer [98] generation increased, which was attributed to the narrowing of the space between the branches thereby improving the interaction with the anion when the generation increased [99, 100]. More recently these receptors have been applied to the recognition of both anions such as ATP and cations such as Pd(II) using 1,2,3-triazolylferrocenyl- [101103] or triazolylbiferrocenyl-terminated dendrimers due to the coordination of Pd(II) to the triazole link that influences the ferrocenyl oxidation potential [49, 104].

8 Concluding Remarks

Ferrocenes and other iron–sandwich complexes are useful stoichiometric and catalytic redox reagents under various oxidation states that are stable. In particular the mixed [FeCp(arene)] involves three stable oxidation states that can be used either as strong oxidants as Fe(III) or strong reductants as Fe(I) with fully permethylated rings. In general iron–sandwich complexes with various amounts of methyl groups on the rings offer flexibility in the modulation of the oxidation potential. The iron–sandwich complexes with fully methylated rings are reliable electrochemical references spanning across the electrochemical scale, whereas the redox potential of the parent ferrocene varies with the nucleophilicity of the solvent and supporting electrolyte anion.

In biferrocenes, the mixed-valency introduces a separation of the CV waves, depending on the structure, of the order of 0.3–0.45 V with [n-Bu4N][PF6] as the supporting electrolyte in CH2Cl2. This separation is due to the mutual electronic effect between the two ferrocenyl moieties and also in a large part to the electrostatic effect, as shown by the splitting increase between these two CV wave potentials with [n-Bu4N][BAr F4 ] up to the order of 0.6–0.7 V as shown by Geiger’s group [105107]. In the absence of electronic effects, the electrostatic factor was shown to be sufficiently significant in tris- and hexa(ferrocenylethyl) benzene using only the latter electrolyte to give rise to stable mixed-valent complexes [108, 109].

In polymers and dendrimers containing biferrocenyl group, oxidation of one of the two ferrocenyl groups by a ferricenium salt is exergonic and convenient, and the mixed-valent biferrocenium moieties that are obtained are much more robust than ferricenium. Stabilization of AgNPs [110, 111] and AuNPs [51, 52] is achieved by reduction of Ag(I) and Au(III) by these biferrocene groups to yield remarkable stable networks encapsulating NPs.

The electrostatic factor becomes negligible, however, when the dendritic tethers are lengthened [112]. Consequently large ferrocenyl dendrimers appear at first sight in CV with a single chemically and electrochemically reversible wave as a single-electron system [113, 114], unlike for instance in Manners’ poly(silylferrocene) oligomers and polymers in which both electronic and electrostatic reasons provoke CV wave splitting [29].

Finally ferrocene-containing macromolecules have been shown to electronically communicate [115] and enhance charge transport through insulating layers [116], and the oxidation of ferrocenes is not only observed upon electron transfer but also upon charge transfer in nanomaterials [117].

In conclusion, it is believed that the redox functions of ferrocenes and other iron–sandwich complexes will become a strategic tool to manipulate nanomaterials including anti-cancer ferrocene and ferricenium compounds that were first developed in oncology by Eberhard W. Neuse [118120].