1 Introduction

The supramolecular redox functions of metallomacromolecules are ubiquitous in the chemistry of life [1] as exemplified by electron transfers in metalloproteins [2], the role of ferredoxins achieving electron transfers in many redox reactions of the metabolism including photosynthesis [3] and nitrogenase [4], and a number of other biocatalytic redox processes of metalloenzymes [5]. Nature has also inspired chemists for the mimicry of biological redox processes in modern technology, for instance in the fields of energy [6, 7], nanomedicine [8], and the metal-tanning process of leather industry [9]. The designs of redox reagents mimicking the redox sites of metalloenzymes range from metalloporphyrins [10] that are found as redox site of oxidoreductase enzymes such cytochromes that are involved in many crucial functions, metallophthalocyanins [11] found in pigments, clusters found in ferredoxins and other metallocyclic nitrogen ligands such as corrolle found in vitamin B12 [1]. Interestingly non-biological redox reagent using metals and ligands that are not found in living system can also achieve remarkable performances, such as polypyridyl-noble metal complexes that are useful in photophysical processes, organic synthesis [12] and polyphenol-metal (e.g. AlIII, CrIII) complexes that are powerful tanning system to improve the hydrothermal stability of tanned leather [9]. Another well-known and current family of non-natural redox systems that could be rationalized and optimized regarding its usefulness is that of the late transition-metal sandwich complexes [13]. This later area has concentrated our attention during several decades and will be discussed in this mini-review essentially based on research conducted in our laboratories. When these non-natural redox systems are engaged in macromolecular materials, their redox switch can exert a variety of functions and processes that are used in synthesis, catalysis, biomedicine and materials science [14]. In biological as well as non-natural processes, the supramolecular aspects involving the redox centers and their close ligand environment play a particularly crucial role due to the hydrogen bonds, electrostatic interaction and redox-induced structural and conformational changes in these metallomacromolecules [15].

2 Electron transfer by Electron-rich Organo-Iron Sandwich complexes

The stabilities of the redox states of a redox reagent within a given timescale are a fundamental criterium toward functions. This story started with the discovery of ferrocene in the late 1940’s and early 1950’s, mainly by Fischer, Wilkinson and Woodward [16], that boosted a number of disciplines including organotransition-metal chemistry, homogeneous catalysis and materials science [17]. In the early days of organotransition-metal chemistry, research on single electron-transfer aspects and their subsequent chemistry was not privileged besides electrochemical studies [18, 19]. In the early days, redox changes were found to be central in transition-metal chemistry, however, as illustrated by Taube when he classified ligand-substitution reactions as a function of the metal oxidation state [20, 21]. In ferrocene chemistry, the oxidation of the parent ferrocene and its alkyl-substituted derivatives is straightforward in a region of potentials that are readily accessible with a large choice of oxidizing reagents and at a convenient oxidation potential around + 0.4 V vs. saturated calomel electrode (SCE) in organic solvents such as THF, 1,2-dichloromethane, and acetonitrile [22]. Under these conditions, the isolation of the oxidized form of ferrocene, ferricenium (a d5 Fe(III), 17-electron complex), is relatively easy using tetrafluoroborate (BF4) or hexafluorophosphate (PF6) as a counter-anion. In turn, the latter salts can be used as mild single-electron oxidants [23]. Due to the absence of an electron in a bonding orbital [24] and the positive charge, however, this cation is somewhat fragile in solution especially in polar solvents, and the situation becomes even more complicated when a ferrocene ring bears an electron-withdrawing substituent such as an acyl group due to the anodic shift of the redox potential and the decreased stability of the oxidized form [25]. On the cathodic side, reduction of ferrocene itself has been observed by cyclic voltammetry, but the cathodic potential is so negative (− 3 V vs. SCE) that the isolation of the ferrocene Fe(I) anion is too difficult and has never been obtained [26].

In 1979, the seminal finding for the first time of the stable and isolable Fe(I) state in a large family of monomeric iron (I) sandwich complexes [27] stable under two or three oxidation states (Scheme 1) opened the possibility of a cascade of stoichiometric and catalytic electron-transfer reactions involving a variety of organic and inorganic substrates, demonstrating the remarkable usefulness of this new family of complexes that were called “Electron Reservoirs” [28,29,30,31,32].

Scheme 1
scheme 1

Synthesis and electron-transfer systems of ferrocene and [CpFeI6-C6Me6)]2+/+/0.

The Cp* series (Cp* = η5-C5Me5) are also known in both families of complexes and present even more robust redox systems as electron reservoirs [17, 21, 23, 29, 33] that have been integrated in polymers and dendrimers (see text).

Neutral Fe(I) sandwich complexes, such as CpFeI6-C6Me6), I, and Cp*FeI6-C6Me6) in which Cp = η5-C5H5 and Cp* = η5-C5Me5 [33] that are Jahn-Teller active [34] have indeed been found by He(I) photoelectron spectroscopy to be the most electron-rich known neutral molecules and remain so to-date based on the extremely low values of their ionization potentials [35]. These complexes are excellent initiators of electron-transfer-chain reactions [36, 37] and redox catalysts for the reduction of nitrate and nitrites to ammonia [38, 39]. Interestingly, they react very fast with O2 or air even under sub-ambient conditions. For instance, the forest-green prototype complex I provides an extremely mild C-H activation upon such a simple contact with O2 yielding the red ferrocene-like cyclohexadienylidene Fe (II) derivative. The superoxide radical anion was detected by its characteristic EPR spectrum as a reaction intermediate. Its formation is due to fast exergonic electron transfer from Fe(I) to O2 leading to a contact ion pair [Fe(II)+,O2.-] in which benzylic deprotonation by superoxide anion activated by the positive charge on the sandwich compound is also fast even at very low temperature. The nucleophilic reactivity of the cyclohexadienylidene complex is very rich and useful, forming a variety of bonds upon reactions with many electrophiles [39, 40]. The complexes [C5R5Fe(η6-C6Me6)]x+ were isolated in the three oxidation states (x = 0–2) corresponding to Fe(II,I,0), and electron-transfer interplay involving various substrates were demonstrated [41]. In addition, the structure-reactivity relationships, established with the family of mixed sandwich complexes C5R5Fe(η6-C6Me6), R = H or Me, have been generalized to first-row late transition metal sandwich complexes for which the neutral forms have 19- or 20 valence electrons, as for instance to the 19-electron cobaltocene derivatives [(η5-C5R5)Co(η5-C5R’5)] (R, R’ = H or Me) [42] and the 20-electron complex [Fe(η6-C6Me6)2, [43]]. Even main group hydrides such as NaBH4 and LiAlH4 can behave as electron-reservoirs able to transfer an electron to substrates at a very negative potential [44,45,46]. Among the useful electron-transfer reactions of these neutral 19-electron complexes, a recently conducted one is remarkable: the reduction of a variety of commercial late transition metal salts quickly leading to very small metal nanoparticles [47,48,49], whereas for instance NaBH4 often produces transition metal borides [50]. Under these conditions the driving force of the reactions is high given the very negative value of the oxidation potential of the neutral 19-electron complexes, resulting in the formation of very small nanoparticles whose catalytic activities are all the higher as they are clean and smaller [47,48,49]. This catalytic activity compares with that obtained using, for the reduction of these salts, the very strong reductant naphthalene anion that is much less easy to manipulate and quantify in stoichiometric amounts [51].

3 From single to simultaneous multi-Electron transfer

A key issue in electron-transfer reactions is the possibility to transfer two or several electrons simultaneously, because such multi-electron transfers are required in most biological energy-related processes [1]. Nature offers a variety of multi-electron transfer catalysts as for instance in photosynthesis [52], nitrogen reduction [53], etc., and unnatural catalysts are designed to mimic Nature in some way as for instance towards water oxidation catalysts [54] and lithium-ion batteries [55]. Thus, it is of interest to examine how multi-electron transfer system may proceed in molecular and solid-state systems [56]. In molecular systems, electrochemical oxidation or reduction of single or multi-redox site systems transfers one electron at a time (to or from the electrode at least for both statistical and electrostatic reasons [57], which brings about a difficult challenge in the design of bi- or polyelectron transfer systems. Cascades of single-electron transfers are currently found in cluster/nanoparticle electrochemistry. With single redox centers see, e.g. the hexanethiol-capped Au147 nanocluster showing 15 high-resolution quantized double-layer charging peaks [58, 59], and with two redox centers see e.g. binuclear transition-metal sandwich complexes displaying a cascade of 4 reversible monoelectronic waves from Fe(II)-Fe(II) to Fe (0)-Fe (0) in the example of the complex [Fe2210-C10H8(C6Me6)2]n+, n = 0–2 [60]. In simple related systems, however, it is possible to observe a two-electron wave between a dication and a neutral species. This is achieved if the redox reaction proceeds with a chemically reversible structural change occurring in the course of the second electron transfer and if the structural reorganization energy is larger than (thus compensates) the electrostatic energy. This is the case in the dicationic binuclear complex [Fe2Cp2212-biphenyl)][PF6]2 that is reduced in a single two-electron wave to the neutral complex [Fe2Cp2210-bicyclohexadinylidene)] (Schemes 2, 61].

Scheme 2
scheme 2

With [Fe2Cp2212-biphenyl)][PF6]2, only a single reversible 2-electron wave is observed by CV, whereas with the Cp* analog [Fe2Cp*2212-biphenyl)][PF6]2 two close single-electron waves are observed, and the average valence 37-electron complex is observable and isolable. In the first case the reorganization energy in the course of the 2nd single electron transfer compensates the electrostatic energy differentiating reduction of a dication from that of a monocation, which does not completely occur in the 2nd case. In both cases a bis-cyclohexadienylidene ligand resulting from intramolecular coupling within the diphenyl ligand is imposed by the 18-electron rule [61].

A related example was shown by Geiger in the octahedral cluster [Os6(CO)18)]2− that is oxidized in a reversible two-electron wave to the monocapped trigonal neutral cluster at 30 °C [62]. Electrochemical reversibility depends on the conditions (electrochemical time scale and temperature) [21, 22, 63]. In natural and unnatural energy-conversion processes, electron transfer often is accompanied by proton transfer. This is the case for many transition-metal-oxo redox catalysts (typically LnRu = O) [64, 65] in which the oxo ligand weakly binds acidic hydrogen atoms of proton donors. In many of these cases proton transfer is coupled with electron transfer, and both events occur simultaneously. Excellent reviews on proton-coupled electron transfer reactions are provided [64,65,66], and this subject will not be further treated here although many essential redox catalysts are concerned.

4 Electron transfers from Hexa-Iron redox centers: A new electrostatic paradigm

Hexa(ferrocenylethynyl)benzene is an interesting example of a six-electron transfer system. Using [n-Bu4N][PF6] 0.1 M as an electrolyte in CH2Cl2, anodic oxidation on Pt shows a single wave with Epa-Epc = 60 mV at 30 °C as expected for a chemically and electrochemically reversible transfer of 6 electrons leading to hexa(ethynylferricenium), meaning that the redox centers are apparently too distant from one another to exert significant electronic and electrostatic factors [67, 68]. Using the Bard-Anson equation [69], the calculated number of transferred electrons is very close to 6. Using [n-Bu4N][BArF4] 0.1 M as an electrolyte in CH2Cl2, however, three well separated two-electron waves are observed (Scheme 3). Note that this later electrolyte, contrary to the former one, contains a large counter anion forming with the cationic ferricenium unit a completely dissociated ion pair in which ion pairing is loose and the cationic charge exacerbated for electrostatic effects. This means that the electrostatic effects should be much more pronounced with this larger electrolyte [70, 71].

Scheme 3
scheme 3

Successive two-electron oxidation steps of hexa(ferrocenylethynyl)benzene with [n-Bu4N][BArF4] in CH2Cl2 based on electrostatic effects related to the conformation of the ferrocenyl group around the benzene plane. Reprinted with permission from ref. [68]. Copyright 2011 American Chemical Society.

Let us consider such enhanced electrostatic effects among the six ferrocenyl centers. For this purpose, the three related bis(ethynylferrocenyl)benzene complexes in which the two substituents around the benzene ring are in ortho, meta and para position, respectively, have been synthesized and analyzed electrochemically. Given the free rotation of the ferrocenyl group around the exocyclic single bond, the two ferrocenyl groups orientate in opposite side from the benzene plane in all three complexes, minimizing the electrostatic effects. The complexes in which the two ferrocenylethynyl groups are in meta or para are oxidized in a single electron wave even with [n-Bu4N][BArF4] as electrolyte, whereas the ortho derivative shows, with this electrolyte, two distinct single-electron waves. In this later case, even if the two ferrocenyl are in transoid conformation, they are close to each other and are the subject of a significant electrostatic effect.

Let us now consider the bis- and tris(ferrocenylethynyl)benzene complexes in which the groups are in meta position relative to one another. With [n-Bu4N][PF6] 0.1 M, these two compounds show a single two-electron wave, but with [n-Bu4N][BArF4] 0.1 M there is a remarkable difference between the electrochemical behaviors of these two complexes. The complex 1,3-bis- (ferrocenylethynyl)benzene shows a single 2-electron wave with Epa-Epc = 60 mV at 30 °C whereas, under these conditions, 1, 3, 5-tris-(ferrocenylethynyl)benzene shows three well distinct single-electron waves. This striking contrast between the bis- and tris-meta substituted complexes is explained by the frustration phenomenon. Whereas the disubstituted complex can bear the two ferrocenyl groups in transoid position to avoid any electrostatic effect, this is no longer the case for the tris complex due to frustration, i.e. the third ferrocenyl group must choose between a cisoid and a transoid conformation and finally probably adopts an intermediate conformation between cisoid and transoid, so that electrostatic repulsion occurs between the three ferrocenyl redox centers [68]. Such a frustration phenomenon has already been encountered in magnetic properties [72]. These experiments clearly illustrate the situation for which electrostatic effects are observed in multi-centered organometallic compounds and the crucial role of supporting electrolytes in exacerbating these effects by variation of counter ion sizes. These experiments allow understanding the three two-electron waves observed for hexa(ferrocenylethynyl)benzene. It is remarkable that the cyclovoltammograms are the same for the 1,3,5-tris- and the hexa-substituted complexes: both compounds show a single wave with [n-Bu4N][PF6] and 3 equal-intensity waves with [n-Bu4N][BArF4]. We know that two ferrocenylethynyl groups in para position are not the subject of electrostatic effect with any electrolyte, which explain why they are oxidized within the same electrochemical wave, whereas both the meta and ortho ferrocenylethynyl groups are the subject of electrostatic effects. This analysis allows one to understand how the ferrocenyl groups are oxidized in hexa(ferrocenylethynyl)benzene with the latter electrolyte in CH2Cl2, i.e. two para ferrocenyl groups at a time (Scheme 3) [68].

Whereas, large cores with six ferrocene termini show simple six-electron anodic oxidation [73], other hexaferrocene complexes are of interest. The electrochemistry of cyclic hexaferrocenes has indeed been reported in which the ferrocene units are directly bonded to one another [74] (Fig. 1) or linked between one another by a SiMe2 preventing significant electronic communication between the redox centers (Fig. 2) [75]. In the former case, both electronic and electrostatic interactions allow a given ferrocene unit to sense the first, second and third neighbors so that four single-electron waves are observed; whereas, the two last CV waves involving high positive charge numbers are beyond the electrochemical window [74]. In the latter case, the electrostatic influence of the first neighbors is observed, but that of the next ones is not, so that only two 3-electron waves are observed. In this family of cyclic ferrocene bridged by a SiMe2 group, however, three waves of different intensities were observed with an odd number of dimethylsilylferrocenyl units [75].

Fig. 1
figure 1

Cyclic hexaferrocene (four single-electron transfer CV waves are observed, the two others being out of scale) and other cyclic poly(ferrocenes), Fcn [74]

Fig. 2
figure 2

Cyclic hexa(ferrocenyldimethylsilane). Electrostatics effect are only observed between a ferrocene group and its two neighbors, but not with further neighbors. As a consequence, two 3-electron waves are seen by cyclic voltammetry [75]

5 Multi-Electron transfers in redox-active polymers

The richest family of redox-active metallopolymers is by far that of ferrocene-containing polymers in which the ferrocene groups are located along the polymer chain [76, 77] or on the side arms [78, 79]. For a long time, polymerization remained uncontrolled, but during the last decade methods of living polymerization [80] greatly improved the field of polymer materials. The most well-known ferrocene polymer family is that developed by Manners by the ring-opening polymerization of dimethylsilyl-bridged ansa-metallocenes, particularly with ferrocene yielding poly(dimethylferrocenyl)silane [76]. Oxidation of these polymers leads to class-II mixed-valence redox systems with an inter-valent absorption band at 1100–1300 nm. Their electrochemistry shows two equal-intensity CV waves observed at 0.00 and 0.24 V vs. ferricenium/ferrocene, respectively, as a result from the electrostatic interactions among the ferricenium centers [81].

The rational determination of the number of ferrocene units in polymers, i.e. the degree of polymerization was shown by Bard and Anson in 1978 with a poly(vinylferrocene) polymer according which the electrons are transferred one by one statistically from independent redox centers [69]. The result of the theoretical treatment is that the CV wave should show the same shape as that of the monomeric species in the same environment, but the diffusion current reflects the number of identical redox centers. Thus, the number of transferred electrons np is provided by the ratio of diffusion current intensities of the polymer ip and monomer im respectively, and also depends on the diffusion coefficient Dm and Dp of the monomer and polymer, respectively. The ratio of the latter is related to that of the molecular weights Mm and Mp of the monomer and polymer, respectively:

Dp/Dm = (Mm/Mp)0.55 [82]

Since oxidation of the vinylferrocene monomer is an one-electron oxidation to vinylferricenium, an equation providing np was empirically provided from dc polarography:

np = (idp Cm /idm Cp) (Mp / Mm)0.275 [69, 83].

This equation can be used in the absence of precipitation, adsorption and slow electron transfer at the electrochemical time scale. Thus, the choice of solvent is important, because it must solubilize both the oxidized and reduced forms of the metallopolymer and insure their chemical stabilities at least during this electrochemical time scale. The equation is useful to determine the number of redox centers np in relatively small- or modest-sized metallomacromolecules, and the risk of departure for incorrect results increases together with the molecular weight due in particular to adsorption. In late transition-metal sandwich complexes, the oxidized and reduced form interconversion undergoes only minute structural variation (minute change of the metal-ligand bond) that are fast on the electrochemical time scale, at least in the range of 17–19 electrons in the valence electron shell, because the full metal-sandwich structure is retained [13].

A particularly efficient living polymerization method is the ring-opening metathesis polymerization (ROMP) developed by the groups of Schrock and Grubbs with their transition metal alkylidene catalysts [84,85,86,87,88]. Recently efficient syntheses using Grubbs’ 3rd-generation ROMP catalyst [89, 90] of such precise ferrocene-containing polymers and co-polymers in which the ferrocene or cobalticenium groups are located on the polymer side arms have been developed [91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107]. Such metallopolymers have also been synthesized in which the electron-reservoir sandwich complex [CpFeII6-C6Me6)][PF6], I, has also been branched on the side chain [108]. Multiblock metallocopolymers have been synthesized containing a combination of blocks of ferrocene, pentamethyl ferrocene, cobalticinium, [CpFeII6-C6Me6)][PF6], and solubilizing polyethylene glycol blocks. Polymers containing the four distinct sandwich blocks have also been fabricated as macromolecular electrochromes. All the different redox systems of the tetrablock metallocopolymer have different colors, and each of them that is redox switchable either by application of a given potential or a redox reagent of specific driving force possesses two redox forms of distinct colors [109]. In all these living metallopolymers, the determination of the numbers of redox units is all the easier as these metallopolymers are monodisperse given the living polymerization process. These numbers where determined by the Bard-Anson formula and confirmed by the end-group analysis by 1H NMR, MALDI TOF mass spectrometry and UV-vis. Spectroscopy, in particular when the degree of polymerization (DoP) was not too high. In the case of high metallopolymers, the solubility must be designed, for instance upon involvement of a PEG block and introduction of a solubilizing block as the first block, such that the 1H NMR yield be quantitative and the final polymer remains soluble. In these cases, and if these conditions are fulfilled the number of redox units is the DoP of the living metallopolymer, as confirmed by the Bard-Anson formula.

Ferrocene-containing polymers, as ferrocene itself, rapidly reduce Au(III) and Ag(I) to Au and Ag nanoparticles (NPs), respectively, although the driving force of these reactions is weak, producing AuNPs and AgNPs of relatively large sizes such as 15 to 25 nm [110, 111]. AuNPs and AgNPs are well-known for their catalytic redox [59, 112, 113] and antimicrobial properties, respectively. AuNP catalysts are efficient if they are very small, thus this mode of Au(III) reduction is not adequate for that purpose, but can be replaced by reduction of Au(III) by NaBH4 in the presence of the polymer. NaBH4 is a strong reducing agent that produces ferrocene polymer-stabilized small AuNPs [59]. Thus, among the above-mentioned ferrocene polymers, several of them have been used in this way and applied to the AuNP-catalyzed nitrophenol reduction [105,106,107, 114].

6 Electron transfers with Metallodendrimers and Metallodendronic polymers

Forty years ago, the advent of the dendrimer field has boosted the area of macromolecules [115,116,117,118,119]. In this context metallodendrimers have been privileged given their application in molecular electronics, sensing, biomedical areas, and catalysis [120,121,122,123,124,125,126]. Redox-active metallodendrimers have included Ru(II)-polypyridine dendrimers largely developed in particular by Balzani’s group; these metallodendrimers can not only be oxidized to their Ru(III) counterparts but also harvest visible light and eventually direct photons like antennas [120]. The other metallodendrimer family of interest is that of ferrocene and other transition-metal-sandwich dendrimers due to the chemically and electrochemically reversibility of ferricenium/ferrocene couple and other related systems in these dendrimers. Reviews are available for the syntheses, structures and electrochemical studies of this very abundant family of macromolecules with ferrocene groups located at the core, branches or peripheries of dendrimers [121, 127, 128]. Two areas involving this redox FeIII/FeII system are of interest here, that of redox recognition/redox sensing and that of molecular electronics and its applications in the materials and biomedical field. The first ferrocene dendrimers as well as their redox sensing properties towards oxo-anions were reported in 1996. This included the very remarkable positive dendritic effect, i.e. the magnitude of the sensing effect of the HSO4 and H2PO4 anions by amidoferrocenyl groups of amidoferrocenyl-terminated dendrimers was all the more important as the dendrimer generation was higher [129, 130]. Three interactions were involved in this positive dendritic sensing effect: (i) the electrostatic attraction between the ferricenium group and the sensed anion (in the monomer; this interaction provokes only little recognition); (ii) the supramolecular interactions (double hydrogen bonding) between the amido group and the anion, and (iii) the dendritic effect encapsulating the anion that is key effect in synergy with the two other effects (Fig. 3).

Fig. 3
figure 3

The three factors acting in synergy explaining the recognition and sensing by amidoferrocene-terminated dendrimers of oxo-anions (HSO4, H2PO4 and ATP2−) and the remarkable positive dendritic effect in anion redox sensing [130]

The sensing effects were pursued using these powerful principles with various iron-sandwich-terminated dendrimers for halide anions [131] and ATP anion with gold nanoparticle-cored stars [132] and covalent [133] or even hydrogen-bonded (supramolecular) dendrimers (Fig. 4) [134].

Fig. 4
figure 4

Sensing of H2PO4 by the supramolecular metallodendrimer: Titration using cyclic voltammetry in CH2Cl2 (Pt, 0.1 M [nBu4N][PF6], reference FeCp*2 (Cp* = η5-C5Me5) of the supramolecular amidoferrocene dendrimer by [nBu4N][H2PO4]: (a) before addition of [nBu4N][H2PO4]; (b) after addition of 0.4 equiv.; (c) 0.5 equiv. The dramatic intensity drop at the equivalent point shows the sharp decrease of the diffusion coefficient reflecting the sudden formation of the large supramolecular assembly proposed here in which each H2PO4 group binds two amidoferrocene groups corresponding to the observed stoichiometry [134]. Copyright 2003 American Chemical Society

Dendrimer syntheses in our research group were based on the general original one-pot CpFe+-induced polyfunctionalization of polymethylarenes by a series of multiple benzylic deprotonation/allylation sequences applied to arenes bearing various number of methyl groups around benzene [135,136,137,138]. In these syntheses, the CpFe+-induced nona-allylation of mesitylene was particularly useful, because it directly introduced 1➔3 branching that was further continued upon dendrimer construction based on related tribranched dendrons [136, 137]. Whereas 1 ➔ 2 branching is dominant in dendrimer chemistry [115, 117,118,119,120,121,122,123,124,125], 1 ➔3 branching, pioneered by Newkome with his [27]-arborol [139, 140], is powerful both in terms of much more rapid dendrimer construction as for its encapsulation properties that are essential in sensing and catalysis with dendrimer templates. Triferrocenyl dendrons were designed in order to form, by 1 ➔ 3 connectivity, ferrocenyl-terminated dendrimers upon Williamson or Cu(I)-catalyzed click chemistry [141,142,143]. Alternatively, azido-terminated dendrimers were “clicked” with alkynyl ferrocene or alkynylcobalticenium to synthesize triazolylmetallocene-terminated dendrimers that proved to be excellent oxo-anion and metal cation electrochemical sensors due to selective supramolecular interactions (Scheme 4) [144,145,146].

Scheme 4
scheme 4

Syntheses starting from ferrocene and mesitylene of 4 generations (9-Fc, 27-Fc, 81-Fc and 243-Fc) of triazolylferrocene-terminated dendrimers. These metallodendrimers cleanly react with aq. HAuCl4 producing large, robust water-soluble gold nanoparticles (AuNPs) stabilized by the ferricenium chloride dendrimers [144].

Electron tunneling from the ferrocene centers through various dendrimers has been demonstrated [147, 148] as well as electronic wiring through isolating nanochannels involving molecular sieving with vertically aligned mesoporous silica films [149].

Despite the relative stability of ferricenium, long-term decomposition of its derivatives for instance in the presence of air and in solution still poses problems (especially with electron-withdrawing linkers), and indeed the oxidation of ferrocene dendrimers has rarely been addressed otherwise than in electrochemistry studies. An excellent solution to stabilize ferricenium, however, is the use of the η5-C5Me5 ligand (Cp*) instead of the parent η5-C5H5 ligand (Cp). Cp* is an extremely efficient stereoelectronic protector in overall organometallic chemistry. Several generations of pentamethylamidoferricenium dendrimers with 1➔3 connectivity have been isolated upon oxidation of their neutral counterparts using ferricinium hexafluorophosphate and characterized and are stable in air even in solution [143]. Giant ferrocenyl- and pentamethylferrocenyl with 1➔3 connectivity up to about 15,000 redox termini have been synthesized under both their Fe(II) and Fe(III) forms (ferricenium dendrimers are obtained by acetylferricenium oxidation of ferrocene dendrimers). Their electrochemistry shows full chemical and electrochemical stabilities, which means that electron transfer among the redox centers and to the electrode proceeds by ultra-fast electron hoping among the redox sites and/or dendrimer rotation occurring faster than the electrochemical time scale. AFM shows that the dendrimer size reversibly increases by about 50% upon oxidation, probably due to repulsion among the multiple cationic ferricenium termini- a type of molecular “breathing” or molecular machine [150]. The full stability of such pentamethylferricenium dendrimers shows that these systems could be used in particular for applications to molecular batteries [151,152,153] and nanomedical devices [154,155,156] (Fig. 5).

Fig. 5
figure 5

Reversible “breathing” (about 50% size increase) of giant ferrocene-terminated dendrimers acting as a molecular machine upon oxidation to polycationic Fe(III) dendrimers due to mutual repulsion of positively charged redox centers in the Fe(III) form

Another way to oxidize ferrocene dendrimers or other ferrocene-containing macromolecules is to conduct oxidation with Au(III) or Ag(I) [96, 98, 110, 111, 157,158,159]. In this case Au(III), for instance, is reduced to Au (0) nanoparticles of relatively large size that are well encapsulated and protected by the ferricenium dendrimers formed in the redox reaction as reported in 2014 [157, 159]. In a recent study, triazolylferrocene dendrimers up to 243 peripheral redox centers for which this number was demonstrated to be 255 ± 25 using the Bard-Anson theory were shown to reduce aqueous HAu(III)Cl4 to purple water-soluble ferricinium chloride-stabilized AuNPs of 30 to 47 nm size. The later was reduced by cobaltocene to ferrocene dendrimer-weakly stabilized AuNPs from which the smaller dendrimers were extracted together with AuNP agglomeration; whereas, the 243-ferrocene dendrimer resisted extraction. This experiment demonstrated the specific ability of ferricenium dendrimer to strongly stabilize AuNPs and AgNPs for catalysis and antimicrobial activities respectively, although click ferrocene dendrimers still provided some stabilization through triazole coordination to the AuNP surface. Most if not all late transition-metal nanoparticles have been stabilized by click dendrimers and used for such functions (Scheme 5).

Scheme 5
scheme 5

Electron flow from and to metallodendrimers. Oxidation of triazolylferrocene-terminated dendrimer by HAuCl4 yielding water soluble triazolylferricenium chloride dendrimer-stabilized gold nanoparticles. Further reduction by cobaltocene regenerates cobaticenium and ferrocene-terminated dendrimers stabilizing the AuNP only in the case of the 243-ferrocene dendrimer [158]. Copyright 2019 Wiley-VCH.

Ferrocene dendrimers, as well as other dendrimers [160], also stabilize small Au NPs if NaBH4 is used in the presence of the dendrimer to reduce HAuCl4 to small AuNPs, the driving force of NaBH4 being considerably superior to that of ferrocene derivatives for Au(III) and Ag(I) reduction, which explains the large difference of AuNP and AgNP size obtained. Likewise, analogous processes are achieved using click ferrocene-containing polymers and dendronized ferrocene polymers [105, 161,162,163,164,165,166,167]. The advantage of the stabilization of small AuNPs is that they are much more efficient as catalysts [114, 168,169,170] for nitrophenol reduction by NaBH4 to aminophenol than large AuNPs formed using only a ferrocene reductant [110, 111]. In these “clicked” dendrimers, the Au(III) and other cation stabilization by the triazole ligand is the key to further intradendritic small AuNP (and other metal NP) stabilization, because the triazole ligand stabilize both the metal cation and (more weakly) the NP [171,172,173,174,175]. A typical example is provided by Wang’s work whereby similar-sized click and non-click dendrimers where compared for Au(III) reduction with AuNP formation: with the click dendrimer small AuNPs were encapsulated within the dendrimer interior, whereas with the non-click dendrimers large AuNPs were stabilized outside the dendrimer by several dendrimers (Fig. 6) [159].

Fig. 6
figure 6

AuNP syntheses from metallodendrimers. Metallodendrimers containing 54 Cp*Fe(II) dppe redox termini were synthesized by two methods: click coupling between azido-terminated dendrimers forming triazole linkers, (a) and Sonogashira coupling that did not provide such intradendritic ligands (b). Two reactions of HAuCl4 were conducted with each of the two dendrimers: 1) reaction with the neutral dendrimers oxidizing them to class-2 mixed-valent polycationic dendrimers; 2) reaction with NaBH4 in the presence of these class-2 mixed-valent polycationic dendrimers. These two reactions produced AuNPs of closely related sizes, but the click dendrimers yielded small AuNPs inside the dendrimer (1.9 resp. 1.4 nm), whereas the non-click dendrimers yielded large AuNPs (3.7 resp. 3.4 nm). This series of experiments clearly shows the Au(III)-triazole coordination and intra-dendritic triazole stabilization of small AuNPs, whereas these phenomena are not occurring with non-click dendrimers that stabilize large AuNPs only outside dendrimers. Reprinted with permission of ref. [159]. Copyright 2015 Nature publishing group

7 Concluding remarks

In conclusion, most electron-transfer processes in metallodendrimers can be compared to some extent to those occurring in metalloproteins [176,177,178,179] and neuronal networks [180]. The most current redox centers used in dendrimers for this purpose have been included in this mini-review. Besides the most commonly encountered ferrocene and related iron-sandwich systems, some other redox-active groups such as cobaltocene, metalloporphyrins, iron-sulfur clusters, ruthenium polypyridines and methylviologen have also sometimes been used in metallomacromolecules [14]. Iron sandwich complexes have been most used and are most useful because of the robustness of the redox systems whose structural design has been optimized for this purpose and due to the ease of their functionalization for linking to macromolecules. Electron-transfer properties found for the monomers can be advantageously extended to macromolecules. This is striking for the sensing properties for which the confinement indispensable inside the network allows boosting redox recognition with a remarkable positive dendritic effect. Electrostatic effects are shown to depend on the design and increase enormously when they are dealing with peripheral chargeable cationic redox centers, i.e. in the presence of large counter anions. Starting with organoiron electron-reservoir complexes applications range from electrochemical references and stoichiometric redox reactions to electrocatalysis (electron-transfer-chain catalysis) and redox mediation (outer-sphere) or redox catalysis (inner sphere). Single 2-electron transfer catalysts are desired for instance in H2 evolution processes. On the molecular level, they are accessible with chemically reversible systems in which the electrostatic energy required for the second electron transfer is compensated by the reorganization energy occurring in this step. In hexanuclear systems the electron-transfer mechanisms are intimately governed by the structure and electrostatic effect. In large and giant metallodendrimers in which the redox groups are located at the branch termini, the electronic and electrostatics effect can become so weak that they are insignificant, and transfer of up to 15,000 electrons in a single reversible CV was indeed observed in the absence of adsorption. Applications are not only the mimicry of electron-transfer proteins, but also more practical ones such as redox sensors, actuators, electrochromes, redox catalysts, catalyst templates, switchable drug delivery systems and batteries. In the metallodendritic assemblies around redox centers, supramolecular interactions such as multiple hydrogen bonding, electrostatic effects, weak coordination, π-π stacking are playing a key role [15, 181,182,183,184,185,186,187,188,189,190].