Mohammad
Zafar
,
Vasudevan
Subramaniyan
,
Françoise
Tibika
and
Yuri
Tulchinsky
*
Institute of Chemistry, The Hebrew University of Jerusalem, Jerusalem 9190401, Israel. E-mail: yuri.tulchinsky@mail.huji.ac.il
First published on 14th June 2024
For a long time, the small group of cationic ligands stood out as obscure systems within the general landscape of coordinative chemistry. However, this situation has started to change rapidly during the last decade, with more and more examples of metal-coordinated cationic species being reported. The growing interest in these systems is not only of purely academic nature, but also driven by accumulating evidence of their high catalytic utility. Overcoming the inherently poor coordinating ability of cationic species often required additional structural stabilization. In numerous cases this was realized by functionalizing them with a pair of chelating side-arms, effectively constructing a pincer-type scaffold. This comprehensive review aims to encompass all cationic ligands possessing such pincer architecture reported to date. Herein every cationic species that has ever been embedded in a pincer framework is described in terms of its electronic structure, followed by an in-depth discussion of its donor/acceptor properties, based on computational studies (DFT) and available experimental data (IR, NMR or CV). We then elaborate on how the positive charge of these ligands affects the spectroscopic and redox properties, as well as the reactivity, of their complexes, compared to those of the structurally related neutral ligands. Among other systems discussed, this review also surveys our own contribution to this field, namely, the introduction of sulfonium-based pincer ligands and their complexes, recently reported by our group.
In a seminal paper, Sidgwick and Bailey2 pointed out the chemical similarities between carbonyl and nitrosyl complexes and suggested that in these complexes nitric oxide species must be treated as an isoelectronic cationic analogue (I) of the neutral carbonyl and anionic cyanide ligands (II and III, Chart 1). A few decades later, Wilkinson3 and Raynor4 dissipated the remaining doubts about the existence of a nitrosyl ligand as a positively charged species, based on infrared spectrum analysis and orbital considerations, respectively.
These studies incontestably showed that the description of a ligand merely as an electron donor to a metal center that acts as an “electron sink” is over-simplified. In fact, transition metals with occupied d-orbitals can behave as electron donors as well. Therefore, today it is unanimously accepted that ligand–metal bonding is due to a combination of donation and back-donation interactions. While coordination of the more common neutral and anionic ligands mostly relies on the L → M donation further enhanced by various degrees of M → L back-donation, with cationic ligands, it is the latter interaction that becomes essential for the bonding.
Generally speaking, most cationic ligands belong to either one of the following two classes – the Z-type ligands (i.e. σ-acceptors, Chart 2a)5 or the more traditional L-type ligands that exhibit poor σ-donor, but strong π-acceptor character (Chart 2b).6 Since even the most electron-rich metal centers are rather poor Lewis bases, the M → L σ-bonding in Z-type complexes is inferior to the conventional L → M σ-donation of traditional L-type ligands. As for the L-type cationic ligands, in most cases, their enhanced π-acidity is still insufficient to fully compensate for their reduced electron donicity. Thus, both classes of cationic ligands often exhibit low binding affinities and small dissociation barriers.
To achieve higher kinetic and thermodynamic stability, it is possible to take advantage of the well-known chelate effect. Here the most commonly employed strategy is flanking a positively charged ligand (E+) with a pair of strongly donating groups (D), such as phosphines, pyridines, aryls, etc., thus constructing the corresponding PE+P, NE+N, and CE+C pincer scaffolds (Chart 2c). In fact, some of the cationic species discussed below, for instance, N-heterocyclic nitrenium or quaternary stibonium cations, show coordinative behavior only when incorporated within such pincer frameworks (see Sections 3.1.2. and 3.3.1., respectively). In these cases, formation of a coordinative bond between the cationic donor moiety and a metal ion or atom is induced by their mutual proximity due to the geometric constraints imposed by the pincer backbone.
The significance of pincer ligand platforms in modern coordination chemistry can hardly be overestimated. Since their introduction by Shaw in the 1970s,7 pincer ligands and their complexes have become ubiquitous in a broad array of chemical fields ranging from chemo-sensing8 and materials science9 to homo-10 and heterogeneous11 catalyses. Due to their high thermal and chemical robustness, pincer scaffolds have found extensive applications as supporting ligands in catalysis, and their structural rigidity has often been found to be indispensable for isolation of elusive species and fleeting reaction intermediates.12 Indeed, very weak ligand–metal interactions, such as agostic C–H13 and C–C14 interactions, along with other unusual types of ligand–metal bonding interactions,15 were observed in pincer complexes. Furthermore, the stabilizing effect of the pincer backbone also allowed isolation of open-shell species of main group elements,16 as well as transition metals exhibiting atypical coordination geometries17 or oxidation states.18
In traditional pincer ligands, the central donor atom is negatively charged or neutral (as in the iconic PCP and PNP pincers, respectively). Pincer platforms featuring a positively charged central donor moiety are still quite rare, although the number of such systems has significantly increased during the last decade. To date, there is a lack of a systematic and comprehensive review of the recent advances in this field. Several reviews on specific types of simple cationic ligands (e.g. α-cationic phosphines)6a or their complexes (e.g. metal-stabilized carbocations)19 have been published, but to the best of our knowledge, no systematic survey dedicated to pincer-type cationic ligands is available. The purpose of this review is to fill this gap, while the pool of such ligands, though rapidly expanding, can still be surveyed within a single article.
To begin with, the very definition of a ligand as a cationic one can sometimes be debatable. This is due to the ambiguity in allocation of the positive charge on the ligand and assigning the oxidation state of the metal center accordingly. In this context, complexes of nitric oxide (NO) are perhaps the most illustrative examples.20 Indeed, the NO ligand can adopt a coordination mode lying in between two idealized cases – the so-called linear nitrosyl (or nitrosonium) with an sp hybridized N atom associated with the formula of NO+ and the so-called bent nitrosyl (or nitroxyl) featuring an sp2 hybridized N atom associated with the formula of NO− (Chart 3a). In addition to the M–N–O angle, linear and bent nitrosyls also differ in their NO bond lengths and the corresponding NO vibration frequency in IR (see Table 1). Therefore, in principle, the charge of the NO ligand in a given metal–nitrosyl complex can be assigned according to its structural or spectroscopic data.20a
Yet, in multiple cases, there is a rather poor correlation between the M–N–O angle and the corresponding N–O bond length (as well as the NO vibration band) and so numerous nitrosyl complexes fall into a grey area, where the formal assignment of ligand charge and consequently the oxidation state of the metal is ambiguous. This gave rise to the Enemark–Feltham notation,21 in which to avoid this ambiguity, the metal and NO are treated together as a single entity, {MNO}Fn, where n indicates the total number of valence electrons.
Appropriate charge assignment can be no less confusing also for the organic cationic ligands discussed in this review, for instance, the phosphenium cations.22 Similar to nitrosyls, they also acquire two binding modes differing in their formal charge – a cationic phosphenium and an anionic phosphide (Chart 3b). The former is usually associated with a trigonal planar geometry at phosphorus and relatively short M–P bonds with a considerable degree of π-bonding, whereas the latter is characterized by a pyramidal geometry with relatively long P–M bonds (Chart 3b). In the absence of an XRD structure, 31P NMR can often be used as a reliable tool for establishing the coordination modes of these ligands, as sp2-hybridized P-atoms in phosphenium species exhibit a huge downfield shift compared to the sp3-hybridized ones in phosphide complexes (Table 2). Unfortunately, this rule of thumb is poorly applicable to the complexes of N-heterocyclic phosphenium species (NHPs), to be discussed in Section 3.2 of this review, since irrespective of the coordination geometry their 31P NMR signals all fall within the same region of 200–300 ppm.
All that being said, for most of the ligands discussed below, attribution of a positive charge to the ligand itself is the most reasonable and sometimes even the only possible choice, in terms of the most accurate representation of electron density distribution within the corresponding complexes.
In the reported cationic ligands, the coordinating atoms all belong to only three groups within the p-block elements, namely, groups IVA, VA, and VIA (Fig. 1). This is not surprising, as most organic compounds of group IIIA elements are Lewis acidic, and hence probably already too electron deficient to become cationic. On the other side of the block, metal-coordinating atoms of group VIIA elements, whether as simple ions or polyatomic halides, tend to have a formal negative charge.
Fig. 1 Position of metal-coordinating elements of the reported cationic ligands in the periodic table. |
In this review, all cationic ligands are organized according to the position of their metal-coordinating elements in the periodic table (Fig. 1). Only cationic species acting as a central moiety of a pincer ligand will be discussed, while those never embedded in a pincer scaffold (including the existing Si-, Sn-, and As-based cationic ligands, as well as the neutral ligands with appended cationic groups) will remain beyond the scope of this review.
Fig. 2 Synthesis and reactivity of a σ-arenium Pt(II) complex 3 (a) and its X-ray structure (b) as a metallated analogue of a Wheland intermediate (c). |
Related Rh(I)-metalated σ-arenium species were later studied in depth by Milstein in PCP pincer systems.27 He found that when an electron-deficient (due to the presence of an electron-withdrawing CO ligand) cationic Rh(I) precursor reacts with a PCP pincer 5, no oxidative addition of C–H bond (into 6) takes place; instead, a stable Rh(I) complex 7 is isolated and characterized by XRD and NMR (Fig. 3a).27a The bonding between the metal center and the aromatic ring in this complex can be described by two resonance structures – either as a dearomatized σ-bonded arenium species (i) or as a neutral arene with a close η2-C–H agostic interaction (ii). The observed short H–Rh distance of 1.950 Å and a large value of the 2JRhH coupling constant of 18.1 Hz were more in favor of the second structural description. Computational studies further confirmed that the contribution of the σ-arenium resonance form is quite small. Thus, complex 7 can be regarded as a rare example of a C–H σ-complex, a “frozen” intermediate on the way toward an oxidative addition of a C–H bond. In a subsequent study,27b a Cipso-alkylated analogue of complex 7 was also prepared (9, Fig. 3b). Although this Rh(I) complex appears to be isostructural with the σ-arenium complex of Pt(II) (3, Fig. 2a), its reactivity did not show the typical reactivity of arenium cations, being inert towards nucleophiles. Thus, similar to the Cipso-protonated complex 7, 9 can be better described as an agostic η2-C–C alkyl-arene (ii), rather than a σ-arenium complex (i) (Fig. 2b).
Fig. 3 Synthesis and resonance structures of the cationic σ-arenium Rh(I) complexes 7 (a) and 9 (b). |
While σ-arenium structures discussed above are all derived from dearomatized carbocycles, formation of fully aromatic σ-bonded arenium pincer complexes is also possible upon expanding the six-membered ring into a seven-membered one. Such tropylium- and benzotropylium-based pincer complexes were obtained by Mayer via hydride abstraction from the corresponding cycloheptatrienyl derivatives (Fig. 4a).28
Fig. 4 Synthesis of tropylium and benzotropylium Ir(III) complexes (a) and comparison of their X-ray structures (b). |
As the hydride abstraction at the ipso position of the carbocycle induces aromaticity, it is preferred to hydride or chloride abstraction from the Ir(III) center. As a result, the central carbocycle switches from a puckered to planar geometry (Fig. 4b), which also manifests in the drastic downfield shift of the 13C NMR signal of the metalated carbon from δ = 46.1 ppm in 10a to δ = 213.3 ppm in 11a. Once formed, complex 11a is surprisingly stable under air as a solid and in solution and can even be refluxed for hours in acetone. However, it readily reacts with excess NaOH, undergoing deprotonation at one of its chelating arms into the asymmetric complex 12 followed by reductive elimination of HCl to form the Ir(I) complex 13.28b The latter two transformations were found to be fully reversible, and thus the tropylium complex 11a can be reformed by treating 13 with HCl (Fig. 4a). Later on, a benzannulated analogue of 11a was also prepared (11b);28c however, no further data on its reactivity were provided. Interestingly, the CO stretching frequency of complex 12 at 2000 cm−1 is identical to that of complex 10a, which shows that the absence of a proton at the ipso carbon has little electronic effect on the metal center. Conversely, the presence of a positive charge on the carbocycle in 11a results in a clear blue shift of its CO frequency to 2030 cm−1.28b
Strictly speaking, classifying the above σ-coordinated arenium species (Chart 4a) as cationic ligands is debatable. To fit into this description these ligands should be regarded as single electron donors (X-type ligands), which implies that the C–M σ-bond in their complexes is covalent. However, in most organometallic compounds C–M bonds are strongly carbon-polarized, and therefore alkyl and aryl ligands are usually classified as monoanionic two electron donors (L-type ligands), isovalent with halides. Following this convention, σ-bonded arenium cations, such as those shown in Fig. 2–4, should not be regarded as cationic, but rather as mesoionic neutral species (Chart 4b).
On the contrary, treating π-coordinated arenium species (Chart 4a) as cationic ligands is less arguable. The tropylium cation mentioned above, for instance, when η7-(rather than η1) coordinated is a classic 6π-electron donor, isoelectronic with benzene and cyclopentadienyl anions. Similarly, a π-coordinated cyclopropenylium cation acts as a 2π-electron donor (Chart 4a). The first such η7-tropylium and η3-cyclopropenylium complexes with electron rich Mo(0) carbonyl fragments, 14 and 15, were prepared by hydride or chloride abstraction in the 1950s and 1960s, respectively (Fig. 5a and b).29 Surprisingly, even when σ-metallated, the tropylium cation retains its ability to form π-arene complexes. Indeed, upon treatment of the previously discussed Ir(III) complex 11a (Fig. 4a) with (η6-p-xylene)Mo(CO)3 an arene exchange occurs and a highly unusual doubly-metallated η1-Ir-η7-Mo-tropylium pincer complex 16 is formed (Fig. 5c).28b To the best of our knowledge, no other examples of pincer complexes containing π-coordinated tropylium or cyclopropenylium cations have been reported.
Fig. 5 Tropylium (a) and cyclo-propenium (b) complexes of Mo(0) carbonyl and a heterobimetallic η1-Ir(III)-η7-Mo(0)-tropylium pincer complex 16 (c). |
Another class of cationic π-arenium cations, which perhaps are of higher relevance to pincer complexes, are the conjugated carbocations, and in particular the benzyl cation. The existence of metal coordinated benzyl cations was first suggested in 1964 by Holms,30 who showed that the solvolysis of complexed benzyl halides was 105 times faster compared to a free benzyl halide. This rate enhancement was attributed to the additional stabilization of the intermediate benzyl cation provided by its complexation to the Cr(CO)3 fragment (Fig. 6a). Attempts to isolate this transient complex by protonolysis of the corresponding Cr(CO)3-coordinated benzyl alcohol failed, but nevertheless, it was characterized in solution by UV-vis spectroscopy by Trahanovsky in 1969.31 Finally, a related Cr(CO)3 complex of a tertiary benzyl cation was isolated and characterized by NMR spectroscopy at low temperature by Olah (Fig. 6b).32 Theoretical studies by Hoffman and others showed that in this complex the arene ring exhibits an η6 coordination mode with its benzylic carbon bending towards the metal center, suggesting an attractive interaction between them (Fig. 6c).33
Fig. 6 Reactivity of a free and metallated benzyl chloride upon hydrolysis (a) and the characterization of an η6-coordinated tertiary benzyl cation (b), along with its optimized primary analogue (c). |
Importantly, one of the resonance structures representing a benzyl cation consists of an arenium cation conjugated to an exocyclic alkene (ii, Chart 5a) and, as discussed above, it is this resonance form that best describes benzyl cations coordinated through their exocyclic methylene group (Chart 5c), rather than through their aromatic ring (Chart 5b). Such η2-coordinated methylene arenium species were studied by Milstein in Rh and Ir pincer systems34 along with the corresponding σ-arenium complexes discussed above. In a seminal paper34c his group reported the synthesis of methylene arenium complexes starting from quinone methide (17) and xylylene (19, 20) pincer complexes by protonation or silylation of the corresponding carbonyl or methylene groups (Fig. 7a and b). Formation of these cationic moieties was apparent from shortening of the Rh–C bond lengths compared to the neutral complexes (for instance, 2.183(5) Å in 18vs. 2.229(4) Å in 17), as well as the drastic increase in the CO stretching frequency of the corresponding carbonyl complexes (2060 cm−1 for 23vs. 2019 cm−1 for 22), due to the increased π-back donation from the metal to the cationic ligand (Fig. 7c).34c
Chart 5 Resonance structures of a benzyl cation (a) and its two possible coordination modes – η6 (b) and η2 (c). |
Comparison of the 13C NMR spectra of complexes 18 and 21 shows that changing the substituent at the para-position (to the methylene) from the strongly electron-donating OH group to the mildly donating CH3 group has the expected effect on the ring carbon signals, shifting them to a lower field, but does not influence the chemical shift of the methylene group. This lack of conjugation validates the description of the η2-coordinated benzyl cation by the methylene arenium resonance structure (i.e., ii on Chart 5c), rather than by the aromatic structure with a metal-localized positive charge.34b,c
Subsequently, the same group also developed a general strategy for the preparation of such complexes via acid promoted metal-to-ligand methyl migration.27b Presumably, this transformation occurs via β-H elimination from a bis-cationic σ-arenium intermediate 25a and consequent hydride abstraction by acid. The resulting π-arenium complex 26 is found to be surprisingly stable both in the solid state and in solution; however, similar to the tropylium complex 11a, in the presence of a base this complex undergoes reversible deprotonation at one of the chelating “arms”, resulting in an asymmetric complex 27 (Fig. 8a).
Fig. 8 Synthesis and reactivity of methylene arenium complexes of Rh and Ir (a) and (b), XRD structures of complexes 27 (M = Rh) and 9* (c) (as the original XRD data of 9 are unavailable, partial structure of an isostructural complex with an ethyl derivative, 9*,27b is shown here). |
The η2-coordinated π-arenium of an analogous complex 28 can be easily converted into the previously mentioned η1-σ-arenium complex 9 (Fig. 3b) by reaction with H2, which acts as a hydride donor. Interestingly, unlike in 26, deprotonation of 9 does not occur at one of its “arms”, but rather at the Rh-methyl group, which results in the formation of an aromatized Rh(I)-benzyl complex 29 (Fig. 8b). In fact, the initial methylene–arenium complex 28 can in principle also be represented as a benzyl derivative of a bis-cationic Rh(III) center (resonance structure ii). However, unlike for 9, the contribution of this resonance form is negligible, due to the presence of a highly uncompensated positive charge on the metal in this bis-cationic species.
Fig. 9 Formation of a protonated CDP (a), its possible resonance structures (b), and the first application as a ligand in a tris-cationic Ag(I) complex (c). |
Mono-protonation of such unusual compounds generates a cationic species with only one remaining lone pair, available for coordination, as a pure σ-donor (Fig. 9a). In this respect, protonated CDPs are quite different from all other ligands discussed herein, as they lack any M → L back donation. Interestingly enough, calculations show that in spite of the protonation, the carbon center in these species still bears a significant negative charge (resonance structure iv in Fig. 9b). This probably explains the surprising stability of the tris-cationic Ag(I) complex 29, reported by Frenking (Fig. 10c).37a Upon coordination the central carbon atom of the monoprotonated ligand undergoes a substantial geometry change, switching from a resonance-stabilized planar sp2 to a pyramidalized sp3 configuration (Fig. 9c).37
After neutral CDPs have been embedded in a pincer framework, resulting in stable pincer complexes with an M–C bond,39 pincer complexes of protonated CDPs have also been prepared either by protonation of the corresponding complexes of the neutral CDP ligand 30a (Fig. 10a and b)40 or by a direct-coordination of a protonated CDP ligand 30b (Fig. 10c).41 In these complexes, the strength of the interaction between the metal center and the protonated C(0) atom is quite variable, as can be estimated not only from its distance to the metal, but also from the extent of its pyramidalization. This is easily noticeable when comparing different M–Cl complexes of the same protonated carbodiphosphorane pincer ligand 30b (Fig. 10a). For instance, in the Au complex 33c (Fig. 10c) this interaction is nearly absent (C–Au = 2.961 Å) and the central carbon atom maintains its trigonal planar geometry (Fig. 10e).41a Likewise, in the analogous Cu complex (33a) a relatively long C–Cu = 2.304(3) Å bond was found, exhibiting only a slight pyramidalization of the carbon atom (estimated from the sum of the bond angles of 352° around the carbon atom).41f On the other hand, in the corresponding complexes of group 10 metals the M–C bond of 2.1 Å lies within the range of typical M–Csp3 covalent bonds (M = Pd, Pt, Fig. 10b).40a,c Accordingly, a pronounced pyramidalization of the central carbon atom, which is similar to or smaller than that for an ideal sp3-hybridized carbon, was found in the XRD structures of those complexes (Fig. 10d).
These geometrical changes upon coordination, apparent from the available XRD structures, are also manifested in the 1H and 13C NMR spectra in solution. In the free protonated CDP, the CH+ moiety resonates at +1.6 ppm and −3.3 ppm in 1H and 13C NMR, respectively.40c However, when the ligand is also involved in an M–C bond (M = Pd, Pt), it exhibits downfield shifts of 4–5 ppm in the 1H NMR and 10–20 ppm in 13C NMR, in parallel to the pyramidalization.40b,c Interestingly, these large downfield values are comparable to the signals obtained for the doubly protonated CDP ligand 30c (Fig. 10a).40c
Yet, in some complexes of protonated CDPs these signals are not always observable.41e This is because of the high lability of the CH+ proton due to its pronounced acidic character. In this respect, this system resembles a coordinated secondary amine that can be deprotonated into the corresponding amide, leading in that case to a negatively charged ligand. Moreover, this acidic CH+ proton can also participate in hydrogen bonding with a suitable ancillary ligand (Cl, OAc, etc.), when the geometry allows it, such as in tetrahedral group 11 complexes. In several cases, this interaction even becomes predominant over the C–M bonding.41f
With the electron rich Ir(I) center, this acidic proton undergoes oxidative addition, resulting in an Ir(III) hydride complex of a neutral CDP (37, Fig. 11a), as this labile proton is transferred to the metal.41e Notably, an Ir(III) complex of a mono-protonated CDP could also be obtained as a mixture of two stereoisomers, 39a and 39b, by an analogous reaction between the same Ir(I) precursor and a doubly-protonated CDP ligand 30c (Fig. 11b).41b With the less electron rich Rh(I) center, however, such an oxidative addition was not observed; instead, as in the previous cases, a Rh(I) complex 38 of a mono-protonated CDP was formed.41e
The ancillary carbonyl ligands present in the above Ir(III) and Rh(I) complexes, as well as in the related Re(I) carbonyl complexes 35a,b (Table 3) prepared by the same group,41c offer the possibility to use IR spectroscopy for estimating the change in the donor ability of the CDP upon its protonation. For the Ir(III), Rh(I) and Re(I) complexes, the νCO IR vibration frequencies of the protonated CDP were found be higher by about 30–50 cm−1 relative to the values of the non-protonated CDP ligands, supporting the electron withdrawing character of the protonated CDP ligand (Table 3).
Interestingly, when the structures of both neutral and protonated CDP complexes of the same metal fragment are available, a rather small elongation of the M–C(0) bond can be observed for the cationic ligand compared to the neutral one, in spite of these large shifts in the IR frequencies. As evident from Table 3, in the Pd(II), Pt(II), and Re(I) systems, this elongation does not exceed 0.06 Å. This might suggests that loss of CDP's π-donor ability upon its protonation does not significantly compromise the strength of the M–C(0) bond.
Soon after the elucidation of the carbodiphosphorane structure, Frenking et al. suggested that rather than with two phosphines, a zero-valent carbon center can also be stabilized by a pair of N-heterocyclic carbenes, challenging synthetic chemists to prepare such a compound.43 This expectation was met a year later by Bertrand who reported the preparation of not only such a carbodicarbene (CDC), but also its complex with Rh(I).44 Similar to the CDPs, later on CDC species were incorporated within pincer systems giving rise to novel coordination compounds with applications in homogeneous catalysis.42
Continuing the analogy to CDPs, one might expect to find quite a few complexes of protonated CDCs, yet only a single Rh(I) complex of such a species (36b, Table 3) has been reported so far.42 Like with the CDP complexes, in carbonyl complexes protonation of the coordinated CDC ligand leads to a significant shift (+30 cm−1) of νCO in the IR spectrum (Table 3). Moreover, the electron withdrawing character of a protonated CDC is also evident from the catalytic activity of its Rh(I) complex as a π-acid catalyst. Although hydroarylation of alkenes is catalyzed by a Rh(I) pincer complex of a neutral carbodicarbene, 36a, the protonated complex 36b exhibited a much higher catalytic performance42 (Table 3).
Interestingly, Langer et al.45 reported the existence of a doubly-protonated bis(phosphine) boronium cation 40, which is isoelectronic to the monoprotonated CDP. This species could be complexed with Pd(II) or Ir(I) through activation of one of its B–H bonds. However, once coordinated, the deprotonated boron center of this pincer ligand is no longer cationic (41, Fig. 12).
Fig. 13 Synthesis of Ge(II) cation complexes (a) and complexes of Ge(II) cations stabilized by chelating frameworks (b). |
Such complexes of cationic Ge(II) species acting as L-type ligands were first reported in 2005 by Barrau in the form of a β-diketiminato-stabilized Ge(II) dication coordinated to a W(CO)5 fragment (Fig. 13a).47a Yet, even when coordinated, the Ge(II) center maintains its strongly Lewis acidic character and is capable of activating inert bonds48b or binding weakly coordinating anions, such OTf (42).47a As a consequence, these complexes could only be isolated when stabilized by an additional donor molecule (43a,b).47a,48 Chelating frameworks, including pincer-like ones possessing flanking imine groups (Fig. 13b), have been employed for further stabilization of positively charged Ge(II) species in their coordination compounds.47b,48d However, no example of a mono- or bis-cationic Ge(II) species incorporated as a central coordinating moiety within a pincer scaffold is known.
It is, however, not the case with the cationic species of Ge(IV), the germylium cation (Chart 6b), which can be regarded as a heavy analogue of a tertiary carbocation. This species lacks a lone pair and therefore can only coordinate as a Z-type ligand. Such a bonding was indeed observed by Gabbai in Pt(II) and Au(I) complexes of cationic Ge(IV)-based pincer ligands, which have so far remained the only examples of metal-coordinated cationic Ge(IV) centers.49 These complexes, featuring prominent M → Ge(IV)+ donor–acceptor interactions, were not synthesized by reacting a cationic PGeP pincer ligand with metal precursors. Instead, they resulted from structural manipulations affecting the character and polarization of Ge–M bonding within the preexisting Ge-coordinated pincer complexes.
Complex 49, for instance, prepared by oxidative addition of dichlorogermane-based pincer ligand 48 to a Pt(0) precursor (Fig. 14a), contains a Ge–Pt σ-bond, which according to natural bond orbital (NBO) analysis, is of covalent nature, with nearly equal contributions of 47.63% for Ge and 52.37% for Pt.49a Yet, oxidation of this complex with PhICl2 results in a profound change in Ge–Pt bonding. Even though the Ge–Pt bond length in the resulting complex 50 shows only a minor elongation (2.389 vs. 2.334 Å), computational analysis suggests that the bonding in this complex is no longer covalent and is best represented by donor–acceptor interactions between a divalent trichloro-platinate anion and a chlorogermylium cation (resonance structure ii in Fig. 14b). In other words, upon oxidation of 50 the electron density within its Pt–Ge bond shifts towards the Pt center. Upon irradiation at 270 nm, this compound was found to undergo reductive elimination of Cl2, leading back to complex 49. Interestingly, elimination of Cl2 was not the only reaction observed for complex 50 under UV irradiation. In addition, it underwent a reversible chloride shift from Ge to Pt, converting into its structural isomer, 51. The latter complex, which could also be obtained directly by reacting ligand 48 with a [PtCl2] precursor, was prone to photoelimination of Cl2 too, and therefore, might have been formed as an intermediate during the photolysis of 50 (Fig. 14a).
Fig. 14 Synthesis and photo-reduction of the chlorogermylium(IV) complex 50 (a), its resonance structures (b), and the X-ray structures of complexes 49–51 (c). |
The photoisomerization of 50 into 51 leads to a significant weakening of the Ge–Pt interaction, evident not only from the significant elongation of the Ge–Pt bond to 3.195 Å, but also from the natural bond orbital (NBO) calculations, with second order perturbative energies dropping from E2 = 83.11 kcal mol−1 in 51 to E2 = 2.44 kcal mol−1 in 50. Thus, even though a neutral dichlorogermane can still be considered a Z-type ligand (with one of its σ* Ge–Cl bonds acting as an acceptor orbital), it binds much more weakly compared to the chlorogermylium cation. The PGeP pincer ligand in complexes 51 and 50 therefore provides a unique example of a Z-type ligand whose binding strength can be manipulated by irradiation.49a
Later on, the same group also prepared a related Au(I) complex of a triarylgermylium cation,49b starting from the chlorogermane-based PGeP pincer ligand 52a (Fig. 15a). Within the so-formed complex 53a the Ge(IV) center is neutral, and according to NBO calculation it acts as a weak Z-type ligand (E2 = 9.1 kcal mol−1). However, upon chloride abstraction from the Ge(IV) center, the Au–Ge interaction in the resulting mono-cationic complex 55a strengthens significantly (E2 = 46.4 kcal mol−1), which also manifests in a drastic shortening of the Au–Ge bond length from 3.030 Å to 2.428 Å. Among the three possible bonding descriptions of this complex (resonance structures i–iii in Fig. 15b) theoretical analysis suggested resonance form i, i.e. the Au(I) → Ge(IV)+ donor–acceptor interaction, as the most accurate representation.
Fig. 15 Synthesis of cationic Ge(IV) complexes of gold (a) and Z, X and L-type resonance structures of the triarylgermylium(VI) complex 55a (b). |
A dual chloride abstraction from both Ge and Au centers of 54a resulted in the formation of a highly Lewis acidic bis-cationic Au(I) complex 56a. Although its presence in solution was clearly detectable by NMR, this intriguing complex could not be isolated in the solid state, as it slowly converted into complex 55a or 58b upon reacting with the chlorinated solvent or the SbF6 anion, respectively (Fig. 15a).49b
Nevertheless, the highly electrophilic nature of the bis-cationic [Ge(VI)–Au(I)]2+ core of 56a could be successfully employed for π-acid catalysis. Indeed, this complex formed in situ showed an excellent performance as an alkyne hydroamination catalyst, capable of converting phenyl acetylene into an aromatic imine in only 20 min (Fig. 16). On the other hand, the activity of the mono-cationic complex 58b (independently prepared starting from a fluorogermane-based PGeP ligand 53b, Fig. 15a) was significantly lower and afforded this imine product only after 6 h, which is no different from the reactivity of a cationic Au(I) complex with triphenyl–phosphine.49b This comparison clearly demonstrates how accumulation of a positive charge on a metal by means of a cationic ligand enhances the activity of the resulting complex in electrophilic catalysis.
Fig. 17 Resonance structure of the N-methyl pyrazinium cation (a) and comparison of pyrazine and pyrazinium complexes with Zd/Cd (b) and Ru(II)/Ru(III) (c). |
The poor donor character of pyrazinium compared to pyrazine is clearly manifested in a series of complexes with Lewis-acidic metal centers, such as Zn(II) (59–60a) and Cd(II) (59–60b), where the outer d shell is too stabilized for an interaction with ligand's π* orbitals (Fig. 17).51 Thus, within two analogous complexes, the L–M bonds with a pyrazinium ligand51a are longer than with a pyrazine one51b,c (Table 4, the 1st and 2nd rows). However, in complexes of late transition metals, such as Ru, where the outer d electrons are available for back-bonding, pyrazinium exhibits shorter M–L bonds than pyrazine (Table 4, the 3rd and 4th rows).52 Moreover, it can even form a surprisingly stable Ru(III) complex, which to the best of our knowledge, is the only reported example of a cationic ligand binding a tris-cationic metal center lacking any structural support by a chelating framework.52a Here too, the M–L bond is shorter compared to the pyrazine analogue (Table 4, the 4th row).52b
The stronger π-acceptor character of the pyrazinium cation compared to pyrazine and other neutral N-heterocycles can also be observed by optical spectroscopy, for instance, in a series of pentaamine Os(II) complexes (Table 5).53 Their optical spectra are dominated by two bands, both corresponding to MLCT transitions.53a Only a single metal d orbital (the one perpendicular to the heterocycle plane) has a suitable symmetry and orientation for π-bonding, forming a dπ orbital, while the other d orbitals remain non-bonding (nd) (Fig. 17a). Therefore, the two observed transitions are dπ → dπ* and nd → dπ* (Fig. 18b).
L | [L(OsII)(NH3)5]2+ | [L(OsII)(NH3)4(N2)]2+ | |
---|---|---|---|
λ 1 (nm) | λ 2 (nm) | ν NN (cm−1) | |
Pyridine | 428 | 555 | No data |
Pyrimidine | 458 | 663 | No data |
Pyrazine | 460 | 770 | 2040 |
N-Methyl pyrazinium | 435 | 1150 | 2098 |
Fig. 18 Graphic representation of the dπ, dπ* and nd molecular orbitals of pyrazinium complexes (a) and the corresponding electron transitions observed for [L(OsII)(NH3)5]2+ complexes (b). |
The energy of the dπ → dπ* transition (ΔE1) is only slightly affected by the nature of the heterocyclic ligand; conversely, the nd → dπ* transition energy (ΔE2) is highly sensitive to the energy of ligand's LUMO. This is because lowering the N-heterocycle's LUMO shifts down both dπ and dπ* levels accordingly, while the energy gap between them (ΔE1) remaining nearly unchanged. At the same time, as the non-bonding nd orbitals are unaffected by the ligand, lowering the LUMO will significantly decrease the gap between them and the antibonding dπ* orbital (ΔE2). As evident from Table 5, the nd → dπ* transition of the pyrazinium complex (λ2) occurs in the near IR region, at significantly lower energies (i.e. longer wavelengths) compared to the related complexes of neutral N-heterocycles, confirming its strong π-acceptor nature.53a
Additional spectroscopic evidence for the difference in electron-withdrawing character between pyrazine and pyrazinium is obtained upon substituting one of the NH3 ligands in their Os(II) pentaamine complexes by N2.54 The N2 stretching frequency of the pyrazinium complex appears to be strongly blue-shifted relative to the pyrazine one (Table 5, last column).
In photoluminescent complexes conversion of a coordinated pyrazine into a pyrazinium strongly affects their emission and absorption spectra. Therefore, complexes of pyrazine ligands, especially those additionally stabilized by pincer-frameworks, can be used as luminescent pH probes. For instance, Hwang reported a homoleptic Ru(II) complex 63a containing a pair of pyrazine bis-carbene pincer ligands55 (Fig. 19). This complex exhibited a strong luminescence at 577 nm, which is somewhat red-shifted relative to the isostructural complex of pyridine bis-carbene ligands (63b) reported earlier (532 nm).56 However, unlike the pyridine analogues, the pyrazine-based ligands can undergo protonation (64a) or alkylation (64b) at their non-coordinating N atom, resulting in pronounced changes in the photophysical properties of this complex.55 As in the case of the Os(NH3)5 complexes, upon conversion of the neutral pyrazine moieties in 63a into the cationic Me-pyrazinium ones in 64a, its MLCT absorption band corresponding to the nd → dπ* transition shifts to a longer wavelength (from 382 to 486 nm), indicating lowering of its dπ* levels. Similar changes are also observed upon pyrazine protonation (64b) (Fig. 19).
The most striking change, however, occurs in the emission spectrum, where the fluorescence intensity is drastically diminished by protonation and fully quenched by alkylation of the pyrazine ring. This fluorescence quenching is also closely related to the low dπ* level of the cationic moieties, as it lowers the overall energy of the potentially emissive 3MLCT excited state. Consequently, this excited state becomes sufficiently close to the ground state allowing a non-radiative decay in accordance to the so-called “energy-gap law”.57 Since this fluorescence quenching is strongly dependent on acid concentration, complex 63a acts as an efficient photoluminescent pH sensor both in aqueous and organic media.
The low lying π* orbitals of pyrazinium responsible for its pronounced π-acidic character also reduce its reduction potential compared to neutral N-heterocyclic ligands (Table 6)53a,54 and render it a redox non-innocent ligand. Indeed, upon 1e− reduction the coordinated pyrazinium becomes a “spin-labeled” ligand, i.e., a relatively stable neutral radical ligand, which can serve as a probe for studying electronic and magnetic properties of a metal center (Fig. 20a).58
N-Heterocycle | E 0red (V vs. SCE) |
---|---|
Pyridine | −2.69 |
Pyrimidine | −2.37 |
Pyrazine | −2.11 |
N-Methyl pyrazinium | −0.73 |
Fig. 20 Pyrazinium-derived radical as a spin labeled ligand (a) and formation of pyrazinium radical complexes (b) and (c). |
For instance, electrochemical reduction of the above mentioned Ru(II) pentaamine pyrazinium complex 62a resulted in a paramagnetic species 62c exhibiting a well-resolved EPR spectrum (Fig. 20b).59 The strong coupling observed with the 1H and 14N nuclei of the pyrazinium ring together with a g-factor very close to that of an organic free radical (2.0021 vs. 2.0034, respectively) proves the presence of a ligand-centered radical, rather than a Ru(I) metalloradical.60
Another example of a persistent pyrazinium radical as a ligand was observed upon reduction of pyrazinium complexes 65a–d (Fig. 20c), prepared soon after the Ru pentaamine complexes.61 Unlike the latter, electrochemical reduction of these Mo(0) and W(0) complexes led to irreversible dissociation of one phosphine ligand and formation of unsaturated 16e− complexes 66a–d. This suggests that upon reduction, the pyrazinium cation undergoes an electronic “umpolung” reaction, transforming from a π-acceptor into a π-donor.61b
Redox non-innocence of methyl–pyrazinium was further studied in pincer systems, specifically, in the Fe(0) carbonyl complexes of a pyrazinediimine (PzDI) ligand (Fig. 21), which was conceived as a modular analogue of a pyridinediimine (PDI) scaffold.62 The latter is well-known for its redox non-innocence due to the low-lying CN π* orbitals that can act as electron reservoirs for coordinated metals.63 It was expected that this unique feature could be enhanced by replacing the backbone pyridine moiety of PDI by the more π-acidic pyrazine and even more so by the pyrazinium cation.62
Fig. 21 Preparation of a pyrazinium-based Fe(0) pincer complex compared to an analogous PDI complex (a) and its redox chemistry (b). |
As in the previous examples, the pyrazinium complex 68 was obtained by N-alkylation of the neutral PzDI complex 67 (Fig. 21a). The nearly identical isomer shifts of 0.0 m s−2 observed in the Mössbauer spectrum of both neutral and cationic complexes confirmed that the presence of a positive charge on the ligand has no influence on the Fe oxidation state. DFT calculations performed on complexes 68 and 69 showed that while in the neutral PzDI complex 67 the LUMO level is only slightly lower than in the corresponding PDI complex 69, it is strongly affected by the N-methylation shifting it down (along with the HOMO) by ca. 3.3 eV (Table 7, the 2nd and 3rd columns). The strong π-acceptor nature of the pyrazinium moiety in 68 is also experimentally confirmed by the significant blue-shift of its carbonyl stretching frequencies (Table 7, 4th column).
Complex | HOMO (eV) | LUMO (eV) | ν CO (cm−1) | E red (V vs. SCE) |
---|---|---|---|---|
a E red values were originally reported vs. Fc/Fc+, but here they are shown vs. SCE for consistency with the electrochemical data in other tables. | ||||
(PDI)Fe(CO)2 (69) | −4.34 | −2.01 | 1946, 1888 | −2.08 |
(PzDI)Fe(CO)2(67) | −4.73 | −2.13 | 1967, 1904 | −1.90 |
(MePzDI)Fe(CO)2(68) | −8.04 | −5.40 | 1999, 1938 | −1.15, −1.72 |
Similar to the PDI complex, both the neutral pyrazine and the cationic pyrazinium complexes exhibited a reduction wave around −2.0 V, which can be assigned to the imine group reduction. For the pyrazinium complex, however, yet another reduction peak was observed at a less negative potential (−1.15 V), presumably corresponding to the reduction of the pyrazinium moiety. Even though this reduction wave was electrochemically reversible, the corresponding radical product (70) was not stable enough for isolation. Instead, upon chemical reduction of 68 with cobaltocene, an immediate intermolecular radical recombination reaction occurs, forming a dimeric species 71 with a very long C–C bond of 1.600(9) Å (as determined by XRD). This bond is quite fragile and can be easily cleaved homolytically upon oxidation with Fc[PF]6, leading back to the cationic monomer 68 (Fig. 21b). It appears, therefore, that conjugation of the pyrazinium cation with additional redox-active species (the imines) does not stabilize the ligand-centered radical in this case.
Chart 7 Reported main group analogues of the NHC ligands (a) and possible resonance structures of an N-heterocyclic nitrenium cation (b). |
On the contrary, the nitrogen analogue, i.e. N-heterocyclic nitrenium (NHN), otherwise known as the 1,2,3-triazolium cation, which was first prepared by Wolff in the early 1900s,73 is indefinitely stable under an ambient atmosphere. Presumably, this stability stems from the fact that the positive charge in NHN is delocalized over all its three N atoms, as represented by the resonance structures i–iii shown in Chart 7b.74 This is quite different from the situation with the heavier NHE congeners (E = P, As, Sb), where due to the size and electronegativity difference between nitrogen and heavier pnictogens, such stabilization occurs to a lesser extent, and therefore the positive charge is mostly localized on the apical atom (as in resonance form iii).74b
The inherent stability of N-heterocyclic nitrenium cations has long been an obstacle for engaging these species in coordinative bonding, despite several attempts.75 Yet, in 2011, in a seminal report published by Gandelman the coordinative behavior of NHNs was finally demonstrated.76 This was achieved by functionalizing triazolium (or benzotriazolium) cations with chelating phosphine arms, forming the corresponding PNP pincer ligands 72 and 73 (Fig. 22). Subsequent work showed that the use of such pincer frameworks was crucial, since appending only a single phosphine arm (75) was found to be insufficient for inducing nitrenium coordination.77d
Reactions of the NHN pincer ligands 72–73 with Ru(II) or Rh(I) precursors resulted in the formation of the corresponding complexes 78 and 79 (Fig. 23a), where formation of the nitrenium–metal bond was unequivocally established by XRD. Furthermore, selective isotopic labeling of the central N atom in ligand 73 allowed monitoring nitrenium coordination in solution by means of 15N NMR spectroscopy. For instance, in the case of complex 79 a significant upfield coordination shift of −92.7 ppm (characteristic of coordinated nitrogen atoms) was revealed, as well as an exceptionally large 15N–103Rh coupling constant of 29 Hz, attributed to an enhanced s-character of the formally sp2-hybridized nitrogen lone-pair orbital.76
Following this initial report, Gandelman demonstrated the versatility of nitrenium pincer ligands 72–75 by preparing an extensive series of NHN complexes with various 2nd and 3rd row transition metals in different oxidation states (Fig. 23).77 In those complexes the metal centers were shown to undergo oxidative addition and auxiliary ligand exchange reactions (sometimes under harsh conditions) with neither the structural integrity of the nitrenium moiety nor the metal–nitrenium bond being compromised.77b Moreover, the structural robustness of those ligands allowed coordination of nitrenium to various mono- and even bis-cationic metal fragments of Rh(I), Pd(II) and Pt(II). Computational studies showed that although formation of such complexes is highly unfavorable thermodynamically, they are kinetically stable due to high N–M dissociation barrier imposed by the pincer framework.77a Later on, the Cu(I) complex of a related NNN nitrenium pincer ligand 76 (Fig. 22) was prepared by Yadav;78 however, in this case, coordination of nitrenium was only supported by AIM calculation performed on the optimized structure of this complex.
Comparison of the carbonyl stretching frequencies in NHN-coordinated carbonyl complexes of Rh(I) and Ir(I) 80a–d76,77c to those of structurally related complexes with neutral pyridine-based PNP ligands (82a,b) (Fig. 23c and Table 8) indicates that NHN is a moderately electron withdrawing ligand. In fact, complexes 80a–d exhibit similar blue shifts of 30–40 cm−1 relative to their analogues with no N–M bond (81a–d), where instead of NHN the metal carbonyl fragment is coordinated to a chloride, a rather poor σ-donor. Therefore, it appears that the electron withdrawing effect of the NHN ligands is mainly due to their stronger π-acceptor ability.
Compound | ν CO (cm−1) | Compound | ν CO (cm−1) |
---|---|---|---|
80a | 1978 | 81a | 2013 |
80b | 1981 | 81b | 2015 |
80c | 1980 | 81c | 2011 |
80d | 1963 | 81d | 2005 |
82a | 1980 | 82b | 1962 |
This was further corroborated by charge decomposition analysis (CDA) performed on a series of model Rh(I) complexes.76 In particular, it showed that although the bond-dissociation energies of the model NHN and pyridine complexes 83a and 83b are nearly the same (Table 9, 1st row), the relative strength of the σ and π bonding interactions in these two complexes is quite different (Table 9, the 2nd and 3rd rows). For the cationic NHN the contribution of the M → L π back-donation is significantly larger than that of the L → M σ donation.
This electron withdrawing character of NHNs suggested the possibility of applying them as ancillary ligands in TM-catalyzed electrophilic transformations, such as N–H and O–H addition to alkenes. In this respect, nitrenium ligand 74 proved to be particularly effective: its Rh(I) complex 84 (Fig. 24a) was quite reactive for intramolecular hydroamination, whereas its Pt(II) complex 85 showed excellent performance in a similar hydroalkoxylation reaction (Fig. 24b, reactions i and ii, respectively).77d
Fig. 24 Catalytically active NHN complexes of Rh(I) and Pt(II) prepared from ligand 74 (a) and their application in intramolecular hydroamination and hydroxylation reactions (b). |
Similar to N-methyl-pyrazinium, the benzannulated NHN ligand 73 was found to be redox-active. When metal-free, it exhibits a reversible reduction wave at −1.16 V, very close to the reduction potential of a simple dimethyl benzotriazolium cation 86 (Fig. 25a and Table 10).79 This indicates that the phosphine substituents have little effect on the redox potential of the central nitrenium moiety and the stability of the corresponding radical. Furthermore, with both cations a deep blue radical species could be obtained by stoichiometric reduction with KC8 (Fig. 25a).77d
Fig. 25 Redox chemistry of a metal free N-heterocyclic nitrenium cation (a) compared to its Rh(I) carbonyl complexes without (b) and with (c) the N–M bond. |
Compound | E 1 (V vs. SCE) | E 2 (V vs. SCE) | Ref. # |
---|---|---|---|
a All Ered values except for 86 were originally reported vs. Fc/Fc+, but here they are shown vs. SCE for consistency with the electrochemical data in other tables. | |||
73 | −1.16 | — | 77d |
77 | −0.58 | — | 80 |
80b | −0.98 | −0.59 | 77d |
81b | −0.19 | — | 77d |
86 | −1.24 | −2.13 | 79 |
88 | −1.37 | −1.31 | 80 |
Compared to free ligand 73, for a nitrenium complex 81b with no Rh–N bond a slightly shifted reduction potential (−0.99 V) was observed. This could be unequivocally attributed to the reduction of nitrenium, since a nitrenium-free complex 87 with an identical coordination sphere (Fig. 25b) is not redox-active within this potential window. On the other hand, the nitrenium coordinated complex 80b shows two reversible reduction waves (at −0.98 V and −0.59 V vs. SCE, Fig. 25c and Table 10), with the first reduction wave occurring at a significantly less negative potential relative to 81b.77d
Interestingly, two reduction waves are also observed for the non-benzannulated NHN complex 80a (shown in Fig. 25c), but this time they are irreversible and occur at significantly more negative potentials (Table 10). The fact that the CO vibration frequencies of 80a and 80b are nearly identical (Table 8) suggests that both benzannulated and non-benzannulated NHN ligands have a similar electronic influence on the Rh(I) center. Therefore, the two reduction events are most likely to occur not on the metal, but on the ligand; otherwise, a similar behavior should be expected for both complexes. These observations not only illustrate the redox non-innocence of NHN ligands, but also emphasize the essential role played by nitrenium–metal coordination in stabilizing the corresponding singly- and doubly-reduced NHN moieties.77d
The redox non-innocence of NHNs was further explored by Ray in a bicyclic ligand system 77 (Fig. 22), which can be viewed as a nitrenium pincer appended to a cyclam macrocycle.80 Surprisingly, reaction of this unusual ligand with a Ni(0) precursor did not lead to the expected NHN–Ni(0) complex, instead, a paramagnetic Ni(I) complex 88 was isolated. As its yield never exceeded 50%, the authors suggested that formation of 88 occurs with a concomitant reduction of the remaining NHN ligand (Fig. 26a). An anisotropic EPR signal recorded in frozen MeCN showed 88 to be a metal-based radical. The assignment of the +1 oxidation state for the Ni center in this complex was further supported by XANES measurement at the Ni K-edge. The XRD structure of 88 is quite intriguing. While the bicyclic framework of ligand 77 forms the anticipated square pyramidal coordination cage around the metal, the NHN plane is tilted at an angle of ca. 142° with respect to the Nnitrenium–Ni bond (Fig. 26b). The nitrenium nitrogen is therefore pyramidalized (Σangles = 332°), which is consistent with its lone pair not being involved in metal coordination. Such a bent binding mode is strikingly different from the planar coordination mode observed in all other NHN complexes, where the nitrenium nitrogen acts as a σ-donor, albeit a weak one. Indeed, the NBO analysis revealed that the Nnitrenium sp2 orbital has nearly the same occupancy in 88 as in the free ligand 77 (1.94e−vs. 1.92e−). Therefore, the NHN moiety here binds as a Z-type (rather than an L-type) ligand, accepting electron density from the Ni dz2 orbital into its π* orbital.
Fig. 26 Synthesis and reactivity of bicyclic NHN Ni complexes (a) and structures of complexes 87 (b) and 89 (c). |
Complex 88 was shown to undergo a 1e− reduction with KC8, resulting in a diamagnetic product 89, as could be expected for a Ni(0) complex (Fig. 26a). Yet, such a description was inconsistent with the XANES results, in particular, the presence of a 1s → 3d pre-edge transition at exactly the same energy, as observed for 87 (8331.4 eV), indicating that the oxidation state of Ni in 88 is +1, rather than 0. Thus, the reduction necessarily involves the NHN ligand, which is also in line with the calculated π-population of the benzotriazolium ring in 89, being 0.74e− higher than in the free ligand. The observed S = 0 state of complex 89 therefore results from an antiferromagnetic coupling between a ligand-centered radical and a metal-centered one.
It is noteworthy that here the reduction of the metal free and Ni-coordinated nitrenium occurs at nearly the same potential (−1.37 and −1.31 V, Table 10), which is more negative than the reduction potential of ligand 77 (−0.58 V) and even much more so compared to complex 81b (−0.19 V). This implies not only that the strain imposed by the bicyclic framework of 77 strongly destabilizes the NHN radical, but also that coordination to the Ni(I) center does not stabilize this radical as much as the Rh(I) center in 80b, due to weaker orbital metal–NHN interaction.
In addition to reduction, complex 87 could also be oxidized with FcBF4 into a paramagnetic (S = 1) product, identified as a complex of Ni(II) by XANES. The so-formed nitrenium-coordinated Ni(II) center was electrophilic enough to abstract a fluoride from the BF4 anion, forming the resulting complex 90 (Fig. 26a). In fact, in the absence of a fluoride source (i.e. when substituting FcBF4 with FcBPh4) the oxidation product could not be isolated. Although no XRD data for 89 were available, its optimized structure (Fig. 26c) revealed a planar coordination mode of nitrenium, indicating that unlike in 88, here the Nnitrenium lone pair is involved in σ-bonding. This was further confirmed by the NBO analysis, which additionally showed that due to the extremely electron poor character of the Ni(II) center, the π-back donation in 90 is negligible. The enhanced electrophilicity of nitrenium-coordinated Ni(II) was also demonstrated by the ability of this complex to oxidize formate into CO2 – a reactivity that was not observed in an analogous Ni(II)–cyclam complex with no appended NHN moiety.
The first binding mode, where the phosphorus adopts a trigonal planar geometry (Chart 8a), is similar to that typical for coordinated NHCs, i.e. lone pair σ-donation to the metal combined with a prominent π back-donation to a vacant phosphorus-centered pπ orbital of NHP. Such NHP complexes usually exhibit short P–M bonds, due to a partially double-bond character.22d
In the second binding mode (Chart 8b) the P lone pair does not participate in bonding, and the major bonding interaction is the electron donation from the metal to vacant pz orbital of the phosphorus atom. As a result, it acquires a pyramidal geometry with an elongated M–P bond.83 However, being highly Lewis acidic, coordinated NHPs can also undergo reduction by an intramolecular M-to-P 2e− transfer (vide infra). In such a case, the NHP is converted from a cationic phosphenium (NHP+) into an anionic phosphide (NHP−) ligand (Chart 8b). In the Introduction, while discussing the ambiguity in ascribing a definite positive charge to a certain ligand, we already mentioned this redox non-innocence of NHPs, along with the analogous nitrosyl NO+/NO− dichotomy.22 In fact, NHPs are often regarded as tunable NO analogues.84 Unfortunately, when exhibiting a pyramidal coordinating mode, discrimination between NHP+ and NHP− requires a thorough computational analysis (vide infra), since no clear-cut structural distinction between the two exists.
Both coordination modes of NHPs were already observed in their first reported complexes obtained by Paine and coworkers (Fig. 27a).82,85 They demonstrated that the reaction of diamino-fluorophosphine 91 with anionic Mo(0) and Fe(0) precursors leads to the formation of the corresponding NHP complexes 92 and 93, with a planar82 and a pyramidal85a geometry, respectively (Fig. 27b). But why does the very same NHP ligand acquire two different binding modes in such structurally similar complexes? The only meaningful difference between the metal fragments of 92 and 93 is their valence electron count: [(Cp)Mo(CO)2] is an unsaturated 16e− fragment, whereas [(Cp*)Fe(CO)2] is a saturated 18e− one. Apparently, in 91 NHP behaves as a 2e− donor supplying the missing electrons, which dictates its planar binding mode. Conversely, with the electronically saturated metal center in 93, the NHP lone pair remains non-bonding. Even though 93 was originally described as an NHP+/Fe(0) complex (resonance form i), yet, in light of similar tetrahedral Fe(II) complexes with the general formula of Cp*Fe(CO)2X, where X is clearly an anionic ligand (X = Cl, CN, OTf, etc.)86 an alternative description of 93 as an NHP−/Fe(II) complex (resonance form ii) cannot be fully discarded.
Fig. 27 Synthesis of the NHP complexes of Mo(0), Fe(0) and Co(−1) carbonyls (a) and comparison between the XRD structures of complexes 92* and 93 (b) (as the original XRD data of 92 are unavailable, the partial structure of an isostructural complex with a di-N-benzyl NHP derivative, 92*,85d is shown here). |
An additional interesting bonding situation, quite relevant to our subsequent discussion on pincer NHP complexes, occurs in an asymmetric homobimetallic Co(–I) complex 94, also obtained by Paine, where the two NHPs bridge between Co(CO)2 and Co(CO)3 centers.85b The P–Co bonds to each of them differ significantly in length (Table 12), indicating a different type of bonding. Indeed, theoretical analysis showed that both NHPs simultaneously act as L-type ligands towards the former center (along with significant π-backbonding), and as Z-type ligands towards the latter. Hence here, like in complex 92, an unsaturated 14e− Co(CO)2 fragment is stabilized by the lone-pairs of the two NHP ligands.
Compound | 31P NMR (ppm) | P–M bond length (Å) | Σ angles (°) | Ref. # | |
---|---|---|---|---|---|
Metal-free NHPs | +264 | — | — | 87 | |
96 (phosphine adduct) | +92 | — | — | 88 | |
NHP complexes | 92 | +271 | 2.213 | 360 | 82 |
93 | +285 | 2.340 | 314 | 85a | |
94 | +307 | 2.05 (av), 2.41 (av) | No data | 85b | |
98a | +257 | 2.0903(6) | 325 | 89 | |
98b | +240 | 2.2424(13) | 310 | 84a | |
98c | +225 | 2.2446(11) | 312 | 84a | |
99a | +273 | 2.2491(5) | No data | 89 | |
2.0437(5) | |||||
99b | +288 | 2.4982(16) | No data | 84a | |
2.1616(15) | |||||
99c | +258 | 2.150(5) | No data | 84a | |
2.482(3) | |||||
99d | +226 | 2.0669(16) | 339 | 90 | |
2.0711(16) | |||||
2.1879(17) | |||||
2.1916(17) | |||||
100a | +249 | 2.0417(9) | 338 | 91 | |
100b | +236 | 2.2535(6) | 326 | 84a | |
100c | +206 | 2.2606(9) | 324 | 84a | |
104 | +286 | 2.2386(6 | 302 | 92 | |
105 | +242 | 1.9922(4) | 356 | 92 | |
106 | +236 | 1.9840(4) | 359 | 93 | |
108 | +204 | 2.0283(5) | 358 | 94 | |
112d | +261 | 2.0957(13) | 333 | 90 | |
114 | +207 | 1.9455(6) | 359 | 92 | |
Phosphide complexes | 102a | +145 | 2.143(1) | 328 | 95 |
102b | +123 | 2.2533(9) | 324 | 95 | |
102c | +68.8 | 2.2573(11) | 321 | 95 |
Following Paine's pioneering work, other NHP complexes were prepared;96 yet, most of them were highly reactive precluding their applications in catalysis.97 Therefore, stabilizing NHP–M bonds by a robust pincer motif appeared to be a reasonable idea. This direct-ion was extensively explored by Thomas who prepared a bis-phosphine NHP pincer ligand, starting from the corresponding chlorophosphine 95 (Fig. 28a).88 Interestingly, because of its highly Lewis acidic character, the phosphenium center strongly interacts with one of the flanking phosphine residues, forming a crystallographically characterized phosphenium–phosphine adduct 96 (Fig. 28d). The triplet multiplicity of the PNHP signal in 31P NMR indicated that the two flanking phosphines quickly interchanged in solution.88 This NHP–phosphine adduct was easily cleaved upon reaction with a PtCl2 precursor. However, instead of the desired NHP complex, chloride migration from a metal to the highly electrophilic phosphenium site resulted in the chlorophosphine complex 97. Such a non-innocent coordination behavior could be avoided by utilizing halide free M(0) precursors (M = Ni, Pd, Pt).84a,89
Surprisingly, with Ni(0)89 or Pd(0)84a this approach exclusively yielded homo-bimetallic dimers 99a,b, where two molecules of 96 span two weakly bonded M(0) centers. XRD structures of 99a and 99b (Fig. 28e) show that each NHP moiety is engaged in two types of M–P interactions, with one of the bonds being noticeably shorter than the other (Table 12), due to its pronounced double bond character. Thus, similar to the binuclear complex 94 discussed above, these semi-bridging NHPs act as both σ-donors and π-acceptors towards one center and σ-acceptors towards the other one. Furthermore, the fact that geometry around the P centers is nearly planar, but close to tetrahedral for both M(0) centers, is quite consistent with the NHP+/M(0) description of 99a,b. These homodimers could also be obtained starting from the chlorophosphine ligand 95 by oxidative addition of the P–Cl bond, followed by chloride abstraction. In fact, this was the only route affording the analogous Pt(0) homodimer 98c (Fig. 28a).84a
The intermediate monomeric MCl complexes 98a–c deserve a separate discussion. The available XRD structures show that their NHP moiety is pyramidalized (Σangles = 325.6° and 310.2° for 98a and 98b, respectively), indicative of a non-bonding lone pair on the PNHP atom. This, together with a square planar geometry around the Pd center in 98b, typical for the 2nd and 3rd row d8 complexes, strongly suggests the NHP−/M(II) type of bonding in these complexes. Furthermore, the higher formal oxidation state of Pt in 98c compared to 99c manifests in a significantly smaller 1JPt–P, dropping from 2161 Hz to merely 663 Hz. The NHP−/M(II) representation is also supported by NBO calculations performed on 98a and the iodo-analogue of 98c, which described the P–M interactions as phosphorus-polarized covalent bonds (56.9% P/43.1% Ni89 and 58.3% P/41.7% Pt,84a respectively).
A similar pyramidal coordination of the NHP moiety was also observed in the cationic monomeric complexes 100a–c (as shown in Fig. 28f for 100b), obtained by addition of PMe3 to the homodimers 99b,c84a or by ligand exchange in 98a.91 Despite the similar geometry around the NHP center observed in those complexes, NBO analysis revealed a striking difference between the Ni complex 100a and its heavier congeners, 100b,c. While in the latter Pd and Pt complexes the NHP–M bonding was quite similar to 98a–c,84a,89 with Ni it was modeled as a strong donor–acceptor Ni → PNHP interaction, and therefore in agreement with the NHP+/Ni(0) representation.91 The inverse polarization of the dative PNHP–M bond in 100avs.100b,c also determines the reactivity of those complexes (Fig. 28b). In the Pt complex 100c, for instance, the NHP ligand could be protonated (101b)98 or alkylated (101c),84a demonstrating the Lewis-basic character of its PNHP center. On the other hand, in the Ni complex 100a, the same PNHP center exhibited an electrophilic behavior, by attacking its own BPh4 anion upon heating (101a),91 a reaction not observed for 100b,c.
Unfortunately, this discrepancy between phosphenium and phosphide character of the PNHP center has little effect on its 31P NMR, as for all complexes 98–100 its chemical shift falls within the same range of 200–300 ppm (Table 12, the middle section). This range is characteristic of sp2-hybridized planar phosphenium species, rather than sp3-hybridized phosphides, observed, for instance, in related bonafide pincer phosphide complexes 102a–c95 (Fig. 28c), resonating at significantly higher fields (Table 12, bottom section). Yet, none of the above monomeric NHP pincer complexes displayed a planar coordination mode, as established in 91 and other complexes of monodentate NHPs.82
This preference of the NHP pincer scaffold for the pyramidal over the planar geometry at the PNHP center was substantiated in a subsequent work with Co90 and Rh,99 where only pyramidal phosphide complexes were obtained. The researchers attempted to rationalize this tendency by the fact that a pincer scaffold inevitably enforces co-planarity of the P–N bonds with the adjacent N-aryl substituents and so intensifies the N lone pair delocalization over the aromatic rings at the expense of their hyperconjugation with the empty pz orbital on the PNHP atom (Fig. 29a). As the positive charge of phosphenium is therefore less stabilized, it becomes more Lewis acidic and/or oxidizing.84a
To test this hypothesis Thomas prepared the chelating chloro-phosphine 103 and compared its reactivity with Co(–I)92 and Ni(0)93 precursors to its pincer analogue 95 (Fig. 29b). Indeed, while the latter afforded the corresponding complex 104 with a pyramidalized NHP, best described by the NHP−/Co(I) formalism (Fig. 29c),90 in complex 105 with the bidentate ligand, the NHP moiety adopts a nearly planar binding mode with a very short P–Co bond (1.992 Å) (Fig. 29d).92 This time NBO calculation clearly identified molecular orbitals corresponding to both σ- and π-interactions. In addition, the computed Co center's natural electron configuration, summing up to the total of 10 valence electrons, was in accordance with the NHP+/Co(–I) formalism. The lower formal oxidation state of Co in complex 105 relative to 104 was also evident from its red-shifted carbonyl frequencies (1967 and 1915 vs. 1981 and 1926 cm−1, respectively). Likewise, a planar NHP+–Ni(0) complex was obtained upon reaction of 104 with a Ni(0) precursor,93 in contrast to the previously discussed complex 100a91 obtained from 95 under similar conditions Fig. 29b).
In both bidentate Co(–I) and Ni(0) complexes, the P–N bond length associated with the mesityl substituent (oriented perpendicularly to the NHP ring) is 0.03–0.04 Å shorter compared with the one associated with a co-planar phosphinoaryl.92 This indeed indicates a stronger PNHP–N conjugation with the perpendicular aryl, and as a result higher stabilization of the phosphenium form, which in turn might explain the preference for a planar coordination mode in these complexes. Yet, later a computational study on model NHP–Co(–I) systems100 showed that hyperconjugation effects alone cannot fully account for the disparate behavior of bidentate and tridentate NHP ligands, because pyramidalized NHP was obtained even in a NHP pincer complex bearing aliphatic phosphine arms (Fig. 29a). Thus, structural rigidity of the tridentate pincer framework also plays an important role in defining the preferred NHP binding geometry.
In light of the above, it came as a surprise, when very recently a planar coordination of NHP was finally obtained within a monomeric complex of ligand 96 (Fig. 30a).94 The XRD structure of this manganese carbonyl complex 108 (Fig. 30b), prepared by reduction of the corresponding chlorophosphine complex 107, revealed a very short Mn–PNHP bond of only 2.0283(5) Å, the shortest reported so far for this pair of elements. Theoretical analysis of the NHP–Mn bonding revealed a highly phosphorus polarized σ-bond intensified by nearly covalent π-interactions, leaving some ambiguity as to the validity of the NHP+/Mn(–I) representation. Nevertheless, the phosphenium character of NHP in complex 108 could be inferred from its reactivity with proton/hydride donors (Fig. 30a). Indeed, contrary to complex 100c (Fig. 28b), protonation of 108 with HCl occurred on the metal, concomitant with a chloride attacking the electrophilic NHP center (109). On the other hand, the latter could be transformed into a phosphine, by reacting 108 with a strong hydride donor (110).
Fig. 30 Synthesis and reactivity of an NHP complex with Mn(–I) carbonyl (a) and its XRD structure (b). |
The redox non-innocence of the NHP+ ligands within their complexes, being so susceptible to reduction by an intramolecular M → L 2e− transfer, raised the question about their reactivity towards external redox agents. It was found that reduction of the monomeric NHP complexes 98a–c with Na amalgam results in symmetrical homodimers 112a–c with a σ-bonded bimetallic core (Fig. 31a). Unlike in the previously mentioned bis-cationic dimers 98a–c, here each PNHP atom is engaged in two nearly identical bonds with both of the metals and therefore, such compounds can best be described as phosphido-bridged M(I) dimers.89,90 An isostructural dimer 112d was also obtained by reduction of a Co(II) chlorophosphine complex 111 with Mg.90
Fig. 31 Electrochemical studies of homodimeric NHP complexes of Ni, Pd, Pt, Co (a) and a monomeric Co/NHP+ complex (b). |
Electrochemical studies performed on the Ni89 and Co90 dimers (112a and 112d) clearly identified two reversible M(I)/M(II) oxidation waves corresponding to successive oxidation of both of their metal centers en route to bis-cationic 99a,d. Furthermore, the intermediate mixed-valence mono-cationic complexes 113a and 113b could be prepared by chemical 1e− reduction/oxidation of 99a or 99d, respectively (Fig. 31a). The EPR signals of these S = 1/2 compounds exhibited no hyperfine coupling to the 31P nucleus and therefore no indication of an NHP-based radical. This was confirmed by DFT calculations showing that most of the unpaired spin densities in these paramagnetic complexes are localized on either of the two metal centers and almost none on the NHP ligand.
On the contrary, the EPR signal of a monomeric S = 1/2 complex 115 (obtained by 1e− oxidation of a chelate phosphenium Co(–I) complex 114) exhibited a complicated pattern due to a strong hyperfine coupling to both 31P and 59Co nuclei (Fig. 31b).92 Computational analysis suggested that although this oxidation is also metal-centered (according to the natural charge on Co decreasing from −1.56 in 114 to −0.67 in 115), here the unpaired spin density is delocalized, imparting the NHP ligand a partial radical character (resonance form ii).
Finally, perhaps the most interesting properties of the pincer-type NHP complexes is their ability to activate E–H bonds (E = H, B, O, S) through ligand–metal cooperativity (Fig. 32a). The Co(I) complex 116a is particularly notable in this respect, providing a rare example of a quantitative splitting of H2 over a P–M bond (117).101 This and other NHP pincer complexes were also shown to activate OH and SH bonds, forming the corresponding 1,2-addition products 118–119a–c (Fig. 32a). Interestingly, opposite regioselectivity of addition was observed for phenol and thiophenol, resulting in the corresponding P–OPh/M–H and P–H/M–SPh products, respectively.90,91,98,99 Furthermore, Ni complex 100a (Fig. 28a) and its Pt analogue 116c exhibited the same reactivity pattern, despite the inverse electronic character of their NHP ligand (Fig. 32a). It therefore appears that what governs addition regioselectivity is not the inherent polarization of the NHP–M bond within a complex, but rather the preferences of hardness/softness.
Examples of such cooperative E–H activation over M–Pphosphido bonds are quite rare,102 in contrast to the analogous reactivity of M–Namido bonds,103 which is widely used in homogeneous catalysis. Therefore, activation of E–H bonds by complexes 114a–d suggested that such NHP complexes might also be employed as efficient hydrofunctionalization catalysts.104 Indeed, a putative Co hydride 121 (Fig. 32b) generated in situ from the previously mentioned chlorophosphine Co(II) complex 111 (Fig. 31a) could be used for catalytic hydroboration of terminal alkenes with excellent anti-Markovnikov selectivity (Fig. 32c, reaction i).104a Recently, it has been found that the same species (this time generated from a different Co(II) precursor, 118) is also capable of promoting hydrogenation of alkenes (Fig. 32c, reaction ii).104b To the best of our knowledge, these are the only successful applications of NHP-based pincer complexes in catalysis, reported so far.
Nevertheless, computational analysis revealed that NHSb/Co(–I) complexes with both pyramidal and planar geometries correspond to local energy minima on a potential surface. Furthermore, comparative calculations performed on those systems83,100 showed that going down the periodic table from P to As and Sb, the corresponding NHE ligands demonstrate a lower tendency for sp orbital hybridization. This results in an increasing degree of the lone pair s-character, and therefore weakening of the σ-donor ability of the heavier pnictogens. On the other hand, as the energy of their LUMO decreases, these ligands become stronger Lewis acids and better π-acceptors.83
The above theoretical considerations obviously suggested that with sufficient kinetic stabilization, NHSb complexes should be attainable. With this in mind, Chiu prepared a pincer-type benzannulated chlorostibine 126 (Fig. 34) as a precursor for a pincer-type NHSb ligand.105 However, the desired stibenium ligand 127 could not be obtained from this precursor by an attempted chloride abstraction with an Ag(I) salt. Instead, the corresponding chlorostibine–(AgOTf) complex 128 was isolated. Although containing no stibenium cation by itself, this complex turned out to be a useful reagent for delivering the stibenium pincer ligand to other transition metals.
Fig. 34 Synthesis of the homobimetallic Rh(I) and Ir(I) NHSb complexes (a) and the XRD structure of complex 129a (b). |
Indeed, reacting this complex with Rh(I) or Ir(I) precursors afforded the first examples of metal-coordinated NHSb complexes, although structurally quite different from the expected. Unlike all previously characterized NHE complexes (either monomeric or dimeric) that invariably exhibited a 1:1 L:M ratio, here only bimetallic complexes 129a,b with the 1:2 L:M ratio were obtained, even with sub-stoichiometric amounts of metal precursors. As determined by XRD crystallography, the Sb atom in these bimetallic complexes acquires a bridging position between two metal centers located above and below the NHSb ligand plane. This bridging coordination mode is somewhat reminiscent of that observed in the previously discussed dimeric NHP complexes (112a–d in Fig. 31a), except that here the Sb center exhibits the same interaction with each of the metal centers. Thus, rather than being a pincer–type ligand, 126 acts as a κ2-P,Sb chelate towards each metal center.105
Theoretical explanation of the bonding in these complexes compounds was provided by NBO analysis, which identified two major orbital interactions within the M–Sb–M fragment. The first stems from the donation of the Sb lone pair to the two empty sd hybrid metal orbitals, giving rise to a 3c–2e− bond. The other is a back-donation from two filled dZ2 orbitals on the metals to an empty p orbital. Calculations also showed that these interactions reduce the aromaticity within the NHSb ring, as evident from the slightly longer N–C and N–Sb bonds, as well as less negative NICS0 and NICS1 values of an optimized metal-coordinated NHSb complex, compared to a metal free one.
This unexpected Lewis acidic character of stibonium species was disclosed by Gabbai in a series of studies on organometallic bis-(1-naphthyl)diphenylstibonium cations, where the naphthyl groups are bridged at their 8th position by a heavy post-transition metal ion, namely, Hg(II)106 or Au(I).107 Thus, these compounds can be regarded as complexes of a C–Sb–C pincer ligand in which rigid naphthyl buttresses bring an electron rich metal ion within ∼3 Å from an electrophilic Sb center, which is close enough for inducing M → Sb interactions. Such interactions were indeed identified by means of electron localization function (ELF) maps generated for the Hg(II) complexes 130a,b (Fig. 35a) that were reported first.106 Non-negligible orbital interactions between Hg and Sb in these compounds were further illustrated by the elongation of the natural localized molecular orbital (NLMO) corresponding to the occupied Hg dz2 orbital in the direction of the Sb atom, especially in complex 130b.
Fig. 35 Reactivity of stibonium complexes of Hg(II) and Au(I) (a), two resonance forms of the Au complex 131a (b) and comparison of its XRD structure to that of the isolobal Hg complex 130a (c). |
In fact, the very formation of the latter complex by reaction of 130a with iodide is quite noteworthy, since common diarylmercury(II) compounds are quite inert towards halides and other Lewis bases. Later it was shown that in addition to halides, 130a readily reacts with other electron donors, even as weak as PF6 (130d,e on Fig. 35).108 Apparently, interaction with the Lewis-acidic stibonium renders the Hg(II) center itself more Lewis acidic. Subsequent computational studies showed that this Lewis-acidity enhancement is due to the ability of stibonium to withdraw electron density from the Hg atom through donor–acceptor interactions. As evident from the XRD structure of 130a (Fig. 35c, top) the Sb atom acquires a trigonal-pyramidal geometry with one of its Sb–C bonds aligned with the Sb–Hg axis. Such a geometry maximizes the donor–acceptor interactions between the corresponding antibonding Sb–CPh σ*-orbital and an occupied sd hybridized Hg orbital.108
Yet, with Sb–Hg distances of 3.06–3.07 Å lying between the sum of covalent and metallic radii of those elements (2.71 Å and 3.17 Å, respectively) the bonding is still relatively weak. Stronger bonding interactions were observed in the Au(I) complex 131a with the Sb–M bond length being 0.3 Å shorter than that in 130a (Fig. 35c), despite similar covalent radii of Au and Hg.107
Unlike in complexes 130a–e, where the divalent state of Hg is quite unequivocal, in the case of 131a the Au–Sb bonding can be represented by two resonance structures, featuring either the anionic Au(I) or cationic Au(III) center (resonance structure i and ii in Fig. 35b). Here X-ray absorption at the Au L3-edge that corresponds to the excitation of Au 2p3/2 electrons into empty 5d orbitals is particularly informative. Although Au(I) is formally a closed shell 5d10 ion, it still possesses a non-negligible d-hole density due to its sd hybridization, which allows for such a transition to occur. It is therefore highly sensitive to even minor changes in 5d-electron density due to oxidation. Yet, the Au L3-edge position of 131a was found to be very similar to a reference Au(I) compound. Accordingly, the Sb K-edge was identical to that of SbPh4+, which was also the case in all stibonium–Hg(II) complexes. Taken together, these results are strongly in favor of the description of 131a as a stibonium–aurate(I) (resonance structure i) complex. Such an assignment was also supported by Boys localization analysis showing that the σ-symmetric orbital oriented along the Sb–Au axis is highly Au polarized.
On the experimental level, the anionic character of the Au center in 130a is apparent from its reluctance to react with halides. It can, however, react with fluoride; yet, as the latter is a hard anion it selectively adds to the hard Sb(V), rather to the soft Au(I) (131b). Interestingly, the dative character of Au → Sb interaction after fluoride addition remains nearly unaltered, since the Sb atom in the resulting fluorostiborane(V) moiety is still quite Lewis acidic.
This redox versatility of antimony-based ligands resulting from their coordination non-innocence111a was demonstrated by Gabbai and coworkers in a series of studies on Sb complexes that display an intricate interplay between Sb(III) (stibine) and Sb(V) (stiborane) states.110a In this respect, it is instructive to discuss in detail several Au and Pt complexes of the same Sb-based triphosphinostibine ligand 132, capable of acquiring both pincer-type and tripodal coordination modes.112
In the case of Au, the oxidation state assignment for Sb in complexes 133 and 134 (Fig. 36a) is relatively straightforward. The Sb(III) center in the pincer-type complex 133, obtained by reaction of 132 with (THT)AuCl, can be regarded as a classic σ-donor, isolobal to phosphine. Oxidation of 133 to 134 with PhICl2 transforms the stibine(III) donor into dichlorostiborane(V), converting it from an L-type into a Z-type ligand.112a Thus, upon oxidation the Sb–Au bond undergoes an umpolung reaction, as evident from the NBO analysis pointing out that the Sb–Au bonding interaction changes from lp(Sb) → pAu in 133 to lp(Au) → σ*Sb-Cltrans in 134 with the corresponding deletion energies (i.e. the rise in electronic energy upon deleting NBOs involved in those donor–acceptor interactions, Edel) reducing from 63.59 to 35.28 kcal mol−1, respectively. Although this oxidation is formally Sb-centered, the Au atom is also affected and changes its geometry from tetrahedral, typical of Au(I), to square planar, typical of Au(III) (Fig. 36b).
Fig. 36 Synthesis Au(I)-stibine and Au(I)-stiborane pincer complexes (a) and comparison of X-ray structures of complexes 133 and 134 (b). |
Unlike with (THT)AuCl, the reaction of ligand 132 with (SEt2)2PtCl2 results in a tripodal complex 135, which can no longer be described as a traditional stibine complex (Fig. 37a).112b If one considers the Pt center in 135 as divalent, like in the precursor, the ligand must contain an anionic Sb(III) center (resonance structure i), akin to rare SbX4 anions (X = Cl, Br, etc).113 Alternatively, this complex can be represented as an anionic [Pt(0)Cl] fragment, which has also been reported,17b coordinated to a chlorostibonium(V) cation (resonance structure iii). According to the NBO calculations, however, 135 contains a Pt-polarized covalent bond (38.0% Sb/53.7% Pt, Table 13, 6th column) better described by the Sb(IV)Pt(I) formalism (resonance structure ii).
Fig. 37 Synthesis of neutral, mono-, and bis-cationic Sb complexes of Pt (a) and (b) and X-ray structures of 136–138 (c). |
Complex | Sb–Pt (Å) | 1 J Pt–P (Hz) | ν NC (cm−1) | NPA charges (Sb/Pt) | NLMO Sb/Pt (%) |
---|---|---|---|---|---|
a Dative Pt → Sb interactions, E2 = 89 kcal mol−1. | |||||
135 | 2.5732 | 2305 | — | No data | 38.0/53.7 |
136 | 2.4706 | 2330 | 2227 | 1.59/0.28 | 38.4/57.8 |
137 | 2.6236 | 2964 | 2196 | 1.84/0.23 | 34.8/56.1 |
138 | 2.6568 | 3351, 2888 | 2181 | 2.08/0.24 | 29.9/60.2 |
144 | 2.4407 | 2566 | — | No data/0.17 | 49.1/45.1 |
145 | 2.5797 | 3462 | — | No data | Dativea |
146 | 2.5044 | 2404 | — | No data | 44.7/49.3 |
147 | 2.4118 | 2450 | — | No data/0.28 | 57.0/38.2 |
148b | 2.5247 | 2635 | — | No data | No data |
149b | 2.5684 | 2464 | 2229 | No data | No data |
Double chloride abstraction from 135 in the presence of isocyanide results in a dicationic complex 136 (Fig. 37b), where assigning the oxidation state to the Sb center becomes even more challenging. At first glance, it appears like a simple Pt(II) stibine complex, yet, according to the NBO analysis, the relative contribution of this classic resonance form is rather negligible compared to structures i and ii, where the Sb center is mono- or bis-cationic, respectively. Indeed, this Pt-coordinated Sb center is highly Lewis acidic and can undergo a stepwise addition of up to two fluoride ions, forming complexes 136 and 137 (Fig. 37b). A thorough theoretical analysis by NBO and quantum theory of atoms in molecules (QTAIM) revealed that with each added fluoride the bonding within the Sb–Pt core shifts from a Pt-polarized covalent Pt(I)–Sb(IV) bond in 136 (resonance form i) to a Pt(0) → Sb(V) donor–acceptor interaction in 138 (resonance form ii). This transformation manifests experimentally in elongation of the Sb–Pt bond length due to the lesser degree of covalency (Table 13, 2nd column). In addition, upon moving from 136 to 138, the calculated natural charge on the Sb atom grows from +1.59 to +2.08 (Table 13, 5th column), despite the decreasing overall positive charge of these complexes. The concomitant opposite change in the Pt atom charge is less pronounced, but the more reduced character of the Pt center is clearly evident from the larger 1JPt–P values and lower CN stretching frequencies (Table 13, the 3rd and 4th columns, respectively). In other words, upon anion coordination, the Sb center experiences a net oxidation, transferring its electron density to the Pt.112b
A very similar coordinative non-innocence was also observed in Au and Pt complexes of a related chlorostibine pincer ligand 139 (Fig. 38).114 Here too both types of Sb(III) to Sb(V) oxidation, i.e. by oxidative addition or by coordination induced intra-molecular Sb → M charge redistribution, were observed. Thus, similar to 133, reaction of complex 140 with PhICl2 converts its Au(I)-coordinated Sb center from chlorostibine(III) into tri-chlorostiborane(V) 141a.114a In this case, however, Sb oxidation does not inverse the polarity of the Sb–Au bond. This is because, as elucidated by NBO calculations, the chlorostibine in 140 already acts a Z-type ligand, due to the low-lying Sb–Cl σ*-orbital orientated along the Sb–Au axis. Nevertheless, oxidation of 140 into 141a, followed by its conversion into the corresponding trifluorostiborane complex 141b, significantly intensifies the Au → Sb donor–acceptor interactions (with Edel values increasing from 70.79 to 148.52 kcal mol−1). In both 140 and 141b the Au-coordinated chloride can be selectively removed, affording the cationic complexes 142 and 143, respectively. Assessing the Lewis acidity of their coordinatively unsaturated Au(I) centers, based on interaction with triphenylphosphine oxide in solution (by the Gutmann–Beckman method) and Au–F contacts with the SbF6 counteranion in the solid state (by XRD), clearly pointed out that the more oxidized Sb atom in 143 renders its Au(I) center significantly more Lewis-acidic. This enhanced Lewis acidity has a crucial influence on its reactivity towards C–C triple bonds, and, specifically, on catalytic hydroamination of alkynes. Indeed, while in the presence of the Sb(III) complex 142 hardly any reaction occurred, with the Sb(V) complex 143 a nearly full conversion of phenylacetylene into the corresponding imine product was observed within 40 minutes (Fig. 39, reaction i).114a
Fig. 39 Catalytic activity of cationic Au and Pt complexes in hydroamination (reactions i and ii) and enyne cycloisomerization (reaction iii). |
In the Pt systems, similar to what was observed in complexes 134–137 (Fig. 37), halide addition to the coordinatively non-innocent Sb(III) center of ligand 139 results in its intramolecular oxidation to the Sb(IV) and Sb(V) states, with the Sb–Pt bond being nearly covalent in 144114b and dative in 145, respectively (Fig. 38).111c Abstracting one of the Sb-bound fluorides in 145 leads to a homodimeric complex 146 that apparently contains a mono-cationic stiboranyl(V) ligand (resonance structure i). However, NBO analysis shows that, perhaps not surprisingly, removal of fluoride from the Sb(V) center of 145 results in a reverse electron flow from Pt to Sb, with the Sb–Pt bond going back from dative to covalent (resonance structure ii).
This electron density withdrawal from the Pt center, as reflected by the drastic drop in the 1JPt–P value from 3462 to 2404 Hz (Table 13, the 3rd column), turns complex 146 into an active catalyst for 1,6-enyne cycloisomerization. Interestingly, an analogous emergence of catalytic activity due to electron density shift from the Pt to Sb could also be achieved without altering the coordination number on either of them, merely by a ligand exchange. It was found that switching all coordinated chlorides in complex 144 into triflates (resulting in complex 147) effectively polarizes its nearly covalent Sb–Pt bond towards Sb (changing it from 49.1% Sb/45.1% Pt to 57.0% Sb/38.2% Pt)115 and consequently increases the local positive charge on the Pt atom from 0.167 to 0.266 (Table 13, the 5th and 6th columns). Judging from its 1JPt–P value, the Pt center in 147 is slightly less electron-depleted than that in 146, and hence its catalytic performance is notably higher, shortening the enyne cycloisomerization time from 4 hours to 10 min (Fig. 39, reaction iii).115
The survey of pincer PSbP ligands with a positively charged Sb center will not be complete without discussing complexes 148a,b116 obtained from complexes 134 and 135 (Fig. 37 and 38) by oxidizing one of their three phosphine donors (Fig. 40a). This effectively transforms their ligand from a tripode into a pincer with an appended Lewis base (phosphine oxide), which can form an intramolecular Lewis-adduct with the Sb center. Depending on the oxidation state attributed to the metal, this ligand can be viewed as either a coordinatively non-innocent neutral stibine(III) (resonance form i) or a donor-stabilized stibonium(V) dication (resonance form ii). The second interpretation is quite exceptional, as it implies that the Sb species in question acts as the first of a kind bis-cationic Z-type ligand.
Fig. 40 Synthesis of Au and Pt of a dicationic Sb(V) ligand (a) and comparison of their XRD structures (b). |
In the case of the Au complex 148a,116a such an unconventional ligand formulation was supported by NBO analysis, which revealed the presence of a vacant 5p orbital on Sb, distinctive of a bis-cationic Sb(V) center, stabilized by donation from the filled Au(I) dz2 and dz2−y2 orbitals, as well as the lp(O) of the appended phosphine oxide. The latter interaction, with a second order perturbation energy of 69.7 kcal mol−1, is quite strong, which is also manifested in a rather short Sb–O bond of only 2.206 Å (Fig. 40b). In Pt complex 149b,116b on the other hand, no such vacant Sb-based p orbital was identified, while the interaction with the phosphine oxide (in this case between lp(O) and the σ*(Sb–Pt)) was found to be significantly weaker, only 8.8 kcal mol−1, in consistency with a longer Sb–O bond (2.432 Å). Thus, 149b is better described as a Pt(II) complex of a coordinatively non-innocent stibine(III) ligand.
Chloride abstraction from 148a,b results in the catalytically active complexes 149a,b (Fig. 40a). The Au(I) complex 149a was found to be a highly efficient catalyst for hydroamination of styrenes (Fig. 39, reaction ii), whereas the Pt(II) complex 149b promoted enyne cycloisomerization with a reaction rate similar to that of 147 (Fig. 39, reaction iii). Thus, both complexes exhibited highly electrophilic character of their metal centers, irrespective of the formal charge distribution within the [Sb–M]n+ core (n = 3 for M = Au; n = 2 for M = Pt).
The first examples of sulfonium complexes were reported by Adams who demonstrated that metal coordinated sulfonium cations can be obtained by S-alkylation of the corresponding sulfide complexes (Fig. 41a).120a–d Comparing the available XRD structures of complexes 150–152a,b (Fig. 41b and c) shows that this alkylation results in a noticeable shortening of the M–S bond length of ligands (2.30 Å vs. 2.48 Å, on average). This is particularly apparent in the structure of a macrocyclic Mo(0) complex 149b (Fig. 41c) obtained later by Kanokogi,120e since it contains both sulfide- and sulfonium-type S-donors coordinated to the same metal center.
In addition to the S–M bond shortening, transformation of a sulfide donor in 150–152a into a sulfonium in 150–152b results in a blue shift of 30–40 cm−1 in the carbonyl stretching frequencies. Furthermore, comparing the IR data of complex 150b with complexes of the general formula CpMn(CO)2L121 places sulfonium cations on par with strongly electron withdrawing ligands such as PPhCl2 and PCl3.
Adams also found that metal-coordination of sulfonium cations increases their proneness to dealkylation, leading back to the sulfide complexes.120a,b,d In addition, due to their decreased donicity, sulfonium ligands are relatively easily displaced by stronger donors, such as phosphines or isonitriles,120a–d even when supported by a chelating arm, as in complex 151b.120c,d
Once again, the idea of using a more robust tridentate pincer scaffold for stabilizing sulfonium complexes comes into mind. It also occurred to us that converting sulfide ligands to sulfonium prior to metalation might provide a more general approach than that proposed by Adams. To explore this direction, we prepared both aliphatic and aromatic sulfonium pincer ligands 154 and 157–158 by S-alkylation/arylation of the corresponding protected PSP-based sulfide ligands 153 or 155–156 (Fig. 42).122 Since sulfur has no useful NMR active nuclei, we decided to study the coordinative behavior of these ligands with Rh(I) and Pt(II), both of which have spin 1/2 isotopes. In this way, magnetic interactions between NMR active nuclei of the ligands (1H, 19F) and the metals (103Rh, 195Pt) could be used for monitoring coordination of sulfonium in solution by multinuclear NMR.122a,b,d Indeed, the 1H signal corresponding to the methylene protons of the ethyl tail in complex 159a (obtained by reaction of the Rh(I) precursor with 154, Fig. 43a) was not only shifted downfield compared to the free ligand, but also showed an additional splitting due to magnetic interaction with the 103Rh nucleus (3JRh–H = 1.3 Hz, Fig. 43b). Furthermore, the striking difference in 1H–195Pt coupling constants of these protons to Pt in 161a and 160a (3JPt–H = 7.7 vs.6JPt–H = 0.2 Hz, respectively) provides a clear-cut distinction between the S-coordinated complex and the S-non coordinated one (Fig. 43a).122a The fluoride substituent in ligand 157 plays a similar role of a “reporter” nucleus, showing a prominent cross-peak in the 19F–195Pt HMBC spectra of 161b,c, but none in those of 160b,c (Fig. 43c).122a,c In addition, the reduced 1JP–M coupling constants (M = 103Rh and 195Pt) observed for the sulfonium complexes 159a,b and 161a–c compared to the analogous sulfide pincer complexes 162a–c123 (Table 14, 3rd column) showcase the electron withdrawing character of sulfonium ligands.
Complex | S–M bond (Å) | 1 J P–M (Hz) | Ref. # | |
---|---|---|---|---|
1 J Pt–P value of complex 168 is the smallest among all sulfonium–Pt(II) complexes we obtained so far (Table 14, the 3rd column). | ||||
Sulfonium complexes | 159a | 2.126(2) | 127.8 | 122a |
159b | 2.112(1) | 126.0 | 122a | |
161a | 2.258(1) | 2736 | 122a | |
161b | 2.261(1) | 2768 | 122a | |
161c | 2.187(14) | 2202.7 | 122b | |
167 | 2.225(3) | 2818 | 122c | |
168 | 2.1632(19) | 2037 | 122c | |
Sulfide complexes | 162a | No data | 147 | 123a |
162b | 2.254(1) | 2547 | 123b | |
162c | 2.336(2) | 2923 | 122a | |
169 | No data | 2280 | 122c |
The obtained XRD structures (Fig. 43d) were fully consistent with the NMR results. In the case of complexes 159a and 159b the S–Rh bond lengths of 2.126(2) and 2.112(1) Å, respectively, are among the shortest reported for this pair of elements. The S–Pt bonds in PtMe complexes 161a,b are somewhat longer, which can mostly be attributed to the strong trans influence of the Me group, rather than to electrostatic repulsion within the dicationic [S–M]2+ core, because in the PtCl complex 161c this bond is only a little longer than that in 159a,b. Yet, even in the presence of a Me group at the trans position, the M–S bond in sulfonium complex 161a is still shorter than that in the corresponding sulfide complex 162c (Table 14, the 2nd column).
The short S–M bond lengths in sulfonium complexes compared to the sulfide ones suggested a high degree of π-back bonding in the former. This was confirmed by a computational analysis we undertook,122a which showed that while bond dissociation energy and σ-donation of sulfonium cations are nearly the same as those in sulfide and sulfoxide ligands, these cations exhibit significantly stronger π-acceptor properties. This was not surprising in light of the earlier ab initio calculations on isoelectronic DMSO and SMe3+ ligands coordinated to an anionic PtCl3− fragment, pointing out that the bonding of the second is predominated by Pt(d) to S–C(σ*) π-back donation.124 What was less anticipated, however, is the profound effect of the pincer framework on the S–M bonding. Comparison between the monodentate and tridentate sulfonium ligands by energy decomposition analysis (EDA-NOCV) showed that perturbation of sulfonium- and metal-based orbitals imposed by the pincer scaffold significantly increases the overall π/σ ratio. Obviously, the π-back donation from the cationic PtMe+ fragment is weaker than from the neutral RhCl, but is still comparable to the σ-donation in magnitude.122a
Formation of complexes 159–160a–c showed that pincer ligands 154 and 157 are hemilabile, i.e. capable of exhibiting both tridentate mer-κ3-PSP and bidentate cis-κ2-PP coordinating modes. A further study revealed that despite the stabilizing effect of the pincer scaffold, the sulfonium moiety can still be displaced from the metal coordination sphere by other donors, such as coordinating solvents, CO or halides.122b,d This hemilability is also determined by the flexibility of the ligand's backbone. The more flexible aliphatic ligand 154 allows not only the formation of bidentate complexes, such as 160a–c, but also homodimers of two different types (Fig. 44). In the first of them, μ2-(PSP)2, the pincer ligands open up forming a bridge between two metal centers (complexes 163 and 164), while in the second, μ2-(Cl)2, the two ligands retain their chelating mode (complex 166). Both types of binuclear complexes are obtained as a mixture of two stereoisomers differing by a mutual orientation of the S-ethyl tails (syn or anti). Conversely, with the more rigid aromatic ligand 157 no binuclear complexes were isolated.122b
Fig. 44 Formation of two possible homodimeric complexes with a hemilabile sulfonium ligand (a) and their X-ray structures (b). |
The easy dissociation of the S–M bond in complexes of ligands 154 and 157 precludes realizing the full potential of their strongly π-acidic sulfonium centers as ancillary ligands for electrophilic catalysis. Since this bond cleavage is accompanied by a trans-to-cis interconversion of the phosphine arms (Fig. 45a) we reasoned that suppressing this process is crucial for better stabilizing the S–M bonding. Therefore, we prepared ligand 158, a rigidified analogue of 157 (Fig. 42b), where the two aromatic rings of the backbone are linked together by a carbonyl bridge. This keeps the phosphine arms further apart, not allowing the cis-κ2-PP coordination (Fig. 45b).
Fig. 45 Cis–trans isomerization of the phosphines in flexible (a) and rigid (b) sulfonium pincer complexes. |
Indeed, the rigid thioxanthone-based scaffold of ligand 158 proved capable of enforcing the coordination of its sulfonium center not only to a mono-cationic PtMe fragment (like in 161b), but also to a bis-cationic Pt(MeCN) fragment (complexes 167 and 168 in Fig. 46a).122c The latter goal could not be achieved with the flexible sulfonium ligands 154 or 157, despite numerous attempts.122b,d XRD structures of complexes 167 and 168 (Fig. 46b) showed that ligand 158 adopts the desired mer-κ3-PSP coordination mode by bending its thioxanthone backbone along the S–CCO axis, which effectively pushes the sulfonium center closer to the Pt. As a result, the S–Pt bond in the bis-cationic complex 167 is ca. 0.04 Å shorter than in its analogue without the carbonyl bridge (161b). The tris-cationic complex 168 exhibits an even shorter S–Pt bond of only 2.163 Å, despite the higher electrostatic repulsion between its sulfonium and Pt(II) centers (Table 14, 2nd column).122c
Computational analysis of sulfonium–metal interactions in this unusual tris-cationic complex followed the same trend of increasing σ/π ratio we observed earlier when comparing sulfonium complexes of mono-cationic Pt(II) and neutral Rh(I) fragments (161a–c and 159a,b, respectively).122a In 168, with an even more electron-deficient bis-cationic Pt(II) center, the σ-donation becomes the dominant bonding interaction, by far stronger than the π-back donation. Thus, in terms of the S–Pt bonding, this sulfonium complex is not very different from an analogous sulfide-based pincer complex 169 (Fig. 46a), which we prepared for comparison. Yet, a weaker σ-donation (89.3 vs. −104.0 kcal mol−1), but a stronger π-back bonding (−36.1 vs. −29.0 kcal mol−1), results in a somewhat more positive natural charge on Pt in the sulfonium complex compared to its sulfide analogue (+0.402 vs. +0.375).
The enhanced electrophilicity of the Pt(II) center in 168 with respect to 169 is also evident from the comparison of the 1JPt–P constants of these two complexes (2037 vs. 2280 Hz). In fact, the presence of a cationic sulfonium moiety within the Pt(II) coordination sphere has a profound effect on its catalytic activity, as illustrated by cycloisomerization reactions i–iii (Fig. 46c). For instance, in presence of 5 mol% complex 168 cycloisomerization of o-binaphthalene (reaction ii) was complete in only 6 hours at 80 °C, which is on par with the best Pt(II) catalysts for this reaction reported by Alcarazo.125 At the same time, less than 10% conversion was observed with the sulfide complex 169 under identical conditions, proving the great utility of sulfonium-based ancillary ligands for π-acid catalysis.
Metal coordinated Te(IV)-based species, such as the anionic Lewis adduct 170 (Fig. 47a), are also known, although they are much rarer.127 Related complexes of Te(IV) ligands could also be obtained by oxidative addition of M–I bonds (M = Co, Fe or Ni) to organotellurium halides (Fig. 47a).128 Although no in-depth computational analysis was performed on the resulting complexes, the observed seesaw geometry of the Te(IV) centers (Fig. 47b) indicates the presence of a non-bonding lone pair. In such a case, these ligands in complexes 172a–c can be regarded as cationic Te(IV) species (dihalotelluronium) acting as Z-type ligands (resonance form ii).
Fig. 47 Transition metal complexes of neutral and cationic Te(IV) ligands (a) with representative XRD structures (b). |
Formation of a related metal-coordinated Te(IV) species with a partial halotelluronium character was observed by Gabbai in a PTeP pincer system.129 Similar to the previously discussed stibines, telluroethers exhibit coordination and redox non-innocence, which blends the border between Te(II) and Te(IV) species.130 Following this tendency, the formally divalent Te center in the cationic complex 174 (Fig. 48a) is engaged in secondary bonding interactions with its Cl− counterion (Te–Cl: 3.117 Å), as evident from its XRD structure (Fig. 48b).129 However, when this complex is oxidized by PhICl2 into 175, the Te–Cl distance shrinks, approaching the length of a typical covalent Te–Cl bond in bonafide Te(IV) compounds (e.g. 2.529 Å in Ph2TeCl2).131 While the Te–Pt bond length shows only a minor change upon oxidation of 173 into 174 (2.635 vs. 2.528 Å), the NBO analysis reveals that this is associated with an umpolung reaction of this bond, which converts from Te-polarized (57% Te/39% Pt) to Pt-polarized (35% Te/65% Pt). Such oxidation-induced polarity inversion appears to be quite similar to what was observed for the PSbP–Au complex 133 (Fig. 36) and is consistent with a description of 174 as a telluronium(IV) platinate(II) complex (resonance structure i in Fig. 48a). However, unlike in the Sb system, here it is the Pt (and not Te) center that undergoes the oxidative addition of Cl2, upon which it changes its geometry from square planar, typical of Pt(II) complexes, to octahedral, typical of the Pt(IV) ones. The Pt-centered nature of this oxidation also manifests in a drastic reduction of the 31P–195Pt coupling constant from 2480 Hz in 173 to 1777 Hz in 174. These structural and spectroscopic features of 174 are more consistent with a TeIIPtIV core (resonance structure iii). In order to reconcile these discrepancies, the authors chose to represent 174 as a covalent TeIII–PtIII complex (resonance structure ii).
Fig. 48 Synthesis and reactivity of a PTeP–Pt(II) complex with a partial telluronium character (a) and the XRD structures of complexes 173–175 (b). |
The highly electron deficient character of the Pt center in 174 determines the reactivity of this complex. Hence, upon irradiation with uv it undergoes photoelimination of Cl2, converting back into 173 (Fig. 48a). On the other hand, converting the Te-bound chloride into fluoride in an alcoholic solution results in a coordination induced Te → Pt 2e− transfer. This intramolecular redox process is obvious from the geometry changes occurring in both Te and Pt centers of the product (175): the former transforms from seesaw to octahedral and the latter from octahedral to square planar (Fig. 48b). The reduction of the Pt center is also evident from the larger 31P–195Pt coupling constant of 175 compared to 173 (2845 Hz vs. 1777 Hz).
While the telluronium character of the Te center in 174 is quite arguable, subsequent work performed by the same group presented more explicit examples of metal coordinated telluronium(IV) cations.132 The most representative case is that of a tris-8-quinolinyl-telluronium species 176, which was characterized both in its metal-free and coordinated forms (Fig. 49a). The XRD structure of complex 177 (Fig. 49b), formed upon a direct reaction of this ligand with an anionic Pd(II) precursor, clearly shows the presence of an elongated Te–Pd bond of 2.920 Å. The Te center exhibits the same seesaw geometry as in complexes 172a–c and 174, implying that here too the telluronium lone pair remains non-bonding.
Fig. 49 Synthesis of Pd(II) complexes of chelating and pincer-type telluronium ligands (a) and their XRD structures (b). |
Comparing the geometries of metal-free and coordinated telluronium species shows that metal-coordination has no effect on the degree of s–p hybridization at the Te atom, and perhaps this is why only a negligible coordination shift was observed in 125Te NMR. Nevertheless, NBO calculations identified substantial donor–acceptor interaction between the occupied dz2 orbital of Pd and a vacant σ* Te–C orbital associated with the quinolinyl substituent situated trans to Pd. Therefore, in stark contrast to the isoelectronic sulfonium-based L-type ligands discussed above,122 telluronium cation 176 binds as a purely Z-type ligand.
Despite the additional stabilization by the chelating quinoline group, the Pd(II) → Te(IV) donor–acceptor interaction in complex 177 is insufficient to prevent its dissociation in coordinative solvents (DMSO). Nevertheless, it was shown that this interaction can be significantly enhanced by replacing one of the three quinolinyl groups with a more electron withdrawing substituent. The pincer-type complex where such a hydroxytelluronium(IV) center is coordinated to Pd(II) was obtained from a neutral bis-8-qunolinyl-Te(IV) species 178 in the form of a bis-cationic μ2-(Cl)2-bridged homodimer 179 (Fig. 49a). This structure is somewhat reminiscent of the previously discussed μ2-(Cl)2-bridged homodimeric complexes of sulfonium 166 (Fig. 44)122b with the crucial difference that here the central chalcogen atom is metal-coordinated. Computational analysis found that with the hydroxyl group trans to Pd the strength of Pd → Te interaction in 179 is nearly 3 times stronger than in 177 (ΔEdel = 69.8 vs. 22.5 kcal mol−1), which also correlates with a shorter bond length (2.783 Å vs. 2.920 Å).
This journal is © The Royal Society of Chemistry 2024 |