True and masked three-coordinate T-shaped platinum(II) intermediates

Although four-coordinate square-planar geometries, with a formally 16-electron counting, are absolutely dominant in isolated Pt(II) complexes, three-coordinate, 14-electron Pt(II) complexes are believed to be key intermediates in a number of platinum-mediated organometallic transformations. Although very few authenticated three-coordinate Pt(II) complexes have been characterized, a much larger number of complexes can be described as operationally three-coordinate in a kinetic sense. In these compounds, which we have called masked T-shaped complexes, the fourth position is occupied by a very weak ligand (agostic bond, solvent molecule or counteranion), which can be easily displaced. This review summarizes the structural features of the true and masked T-shaped Pt(II) complexes reported so far and describes synthetic strategies employed for their formation. Moreover, recent experimental and theoretical reports are analyzed, which suggest the involvement of such intermediates in reaction mechanisms, particularly C–H bond-activation processes.

tron Pt(II) complexes are believed to be key intermediates in a number of reactions, e.g., β-hydrogen elimination, thermal decomposition of dialkyls, insertion of olefins into M-H bonds, electrophilic attack at Pt-C bonds, and ligand cycloplatination [2]. Likewise, related Pd(II) complexes [3,4] are relevant in cross-coupling reactions and C-H bond-activation processes. The accessibility of three-coordinate Pd(II) species have been recently discussed [5].
Three-coordinate Pt(II) intermediates are the focus of this review. Despite the kinetic perception of the intermediacy of these coordinatively unsaturated species in important organometallic processes, direct proofs of 14-electron Pt(II) complexes have been difficult to find. The strong readiness to alleviate the unsaturation makes them very reactive species but hampers their isolation. Low-coordinate Pt(II) complexes have been generated in gas-phase experiments [6,7] but have remained elusive in solution. With a few exceptions the fourth coordination site is occupied by a weak ligand, e.g., an agostic interaction, a counteranion or a solvent molecule. However, if this additional interaction is weak and labile enough, the threecoordinate species is very accessible and the complex can be considered as "operationally unsaturated" in a kinetic sense [8]. We will name these compounds "masked" three-coordinate complexes to distinguish them from the true low-coordinate complexes. In this review, we will summarize recent advances in true and masked three-coordinate Pt(II) complexes, highlighting both their structural features and their possible participation as reaction intermediates. Computational studies have become an invaluable tool for the investigation of short-lived elusive intermediates and will be quoted throughout the article. The review is organized as follows. First, a general picture of the electronic and geometrical structure of three-coordinate Pt(II) complexes will be presented. Then, the structural features of the main families of these compounds will be summarized. The next sections will be devoted to the spectroscopic tools for their detection and the synthetic strategies employed to their formation. Afterwards, the rearrangement processes exhibited by the low-coordinate complexes in solution will be discussed. Finally, participation of three-coordinate Pt(II) intermediates in reactions, mainly C-H bond-activation processes and ligand exchanges, will be analyzed. A thorough review on the bonding and stereochemistry of three-coordinate transition-metal compounds was published several years ago [9].

Electronic and geometrical structure
As we will discuss later on, three-coordinate, 14-electron Pt(II) d 8 complexes display a T-shaped structure, i.e., a structure with two ligands mutually trans and the third ligand trans to the vacant position. A qualitative molecular orbital scheme of the d block of a T-shaped d 8 metal complex can be easily derived from that of a square-planar complex by removing one of the ligands ( Figure 1) [10,11]. The three nonbonding orbitals in square-planar complexes (d xz , d xy and d yz ) are not affected by the removal of a ligand. The d z 2 orbital is slightly stabilized due to the disappearance of a small antibonding interaction. However, the most striking difference is observed in the lowest unoccupied d x 2 −y 2 orbital; its energy strongly decreases by removing an antibonding interaction with one ligand and its shape changes by mixing with the p y orbital. This hybridization takes place so that the orbital is directed away from the three ligands toward the empty coordination site, making this position suitable for the approach of an incoming ligand. This simple picture has been corroborated by DFT calculations on T-shaped [Pt(alkyl)(PMe 3 ) 2 ] + complexes [2].
The splitting of the d-block orbitals favors a singlet ground state for T-shaped d 8 complexes and these species usually exhibit a low-spin reactivity. Recent DFT calculations on the 14-electron [Pt( F PNP)] + complex ( F PNP = (4-F-2-(iPr 2 P)C 6 H 3 ) 2 N) predict a relatively small (10-12 kcal mol −1 ) separation between the singlet and the triplet states of this intermediate [12]. It is worth pointing out that a d 8 C 2v ML 3 fragment is isolobal with CH 2 , although the ordering of the two valence orbitals a 1 and b 2 differs. For a singlet 14-electron T-shaped ML 3 , the two electrons fill the b 2 level (d yz ), while for a singlet CH 2 fragment they occupy the a 1 orbital (empty d x 2 −y 2 in ML 3 , Figure 1) [13]. Steric reasons should favor trigonal-planar D 3h -like structures for three-coordinate Pt(II) complexes instead of the sterically unfavorable T-shaped structure. The preference of these Pt(II) compounds (with a low spin d 8 configuration) for T-shaped structures is due to electronic effects and can be understood from the Walsh diagram shown in Figure 2. The Walsh diagram describes the variation of the L-M-L angle from a T-shaped structure through D 3h to a Y-shaped structure. A d 8 , 14-electron ML 3 complex would have the lowest four levels filled in Figure 2. In a D 3h geometry, the degeneracy of e' orbitals will promote a Jahn-Teller distortion towards T or Y geometries [10,14]. DFT calculations on asymmetric cis-[Pt(alkyl)(PMe 3 ) 2 ] + show that, in spite of the reduced symmetry, the bending energy profile maintains the same basic features: the T-shaped configurations of the 14-electron species are energy minima, and their cis-like to trans-like interconversion occurs via transition states with Y-shaped configurations [2].
The X-ray structure of the recently reported three-coordinate platinum complex [Pt(SiMe 2 Ph) 2 (IPr)] (IPr = 1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene) shows a unique Y-shaped geometry in which the Si-Pt-Si angle is very acute (80.9°) and far from the ideal values for both trigonal-planar and T-shaped structures (Y1, Figure 3) [15]. Computations on non-sterically demanding models [Pt(R) 2 (Im)] (R = SiMe 3 , Me; Im = imidazol-2-ylidene) appealed to the trans influence of both NHC and silyl ligands to explain the structure. However, a recent DFT investigation concluded that Y1 is better described as a Pt(0) σ-disilane complex [16] than as a Pt(II) disilyl species. Thorough geometrical and electronic analyses support a Pt(0)···disilane coordination via donation and back-donation interactions. This suggestion also explains the experimentally observed 195 Pt NMR chemical shift, which is closer to the Pt(0) rather than the Pt(II) species.
The nonequivalence of axial and equatorial positions in ML 3 complexes raises the question of which positions are preferred by the ligands in such compounds. Although in most of the experimentally characterized systems the steric demands dictate the disposition of the ligands, electronic effects are also at work, and they were analyzed for simple model systems with theoretical methods. An early study based on perturbation theory concluded that in T-shaped ML 3 d 8 complexes both axial and equatorial bonds should have similar strength and that the most electronegative atoms will substitute axially [17]. DFT calculations on [PtXY(PH 3 )] (X, Y = Cl, CH 3 , SiH 3 , Si(OH) 3 ) demonstrated the importance of the trans influence in governing the stability of T-shaped isomers. In the most stable isomer, the ligand with the smaller trans influence is located trans to the PH 3 ligand [18].

Structurally characterized compounds
The T-shaped structure is easily recognized by the absence of one ligand in a square-planar disposition. Nevertheless, the number of well-characterized three-coordinate Pt(II) complexes is very low. To achieve a true T-shaped structure, the vacancy at the metal center must be blocked to avoid intra-and intermolecular interactions, such as agostic bonds and counteranion or solvent coordination. These interactions mask the T-shaped structure, but due to their potential labile nature, the three-coordinate species are still accessible.
In this section we will describe the main families of T-shaped Pt(II) complexes, which are structurally characterized ( Figure 4). First, true T-shaped structures with no stabilization at the vacant site are compiled. Then, complexes involving agostic, counteranion and solvent interactions will be summarized. From now on, true T-shaped complexes are labeled as T complexes, while agostic, counteranion and solvent-stabilized complexes are designated as A, C and S complexes, respectively.  True T-shaped The isolation and characterization of such highly electrophilic species is challenging. Accordingly, the number of well-characterized true T-shaped Pt(II) complexes is scarce. The empty site should be blocked somehow to prevent undesirable contacts, and in most cases, bulky ligands are required to protect the empty coordination site. Strong electron-donor ligands also help stabilizing the formally 14-electron compounds. It is worthy of note that almost all of them are cationic.
The first true T-shaped complexes were prepared by using phosphine ligands ( Figure 5). In the 80s, Goel et al. proposed, from nuclear magnetic resonance (NMR) data, a T-shaped structure for the cationic hydride trans-[(t-Bu 3 P) 2 Pt(H)] + T1 containing bulky phosphine ligands [19]. However, it was not until 2005 when Braunschweig et al. successfully characterized a true 14-electron T-shaped Pt(II) boryl complex T2a by means of X-ray studies [20]. By halide abstraction, the cationic trans-    [22]. DFT-based electron localization function (ELF) analyses [20] and geometry optimizations [22] supported the lack of agostic interactions in these complexes. This absence has been attributed to the strong trans influence exerted by the ligand in trans position with respect to the vacant site [23,24].
Moreover, in the past few years the synthesis of 14-electron Pt(II) complexes was extended by using N-heterocyclic carbene (NHC) ligands ( Figure 6), which have been proven to be useful stabilizing electron-deficient transition-metal species [25][26][27]. In this regard, recent studies state that the use of IMes* (4,5dimethyl-1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene) and IMes (1,3-bis(2,4,6-trimethylphenyl)imidazol-2-ylidene) ligands in [Pt(Me)(NHC) 2 ] + T4 and [Pt(NHC')(NHC)] + T5 (NHC = IMes*, IMes; NHC' = cyclometalated ligand) provides pure T-shaped species with no agostic stabilization [28]. Additionally, the resulting [Pt(Ar)(IMes*) 2 ] + T6 formed after C-H bond activation has also proven to be a three-coordinate species with no agostic interactions according to the X-ray structure of the derivative T6d, in which the closest Pt-H contact is located at 3.117 Å. The absence of agostic interactions has been attributed to geometrical constraints, the limited flexibility of the mesityl groups in IMes* and IMes hampering the approach of the CH bond to the metal center. Once again, Lewis acids as acetonitrile can access the empty site, which means that a potential reactant molecule can coordinate to the metal for further reactions.

Masked T-shaped via agostic interaction
The agostic interaction [29,30] is usually explained as an intramolecular 3-center-2-electron interaction between a metal M and a C-H bond. This type of contact is a recurrent event in unsaturated transition-metal complexes [31] and it can be characterized by structural and spectroscopic techniques [29,30] together with computational tools [32,33]. The agostic interaction shows different behaviors. It can be a transient species prior to the C-H bond breaking or stabilize low electron count situations. In this section we will focus on the latter aspect.
Since this intramolecular interaction is difficult to avoid, even in the presence of bulky ligands, it is the most common motif for the stabilization of three-coordinate complexes.
Early works on T-shaped Pt(II) species are related to complexes stabilized with phosphine ligands (Figure 7). Pioneering studies on alkyl complexes from Orpen's and Spencer's research groups reported the synthesis of cationic [Pt(norbornyl)(P-P)] + complexes A1a-e (P-P = bidentate phosphine ligand) [34,35]. Further studies involved other alkyl ligands such as ethyl (A2b-c), 3,3-dimethylbutyl (A3b-c) and 2,3,3-trimethylbutyl (A4) [36][37][38]. NMR spectroscopic data and X-ray structures of A1a [34] and A2b [36] demonstrate that these compounds display a β-agostic interaction filling the fourth coordination site. This interaction is strong enough to cleave the C-H bond, creating a chemical equilibrium with the corresponding hydrido-alkene derivatives. Indeed, the detected ground state of the compounds in brackets (A2a, A2d, A2e and A3a in Figure 7) is the hydrido-alkene isomer instead of the agostic-alkyl one. Later, Baratta et al. succeeded in the isolation of 14-electron Pt(II) complexes bearing bulky phosphine  ligands [39]. The usage of trans-[Pt(Me)Cl(PR 3 ) 2 ] as a starting material, followed by halide removal and the release of methane by intramolecular C-H bond activation provided the cationic T-shaped cyclometalated complexes A5a-b. The X-ray structure of A5b shows that the platinum atom exhibits a δ-agostic interaction with one hydrogen atom of the methyl group of the non-cyclometalated phosphine ligand. Subsequent hydrogenation generates the corresponding hydride complexes A6a-b. On the basis of NMR studies ( 1 J Pt,H values of ca. 2000 Hz, see next section), a weak interaction trans to the hydride ligand, typically an agostic contact, is supposed to exist. Carmona and co-workers also prepared quite similar complexes, A5c and A6c, where R labels represent isopropyl groups [40]. Weller and co-workers reported the formation of complex trans-[Pt(Me)(PiPr 3 ) 2 ] + A7 in which a γ-agostic interaction is located in the fourth coordination site [41]. The addition of tetrahydrofuran (THF) rapidly forms the corresponding adduct. Braunschweig et al. obtained the agostic structure A8 by employing BCat as a ligand (BCat = catecholatoboryl) in three-coordinate Pt(II) complexes [21]. Unlike the previous pure T-shaped boryl complexes T2 ( Figure 5), the weaker trans influence of the BCat ligand allows the formation of an agostic contact. Treatment of A8 with Lewis bases also removes the agostic interaction, generating the adduct complex.
Natural Bond Orbital (NBO) and Atoms In Molecules (AIM) calculations on complexes A1, A2b, A5b and A7 classified the Pt···H-C contacts as agostic interactions for all the species with the only exception being A7, in which a H-bond character can also be ascribed [42].
Concerning other types of ligands (Figure 8), the phenylpyridyl complex A9 reported by Rourke and co-workers exhibits a bifurcated δ-agostic interaction that has been determined by X-ray studies [43]. This is one of the rare neutral T-shaped  [44]. In sharp contrast with the analogous T5 and T6 ( Figure 6) [28], δ-and ζ-agostic interactions at the fourth coordination site were detected by X-ray and NMR studies for complexes containing IPr (A11a) and It-Bu (A11b), respectively.
Selected Pt···H-C parameters from the X-ray structures of agostic compounds are collected in Table 1. As it can be expected, short Pt-H and Pt-C distances are observed for the β-agostic interactions in A1a and A2b. In addition, Pt-H-C angles larger than 100° correlate with large Pt-C distances as shown in A7, A8 and A11a. Although a remote contact is described in A11a, the largest distances correspond with the γ-agostic A8. In other words, the strength of the agostic interaction is not only governed by geometrical constraints and the surroundings of the CH group, but also by the trans influence of the ligand in trans position with respect to the agostic interaction [23,24]. In this case, the boryl ligand in A8 exhibits a higher trans influence than the alkyl group in A11a [45,46], and therefore the agostic interaction in the former is weaker.

Masked T-shaped via counteranion interaction
Most of the T-shaped Pt(II) complexes are cationic. Thus, there is the possibility of stabilizing the unsaturated structure by nonbulky, weakly coordinating counteranions. Triflate (OTf = SO 3 CF 3 ) and tetrafluoroborate are the best candidates to play such a role ( Figure 9). All the well-characterized compounds bear a coordinated triflate anion. In the 80s, Whitesides et al.

Masked T-shaped via solvent coordination
In solution, a solvent molecule can occupy the vacant position of a T-shaped three-coordinate Pt(II) compound via σ-interaction. A number of phosphine-based complexes with different solvent molecules coordinated to the T-shaped frame have been characterized ( Figure 10). In this regard, Kubas and co-workers prepared a set of hydride trans-[PtH(PiPr 3 ) 2 (solv)] + compounds where "solv" stands for η 1 -ClCH 2 Cl (S1a), OEt 2 (S1b) and THF (S1c) solvent molecules [54]. Interestingly, the closely related species T1 ( Figure 5) does not include any solvent molecule filling the vacancy [19]. Moreover, the structures of S1a and S1c were fully confirmed by X-ray studies. The similar dichloromethane adduct S1d was later detected by NMR spectroscopy [55]. This work was extended to methyl complexes  with the formula trans-[Pt(Me)(PR 3 ) 2 (solv)] (S2) [56,57]. Similar structures have been obtained by the use of chelating phosphine ligands in diethylether (S3) [58]. In this line, Peters and co-workers developed several bidentate phosphine ligands to yield complexes S4 in which one THF molecule is coordinated to the metal center [59,60]. Romeo's group was actively working on solvento Pt(II) complexes S5. For instance, they employed an extended series of phosphine ligands to obtain cis and trans-solvento complexes containing methanol [61]. The formation of acetonitrile-d 3 adducts with triethylphosphine ligands was also reported [2,62].
Pincer ligands [82,83] naturally define a three-coordinate environment and, in this way, they are well-suited as ligands in T-shaped compounds. Accordingly, a number of T-shaped Pt(II) pincer complexes with an additional solvent molecule as a fourth ligand have been reported (S11-19, Figure 12). Van Eldik and co-workers studied the aquo complexes S11 by using several tridentate nitrogen-based ligands (NNN) [84]. Peters and co-workers isolated [Pt(OTf)(NNN)] species, though NMR data were collected in acetonitrile-d 3 solution, in which the solvento adduct S12 is favored [50]. Extensive work carried out by van Koten and co-workers [85,86] concerned NCN aquo compounds such as S13. Unsaturated complexes bearing NCP ligands were characterized by Milstein and co-workers, including water (S14a) [51] and acetone (S14b) [87] solvent molecules. Isolation of compound S15 in acetone/pyridine solution provided suitable crystals for X-ray studies, which showed that a pyridine molecule has been added [87]. Other investigations have been devoted to complexes with PCP ligands. For instance, Bullock and co-workers prepared a dihydrogen adduct of the type [PtH 2 (PCP)] + in dichloromethane, but NMR data suggest a mixture of the former compound and the corresponding solvento complex S16a [88]. Milstein et al. proposed that the closely related complex S16b exists in THF solution, since an interaction with the BF 4 − counteranion is not observed, although the elemental analysis evidences the lack of the THF molecule in the solid state [89]. Contrarily, the naphthyl deriva-tive S17 was found together with the BF 4 − adduct C8 (Figure 9) [53]. For silicon-related ligands, Turculet and co-workers successfully crystallized S18 in a diethylether solution at low temperature, showing one diethylether molecule directly coordinated to the platinum center [90]. A set of dicationic complexes based on carbene groups have been developed by Limbach and co-workers, forming pyridine S19a and acetonitrile S19b,c adducts [91].
The relevance of solvent-stabilized T-shaped Pt(II) complexes extends beyond the organometallic field. Their presence in cisplatin-protein adducts has also been proposed concerning the bovine Cu, Zn superoxide dismutase [92] and the hen egg white lysozyme [93]. The corresponding crystallographic structures show the platinum coordinated to a histidine residue of the protein and two ligands (Cl − [92] or NH 3 [93]). However, in both situations the fourth ligand was not fully detectable, so that the authors suggested the participation of a coordinating water molecule. Indeed, quantum mechanics/molecular mechanics (QM/MM) calculations on the cisplatin-hen egg white lysozyme adduct confirmed the facile inclusion of a solvent water molecule in the first coordination shell of the platinum complex ( Figure 13) [94].
NMR coupling constants to 195 Pt as sensitive probes for coordinatively unsaturated Pt(II) complexes NMR spectroscopy has provided additional evidence for the existence of low electron-count Pt(II) complexes not only Figure 13: Structure of the QM/MM optimized cisplatin-protein adduct [94].
through the observation (in some cases) of NMR signals for the C-H bonds involved in agostic interactions, but also by the magnitude of the coupling constant of some of the ligands around the metal center with the NMR active 195 Pt nuclei (33.7% natural abundance). This is actually not the case with 31 (Figure 10, R = iPr S1a, Cy S1d) exhibit a smaller J Pt,H of 1480 (R = Cy) or 1852 Hz (R = iPr) in agreement with the presence of a ligand trans to the hydride. With regard to Pt-H two-bond and Pt-C one-bond coupling constants in Pt-C alkyl complexes ( Figure 15 and  In those cases where solvent molecules (THF, acetonitrile) have been reported to coordinate to some of the complexes shown in Figure 15, smaller values of both the 2 J Pt,H and the 1 J Pt,C coupling constant are observed. For example, the 2 J Pt,H of the methyl ligand decreases to 98 Hz in the THF adduct of derivative A7, whereas the acetonitrile adduct of complex A11b Figure 15: The chemical formula of the complexes discussed in Table 2.
Synthetic routes to stable, solvent-stabilized and transient 14-electron Pt(II) species True and agostic T-shaped Pt(II) complexes Several methods have been described to prepare coordinatively unsaturated Pt(II) complexes with a T-shaped geometry. Although the number of species that have been authenticated by crystallographic or spectroscopic methods is still very limited, the best and most general method for obtaining them is by removing a halogen ligand (Cl, Br, I) from the platinum coordination center in 1 by using a halogen abstractor with a poor coordinating anion, such as tetrakis[3,5-(trifluoromethyl)phenyl]borate or hexafluoroantimonate (Scheme 1) [19][20][21]28,39,44].
This procedure has been successfully employed for the generation of complexes stabilized by two phosphine PR 3 or two NHC ligands, in which the third coordination site is occupied by alkyl, hydride or boryl ligands. Therefore, high trans-influence ligands are present in all these cases, favoring the dissociation of the Pt-X bond. The generality of this method is so powerful that it has provided access to an intriguing dicationic borylene Pt(II) complex T3 which is not stabilized by agostic interactions (Scheme 2) [22].

Scheme 2:
Halogen abstraction from 2 forming the dicationic complex T3 [22]. Alternatively, Carmona and co-workers have recently described that the hydrogenation of the 16-electron Pt(II) carbene 3 bearing a cyclometalated phosphine ligand (Scheme 4) resulted in the formation of the cyclometalated complex A5c, which can be further hydrogenated to give the Pt(II) hydride A6c [40], similar to those reported by Baratta [39].

Pt(II) complexes stabilized by solvent molecules
Although some of these compounds have been prepared by halogen [58] or methide [74]

Solution behavior
In masked three-coordinate d 8 complexes, the fourth coordination site is occupied by a weakly bound ligand, which can be easily displaced. In this way, in solution the ligands of these compounds are opened to conformational events that will be discussed in this section. The empty site can also be available for incoming substrates, allowing the participation of these complexes as reaction intermediates. This aspect will be discussed in the next section.
Most of the reports of fluxional processes correspond to agostic bonded complexes. The equivalence of 1 H and 13 C chemical shifts observed in solution for the γ-, δ-and remote-agostic contacts is a common feature of the complexes collected in Figure 15 and arises from the rapid intramolecular exchange of the C-H bond involved in the agostic interaction. In several cases, such as A11a [44], the overlapping of the proton signals and signal averaging of the ligands avoid drawing a conclusion about the nature of the agostic interaction at work in solution.
Recently, ab initio molecular dynamics (AIMD) simulations of some representative T-shaped Pt(II) complexes (T5b, A2b and A11a), performed in explicit dichloromethane solvent, have provided a detailed description of the mechanism by which the equivalence of signals takes place [106]. Simulations showed that the dynamics of the agostic interaction in solution strongly depends on the complex. Contingent upon the strength of the agostic interaction and the flexibility of the ligand, several events related with the occupation of the fourth coordination site by an agostic bond could happen: (i) the same C-H bond maintains the agostic interaction with the platinum atom for the entire simulation (A2b); (ii) an agostic interaction is present throughout the simulation, though the C-H bond changes by a rotation of alkyl groups (A11b); (iii) the agostic interaction moves on and off (A11a) [106]. (S12) [50]. Similar behavior has been observed for other pincer complexes. The naphthyl-based PCP-Pt(II) complex was found in THF as a 1:2 mixture of the counteranion (BF 4 − , C8) and solvent (S17) forms [53].
For the latter scenario, 14-electron T-shaped species involving phosphine ligands can be envisaged as feasible intermediates. The dissociative pathway fully prevails when overcrowded PR 3 ligands are employed, where R stands for o-methoxyphenyl 11 [110] and o-tolyl 12 [111]. The detailed mechanism (Scheme 11) is described as follows: (i) dissociative nitrogen decomplexation, (ii) isomerization between two nonequivalent exchanging sites, and (iii) nitrogen coordination recovering the initial chelating situation. The fluxional motion of dmphen is not affected by the counteranion, solvent, or the presence of weak nucleophiles deliberately added. In addition, for 11a and 12a both flipping of dmphen and phosphine rotation motions are synchronized, behaving as molecular gears. The calculated activation barriers are ca. 13.3 and 16.4 kcal mol −1 for the motions involving 11 and 12, respectively. Concerning 14-electron intermediates b and b', the assistance of methoxy groups in 11b,b' [110] as well as the presence of agostic interactions in 12b,b' [111] can be postulated ( Figure 16). Indeed, 12a easily undergoes a cyclometalation process, probably assisted by an agostic contact via intermediate 12b.

Three-coordinate Pt(II) species as reaction intermediates
Coordinatively unsaturated Pt(II) species are considered as intermediates in many organometallic processes. Early studies by Whitesides and co-workers dealing with the thermal decomposition of [Pt(R) 2 L 2 ] complexes demonstrated that the dissociation of a phosphine ligand is a preliminary requisite for the reaction to occur [112]. Since then, T-shaped three-coordinate 14-electron intermediates have been proposed in a large number of reaction mechanisms. A review concerning dissociative pathways in Pt(II) complexes was published in 1990 [113].
Due to both the low electron count and the presence of a vacant site in the coordination sphere, T-shaped Pt(II) species are suitable intermediates in ligand-exchange and bond-breaking processes of unreactive bonds, such as C-H bond activations. A selection of results from the past few years regarding these two issues are collected in this section to illustrate the growing importance of three-coordinate Pt(II) species as reaction intermediates.

Intramolecular C-H bond activation
Intramolecular C-H bond activation is a common reaction of unsaturated Pt(II) complexes. Cyclometalation processes, sometimes involving agostic situations [114], have been thoroughly reviewed recently [115].
According to Baratta and co-workers, the abstraction of Cl from 13a,b with Na[BAr F ] directly generates the cyclometalated products A5a,b and methane (Scheme 12) [39]. Unsaturated intermediates such as [Pt(Me)(PR 3 ) 2 ] + , probably stabilized by agostic contacts, might be considered.
Weller et al. reported some reactivity features of complex A7 [41]. Although it shows an agostic interaction, no cyclometalation via C-H bond activation is observed in CD 2 Cl 2 (40 °C, 7 days). However, the addition of THF rapidly traps the T-shaped complex forming the adduct 14 (Scheme 13). This species does undergo cyclometalation in the presence of trace Scheme 12: Halogen abstraction from 13a,b and subsequent cyclometalation to yield complexes A5a,b [39].
amounts of acids to yield 16 [116]. The unsaturated agostic species 15 has been proposed as a reaction intermediate.
Examples of cyclometalation processes preceded by ligand dissociation have been described by Romeo and co-workers [111,117,118].  [120] as experimentally reported in [118]. They demonstrated that the reaction involves coordinatively unsatu- After methane releasing, the cyclometalated T-shaped structure 27 is obtained, once again stabilized by an agostic interaction. The corresponding transition states for OA and RE processes are isoenergetic with respect to 25. Finally, the agostic coordination mode in 27 is displaced by a DMSO molecule forming the cyclometalated product 28. Moreover, they showed that the reaction mechanism starting from the four-coordinate 16-electron complex 23 does not provide low-energy paths for OA, demonstrating the kinetic inertness of these types of compounds.
Three-coordinate species have been invoked by Nolan and co-workers to account for the reaction of cis-[Pt(NO 3 ) 2 L 2 ] 29 with tetramethylthiourea (tmtu) (Scheme 16) [121]. The sulfur atom of tmtu displaces one nitrate ligand of 29 forming 30, which eventually evolves to the cyclometalated product 33 via C-H bond activation. The increase of nitrate concentration as well as the addition of coordinating agents, such as triethylamine and triphenylphosphine, clearly retard the process. Therefore, a dissociative mechanism has been proposed to T-shaped Pt(II) complexes bearing NHC ligands can be prepared starting from the pertinent iodo-precursors by halide removal (Scheme 17) [44]. For IPr [44], IMes* and IMes Scheme 17: Cyclometalation process of NHC-based Pt(II) complexes [28,44].
ligands [28], the corresponding methyl complexes A10a and T4a,b can be isolated. Interestingly, cyclometalation involving a methyl group of the carbene arm was observed upon heating these methyl derivatives. It is noteworthy that the process barely depends on the nature of the fourth coordination site, either agostic A10a or pure empty site T4a,b. In the case of It-Bu ligand [44], a similar cyclometalation reaction is observed even at low temperatures (−70 °C), but the putative methyl intermediate could not be detected.
The agostic complex A9 can exchange the site of cyclometalation by C(sp 3 )-H bond activation (Scheme 18) [43]. The addition of one equivalent of L (L = DMSO, PPh 3 or pyridine) to A9 in chloroform or acetone at room temperature yields the products 37, in which one methyl group (previously agostic) has been cyclometalated, and L has entered into the coordination sphere of the platinum atom. The authors reasonably propose a σ-bond metathesis (σ-CAM) mechanism via intermediate 36 invoking the well-known capacity of agostic interactions to facilitate the C-H bond cleavage.
The "rollover" process is a class of cyclometalation, in which a heteroaryl ligand undergoes decoordination and bond-rotation processes prior to C-H bond activation. A recent review on this topic collects the most important features of this reaction [122]. Early work by Young and co-workers [123] proposes the formation of coordinately unsaturated species 39 and 39' as intermediates in the "rollover" reactions of 38, leading to polymeric species (Scheme 19).
Due to the steric congestion between the R group of the bipyridine and the methyl ligand 42 becomes unstable promoting the decoordination of the nitrogen atom. Subsequent C-H bond activation is proposed to take place through an agostic intermediate 43 generated by rotation around the 2,2'-C-C bond of the bipyridine ligand. After the release of methane, the vacant site is easily occupied by one DMSO ligand yielding 44. Interestingly, from 44 (R = t-Bu, Ph) the corresponding hydride compounds can be prepared [125]. It is worth pointing out that, depending on the ligand present in solution, both 16-electron and 14-electron species are obtained, though the latter are not stable, and only oligomers with bridging hydrides can be detected [125].
Wang and co-workers disclosed the spontaneous self-assembling of [Pt(Me) 2 (NPA)] 45 by a "rollover" cyclometalation process (Scheme 21) [126]. The suggested mechanism begins with a C-N bond rotation by chelating ligand dissociation forming species 46, which is stabilized by solvent coordination. Eventually, an agostic interaction in 47 prior to the rate-determining C-H bond cleavage should displace the solvent molecule. An oxidative addition and reductive elimination (OA/RE) scenario via hydride Pt(IV) 48 and subsequent methane release yield the corresponding solvent adduct 49, from which selfassociation generates the cyclic tetramer 50. Indeed, when good coordinating agents such as acetonitrile are added, the reaction slows down.
On the other hand, by switching the DMSO solvent to the less polar chloroform the expected cyclometalated product 37a (Scheme 18) is obtained. DFT calculations correctly explain the relative stabilities Scheme 23: Proposed mechanisms for the "rollover" cyclometalation of 52 in gas-phase ion-molecule reactions [128].
of 51 with respect to 37a depending on the polarity of the solvent [127].
This kind of reactivity has also attracted attention in gas-phase conditions [122]. The gas-phase behavior of cationic species 52 has been analyzed by means of combined experimental and computational studies (Scheme 23) [128]. DFT calculations discourage the initial loss of dimethylsulfide. Instead, decomplexation and C-C bond rotation processes starting from the four-coordinate complex 52 are favored. The resulting isomeric compounds exhibit an empty coordination site that is filled by an agostic interaction prior to the C-H bond activation. Intermediate 53 evolves through a σ-CAM process, whereas intermediate 54 undergoes oxidative addition and reductive elimination processes. The release of methane and dimethylsulfide yields 55. Further studies including labeling experiments support the reversibility of these "rollover" reactions. The highly unsaturated species 55 is still reactive and can coordinate and decompose XMe 2 molecules (X = S [128] and O [129]) and dehydrogenate alkanes [130]. Finally, other cyclometalation processes including "rollover" reactions have also been observed for the complex [Pt(Me)L(SMe 2 )] bearing a diimine ligand instead of the ubiquitous bipyridyl backbone [131].

β-H elimination
As previously noted, the agostic complexes A1-4 ( Figure 7) are in equilibrium with the hydrido-alkene isomers [34][35][36][37][38]. Experimental evidence points out that substituted alkyls and large chelate ring-size diphosphine ligands favor the β-agostic isomer. As a representative example, Scheme 24 shows the β-H elimination for A1d together with the reverse 1,2-insertion for 56. Interestingly, upon crystallization in dichloromethane solvent, A1d eliminates norbornene from 56 and generates the dinuclear complex 57 [35]. The presence of chloride as a bridging ligand suggests that solvent molecules are involved in the reaction. Therefore, the participation of T-shaped intermediates, probably stabilized as solvento adducts, might be relevant in the overall process. Scheme 27: Intermolecular C-H bond activation from 62 [95].
to give the complex 58. The subsequent β-H elimination process shall provide the expected hydride complex 59. However, solvent molecules come into play, so that the unsaturated intermediate 58 activates the C-D bond of benzene-d 6 forming 60.
Cyclometalation and subsequent β-H elimination processes generate D-59, which has fully incorporated the corresponding deuterium atoms. This latter evidence suggests that the intermolecular C-D bond activation of benzene-d 6 is indeed faster than the β-elimination. On the other hand, when the reaction is conducted in a cyclohexane-d 12 solution, D-59 is hardly obtained and 59 prevails. It means that, in sharp contrast to the arene solvent, the intermolecular C-D bond activation of alkanes becomes slower than the β-elimination.
This type of reaction has also been observed for the agostic complex A11a [28], although the CH group of the agostic contact is not involved. Upon heating or under UV irradiation (Scheme 26), one hydrogen of the cyclometalated isopropyl group in A11a undergoes a β-H elimination process yielding 61, in which the alkene and the hydride ligands are located mutually trans.

Intermolecular C-H bond activation
Three-coordinate T-shaped Pt(II) complexes have been postulated in hydrocarbon C-H bond activations, particularly in the Scheme 26: β-H elimination process of A11a [28].
Labile ligands in masked T-shaped compounds allow further reactivity. Although some solvent complexes exhibit associative pathways eluding 14-electron species [73,138], the participation of such coordinatively unsaturated intermediates should be taken into account for other systems [135,136]. A good example is the investigation reported by Wick and Goldberg (Scheme 27) [95]. From the anionic species [Pt(Me) 2 Tp'] − 62, they attempted to generate the unsaturated species 63 through the abstraction of one methyl ligand. Indeed, the treatment of 62 with B(C 6 F 5 ) 3 in benzene, cyclohexane and n-pentane provides Scheme 28: Reductive elimination of methane from 65 followed by CD 3 CN coordination or C-D bond-activation processes [105]. the products 64 arising from the C-H bond activation of solvent molecules. Interestingly, the activation of pentane molecules only occurs at the primary carbon atom.
A similar situation has been observed for the reductive elimination of methane in the complex 65 (Scheme 28) [105]. Experimental evidence is consistent with a dissociative methane loss from 66 as the rate-determining step. Therefore, as shown in Scheme   Recently, arene activation has been reported regarding true T-shaped species with NHC ligands [28]. The cyclometalated complexes A11a, A11b, T5a and T5b were tested toward C-H bond-activation processes by using benzene as a solvent (Scheme 29). No reaction was observed for A11a and A11b even under drastic conditions (high temperatures and long reaction times). On the other hand, T5a yields the phenyl product T6a, whereas T5b barely reacts.
DFT calculations suggest an oxidative addition and reductive elimination scenario via Pt(IV) hydride intermediates 70 ( Figure 19). The steric environment of the agostic complexes A11a and A11b complicates the reaction, which is reflected in the high energy barriers (more than 40 kcal mol −1 , red line). In sharp contrast, the true T-shaped species T5a and T5b, with no agostic bonds, show lower energy profiles (ca. 30 kcal mol −1 , blue line) and the reaction thermodynamics (ΔE r values, Figure 19) accounts for the observation of T6a and the poor detection of T7.
Other arenes can also be activated by the use of T5a to afford T6 ( Figure 6). In toluene solvent, only the two products m-T6b and p-T6b corresponding to metaand para-site activations are observed in a 5:1 molar ratio. Neither orthonor benzylic C-H bond activations are detected. Once again, DFT calculations provide reasonable ΔE ‡ for both meta and para-routes (ca. 30 kcal mol −1 ), and the thermodynamic effects, i.e., that only m-T6b and p-T6b are slightly more stable than T5a, explain the experimental evidence.
Some pincer complexes can indeed activate the C-H bonds of benzene, though the mechanism is not fully understood [50,140]. In this line, Ozerov and co-workers have attempted to access unsaturated species by means of abstraction of triflate ligand from 71 (Scheme 30) [12].  [12].
In some cases, C-X bond-activation processes involving PhBr and CH 2 Cl 2 molecules have also been registered.
Complexes of the type [Pt(Me)(NN)] + 74 (Scheme 31) have shown reactivity toward the C-H bond activation of benzene in gas-phase ion-molecule reactions [7]. Suitable labeling experiments disclose the reversibility of the C-H bond activation. Interestingly, the relative rate constants reveal that 74d reacts slower than 74a-c. In the absence of solvent, both reactants 74a-c and phenyl products 76a-c are believed to exhibit true T-shaped structures. On the other hand, the lower reactivity of 74d has been attributed to an interaction of the empty site of the platinum with one ortho-chlorine atom of the (o,o'-Cl 2 C 6 H 3 )N=C(CH 3 )-C(CH 3 )=N(o,o'-Cl 2 C 6 H 3 ) ( Me DAB DCP ) ligand [141], which blocks the approach of a benzene molecule. Oxidative addition and reductive elimination (OA/RE) as well as σ-CAM scenarios have been evaluated for 74b, but unfortunately the results are functional-dependent. B3LYP cannot Scheme 32: Dihydrogen activation through complexes A5a, A5b [39], A5c [40] and S1a [54].

H 2 activation
Some agostic complexes previously reported can activate small molecules such as H 2 (Scheme 32). For instance, the complexes A5a and A5b react with dihydrogen (1 atm H 2 , 20 °C) in dichloromethane solution to yield the corresponding agostic hydride products A6a and A6b [39]. Both species were found in equilibrium, exhibiting A5/A6 ratios of 4:1 and 1:2 for species a and b, respectively. A similar behavior is observed for the Scheme 35: Detection and isolation of the Pt(III) complex 81a [143]. analogous reaction of A5c (1 bar H 2 , −30 °C to room temperature), in which A6c is obtained in a 1:3 ratio [40]. These results suggest that hydride species seem to be favored according to the basicity of the phosphine ligand. In previous works, the dichloromethane molecule in the solvento complex S1a was labile enough to allow H 2 activation but, unlike A6, the corresponding dihydrogen adduct 77 was observed (Scheme 32) [54]. The latter undergoes a fluxional process which can be slowed down by cooling at −60 °C.
Weller and co-workers reported an alternative procedure to obtain 77 from A7 by addition of dihydrogen in CD 2 Cl 2 solution (Scheme 33) [41]. The agostic interaction in A7 is supposed to be displaced by one dihydrogen molecule forming the putative intermediate 78.
The concomitant release of methane may be the driving force of the reaction, thus the entropic factors together with the nature of the phosphine ligands should be taken into account. Additionally, the related compound 16 (Scheme 13) can also activate dihydrogen mole-cules; the cyclometalated ring is opened and the corresponding hydride complex S1c is obtained [54].

X 2 activation
The cyclometalated compounds A11a and T5a can activate X 2 molecules (X = Br, I) in dichloromethane solution affording 79 and 80 in which one C(sp 3 )-X bond is constructed (Scheme 34) [143]. It is noteworthy that the reaction occurs in the presence (A11a) or the absence (T5a) of agostic interactions in the starting materials.
Interestingly, during the reaction of A11a with Br 2 at low temperature (−78 °C) a paramagnetic, see-saw Pt(III) alkyl intermediate 81a could be isolated and characterized (Scheme 35), although the iodo-analogue 81b could not. DFT calculations correctly predicted the feasibility of 81a (−9.5 kcal mol −1 below reactants) and explained the nondetection of 81b (+4.5 kcal mol −1 above reactants) in terms of higher relative energies.
A related Cl 2 activation process involving unsaturated species has been recently reported by Rourke and co-workers (Scheme 36) [144]. Under certain conditions, the reaction of 82 with PhI·Cl 2 yields 83. Although suitable crystals of 83 could not be obtained, the structure was elucidated by NMR studies showing the formation of one C(sp 3 )-Cl bond. Similarly to the Scheme 36: Cl 2 bond activation through complexes 82 and 83 [144].

cis-trans Isomerization
The cis-trans isomerization of several solvento species S5 has been studied and the formation of low electron-count Pt(II) complexes 85 through dissociative pathways accounts for the experimental observations [2,61]. The general mechanism is depicted as follows: (i) rate-determining dissociation step of a solvent molecule, (ii) isomerization process, and (iii) fast association of a solvent molecule (Scheme 37).
As shown in Scheme 37, the 14-electron structures 85 are involved in the process. Due to their high tendency to fulfill the empty coordination site, these species strongly favor intramolecular contacts. Indeed, it has been experimentally observed that the complexes S5 bearing R = Et, n-Pr and n-Bu groups react much faster than other derivatives with R groups without β-hydrogen atoms [61]. This phenomenon has been called the β-hydrogen kinetic effect [2]. It is defined by an incipient β-agostic interaction that can stabilize transient T-shaped intermediates and transition states, and therefore, an increase of the reaction rate would be expected. Further computational studies were performed on complexes [Pt(R)(PMe 3 ) 2 (solv)] + 85 describing the isomerization process in different solvents [2]. The β-hydrogen kinetic effect can be detected during the first steps of the isomerization energy profiles ( Figure 20) for both the methyl (85a, red line) and the ethyl (85b, blue line) complexes in acetonitrile. Note that the agostic contact in cis-85b decreases the overall energy requirement with respect to cis-85a, but at the same time, it increases the relative energy barrier to reach the Y-shaped transition state TS85, i.e., 4.3 kcal mol −1 for the ethyl (TS85b) and 2.3 kcal mol −1 for the methyl (TS85a) derivatives. By removal of the agostic interaction in cis-85b (through a rotation of the C-C bond of the ethyl ligand), this energy stabilization was estimated to be 5.9 kcal mol −1 .
Other solvents such as methanol, dimethylsulfoxide and benzene were also considered. The donor ability strongly affects the dissociation step, although it has a little impact on the strength of the agostic interaction.
In the presence of benzyl ligands, a similar cis-trans isomerization via the unsaturated species [Pt(R)(PEt 3 ) 2 ] + (Scheme 37) has been claimed [62]. DFT calculations on the model derivative [Pt(Bn)(PMe 3 ) 2 ] + predict a Pt···η 2 -benzyl coordination mode in 86 involving the ipso-carbon of the benzyl ligand ( Figure 21). This stabilization accounts for 7.6 kcal mol −1 , i.e., 1.7 kcal mol −1 stronger than the above-mentioned agostic contact for the same family of compounds (5.9 kcal mol −1 , [2]). The interaction is favored by electron-donating groups on the phenyl ring and, as a consequence, the reaction rate is enhanced. (ii) subsequent ligand addition (89), and (iii) displacement of the sulfur ligand (90) [117,119]. The electron-rich metal, the Pt-S bond weakening due to the trans-influence of the R groups, and the stabilization of T-shaped intermediates favor this dissociative pathway. Interestingly, for the complex cis-[Pt(Ph) 2 (CO)(SMe 2 )], in which one thioether ligand has been replaced by CO, the operating mechanism becomes associative [145]. Steric and β-hydrogen kinetic effects [2] have been invoked to explain the fast reaction of [Pt(Hbph) 2 (DMSO) 2 ] (Hbph = η 1 -biphenyl monoanion) compared to species containing other R groups [117]. Puddephatt and co-workers have demonstrated that, depending on the nature of the R group, dissociative mechanisms can also facilitate the substitution of dimethylsulfide by phthalazine in the dinuclear species 91 (Scheme 39) [146]. When alkyl derivatives are employed, kinetic evidence mainly points to an associative mechanism via 92, although a minor part of the reaction has been related to a dissociative pathway. The inclusion of phenyl and p-tolyl ligands in 91 completely changes the mechanism. A first-order reaction is observed and large negative entropies of activation are collected. The proposed mechanism involves the dissociation of dimethylsulfide generating the puta-tive unsaturated intermediate 93. This species is supposed to be stabilized by a donor-aceptor Pt-Pt bond, i.e., the electron-rich four-coordinate platinum atom interacts with the electron-deficient three-coordinate counterpart. Consequently, the electron density of the former platinum atom decreases, thereby favoring the subsequent attack of the N-N ligand to yield 94.

Conclusion
There are only a very few authenticated three-coordinate Pt(II) complexes. Yet, T-shaped intermediates play an important role in a number of organometallic transformations. Given the electronic and coordinative unsaturated nature of such compounds, they can only be isolated when the vacant coordination site is blocked to avoid intra-and intermolecular interactions. Steric effects are important in preventing the coordination of a fourth external ligand. The use of bulky species, which hamper the entry of ligands into the platinum coordination sphere, is a good strategy toward this goal. The presence of shorter or less flexible alkyl chains in bulky ligands could avoid agostic interactions. Electronic effects, as well as the involvement of strong electron-donating ligands, also have a stabilizing impact.
Overall, the number of true T-shaped complexes that are wellcharacterized is still very low. Nevertheless, a much larger number of Pt(II) complexes can be described as operationally three-coordinate in a kinetic sense. This happens when the fourth position in a square-planar complex is occupied by a very weak ligand, which can be easily displaced. We have named these compounds masked T-shaped complexes. An intramolecular agostic interaction, a weakly coordinated solvent molecule or a counteranion can play such a role. Whether or not the interaction of the platinum atom with the fourth ligand is very weak, the three-coordinate complex should be very close in energy to the square-planar ground state. Consequently, in practice the masked T-shaped complexes can be devised as a resting state of the three-coordinate species.
The accessibility of three-coordinate complexes makes them suitable intermediates in Pt(II) chemistry. In spite of the difficulty in detecting these intermediates, this review gathers a number of recent experimental and theoretical reports, which claim the involvement of T-shaped Pt(II) intermediates in reaction mechanisms. In the near future, the acquired knowledge on the synthetic routes and structural features, as well as the advances in the detection techniques and computational methods, should lead to an improved design of such compounds and to a wider recognition of the role they play in chemical transformations.