POP-type ligands: Variable coordination and hemilabile behaviour

Hemilabile ligands – ligands containing two or more potential donors to a metal centre, of which one or more can dissociate – have the ability to provide a transition metal complex with open coordination sites at which reactivity can occur, or stabilise low coordinate intermediates along reaction pathways. POP- type ligands and in particular POP, Xantphos, DBFphos and DPEphos-based ligands contain three possible binding sites: two phosphines and an ether linker, thus have the potential to show j 1 -, j 2 - or j 3 -binding modes. This review summarises the examples where POP-type ligands display hemilabile, or closely related variable coordination, characteristics in either synthesis or catalysis.


Introduction
The importance of transition-metal complexes in homogeneous catalysis cannot be over-stated. The ability to precisely control at the atomic level the structure of a catalyst through manipulation of both the metal and -perhaps more importantly -the supporting ligands is central to the development and utilisation of such catalysts in a myriad of applications for the synthesis of fineor commodity-chemicals or new materials. [1]. Bidentate diphosphine-based ligands have played a particularly important role in processes catalysed by later transition metals, such as alkene hydrogenation, hydroformylation, cross-coupling and carbonylations [2]. Fine control of the metal-ligand environment through bite angle effects or the steric profile of phosphine substituents provides the ability to control turnover-and productdetermining steps [3,4]. Tridentate pincer-type ligands have also played an increasingly dominant role in catalysis [5,6], owing to their stability, ability to define a particular coordination geometry in which the pincer ligand occupies three meridional sites around a metal centre, and in some cases ability to take part in metal-ligand cooperative reactivity [7]. Although, again, there are many examples of the use of pincer-ligand complexes, those based around ''PCP", ''PNP" or ''PONOP" (Fig. 1) have found particular applications in alkane dehydrogenation when partnered with iridium, or in the activation of strong bonds and subsequent catalytic transformations more generally [8][9][10]. These ligands occupy such a ''privileged" volume of chemical space due, in large part, to the robust coordination framework that the PNP or PCP ligand motif provides, although the role that coordination flexibility has in certain pincerligand classes is also being recognised as being potentially important [11].
The coordination chemistry of tridentate ligands based upon diphosphine ligand frameworks with an additional ether linkage, the so-called ''POP" ligands (Fig. 1), has been known for some decades. Their use in catalysis has been significant, especially ligand frameworks based around Xantphos and DPEphos -originally developed as wide bite angle bidentate ligands for rhodiumcatalysed hydroformylation [12]. Although principally bidentate in coordination mode, their ability to coordinate the central ether gives ligand geometries that are related to classical tridentate PCP or PNP pincer ligands. If coordination of any of the three ligating atoms was flexible, this offers the ability of the ligand set to adapt to the specific requirements of the metal (i.e. variable oxidation states and resulting geometric constraints), including the binding/release of co-ligands, substrates or products during synthesis, or more importantly during a catalytic cycle. When such flexibility also involves the reversible decoordination of a donorgroup, this leads to hemilabile behaviour, in which a metal centre is rendered operationally unsaturated by the temporary decoordination of a ligand, but one that still remains associated with the coordination sphere of the metal. This flexibility, potentially, leads to a number of coordination modes as outlined in Fig. 2. The term hemilabile was introduced by Jeffrey and Rauchfuss in 1979 to describe coordination modes of ortho-substituted diphenylanisole ligands [13]. Since then many examples of hemilabile ligand behaviour have been reviewed, amongst others by Lindner in 1991 [14], Mirkin in 1999 [15], Braunstein in 2001 [16], Bassetti in 2006 [17], Mirkin in 2008 (so-called weak link approach to supramolecular complexes) [18] and Hor in 2007 [19]. Interestingly, the potential for hemilability in POP, Xantphos or DPEphos-type ligands was not generally commented upon, but in the last decade this variability in ligand binding has become increasingly recognised. In this contribution we survey the coordination chemistry of such ligands and outline cases where changes in coordination geometry lead to changes in ligand denticity that are directly related to the potential for these ligands to behave in a hemilabile manner. Although when rigorously defined hemilabile refers to the ability to reversibly decoordinate a donor group [16], we also survey examples where decoordination may also be irreversible, but focus in these cases on situations where the POP ligand changes its binding mode to accommodate the reaction course. We also briefly comment upon the coordination modes of PSP-type ligands (Fig. 1), however examples of transition metal complexes using this ligand type are much rarer than their POP-type analogues.
Coordination of the ether linker of the POP-R ligand to a transition metal centre was also noted by Gusev [23]. Replacement of the phenyl substituents on the phosphines with alkyl groups led to interesting substituent effects, as demonstrated by the reaction of [RuCl 2 (p-cymene)] 2 with POP-t Bu versus reaction with POP-i Pr (Scheme 2). The complex derived from POP-t Bu formed a monomeric, formally 16-electron species, mer-Ru(Cl) 2 (j 3 -P,O,P-POP-t Bu) (4), which showed approximately octahedral geometry around the Ru centre, stabilised by mer-j 3 -P,O,P coordination of POP-t Bu, with the sixth coordination site occupied by a CAHÁ Á ÁRu agostic bond from one of the t Bu groups of the ligand (RuAO = 2.123(2) Å, PARuAP = 161.1(1)°). In comparison, the equivalent reaction with the less sterically bulky POP-i Pr ligand yielded a bimetallic complex, fac-[Rh 2 (m-Cl) 3 (j 3 -P,O,P-POP-i Pr) 2 ] [Cl] (5), which was crystallographically characterised as its [PF 6 ] À salt, and showed fac-j 3 -P,O,P binding of the POP-i Pr ligand (RuAO = 2.136(6) Å, PARuAP = 101.8(1)°). Whilst both complexes allowed tridentate donation by the ligands, evidenced by the short RuAO bonds in each case, the wider PARuAP bite angle characteristic of mer-j 3 -P,O,P coordination of POP-t Bu in complex 4 was favoured over fac-j 3 -P,O,P, presumably due to the increased steric bulk of t Bu compared to i Pr. The authors postulated that the monomeric j 3 -P,O,P-POP-t Bu complex 4 was formed by dissociation of an initially formed (but not observed) bimetallic POP-t Bu analogue of complex 5, fac-[Rh 2 (m-Cl) 3 (j 3 -P,O,P-POP-t Bu) 2 ][Cl]. The dissociation of the dinuclear species via chloride association to two mer-Ru (Cl) 2 (j 3 -P,O,P-POP-t Bu) monomers was driven by the dimer's inherent instability, caused by the large t Bu substituents. Further reactivity studies of both complexes 4 and 5 with H 2 and N 2 gave rise to octahedral mer-j 3 -P,O,P-coordinated complexes.
Duatti and co-workers have noted the existence of fac-mer isomerisation in group-7 nitrido complexes M(N)(Cl) 2 (j 3 -P,O,P-POP-Ph) complexes (M = Tc, Re), Scheme 3 [24,25]. The kinetically favoured fac-isomer (6) was synthesised at relatively low temperatures, however at elevated temperatures in acetonitrile solvent, the mer-isomer (7) was irreversibly formed. This was confirmed by NMR spectroscopy, and supported by computational studies of the Tc complexes in which mer-Tc(N)(Cl) 2 (j 3 -P,O,P-POP-Ph) was thermodynamically favoured over the fac-isomer by 5.9 kcal mol À1 . Whilst neither of the fac-isomers of Tc or Re could be characterised by X-ray diffraction studies, hence the presence of a MAO bond in a fac-j 3 -P,O,P conformation could not be confirmed, the crystal structures of both mer-Tc(N)(Cl) 2 (j 3 -P,O,P-POP-Ph) [24] and mer-Re(N)(Cl) 2 (j 3 -P,O,P-POP-Ph) [25] were obtained. The PAMAP bond angles of the Tc and Re complexes were 152.1(1)°and 152.7(1)°respectively, with the phosphine ligands trans to one another. However, the MAO bond distances were considerably longer than expected: 2.500(4) Å for the Tc complex and 2.505(9) Å for Re. This was proposed to be due to the strong trans-labilising effect exerted by the nitrido group, resulting in, an at best weak, MAO interaction. The fac-mer isomerisation process was proposed to involve dissociation of the ether linker and rearrangement of the resulting square pyramidal structure. Further reactivity of both isomers showed that the facisomer can undergo a ligand substitution reaction with S-methyl 2-methyldithiocarbazate and Na [BF 4 ] to form complex 8 (Scheme 3) in refluxing CH 2 Cl 2 /ethanol, whereas mer-M(N) (Cl) 2 (j 3 -P,O,P-POP-Ph), 7, was found to be unreactive under the same conditions. This difference in reactivity was suggested to be due to a very low lying LUMO, coupled with steric constraints, for the mer-isomer.
van Leeuwen et al. utilised the POP-Ph ligand (and derivatives thereof) complexed with rhodium in hydroformylation catalysis [26]. Interestingly, whilst the activity of these complexes were comparable to those of the well-studied Xantphos-Ph ligand (though their selectivity for linear versus branched aldehyde products was moderate in comparison), very little isomerisation ( 0.8%) of the starting 1-alkene was observed when using POP-Ph instead of Xantphos-Ph. The authors speculated that this occurred because the POP-Ph ligand could act in a hemilabile manner [13], and utilise its ability to bind strongly in both bidentate and tridentate fashions in order to occupy any vacant sites formed during the catalytic cycle. The binding of the ether linker of POP-Ph was suggested to be capable of suppressing any beta-hydrideelimination events which would form unwanted internal alkenes (Scheme 4).
The study of potential for hemilability in POP-Ph was investigated by Weller and co-workers during their studies of the hydroacylation reaction, which involves the formal addition of a CAH bond across an alkene or alkyne [27]. Previous work had demonstrated that a hemilabile DPEphos ligand (see Schemes 28 and 29) enabled the stabilisation of important intermediates during catalysis, as well as suppressing unwanted reductive decarbonylation and subsequent deactivation of the catalyst [28,29]. Although it was shown that initial oxidative addition of the aldehyde occurred to give a POP-Ph-containing complex mer-[Rh (COCH 2 CH 2 SMe)(H)(j 3 -P,O,P-POP-Ph)][BAr F 4 ] (Ar F = 3,5-C 6 H 3 (CF 3 ) 2 ) (9), this complex showed no onward reaction with the alkene, methyl acrylate, at 10 mol% catalyst loading (Scheme 5). Interest-ingly, the Xantphos-Ph analogue also showed the same reactivity profile. This was postulated to be due to the unavailability of a vacant site on the metal due to the strong binding of the ether linker of POP-Ph in complex 9 (RhAO = 2.302 (1) (10), was not achieved. This behaviour of POP-Ph contrasted van Leeuwen's Rh-system used for closely related hydroformylation, which did turnover in catalysis (Scheme 4) [26]. Weller's system thus appears to exhibit stronger binding of the POP-Ph ligand, and in particular of the binding of the ether linker through mer-j 3 -P,O,P-coordination, compared to the hemilabile fac-j 3 -P,O,P-POP-Ph ligand proposed in van Leeuwen's work. The difference in reactivity between fac and merisomers had been presented previously by Duatti [25], and also has relevance for catalysis which involved Xantphos-Ph and DPEphos ligands as described in Schemes 17 and 30 respectively [30,31].

Xantphos-type ligands with aryl substituents
The Xantphos-Ph ligand (4,5-bis(diphenylphosphino)-9,9-dime thylxanthene) was contemporaneously reported in 1995 by the groups of Haenel [32] and van Leeuwen [12]. Other ligands based on the xanthene backbone, also containing diarylphosphine groups and an ether linker, namely Sixantphos (4,6-bis(diphenylpho sphino)-10,10-dimethylphenoxasilin) and Thixantphos (2,8-dime thyl-4,6-bis(diphenylphosphino)phenoxathiin), have also been detailed [12]. These ligand types were originally designed to provide bidentate binding to rhodium-based catalysts with varying bite angles for use in hydroformylation catalysis [12]. Whilst in this initial study no hemilabile character of the ligands was reported, nor any sign of ether binding to the metal centre observed, it was shown that increasing the bite angle from Sixantphos (108.7°) to Xantphos (111.7°) gave improved selectivity for linear versus branched aldehyde products -showing possible advantages for their use in catalysis. The potential hemilabile, or at least variable, coordination mode of Xantphos-Ph was largely overlooked until recently. In 2011, Haynes and co-workers presented a survey of crystallographically characterised Xantphos-Ph complexes, and plotted the values of MAO distance (Å) versus PAMAP bite angle (°) (Fig. 3), which elegantly showed, graphically, the wide range of bite angles and coordination modes available to the Xantphos-Ph ligand [33].
An early example of Xantphos-Ph acting as a j 3 -P,O,P pincerlike ligand came from Sandee et al., who immobilised a rhodium complex containing the Xantphos-based ligand, Siloxantphos   (N-(3-trimethoxysilane-n-propyl)-4,5-bis(diphenylphosphanyl)phenoxazine), onto a silicate matrix via sol-gel synthesis (Scheme 6), for use as a heterogeneous catalyst in hydroformylation [34]. Due to the problems associated with characterisation of this supported catalyst, a model complex containing an unsupported Xantphos-Ph ligand was also exposed to the solgel process, but its relative mobility allowed for NMR characterisation in the sol-gel as well as independent single crystal X-ray analysis. During the sol-gel process, the cis-Rh(acac)(CO)(j 2 -P,P-Xantphos-Ph) (Siloxantphos analogue = 11) precursor complex was transformed to the cationic species, mer-[Rh(CO)(j 3 -P,O,P-Xantphos-Ph)] + (Siloxantphos analogue = 12), the structure of which was confirmed by X-ray crystallography (RhAO = 2.126 (3) Å, PARhAP = 164.42(4)°). On further reaction with CO/H 2 , a j 2 -P,P-ligated structure, cis-Rh(CO) 2 (H)(j 2 -P,P-Xantphos-Ph) (Siloxantphos analogue = 13), was formed, which is a catalyst precursor in hydroformylation [12]. The authors suggested that the j 3 -P,O,P-chelating nature of Xantphos-Ph, and by analogy, Siloxantphos, stabilised the complex during the sol-gel process (Scheme 6), with the hemilability of the ether oxygen allowing the catalytic site to function in the silicate matrix. No comment was made about the change in overall charge between 12 and 13 but it is likely that in such a sol-gel matrix deprotonation of a, rather acidic, bound dihydrogen ligand is possible [35].
The potential for variable interaction between the Xantphos-Ph ether group and a transition metal was also postulated to be important in allyl-palladium complexes reported by van Haaren et al. that showed fluxional behaviour [36]. A p-r rearrangement was proposed to be responsible for exchange of the endo and exo forms of cis-Pd(g 3 -1,1-(CH 3 ) 2 C 3 H 3 )(j 2 -P,P-Xantphos-Ph) (14), which could not be frozen out at low temperatures. Two intermediates for the rearrangement were proposed, a T-shaped intermediate and a Y-shaped intermediate which contained a stabilising PdAO interaction (Scheme 7).
Hartwig and co-workers noted that the Xantphos-Ph CMe 2 groups showed as a single resonance in the 1 H NMR spectrum of cis-Pd(Cl)(g 3 -allyl)(j 2 -P,P-Xantphos-Ph), and proposed that one explanation for the interconversion of the methyl groups was through reversible dissociation of one of the phosphine arms of Xantphos-Ph, which presumably led to transient j 1 -P-coordination of the ligand and subsequent rearrangement of the 3-coordinate intermediate [37]. j 1 -P-bound Xantphos-Ph ligands have also been proposed from DFT calculations of various catalytic transformations (vide infra). An even more unusual binding mode of Xantphos-Ph was proposed by Momeni et al. [38], where j 1 -O-binding through the Xantphos-Ph oxygen was postulated to form the 5-coordinate complex Sn(Me) 2 (Cl) 2 (j 1 -O-Xantphos-Ph) in CDCl 3 solution. The 1 H NMR spectrum of the complex indicated penta-coordinate geometry of ligands around the tin centre, whilst the 31 P NMR spectrum showed a single resonance very close in chemical shift to that of free Xantphos-Ph; which led the authors to suggest that the ligands are unlikely to be bound through either of the phosphine moieties.
The Weller group reported upon iridium complexes of Xantphos-Ph, and found that different coordination modes of the ligand could be obtained [39]. When cis-[Ir(COD)(j 2 -P,P-Xantphos-Ph)][BAr   [41]. The solution room temperature 31 P NMR spectra for each complex showed a broad singlet, which resolved into two resonances on cooling to À40°C, indicating a rapid equilibrium between the cis and transisomers, with the trans-isomer favoured at low temperature. Recrystallisation, however, favoured trans-Pd(p-C 6 H 4 (CN))(Br)(j 2 -P,P-L) (e.g., 23, Scheme 10), and this isomer could be crystallographically characterised for each ligand. Single crystal X-ray diffraction data for trans-Pd(p-C 6 H 4 (CN))(Br)(j 2 -P,P-Xantphos-Ph) (23) showed trans-coordination of the ligand to the metal centre, with a large PAPdAP bite angle of 150.35(3)°. Interestingly a weak, at best, PdÁ Á ÁO interaction, 2.698(2) Å, was present, as evidenced by a close to square planar geometry of the complex. The authors suggested that this extra PdÁ Á ÁO interaction between the metal and the Xantphos-Ph ligand may be a factor in stabilising the trans-isomer over the cis. For the complexes Pd(CH 3 )(Cl)(j 2 -P,P-Xantphos-Ph) (24) and Pd(CH 3 )(Br)(j 2 -P,P-Xantphos-Ph) (25), cis and transisomers were also found to be fluxional at room temperature, but static at À60°C, and despite being neutral complexes, showed high conductivity values. The authors postulated that this observation may be due to an ionic intermediate, such as mer-[Pd(CH 3 )(j 3 -P, O,P-Xantphos-Ph)][X] (X = Cl (26), X = Br (27)), that forms during the cis-trans isomerisation pathway, which occurs through coordination of the Xantphos-Ph oxygen to palladium, with chloride or bromide dissociation (Scheme 10). It was also noted that the rate of interconversion between the isomers was increased in polar solvents, which would allow further stabilisation of the proposed ionic intermediate which contains a j 3 -P,O,P-Xantphos-Ph ligand.
Jamison and co-workers also observed an isomerisation process in their study of air-stable nickel complexes which contained the Xantphos-Ph ligand [43]. The solid-state structure of trans-Ni(o-C 6 H 4 (CH 3 ))(Cl)(Xantphos-Ph) (31, Scheme 12) showed a structure that was described as a distorted square pyramidal geometry in the solid-state, but had a rather long apical NiAO distance (2.5408(6) Å), and a large PANiAP bite angle (156.01(1)°), which suggested trans orientation of the phosphine groups. In solution, however, the authors observed another isomer by NMR spectroscopy, which they speculated to be the analogous complex which lacked oxygen coordination of the Xantphos-Ph ligand at nickel, shown in Scheme 12, hence suggesting hemilability. It is interesting to note that interconversion of simple rotamers of the o-tolyl group in which the methyl was syn or anti (as suggested crystallographically) to the Xantphos ligand would also provide similar data to those reported.
A common theme of Xantphos-Ph complexes is their ability to promote reductive elimination at transition metal centres [4]. Buchwald noted that the trans-Pd(p-C 6 H 4 (CN))(Br)(j 2 -P,P-Xantphos-Ph) complex, which was contemporaneously prepared by van Leeuwen and co-workers [41], (complex 23, Scheme 10) (containing a large PAPdAP angle of 150.70 (6)°and weak at best interaction between the palladium and oxygen atoms with a distance of 2.696(4) Å) was a competent catalyst for the amidation of 4-bromobenzonitrile with benzamide [44]. Given that Scheme 11. Left: Postulated mechanism for the cis-j 2 -P,P/mer-j 3   the aryl halide precatalyst 23 was predominantly trans in solution, it was postulated that the proposed intermediate prior to reductive elimination, Pd(p-C 6 H 4 (CN))(NHCOPh)(j 2 -P,P-Xantphos-Ph), would also be trans, and therefore that dissociation of one of the phosphine arms could take place to allow reductive elimination from the tricoordinate {Pd(j 1 -P-Xantphos-Ph)} fragment (similar to 33, Scheme 13), or alternatively, that isomerisation to the cis-isomer could occur via breaking of the PdAO interaction and subsequent rearrangement of the phosphine groups. A possible low coordinate j 1 -P-Xantphos-Ph intermediate from which reductive elimination may occur was postulated in later work by Stockland and coworkers who suggested that fast reductive elimination observed during the formation of methylphosphonates using a palladium/ Xantphos-Ph system could be due to the dissociation of one of the Xantphos-Ph phosphines immediately prior to reductive elimination, analogous to that of complex 33 (Scheme 13) [45].
Following on from Grushin's initial report on facile ArACF 3 bond formation from the corresponding cis-j 2 -P,P-Xantphos-Ph complexes of palladium [46], in which the flexible nature of Xantphos-Ph was suggested to be critical, a more detailed study was undertaken in 2012 by Bakhmutov, Grushin, Macgregor and co-workers [47]. Pd(Ph)(CF 3 )(j 2 -P,P-Xantphos-Ph) (32) was found to be predominantly cis in solution, and evidence from both experimental and computational studies suggested that reductive elimination occurred directly from this isomer without the need for breaking of one of the PdAP bonds, that would result in a higher energy 3-coordinate j 1 -P-Xantphos-Ph complex (33) (Scheme 13).
Interestingly, when excess Xantphos-Ph was present, the rate of reductive elimination decreased slightly but this effect plateaued at higher ratios of free Xantphos compared to the palladium complex. This was accounted for by the formation of cis-Pd(Ph)(CF 3 ) (j 1 -P-Xantphos-Ph) 2 (34) -where two Xantphos-Ph ligands bind to the metal by one phosphine arm each -from which reductive elimination could also take place with a relatively low energetic barrier (Scheme 13). This finding suggested that the chelating nature of Xantphos-Ph, with its wide bite angle, is not necessarily required for reductive elimination and monodentate ligands with a steric profile that allows for both cis-coordination but remain bulky enough to promote reductive elimination may also be competent ligands for such processes.
A kinetic study on the palladium-catalysed amination of aryl triflates and aryl bromides was reported by van Leeuwen and coworkers, and utilised Xantphos-Ph, DPEphos, Sixantphos, Thixantphos, Thioxantphos and Homoxantphos ligands (L) [48]. Neutral complexes Pd(p-C 6 H 4 (CN))(Br)(L) were synthesised, where the DPEphos and Homoxantphos ligands provided cis-j 2 -P,Pcoordination to the metal centre only, whilst Sixantphos, Thixantphos and Xantphos-Ph ligands showed cis-trans isomerisation in solution, with the trans-complexes shown to be the crystallographically characterised form (i.e. for trans-Pd(p-C 6 H 4 (CN))(Br)(j 2 -P,P-Xantphos-Ph) (trans-23), see Scheme 10). For these complexes, a weak PdÀO interaction was proposed, with a PdAO distance of 2.698(2) Å found for complex trans-23 [41]. Only the neutral complexes capable of cis-trans isomerisation gave appreciable conversions during the amination of bromobenzene, which suggested that the ability of the ligand to adopt different coordination motifs was important. The cis/trans-Pd(p-C 6 H 4 CN)(Br)(j 2 -P,P-Xantphos-Ph) (cis/trans-23) complex was studied as a precatalyst for this transformation in an in-depth mechanistic study, and the proposed catalytic cycle is detailed in Scheme 14 (only the neutral cis-isomer is shown). The pre-equilibrium prior to the rate determining step was controlled by the ability of the Xantphos-type ligand in cis-Pd(Ph)(Br)(j 2 -P,P-L) (e.g., 35) to coordinate the ether oxygen and act in an associative hemilabile manner, which triggered dissociation of bromide to form cationic mer-[Pd(Ph)(j 3 -P,O,P-L)][Br] (e.g., 36). For example, the larger bite angle of Xantphos-Ph compared to Sixantphos, resulted in closer proximity between the metal centre and ether oxygen in the Xantphos-Ph complex 35, which subsequently showed higher reactivity in the amination of aryl bromides relative to the Sixantphos analogue. In comparison, the amination of aryl triflates was achieved by using cationic trans-[Pd(p-C 6 H 4 (CN))(j 3 (38), which placed the aryl and the amine cis, was proposed to be important -which is influenced by ligand flexibility and bite angle. As Sixantphos has a smaller bite angle than Xantphos-Ph, it favours cis-coordination geometry (i.e. 38) more so than Xantphos-Ph and thus showed greater reactivity in the amination of aryl triflates. These observations demonstrate the subtle interplay between ligand geometries, substrate binding and turnover limiting steps.
The work carried out by Bakhmutov et al. [47], in which excess Xantphos-Ph was found to cause only a slight decrease in the rate of reaction of reductive elimination in palladium complexes was detailed earlier in Scheme 13. However, Eisenburg and coworkers in their study of {Ir(Xantphos-Ph)}-catalysed hydroformylation of 1-hexene [49] found that excess Xantphos-Ph caused deactivation of the catalyst. They also postulated that a j 2 -P,P to j 1 -P-coordination mode change (Scheme 15) may be important for the oxidative addition of dihydrogen during the hydroformylation process. The mechanism for the oxidative addition of H 2 to (crystallographically characterised) Ir(Cl)(CO) 2 (j 2 -P,P-Xantphos-Ph) (39) that results in cis-Ir(Cl)(CO)(H) 2 (j 2 -P,P-Xantphos-Ph) (40) was proposed to occur by the steps shown in Scheme 15. No further details of the rate law were obtained.
A study by Macgregor and Grushin in 2014 described the mechanism of the azidocarbonylation of iodobenzene using CO and NaN 3 with an in situ formed Pd(CO) 2 (j 2 -P,P-Xantphos-Ph) precata-Scheme 13. Possible structures of intermediates from which reductive elimination of Ar-CF 3 could occur from in {Pd(Xantphos-Ph)} complexes. Reductive elimination from 33 was found to be energetically disfavoured.
lyst [50]. They proposed that j 1 -P and j 2 -P,P-Xantphos-Ph species were necessary for the aromatic azidocarbonylation to occur, as one of the phosphine arms must dissociate from the palladium centre to allow oxidative addition of the aryl halide substrate. In 2013, Fu and co-workers proposed, using DFT studies, that the borylation of nitriles using catalysts based upon M(BOR 2 )(j 3 -P,O, P-Xantphos-Ph) (M = Rh, Ir) required pathways which involved j 1 -P, j 2 -P,P and j 3 -P,O,P-binding of Xantphos-Ph to the metal centre [51]. In contrast, a similar report by Tobisu, Mori and Chatani published a year later, suggested that only j 2 -P,P and j 3 -P,O,Pcomplexes were required for a similar transformation [52]. In 2016, Slaughter and co-workers calculated that the CAH arylation of binaphthyl triflates could be promoted using a {Pd(Xantphos-Ph)} + catalyst via an inner-sphere concerted metalation-deprotona tion (CMD) mechanism, which involved j 3 -P,O,P (41), j 2 -P,P (42) and j 1 -P-Xantphos-Ph species (43) (Scheme 16) [53]. This was found to be the lowest energy pathway, with the other option an outer-sphere CMD arylation pathway in which the ligand remained j 2 -P,P-bound. The ability of the Xantphos-Ph ligand to display hemilabile behaviour therefore enables the inner-sphere mechanism, and the authors also note that the wide angle possessed by Xantphos-Ph was responsible for facile reductive elimination of the helicene product of CAH arylation.
As shown, j 1 -P-binding of the Xantphos-Ph ligand has been a proposed mode of binding in intermediates of catalytic reactions. Whilst such behaviour has been supported by computation, there is no direct experimental evidence for a j 1 -P/j 2 -P,P-coordination mode change in transition metal complexes during catalysis. Isolated examples where Xantphos-Ph binds in a j 1 -P-manner show however that such hemilability is plausible (e.g., complexes 22 and 65).
The catalytic carbothiolation of phenylacetylene with 2-(methylthio)acetophenone using a mer-[Rh(PhCCPh)(j 3 -P,O,P-Xantphos-Ph)][BAr F 4 ] precatalyst (44) was reported by Weller and Willis, in which the Xantphos-Ph ligand is proposed to occupy cisj 2 -P,P, fac-j 3 -P,O,P and mer-j 3 -P,O,P coordination sites during the catalytic cycle, shown in Scheme 17 [30]. Interestingly, the reaction from the Rh(I) precatalyst (complex 44) showed two catalytic regimes at 373 K: a fast region that gave 15% conversion after 20 min which was followed by a much slower rate (Fig. 4), suggesting that an initially fast catalyst was formed which then converted to a slower less-active species. The mer-Rh(III) aryl sulphide complex (45) was the only organometallic species observed in the reaction mixture by 31 P{ 1 H} NMR spectroscopy. This complex could be prepared independently at room temperature as the thermodynamic product of the reaction of the Rh(I) precatalyst 44 with 2-(methylthio)acetophenone after 32 h, which initially formed a 20:1 ratio of fac-j 3 -P,O,P and mer-j 3 -P,O,P isomers. On heating this mixture to 373 K, full conversion to the thermodynamically favoured mer-isomer was achieved after 40 min. The mer-j 3 -P,O,P complex (45) could be isolated, and its crystallographically-determined structure showed a tridentate pincer geometry of the ligand (PARhAP = 165.22(5)°, RhAO = 2.215(3) Å). The fac-j 3 -P,O,P isomer (46) could not be structurally characterised. When pure mer-j 3 -P, O,P complex 45 was used as the precatalyst the rate of reaction was slow at 373 K, with no fast regime observed (Fig. 4). This suggested that the mer-complex was responsible for the later, slow, catalytic regime when starting from Rh(I) precatalyst 44, and that the kinetically favoured fac-j 3   Haynes and co-workers proposed a mechanism for methanol carbonylation using {Rh(Xantphos-Ph)} + complexes in which j 2 -P,P to j 3 -P,O,P-rearrangements in Xantphos are proposed to be important [33]. Whilst throughout the catalytic cycle, the Xantphos-Ph ligand is shown to be predominantly mer-j 3 -P,O,Pbound (Scheme 19), the initial oxidative addition of MeI occurs at what is postulated to be a trans-j 2 -P,P complex, trans-Rh(CO) (I)(j 2 -P,P-Xantphos-Ph) (54). Whilst structural data for this Rh(I) complex could not be obtained, the Xantphos-(o-tol) analogue of complex 54, trans-Rh(CO)(I)(j 2 -P,P-Xantphos-(o-tol)) (54-o-tol), was crystallographically characterised, which showed the RhAO distance to be reasonably short (2.334(4) Å). However, they noted that MeI oxidative addition to this complex was disfavoured, possibly due to the increased steric bulk from the o-methyl groups, therefore 54-o-tol could not accurately describe the catalytic cycle proposed for Xantphos-Ph. DFT calculations suggested that there was heightened nucleophilicity of complex 54 because the Xantphos-Ph ether oxygen could associatively stabilise the S N 2  (Fig. 3), they also noted that the j 3 -P,O,P-intermediates postulated during the reaction could alternatively adopt fac-j 3 -P,O,P geometries instead, although these were not specifically detailed.

Xantphos-type ligands with alkyl substituents
In comparison to the Xantphos-Ph parent ligand, which formed Pd(Ph)(CF 3 )(j 2 -P,P-Xantphos-Ph) (32, Scheme 13) with both cis and trans-isomers observed in solution (the cis-isomer is the crystallographically characterised complex), the analogous Xantphos-i Pr complex was found to be exclusively trans-j 2 -P,Pbound in both solid-state and in solution [47]. In addition, reductive elimination from trans-Pd(Ph)(CF 3 )(j 2 -P,P-Xantphos-i Pr) was much slower in comparison to the Xantphos-Ph analogue, presumably due to the lower propensity for this complex to form ciscomplexes required for this process. Whilst the PAPdAP bond angle was in the range expected for trans-coordination of Xantphos-i Pr at the metal centre (144. 46(2)°), a possible PdAO interaction was less likely, as the distance between the two atoms is 2.784(1) Å -much longer than the lengths of PdAO bonds present in other Xantphos-containing complexes [40,41,57]. The ability of the Xantphos-i Pr to coordinate to a metal centre in both j 2 -P,P and j 3 -P,O,P-fashions has been demonstrated by Esteruelas and co-workers in osmium complexes [58]. The hexahydride complex, cis-Os(H) 6 (j 2 -P,P-Xantphos-i Pr) (56, Scheme 20), where the ligand was bound in a cis-bidentate manner, was stable in solid form or in solution under an atmosphere of H 2 . Although the solid-state structure could not be obtained using single crystal X-ray diffraction, the DFT-optimised structure of the complex revealed a PAOsAP angle of 105.4°, a structure consistent with cis-j 2 -P,P-Xantphos-i Pr coordination. However, when the hexahydride complex was left in MeOH solution under an argon atmosphere, one equivalent of H 2 was lost slowly over time, forming the tetrahydride complex, mer-Os(H) 4 (j 3 -P,O,P-Xantphos-i Pr) (57,Scheme 20), where the overall coordination mode switches from cis-j 2 -P,P to mer-j 3 -P,O,P. The j 3 -P,O,P-ligated complex was characterised by single crystal X-ray crystallography, and showed a short OsAO bond (2.222(3) Å) and a wide PAOsAP angle (164.53 (5)°) (data for one of the four crystallographically independent molecules in the asymmetric unit), which confirmed the pincerlike, mer-geometry of the ligand. From this initial work, a wider variety of variable coordination modes of Xantphos-i Pr were reported in 2017, again by the group of Esteruelas [59]. A number of dihydrogen and hydride complexes were stabilised by cis-j 2 -P,P, fac-j 3 -P,P and mer-j 3 -P,O,P binding of this ligand with Os, by stepwise addition of H + and H À sources, and overall loss of one equivalent of H 2 (Scheme 20). Completion of the cycle which involved H 2 loss from complex 56 to form 57 was achieved by hydride addition to 57 to form anionic cis-[Os(H) 5 (j 2 -P,P-Xantphos-i Pr)] À (58) which was characterised by a single crystal X-ray diffraction study. It was noted that mer-j 3 -P,O,Pcoordination of Xantphos-i Pr appeared to be favoured over fac-j 3  The bulkier Xantphos-t Bu ligand was also shown to adopt both trans-j 2 -P,P and mer-j 3 -P,O,P-coordination modes with a few crystallographically characterised examples reported, including mer-{Rh(j 3 -P,O,P-Xantphos-t Bu)} complexes prepared by Goldman and co-workers in 2013 [61]. In 2016, Sun, Zhang and co-workers synthesised gold complexes of Xantphos-t Bu for use in catalytic CAF to CAX (X = O, S or N) bond transformation studies [62]. Au (Cl)(j 2 -P,P-Xantphos-t Bu) (62) was shown to be a tricoordinate   (67)). The pincer-type coordination mode observed by the Xantphos-t Bu ligand was proposed to occur due to the greater rigidity of the ligand, combined with the bulky t Bu groups which did not allow a short enough P-P distance to form within the complex to accommodate a stable Au 2 pair, as occurs for the less sterically bulky Xantphos-Ph, DPEphos and DBFphos ligands.
Esteruelas studied the Xantphos-i Pr ligand in catalytic ammonia borane and dimethylamine-borane dehydrogenation [65]. The precatalyst of choice was a neutral mer-Rh(H)(j 3 -P,O,P-Xantphos-i Pr) complex (68), which DFT calculations showed to dissociate the oxygen atom of the Xantphos-i Pr ligand to form a tricoordinate T-shaped intermediate (69) to which the amine-borane could bind to form a sigma-complex (70) (Scheme 22). The mechanism suggested showed that the ligand remained cis-j 2 -P,P-coordinated to the rhodium centre throughout the catalytic cycle (Scheme 22), although there was no experimental evidence for this reported.
A computational report from Yu and Fu demonstrated that the alkylboration of 1-butene using (Bpin) 2 and 1-bromopropane, catalysed by Cu(Bpin)(Xantphos-R) complexes (R = Ph or Cy) operated by a mechanism that diverged at the oxidative addition step of the alkyl halide, the route of which was dependent on the substituent at the Xantphos-R phosphine groups [66]. Whilst Xantphos-Ph remained j 2 -P,P-bound to the metal centre at the rate-determining oxidative addition transition state (71), the equivalent complex containing Xantphos-Cy dissociated one of the phosphorus arms to form a j 1 -P-Xantphos-Cy transition state (72) (Scheme 23). Interestingly, the use of Xantphos-Cy ligand compared to the Xantphos-Ph also promoted differences in regioselectivity, favouring the Markovnikov and anti-Markovnikov products respectively. The authors explained this difference by the comparison of steric bulk of Cy versus Ph -the lower steric bulk of Xantphos-Ph enabled the ethyl substituent to lie close to the Xantphos-Ph ligand in the transition state (73) giving the anti-Markovnikov product, whereas the Xantphos-Cy ligand favoured the ethyl substituent on the alkene to be closer to the Bpin ligand in the corresponding transition state (74), which facilitated Markovnikov-selective alkene insertion.

DBFphos-type ligands
The synthesis of DBFphos (4,6-bis(diphenylphosphino)dibenzo furan) was first published in 1991 by Haenel et al. [67], and was found to have a wide natural bite angle of 131.1°, much larger than that of Xantphos-type ligands [12]. Initially, this was proposed to be of use in the stabilisation of bimetallic complexes through j 1 -P-binding of each phosphine to a transition metal centre, for example Co 2 (CO) 6 (j 1 -P,j 1 -P-DBFPhos) (75) (Fig. 5) [68]. The authors also discussed the possibility of oxygen coordination from the dibenzofuran motif to one of the cobalt centres, as this may have an effect on the potential catalytic activity of the complex, however the crystal structure indicated that no such CoAO interaction existed, as the distances between each Co and oxygen being long, i.e. 3.345(3) Å or 3.302(2) Å. Similarly, work by the groups of Lagunas [64], and Gray [63], showed that the DPFphos ligand also behaved in a j 1 -P,j 1 -P-binding mode to stabilise Au 2 complexes, (AuX) 2 (j 1 -P,j 1 -P-DBFphos) where X = Cl, Br, I (e.g., 66, Scheme 21). Interestingly, for the complexes where X = Cl or Br, an intramolecular aurophilic interaction between the gold atoms was not observed, unlike for Xantphos-Ph analogues (e.g., 65). This was possibly due to the wide bite angle of the ligand which allowed for intermolecular AuÁ Á ÁX stabilising interactions, rather than close, intramolecular AuAAu interactions which are generally considered to be weak. A report by Walton suggested that DBFphos could behave as a j 1 -P-ligand bound to one metal centre only in [Re 2 (Cl) 7 (j 1 -P-DBFphos)][N n Bu 4 ] (complex 89 shows the analogous DPEphos structure) [69], as evidenced by two separate resonances observed by 31 P NMR spectroscopy for the complex, which indicated one phosphine environment was bound to the rhenium metal centre, whereas the second phosphine displayed a chemical shift indicative of the free ligand.  Vogl et al. utilised the DBFphos ligand in their synthesis of mer-Ru(Cl) 2 (j 1 -P-phosphine)(j 3 -P,O,P-DBFphos) complexes (j 1 -Pphosphine = DBFphos (76), PPh 3 , p-tol), and found that the DBFphos ligand was surprisingly bound to the metal in both j 1 -P and j 3 -P,O,P-coordination modes in the same molecule (Fig. 5) [70]. Unfortunately, crystals of sufficient quality of complex 76, Ru(Cl) 2 (j 3 -P,O,P-DBFphos)(j 1 -P-DBFphos), could not be obtained, therefore a detailed structural analysis of the complex was not performed. However the solution-phase 31 P NMR spectrum of the complex showed three phosphorus environments, as expected for the simultaneous j 1 -P,j 3 -P,O,P-binding of the ligand, of which one had a chemical shift value close to that of uncoordinated phosphine. The tridentate nature of the DBFphos ligand was unambiguously confirmed during the synthesis and subsequent X-ray analysis of mer-Ru(Cl) 2 (PR 3 )(j 3 -P,O,P-DBFphos) (R = Ph and p-tol), which showed a wide bite angle (155.2(1)°) and close RuAO distance of 2.101(2) Å for the R = Ph derivative.

DPEphos-type ligands
DPEphos was initially synthesised by van Leeuwen and coworkers in 1995 for use as a j 2 -P,P-bidentate ligand [12]. Its natural bite angle of 102.2°was calculated to be lower than that of Xantphos-type ligands. Many reports have also suggested that due to the lack of a CMe 2 linker in its backbone, the DPEphos ligand is more flexible than its Xantphos-type counterparts.
The complex [Cu(DPEphos) 2 ][BF 4 ] (77), synthesised by Venkateswaran et al. [71], revealed one 31 P environment for the four phosphine groups at room temperature. However at À50°C the signal broadened, indicating fluxional behaviour in solution. In fact, the solid-state structure of the complex showed a tricoordinate Cu centre, with three short CuAP bond lengths of 2.2686 (8) Å, 2.2719(8) Å and 2.2782(9) Å, and one much longer CuAP distance demonstrating an unbound phosphine ligand (3.9576(9) Å) ( Fig. 6), i.e. DPEphos behaves as both a j 2 -P,P and j 1 -P-ligand in the same molecule. It was postulated that the fourth phosphine ligand was unable to establish a CuAP bond due to the relative bulkiness of the ligand in comparison to the analogous complex of the diphenylphosphinoethane (dppe) ligand, which does form a tetracoordinated solid-state structure with copper. It was proposed that the solution-phase fluxional behaviour demonstrated by the DPEphos ligand involved the coordination/decoordination of each phosphine arm. j 1 -P-coordination of the DPEphos ligand was also observed in the formation of bimetallic gold complexes, where the two phosphines on the same ligand each bond to different gold atoms (e.g., complex 67, Scheme 21) [63,64].
In a report by Balakrishna and co-workers, ruthenium complexes of DPEphos, incorporating cis-j 2 -P,P, fac-j 3 -P,O,P and mer-j 3 -P,O,P coordination were crystallographically characterised (Scheme 24) [72]. The variable coordination-mode ability of the DPEphos ligand was observed on addition of two equivalents of pyridine to complex 78, yielding trans,cis-Ru(Cl) 2 (py) 2 (j 2 -P,P-DPEphos) (80). The authors proposed that the first step of this reaction was through dissociation of the DPEphos ether linkage to reveal a vacant site to which the first pyridine molecule could bind (Scheme 24). The lack of RuAO bond in the cis-j 2 -P,P-complex was shown by a distance of 3.287(2) Å present between the two atoms.
The Weller group reported that on placing fac-[Rh{PCyp 2 (g 2 -C 5 H 7 )}(j 3   ing. The MeCN adduct, analogous to 84, could be formed, though. Equilibrium studies showed that the fac-isomer 86 was slightly favoured over the mer (87), due to entropic factors. The Xantphos isomers 86 and 87 also lost H 2 to form the corresponding mer-Rh (I) complex, mer-[Rh(PCyp 3 )(j 3 -P,O,P-Xantphos)][BAr F 4 ], analogous to complex 85, and this was observed to be faster than for the DPEphos variants.
As for Xantphos-Ph complexes described previously, the reactivity of DPEphos complexes may also require the dissociation of one of the phosphine arms, to form intermediates that contain a j 1 -P-DPEphos binding mode. Such behaviour was reported by Walton and co-workers, in which the formation of mer-Re 2 (Cl) 6 (j 3 -P,O,P-DPEphos) (88) from ( n Bu 4 N) 2 Re 2 Cl 8 was proposed to proceed through a [Re 2 (Cl) 7 (j 1 -P-DPEphos)][ n Bu 4 N] intermediate (89), as shown in Scheme 26 [69]. The final complex 88 was suggested to contain a mer-j 3 -P,O,P-bound DPEPhos ligand with a weak ReAO interaction; however the crystal structure of the complex could not be obtained. DPEphos was postulated to be flexible enough to allow for both mer-j 3 -P,O,P and cis-j 2 -P,P complexes to form over the course of the reaction, as the association of methanol was rate determining, and therefore the switch of ligand binding mode between complexes 90 and 91 was important for the overall rate. DG à for this transformation was estimated to be +83 kJ mol À1 , whereas the analogous system involving the Xantphos-Ph ligand had a significantly higher barrier, DG à = +92 kJ mol À1 . This increased barrier suggests that oxygen dissociation and ligand reorganisation of the Xantphos-Ph ligand was less favourable than for DPEphos.
The hemilabile nature of DPEphos was exploited by Moxham et al. in studies centred on the hydroacylation reaction of alkenes using aldehydes by the in situ formed precatalyst cis-[Rh (OCMe 2 ) 2 (j 2 -P,P-DPEphos)][CB 11 H 6 Cl 6 ] (94, Scheme 28) [28]. Hydroacylation involves a CAH activation of an aldehyde at a Rh(I) centre, followed by insertion of an olefin or alkyne and reductive elimination of a ketone product. There are therefore oxidation state changes and a requirement for formation of latent coordination sites during catalysis. A problem that persists during catalysis is the reductive decarbonylation from a low-coordinate acyl hydride intermediate that arises from oxidative addition of the aldehyde, which results in catalyst deactivation. The DPEphos ligand was shown to be able to stabilise the acyl hydride intermediate formed by alde-hyde addition to 94 by coordination of the ether linker, forming mer-[Rh(H)(MeSCH 2 CH 2 CO)(j 3 -P,O,P-DPEphos)][CB 11 H 6 Cl 6 ] (95) which was characterised in situ using NMR spectroscopy. As reductive decarbonylation of this species to form 96 was much slower than with other, non-hemilabile ligands, such as dppe, this allowed for more challenging substrates to be tackled in hydroacylation catalysis. For productive hydroacylation to occur, however, dissociation of the bound oxygen is required prior to binding of alkene, which was expected to form cis-[Rh(H)(g 2 -H 2 C@CHCOOMe) (MeSCH 2 CH 2 CO)(j 2 -P,P-DPEphos)][CB 11 H 6 Cl 6 ] (97). To support this, acetonitrile was added to the acyl hydride intermediate as a proxy for alkene, and the resultant solid-state structure showed cis-[Rh (H)(NCMe)(MeSCH 2 CH 2 CO)(j 2 -P,P-DPEphos)][CB 11 H 6 Cl 6 ] (98), in which the phosphine ligands were bound cis (PARhAP = 100.37 (3)°) and the oxygen atom was no longer coordinated [RhÁ Á ÁO = 4.271(1) Å]. The overall proposed catalytic cycle for this transformation is detailed in Scheme 28 and shows the importance of the ability for the DPEphos ether group to bind to the metal centre to stabilise intermediates, as well as dissociate from rhodium in order to allow further reactivity to occur. Notably, the use of the ligands POP-Ph (Scheme 5) and Xantphos-Ph in place of DPEphos did not lead to productive hydroacylation catalysis [27], as the central oxygen atoms were too strongly bound in the mer-j 3 -P,O,P geometry (similar to complex 95) to allow for alkene complexation.
As shown in Scheme 29 the stabilisation of fac-j 3 -P,O,P and cis-j 2 -P,P intermediate complexes demonstrated the importance of the hemilability of the ligand during such transformations. In particular, the isolation of the branched alkenyl intermediate (complex 99) which preceded turnover-limiting reductive elimination enabled the mechanism of this reaction to be studied in detail. The fac-j 3 -P,O,P binding of the ligand in this complex was demonstrated by the solid-state structure as confirmed by single crystal X-ray diffraction studies (PARhAP = 95.41(2)°, Scheme 28. Proposed catalytic cycle for the hydroacylation of alkenes using the {Rh(DPEphos)} fragment. CB 11 H 6 Cl 6 anion omitted for clarity. RhAO = 2.3458(17) Å). The a,b-unsaturated ketone product was found to bind to the metal centre after a slow reductive elimination event from this alkenyl complex, which yielded a 5-coordinate complex (100) -where the crystal structure showed a cis-j 2 -P,P binding mode of the DPEphos ligand, indicated by the PARhAP bite angle of 97.87(5)°and long RhÁ Á ÁO distance (3.395(5) Å). The ketone product was released from the metal centre with oxidative addition of the starting aldehyde and coordination of the DPEphos ether group, and reformed the mer-j 3 -P,O,P acyl hydride species (101).
Another example of catalysis in which the DPEphos was used with success, and the flexible nature of the ligand shown, was in alkyne carbothiolation [31]. The Rh(I) j 2 -P,P-DPEphos precatalyst (102) oxidatively added an aryl methyl sulphide to form a merj 3 -P,O,P-DPEphos complex (103), which was in equilibrium with its analogous fac-j 3 -P,O,P complex (104) (Scheme 30). This equilibrium was proposed to operate through a 5-coordinate intermediate where the DPEphos oxygen dissociated from the metal in order to access a conformationally flexible species required for fac-mer isomerisation. Similar behaviour was noted with the Xantphos-Ph ligand in which it was demonstrated that the facj 3 -P,O,P-Xantphos-Ph catalyst was a more active catalyst than the mer (Scheme 17) [30]. j 2 -P,P to j 1 -P-coordination of the DPEphos ligand was proposed to be important in a study by Ohshima et al. in the platinum-catalysed amination of allylic alcohols which utilised an in situ formed catalyst from Pt(COD)Cl 2 /DPEphos [76]. It was found that a 1:1 ratio of Pt:DPEphos gave rise to the formation of a black precipitate during catalysis -a result of decomposition of the active Pd(0) species. To combat this, two equivalents of DPEphos with respect to the metal were employed in an attempt to stabilise the active species by formation of Pt(j 2 -P,P-DPEphos) 2 . This coordinatively saturated complex was an active precatalyst, and it was speculated that complete dissociation of one of the DPEphos ligands, or alternatively decoordination of just one of the DPEphos phosphine arms to form a coordinatively unsaturated platinum metal centre with a j 1 -P-DPEphos ligand, allowed a coordinatively unsaturated, active Pt(0) catalyst to form. This method of stabilising the active species towards decomposition was also attempted using the Xantphos-Ph ligand, however the corresponding Pt(j 2 -P,P-Xantphos-Ph) 2 complex showed limited solubility in toluene, the reaction solvent of choice.
A very recent study carried out by Goldberg and co-workers [77], investigated the use of rhodium complexes containing the DPEphos ligand for the reductive elimination of products with C Scheme 29. Proposed catalytic cycle for the hydroacylation of alkynes to form the branched alkenyl product using the {Rh(DPEphos)} fragment. CB 11 H 12 anion omitted for clarity.
Scheme 30. fac-mer isomerisation of the catalyst resting state during the carbothiolation of alkynes using a {Rh(DPEphos)} + fragment. BAr F 4 anions omitted for clarity.
(sp 3 )AX bonds (X = N, S, I). The elimination of MeI from mer-Rh (Me)(I) 2 (j 3 -P,O,P-DPEphos) (105) was found to occur via two competing pathways which both involved S N 2 attack of the iodide on the methyl group of either a neutral or cationic species. The mechanism detailed in Scheme 31 was suggested to involve initial decoordination of the DPEphos ether linkage from the metal in complex 105 to form complex 106, trans-Rh(Me)(I) 2 (j 2 -P,P-DPEphos) that then underwent attack by iodide, and subsequent elimination of MeI. The hemilability of the DPEphos was suggested to increase the electrophilic nature of the methyl group and encourage nucleophilic attack of the iodide to eventually form the final product mer-Rh(I)(j 3 -P,O,P-DPEphos) (107) with loss of MeI. Reaction with the solvent, iodobenzene, formed mer-Rh(Ph)(I) 2 (j 3 -P,O,P-DPEphos) (108). Addition of excess sodium thiophenolate to complex 105 gave the corresponding MeSPh product, square pyramidal cis-Rh(Ph)(SPh) 2 (j 2 -P,P-DPEphos) (109). A DFT study on alkene hydroamination reactions catalysed by the {Rh(DPEphos)} + catalyst [78] was published by Couce-Rios et al. in 2016 [79]. In this, structural isomers (j 2 -P,P-DPEphos versus j 3 -P,O,P-DPEphos) of proposed intermediates along the calculated reaction pathway were studied. Interestingly, the structures for which DPEphos occupied j 3 -P,O,P coordination sites were more than 6 kcal mol À1 higher in energy than their j 2 -P,P-DPEphos analogues.

Xantphos-type ligands containing aminophosphine groups
Xantphos-type ligands containing aminophosphine groups have been developed by van Leeuwen [80] and Hartwig (e.g., compounds 110 and 111, Fig. 7) [81]. In particular, Hartwig reported their use in rhodium-catalysed intramolecular hydroamination of aminoalkenes, which proved to be advantageous compared to the same reactions with the more ubiquitous Xantphos-Ph or Xantphos-Cy analogues. The Xantphos-NEt 2 ligands (e.g., 110) led to the best conversions, and also showed the greatest selectivity for hydroamination over oxidative amination, attributed to the mer-j 3 -P,O,P binding mode of the ligand to the rhodium centre, which may inhibit any possible beta-hydride-elimination (Scheme 4) and therefore prevented formation of unwanted oxidative amination products. The j 3 -P,O,P-tridentate binding mode of the Xantphos-NEt 2 ligand was confirmed by an X-ray diffraction study of the precatalyst, mer-[Rh(NCMe)(j 3 -P,O,P-Xantphos-NEt 2 )][BF 4 ] (112, Fig. 7), and showed a short RhAO bond of 2.129 (8) Å and a wide PARhAP angle of 168.0(1)°, whilst the proposed catalytic cycle retained j 3 -P,O,P-coordination of the ligand in the formation of intermediates. The activity of the aminophosphine ligand 110 in hydroamination catalysis was around twice that of the Xantphos-Cy and Xantphos-Ph ligands, therefore the electronic and steric effects of the ligands were compared. The complexes {Rh (CO)(Xantphos-NEt 2 )} + and {Rh(CO)(Xantphos-Cy)} + were synthesised, and their m CO values were found to be 1993.3 cm À1 and 1992.7 cm À1 respectively, indicating that both the amino and alkyl Xantphos ligands have similar electron-donating effects to the metal centre. In addition, a Xantphos-type ligand containing piperidinyl groups (111) was proposed to be isosteric to Xantphos-Cy, but showed increased reactivity similar to that of Xantphos-NEt 2 , therefore steric factors were also speculated to not be of importance. The authors suggested that the effect of the substituents cannot be attributed to steric or electronic demands, though other reasoning was not explored. Aminophosphine Xantphos-NR 2 ligands such as 110 and 111 have not yet been explored for their hemilabile capabilities.
Interestingly, complex 119 was shown to be inactive in the amination of aryl triflates (see Scheme 14 for analogous Xantphos reactivity), postulated to be due to strong coordination of the soft sulphur atom to the palladium metal centre which is not readily dissociated to allow for onward reactivity [48].

Concluding remarks
This review has explored examples of hemilabile-type behaviour and flexible coordination modes of POP-type ligands in transition metal complexes. The subtle interplay between sterics (imposed by the appended phosphine groups) and ligand flexibility or bite angle (stemming from the nature of the ligand backbone), plays an important role in determining the j 1 -P, j 2 -P,P or j 3 -P,O, P-coordination mode of the POP-ligand with the metal. In particular, the ether linker has the capacity to occupy vacant sites in lowcoordinate complexes, but may remain easy enough to dissociate  Scheme 32. Left: Structure of complex 120 which has been crystallographically characterised to show the fac-j 3 -P,S,P coordination of the PSP-type ligand. Right: Variable coordination modes of a derivative of ligand 117 (Fig. 8), both the cis-j 2 -P,P in 121 and mer-j 3 -P,S,P coordination in 122 was shown by X-ray diffraction studies. from the metal centre to allow the coordination of incoming substrates. Perhaps surprisingly, but persuasively, calculations have also demonstrated that the phosphine groups may have the capacity for such hemilabile behaviour as well.
Throughout the current literature and discussed in this review, the MAO bond distance and PAMAP bond angle of crystallographically characterised complexes containing POP-type ligands have been helpful in determining the coordination geometry of the ligand. Building on the graphical representation of MAO bond distance versus PAMAP bond angle for Xantphos-Ph complexes reported by Haynes and co-workers (Fig. 3) [33], we now construct the analogous graph for all complexes contained in Cambridge Structural Database (CSD) containing a POP-type ligand (Fig. 10) [90][91][92]. The clustering of data in Fig. 10 allows for demarcation of the four possible coordination geometries of POP-type ligands in transition metal complexes, although data which sit close to a boundary should be more closely interrogated to determine the precise bonding mode. In general, complexes with long MAO distances and small bite angles (MAO > 2.5 Å, PAMAP < 125°) are considered to be cis-j 2 -P,P-POP-type complexes with no formal metal-oxygen bond and lie at the top left of the plot. trans-j 2 -P, P-POP-type complexes where the bite angle is large, but the MAO distance remains long, are situated to the top right of the plot (MAO > 2.4 Å, PAMAP > 130°). There may be an inherent, at best weak, metal-oxygen interaction present in some of these complexes due to the close proximity of the metal and oxygen atoms caused by the trans-spanning nature of the POP-type ligand. Complexes occupying the bottom right of the graph are mer-j 3 -P,O,Pcomplexes, in which MAO distances are shorter and the phosphine ligands are situated trans (MAO < 2.4 Å, PAMAP > 145°). This subset of ligand geometry can be considered to be pincer-like. The final set of coordination complexes are found at the bottom left of the figure, containing complexes with both short MAO distances and narrow bite angles (MAO < 2.4 Å, PAMAP < 130°), which are examples of fac-j 3 -P,O,P structures, where the phosphine ligands lie cis to one another, and the ether linker lies in the apical position of the complex, and cis to both phosphines.
The ability of POP-type ligands to adapt their denticity in a variety of ways to respond to the requirements of the metal centre shows their versatility for the fine-tuning of reactivity of their asso-ciated transition metal complexes. Given the body of work that already exists, there is no doubt that the flexible coordination modes and hemilability of POP-type ligands will lead to their increased use in the synthesis of tailor-made homogeneous catalysts, which often -of course -require variable and accessible coordination geometries throughout the catalytic cycle.