Magnesium, zinc, aluminium and gallium hydride complexes of the transition metals† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c6cc05702k

Here we survey and organise the state-of-the-art understanding of the TM–H–M linkage (M = Mg, Zn, Al, Ga). We discuss the structure and bonding in these complexes, their known reactivity, and their largely unrealised potential in catalysis.


Introduction
The catalytic practices of C-H bond functionalisation, dehydrocoupling (for hydrogen storage), hydroboration and hydrosilylation are all attractive prospects for the future chemical economy. The modern development in these methodologies continues to be enhanced by the perception of borane and silane σ-complexes as intermediates in reaction mechanisms (Figure 1). A σ-complex can be described as an η 2 -binding of the σ-E-H bond to a transition metal centre (E = C, Si, B, H). 1,2 Along with dihydrogen complexes, [3][4][5][6] σ-silanes are the most comprehensively studied type of this bonding mode. [7][8][9][10][11][12] While the latter appear as potential intermediates in alkene hydrosilylation via the Chalk−Harrod mechanism, 13,14 the former bear significance for a range of industrially relevant hydrogenation reactions. 15,16 In C-H borylation, stabilisation of catalytic intermediates by a TM-H-B (TM = transition metal) interaction has been supported by significant experimental mechanistic studies. 17  The TM-H-M motif is one way of adjoining two metal centres bearing at least one reactive hydride ligand (Figure 2a). This motif can also be obtained by coordinating a transition metal hydride to a neutral main group metal fragment (Figure 2b), and multiply bridged species formed by a combination of the two aforementioned donor-acceptor interactions (Figure 2c).
Herein we survey the known heterobimetallic complexes of transition metal and main group hydrides (M = Mg, Zn, Al, Ga). The coordination chemistry of heterobimetallic transition metal hydrides, [20][21][22] and of Al, Ga, In and Zn-based ligands at transition metal centre have been summarised previously. [23][24][25][26] To focus the discourse, a loose definition 'heterobimetallic hydride complexes' is in order to represent charge neutral species rather than an accurate representation of the bonding within the TM-H-M group. The literature survey is followed by discussion of the "continuum" of bonding descriptions, reactivity and the potential these complexes hold for catalysis.

σ-Complexes (TMßH-M)
We have reported 1, a Zn congener of structurally related σ-alane complexes (Figure 3). 27 The binding of the zinc hydride to the Cu I centre is weak and reversible. In toluene or benzene solution, an equilibrium exists between the heterobimetallic complex and the η 2 -arene complex of Cu I . The electronic structure of the three-centre linkage has been investigated by DFT calculations. The analysis suggests that the formal L donation of the M-H σ-bond to the 4s orbital on Cu I is accompanied by weak CuàM back-donation into the M-H σ*-orbital (see discussion section). In combination, the data allow 1 to be described as a weakly bound σ-complex. DFT calculations, the coordination of exogenous ligands to the Cu I fragment was found be increasingly exergonic across the series C 6 F 6 < H-B < H-Si < C 6 H 6 < H-Zn < H-Al -a trend that is manifest in the experimental data. 27 In 2 the Cu---Al vector lies outside of the H-Al-H wedge -the only example of this structural feature we are aware of in main group metal σ-complexes. 27 The 'lee side' coordination of the Al-H bond and very long intermetallic distance in 2 are notable. For comparison, σ-complexes of HBpin (pinacolborane) or HBcat (catecholborane) show different coordination geometries to those of fourcoordinate boranes BH 3 •EMe 3 (E = N, P). 28,29 The discrepancy has been rationalised by disruption of the TM---B back-donation due to the absence of a vacant orbital of suitable energy in BH 3 •EMe 3 , be it the boron p-orbital or the σ*-orbital of the B-H bond. 28 While steric factors are undoubtedly important, a similar effect may explain the solid state structures of 2 and 3. DFT calculations are consistent with reduced back-donation from d 10 Cu I into the Al-H σ*-orbital of the five-coordinate species when compared to the four-coordinate analogue. 27 Higher nuclearity species containing Cu-H-Al interactions are known and the intermetallic cluster [(Cp*AlCu) 6   Aldridge and co-workers pioneered this area of research and have isolated a series of σ-alane and σ-gallane complexes of groups 6 and 7 transition metal carbonyls. 31 isolated in an approximate 9:1 mixture with the minor component containing the η 2 :η 2 -coordination mode. 34 The adduct 8-W/Al could not be obtained by photoejection of CO from [W(CO) 6 ] and was finally obtained under thermal conditions using [W(CO) 4 (1,5-COD)] as a precursor (COD = cyclooctadiene). 34 The M---Al distance in 8-W/Al is shorter than would be expected based on the lighter members of the series, presumably due to a tighter binding of the σ-alane to the more expanded 5d orbitals of W. In contrast to the W analogue, 8-Cr/Al may be formed directly upon heating 7-Cr/Al: Eyring analysis and the first order kinetics of this reaction have led Aldridge and coworkers to suggest it proceeds by an associative pathway. 34 All TM-H-Al heterobimetallic complexes characterised by Aldridge and co-workers show slow exchange between the bridging and terminal hydride ligands on the NMR timescale at ambient temperature. [31][32][33][34]36 From structural and spectroscopic evaluation of these complexes (4-9), it appears that back-donation into the M-H σ* orbital is negligible. While the close TM---M contacts (especially in 9-Ga) are short enough to hint at TM-M interaction, the four-membered ring imparted by the η 2 :η 2 -coordination mode in 8-9 demands such a short contact. 33 The weaker nature of the  37 Ueno and co-workers provided evidence that the η 2 :η 2 -coordination mode is not necessary for the stabilisation of σ-gallanes, and reported 12 and 13. 38 40 An alternative approach to TM-H-Al groups has been discovered by Fischer and co-workers.
Addition of [Cp*Al] 4 to transition metals with labile ligands is proposed to generate intermediates of the form [TM(AlCp*) n ] which react further, effecting the inter-or intramolecular C-H activation of arenes or alkanes. 41,42 Complexes 15-16 are formed through this route and possess geometries that are consistent with the σ-alanes described above (Figure 4). For example, the Al-H distance in 15 of 1.76(3) Å lies within the range established σ-alanes. While 16 possesses elongated Al-H distances ranging from 1.88(8) to 1.89(7) Å, these are still substantially shorter than the >2.0 Å separation required to suggest oxidative addition (vide infra). The latter may be described as a complex containing stretched σ-alane ligands with an Al-H-TM geometry somewhere between coordination and oxidative addition.

Oxidative Addition / Hydride Transfer (H-TM-M)
We  analogues. As such, while these Zn and Mg complexes could be described as oxidative addition products, hydride transfer to form an is also a fair description. 43 The structure of 17-Al, an analogue of 18 that incorporates more sterically demanding substituents on the β-diketiminate ligand, shows a geometry with familiar trans-disposed hydrides, short TM-Al distance and long Al---H distances ( Figure   leads to the dimer 21 ( Figure 7). 43 Complex 21 contains a Rh 2 Al 2 H 4 core. Supporting the argument for Al I is the deviation of the alumocycle from planarity, suggestive of decreased π-donation by the N atoms into the 3p orbital of Al. This allows for Al I to act as both a Z-and L-type ligand with respect to Rh. 43 50,51 as are higher nuclearity species in which multiple main group fragments act as ligands for the transition metal (see supporting information, Figure S1). [52][53][54][55][56][57] The work on cluster complexes supported by organozinc, organoaluminum and organogallium ligands has been reviewed before. 23 Coordination of a related aluminium dihydride to a 14-electron {Co I (CO) 3 } + synthon gives 22 ( Figure   7). 34 This latter species appears to be a product of simultaneous addition of both Al-H bonds to Co.   The TM---M distances in 27-29 are in range of the sum of the covalent radii, and the coordination at iridium in 27 is such that the Ir-Al bond is distorted away from the IrH 2 plane by 37°. 62 The geometries contrast with those of 30-W and 30-Mo. First reported by Wailes et al. 63 and Storr et al., 64 30-W has been characterised by X-ray diffraction and analysed by DFT methods. 65 63 The potential for reversible H/D exchange between the hydrides and protons of the cyclopentadienyl ring of 30-W has been highlighted and proposed to occur by a mechanism involving anchimeric assistance. 65 Non-reversible intramolecular deprotonation of the cyclopentadienyl ligands is also well established, and often leads to high nuclearity species such as complex 31. [68][69][70][71] When [Cp 2 MoH 2 ] was treated with ethylzinc bromide, 32 was isolated and presumed to derive from ZnBr 2 formed from a Schlenk-type equilibrium. 72 An η 2 :η 2 -bonding mode of the Mo hydrides is implied by the narrowing of the MoH 2 wedge upon adduct formation, data that contrast those of 27.

Late Transition Metal Adducts
A similar adduct, 33, is formed from addition of a bis-aryloxy zinc solvate to a Rh trihydride complex. 73 Reaction of a mixture of NbCl 5 , sodium cyclopentadienyl and zinc powder in THF under an atmosphere of CO, and subsequent treatment with NaBH 4 gives 34. 74 The metal---metal distance in   (Figure 9). [78][79][80][81] In more recent work, Bourissou, Uhl and co-workers have shown that hydrogenation of an intramolecularly coordinated PtàAl adduct leads to the heterobimetallic hydride 39 ( Figure 10). 82 Calculations suggest that H 2 addition occurs across the PtàAl bond. There is precedent for this intramolecular coordination mode: Fischer and co-workers have reported the gallium adduct 40. 83 Related hydrogenation reactions of platinium diene complexes either bearing a Z-type Al ligand or in the presence of Ga I co-ligands lead to the formation of heterobimetallic complexes bearing terminal hydride ligands on the transition metal ( Figure S1). 46 Figure 11). 92 Alanes react with yttrocene hydride or carboxylate complexes to form YßH-Al adducts (55)(56). 95 While investigating the salt metathesis of a β-diketiminato-supported Sc dichloride complex with LiAlH 4 , Piers and co-workers isolated the alane adduct, 57. 107 In 57, both metals are six coordinate, and this is the only time a (µ-H) 3 bridging motif seen in the absence of a late TM. Heterobimetallic hydrides of rare earth metals are not limited to those in which the heavy metal is in the +3 oxidation.

Bulychev et al. found that [(1,3-t Bu 2 -C 5 H 3 ) 2 Sm] partially decomposes into an octanuclear aggregate
of Sm III and alane, when treated with AlH 3 in the presence of TMEDA. Upon substitution of the alane for AlD 3 , however, 58 may be isolated: an apparent effect of isotopic substitution. 108 This may be a result of tighter binding due to an increased ionic contribution to the donor-acceptor linkage ( Figure 13). 120 Grignard reagents or Mg 0 powder in etheric solvents. [135][136][137][138][139][140][141][142] For preparations in which a main group hydride is not used as a reagent the hydride ligands result from either: i) C-H atom abstraction from the solvent (THF), ii) β-hydride elimination group derived from a main group alkyl, or iii) C-H bond activation of the cyclopentadienyl ligand (or its substituents). Ligand activation is common in these complexes and intramolecular deprotonation to form dianionic Cp ligands, including "tuck-in" complexes has been observed to lead to both diamagnetic TM IV and paramagnetic Ti III complexes (72 & 78-79). 126,133,134,143,144  Common themes emerge in the coordination geometries of these complexes. 170 Unlike the group 4 analogues described above, the (µ-H) 3  The low-spin ground state structure of 100 has been thoroughly investigated by computational methods. 174

Discussion
An exhaustive account of the preparation and structures of heterobimetallic hydride complexes reported over the last half a century is presented above. During our own research in this area and through analysis of the material above, we have alighted upon recurring issues in the understanding this family of complexes. For the benefit of potential future studies, these issues are discussed below.  Considering the structures of the isolated complexes in sections 2 and 3, it becomes clear that the well understood continuum between σ-complexes and oxidative addition products detailed for addition of E-H bonds to transition metals also applies to M-H bonds.

Sigma-Complexes and Oxidative
The triangular TM-H-M unit is subject to structural changes as the electron density changes at the TM ( Figure 17). The formal shortness ratio (fsr) normalises the metal---metal distance and has been used to evaluate the intermetallic interaction in complexes containing two metals in close proximity.
For data collected to date, this metric appears to conveniently describe the extremes of the reaction coordinate: σ-complexes (fsr approx. > 1) and products of oxidative addition / hydride transfer (fsr approx. ≤ 1). 179 One caveat of this approach is that short metal---metal distances can arise due to the geometric constraints imposed by multiple bridging ligands, to date the analysis has only been applied to molecules containing a single TM-H-M unit. In the case of the analysis presented below, this concern is circumvented by performing DFT calculations to confirm the location of the hydride.  43 In combination the data suggest that the ionic contribution becomes more significant for the more electropositive elements.  While, in general, the higher electronegativity of the main group element in silanes (and to an extent boranes) means that TM-HàER n bonding descriptions are less common, there is growing appreciation that these descriptions may be relevant in σ-silane and σ-borane chemistry.   By considering the properties of the TM-H-Zn complexes on the reaction trajectory presented in Figure 18, DFT studies have shown that as the hydride is transferred from zinc it becomes less hydridic and more acidic. 37 While the data would suggest that a rich acid/base chemistry may be possible with heterobimetallic hydride complexes, the reactivity of these species remains understudied. The same with insertion of unsaturated substrates (e.g. alkenes, alkynes, CO, CO 2 , carbonyls, etc.) into 3-centre 2-electron TM-H-M bonds: it remains unclear how this chemistry will compare to the well-studied reactions of transition metal hydrides.

iii) Potential in Catalysis
Transition metal hydride complexes are known intermediates in the hydrogenation of unsaturated hydrocarbons, CO (Fischer-Tropsch), and CO 2 along with numerous other small molecules. [185][186][187] They also play a key role in hydrogenase enzymes and related catalysts for H + reduction to H 2 and have been invoked as (off-cycle) intermediates during a number of important polymerisation reactions including Ziegler-Natta catalysis. 85,109 To date however, defined catalytic reactions believed to involve heterobimetallic hydride complexes are limited. Systems that are catalytic in both the transition and main group metal are the most attractive. Lu and co-workers have reported the heterobimetallic complexes 107 as catalysts for alkene hydrogenation and isomerisation and demonstrated that the nature of the main group metal, M, effects the activity (Figure 24). 188 Dihydrogen activation by addition across a NiàM bond has been proposed as a key step in hydrogenation catalysis.

Figure 24. Catalytic alkene hydrogenation and isomerisation by heterobimetallic complexes
The importance of transition metal hydrides to numerous aspects of catalysis is not up for debate.
Whether or not heterobimetallic complexes will be able to offer advantages over existing monometallic systems remains an open question for the community. It is clear that substantial developments in the stoichiometric and catalytic reactivity of heterobimetallic hydrides need to be made. Nevertheless, the co-location of two metals by direct metal-metal bonds and/or bridging ligands offers two opportunities in catalysis: new fundamental reactivity, and fine tuning of selection events in known reactions.
The task for the current generation of chemists is to transcribe the known methods for preparation, and the structural understanding of, heterobimetallic hydride complexes into new reactivity. While progress in this area seems to have been made in stops and starts, given the renewed interest in heterobimetallics and main group complexes for catalysis, the challenge appears timely. [189][190]