Alkali Metal (Li, Na, K, Rb, Cs) Mediation in Magnesium Hexamethyldisilazide [Mg(HMDS)2] Catalysed Transfer Hydrogenation of Alkenes

Though s‐block organometallics are indispensable tools in synthetic chemistry, their impact in homogeneous catalysis has been small compared to that for transition metals. Surprisingly, however, Group 1 and 2 compounds have recently been utilised in hydroelementation catalysis. For example, transfer hydrogenation of alkenes to alkanes has been catalysed by Group 2 (Ca, Sr, Ba) bis‐hexamethyldisilazides (HMDS), though reactions failed with magnesium. Stimulated by the knowledge that alkali metals can mediate applications of other compounds, this study reports successful catalytic transfer hydrogenation of alkenes with magnesium bis‐hexamethyldisilazide when combined with alkali metals in heterobimetallic complexes. By screening all Group 1 metals (Li−Cs), the efficiency of these catalyses gives an indication of the relative power of the synergistic effect exerted on magnesium by the alkali metal. The Cs−Mg partnership shows the best efficiency with styrene, while K−Mg works best with 1,1‐diphenylethylene.

Though s-block organometallics are indispensable tools in synthetic chemistry, their impact in homogeneous catalysis has been small compared to that for transition metals. Surprisingly, however, Group 1 and 2 compounds have recently been utilised in hydroelementation catalysis. For example, transfer hydrogenation of alkenes to alkanes has been catalysed by Group 2 (Ca, Sr, Ba) bis-hexamethyldisilazides (HMDS), though reactions failed with magnesium. Stimulated by the knowledge that alkali metals can mediate applications of other compounds, this study reports successful catalytic transfer hydrogenation of alkenes with magnesium bis-hexamethyldisilazide when combined with alkali metals in heterobimetallic complexes. By screening all Group 1 metals (LiÀ Cs), the efficiency of these catalyses gives an indication of the relative power of the synergistic effect exerted on magnesium by the alkali metal. The CsÀ Mg partnership shows the best efficiency with styrene, while KÀ Mg works best with 1,1-diphenylethylene.
Homogeneous hydrogenation is a synthetic procedure of fundamental importance to the chemical industry as well as being one of the most extensively studied reactions in homogeneous catalysis. [1][2][3] Classic compounds such as the Wilkinson catalyst or the asymmetric hydrogenation catalysts from Noyori or Knowles exhibit excellent performance under mild conditions alongside high selectivity and good functional group tolerance. [4][5][6] Besides the aforementioned benefits these catalysts are centred on late transition metals and the growing awareness of their toxicity and high costs has prompted chemists to seek to develop alternatives. Potential substitutes for these precious metals are the elements of the s-block. They are (mostly) biocompatible, environmentally benign and with their high abundance in the crustal layer and oceans of earth, a low price per mole is guaranteed. Activity in this area has been steadily rising since the turn of the millennium with pioneering contributions by Harder, Hill, Okuda, Cheng and Stock, with several catalytic hydrogenations published.  In most cases the catalytically active species is a highly reactive metal hydride complex. The polar nature of the bond between the electropositive s-block metal and the electronegative hydride ligand equips the latter with strong nucleophilicity, which facilitates the insertion into unsaturated bonds. Although, such high reactivity makes the isolation of these compounds challenging, suitable synthetic procedures are known. The first fully characterised magnesium hydride complex is an inverse crown reported in 2002 by the Mulvey group. Inverse crowns are bimetallic compounds in which cationic rings host anionic guests. [28] This sodium-magnesium inverse crown was synthesised by heating a 1 : 1 : 3 mixture of nbutylsodium, n-secdibutylmagnesium and diisopropylamine in toluene resulting in a β-hydride elimination reaction of one of three isopropyl amide ligands, with the released hydride stored within the sodium and magnesium bimetallic scaffold (Scheme 1a). [29,30] The analogous potassium magnesiate could be isolated via the same methodology using benzyl potassium as potassium source. [31] A frequently applied synthetic route to hydride complexes is to exchange a reactive group from the metal precursor with a hydride ligand from the silane PhSiH 3 via σ-bond metathesis. Hill and co-workers used this procedure to good effect to access other heterobimetallic s-block hydrides. Their in situ generation of [KMg( n Bu)(HMDS) 2 ] [HMDS=N(SiMe 3 ) 2 ] and subsequent treatment with PhSiH 3 resulted in a related inverse crown stabilised by bulky HMDS ligands (Scheme 1b). [32] In 2018 Harder reported a simple alkene hydrogenation by AEM (HMDS) 2 (AEM=Ca, Sr, Ba) (Scheme 1c). [13] Although no hydride complex was isolated from this study, the catalytically active species were proposed to be in situ generated hydride clusters reported two years earlier. [33,34] The same compounds were proposed to catalyse alkene transfer hydrogenation (Scheme 1d). Deprotonation of 1,4-cyclohexadiene (1,4-CHD) by AEM(HMDS) 2 (AEM=Ca, Sr, Ba) gives an unstable Meisenheimer complex, which then undergoes β-hydride elimination to the corresponding hydride and benzene. [14] A low-cost route to heterobimetallic s-block metal hydrides was presented very recently by the Guan group (Scheme 1e). Barely soluble potassium hydride was mixed together with AEM(HMDS) 2 (AEM=Mg, Ca) resulting in formation of an effective hydro-genation catalyst for olefins. [35] Although the active species itself could not be isolated, the group found significant evidence that (in the case of magnesium) the compound is presumably the aforementioned [KMg(H)(HMDS) 2 ] 2 reported by Hill et al. [32] In our continued development of synergistic alkali-metalmediated bimetallic chemistry of main group metals, [36] we recently started to investigate their performance in catalysis. Our initial investigations have found that certain bimetallic systems can operate more efficiently compared to their monometallic counterparts. This is evident in aluminium catalysed hydroboration of aldehydes, ketones and imines, where the key reaction intermediate being more polarised in the bimetallic compound than in the neutral complex displays a marked higher reactivity. [37] In lithium-aluminate catalysed hydrophosphination the lithium boosts the Lewis acidity of the reactive aluminium centre. [38] A comprehensive summary about alkali metals in bimetallic catalysed reactions and their excep-tional ability to mediate a plethora of reactions was recently reviewed separately by the Hevia and Mulvey groups. [39,40] Studies comprising the complete Group 1 (Li, Na, K, Rb and Cs) are still exceptionally rare in organometallic chemistry and lesser still is the mediation efficiency of the larger siblings rubidium and caesium probed. With their higher electropositivity and preference towards electron rich systems they have the potential in theory to exceed the reactivity limits of the smaller metals lithium, sodium and potassium. Herein, we report the catalytic transfer hydrogenation of alkenes by the whole series of alkali metal magnesiates including a detailed evaluation of the influence of the alkali metal supported by solid state structures. Moreover, we reveal the first example of a rubidium inverse crown of any type, the significance of which is magnified since it incorporates a rubidium hydride moiety.
As alluded to in the introduction, in 2019 Harder and coworkers reported the transfer hydrogenation of olefins catalysed by the heavier alkaline earth metal bisamides AEM (HMDS) 2 (AEM=Ca, Sr, Ba). The best performance was obtained with the largest metal, while ascending the group the reactivity decreased from barium to calcium. It sparked our interest that magnesium showed no activity at all. [14] Our first experiment was therefore to test the reduction of styrene to ethylbenzene with Mg(HMDS) 2 using 1.5 equivalents of 1,4-CHD as the hydrogen source. At 75°C and 10 mol % catalyst loading no reaction was observed (Table 1, entry 1). Repeating the experiment with additional 10 mol % of K(HMDS) resulted in complete conversion to ethylbenzene within 3.5 h ( Table 1, entry 2). To clarify whether a synergistic effect was in operation we tested K(HMDS) as the sole catalyst (Table 1, entry 3). Heating to 75°C immediately produced strong gas evolution. 1 H NMR spectroscopy revealed that hydrogen was developed and only 36 % of styrene was converted after 3 hours, although the majority of CHD was consumed. This modest yield is due to known competing side reactions (Scheme 2, eqn. 1). Thus, in situ generated potassium hydride can either insert into the olefinic double bond of styrene (desired-path A) or undergo two potential side reactions: It can either react with the prior formed protonated amine to generate hydrogen and starting material (side-path B) or dehydrogenate 1,4-CHD (side-path C). These side reactions appear less of a problem for the magnesiates, as the bimetallic compounds display high selectivity as well as high reactivity. The fact that both metal precursors on their own are not (or only partially) able to reduce the alkene catalytically indicates a synergistic effect in the bimetallic case.
To further investigate this phenomenon, we synthesised a series of magnesiates with the whole group of alkali metals (Li, Na, K, Rb, Cs). For the K, Rb and Cs congeners the synthetic procedure is similar to the synthesis of KMg(HMDS) 3 reported in 2002 by our group, [41] but significantly benzene was used as solvent instead of toluene (Scheme 2, eqn. 2). The advantage of benzene is that the products exhibit solubility at higher temperatures (e. g. 60°C), but nearly no solubility at room temperature. This facilitates product isolation after filtration giving the target tris-HMDS compounds in good yields as white solids. Working with isolated, rather than in situ generated, mixed-metal compounds should be advantageous since inac-Scheme 1. Various synthetic routes to s-block metal hydrides. curacies in stoichiometry and thus the unwanted side reactions (Scheme 2, eqn. 1, path B + C) may be diminished. We selected the reduction of styrene with 1.5 equivalents of CHD as our benchmark reaction (Scheme 2, eqn. 3) and started to vary the reaction conditions. Lowering the KMg(HMDS) 3 loading to 5 or 1 mol % resulted in prolonged reaction time and gave complete conversion at 75°C after 7 ( , indicating a negative effect of a polar donor solvent. Additionally, in THF oligomerization of styrene was observed lowering the yield of ethylbenzene to 15 %. A significant effect was exerted on the reactivity by changing the alkali metal. Using the smaller, less electropositive metals sodium and lithium gave after 16 hours only 4 % (entry 8) and no conversion at all (entry 9), respectively. A plausible factor is that a larger alkali metal interacts more with the π-system of the aromatic ring as was described for a series of monomeric alkali metal benzyl complexes by Robertson and co-workers. [42,43] This trend has also been observed for the catalytic intramolecular hydroalkoxylation reactions utilizing other alkali-metal magnesiates as catalysts. [44] As a result, the conjugated double bond is polarised and the insertion of the nucleophilic hydride becomes more favourable. Prompted by the reactivity enhancement seen on descending the first three alkali metals we eagerly investigated the behaviour of the two larger alkali metals rubidium and caesium. We observed that the trend is followed for the whole group. Thus, the rubidium magnesiate converted all styrene within 45 minutes (Table 1, entry 10). The enhanced reactivity comes along with a small but not negligible drawback: besides the high yield of 93 % ethylbenzene a small amount of oligomerization was observed. The best performance came when using the largest, most electropositive (non-radioactive) alkali metal. Full conversion with 10 or 5 mol % of CsMg(HMDS) 3 at 75°C was observed after just half an hour (Table 1, entries 11 and 12). Decreasing catalyst loading to 1 mol % led to an extension of reaction time to only 1 hour and 30 minutes (entry 13). The increased reactivity resulted again in a small decrease of yield (89-92 % ethylbenzene) alongside minor oligomerization. Contrary to our observations, Guan reported that the hydrogenation of styrene with KH and Mg(HMDS) 2 under 6 bar of H 2 pressure gave only 5 % of ethylbenzene since most of the substrate immediately undergoes polymerization. [35] Assuming that the active catalyst is in both cases a hydride complex, a reasonable explanation is that the protonation of the alkyl intermediate is facilitated with the prior formed HÀ HMDS (pK a � 26) [45] than with H 2 (pKa � 49) [46] and thus liberates the product prior to CÀ C bond formation. In all catalytic experiments using potassium, rubidium or caesium magnesiates only small amounts of H 2 were observed as side product, indicating that the hydrogenation (Scheme 2, eqn. 1, path A) is strongly favoured over the dehydrogenation (Scheme 2, eqn. 1, path B). A similar observation was made by Harder and co-workers with Ba(HMDS) 2 as catalyst. [14] To further investigate the influence of the alkali metal we added 18crown-6 to a catalytic reaction with KMg(HMDS) 3 (Table 1, entry 14). As crown ethers are well known to sequester alkali metals and form solvent-separated ion pairs we expected a decrease or even shutdown of reactivity, [47] though it is known that 18-crown-6 does not always completely fill the coordination sphere of potassium. [44] Surprisingly the substrate was fully consumed within only 30 minutes, however, only 17 % of the styrene was converted to ethylbenzene while the rest underwent oligomerization. A possible explanation is that after the addition of Lewis bases like THF or a crown ether smaller aggregates are formed resulting in an increased reactivity. [15] Again there is no sign of H 2 , underlining the higher selectivity of the ate complexes by suppressing the dehydrogenation. To circumvent the undesired side reaction, we changed the substrate to 1,1-diphenylethylene (DPE) and investigated its reduction with the larger alkali metal (K, Rb, Cs) catalysts. Here, the reactivity follows an opposite trend. For potassium, full conversion of the substrate was obtained after 1.5 hours (  Figure S20), which might be formed via Brønsted base catalysed CÀ C addition. The desired product, 1,1-diphenylethane can be deprotonated at the benzylic position and add to the double bond of DPE. An experiment with twice the amount of DPE gave full consumption of the substrate although an under stoichiometric amount of the hydride source (CHD) was present. HMDS based catalysts have been proven to be efficient catalysts in Brønsted base catalysed CÀ C addition in the past, [40] for example the Guan group reported benzylic CÀ H bond addition of diarylmethanes to styrenes catalysed by our KZn (HMDS) 2 (benzyl) system, [48] which is ultimately formed via benzylic cleavage of the methyl group of toluene by KZn (HMDS) 3 . [49] During the catalyses with AMMg(HMDS) 3 (AM=Na, K, Rb, Cs) further side products were observed: 1,3-cyclohexadiene was found as a result of isomerization of 1,4-cyclohexadiene. For RbMg(HMDS) 3 and CsMg(HMDS) 3 also cyclohexene was detected. On the positive side this indicates that the substrate scope might be extended to more challenging-to-hydrogenate alkenes, on the negative side reduction of CHD and substrate would compete with each other. Attempted reduction of 1hexene, a terminal, non-activated alkene, resulted mainly in the formation of H 2 and reduction of a small quantity of cyclohexadiene to cyclohexene (see Figure S23).
Intriguingly, during the reduction of styrene with NaMg(HMDS) 3 we could observe precipitation of crystalline material at the liquid/gas interface. The crystals could be identified via X-ray diffraction experiments as the hydride complex [{NaMg(HMDS) 2 H} 2 ] 1 , which is proposed to be an intermediate in the catalytic cycle. Prompted by this observation we tried to rationally synthesise this compound. Thus, we suspended NaMg(HMDS) 3 with 10 equivalents of 1,4-CHD in benzene. Heating at 60°C (without stirring and placing the flask into the oil bath where a quarter of the suspension is covered by oil) gave the desired compound after 2 days as colourless crystals in 72 % yield. With this result we have uniquely managed to isolate an intermediate in the catalytic transfer hydrogenation with CHD and thus give support to the proposed catalytic cycle of Harder et al.. [14] Next, we investigated whether this synthetic procedure is suitable for the other alkali metal magnesiates. Carrying out the reaction with KMg(HMDS) 3 was also successful (59 % yield). However, small amounts of H 2 were observed in the 1 H-NMR spectrum, indicating that the formed hydride complex [KMg(HMDS) 2 H] 2 can undergo the unwanted side reactions (Scheme 2, eqn. 1, path B + C) to H 2 and starting material. For larger alkali metal magnesiates this procedure seems to be unsuitable, since due to the higher reactivity of the hydride complexes [AMMg(HMDS) 2 H] 2 (AM=Rb, Cs) the side reactions (Scheme 2, eqn. 1, path B + C) becomes more dominant. Moreover, reduction of CHD to cyclohexene was observed in both cases, implying that the desired hydride complexes were consumed during the reaction (see Supporting information Figure S24). We therefore investigated the widely used ligand exchange reaction with a commonly utilised hydride donor source. Treatment with PhSiH 3 at 60°C in C 6 H 6 gave full consumption of RbMg(HMDS) 3 and CsMg(HMDS) 3 and formation of an insoluble white powder alongside ligand exchange products (e. g. PhSiH 2 (HMDS), PhSiH(HMDS) 2 ) as revealed by 1 H NMR spectroscopy. However, no evidence for a hydride compound of the form [(AM)Mg(HMDS) 2 H] 2 (AM=Rb, Cs) can be found in the 1 H NMR spectra. The powder, which is even insoluble in THF, reacts immediately with isopropanol under strong gas evolution and changes instantly a colourless solution of DPE in C 6 H 6 to dark red, which is likely to be an indicator for the presence of a reactive hydride component.  [32] Instead of a complex consisting of a sodium to hydride ratio of 1 : 1, a remarkable compound with six sodium and ten hydride moieties was obtained. Considering our sodium hydride complex [{NaMg(HMDS) 2 H} 2 ] 1 consists of a 1 : 1, sodium to hydride ratio, we want to emphasise the importance of the hydride source. In comparison to PhSiH 3 , where several ligand exchanges can occur, the usage of CHD provides a more controlled reaction.
The molecular structure of the potassium analogue [(C 6 H 6 ) KMg(HMDS) 2 H] 2 (note: there is additional C 6 H 6 of crystallisation present in the crystal lattice) is very similar to the one published by Hill and co-workers, [32] but there is one distinct difference: whereas the latter builds up to a polymeric, 2 dimensional framework via four anagostic C···K interactions between SiMe 3 groups and a potassium, in our version each dimer is connected via interactions of each potassium with co-coordinated benzene. [(C 6 H 6 )KMg(HMDS) 2 H] 2 shows poor solubility in C 6 D 6 but dissolves well in toluene-d 8 . The hydride signal appears at 3.86 ppm in the 1 H NMR spectrum and the HMDS groups were detected at 0.30 ppm ( 1 H NMR) and 7.21 ppm ( 13 C NMR).
To shed more light on the mechanism we carried out catalytic experiments with the isolated inverse crown hydride complex [(C 6 H 6 )KMg(HMDS) 2 H] 2 . Attempted reduction of styrene resulted in immediate precipitation of a white solid, which is the result of polymerization of the substrate. This observation highlights the importance of the HMDS group, since the acidic proton of the liberated HÀ HMDS enables protonation of the alkyl intermediate and prevents the chain growth. Changing the substrate to 1,1-DPE gave full conversion of the substrate after 3 hours in a quantitative yield (see Figure S25). We attribute the prolonged reaction time to the low solubility of the hydride complex in benzene. In Scheme 3 a catalytic cycle, adapted from the alkaline earth metal transfer hydrogenation from Harder, is proposed. The tris-HMDS complex deprotonates CHD under the liberation of HÀ HMDS and formation of an unstable Meisenheimer complex, which immediately undergoes ß-hydride elimination to form the catalytically active mixed metal hydride compound and benzene. The formation of the latter is proposed to be the driving force, since creating an aromatic compound is energetically favoured. The hydride complex then undergoes insertion into the double bond of styrene to form an intermediate heterobimetallic alkyl complex, which upon deprotonation of the previously liberated HÀ HMDS forms the desired product (ethylbenzene) and regenerates the catalyst again to complete the cycle.
In summary, we reported the first example of transfer hydrogenation of alkenes with heterobimetallic s-block complexes where the whole series of alkali metal magnesiates (LiÀ Cs) was investigated. A powerful synergistic effect is present between the alkali metal and magnesium as the monometallic compounds on their own are not or only partially capable of performing this transformation. Clearly the bimetallic partnership brings in advantageous features unavailable to using magnesium on its own: the alkali metal gives anionic activation to the magnesium as it forms a charged ate complex as opposed to a neutral species, and the divergence in the performance of the alkali metals intimates that the propensity for π-coordination favoured by the larger alkali metals can be another advantageous feature. This is consistent with the reduction of styrene being better with increasing size of the alkali metal. Since addition of 18-crown-6 boosted the reactivity for the potassium analogue but impeded the reaction for the rubidium and caesium compounds, other more subtle factors must also be involved. A novel synthetic procedure to heterobimetallic hydride complexes facilitated the isolation and characterization of complexes containing a sodium hydride or a potassium hydride moiety, which gave support for the catalytic cycle and for the first time the synthesis and molecular structure of a rubidium inverse crown incorporating a hydride group is reported. As the most potent catalysts were capable of reducing the hydride source CHD, we are currently investigating their performance in the hydrogenation of olefins with H 2 .

Experimental Section General Procedures
Hexane, THF, C 6 H 6 and toluene were dried by heating to reflux over sodium benzophenone ketyl and then distilled under nitrogen prior to use. C 6 D 6 , toluene-d 8 2 were collected on an Oxford Diffraction Gemini S instrument with graphite-monochromated CuÀ Kα (λ 1.54184 Å) radiation. All structures were solved and refined to convergence against F2 for all independent reflections by the fullmatrix least-squares method using SHELXL-2018 implemented within WINGX. [53,54] All non-hydrogen atoms were refined using anisotropic thermal parameters. Selected crystallographic data are shown in Tables S1, S2 and S3 and full details in cif format are available from CCDC (2058374-2058376). Data used within this publication can be accessed at https://doi.org/10.15129/60411ebe-e3a1-41f8-a6b3-947815e05b86.

Experimental Procedure and Product Characterisation
All reactions and manipulations were performed under a protective argon atmosphere using standard glovebox techniques, or under a protective nitrogen atmosphere using standard Schlenk techniques. Mg(HMDS) 2 was prepared by heating a commercially available heptane solution of MgnBu 2 with 2 molar equivalents of (HMDS)H for 2 h at 60°C. Rb(HMDS) and Cs(HMDS) was prepared according to the literature procedure. [55] The ate complexes AMMg(HMDS) 3 were prepared by a modified literature procedure where toluene was replaced by benzene as a solvent. [41]

Catalytic reactions
The catalyst (0.04 mmol for 10 mol %; 0.02 mmol for 5 mol %; 0.005 mmol for 1 mol % catalyst loading) was filled into a NMR J-Youngs tube and 500 μL of the desired solvent (C 6 D 6 or THF-d 8 ) was added. The alkene (0.4 mmol for 10 and 5 mol % catalyst loading; 0.5 mmol for 1 mol % catalyst loading),1.5 eqv. of 1,4-cyclohexadiene (0.6 mmol for 10 and 5 mol % catalyst loading; 0.75 mmol for 1 mol % catalyst loading) and a suitable standard (Si(SiMe 3 ) 4 or Adamantane 0.01 mmol) were added and the sealed NMR tube was heated in NMR spectrometer at the desired temperature for 16 h or until completion was observed (a spectrum was measured every 5 minutes).