Ligand Effects in Calcium Catalyzed Ketone Hydroboration

The first “naked” (Lewis base-free) cationic Ca amidinate complex [AmCa(C6H6)][B(C6F5)4] was prepared in 62 % yield {tBuAmDIPP = tBuC(N–DIPP)2; DIPP = 2,6diisopropylphenyl} by reaction of [AmCaH]2 with [Ph3C][B(C6F5)4] in chlorobenzene. The ether-free complex AmCaN(SiMe3)2 was obtained by removal of diethyl ether from its ether adduct. Crystal structures show that the amidinate ligand in both complexes is N,Aryl-chelating. In this coordination mode the bulk of the amidinate ligand is comparable to that of a DIPP-substituted -diketiminate ligand. Isomers with N,N-coordinating amidinate ligands are circa 15 kcal/mol higher in energy and this coordination mode is only present in case additional ether ligands compensate for energy loss or in


Introduction
Over the last decades, research on alkaline earth metal based homogeneous catalysis gained momentum and conquered fields, which were long thought to be the exclusive domain of transition metal catalysis. [1][2][3][4] Although often still not on par with their classical transition metal based counterparts, alkaline earth metals make up for lower catalytic activity of their complexes by price, availability and non-toxicity, at least in case of magnesium and calcium. Calcium catalyzed reactions include inter alia polymerizations, [5] alkene hydrogenations, [6] alkene and imine hydrosilylation, [7] intramolecular alkene hydroamination, [8] alkene and alkyne hydrophosphination, [9] hydroboration [10] or Mannich-type reactions. [11] Notwithstanding those successful applications, the conceptual foundation to predict whether a calcium catalyst is highly active for a certain reaction is so far unknown. This is due to the fact that the limited number of reports in calcium catalysis often describe results with drastically different catalysts. Nevertheless, possible factors to influence the reactivity of a catalyst in a given environment are in principle well known and include, but are not limited to, steric demand, charge, donor capacity and donor atom type of spectator ligands, their number and , catalyst performance increases with the Lewis acidity of the metal and a mechanism is proposed in which HBpin and ketone coordinate to the Ca 2+ ion which is followed by direct hydroboration. The more active catalysts with X -= (Me 3 Si) 2 Nor Hlikely operate through a mechanism which involves intermediate metal hydride (or borate) complexes.
the resulting coordination number of the metal ion, the nuclearity of the resulting complexes and the counterions present. In general, catalysts of type LCaR consist of a passive spectator L in combination with a reactive group R. The catalytic reaction is based on a combination of substrate activation by the Lewisacidic Ca 2+ center and the high nucleophilicity or basicity of R. In some cases, also catalysts that only rely on Lewis acid activation have been reported. [11,12] E.g. the Sen group introduced the amidinate calcium iodide catalyst I in the hydroboration of ketones and aldehydes. [12] It is unlikely that the highly stable iodide ligand actively takes part in catalysis. In an effort to address the importance of this second anionic ligand, we chose to study the hydroboration of ketones (and aldehydes) as a function of the ligands. We present here a series of catalysts with the bulky amidinate spectator ligand tBu Am DIPP { tBu Am DIPP = tBuC(N-DIPP) 2 ; DIPP = 2,6-diisopropylphenyl} that allowed for the synthesis and isolation of complexes tBu Am DIPP CaX with X -= I -(1), [B(C 6 F 5 ) 4 ] -(2), (Me 3 Si) 2 N -(3) or H -(4); see Scheme 1. The performance of these catalysts will be directly compared with results reported earlier by Sen and co-workers for PhC(NiPr) 2 CaI(THF) 3 {I, Ph Am iPr CaI(THF) 3 }. [12] While synthetic strategies for [ tBu Am DIPP CaI(thf ) 2 ] 2 (1), [13] tBu Am DIPP CaN(SiMe 3 ) 2 (Et 2 O) [14] and [ tBu Am DIPP CaH] 2 (4) [14] are known, synthetic routes had to be developed for [ tBu Am DIPP Ca(C 6 H 6 )] + [B(C 6 F 5 ) 4 ] - (2) and ether-free Scheme 1. Previously used calcium based catalyst for the hydroboration of ketones (I) and calcium catalysts used in this investigation (1)(2)(3)(4).
sen and co-workers [15] and previously used in our group for the stabilization of highly reactive calcium hydride complexes, [14] and for Ca derivatives featuring stilbene dianions [13] or novel anionic N-heterocyclic olefins. [16] This ligand choice might seem counterintuitive, since tBu Am DIPP has a higher steric demand and lower basicity of its nitrogen atoms when compared to Ph Am iPr , used by Sen and co-workers [12,17] {buried volume V B = 34.0 % in [ tBu Am DIPP CaI(THF) 2 ] 2 (1) vs. V B = 28.1 % in Ph Am iPr CaI(THF) 3 }, thus resulting in a less accessible and more electron deficient calcium center. However, its superior adaptability to the changing needs of a bound calcium ion, which is related to a facile interconversion of its N,N-and N,Aryl-coordination mode, [15,18] Table S1, Supporting Information). For the previously unpublished structure of ( tBu Am DIPP ) 2 Ca (see Figure 1), the so far highest value of V B = 42.4 % is found for one of the ligands, but the steric shielding provided by tBu Am DIPP in this compound is still significantly lower than for Me BDI DIPP in the above mentioned complexes. This changes drastically, when the ligand adopts a N,Aryl-coordination mode. In this conformation, the buried volume of the ligand ranges from 54.7 % in [ tBu Am DIPP Ca(NBO-H)] 2 [16] (NBO-H = deprotonated 1,3-dimethyl-2-methylene-2,3-dihydro-1H-imidazole) to 57.6 % in [ tBu Am DIPP Ca] 2 (SD) [13] (SD = stilbene dianion), making tBu Am DIPP competitive to Me BDI DIPP when it comes to steric demand (see Table S2, Supporting Information).
[b] See ref. [13] [c] See ref. [16] [14] in chlorobenzene led overnight to a slow color change from orange-red to brown. After removal of chlorobenzene, a brown foam was obtained which formed a biphasic system upon addition of benzene. The lower phase was washed with benzene until colorless crystals in a sticky brown residue grew. These crystals were suitable for X-ray analysis, but further purification was necessary. The crude product could be crystallized by thermal diffusion in a hexane/benzene (2:1) mixture in good yield (62 %) (see Figure S11, Supporting Information). XRD structure determination (see Figure 1, Table 1) revealed the retention of the (N,Aryl)-coordination mode of the starting material [ tBu Am DIPP CaH] 2 in [ tBu Am DIPP Ca(C 6 H 6 )] + [B(C 6 F 5 ) 4 ] -(2), as well as complete separation of cation and anion. This contrasts with the structure of the -diketiminate complex [ Me BDI DIPP Ca(C 6 H 6 )] + [B(C 6 F 5 ) 4 ]in which a Ca···F contact to the anion persisted. Similar cation-anion separation was earlier observed when Krossing's even weaker coordinating anion [Al{OC(CF 3 ) 3 } 4 ]was employed. [19] In related magnesium complexes, containing the B(C 6 F 5 ) 4 anion, it was necessary to further increase the steric bulk of the BDI ligand by an exchange of Me groups for tBu groups in the ligand backbone, to break the Mg···F interaction. [23] These findings indicate that the tBu Am DIPP ligand in N,Arylcoordination mode is at least as bulky as the Me BDI DIPP ligand with N,N-coordination. This assumption is supported by almost identical values for the volume buried by those ligands in the three-coordinate cations of [ tBu Am DIPP Ca(C 6 H 6 )] + [B(C 6 F 5 ) 4 ] -(2: 63.2 %) and [ Me BDI DIPP Ca(C 6 H 6 )] + [Al{OC(CF 3 ) 3 } 4 ] -(64.1 %).
The complete separation of cation and anion in 2 clearly leads to a much higher metal Lewis acidity and consequently shorter bonds to both, N2 and the aryl ring are observed, when compared to contact ion pair [ tBu Am DIPP Ca] 2 (SD). Expectedly, the effect is stronger for the negatively charged nitrogen {2: than for the η 6 -coordinated aryl ring (Ca-C av. 2.8018 Å vs. Ca-C av. 2.839 Å) (see Table 1).
Despite the very strong metal-ligand interaction in 2, exchange between coordinated and non-coordinated DIPP substituents is not prevented. While at ambient temperature, two distinct sets of 1 H NMR signals for the different DIPP moieties are observed (four doublets and two heptets for the iPr substituents), those signals show coalescence upon heating. The activation energy for fast exchange between the two different sides of the amidinate ligand has been estimated from the coalescence temperature of 337 K as ΔG ‡ = 16.1 kcal/mol. This value is in the same range as observed for [ tBu Am DIPP CaH] 2 (4, ΔG ‡ = 16.8 kcal/mol). [14] Dissolving complex 2 in [D 5 ]bromobenzene, led to loss of the coordinated benzene ligand and likely coordination of bromobenzene. This is evident from the benzene chemical shift of 7.21 ppm, which is the value of free benzene in this solvent.
Complex tBu Am DIPP CaN(SiMe 3 ) 2 (3) shows similar behavior. The flexible coordination mode of the tBu Am DIPP ligand in 3 is nicely illustrated by its synthesis from the corresponding diethyl ether adduct tBu Am DIPP CaN(SiMe 3 ) 2 (Et 2 O). The remarkably facile removal of ether in vacuo is accompanied by a change of the coordination mode from N,N to N,Aryl, as confirmed by XRD (see Figure 2). Structural features of 3 are similar to [ tBu Am DIPP Ca] 2 (SD), but the Ca···Aryl contact is somewhat longer (see Table 1). Exchange of the two inequivalent DIPP groups in solution has a significantly smaller activation barrier of ΔG ‡ = 14.3 kcal/mol (determined by NMR spectroscopy, see Supporting Information) than observed for 2, 4 and [ tBu Am DIPP Ca] 2 (SD).
The differences in coordination modes have been evaluated by DFT calculations (ωB97XD/def2tzvpp). Calculations on the cationic complex [ tBu Am DIPP Ca(C 6 H 6 )] + show that the N,Aryl coordination mode is favored over N,N-coordination by ΔH = 13.0 kcal/mol. In agreement with experiment a somewhat lower energy difference is found for tBu Am DIPP CaN(SiMe 3 ) 2 (3): ΔH = 10.3 kcal/mol. By comparison of the crystal structures of I, 1-4 and tBu Am DIPP CaN(SiMe 3 ) 2 (Et 2 O), it is clear that the less favorable N,N-coordination can only exist when coordinating solvents like Et 2 O or THF are present. In these cases, the switch from N,Aryl-to N,N-coordination creates free coordination sites and the energy needed for this process is compensated for by additional Ca···ether interaction.
(i) Complex [ tBu Am DIPP CaI(THF) 2 ] 2 (1), which exists as a iodobridged dimer in the solid state, already shows superior performance in comparison to Ph Am iPr CaI(THF) 3 , used by Sen and co-workers (compare entries 1-2, 6-7 and 11-12, Table 2). Since the additional THF ligands in 1 have no influence on the catalysis, because they are rapidly replaced by the ketone substrates and HBpin, which are present in large excess, the difference in reactivity can be solely attributed to the different amidinate spectator ligand. This could be due to difference in ligand bulk between tBu Am DIPP (V B = 34.0 %) and Ph Am iPr (V B = 28.1 %); both in N,N-coordination mode. Another difference is the fact that aryl-substituted N′s in tBu Am DIPP are much less electron-donating than the alkyl-substituted N′s in Ph Am iPr thus making the metal center in [ tBu Am DIPP CaI(THF) 2 ] 2 (1) more electrophilic.
(ii) In case high electrophilicity of the Ca center is needed for activity, the exchange of the iodide anion for B(C 6 F 5 ) 4 should further increase the performance of the system. Indeed, the activity of the cationic Ca complex 2 is for all substrates consistently higher than that of 1. Ketones with electron donating or withdrawing groups were rapidly consumed (> 94 % conversion) in presence of catalyst 2 and the desired hydroboration products formed even with very low catalyst loadings down to 0.05 mol-% (see Table 2). Similar to the calcium amidinate complex Ph Am iPr CaI(THF) 3 by Sen and co-workers, [12] the system showed a reasonable functional group tolerance. The high TOF's found for this system (see Table S3 N,N), a significant influence of the N,Aryl-coordination mode on catalysis is unlikely. It could be shown that addition of benzaldehyde (as a model substrate) led to a replacement of benzene in 2 and subsequently to a change of the initial N,Aryl-coordination mode to a symmetrical N,Ncoordination, when an excess of substrate is present, as it is during catalysis (see Figure S12, Supporting Information). (iii) The performance in catalysis of tBu Am DIPP CaN(SiMe 3 ) 2 (3) is again clearly higher than that of highly Lewis acidic 2 and an activity similar to that of previously investigated magnesium catalysts was found. For instance, the TOF of 600 h -1 in case of benzophenone, which often serves as benchmark substrate, is in the same order of magnitude as Hill's Me BDI DIPP MgBu (500 h -1 ), [24] Okuda's [Mg(THF) 6 ] 2+ [HBPh 3 ] -2 (1000 h -1 ) [25] or the phosphinoamido stabilized magnesium hydride used by Stasch (1760 h -1 per magnesium center). [26] The much lower TOF's of 8.6 h -1 for Ph Am iPr CaI(THF) 3   (iv) Our earlier reported Ca hydride complex [ tBu Am DIPP CaH] 2 (4) showed activities which are very similar to those of tBu Am DIPP CaN(SiMe 3 ) 2 (3). It is therefore likely that catalysts 3 and 4 operate through a metal hydride mechanism that is generally accepted for Mg catalysts of type LMgR (L = spectator ligand and R = active group). [3,24,26] The intermediacy of a hydride complex is obvious in case of 4 or Stasch's Mg hydride catalyst, where the hydride is already present, or in case of [ Me BDI DIPP MgBu], where the formation of [ Me BDI DIPP MgH] 2 upon reaction with HBpin was conclusively proven. [24] In case of Okuda's [Mg(thf ) 6 ][HBPh 3 ] 2 catalyst, transfer of a hydride from the boron center of the anion to magnesium (or directly to the substrate) seems feasible. [25] In [Mg(THF) 6 ][HB(C 6 F 5 ) 3 ] 2 , however, such transfer is impeded by the higher Lewis acidity of the boron center, which is likely the reason for the inferior catalytic activity of this system. [25] Calcium complexes containing a (Me 3 Si) 2 Ngroup are also known as excellent precursors for the formation of calcium hydride complexes, e.g. by reaction with PhSiH 3 , and it may be envisioned that the well-known complex [ tBu Am DIPP CaH] 2 forms under catalytic conditions as well.
Alternative to a hydride cycle is a pathway in which the hydride is not transferred from the metal to the ketone but directly from the borate (Scheme 3, far left). Indications that hydroboration not necessarily proceeds through the intermediacy of a metal hydride complex come from our group's previous studies of pyridine hydroboration. [27] This conclusion was based on differences in regioselectivity between stoichiometric metal hydride reactions and catalytic conversions.
Catalysts I, 1 and 2 do not contain active groups and it is a priori not clear how in this case intermediate hydride or borate species could be formed. Since the activity for this groups of catalysts increases with increasing Lewis acidity, we propose a mechanism in which the metal's Lewis acidity plays a central role. It could be envisioned that HBpin and the ketone both bind to the Ca 2+ metal center. Polarization of the C=O bond subsequently leads to hydride transfer and concomitant B-O bond formation. This direct B-H/C=O addition mechanism is Eur. J. Inorg. Chem. 2020, 1728-1735 www.eurjic.org similar to that proposed for catalyst-free ketone hydroboration. [28] Ketone hydroboration by Lewis acidic Ca complexes could best be interpreted by considering the Ca 2+ metal as a connector that brings both substrates in close vicinity. This compensates for the considerable entropy loss in ketone hydroboration.
Hydroboration of aldehydes was also briefly tested, but due the ease of this transformation and the resulting higher reaction rates, differences between the different catalysts are less pronounced (compare Supporting Information, Table S3).

Conclusion
We have prepared a series of Ca amidinate complexes with the amidinate ligand tBu Am DIPP . This bulky ligand is able to saturate the coordination sphere of large metal ions like Ca 2+ by N,Arylor N,N-chelation. N,N-coordination is typically observed when coordinating solvents are present, e.g. in [ tBu Am DIPP CaI(THF) 2 ] 2 (1) or tBu Am DIPP CaN(SiMe 3 ) 2 (Et 2 O), or when there is not enough space available for N,Aryl-coordination, e.g. in homoleptic ( tBu Am DIPP ) 2 Ca. The buried volume for the ligand with N,Arylcoordination is comparable to that of the widely known -diketiminate ligand Me BDI DIPP . Using tBu Am DIPP we achieved the isolation of the first "naked" (Lewis base-free) cationic Ca amidinate complex [ tBu Am DIPP Ca(C 6 H 6 )] + [B(C 6 F 5 ) 4 ] - (2) in which the ligand is bound by N,Aryl-chelation. This coordination mode was also found in ether-free tBu Am DIPP CaN(SiMe 3 ) 2 (3). In solution, the fast exchange between N,Aryl-and N,N-coordination modes is more facile in tBu Am DIPP CaN(SiMe 3 ) 2 (ΔG = 14.3 kcal/ mol) than in the cation [ tBu Am DIPP Ca(C 6 H 6 )] + (ΔG = 16.1 kcal/ mol). This likely originates from the higher Lewis acidity of the metal in the cationic complex.
We demonstrated that calcium complexes bearing the highly flexible amidinate ligand tBu Am DIPP are suitable catalysts for the hydroboration of ketones and aldehydes. Since catalysts 1-4 carry the same spectator ligand the influence of the second anionic ligand or counter anion could be evaluated. The anion or counterion Xin the catalysts tBu Am DIPP CaX significantly influences the performance of the system and activities increase along the series I -< B(C 6 F 5 ) 4 -< (Me 3 Si) 2 N -≈ H -. For the first two catalysts with Ior B(C 6 F 5 ) 4 -, catalyst activities increase with the Lewis acidity of the metal. However, compared to these catalysts, Ca complexes with X -= (Me 3 Si) 2 Nor Hare by far superior. They could be considered being competitive with previously reported magnesium based catalysts.
We propose two independently operating mechanisms. For tBu Am DIPP CaX complexes with an unreactive ligand X -, like Ior B(C 6 F 5 ) 4 -, a Lewis acid mechanism is likely while catalysts with a reactive ligand X -, like (Me 3 Si) 2 Nor Hmust operate through a mechanism which involves intermediate metal hydride (or borate) complexes.

Synthesis of [ tBu
Crystal Structure Determinations: Using Olex2, [30] the structure was solved by Intrinsic Phasing (ShelXT) [31] and refined with ShelXL [32] using Least Squares minimization. All non-hydrogen atoms were refined with anisotropic displacement parameters. All hydrogen atoms were placed in ideal positions and refined as riding atoms with relative isotropic displacement parameters. Crystal data and experimental methods can be found in the Supporting Information.  2 Ca} contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre.