Base-induced reversible H2 addition to a single Sn(ii) centre

A ‘frustrated Lewis pair’-type mechanism allows the first observation of reversible H2 addition to a single-site main-group complex.


Introduction
In the past decade there has been signicant interest in transition metal (TM) free systems which activate H 2 . 1 Two main strategies have emerged to facilitate this reactivity: the use of lowvalent main group (MG) compounds, 2 and so-called 'frustrated Lewis pairs' (FLPs). 3 In both cases, reactivity arises from simultaneously having access to a high-lying HOMO and low-lying LUMO (Fig. 1). Various low-valent MG compounds containing multiple E-E bonds (E ¼ Al, Si, Ga, Ge, Sn), 4,5 or single-site low-valent centres such as carbenes and heavier tetrylene analogues, have been shown to react with H 2 . 6 The scope of Lewis bases (LBs) and, to a lesser extent, Lewis acids (LAs), which can be used in H 2 -activating FLPs has expanded to include a number of elements from across the periodic table. This is principally due to the readily tuneable steric and electronic proles of the individual LA and LB sites. [7][8][9] Many FLP systems display reversible H 2 cleavage, which has facilitated their rapid expansion into the eld of catalytic hydrogenation. 10 The same is not true for low-valent MG compounds; examples of reversible H 2 activation are very rare and limited to antiaromatic boracycles, 11 a phosphorusbased singlet biradicaloid, 12 and only one low-valent group 14 compound: a dinuclear Sn(I) distannyne. 13 The design of singlesite MG systems which are ergoneutral for H 2 activation requires ne-tuning of thermodynamic (e.g. weak E-H bond strengths promoting an accessible formal E n+2 /E n couple) and kinetic factors, both of which are constrained to a mononuclear species, and is hence especially challenging.
The ability of L 2 Sn(II) compounds to undergo OA has been inversely correlated with the size of the singlet-triplet (HOMO-LUMO) gap, which may be diminished through the use of extremely strong s-donor ligands. Aldridge et al. have employed a bis(boryl)tin(II) system to achieve the only example of direct OA of H 2 to a mononuclear Sn(II) centre, irreversibly forming the Sn(IV) dihydride; boryl ligands are even stronger s-donors than hydride or alkyl ligands, permitting a successful reaction outcome. 6d Fig. 1 Representative orbital interactions between H 2 and main group compounds: (a) unsaturated E-E compounds e.g. distannynes (Ar ¼ C 6 H 2 -2,6-(C 6 H 3 -2,6-i Pr 2 ) 2 -4-X; X ¼ H, SiMe 3 , F); for X ¼ H, the reaction is reversible at 80 C; 5a,13 (b) single site low-valent centres e.g. carbenes; 6a (c) sterically hindered LAs and LBs (FLPs); (d) this work.
Conversely, the irreversible base-induced RE of H 2 from organostannanes is well-known. 14 Wesemann and others have studied RE from ArSnH 3 and [(Me 3 Si) 2 CH]SnH 3 compounds to yield various mononuclear Sn and Sn-Sn bound species (Ar ¼ terphenyl). 15 Nevertheless, there has yet to be a report of reversible OA and RE occurring on a single Sn(II) scaffold. Lappert's stannylene [(Me 3 Si) 2 CH] 2 Sn (1), which can act as both Lewis acid (LA) and base (LB), is a paradigmatic system for investigating OA to low-valent MG centres, yet to date its reactivity with H 2 has been unexplored. 16 Herein we report the use of FLP methodology to promote formal OA of H 2 to this simple dialkylstannylene. Furthermore we document the rst example of reversible H 2 addition to a single-site MG complex, which accesses an FLP via reversible dissociation of a classical 1$LB adduct; formation of the latter renders OA of H 2 to 1 energetically less favourable, enabling RE to occur from the Sn(IV) dihydride and reform 1, which is in equilibrium with 1$LB. 17

Results and discussion
1 is in a rapid solution-phase equilibrium with its dimer [1] 2 , which has been crystallographically characterised and contains a formal Sn]Sn double bond. 18 When a d 8 -toluene solution of 1/[1] 2 was placed under an atmosphere of H 2 (4 bar) in a sealed NMR tube, no change was observed in the 1 H NMR spectrum, even aer prolonged periods (>48 h), conrming that neither 1 nor [1] 2 can react with H 2 alone. Separately, addition of Et 3 N (20 mol%) to a solution of 1 resulted in no perturbation of their 1 H NMR resonances, suggesting no interaction between the components; i.e. the formation of an FLP. 19

Isotopic investigation
When D 2 was used in place of H 2 , the methine peak present in the 1 H NMR spectrum of the product mixture resolved as a singlet, while the Sn-H signal was absent and replaced by a Sn-D signal at d ¼ 5.11 ppm [ 1 J( 117 Sn-2 H) ¼ 262 Hz, 1 J( 119 Sn-2 H) ¼ 274 Hz] in the 2 H NMR spectrum. These results demonstrate the formation of dideuteride 2-D 2 , 20 and that the Sn-bound protons in 2 must originate from the hydrogen atmosphere.
In order to probe the mechanism further, a d 8 -toluene solution of 1/[1] 2 and Et 3 N was reacted with a 1 : 1 mixture of H 2 /D 2 . The resultant 1 H NMR spectrum was very similar in appearance to that of 2, with two exceptions: the relative integration of the Sn-H peak did not match that of the methine signal (1.2 : 2; consistent with the faster rate of reaction with H 2 vs. D 2vide infra), and the C-H resonance was composed of overlapping peaks commensurate with a mixture of 2 and 2-D 2 . No spectroscopic evidence was seen for the formation of 2-HD, which was independently and selectively obtained by analogous reaction of 1/[1] 2 under an HD atmosphere. These observations provide strong evidence that delivery of both atoms from H 2 /D 2 / HD to a single Sn centre occurs either simultaneously, or in a near-concerted fashion.

Kinetic analysis
By analogy with established FLP systems, and the microscopic reverse of the polar mechanism by which dehydrogenation of ArSnH 3 species is proposed to occur, 15a we envisaged a reaction mechanism in which 1 and Et 3 N form a weakly associated 'encounter complex' which subsequently reacts with H 2 (Scheme 1). 21 Assuming that encounter complex formation is a rapid pre-equilibrium prior to rate-limiting H 2 activation gives the expected rate law: rate Calorimetric studies on H 2 activation by the FLP Mes 3 P/ B(C 6 F 5 ) 3 (Mes ¼ 2,4,6-C 6 Me 3 H 2 ) found the rate to be very accurately modelled as a single, termolecular step, which formally gives the same rate law. 22 To conrm the order of catalytic Et 3 N, the method of time (t) scale normalisation was used; 23 normalisation to the scale of t$ [Et 3 N] x resulted in the superposition of all reactant traces only when x ¼ 1, conrming the rate to be rst order with respect to the amine (Fig. 3a). Determination of reaction order with respect to 1 requires its concentration to be known accurately at any given time in a reaction mixture. However, since the  conditions, simple observation of the concentration of 1 is not directly possible by 1 H NMR spectroscopy. 18a The concentration of 1 can, however, be calculated from the total concentration of "R 2 Sn" species in solution, [Tot], present as either monomer or dimer, which are related to the concentrations of 1 and [1] 2 by: (1) The dimerisation equilibrium of 1 can be expressed as: Combining eqn (1) and (2) and solving for [1] yields: Inserting eqn (3) into the expected rate law (vide supra) gives: where, if the amount of H 2 is sufficiently high that its concentration remains approximately constant: Rearrangement and integration by substitution of eqn (4) (see ESI †) gives: Therefore, plotting the variable portion of the LHS of this expression against t gives a straight line of gradient Àk*, con-rming the proposed rst-order dependence on 1 (Fig. 3b).
Scheme 1 Proposed reaction mechanism for H 2 heterolysis by 1, catalysed by Et 3 N.

Coordinating bases
When the less sterically bulky 1, 8-diazabicyclo[5.4.0]undec-7ene (DBU) is used, an interaction with 1 can be clearly seen in the 13 C{ 1 H} NMR spectrum: upon gradual addition of DBU to 1/ [1] 2 , the methine resonance undergoes a substantial upeld shi, reaching a limiting value of d ¼ 18.5 ppm (10-fold excess of DBU). Using the established 13 C NMR chemical shi values for 1 and [1] 2 (60.0 ppm and 28.7 ppm, respectively), 18a this is consistent with a fast equilibrium between 1$DBU, 1 and [1] 2 (Scheme 2; see ESI † for full details). A value of DG ¼ À3.7 AE 0.2 kcal mol À1 for the formation of 1$DBU from [1] 2 was obtained from a van't Hoff analysis of variable temperature UV-Vis spectra.
While the reaction of 1/DBU mixtures (containing 0.1-10 equivalents of DBU) with H 2 proceed rapidly, they do not reach completion, indicative of a reversible process (see Fig. S7 in ESI †).
The reversibility can be explicitly demonstrated by the (CH 3 ) 3 Si region of the 1 H NMR spectrum, whereby addition of DBU to a solution of 2 led to the appearance of a signal corresponding to the dehydrogenated mixture 1$DBU 4 1 4 [1] 2 ; this increased in intensity at the expense of the (CH 3 ) 3 Si peak of 2 (Fig. 4a-c). No H 2 is observed in the 1 H NMR spectrum as the solution was degassed multiple times in order to accelerate the reactionhowever, the very small amount of H 2 generated (approx. 0.3 bar) would likely hamper detection. Furthermore, the methine resonance of the 1$DBU 4 1 4 [1] 2 mixture is subject to a signicant upeld shi compared to [1]/[1] 2 (dependent upon the DBU concentration), and so is obscured beneath the relatively intense (CH 3 ) 3 Si region. Upon charging this reaction with H 2 , restoration of 2 was rapidly observed (Fig. 4d). For the equilibrium involving H 2 (Scheme 2), an equilibrium constant, K eq ¼ 164 AE 5, in favour of 2 can be calculated from the relative intensities of the (CH 3 ) 3 Si resonances, providing DG ¼ À3.0 kcal mol À1 (1 bar H 2 ).
Using the similarly unhindered but less basic 4-(dimethyl amino)pyridine (DMAP) also gave an adduct 1$DMAP, but no reaction with H 2 at room temperature. However, heating a solution of 1 with excess DMAP (4 bar H 2 , 2 h, 100 C) yielded 2 in 31% conversion.

Computational investigation
To gain further insight into the mechanism of H 2 activation, DFT calculations were performed for various 1/LB pairs; 31 the computed reaction proles for both the Et 3 N-and DBU-mediated reactions are depicted in Fig. 5. When LB ¼ Et 3 N, the reaction was found to proceed via initial H 2 heterolysis leading to a tight ion pair intermediate [1H] À [Et 3 NH] + (int 1 ). Facile rearrangement to int 2 and subsequent delivery of the H + to the lone pair on the [1H] À moiety furnishes 2 (Fig. 6a); a very similar mechanism was found when LB ¼ DBU. In support of this polar mechanism, the rate using Et 3 N as the LB was found to be faster in THF ðk 0 ðTHFÞ =k 0 ðtolueneÞ ¼ 1:97 AE 0:04Þ. The low Scheme 2 Equilibrium between product 2 + DBU and the dehydrogenated mixture 1$DBU 4 1 4 [1] 2 .  barriers to rearrangement of the intermediates also offer an explanation as to why H/D exchange is not observed upon reaction with an H 2 /D 2 mixture or HD: collapse of the ion pairs is likely much faster than solvent cage escape. Although the located transition states (TSs) are energetically close-lying, the overall reaction barrier appears to be determined by the H 2 splitting step, which is in line with kinetic measurements. Free energy data computed for the H 2 splitting step for reactions with different bases are compiled in Table 1 alongside other properties. For Et 3 N, TBTMG and PMP, no favourable adduct formation was found with 1, and the DG ‡ values follow the order TBTMG < Et 3 N < PMP, which is consistent with experimental reaction rates. For the coordinating bases DBU (Fig. 6b) and DMAP, adducts favourable relative to free [1] 2 and base were computationally determined. This reduces the absolute value of DG reaction such that an equilibrium is experimentally observed in the case of DBU. For DMAP, the activation barrier is found to be much higher, paralleling results seen by experiment where elevated temperatures are required to obtain product 2.
The energies of all intermediates int 1 are computed to be well above the reference state, which follows from the weak Lewis acidity of 1. The stabilities of int 1 species correlate very well with the general trend in PA and pK a , but this is not strictly true for the TSs, where steric factors are more important. Unstable int 1 intermediates imply late TSs for the H 2 activation step, which is shown by signicantly elongated H-H distances in the TS structures. The experimentally observed KIE (1.51 AE 0.04) supports this nding, which is commensurate with ratelimiting H 2 /D 2 activation involving considerable H-H/D-D bond breaking. 32

Conclusions
In conclusion, we have demonstrated the ability of FLPmediated reactivity to enable the formal oxidative addition of H 2 to an otherwise inert MG centre, and in doing so have also observed the rst example of reversible H 2 addition to a singlesite MG complex. We have utilised experimental and computational means to comprehensively explore the mechanism of this transformation and found that H 2 activation in this system differs from those based on more typical FLPs, due to the highenergy nature of the immediate H 2 splitting products, resulting in rare examples of late TSs. The development of methods to harness this FLP-promoted OA/RE H 2 reactivity for hydrogenation catalysis is currently underway.

Conflicts of interest
There are no conicts to declare.