Bifunctional activation of amine-boranes by the W/Pd bimetallic analogs of “frustrated Lewis pairs”

The reaction between basic [(PCP)Pd(H)] (PCP = 2,6-(CH2P(t-C4H9)2)2C6H4) and acidic [LWH(CO)3] (L = Cp (1a), Tp (1b); Cp = η5-cyclopentadienyl, Tp = κ3-hydridotris(pyrazolyl)borate) leads to the formation of bimolecular complexes [LW(CO)2(μ-CO)⋯Pd(PCP)] (4a, 4b), which catalyze amine-borane (Me2NHBH3, tBuNH2BH3) dehydrogenation. The combination of variable-temperature (1H, 31P{1H}, 11B NMR and IR) spectroscopies and computational (ωB97XD/def2-TZVP) studies reveal the formation of an η1-borane complex [(PCP)Pd(Me2NHBH3)]+[LW(CO3)]− (5) in the first step, where a BH bond strongly binds palladium and an amine group is hydrogen-bonded to tungsten. The subsequent intracomplex proton transfer is the rate-determining step, followed by an almost barrierless hydride transfer. Bimetallic species 4 are easily regenerated through hydrogen evolution in the reaction between two hydrides.


Introduction
Catalytic processes occurring under the action of "frustrated Lewis pairs" (FLPs) have been intensively sought aer during the past decade. 1,2 Heterolytic cleavage of H 2 by main group intermolecular FLPs has been proposed to occur through an 'encounter complex' where the Lewis acid and Lewis base are in close proximity, but non-bonding, and H 2 is accommodated between the two centers prior to heterolysis. DFT metadynamics studies for the prototypical FLP system P(Mes) 3 /B(C 6 F5) 3 gave evidence for the H 2 polarization followed by the ratedetermining hydride transfer to B and proton transfer to P. 3 The heterolytic cleavage of H 2 into a proton and a hydride is a crucial step in many chemical and biochemical processes such as, e.g., hydrogen oxidation by hydrogenases, transition metal-catalyzed hydrogenation of C]O bonds, or metalcatalyzed hydrogen oxidation in energy conversion reactions. From this point of view, the nomenclature of FLPs has many parallels in metal-ligand cooperation (bifunctional) catalysis, where the two sites of cooperation are typically metal-and ligand-based. 1,4,5 Closely related to this concept is also an area of bimetallic catalysis, wherein two metal sites demonstrate cooperativity in fundamental catalytic reactions. 6 The species featuring an H 2 molecule poised between the Lewis acid and Lewis base centers have been found by us in previous work as the intermediate 3 (1a); PCP ¼ 2,6-C 6 -H 3 (CH 2 P t Bu 2 ) 2 ) precedes the concerted proton and hydride transfer, 7 eventually yielding the bimetallic isocarbonyl bridging M/W bimetallic complex 4 (Scheme 1). 8,9 Up to date, there are only a few examples of the attempted use of such systems in catalysis. CpMoH(CO) 3 has been used as a proton source in Rh-catalyzed hydroformylation. 10 More recently, the switch of selectivity between migratory insertion versus C-H activation has been shown for the reaction of [( iPr POCOP)Ni] + ( iPr POCOP ¼ 2,6-C 6 H 3 (OP i Pr 2 ) 2 ) with phenylacetylene in the presence of [CpW(CO) 3 ] À presumably acting as a proton shuttle. 11 However, this reaction was performed only in a stoichiometric regime.
In the eld of chemical hydrogen storage, lightweight B/Ncontaining inorganic hydrides (ammonia borane, amineboranes, and hydrazine boranes) have gained increasing attention in the last few years as sources of carbon-free and high-purity H 2 for renewable energy applications. [12][13][14] Metal-catalyzed dehydrogenation became a well-developed route for H 2 production from amine-boranes 15 and new studies keep appearing in the literature aiming to optimise the reaction conditions and to control the selectivity in product distribution as dehydrogenative coupling not only produces H 2 a A.N. Nesmeyanov Institute of Organoelement Compounds, Russian Academy of Sciences (INEOS RAS), Vavilova Str. 28 21 The simultaneous presence of protic and hydridic H atoms within the same amine-borane molecule makes them   alternative reagents in transfer hydrogenation which attracts increased research interest. 22 It also triggers an extended dihydrogen bonding (DHB) throughout their solid-state structure. 23 DHB is preliminary to H 2 evolution occurring aer a simple thermal treatment or in a catalytic fashion. 15,24 Given the analogy between protic and hydridic H atom co-existence and heterolytic H 2 cleavage, the latter is intimately related to B/N inorganic hydride activation. Herein, we show for the rst time that the bimetallic systems reported above act as homogeneous amine-borane dehydrogenation catalysts and explore the impact of hydride and proton transfer steps in the reaction mechanism using 1a in comparison with its more acidic analog TpWH(CO) 3 (1b; Tp ¼ k 3 -hydridotris(pyrazolyl)borate) as proton donating hydrides (Scheme 1).

Pairwise interaction of two metal hydrides
The reaction of TpWH(CO) 3 (1b) with (PCP)PdH follows the same mechanism (Scheme 1) as that of CpWH(CO) 3 (1a). The IR spectra obtained for the mixture of 1b with a 1.5-fold excess of (PCP)PdH in pure THF at 200 K show two new n CO bands of the intermediate 3b at 1884 and 1754 cm À1 , which disappear gradually upon temperature increase (Scheme 1 and Fig. S1 †). At the same time, the bands at 1650, 1796, and 1901 cm À1 of the H 2 evolution productisocarbonyl complex 4bincrease in intensity. In a closed system, the reaction goes to equilibrium (Fig. 1). When the reaction mixture was allowed to reach equilibrium at, e.g., 240 K and then warmed to 290 K, the intensity decrease of the n CO bands of the ionic intermediate 3b and a slight increase of the n CO bands of the initial hydride 1b are observed. This indicates the shi of the proton transfer equilibrium 1 4 3 to the le. At the same time, the n CO intensity of complex 4b increases due to the hydrogen evolution and right shi of the 3 4 4 equilibrium (Fig. S2 †). 25 The IR monitoring of the reaction kinetics showed that the proton transfer is the ratedetermining step of the overall process and the rate constant values are larger for TpWH(CO) 3 (1b) than those with CpWH(CO) 3 (1a): at 240 K k obs ¼ 0.202 mol À1 s À1 for 1b and 0.140 mol À1 s À1 for 1a. This gives Gibbs free energy values of DG ‡ 298 K ¼ 16.5 AE 0.1 kcal mol À1 for 1b and 16.9 AE 0.1 kcal mol À1 for 1a. Complexes 4 reversibly capture molecular hydrogen similarly to FLPs. In a closed system, the reverse cooling from 290 K to 240-200 K restores the original spectroscopic picture. The crystal structures of both bimetallic complexes 4 ( Fig. 2) appeared to be very similar to those reported for the related Nicontaining complexes. 8,9,11 The larger steric volume of the Tp ligand leads to a different orientation of the [TpW(CO) 3 ] fragment: rotation around the Pd-O bond places terminal CO ligands on one side of the [(PCP)Pd] plane.

Interaction with dimethylamine-borane (DMAB)
Since the bifunctional ionic pairs [LW(CO) 2 (m-CO)/M(PCP)] (4) reversibly bind hydrogen, we hypothesised that they could interact with amine-boranes which are typical bifunctional molecules (RR 0 NHBH 3 ; R ¼ H, Me; R 0 ¼ H, Me, t Bu). We used substituted amine-boranes for these studies because they produce boron-containing reaction products and intermediates soluble in organic solvents. 26 Besides, Me 2 NHBH 3 usually reacts slower than ammonia borane (NH 3 BH 3 ). 27 The addition of excess Me 2 NHBH 3 (DMAB, 3.3 equiv.) to the bimetallic complexes 4 in THF at 270 K disrupts the inter-ion interaction due to the simultaneous coordination of DMAB molecules between the two metals. The decrease of n CO bands of 4 and the appearance of two new n CO bands of a new ionic complex 5 (Table 1) are observed in the IR spectra ( Fig. S3 †). These changes are reversible; the equilibrium shis toward 5 upon cooling. Changing the solvent to less polar toluene allows the reaction to move one step forward: the bands of LWH(CO) 3 (1) and (PCP)PdH (n PdH 1717 cm À1 ) appear in the IR spectrum ( Fig. S4 †), conrming the occurrence of proton and hydride transfer and the formation of neutral molecules. The hydride transfer from the boron atom to palladium is conrmed by the synchronous decrease of the DMAB band intensity (n BH 2364 cm À1 ) and the n PdH growth on going from 260 K to 190 K (Fig. S5 †). Warming the mixture from 190 to 260 K restores the original spectral picture, further conrming the existence of the equilibrium shown in Scheme 2. Denitely, the neutral trimolecular complex should be unfavorable due to the high entropy effect (see DFT calculations below), but it is conserved at low temperatures when TDS contribution to the Gibbs energy is diminished. The behavior of two ionic complexes 4a and 4b is qualitatively the same. However, the proton transfer from DMAB to 4a is easier; the tungsten hydride 1a is formed in a larger quantity, in agreement with the higher basicity of the [CpW(CO) 3 ] À anion: pK a of the conjugated acid LWH(CO) 3

Catalytic DMAB dehydrogenation
At ambient temperature in the presence of up to 50-fold excess of DMAB over 4, the catalytic dehydrogenation reaction occurs. To deeper understand the mechanism, the reaction was followed spectroscopically and volumetrically. IR monitoring shows the progressive decrease of n BH (2360-2260 cm À1 ) and n NH (3200 cm À1 ) bands of DMAB in the presence of 4a or 4b until their complete disappearance (Fig. S6 †). This observation provides strong conrmatory evidence for the cleavage of B-H and N-H bonds. A decrease of the typical DMAB signal (d B À15.3) and the accumulation of the dehydrogenation producti.e. the cyclic dimer (Me 2 NBH 2 ) 2 (d B 3.4), as well as the appearance of the steady-state intermediateaminoborane Me 2 N] BH 2 (d B 35.8, Fig. 3)was observed in the 11 B{ 1 H} NMR spectra, evidencing the off-metal dimerization mechanism. 30 Under these conditions, 1 H NMR spectroscopy shows the initial appearance of metal hydride resonances (d WH and d PdH , Fig. S7 and S8 †). Then, the tungsten hydride singlet (d WH À7.4 and À2.3 for 1a and 1b, respectively) disappears, but the palladium hydride triplet (d PdH À4.2) still accumulates. 31 P{ 1 H} NMR spectra show the corresponding growth of the (PCP)PdH  signal (d P 92.0), whereas the bimetallic complex (d P 75.6) is rst consumed and then restored (Fig. S9 †). Changes in the IR spectra show the same trend: the ionic complex 5 is formed at the expense of 4, whereas the n PdH band (1718 cm À1 ) increases in intensity ( Fig. S10 and S11 †). The bands (n CO , n PdH ) of 4, 5 and (PCP)PdH restore their initial intensity aer the complete conversion of DMAB (Fig. 4, top). An increase in DMAB loading increases the length of the quasi-stationary stage of the reaction when the concentration of both the catalyst and reaction intermediates is nearly constant (Fig. 4, bottom).
The reaction kinetics was measured following the decrease of n BH bond intensity of DMAB. The observed initial rate corresponds to the substrate disappearance (Fig. S6 †). In the presence of excess DMAB, the kinetic curves (c(DMAB) vs. t) are linear (Fig. S12 †). DMAB dehydrogenation occurs faster in the presence of the bimetallic complex 4a with the Cp-ligand: the initial reaction rate n 0 is 5.8 Â 10 À6 M s À1 for 4a and 1.7 Â 10 À6 M s À1 for 4b at 20 mol% catalyst loading. Keeping the substrate concentration constant, the increase of the catalyst 4a concentration from 0.003 M to 0.006 M leads to the doubling of the initial reaction rate. It should be noted that the kinetic curves obtained by the integration of proton NMR spectra have very similar behavior to the IR curves ( Fig. S13 †). According to the spectral data obtained, the bimetallic species [LW(CO) 2 (m-CO)/Pd(PCP)] (4) are regenerated at the end of the catalytic reaction and can be reused for the next substrate loading (Fig. 5). The initial reaction rates n 0 are similar when DMAB is added to the mixture of two neutral hydrides or to the pregenerated bimetallic complex 4, revealing that the reagents' mixing sequence does not affect the reaction rate ( Fig. S11 and S12 †).
The H 2 production in the reaction of DMAB with [CpW(CO) 2 (m-CO)/Pd(PCP)] (4a) in THF was monitored using the Man on the Moon X103 kit at ambient temperature and 313 K ( Fig. 5 and S14 †). At 313 K and 1 : 5 catalyst : DMAB ratio, complete DMAB conversion is achieved in 4 h ( Fig. 7 and Table  S3 †) with an initial reaction rate of 2.5 Â 10 À6 M s À1 . The increase of substrate loading to 50 equivalents at the same concentration of 4a (c ¼ 0.003 M, 2 mol%; T ¼ 313 K) leads to complete DMAB conversion in less than 3 hours (Fig. 7), and the initial reaction rate is 2.1 Â 10 À5 M s À1 . Under these conditions, the TOF value reaches 26 h À1 at a half-conversion time (Table  S3 †). The initial reaction rate at a catalyst : DMAB ratio of 1 : 50 is higher than that at 1 : 5, since the increase of DMAB amount at a constant catalyst concentration shis the equilibrium of complex 5 formation (Scheme 2 and Fig. 6).

Mechanism of R 0 RNHBH 3 dehydrogenation by bimetallic ion pairs
According to the experimental data, we suggest the reaction mechanism shown in Scheme 3 involving three important steps: (1) amine-borane molecule insertion between the two units of the bimetallic complex, (2) proton and hydride transfer, yielding aminoborane (Me 2 N]BH 2 ) and neutral metal hydrides, (3) dihydrogen release as a result of the reaction between palladium and tungsten hydrides via the DHB complex. The insertion of the amine-borane molecule into the bimetallic complex 4 yields the h 1 -borane complex 5 (Scheme   3). The latter is also an ionic pair with the tungsten anion [LW(CO) 3 ] À , which could explain the accumulation of LW(CO) 3 À in the reaction mixture (Fig. 4). In the next stage, the proton transfer from the NH group of coordinated amineborane to tungsten atom gives the neutral hydride 1 and the zwitterionic complex 6. The hydride transfer from the BH group to the palladium atom inside complex 6 leads to Me 2 N]BH 2 elimination and the formation of (PCP)PdH. To close the catalytic cycle, the two neutral metal hydrides react with each other, regenerating the catalytically active complex 4.

DFT calculations
This mechanism was supported by DFT calculations at the uB97XD/def2-TZVP theory level taking the real system for the structure optimization in toluene, which was introduced within the SMD model. Complex 5 is 7.7 kcal mol À1 above 4 on the Gibbs free energy scale (Fig. 8); however, its formation is possible (electronic energy difference DE(4-5) ¼ À7.5 kcal mol À1 ). The formation of 5 from 1 + Pd + DMAB is nearly ergoneutral (DG(1-5) ¼ +1.9 kcal mol À1 ) and thus this equilibrium can be effectively shied toward 5 in the presence of excess DMAB. The key bonds in complex 5 (  (Fig. S15 †). As a result, the hydride transfer goes faster than the proton transfer (k 3 [ k 2 ); therefore, complex 6 could not be observed experimentally. The ensemble of the reaction products ((PCP) PdH, LW(CO) 3 H and H 2 B]NR 2 ) is only slightly higher in free energy than the starting adduct 5 (+1.0 kcal mol À1 ), and the reaction is still reversible. However, since H 2 evolution from the two hydrides (k 1 , Scheme 3) 8,9,33 and the off-metal B]N dimerization 34,35 are featured with a comparable barrier, the overall reaction of dehydrogenative DMAB coupling becomes irreversible.

Catalytic TBAB dehydrogenation
Dehydrogenation of mono-substituted tert-butylamine-borane ( t BuNH 2 BH 3 , TBAB) was also studied using complex 4a as a catalyst. However, in this case, the IR spectroscopic picture and kinetic curves are not similar to those with DMAB ( Fig. S16 †). On the quasi-stationary stage, the concentration of neutral tungsten hydride 1a is higher, and that of the h 1 -borane complex 5a is lower than in DMAB dehydrogenation. per t BuNH 2 BH 3 molecule was also conrmed by volumetric measurements (Fig. S18 †).
To explain the different ratios of ionic and molecular forms of catalyst units in the reaction with DMAB and TBAB, we considered the behavior of the rst dehydrogenation product that accumulates in an unexpectedly high amount in the case of TBAB. Monomeric aminoborane H 2 B]NR 2 is potentially basic 36 and appears to be able to deprotonate the tungsten hydride (eqn (1)).
The isolated [H 2 B-NHR 2 ] + cation is not stable, but it can be stabilized by interacting with nucleophilic atoms like the carbonyl or the THF's oxygen atoms. Indeed, the neutral CpWH(CO) 3 /H 2 B]NMe 2 complex appears to be almost ergoneutral relative to the ion pair CpW(CO) 3 À /H 2 B-NMe 2 H + (DE ¼ À1.3 kcal mol À1 in favor of the neutral form, Fig. S19 †) stabilized by B/OC W interactions. Introducing a THF molecule into the system also allows stabilizing the ion pair via B/O THF interactions (DE ¼ +0.2 kcal mol À1 ; Fig. S19 †), whereas CO groups of the CpW(CO) 3 À anion appear unbounded, having a geometry similar to that of the CpW(CO) 3 À /R 3 NH + ion pair. 37 It should be noted that H 2 B]NH t Bu appears to be a weaker base than H 2 B]NMe 2 ; its neutral form is preferred by 3.8 kcal mol À1 . That suggests a lower deprotonation extent of LWH(CO) 3 when H 2 B]NH t Bu is formed (Fig. S19 and S20 †). Based on these computational data, we could expect the equilibrium side process of LW(CO) 3 H deprotonation, yielding the ion pair CpW(CO) 3 À /H 2 B-NMe 2 H + with CO/B interaction for DMAB dehydrogenation in toluene. When THF is used as a solvent, this ion pair is likely converted into the THF stabilized one. Consequently, the IR spectra in the n CO range in THF resemble those of the tungsten anion, while the IR spectra in toluene would rather resemble those of the ion pair 4 (Fig. S21 †). Thus, this side process (eqn (1)) should affect the solution composition at the quasi-stationary stage of DMAB dehydrogenation, increasing the relative amount of LW(CO) 3 À . Switching DMAB to TBAB diminishes the impact of LW(CO) 3 H deprotonation leading to the presence of both metals in hydridic forms in the reaction mixture (Fig. S16 †).

Reaction kinetics
Both DFT calculations and experimental data suggest that the NHbond cleavage is the rate-determining step of the catalytic reaction. Under these conditions, the overall reaction rate is determined by the rate of proton transfer in complex 5 (5 / 6) r 2 ¼ k 2 $ [5]. As shown by DFT calculations, the activation energy values for stepwise proton and hydride transfer (DG ‡ PT ¼ 21 kcal mol À1 and DG ‡ HT < 2 kcal mol À1 ) indicate that k 3 [ k 2 (Scheme 3). Thus, the hydride transfer goes much faster than proton transfer and conversion of 6 to products has no inuence on the reaction rate. When DMAB is in excess, the reaction is pseudo-zero order in DMAB. Taking into account the pre-kinetic step of DMAB coordination to bimetallic complex 4 in the presence of excess DMAB, the reaction rate is r ¼ k 2 Kc 0 (DMAB)$ [4] Thus, analysis of the experimental data gives the k 2 values of the rate-limiting step equal to 0.17 s À1 and 0.03 s À1 for 4a and 4b, respectively (Table S4 †). These values correspond to DG ‡ 298 K 18.5 AE 0.1 and 19.5 AE 0.1 kcal mol À1 , in reasonable agreement with DFT calculations. For TBAB dehydrogenation by complex 4a, the k 2 value is 0.10 s À1 (DG ‡ 298 K 18.8 AE 0.1 kcal mol À1 ), indicating faster proton transfer. The overall activation free energy DG ‡ 298 K for the conversion of 4 to 6 is ca. 25 kcal mol À1 that is in good agreement with DFT calculations (Fig. 8).
The use of deuterated amine-boranes NDMe 2 BH 3 and NDMe 2 BD 3 proves that the rate-limiting stage is the N-H bond cleavage, as is oen for AB dehydrogenation. 38 The rate constant k eff for NDMe 2 BH 3 dehydrogenation catalyzed by 4a (Fig. S22 and S23 †) is substantially lower (k ND eff ¼ 2.8 Â 10 À7 M s À1 ) than that for DMAB (k NH eff ¼ 5.8 Â 10 À6 M s À1 ) giving the kinetic isotope effect (KIE ¼ k NH /k ND ) of 20.6 AE 0.3. The use of the fully deuterated analogue NDMe 2 BD 3 does not lead to a further decrease in the reaction rate, giving KIE ¼ k H /k D of 20.6 AE 0.7. The KIE obtained is substantially higher than typical KIEs in metal-catalyzed amine-borane dehydrogenation. [39][40][41][42][43] This value also exceeds the KIEs for self-exchange in CpW(CO) 3 H/[CpW(CO) 3 ] À or for proton transfer of CpW(CO) 3 H to aniline. 44,45 Such high KIE values suggest a proton tunneling that is likely to occur when there is  a minimal geometry distortion along the reaction coordinate. 46 Our computations partially reproduce the observed magnitude of the isotope effect predicting increasing of the barrier upon NH to ND substitution by DDG ‡ 298 ¼ 1.5 kcal mol À1 (KIE ¼ k NH /k ND ¼ 13). One of the reasons for large KIE magnitudes (>10) is long (large acid-base separations) and strong H-bonded complexes. 47

Conclusions
In summary, bimetallic complexes [LW(CO) 2 (m-CO)/Pd(PCP)] (4) act as "metallic-analogs" of typical main group FLPs. The presence of acidic and basic metal centers in these ionic pairs triggers the cooperative BH/NH bond activation in amineboranes. In the rst reaction stage, the h 1 -borane complex [(PCP)Pd-(s 1 -HBH 2 -NR 2 H)/W(CO) 3 L] (5) is formed, in which BH is strongly bound to the palladium atom and the amine group is hydrogen-bonded to the tungsten atom. The step-wise proton transfer to W and hydride transfer to Pd yield the unsaturated B]N fragment and neutral metal hydrides. Molecular hydrogen evolution is the result of two metal hydrides interacting, regenerating bimetallic species 4. One equivalent of H 2 can be produced from Me 2 NHBH 3 , whereas dehydrogenation of t BuNH 2 BH 3 gives two equivalents of hydrogen per monomer at room temperature. The catalytic system can be easily generated through direct mixing of LWH(CO) 3 , (PCP)PdH and amine-borane, without preliminary synthesis of the ionic catalyst 4. The presence of ionic intermediates during the dehydrogenation cycle requires the use of polar solvents to achieve effective catalysis. The mechanistic study showed that proton transfer is the rate-determining step; therefore, the reaction is accelerated by a more basic anion ([CpW(CO) 3 ] À > [TpW(CO) 3 ] À ) while the hydride transfer is almost barrierless. The dehydrogenation process starts only in the presence of excess amine-borane due to the shi of the prekinetic equilibrium (H 3 BNHR 2 + 4 ¼ 5) to the right, which in turn causes the increase of the initial reaction rate. Yet another notable observation is that the dimethylaminoborane monomer H 2 B]NR 2 is able to deprotonate tungsten hydride in competition with H 2 evolution from two hydrides and H 2 B]NR 2 oligomerization. Different basicities of dehydrogenated DMAB and TBAB monomers lead to a different impact of this side process to the overall catalytic reaction.
Prior studies have shown that bimetallic systems featuring a metal-metal bond 48 can indeed operate in a concerted way activating H 2 , C-H, B-H, and other bonds. [49][50][51][52] These complexes activate the ONE bond, splitting it between two metals, oen in an oxidative addition fashion. In our case, two transition metalbased building units of a bimetallic complex do not interact directly but act cooperatively as a Lewis acid and a Lewis base, splitting the N-H and B-H bond without changing the metals' oxidation state. So far, such reactivity has been reported only for [Cp 2 ZrOC 6 H 4 P t Bu 2 ] + which can be described as an early transition-metal-containing linked FLP. 21 This behavior is similar to that of Stephan's main group FLPs 18 and can be exploited for other catalytic conversions; these studies are underway in our laboratories.

Experimental section
All reactions were performed using standard Schlenk procedures under a dry argon atmosphere. Commercial reagents (dimethylamine-borane, tert-butylamine-borane) were purchased from Aldrich and used aer preliminary sublimation. Tetrahydrofuran (THF) and toluene were dried over Na/ benzophenone and distilled under an argon atmosphere. THF-d 8 (Aldrich) was stored over 4Å molecular sieves and degassed before use by three freeze-pump-thaw cycles. NDMe 2 BH 3 and NDMe 2 BD 3 were prepared as described in the literature. 40,53 Variable-temperature (VT) NMR spectra were recorded on Bruker AVANCE II and Varian Inova FT-NMR spectrometers operating at 300 and 400 MHz in the 200-320 K temperature range. 1 H chemical shis are reported in parts per million (ppm) downeld of tetramethylsilane (TMS) and were calibrated against the residual resonance of the deuterated solvent, while 31 P{ 1 H} chemical shis were referenced to 85% H 3 PO 4 with downeld shi taken as positive. 11 B and 11 B { 1 H} were referenced to BF 3 $OEt 2 . IR spectra were recorded at different temperatures (190-293 K) using a home-modied cryostat (Carl Zeiss Jena) with a Nicolet 6700 spectrometer using 0.05-0.2 cm CaF 2 cells. The accuracy of the experimental temperature was AE0.5 C. The cryostat modication allows transferring the reagents (premixed at either low or room temperature) under an inert atmosphere directly into the cells.
Elemental analyses were carried out in the Laboratory of Microanalysis of INEOS RAS. The classic manual technique was used. The sample was burned in a platinum crucible in a stream of oxygen at 950 C followed by trapping CO 2 and water with Ascaris (asbestos impregnated with NaOH) and Anhydrone (anhydrous magnesium perchlorate), respectively, and the analysis of the mass changes. Solid LW(CO) 3 H (0.03 mmol) was placed in a Schlenk ask lled with argon (10 ml) together with 1 ml of tetrahydrofuran. Then, 1 ml of a (PCP)PdH THF solution (0.03 mmol of hydride in 2 ml of the solvent) was added to this solution. The resulting mixture of two colorless hydrides instantly became yellow colored. Aer an hour of stirring, the solvent was concentrated to ca. 0.1 ml under vacuum. Then, the Schlenk ask was lled with argon and le for several days standing at ambient temperature until a crystalline precipitate was obtained. The resulting solid (yellow needle-shaped crystals) was washed with a small amount of cold THF (2 Â 0.2 ml) and dried in a vacuum. Yield: 90%.
For variable temperature IR studies. Complex 4a or 4b was generated in situ by the reaction of (PCP)PdH (c ¼ 0.003-0.0039 M) with 1a or 1b (c ¼ 0.003 M), respectively. The reagents were dissolved in THF or toluene at 270 K and allowed to react for 30 min. Then, a chosen amount of DMAB was added, and the obtained mixture was monitored in the temperature range 190-290 K.
For kinetic IR studies. Reagents were prepared via three methods.
Method I. A portion of isolated complex 4a or 4b was dissolved in THF (c ¼ 0.003 M). Then, a chosen amount of DMAB (5 equiv.) was added and the IR spectra were monitored until full catalyst regeneration.
Method II. Complex 4a or 4b was generated in situ by the reaction of (PCP)PdH (c ¼ 0.003-0.0036 M) with 1a or 1b (c ¼ 0.003 M), respectively. The reagents were dissolved in THF at 290 K and allowed to react for 20 min. Then, a chosen amount of DMAB (1-5 eq.) was added, and the mixture obtained was monitored until full catalyst regeneration.
Volumetric studies of amine-borane dehydrogenation. Hydrogen evolution during dehydrogenation of amine-boranes was monitored using the Man on the Moon X103 kit. The volume of the system is 32 ml (two-necked round-bottom ask -30 ml, three-way valve -2 ml). The monitored solutions were prepared via three methods.
Method I. A portion of the isolated complex 4a or 4b (0.006 mmol) was dissolved in THF (1 ml) in an argon-lled ask of the device connected to a three-way valve. Then the ask was tightly closed with a septum cap, and the valve was opened to the pressure sensor. The chosen amount of DMAB (5-50 equiv.) in 1 ml of THF was injected with a syringe through a septum cap.
Method II. Complex 4a or 4b (c ¼ 0.003-0.01 M) was generated in situ by mixing the solutions of (PCP)PdH (1-1.2 eq.) and 1 (c ¼ 0.003-0.01 M, 1 eq.) in THF in an argon-lled ask of the device connected to a three-way valve. Then, the ask was tightly closed with a septum cap, and the valve was opened to the pressure sensor. A chosen amount of DMAB (5-50 eq.) in THF was injected with a syringe through a septum cap.
Method III. The solution of (PCP)PdH in THF (1-1.2 eq.) was prepared in an argon-lled ask of the device connected to a three-way valve. Then the ask was tightly closed with a septum cap, and the valve was opened to the pressure sensor.
The resulting mixture was stirred at 295 K. Data from a pressure sensor connected via a wireless network to a computer were recorded as a function of pressure versus time for 3-20 hours. The values accumulated were referenced by the pressure of THF in a blank experiment at 295 K and used for calculations of the H 2 equivalents evolved. The calculations were performed in the ideal gas approximation (pV ¼ nRT).

Computational details
Calculations were performed with the Gaussian 09 (ref. 54) package at the DFT/uB97XD 55 level without any ligand simpli-cation. For all atoms, the Def2-TZVP 56 basis set was applied, supplemented with an effective core potential 57 in the case of Pd and W. The structures of all complexes and transition states were fully optimized in toluene (3 r ¼ 2.3741) described by the SMD model, 58 without any symmetry restrictions. The nature of all the stationary points on the potential energy surfaces was conrmed by vibrational analysis. Transition state (TS) structures showed only one negative eigenvalue in their diagonalized force constant matrices, and their associated eigenvectors were conrmed to correspond to the motion along the reaction coordinate under consideration using the Intrinsic Reaction Coordinate (IRC) method. 59

X-ray crystallography
For crystals of 4a and 4b, X-ray diffraction data were collected at 120 K with a Bruker ApexII DUO diffractometer using graphite monochromatic Cu-Ka and Mo-Ka radiation, respectively. Using Olex2, 60 the structures were solved with the ShelXT 61 structure solution program using intrinsic phasing and rened with the XL 62 renement package using least-squares minimization. Hydrogen atom of the BH group in 4b was found in the difference Fourier synthesis while positions of other hydrogen atoms were calculated, and they all were rened in the isotropic approximation within the riding model. Crystallographic data and structure renement parameters are given in Table S1. † CCDC 2020560 and 2020559 contain the supplementary crystallographic data for 4a and 4b, respectively. †

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