Nature of hydrogen interactions with Ni(II) complexes containing cyclic phosphine ligands with pendant nitrogen bases

Abstract

Studies of the role of proton relays in molecular catalysts for the electrocatalytic production and oxidation of H₂ have been carried out. The electrochemical production of hydrogen from protonated DMF solutions catalyzed by [Ni(P₂^PhN₂^Ph)₂(CH₃CN)](BF₄)₂, 3a (where P₂^PhN₂^Ph is 1,3,5,7-tetraphenyl-1,5-diaza-3,7-diphosphacyclooctane), permits a limiting value of the H₂ production rate to be determined. The turnover frequency of 350 s⁻¹ establishes that the rate of H₂ production for the mononuclear nickel catalyst 3a is comparable to those observed for Ni-Fe hydrogenase enzymes. In the electrochemical oxidation of hydrogen catalyzed by [Ni(P₂^CyN₂^Bz)₂](BF₄)₂, 3b (where Cy is cyclohexyl and Bz is benzyl), the initial step is the reversible addition of hydrogen to 3b (K_eq = 190 atm⁻¹ at 25°C). The hydrogen addition product exists as three nearly isoenergetic isomers 4A–4C, which have been identified by a combination of one- and two-dimensional ¹H, ³¹P, and ¹⁵N NMR spectroscopies as Ni(0) complexes with a protonated amine in each cyclic ligand. The nature of the isomers, together with calculations, suggests a mode of hydrogen activation that involves a symmetrical interaction of a nickel dihydrogen ligand with two amine bases in the diphosphine ligands. Single deprotonation of 4 by an external base results in a rearrangement to [HNi(P₂^CyN₂^Bz)₂](BF₄), 5, and this reaction is reversed by the addition of a proton to the nickel hydride complex. The small energy differences associated with significantly different distributions in electron density and protons within these molecules may contribute to their high catalytic activity.

The catalytic interconversion of H₂ with two protons and two electrons plays an important role in the metabolism of various bacteria and algae, and it is important to the future development of hydrogen-based fuel cells and solar hydrogen production technologies. Recent structural studies of Fe-only and Ni-Fe hydrogenase enzymes have demonstrated that complexes of these relatively inexpensive and common metals can display high activities for this reaction (1–8). This has led to the hope that replacement of platinum in fuel cells for the hydrogen oxidation reaction could be achieved with simple synthetic catalysts based on iron or nickel, and many synthetic dinuclear iron complexes, developed as structural models for the enzyme active site, have been studied to explore their fundamental properties and catalytic potential (9–20). Structural features of the Fe-only hydrogenase catalytic site are depicted in structure 1 of Fig. 1, and a dihydrogen molecule is shown in the putative binding site on the distal iron atom. It has also been proposed that the central atom of the three atom backbone of the dithiolate bridge is a N atom, and that this amine plays a central role in the heterolytic cleavage reaction (1). In this process dihydrogen is split to form a hydride ligand coordinated to iron and a proton coordinated to the amine as shown in structure 2.

Fig. 1.

Structural features of the active site of Fe-only hydrogenases and the proposed activation of hydrogen.

The structural features of the hydrogenase active site and its proposed mechanism of operation suggest that the following considerations should be important in designing molecular catalysts for this reaction. (i) The heterolytic cleavage of H₂ should be at or near equilibrium to avoid high-energy intermediates. This implies the hydride (H⁻) acceptor ability of the metal and the proton (H⁺) acceptor ability of the base must be energetically matched to provide enough energy to drive the heterolytic cleavage of H₂, but this reaction should not be strongly exergonic. (ii) A pendant base should be incorporated into the second coordination sphere of the catalyst to serve as a proton relay to shuttle protons from the central metal to the exterior of the catalyst molecule. This can minimize reorganization energies associated with the approach of an external base for proton transfer. (iii) The nitrogen atom of the proton relay should be precisely positioned to assist the heterolytic cleavage of H₂ as shown in Fig. 1.

In previous studies we have attempted to sequentially incorporate each of these features into synthetic molecular catalysts for H₂ oxidation and production (21, 22). This resulted in the synthesis of nickel complexes with the general structure shown for 3 (Fig. 2). Depending on the substituents on nitrogen and phosphorus, these complexes are very active catalysts for either H₂ production {3a, [Ni(P₂^PhN₂^Ph)₂(CH₃CN)](BF₄)₂, R = R′ = phenyl} or H₂ oxidation {3b, [Ni(P₂^CyN₂^Bz)₂](BF₄)₂, R = cyclohexyl, R′ = benzyl} (22). These nickel-based synthetic catalysts allow a more detailed understanding of the relationship between structure and activity to be developed for this simple but important redox reaction.

Fig. 2.

Structure 3.

In this article, we examine additional aspects of both the catalytic hydrogen evolution reaction and the hydrogen oxidation cycle. In our previous report, our study of the rate of hydrogen production was limited by the instability of the catalyst at high concentrations of triflic acid. We now report kinetic studies of the catalytic reaction carried out in the presence of a milder acid in which the catalyst displays long-term stability. These studies permit a limiting value of the H₂ production rate to be determined and establish that the rate of H₂ production for the mononuclear nickel catalyst 3a is comparable to those observed for Ni-Fe hydrogenase enzymes. To obtain further insights into possible structures of the doubly protonated and doubly reduced intermediate formed during the catalytic production of H₂ by 3a, we examine the initial step in the catalytic oxidation of hydrogen and report our characterization of the isomeric products formed upon the addition of hydrogen to 3b. The structures illustrate a novel pathway for hydrogen oxidation/production facilitated by multiple pendant bases positioned near a redox-active metal center. Our results suggest that both the positioning and the number of pendant bases may be important for achieving high catalytic rates for H₂ oxidation and production for these nickel complexes.

Results

Rate of Electrocatalytic Hydrogen Formation.

In previous studies we have shown that 3a catalyzes H₂ production from triflic acid, CF₃SO₃H, in acetonitrile solutions (22). However, at triflic acid concentrations above ≈0.1 M, the complex is unstable. Dissociation of the protonated diphosphine ligands is a likely mode of decomposition under these strongly acidic conditions. This behavior prevented an accurate assessment of the intrinsic turnover frequency for 3a, but a value of 130 sec⁻¹ could be observed for a 0.1 M acid concentration. To achieve the optimum rate and performance of an electrocatalyst for proton reduction, the acid strength should be selected to match the pK_a of the catalyst. Although the pK_a of the phenylamine base in 3a has not been measured directly, the pK_a value can be estimated to be ≈7 in acetonitrile on the basis of thermodynamic studies of related complexes (K. Fraze, A.D.W., M.R.D., and D.L.D., unpublished data). Protonated dimethylformamide in acetonitrile (pK_a = 6.1) (23) is a significantly weaker acid than triflic acid in the same solvent (pK_a = 2.6) (23), and provides a good match in pK_a values with catalyst 3a. Complex 3a was found to be quite stable at high concentrations of H⁺-DMF/DMF in acetonitrile, where the pH approaches the pK_a value. Less than 10% decomposition of 3a has been observed by ¹H and ³¹P NMR spectroscopy after 10 days in a mixture of 0.62 M H⁺-DMF and 0.62 M DMF in acetonitrile-d₃.

The catalytic activity of 3a using a buffer solution of protonated dimethylformamide/dimethylformamide (H⁺-DMF/DMF, 1:1) in acetonitrile was studied by cyclic voltammetry. Fig. 3 shows a series of cyclic voltammograms recorded at increasing H⁺-DMF concentrations. The peak current (i_p) associated with the Ni(II/I) couple in the absence of acid (not resolved in initial scan at scale shown here) is much smaller than the catalytic current (i_c) measured in the presence of acid. A plot of the i_c/i_p ratio vs. acid concentration is shown in the inset of Fig. 3. It can be seen that as the acid concentration increases, the catalytic current initially increases and then becomes independent of acid concentration. The linear region observed at low acid concentrations indicates a second order process in acid, and the acid independent region indicates saturation with acid and a rate-limiting step such as H₂ elimination or an intramolecular proton transfer. As described previously (22), the i_c/i_p ratio can be used to calculate a turnover frequency (24–27). For the acid independent region of the plot, a limiting value of 350 s⁻¹ has been determined for this system at 22°C.

Fig. 3.

Cyclic voltammograms of a 3.2 × 10⁻⁴ M solution of [Ni(P^Ph₂N^Ph₂)₂(CH₃CN)](BF₄)₂, 3a, with increasing concentrations of H⁺-DMF(OTF)/DMF (1:1) in acetonitrile. Conditions were as follows: scan rate of 50 mV/s, acetonitrile solvent, 0.3 M NEt₄BF₄ as supporting electrolyte, glassy carbon working electrode. Inset shows values of i_c/i_p vs. the concentration of the buffer H⁺-DMF(OTF)/DMF in acetonitrile.

Characterization of H₂·[Ni(P^Cy₂N^Bz₂)₂](BF₄)₂.

The preceding studies are consistent with H₂ elimination from the catalyst as the rate-determining step in the production of H₂, but intermediates before hydrogen elimination were not directly observable by spectroscopic methods. To obtain more information on the nature of the intermediate steps, we turned to the study of intermediates observed for the reverse reaction, H₂ oxidation. We have recently reported that the Ni(II) complex [Ni(P^Cy₂N^Bz₂)₂](BF₄)₂, 3b, serves as a catalyst for the electrochemical oxidation of hydrogen in the presence of a base (22). The proposed catalytic cycle for this complex together with intermediate species to be discussed in this article are summarized in Scheme 1.

Scheme 1.

When one atmosphere of hydrogen is added to an acetone or acetonitrile solution of 3b in the absence of an external base, spectroscopic data support the formation of a hydrogen addition product, 4. The hydrogen addition is reversible and an equilibrium constant of 190 ± 20 atm⁻¹ at 21°C in acetonitrile has been measured (22).

The ³¹P NMR spectrum of the hydrogen addition product, 4, indicates that a mixture of isomers is present, but the positions of the added hydrogens were not initially determined. Although the product isomers are not stable enough for isolation, it has been possible in further studies of this system to obtain detailed information about the nature of the isomers by heteronuclear and two-dimensional NMR techniques.

The ³¹P NMR spectrum of the hydrogen addition product in acetone-d₆, shown in Fig. 4a, shows both broad and sharp resonances. When this NMR solution is cooled to −70°C, the peaks become sharper, Fig. 4b, and resonances for several isomers are observed. The spectrum includes two AB patterns centered at 19.5 and −7.5 ppm and two singlets at 16.6 and −10.2 ppm. In the ¹H NMR spectrum of this mixture recorded at −70°C, no nickel hydride resonances are observed at characteristic upfield shifts. Although the resonances for the P₂N₂ ligands obscure many regions of the proton spectrum, new broad singlets are observed at 6.8 ppm and near 15 ppm.

Fig. 4.

400-MHz ³¹P NMR spectra of 4 formed at room temperature in acetone-d₆ (a) and the same sample solution cooled to −80°C (b).

Synthesis and Characterization of a Single Isomer, 4A.

Further interpretation of these data is facilitated by the observation that when an acetone-d₆ solution of 3b is first cooled to −70°C and then reacted with hydrogen, as described in the Experimental, only one isomer of the hydrogen addition product is formed. This isomer 4A displays the AB pattern in the ³¹P NMR spectrum at 22.4 and 17.2 ppm with J_P-P = 34 Hz. Simulation of the spectrum has established that the pattern is actually consistent with the expected AA′BB′ spin system and a listing of coupling constants is given in supporting information (SI) Table 1. Once again no nickel hydride resonances are detected in the upfield region of the ¹H NMR spectrum of 4A, but the broad singlet at 6.8 ppm is observed in this spectrum.

Additional experiments in which HD and D₂ are added to precooled solutions of 3a at −70°C provide more information on this system. The addition of deuterium resulted in the formation of the single isomer 4A(D₂). The ³¹P NMR shows a similar AB pattern as 4A(H₂), but the two doublets are shifted to slightly lower chemical shifts of 21.8 and 16.9 ppm (see Fig. 5a for H₂ and 5b for D₂). When HD is used instead of D₂, the ³¹P spectrum of the product shows two AB patterns for phosphorus chemical shifts influenced by both hydrogen and deuterium (Fig. 5c). The observation of two AB patterns indicates that rapid HD exchange is not occurring in this system at −70°C, as this should lead to one averaged AB spectrum. Consequently, the absence of a nickel hydride signal in the ¹H NMR does not appear to be a result of Ni-H/N-H exchange. The resonance at 6.8 ppm in the ¹H NMR spectrum of 4A(H₂) decreases to about half its intensity in the spectrum of 4A(HD) and disappears in the spectrum of 4A(D₂) (SI Fig. 8 a–c). The chemical shift of this resonance is consistent with an NH functional group.

Fig. 5.

³¹P NMR spectra of 4A(H₂) (a), 4A(D₂) (b), and 4A(HD) (c) recorded at 400 MHz in acetone-d₆ at −70°C.

On the basis of these data, 4A is proposed to be a tetrahedral Ni(0) complex with a protonated amine in each cyclic ligand, as shown in Scheme 1. The proposed structure is symmetric along a C2 axis bisecting the P^B-Ni-P^B′ and P^A-Ni-P^A′ angles and is consistent with both the ³¹P and ¹H NMR data. To provide additional support for this structure, the ¹⁵N labeled nickel complex was synthesized by using a labeled form of the ligand P^Cy₂N^Bz₂ prepared from ¹⁵N-benzylamine. The ¹⁵N NMR spectrum of the hydrogen addition product formed at −70°C shows two resonances of about equal intensity at 74.7 and 43.5 ppm. The chemical shifts are consistent with assignment of the resonances to quaternary ammonium and tertiary amine nitrogens, respectively (28). A two-dimensional ¹H/¹⁵N NMR [gradient heteronuclear single quantum coherence (gHSQC)] experiment (SI Fig. 9) established that the resonance at 74.7 ppm is coupled to a proton resonance near 6.8 ppm, and in the ¹H NMR spectrum, this resonance is observed as a doublet with ¹J_NH = 71.6 Hz, SI Fig. 8d. No proton coupling was observed in the gradient heteronuclear single quantum coherence spectrum for the ¹⁵N resonance at 43.5 ppm. The combined spectroscopic data are all consistent with the proposed structure of 4A. It is possible that in this structure the Ni(0) center is further stabilized by N-H–Ni(0) interactions. We have no direct spectroscopic evidence for this, but the relative ³¹P NMR chemical shifts are suggestive of such interactions, as discussed in the next section.

Characterization of Additional Isomers.

As the solution of 4A is slowly warmed above −70°C, resonances for additional isomers are observed in the ³¹P NMR spectrum. The isomers are all similar in energy and are also assigned as diprotonated Ni(0) complexes, related to 4A by proton transfers, conformational changes of the ligand chelate rings, and/or inversions at the amine nitrogens. The heteronuclear NMR data, discussed below, provide support for the structures shown below for 4B and 4C in Fig. 6.

Fig. 6.

Proposed structures for isomers 4B and 4C.

As the solution is warmed, in addition to the resonances of 4A, the AB pattern near −7 ppm and the singlet at 17.4 ppm are also observed in the ³¹P NMR spectrum. In acetone-d₆ the integrations for these two resonances are very close to 1:1, suggesting that they correspond to a single compound, 4B. This conclusion is also supported by the fact that the relative integrations for these two resonances remained relatively constant when hydrogen addition was carried out in acetonitrile, dichloromethane, and dimethylformamide at room temperature, whereas ratios of isomers 4A and 4C varied significantly. The large difference in the ³¹P NMR chemical shifts of the two ligands in isomer 4B provides support for a N-H–Ni interaction involving the ligand chelate in the boat conformation. Such an interaction results in the formation of two five-membered rings, and such ring systems are known to result in significant downfield shifts of the ³¹P NMR resonances compared with those in six-membered ring systems (29).

The ¹⁵N NMR spectrum for 4B in acetone-d₆ shows two resonances that are very close in chemical shifts to those of 4A, as summarized in SI Table 2. In addition, a third ¹⁵N resonance is observed at 47.0 ppm. In the ¹H NMR spectrum at −70°C, singlets at 7.0 and 14.6 ppm are assigned to the N-H protons in the two inequivalent ligands. For the ¹⁵N-labeled complex, the splitting of the proton resonance near 7 ppm is obscured by phenyl resonances, but the downfield resonance is split into a triplet at 15.4 ppm with a coupling constant of 31 Hz. The coupling between the ¹H and ¹⁵N resonances is confirmed by the gradient heteronuclear single quantum coherence spectrum. Similar downfield ¹H chemical shifts and J_N-H values have been observed previously for the proton stabilized by two amines in protonated Proton Sponge [1,8-bis(dimethylamino)naphthalene] and related derivatives (30).

The last isomer to form as the solution is warmed to room temperature is 4C, which shows a singlet in the ³¹P NMR spectrum at −9.2 ppm. The ¹H and ¹⁵N NMR data, summarized in SI Table 2, are consistent with a tetrahedral structure with D_2d symmetry in which each proton is stabilized by two amines of the chelate rings in chair conformations as shown in Fig. 6.

The broad resonances observed at 19.5 and −7.5 ppm in the room temperature ³¹P NMR spectrum shown in Fig. 4a indicate that exchange processes are occurring for isomers 4A and 4B. In addition, the coupling of a proton to two ¹⁵N nuclei in isomers 4B and 4C may indicate that each NH proton is exchanging rapidly between a single pair of ¹⁵N atoms, or it could indicate structures in which the proton symmetrically bridges the two ¹⁵N atoms. Although further information regarding the nature and mechanism of these exchange processes is certainly of interest, it is beyond the scope of the present work.

Deprotonation of 4.

The isomers of 4 are rapidly converted to the Ni(II) hydride [HNi(P^Cy₂N^Bz₂)₂](BF₄), 5, upon deprotonation with triethylamine in acetonitrile, as shown by the bottom reaction in Scheme 1. The deprotonation reaction is reversible, and the hydride complex 5 is converted to the doubly protonated Ni(0) derivatives, 4A–4C, upon reaction with tetrafluoroboric acid or anisidinium tetrafluoroborate. This equilibrium represents an unusual pH-dependent intramolecular redox event in which protonation leads to a formal two-electron reduction of the metal while deprotonation results in a two-electron metal oxidation. We propose that protonation of one ligand leads to a significant decrease in electron density at nickel. As a result the nickel hydride becomes more acidic and the hydrogen is transferred as a proton to an adjacent base, forming the isomers of the Ni(0) product. Consistent with this interpretation is the observation that the cyclic voltammogram for 4 in acetonitrile shows a significant anodic shift in potentials relative to 3b. The cyclic voltammogram for the isomers of 4 shows two irreversible waves with peak potentials at 0.0 and −0.4 V vs. ferrocene, whereas the reduction potentials for the Ni(II/I) and (I/0) couples for 3b are observed at −0.80 and −1.28 V (22). Although the waves for 4 are irreversible, they indicate that large changes in electron density occur at nickel as a result of protonating the nitrogen atoms of the diphosphine ligands.

Discussion

Complex 3a is an exceptionally effective catalyst for the electrochemical production of hydrogen, displaying high rates and long lifetimes. The turnover frequency of 350 s⁻¹, determined at 22°C, is comparable to the catalytic rates of 500–700 s⁻¹ reported for H₂ production at 30°C for Ni-Fe hydrogenases (31, 32). At the higher acid concentrations shown in Fig. 3, H₂ elimination or an intramolecular proton transfer appears to be the rate-limiting step in the catalytic cycle. Similarly, hydrogen addition appears to be the rate-determining step in the catalytic oxidation of H₂ by 3b. In the latter case, oxidation of H₂ occurs readily within the coordination sphere of catalyst 3b to give an isomeric series of Ni(0) products containing a protonated nitrogen atom in each of the cyclic ligands. The nature of the first observable intermediate, 4A, presents a surprising contrast to that observed previously for [Ni(PNP)₂]²⁺, 6, where PNP is Et₂PCH₂N(Me)CH₂PEt₂ (21). In the latter case, an intramolecular heterolytic cleavage of H₂ occurs when 6 reacts with H₂ to form [HNi(PNHP)(PNP)]²⁺, 7, as shown in Eq. 1.

The differences in structures 4 and 7 demonstrate that seemingly subtle differences in the diphosphine ligand structure that involve positioning of the nitrogen bases can result in significantly different electron distributions during hydrogen addition.

In our spectroscopic studies that led to the identification of 4 and 7, no evidence for the initial interaction with hydrogen to form a nickel-dihydrogen intermediate has been obtained for either system. However, DFT calculations on model complexes of both [Ni(PNP)₂]²⁺ and [Ni(P₂^RN₂^R′)₂]²⁺ derivatives (where phosphine and nitrogen substituents are replaced with hydrogens) have indicated that an initial intermediate is a Ni(II) dihydrogen complex. In the case of the PNP complex, the dihydrogen ligand is somewhat unsymmetrically coordinated with one hydrogen showing a closer approach to the pendant amine of the ligand (22). In the case of the P₂N₂ complex, a completely symmetrical dihydrogen complex, as shown in Scheme 1, is observed computationally as the initial intermediate. This dihydrogen intermediate lies 2.1 kcal/mol above the energy of the reactants, whereas the corresponding dihydride lies 15.0 kcal/mol higher in energy. The Ni(0) complex with two protonated nitrogen atoms analogous to 4 has a calculated energy of −2.5 kcal/mol with respect to the reactants in comparison with an experimental free energy of −3.1 kcal/mol observed for H₂ addition to 3b.

The heterolytic cleavage of H₂ is generally thought to occur by an asymmetric polarization of a dihydrogen molecule to form an H⁺ and an H⁻ species. The heterolytic cleavage of H₂ by 6 to form 7 shown in Eq. 1 is an example of such a reaction. A transition state for this reaction has been calculated [structure 8 (Fig. 7)] in which the H-H bond of the coordinated dihydrogen ligand is weakened as the new bonds with the proton and hydride acceptor sites are formed (22). The observation that 4A is the first detectable intermediate in the addition of H₂ to 3b is interesting because it suggests that a distinctly different mechanism for H₂ activation may occur for this complex. A novel feature of 3b is that two amine nitrogens are held in positions that could allow for simultaneous interaction of both bases with the incoming H₂ molecule, as shown by the proposed dihydrogen intermediate in Scheme 1. If such a symmetric transition state occurs, then the heterolytic cleavage achieved by 3b would not involve a H₂ molecule that is strongly polarized as H⁺H⁻. Instead the polarization would occur by the removal of two electrons by nickel from a symmetric transition state in which both of the hydrogen atoms gradually develop a more positive charge. Similarly, the production of H₂ by complex 3a may involve the reverse process. The proposed symmetric mechanism of H₂ oxidation/production implies that cooperative interactions of the dihydrogen ligand with both the metal center and multiple proton relays incorporated in the second coordination sphere contribute to the high activity observed for these nickel-based molecular catalysts.

Fig. 7.

Structure 8.

Summary.

Complex 3a is an extremely effective catalyst for the electrochemical production of hydrogen with a turnover frequency (350 s⁻¹) that is comparable to that of the nickel-iron hydrogenase enzymes. Kinetic studies indicate that the rate-determining step involves a doubly protonated and doubly reduced intermediate. To provide insight into possible structures of this intermediate, H₂ addition to 3b has been studied. Complex 3b oxidizes H₂ to form a mixture of Ni(0) products in which an amine in each of the cyclic ligands is protonated. Three nearly isoenergetic Ni(0) isomers, 4A–4C, are suggested on the basis of heteronuclear and two-dimensional NMR techniques. The structure of the first intermediate 4A suggests that both of the pendant nitrogen atoms in 3b may simultaneously interact in a symmetric manner with H₂. This may be important for stabilizing the dihydrogen intermediate and for the subsequent cleaving of H₂ with a concerted reduction of Ni(II) to Ni(0). The contrast between these products and the closely related PNP structure 7, which contains both a nickel hydride and a protonated nitrogen atom, demonstrates that subtle differences in the diphosphine ligand structure can result in very different proton and electron distributions. The cooperative interactions that lead to a delicate balance in the distribution of protons and electrons between nickel and the pendant bases of the second coordination sphere are believed to be important factors in the high catalytic activity of these systems.

Materials and Methods

Complexes 3a and 3b were synthesized according to a published procedure (22). ¹⁵N (95%) labeled benzylamine was purchased from Aldrich. NMR spectra were recorded on a Varian Inova 400-MHz spectrometer, operating at 400.159 MHz for ¹H observation. ¹H NMR chemical shifts are reported relative to tetramethylsilane using residual solvent protons as a secondary reference. ³¹P chemical shifts are reported relative to external phosphoric acid, and ¹⁵N chemical shifts are reported relative to external ammonia, referenced indirectly to the frequency of the deuterated solvent. Two-dimensional ¹H-¹⁵N field gradient heteronuclear single quantum coherence experiments were acquired in pure-phase mode with broadband ¹⁵N decoupling during detection by using pulsed-field gradients for coherence selection, optimized for an average ¹H-¹⁵N heteronuclear coupling of 80 Hz. Reported sample temperatures for low-temperature experiments were calibrated by using a standard 100% methanol sample and calculated by using utilities provided in the VNMR 6.1C software (Varian). One-dimensional ¹⁵N NMR spectra were acquired with continuous broadband irradiation of the protons to provide sensitivity enhancement from both the NOE interaction and from decoupling J_NH.

Cyclic voltammetry experiments were carried out on a Cypress Systems computer-aided electrolysis system under an N₂ or H₂ atmosphere on acetonitrile solutions containing 0.3 M Bu₄NBF₄. The working electrode was a glassy carbon disk, and the counter electrode was a glassy carbon rod. A silver wire was used as a pseudoreference electrode. Ferrocene was used as an internal standard, and all potentials are referenced to the ferrocene/ferrocenium couple. To determine the rate of proton reduction by 3a, cyclic voltammograms were recorded at 50 mV/s as aliquots of a 1:1 buffer solution of [H⁺-DMF]OTF/DMF were added (up to ≈0.3 M) to 3a (3.2 × 10⁻⁴ M) in acetonitrile. The values of i_c/i_p vs. [H⁺-DMF] were plotted as shown in Fig. 3.

Synthesis of P^Cy₂¹⁵N^Bz₂.

A procedure slightly modified from the ligand synthesis reported previously was used. A 250-ml Schlenk flask was charged with cylohexylphosphine (2.32 g, 0.02 mol), fresh paraformaldehyde (1.21 g, 0.04 mol), and degassed ethanol (100 ml). The resulting suspension was immersed in a hot oil bath (80°C) and stirred for 15 min resulting in a clear solution. Under magnetic stirring a solution of ¹⁵N-benzylamine (2.2 ml, 0.02 mol) in ethanol (30 ml) was added dropwise to the hot solution over a period of 60 min. The reaction mixture turned slightly cloudy when 25 ml of the benzylamine solution had been added. Completion of the addition and further heating resulted in a clear solution that was heated overnight at 75°C to form a white precipitate. The solution was cooled to room temperature, solvent volume was reduced on a vacuum line to ≈50 ml, and product was filtered via a cannula stick and dried on a vacuum line. The yield was 2.75 g, 55%. Reducing the volume of the filtrate solution gave another fraction of product. ³¹P NMR(CDCl₃): 41 ppm (s).

Low-Temperature NMR Studies.

A solution of [Ni(P^Cy₂N^Bz₂)₂](BF₄)₂ (15 mg, 0.012 mmol) in actone-d₆ was cooled to −78°C, and H₂ (3 ml, 0.11 mmol) was added through a gas-tight syringe. The solution was shaken and allowed to warm to approximately −10°C until it reacted. After the solution changed from purple to colorless, it was cooled again to −78°C. Within minutes, the solution was monitored by ³¹P NMR at −70°C.

DFT Calculations.

The all-electron DFT calculations were carried out with the Gaussian 03 program by using the 6–31G(d,p) basis and the hybrid B3LYP method.