Acidic ionic liquid/water solution as both medium and proton source for electrocatalytic H₂ evolution by [Ni(P₂N₂)₂]²⁺ complexes

Abstract

The electrocatalytic reduction of protons to H₂ by (where in the highly acidic ionic liquid dibutylformamidium bis(trifluoromethanesulfonyl)amide shows a strong dependence on added water. A turnover frequency of 43,000–53,000 s^-1 has been measured for hydrogen production at 25 °C when the mole fraction of water (χ_H2O) is 0.72. The same catalyst in acetonitrile with added dimethylformamidium trifluoromethanesulfonate and water has a turnover frequency of 720 s^-1. Thus, the use of an ionic liquid/aqueous solution enhances the observed catalytic rate by more than a factor of 50, compared to a similar acid in a traditional organic solvent. Complexes (X = H, OMe,CH₂P(O)(OEt)₂, Br) are also catalysts in the ionic liquid/water mixture, and the observed catalytic rates correlate with the hydrophobicity of X.

Hydrogenase enzymes are active and efficient catalysts for H₂ oxidation and production; for example, [FeFe] hydrogenase exhibits a turnover frequency (TOF) as high as 9,000 s^-1 at 30 °C, and operates at significantly lower overpotentials than most synthetic catalysts (1). In the proposed active site (Fig. 1, structure 1), the noncoordinating amine near the vacant coordination site of the distal iron atom Fe_d is thought to assist both the heterolytic cleavage of H₂ and the movement of protons between the active site and the bulk solution (2). This functionality has inspired the exploration of the role of pendant amines in the electrocatalytic production and oxidation of H₂ (3–7).

Fig. 1.

The proposed active site of [FeFe] hydrogenase (2).

The cyclic diphosphine ligands shown in structure 2 (Fig. 2) have positioned pendant amines in proximity to the Ni. These functional mimics of hydogenase have turnover frequencies of up to 1,850 s^-1 for the production of hydrogen, faster than some naturally occurring hydrogenases (2, 4, 8–11). While variants of 2 exhibit rapid turnover in acetonitrile, computational results indicate that much faster turnover is possible. The rate constant for release of H₂ from intermediate 4 has been estimated to exceed 10⁶ s^-1 at 25 °C (12). However, 4 can only form following two endo protonation events (Fig. 2), and exo protonation limits the rate at which this intermediate is produced. The turnover frequency is then determined by the relative rates of formation and interconversion of protonated isomers 3. Experimental observations support this model: Smaller acids and the addition of water result in higher rates, allowing more ready access to endo protonation and more facile interconversion from the exo isomers (9, 13). These considerations led to the development of a Ni catalyst with cyclic diphosphine ligands having one pendant amine rather than two, precluding the N-H-N pinching interaction shown in structures 3b and 3c and affording a turnover frequency of 106,000 s^-1 at 22 °C in acetonitrile, albeit with a larger overpotential (0.6 V) than seen for variants of 2 (0.28–0.37 V) (14).

Fig. 2.

Catalytic cycle for hydrogen production mediated by complexes with structure 2.

These complexes are designed to control the movement of protons between the metal and the pendant amines of the second coordination sphere. Control of proton movement beyond the second coordination sphere, as manifested in the proton conduction channels of the hydrogenase enzymes, constitutes a natural extension of this approach. In this work, we use a protic ionic liquid with pK_a values closely matching those of the catalyst to accomplish this control. Observations of rapid catalysis with [(DMF)H]OTf (DMF, dimethylformamide; Otf, trifluoromethanesulfonate) in acetonitrile led us to select [(DBF)H]NTf₂ (DBF, di--nbutylformamide; NTf₂, bis(trifluoromethanesulfonyl)amide) (Fig. 3), which serves as electrolyte, acid, and solvent (15). Catalytic currents observed in this medium increase dramatically with added water, and the highest rates are obtained with a saturated aqueous solution.

Fig. 3.

Dibutylformamidium bis(trifluoromethanesulfonyl)amide.

Results

Here, we report a variant of complex 2 developed specifically to favor interaction with hydrophobic media, presenting first the characterization of this complex and its electrocatalytic response in conventional media. This is followed by an examination of its chemical behavior in the protic ionic liquid as a function of water concentration. Lastly, we describe the electrocatalytic results with this and related complexes in ionic liquid/water mixtures and present the methods employed for quantifying both turnover frequencies and overpotentials.

Synthesis and Characterization of and (5).

The ligand 1,5-di(4-n-hexylphenyl)-3,7-diphenyl-1,5-diaza-3,7-diphosphacyclooctane () and the Ni(II) complex, (5), were prepared as shown in Fig. 4 (9, 10). The ³¹P{¹H} NMR spectrum of 5 in CD₃CN features a single peak at 5.3 ppm, consistent with similar [Ni(P₂N₂)₂]²⁺ complexes (9). Spectroscopic data and elemental analysis are consistent with the structure shown; details are given in the SI Text.

Fig. 4.

Synthesis of complex 5.

Cyclic voltammetry of 5 in benzonitrile (PhCN; 0.1 M NBu₄PF₆) shows reversible Ni(II/I) and Ni(I/0) redox couples (E_1/2 = -0.81 and -1.04 V vs. the ferrocenium/ferrocene couple, Fc⁺/Fc; SI Appendix, Fig. S1A), similar to those reported for (9). Voltammetry in acetonitrile reveals a desorption wave on the oxidative scan following reduction to Ni(0), indicating electrodeposition (SI Appendix, Table S1 and Fig. S1B (16).

Hydrogen Production Electrocatalysis by 5 in Acetonitrile.

Adding the acid [(DMF)H]OTf to 5 in acetonitrile (0.1 M NBu₄PF₆) affords a sigmoidal trace consistent with catalytic reduction of protons to form H₂ (Eq. 1 with B = DMF, Fig. 5). The catalytic current i_cat increases with added acid until substrate saturation is attained at approximately 0.16 M. Under these conditions, the turnover frequency (TOF) is equal to k_obs (200 s^-1), which can be calculated using Eq. 2 from i_cat/i_p where i_p is the peak current in the absence of substrate, n is the number of electrons per turnover (two), υ is the scan rate, and F, R, and T are Faraday’s constant, the gas constant, and the temperature (10).

[1]

[2]

Fig. 5.

Cyclic voltammograms of 5 (0.9 mM) in acetonitrile (0.1 M NBu₄PF₆) without added acid (red trace, SI Appendix, Fig. S1) and with added acid (0.28 M [(dimethylformamide)H]OTf; blue trace). One millimeter glassy carbon working electrode, scan rate υ = 0.05 V s^-1. Referenced to the Fc⁺/Fc couple.

The half-peak potential E_p/2 for the catalytic wave is -0.73 V, corresponding to an overpotential of 0.23 V, using the method of Evans and coworkers (17). Adding water gives a maximum turnover frequency of 740 s^-1 with a E_p/2 of -0.79 V and an overpotential of 0.28 V at a water concentration of 1.4 M. These rates are comparable to those observed for under the same conditions (720 s^-1 with water) (9).

NMR Spectroscopy of 5 in [(DBF)H]NTf₂ and Its Mixtures with Water.

Compound 5 gives a viscous orange-red solution in [(DBF)H]NTf₂ which darkens with added water, up to a mole fraction (χ_H2O) of 0.78, beyond which two phases are observed. ³¹P{¹H} NMR spectra (Fig. 6) show a broad singlet at -14 ppm and smaller peaks at 0 and 7 ppm. With added water, these resonances diminish and a singlet at 3 ppm emerges with increasing intensity. Adding DBF to 5 in neat [(DBF)H]NTf₂ produces a similar effect, with the ³¹P NMR resonance of 5 approaching its value in free DBF (6 ppm) (SI Appendix, Fig. S2).

Fig. 6.

³¹P{¹H} NMR spectra of 5 in [(DBF)H]NTf₂ with χ_H2O ranging from 0 to 0.75.

Adding water results in the removal of one or both protons from the diprotic Ni(II) species (Fig. 6). The singlet at -14 ppm is assigned to diprotic 5. For comparison, in CD₃CN, the closely related doubly pinched diprotic isomer of (where Bn is benzyl) appears at -15.3 ppm in the ³¹P NMR spectrum (8). The two equally intense singlets are assigned to the inequivalent P environments of a monoprotic species, and the singlet at 3 ppm is from the aprotic Ni(II) species. Cyclic voltammetry substantiates deprotonation with added water, and open circuit potential measurements show a decrease in solution acidity upon addition of water.

³¹P NMR spectra obtained over one month show that complex 5 is moderately stable in neat [(DBF)H]NTf₂ (a 25% decrease in signal was observed). Complex 5 in [(DBF)H]NTf₂ (χ_H2O = 0.72) decomposes with t_1/2 ≈ 1 week. With HNTf₂ in CD₃CN, 5 decomposes rapidly to afford unbound, protonated ligand, and in neat DBF, 5 undergoes slow ligand displacement (t_1/2 ≈ 2 wk).

Cyclic Voltammetry of 5 in [(DBF)H]NTf₂.

Fig. 7 shows voltammograms with [5] = [ferrocene] = 1.1 mM in DBF (0.1 M NBu₄PF₆; red trace) and in [(DBF)H]NTf₂ (blue trace). The ionic liquid is a proton source for hydrogen production, so only a catalytic wave is observed. This wave has E_p/2 = -0.38 V, considerably positive of the Ni(II/I) couple in DBF (E_1/2 = -0.82 V) reflecting protonation of 5 (8, 18), consistent with the NMR results discussed above. Diffusion is slowed in the more viscous ionic liquid, attenuating both the Fc⁺/Fc wave and the catalytic current (19).

Fig. 7.

Cyclic voltammograms of 5 (1.1 mM) and ferrocene (1.1 mM) in DBF (0.2 M NBu₄PF₆, red trace) and in [(DBF)H]NTf₂, (blue trace). 1 mm glassy carbon working electrode, scan rate υ = 0.05 V s^-1. Referenced to the Fc⁺/Fc couple.

Cyclic Voltammetry of 5 in [(DBF)H]NTf₂ with Added Water.

As water is added to 5 in [(DBF)H]NTf₂, i_cat increases and E_p/2 shifts to more negative values (Table 1). Fig. 8 shows the cyclic voltammogram of 5 in [(DBF)H]NTf₂ with χ_H2O ranging from 0 to 0.75. At χ_H2O = 0.57, the conversion of protic Ni(II) to aprotic 5 is essentially complete (Fig. 6). Ultimately, i_cat increases by a factor of 90, and E_p/2 shifts by -0.25 V. Controlled-potential coulometry of 5 in [(DBF)H]NTf₂ (χ_H2O = 0.72) confirms the production of hydrogen, giving a current efficiency of 92 ± 5% with a measured turnover number 13. Without added catalyst, onset of hydrogen production is observed at -1.2 V (SI Appendix, Fig. S3). Viscosity decreases with added water, increasing Fc⁺/Fc redox currents (SI Appendix, Fig. S4). However, the increase in i_cat is much larger than expected due to viscosity changes alone. This is discussed in detail in the SI Text.

Table 1.

Voltammetric data for 5 in [(DBF)H]NTf₂ with water added

[Ni2+] (mM)	χH2O ([H2O], M)	icat (μA)	Ep/2 (V)*	OCP (V)*†	OP (V)
0.66	0 (0)	0.4	-0.38	0.041	0.42
0.64	0.30 (1.4)	0.8	-0.40	-0.012	0.39
0.63	0.47 (2.6)	3.4	-0.49	-0.058	0.43
0.62	0.57 (3.9)	5.6	-0.55	-0.107	0.44
0.60	0.64 (5.1)	15.2	-0.59	-0.152	0.44
0.59	0.69 (6.2)	28.0	-0.61	-0.184	0.43
0.58	0.72 (7.2)	29.5	-0.62	-0.207	0.41
0.56	0.75 (7.8)	35.8	-0.63	-0.219	0.41

Voltammograms are shown in Fig. 8. Overpotentials (OP) are based on open circuit potential (OCP) measurements (SI Appendix, Fig. S5) and E_p/2.

^*Referenced to Fc⁺/Fc.

^†Interpolated from [H₂O] vs. OCP data (SI Appendix, Fig. S6).

Fig. 8.

Cyclic voltammograms of 5 (0.6 mM) in 0.83 mL of [(DBF)H]NTf₂, 1 mm glassy carbon working electrode, υ = 0.1 V s^-1, initially and after each of seven additions of 20 μL water. Referenced to Fc⁺/Fc.

Cyclic Voltammetry of 5 in [(DBF)H]NTf₂ (χ_H2O = 0.72). Catalyst Concentration Effects.

Catalytic currents for hydrogen production mediated by catalysts typically show a linear dependence on [catalyst] (9, 11). The dependence of i_cat on [5] in [(DBF)H]NTf₂ (χ_H2O = 0.72) is linear over 3 orders of magnitude (SI Appendix, Fig. S7), with [5] < 1 mM, with currents falling off at higher [5]. Data collected with low catalyst concentrations were employed to obtain turnover frequencies.

Voltammetry of Other [Ni(P₂N₂)₂](BF₄)₂ Complexes in Anhydrous [(DBF)H]NTf₂ and [(DBF)H]NTf₂/Water Mixtures.

The complexes having X = H (6), OMe (7), CH₂P(O)(OEt)₂ (8), Br (9), and CF₃ (10), all catalysts in MeCN, (9) also show catalytic waves in dry [(DBF)H]NTf₂ that increase with added water, and show a linear dependence of i_cat on [cat] (SI Appendix, Figs. S8 and S9). With 10, low catalytic currents and poorly defined plateaus obscured i_cat (SI Appendix, Fig. S10). As with 5, the catalytic wave for 6 shifts negative with added water. The CF₃ substituted complex 10 shows little variation in E_p/2 with χ_H2O, as expected from NMR evidence that 10 is not protonated in neat [(DBF)H]NTf₂ (SI Appendix, Fig. S11).

Determination of Turnover Frequencies.

Steady-state catalytic currents i_cat can be used to determine turnover frequencies (k_obs; s^-1) using Eq. 3, where n is the number of electrons consumed per turnover, A is the electrode area, and D_cat is the catalyst diffusion coefficient (20, 21). The peak current i_p of an n^′-electron wave for the same catalyst without substrate, when observable, will also depend on A, [cat], and D_cat (Eq. 4) (19), and k_obs is given by i_cat/i_p, as shown in Eq. 2.(9–11, 14) Since [(DBF)H]NTf₂ is a substrate for electrocatalytic H₂ evolution, i_p cannot be measured, and Eq. 2 cannot be used. Determination of k_obs using Eq. 3 requires the measurement of A and D. Determination of A is described in the SI Text.

[3]

[4]

Catalysts were evaluated using [(DBF)H]NTf₂ (χ_H2O = 0.72) to ensure solution homogeneity. Catalyst diffusion coefficients in this medium were estimated voltammetrically using a noncatalytic model complex [Ni(dppb)₂](BF₄)₂ (11; dppb = 1,2-bis(diphenylphosphinyl)benzene), and by ¹⁹F pulsed gradient spin-echo (PGSE) NMR spectroscopy using (10). Steady-state linear sweep voltammetry and chronoamperometry experiments of 5 and 11 in MeCN (0.1 M NBu₄PF₆) gave D₅ = 6 × 10^-6 and D₁₁ = 9 × 10^-6 cm² s^-1. Chronoamperometry experiments of 11 in [(DBF)H]NTf₂ (χ_H2O = 0.72) gave D₁₁ = 2.4 × 10^-7 cm² s^-1. Assuming D₅/D₁₁ is medium-independent, D₅ is 1.6 × 10^-7 cm² s^-1 in this medium. The value of D₁₀ in [(DBF)H]NTf₂ (χ_H2O = 0.72) is 1.3 × 10^-7 cm² s^-1 by ¹⁹F PGSE, in agreement with D₅ in [(DBF)H]NTf₂ (χ_H2O = 0.72) estimated using electrochemical methods. Turnover frequencies were calculated for each catalyst using Eq. 3 and D₅ and D₁₀ as lower and upper limits for the catalyst diffusion coefficients (Table 2).

Table 2.

Comparison of catalytic turnover frequencies (TOF) and overpotentials (OP) for in [(DBF)H]NTf₂ (χ_H2O = 0.72) and MeCN (0.1 M NBu₄PF₆) with added [(dimethylformamide)H]OTf and water

	[(DBF)H]NTf2 (χH2O = 0.72)		MeCN*
X	TOF† (s-1)	OP (V)	TOF (s-1)	OP (V)‡
n-hexyl (5)	4.3–5.3 × 104	0.40	7.4 × 102	0.28
H (6)	5.5–6.8 × 103	0.42	7.2 × 102	0.32
OMe (7)	6.3–7.8 × 102	0.37	4.8 × 102	0.33
CH2P(O)(OEt)2(8)	1.0–1.3 × 102	0.41	1.9 × 103	0.37
Br (9)	4.4–5.4 × 103	0.44	1.0 × 103	0.29
CF3 (10)	—	—	1.2 × 102	0.30

In amounts affording maximum catalytic currents.

^*See ref. 12 for 6–9.

^†Lower and upper limits calculated using D₅ and D₁₀, respectively.

^‡Calculated according to ref. 20. See Discussion for comparison of overpotentials for the different media.

Determination of Equilibrium Potentials for Hydrogen Production.

The thermodynamic potentials for the interconversion of protons and electrons with H₂ and base (Eq. 1), required for the determination of overpotentials, were obtained for [(DBF)H]NTf₂/water mixtures with χ_H2O from 0 to 0.72 under 1 atm H₂. For each solution, the open circuit potential (OCP) (SI Appendix, Fig. S5) between a platinum electrode and a frit-separated AgCl/Ag reference electrode was recorded (22) and referenced to the Fc⁺/Fc couple, placing the OCP on the scale used for the catalytic experiments and eliminating the unknown junction potential between the reference and analyte compartments. Adding water to the ionic liquid shifts the OCP to more negative values, as expected for addition of a base (SI Appendix, Fig. S6). Experiments using 5% H₂ in N₂ rather than pure H₂ shifted the OCP to more positive values, also as expected. Table 1 provides OCP values and overpotentials |E_p/2 - OCP| for catalyst 5, corresponding to the cyclic voltammograms shown in Fig. 8. Although E_p/2 values shift negative with added water, the overpotential for 5 stays fairly consistent across the range of water concentrations.

Discussion

Control of the movement of protons over different distance scales is accomplished in hydrogenase enzymes by coupling the active site of the catalyst with the reaction medium through proton channels, and in polymer electrolyte membrane fuel cells by placing the catalyst in contact with a proton-conducting membrane. The pendant amines built into electrocatalysts with structure 2 shuttle protons between the metal and the catalyst periphery, enabling rapid hydrogen production and oxidation in acetonitrile (4, 11). Fast proton movement in protic ionic liquids (23, 24) has guided the development of media for hydrogen production at conventional platinum and other metallic electrodes (22, 23, 25, 26).

These considerations prompted us to study catalysts with structure 2, designed to control the intra- and intermolecular movement of protons, in acidic ionic liquid media exhibiting high proton mobility. Our principal findings are these: [(DBF)H]NTf₂ is an excellent electrolyte and proton source for electrocatalytic production of H₂. In this medium, complex 5 exhibits a strong rate enhancement upon the addition of water, showing that water influences the kinetics of protonation and deprotonation; however, the overpotentials are invariant with added water. Turnover is substantially faster with this complex in [(DBF)H]NTf₂ (χ_H2O = 0.72) than observed in acetonitrile:water with [(DMF)H]⁺. The TOF values for complexes 5–9 appear to track the hydrophobicity of X, indicating that specific catalyst-solvent interactions may be important in controlling proton movement.

Complex 5 dissolves in [(DBF)H]NTf₂ to produce species having ³¹P{¹H} NMR resonances consistent with the diprotic and monoprotic Ni(II) complexes shown in Fig. 6. The N-H-N pinching interaction shown in these structures has been characterized for a similar Ni(II) system by isotopic labeling (8). The cyclic voltammogram of 5 in the neat ionic liquid shows a catalytic wave having a half-peak potential E_p/2 = -0.38 V vs. Fc⁺/Fc (Fig. 8).

The rate enhancements with added water shown in previous studies of these complexes in acetonitrile are also observed in the ionic liquid system. With complex 5 in particular, adding water affords a 90-fold enhancement in the catalytic current (Fig. 8). As i_cat increases, the catalytic wave shifts from -0.38 to -0.63 V vs. Fc⁺/Fc. Aprotic Ni(II) should be reduced at more negative potentials than the protonated Ni(II) species shown in Fig. 6, so the shift of E_p/2 with added water to more negative potentials is consistent with the deprotonation observed by NMR spectroscopy. These findings suggest that exo protonation of Ni(II) is slowing catalysis by 5 under dry conditions, and that the increase in pH on addition of water precludes this preprotonation. A medium of sufficient acidity to protonate 5 in the Ni(II) oxidation state is unlikely to promote isomerization among protic intermediates in reduced states (intermediates 3, Fig. 2), since these states will be much less acidic than their Ni(II) congeners.

The deprotonation of exo-protonated Ni(II) species is one factor in the substantial water effect observed with 5 in [(DBF)H]NTf₂. However, currents continue to increase significantly even after this deprotonation is essentially complete (χ_H2O≥0.64, Table 1 and Fig. 6). Solution viscosity decreases with added water, increasing diffusion rates and thus catalytic currents [Eq. 3]; this may largely account for the moderate increase in current with 10, however the magnitude of the increase in current with complex 5 is greater than can be attributed to viscosity effects alone. The effect of water on the kinetics of proton transfer is perhaps the most important contributing factor. This rate enhancement is attributed to increased rates of endo protonation of reduced Ni species during catalysis, leading to faster production of isomer 4 and thus faster H₂ evolution (Fig. 2) (9, 13).

The turnover frequency of 5 in [(DBF)H]NTf₂ (χ_H2O = 0.72) is 4.3–5.3 × 10⁵ s^-1, 60–70 times faster than observed for this complex in acetonitrile using [(DMF)H]⁺ with water added. TOF values for [ complexes in [(DBF)H]NTf₂ (χ_H2O = 0.72) are highest for X = hexyl (5) and lowest for X = CH₂P(O)(OEt)₂ (8) and span two orders of magnitude. These values were calculated for each catalyst using Eq. 3 and D₅ and D₁₀ as bracketing estimates for the catalyst diffusion coefficients. Larger solutes diffuse more slowly, and with the exception of (6), all of the catalysts are at least as large as complex 10. Since overestimating diffusion coefficients will lead to smaller estimates for TOF [Eq. 3], the upper limits listed in Table 2 are conservative.

The turnover frequencies for complexes 5–9 do not track the driving force for H₂ release from intermediate 4 (Fig. 2) as observed for these complexes in acetonitrile (9), but instead appear to track the hydrophobicity of X, suggesting that ligand interactions with DBF or [(DBF)H]⁺ are important. Self-assembly via interaction of similarly hydrophobic moieties is well-known for ionic liquids (27), and may play a role in directing proton delivery. In particular, interactions between the hexyl tails of 5 and the butyl groups of dibutylformamide or [(DBF)H]⁺ may provide a high local concentration of protic and basic sites, possibly facilitating isomerization or favoring endo over exo protonation. Water could promote this self-assembly, and also furnish a conduit for rapid proton movement between the bulk solution and the catalyst proton relays. This structuring would likely be more significant for 5 than for CF₃-substituted 10 and would slow diffusion, so the upper limit to turnover frequency for 5 calculated using D₁₀ may be underestimated.

The overpotentials for complexes 5–9 in [(DBF)H]NTf₂ (χ_H2O = 0.72) are all near 0.4 V (Table 2, column 4), somewhat larger than the values listed for acetonitrile (column 6), due in part to the difference in the (BH⁺)/(H₂ + B) equilibrium potentials measured directly for [(DBF)H]NTf₂ (χ_H2O = 0.72), compared to those calculated for acetonitrile with [(DMF)H]⁺ by a different method (17). The large increases in i_cat as water is added to 5 in [(DBF)H]NTf₂ are not accompanied by changes in overpotential (Table 1), indicating that faster turnover frequencies can be achieved without increasing overpotentials, in this case by optimization of the reaction medium and its interactions with the catalyst.

Conclusions

The results presented here demonstrate rapid, efficient production of H₂ mediated by a molecular electrocatalyst in a medium consisting of water and acid. The catalyst (5) dissolved in the highly acidic protic ionic liquid dibutylformamidium bis(trifluoromethanesulfonyl)amide and water (χ_H2O = 0.72) exhibits a TOF of at least 4 × 10⁴ s^-1, far greater than observed for related systems in acetonitrile. Water increases rates without significantly increasing overpotentials, a finding of particular significance in the context of electrocatalysis for energy conversion. Of the catalysts studied here, rates correlate with ligand hydrophobicity, suggesting that catalyst-medium interactions are essential in determining catalytic properties. Interactions between the hexyl tails of 5 and the butyl substituents of DBF and [(DBF)H]⁺ are thought to promote rapid protonation of the catalyst at endo positions, leading to faster turnover. These results illustrate that control of proton movement as a design principle extends beyond the catalyst itself to encompass the catalyst, the medium, and their interactions.

Materials and Methods

The sourcing and purification of commercially available chemicals, references for known compounds, synthetic details for 5, and the instrumentation and methods for routine NMR experiments, elemental analysis, and electrochemistry are detailed in the SI Text. All voltammetry experiments were conducted at ambient temperatures (23–26 °C).

Cyclic Voltammetry of 5–10 in [(DBF)H]NTf₂. Addition of H₂O.

The Ni complexes were dissolved in 1 mL of [(DBF)H]NTf₂ with stirring. In a representative experiment, a cyclic voltammogram was recorded (υ = 0.1 V s^-1) with 5 (0.66 mM) and ferrocene (< 5 mM) in [(DBF)H]NTf₂. Water was added in 20–25 μL aliquots. After each addition, the solution was stirred briefly, and a cyclic voltammogram was recorded.

Diffusion NMR Experiments.

The standard Varian stimulated echo PGSE pulse sequence was used (28). Experimental parameters included 4–64 scans, 4–7 ms gradient pulse lengths and diffusion delays of 0.01–1.0 s. Longer diffusion delays and larger gradient strengths were needed for the ionic liquid samples due to the increased viscosity. The gradient strengths were varied from 0–20 G cm^-1. Diffusion measurements on the sample of 10 in [(DBF)H]NTf₂ (χ_H2O = 0.72) employed ¹⁹F nuclei due to the increased sensitivity compared to ³¹P nuclei, which showed poor signal-to-noise ratios due to the fast relaxation times and increased internal gradients (29). The ¹⁹F 90° pulse was 7.5 μs and relaxation delays of 20 s (10 × T₁) were used. The temperature was maintained at 26 °C using an XRII852 Sample Cooler (FTS Systems).

Electrochemical Determination of D for 5 and 11 in MeCN (0.1 M NBu₄PF₆) and [(DBF)H]NTf₂ (χ_H2O = 0.72).

[Ni(dppb)₂](BF₄)₂ (11; dppb = 1,2-bis(diphenylphosphinyl)benzene) showed reversible Ni(II/I) and Ni(I/0) couples in MeCN (0.1 M NBu₄PF₆) (30). In [(DBF)H]NTf₂ (χ_H2O = 0.72), 11 showed a diffusion-controlled Ni(II/I) couple with E_1/2 = -0.52 V (SI Appendix, Fig. S12). The Ni(I/0) couple was irreversible (E_p,red = -0.8 V), but well separated from the Ni(II/I) couple. Steady-state linear sweep voltammetry and chronoamperometry experiments using solutions of 5 and 11 in MeCN (0.1 M NBu₄PF₆) with known concentrations gave D = 6 × 10^-6 and 9 × 10^-6 cm² s^-1, respectively. Chronoamperometry of 11 using a potential step from -0.3 to -0.7 V vs. Fc⁺/Fc gave D = 2.4 × 10^-7 cm² s^-1 for 11 in [(DBF)H]NTf₂ (χ_H2O = 0.72).

Open Circuit Potential Determinations in [(DBF)H]NTf₂ (χ_H2O = 0 to 0.75).

A 1 mm platinum wire electrode was immersed in aqua regia for 30 min, rinsed in flowing 18 MΩ H₂O for several minutes, then heated to an orange glow in a hydrogen/air flame, cooled in a stream of hydrogen, and transferred under nitrogen to the glovebox. Open circuit potentials were measured between this electrode and a AgCl/Ag pseudoreference electrode containing MeCN (0.1 M NBu₄PF₆), separated from the analyte compartment by a Vycor frit. The analyte solution consisted of [(DBF)H]NTf₂ (0.6732 g, 0.498 mL) and ferrocenium tetrafluoroborate (approximately 1 mg), sparged with H₂ for 10 min prior to measurement. The open circuit potential was recorded with stirring for 30 s; the stirring was turned off, and the potential of the pseudoreference electrode vs. the Fc⁺/Fc couple was established voltammetrically, using glassy carbon working and counterelectrodes in a three-electrode configuration. Hydrogen-sparged water was added in 10 μL aliquots, and the measurement sequence was performed twice after each addition. In general, the open circuit potential changed by less than 1 mV over 30 s. Chronopotentiograms are given in SI Appendix, Fig. S11, and plots of the open circuit potential vs. [H₂O] and ln([H₂O]) are given in SI Appendix, Fig. S12.