Using a Two-Step Hydride Transfer to Achieve 1,4 Reduction in the Catalytic Hydrogenation of an Acyl Pyridinium Cation

Abstract

The stoichiometric reduction of N-carbophenoxypyridinium tetraphenylborate (6) by CpRu(P–P)H (Cp = η⁵-cyclopentadienyl; P–P = dppe, 1,2-bis(diphenylphosphino)ethane or dppf, 1,1′-bis(diphenylphosphino)ferrocene) and Cp*Ru(P–P)H (Cp* = η⁵-pentamethylcyclopentadienyl; P–P = dppe) gives mixtures of 1,2- and 1,4-dihydropyridines. The stoichiometric reduction of 6 by Cp*Ru(dppf)H (5) gives only the 1,4-dihydropyridine, and 5 catalyzes the exclusive formation of the 1,4-dihydropyridine from 6, H₂, and 2,2,6,6-tetramethylpiperidine. In the stoichiometric reductions, the ratio of 1,4 to 1,2 product increases as the Ru hydrides become better one-electron reductants, suggesting that the 1,4 product arises from a two-step (e⁻/H•) hydride transfer. Calculations at the UB3LYP/6-311++G(3df,3pd)//UB3LYP/6-31G* level support this hypothesis, indicating that the spin density in the N-carbophenoxypyridinium radical (13) resides primarily at C4, while the positive charge in 6 resides primarily at C2 and C6. The isomeric dihydropyridines thus result from the operation of different mechanisms: the 1,2 product from a single-step H⁻ transfer and the 1,4 product from a two-step (e⁻/H•) transfer.

Introduction

In 1972 Stout, Takaya, and Meyers reported a synthesis of 1,4-dihydropyridines via the addition of an azide to a 2,3-diazabicycloheptene to form an aziridine (followed by hydrolysis, oxidation, and N₂ extrusion).^1,2 Dihydropyridines may also be prepared by the reduction of pyridinium salts. However, in a 1982 review with Stout entitled Recent Advances in the Chemistry of Dihydropyridines, Meyers noted that “[t]he formation of dihydropyridines by the reduction of pyridines and pyridinium salts is complicated by the fact that mixtures of 1,2- and 1,4-dihydropyridines result.”³ For example, Fowler obtained both N-carbomethoxy-1,2-dihydropyridine and N-carbomethoxy-1,4-dihydropyridine by treating a mixture of pyridine and sodium borohydride with methyl chloroformate (eq 1).⁴ Comins concluded in 1984 that “a 1-acyl substituent stabilizes the dihydropyridine system” and found that both N-(alkylcarbonyl)- and N-(alkoxycarbonyl)-1,4-dihydropyridines can be prepared by a completely regioselective (but stoichiometric) reduction with Li(^tBuO)₃AlH/CuBr (eq 2).⁵

(1)

(2)

In general a 1,4-dihydropyridine is more stable thermodynamically than its 1,2 isomer.^6–9 Fowler measured in 1972 a substantial positive ΔG°, 2.3 kcal/mol, for eq 3 at 91.6° in dmso,¹⁰ and Eisner and Sadeghi reported in 1978 the Rh-catalyzed isomerization of 1a to 1b.¹¹ The reduction of a pyridinium cation to a 1,2-dihydropyridine is thus usually the result of kinetic control.

(3)

We have previously shown that half-sandwich ruthenium hydride complexes catalyze the hydrogenation of the C=N bonds in iminium cations (eq 4)¹² and the rings in aziridinium cations (eq 5).¹³ In the present work we have investigated the reactivity of the Ru hydride complexes 2–5 toward the acyl pyridinium cation 6,¹⁴ and the ability of 5 to catalyze the hydrogenation of 6. The water-soluble Ru catalyst 7 and the Rh catalyst 8 have been shown to catalyze the 1,4 hydrogenation of NAD⁺ models as well as NADP⁺.^15–17

(4)

(5)

While investigating the hydrogenation of aziridinium cations we observed the formation of the radical cation 5^•+ in eq 6,¹³ suggesting electron transfer. Multi-step mechanisms beginning with e⁻ transfer have often been suggested as an alternative to the single-step transfer of hydride, particularly for H⁻ transfers from NAD(P)H and its analogs.¹⁸ Scheme 1 shows a possible multi-step pathway, involving an H• transfer (HAT)^19,20 after the initial single-electron transfer (SET);^21,22 the possibility that the “H•” step involves the separate transfer of H⁺ and e⁻ has also been considered.^22–25 We have therefore considered the possibility of e⁻ transfer during the reaction of the Ru hydrides 2–5 with the pyridinium cation 6, and have examined the electrochemical oxidation of 2–5 and the electrochemical reduction of 6.

SCHEME 1.

(6)

Results and Discussion

Stoichiometric and Catalytic Reactions

All four Ru–H complexes quantitatively transfer H⁻ to 6, giving the 1,2- and 1,4-dihydropyridines 9a and 9b (Table 1). The resulting 16-electron Ru cations are trapped by excess CH₃CN in entries 1–4 or solvent CD₃CN in entries 5–8. These stoichiometric reactions proceed readily at room temperature, although entries 6 and 8 require heating to dissolve the sparingly soluble 3 and 5. The trend in the product distribution among the different hydride complexes is the same in CD₂Cl₂ and CD₃CN. In both solvents 5 (Hembre’s hydride,²⁶ Cp*Ru(dppf)H) is remarkably selective, giving only the 1,4-dihydropyridine.

TABLE 1.

Product Distributions of the Stoichiometric Reduction of 6 with Ruthenium Hydrides^a

Entry	Complex	Solvent	9a:9b
1b	2	CD2Cl2	52	48
2b	3	CD2Cl2	30	70
3b	4	CD2Cl2	4	96
4b	5	CD2Cl2	-	100
5c	2	CD3CN	23	77
6d	3	CD3CN	15	85
7c	4	CD3CN	2	98
8d	5	CD3CN	-	100

^aGeneral reaction conditions: 0.02 mmol 6, 0.02 mmol hydride complex, 700 μL solvent, product ratio determined by ¹H NMR integration.

^b0.08 mmol CH₃CN, room temperature.

^cRoom temperature.

^d75 °C, 2 h.

The product ratios in Table 1 are different for each hydride, remain constant after each reaction, and are obviously the result of kinetic control. Thermodynamic control (recall the reported slow isomerization of 1a to 1b¹¹) would give the 1,4 product 9b. Only the formation of 9b (and none of 9a) was observed when the reaction in entry 4 was monitored by ¹H NMR at 228 K.

The iminium and aziridinium cations in eqs 4 and 5 are hydrogenated by an ionic mechanism, where H⁻ and H⁺ are transferred in separate steps.^12,13 In eqs 4 and 5, the tertiary amine product removes H⁺ from an H₂ complex, regenerates the hydride complex, and makes the reaction catalytic in Ru, with H₂ as the ultimate reductant. If we attempt to make the reduction of the pyridinium cation 6 in Table 1 catalytic, the dihydropyridine products 9a and 9b are not sufficiently basic to deprotonate the H₂ complex (10a), so a non-nucleophilic base must be added (Scheme 2). TMP (2,2,6,6-tetramethylpiperidine) fulfills this requirement and makes the reaction catalytic; DBN (1,5-diazabicyclo[4.3.0]non-5-ene) and NEt₃ react with 6.

SCHEME 2.

Isomerization of the cationic dihydrogen complex Cp*Ru(dppf)(H₂)⁺ 10a to the trans-dihydride complex (10b)²⁷ competes with the catalytic cycle, as TMP is not basic enough to deprotonate 10b. Such isomeric dihydrogen and dihydride complexes are readily distinguished by the T₁(min) of their 1H NMR hydride resonances;28 the dihydrogen resonance of 10a has a typically short T₁(min) of 11.5 ms (218.5 K, 300 MHz), while that of the dihydride resonance of 10b is much longer, 0.151 s (195.2 K, 300 MHz).

An attempt to catalyze the hydrogenation of 6 with 5 at room temperature in CH₂Cl₂ gave only 28% conversion to 9b (Table 2). The ¹H NMR analysis of a reaction aliquot showed the presence of 10b, confirming that the isomerization reaction (10a → 10b) is responsible for stopping catalysis.

TABLE 2.

Catalytic Reactions in CH₂Cl₂^a

Temperature, °C	% Conversionb
RT	28
10	58
0	38

^aGeneral reaction conditions: 0.50 mmol 6, 0.60 mmol TMP, 0.01 mmol 5, 10 mL CH₂Cl₂, 80 psi H₂, 24 h.

^bDetermined by ¹H NMR integration.

Because the 10a → 10b isomerization shuts down the catalytic cycle, it is important to know its rate. The protonation of 5 at room temperature gives 10a, which rapidly isomerizes to 10b (Scheme 3). At lower temperatures, the isomerization (a first-order reaction) can be followed by ¹H NMR. At 0 °C, k = 1.32(2) × 10⁻³ s⁻¹ in CD₂Cl₂, giving 10a a half-life of about 9 min at this temperature. Belkova has recently reported k = 1.53 × 10⁻³ s⁻¹ for the isomerization of protonated 4 (Cp*Ru(dppe)(H₂)⁺ → trans-Cp*Ru(dppe)(H)₂⁺) at 260 K.²⁹

SCHEME 3.

One would expect lower temperatures to decrease the rate of catalyst deactivation, slowing the unwanted (intramolecular) isomerization of 10a → 10b relative to the needed (intermolecular) proton transfer from the H₂ ligand of 10a (note the relatively large ΔH^‡ for isomerization reported by Belkova).²⁹ Unfortunately, 6 is not very soluble in CH₂Cl₂ below room temperature; a reaction at 0 °C in CH₂Cl₂ gave only a modest improvement in conversion over the room temperature reaction (Table 2). An intermediate temperature of 10 °C gave a significant improvement in the conversion of 6 to 9b in CH₂Cl₂.

The solubility of 6 is much greater in THF than in CH₂Cl₂. Using THF at 10 °C gave 86% conversion to 9b (eq 7); aqueous workup and flash chromatography gave pure 9b in 75% yield (with respect to the initial 6). To the best of our knowledge, this is the first catalytic hydrogenation of an N-acyl pyridinium cation that gives a 1,4 product exclusively.

(7)

Probing the Mechanism of H⁻ Transfer

The reactivities of the hydride ligands in the Ru complexes 2–5 are influenced by the nature of the chelating bisphosphine and by the extent of substitution on the Cp ring. For example, 2 readily transfers H⁻ to the iminium cation in eq 4, while the hydride ligand of 4 is basic enough to deprotonate that cation.¹² The dppf ligand makes 3 and 5 unusually good one-electron reducing agents,²⁶ but it does not make them better hydride donors: complexes 2, 3, and 4 (but not 5) react with the iminium cation 11 (eq 8).

(8)

The difference in reactivity between 5 with 6 and 5 with 11 led us to suspect a two-step mechanism for H⁻ transfer from 5 to 6, with e⁻ transfer preceding H• transfer. Hembre found that 5 is easily oxidized by ferrocenium or trityl cation.^26,27

Although the potentials of 2,^30,31 4,²⁹ and 5²⁶ have been reported, they were obtained in THF. We therefore performed cyclic voltammetry on 2–5 (and 6 and 11 for comparison) in CH₂Cl₂ (Table 3).

TABLE 3.

Cyclic Voltammetry

Compound	Potentiala	Process
CpRu(dppe)H (2)	−0.16b	Ru(III/II)
CpRu(dppf)H (3)	−0.31c	Ru(III/II)
Cp*Ru(dppe)H (4)	−0.51d	Ru(III/II)
Cp*Ru(dppf)H (5)	−0.63d	Ru(III/II)
6	−0.87e	reduction
11	−1.42e	reduction

^aPotentials in Volts (V) in CH₂Cl₂ vs. Fc/Fc⁺.

^bE_pa, irreversible, 50 mV/s.

^cE_pa, quasireversible, I_a/I_c = 1.25, E_pa − E_pc = 0.09, 50 mV/s.

^dE_pc, irreversible, 50 mV/s.

^eE_pc, irreversible, 200 mV/s.

Under our conditions the cyclic voltammogram of 2 is sufficiently irreversible (see Supporting Information) to leave some doubt about its thermodynamic potential relative to that of 3. However, the order of the reversible potentials of 4 and 5 implies that the thermodynamic potential of 3 must be more negative than that of 2, and that the order of the reduction potentials of the hydride complexes, from least to most negative, is 2, 3, 4, and 5. This order correlates with the increase in 9b/9a as we go from 2 to 5 in Table 1.

From their irreversible potentials in Table 3 it appears that 6 is more easily reduced than 11, and that an SET mechanism for H⁻ transfer is much more plausible with 6 than with 11. The E_1/2 of 5 is close to the E_pc of 6 but not to the E_pc of 11, suggesting that an e⁻ transfer mechanism is available for the reaction of 5 with 6 (which occurs) but not for the reaction of 5 with 11 (which does not occur).

Indirect evidence for an initial electron transfer in the reaction of 5 with 6 (Scheme 4) was obtained by ¹H NMR. At 228 K, the hydride signal of 5 was broadened, and after 1 h a significant amount of 10a was present (see Supporting Information). The broadening arose from self-exchange with Ru–H^•+ (5^•+).¹³ The 10a came from disproportionation (eq 9) and proton transfer (eq 10), a sequence known to result in the formation of dihydrogen complexes from Ru–H^•+.³¹

SCHEME 4.

$$2\text{Ru}-{\text{H}}^{•+}\to \text{Ru}-\text{H}+\text{Ru}-{\text{H}}^{2+}$$ (9)

$$\text{Ru}-\text{H}+\text{Ru}-{\text{H}}^{2+}\to {\text{Ru(H}}_{\text{2}}{)}^{+}-{\text{Ru}}^{+}$$ (10)

EPR evidence for the formation of 5^•+ was obtained by mixing CH₂Cl₂ solutions of 5 and 6 at 223 K. The reaction mixture, initially orange, became green after 10 min. The X-band EPR, recorded at 77 K after quenching the reaction in liquid nitrogen, clearly showed the presence of 5^•+ (Figure 1).³² Apparently at this temperature, the second step (HAT) in Scheme 4 is slow relative to the first step (SET), permitting the accumulation of some 5^•+.

FIGURE 1.

X-band EPR spectrum of 5^•+ at 77 K (from the reaction of 5 with 6 in CH₂Cl₂ at 223 K).

Computational Results

A two-step mechanism for the reaction of 5 with 6 involves not only 5^•+ but also the radical 13 as a short-lived intermediate (Scheme 4). Further evidence for this mechanism is available from various quantum mechanical calculations. Mulliken population analysis³³ provides a means of assessing radical character as well as charge at any atomic center. In the case of radical character, the atomic spin density (S) is given by the difference in α and β electron density (D_α and D_β) at the atomic center of interest. By definition, α-electrons are assigned positive spin and β-electrons are assigned negative spin, such that the atomic spin density at a particular atomic center is given by eq 11. One unpaired electron on an atomic center is ideally given by S = 1, while a closed shell atom is ideally given by S = 0. However, atomic spin densities are often non-integer and deviate from these ideal numbers.

$$S=D_{\alpha }-D_{\beta }$$ (11)

For the radical 13, different resonance structures place the unpaired electron on C2, C4, or C6. Mulliken population analysis at the UB3LYP/6-311++G(3df,3pd)//UB3LYP/6-31G* level shows that the radical resides primarily para to nitrogen, at C4. In contrast, the positive charge in the pyridinium cation 6 resides primarily at C2 and C6 (Figure 2). Our results vary little across basis sets and regardless of whether implicit solvation is included. Our results are consistent with calculations performed as early as 1970 in which Hückel theory and SCF quantum calculations were used to show that the electron density of the 7-(π-electron) pyridine anion resides primarily at the 4 position.³⁴ Semi-empirical methods AM1 and MNDO have also been used to show that the kinetic regioselectivity for nucleophilic attack on the pyridinium ring is governed by the electron density at each carbon.⁸

FIGURE 2.

Calculated Atomic Spin Densities (13) and Charges (6) at the UB3LYP/6-311++G(3df,3pd)//UB3LYP/6-31G* level.

Single- and Multi-Step Mechanisms

The first explanation of the regioselectivity of nucleophilic attack on pyridinium cations^8,35 was offered by Kosower in 1956. He noted the formation of charge-transfer complexes between I⁻ and pyridinium cations,³⁶ and suggested that nucleophiles which could form such complexes added to the 4 position while other nucleophiles preferred the 2/6 position.³⁷ Later, Klopman suggested that the regioselectivity was the result of the hard/soft character of the nucleophile, with the total charge density (charge control) directing hard nucleophiles to the 2/6 position while the coefficient of C4 in the LUMO (frontier orbital control) directed soft nucleophiles to the 4 position.³⁸ Doddi, Ercolani, and Mencarelli have noted⁸ the frequent misuse of Klopman’s analysis to explain results that have arisen from thermodynamic control rather than from the kinetic control he assumed.

The regioselectivities we obtain for the transfer of H⁻ from Ru hydrides to 6 arise entirely from kinetic control. Our results suggest that, at least with Ru hydrides, the 1,2 and 1,4 products arise from the operation of different mechanisms. Mulliken population analysis confirms that the positive charge in the cation 6 resides predominantly at C2 and C6, whereas the spin density in the radical 13 resides predominantly at C4. A single-step H⁻ transfer is likely to be charge controlled and reduction at C2/C6 will be electronically favored, although steric factors may result in a mixture; the e⁻ transfer at the beginning of a multi-step mechanism will favor H• transfer to C4 (Scheme 5). As our Ru hydrides become better one-electron reductants, they give greater percentages of the 1,4 reduction product, until 5 gives only the 1,4 product.

SCHEME 5.

Other reductants that result exclusively in 1,4-dihydropyridines are ones we expect to be particularly good at single-electron transfer. Examples include sodium dithionite,⁷ the copper hydride (probably polynuclear) formed from Li(^tBuO)₃AlH and CuBr,⁵ and the formyl complex [Ru(bpy)₂(CO)(CHO)]⁺.³⁹

Indeed, the importance of single-step or multi-step mechanisms in determining the regiochemistry of nucleophilic attack on pyridinium cations is implied by much previous literature. In his 1995 review of the “Regioselectivity of the Reactions of Pyridinium … Salts with Various Nucleophiles”, Poddubnyi cited various calculations as predicting “the highest spin density for the γ[4] position” of the radicals formed by e⁻ transfer, and concluded that an “SET mechanism … gives rise to γ[4]-selectivity” while a “polar … mechanism … is characteristic of α[2]-selective addition”.³⁵

Experimental Section

General Procedures

All air-sensitive compounds were prepared and handled under an N₂/Ar atmosphere using standard Schlenk and inert-atmosphere box techniques. N-Carbophenoxypyridinium tetraphenylborate (6) was prepared by the method of King¹⁴ and recrystallized from CH₂Cl₂–Et₂O. N-Benzylidenepyrrolidinium tetrafluoroborate (11),⁴⁰ Cp*Ru(dppf)H (5),²⁶ CpRu(dppf)H (3),⁴¹ Cp*Ru(dppe)H (4),²⁹ and CpRu(dppe)H (2)⁴² were prepared by the literature methods. CD₂Cl₂ and CD₃CN were degassed and stored over 3 Å molecular sieves. CH₂Cl₂ was deoxygenated and dried by two successive columns (Q-5, activated alumina). THF was distilled from sodium/benzophenone under an N₂ atmosphere.

General Electrochemical Procedure

Cyclic voltammetry was performed with a BAS CV-50W Potentiostat. The supporting electrolyte for all solutions except the reference electrode was 0.10 M [Bu₄N]PF₆ in CH₂Cl₂. The cell consisted of a 1.6 mm diameter platinum disk working electrode, a platinum wire auxiliary electrode, and a silver wire reference electrode (0.01 M AgNO₃ + 0.10 M [Bu₄N]PF₆ in CH₃CN). The reference electrode was separated from the sample solutions with a porous Vycor tip (Bioanalytical Systems, MF-2042). Fc/Fc⁺ was used as an external reference and was found to be +0.22 V with respect to our reference electrode. All samples were prepared under an N₂/Ar atmosphere and further purged with Ar before measurement. Analyte concentrations were 0.001 M. Cyclic voltammograms of the hydride complexes (2–5) were recorded at 50 mV/s. Cyclic voltammograms of 6 and 11 were recorded at 200 mV/s. All potentials are reported in volts (V) vs. Fc/Fc⁺.

General Hydrogenation Procedure

CAUTION! Always shield pressurized vessels! Under an inert atmosphere, 6 (0.26 g, 0.50 mmol) and 5 (7.9 mg, 0.01 mmol) were combined in a Fischer-Porter bottle. THF or CH₂Cl₂ (10 mL) and 2,2,6,6-tetramethylpiperidine (TMP, 0.10 mL, 0.6 mmol) were added and the apparatus was charged with H₂ (80 psi). The reaction mixture was stirred rapidly for 24 h at room temperature. For lower temperature reactions, the sealed apparatus was cooled in a salt-water bath at 0 or 10 °C for 5 min prior to charging with H₂. The temperature was maintained for 24 h by placing the sealed apparatus, salt-water bath, magnetic stir plate, and shielding inside an appropriately set refrigerator. A 1 mL aliquot of the reaction mixture was evaporated and the residue dissolved in CD₂Cl₂. The percent conversion was determined by comparing the ¹H NMR integrations of the product peaks with that of 3.0 μL of added CH₃CN.

Variable Temperature NMR

Probe temperatures were calibrated with an ethylene glycol or methanol (99.97% MeOH + 0.03% HCl) chemical shift thermometer.^43,44

Cp*Ru(dppf)(H₂)⁺ (10a)

HBF₄·OMe₂ (0.01 mmol) and 400 μL of CD₂Cl₂ were added to a screw-cap NMR tube with a Teflon-coated septum insert. The NMR tube was cooled in an acetone/CO₂ bath while connected to an N₂ bubbler. Separately, 5 (0.01 mmol) was dissolved in 400 μL of CD₂Cl₂ and then added slowly to the cold NMR tube with a syringe. The tube was quickly shaken and inserted into a pre-cooled NMR probe. ¹H NMR (300 MHz, 195.2 K, CD₂Cl₂): δ −8.12 (s, br, Ru(H₂), 2H), 1.21 (s, Cp*, 15H), 4.11 (s, dppf Cp, 2H), 4.22 (s, dppf Cp, 2H), 4.29 (s, dppf Cp, 2H), 4.49 (s, dppf Cp, 2H), 7.30–7.70 (m, Ar, 20H). ³¹P {¹H} NMR (121.5 MHz, 195.2 K, CD₂Cl₂): δ 55.09. T₁ measurements (300 MHz) of the dihydrogen resonance: T₁ = 12.6(2) ms, 195.2 K; 11.5(2) ms, 218.5 K; 12.8(2) ms, 238.9 K.

trans-Cp*Ru(dppf)(H)₂⁺ (10b)

The title dihydride complex²⁷ may be prepared by treating 5 with HBF₄·OMe₂ at room temperature, or by warming a solution of 10a to room temperature. ¹H NMR (300 MHz, 279.5 K, CD₂Cl₂): δ −7.80 (t, Ru(H)₂, J_P–H = 25.8 Hz, 2H), 1.24 (s, Cp*, 15H), 4.19 (s, dppf Cp, 4H), 4.21 (s, dppf Cp, 4H), 7.58–7.66 (m, Ar, 12H), 7.85–7.96 (m, Ar, 8H). ³¹P {¹H} NMR (121.5 MHz, 279.5 K, CD₂Cl₂): δ 58.33. T₁ measurements (300 MHz) of the dihydride resonance: T₁ = 0.224(4) s, 178.0 K; 0.151(2) s, 195.2 K; 0.223(5) s, 218.5 K; 0.253(5) s, 238.9 K; 0.323(5) s, 258.9 K; 0.439(7) s, 279.5 K.

Isomerization Kinetics

A solution of 10a (0.04 M in CD₂Cl₂) was prepared as described above and inserted into an NMR probe pre-cooled to 0 °C. The reaction was followed by the integration of the dihydrogen complex peak at δ 4.59 (s, dppf Cp, 2H) in comparison with the integration of the residual solvent peak at δ 5.32. The average of three experiments gave a first order rate constant of 1.32(2) × 10⁻³ s⁻¹ for the disappearance of 10a.

N-Carbophenoxy-1,2-dihydropyridine (9a) as prepared by the literature method⁴⁵ was contaminated with 6% of the isomeric 1,4-dihydropyridine (9b). Two conformers of 9a (in a 3/4 ratio) were observed at room temperature. ¹H NMR (300 MHz, 298 K, CD₃CN): δ 4.36 (s, CH₂, major), 4.56 (s, CH₂, minor), 5.29 (m, 1H), 5.62 (m, 1H), 5.90 (m, 1H), 6.65–6.95 (m, N–CH, 1H), 7.15 (m, Ar, 2H), 7.26 (m, Ar, 1H), 7.41 (m, Ar, 2H). An averaged spectrum was observed at 340 K: δ 4.46 (s, br, CH₂, 2H), 5.31 (m, 1H), 5.64 (m, 1H), 5.92 (m, 1H), 6.83 (m, N–CH, 1H), 7.15 (m, Ar, 2H), 7.26 (m, Ar, 1H), 7.41 (m, Ar, 2H).

N-Carbophenoxy-1,4-dihydropyridine (9b).⁵

The hydrogenation of 0.50 mmol of 6 (see General Hydrogenation Procedure) in 10 mL of THF at 10 °C for 24 h gave a yellow solution. An aliquot (1 mL) was evaporated and ¹H NMR (in CD₂Cl₂ + 3.0 μL CH₃CN internal standard) indicated 86% conversion to 9b. The remainder of the reaction solution was evaporated to give a yellow residue. The residue was extracted with 4 × 5 mL of Et₂O. The Et₂O solution was washed with 1 M NH₄Cl (2 × 10 mL), satd NaHCO₃ (2 × 10 mL), dried over MgSO₄, and evaporated to give a yellow solid. The solid was dissolved in an 8/2 mixture of hexanes/Et₂O and loaded on a flash column (230–400 mesh silica, 14 cm × 1 cm diameter). The product was eluted with an 8/2 mixture of hexanes/Et₂O. Evaporation of the solvent gave the product, a white crystalline solid, in 75% yield (corrected for the aliquot removed, and with respect to 6). ¹H NMR (300 MHz, 298 K, CD₃CN): δ 2.87 (m, CH₂, 2H), 5.02 (br, β-CH, 1H), 5.07 (br, β-CH, 1H), 6.76 (d, α-CH, J = 7.8 Hz, 1H), 6.91 (d, α-CH, J = 8.7 Hz, 1H), 7.17 (m, Ar, 2H), 7.27 (m, Ar, 1H), 7.42 (m, Ar, 2H). An averaged spectrum (due to fast rotation about the N–C(O) bond) was observed at 340 K: δ 2.88 (m, CH₂, 2H), 5.06 (br, β-CH, 2H), 6.84 (br, α-CH, 2H), 7.19 (m, Ar, 2H), 7.27 (m, Ar, 1H), 7.42 (m, Ar, 2H). FAB⁺ MS (m-NBA): m/z for [M+1]⁺ Calcd 202.0868; Found 202.0867.

Stoichiometric Hydride Transfer from Ruthenium Hydrides to 6 in CD₂Cl₂

The Ru hydride (0.02 mmol) and 6 (0.02 mmol) were dissolved in 700 μL of CD₂Cl₂ at room temperature. Then, CH₃CN (0.08 mmol) was added and the product ratio was measured by ¹H NMR integration.

Stoichiometric Hydride Transfer from Ruthenium Hydrides to 6 in CD₃CN

The Ru hydride (0.02 mmol) and 6 (0.02 mmol) were dissolved in 700 μL of CD₃CN at room temperature; the product ratio was measured by ¹H NMR integration. The hydride complexes 3 and 5 are sparingly soluble in CD₃CN. For reactions with 3 or 5, the NMR tube was heated in a 75 °C oil bath and shaken to mix the contents every 20 min. Once most of the yellow 3 or 5 had dissolved (about 2 h), the tube was removed from the bath and the product ratio determined by ¹H NMR integration.

Stoichiometric Hydride Transfer from Ruthenium Hydrides to 11 in CD₂Cl₂

The Ru hydride (0.02 mmol) and 11 (0.02 mmol) were dissolved in 700 μL of CD₂Cl₂ at room temperature. Then, CH₃CN (0.08 mmol) was added and the ¹H NMR spectrum was recorded. Complexes 2, 3, and 4 transferred hydride to 11, giving the tertiary amine (12). In these cases, the CD₂Cl₂ solution was evaporated, the residue extracted with Et₂O, the Et₂O solution evaporated, and the resulting residue dissolved in CDCl₃. The ¹H NMR spectra matched the reported spectrum of 12.⁴⁶ The hydride complex 5 did not react with 11 in CD₂Cl₂.

Low Temperature Reaction of 5 with 6 (¹H NMR)

CH₃CN (0.08 mmol), 6 (0.02 mmol), and 550 μL of CD₂Cl₂ were added to a screw-cap NMR tube with a Teflon-coated septum insert. The NMR tube was cooled in an acetone/CO₂ bath while connected to a N₂ bubbler. Separately, 5 (0.02 mmol) was dissolved in 250 μL of CD₂Cl₂ and then added slowly to the cold NMR tube with a syringe. The tube was quickly shaken and inserted into a pre-cooled NMR probe. The reaction was monitored by 300 MHz ¹H NMR at 228.0 K for 1 h, during which time the only dihydropyridine product formed was 9b.

Low Temperature Reaction of 5 with 6 (EPR)

An EPR tube (quartz, 3 mm) was charged with 6 (200 μL, 0.02 M in CH₂Cl₂). The tube was sealed with a septum and cooled in a hexanes/CO₂ bath at 223 K while connected to a N₂ bubbler. Separately, 5 (200 μL, 0.02 M in CH₂Cl₂) was slowly added to the cold EPR tube with a syringe. The mixture, initially orange, became green after 10 min and was then quenched in liquid N₂. The X-band EPR spectrum was obtained at 77 K with a Bruker EMX EPR spectrometer with a TE₁₀₂ rectangular cavity.

Computational Methods

Both the radical (13) and the pyridinium cation (6) were subjected to conformational searching using Macromodel 6.0⁴⁷ and the OPLS 2001 force field.⁴⁸ The lowest energy structures were subsequently minimized at the DFT-UB3LYP/6-31G* level^49–51 in both vacuum and implicit solvent (dichloromethane) using Jaguar 7.0.⁵² Single point calculations were also performed at the UB3LYP/6-311++G(3df,3pd)//UB3LYP/6-31G* level. Spin densities and atomic charges were determined by Mulliken population analysis.³³

Supplementary Material

2. Supporting Information Available.

lists of ¹H NMR peaks for 6 and 11, ¹H NMR spectra of 9a and 9b (298 and 340 K) and 10a (195.2 K), a ¹H NMR spectrum from (and detailed comments on) the low temperature reaction of 5 with 6, kinetic data for 10a → 10b, computational details for 6 and 13, and cyclic voltammograms of 2, 3, 4, 5, 6, and 11. This material is available free of charge via the Internet at http://pubs.acs.org.