Diazaphosphinanes as hydride, hydrogen atom, proton or electron donors under transition-metal-free conditions: thermodynamics, kinetics, and synthetic applications.

Exploration of new hydrogen donors is in large demand in hydrogenation chemistry. Herein, we developed a new 1,3,2-diazaphosphinane 1a, which can serve as a hydride, hydrogen atom or proton donor without transition-metal mediation. The thermodynamics and kinetics of these three pathways of 1a, together with those of its analog 1b, were investigated in acetonitrile. It is noteworthy that, the reduction potentials (E red) of the phosphenium cations 1a-[P]+ and 1b-[P]+ are extremely low, being -1.94 and -2.39 V (vs. Fc+/0), respectively, enabling corresponding phosphinyl radicals to function as neutral super-electron-donors. Kinetic studies revealed an extraordinarily large kinetic isotope effect KIE(1a) of 31.3 for the hydrogen atom transfer from 1a to the 2,4,6-tri-(tert-butyl)-phenoxyl radical, implying a tunneling effect. Furthermore, successful applications of these diverse P-H bond energetic parameters in organic syntheses were exemplified, shedding light on more exploitations of these versatile and powerful diazaphosphinane reagents in organic chemistry. This journal is © The Royal Society of Chemistry.

Introduction

class="Chemical">Hydrogen transfer class="Chemical">plays a very imclass="Chemical">portant role in chemical science and related fields. This class="Chemical">process occurs through the cleavage of the targeted R–H bond in three class="Chemical">possible class="Chemical">pathways (Scheme 1), i.e., class="Chemical">pan class="Chemical">hydride (H−),[1] hydrogen atom (H˙)[2] and proton (H+)[3] transfers, which, from a thermodynamic aspect, depends on the R–H bond strength and the nature of the atom bound to the transferred hydrogen atom. Accordingly, knowledge of the relevant thermodynamics of hydricity ΔGH,[1] bond-dissociation free-energy, BDFE[5] and acidity, pKa[3] as well as the kinetics[7] relevant to these processes would facilitate rational exploitations of new transformations. In recent years, considerable attention has been paid to the development of new hydrogen sources with versatile hydrogen donor reactivities, with particular interests in transition-metal hydrides M–H,[1] such as hydrogenase enzyme analogs[8] and hydrogenation catalysts,[9] and main-group organic hydrides X–H,[1] such as nicotinamide coenzyme models[10] and Hantzsch esters,[11] and so on.

Scheme 1

Possible pathways of hydrogen transfers for transition-metal and transition-metal-free systems as well as corresponding thermodynamic driving forces in acetonitrile.

It is well known that class="Chemical">metal hydrides (M–H) could serve as H−, H˙ and H+ donors (Scheme 1a).[1] Their diverse class="Chemical">pan class="Chemical">hydrogen reactivities originate from the readily modulated polarity of M–H bonds through electronic communication between the metal centers and coordinated ligands. Compared to an M–H system, the X–H hydride manifested some advantages in mild reaction conditions, good functional-group compatibility, easy modification, etc. Integration of these three dissociation possibilities into one X–H covalent bond is a substantial challenge, due to the great disparity in the electronegativity of the X and H fragments. Nevertheless, Waymouth et al. very recently reported an excellent new example of such a X–H bond where the N–H bond of the phenylazopyridine ligand in the Co(i) complex can serve as either a H−, H˙ or a H+ donor (Scheme 1b).[13] It is noted, however, that the diverse reactivity of this N–H bond may require latent electronic communication between the azopyridine ligand and cobalt atom within the five-membered ring, so it may be viewed as a quasi-M–H system.

For a true transition-class="Chemical">metal-free system, till now only class="Chemical">pan class="Chemical">triarylmethane analogs, where there exist steric as well as resonance stabilization effects on the incipient radical or charges, were reported to have α-C–H bond-cleavage energies determined for all three pathways (Scheme 1c).[14] However, their poor hydrogen donability and severe steric hindrance make these dissociation processes not very useful for synthesis. This, in turn, stimulated our interest to find more useful transition-metal-free X–H systems that promise both the desired energetic measurement and new synthetic applications.

Recently, class="Chemical">diazaphospholenium hydrides (class="Chemical">pan class="Chemical">P–H) have attracted substantial research interest due to their superior hydricity endowed by the unique diazaphospholene skeleton.[15] Many diazaphospholenes were developed and successfully used in various hydridic reductions.[16] However, the superior hydricity resulted in their other promising applications, such as serving as hydrogen atom and proton donors, as well as the precursors of strong electron donors, being greatly overlooked. Considering the similar electronegativity of the hydrogen (χAR = 2.20) and phosphorus (χAR = 2.06) atoms,[15] we reckoned that the P–H bond, beside being a good hydride donor,[17] may have the potential to release proton and hydrogen atoms[18] as well, by fine-tuning the electrical properties of the P fragment. And indeed luckily enough, this was realized in the present work.

Herein, we reported a new class="Chemical">N-heterocyclic phosphine 1a, which could act as a H−, H˙ and H+ donor (Scheme 2). Thermodynamics and kinetics class="Chemical">pertinent to these three class="Chemical">processes were examined in detail. For comclass="Chemical">parison, class="Chemical">pan class="Chemical">1b was investigated as well. Based on the P–H bond energetic studies, we also exploited their versatile applications in the reduction of pyridines, synthesis of bisphosphines, H/D exchange, and activation of carbon–halogen bonds as original examples.

Scheme 2

P–H reagents 1a and 1b serving as diverse hydrogen donors. A+, B− and C˙ are the hydride, proton and hydrogen atom acceptors, respectively.

Results and discussion

Synthesis

The substrate 1a was prepared via a H/Cl exchange between the corresponding class="Chemical">phosphoric chloride and class="Chemical">pan class="Chemical">LiAlH4 in THF according to the literature[15,17] (see the ESI† for details). LiAlD4 was used for deriving the deuterated substrates 1a-D and 1b-D. The crystal structure of 1a was shown in Fig. 1 and features an envelope configuration with the P atom standing out of the naphthalene plane and the P–H bond adopting a flagpole position, similar to the five-membered ring analogs disclosed previously.[15] And a P–H distance of 1.42(2) Å is slightly shorter than that of the five-membered ring analog.[15]31P NMR spectra of 1a in CD3CN show a dt peak at 25.66 ppm that is split by the hydrogen atoms on the P atom and isopropyl groups. It moves to the higher magnetic field in comparison with other N-heterocyclic phosphines.[17] Under an inert atmosphere, 1a is stable enough for over 12 hours in tetrahydrofuran, acetonitrile, alcohols, toluene, dimethylsulfoxide and chloroform. On the other hand, 1b is stable in these solvents as well except in alcohols and chlorinated hydrocarbons, where it decomposes rapidly to yield phosphonites (RO)2PH[19] or 2-chloro-1,3,2-diazaphosphinane [P]-Cl (Scheme 3), showing that 1b is more air- and moisture-sensitive than 1a.

Fig. 1

The crystal structure of 1a (50% probability thermal ellipsoids).[20] Selected bond lengths (in Å): P1–H1 1.42(2), P1–N1 1.747(6), N1–C1 1.329(9), N1–C7 1.440(8), and C1–C6 1.413(8).

Scheme 3

Decomposition of 1b in alcohols and chlorinated hydrocarbons. [P] = (CH2)3(NBu)2P, R = Me or Et.

Redox properties

The oxidation potentials of 1a and class="Chemical">1b were examined by cyclic voltammetry (CV), exhibiting a class="Chemical">partially reversible oxidation class="Chemical">peak of 1a to 1a˙+ at 0.23 V (vs. Fc+/0, Fig. S1a†) and an irreversible oxidation wave of class="Chemical">pan class="Chemical">1b at 0.47 V (vs. Fc+/0, Fig. S1b†) in acetonitrile. The low oxidation potentials of 1a and 1b indicate their good reducing capacity. It is interesting to find that 1a is a better electron donor, in spite of being a far poorer hydride donor than 1b (vide infra). Besides, the redox behaviors of the respective phosphinyl radicals (1a-[P]˙ and 1b-[P]˙) were also examined through the reduction of the corresponding phosphenium cations (1a-[P]+ [21] and 1b-[P]+), resulting in the CV peaks at −1.94 V and −2.39 V (vs. Fc+/0) in acetonitrile, respectively (Fig. S1c and 1d†). The extremely low potentials, especially for 1b-[P]+, indicate the very high reducing capacity of the corresponding 1a-[P]˙ and 1b-[P]˙ radicals. This suggests that they both have a very promising potential to serve as a super electron donor in organic synthesis to reduce aryl halides.[22] And indeed, this was verified in the tentative application of the present work in radical hydrodebromination of bromobenzene (see Application part, vide infra).

Thermodynamic driving force and reactivity

Hydride transfer (HT)

class="Chemical">Hydride transfer from 1a or class="Chemical">pan class="Chemical">1b [17] to N-methylacridinium iodide A1+ yielded equimolar products (Fig. S3†). To exclude the suspected interruption of the hydridic isopropyl α-C–H, the deuterated substrate 1a-D was employed, and over 95% deuterated A1D was derived (eqn (1) and Fig. S4†). This confirmed that the P–H hydride of 1a is a stronger donor than that of the isopropyl α-C–H.

The hydricity (ΔGH, defined by eqn (2)) of 1a and class="Chemical">1b in class="Chemical">pan class="Chemical">acetonitrile was determined by measuring their equilibrium constants with the acceptors of known hydricities (eqn (3)). For example, the measured hydride transfer equilibrium constant Keq(1a) of 0.48 (equivalent to 0.43 kcal mol−1) between 1a and phenanthridinuim trifluoromethanesulfonate A2+ (ΔGH(A2H) = 61.4 kcal mol−1)[1] provided the hydricity ΔGH of 1a to be 61.8 kcal mol−1 (see Fig. S5 in the ESI† for details) and other equilibria (in Fig. S6 and S7†) verified this value, giving a mean hydricity of 62.2 ± 1.0 kcal mol−1. This demonstrates 1a to be a moderate hydride donor, similar to the hydride-donating ability of the NADH analogs.[1] Thus, it should be capable of reducing commonly used hydride acceptors, such as N-methylacridinium cations (ΔGH = 76 kcal mol−1), 9-phenylxanthylium cations (89 kcal mol−1) and trityl cations (99 kcal mol−1). Similarly, a much lower equilibrium constant Keq (1b) of 1.6 × 10−3 (equivalent to 3.8 kcal mol−1, eqn (5) and Fig. S8†) established between 1b and benzimidazolium perchlorate A3+ (ΔGH(A3H) = 45.0 kcal mol−1)[4] resulted in ΔGH(1b) being 48.8 ± 1.0 kcal mol−1, which is comparable to that of the commonly used good hydride donor, 2,3-dihydrobenzo[d]imidazoles.[1] The good hydridic reactivity of 1b was illustrated in our recent work in the reduction of various unsaturated compounds.[17] It is interesting to note that 1b is about 13 kcal mol−1 stronger than 1a in hydricity, but is 0.24 V weaker in electron-donating ability.

The hydricity can be used to rationalize the disparate reactivity between 1a and some reported class="Chemical">diazaphospholenium hydrides. The latter comclass="Chemical">pounds were reclass="Chemical">ported to react with strong acid class="Chemical">pan class="Chemical">HBF4 to give H2 and the corresponding phosphenium cations.[15] However, when 1a was treated with HBF4 (or HOTf), no gaseous product was generated. Instead, a new phosphorus species with a doublet peak at about 80 ppm was detected in the 31P NMR spectrum (Fig. S9a and S9b†). To identify this newly formed species, a base pyridine was added to the reaction mixture, which rendered 1a the primary product with a small amount of unidentified impurities (Fig. S9c†). Consequently, this mysterious P species was speculated to be protonated 1a (1aH+, Scheme 4). Based on the doublet peak of the phosphorus atom in 31P NMR spectra (if the extraneous proton was connected to the P atom, a triplet 31P peak would be observed) and the unsymmetrical peaks of naphthyl and isopropyl hydrogen atoms in 1H NMR spectra (Fig. S10†), the protonated site was thus assigned to the N atom (Scheme 4). We thought that their dissimilarity in reactivity between 1a and some reported diazaphospholenium hydrides should originate from the relatively weak hydricity of 1a (ΔGH = 62.2 kcal mol−1). As a consequence, it makes the H2 release unable to compete with its protonation even under the condition of heating (80 °C for 5 hours). The moderate hydricity of 1a endows it with good tolerance to even strong acids, avoiding direct elimination of dihydrogen. This in turn offers a possibility for 1a to act as a hydride donor in acid-catalyzed reduction. Lots of examples in this connection[1] can be found, especially in the hydrogenation of imines by hydride donors with comparable hydricity to 1a (∼60–70 kcal mol−1, like Hantzsch ester,[23] benzothiazoline[24] and cyclohexadiene[25]) under the catalysis of Brønsted acids (for examples BINOL-phosphoric acid, trifluoroacetic acid and Tf2NH).

Scheme 4

Protonation of 1a by HBF4 and recovery of it by adding pyridine in CD3CN.

The kinetics of the class="Chemical">hydride transfer from 1a to A1+ was conducted by following the decay of the absorclass="Chemical">ption of A1+ in class="Chemical">pan class="Chemical">CH3CN at 430 nm with a stopped-flow spectrophotometer (Fig. 2, see the ESI† for details). The second-order rate constant kHT(1a) was found to be 5.94 M−1 s−1 (Inset in Fig. 2 and Table S1†). Moreover, the nucleophilicity of 1a can be estimated by the second-order rate constant of the reaction of 1a with A1+. Because 1a has a similar structure with 1b, based on the nucleophile-specific sensitivity parameter sN of 1b (0.52) and the electrophilicity of A1+ in our recent work,[17] the nucleophilicity N, of 1a is estimated to be 8.64. Similarly, the kinetics of 1a-D with A1+kHT(1a-D) was determined to be 2.34 M−1 s−1 (Table S2†). These gave a primary kinetic isotope effect KIE [kHT(1a)/kHT(1a-D)] of 2.5. A comparison with the previously determined kHT(1b) of 5.41 × 102 M−1 s−1 for the reaction of 1b with A1+ in our earlier work[17] indicates that 1b is about 100 times more reactive than 1a. This is exactly in accordance with the order of their hydricities (ΔGH) of 1avs.1b found here.

Fig. 2

Monoexponential decay of the absorbance Abs (at 430 nm) with the time t (s) for the reaction of 1a (3.03 × 10−3 M) with A1+ (5.00 × 10−5 M) in CH3CN at 20 °C. Inset: Correlation of kobs with [1a].

Hydrogen atom transfer (HAT)

To examine the possibility of a HAT reaction from class="Chemical">diazaphosphinane substrates 1, the reactions of 1a and class="Chemical">pan class="Chemical">1b with the 2,4,6-tri-tert-butyl-phenoxyl radical O˙ were conducted, and yielded 2,4,6-tri-tert-butylphenol OH and the corresponding phosphorus species (eqn (6)) without generation of bisphosphines (compared to the reaction of 1a or 1b with AIBN, Table 1). A stoichiometric study found a [1]/[O˙] ratio of 1 : 2, consistent with the product analysis (Fig. S11 and S12†).

Applications of 1a and 1b in organic syntheses

Entry	Reactant	Condition	Product	Yield	Type
(1)		1a-[P]+ (15 mol%), HBpin (1.5 equiv.), CH3CN, 80 °C, 36 h		40%a	HT
(2)		tBuOK, CD3CN, 20 °C, 10 min		Quant.b	PT
(3)		AIBN (1.5 equiv.), C6D6, 80 °C, 3 h		>90%b	HAT
(4)		1b (1.5 equiv.), AIBN 15 mol%, toluene-d8, 90 °C, 5 h		>90%c	HAT&ET

Isolated yield.

NMR yields determined by the amount of phosphorus species.

NMR yields with 1,3,5-trimethoxybenzene as the internal standard.

NMR yields determined by the amount of pan class="Chemical">phosphorus sclass="Chemical">pecies.

NMR yields with pan class="Chemical">1,3,5-trimethoxybenzene as the internal standard.

To evaluate the class="Chemical">hydrogen atom donability of class="Chemical">pan class="Chemical">diazaphosphinanes (i.e., P–H BDFE), the thermodynamic cycle was applied on the basis of the ΔGH and Ered([P]+) in hand (Scheme 5 and eqn (7); FEox(H−) is −26.0 kcal mol−1[1] in acetonitrile vs. Fc0/+).[26] Substituting the known values into eqn (7) [26] gives P–H BDFEs of 80.9 ± 2.0 and 77.9 ± 2.0 kcal mol−1 for 1a and 1b, respectively, being comparable with those of phenol and thiophenol antioxidants. Therefore, they could transfer hydrogen atoms smoothly to acceptors with X–H BDFEs around or larger than 78 kcal mol−1, such as 2,4,6-tri-tert-butyl-phenoxyl O˙ (77.1 kcal mol−1), CN(CH3)2C˙ (about 88 kcal mol−1) and phenyl radicals (104.7 kcal mol−1).[5] Compared to 1b, though the fused aryl groups attenuate the hydricity of 1a by 13 kcal mol−1 thermodynamically, they show a negligible effect on the HAT reactivity.

Scheme 5

Thermodynamic cycle for deriving the P–H BDFEs of 1a and 1b.

The reaction mechanism of 1a with radical O˙ (eqn (6)) is proposed in Scheme 6 based on the derived thermodynamic data (the reaction of class="Chemical">1b with O˙ is similar). Since class="Chemical">pan class="Chemical">hydrogen atom transfer from 1a to O˙ is slightly endothermic (∼4 kcal mol−1), it implies the possibility for a reversible HAT. This reversible HAT could be further driven to the far right by the largely exothermic electron transfer, ET of 1a-[P]˙ radical (Eox(1a-[P]˙) = −1.94 V) or a follow-up radical coupling. The observed stoichiometric ratio of 1 : 2 for the reactants [1] vs. [O˙] is in exact accordance with the HAT mechanism proposed. On the other hand, a suspected proton transfer between 1a and the O˙ radical can be easily ruled out due to the very weak acidity of 1a (vide infra) and strong acidity (pKa of ∼−3)[5] reported for the phenol radical cations (eqn (8)). The other suspected possibility, i.e. the ET path between 1a and the O˙ radical (Ered = −0.70 V [5]) can also be excluded by a similar strategy on the basis of their respective thermodynamics determined, that indicates that the ET from 1a to O˙ is remarkably uphill (ΔGET = 21.4 kcal mol−1, eqn (9)).[13,29] It is thus safe to conclude that the initial HAT should be the rate-determining step (RDS) for the reaction expressed in eqn (6).

Scheme 6

The possible mechanism for the reaction of 1a with O˙. Ar = 2,4,6-tri-tert-butyl-phenyl group.

The rate constants of this class="Chemical">hydrogen atom transfer from 1a and 1a-D to radical O˙ were measured to be a kHAT(1a) of 0.70 M−1 s−1 and a kHAT(1a-D) of 2.24 × 10−2 M−1 s−1, resclass="Chemical">pectively (Tables S3 and S4†). This intermediately class="Chemical">points out an extremely large KIE(1a) of 31.3, imclass="Chemical">plying a tunneling effect in this HAT class="Chemical">process. Such a large KIE has been found in biochemical[30] and transition-class="Chemical">pan class="Chemical">metal systems,[31] but seldom observed in organic HAT processes. Similarly, the rate of the analogous HAT of 1bkHAT(1b) was determined to be 0.74 M−1 s−1 (Table S5†). The quite close kHAT for 1a and 1b agrees well with their comparable P–H BDFEs (80.9 vs. 77.9 kcal mol−1, vide supra), verifying their similar HAT reactivity. However, the same kinetic study for 1b-D gave a kHAT(1b-D) of 0.12 M−1 s−1 (Table S6†), resulting in a primary KIE(1b) of 6.2 with no tunneling phenomenon detected.

To understand this discrepancy, temperature-dependent kinetics were further performed (Tables S7–S10†). The ratio of Arrhenius pre-factors [A(1a)/A(1a-D)] was obtained to be 0.021, and the activation energy difference [Ea(1a) − Ea(1a-D)] was −5.53 kcal mol−1 (Fig. S2 and Table S11†). These kinetic characteristics [A(1a)/A(1a-D) ≪ 1 and Ea(1a) − Ea(1a-D) ≪ 0] are in fact in line with the tunneling model illustrated by Klinman,[30] which features a “through-the-barrier” transition state near the top of the classical barrier. In the class="Chemical">1b system, on the other hand, the normal class="Chemical">primary KIE of 6.2, A(class="Chemical">pan class="Chemical">1b)/A(1b-D) of 3.39 and [Ea(1b) − Ea(1b-D)] of −0.38 kcal mol−1 all support a classical “over-the-barrier” transition state. A more comprehensive investigation of the tunneling issue is presently underway.

Proton transfer (PT)

It's known that a good class="Chemical">hydride donor usually exhibits class="Chemical">poor acidity. This exclass="Chemical">plains why the acidity of widely-used X–H tyclass="Chemical">pe class="Chemical">pan class="Chemical">hydride donors, e.g. 2,3-dihydrobenzo[d]-imidazoles, has never been determined in the absence of metal-coordination. The hydricity of the present N-heterocyclic phosphines is also quite good (vide supra), hence an investigation of their acidic reactivity cannot be expected to be too straightforward. Nevertheless, as we learned from the initial failures in deprotonating 1a or 1b with some ordinary strong bases, such as 1,8-diazabicyclo[5,4,0]-7-undecene (DBU, pKa = 24.34 in CH3CN), 1,3,4,6,7,8-hexahydro-2H-pyrimido[1,2-a]pyrimidine (TBD, pKa = 26.03) and (tert-butylimino)tris(pyrrolidino)-phosphorane (BTPP, pKa = 28.42)[28] (see the ESI† for details), we finally found that the H/D exchange of the 1a P–H proton in the presence of stoichiometric BuOK in CD3CN (eqn (10) and Fig. S13†) was accomplished in about 10 minutes at room temperature,[32] whereas the same operation for 1a-D in CH3CN was finished in 8 hours at room temperature (eqn (11) and Fig. S14†). The reversible H/D exchange of 1a in acetonitrile under the promotion of BuOK suggested that the acidity of 1a reaches the measurable acidity limitation in acetonitrile (see the ESI† for details).[33] As for 1b with much stronger hydricity, its acidity is, unfortunately, too weak to examine in solution by any existing experiment. However, it is worth noting that the acidity of a P–H bond can be significantly enhanced upon activation by coordination with transition metals[34] or other Lewis acids.[35]

Applications in syntheses

Based on the above thermodynamic and kinetic investigations, the potential of applying 1a and pan class="Chemical">1b in organic syntheses, encomclass="Chemical">passing the diverse class="Chemical">pan class="Chemical">hydrogen (H+, H˙ and H−) or electron transfer, was preliminarily explored and showcased herein.

Catalytic reduction of class="Chemical">pyridines is a class="Chemical">prevailing strategy for the synthesis of class="Chemical">pan class="Chemical">dihydropyridines. Recently, Kinjo et al.[16]and Speed et al.[36] exploited 1,3,2-diazaphosphenium-catalyzed hydroboration of pyridines with good regio- and chemo-selectivity.[16] However, in their system, the substrate 3-CN-pyridine gave a mixture of three dihydropyridine isomers. To our delight, we found that under the catalysis of 15 mol% of our 1a-[P]+,[37] the sole 1,4-hydroborated isomer of this reaction was selectively produced in a moderate yield of 40% (entry 1 in Table 1 and Fig. S15†). The good regio-selectivity may be ascribed to the steric factors of the isopropyl group on the nitrogen atom of 1a based on Kinjo's DFT calculations.[16] This good regio-selectivity exhibits the catalytic potential of 1a in the reduction of unsaturated compounds.

Moreover, the acidic character of 1a (class="Chemical">vide suclass="Chemical">pra) offers a more economic method for the synthesis of class="Chemical">pan class="Chemical">deuterated reagent 1a-D (entry 2 in Table 1) by employing the much cheaper deuterated solvent CD3CN instead of expensive LiAlD4 in an almost quantitative yield (see the ESI† for details about the synthetic method). Meanwhile, it also provides a new approach for isotope labeling of many other hydridic species. The generation of the hydridic hydrogen from a “protic” solvent as illustrated here may serve as an example of Umpolung which was also proposed by Gudat et al. for the P–H bond.[15]

Since class="Chemical">bisphosphines can easily decomclass="Chemical">pose into class="Chemical">pan class="Chemical">phosphinyl radicals,[38] they could serve as radical reservoirs for relevant studies and applications. Based on thermodynamic analysis, CN(CH3)2C˙ (C–H BDFE ≈ 85 kcal mol−1), generated from the homolysis of azo-bis-isobutyronitrile (AIBN), could abstract the hydrogen atoms from 1a (P–H BDFE = 80.9 kcal mol−1). Then, the generated phosphinyl radical 1a-[P]˙ coupled quickly with each other to yield bisphosphines (entry 3 in Table 1, Fig. S16† for 1a and Fig. S17† for 1b). The targeted dimers[39] are readily extracted from reaction mixtures in a high yield. This transformation, together with Gudat's photochemical dehydrocoupling,[18] offers easy access to bisphosphines, avoiding complicated and harsh operations and workups required in other synthetic processes using metals such as sodium[38] and magnesium[38].

Furthermore, we attempted to design a new radical transformation by an integrated use of class="Chemical">hydrogen atoms and electron transfers from class="Chemical">pan class="Chemical">N-heterocyclic phosphines. This was inspired by the reactions of AIBN with 1a and 1b, where phosphinyl radicals can be easily generated from their P–H precursors through hydrogen atom release. The highly negative oxidation potentials of phosphinyl radicals imply their competency in the reduction of challenging substrates. If a substrate in the reaction system is sufficiently oxidative, its single-electron capture from phosphinyl radicals may be able to compete with radical coupling. In such cases, subsequent transformations may be triggered. Based on the redox potentials of 1b-[P]˙ (Eox = −2.39 V vs. Fc+/0, the same with the Ered(1b-[P]+)) and bromobenzene (Ered = −2.8 V),[40] hydrodehalogenation of bromobenzene by 1b was performed in toluene with 15 mol% AIBN as the initiator (entry 4 in Table 1, Fig. S18†). After 5 hours, the expected reduction product, benzene was obtained in over 90% yield. Due to the poor reduction ability of 1a-[P]˙, the reaction could not proceed when 1a was used.[41] Therefore, it could be anticipated that these N-heterocyclic phosphines may open up a promising avenue for the development of super electron donors.

Conclusions

class="Chemical">Hydrogen and electron transfers from class="Chemical">pan class="Chemical">phosphines 1a and 1b were thermodynamically and kinetically investigated in acetonitrile. The P–H bond of 1a can participate in reactions involving formal transfer of a hydride/hydrogen atom/proton to a substrate under respective conditions in the absence of the usually required transition-metal- or Lewis acid-mediation. It was observed that 1a is a moderate hydride and hydrogen atom donor, but a poor Brønsted acid. In comparison, 1b is a good hydride donor and a moderate hydrogen atom donor; but did not show a detectable acidic reactivity even under the strongest basic solution. With regard to electron donability, 1a (Eox(1a) = 0.23 V) is superior to 1b (Eox(1b) = 0.47 V), contrasting with the order of their hydridic reactivity. Their corresponding phosphinyl radicals are very strong electron donors as reflected by their extremely negative redox potentials. Kinetic studies revealed a tunneling contribution (KIE = 31.3) to the hydrogen atom transfer from 1a to the phenoxy radical. Based on the derived energetic parameters, their synthetic applications were tentatively exploited. Preliminary results from the synthesis attempts by using the new reagents 1a and 1b confirmed their competence as effective hydrogen and electron donors in various redox transformations. The effectiveness of applying the relevant physicochemical parameters in the analysis and design of hydrogen transfer reactions was also shown.

Conflicts of interest

There are no conflicts to declare.