Borane B-C Bond Cleavage by a Low-Coordinate Iron Hydride Complex and N-N Bond Cleavage by the Hydridoborate Product

Abstract

The iron(II) hydride dimers [L^RFe(μ-H)₂FeL^R] (L^Me = 2,4-bis(2,6-diisopropylphenylimino) pent-3-yl; L^tBu = 2,2,6,6-tetramethyl-3,5-bis(2,6-diisopropylphenylimino)hept-4-yl) abstract hydrocarbyl groups from BR′₃ (R′ = Et, Ph) to give L^RFeR′ and L^RFe(μ-H)₂BR′₂. Mechanistic studies with R = R′ = Me are consistent with a process in which the hyride dimer opens one Fe-H bond, and subsequent B-H bond formation is concerted with dissociation of an Fe-H unit. Cleavage of boron-carbon bonds is likely to proceed at least in part from transient quaternary borate anions. In a separate bond-breaking reaction, L^MeFe(μ-H)₂BEt₂ reacts with N₂H₄ to eject H₂ from the bridging hydrides and cleave the N-N bond in the diaminoborate complex L^MeFe(μ-NH₂)₂BEt₂. These novel bond-breaking reactions are facilitated by the low coordination number at the iron(II) center.

Introduction

In addition to their classic roles as intermediates in homogeneous catalytic reactions of organometallic complexes,¹ iron hydrides have been postulated to be reactive intermediates in the mechanisms of hydrogenase ² and nitrogenase ³ enzymes. We have become especially interested in exploring the chemistry of unsaturated hydride complexes of the late transition metals because of the juxtaposition of the reactive M-H functionality and an open coordination site for substrate binding.

We have used bulky β-diketiminate ligands (Figure 1) to enable the synthesis of the only known three-and four-coordinate iron hydride complexes.^4,5 These unsaturated iron hydrides, L^RFe(μ-H)₂FeL^R, are prepared from reactions between iron(II) chloride complexes and potassium triethylborohydride (eq 1a and 1b).

Figure 1.

Diketiminate ligands L^R, where R indicates the substituent on the N₂C₃ backbone.

$${[{\text{L}}^{\text{Me}}\text{FeCl}]}_2+2{\text{KHBEt}}_3\to {\text{L}}^{\text{Me}}\text{Fe}{(\mu -\text{H})}_2{\text{FeL}}^{\text{Me}}(\mathbf{1}\mathbf{a})+2{\text{BEt}}_3+2\text{KCl}$$ (1a)

$${\text{L}}^{\text{tBu}}\text{FeCl}+2{\text{KHBEt}}_3\to 0.5{\text{L}}^{\text{tBu}}\text{Fe}{(\mu -\text{H})}_2{\text{FeL}}^{\text{tBu}}(\mathbf{1}\mathbf{b})+{\text{BEt}}_3+\text{KCl}$$ (1b)

Although both iron hydride compounds exist as dimers in the solid state, they behave differently in solution: L^tBuFe(μ-H)₂FeL^tBu dissociates into monomers in solution,^4a while only dimer is observed in solutions of the L^MeFe(μ-H)₂FeL^Me.4b One phenomenon caught our attention during the preparation of the iron hydride complexes: long reaction time (> 30 min at room temperature) or heating leads to the formation of red products instead of the brown iron hydride complexes. Here, we report that the iron hydride complex reacts with the byproduct BEt₃ to give an iron dihydridoborate complex and an iron ethyl complex (Scheme 1).

Scheme 1.

The reaction in Scheme 1 features migration of a boron-bound hydrocarbyl group. Boron-to-carbon migrations have precedent in the organic chemistry literature, where alkylboranes have been utilized for forming N-C or C-C bonds to nucleophiles that contain a potential leaving group (e.g. deprotonated α-haloesters or carbon monoxide).⁶ In these reactions, trialkylborane initially attacks the nucleophile to form a quaternary adduct, and one of the boron R groups subsequently migrates to the adjacent N or C, displacing a leaving group. The migration of hydrocarbyl groups from boron to a transition metal is also a key step of the Suzuki-Miyaura reaction, the transmetallation of the organoboron compound to form organopalladium species (Scheme 2).⁷ Interestingly, in both organic and organometallic precedents, quaternary anionic boron species are the actual alkyl/aryl group donor. For example, transmetallation from boron to palladium in the Sukuzi-Miyaura reaction is typically promoted by adding bases to form a quaternary borate anion (Scheme 2, bottom).^{6a, 8,9} Since the B-C bond cleavage in Scheme 1 requires no added base, and occurs in non-coordinating solvents, we were curious to learn whether this reaction is mechanistically distinct from the palladium reaction, possibly leading to insights on strategies for milder cross-coupling reactions.

Scheme 2.

Few boron alkylations of any sort are understood in great mechanistic detail.¹⁰ Below, we describe kinetic studies that give insight into the mechanism of this transformation. The hydridoborate product also undergoes an interesting reaction with hydrazine to cleave the N-N bond to yield a diaminoborate complex.

Results and Discussion

Formation of the dihydridoborate complex from L^RFe(μ-H)₂FeL^R and BEt₃

Reaction of the iron hydride dimers L^RFe(μ-H)₂FeL^R (L^R = β-diketiminate ligands shown in Figure 1; 1a with R = Me and 1b with R = tBu) with one molar equivalent of BR₃ (R = Et, Ph) in toluene at 80 °C for 22 h leads to an equimolar mixture of a four-coordinate hydridoborate complex, L^RFe(μ-H)₂BR′₂ (R = Me, R′ = Et 2a or Ph 2b; R = ^tBu, R′ = Et 2c), and an iron alkyl complex L^RFeR′ (R = Me, R′ = Et 3a or Ph 3b; R = ^tBu, R′ = Et 3c).¹¹ These reactions are shown in Scheme 3. Reaction of 1a with B(OEt)₃ and B(iBu)₃ gave multiple unidentified products, and these reactions were not pursued further.

Scheme 3.

The alkyliron compounds 3a and 3c were identified by comparison to their known ¹H NMR spectra (they can be prepared independently from [L^RFeCl]_n (n = 1 or 2) and Grignard reagents).¹² The new complex L^MeFePh (3b) was previously unknown, and so it was generated from the reaction of [L^MeFe(μ-Cl)₂FeL^Me] with PhMgCl. Its spectroscopic features and X-ray crystal structure (see Experimental Section for spectroscopy and Supporting Information for solid-state structure) are analogous to those for the three-coordinate alkyl complexes. The ¹H NMR spectrum of independently prepared 3b verifies that it is identical to one of the products from reaction of L^MeFe(μ-Cl)FeL^Me and BPh₃.

The new (dihydridoborate)iron(II) complexes 2a, 2b, and 2c were isolated in 59–72% yield after fractional crystallization away from the iron(II) alkyl/aryl complexes. The reaction of 1a with BEt₃ was very clean: 1a, 2a, and 3a are the only compounds observed in ¹H NMR spectra of the reaction mixture. In the other reactions, there were a few unidentified paramagnetic peaks in the crude mixtures, representing side products that were removed by crystallization. Each dihydridoborate complex exhibits a magnetic moment of 4.0(1) to 5.7(1) μ_B in C₆D₆ solution, indicating a high-spin iron(II) center (S = 2). Each ¹H NMR spectrum covers an extremely wide range from 100 ppm to −310 ppm, with large chemical shift dispersion as a result of the paramagnetic iron ion. The proton resonances were generally assigned from the relative integrations (see Experimental Section). The six-proton peaks for the BEt₂ group and the L^Me backbone could be distinguished because the ¹H NMR spectra of L^MeFe(μ-H)₂BEt₂ and L^tBuFe(μ-H)₂BEt₂ differ only by the 6-proton resonance in the L^Me compound and the 18-proton resonance in the L^tBu compound. The bridging hydride protons are not observed because of fast relaxation.

A single crystal of each iron dihydridoborate complex was grown from pentane. The solid-state structure of 2a is shown in Figure 2 as an example, and others may be found in the Supporting Information. The bridging hydride ions were located and refined with an isotropic thermal parameter in each structure (the two hydrides are crystallographically distinct, except in the case of L^MeFe(μ-H)₂BEt₂, where the hydride positions are related by a C₂ axis). Including the two Fe-H bonds, each iron center has a pseudotetrahedral geometry with a Fe-H distance of about 1.7 Å (Table 1). Although the Fe^···B distances are relatively short (ca. 2.24 Å), we do not postulate any direct Fe-B bonding interaction because the two atoms are held in proximity by the bridging hydride.¹³ The B-H distances range from 1.19(2) Å to 1.35(2) Å, which is comparable to the average M-H-B distance of 1.24 Å in the Cambridge Structural Database.

Figure 2.

Thermal-ellipsoid plot of the molecular structure of L^MeFe(μ-H)₂BEt₂ (2a). A crystallographic C₂ axis passes through Fe and B. Other crystallographic data are in the Supporting Information.

Table 1.

Important bond distances and angles for dihydridoborate complexes in this work.

Complex	Fe-H/Å	B-H/Å	Fe-B/Å
LMeFe(μ-H)2BEt2 (2a)	1.74(2)	1.35(2)	2.237(3)
LMeFe(μ-H)2BPh2 (2b)	1.72(2), 1.72(2)	1.28(2), 1.19(2)	2.239(2)
LtBuFe(μ-H)2BEt2 (2c)	1.74(2), 1.73(2)	1.26(2), 1.21(2)	2.232(1)

The formation of 2 and 3 is the result of an interesting B-C bond cleavage. Previous examples of B-C cleavage are most often from nucleophilic borates, rather than from boranes.¹⁴ In complexes with covalently bound hydridoborate ligands, BH_nR_4−n⁻ (n = 1–4, R = H or alkyl) complexes are much less common than borohydride (BH₄⁻) complexes.¹⁵ Known examples of MH₂BR₂ complexes typically come from double metathesis reactions of a metal chloride complex with an alkali metal salt of H₂BR₂ (R₂ = H₂, ¹⁶ C₅H₁₀, C₈H₁₄ ¹⁷ and Et₂ ^{15b, 18}). In some literature cases, the H₂BEt₂⁻ group was present from impurities in commercial HBEt₃^{−.15b, 1819} Here, the hydridoborate product is formed from cleavage of the B-C bond of a trialkylborane, and not from a dihydridoborate impurity in the starting material (see below).

Reaction of L^MeFe(μ-H)₂FeL^Me with BR₃: Rate Law and Activation Parameters

We used the reaction of L^MeFe(μ-H)₂FeL^Me (1a) with BEt₃ as the subject of kinetic studies because of the simple ¹H NMR spectra of the starting material and the exceptionally clean reaction. Using an internal integration standard (L^tBuFeCl) in a sealed capillary, we verified that the spectroscopic yields of products are 95% and 99% for L^MeFe(μ-H)₂BEt₂ (2a) and L^MeFeEt (3a) respectively. In the kinetics experiments, various amounts of 1a and BEt₃ (>10 equiv relative to 1a) were mixed in C₆D₆ and held at various temperatures between 6.5 °C and 59.7 °C. Complexes 2a and 3a were the only products observed by ¹H NMR spectroscopy. The integrations of the peaks at 13 ppm (1a), 17 ppm (2a), and −11 ppm (3a) were followed as a function of time, and invariably showed exponential decays. The rate constant was independent of [Fe] when a drop of liquid mercury was added to the reaction mixtures. Without the mercury, we observed a slight decrease in k_obs from 3.97(7) × 10⁻³ s⁻¹ to 2.51(2) × 10⁻³ s⁻¹ with an increase of [1a] from 11.1 mM to 42.3 mM. With added Hg⁰, the derived k_obs showed no clear trend while changing [1a] from 20 mM to 40 mM. Therefore, subsequent trials were run with a drop of mercury, which presumably amalgamates trace metallic iron.

For consistency, the following use exclusively the rate constants derived from following the decrease of 1a (Table 2). The rate has a first-order dependence on [1a] (as evident from its exponential decay, and the constancy of the rate constant upon variation of [1a]) and a first-order dependence on [BEt₃] (from the linear dependence of the pseudo-first-order rate constant on BEt₃ concentration). Therefore, the rate law is

Table 2.

Effect of [BEt₃] on observed rate constant

Entry	Initial concentrations (mM)		Observed rate constant kobs (s−1)
	1	BEt3
	1	6.4	64.8	2.2(4)×10−3
2	6.4	129.6	4.7(2)×10−3
3	6.4	259.0	9.4(2)×10−3

$$\text{rate}=-d[\mathbf{1}\mathbf{a}]/dt=k[\mathbf{1}\mathbf{a}][{\text{BEt}}_3]$$ (2)

The second-order rate constant at room temperature (21.5 °C) was 4.9(1) × 10⁻⁴ M⁻¹s⁻¹. Activation parameters were calculated from an Eyring plot of second-order rate constants from 6.5 °C to 59.7 °C, giving ΔH^‡ = 21.8 ± 0.8 kcal·mol⁻¹ and ΔS^‡ = − 1 ± 2 eu.²⁰

Table 4 shows that varying the steric demands of the diketiminate ligand and BR₃ gives rate constants of the same magnitude, which slightly increase with a larger borane (BPh₃) or diketiminate ligand (L^tBu). Therefore, there is only a small steric influence on the reaction rate.

Table 4.

Rate constants with variation of R groups on the diketiminate and borane^a

Iron hydride reagent	BR3 reagent	Pseudo first-order rate constant kobs/s−1
LMeFe(μ-H)2FeLMe (1a)	BEt3	3.09(2) × 10−3
LMeFe(μ-H)2FeLMe (1a)	BPh3	8.91(5) × 10−3
LtBuFe(μ-H)2FeLtBu (1b)	BEt3	4.8(2) × 10−3

^a[BR₃] = 0.419 M, [L^RFe(μ-H)₂FeL^R] = 26 mM, T = 40 °C.

Finally, the reaction of 1a with BEt₃ was repeated in different polar solvents (75% o-difluorobenzene/25% C₆D₆ or THF-d₈). We observed a small increase in rate constant using THF, but no increase in a more polar mixture of o-difluorobenzene and benzene (Table 5). The similarities in rate with a more polar solvent mixture suggests that charged species are not formed in the rate-limiting transition state (see below). The rate increase in THF may be from coordination to THF to intermediates, but the nature of this interaction was not queried further.

Table 5.

Solvent effect on rate constant^a

Solvent	Dielectric Constant	[BEt3]	[1a]	kobs (s−1)
C6D6	2.2521	419 mM	21.1 mM	3.33(1) × 10−3
THF-d8	7.4722	419 mM	23.2 mM	5.77(5) × 10−3
75% o-C6H4F2 +25% C6D6	11.26b	419 mM	24.5 mM	2.57(3) × 10−3

^aMonitored at 40 °C. These experiments were performed without mercury, and therefore the rate constants differ slightly from those in Table 4.

^bThe dielectric constant for the mixed solvent is estimated as the weighted average of 14.26 (pure o-C₆H₄F₂)²³ and 2.25 (pure C₆D₆).

B-C Bond Cleavage: Mechanistic Hypotheses

Several reasonable mechanisms can be proposed for the B-C bond cleavage reaction. We do not consider mechanisms in which BR₃ undergoes β-hydride elimination because the reaction is equally facile with BPh₃, for which β-hydride elimination is impossible. Moreover, independent experiments show that BEt₃ and 1-hexene do not react when heated for 2 h at 80 °C, indicating that borane alkyl groups do not undergo substantial β-hydride elimination under the conditions of the reaction in Scheme 1.

The mechanisms considered here are grouped into categories: dissociation of the iron hydride dimer into monomers before reaction with borane (A), concerted Fe-H and B-C cleavage with B-H formation (B, C, D), and formation and disproportionation of a transient iron triethylborohydride species (E, F).

Dissociation of 1a into monomer

Mechanism A (Scheme 4) illustrates the potentialdissociation of the iron hydridedimer as part of the mechanism. A mechanism of this type was strongly implicated in the reaction of L ^tBuFe(μ-H)₂FeL^tBu with 3-hexyne, where a zero-order dependence on alkyne concentration indicated an intramolecular rate-limiting step prior to any interaction with alkyne.⁴ However, in the reaction with borane described here, hydride dissociation can be ruled out because it does not predict the observed rate law. Mechanism A would lead to a zero-order dependence on [BEt₃] if k₁ (dimer dissociation) were rate-limiting, or a half order dependence on [1a] if k₂ (attack of borane) were rate-limiting.

Scheme 4.

$$\underset{\_}{\mathbf{A}},k_1\text{rate}\text{limiting}:\text{rate}=k_1[\mathbf{1}\mathbf{a}]$$ (8)

$$\underset{\_}{\mathbf{A}},k_2\text{rate}\text{limiting}:\text{rate}=k_2[\text{LFeH}][{\text{BEt}}_3]=k_2{K_{\text{eq}}}^{1/2}{[\mathbf{1}\mathbf{a}]}^{1/2}[{\text{BEt}}_3]\text{where}{K}_{\text{eq}}=k_1/k_{-1}$$ (9)

The time course of the concentration of 1a is inconsistent with a half-order dependence of the rate on [1a].

Concerted mechanisms

In mechanism B, 1a reacts with BEt₃ in an initial rate-limiting step. While the empty boron p-orbital interacts with the bridging hydride, the nucleophilic boron-bound alkyl group simultaneously approaches the iron atom of 1a. This resembles the “σ-bond metathesis” mechanism that explains C-H activation reactions by d⁰ metal centers.²⁴ This pathway leads to the observed products after dissociation of a bridging hydride, and is the microscopic reverse of the reaction of a ruthenium alkyl complex with catecholborane through a σ-bond metathesis mechanism.²⁵

Mechanism C, on the other hand, begins with Fe-H bond breaking. After dissociating one of the Fe-H bonds, the “open” isomer leads to a six-membered transition state for a pericyclic reaction. As long as the initial Fe-H opening is rapid and reversible, mechanism C agrees with the experimental rate law, rate = k[1a][BEt₃].

Mechanism D (Scheme 7) starts with rate-limiting abstraction of “H⁻” from 1a by the borane, breaking two of the Fe-H bonds and forming free HBEt₃⁻. Previous researchers have observed alkyl group transfers from a borate to a metal center, which is more common compared to group transfer from borane species.¹⁴ In analogy, mechanism D concludes with transfer of an ethyl group from HBEt₃⁻ to iron. The ion pair undergoes a σ-bond metathesis to give the final products. Whether the hydride abstraction (D1) or the recombination of the ion pair (D2) is rate limiting, the predicted rate law is the same as mechanisms B and C: a first order dependence on both [1a] and [BEt₃], in concord with the observed rate law.

Scheme 7.

Notably, each of these mechanisms suffers from inconsistencies with other mechanistic data. First, the entropy of activation near zero is difficult to rationalize in the light of the bimolecular transition states in these mechanisms. In order to outweigh the expected large negative entropy change from bringing two molecules together, it is necessary to have a roughly equivalent release of entropy prior to the rate-limiting step. This situation is provided only by mechanisms C and D2, where the observed rate constant k_obs is actually a product of the equilibrium constant for cleavage one or two Fe-H bond(s) in 1a (K_FeH) and the rate constant for the elementary step consisting of the pericyclic reaction, k_peri; k_obs = K_FeHk_peri. Therefore ΔS^‡_obs = ΔS_FeH + ΔS^‡_peri. We cannot determine either of these individual entropy values quantitatively, but it is reasonable to postulate that the entropy increase from partly opening the crowded dimer structure of 1a (ΔS_FeH) could roughly equal the entropy decrease of bringing in the borane or borate to reach the transition state for the pericyclic reaction (ΔS^‡_peri). Molecular models of the reactants suggest that only an intermediate generated by cleavage of the Fe₂H₂ ring (as in mechanism C and D2) would have sufficient space for approach of the borane or borate B-C bond.

However, it is difficult to rationalize the similar observed reaction rate between BEt₃ and BPh₃ if mechanism C is followed. In the six-membered ring transition state, one of the R groups from the borate sits between the BR₂ fragment and the L^MeFe fragment. Considering the huge hindrance from the β-diketiminate ligand, this conformation is expected to have different steric effects for ethyl and phenyl groups. This argument, as well as the near-zero entropy of activation, makes mechanism C less plausible.

In mechanisms D, we note that an ionic intermediate is formed in or before the rate-limiting transition state. The rate of disappearance of 1a has very little dependence on solvent polarity, though polar solvents are expected to stabilize transition states in which charge is developing. Therefore, the solvent effect is inconsistent with mechanism D.

Borohydride disproportionation mechanisms

Commercial solutions of Super-Hydride (LiHBEt₃) have been reported to undergo disproportionation to LiBEt₄ and Li[BH₂Et₂] at room temperature over long time periods (eq 3).^{15b,18b,18c, 26} The resulting Li[BH₂Et₂] can lead to H₂BEt₂ complexes by metathesis with a metal chloride. In addition, solutions of LiHBEt₃·THF are known to react with BEt₃ to give LiBEt₄ and (HBEt₂)₂, and subsequently Li[BH₂Et₂] ^18c,26 (eq 3, 4).

$$2{\text{LiHBEt}}_3\to \text{Li}({\text{H}}_2{\text{BEt}}_2)+{\text{LiBEt}}_4$$ (3)

$$2{\text{BEt}}_3+2{\text{LiHBEt}}_3\to {({\text{HBEt}}_2)}_2+{\text{LiBEt}}_4$$ (4)

However, the spurious presence of diethylborohydride anions is not likely to be involved in the reaction of 1a with BR₃ for the following reasons. First, our KHBEt₃ was prepared in-house from KH and BEt₃ in toluene (eq 4) and stored as a solid. The ¹¹B NMR spectrum of this isolated KHBEt₃ in C₆D₆ has a single peak at −16 ppm, with no signs of dihydridoborate or BEt₃ impurities. The ¹¹B NMR spectrum of KHBEt₃ did not show any change when heated to 80 °C for 2 h in benzene-d₆ (mimicking the conditions of our kinetic experiments). An even more convincing piece of evidence is our ability to synthesize L^MeFeH₂BEt₂ from 1a and BEt₃, a reaction in which KHBEt₃ is not present. Therefore, the presence of potassium diethylborohydride in our reaction is unlikely, and there is no necessity for hydridoborate salts in for the reaction to occur.

It is still necessary to consider the transient formation of trialkylborohydride species later in the pathway. For example, the first step of the reaction might resemble the reaction of alkali metal hydrides with triethylborane²⁷ (eq 5).

$$\text{KH}+{\text{BEt}}_3\to {\text{KHBEt}}_3$$ (5)

The analogous reaction at iron would form a transient iron triethylborohydride complex, L^MeFe(HBEt₃), either through monomer (mechanism E) or through a partially open dimer (mechanism F). Because L^MeFe(HBEt₃) is the presumed intermediate in the reaction of L^MeFeCl and KHBEt₃ to form L^MeFe(μ-H)₂FeL^Me + KCl + BEt₃ (eq 1a, above), steps preceding the formation of this intermediate are assumed to be reversible (Scheme 8)

Scheme 8.

In Scheme 8, the transient iron triethylborohydride species lies in the center. Since L^MeFe(HBEt₃) is not observed by NMR spectroscopy during any reaction, this intermediate must be consumed very quickly (k_d, k₋₂ and k′₋₂ are large). Assuming that L^MeFe(HBEt₃) is an intermediate on the way to [L^MeFe(μ-H)₂FeL^Me]₂ in eq 1, then k₋₂ and k₋₁ in mechanism E or k′ ₋₁ in mechanism F must be larger than k_d (or else 1a could not be isolated in the reaction of [L^MeFeCl]₂ and KHBEt₃). Therefore k₁ or k₂ (in E) or k′₁ (in F) is rate-limiting. In other words, the rate determining step (RDS) must happen before the formation of L^MeFe(HBEt₃).

E if RDS is k₁,

$$\text{rate}=k_1[\mathbf{1}\mathbf{a}]$$ (6a)

if RDS is k₂,

$$\text{rate}=k_2{K_{\text{eq}}}^{1/2}{[\mathbf{1}\mathbf{a}]}^{1/2}[{\text{BEt}}_3]\text{where}{K}_{\text{eq}}=k_1/k_{-1}$$ (6b)

Mechanism E predicts a zero-order dependence on [BEt₃] if k₁ is the RDS or a half-order dependence on [1a] if k₂ is the RDS (eq 6). Neither situation agrees with the observed rate law.

In F, as shown in Scheme 8, the hydride dimer first reacts with one molecule of BEt₃, going through transition state a* to give L^MeFeHBEt₃ and L^MeFeH. The L^MeFeH monomer rapidly reacts with another molecule of BEt₃ to form a second equivalent of L^MeFeHBEt₃. The two L^MeFeHBEt₃ species then undergo disproportionation to give the final products. If k′₁ is rate-limiting the rate law is F

$$\text{rate}=k_1\text{'}[\mathbf{1}\mathbf{a}][{\text{BEt}}_3]$$ (7)

The first order dependence predicted for [1a] and [BEt₃] is consistent with the experimental rate law. Moreover, the transition state a* involves the dissociation of one Fe-H bond and the formation of one B-H bond. As such, it bears a resemblance to the transition state for an interchange mechanism for substitution at a transition-metal complex (the intermediate stage between associative and dissociative substitution reactions). For I_a or I_d mechanisms, values of ΔS^‡ near zero are common.²⁸

Mechanism F is also consistent with other mechanistic data. With the opening of one iron-hydride bond, the β-diketiminate has more space to avoid steric interactions with the borane R group. This explains the similar observed reaction rate for BEt₃ and BPh₃. Finally, there are no ionic intermediates prior to the rate-determining-step, which agrees with the lack of a solvent polarity effect during the reaction. Based on the above analysis, mechanism F is the one most consistent with all experimental data.

Because the key disproportionation step lies after the rate-limiting step, kinetic data are not useful in elucidating its characteristics. Therefore, we endeavored to independently synthesize L^MeFe⁺BEt⁻₄, the putative intermediate in the reaction.²⁹ First, reaction of [L^MeFeCl]₂ and NaBEt₄ in Et₂O leads cleanly to the formation of L^MeFeEt, as observed by ¹H NMR spectroscopy. Also, L^MeFeEt does not react with BEt₃ up to 120 °C. This is a thermodynamic phenomenon rather than a kinetic barrier, because heating L^MeFePh to 80 °C for 3 h with one equiv of BEt₃ gives a mixture of L^MeFePh and L^MeFeEt. These experiments strongly suggest that transient L^MeFe⁺BR₄⁻ species are unstable with respect to dissociation of borane, as required in the last step of mechanism F. In benzene, the ionic species may exist as a contact ion pair or as a tight ion pair, analogous to one that was previously observed.³⁰

In summary, the available kinetic and mechanistic evidence are most consistent with mechanism F, in which a transient ring-opened isomer of 1a reacts with BR₃ to give LFeHBR₃, which converts into the iron hydridoborate and iron alkyl products through the ionic intermediate LFeBR₄.

L^MeFe(μ-H)₂BEt₂ reactivity with hydrazine: N-N bond cleavage

Exploration of the reactivity of the dihydridoborate complex L^MeFe(μ-H)₂BEt₂ (2a) has shown an interesting series of reactions upon the addition of hydrazine. At room temperature, complex 2a reacts rapidly with H₂NNH₂ to give a hydrazine adduct, L^MeFe(η¹-H₂NNH₂)(μ-H)₂BEt₂ (4), isolated in 90% yield. Titration of hydrazine into a solution of 2a gave solutions with ¹H NMR spectra similar to 2a but variable chemical shifts, showing that the bound hydrazine is in rapid exchange with free hydrazine. Proton NMR resonances for the N-bound protons are not seen because of their proximity to the paramagnetic metal center.

The binding stoichiometry in solution can be determined by Job plot analysis, where small aliquots of hydrazine are added to 2a and the ¹H NMR spectrum is recorded after each addition. The chemical shift of peaks near −260 ppm and 19 ppm vary depending on the ratio of H₂NNH₂ and 2a (Figure 4). The intersection of the initial line (deficiency of hydrazine) and final line (deficiency of 2a) shows that the two molecules bind in a 1:1 manner.

Figure 4.

Titration curve and Job plot, using the chemical shift of the ¹H NMR resonance near −260 ppm.

Figure 5 displays the X-ray crystal structure of 4. The hydrazine ligand binds to the iron atom in a η¹ mode and points towards the diketiminate. The bridging hydrogen atoms were again located in Fourier difference maps, and their positions were refined. The geometry of the five-coordinate iron atom is intermediate between ideal square-pyramidal and trigonal-bipyramidal with τ = 0.43³¹. With the higher coordination number at iron, the Fe-H distances increase to 1.82(2) Å and 1.91(2) Å, which are about 0.1 Å longer than those in 1. Accordingly, the H-B distances decrease by 0.12 Å to 1.22(2) Å, comparable to the average value for literature hydridoborate complexes (see Supporting Information). The FTIR spectrum of 4 shows two bands at 3368 cm⁻¹ and 3333 cm⁻¹, which shifted to 2519 cm⁻¹ and 2493 cm⁻¹ when N₂D₄ was used (calculated from harmonic oscillator: 2460 cm⁻¹ and 2434 cm⁻¹). Therefore, these are assigned as N-H (N-D) stretching vibrations. ³²

Figure 5.

ORTEP diagram of the molecular structure of L^MeFe(η¹-H₂NNH₂)(μ-H)₂BEt₂ (4). Aryl groups and C-H are omitted for clarity. Important bond distances (Å) and angles (°): Fe–H1 1.91(2), Fe–H2 1.82(2), Fe–N15 2.214(2), B–H1 1.22(2), B–H2 1.22(2), N15–N25 1.441(2), Fe–B 2.350(2), H1-Fe-H2 61.2(9), H1-B-H2 103(1).

Well-characterized complexes of iron with N₂ and its partial reduction products N₂H₂ and N₂H₄ are rare. A search of the Cambridge Structural Database showed only eight examples of Fe-N₂H₄ species.^32,33 The Fe-N distance in our complex is 2.214(2) Å, which is comparable to the average value of 2.14(3) Å for the known iron-hydrazine adducts. The N-N distance in 4 also agrees well with the average literature value of 1.448(3) Å for bound N₂H₄.

After heating a solution of 4 in toluene at 60 °C for 22 h, the new complex L^MeFe(μ-NH₂)₂BEt₂ (5) can be isolated in 57% yield (Scheme 9). The solid-state structure of 5 is shown in Figure 6. Interestingly, the N-N bond has been cleaved (N^···N = 2.459(3) Å) to yield a diaminoborate complex. The IR spectrum shows a single N-H stretching band at 3393 cm⁻¹, which shifts to 2494 cm⁻¹ in the compound synthesized from N₂D₄. Most complexes containing M(μ-NH₂)₂M fragments exhibit two N-H stretching bands between 3100–3300 cm^−1.34

Scheme 9.

Figure 6.

ORTEP diagram of the molecular structure of L^MeFe(μ-NH₂)₂BEt₂ (5) (aryl isopropyl groups and hydrogens are omitted for clarity). There is a crystallographic C₂ axis through Fe and B. Important bond distances (Å) and angles (°): Fe-N1 2.064(2), Fe-N1A 2.064(1), B-N1 1.571(8), B-N1A 1.577(8), N1-N1A: 2.4589(31), N1-Fe-N1A: 73.12(9), N1-B-N1A: 102.5(3).

The observation of one N-H stretching band suggests a possible alternative formulation for 5, L^MeFe(μ-NH)₂BEt₂. However, this formally iron(IV) complex seems less likely for several reasons. First, all examples of tetrahedral iron(IV) in the literature are low-spin and supported by strongly donating terminal nitride or imide ligands.³⁵ The solution magnetic moment of 5 is 4.8(1) μ_B, which is more consistent with a high-spin Fe(II) complex. Second, each of the four N-bound H atoms was clearly visible in Fourier maps from the X-ray diffraction experiment and was refined.³⁶ Finally, the alternative formulation implies that 2 equiv of H₂ are produced in the reaction. Using gas chromatography, we determined that only 1.67 ± 0.03 equiv of H₂ is produced (see Supporting Information for details), inconsistent with the diimidoborate formulation.

We have no definitive explanation for the observation of H₂ in excess of the amount predicted by the stoichiometry of the reaction in Scheme 9. No substantial amount of H₂ was produced in control experiments where 2a was heated with hydrazine. However, it is important to note that the spectroscopic yield of 5 is only 60% (from an NMR experiment with internal standard). A substantial amount of a black precipitate is formed along with 5. It seems possible that the remaining 0.4 equiv of hydrazine is disproportionated to H₂ and N₂ by the iron byproducts, with complete disproportionation giving the limiting stoichiometry in Scheme 10. The rough agreement with the observed H₂ production supports this speculation.

Scheme 10.

Because 4 has hydrogens in the hydride bridges as well as the bound hydrazine, a final question regards the source of the N-H protons in 5. Importantly, the synthesis of 5 from N₂D₄ gave a product with no N-H peak and a new N-D peak in the infrared spectrum. This observation indicates that all N-H protons in 5 derive from hydrazine, and that the bridging hydrides in 4 are released as H₂ upon heating. Although the unknown nature of some products dissuades us from further mechanistic studies, one can speculate a mechanism where loss of H₂ from 4 generates unsaturation at iron and boron. This would enable coordination of the free lone pair on the bound hydrazine, followed by N-N cleavage to give 5. However, we stress that other mechanisms are also consistent with the limited data.

A limited number of literature examples report the cleavage of hydrazine N-N bonds with metal-containing clusters, often through formation of bridging NH or NH₂.³⁷ There are even fewer complexes that cleanly perform this reaction with two or less metals.³⁸ Here, the Fe-N interactions are reinforced by the low-coordinate environment, which creates strong bonds to electronegative groups, as we have demonstrated previously.¹²

Conclusions

We have discovered an interesting series of borane alkylations/arylations of iron hydride complexes, which give iron hydridoborate products through attack of a ring-opened iron hydride species on a borane. In this reaction, the hydride ligand acts as a nucleophile, apparently activating the borane toward transfer of the R group to iron. A combination of kinetics studies has been unusually informative with respect to the mechanism, elucidating the nature of the rate-determining transition state. In addition, the hydridoborate product of the reaction reacts to cleave the N-N bond of hydrazine, releasing the two bridging hydrides as H₂. The B-C and N-N cleavage reactions demonstrate the ability of low-coordinate iron complexes to perform difficult bond transformations under mild conditions.

Experimental Section

General Considerations

All manipulations were performed unde r a nitrogen atmosphere using standard Schlenk techniques or in an M. Braun glovebox maintained at or below 1 ppm of O₂ and H₂O. Glassware was dried at 150 °C overnight. ¹H NMR spectra were recorded on a Bruker Avance 500 spectrometer (500 MHz) at 22 °C and referenced internally to residual protiated solvent (C₆D₅H at 7.15 ppm). Resonances are broad singlets unless otherwise specified. Infrared spectra (450–4000 cm⁻¹) were recorded on KBr pellet samples in a Shimadzu FTIR spectrophotometer (FTIR-8400S). A total of 32 scans at 2 cm⁻¹ resolution were collected in each case. Electronic spectra were recorded between 280 nm and 1000 nm on a Cary 50 UV-visible spectrophotometer, using screw-cap quartz cuvettes of 1 cm optical path length. Elemental analyses were determined by Desert Analytics (Tucson, AZ). Pentane, tetrahydrofuran (THF), and toluene were purified by passage through activated alumina and “deoxygenizer” columns obtained from Glass Contour Co. (Laguna Beach, CA). Deuterated benzene and THF were dried over CaH₂, then over Na/benzophenone, and then vacuum transferred into a storage container. Before use, an aliquot of each solvent was tested with a drop of sodium benzophenone ketyl in THF solution. Celite was dried overnight at 200 °C under vacuum. Hydrazine, purchased from Aldrich, was dried over KH and vacuum transferred before use. KH, purchased for Aldrich as suspensions in oil, was washed with pentane three times and dried. BEt₃ (1.0 M in hexane) was purchased from Aldrich and used as it arrived without any further purification. The preparation and properties of [L ^MeFeCl]₂, [L ^MeFeH]₂ (1) and KHBEt₃ were previously reported.³⁹

L^MeFe(μ-H)₂BEt₂ (2a) from 1

A sample of 1 (139 mg, 0.147 mmol) was dissolved in 15 mL of toluene. Triethylborane (0.29 mL of a 1.0 M solution in hexane) was added via syringe to the bright red solution. The solution was heated at 70 °C for 14 h. Volatile materials were removed under vacuum and the residue was extracted with pentane (15 mL), filtered and concentrated to 5 mL. Crystallization from pentane at −35 °C gave bright red needles (68 mg, 72%).

L^MeFe(μ-H)₂BEt₂ (2a) from FeCl₂

The compound was typically synthesized in one flask from L^MeLi and Fe(THF)_1.5Cl₂ as follows. A Schlenk flask was loaded with a mixture of L^MeLi (1.69 g, 3.98 mmol) and Fe(THF)_1.5Cl₂ (0.94 g, 4.0 mmol) in toluene (50 mL). The mixture was stirred at 80 °C for 22 h. All solvent was removed under vacuum and a solution of KHBEt₃ (0.56 g, 4.0 mmol) in toluene (50 mL) was added. The mixture was heated at 80 °C for 4 h. Volatile materials were removed under vacuum and the residue was extracted with pentane (100 mL), filtered and concentrated to 20 mL. Crystallization at −35 °C gave bright red needles (777 mg, 72%). ¹H NMR (400 MHz, C₆D₆): 81 (6H, BEt₂ CH₃), 46 (6H, backbone CH₃), 17 (4H), −4 (12H, iPr CH₃), −36 (12H, iPr CH₃), −40 (4H), −50 (2H, p-H), −208 (4H) ppm. (Peaks integrated as 4H could be BEt₂ CH₂, m-H or ⁱPr methine. Peaks for the backbone H and bridging H were not observed.) UV-vis (pentane): 299 (ε= 27.3 mM⁻¹cm⁻¹), 398 (ε = 15.0 mM⁻¹cm⁻¹), 554 (ε = 2.9 mM⁻¹cm⁻¹) nm. μ_eff (C₆D₆, 25 °C): 5.1(1) μ _B. Elem. Anal. Calcd for C₃₃H₅₃N₂BFe: C, 72.80; H, 9.81; N, 5.15. Found: C, 73.52; H, 9.52; N, 5.62.

L^MeFe(μ-H)₂BPh₂ (2b)

1 (127 mg, 0.13 mmol) and triphenylborane (32 mg, 0.13 mmol) were dissolved in 20 mL of toluene. The solution was heated at 80 °C for 18 h. Volatile materials were removed under vacuum and the residue was extracted with toluene (15 mL), filtered and concentrated to 5 mL. Crystallization from pentane at −26 °C gave red blocks (56 mg, 67%). ¹H NMR (500 MHz, C₆D₆): 78 (6H, backbone CH₃), 32 (4H), 17 (4H+4H), −4 (12H, iPr CH₃), −8 (2H), −40 (12H, iPr CH₃), −46 (4H), −52 (2H) ppm. (Peaks integrated as 4H could be iPr methine, aryl m-H, phenyl o-H or phenyl m-H. Peaks integrated as 2H could be aryl p-H or phenyl p-H. Peaks for the backbone H and bridging H were not observed.) UV-vis (toluene): 291 (ε = 14.6 mM⁻¹cm⁻¹), 398 (ε = 6.3 mM⁻¹cm⁻¹), 567 (ε = 1.2 mM⁻¹cm⁻¹) nm. μ_eff (C₆D₆, 25 °C): 4.0(1) μ _B. Elem. Anal. Calcd for C₄₁H₅₃BFeN₂: C, 77.12; H, 8.05; N, 4.39. Found: C 76.88, H 7.85, N 4.38,.

L^tBuFe(μ-H)₂BEt₂ (2c)

A Schlenk flask was loaded with a mixture of L^tBuFeCl (722 mg, 1.21 mmol) and KHBEt₃ (167 mg, 1.21 mmol) in toluene (50 mL). The mixture was stirred at 80 °C for 18 h. Volatile materials were removed under vacuum and the residue was extracted with pentane (50 mL), filtered and concentrated to 10 mL. Crystallization at −26 °C gave dark red blocks (224 mg, 59%). ¹H NMR (500 MHz, C₆D₆): 71 (6H, BEt₂ CH₃), 42 (1H, backbone), 35 (18H, backbone CH₃), 16 (4H), −13 (12H, iPr CH₃), −61 (12H, iPr CH₃), −66 (4H), −89 (2H, p-H), −313 (4H) ppm. (Peaks integrated as 4H could be BEt₂ CH₂, m-H or iPr methine. Peaks for the bridging H were not observed.) UV-vis (pentane): 333 (ε = 9.42 mM⁻¹cm⁻¹), 414 (ε = 6.82 mM⁻¹cm⁻¹), 597 (ε = 1.27 mM⁻¹cm⁻¹) nm. μ_eff (C₆D₆, 25 °C): 5.7(1) μ _B. Elem. Anal. Calcd for C₃₉H₆₅N₂BFe: C, 74.52; H, 10.42; N, 4.46. Found: C, 74.60; H 10.20,; N, 4.34.

Complexes 3a and 3c have been reported in ref 12.

L^MeFePh (3b)

The compound was typically synthesized in one flask from L^MeLi and FeCl₂(THF)_1.5 as follows. A resealable flask was loaded with a mixture of L^MeLi (592 mg, 1.4 mmol) and FeCl₂(THF)_1.5 (329 mg, 1.4 mmol) in toluene (20 mL). The mixture was stirred at 80 °C for 22 h. Phenylmagnesium chloride (2.0 M in THF, 0.7 mL, 1.4 mmol) was added, and the mixture was stirred for 4 h at room temperature. Solvent was removed under vacuum. The residue was extracted with pentane (20 mL), filtered through Celite, concentrated to 3 mL, and cooled to −45 °C to give bright yellow crystals (487 mg, 63% yield). ¹H NMR (C₆D₆, 21 °C): 163 (2H, phenyl m-CH), 126 (1H, phenyl p-CH), 47 (6H, backbone CH₃), 28 (1H, backbone CH), −10 (4H, m-CH), −21 (12H, iPr CH₃), −82 (2H, p-CH), −118 (4H, iPr CH), −125 (12H, iPr CH₃). (Phenyl o-CH was not observed. Peak assignments were based on integrations and ¹H NMR spectra of other known L^MeFeR complexes.) UV-vis (pentane): 325 (ε = 15.6 mM⁻¹cm⁻¹), 368 (ε = 11.4 mM⁻¹cm⁻¹), 494 (ε = 1.0 mM⁻¹cm⁻¹) nm. μ_eff (C₆D₆, 25 °C): 5.2(1) μ _B. Elem. Anal. Calcd for C₃₅H₄₆N₂Fe: C 76.35, H 8.42, N 5.09. Found: C, 76.06; H, 8.20; N, 4.99.

L^MeFe(H₂NNH₂)(μ-H)₂BEt₂ (4)

L^MeFe(μ-H)₂BEt₂ (221 mg, 0.407 mmol) was dissolved in diethyl ether (10 mL). Hydrazine (13 μL, 0.41 mmol) was added via syringe to the bright red solution. The solution was shaken, causing an immediate color change from bright red to orange-pink. The solution was concentrated to 2 mL and cooled to −35 °C to give bright pink-yellow needles of 3 (212 mg, 90%). ¹H NMR (400 MHz, C₆D₆): 24 (6H), 19 (4H), 3 (12H, iPr CH₃), 1 (6H), −15 (12H, iPr CH₃), −24 (4H), −40 (2H, p-H), −62 (1H, backbone C-H), −260 (4H) ppm. (Peaks integrated as 4H could be BEt₂ CH₂, m-H or iPr methine. Peaks integrated as 6H could be BEt₂ CH₃ or backbone CH₃. Peaks for the bridging H and hydrazine H were not observed.) IR (KBr pellet): 3368 cm⁻¹, 3338 cm⁻¹ (N-H). UV-vis (pentane): 298 (ε = 13.4 mM⁻¹cm⁻¹), 397 (ε = 7.5 mM⁻¹cm⁻¹), 552 (ε = 1.5 mM⁻¹cm⁻¹) nm. μ_eff (C₆D₆, 25 °C): 3.8(1) μ_B. Elem. Anal. Calcd for C₃₃H₅₇N₄BFe: C, 68.75; H, 9.97; N, 9.72. Found: C, 68.01; H, 8.83; N, 5.70. The bound hydrazine could be removed by extended pumping under vacuum (i.e. 50% hydrazine was removed after 26 h at RT), and we suspect that this is the reason for the low nitrogen analysis.

L^MeFe(μ-NH₂)₂BEt₂ (5)

A sample of L^MeFe(μ-H)₂BEt₂ (201 mg, 0.371 mmol) was dissolved in toluene (15 mL). Hydrazine (14 μL, 0.45 mmol) was added via syringe to the bright red solution. The solution was shaken and then heated at 60 °C for 22 h. Volatile materials were removed under vacuum and the residue was extracted with pentane (15 mL), filtered and concentrated to 2 mL. Crystallization from pentane at −35 °C gave brown blocks (121.6 mg, 57%). ¹H NMR (500 MHz, C₆D₆): 57 (4H), 33 (6H), 19 (4H), 2 (12H, iPr CH₃), −20 (12H, iPr CH₃), −35 (4H), −42 (1H, backbone C-H), −49 (2H, p-H), −71 (6H) ppm. (Peaks integrated as 4H could be BEt₂ CH₂, m-H or iPr methine. Peaks integrated as 6H could be BEt₂ CH₃ or backbone CH₃. Peaks for the bridging NH₂ groups were not observed.) IR (KBr pellet): 3393 cm⁻¹ (N-H). UV-vis (pentane): 331 (ε = 16.0 mM⁻¹cm⁻¹). μ_eff (C₆D₆, 25 °C): 4.8(1) mu;_B. Elem. Anal. Calcd for C₃₃H₅₅N₄BFe: C, 68.99; H, 9.65; N, 9.75. Found: C, 68.68; H, 9.45; N, 9.35.

Kinetic experiments: effect of BEt₃

[L^MeFeH]₂, (16.4 mg, 17.3 mmol), was dissolved in C₆D₆ (2.0 mL). A J. Young NMR tube, which was pre-dried at 180°C, was loaded with 0.4 mL of the above iron hydride complex solution and 0.14 mL of BEt₃ (1.0 M in hexane) C₆D₆ solution (with various BEt₃ concentrations - see Supporting Information). The NMR tube was placed into the NMR probe, which was equilibrated at the appropriate temperature (typically 319.2 K). After the sample had equilibrated for 2 min, ¹H NMR spectra were recorded at preset times using an automated program. After Fourier transform, phasing, calibrating and integrating each spectrum using MestRec, a plot of normalized [1] (y) vs. reaction time (x) was generated. Then these experimental data were fitted into the general, integrated equation: y₀+A₁*exp(−x/t₁), where y₀, A₁, t₁ are variables, t₁ being the reciprocal of the first order rate constant. Origin 6.1 was employed for the data fitting. The error bars on the rate constant came from the least-squares fit to the data.

Kinetic experiments: effect of [L^MeFeH]₂

BEt₃ (1.34, 1.0 M hexane solution) was mixed with C₆D₆ (1.86 mL) and used as a stock solution. A pre-dried J. Young NMR tube was charged with different amounts of 1 and BEt₃ (0.3 mL of standard solution), with a drop of metallic mercury (see text).

Kinetic experiments: activation parameters

In the range 280–300 K, temperatures in the NMR probe were calibrated using 100% methanol; in the range 300–380 K, 100% ethylene glycol was used for calibrations (Bruker Instruments, VT-Calibration manual). 1a (73.7 mg, 77.7 mmol) was dissolved in C₆D₆ (1.86 mL) then BEt₃ (1.34 mL, 1.0 M hexane solution) was added to make the reaction solution. For each temperature, 0.3 mL of the reaction solution was injected into a J. Young NMR tube with one drop of Hg⁰. The reaction solution was stored at −35 °C when it was not in use.

Supplementary Material

File0021. Supporting Information Available.

Synthetic, analytical, and kinetic details (PDF), and crystallographic data (CIF). This material is available free of charge via the Internet at http://pubs.acs.org.

Figure 3.

Eyring plot for the reaction of 1a + BEt₃ to give 2a and 3a.

Scheme 5.

Scheme 6.

Table 3.

Effect of [1a] on observed rate constant^a

Entry	Initial concentrations (mM)		Observed rate constant kobs (s−1)
	1a	BEt3
1	11.6	418.8	6.58(4)×10−3
2	20.8	418.8	4.59(2)×10−3
3	30.8	418.8	4.52(4)×10−3
4	41.6	418.8	4.27(8)×10−3

^aFor the effect of changes in [1a] in the absence of mercury, see SI.