Ammonia Synthesis at Room Temperature and Atmospheric Pressure from N₂: A Boron‐Radical Approach

Abstract

Ammonia, NH₃, is an essential molecule, being part of fertilizers. It is currently synthesized via the Haber–Bosch process, from the very stable dinitrogen molecule, N₂ and dihydrogen, H₂. This process requires high temperatures and pressures, thereby generating ca 1.6 % of the global CO₂ emissions. Alternative strategies are needed to realize the functionalization of N₂ to NH₃ under mild conditions. Here, we show that boron‐centered radicals provide a means of activating N₂ at room temperature and atmospheric pressure whilst allowing a radical process to occur, leading to the production of borylamines. Subsequent hydrolysis released NH₄ ⁺, the acidic form of NH₃. EPR spectroscopy supported the intermediacy of radicals in the process, corroborated by DFT calculations, which rationalized the mechanism of the N₂ functionalization by R₂B radicals.

Boron‐centered radicals, generated by reduction of R₂BCl derivatives, react with N₂ all the way to borylamine formation. DFT calculations rationalize the reduction/functionalization process, involving N−N bond splitting. Hydrolysis of the mixture yields NH₄ ⁺. Radical addition to N₂ provides a new strategy for N₂ fixation.

Introduction

Ammonia is one of the most abundantly produced chemicals worldwide. This compound is an essential component of fertilizers, providing nitrogen for plants to grow. Without fertilizers, the growing world's population could not be fed. Currently, 150 million t/y of NH₃ is produced worldwide; this figure is expected to more than double by 2050. ^[1] Ammonia is also a promising material as a transportation fuel. ^[2] Ammonia is a prime candidate for energy storage in the hydrogen economy. Indeed, the energy density of liquid ammonia (11.5 MJ L⁻¹), is higher than for liquid hydrogen (8.491 MJ L⁻¹) and for compressed H₂ at 690 bar and 15 °C (4.5 MJ L⁻¹). ^[3]

Around 78 % of our planet's atmosphere is made up of nitrogen gas (N₂) but its activation and conversion to NH₃ remains a challenging task due to the inertness of the dinitrogen. Plants have developed mechanisms to activate and fix nitrogen from N₂ using nitrogenase enzymes, but the process is slow and does not lend itself to industrialization. The industrial production of NH₃ uses the Haber–Bosch (HB) process, during which hydrogen and nitrogen react at a high temperature and pressure over a metal catalyst.[ ⁴ , ⁵ ] The process uses natural gas as raw material and has a heavy CO₂ footprint of 1.9 t of CO₂ per t NH₃. The HB process contributes 1.4 % of the world's CO₂ emissions. ^[6] The process can only be run economically in large, capital‐intensive units.

Decarbonizing the HB process can potentially be done by sequestering and storing the generated CO₂, ^[7] or by using the hydrogen gas produced by water splitting from renewable energy. ^[8] However, the primary step of the process—the actual HB ammonia production—remains energy and capital intensive. The process does not lend itself easily to medium‐sized production units that are aimed at using locally available renewable energy.

A clear need exists for a process to convert nitrogen to ammonia without the use of high temperatures and high pressures.

Researchers have been looking into developing processes for N₂ transformation into N‐containing derivatives (i.e., N₂ fixation) under mild, homogeneous conditions for several decades, especially processes that use carefully designed transition metal complexes (Figure 1, A). ^[9] The classical method for N₂‐activation at a metal center relies on a two‐way electron flow. Donation of electron density from N₂ to an empty orbital at the metal and back donation from a filled orbital of the metal to π* orbitals of N₂, weakens the strong N≡N bond.[ ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ ] Despite impressive results obtained recently, only a few complexes have been demonstrated to efficiently reduce N₂ to NH₃ using carefully optimized sources of protons and electrons.[ ¹⁵ , ¹⁶ , ¹⁷ ] Notably, this strategy was successfully applied most recently to main group boron species (borylene derivatives).[ ¹⁸ , ¹⁹ , ²⁰ ]

Figure 1.

A: Known catalytic N₂ to NH₃ transformations. B: Radical approach to N₂‐activation: our strategy. C: Computed additions of R₂B radicals on N₂. D: Successful N₂ to NH₄ ⁺ transformation via B‐centered radicals.

On the other hand, theoretical calculations show that the incorporation of N₂ can take place in some organic molecules.[ ²¹ , ²² ] The lack of experimental proof demonstrating the feasibility of such a direct insertion process is obviously due to kinetic limitations, i.e., the involvement of high‐energy transition states and intermediates.

For example, the addition of H₂ across N₂ to afford NH₃ implies the formation of diazene, HN=NH, as the first intermediate. Although this step leads to the formation of two strong NH bonds, calculations show that the reaction is far too endothermic to become feasible (ΔH°=56.0 and 51.0 kcal mol⁻¹ for cis and trans diazene, respectively).

We reasoned that the use of high‐energy radicals might provide a kinetically and thermodynamically favorable pathway to N₂ functionalization. The reaction between N₂ and the high‐energy radical H⋅ is, however, known to be endothermic (ΔH°=9.0 kcal mol⁻¹), which precludes a radical chain process initiated by H atom transfer. ^[23] To develop the first N₂ functionalization process based on radical addition, we performed a computational study (using density functional theory, DFT) to identify radicals (R⋅) forming compounds of the type RN₂⋅ in an exothermic (and exergonic) reaction (Figure 1C). We inferred that, in addition to creating a strong σ‐interaction, delocalization of the radical onto the π* system of N₂ could be the key to such an exergonic process (Figure 1B). ^[24] Calculations, therefore, focused on various R₂B⋅ derivatives. Experimental work was subsequently dedicated to generating such radicals under mild conditions. We report here the first functionalization of N₂ with three radical derivatives, namely Cy₂B⋅, (DIP)B⋅ (DIP=Di‐isopinocampheyl) and Bis(bicyclo[2.2.1]‐2‐heptyl)B⋅, generated at room temperature in the presence of potassium (K). NMR and EPR spectroscopies and DFT calculations have unambiguously established the in situ formation of borylamines from N₂ and boron‐centered radicals. NH₄ ⁺ was then obtained by simple hydrolysis of the borylamine intermediates, thereby providing an unprecedented strategy for the transformation of dinitrogen to ammonia under mild conditions.

Results and Discussion

Reaction Optimization: NH₄ ⁺ Synthesis from N₂

As a first step, the additions of CatB⋅ (Cat=catechol) and Cy₂B⋅ on N₂ were computed by DFT (Figure 1, C). ^[25] Both radicals are computed to have an exothermic (ΔH=−16.9 kcal mol⁻¹, CatB; ΔH=−32.1 kcal mol⁻¹, Cy₂B) and exoergic (ΔG=−7.5 kcal mol⁻¹, CatB; ΔG=−22.1 kcal mol⁻¹, Cy₂B) reaction with N₂. According to these calculations, both radicals were promising candidates to react with N₂ and we set out to generate them in solution under N₂ upon one‐electron reduction of the corresponding R₂BCl derivative. It should be noted, however, that few R₂B−BR₂ compounds are known to be synthesized via this reductive strategy under N₂, implying favored radical coupling rather than addition on N₂.[ ^26a , ^26b , ^26c ] The efficiency of the BB bond formation appeared to depend on the substituents at B, with the presence of donor atoms being very favorable. ^[27] The donor atoms likely stabilize the radical, thereby increasing its concentration and thus the radical coupling. Accordingly, several (RO)₂B−B(OR)₂ compounds have been synthesized in good to excellent yields from (RO)₂BCl and Na/Hg under N₂ in toluene at 90 °C. ^[28] Therefore, the reduction of Cy₂BCl to form a higher‐energy non‐stabilized boryl radical was chosen as the more promising candidate. ^[29] Thus, as a preliminary investigation a stoichiometric mixture of Cy₂BCl and K (2.0 mmol) in THF (12 mL) was stirred for 12 h under N₂ (1 bar), followed by hydrolysis with an excess of HCl in Et₂O to convert all N containing species into NH₄Cl. Quantification was achieved by dissolution of the solid residue and integration vs. 1,3,5‐trimethoxybenzene as internal standard by ¹H NMR spectroscopy (see ESI for details on the quantification method). It revealed that NH₄ ⁺ is indeed produced, albeit in a small yield (0.046 mmol, 7 % yield, Table 1, entry 1). Further studies then showed that this yield could be significantly improved, up to 41 %, by lowering the concentration of the reagents and by using an excess of K (entry 2). A thorough screening of solvents and reducing agents was carried out under these conditions.

Table 1.

Exploring the reaction parameters: step 1) N₂+R₂BCl+x red. (6 h, solvent, temperature, N₂ pressure); step 2: hydrolysis with excess HCl; step 3 quantification of NH₄ ⁺. Standard reaction conditions: 0.16 mmol Cy₂BCl unless otherwise specified; stirring at 700 rpm at room temperature for 6 h, open flask in glove box. Yield is calculated as 3×n(NH₄ ⁺)/n(Cy₂BCl)×100 thus considering only N(BCy₂)₃ to be formed during the reaction (see text). All reactions were reproduced at least twice (entry 2, done 6 times). Yield accuracy was estimated at ±4 %.

	[R2BCl] [mol L−1]	Solvent	Reducing agent	Stoich. Red	PN2 [bar]	Yield
1	0.17	THF	K	1	1	7
2	0.04	THF	K	2.5	1	41
3	0.04	MeTHF	K	2.5	1	0
4	0.04	Pentane	K	2.5	1	0
5	0.04	Toluene	K	2.5	1	0
6	0.04	Et2O	K	3.3	1	0
7	0.04	Dioxane	K	3.3	1	Trace
8	0.04	THF	Na	2.5	1	0
9	0.04	THF	Na/Hg	2.5	1	0
10	0.04	THF	Na/naphtalene	2.5	1	0
11	0.04	THF	KC8	2.5	1	0
12	0.04	THF	SmI2	2.5	1	0
13[a]	0.04	THF	K	2.5	1	0
14[b]	0.04	THF	K	2.5	1	0
15[c]	0.02	THF	K	2.5	1	56
16	0.02	THF	K	1	1	30
17[d]	0.04	THF	K	2.5	1	56
18	0.04	THF	K	2.5	20	60
19	0.04	THF	K	2.5	40	76
20	0.04	THF	K	2.5	80	94
21[e]	0.04	THF	K	2.5	1	48
22[f]	0.08	THF	K	2.5	1	38
23[g] CatBCl	0.04	THF	K	2.5	1	0
24[h] DIP‐Cl	0.04	THF	K	2.5	1	28
25[i] BCHBCl	0.04	THF	K	2.5	1	43

[a] reaction at 60 °C; [b] reaction at −70 °C; [c] reaction conducted on a 0.08 mmol scale; [d] K was cut into 5 pieces; [e] reaction conducted on a 10 times scale compared to standard; [f] reaction conducted on a 100 times scale compared to standard; [g] CatBCl=catechol‐chloroborane; [h] DIP‐Cl=Di‐isopinocampheyl‐chloroborane; [i] BCHBCl=Bis(bicyclo[2.2.1]‐2‐heptyl)chloroborane.

No NH₄ ⁺ production was observed in toluene, Et₂O, pentane or 2Me‐THF, no matter what the reducing agent was, while trace amounts were observed in 1,4‐dioxane (entries 3–7). We then tested the use of several strong one‐electron reducing agents, such as Na, Na/Hg, Na/naphthalene, KC₈ and SmI₂ in THF. Despite the fact that the chloroborane is reduced, as attested by the formation of BH species in the crude mixture by ¹¹B NMR (vide infra), in none of cases, NH₄ ⁺ was observed after hydrolysis of the crude mixture (entries 8–12). To ascertain that all reagents were needed for the reaction to proceed, the control reactions (under Ar; without K; and without Cy₂BCl) were carried out. As expected, in none of these cases was NH₄ ⁺ observed after hydrolysis. Further optimizations were carried out to increase the efficiency in N−B bond formation, and thus the yield of NH₄ ⁺ after hydrolysis. Due to the use of solid K, the stirring efficiency was found to be particularly important. A strong agitation was most likely required to prevent passivation of the K surface.

The first parameter evaluated was the stoichiometry between K and Cy₂BCl (entries 16 vs. 1). A 2.5‐fold excess of K was found to be favorable, although some of it remained unconsumed at the end of the reaction. Decreasing the concentration of chloroborane from 0.04 to 0.02 mol L⁻¹ led to a significant increase in yield (entries 15 vs. 2), which was, however, counterbalanced by cutting the K into several pieces (entries 17 vs. 15).

The effect of N₂ pressure was also tested. Increasing the pressure to 20 and 40 bars resulted in major improvements in the yields: up to 60 % and 76 %, respectively (entries 18 and 19 vs. 2). At 80 bar, a 94 % yield was obtained (entry 20), attesting a highly efficient N₂ functionalization by the Cy₂B radicals. Running the reaction on a ten‐fold scale provided a slightly improved yield (entry 21, 48 % vs. 41 % entry 2), while a hundred‐fold scale reaction (with doubled concentration) showed a very moderate decreased yield (entry 22, 38 % vs. 41 % entry 2). Finally, two experiments were carried out with ¹⁵N₂ gas, either mixed with ¹⁴N₂ or pure (ESI). ¹⁵NH₄ ⁺ was observed after hydrolysis when pure ¹⁵N₂ was used, while both ¹⁵NH₄ ⁺ and ¹⁴NH₄ ⁺ were obtained as expected from the mixture providing definitive evidence of the direct functionalization of N₂ under these conditions. The kinetics of the reaction were measured by hydrolysis and ¹H NMR quantification of aliquots under conditions of entry 2. These studies led to the conclusion that the reaction is essentially completed after 2 h (ESI), even though the color of the medium continues to darken over time.

NMR Monitoring: Identification of Borylamines and Boron Containing Side Products

All of the above reactions were monitored by ¹¹B NMR with the aim to characterize the products prior to hydrolysis. Unfortunately, the N−B‐containing products were not observed, most likely because of signal broadening due to N−B coupling. On the other hand, side products corresponding to the known Cy₂BH₂ ⁻ (δ=−9.0 ppm, triplet) and Cy₂BH (δ=17.0 ppm, doublet) could be observed in variable proportions, depending on the applied conditions. When stoichiometric amounts of K and Cy₂BCl were used (entry 16), only Cy₂BH was observed, forming colorless crystals of (Cy₂BH)₂ upon concentration of crude mixtures. On the other hand, the anion Cy₂BH₂ ⁻ was formed when an excess of K was used for the reaction. The observation of signals attributed to Cy₂BH₂ ⁻ and (Cy₂BH)₂ in the ¹H coupled ¹¹B spectrum of the reaction carried out in THF‐D8 demonstrated that intramolecular H migration to the B‐centered radical is faster than the intermolecular abstraction of D atom from the solvent. ^[29]

We also found that the intensity of these two signals is inversely proportional to the yield in NH₄ ⁺ obtained after hydrolysis, indicating that the formation of B−H derivatives competes with the coupling of the B‐based radical with N₂.

Multinuclear NMR experiments (¹H, ¹⁵N, ¹⁵N‐¹H HMBC) demonstrated that a mixture of two borylamine compound(s), N(BCy₂)₃ and NH(BCy₂)₂ were generated by the reduction process, with relative ratios also depending on the experimental conditions (see ESI). Above all, NMR analyses carried out on samples taken before hydrolysis of a reaction performed under the conditions of entry 20, previously found to give the best yield in NH₄ ⁺ without BH‐type secondary products, revealed that N(BCy₂)₃ is formed with a yield exceeding 90 %. The ¹⁵N‐¹H HMBC spectrum of a reaction carried out in THF‐D8 showed a cross‐signal for the NH(BCy₂)₂ compound, thereby pointing to a faster intramolecular HAT (H atom transfer) process at a generated N radical, than D abstraction from the solvent.

With these results in hand, other halogeno boranes were tested under optimized conditions. The chlorocatecholborane did not lead to formation of any NH₄ ⁺ after hydrolysis (entry 23), which implies, in accord with earlier reports, ^[28] that the B−B bond forming process is favored over that of the N−B bond. On the other hand, two other dialkylchloroborane were used successfully to achieve this transformation. Using DIP‐Cl under the standard conditions used in entry 2 resulted in the formation of NH₄ ⁺ in 28 % yield (entry 24), to be compared with the 41 % and 43 % yields obtained under the same conditions with Cy₂BCl and Bis(bicyclo[2.2.1]‐2‐heptyl)chloroborane (entry 25), respectively. These results clearly bring to light the major influence of the alkyl substituents at B, an effect which will be the subject of a subsequent study.

EPR Monitoring: Evidence of Organic Radical Formation

The proposed reaction scheme, involving formation of organic radicals, was further confirmed by EPR measurements (X band) carried out either at room temperature or at 110 K. The samples were prepared under the same conditions as above, i.e., in a dedicated glove box under N₂, upon mixing at room temperature under vigorous stirring Cy₂BCl and K in dry THF (entry 17). EPR analyses were conducted on aliquots collected at different reaction times. The spectra collected after 90, 150 and 210 minutes of stirring are shown in Figure 2.

Figure 2.

EPR spectra recorded at RT after stirring a mixture of Cy₂BCl/K in THF under N₂ for a) 90; b) 150, c) 210 minutes. d) Experimental spectrum recorded at 110 K after 4 h of stirring; e) simulated spectrum using easyspin (Matlab toolbox).

The spectra in Figure 2, A reveal that stable organic radical(s) accumulate in solution after one hour of reaction. At first sight, those spectra seem to consist of three lines of relative intensity: 1/2/1. Further studies on longer reaction times, however, revealed that the intensity of the central signal at g=2.0034 increases more rapidly than the two peripheral ones. This outcome is consistent with the conclusion that the apparent triplet developing over time in fact results from the presence of at least two different organic radicals. We also found that the temperature (RT vs. 110 K) of the resonator has only a limited effect on the number, shape and relative intensity of those signals and that those signals eventually disappear over time. Key insights into the nature of the radicals involved in solution were provided by simulation of the experimental data. The three‐line signal showing a large peak‐to‐peak line width could be readily simulated with a 1 : 1 : 1 triplet at g=2.0036 attributed to a ¹⁴N (I=1)‐centered radical featuring a hyperfine constant, a_N=11.8 G (see Figure S25, ESI). ^[30] The broad and unsmooth character of the experimental lines (Figure 2) is, moreover, consistent with the existence of low, superhyperfine, and thus unresolved, interactions (<1 G) with surrounding hydrogen atoms. Such features could correspond to the computed intermediate radicals I, or more likely M proposed in Figure 4 (vide infra). Attributing the additional central singlet at g=2.0034 remains hypothetical at this stage. Based on relevant previous studies involving similar organic radicals,[ ³¹ , ³² , ³³ , ³⁴ , ³⁵ ] we propose an attribution of that additional signal either to a C‐ or B‐centered radicals featuring low, and thus unresolved, hyperfine coupling constants (I[B¹¹]=3/2, I[B¹⁰]=3). This attribution was further supported by EPR measurements performed under an Ar atmosphere. In the absence of N₂, we found that the colorless solution remains EPR silent until a brown red color appears leading to the development of one intense singlet signal at g=2.0034 (see Figure S26, ESI) consistent with the formation of the radical discussed above. Taken together, these measurements provide unambiguous evidence supporting the existence of in situ generated organic radicals, most notably nitrogen‐centered species as intermediates in the direct chemical reduction of Cy₂BCl by K in THF under N₂.

Figure 4.

Computed pathway: the energies are relative to “6 A+N_2′′. Functionalization to the hydrazine (Cy₂B)₂NN(BCy₂)₂ derivative (K) followed by reductive NN bond splitting process leading to bis‐borylamide N.

DFT Calculations: Reaction Mechanism

To shed light on the reaction mechanism, DFT calculations were carried out. Only the lowest energy pathway leading to N(BCy₂)₃ is presented here. Not surprisingly, in light of the poor Lewis basic properties of N₂, attempts to locate an adduct with Cy₂BCl failed. The reactivity between N₂ and the reduced radical anion Cy₂BCl⋅⁻ (A), generated through one‐electron reduction of Cy₂BCl by potassium, was investigated (Figure 3). Two pathways were computed from A: coordination of N₂ vs. Cl⁻ elimination. The latter pathway is kinetically unfavorable (ΔΔG ^≠=8.9 kcal mol⁻¹), as depicted in Figure 3. A transition state TS_A‐B for N₂ coordination was located at ΔG ^≠=7.9 kcal mol⁻¹ above A. N₂ coordination is exoergic (ΔG=−8.2 kcal mol⁻¹) and leads to the radical anion B with a clear nitrogen‐boron interaction (N−B=1.558 Å). Upon dinitrogen coordination to A, the N−N bond elongates from 1.097 Å to 1.174 Å. Analysis of the electronic structure of B indicates that the N−B bond results from the donation of the B‐centered electron into one π* orbital of N₂, in agreement with the elongation of the N−N bond upon coordination. Dissociation of Cl⁻ from B to form the radical Cy₂BNN⋅ C is computed to be a favorable pathway (ΔG=−5.0 kcal mol⁻¹), resulting in an even shorter N−B bond (1.413 Å). In C, the spin density has significant weight on the terminal nitrogen atom (see Figure S27). Therefore, the creation of a second B−N bond upon interaction with the radical anion A, is computed to be strongly exoergic (ΔG=−44.9 kcal mol⁻¹) to form E. Diazene F, the result of a two‐electron reduction of N₂ with concomitant formation of two N−B bonds, is obtained upon chloride dissociation from E. These two electrons have populated one π* orbital on N₂ and, due to the unsaturated nature of boron, a two‐electron delocalized π‐type interaction develops between the B−N−N−B atoms. This interaction is illustrated by the linear geometry for the four atoms with rather short B−N bonds, an elongated N−N bond (B−N=1.37 Å and N−N=1.18 Å; N−N=1.25 Å in free H₂N₂) and the nature of the HOMO (see Figure S30). ^[36]

Figure 3.

Computed pathway: the energies are relative to “6 A+N₂”. Addition of A (Cy₂BCl⋅⁻) to N₂ up to the formation of three N−B bonds (compound H, three‐electron reduction of N₂). Note that the energy positioning of the various species does not take into account any potential stabilizing interactions developing between K⁺ and Cl⁻ in the experimental situation.

One‐electron reduction of F by A populates the remaining N₂‐centered π* orbital in an exoergic process (ΔG=−23.3 kcal mol⁻¹) to yield the radical anion G featuring a zig‐zag geometry for the B−N−N−B linkage (B−N=1.380 Å, N−N=1.284 Å and B−N−N=136.5°). Coordination of Cy₂BCl affords the radical anion H (ΔG=−12.4 kcal mol⁻¹), from which chloride readily dissociates to yield the neutral radical I featuring three N−B bonds. A strongly exoergic (ΔG=−37.6 kcal mol⁻¹) one‐electron reduction of I by A populates the π* orbital between the two nitrogen atoms and yields the anionic system J (Figure 4) featuring an N‐centered lone pair on the nitrogen with only one N−B bond (see Figure S31). Subsequent nucleophilic substitution at Cy₂BCl (ΔG=−31.7 kcal mol⁻¹) forms the four boron‐substituted hydrazine K.

The bond‐dissociation energy of the N−N bond in hydrazine K is 46.1 kcal mol⁻¹; reduction is required for subsequent N−N bond breaking. Thus, one‐electron reduction of K by A is computed also to be endoergic but by only ΔG=8.0 kcal mol⁻¹ to yield L. The transition state TS_{L −M+N} associated with N−N bond cleavage is computed with a very small activation barrier (ΔG ^≠=1.5 kcal mol⁻¹) to form the radical (Cy₂B)₂N⋅ M and the anion (Cy₂B)₂N⁻ (N) with ΔG=−47.0 kcal mol⁻¹. The one‐electron reduction of M by A to form N and Cy₂BCl is computed to be strongly exoergic (ΔG=−49.7 kcal mol⁻¹). This step concludes the six‐electron reduction of N₂ to form two equivalents of the anion N. This boron‐substituted amide features a linear geometry with two perpendicular B−N π‐orbitals similar to the allene structure. Finally, tris‐borylamine N(BCy₂)₃ P is formed by a nucleophilic substitution between N and Cy₂BCl (Figure S32).

In summary, these computations highlight the facile and favorable coordination of N₂ to the Cy₂BCl radical anion A obtained by one‐electron reduction by K, as the first and crucial step in the functionalization process. Notably, coordination of N₂ to this radical anion prior to chloride elimination appears to be kinetically preferred over the reverse process. The overall transformation is strongly exoergic because of the high reactivity of the radical anion A, Cy₂BCl⋅⁻.

Having rationalized the formation of N(BCy₂)_3, as well as NH(BCy₂)₂ (Figure S34), the competing formation of BH‐containing byproducts required investigation. The intramolecular H migration from the cyclohexyl to B was thus computed (Figure S33). In accordance with the literature, ^[28] this migration proved both facile and thermodynamically favorable. The lowest energy pathway was found via the generation of the radical Cy₂B⋅ D. The transition state was found to be only 13.3 kcal mol⁻¹ higher than Cy₂B⋅, and thus 22.2 kcal mol⁻¹ above A, readily accessible at room temperature. The resulting C‐centered radical is computed at −18.5 kcal mol⁻¹ relative to D (−9.6 kcal mol⁻¹ vs. A). Most importantly, the addition of N₂ to the in situ generated R₂BCl radical anion A to form the first BN bond is lower in energy than the competing HAT from the neighboring Cy to B that creates the BH bond leading to side‐product formation. This kinetic preference is crucial in the successful N₂ functionalization.

Conclusion

In conclusion, we present the first successful functionalization of N₂ using boron‐centered radicals, at room temperature and low pressures (from one atmosphere). These radicals are efficiently generated in situ by reduction of the corresponding, readily available chloroborane derivatives. We believe that these results open new avenues to N₂ functionalization notably different from the high‐energy heterogeneous HB process and the bio‐inspired homogeneous transformations. These findings could drastically affect the carbon footprint of ammonia production and pave the way for new applications for ammonia, particularly in renewable energy storage and the hydrogen economy. ^[37]

Supporting Information

See Supporting Information for experimental section, characterization data, spectra, as well as EPR measurements and DFT calculations (details of computations, mechanisms leading to NH and BH bond formations).

Conflict of interest

A patent based on the present work has been filed on May 20, 2021 under the number: FR2105295, with inventors S. Bennaamane and N. Mézailles. (mentioned in our article as ref [37]) The patent has been licensed to SWAN‐H. N.M. is co‐founder and president of SWAN‐H.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

In memory of Pascal Le Floch

Bennaamane S., Rialland B., Khrouz L., Fustier-Boutignon M., Bucher C., Clot E., Mézailles N., Angew. Chem. Int. Ed. 2023, 62, e202209102; Angew. Chem. 2023, 135, e202209102.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.