Catalyst‐Free Selective Reduction of Nitrogen Dioxide to Nitric Oxide

ABSTRACT

Nitrogen dioxide (NO₂) is a highly toxic gas and a notorious respiratory irritant that contributes to the greenhouse effect and acid rain. Herein, we study the catalyst‐free reduction of NO₂ with boranes, silanes, and the hydrogen surrogates 1,4‐cyclohexadiene and isopropyl alcohol, resulting in the selective formation of nitric oxide (NO). Selective NO generation was confirmed by electron paramagnetic resonance spectroscopy and low‐temperature ¹⁵N{¹H} nuclear magnetic resonance spectroscopy. Additionally, density functional theory calculations have been employed to probe the mechanism of these reductions.

Common main group reductants selectively convert nitrogen dioxide to nitric oxide in a catalyst‐free manner. Computational studies suggest divergent reaction pathways depending on the reductant employed. Where boranes and silanes proceed via a two‐electron process, and hydrogen surrogates undergo stepwise one‐electron processes.

1. Introduction

Combustion engines are pervasive in industry and transportation, and at elevated temperatures, they generate nitrogen oxides, namely nitrogen dioxide (NO₂) and nitric oxide (NO). The generation of NO₂ is particularly undesirable due to its deleterious environmental and health effects [1]. Exposure to NO₂ has been associated with increased mortality and heightened susceptibility to cardiovascular and respiratory disease [2]. Additionally, NO₂ has been implicated in the formation of tropospheric ozone [3], photochemical smog [4], acid rain [5], and the indirect enhancement of the greenhouse effect [6]. Current technologies that degrade NO_x emissions require the use of noble metal catalytic converters [7]. These catalysts operate at elevated temperatures (> 200 °C) and rely on scarce, expensive, and depleting transition metals.

NO₂ can be degraded by single‐electron reduction to nitrite salts or nitrous acid (HONO), or via a two‐electron reduction to NO (Figure 1). In contrast to NO₂, NO is considerably less toxic and is stable in vivo [8]. NO plays a critical role as a biochemical signaling molecule, responsible for mediating a wide array of physiological responses. Mineral aerosol substrates [9, 10], polycyclic aromatic hydrocarbons [11], and hydrated bulk minerals [12] are all known to mediate single‐electron pathways for NO₂ degradation, while two‐electron pathways are less widely explored. In 2012, Mullins reported the complete and selective reduction of NO₂ to NO on hydrogen‐precovered Au at cryogenic temperatures [13]. Later, in 2021, Grützmacher reported a homogeneous Rh(I)–Pt(II) based heterobimetallic catalyst capable of hydrogenating NO_x gases, including NO₂ [14]. Furthermore, a recent report by Chen and co‐workers illustrates how mixtures of oxalic acid and Fe₂O₃ can also facilitate this transformation [15]. There has been global interest in developing catalyst‐free protocols as sustainable alternatives to transition‐metal mediated processes. Herein, we study the reactivity of organo‐main group reductants, including boranes, silanes, cyclohexadiene, and isopropyl alcohol, with NO₂ in the absence of a catalyst and demonstrate the selective production of NO.

FIGURE 1.

(a) Products from NO₂ reduction. (b) Transition metal‐mediated reductions of NO₂. (c) This work, catalyst‐free reduction of NO₂ by main group reductants.

2. Results and Discussion

First, excess NO₂ (∼1 atm, 25°C) was added to a degassed C₆D₆ solution of 4,4,5,5‐tetramethyl‐1,3,2‐dioxaborolane (HBpin), yielding an orange solution. Rapid gas evolution was observed, and the reaction mixture gradually turned green over the course of 5 min. Freezing the reaction mixture in liquid nitrogen (−196°C) after 5 min resulted in the formation of a bright blue solid. The observed thermochromism is consistent with the formation of dinitrogen trioxide (N₂O₃), a known product formed from NO and NO₂ coupling [16, 17]. After 30 min, the reaction mixture was analyzed by ¹H and ¹¹B{¹H} NMR spectroscopy, which revealed the complete consumption of HBpin and formation of HOBpin and pinBOBpin (88% and 12%, respectively, by ¹H NMR using toluene as an internal standard) and H₂ (Figure 2). The reaction mixture was then cooled to −20°C, to freeze any residual NO₂ (which has a melting point of −11.2°C) [17], and the remaining volatile products were vacuum transferred to an electron paramagnetic resonance (EPR) tube. Initially, no resonances could be detected by EPR spectroscopy at room temperature or −123°C (150 K). A separate control experiment of only NO₂ in C₆D₆ confirmed that NO₂ gives rise to an EPR resonance as a broad singlet at g = 2.001. Following the addition of excess HBpin to a solution of NO₂, no resonances could be detected by EPR spectroscopy, indicating the complete consumption of NO₂. It has previously been reported that it can be challenging to directly observe NO by EPR spectroscopy, due to rapid spin relaxation arising from large spin‐orbit coupling and collisional broadening effects associated with gaseous radicals [18, 19]. Thus, the reaction was repeated as initially described and, after freezing the solution, the headspace was transferred to an EPR tube containing a solution of [Fe(detc)₃] (detc = diethyldithiocarbamate), a known NO spin trap [20]. Analysis of this reaction mixture by continuous wave (CW) EPR spectroscopy successfully detected the EPR active complex [Fe(NO)(detc)₂]. Additionally, known NO trapping reagent [T(O‐Me)PP(Co)] was employed in a similar manner to further confirm the generation of NO. Analysis of the reaction mixture by ¹H NMR spectroscopy confirmed the formation of [T(O‐Me)PP(Co)(NO)], as expected (see Section S6.2). These experiments confirm the formation of NO in the headspace when HBpin is allowed to react with NO₂.

FIGURE 2.

(a) Reduction of NO₂ with HBpin. (b) Spin trapping of NO by [Fe(detc)₃] to generate the EPR active species [Fe(NO)(detc)₂]. (c) Experimental CW EPR spectrum of spin‐trapped NO by [Fe(detc)₃] at room temperature (experimental parameters: solvent: toluene, frequency: 9.367292 GHz, receiver gain: 30 dB, g = 2.038, blue). Simulation of experimental CW EPR spectrum (1 x A_N = 35.5 Hz) (grey). CW EPR spectrum of independently synthesized [Fe(NO)(detc)₂] from the reaction [Fe(detc)₃] with NO at room temperature (experimental parameters: solvent: toluene, frequency: 9.372613 GHz, receiver gain: 30 dB, g = 2.038, red).

Isotopically labelled ¹⁵NO₂ was generated by reacting ¹⁵N‐labelled nitric acid with elemental copper [21]. The gaseous product was then vacuum transferred to a separate ampoule and repeatedly degassed to remove any traces of ¹⁵NO (see Sections S5 and S8). A degassed solution of HBpin in d₈‐toluene was then pressurized with ¹⁵NO₂, and was allowed to react at room temperature for 10 min. The reaction mixture was then cooled to −50°C, at which point a distinct blue color was observed. The reaction mixture was subsequently analyzed by ¹⁵N{¹H} NMR spectroscopy at −50°C. At this temperature, NO₂ is known to dimerize to form the diamagnetic product N₂O₄. Additionally, the diamagnetic adduct of NO and NO_2, N₂O₃, is also known to form. It is worth noting that the dimer of NO, N₂O₂, is significantly weaker and more unstable [22], and thus was not anticipated to be detectable by NMR spectroscopy in these experiments. Low‐temperature ¹⁵N{¹H} NMR spectroscopy confirmed the presence of N₂O₃ with two broad resonances at 671 and 439 ppm, and N₂O₄ at 360 ppm (Figure 3). Upon warming to room temperature, no resonances could be detected by ¹⁵N{¹H} NMR spectroscopy, in line with the formation of diamagnetic adducts at low temperature as opposed to persistent diamagnetic species. These experiments also show that NO₂ and NO, along with their corresponding dimers at low temperature, are the only nitrogen‐containing products in the reaction mixture. The reaction mixture was allowed to react for a further 15 min at room temperature, after which it was cooled to −50°C and reanalyzed by ¹⁵N{¹H} NMR spectroscopy. No resonances could be detected, suggesting the complete hydroborative reduction of ¹⁵NO₂ to ¹⁵NO. The reaction headspace was vacuum transferred to a J‐Young EPR tube charged with a solution of [Fe(detc)₃] in toluene, and EPR spectroscopy again confirmed the formation of [Fe(¹⁵NO)(detc)₂] (with a distinct N hyperfine coupling x1 A_N = 45.5 Hz), the ¹⁵NO spin trapped product. The larger hyperfine coupling is in agreement with ¹⁵N's larger gyromagnetic ratio [23].

FIGURE 3.

(a) Low‐temperature (−50°C) ¹⁵N{¹H} NMR spectrum showing N₂O₃ and N₂O₄. (b) Spin trapping of ¹⁵NO and CW EPR spectrum at room temperature (experimental parameters: solvent: toluene, frequency: 9.739740 GHz, receiver gain: 30 dB, g = 2.038, blue) and simulation of [Fe(¹⁵NO)(detc)₂] (1 x A_N = 45.5 Hz, grey).

In order to gain insight into the mechanism, several pathways were postulated and computationally assessed using r²SCAN‐3c composite method with conductor‐like polarizable continuum model (CPCM) benzene solvation. Given the radical nature of NO₂, single‐electron mechanisms were considered. However, mechanisms involving H atom abstraction or the homolytic cleavage of HBpin were found to be energetically inaccessible (see Section S9.1). Furthermore, experimental observations indicate that pathways involving ^•Bpin are unlikely, as evidenced by B₂pin₂ showing no reaction with NO₂. Thus, two‐electron hydride mechanisms were explored computationally. Figure 4 summarizes a computationally viable mechanism, with adduct formation between NO₂ and HBpin being rate‐determining to yield Int‐1. Next, Int‐1 can rearrange into TS‐2 in a near‐barrierless fashion, followed by hydride migration to give HOBpin and NO via TS‐3. Following on from Int‐1, both three‐ and five‐membered transition states were also considered and discounted due to their higher energetic barriers (See Sections S9.2 and S9.3).

FIGURE 4.

Calculated reaction pathway for the formation of HOBpin, pinBOBpin, and H₂ gas from the reduction of NO₂ with HBpin. Gibbs energies in kcal mol⁻¹ (r²SCAN‐3c composite method with CPCM benzene solvation). Intermediates have been drawn as Lewis structures, with spin density calculations provided in Section S9.5.

Since both HOBpin, pinBOBpin, and H₂ are observed in the reaction mixture, we postulated that HOBpin could react with excess HBpin in the reaction mixture to give pinBOBpin and H₂. However, when independently synthesized HOBpin was reacted with one equivalent of HBpin in C₆D₆, no reaction was observed at room temperature over the course of a day. Furthermore, the direct dehydrocoupling between HOBpin and HBpin was computed to be prohibitively high (TS‐4’ ΔG^‡ = 45.3 kcal mol⁻¹). Although, it should be noted that analogous dehydrocoupling reactions can be mediated by Brønsted‐Lowry bases or Lewis acids [24, 25, 26, 27, 28], and thus these pathways cannot be completely ruled out. Nonetheless, we considered a separate mechanism which could lead to the formation of pinBOBpin and H₂ (shown in Figure 4). Instead, Int‐2 can react with a second equivalent of HBpin via TS‐4 (ΔG^‡ = 32 kcal mol⁻¹) to give Int‐4, a diborylated aminyl‐type radical, together with H₂. The relatively high barrier is consistent with the observed selective formation of HOBpin over pinBOBpin in this reaction (88% vs. 12%).

Next, various main group reducing agents were screened to probe the generality of this method. HB(C₆F₅)₂, catecholborane, and ammonia‐borane were tested and found to yield a complex mixture of products consistent with decomposition products. While triethylsilane (Et₃SiH) was found to react in an analogous fashion to HBpin, albeit not to completion, affording Et₃SiOH (74%) and Et₃SiOSiEt₃ (10%) over the course of 4 h (Table 1). Triphenylsilane (Ph₃SiH) displayed similar reactivity, forming only Ph₃SiOH (74%) after 12 h. Whereas reaction of hexamethyldisilane (Si₂Me₆) with NO₂ showed no reaction even after 12 h. Hydrogen‐transfer reagents were also investigated; these reagents are attractive reductants as they bypass the safety hazards associated with using flammable H_2(g) [29, 30]. A degassed solution of 1,4‐cyclohexadiene (CHD) in C₆D₆ was reacted with excess NO₂ (∼1 atm, 25°C), resulting in the immediate formation of a persistent bright green solution. ¹H and ¹³C{¹H} NMR spectroscopy confirmed the clean and quantitative conversion of 1,4‐cyclohexadiene to benzene. No other organic products were detected by ¹H or ¹³C{¹H} NMR spectroscopy. Water was not observed in the ¹H NMR spectrum; however, a broad signal at ∼6 ppm was detected, in line with the expected resonance of nitrous acid (HONO) [31], possibly formed by water reacting with N₂O₃ [16]. Spin trapping experiments using [Fe(detc)₃] again confirmed the generation of NO from this reaction mixture. EPR spectroscopy was employed to assess whether the final reduction product was NO or HONO. HONO is known to be unstable and decomposes into NO₂, NO, and water. If HONO were the final reduction product, we postulated that the regeneration of NO₂ would be observable by CW EPR spectroscopy. Thus, an EPR tube charged with a solution of excess 1,4‐cyclohexadiene in C₆D₆ was degassed and pressurized with NO₂. The reaction was monitored daily by EPR spectroscopy, and regenerated NO₂ was not detected over the course of one week. Therefore, it is likely that 1,4‐cyclohexadiene reacts with NO₂ to give HONO as an intermediate, which is then further reduced to NO. Both radical and proton‐hydride mechanisms were probed computationally, and proton‐hydride step‐wise transfer mechanisms were found to be energetically infeasible (see Section S9.6).

TABLE 1.

Scope of main group reductants tested in the catalyst‐free reduction of NO₂.

Reductant	t	Reductant consumption (%) a
HBpin	30 min	100
B2pin2	12 h	0
Et3SiH	4 h	84
Ph3SiH	12 h	74
Si2Me6	12 h	0
CHD	5 min	100
IPA	12 h	100

^a Determined by ¹H NMR spectroscopy using toluene as an internal standard. CHD = 1,4‐cyclohexadiene; IPA = isopropyl alcohol.

Figure 5 presents a plausible single‐electron transfer mechanism for how 1,4‐cyclohexadiene reduces NO₂. Significant radical spin density in NO₂ is located on the N atom [32, 33], which can abstract a H^• from 1,4‐cyclohexadiene to give O(H)NO (Int‐7) via TS‐7, which is slightly more energetically favorable than TS‐7’. Intramolecular isomerization of ON(H)O to HONO was found to be kinetically disfavored (see Section SS9.7, TS‐d: ΔG^‡ = 47.2 kcal mol⁻¹). However, isomerization via a six‐membered intermolecular transition state, TS‐8, is energetically feasible and converts Int‐7 to HONO (Int‐8). Finally, HONO reacts with cyclohexadienyl radical to afford NO, benzene, and water. This computed mechanism is in agreement with Mullins’ gold‐mediated hydrogenation mechanism, where hydrogen atoms first reduce NO₂ to HONO, which is then further reduced to NO [13].

FIGURE 5.

Calculated reaction pathway for the reduction of NO₂ by 1,4‐cyclohexadiene. Gibbs energies in kcal mol⁻¹ (r²SCAN‐3c composite method with CPCM benzene solvation).

Isopropyl alcohol (IPA) was also tested as a hydrogen‐transfer reductant. When reacted with excess NO₂, IPA was fully consumed within 12 h, forming only acetone (27%) and isopropyl nitrite (73%), confirmed by ¹H and ¹³C{¹H} NMR spectroscopy. A broad resonance at ∼6 ppm was again observed in the ¹H NMR spectrum, suggesting HONO as an intermediate. Furthermore, the formation of isopropyl nitrite is consistent with the presence of HONO in the reaction mixture [34]. Additionally, N₂O₃ (the NO₂ and NO adduct) is also known to be an active nitrosating agent for alkyl alcohols [35]. Spin trapping experiments using Fe(detc)₃confirmed the formation of NO when IPA was employed as a reductant.

3. Conclusion

In summary, we report the selective reduction of NO₂ to NO using common main group reducing agents, including boranes, silanes, and hydrogen‐transfer reagents, in a catalyst‐free manner. Selective NO formation was confirmed by EPR studies, including the use of NO‐trapping reagents, and labelled ¹⁵N NMR experiments. The mechanisms for these reductions were investigated computationally. The borane and silane mediated pathways are believed to differ from those involving hydrogen‐transfer reagents such as 1,4‐cyclohexadiene and isopropyl alcohol, with the latter possibly proceeding via HONO as an intermediate.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: The data supporting this article have been included as part of the Supplementary Information (.pdf).

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.