On Demand Flow Platform for the Generation of Anhydrous Dinitrogen Trioxide and Its Further Use in N‐Nitrosative Reactions

Abstract

Dinitrogen trioxide (N₂O₃) is a powerful and efficient nitrosating agent that comes with an unprecedented atom economy. However, the synthetic application of N₂O₃ is still underdeveloped mostly due to its inherent instability and the lack of reliable protocols for its preparation. This paper presents an open‐source setup and procedure for the on‐demand generation of anhydrous N₂O₃ solution (up to 1 M), which can be further used for reactions under batch and flow conditions. The accuracy and stability of N₂O₃ concentration are guaranteed with the absence of head‐space in the setup and with the synchronization of the gas flows. The reliability of this protocol is demonstrated by >30 worked examples in the nitrosative synthesis of heterocycles—a library of structurally diverse benzotriazoles and sydnones. Kinetic and mechanistic aspects of the N‐nitrosative steps are also explored.

A versatile continuous flow platform for the generation of concentrated dinitrogen trioxide (N₂O₃) solution with accurate and tunable concentration is reported. It unlocks the use of N₂O₃ in the nitrosative synthesis of heterocycles in flow and batch.

Introduction

Nitric oxide (NO) is mostly known either as a common nitroxide pollutant from vehicle exhausts or as a key vertebrate cell messenger. Besides, NO is the primary product of the Ostwald process, ^[1] and therefore a central intermediate for the industrial production of nitric acid, nitro and nitroso compounds (Figure 1a). ^[2] Nitrosation is a versatile means of introducing a NO moiety onto organic scaffolds. ^[3] Although NO is not reactive towards organic molecules in the absence of other reactive species, several downstream products of NO have been employed as nitrosating agents in organic synthesis (Figure 1a). ^[4] Sodium nitrite (NaNO₂) ^[5] is commonly used as the source of nitrous acid (HNO₂) in nitrosations, ^[6] whereas alkyl nitrites (R‐ONO),[ ⁷ , ⁸ , ⁹ ] nitrosyl chloride (Cl‐NO),[ ¹⁰ , ¹¹ ] and nitrosonium salts (NO⁺X⁻)[ ¹² , ¹³ , ¹⁴ ] are employed in anhydrous media. Dinitrogen trioxide (N₂O₃), a powerful nitrosating agent which comes with an appealing atom economy, is frequently involved in NO chemistry under aerobic conditions; ^[15] it is also believed to be the reactive species in the nitrosation with HNO₂.[ ¹⁶ , ¹⁷ , ¹⁸ ] Pure N₂O₃ is a dark blue liquid, which is stable only at low temperature under NO atmosphere.[ ¹⁹ , ²⁰ ] Dinitrogen trioxide decomposes spontaneously to NO, N₂O₄, and NO₂ at room temperature[ ²¹ , ²² ] or upon mixing with a solvent. ^[23] Nitrosation using pure N₂O₃ is highly exothermic and has to be handled with care under cryogenic conditions. ^[24] Due to both its instability and high reactivity, N₂O₃ is therefore rarely applied as a stock chemical in laboratories and in industry.

Figure 1.

The preparation of nitrosating agents and dinitrogen trioxide (N₂O₃).

Neat N₂O₃ can be prepared with >98 % purity by dissolving NO in liquid N₂O₄ at low temperature [Figure 1b, Eqs. (3) and (2)] or by slowly passing an 8 : 1 (v/v) mixture of NO and O₂ through a cold trap [Figure 1b, Eqs. (1) and (2)]. ^[20] The manipulation of low‐boiling toxic compounds requires sophisticated manual operation and is time‐consuming because of the slow mass transfer across the gas/liquid (g/l) interface. In some cases, N₂O₃ was generated in situ from NaNO₂ and acetic acid ^[25] or by bubbling NO and O₂ through the substrate solution.[ ²⁶ , ²⁷ ]

In order to utilize N₂O₃ efficiently for organic synthesis, the preparation of its concentrated solution (0.1–1 M) is of practical significance. Bohle and co‐workers prepared the N₂O₃ solutions in Brønsted‐acid‐free media by manually injecting NO and O₂ from gas‐tight syringes into a closed test tube containing a solvent at 0 °C. ^[28] Such operation is slow and cumbersome. It provides only a few milliliters of N₂O₃ solution, and only finds utility for small‐scale qualitative experiments, which do not require an accurate concentration of N₂O₃.

The preparation of N₂O₃ solution is always entangled with three problems associated with batch setups. Firstly, mass transport is a persistent issue accompanying g/l reactions. Because of the low solubility of NO[ ²⁹ , ³⁰ ] and O₂[ ³¹ , ³² ] in organic solvents, the reaction between NO and O₂ mainly takes place in the gas phase. The mass transport across a limited g/l interface renders the operation laborious and time‐consuming. ^[28] Secondly, the gas phase in the head space could influence the composition in the liquid phase. Similar to the preparation of pure N₂O₃, ^[20] an excess of NO in the gas phase is always required to stabilize the N₂O₃ in the liquid phase, which is wasteful and becomes problematic either upon addition of other reactants or upon workup. Lastly, in the biphasic system, the concentration of N₂O₃ in the liquid phase (c _l) depends on several parameters that are not always independent from one another. Therefore, the individual optimization of the reaction parameters is prohibited by their interrelations.

The g/l mass transport can be accelerated by performing the mixing in a meso‐ or microfluidic channel.[ ³³ , ³⁴ ] The segmented flow pattern in flow channels provides the transport process with immense specific interface,[ ³⁴ , ³⁵ ] which has been successfully exploited for reducing the reaction time of many g/l organic transformations.[ ³⁶ , ³⁷ , ³⁸ ]

The phase rule derived by J. W. Gibbs (Figure 2a) ^[39] can be introduced to elucidate the last two problems.[ ⁴⁰ , ⁴¹ ] This rule relates the number of independent intensive variables (concentrations, temperature, and pressure) (f) in a closed system with the number of independent components (C) and the number of phases (P). There are 5 compounds in a closed system containing an inert solvent and nitroxides with overall composition NO_1.5 (Figure 2b). Because of the restriction of 2 reactions and 1 constraint, the number of independent components is 2.[ ⁴⁰ , ⁴¹ ] In batch reactors, where the gas phase always exists in the head space (P=2), only 2 among the 4 variables in such system (T and p) can be varied independently (Figure 2c). Regarding the mass balance, the concentration in the liquid phase (c _l) is governed by the g/l equilibrium and the volume ratio of the 2 phases. However, the head space volume (V _g), that determines the amount of NO required to stabilize N₂O₃, is usually unknown and difficult to measure. In a flow reactor, where the g/l mixture is continuously moving through a micro‐channel, if the gases are dosed in stoichiometric ratio in the formation of N₂O₃ (Figure 1b), the gas phase will disappear after the dissolution of the gases (V _g=0, P=1) (Figure 2d). In this manner, 1 variable (c _g) vanishes while 1 degree of freedom (f) is gained, the 3 remaining variables (c _l, T, p) become independent, allowing their individual adjustment during the optimization.

Figure 2.

Phase rule and the control of N₂O₃ concentration in the liquid phase.

With the aforementioned 3 problems solved by using flow conditions, we could in theory develop a chemical generator ^[42] to produce N₂O₃ solution from gases and solvent. When the gas phase is eliminated, the N₂O₃ concentration in the liquid phase (c _l) can be precisely controlled by metering devices: V _l by pumps and n by mass flow controllers (Figure 2d). Besides, the continuous flow operation of the generator enables the concatenation with a downstream process that consumes the freshly generated N₂O₃ (Figure 1c). As the gases could be dissolved completely, c _l can be adjusted on the go by tuning the flow rates of both gas feeds without affecting the input flow rate to the downstream reactor.

We herein report our efforts towards a versatile generator of anhydrous N₂O₃ in solution (Figure 1c) along with safe and reliable protocols for its nitrosative reactions. The synthetic utility of N₂O₃ solution is demonstrated through the preparation of the 2 classes of important N‐heterocycles: benzotriazole (2) (Figure 3a) and sydnone (5) (Figure 3b). For both families of N‐heterocycles, a preliminary N‐nitrosation of an aniline precursor is the key step. In the prior Art, this step is accomplished either by NaNO₂ (ref. 2;[ ⁴³ , ⁴⁴ ] 5[ ⁴⁵ , ⁴⁶ ]) or by alkyl nitrite (ref. 2; ^[47] 5 ^[48] ). Unsubstituted benzotriazole and its simple derivatives are well established as useful reagents in chemical synthesis,[ ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ ] especially as peptide coupling reagents.[ ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ ] Sydnones are mesoionic compounds containing an 1,2,3‐oxadiazole ring,[ ⁵⁹ , ⁶⁰ , ⁶¹ ] that have found utility as ligands in organometallic complexes.[ ⁶² , ⁶³ ] In the synthesis of other N‐heterocycles, sydnones are often applied as reaction partners in the [3+2] dipolar cycloaddition.[ ⁴⁸ , ⁶⁴ , ⁶⁵ ] Besides, the benzotriazole and sydnone scaffolds are central to a selection of active pharmaceutical ingredients (APIs) (Figure 3c) and other industrially relevant materials (see below). As both N‐heterocycles are slightly soluble in water, it will be beneficial to synthesize them in anhydrous media using a volatile nitrosating agent to avoid aqueous workup and extraction.

Figure 3.

Benzotriazoles and sydnones: synthesis and related APIs.

Results and Discussion

A continuous flow setup for the first trial (Figure 1c) is constructed with commercially available fluidic elements (see Supporting Information for technical details). The reactor coils, injection loops, and other conduits are made from perfluoroalkoxy alkane (PFA) tubing. The pumps and mass flow controllers (MFC) have been calibrated by volumetric methods. The pressure of the flow system is controlled by a membrane back pressure regulator (BPR) and measured with a piezoresitive pressure sensor. An in‐line IR spectrometer is connected near the outlet of the flow system to monitor the concentration of N₂O₃.

Nitrogen dioxide was discarded as a gaseous feed stock because of its corrosiveness and high boiling point (21 °C). Instead, we relied on the reaction of NO and O₂ for the generation of N₂O₃. Both gases and an aprotic solvent are brought to contact in an X‐mixer (M1) and admixed at 0 °C temperature (reactor I), under which the formation of N₂O₃ is thermodynamically favored (vide etiam Figure 1b). ^[21] When the flow rates of NO and O₂ reaches 4 : 1 ratio, bubble‐free blue solution was observed in reactor I (Figure 4a), yet it evolved to a colorless g/l segmented flow. The alternating mono‐ and biphasic flow can be attributed to the very different flow rates of two insoluble gases. Because the flow rate of O₂ is only 1/4^th of that of NO, O₂ can be blocked by liquid over a longer period of time than NO in response to the change of pressure, releasing a biphasic mixture of NO and solvent (low viscosity) into the reactor coils (Figure 4a). When the O₂ flow resumes, the generated homogeneous N₂O₃ solution (high viscosity) will trigger a new round of pressure fluctuation when it reaches the BPR. Such vicious cycle renders the concentration of N₂O₃ always fluctuating in 5–17 % amplitude (Figure 4b).

Figure 4.

Stabilizing the concentration of N₂O₃ by using the gas flow synchronizer.

After trying out several plausible reactor configurations attempting to generate a homogeneous liquid flow, we moved on with an X‐mixer (M1) fitted with two inlet tubing for gases having well‐defined internal volumes (the gas flow synchronizer) (Figure 4c). These volumes are separated from the upstream by check valves with 2.8 bar back pressure, thus preventing the backflow of the fluid. The internal volumes of the inlet tubing are set proportional to the flow rates of gases (V ₀₁/V ₀₂=4 : 1). This mathematically derived proportionality [Eq. (5)] maintains the equality of the relative rates of volume changes of the gas phase in these tubing [Eq. (4)], enabling the gas feeds to be blocked and resumed simultaneously in response to pressure variations. As a result of the gas flow synchronization, the flow system is always filled with liquid, which could be either solvent (colorless) or N₂O₃ solution (blue). Since the viscosity‐related pressure fluctuation arises from the flow pattern change between two liquids is significantly smaller than that caused by the change between g/l mixture and a liquid, the concentration of N₂O₃ can be stabilized without forming new bubbles (Figure 4d).

Apart from synchronizing the gas flow, additional adjustments were implemented to shorten the startup protocol: since pressurizing the g/l mixture is faster in small reactors, the internal volume of reactor II is made at least 10 times larger than that of reactor I; a six‐port valve (6PV II) was inserted in between the two coil reactors for rapid connection and disconnection. When the flow system is thoroughly purged with solvent, reactor II is disconnected by switching the six‐port valve before starting the gas supply; after the g/l mixture is pressurized in reactor I to the desired pressure and the blue color of N₂O₃ is observed at the outlet, the reactor II filled with solvent is then switched back to the flow system. This manipulation enables a fast stabilization of the flow pattern (<1 h) by limiting the volume of the gas phase below 9 %. Combining the merits of the gas flow synchronizer and the disconnectable large reactor, the modified flow setup (Figure 5) can be started and stabilized within 40 min, producing homogeneous N₂O₃ solution in a stable concentration (fluctuation <0.5 %) (Figure 4d).

Figure 5.

The continuous flow setup for the generation and the reaction of N₂O₃.

Equipped with a flow setup capable of generating bubble‐free N₂O₃ solution with stable concentration (Figure 5), we are able to investigate the property of N₂O₃ solution in organic solvents with process analytical technologies (PAT). ^[66] As we have discussed previously, the absence of gas phase permits the independent control of the temperature, pressure, and concentration (Figure 2d), the concentration of N₂O₃ is adjustable by changing the flow rate ratio of gas and liquid feeds while maintaining NO/O₂=4 (Figure 1b); hence, the composition of the dissolved nitroxide is always NO_1.5 (Figure 4a). The infrared absorption spectrum of the N₂O₃ solution is recorded by an in‐line IR spectrometer at different concentrations up to 1 M (Figure 6a). Four peaks corresponding to the vibrations of nitroso and nitro groups in the N₂O₃ molecule can be observed (blue solid arrows in Figure 6a): 1857 cm⁻¹ (NO stretching), 1592 cm⁻¹ (NO₂ asymmetric stretching), 1298 cm⁻¹ (NO₂ symmetric stretching), and 783 cm⁻¹ (NO₂ deformation).[ ⁶⁷ , ⁶⁸ ] Among them, the peak of the NO stretching (≈1850 cm⁻¹) was used to monitor the concentration of N₂O₃ (Figure 6b, as well as Figure 4b and d), since it does not overlap with the absorptions of common functional groups in organic molecules. At 20 °C, under which N₂O₃ is thermodynamically unstable (Figure 1b), the peaks of N₂O₄ (orange dashed arrows in Figure 6a) ^[69] are only visible after subtracting the background absorption of the solvent. Nitrogen dioxide ^[69] is detectable only under a significant excess of O₂ (NO/O₂ <4, Figure 4a). From the recorded IR spectra (Figure 6a), we can see that dissolved N₂O₃ is quite stable at room temperature in a pressurized monophasic system. The attempt to record the spectra on an FT‐IR spectrometer for solid samples failed due to the rapid loss of nitroxides into the atmosphere (V _g≫V _l, c _g/c _l=constant; Figure 2c).

Figure 6.

Observation of dissolved N₂O₃ with in‐line IR.

The solution of N₂O₃ is characteristic for its intense blue color, which arises from its absorption of visible light at ≈700 nm. ^[70] The blue color provides a strong contrast to the reaction mixture, which is normally light yellow or colorless, indicating the start‐ and endpoint in the collection of the product. The water in the solvent could brighten the color of N₂O₃ solution by hydrolysis. The content of water in the solvent was determined by Karl‐Fischer titration before carrying out a reaction with N₂O₃. Solvents containing less than 200 ppm of water (a newly opened bottle) were used.

The disconnectability of reactor II enables the flow setup (Figure 5) to be employed for the quantified reactions in both flow and batch. When reactor II is connected (the inject mode, Figure 5a), the substrate solution is injected from the injection loop via a six‐port valve (6PV I) and mixed with N₂O₃ solution in the second X‐mixer (M2) for the reaction in reactor II. When no substrate is injected, reactor II will be collecting the freshly generated N₂O₃ solution; the solution therein could be collected for batch chemistry by switching the six‐port valve (6PV II) to the load mode (Figure 5b) and purging the solution into a dry syringe with nitrogen. The collected solution with known volume is then injected immediately into the substrate solution in a batch reactor. Meanwhile, the N₂O₃ solution is always subjected to positive pressure and minimum exposure to the gas phase; in this manner, the accuracy of N₂O₃ concentration is guaranteed. In the applications in both flow and batch, the stoichiometry can be easily adjusted and precisely controlled by changing flow rates and the liquid phase volume; this feature is not achievable by any existing protocols for the preparation of N₂O₃.

With the operating procedure for the reactions in flow and batch established, we are ready to explore the nitrosation chemistry of anhydrous N₂O₃ solutions (Figure 7). The nitrosation of o‐phenylenediamine (1) can be performed in different aprotic solvents (Figure 7a). Substrate 1  b was converted to the corresponding benzotriazole 2  b in 10 min at 20 °C; the best yield (95 %) was given by acetonitrile (MeCN). A 50 : 50 (v/v) mixture of dichloromethane (DCM) and MeCN has been employed for some of the C‐substituted o‐phenylenediamines (1  g–y) and their nitrosation products (2  g–y), which are difficult to dissolve in MeCN alone; N,N‐Dimethylformamide (DMF) appeared as a suitable solvent for most substrates (Figure 8A). Dinitrogen trioxide is so reactive that the reaction completes in 10 min at 0 °C (Figure 7b). We found that 5 bar back pressure is enough to stabilize N₂O₃ solution at 20 °C (Figure 7c). Reducing the pressure to 2.5 bar could result in the formation of gas bubbles and consequently a shorter retention time (t ₂). So far, the optimization was performed with 103 mol % N₂O₃; considering the hydrolysis of 1 mol N₂O₃ gives 2 mol HNO₂, we envisaged the stoichiometry of this N‐nitrosative reaction, seeking the possibility of further enhancing the atom‐economy (Figure 7d). The preliminary results in flow turned out to be quite disappointing: 103 mol % N₂O₃ is necessary for a full conversion in 10 min.

Figure 7.

The optimization of reaction conditions: A) the synthesis of benzotriazole; B) the N‐nitrosation of N‐phenylglycine. The yields (%) of the products are calculated from the concentrations measured by HPLC‐UV (254 nm) with internal standard. For a full list of explored conditions, see Supporting Information.

Figure 8.

The substrate scope of benzotriazoles (A) and sydnones (B). All yields in this scope are based on isolated pure products after flash chromatography unless otherwise mentioned: * The compound was purified by recrystallization; ^# the yield is measured by HPLC‐UV (254 nm) with internal standard (PhCF₃).

In order to explore the reactivity over a longer period, we performed the reaction at 5 °C in a jacketed batch reactor equipped with a digital thermometer. When the solution of 4‐methyl‐o‐phenylenediamine (1  p) was treated with 103 mol % N₂O₃, complete conversion was reached within 5 min (Figure 7e) accompanied by a +24 °C temperature increase (Figure 7f). The reaction with N₂O₃ is very fast and exothermic (cf. ref. [24]), that would be dangerous and difficult to control in large scales in batch.[ ⁷¹ , ⁷² ] When the amount of the N₂O₃ is halved (51.5 mol %), the kinetics exhibits a discontinuous behavior (Figure 7g): ca. 50 % of 1  p was consumed in the first 5 min, and then the reaction proceeded sluggishly over the next 95 min to approach completion. This discontinuity indicates the presence of two nitrosating agents of different reactivity, namely N₂O₃ (0–5 min) and HNO₂ (5–100 min). The reaction with HNO₂ obeys a second order kinetics demonstrated by the linear 1/c−t profile (Figure 7h), which means that N₂O₃ is still the active nitrosating species whose formation is rate‐limiting.

The nitrosation of N‐substituted glycines (3) was performed in DMF because of their low solubility in many aprotic solvents. The reaction of N‐phenylglycine (3  a) with 103 mol % N₂O₃ is fast and clean (Figure 7i). In contrast to 1 (Figure 7d), 51.5 mol % of N₂O₃ is enough to attain a 98 % yield of the N‐nitroso‐N‐phenylglycine (4  a) in 10 min (Figure 7j). This stoichiometry was confirmed in the batch experiment (Figure 7k), while a weaker exotherm (+7.5 °C) was observed (Figure 7l). The formation of nitrosyl carboxylate (−COONO) (3′) and its subsequent intramolecular N‐nitrosation ^[73] are supposed to be responsible for the fast reaction with HNO₂. The nitrosation mixture collected from the outlet of the flow reactor can be directly treated with an acid anhydride without isolating the carcinogenic N‐nitroso compound (4) to afford the sydnone (5) (Figure 8B).Using the conditions optimized in Figure 7, a library of benzotriazoles (2) and 3‐substitued sydnone (5) have been successfully synthesized with good yields. The final reaction mixtures were evaporated under reduced pressure and purified by flash chromatography or, sometimes, recrystallization. In the preparation of the unsubstituted benzotriazole (2  a), the 1 M solution of N₂O₃ has been employed for more concentrated substrate solution (108 g L⁻¹) (Figure 8a).

The substrates in the synthesis of 1‐substituted benzotriazoles (Figure 8b) can be prepared from the N‐functionalization of 2‐nitroaniline followed by hydrogenation. This approach provides an alternative to the direct 1‐substitution of 2  a. Noteworthy, the acid‐sensitive tert‐butoxycarbonyl (Boc) group (2  e) survives the treatment with anhydrous N₂O₃ solution, which is similar to the result provided by 2 equiv tert‐butyl nitrite; ^[47] whereas the reaction with NaNO₂/AcOH gives only deprotected product (2  a). ^[47] Generally, the C‐substituted o‐phenylenediamines can be cleanly transformed to benzotriazoles (Figure 8c–e). Electron‐rich diamine 2  o can be partially oxidized by nitroxides, generating intensely colored sticky side products. Among the synthesized compounds, 2  a,[ ⁷⁴ , ⁷⁵ , ⁷⁶ ] 2  p, ^[77] and 2  r ^[78] are used as corrosion inhibitors for metals; the structural motives of 2  u, 2  v, and 2  y can be found in the API molecules shown in Figure 3c.

For the synthesis of 3‐substituted sydnones (5), the N‐nitrosation step was performed on the flow setup (Figure 5); whereas the cyclization was carried out in batch using conditions adapted from the literature.[ ⁴⁵ , ⁴⁸ ] The N‐aryl glycines (3) except 3  a were prepared according to literature procedures (see Supporting Information). Glycine 3  a was converted to the N‐nitroso compound (4  a) almost quantitatively by 51.5 mol % N₂O₃ (cf. Figure 7j). Although 103 mol % N₂O₃ could do the same job (cf. Figure 7i), the stoichiometric amount of byproduct HNO₂ will invoke the oxidative degradation of the product in the subsequent step, releasing a red‐colored gas (NO₂). After collecting the DMF solution of the N‐nitroso intermediate (4) in a round‐bottomed flask, acid anhydride was added to induce ring closure to sydnone (5). The reaction of 4 with trifluoroacetic anhydride (TFAA) is fast at room temperature, whereas acetic anhydride (Ac₂O) is much less reactive—it gives no conversion in 1 h at 60 °C. Even though, Ac₂O provides good result for 5  a comparable with TFAA. Concerning that DMF (b.p. 153 °C) has to be evaporated after the reaction, Ac₂O is favored in this case because of its mild reactivity as well as its low vapor pressure; while TFAA gives rise to the formation of impurities when the reaction mixture is heated on the water bath.

The mechanistic background of the stoichiometry in the nitrosation of 1 and 3 was investigated in a computational approach. The geometries and the energies of several key transition states (TS) have been calculated using Gaussian 16 software ^[79] (Figure 9C). From the computed molar free enthalpy of activation (ΔG _m ^≠), the kinetic constants (k) are calculated (Figure 9A−B). In the reaction of 1  a with N₂O₃ (Figure 9A left), a fast nitrosation takes place on one of the amino groups via a five‐membered cyclic transition state (TS1);[ ⁸⁰ , ⁸¹ ] the resultant N‐nitroso intermediate (1  a′) undergoes dehydration immediately to afford 2  a. However, the byproduct HNO₂ is unable to turnover a second 1  a molecule, since a proton transfer between a base (1  a) and a Brønsted acid (HNO₂) is more favorable than going through the four‐membered ring transition state (TS2) ^[82] towards 1  a′. Therefore, 100 mol % of N₂O₃ is required to complete the conversion in 10 min. In the treatment with 50 mol % N₂O₃ (Figure 7g), the reactivity after 50 % conversion can be attributed to the slow reversing of HNO₂ to N₂O₃.

Figure 9.

The computational study of the reaction mechanism (M08HX/6‐311++G*//B3LYP‐D3/6‐31+G*, 25 °C in DMF): initial concentration 0.325 M is used in the calculation of the half‐lives. For more details, see Supporting Information.

Regarding the reaction of 3  a with 50 mol % N₂O₃ (Figure 9B), both the amino group and the carboxylic group can be nitrosated by N₂O₃ via cyclic transition states (TS6 and TS8). The primary product of O‐nitrosation (3  a′) is unstable and readily rearranges to N‐nitroso product 4  a via a fast and spontaneous intramolecular process ^[73] (TS10) (Δ_r G _m=−19.86 kcal mol⁻¹), which exhibits no activation barrier. The N‐nitrosation of 3  a by HNO₂ (TS7) is prohibited by a high ΔG _m ^≠; the O‐nitrosation with HNO₂ (TS9) is significantly slower than those with N₂O₃, which seemingly cannot account for the fast and quantitative reaction (Figure 7j). The carboxylic group could accelerate the nitrosation with HNO₂ by enabling two processes: the regeneration of N₂O₃ (as Brønsted acid) and the formation of 3  a′ (as Lewis base). ^[73]

In order to find out the process that contributes to the formation of 4  a, compound 1  v, in which the carboxylic group is fixed far away from the amino group, was treated with 51.5 mol % N₂O₃, giving 2  v in 36 % yield (Figure 8e). Since no intramolecular reaction is allowed, the experimental low yield (<50 %) indicates that the formation of 2  v cannot be promoted by the faster generation of N₂O₃, nor by the accelerated O‐nitrosation of 1  v (TS4) (Figure 9A right). Further computation revealed the existence of a prohibitively high activation barrier (ΔG_m ^≠=+39.11 kcal mol⁻¹) in the intermolecular reaction between 1  v and 1  v′ (TS5), which ultimately blocks the ways to 2  v via O‐nitrosation. Consequently, the synthesis of 2  v still requires 100 mol % N₂O₃. Based on the discussion above, we could conclude that the accelerated O‐nitrosation (TS9) with HNO₂ coupled with the intramolecular nitroso migration (TS10) permits the fast reaction with HNO₂, requiring only 50 mol % N₂O₃ for the full conversion of 3  a in 10 min (Figure 7j–k).

Computations also show that the hydrolysis of N₂O₃ proceeds very slowly in the presence of 6000 ppm (1 equiv) of H₂O (TS11, t _1/2(0.325 M)=2.73×10² s). This result accounts for the insensitiveness of the N₂O₃ reactions towards the water in the solvent as well as the water generated from the nitrosative reactions. As experimentally determined earlier (Figure 7h), the rate of the regeneration of N₂O₃ from HNO₂ is approximately 1/10 of the rate of its hydrolysis. A carboxylic group cannot increase the concentration of N₂O₃ to accelerate the overall reaction significantly; as a result, the fast reaction with HNO₂ is almost irrelevant with the regeneration of N₂O₃.

Conclusion

A continuous flow protocol for the generation and reaction of anhydrous dinitrogen trioxide (N₂O₃) solution is presented. Using an open source flow setup and process analytical technologies (PAT) (in‐line IR and barometry), concentrated N₂O₃ solution (up to 1 M, 75 g L⁻¹) can be prepared in stable concentration for the nitrosative reactions in flow and in batch. The elimination of head space in the flow reactors enables the independent variation of the reaction parameters (concentration, temperature, and pressure) during the optimization of reaction conditions. Safety concerns associated with the toxic gas and exothermic reactions are minimized by using continuous flow reactors. N‐Nitrosative steps in the synthesis of two types of N‐heterocycles (benzotriazole and 3‐substituted sydnone) were accomplished through the efficient and unprecedented use of a stoichiometric amount of N₂O₃, hence demonstrating an elegant and robust technological solution to an unaddressed problem in synthetic organic chemistry. Finally, insights on the reactivity of N₂O₃ and nitrous acid (HNO₂), reaction mechanism and stoichiometry were provided through a computational study.

Conflict of interest

The authors declare no conflict of interest.

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 Professor Alan R. Katritzky

Y. Chen, S. Renson, J.-C. M. Monbaliu, Angew. Chem. Int. Ed. 2022, 61, e202210146; Angew. Chem. 2022, 134, e202210146.

Data Availability Statement

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