Benzotriazolium Salts: Emergent Readily Accessible Bench-Stable Lewis Acid Catalysts.

In this work, benzotriazolium salts have been introduced as efficient, readily accessible, bench-stable Lewis acid catalysts. Though these sorts of N-heterocyclic compounds have found wide applications as ionic liquids or electrolytes, their Lewis acid catalytic activity remained unexplored. Herein, their potential as Lewis acid catalysts was demonstrated in two prototypical allylic and Nazarov cyclization reactions, showing a matching reactivity and allowing low catalytic loadings (down to 0.5 mol %).

The ability of Lewis acids (LAs) to form acid–base adducts and activate small molecules has been widely used in synthesis and catalysis,[1] hence being established as a powerful tool in a large variety of chemical processes. Although classical main-group Lewis acids containing elements with low-lying free orbitals like boron, aluminum, silicon, or phosphorus have been studied intensively,[1,2] nitrogen-centered Lewis acids have just recently evoked a rising interest. This field was most probably long neglected since the common compounds showing an N-centered electrophilic behavior, such as nitrenes, azides, or diazo derivatives, led to rather unstable Lewis acid–Lewis base adducts or presented strong oxidizing properties like nitronium salts, hampering their application as Lewis acids.[3] Alternatively, the carbene-isoelectronic cationic nitrenium ions have proved more suitable for the design of novel nitrogen Lewis acid structures. Hence, although previously prepared, since the study on the electronic features of stable triazolium ions by the group of Boche in 1996,[4] this type of nitrenium salt became an important building block for the synthesis of organocatalysts, serving as precatalysts,[5] dual catalysts,[6] or ligands.[7] Nevertheless, stable adduct formation between triazinium ions and several Lewis bases was reported by Gandelman et al. in 2017 (Scheme A, left),[8] which resembled the Lewis acidic behavior that was already known for its heavier phosphonium[9,10] and arsenium[11] analogues. This Lewis acid reactivity was then further demonstrated by the group of Stephan using indazolium salts (Scheme A, middle)[12] and extended to frustrated Lewis pair (FLP)[13] driven chemistry (Scheme A, right).[14]

Scheme 1

(A) Nitrenium-Based LA Structures Employed in Acid–Base Adduct Formation and FLP Chemistry, (B) First Nitrenium LA Catalysis, and (C) Benzotriazolium Salts (BZTs) as Novel LA Catalysts

Furthermore, only very recently has the unprecedented use of nitrenium-based Lewis acids in catalysis been described. Thus, in 2020 Goicoechea et al. employed catalytic amounts of a triazinium cation to promote several organic reactions, including hydrogenations, reductions, and a Friedel–Crafts reaction (Scheme B).[15] However, in this pioneering report, only six-membered triaziniums have been proven as efficient N-based Lewis acids, and a few limited types of reactions involving mainly carbonyl and silane activation could be accomplished to date. Therefore, the development of further nitrenium-based catalysts with enhanced and/or adjusted Lewis acid properties to allow for a broader synthetic application is still highly desirable.

Inspired by state-of-the-art and encouraged to solve some of the current reactivity limitations, we envisioned benzotriazolium salts, which are widely used as ionic liquids (ILs), electrolytes, or fluids,[16] as suitable new platforms to provide alternative potent N-based Lewis acid catalyst structures. Despite their ready accessibility and easy modulability that make them excellent candidates for this purpose, their catalytic activity as Lewis acids remained rather unexplored.[17] We herein present the first example of benzotriazolium-based Lewis acids as efficient, bench-stable catalysts for two benchmark reactions implying a carbonyl and hydroxyl group activation, such as allylic and Nazarov cyclizations toward synthetically valuable chromenes and cyclopentenones, respectively (Scheme C).

We started our investigation by preparing a series of 1-methyl-3-methyl (1a–d), 3-(2,2,2-trifluoroethyl) (1e–f), and 3-aryl (1g–i) triazolium salts with different counteranions upon methylation of the corresponding benzotriazole with MeOTf or Me3OBF4, followed, when needed, by a subsequent counteranion exchange (Figure A, see SI for details). Alternatively, the iodide salt was also prepared by reaction with MeI, and the anion was later exchanged to BF4– and a BArF-type anion with AgBF4 and NaBArF4 or Na/KB(C6F5)4, respectively. Similarly, various methoxytriazoliums 2a–d and previously reported triaziniums 3a–c[8,15] were also synthesized. To evaluate the relative Lewis acidity of the triazolium salts, the Gutmann–Beckett test (GBT)[18] was then performed in acetone-d6, and the acceptor number (AN) was calculated (see Table S1 in the SI), revealing an apparent moderate Lewis acidity (≤20 AN).[19] Additionally, the global electrophilicity index (GEI) and fluoride ion affinity were calculated for a series of 1,3-dimethyl benzotriazoles (1a, 1b, and 1d) with their corresponding counteranion at the DFT level of theory, supporting the postulation of a negligible influence of the anion on the Lewis acidity (see SI for details). Hence, the exchange of BF4– with Na or K[B(C6F5)4] and NaBArF4 to bulkier counteranions resulted in a counter-intuitively slight decrease of Lewis acidity. Moreover, the analysis of the X-ray structures of the benzotriazolium series 1a–d[21] revealed the expected increasing benzotriazole–anion distance trend upon an increase of the bulkiness of the anion from OTf– to the less coordinative barfates (Figure B, see SI for details).

Figure 1

(A) Synthesis of benzotriazolium salts 1–2. (B) X-ray structures of representative derivatives 1a–d (ellipsoid contours given at the 50% probability level).

Next, we focused on the evaluation of their reactivity toward cyclization reactions. We chose the allylic cyclization upon the activation of the alcohol 4a to form 2H-chromene (5a)[22] in THF as a benchmark reaction to test the ability of benzotriazolium salts to act as Lewis acid catalysts (Scheme ). This cyclization did not proceed in the absence of a Lewis acid, even at elevated temperatures (70 °C).

Scheme 2

Catalyst Screening for the Allyl Cyclization Reaction of 4a

Conditions: 4a (0.2 mmol) and LA catalyst (10 mol %) in THF (0.1 M) for 18 h at 70 °C under Ar. Yield determined by 1H NMR using CH2Br2 as an internal standard. (See SI for full optimization.)

Same yield for LA catalyst obtained from methylation with Me3OBF4 and exchange from I– with AgBF4, respectively.

Reaction performed under air.

The nitrenium series were then tested using 10 mol % of catalytic loading under an argon atmosphere. Dismissed yields or decompositions were generally observed for most of the Lewis acid catalysts 1–3, as well as for the strong Lewis acid BF3·Et2O. Gratifyingly, the LA catalysts 1b, 2b, and 3c performed well in this benchmark reaction, showing up to 77% yield with catalyst 1b. This result could not be improved by changing the temperature or solvent, which both caused a decrease in yield (see SI). Moreover, the presence of oxygen led to a detriment of the yield, providing the product 5a in only 33% yield when the reaction was run under air with 1b, most probably due to partial oxidation/aromatization of the product.

To confirm that the catalytic activation is induced by our triazolium species, we carried out some control experiments. Since this kind of reaction can also be promoted by strong Brønsted acids that might be formed in the media, we tested our system in the presence of 1 mol % of triflic acid or Brookhart’s-type acid [H(OEt2)2]+[B(C6F5)4]−[23] as catalysts at both 70 °C and rt, but only decomposition could be detected (see SI, Table S6 for more details). This is in line with the observed beneficial moderate Lewis acidity to avoid decomposition while allowing for an efficient promotion of this cyclization reaction. Moreover, we were able to reproduce the good results with 1b obtained through a different synthetic approach by I– anion exchange using AgBF4 instead of direct dimethylation with an excess of Me3OBF4 to rule out any traces of HBF4 or Me3OBF4. Furthermore, we used NaBF4 in catalytic amounts to exclude any influence of the counteranion on the activation process of this reaction. As expected, the ring closure did not proceed, proving that the triazolium cation is indeed inducing the ring closure.

With the optimized conditions in hand, the scope of the allylic cyclization was explored next (Scheme ). First of all, it is worthy to note that the reaction could be scaled up to 1 mmol without any significant detriment on the yield (65% vs 77%). Moreover, the formation of the products bearing different substitution patterns was pleasantly observed in every case. The methyl substitution at both the aryl and allylic moiety of the substrate gave the desired products 5b–d in moderate to good yields (53–76%), while the π-extended compound 5e showed a high stability, leading to a good 88% isolated yield. Moreover, strong electron-donating groups such as methoxy were well tolerated, providing 6- and 5-substituted chromenes 5f and 5g in 83% and 53% yield, respectively. Finally, less reactive substrates bearing electron-withdrawing groups such as halogen or nitro substituents could also be enrolled in the reaction, providing chromenes 5h–j in moderate yields (30–50%),[24] while substrates with internal olefins were unreactive.

Scheme 3

Scope of the Allylic Cyclization to 2H-Chromenes 5

1.0 mmol scale.

Yield determined by 1H NMR using CH2Br2 as an internal standard. Isolated yield given in brackets.

Due to the already known nitrenium activation of carbonyl groups for deoxygenation reactions (see Scheme B), we decided to further investigate the nitrenium series 1–3 in the Nazarov cyclization. This reaction is known to be activated by either LA metal-based catalysts[25] or phosphonium cations[10] and represents one of the most versatile methods for the synthesis of natural products and bioactive molecules containing cyclopentenone moieties.[26] In particular, we chose dienone 6a as the model substrate[27] to determine the activity and requirements of the Lewis acid nitrenium salts (Table ).

Table 1

Catalyst Screening in the Model Nazarov Cyclization of 6a

entry	LA catalyst (mol %)	yield 7a (%)b
1	1b (1)	47
2	1c (1)	30
3	1d (1)	99(86)c
4	2d (1)	15
5	3a (1)	7
6	1d (1)	(77)d,e
7	1d (5)	>99
8	1d (0.5)	69f

Conditions: 6 (0.2 mmol) and LA catalyst in CH2Cl2 (0.1 M) at rt for 24 h under Ar.

Yield determined by 1H NMR using CH2Br2 as an internal standard.

>20:1 dr determined by 1H NMR for all catalytic reactions.

Isolated yield in brackets.

Reaction performed under air.

0.4 mmol scale reaction.

Aiming at identifying highly efficient catalyst structures, 1 mol % of the salts 1–3 was initially employed at room temperature in CH2Cl2 as standard conditions. To our delight, LA catalysts 1b–d, 2d, and 3a were able to promote the cyclization ranking from 7 to 99% conversion (entries 1–5), whereas the other nitrenium salts, including the already established triazinium catalyst 3c, showed no catalytic activity in this reaction (see SI for full screening). However, there was not a perfect correlation between the LA strength and yield of the cyclopentanone 7. Thus, catalyst 1d with a slightly lower LA strength than the other catalysts led to full conversion and an excellent 86% isolated yield (entry 3). Moreover, performing this reaction under air instead of an inert atmosphere resulted in an insignificant drop of yield, which also shows the excellent bench, moisture, and air stability of this catalyst (entry 6). Additionally, while increasing the amount of catalyst to 5 mol % resulted in a similar nearly quantitative yield (entry 7). The remarkable performance of triazolium 1d allowed the activation of the dienone system 6a even at a 0.5 mol % loading to provide 7a in a good 69% yield (entry 8).

These results encouraged us to further investigate the relative kinetics and potency of our catalytic system by monitoring the reaction with selected nitrenium LAs by 1H NMR (Figure ; shown 3 h, see SI for 15 h monitoring). For this purpose, we chose 5 mol % of catalyst loading due to the low conversion for most of the studied nitrenium salts during the catalyst screening with 1 mol %. The highly activating triazolium 1d showed full conversion after only 1 h. Due to this rapid evolution, the reaction with 1 mol % of 1d was also investigated, in which a good 70% yield could still be observed after 3 h. As expected, for the less activating triazinium 3a or catalysts 1b and 2d, a low linear conversion with less than 20% or 10% yield after 3 h was observed. Moreover, the control experiments with Na[B(C6F5)4] and 1a showed no catalytic activity at all.

Figure 2

Kinetic experiment with an exemplary choice of catalysts 1–3 (5 mol %) in the Nazarov cyclization (see SI for full study).

Lastly, the scope and limitations of the Nazarov cyclization were explored (Scheme ). The upscaling of the standard reaction to 1 mmol using 1 mol % of the Lewis acid catalyst 1d was first performed, showing the robustness of the method by providing a similar 75% yield. A different substitution on the ester moiety was well-tolerated (Me, 7b and tBu, 7c), in which tert-butyl led to the best result with a 98% yield on the product 7c. Furthermore, the different substitution on the R2 group was also investigated. While the less reactive dienones bearing neutral Ph or weak activating p-tolyl rests (products: 7d, 81%; 7e, 62%) or electron-deficient aryl groups (product 7f, 46%) required 5 mol % of the catalyst to provide synthetically useful conversions, electron-rich systems such as 3,4,5-triMeOPh provided the corresponding product 7g in almost quantitative yield.

Scheme 4

Scope of the Nazarov Cyclization

Results with 5 mol % of 1d in brackets.

Yields of the isolated products are given.

In summary, we have introduced benzotriazolium salts as a new class of bench-stable and efficient N-centered Lewis-acid catalysts. The new structures were benchmarked with already known nitrenium Lewis acids such as triazinium cations, showing a remarkably higher activity in the studied allylic cyclization and Nazarov reactions, implying a hydroxyl and carbonyl activation, respectively. Moreover, kinetic reaction monitoring of the Nazarov cyclization demonstrated the high reactivity of these N-based Lewis acid structures, which was confirmed by the low 1 mol % catalyst loading required and already shows their potential for a wide-ranging field of applications in Lewis acid catalysis.