2-Ethyl-1-phenylindazolium hexafluorophosphate. N-heterocyclic carbene formation, rearrangement, ring-cleavage reactions, and rhodium complex formation

1 Introduction

The number of publications dealing with indazoles has increased considerably during the last decade. Progress in the field of indazole chemistry concerning syntheses [1–4], structures [5], metalations [6, 7], and biological activities [8–11] has been summarized in recent review articles and book chapters. To date, only very few examples of natural products possessing the indazole ring have been discovered [12–14]. Among those, nigellicin (1) has been isolated from Nigella sativa (Scheme 1) [14]. This alkaloid belongs to the class of pseudo-cross-conjugated heterocyclic mesomeric betaines known to provide versatile starting materials for the generation of N-heterocyclic carbenes (NHC). Indeed, indazol-3-ylidenes 4 are formed on thermal decarboxylations of indazolium-3-carboxylates 2, when stabilizing effects such as hydrogen bonds to protic solvents or water of crystallization are absent [15]. The pseudo-cross-conjugated mesomeric betaines, pyrazolium-3-carboxylates [16] and imidazolium-2-carboxylates [17–22], behave similarly, and normal N-heterocyclic carbenes (nNHC) are formed effectively under relatively mild conditions. By contrast, decarboxylations of imidazolium-4-carboxylates, which are members of the class of cross-conjugated heterocyclic mesomeric betaines, to the abnormal N-heterocyclic carbene (aNHC) imidazole-4-ylidene require harsh conditions [22]. The relationship between N-heterocyclic carbenes and mesomeric betaines has been the subject of recent review articles [23, 24]. The deprotonation of indazolium salts 3 is an alternative pathway for the generation of indazol-3-ylidenes 4. In general, indazol-3-ylidenes (and pyrazol-3-ylidenes) have a chemistry of their own that sets them apart from the chemistry of N-heterocyclic carbenes of other ring systems [25–29]. Whereas the majority of N-heterocyclic carbenes are strong σ donors in transition metal-catalyzed cross-coupling reactions, indazol-3-ylidenes can be applied in coordination chemistry [30] and heterocycle synthesis [23–26]. In continuation of our studies of N-heterocyclic carbenes in synthesis [27–29] and catalysis [31], we describe here an indazolethione, an indazolone, and the formation of a rhodium complex as well as an indazole-acridine rearrangement of 2-ethyl-1-phenyl-indazol-3-ylidene. In addition, we report a ring-opening reaction of the indazole ring to a benzamide and a benzothioamide.

Scheme 1

Betaines, salts and N-heterocyclic carbenes of indazole.

2 Results and discussion

2-Ethyl-1-phenylindazolium hexafluorophosphate 5, prepared by ethylation of 1-phenylindazole using diethyl sulfate, was deprotonated with t-BuOK in toluene to give the 9-ethylaminoacridine 6 (Scheme 2). The mechanism of this reaction can be formulated in analogy to related transformations starting from indazol-3-ylidenes [26] as well as pyrazole-3-ylidenes [16] as follows: Deprotonation of the indazolium salt 5 gives the indazol-3-ylidene 5A, which undergoes a ring opening by N–N bond cleavage to ketenimine 5B. Subsequent ring–closure results in the formation of 5C, which tautomerizes to yield the acridine 6. DFT calculations have revealed that the ring–closure from 5B to 5C is predominantely pericyclic in nature [16, 26].

Scheme 2

Ring transformation of indazolium salt 5 to acridine 6 via the N-heterocyclic carbene of an indazole.

In the present studies, we investigated some interception reactions. Thus, reaction of indazolium salt 5 with t-BuOK in toluene at reflux temperature in the presence of sulfur gave indazole-3-thione 7, although in low yield (Scheme 3). Thione formations are supposed to be typical trapping reactions of N-heterocyclic carbenes.

Scheme 3

Trapping of the NHC derived from 5 by sulfur.

The corresponding oxygen analogue of thione 7, 2-ethyl-1-phenylindazolone 8, could be obtained on treatment of 5 with silver(I) oxide in anhydrous dichloromethane in excellent yields (Scheme 4). On examination of the reaction mixture by electrospray ionization mass spectrometry (ESIMS), the silver adduct of indazol-3-ylidene was unambiguously detected at m/z=551.0 (2 carbene + Ag⁺) when a sample was sprayed from MeOH at 0 V fragmentor voltage. A similar reaction, but in the presence of copper(I) chloride, has been observed with a triazolylidene that reacted with CsOH to give the corresponding triazolium-olates [32].

Scheme 4

Indazolone formation starting from indazolium salt 5.

The indazolone 8 dissolved in THF displays a strong, broad but structureless fluorescence emission in the range 378–570 nm with a maximum at 425 nm (excitation wavelength 317 nm). Upon excitation at 248 nm, an additional broad, slightly structured fluorescence peak centered at 330 nm appears. This weaker emission can probably be attributed to the phenyl substituent. Single crystals of indazolone 8 suitable for an X-ray structure determination (Fig. 1) were obtained by slow evaporation of a concentrated solution in diethyl ether. The compound crystallizes in the monoclinic space group P2₁/c (no. 14) with Z=4. The phenyl ring is twisted with respect to the plane of the indazolone by 79.3° (torsion angle C2–N5–C6–C7; crystallographic numbering). The C–O bond length (C3 and O17; crystallographic numbering) was determined to be 123.5(4) pm.

Fig. 1

Molecular structure of indazolone 8 in the solid state. Displacement ellipsoids are drawn at the 50 % probability level, H atoms as spheres with arbitrary radius.

Reaction of the indazolium salt 5 with sodium hydrogen sulfide hydrate gave a mixture of N-ethyl-2-(phenylamino)benzamide (9) as the main product and the corresponding thioamide 10 as a result of two competing nucleophiles (Scheme 5).

Scheme 5

Competing ring-opening reactions on treatment of indazolium salt 5 with NaSH·H₂O.

The formation of the products 9 and 10 can either be explained by the nucleophilic attack of hydroxide or hydrogen sulfide, respectively, to the iminium bond of the indazolium salt 5, or by interception of the ketenimine 5B (Scheme 6). An interception of a ketenimine intermediate, although formed under inert conditions, with water has been reported [33]. Its formation as a byproduct was also observed on treatment of an N-ethyl indazolium-carboxylate with lithium silylcuprate [34]. Benzothioamide 10 is a known compound. It has been prepared before by treatment of a 3,1,2-benzothiazaphosphorine-4-thione with ethylamine [35].

Scheme 6

Proposed mechanisms for the formation of 9 and 10.

Correspondingly, 4-methylthiophenol and phenol converted the indazolium salt 5 into p-tolyl N-ethyl-2-(phenylamino)benzimidothioate (11) and phenyl N-ethyl-2-(phenylamino)benzimidate (12) (Scheme 7).

Scheme 7

Ring cleavage of indazolium salt 5 to form a benzimidate (12) and a benzthioimidate (11).

Finally, we trapped the N-heterocyclic carbene indazol-3-ylidene as a rhodium complex. Deprotonation of 2-ethyl-1-phenyl-indazolium hexafluorophosphate (5) with potassium 2-methyl-butan-2-olate in anhydrous THF at –80°C resulted in the formation of indazol-3-ylidene which was trapped by carbonylbis(triphenylphosphane)rhodium(I) chloride to give the rhodium complex 13 in 54% yield as yellow crystals (Scheme 8).

Scheme 8

Synthesis of the rhodium complex 13.

In the ¹³C NMR spectra of 13, measured in CDCl₃, the signal of the former carbene carbon atom of the indazole ligand appears at δ=186.6 ppm. The signal of the CO ligand was detected at 191.3 ppm. In comparison to salt 5, in the ¹H NMR spectrum, the resonance of 5-H is shifted upfield by Δδ=0.54 ppm on complex formation. However, this value is also influenced by a solvent change from CDCl₃ to [D₆]DMSO, in which the salt is soluble. The CO frequency can be seen in the IR spectrum at 1974 cm^–1. A comparison with the stretching frequencies of the CO ligands in comparable complexes (2009 and 1994 cm^–1) [25] indicates very strong donor strengths of these indazol-3-ylidenes as already postulated earlier on comparing different stretching frequencies of selected 5-membered NHCs [30] or ¹³C NMR resonance frequencies of several palladium carbene complexes [36]. ESIMS were taken in MeOH solution. Samples were also sprayed from MeOH at fragmentor voltages between 0 and 150 V at 300°C drying gas temperature. At 0 V fragmentor voltage, the base peak corresponds to the molecular ion at m/z=877.2. The peak at m/z=849.2 at voltages above 100 V is due to the loss of CO, and the peak at m/z=615.0 is due to the loss of one molecule of PPh₃. Additional peaks were detected at m/z=587.0 (–CO, –PPh₃) and m/z=223.1, which corresponds to the carbene.

Single crystals of 13 were obtained by slow evaporation of a concentrated solution in CH₂Cl₂-i-PrOH, which allowed the determination of the solid-state structure (Fig. 2). The complex crystallizes in the triclinic space group P1̅ (no. 2) with Z=2. The Rh–C_carbene and Rh–C_CO bond lengths (Rh–C9 and Rh–C17; crystallographic numbering) were determined to be 208.3(2) and 186.9(3) pm, respectively. The phenyl substituent is rotated by 71.5(4)° out of the plane of the indazole ring (C4–N7–C10–C15).

Fig. 2

Molecular structure of complex 13 in the solid state. Displacement ellipsoids are drawn at the 50 % probability level, H atoms as spheres with arbitrary radius.

In summary, we generated the N-heterocyclic carbene of 2-ethyl-1-phenylindazole by deprotonation of the corresponding indazolium salt. The carbene can be trapped as thione, oxide, or rhodium complex. It rearranges to 9-aminoethylacridine upon warming. Treatment of the indazolium salt with phenolate and thiophenolate resulted in a ring cleavage of the indazole ring to give a benzimidate and a benzthioimidate, respectively.

3 Experimental section

Nuclear magnetic resonance (NMR) spectra were measured with Bruker Avance 400 MHz and Bruker Avance III 600 MHz instruments. ¹H NMR spectra were recorded at 400 or 600 MHz, ¹³C NMR spectra at 100 or 150 MHz, with the solvent peak or tetramethylsilane used as the internal reference. Multiplicities are described by using the following abbreviations: s=singlet, d=doublet, t=triplet, q=quartet, and m=multiplet. The mass spectra were measured with a Varian 320 MS Triple Quad GC/MS/MS instrument. The ESIMS were measured with an Agilent LCMSD instrument series HP 1100. Samples for ESI mass spectrometry were sprayed from methanol at 30 V fragmentor voltage unless otherwise noted. Melting points are uncorrected and were determined in an apparatus according to Dr. Tottoli (Büchi). Yields are not optimized.