A new access to diazaphospholes via cycloaddition–cycloreversion reactions on triazaphospholes

Abstract

A novel bis-CF₃-substituted diazaphosphole was synthesized selectively from hexafluoro-2-butyne and a 3H-1,2,3,4-triazaphosphole derivative. The [4+2] cycloaddition and subsequent cycloreversion reaction under elimination of pivaloyl nitrile affords the product in high yield. The heterocycle coordinates via the phosphorus atom to a W(CO)₅-fragment and shows stronger π-accepting properties than the triazaphosphole.

---

3,5-Disubstituted 3H-1,2,3,4-triazaphospholes (B) are the phosphorus congeners of the well-studied 1,2,3-triazoles (A), according to the isolobal relationship between a trivalent P-atom and a C–H fragment (Chart 1).

Chart 1 Triazole A, triazaphosphole B and possible coordination modes C. Selected examples (D and E) for the reactivity of B.

These λ³, σ² phosphorus heterocycles have a conjugated π-system with a high degree of aromaticity.¹ They can easily be prepared regioselectively by a modular [3 + 2] cycloaddition reaction, starting from various aryl/alkyl-azides and phosphaalkynes.^2,3 Despite the fact that 3H-1,2,3,4-triazaphosphole derivatives have been synthesized independently by Carreé and Regitz already in 1984, the first reports on their coordination chemistry have not appeared in literature before 2010.^2,4 As ambidentate ligands the coordination of the heterocycle to a metal center might proceed either via the phosphorus atom or the nitrogen donors N¹ or N² (Chart 1, C). However, the η¹(P)-coordination mode has so far only been observed in a Pt(0)-complex.^4b

Even less is known about the chemical reactivity of 3H-1,2,3,4-triazaphosphole derivatives. We could demonstrate that the cationic phosphorus analogues D of neutral mesoionic carbenes (1,2,3-triazolylidenes) can be obtained by quaternization of the N1 atom in B with Meerwein salts.⁵ Moreover, we noticed that the introduction of electron-withdrawing N-sulfonyl groups at the N³-atom changes the reactivity of the corresponding triazaphosphole considerably. In the presence of stoichiometric amounts of AuCl·S(CH₃)₂, loss of N₂ and the formation of cyclo-1,3-diphospha(III)-2,4-diazane-Au(I) complexes of type E were observed.⁶ Inspired by the fact that 6-membered azaphosphinines and 5-membered azaphospholes can undergo [4 + 2] cycloaddition reactions with various alkynes under subsequent nitrile elimination, we decided to investigate the reactivity of B towards alkynes in more detail with the aim to synthesize 2H-1,2,3-diazaphosphole derivatives (G) directly in one step (Chart 2).⁷

Chart 2 Attempted synthesis of G, starting from B and an alkyne.

These heterocycles are otherwise only accessible by multistep synthetic procedures.⁸ In fact, similar reactions with RC≡P elimination from oxadiphospholes and selenadiphospholes via a concerted mechanism have been reported.⁹ Moreover, an imino-substituted diazaphosphole biradicaloid showed facile isonitrile cycloaddition, but no subsequent cycloreversion.¹⁰

The 3,5-disubstituted triazaphosphole 1 was prepared according to literature procedures from PhN₃ and ^tBu-C≡P.^2a Triazaphosphole 3 does not react with dimethyl acetylenedicarboxylate (DMAD) to diazaphosphole 2 (Scheme 1a). Using the stronger dienophile hexafluoro-2-butyne, however, elimination of ^tBu–C≡N and, according to ³¹P{¹H} NMR spectroscopy, quantitative formation of diazaphosphole 3 was observed (Scheme 1b). We could not detect the apparent intermediate F (Chart 2) during the course of the reaction. Interestingly, triazoles, such as 4, did not react with CF₃C≡CCF₃ to the CF₃-substituted pyrazole 5, although cycloaddition/cycloreversion reactions on 1,2,3-triazoles with DMAD have been reported in the literature (Scheme 1c).¹¹ This is particularly intriguing as 1-aryl-3,4-bis(trifluoromethyl)-substituted pyrazole motifs (5), are present in numerous pharmacologically relevant and bioactive nitrogen heterocycles and have to be prepared via a multistep synthesis.¹² Our novel diazaphosphole 3 thus represents a phosphorus derivative of this compound class.

Scheme 1 Reactivity of 1 and 4 towards electron-withdrawing alkynes.

Diazaphosphole 3 was obtained as an off-white solid in 87% isolated yield and shows a signal at δ(ppm) = 234.4 (q, ³J_P–F = 25.5 Hz) in the ³¹P{¹H} NMR spectrum (starting material 1: δ(ppm) = 174.3). For the CF₃-groups, resonances at δ(ppm) = −53.3 (dq, ³J_F–P = 25.5, ⁵J_F–F = 7.4 Hz) and δ(ppm) = −61.8 (qd, ⁵J_F–F = 7.4 Hz, ⁴J_F–P = 1.2 Hz) were observed in the ¹⁹F{¹H} NMR spectrum. Single crystals of 3 suitable for X-ray diffraction were obtained by slow evaporation of a dichloromethane solution and the molecular structure of 3 in the crystal is depicted in Fig. 1 along with selected bond lengths and distances.

Fig. 1 Molecular structure of 3 in the crystal. Displacement ellipsoids are shown at the 50% probability level. Selected bond lengths (Å) and angles (°): P(1)–N(2): 1.693(2), P(1)–C(8): 1.712(2), C(8)–C(7): 1.406(3), C(7)–N(1): 1.323(3), N(1)–N(2): 1.349(2), N(2)–C(1): 1.436(3). N(1)–N(2)–C(1)–C(2): 145.9(2).

Fig. 1 represents the first crystallographically characterized CF₃-substituted diazaphosphole. From the X-ray data it is evident that the heterocycle is fully planar and that the P(1)–C(8) and N(1)–N(2) bond distances in 3 are very similar to the ones observed in the starting material 1,¹³ with P–C and C–C bond lengths characteristic for aromatic compounds. The significantly negative NICS(1) values (see Table S1 in the ESI†) are in accordance with aromaticity. Apparently, exchanging pivaloyl nitrile by a perfluorobutyne-moiety does not cause a significant structural change within the heterocycle. The same holds for the inter-ring N(2)–C(1) distance. Also, the N(2)–P(1)–C(8) and P(1)–N(2)–N(1) angles as well as the torsion angle N(1)–N(2)–C(1)–C(2) in 3 are very similar compared to the data found for triazaphosphole 1.

In order to understand the reaction mechanism, ωB97X-D/6-311 + G** DFT calculations (see ESI†) were performed after validating the optimized geometries with the X-ray data of 3 (see Table S2, ESI†). This level of theory was used successfully for cycloaddtion reactions before.¹⁴

The concerted cycloaddition–cycloreversion process (Fig. 2, Chart 2 and ESI†) is in full agreement with all experimental observations. The cycloaddition step B→F (Chart 2) is nearly thermoneutral, while the ^tBuC≡N eliminating cycloreversion (forming G) is highly exergonic. Accordingly (see Hammond principle), the rate determining step of the overall reaction is TS1, that allows the formation of 3 (27.3 kcal mol⁻¹ activation Gibbs free energy) but not of 2 and 5 (barriers 31.9 kcal mol⁻¹, 36.9 kcal mol⁻¹, respectively). It is noteworthy that IRC calculations reveal, that the Ph substituent at the nitrogen atom should be in endo position with respect to the approaching/leaving group for any [4 + 2] cycloaddition step. The interconversion F→F′ is needed prior to the retro cycloaddition step, by flattening the pyramidal nitrogen atom via a small barrier. The fact that no intermediate F/F′ could be detected is in accordance with the very small barrier for the cycloreversion step.

Fig. 2 ωB97X-D/6-311 + G**(PCM = toluene) Gibbs free energy (T = 130 °C) profiles for the reactions in Scheme 1. Relative energies (in kcal mol⁻¹) are compared to the initial van der Waals complex of the reactants.

A comparison of the Kohn–Sham orbitals of the parent CF₃-substituted diazaphosphole, the parent 2H-1,2,3-diazaphosphole and the parent 3H-1,2,3,4-triazaphosphole (Fig. S1, ESI†) shows, that in all three heterocycles, the π-type LUMO has a large coefficient at the phosphorus atom, indicating good π-acceptor properties when coordinated via the phosphorus atom to a metal center. While the orbital energies of the unsubstituted diazaphosphole are generally destabilized with respect to the triazaphosphole (Fig. S1, ESI†), in accordance with the observed ionization energies,¹ CF₃-substitution acts strongly stabilizing (Fig. S1, ESI†). Altogether, 3 should be a stronger π-acceptor than 1. In all three compounds, the lone pair at the phosphorus atom (mixed with the nitrogen in-plane lone-pair) is represented by the HOMO−2 (CF₃-diazaphosphole: E = −11.24 eV; 3H-1,2,3,4-triazaphosphole: E = −10.92 eV; 2H-1,2,3-diazaphosphole: E = −10.21 eV). Consequently, triazaphospholes and diazaphospholes are expected to be rather weak σ-donors, as anticipated for low-coordinate phosphorus compounds. The π-donor properties of triazaphospholes and diazaphospholes are evident from the HOMOs, each having a large π-coefficient at the phosphorus atom, as it is known for other electron-rich phosphorus heterocycles.¹⁵ Again, due to the energetically higher HOMO, triazaphosphole 1 should show stronger π-donor properties than the CF₃-substituted diazaphosphole 3.

The interplay between the above described effects makes the coordination behavior of compound 3 highly interesting, also with respect to triazaphospholes. As a matter of fact, the coordination chemistry of 2H-1,2,3-diazaphospholes is largely unknown and only a few examples can be found in the literature. Chart 3 shows the possible coordination modes for this class of compounds. Analogous to triazaphospholes, diazaphospholes are ambidentate ligands and can coordinate to a metal center either via the phosphorus lone pair (H) or the nitrogen donor (I). This has been demonstrated in a few cases by van Koten, Schmidpeter and co-workers by using suitable Pt(II) and Pd(II) complexes as metal precursor.¹⁶ The simultaneous coordination of a diazaphosphole to two metal fragments (J) has so far not been observed. Only recently, Erben and co-workers have investigated the synthesis and coordination chemistry of Si-bridged, chelating diazaphospholes.¹⁷

Chart 3 Possible coordination modes (H–J) of diazaphospholes.

We decided to focus on the synthesis of a tungsten carbonyl complex of 3, as it can provide valuable information on the electronic ligand properties via IR spectroscopy. Moreover, ³¹P NMR spectroscopy would immediately reveal, whether the coordination of the ligand to the W(CO)₅ fragment occurs via the phosphorus or the nitrogen donor. 3 was reacted with one equivalent of W(CO)₆ in THF at room temperature and under UV irradiation (Scheme 2). After only a short time, the formation of a single new resonance at δ(ppm) = 217.3 was observed in the ³¹P{¹H} NMR spectrum, which corresponds to a coordination shift of Δδ(ppm) = 17.1 compared to the starting material.¹⁸ The selective reaction towards product 6 was complete within 68h. Interestingly, the signal of the product at δ(ppm) = 217.3 shows tungsten satellites with a coupling constant of ¹J_W–P = 326.5 Hz (Fig. S9, ESI†). This indicates that coordination of the ligand to the metal center occurs via the phosphorus atom, in agreement with the calculated 10.6 kcal mol⁻¹ preference of the coordination at phosphorus over nitrogen. For comparison reasons, we also reacted triazaphospholes 1 and 8 (Ar = 2,5-diisopropylphenyl, Dipp) with W(CO)₆ in THF at room temperature and under irradiation with UV light. The course of the reaction was again followed by means of NMR spectroscopy, which revealed a selective and quantitative formation of a new species within 5d. The new compounds (7, 9) show a signal at δ(ppm) = 136.1, respectively δ(ppm) = 160.6 in the ³¹P{¹H} NMR spectrum (Δδ(ppm) = 38.2, 39.7). Much to our surprise, these signals also show tungsten satellites (¹J_P–W = 262.1 Hz; 285.6 Hz), which verifies that also 1 and 8 coordinate via the phosphorus atom to the metal center. This is particularly interesting taking into account that a coordination via N¹ or N² (Chart 1) has so far been observed for the majority of triazaphosphole-based complexes.^4,19 The calculated 0.6 kcal mol⁻¹ energy difference between the two complexation modes of 1 indicates that subtle steric effects determine the complexation site in triazaphospholes.

Scheme 2 Synthesis of W(0)-complexes 6 and 7.

A comparison of the IR spectra of 6, 7 and 9 further shows, that the wavenumbers of the CO stretching frequencies are shifted to higher values in 6 compared to the ones found for 7 and 9 (Table 1). This is in line with the expected lower net-donor properties of 3 compared to 1 and 8.

Table 1 Experimental wavenumbers [in cm⁻¹] for the CO stretching modes. These data were also supported by DFT calculations (see Table S3 in the ESI)

ṽ (CO) [cm^−1]
6	2089	2017	2002	1934
7	2077	2023	1980	1885
9	2081	2000	1954	1934

---

Thus, the CF₃-substituted diazaphosphole 3 is a stronger π-accepting ligand than triazaphospholes 1 and 8, if coordination to the metal center proceeds via the phosphorus donor.

Finally, single crystals of 6 and 9, suitable for X-ray diffraction, could be obtained by slow evaporation of the solvent of a saturated solution of 6 and 9 in n-pentane. Fig. 3 shows the molecular structures of 6 and 9 in the crystal, along with selected bond lengths and angles. The W(0) complexes 6 and 9 show a slightly distorted octahedral coordination geometry and unequivocally confirm that the heterocycles coordinates via the phosphorus atom to the W(CO)₅ fragment. Compared to the solid state structure of the free ligand 3 (Fig. 1), the P(1)–C(8) and P(1)–N(2) bonds in 6 are slightly shortened upon coordination of the ligand to the metal center (1.707(2) Å and 1.677(2) Å in 6vs. 1.712(2) Å and 1.693(2) Å in 3). For steric reasons, the aryl rings in 6 and 9 are rotated out of the heterocyclic plane (see also Fig. 1).

Fig. 3 Molecular structures of 6 (left) and 9 (right) in the crystal. Displacement ellipsoids are shown at the 50% probability level. Selected bond lengths (Å) and angles (°): 6: P(1)–N(2): 1.677(2), P(1)–C(8): 1.707(2), C(8)–C(7): 1.413(3), C(7)–N(1), N(1)–N(2): 1.361(3), P(1)–W(1): 2.3890(6), N(2)–C(1): 1.441(3). N(1)–N(2)–C(1)–C(2): 64.1(3). 7: Only one independent molecule is shown. P(1)–N(2): 1.675(2), P(1)–C(1): 1.712(2), C(1)–N(3): 1.357(3), N(3)–N(2): 1.303(2), N(2)–N(1): 1.358(2), N(1)–C(2): 1.446(2), C(1)–C(14): 1.521(3), P(1)–W(1): 2.4512(5). N(1)–P(1)–C(1): 88.67(9), N(2)–N(1)–C(2)–C(3): 86.3(2).

We could demonstrate for the first time that a 3H-1,2,3,4-triazaphosphole derivative undergoes a selective [4 + 2] cycloaddition with hexafluoro-2-butyne with subsequent elimination of pivaloyl nitrile to afford a bis-CF₃-substituted diazaphosphole in high yield. According to the isolobal relationship between a trivalent phosphorus atom and a C–H fragment, this heterocycle represents a phosphorus congener of a bis-CF₃-substituted pyrazole, which finds applications as a bioactive nitrogen heterocycle. The novel diazaphosphole forms an (L)W(CO)₅-complex, in which the ligand coordinates via the phosphorus atom to the metal center. In combination with DFT-calculations, the experimental results show that the bis-CF₃-substituted diazaphosphole is a stronger π-acceptor than the corresponding triazaphosphole, which was used as a starting material. Our results demonstrate that bis-CF₃-substituted diazaphospholes are accessible in a facile manner. Their use as novel π-accepting ligands in coordination chemistry and homogeneous catalysis as well as the investigation of their potential bioactive properties is currently explored.

The authors are grateful for financial support provided by Freie Universität Berlin. Z. K. is grateful for the general support of Hungarian Academy of Science under the Premium Postdoctoral Research Program 2019.

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. L. Nyulászi, T. Veszprémi, J. Réffy, B. Burkhardt and M. Regitz, J. Am. Chem. Soc., 1992, 114, 9080.
2. (a) Y. Y. C. Yeung Lam, Ko, R. Carrié, A. Muench and G. Becker, J. Chem. Soc., Chem. Commun., 1984, 1634; (b) W. Rösch and M. Regitz, Angew. Chem., Int. Ed. Engl., 1984, 23, 900.
3. L. Nyulászi, Chem. Rev., 2001, 101, 1229.
4. (a) S. L. Choong, A. Nafady, A. Stasch, A. M. Bond and C. Jones, Dalton Trans., 2013, 42, 7775; (b) S. L. Choong, C. Jones and A. Stasch, Dalton Trans., 2010, 39, 5774.
5. M. Papke, L. Dettling, J. A. W. Sklorz, D. Szieberth, L. Nyulászi and C. Müller, Angew. Chem., Int. Ed., 2017, 56, 16484.
6. E. Yue, L. Dettling, J. A. W. Sklorz, S. Kaiser, M. Weber and C. Müller, Chem. Commun., 2022, 58, 310.
7. (a) N. Avavari, P. Le Floch and F. Mathey, J. Am. Chem. Soc., 1996, 118, 11978; (b) A. Schmidpeter and H. Klehr, Z. Naturforsch., B: Anorg. Chem., Org. Chem., 1983, 38, 1484.
8. (a) G. Baccolini, R. Dalpozzo and E. Errani, Tetrahedron, 1987, 43, 2755; (b) J. H. Weinmaier, G. Brunnhuber and A. Schmidpeter, Chem. Ber., 1980, 113, 2278; (c) F. Armbruster, U. Klingebiel and M. Noltemeyer, Z. Naturforsch., 2006, 61b, 225.
9. (a) A. Mack, U. Bergsträßer, G. J. Reiß and M. Regitz, Eur. J. Org. Chem., 1999, 587; (b) S. Asmus, L. Nyulászi and M. Regitz, J. Chem. Soc., Perkin Trans. 2, 2001, 1968.
10. H. Beer, J. Bresien, D. Michalik, A.-K. Rölke, A. Schulz, A. Villinger and R. Wustrack, J. Org. Chem., 2020, 85, 14435.
11. Á. Díaz-Ortiz, A. de Cózar, P. Prieto, A. de la Hoz and A. Moreno, Tetrahedron Lett., 2006, 47, 8761.
12. (a) S. Eguchi. Bioactive Heterocycles I, Springer, Heidelberg, 2006; (b) Y. L. Yagupolskii, N. V. Pavlenko, I. I. Gerus, S. Peng and M. Nappa, ChemistrySelect, 2019, 4, 4604.
13. J. A. W. Sklorz, S. Hoof, N. Rades, N. De Rycke, L. Könczöl, D. Szieberth, M. Weber, J. Wiecko, L. Nyulászi, M. Hissler and C. Müller, Chem. – Eur. J., 2015, 21, 11096.
14. S. Giese, D. Buzsáki, L. Nyulászi and C. Müller, Chem. Commun., 2019, 55, 13812.
15. C. Batich, E. Heilbronner, V. Hornung, A. J. Ashe III, D. T. Clark, U. T. Gobley, D. Kilcast and I. Scanlan, J. Am. Chem. Soc., 1973, 95, 929.
16. (a) J. G. Kraaijkamp, D. M. Grove, G. van Koten and A. Schmidpeter, Inorg. Chem., 1988, 27, 2612; (b) J. G. Kraaijkamp, G. van Koten, K. Vrieze, D. M. Grove, E. A. Klop, A. L. Spek and A. Schmidpeter, J. Organomet. Chem., 1983, 256, 375.
17. P. Kozáček, L. Dostál, A. Růžička, I. Císařová, Z. Černošek and M. Erben, New J. Chem., 2019, 43, 13388.
18. See also: (a) A. B. Grimm, S. Evariste, A. L. Rheingold, C. E. Moore and J. D. Protasiewicz, J. Organomet. Chem., 2017, 851, 9; (b) J. Heinicke, N. Gupta, S. Singh, A. Surana, O. Kühl, R. K. Bansal, K. Karaghiosoff and M. Vogt, Z. Anorg. Allg. Chem., 2002, 628, 2869.
19. J. A. W. Sklorz, S. Hoof, M. G. Sommer, F. Weißer, M. Weber, J. Wiecko, B. Sarkar and C. Müller, Organometallics, 2014, 33, 511.

---

Footnotes

† Electronic supplementary information (ESI) available. CCDC 2163856–2163858. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d2cc02269a

‡ These authors have contributed equally.

---