Di-π-methane rearrangement in 1-phosphabarrelenes: formation and reactivity of an unprecedented 2-phosphasemibullvalene

Abstract

The photolysis of 1-phosphabarrelenes, generated from 3,5-diarylphosphinines and benzyne in a [4 + 2] cycloaddition reaction, affords hitherto unknown 2-phosphasemibullvalenes via di-π-methane rearrangement reaction. These compounds occur only as intermediates, while subsequent and rapid dimerization to 6-membered, cyclic diphosphanes with a P–P bond was observed. The results are in stark contrast to the photochemical conversion of 1-phosphabarrelenes, obtained from 2,4,6-triarylphosphinines and a strong dienophile. In this case, the corresponding 5-phosphasemibullvalenes are formed selectively and exclusively. Our results nicely demonstrate the strong impact of the substitution pattern of the starting material on the outcome of the di-π-methane rearrangement reaction.

---

The synthesis of barrelene (C₈H₈) with the systematic name bicyclo[2.2.2]octa-2,5,7-triene was first reported by Zimmerman and Paufler in 1960 (A, Fig. 1).¹ One of the most notable features of barrelene is its photochemical reactivity, particularly the di-π-methane rearrangement. Upon UV irradiation, barrelene undergoes a light-induced 1,2-shift of one π-system, forming semibullvalene B.² This reaction proceeds through a biradical intermediate formed upon excitation, which then undergoes a series of bond breaking and bond formation steps. Semibullvalene itself exhibits a rapid, degenerate Cope rearrangement (Fig. 1).^2,3

Fig. 1 Photolytic di-π-methane rearrangement of barrelene A to semibullvalene B and degenerate Cope-rearrangement to give enantiomer B′′. Achiral (C) and (D) and chiral (E) 1-phosphabarrelenes, photochemical di-π-methane rearrangement of 2,4,6-triphenyl-benzophosphabarrelene F to 5-phosphasemibullvalene G, pyridyl-functionalized phosphasemibullvalene-Rh-complex H and brief summary of this work.

Replacing one CH-group of barrelene by an isolobal, trivalent phosphorus atom leads to phosphabarrelenes. Among the series of known derivatives, the 1-phosphabarrelene (C) has first been described by Märkl and Lieb in 1968.⁴ C was prepared by [4 + 2] cycloaddition reaction of 2,4,6-triphenylphosphinine with the strong dienophile hexafluoro-2-butyne. Later on, a series of benzophosphabarrelenes (D) have been prepared using in situ generated benzyne as highly reactive dienophile.⁵ 1-Phosphabarrelenes have emerged as fascinating stereorigid phosphines (PR₃), which are used as powerful ligands in several catalytic reactions.^6–11 As an example, the enantiomerically pure pyridyl-functionalized 1-phosphabarrelene (E) was recently used as chiral P,N-hybrid ligands in Rh- and Ir-catalyzed hydrogenation reactions, leading to ee′s of up to 95%.¹²

As we could demonstrate for the first time, also 2,4,6-triphenyl-benzophosphabarrelene F undergoes a photochemical di-π-methane rearrangement upon UV irradiation.¹³ The light-induced reaction is highly selective and the 5-phosphasemibullvalene derivative G, along with its enantiomer, is formed quantitatively. Most importantly, it turned out, that the adjacent phenyl-groups in 2- and 6-position of the phosphorus-heterocycle are important for the stabilization of one of the radicals, formed upon excitation. This leads to the exclusive formation of the observed product G, which is formed as a racemate. Later, we could demonstrate that the photochemical di-π-methane rearrangement can also occur in transition metal complexes based on pyridyl-functionalized 1-phosphabarrelenes. In this way, coordination compounds, such as H, could be obtained in up to 90% yield.¹⁴

Inspired by this highly selective di-π-methane rearrangement of 1-phosphabarrelenes to 5-phosphasemibullvalenes and the influence of additional phenyl-groups on the product formation, we anticipated that aryl groups located in the 3,5-position of the phosphorus heterocycle should lead to yet unknown 2-phosphasemibullvalenes. In this isomer, the phosphorus atom is incorporated in the three-membered ring of the semibullvalene core. We envisaged two possible routes towards 3,5-diaryl-benzophosphabarrelenes 4a/b (a: Ar = Ph, b: Ar = para-tolyl), both starting from phosphinines 1a/b. These aromatic phosphorus heterocycles can easily be synthesized from the corresponding diazaphosphinine (Scheme 1).¹⁵

Scheme 1 Desilylation of 1a with HCl/Et₂O to yield phosphinine 2a, which can be converted with 1-bromo-4-fluorobenzene/Mg to 1-phosphabarrelene 4a. Alternative conversion of 1a/b with 1-bromo-4-fluorobenzene/Mg to yield the barrelenes 3a/b and subsequent desilylation using TBAF·nH₂O.

First, protodesilylation of 1a with HCl/Et₂O in dichloromethane affords the literature-known 3,5-diarylphosphinine 2a, which shows a triplet in the ³¹P NMR spectrum at δ(ppm) = 205,7 (²J_H–P = 37.0 Hz).¹⁶ Phosphinine 2a was further characterized crystallographically, and the molecular structure of this compound, along with selected bond lengths and angles, is depicted in Fig. 2. It turned, however, out that the [4 + 2] cycloaddition of 2a with in situ generated benzyne towards the target compounds 4a is not very selective, as several by-products could be observed in the corresponding ³¹P{¹H} NMR spectrum, which could not be separated by conventional methods from the 1-phosphabarrelene. We therefore reacted phosphinines 1a/b first with in situ prepared benzyne. In this case, 1-phosphabarrelenes 3a/b could be obtained in reasonable yields as colorless compounds after recrystallization.¹⁷ Both compounds show a single resonance in the ³¹P{¹H} NMR spectrum at δ(ppm) = −58.5 (3a)¹⁷ and δ(ppm) = −58.2 (3b), respectively. The TMS-substituted 1-phosphabarrelene 3b was additionally characterized crystallographically (see SI). Subsequently, desilylation of 3a/b was achieved with tetrabutylammonium fluoride (TBAF·nH₂O) in THF. After workup and filtration of the reaction mixture over a silica plug with of n-pentane/toluene (see SI) the desired 1-phosphabarrelenes 4a/b could be obtained in yields of 76% (4a) and 46.7% (4b), respectively.

Fig. 2 Molecular structure of 2a in the crystal. Displacement ellipsoids are shown at the 50% probability level. Selected bond lengths (Å) and angles (°): P(1)–C(1): 1.712(6), C(1)–C(2): 1.410(8), C(2)–C(3): 1.406(6). C(1)–P(1)–C(1′): 101.0 (4)°.

The successful desilylation is apparent from the occurrence of a triplet resonance at δ(ppm) = −78.1 (both for 4a and 4b) in the proton-coupled ³¹P NMR spectra. Moreover, we could further characterize the desilylated 3,5-diaryl-benzophosphabarrelene 4b by means of single crystal X-ray diffraction (Fig. 3). Structural information of 4a was obtained by single crystal X-ray diffraction of the corresponding 1-phosphabarrelene-selenide 4a = Se (see SI).

Fig. 3 Molecular structure of 4b in the crystal. Displacement ellipsoids are shown at the 50% probability level. Selected bond lengths (Å) and angles (°): P(1)–C(1): 1.837(3), P(1)–C(5): 1.829(3), P(1)–C(6): 1.849(3), C(1)–C(2): 1.341(4), C(2)–C(3): 1.535(4), C(3)–C(4): 1.535(4), C(4)–C(5): 1.334(4), C(3)–C(7): 1.528(4). Σ∢(C–P–C): 282.81°.

J. Young NMR tube samples of 4a/b in THF were subsequently irradiated with UV-light (λ = 365 nm) for up to 16 h and the reactions were monitored by means of NMR spectroscopy. Fig. 4 shows the ³¹P NMR spectra for the photolysis of 4a in THF over a time interval of 16 h. The 1-phosphabarrelene 4a (δ(ppm) = −78.1) is gradually consumed upon irradiation with UV light. Moreover, a transient intermediate (5a) of low signal intensity is observed during the course of the reaction, which shows a doublet of doublet at δ(ppm) = −111.1.

Fig. 4 ³¹P NMR spectra of the photolysis of 4a in THF. Time interval: 16 h.

This species, however, disappears with prolonged reaction time. Additionally, a new species (6a) is formed, which shows a single resonance at δ(ppm) = −11.6. This compound is at the same time the only species that remains in the reaction mixture after full consumption of 4a. A similar product distribution was also observed for the photochemical conversion of 4b to 6b, with 5b as the transient intermediate (Scheme 2).

Scheme 2 Irradiation of the 3,5-diaryl-benzophosphabarrelenes 4a/b with UV-light yielding 2-phosphasemibullvalenes 5a/b, which dimerize to the cyclic diphosphane 6a/b.

We were able to grow single crystals of the product 6b, suitable for an X-ray crystal structure determination and the molecular structure of this compound is depicted in Fig. 5, along with selected bond lengths.

Fig. 5 Molecular structure of 6b in the crystal, Ar groups are displayed as wireframes. Displacement ellipsoids are shown at the 50% probability level. Selected bond lengths (Å): P(1)–P(2): 2.2301(4), P(1)–C(1): 1.8867 (11), C(1)–C(2): 1.5648(15), C(2)–C(3): 1.5005(15), C(3)–C(4): 1.3409(15), C(4)–C(5): 1.5296(15), C(5)–C(1): 1.5820(15), C(2)–C(22): 1.5460 (15).

Much to our surprise the crystallographic characterization of 6b unequivocally confirms the presence of a cyclic diphosphane, rather than the anticipated 2-phosphasemibullvalene 5b. Upon inspection of the molecular structure, however, it is clear that 6b was formed by cleavage of the P(1)–C(2) (respectively P(2)–C(22)) bond (Fig. 5) and subsequent dimerization, as half of the molecule represents an almost complete 2-phosphasemibullvalene. In order to investigate whether the final product has been formed by dimerization of two 2-phosphasemibullvalenes, we irradiated a 1 : 1 mixture of 4a and 4b with UV light. If the intact 2-phosphasemibullvalenes 5a and 5b are formed by photochemical di-π-methane rearrangement of 4a and 4b, the subsequent dimerization should afford a mixture of the homo-dimers 6a and 6b, as well as the heterodimer 6a,b in a ratio that is reasonably close to the statistically expected ratio of 0.25 : 0.5 : 0.25 (along with the corresponding enantiomers, Scheme 3 and SI).

Scheme 3 Irradiation of a 1 : 1 mixture of 1-phosphabarrelenes 4a and 4b yielding the 2-phosphasemibullvalenes 5a and 5b as intermediates, which dimerize to form the homo-dimerization products 6a and 6b as well as the mixed hetero-dimer 6a,b.

Much to our delight, the analysis of the final product composition by means of ³¹P NMR spectroscopy as well as mass spectrometry indeed verified the presence of the three dimeric species 6a, 6a,b and 6b (see SI). This confirms, that the 2-phosphasemibullvalenes 5a and 5b are formed first by di-π-methane rearrangement of 4a/b, followed by P–C bond cleavage and dimerization by C–C and P–P bond formation reactions. The P–P bond distance was found to be 2.2301(4) Å, which is similar to the values reported by Russell et al. for P₂-cages.^18,19

Our findings suggest, that the transient species, observed during the photolysis of 4a/b (Fig. 4) can be assigned to the 2-phosphasemibullvalenes 5a/b. Since we could not isolate these compounds, we decided to verify the nature of compound 5b by its calculated phosphorus chemical shift. For the quantum chemical calculations of the ³¹P NMR shifts, the structures of 4b, the possible intermediate 5b, and the product 6b were optimized with density functional methods as implemented in TURBOMOLE (see SI).²⁰ Overall, it turned out that the local hybrid CHYF-PBE and the NMR-tailored pcSseg-3 basis set lead to the best agreement with the experiment.^21,22 This results in NMR shifts of −80.9 ppm (4b), −109.2 ppm (5b), and −26.4 ppm (6b). The NMR shifts are sensitive to the molecular structure, e.g., using the structure at the r²SCAN-D4 level leads to NMR shifts of −73.9 ppm (4b), −98.5 ppm (5b), and −17.7 ppm (6b), respectively. For the dimer 6b, the sterically demanding aryl groups lead to larger effects from dispersion.²³ Optimizations of r²SCAN-D4 vs. r²SCAN show that D4 leads to a shorter P–P bond length (223.59 pm vs. 224.12 pm), which is in better agreement with the crystallographic analysis. Currently, D4 parameters are not available for CHYF-PBE. Therefore, the structure optimized with r²SCAN-D4 is possibly more accurate than the one optimized with CHYF-PBE for 6b. This may rationalize the larger deviation towards the experiment. Results are, however, in excellent agreement with the experimental findings of δ(ppm) = −78.1 (4b), δ(ppm) = −111.7 ppm (5b), and δ(ppm) = −11.6 (6b).

We anticipate that the photochemical conversion of 1-phosphabarrelenes 4a/b to the 2-phosphasemibullvalenes 5a/b proceeds in an analogous manner as the di-π-methane rearrangement of benzobarrelene to benzosemibullvalene, investigated by Zimmerman and co-worker.²

The aryl-groups in 3- and 5-position of the 1-phosphabarrelene play an important role in the stabilization of the intermediate radicals, which leads exclusively to a racemic mixture of the novel 2-phosphasemibullvalenes. The proposed mechanism for the formation of one enantiomer of 5a from 4a is shown in Scheme 4.¹³ The other enantiomer is formed, when the radical is stabilized by the phenyl-group on the opposite side of the heterocycle.

Scheme 4 Di-π-methane rearrangement of 4a to 5a, based on the generally accepted mechanism for the photochemical conversion of barrelene to semibullvalene.

In conclusion, this study presents the first example of a photochemical di-π-methane rearrangement in 1-phosphabarrelenes, leading to the transient formation of previously unknown 2-phosphasemibullvalene derivatives. These highly reactive species undergo rapid dimerization to form six-membered, cyclic diphosphanes. Crucially, the substitution pattern on the phosphorus heterocycle significantly influences the rearrangement pathway: while 2,4,6-triaryl-substituted 1-phosphabarrelenes yield exclusively 5-phosphasemibullvalenes, the 3,5-substitution directs the process toward the 2-isomer. Although the 2-phosphasemibullvalene intermediates could not be isolated, their presence was confirmed by NMR spectroscopy and supported by quantum chemical calculations of phosphorus chemical shifts. X-ray crystallographic analysis of the final products confirmed the formation of cyclic diphosphanes via P–C bond cleavage and subsequent dimerization. These findings highlight how subtle changes in substitution can fundamentally alter photochemical reaction outcomes, offering new opportunities for tuning the reactivity of phosphorus-containing systems.

The authors thank the Deutsche Forschungsgemeinschaft (DFG) for their generous financial support. We thank Dr. Massimo Rigo and Dr. Friedrich Wossidlo for initial work on this project. Y. J. F. gratefully acknowledges support via the Walter–Benjamin programme funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 518707327.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the SI. See DOI: https://doi.org/10.1039/d5cc04015a

CCDC 2471357–2471361 (2a, 3b, 4b, 6b, 4a=Se) contain the supplementary crystallographic data for this paper.^24a–e

Notes and references

1. H. E. Zimmermann and R. M. Paufler, J. Am. Chem. Soc., 1960, 82, 1514.
2. H. E. Zimmerman and G. L. Grunewald, J. Am. Chem. Soc., 1966, 88, 183.
3. H. E. Zimmerman, R. W. Binkley, R. S. Givens, G. L. Grunewald and M. A. Sherwin, J. Am. Chem. Soc., 1969, 91, 3316.
4. G. Märkl and F. Lieb, Angew. Chem., Int. Ed. Engl., 1968, 7, 733.
5. G. Märkl, F. Lieb and C. Martin, Tetrahedron Lett., 1971, 17, 1249.
6. M. Fuchs, M. Keller and B. Breit, Chem. – Eur. J., 2006, 12, 6930.
7. M. Blug, X.-F. Le Goff, N. Mézailles and P. Le Floch, Organometallics, 2009, 28, 2360.
8. M. Blug, C. Guibert, X.-F. Le Goff, N. Mézailles and P. Le Floch, Chem. Commun., 2009, 201.
9. P. S. Bäuerlein, I. Arenas Gonzalez, J. J. M. Weemers, M. Lutz, A. L. Spek, D. Vogt and C. Müller, Chem. Commun., 2009, 4944.
10. O. Piechaczyk, M. Doux, L. Ricard and P. Le Floch, Organometallics, 2005, 24, 1204.
11. B. Breit and E. Fuchs, Chem. Commun., 2004, 694.
12. J. Saltó, D. Tarr, A. Petrov, D. Reich, M. Weber, M. Biosca, O. Pàmies, C. Müller and M. Diéguez, Adv. Synth. Catal., 2024, 366, 813.
13. M. Rigo, M. Weber and C. Müller, Chem. Commun., 2016, 52, 7090.
14. M. Bruce, M. Papke, A. W. Ehlers, M. Weber, D. Lentz, N. Mézailles, J. C. Slootweg and C. Müller, Chem. – Eur. J., 2019, 25, 14332.
15. N. Avarvari, P. Le Floch and F. Mathey, J. Am. Chem. Soc., 1996, 118, 11978.
16. M. Blug, O. Piechaczyk, M. Fustier, N. Mézailles and P. Le Floch, J. Org. Chem., 2008, 73, 3258.
17. M. Rigo, E. R. M. Habraken, K. Bhattacharyya, M. Weber, A. W. Ehlers, N. Mézailles, J. C. Slootweg and C. Müller, Chem. – Eur. J., 2019, 25, 8769.
18. O. C. Presly, T. J. Davin, M. Green, R. J. Kilby, S. M. Mansell, J. E. McGrady and C. A. Russell, Eur. J. Inorg. Chem., 2008, 4511.
19. O. C. Presly, M. Green, J. C. Jeffery, E. Leiner, M. Murray, C. A. Russell, M. Scheer and U. Vogel, Chem. Commun., 2006, 4542.
20. Y. J. Franzke, et al., J. Chem. Theory Comput., 2023, 19, 6859.
21. C. Holzer and Y. J. Franzke, J. Chem. Theory Comput., 2025, 21, 202.
22. F. Jensen, J. Chem. Theory Comput., 2015, 11, 132.
23. J. W. Furness, A. D. Kaplan, J. Ning, J. P. Perdew and J. Sun, J. Phys. Chem. Lett., 2020, 11, 8208.
24. (a) L. Steffen, et al., CCDC 2471357: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2nyn6h; (b) L. Steffen, et al., CCDC 2471358: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2nyn7j; (c) L. Steffen, et al., CCDC 2471359: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2nyn8k; (d) L. Steffen, et al., CCDC 2471360: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2nyn9l; (e) L. Steffen, et al., CCDC 2471361: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2nynbm.

---