Arbuzov meets 1,2-oxaphosphetanes: transient 1,2-oxaphosphetan-2-iums as an entry point to beta-halo phosphane oxides and P-containing oligomers

Abstract

Herein, we describe the synthesis of a 1,2σ³λ³-oxaphosphetane from ethylene oxide and its reactions with alkyl halides to form β-halo phosphane oxides in an Arbuzov-type reaction. When methyl triflate was used as a hard electrophile, cationic oligomerisation of 1,2-oxaphosphetanes was observed. DFT calculations indicate 1,2-oxaphosphetan-2-iums as intermediates and reveal differences between the Arbuzov and the potential Perkow reaction pathway.

---

Strained organic and inorganic ring systems¹ are highly interesting due to their special bonding situation and high reactivity. For example, oxetanes (I) (Fig. 1) are important building blocks for synthesising more complicated molecules² and polymers.³ The phosphorus-containing four-membered rings such as phosphetanes (II) drew attention because of their use as steering ligands in transition metal catalysis,⁴ and, more recently, as organocatalysts.^5–10 Oxaphosphetanes (III, IV) represent an unusual combination of oxetanes and phosphetanes but have been scarcely studied, so far. It can be anticipated that 1,3-oxaphosphetanes (III)¹¹ and 1,2-oxaphosphetanes (IV)¹² will behave quite differently whereby III should possess features more similar to I and II. Interestingly, higher coordinate derivatives of III and IV have been investigated more often. For example, 1,3σ⁴λ⁵-oxaphosphetanes are available through intramolecular Mitsunobu reactions,¹¹ and bi- and tricyclic 1,3σ³λ³-oxaphosphetanes have recently been proposed in the decomposition of HPCO.¹³ The 1,2σ⁵λ⁵-oxaphosphetanes (IVb) were discussed as intermediates in the lithium-free Wittig reaction,^12,14,15 and also in the deoxygenation of epoxides.^15–17 Recent DFT calculations showed that they also occur in the phosphite-initiated reductive dimerization of ketones.¹⁸ Until now, very few crystal structures of 1,2σ⁵λ⁵-oxaphosphetanes (IVb) have been reported.^19–22 In contrast, the lower coordinate 1,2σ³λ³-oxaphosphetanes (IVa) were only proposed for a long time, with no known stable derivatives.²³ In 2016, we reported the synthesis of pentacarbonylmetal(0) complexes (M = Cr, Mo, and W) (V) either through ring expansion of epoxides using highly reactive Li/Cl phosphinidenoid complexes or ring formation through intramolecular nucleophilic attack in a α,ω-bifunctional phosphane complex.^24–26 Recently, we also gained access to the free ligand²⁷ (IVa) using thermal decomplexation²⁸ from complexes [M(CO)₅L] based on the chelating properties of bis(diphenylphosphino)ethane (DPPE). The first SC-XRD structure of a non-ligated 1,2σ³λ³-oxaphosphetane (IVa) was reported together with examples of P-oxidation reactions using ortho-chloranil and P-complexation by gold(I) chloride.²⁷ Subsequently, we reported on an elusive 1,2σ⁴λ⁵-oxaphos-phetane P-oxide (VI) and also on a stable P-sulfide and a P-selenide, formerly unknown.²⁹

Fig. 1 Oxetane (I), phosphetanes (σ³λ³IIa, σ⁵λ⁵IIb), 1,3-oxaphosphetanes (σ³λ³IIIa, σ⁵λ⁵IIIb), 1,2-oxaphosphetanes (σ³λ³IVa, σ⁵λ⁵IVb), 1,2σ³λ³-oxaphosphetane metal complexes (V), 1,2σ⁴λ⁵-oxaphosphetanes (VI) and 1,2^σ4λ⁴-oxaphos-phetan-2-ium (VII).

We became attracted to test the usability of 1,2σ³λ³-oxaphosphetanes IVa as a building block for functional organophosphorus compounds via the Arbuzov reaction. The latter is an important and well-studied reaction for acyclic phosphanes,^30,31 bearing a P-alkoxy substituent, with alkyl halides, thus rendering functional phosphane oxides as final products (Scheme 1).^30,32 Cyclic versions of this reaction have been studied so far only for 1,2-oxaphospholanes, but not for smaller P-heterocyclic rings such as in IVa.^33,34

Scheme 1 Examples of Arbuzov reactions for: (a) general, acyclic compounds VIII; (b) 1,2σ³λ³-oxaphosphetanes IVa, presented in this work.

Herein, we report on the synthesis and reactions of 1,2-oxaphosphetanes with various alkyl halides thus rendering a variety of β-halo phosphane oxides XII involving transient 1,2-oxaphosphetan-2-ium species VII as indicated by DFT calculations. Furthermore, preliminary results of a cationic oligo-merization of a 1,2-oxaphosphetane is presented.

To avoid complications from the use of diastereomeric 1,2σ³λ³-oxaphosphetanes, we first synthesized C⁴-unsubstituted and symmetrically C⁴-disubstituted 1,2-oxaphosphetane complexes. Li/Cl phosphinidenoid molybdenum complex 1 reacted with oxirane or 1,1′-dimethyloxirane to yield selectively the corresponding 1,2-oxaphosphetane molybdenum(0) complexes 4 (65% yield) and 5 (77% yield) (Scheme 2). To our surprise, the subsequent treatment of toluene solutions of 4 and 5 with DPPE at 80 °C led to partial decomposition in the case of 5, whereas selective formation of unligated 1,2-oxaphosphetane 6 (70% yield) was observed for 4.

Scheme 2 Synthesis of 1,2-oxaphosphetanes molybdenum complexes 4 and 5 and free 1,2-oxaphosphetanes 6 and 7.

The outcome for 5 could be somewhat improved by reducing the reaction time from 4 h to 2 h thus enabling to detect a resonance signal at 161.6 ppm in the ³¹P-NMR spectrum of the reaction mixture which was assigned to 7 (content estimated 18%, by ³¹P{¹H} NMR integration). Our established protocol enabled us to isolate 1,2-oxaphosphetane 6, but 7 could only be observed as a mixture in solution (Scheme 2).^27,29 Complexes 4 and 5 as well as the “free” 1,2-oxaphosphetane 6 were structurally confirmed via single crystal X-ray diffraction (SC-XRD) studies. Interestingly, the structural features of 4 and 6 (Fig. 2) are very similar, i.e., an almost planar ring system and similar bond lengths (see the ESI‡). But the ³¹P{¹H} NMR chemical shifts vary distinctively for both complexes (219.5 ppm (4) and 183.9 ppm (5)) and the unligated 1,2-oxaphosphetanes (212.1 ppm (6) and 161.6 ppm (7)).

Fig. 2 Molecular structures of 4 (left) and 6 (right) in the solid state. Hydrogen atoms are omitted, and the thermal ellipsoids are set at the 50% probability level. For solid state structures of 5 see the ESI.‡ Selected bond lengths [Å] and angles [°]: 4: Mo–P 2.487, P–O 1.661(3), P–C1 1.845(5), P–C3 1.895(5), and C1–P–O 80.81(19). 6: P–O 1.670(3), P–C1 1.849(4), P–C3 1.923(5), and C1–P–O 80.47(18).

To explore the reactivity of 6, it was treated with various activated alkyl halides 8a–d to achieve alkylation at phosphorus to give 1,2-oxaphosphetan-2-ium salts 9a–d as intermediates. As acyclic β-halo phosphane oxides 10a–d were obtained in an Arbuzov-type reaction (Scheme 3), we concluded that a rapid cleavage of the O–C⁴ bond upon nucleophilic attack of the counter anion had followed the P-alkylation step instead of forming P-functional 1,2σ⁴λ⁵-oxaphosphetanes. The reactions of 6 with 8a–d were selective and products 10a–d were isolated in moderate yields (Table 1). For 10a–c SC-XRD measurements (Fig. 3) confirmed all proposed structures.

Scheme 3 Arbuzov reactions of 6 yielding β-halo phosphane oxides 10a–d.

Table 1 ³¹P{¹H}-NMR chemical shift in CDCl₃ and yield of 10a–d

	10a	10b	10c	10d
δ [ppm]	48.4	49.7	48.5	53.6
Yield [%]	42	53	6	57

---

Fig. 3 Molecular structures of 10a (left) and 10c (right) in the solid state. Hydrogen atoms are omitted, and the thermal ellipsoids are set at the 50% probability level. For solid state structures of 10b, see the ESI.‡ Selected bond lengths [Å] and angles [°]: 10a: P–O 1.4841(10), P–C1 1.8254(14), P–C3 1.8290(14), P–C6 1.8809(14), and C1–P–C3 104.05(7). 10c: P–O 1.488(3), P–C1 1.823(4), P–C3 1.838(4), P–C6 1.886(3), and C1–P–C3 106.16(17).

The effect of the counter anion was then examined. While a clean reaction occurred with methyl iodide (8d), leading to 10d, methyl triflate (8e) did not enable the reaction of transiently formed 9e with the counter anion at low temperature, thus behaving as a weakly coordinating anion (WCA).^35,36 Therefore, no “monomeric” molecular compound was formed, and the cationic intermediate 9e reacted with further units of 6 in propagation steps to give a mixture of oligomeric compounds “11^n” (Scheme 4). Analysis of the solid residue of the reaction mixture by MALDI-MS revealed oligomers with up to ten units of 6 incorporated (m/z 318.1). Repeating the experiment with different equivalents of MeOTf (Fig. 4) showed that, in general, a lesser amount of MeOTf (0.25 eq., 0.05 eq.) led to an increase in heavier oligomers 11,^6–10 but no oligomers with a chain length longer than n = 10 were detected. Dimer 11² formation was favoured for 0.5 eq. of reagent, whereas a pentamer 11⁵ was preferred for all other sub-stoichiometric ratios. The largest relative ratio of heavier oligomers 11^6-10 was observed for 0.05 eq. of MeOTf.

Scheme 4 Arbuzov reaction of 6 with methyl triflate leading to a mixture of oligomers 11ⁿ (n = 2–10).

Fig. 4 Relative distribution of chain length of oligomers of 11ⁿ, depending on equivalents of MeOTf with constant concentration of 6, as detected by MALDI mass spectrometry.

VT-NMR monitoring showed a complex reaction progress. In the ³¹P{¹H}-NMR spectra, a signal at 130.4 ppm was detected between −80 and −20 °C. This signal is tentatively assigned to the 2-methyl-1,2-oxaphosphetan-2-ium salt 9e; this was further supported by the calculated ³¹P chemical shift of 132.5 ppm for the same species (see the ESI‡). Broader signals in the same area were attributed to the propagating 1,2-oxaphosphetan-2-ium end units of growing oligomers 11ⁿ. A broad signal group from 50 to 58 ppm is assigned to the various open chain phosphane oxide units of the oligomers 11ⁿ.

While knowing that 1,2-oxaphosphetanes are moderately strained (ring strain energy of the parent compound, RSE = 19.09 kcal mol^−1 27), DFT calculations on the Arbuzov reaction pathway of 6 were performed and, hence, the reaction of the model (S)-2-methyl-1,2-oxaphosphetane 6* with methyl iodide (8d) was computed. The primary step is the exergonic P-methylation, affording 2,2-dimethyl-1,2-oxaphosphetan-2-ium iodide 9d* (Scheme 5 and Fig. 5), which led to a significant increase of the RSE from 18.66 to 26.24 kcal mol⁻¹ at the (gas phase) working level of theory,³⁷ in agreement with the enlarged Lagrangian of the kinetic energy density per electron computed at the ring critical point³⁸ (G/ρ = 1.1168 and 1.1392 a.u. for 6* and 9*, respectively), thus paralleling the observed tendency of oxaphosphiranes.³⁹ The Arbuzov-type reaction of the iodide counter anion in 9d* is largely facilitated due to the RSE relief, thus proceeding exergonically over a low energy barrier (Fig. 5) and affording β-iodoethyl phosphane oxide 10d*. In the case of using methyl triflate (8e) as the methylating agent, the energy barrier is slightly lower (ΔG_rel = 23.46 kcal mol⁻¹) and the resulting 1,2-oxaphosphetan-2-ium triflate 9e* (ΔG_rel = –37.35 kcal mol⁻¹) is much more stable than the iodide analogue 9d*. This is due to the presence of a rather strong anion⋯P bonding interaction with the triflate O⁻ centre in 9e* (d_P⋯O = 2.385 Å) than with I⁻ in 9d* (d_P⋯I = 3.314 Å). The former highlights the ambiguity of the WCA character of the triflate anion and its inability to promote Arbuzov-type ring opening in 9e*.

Scheme 5 Reaction of 6* with methyl bromoacetate 8a.

Fig. 5 Computed (CPCM_solv/PWPB95-D3/def2-QZVPP//CPCM_solv/PBEh-3c) Gibbs energy profile for the reaction of model 6* with 8a, d, and e. Solvent: THF for reaction with 8a and d but CH₂Cl₂ for 8e.

Similarly, the reaction of 6 with methyl bromoacetate 8a occurred selectively giving a β-bromoethyl phosphane oxide, like compound 10a* in the computed model reaction (Fig. 5 and Scheme 5). Intermediate 9a* also displays a remarkable pnictogen bonding interaction with the Br⁻ formal anion (d_P⋯Br = 2.665 Å), but the value is well below the sum of the van-der-Waals radii (1.80 + 1.85 = 3.65 Å). However, α-halocarbonyl compounds feature a second electrophilic site at the carbonyl C atom which can allow for the formation of the Perkow product, in general,^40,41 that was shown to proceed through initial formation of a σ⁵λ⁵-oxaphosphirane intermediate.⁴² In the case of the reaction of 6* with methyl bromoacetate (8d), the alternative Perkow route was investigated computationally (see the ESI‡) which turned out to be both kinetically and thermodynamically disfavoured compared to the Arbuzov route. The kinetic preference for the Michaelis–Arbuzov pathway can be explained in terms of most favoured frontier molecular orbital (FMO) interactions (i.e. the phosphorus lone pair of 6 and the σ*(C–Br)-orbital of 8a, see the ESI‡). For 8c, the only observed product is again the thermodynamically most stable Arbuzov product 10c.

In total, reactions of a C,C-unsubstituted 1,2σ³λ³-oxaphosphetane with various organic bromo and iodo derivatives were studied, including methyl triflate, to probe the effect of the counter anion. While bromo and iodo derivatives led cleanly to Arbuzov-type products, the β-halo phosphane oxides, methyl triflate led to a mixture of low-weight oligomers. In the latter case, VT-NMR studies provided evidence for the first 1,2-oxaphosphetan-2-ium salt formed at low temperature, which was also supported by DFT calculations.

FG carried out experiments and evaluated analytical data. GS is responsible for the crystallographic data. RS provided concept and the supervised the investigation. FG, AEF and RS wrote the manuscript. The ESI‡ was prepared by FG, AEF and RS.

We gratefully acknowledge financial support from the Deutsche Forschungsgemeinschaft and the University of Murcia for the computational resources of Servicio de Cálculo Científico (SCC).

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. G. He, O. Shynkaruk, M. W. Lui and E. Rivard, Chem. Rev., 2014, 114, 7815.
2. J. A. Bull, R. A. Croft, O. A. Davis, R. Doran and K. F. Morgan, Chem. Rev., 2016, 116, 12150.
3. R. Klein and F. R. Wurm, Macromol. Rapid Commun., 2015, 36, 1147.
4. A. Marinetti and D. Carmichael, Chem. Rev., 2002, 102, 201.
5. W. Zhao, P. K. Yan and A. T. Radosevich, J. Am. Chem. Soc., 2015, 137, 616.
6. K. D. Reichl, N. L. Dunn, N. J. Fastuca and A. T. Radosevich, J. Am. Chem. Soc., 2015, 137, 5292.
7. T. V. Nykaza, T. S. Harrison, A. Ghosh, R. A. Putnik and A. T. Radosevich, J. Am. Chem. Soc., 2017, 139, 6839.
8. L. Longwitz, A. Spannenberg and T. Werner, ACS Catal., 2019, 9, 9237.
9. L. Longwitz and T. Werner, Pure Appl. Chem., 2019, 91, 95.
10. L. Longwitz and T. Werner, Angew. Chem., Int. Ed., 2020, 59, 2760.
11. B. Kaboudin, H. Haghighat and T. Yokomatsu, Synthesis, 2011, 3185.
12. G. Wittig and G. Geissler, Justus Liebigs Ann. Chem., 1953, 580, 44.
13. A. Hinz, R. Labbow, C. Rennick, A. Schulz and J. M. Goicoechea, Angew. Chem., Int. Ed., 2017, 56, 3911.
14. G. Wittig and U. Schöllkopf, Chem. Ber., 1954, 87, 1318.
15. E. Vedejs, G. P. Meier and K. A. J. Snoble, J. Am. Chem. Soc., 1981, 103, 2823.
16. G. Wittig and W. Haag, Chem. Ber., 1955, 88, 1654.
17. A. Espinosa Ferao, ChemPlusChem, 2023, e202300474.
18. A. Espinosa Ferao, Inorg. Chem., 2018, 57, 8058.
19. U. Dieckbreder, E. Lork, G.-V. Röschenthaler and A. A. Kolomeitsev, Heteroat. Chem., 1996, 7, 281.
20. M. Hamaguchi, Y. Iyama, E. Mochizuki and T. Oshima, Tetrahedron Lett., 2005, 46, 8949.
21. T. Kawashima, K. Kato and R. Okazaki, Angew. Chem., Int. Ed. Engl., 1993, 32, 869.
22. Mazhar-ul-Haque, C. N. Caughlan, F. Ramirez, J. F. Pilot and C. P. Smith, J. Am. Chem. Soc., 1971, 93, 5229.
23. E. N. Dianova, E. Y. Zabotina, I. Z. Akhmethkhaova and Y. D. Samuilov, Zh. Obshch. Khim., 1991, 1063.
24. A. W. Kyri, G. Schnakenburg and R. Streubel, Chem. Commun., 2016, 52, 8593.
25. A. W. Kyri, Investigations on 1,2-Oxaphosphetane Complexes. PhD Thesis, University of Bonn, Germany, 2017.
26. A. W. Kyri, V. Nesterov, G. Schnakenburg and R. Streubel, Angew. Chem., Int. Ed., 2014, 53, 10809.
27. A. W. Kyri, F. Gleim, A. García Alcaraz, G. Schnakenburg, A. Espinosa Ferao and R. Streubel, Chem. Commun., 2018, 54, 7123.
28. A. Espinosa Ferao, B. Deschamps and F. Mathey, Bull. Soc. Chim. Fr., 1993, 695.
29. F. Gleim, A. García Alcaraz, G. Schnakenburg, A. Espinosa Ferao and R. Streubel, Molecules, 2022, 27, 3345.
30. B. A. Arbusow, Pure Appl. Chem., 1964, 9, 307.
31. A. Michaelis and R. Kähne, Ber. Dtsch. Chem. Ges., 1898, 31, 1048.
32. A. K. Bhattacharya and G. Thyagarajan, Chem. Rev., 1981, 81, 415.
33. S. Kobayashi, M. Suzuki and T. Saegusa, Macromolecules, 1984, 17, 107.
34. F. Mathey and F. Mercier, J. Chem. Soc., Chem. Commun., 1980, 0, 191.
35. G. A. Lawrance, Chem. Rev., 1986, 86, 17.
36. I. Krossing and I. Raabe, Angew. Chem., Int. Ed., 2004, 43, 2066.
37. RSE 20.19 (6*) and 27.62 kcal mol⁻¹ (9*), at the standard DLPNO-CCSD(T)/def2-QZVPP//B3LYP-D4/def2-TZVP level of theory.
38. A. Espinosa Ferao, Tetrahedron Lett., 2016, 57, 5616.
39. A. Espinosa Ferao, A. Rey Planells and R. Streubel, Eur. J. Inorg. Chem., 2021, 348.
40. W. Perkow, K. Ullerich and F. Meyer, Sci. Nat., 1952, 39, 353.
41. W. Perkow, Chem. Ber., 1954, 87, 755.
42. A. Espinosa Ferao, J. Phys. Chem. A, 2017, 121, 6517.

---

Footnotes

† Dedicated to Prof. Dr Hansjörg Grützmacher on the occasion of his 65th birthday and in memory of Prof. Dr Edgar Niecke, one of the pioneers of modern chemistry of small P-heterocycles.

‡ Electronic supplementary information (ESI) available: Experimental protocols, NMR and theoretical data. CCDC 2290099–2290104. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4cc00254g

---