Introduction

The interest in organophosphorus compounds in life science, material science, and especially in ligand design for catalysis is an omnipresent topic1,2,3,4,5. Besides the use of phosphines as ligands4,5,6, the resulting complexes are widely used as catalysts in all areas of organic and organometallic chemistry7. One of the most prominent examples of a catalyst containing phosphines represents the Wilkinson’s catalyst [RhCl(PPh3)]8, which catalyses, e.g. the hydrogenation of olefins. Over the last decades, a plethora of organophosphorus compounds was synthesised and investigated9,10,11,12,13. Although organophosphorus chemistry is a well-established area, the synthesis of asymmetrically substituted organophosphorus compounds is a crucial and challenging topic2,6,14,15. Its importance in catalysis was awarded, e.g. with the Nobel prize16. The classical way to synthesise phosphines is via salt metathesis or hydrophosphination starting from PCl3 or PH3, respectively (Fig. 1)12,13,17. In both cases, the targeted variation of the organic substituents is synthetically very challenging. Moreover, both PCl3 and PH3 are synthesised from P4 either by reaction with chlorine gas or by hydrolysis in basic or acidic aqueous media18. Note that, aryl phosphines cannot be synthesized via hydrophosphination19. Interestingly, first attempts to synthesise organophosphorus compounds directly from phosphates have already been made by reduction with HSiCl3 followed by reaction with RCl20, although this development is still in its infancy21. Since both Cl2 and PH3 are toxic gases and because the synthesis of phosphines from PCl3 is associated with large (stoichiometric) amounts of side products, a more sustainable and atom-efficient process for a direct conversion of P4 into organophosphorus reagents is desirable. Besides these classical synthetic strategies, a few protocols have been developed for the direct synthesis of phosphines from P4. The first one of these protocols involves the generation of aryl radicals from aryl halides induced by an unsaturated Ti(III) complex22. The fundamental synthetic approach of all these protocols involves either an electrochemical approach23 or, more often, the generation of organo radicals from the corresponding halides induced by transition metal complexes22,24,25,26 as well as, quite recently, organotin-compounds27. These radical-promoted routes can be divided into stochiometric22,24,27 and/or (photo)catalytic approaches25,26.

Fig. 1: Different approaches for the synthesis of ternary phosphines.
figure 1

Left: conventional way, hydrophosphination and salt metathesis. Right: Radical promoted, stochiometric and catalytic, as well as the ‘semi-catalytic’ transition metal (TM) mediated synthesis of ternary phosphines.

Full size image

The major drawback to all of these processes is that they do not lead to the formation of asymmetrically substituted phosphines, which are crucial for catalysis. In this report, we present a conceptually innovative strategy for a controlled and directed synthesis of symmetrically and asymmetrically substituted phosphines starting from P4 and carbon-centred nucleophiles and electrophiles by using [Cp*Fe(η5-P5)] (1) (Cp* = η5-C5(CH3)5) as a P-atom carrier.

Results

Synthesis and structural characterization of the mono-substituted complexes 2c-e

In our previous works, we were able to show that [Cp*Fe(η5-P5)] (1) readily reacts with main-group nucleophiles such as Me3SiCH2− or Me2N− via the formation of a P–C/P–N bond28, leading to the complexes [Cp*Fe(η4-P5R′)]− (R′ = CH2SiMe3 (2a), NMe2 (2b)). However, we found now that the nucleophile used can be freely varied, and basically any alkali metal organyl can be used (Fig. 2). The reaction of [Cp*Fe(η5-P5)] (1) with MeLi, tBuLi and PhLi, respectively, at −80 °C leads to an immediate colour change from green to brown. After workup, the complexes [Li(dme)3][Cp*Fe(η4-P5Me)] (2c), [Li(12c4)2][Cp*Fe(η4-P5tBu)] (2d), [Li(12c4)(thf)][Cp*Fe(η4-P5Ph)] (2e) can be isolated in crystalline yields of 84%, 86% and 75%, respectively.

Fig. 2: Synthesis of asymmetric phosphines 4, via successive nucleophilic, electrophilic, nucleophilic attack.
figure 2

Synthesis of the anionic precursor complexes 2 by nucleophilic attack; electrophilic quenching of 2 (synthesis of 3); asymmetric phosphine abstraction by nucleophiles (synthesis of 4, 5); regeneration of 1 by thermolysis with P4; alternative synthesis of 3 by electrophilic quenching of 1′.

Full size image

The single-crystal X-ray structure analysis of 2 shows anionic complexes with a folded P5R′-ligand in an envelope conformation (Fig. 3, Supplementary Figs. 40–43). The four coordinating phosphorus atoms of the P5R′ ligand build a nearly square planar P4 unit which coordinates symmetrically to the Cp*Fe fragment. The P1 atom to which the organic group is attached (P1–C1 bond length of 1.849(2) Å (2c), 1.898(3) Å (2d) and 1.841(3) Å (2e)) deviates from this plane. All P2–P5 bond lengths (1.993(13)–2.1575(10) Å)) are in the range between a single and a double bond29,30.

Reactivity of 2 towards electrophiles

The presence of the negative charge renders the compounds 2 nucleophilic, which is why they can be quenched with electrophiles such as alkyl or aryl halides. Thus, the compounds 2 were treated with alkyl halides R″X (R″ = Me, iPr; X = Br, I) as common carbon-centred electrophiles. Indeed, the neutral organo-substituted polyphosphorus complexes [Cp*Fe(η4-P5R′R″)] (3; Fig. 2) were obtained in almost quantitative yields according to 31P NMR spectroscopy. The isolated compounds 2 and 3 were characterized by NMR spectroscopy (Supplementary Figs. 1–9; 17–26), mass spectrometry, elemental analysis, and single-crystal XRD (Fig. 3, Supplementary Figs. 40–50). The latter shows that, in 3, the folded η4-P5 unit remains intact and the introduced electrophile is attached to the phosphorus atom, which bears the first organo-substituent (former nucleophiles; R′, Fig. 3), leading to a 1,1-substitution pattern with a P1–C2 bond length (1.785(5)–1.922(11) Å) corresponding to single bonds29. The functionalized phosphorus atoms P1 carry two organic substituents and possess a phosphonium ion-like character (Fig. 4). Of particular interest are the compounds 3a,b and 3f since they represent precursors for quite rare asymmetrically substituted organo-phosphine derivatives2,31. Due to the large diversity of alkali metal organyls and electrophilic organo halides, the shown procedure gives access to the synthesis of a plethora of complexes of type 3. An alternative way for the synthesis of 3 is the reaction of the doubly reduced derivative of 1, [K(dme)2]2[Cp*Fe(η4-P5)] (1′)32, with alkyl halides. This was demonstrated by the synthesis of 3c and 3e, which proceeds according to 31P NMR in a quantitative manner (Supplementary Figs. 69, S70) and shows that possible limitations of organolithium reagents needed in the former procedure can be avoided. The reaction of 1′ with two equivalents of MeI leads smoothly to 3c, while the addition of one equivalent of tBuI followed by the addition of one equivalent of MeI leads to the formation of 3e (Fig. 2). In the 31P NMR spectra of 2 (Supplementary Figs. 17–20) and 3 (Supplementary Figs. 21–26), AMM′XX′ spin systems were detected from which the coupling constants were obtained by simulations (Supplementary Tables 1–10). The related signals of the phosphorus atoms P1 of 3 are downfield shifted compared to 2. This is in agreement with the phosphonium-like nature of the phosphorus atom P1 in 3.

Fig. 3: Molecular structures of the anions in 2d, 5 and of the neutral complexes 3f, 4a′ with thermal ellipsoids at 50% probability level.
figure 3

Cations and hydrogen atoms are omitted for clarity. The Cp* ligands are drawn in the wire frame model.

Full size image
Fig. 4: Electrostatic potential mapped on electron density (isovalue = 0.001) for 3c.
figure 4

Colour code (blue = positive, red = negative) in kJ·mol−1.

Full size image

Synthesis of (a)symmetric phosphines—reactivity of 3 towards nucleophiles

The electrostatic potential surface of compound 3c (B3LYP/def2-TZVPP level of theory; Fig. 4) shows a rather localized positive potential on the phosphorus atom bearing the organo-substituents (P1). Therefore, the attack of a nucleophile is expected to occur at the position P1. Indeed, the reaction of 3d with benzyl potassium lead to the formation of the asymmetrically substituted phosphine PMeiPrBnz (4a) in 82% yield (Supplementary Fig. 61) and the anion [Cp*Fe(η4-P4)]− (Supplementary Fig. 55). This strategy can be extended for the synthesis of a series of asymmetric phosphines, with PR′R″R‴ (4a-c (a: R′ = Me, R″ = iPr, R‴ = Bnz; b: R′ = tBu, R″ = Me, R‴ = Bnz; c: R′ = Ph, R″ = Me, R‴ = Bnz)) being isolated as air-sensitive viscous liquids in 82% (4a), 55% (4b), 80% (4c) yields, respectively. The identity of the phosphine was proven by NMR spectroscopy and, after oxidation with sulfur to the corresponding phosphine sulfides (Supplementary Figs. 10–16; 27–32; 34–39), also by single-crystal X-ray diffraction analysis (Fig. 3, Supplementary Figs. 51–54). Furthermore, by changing the nucleophile to MeLi and nBuLi, the corresponding phosphines PMe3 and PMe2nBu can be obtained in the reaction with 3c (Supplementary Figs. 56, 57). The latter is less selective, PMe3, however, can be synthesized via this route in 84% yield (according to NMR).

To identify potential intermediates in the reaction of 3 with nucleophiles, the reaction of 3c with MeLi at −80 °C in THF-d8 was monitored by 31P{1H} NMR spectroscopy at this temperature. The 31P NMR spectrum clearly shows the full conversion of 3 and quantitative formation of 4 and 5 (Fig. 5A, Supplementary Fig. 68). These very fast reactions agree with the results of the DFT calculations which show a very low activation barrier for this process (Figs. 5B, S71).

Fig. 5: Identification of the reaction pathway of phosphine abstraction.
figure 5

A Experimental 31P{1H} NMR (242.90 MHz, THF-d8) spectrum of the crude reaction solution (low-temperature reaction) of 3c with KBnz at 180 K. B Gibbs free energy profile of the reaction of 3c with KBnz, calculated at the B3LYP-D3(BJ)/def2-TZVPP (PCM = THF) level of theory.

Full size image

The other product of the reaction of 3 with nucleophiles, i.e. [M][Cp*Fe(η4-P4)] (M = K(dme)3; 5) is a rare representative of the anionic complexes with a cyclo-P4 ligand as an end-deck33,34,35. The addition of 18-crown-6 considerably increases its stability and 5 can be isolated in 81% crystalline yield, which is a major improvement over the previously reported complex [K(18c6)][CpArFe(η4-P4)]− (CpAr = C5(C6H4-4-Et)5) obtained in 4% yield33. In the solid state, the anion [Cp*Fe(η4-P4)]− (Fig. 3, Supplementary Fig. 55) features a square planar cyclo-P4 ligand with bond lengths between 2.1666(13) and 2.175(13) Å, which are in the range between a single29 and a double bond30. In the 31P{1H} NMR spectrum of 5, a singlet is observed at 118.9 ppm (Supplementary Fig. 39). The lithium derivative of 5 can be detected in all reactions of 3b-e with organolithium reagents but decomposes partly over time in solution.

Formation of phosphines and regeneration of the carrier platform [Cp*Fe(η5-P5)] (1) in a ‘semi-catalytic’ cyclic process

The phosphines generated in the reaction of 3 with nucleophiles can be easily isolated either by extraction with n-hexane (5 is not soluble in n-hexane) or, in the case of volatile phosphines, by condensation. That is why we were interested in designing a closed-cycle-process for the generation of phosphines in which 1 represents the platform for the phosphorus atom transfer. To close the circle, 5 was reacted with one equivalent of P4 with 1 being formed in almost quantitative yield, among KP5, which partly decomposes to polyphosphides36 (Supplementary Fig. 65). After workup, 1 can be isolated in 81% crystalline yield (Fig. 6). If required, KP5 can be transformed into the corresponding pentaphosphaferrocene by reacting with [Cp*FeBr]237 to increase the atom efficiency of this process. Moreover, it is possible to perform a larger scale synthesis of the phosphines in a one-pot reaction to easily reuse 1 and to avoid the workup of the reaction solution. In a one-pot reaction, compound 1 was dissolved in 2,5,8,11,14-pentaoxapentadecane (tetraglyme) and stoichiometric amounts of MeLi and MeI were successively added at r.t. with 5 min stirring in between (Fig. 6, steps I and II). Subsequently, the solution was cooled to −30 °C and one equivalent of KBnz in tetraglyme was added (Fig. 6, step III). The cooling bath was removed, the formed phosphine PMe2Bnz was distilled off under reduced pressure (1 × 10−3 mbar, 55 °C) and was obtained in 87% yield (Fig. 6, step III, Supplementary Fig. 67). To the remaining solution, one equivalent of white phosphorus was added and heated under reflux for 1 h at 275 °C (IV). The 31P{1H} NMR spectrum (Supplementary Fig. 63) shows the almost quantitative regeneration of 1. The same protocol was repeated two more times on the same reaction solution (see Supplementary Fig. 64). The phosphine PMe2Bnz was isolated in 82–67% yield (overall isolated yield 79%, see Supplementary Fig. 67, Supplementary Table 40). After these three cycles, 1 can be isolated from the reaction solution in 69% yield (Supplementary Fig. 66), indicating that this process could be carried on for many more cycles.

Fig. 6: ‘Semi-catalytic’ cycle for the synthesis of asymmetric phosphines.
figure 6

One-pot reaction of [Cp*Fe(η5-P5)] (1) with MeLi (I, R′ = Me), quenching with MeI (II, R″ = Me), reaction with KBnz (III, R‴ = Bnz) and subsequent thermolysis with white phosphorus (IV).

Full size image

In summary, the polyphosphorus compound [Cp*Fe(η5-P5)] (1) can be used as a recyclable platform for the targeted synthesis of symmetric or asymmetric phosphines, via a sequence of nucleophilic-electrophilic-nucleophilic reactions, directly from white phosphorus, avoiding the use of intermediate products such as PCl3 or PH3. With this modular system, asymmetric phosphines are obtained in high yields in significant preparative scales and 1 can be regenerated and reused in a ‘semi-catalytic’ cyclic process, which can be run for several cycles in a one-pot reaction. The presented results pave the way for a selective and easy synthetic route to asymmetric and (in future work potential) chiral phosphines based on white phosphorus. This conceptual innovative approach avoids radicals and is not limited to aryl or alkyl substituents, but phosphines with a variety of different substitution patterns are now accessible in high yields by a very simple approach.

Methods

General methods

All manipulations were carried out under an inert atmosphere of dried argon using standard Schlenk and glove box techniques. 1,2-dimethoxyethane (DME), 2,5,8,11-tetraoxadodecane (triglyme) and 2,5,8,11,14-pentaoxapentadecane (tetraglyme) were dried and deoxygenated by distillation under argon atmosphere from sodium (DME) or calcium hydride. All other solvents were dried using a MB SPS-800 device of the company MBRAUN and stored over molecular sieve. NMR spectra were recorded on a Bruker Avance III 400/600 MHz NMR spectrometer. Chemical shifts were measured at ambient temperature and are given in ppm; they are referenced to TMS for 1H and 85% H3PO4 for 31P as external standard. Signal multiplicities are described using common abbreviations: s (singlet), d (doublet), t (triplet), q (quartet), quint (quintet), m (multiplet) and br (broad). LIFDI-/FD-/EI-MS spectra (LIFDI = liquid injection field desorption ionization, FD = field desorption, EI = electron ionization) were measured on a JEOL AccuTOF GCX. ESI-MS spectra (ESI = electrospray ionization) were measured on an Agilent Q-TOF 6540 UHD. Elemental analysis (CHN) was determined using a Vario micro cube instrument. A glass stirring bar was used in reactions with potassium benzyl. DFT calculations were carried out using the Gaussian 16 program38. The geometries were optimised using the B3LYP39,40,41,42 functional together with the def2-TZVPP basis set43. Solvation effects were incorporated via the polarizable continuum model44 using the dielectric constant of THF. Dispersion effects were incorporated via Grimme’s correction with Becke-Johnson damping45.

Unless otherwise stated, all other chemicals were obtained from commercial sources

Crystals suitable for single-crystal X-ray diffraction analysis were obtained as described in the corresponding synthetic protocols below. The diffraction data were collected either on a Gemini Ultra diffractometer equipped with an AtlasS2 CCD detector and with a fine-focus sealed Cu-Kα X-ray tube, on a XtaLAB Synergy R, DW system diffractometer equipped with a HyPix-Arc 150 detector and a rotating-anode Cu-Kα X-ray tube or a GV50 diffractometer equipped with a TitanS2 CCD detector and a micro-focus Cu-Kα X-ray tube. Data collection and reduction were performed with CrysAlisPro software package. The structures were solved with Olex246, using ShelXT47 and a least‐square refinement on F2 was carried out with ShelXL48. All non‐hydrogen atoms were refined anisotropically. Hydrogen atoms at the carbon atoms were located in idealized positions and refined with isotropic displacement parameters according to the riding model.

The images of the molecular structures were made using Olex246.

Synthetic protocols

Synthesis of [Cp*Fe(η5-P5)] (1)49

The complex [Cp*Fe(CO)2]2 (10 g, 20.0 mmol) and white phosphorus (7 g, 56.5 mmol) are dissolved in 1,3-diisopropylbenzene (500 ml) and heated to reflux for 7 h. The solvent was removed in vacuo. The residue was dissolved in dichloromethane, silica was added and the solvent was removed under reduced pressure. The preabsorbed reaction mixture was purified via column chromatography (SiO2, hexane, 10 × 3 cm). Using n-hexane, compound 1 (12.3 g, 35.6 mmol, 89%) can be eluted as a dark green fraction. Complex 1 can be obtained after 1 day from a concentrated solution stored at −30 °C as green needles.

Synthesis of [K(dme)2]2[Cp*Fe(η4-P5)] (1′)32

A solution of 1 (0.4 g, 1.16 mmol, 1 eq) in DME (30 ml) was added to potassium (0.136 g, 3.47 mmol, 3 eq). The solution was stirred at ambient temperature for 4 h. The solution was filtered; the product was extracted from an insoluble residue with DME (70 ml). The volume of the combined solution was reduced to ca. 30 ml and Et2O (30 ml) was added. The suspension was stirred overnight and the product was collected on a frit, washed with Et2O (30 ml) and dried in vacuum to give 1′ (0.74 g, 0.95 mmol) as dark brown-green powder.

Synthesis of [Li(thf)3][Cp*Fe(η4-P5NMe2)] (2a)28

To a solution of 1 (176 mg, 0.51 mmol) in 5 mL THF a solution of LiNMe2 (26 mg, 0.51 mmol) was added at −35 °C. The colour of the solution turned immediately from green to brown. The reaction mixture was stirred over 20 h, the solvent was concentrated to 3 mL and finally layered with n-hexane. After 4 days on room temperature light brown square-shaped crystals of 2a (150 mg, 0.25 mmol, 50%) were formed.

Synthesis of [Li(dme)3][Cp*Fe(η4-P5Me)] (2b)49

To a solution of ((trimethylsilyl)methyl)lithium (68 mg, 0.72 mmol) in Et2O (5 mL) at −35 °C was added a solution of 1 (250 mg, 0.72 mmol) in Et2O (15 mL). Immediate colour change was observed. The solution was slowly warmed up to ambient temperature and volume of the reaction mixture was reduced to ca. 15 mL and layered with n-hexane (20 mL). Black crystals of 2b (342 mg, 0.66 mmol, 92%) were formed after 2 days at 0 °C.

Synthesis of [Li(dme)3][Cp*Fe(η4-P5Me)] (2c)

To a solution of 138.0 mg (0.4 mmol, 1 eq) of 1 in DME at −60 °C, a 1.6 molar solution of MeLi (0.4 mmol, 0.27 mL, 1 eq) in Et2O (diethylether) was added and an immediate colour change to brown could be observed. The solution was stirred for 4 h, the solvent was reduced to 5 mL and layered with n-hexane (25 mL). [Li(dme)3][Cp*Fe(η4-P5Me)] (2c) was isolated as black truncated prisms. Yield: 216.0 mg (0.34 mmol, 83%). 1H NMR (thf-d8, 293 K): δ [ppm] = 1.49 (s, 15H, C5(CH3)5), 0.39 (m, 3H, -(CH3)). 31P{1H} NMR (thf-d8, 293 K): δ [ppm] = 71.3 (m, 1P, PA), 13.7 (m, 2P, PM,M′), −71.5 (m, 2P, Px,x′). 31P NMR (thf-d8, 293 K): δ [ppm] = 71.3 (m, 1P, PA), 13.7 (m, 2P, PM,M′), −71.5 (m, 2P, Px,x′); For coupling constants, see Supplementary Tables 1–10. ESI-MS (anion, DME): m/z = 360.95 (6%, [M]−), 376.94 (8%, [MO]−), 392.93 (100%, [MO2]−); analysis (calcd., found for C23H48FeLiO6P5): C (43.28, 42.84), H (7.58, 7.42).

Synthesis of [Li(12c4)2][Cp*Fe(η4-P5 tBu)] (2d)

Compound 1 (0.2 mmol, 69.2 mg, 1 eq) was dissolved in 10 mL THF (tetrahydrofuran) and cooled to –80 °C. A 1.7 molar solution of tBuLi (0.2 mmol, 0.12 mL, 1 eq) in n-pentane was added dropwise. An immediate change in colour from green to brown occurred. The mixture was allowed to stir overnight and reach room temperature. 12-crown-4 (12c4) (0.4 mmol, 0.07 mL, 2 eq) was added to the brown solution and stirred for 30 min at room temperature. The solvent was removed under reduced pressure. The brown residue was washed with 2 × 10 mL n-hexane, dissolved in 2 mL THF and layered with 5 mL toluene. The mixture was stored at 8 °C. Compound 2c was isolated after 1 day as dark green plates. Yield: 110.2 mg (0.17 mmol, 86%). 1H NMR (THF-d8, 293 K): δ [ppm] = 3.78 (s, 33H, 12c4), 1.45 (s, 15H, C5(CH3)5), 0.03 (d, 9H, 3JP-H = 10.8 Hz, -C(CH3)3). 31P{1H} NMR (THF-d8, 293 K): δ [ppm] = 105.6 (m, 1P, PA), 19.6 (m, 2P, PM,M′), − 70.2 (m, 2P, Px,x′). 31P NMR (THF-d8, 293 K): δ [ppm] = 105.6 (m, 1P, PA), 19.6 (m, 2P, PM,M′), −70.2 (m, 2P, Px,x′); (for coupling constants, see Supplementary Tables 2, 3). ESI-MS (anion, DME): m/z = 403.00 (100%, [M]−); analysis (calcd., found for C30H56FeLiO8P5): C (47.26, 47.61), H (7.40, 7.49).

Synthesis of [Li(12c4)(thf)][Cp*Fe(η4-P5Ph)] (2e)

Compound 1 (0.2 mmol, 69.2 mg, 1 eq) was dissolved in THF, cooled to −80 °C and a 1.9 molar solution of PhLi in dibutyl ether (0.2 mmol, 0.11 mL, 1 eq) was added. A colour change from green to brown occurred. Two equivalents of 12-crown-4 (0.4 mmol, 0.12 mL) were added and the solution was stirred overnight and allowed to reach room temperature. The solvent was removed under reduced pressure. The resulting brownish-green residue was washed with 10 mL of n-hexane and again dissolved with 3 mL of THF layered with 10 mL of n-hexane and stored at room temperature. Dark green blocks of 2e were isolated after 1 week. Yield: 102.0 mg (0.15 mmol, 75%). 1H NMR (THF-d8, 293 K): δ [ppm] = 6.56 (m, 5H, -C6H5), 3.72 (s, 17H, 12c4), 3.60 (m, 3H, THF-H8), 1.77 (m, 3H, THF-H8), 1.51 (s, 15H, C5(CH3)5). 31P{1H} NMR (C6D6, 293 K): δ [ppm] = 79.3 (m, 1P, PA), 21.1 (m, 2P, PM,M′), −61.84 (m, 2P, Px,x′). 31P NMR (C6D6, 293 K): δ [ppm] = 79.3 (m, 1P, PA), 21.1 (m, 2P, PM,M′), −61.84 (m, 2P, Px,x′); (For coupling constants, see Supplementary Table 4). ESI-MS (anion, DME): no signal can be assigned; analysis (calcd., found for C28H44FeP5LiO5): C (49.58, 49.80), H (6.54, 6.53).

Synthesis of [Cp*Fe{η4-P5(NMe2)Me}] (3a)

Compound 2a (0.3 mmol, 168.0 mg, 1 eq) was dissolved in 50 mL THF. A 1.11 molar solution of MeI in diethylether (0.3 mmol, 0.27 mL, 1 eq) was added. The mixture was allowed to stir overnight. The solvent was removed under reduced pressure and a brown solution was extracted with 20 mL n-hexane and filtered over diatomaceous earth. The solvent was reduced to 5 mL and stored at −30 °C. 3a was formed as dark prisms after 1 day. Yield: 69.1 mg (0.17 mmol, 57%). 1H NMR (C6D6, 293 K): δ [ppm] = 1.86 (dt, 3H, 2JP-H = 11.7 Hz, 3JP-H = 5.8 Hz, -(CH3), 1.53 (d, 3JP-H = 10.4 Hz, 6H, N(CH3)2). 31P{1H} NMR (C6D6, 293 K): δ [ppm] = 130.8 (m, 1P, PA), −65.1 (m, 2P, Px,x′), 37.7 (m, 2P, PM,M′). 31P NMR (C6D6, 293 K): δ [ppm] = 130.8 (m, 1P, PA), −65.1 (m, 2P, Px,x′), 37.7 (m, 2P, PM,M′) (for coupling constants, see Supplementary Table 4). LIFDI-MS (toluene): 405.02 (100%, [M]+); analysis (calcd., found for C13H24FeP5N): C (38.55, 38.14), H (5.97, 6.03).

Synthesis of [Cp*Fe{η4-P5(CH2SiMe3)Me}] (3b)

Compound 2b (1.33 mmol, 686.0 mg, 1 eq) was dissolved in 100 mL THF. 1.2 mL of a 1.110 molar solution of MeI in Et2O (1.33 mmol, 1 eq) was added under strong stirring at room temperature and the solution was stirred overnight. The formation of a white precipitate was observed on the glass wall. The solvent was removed, 3b extracted with 40 mL of n-hexane and filtered over diatomaceous earth. The solvent was reduced to 10 mL. Complex 3b was isolated as black plates. Crystals for X-ray structure analysis were obtained from a concentrated n-hexane solution at −30 °C. Yield: 426 mg (0.94 mmol, 71%). 1H NMR (C6D6, 293 K): δ [ppm] = 1.80 (dt, 3H, 2JP-H = 11.9 Hz, 3JP-H = 5.8 Hz, -CH3), 1.67 (s, 15H, C5(CH3)5), 0.25 (d, 2H, 2JP-H = 14.9 Hz, CH2Si(CH3)3), −0.18 (s, 9H, CH2Si(CH3)3). 31P{1H} NMR (C6D6, 293 K): δ [ppm] = 129.4 (m, 1P, PA), 37.0 (m, 2P, PM,M′), −96.7 (m, 2P, Px,x′). 31P NMR (C6D6, 293 K): δ [ppm] = 129.4 (m, 1P, PA), 37.0 (m, 2P, PM,M′), −96.7 (m, 2P, Px,x′); (for coupling constants, see Supplementary Table 5). LIFDI-MS (toluene): 448.03 (100%, [M]+); analysis (calcd., found for C15H29FeP5Si): C (40.20, 39.64), H (6.52, 6.48).

Synthesis of [Cp*Fe(η4-P5Me2)] (3c)

Method A: Compound 2c (3.53 mmol, 2.26 g, 1 eq) was dissolved in DME and 3.18 mL of a 1.11 molar solution of MeI (3.53 mmol, 1 eq) in Et2O was added at room temperature and stirred overnight. The solvent was removed in vacuo, a brown residue was extracted with n-hexane and filtered over silica gel. The solvent was removed under reduced pressure, dissolved in 5 mL of CH2Cl2, layered with 20 mL of acetonitrile and stored at −30 °C. Dark green blocks of 3c were isolated after 1 week. Yield: 938 mg (2.49 mmol, 71%).

Method B: [K(dme)2]2[Cp*Fe(η4-P5)] (1′) (0.2 mmol, 165.9 mg, 1 eq) was dissolved in 10 mL THF and 0.36 mL of a 1.11 molar solution of MeI (0.4 mmol, 2 eq) in Et2O was added at room temperature. The reaction mixture was stirred for five min and the solvent was partly reduced in vacuo. The performed 31P{1H} NMR experiment (Supplementary Fig. 69) of the reaction in THF/C6D6-capillary matches the spectroscopic data of method A and shows a quantitative formation of 3c. 1H NMR (C6D6, 293 K): δ [ppm] = 1.61 (s, 15H, C5(CH3)5), 1.51 (dt, 3H, 2JP-H = 11.1 Hz, 3JP-H = 5.5 Hz, -CH3(exo)), 0.06 (d, 3H, 2JP-H = 11.5 Hz, -CH3(endo)). 31P{1H} NMR (C6D6, 293 K): δ [ppm] = 117.2 (m, 1P, PA), 35.4 (m, 2P, PM,M′), −122.7 (m, 2P, Px,x′). 31P NMR (C6D6, 293 K): δ [ppm] = 117.2 (m, 1P, PA), 35.4 (m, 2P, PM,M′), −122.7 (m, 2P, Px,x′); (for coupling constants, see Supplementary Table 6). LIFDI-MS (toluene): 376.97 (100%, [M]+), 360.95 (40%, [M]+-CH3), 345.92 (20%, [M]+-(CH3)2), 314.95 (10% [M]+-P(CH3)2); analysis (calcd., found for C12H21FeP5): C (38.33, 38.37), H (5.63, 5.63).

Synthesis of [Cp*Fe(η4-P5MeiPr)] (3d)

To a solution of 2c (0.15 mmol, 90 mg, 1 eq) in 10 mL DME. 0.03 mL of isopropyl iodide (0.15 mmol, 1 eq) was added. The brownish solution was stirred overnight, all volatiles were removed in vacuum and the brown residue was dissolved in n-hexane and filtered through a frit. The resulting solution was concentrated to 3 mL and stored at −30 °C. Dark red blocks of [Cp*Fe(η4-P5iPrMe)] (3d) could be isolated after 2 days. Yield: 43 mg, (0.106 mmol, 72% yield). 1H NMR (C6D6, 293 K): δ [ppm] = 2.40 (m, 1H, -CH(CH3)2), 1.59 (s, 15H, C5(CH3)5), 1.01 (dd, 6H, -CH(CH3)2, 3JH-H = 7.1 Hz, 3JP-H = 15.0 Hz), 0.05 (d, 3H, -CH3, JP-H = 10.5 Hz). 31P{1H} NMR (C6D6, 293 K): δ [ppm] = 151.8 (m, 1P, PA), 33.3 (m, 2P, PM,M′), −135.8 (m, 2P, Px,x′). 31P NMR (C6D6, 293 K): δ [ppm] = 151.8 (m, 1P, PA), 33.3 (m, 2P, PM,M′), −135.8 (m, 2P, Px,x′); (for coupling constants see Supplementary Table 7). FD-MS (toluene): 404.00 (100%, [M]+); analysis (calcd., found for C14H25FeP5): C (41.61, 41.60), H (6.24, 6.11).

Synthesis of [Cp*Fe(η4-P5 tBuMe)] (3e)

Method A: A solution of 2d (0.2 mmol, 152.5 mg, 1 eq) in 10 mL THF was cooled to −80 °C and a 1.005 molar solution of MeI in DME (0.2 mmol, 0.20 mL, 1 eq) was added. A colour change from green to brown occurred and a colourless solid was formed. The mixture was allowed to stir overnight and reach room temperature. The solvent was removed under reduced pressure and a brown solution was extracted with 10 mL toluene and filtered over diatomaceous earth. The solvent was removed in vacuo. The brown residue was dissolved in 2 mL THF, layered with 4 mL n-hexane and stored at –30 °C. [Cp*Fe(η4-P5tBuMe)] 3e was formed as dark green blocks after 1 day. Yield: 71.1 mg (0.17 mmol, 85%).

Method B: [K(dme)2]2[Cp*Fe(η4-P5)] (1′) (0.4 mmol, 313.8 mg, 1 eq) was dissolved in 10 mL THF, cooled to −80 °C and 0.05 mL tBuI (0.4 mmol, 1 eq) was added. The reaction mixture was stirred for 10 min and 0.18 mL of a 1.11 molar solution of MeI (0.2 mmol, 1 eq) in Et2O was added. The resulting brown solution was stirred overnight and the solvent was partly reduced in vacuo. The performed 31P{1H} NMR experiment (Supplementary Fig. 70) of the reaction in THF/C6D6-capillary matches the spectroscopic data of method A and shows a quantitative formation of 3e. 1H NMR (C6D6, 293 K): δ [ppm] = 1.74 (m, 3H, -CH3), 1.62 (s, 15H, C5(CH3)5), 0.26 (d, 9H, 3JP-H = 14.5 Hz, -C(CH3)3). 31P{1H} NMR (C6D6, 293 K): δ [ppm] = 151.9 (m, 1P, PA), 47.0 (m, 2P, PM,M′), −117.2 (m, 2P, Px,x′). 31P NMR (C6D6, 293 K): δ [ppm] = 151.9 (m, 1P, PA), 47.0 (m, 2P, PM,M′), −117.2 (m, 2P, Px,x′): (for coupling constants, see Supplementary Table 8). LIFDI-MS (toluene): 418.02 (100%, [M]+); analysis (calcd., found for C15H27FeP5): C (43.09, 43.63), H (6.51, 6.36).

Synthesis of [Cp*Fe(η4-P5PhMe)] (3f)

Compound 1 (0.4 mmol, 138.4 mg, 1 eq) was dissolved in 20 mL of DME, cooled to −50 °C and 0.22 mL of a 1.9 molar solution of PhLi (0.4 mmol, 1 eq) in dibutyl ether was added. The solution was stirred for 2 h and allowed to reach room temperature. To the brown solution, a 1.11 molar solution of MeI (0.4 mmol, 0.36 mL, 1 eq) in Et2O was added. The solution was stirred for 18 h, the solvent removed under reduced pressure, the remaining brown residue extracted with 3 × 5 mL of n-hexane and filtered over diatomaceous earth. The volume was reduced in vacuo and the solution stored at −30 °C. Compound 3f was isolated after 3 days as dark green plates. Yield: 132.0 mg (0.30 mmol, 75%). 1H NMR (C6D6, 293 K): δ [ppm] = 6.82 (m, 2H, -C6H5), =6.71 (m, 3H, -C6H5), 1.95 (dt, 3H, 2JP-H = 10.8 Hz, 3JP-H = 5.4 Hz, -CH3), 1.64 (s, 15H, C5(CH3)5). 31P{1H} NMR (C6D6, 293 K): δ [ppm] = 117.7 (m, 1P, PA), 40.5 (m, 2P, PM,M′), −109.5 (m, 2P, Px,x′). 31P NMR (C6D6, 293 K): δ [ppm] = 117.7 (m, 1P, PA), 40.5 (m, 2P, PM,M′), −109.5 (m, 2P, Px,x′); (for coupling constants, see Supplementary Table 9). LIFDI-MS (toluene): 437.96 (100%, [M]+); analysis (calcd., found for C17H23FeP5): C (46.61, 47.09), H (5.29, 5.26).

Synthesis of PR′R″R‴ (4a-c)

Compound 3d (0.1 mmol, 40.4 mg, 1 eq), 3e (0.2 mmol, 83.6 mg, 1 eq) or 3f (0.1 mmol, 43.8 mg, 1 eq) were dissolved in DME and cooled to −50 °C. To the brownish-green solution, a −50 °C cold solution of KBnz50 (0.1 mmol, 13.0 mg, 1equ. for 3d, 3f, 0.2 mmol, 26.0 mg, 1equ. for 3e) in DME was added. The colour changed to dark red and rapidly back to brownish-green. The solution was stirred overnight and allowed to reach room temperature. The solvent was slowly removed under reduced pressure (note: in order to avoid the loss of the desired phosphine by removal of the solvent, lower boiling solvents like THF can be used for the reaction). The oily residue was extracted with n-pentane (3 × 5 mL) and decanted off from the remaining solid (compound 5, see below). The solvent was removed in vacuo and compound 4a-c (a: R′ = Me, R″ = iPr, R‴ = Bnz; b: R′ = tBu, R″ = Me, R‴ = Bnz; c: R′ = Ph, R″ = Me, R‴ = Bnz) can be isolated as a viscous liquid. Yield: 4a: 14.8 mg (0.082 mmol, 82%), 4b: 21.5 mg (0.11 mmol, 55%), 4c: 17.1 mg (0.08 mmol, 80%).

1H NMR: 4a (C6D6, 293 K): δ [ppm] = 7.07 (m, 5H, -CH2-C6H5), 2.75 (dd, 1H, 3JP-H = 13.14 Hz, 3JH-H = 3.51 Hz, -CH2-Ph), 2.41 (dd, 1H, 3JP-H = 13.18 Hz, 3JH-H = 1.69 Hz, -CH2-Ph), 1.28 (m, 1H, -(CH)-CH3), 0.99 (dd, 3H, 3JP-H = 14.1 Hz, 2JH-H = 7.0 Hz, -(CH)-(CH3)2, 0.91 (dd, 3H, 3JP-H = 12.2 Hz, 2JH-H = 6.9 Hz, -(CH)-(CH3)2), 0.73 (d, 9H, 2JP-H = 3.5 Hz, -CH3); 4b (C6D6, 293 K): δ [ppm] = 7.13 (m, 8H, -CH2-C6H5), 2.76 (dd, 1H, 3JP-H = 12.2 Hz, 3JH-H = 3.9 Hz, -CH2-Ph), 2.33 (d, 1H, 3JP-H = 13.0 Hz, 3JH-H = 4.1 Hz, -CH2-Ph), 0.94 (d, 9H, 3JP-H = 11.2 Hz, -C(CH3)3), 0.71 (d, 3H, 2JP-H = 3.5 Hz, -CH3); 4c (C6D6, 293 K): δ [ppm] = 7.21 (m, 5H, -CH2-C6H5), 7.21 (m, 5H, -CH2-C6H5), 2.95 (dd, 2H, 2JP-H = 13.2 Hz, 3JH-H = 2.9 Hz, -CH2-Ph), 2.79 (d, 1H, 3JP-H = 13.0 Hz, -CH2-Ph), 1.10 (d, 3H, 2JP-H = 4.0 Hz, -CH3).

31P{1H} NMR: 4a (C6D6, 293 K): δ [ppm] = −18.5 (s); 4b (C6D6, 293 K): δ [ppm] = −7.00 (s); 4c (C6D6, 293 K): δ [ppm] = −29.6 (s).

31P NMR: 4a (C6D6, 293 K): δ [ppm] = −18.5 (s); 4b (C6D6, 293 K): δ [ppm] = −7.00 (s, br); 4c (C6D6, 293 K): δ [ppm] = −29.6 (s, br).

13C{1H} NMR: 4a (C6D6, 293 K): δ [ppm] = 138.5 (d, 2JP-C = 4.1 Hz, Ph), 129.1 (d, 3JP-C = 5.4 Hz, Ph), 128.2 (s, Ph), 125.5 (d, 4JP-C = 2.1 Hz, Ph), 34.8 (d, 1JP-C = 18.64 Hz, -(CH2)-Ph), 26.6 (d, 2JP-C = 11.71 Hz, -(CH)-(CH3)2), 19.2 (d, 1JP-C = 17.1 Hz, -(CH)-(CH3)2), 18.3 (d, 2JP-C = 12.8 Hz, -(CH)-(CH3)2), 8.1 (d, 1JP-C = 20.2 Hz, -CH3); 4b (C6D6, 293 K): δ [ppm] = 139.1 (d, 2JP-C = 6.9 Hz, Ph), 129.3 (d, 3JP-C = 6.3 Hz, Ph), 128.3 (d, 5JP-C = 0.7 Hz, Ph), 125.5 (d, 4JP-C = 2.3 Hz, Ph), 33.0 (d, 1JP-C = 20.8 Hz, -(CH2)-Ph), 27.0 (d, 1JP-C = 13.6 Hz, -C(CH3)3), 26.5 (d, 2JP-C = 13.6 Hz, -C(CH3)3), 6.2 (d, 1JP-C = 23.0 Hz, -CH3); 4c (C6D6, 293 K): δ [ppm] = 140.1 (d, 1JP-C = 17.9 Hz, Ph), 137.7 (d, 2JP-C = 4.5 Hz, Ph), 131.6 (d, 1JP-C = 18.6 Hz, Ph), 129.1 (d, 3JP-C = 5.4 Hz, Ph), 128.2 (d, 2JP-C = 11.2 Hz, Ph), 128.1 (s, Ph), 128.1 (s, Ph), 125.7 (d, 3JP-C = 5.4 Hz, Ph), 38.3 (d, 1JP-C = 17.1 Hz, -(CH2)-Ph), 10.2 (d, 1JP-C = 17.1 Hz, -CH3).

Synthesis of SPR′R″R‴ (4a′-c′)

Compound 4a-c (0.1 mmol, 1 eq) were synthesized according to the previous procedure, dissolved in 10 mL of n-pentane, respectively, added to a solution of sulfur (0.1 mmol, 3.2 mg, 1 eq) in 10 mL n-pentane and stirred for 4 h at room temperature. The volume was reduced under reduced pressure and stored at −30 °C. Colourless crystals of 4a′-c′ (a: R′ = Me, R″ = iPr, R‴ = Bnz; b: R′ = tBu, R″ = Me, R‴ = Bnz; c: R′ = Ph, R″ = Me, R‴ = Bnz) were formed after 2 days. Yield: 4a′: 16.0 mg (0.089 mmol, 89%), 4b′: 17.5 mg (0.08 mmol, 80%), 4c′: 17.1 mg (0.08 mmol, 80%).

1H NMR: 4a′ (C6D6, 293 K): δ [ppm] = 7.05 (m, 5H, -CH2-C6H5), 2.89 (t, 1H, 3JP-H = 14.7 Hz, -CH2-Ph), 2.74 (dd, 1H, 3JP-H = 13.8 Hz, 3JH-H = 1.8 Hz, -CH2-Ph), 1.39 (m, 1H, -(CH)-CH3), 0.99 (dd, 3H, 3JP-H = 17.8 Hz, 2JH-H = 7.0 Hz, -(CH)-(CH3)2), 0.97 (d, 9H, 2JP-H = 11.9 Hz, -CH3), 0.84 (dd, 3H, 3JP-H = 17.5 Hz, 2JH-H = 6.9 Hz, -(CH)-(CH3)2); 4b′ (C6D6, 293 K): δ [ppm] = 7.27 (m, 2H, -CH2-C6H5), 7.12 (m, 3H, -CH2-C6H5), 2.79 (m, 2H, -CH2-Ph), 0.97 (d, 3H, 2JP-H = 11.8 Hz, -CH3), 0.95 (d, 9H, 3JP-H = 15.5 Hz, -C(CH3)3); 4c′ (C6D6, 293 K): δ [ppm] = 7.60 (m, 2H, -Ph), 7.00 (m, 6H, -Ph), 6.91 (m, 2H, -Ph), 3.01 (d, 2H, 2JP-H = 14.1 Hz, -CH2-Ph), 1.39 (d, 3H, 2JP-H = 12.7 Hz, -CH2-Ph).

31P{1H} NMR: 4a′ (C6D6, 293 K): δ [ppm] = 51.3 (s); 4b′ (C6D6, 293 K): δ [ppm] = 58.6 (s); 4c′ (C6D6, 293 K): δ [ppm] = 38.0 (s).

31P NMR: 4a′ (C6D6, 293 K): δ [ppm] = 51.3 (s, br); 4b′ (C6D6, 293 K): δ [ppm] = 58.6 (m); 4c′ (C6D6, 293 K): δ [ppm] = 38.0 (m).

13C{1H} NMR: 4a′ (C6D6, 293 K): δ [ppm] = 132.7 (d, 2JP-C = 7.9 Hz, Ph), 129.7 (d, 3JP-C = 4.9 Hz, Ph), 128.2 (d, 4JP-C = 2.8 Hz, Ph), 126.8 (d, 3JP-C = 3.3 Hz, Ph), 39.4 (d, 1JP-C = 45.1 Hz, -(CH2)-Ph), 28.7 (d, 1JP-C = 52.3 Hz, -(CH)-(CH3)2), 15.6 (d, 2JP-C = 21.3 Hz, -(CH)-(CH3)2), 15.5 (d, 1JP-C = 52.5 Hz, -CH3 15.3 (d, 2JP-C = 26.8 Hz, -(CH)-(CH3)2); 4b′ (C6D6, 293 K): δ [ppm] = 132.9 (d, 2JP-C = 7.8 Hz, Ph), 130.6 (d, 3JP-C = 5.7 Hz, Ph), 127.1 (d, 5JP-C = 2.9 Hz, Ph), 127.1 (d, 4JP-C = 3.2 Hz, Ph), 36.1 (d, 1JP-C = 43.1 Hz, -(CH2)-Ph), 33.3 (d, 1JP-C = 49.4 Hz, -C(CH3)3), 24.6 (d, 2JP-C = 1.4 Hz, -C(CH3)3), 13.4 (d, 1JP-C = 50.8 Hz, -CH3); 4c′ (C6D6, 293 K): δ [ppm] = 131.0 (d, 2JP-C = 9.9 Hz, Ph), 130.8 (d, 4JP-C = 2.8 Hz, Ph), 130.0 (d, 3JP-C = 5.2 Hz, Ph), 126.8 (d, 3JP-C = 3.7 Hz, Ph), 43.8 (d, 1JP-C = 49.0 Hz, -(CH2)-Ph), 18.9 (d, 1JP-C = 57.3 Hz, -C(CH3)3).

4a′: EI-MS (CH2Cl2): 212.08 (37%, [M]+), 91.05 (100% [Bnz]+), 170.03 (46% [M-C3H6]+).

4b′: GC-MS (CH2Cl2): 226.09 (20%, [M]+), 91.06 (100% [Bnz]+),170.03 (44% [M-tBu + H]+).

4c′: EI-MS (CH2Cl2): 246.06 (44%, [M]+), 155.01 (100% [M-Bnz]+), 91.06 (25% [Bnz]+).

Targeted synthesis of [K(18c6)(thf)][Cp*Fe(η4-P4)] (5)

Compound 3c (0.2 mmol, 75.2 mg, 1 eq), KBnz (0.2 mmol, 26.0 mg, 1 eq) and 18-crown-6 (0.2 mmol, 52.9 mg, 1 eq) were dissolved separately in THF and cooled to −80 °C. The potassium benzyl solution is added to compound 3c, stirred for 2 min and 18c6 is added afterwards. The mixture was allowed to stir overnight and reach room temperature. The solvent of the resulting brown solution was removed in vacuo. The by-product PMe2Bnz was removed by washing the oily residue three times with 5 mL of n-pentane, resulting a greenish powder, which was dissolved in THF and layered with n-hexane. The solution was stored at −30 °C and compound 5 was isolated as highly air-sensitive green/turquoise plates/blocks after 1 week. Yield: 111.9 mg (0.16 mmol, 81%). 1H NMR (THF-d8, 293 K): δ [ppm] = 3.73 (s, 26 H, 18-crown-6), 2.26 (s, 15H, C5(CH3)5). 31P{1H} NMR (THF-d8, 293 K): δ [ppm] = 118.9 (s), 31P NMR (THF-d8, 293 K): δ [ppm] = 118.9 (s). ESI-MS (anion, DME): m/z = 314.95 (100%, [M]−); analysis (calcd., found for C22H39FeP4KO6 ∙ (THF)0.75: C (44.65, 44.28), H (6.74, 6.67).

Synthesis of PMe2Bnz and regeneration of [Cp*Fe(η5-P5)] (1) in a cyclic process

Compound 1 (3.00 mmol, 1.04 g, 1 eq) was dissolved in 100 mL tetraglyme, 1.88 mL of a 1.6 molar solution of MeLi (3.00 mmol, 1 eq) in Et2O was added and the resulting solution was allowed to stir for 5 min. To the reaction mixture, 2.80 mL of a 1.07 molar solution of MeI (3.00 mmol, 1 eq) in Et2O was slowly added at room temperature and stirred for 5 min. The diethylether of the stock solutions was removed under reduced pressure. The reaction mixture was cooled to −30 °C, a solution of KBnz (3.00 mmol, 390.7 mg, 1 eq) in 50 mL tetraglyme was added and the reaction was stirred for 30 min. The cooling bath was removed and the phosphine PMe2Bnz was distilled off under reduced pressure (1 × 10−3 mbar, 55 °C). White phosphorus (3.00 mmol, 371.7 mg, 1 eq) was added to the remaining solution and the reaction mixture was heated under reflux for 1 h at 275 °C. The described steps were repeated two more times. The starting material 1 was isolated after column chromatographic workup (silica gel, n-hexane, 20 × 2 cm) in overall 69% yield (2.07 mmol, 2.07 g) (Supplementary Fig. 66). The phosphine PMe2Bnz was isolated as a pyrophoric colourless liquid in an overall yield of 79% (2.37 mmol, 315.0 mg) (see Supplementary Fig. 67, Supplementary Table 40).