Tetramethylphosphinane as a new secondary phosphine synthon

Abstract

Secondary phosphines are important building blocks in organic chemistry as their reactive P—H bond enables construction of more elaborate molecules. In particular, they can be used to construct tertiary phosphines that have widespread applications as organocatalysts, and as ligands in metal-complex catalysis. We report here a practical synthesis of the bulky secondary phosphine synthon 2,2,6,6-tetramethylphosphinane (TMPhos). Its nitrogen analogue tetramethylpiperidine, known for over a century, is used as a base in organic chemistry. We obtained TMPhos on a multigram scale from an inexpensive air-stable precursor, ammonium hypophosphite. TMPhos is also a close structural relative of di-tert-butylphosphine, a key component of many important catalysts. Herein we also describe the synthesis of key derivatives of TMPhos, with potential applications ranging from CO₂ conversion to cross-coupling and beyond. The availability of a new core phosphine building block opens up a diverse array of opportunities in catalysis.

Di-tert-butylphosphine is an important substituent in many phosphine ligands, however, its heterocyclic analogues have not yet been isolated. Here, the authors report the synthesis and isolation of 2,2,6,6-tetramethylphosphinane on a gram scale starting from ammonium hypophosphite, and its facile use as a building block in ligand construction.

Introduction

Over a century ago, the sterically hindered base 2,2,6,6-tetramethylpiperidine (TMP) was first isolated by Franchimont and Friedmann from aqueous ammonia and phorone¹. TMP and its derivatives would turn out to have widespread use in organic chemistry including the oxidant and radical trap TEMPO^2,3, the superbase LiTMP⁴, and hindered amine light stabilisers (HALS)⁵. Recently, it has been used as the base component of certain frustrated Lewis Pairs⁶. Yet surprisingly, the synthesis of the phosphorus congener, 2,2,6,6-tetramethylphosphinane (TMPhos, Fig. 1), has not been described until now.

Fig. 1. Target phosphine synthon TMPhos.

a TMPhos (2,2,6,6-tetramethylphosphinane) as the P-analogue of TMP and structurally related to ^tBu₂PH. b Well-known phosphine ligands containing -P^tBu₂.

Structurally related to TMPhos is di-tert-butylphosphine (^tBu₂PH), a bulky secondary phosphine synthon and an important component in many well-known ligands for homogeneous catalysis (Fig. 1). For example, the Alpha process for the commercial production of methyl methacrylate employs di-tert-butylphosphino-o-xylene palladium catalyst (Pd-DTBPX) in one of the key steps: methoxycarbonylation of ethylene to methylpropanoate⁷. JohnPhos from the Buchwald ligand family has multiple applications in Pd–catalysed cross-coupling reactions, such as Suzuki–Miyaura reactions between boronic acids and aryl halides⁸, amination of aryl halides and triflates^9,10, as well as arylation of thiophenes¹¹. Pincer ligands¹² also have widespread applications in homogeneous catalysis and have undergone a renaissance in recent decades due to their ability to participate in metal-ligand cooperation¹³. In many examples the ^tBu₂P motif is an important component of such pincer ligands^14,15.

^tBu₂PH is a widely used building block in the design of many ligands. Yet its 5- and 6-membered heterocyclic secondary phosphine analogues, 2,2,5,5-tetramethylphospholane and TMPhos have not yet been isolated. Sulfide derivatives of the former were detected from the reaction between the corresponding di-Grignard and PhPCl₂ (Fig. 2a), but the secondary phosphine has thus far remained elusive¹⁶. McNulty and Capretta obtained a series of substituted tertiary phosphinanes from the reaction between R-PH₂ and phorone, followed by reduction of the ketone group, Fig. 2b¹⁷. They found that this family of ligands was tunable, cheap and efficient in cross-coupling reactions, like popular ligands such as ^tBu₃P. The six-membered heterocycle 2,2,6,6-tetramethylphosphinan-4-one has been prepared by a lithium cleavage of the corresponding phenyl derivative, Fig. 2c. This bulky secondary phosphine is probably the closest known structure to TMPhos, however, the yield of the reaction was low (~10%) most likely due to competing lithium cleavage of the P-tert-alkyl bonds¹⁸. Van Meurs and co-workers have also recently demonstrated that the bulky 1,2-bis(2,2,6,6-tetramethylphosphorinan-4-one)xylene (BPX, Fig. 2d), is a more effective ligand in the isomerising carbonylation of alkenes, compared to its acyclic analogue¹⁹.

Fig. 2. Known bulky phospholanes and phosphinanes.

a Phospholane sulfide, b phosphinanes, c phosphorinones, d bis(phosphorinone)xylene, BPX.

Heterocyclic phosphines offer unique properties that can be tuned by ring size, substituents and functional groups^20–23. In order to access more ligands incorporating bulky phosphinanes, we targeted the synthesis of the secondary phosphine synthon, TMPhos. Herein, we report the synthesis and isolation of TMPhos on a multigram scale starting from ammonium hypophosphite (NH₄H₂PO₂), an abundant low-cost reducing agent used in metallurgy²⁴. We demonstrate the facile use of TMPhos as a building block in the construction of various ligands and compare them to structurally similar commercial counterparts.

Results and discussion

Initial synthesis of TMPhos

2,2,6,6-Tetramethylpiperidine can be made via conjugate addition of ammonia and phorone to give 2,2,6,6-tetramethyl-4-piperidinone, followed by a Wolff–Kishner reduction of the ketone⁴. Our approach was based on a similar strategy: a ring forming reaction between phorone and a suitable phosphorus precursor; followed by the subsequent reduction of the ketone group to furnish TMPhos. Phorone can be readily obtained by the aldol condensation of acetone²⁵. However, Welcher and Day had attempted the Michael addition of phosphine (PH₃) to phorone 60 years ago and reported that no reaction occurred²⁶. Furthermore, phosphine gas is highly toxic, and pyrophoric, making further investigations into this reaction hazardous, in spite of recent progress being made in the development of procedures using in situ generated PH₃²⁷.

Aiming to avoid handling PH₃ gas we first attempted to obtain TMPhos via a Li–cleavage of 1-phenyl-2,2,6,6-tetramethylphosphinane in an adaptation of Pastor’s procedure¹⁸, see Fig. 3a. We first obtained phenyl phosphorinone (i) from the condensation of phorone and PhPH₂²⁶, followed by Wolff–Kishner reduction to produce phenyl phosphinane (ii)¹⁷. Reductive cleavage of aryl-phosphorus bonds using alkali metals can be used to generate lithium phosphide species which upon hydrolysis will generate secondary phosphines²⁸. The reaction between (ii) and Li at 5 °C furnished the desired secondary phosphine (iii) (³¹P δ = −9.1 ppm) but only as a minor product (~26% yield). The major product, 2,6-dimethylheptan-2-yl(phenyl)phosphine, has a slightly upfield ³¹P chemical shift of −11.0 ppm. This product arises from a competing ring opening reaction, as Li preferentially cleaves phosphorus bonded to the tertiary carbon atom²⁸. The ring-opened compound undergoes a second cleavage to generate PhPH₂ as well as 2,6-dimethylheptane. Despite being only a minor product, the desired secondary phosphine, TMPhos, could be isolated as the borane adduct by first performing vacuum distillation to separate it from the high boiling phosphines, followed by addition of borane dimethylsulfide complex to generate TMPhos·BH₃, (iv).

Fig. 3. Initial attempts to isolate TMPhos.

a Synthesis of TMPhos via Li cleavage of 1-phenyl-2,2,6,6-tetramethylphosphinane. b Attempted synthesis of (v).

Improved synthesis of TMPhos

Since the atom economy and reaction yield from the Li cleavage of (ii) were both poor, this route was not attractive for obtaining sufficient amounts of TMPhos to explore its downstream chemistry. Therefore, we continued to investigate alternative routes for the synthesis of TMPhos. There are a number of primary phosphine surrogates that mimic the reactivity of PH₃, such as methyl hypophosphite (H₂PO₂Me) and bis(trimethylsilyl)phosphonite ((Me₃SiO)₂PH) that can both be derived from commercially available and relatively benign hypophosphorous acid. By this route, we obtained H₂PO₂Me from the alkylation of hypophosphorous acid with trimethyl orthoformate²⁹. However, in the presence of NEt₃ we observed no reaction between phorone and H₂PO₂Me, but instead a base-catalysed disproportionation to hypophosphorous acid and dimethoxyphosphine occurred, see Fig. 3b.

We therefore turned our attention to the more reactive (Me₃SiO)₂PH, which had been shown to undergo Michael additions to conjugated alkenes^30–32, and nucleophilic substitutions with alkyl halides³³ to furnish mono- or di-substituted phosphinic acids and even heterocycles. As (Me₃SiO)₂PH is pyrophoric it is typically generated in situ. We generated (Me₃SiO)₂PH either by the reaction between NH₄H₂PO₂ and hexamethyldisilazane (HMDS)³¹ or with NH₄H₂PO₂ and trimethylsilyl chloride (TMSCl) in the presence of Hünig’s base (Fig. 4)³⁰. The formation of (Me₃SiO)₂PH is evidenced by a doublet in the ³¹P NMR spectrum (δ = 141.7 ppm, ¹J_HP = 175.4 Hz). In both cases the (Me₃SiO)₂PH generated in situ reacted readily at RT with phorone producing intermediate 1. In this intermediate, one of the TMS groups has migrated to form a silyl enol ether. This was reflected by two inequivalent TMS groups in ²⁹Si{¹H} NMR, of which only one is coupled to ³¹P (d, 11.0 Hz) while the other remains a singlet, and was further confirmed by ¹H-²⁹Si 2D HMBC. The Z-isomer was determined to be the major product from 2D NOESY experiment, in which a distinctive NOE was observed between one terminal methyl group and the (distal) alkene proton of the enol ether (Supplementary Fig. 39 and Supplementary Data 1). In the ¹H NMR spectrum the PH appears as a doublet at 6.96 ppm, with a large coupling constant (¹J_HP = 544.2 Hz), consistent with that of a hypophosphite³⁴.

Fig. 4. Synthesis of TMPhos (8).

a Hünig’s Base/TMSCl/NH₄H₂PO₂, b HMDS/NH₄H₂PO₂, c 125 °C, 72 h, d microwave heating at 220 °C, 90 min.

The enolisation of the ketone group prevents a second Michael addition and ring closure from occurring due to the disruption of the conjugation. It could, however, be easily deprotected using 2 M aq. HCl to furnish enone-phosphinic acid 2 (³¹P δ = 46.1 ppm) in 93% yield on a multigram scale (>40 g). SC-XRD of compound 2 confirmed the expected structure (see Fig. 4 and Supplementary Fig. 3). We achieved the ring closure of 2 by addition of HMDS while monitoring the reaction by ³¹P NMR spectroscopy. Upon addition of HMDS to 2 the silyl ester 3 forms immediately, and upon heating this gradually forms the cyclised silyl intermediate 4.

The ring closure (6-endo-trig) is sluggish likely due to the steric constraints. Initially attempts at ring closure in 1,2-dichloroethane (DCE) at 70 °C required two weeks to approach full conversion. Later we changed the solvent to xylenes, which allowed for a reaction temperature of 125 °C and enabled good conversion (>80%) in 3 days. Alternatively, full conversion can be obtained in just 90 min using a microwave reactor at 220 °C, with a comparable isolated yield of 40%. The reaction was typically carried out at a concentration of ~0.2 M of phosphinic acid in xylenes) in order to minimise potential intermolecular side reactions; at higher concentration (0.6 M) a polymeric precipitate was observed. We also found that intermediate 3 can be formed via a partial hydrolysis of intermediate 1, since the silyl enol ether will preferentially hydrolyse in the presence of one equivalent of protic solvent such as ethanol. This enables a one-pot synthesis of 2,2,6,6-tetramethylphosphorinic acid, 5, however, we found that less side-products were obtained if the phosphinic acid 2 was first isolated and purified. After acid hydrolysis of intermediate 4, bifunctional bulky heterocyclic 5 was isolated in 58% yield from 2 at >10 g scale. Single crystals of 5 were grown by evaporation of a solution of the compound in acetone and a representation of the molecular structure is shown in Fig. 4 and Supplementary Fig. 4.

The conversion of 5 to TMPhos involves the reduction of both the ketone and phosphinic acid functional groups. We first reduced the ketone group using a standard Wolff–Kishner procedure to give 2,2,6,6-tetramethylphosphaninic acid, 6, in good yield (77%). Direct reduction of phosphinic acids to phosphine has been reported using Ph₂SiH₂³⁵ or PhSiH₃³⁰. However, these approaches gave low or no yield of the desired secondary phosphine when applied to compound 6, presumably due to either the increased steric bulk or the more basic P character (as compared to aromatic P compounds). Conversion to TMPhos was instead achieved via reduction of the corresponding phosphinic chloride 7, obtained by chlorination using (COCl)₂ in the presence of catalytic DMF. After LiAlH₄ reduction, TMPhos, 8, was obtained by distillation in moderate yield (53%). We have performed all synthetic steps to TMPhos on at least 5 g scale (reactant) with some on significantly larger scales. For safety reasons, at lab scale some steps were difficult to scale beyond certain thresholds (e.g. Wolff–Kishner reduction and LiAlH₄ reduction) and we are working on the process development of these steps.

Properties and reactivity of TMPhos

TMPhos is a colourless liquid with a ³¹P chemical shift of −9.1 ppm (¹J_PH = 200.0 Hz), significantly upfield of ^tBu₂PH (δ = 20.6 ppm, ¹J_PH = 193.0 Hz). Six-membered phosphinanes typically adopt ring shapes typical of cyclohexanes, with chair confirmations normally observed in solid state structures (vide infra)²². Compared to ^tBu₂PH, TMPhos is notably more resistant to oxidation. When a solution of TMPhos in CDCl₃ was exposed to air, remarkably no oxidation was detected in the first 24 h. The slower oxidation of TMPhos was gratifying since air oxidation is a categorical weakness of alkyl phosphines and may arise from a less basic P in TMPhos compared with ^tBu₂PH (vide infra). Even after exposure to air for 5 days only 20% of TMPhos had oxidised to the phosphine oxide 9 as the sole product (which could also be generated cleanly using m-CPBA, Fig. 5). In contrast, 80% of ^tBu₂PH had decomposed to a mixture of products resulting from oxygen insertion into a P-^tBu bond³⁶ forming t-butylphosphinate which hydrolyses to t-butylphosphinic acid (Supplementary Figs. 1 and 2). The heterocyclic conformation of TMPhos presumably prevents oxygen insertion from occurring which is a clear demonstration of its unique properties.

Fig. 5. Conversion of TMPhos to derivatives and its coordination compounds.

Conversion of TMPhos to the corresponding oxide (9), borane adduct (10), chlorophosphine (11), selenide (12) and rhodium (13), and palladium complexes (14, 15).

To demonstrate the versatility of TMPhos as a building block in organic synthesis, several TMPhos derivatives 10 and 11 were also synthesised (Fig. 5). For benchtop use, air-stable tetramethylphosphinane borane complex TMPhos·BH₃ (compound 10) was obtained by stirring equimolar amounts of TMPhos with BH₃·SMe₂. Dialkylchlorophosphines are also synthetically versatile synthons, and chlorination of TMPhos with one equivalent of methyl trichloroacetate affords the chlorophosphine 11 in high yield (72%).

We synthesised the phosphorus selenide 12 since ¹J_PSe coupling constants can give an indication of the basicity of phosphines³⁷. The ¹J_PSe of compound 12 (717.2 Hz) is larger than reported for the corresponding ^tBu₂P(H) = Se (704 Hz) which suggests TMPhos is less basic³⁸. This was supported by a ν(C‍ ≡ ‍O) of 1971.0 cm⁻¹ for [(TMPhos)Rh(CO)(acac)] 13 compared to 1963.4 cm⁻¹ for [(^tBu₂PH)Rh(CO)(acac)] (see Supplementary Methods). This is consistent with observations for similar compounds where it was postulated that a smaller C-P-C angle resulting from the ring conformation contributes to a less basic P¹⁹.

In comparison with tertiary phosphines, palladium complexes bearing secondary phosphines are relatively uncommon owing to their reactive P–H bond. Homoleptic complexes of secondary phosphines are rare, with [(^tBu₂PH)₃Pd⁽⁰⁾] and [(Ph₂PH)₄Pd⁽⁰⁾] being the only reported examples for palladium^39,40. We were therefore interested to obtain a homoleptic complex of Pd bearing TMPhos ligands. [(TMPhos)₃Pd⁽⁰⁾] 14 was synthesised following the same procedure reported for [(^tBu₂PH)₃Pd⁽⁰⁾]³⁹. The addition of excess TMPhos and allyl(cyclopentadienyl)palladium(II) was accompanied by the reductive elimination of 5-allyl-1,3-cyclopentadiene to give 14 as a yellow solid in 67% yield. In solution, 14 gave a broad doublet at 42.1 ppm (¹J_PH = 246 Hz) in its ³¹P NMR spectrum, upfield of the corresponding ^tBu₂PH complex (δ = 54.5 ppm, ¹J_PH = 256 Hz). In solution Leoni found evidence of a rapid equilibrium between [(^tBu₂PH)₃Pd] and [(^tBu₂PH)₂Pd]³⁹. The latter thermally transforms to dimeric [Pd(μ-P^tBu₂)(HP^tBu₂)]₂ with the concomitant loss of H₂. We found that complex 14 underwent a similar transformation. A solution of 14 kept at RT formed red single crystals of bimetallic complex 15, its identity being confirmed by SC-XRD (Fig. 6 and Supplementary Fig. 5). The structure contains two bridging TMPhos ligands and two terminally bound TMPhos ligands with H atoms in the axial position. The Pd1-Pd1ⁱ bond length of 2.61(1) Å is statistically comparable with corresponding [Pd(μ-P^tBu₂)(HP^tBu₂)]₂ complex where it was 2.60(1) Å⁴¹. Likewise the Pd-P1 and Pd-P2 bond lengths of 2.27(1) and 2.31(1) Å, respectively, were also comparable to the ^tBu₂PH complex, where they were 2.29(1) and 2.34(1) Å. The most noticeable difference is the anticipated C-P-C bond angle contraction in the phosphinane rings. In the case of terminal phosphinane the C1-P1-C5 angle was 105.3(1)° whereas the bridging phosphinane had an even smaller C10-Pd-C14 angle of 104.6(1)°. These angles are noticeably smaller than the corresponding angles in the ^tBu₂PH complex which were 112.0(3)° and 111.3(3)°, respectively. These differences are reflected in a smaller cone angle calculated for the terminal TMPhos-Pd (126°) compared with ^tBu₂PH-Pd (134°), see Supplementary Methods Section 5.

Fig. 6. Molecular structure of compound 15.

Selected bond lengths (Å) and angles (°): Pd1-Pd1ⁱ 2.6053(3), Pd1-P1 2.2686(3) Pd1-P2 2.3124(3), Pd1-P2ⁱ 2.3426(3), P1-Pd1-P2 131.267(9), P1-Pd1-P2ⁱ 116.715(9), P2-Pd1-P2ⁱ 112.017(7), Pd1-P2-Pd1ⁱ 67.982(7), P1-Pd1-Pd1ⁱ 172.166(8), P2-Pd1-Pd1ⁱ 56.562(7), P2ⁱ-Pd1-Pd1ⁱ 55.455(7), C1-P1-C5 105.34(5), and C10-P2-C14 104.64(4).

Using TMPhos we were also able to readily construct a variety of pro-ligands based on well-known phosphines as shown in Fig. 7. Simple alkyl phosphinanes, such as those shown in Fig. 2b, have been demonstrated to be effective ligands in palladium catalysed cross-coupling reactions¹⁷, and JohnPhos is commercialised for the same purpose⁴². We therefore synthesised the monophosphine ^TMPhos(Biphenyl), 16 which is a combination of a phosphinane ring and a classic Buchwald biphenyl substituent. To showcase TMPhos-based bidentate ligands, we first prepared the simple bis-phosphines BTMPPr 17 and BTMPBu 18 with propyl and butyl backbones respectively. The ^tBu₂P versions of these ligands have been applied in carbonylation of alkenes⁴³ and polymerisation of phenylacetylene⁴⁴. Next we constructed a bis-phosphine bearing the well-known xanthene backbone ^TMPhos(Xantphos) 19 as Xantphos has many applications in coordination chemistry and catalysis⁴⁵. For example, ^tBu-Xantphos has been shown to stabilise Ni^I alkyl complexes that rapidly insert CO₂ to form the corresponding Ni-carboxylate species⁴⁶. We also made bis(tetramethylphosphinane)xylene (BTMPX) 20 and its SC-XRD data is shown in Supplementary Fig. 6. This bis(phosphine) is analogous to DTBPX, a ligand used industrially and recently reviewed⁴⁷. Finally we synthesised the pincer ligands ^TMPhos(PCP) and ^TMPhos(PNP), (21 and 22), as pincer ligands are now well established in homogenous catalysis with an abundance of applications^48,49. The above examples were chosen as their ^tBu₂P analogues are well known, in many cases affording highly active catalysts in a variety of different reactions. In all cases the TMPhos derivatives possess significantly upfield ³¹P chemical shifts: between 16 to 24 ppm more negative than their ^tBu₂P counterparts. This is likely a consequence of the γ-substituent effects imposed by the ring, rather than a simple reflection of the electron-donating power of the phosphine⁵⁰.

Fig. 7. Ligands and complexes incorporating TMPhos synthons.

Ligands incorporating TMPhos (16–22) and associated metal complexes (23–25) with potential catalytic applications (in green).

We went on to explore the coordination chemistry of selected pro-ligands, targeting variations of well-known pre-catalysts. The first example, [(BTMPX)Pd(O₂CCF₃)₂] 23 was synthesised from [(BTMPX)PdCl₂] (S3, Supplementary Methods and Supplementary Fig. 7) and investigated due to its similarity to the commercially relevant Mitsubishi–Lucite catalyst. Its molecular structure is shown in Fig. 8 and Supplementary Fig. 8, together with selected key bond lengths and angles in Table 1. The complex adopts a four-coordinate square planar geometry. The P-Pd-P angle of 101.9(1)° is slightly larger than for similar complexes like [(DTBPX)Pd(O₃SMe)₂]⁵¹ and [(BPX)Pd(O₂CCF₃)₂]¹⁹, where the bite angles are 100.6(1)° and 100.1(1)°, respectively. As seen in complex 15, smaller C-P-C angles were observed in the phosphinane ring. The C(^tBu)-P-C(^tBu) angle in the DTBPX analogues is about ~110°⁵², whereas for 23 the bond angles for C18-P1-C22 and C9-P2-C13 are 105.8(1)° and 106.8(1)° respectively. Taken together, it follows that complex 23 has a marginally smaller buried volume (%V_Bur = 50.0%) compared to that calculated for structurally related [(DTBPX)Pd(O₃SMe)₂] (%V_Bur = 53.1%), see Supplementary Figs. 11 and 14 for topographic steric maps^51,53.

Fig. 8. Molecular structures of TMPhos-containing metal complexes.

ORTEP representations for the crystal structures of [(BTMPX)Pd(O₂CCF₃)₂] 23 and [^TMPhos(PNP)Ru(CO)(Cl)H] 24, drawn at 50% probability. The fluorine atoms of the CF₃ groups exhibit rotational disorder in the crystal structure.

Table 1.

Selected bond lengths (Å) and angles (°) of [(BTMPX)Pd(O₂CCF₃)₂], 23, and [^TMPhos(PNP)Ru(CO)(Cl)H], 24.

[(BTMPX)Pd(O2CCF3)2], 23				[TMPhos(PNP)Ru(CO)(Cl)H], 24
Bond lengths (Å)		Bond angles (°)		Bond lengths (Å)			Bond angles (°)
					Molecule-1	Molecule-2		Molecule-1	Molecule-2
Pd1-P1	2.2964(6)	P1-Pd-P2	101.90(2)	Ru1-P1	2.3285(8)	2.3412(8)	P1-Ru1-P2	159.92(3)	159.68(4)
Pd1-P2	2.3147(7)	P1-Pd1-O1	87.64(5)	Ru1-P2	2.3289(8)	2.3379(9)	N1-Ru1-C52	175.52(13)	173.22(14)
Pd1-O1	2.0698(17)	P1-Pd1-O3	168.17(5)	Ru1-Cl1	2.5463(8)	2.5471(9)	C17-P1-C21	105.23(19)	104.77(18)
Pd1-O3	2.0974(17)	P2-Pd1-O1	170.44(5)	Ru1-N1	2.154(3)	2.176(3)	C7-P2-C11	104.46(18)	105.1(2)
		C22-P1-C18	105.80(11)	Ru1-C52	1.843(4)	1.836(4)
		C9-P2-C13	106.82(11)

We also prepared the ruthenium hydride complex [^TMPhos(PNP)Ru(CO)(Cl)H] 24. The ^tBu₂P variant of this complex is an effective catalyst for the reversible hydrogenation of CO₂ to formates, giving a TOF in excess of 1 million⁵⁴. In our complex, the hydride signal appears as a triplet at −15.05 ppm (²J_HP = 19.2 Hz) in the ¹H NMR spectrum, very similar to the hydride signal in the corresponding ^tBu₂P complex (−15.22 ppm, ²J_HP = 19.4 Hz). Single crystals were obtained from the vapour diffusion of pentane into a saturated solution of the complex in CH₂Cl₂. The crystal structure contained two crystallographically independent complexes per unit cell (Fig. 8 and Supplementary Figs. 9-10). In one molecule there is one phosphorus above and below the plane of the pyridine ring, by 0.77 Å and 0.57 Å respectively. By contrast in the second molecule both P atoms lie on the same side of the pyridine ring by 0.55 Å and 0.91 Å respectively. The P-Ru-P angles of 159.9(1)° and 159.7(1)° are slightly larger than that for the corresponding ^tBu₂P complex at 158.4(1)°⁵⁵, whereas the N-Ru-CO angles of 175.5(1)° and 173.2(1)° are quite compressed as compared to an angle of 178.5(1)° in the ^tBu₂P version. Complex 24 has a slightly smaller %V_Bur of 51.9% compared to 53.3% for the analogous ^tBu₂P complex⁵⁵, see Supplementary Figs. 12, 13, and 15.

Finally, we prepared the TMPhos analogue of the classic Shaw Ir-^tBuPCP pincer complex, [^TMPhos(PCP)Ir(H)Cl], complex 25. The TMPhos variant, 25, possesses a broad hydride resonance in its ¹H NMR spectrum (−40.44 ppm) which is downfield compared to the ^tBu₂P complex (−43.37 ppm in CD₂Cl₂). This suggests a decreased electron density at the metal centre when TMPhos is used. Nearly half a century ago, Shaw’s seminal work on PCP pincers recognised the special properties conferred by bulky tertiary phosphine ligands⁵⁶. These include the ability to promote hydride formation and metalation reactions, as well as the stabilisation of coordinative unsaturation⁵⁶, anticipating the huge contribution bulky pincer ligands have subsequently made in catalysis. We therefore expect TMPhos ligands to also have useful catalytic applications and work is ongoing to investigate the use of 16–25 in this regard.

Conclusion

We have developed a multigram synthetic route to a bulky secondary heterocyclic phosphine synthon, TMPhos, starting from an inexpensive and air-stable phosphine precursor. Remarkably, this phosphorus heterocycle has been only described now, almost 120 years after its congener, the widely-used TMP. We have successfully demonstrated its use as a synthon by constructing a variety of tertiary phosphine ligands as well as several metal complexes incorporating the TMPhos substituent. We believe that TMPhos could find similar applications as the important acyclic di-tert-butylphosphine substituent in ligand design and catalysis as it offers a different steric environment, restricted rotation, and different electronic properties to previously known phosphine ligands.

Methods

Synthetic procedures

See Supplementary Information Section 1.

Phosphine oxidation tests

See Supplementary Figs. 1-2

NMR spectra

See Supplementary Figs. 17–123 in Supplementary Data 1.

Single crystal X-ray diffraction data

See Supplementary Figs. 3–10 and Supplementary Tables 1–14.

Buried volume calculations

See Supplementary Figs. 11–15.

Calculation of Tolman cone angles

See Supplementary Fig. 16 and Supplementary Tables 15–16.

Author contributions

J.D.N. is the lead author of manuscript and carried out and directed experiments. S.S. synthesised majority of compounds in the manuscript and was first to isolate TMPhos. X.Y.S. pioneered early work, isolating a key intermediate and revised manuscript. C.B.C. performed structural elucidation, spectral assignments and revised manuscript. S.A. carried out single-crystal X-ray crystallography for this work and revised manuscript. L.P.S. and M.v.M. provided ideas, directed work and revised manuscript.

Peer review

Peer review information

Communications Chemistry thanks the anonymous reviewers for their contribution to the peer review of this work.

Data availability

The authors declare that the data supporting the findings of this study are available within the article and Supplementary Information. For experimental details and compound characterisation data, see Supplementary Information. For NMR spectra, see Supplementary Data 1. The X-ray crystallographic data for compounds can be found in Supplementary Data 2–8 or obtained free of charge from The Cambridge Crystallographic Data Centre with the accession codes CDCC #2182394 (2), #2182395 (5), #2182396 (15), #2182397 (20), #2182398 (23), #2182399 (24) and #2182393 (S3) via www.ccdc.cam.ac.uk/data_request/cif.

Competing interests

The authors declare the following competing interest: some of the authors are inventors on a patent application that has recently been published⁵⁷. All other authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42004-023-00876-8.