Aminophobanes: hydrolytic stability, tautomerism and application in Cr-catalysed ethene oligomerisation.

9-Amino-9-phosphabicyclo[3.3.1]nonanes, (PhobPNHR ’ ; R = Me or i Pr) are readily prepared by aminolysis of PhobPCl and are signi ﬁ cantly less susceptible to hydrolysis than the acyclic analogues Cy 2 PNHR ’ . Treatment of Cy 2 PNHMe with Cy 2 PCl readily gave Cy 2 PNMePCy 2 . By contrast, treatment of PhobPCl with PhobPNHMe in the presence of Et 3 N does not a ﬀ ord PhobPNMePPhob but instead the salt [PhobP-( v NMeH)PPhob]Cl is formed which, upon addition of [PtCl 2 (NC t Bu) 2 ] gives the zwitterionic complex [PtCl 3 (PhobP( v NMeH)PPhob)]. The neutral PhobP( v NMe)PPhob is accessible from PhobNMeLi and is converted to the chelate [PdCl 2 (PhobPNMePPhob)] by addition of [PdCl 2 (cod)]. The anomalous preference of the PhobP group for the formation of PPN products is discussed. The unsymmetrical diphos ligands PhobPNMePAr 2 (Ar = Ph, o -Tol) are prepared, converted to [Cr(CO) 4 (PhobPNMePAr 2 )] and shown to form Cr-catalysts for ethene oligomerisation, producing a pattern of higher alkenes that corresponds to a Schulz-Flory distribution overlaid on selective tri/tetramerisation.


Introduction
Significant differences between the donor properties of phosphacycles and their acyclic analogues are to be expected because of the effects that ring constraints can have on the frontier orbital energies and the steric properties of the P-donor. 1 The molecular manifestations of these ring effects include stability (thermodynamic and kinetic) and structural rigidity which can be desirable qualities when considering the design of ligands. As a result, the coordination chemistry of phosphacycles and their applications in catalysis have attracted much academic and industrial attention. 2 Phobanes (PhobPZ, in Chart 1) are examples of rigid phosphacycles which have found important applications in homo-geneous catalysis 3,4 most notably in Co-catalysed hydroformylation. 5 We are interested in heterophobanes (PhobPZ where Z = a non-hydrocarbyl group) as ligands and particularly the effect that the phobyl group has on the reactivity of the P-Z bond. For example, fluorophobane (PhobPF) was shown to be a rare example of a fluorophosphine that is thermodynamically more stable to disproportionation and kinetically more stable to hydrolysis than acyclic fluorophosphine analogues; moreover PhobPF shows promise as a ligand for hydroformylation and hydrocyanation catalysis. 6 Aminodiphosphines R 2 PNR′PR 2 (known as PNP ligands, Chart 1) are excellent ligands for Cr-catalysed ethene tri/tetramerisation. As illustrated in Table 1, the characteristics of the R and R′ groups in R 2 PNR′PR 2 have a decisive effect on the chemoselectivity, productivity and therefore the potential industrial utility of the oligomerisation catalyst. 7,8 Increased steric bulk serves to lower the ratio of 1-octene to 1-hexene obtained, whilst changing from aryl to alkyl substituents on phosphorus dramatically reduces activity and increases polymer formation. The data in Table 1 highlight the impact of process conditions such as solvent, temperature and pressure upon the catalysis.
The industrial interest in PNP ligands 9 makes it important to have reliable methods for their preparation. As summarised in Scheme 1, the most general route to PNP ligands is the reaction of a primary amine with a chlorophosphine in the presence of a base. 10 The monophos R 2 PNHR′ species are presumed intermediates and when R or R′ is bulky, they are readily isolated and are potential intermediates to unsymmetrical PNP ligands. 11 When the substituents in either of the reactants R 2 PCl or R′NH 2 are bulky, a complication is the formation of the phosphinimine PPN compounds (Scheme 1); Maumela et al. 10 have shown that when R = Ph and R′ = t Bu, the PPN product is the kinetic product whose isomerisation to the thermodynamic PNP product is catalysed by Ph 2 PCl.
We were interested in investigating PNP ligands such as L a -L d where a phobyl group has been incorporated (Chart 2). It is shown here that the monophosphines L a and L b are readily prepared but their conversions to L c and L d has not been achieved. However the mixed diphosphines L f and L g are accessible and are shown to be ligands for Cr-catalysed ethene tri/ tetramerisation.

Results and discussion
Stereoelectronically, a Cy 2 P group can be viewed as an acyclic analogue of a PhobP group since ostensibly, they are similarly bulky dialkylphosphino groups. However, we have shown previously that the rigidity of the PhobP moiety leads to a larger steric profile than expected 12 and the approximately 90°C-P-C bridgehead angle in PhobP has the effect of lowering the HOMO and LUMO energies. 13 Ligands L 1 -L 4 (Chart 3) were targeted in the belief that a comparison of their chemistry with the phobane analogues L a -L d (Chart 2) would provide insight into the effect of the bicycle.

Monodentate aminophobanes
The monophobanes L a and L b were readily prepared by aminolysis of PhobPCl. The relative lability of L a and L b to hydrolysis (eqn (1)) was gauged by treatment of L a , L b , L 1 and L 2 with aqueous solutions under the same conditions and monitoring the formation of R 2 P(vO)H by 31 P NMR spectroscopy. All four aminophosphines eventually underwent complete hydrolysis but at different rates. Comparison of the extents of hydrolysis a Catalysis data taken from ref. 7 and 8. All wt% values are of total product slate. C 6 and C 8 refers to the entire C 6 and C 8 fractions and 1-C 6 and 1-C 8 refer to the proportion of the linear α-olefin within that fraction. b Catalysis conditions: 33 μmol CrCl 3 (THF) 3 ; 2 eq. ligand; 300 eq. MMAO-3A; 100 mL toluene. c Catalysis conditions: 33 μmol Cr(acac) 3 ; 2 eq. ligand; 300 eq. MMAO-3A; 100 mL toluene. d Catalysis conditions: 2.5 μmol Cr(acac) 3 ; 1.2 eq. ligand; 300 eq. MMAO-3A; 100 mL methylcyclohexane.
after 16 h (Table 2) shows that the NH i Pr group provides more protection from hydrolysis than the less bulky NHMe. Moreover, the bicyclic compounds PhobPNHR are significantly kinetically stabilised to hydrolysis with respect to the acyclic Cy 2 PNHR analogues. The resistance to hydrolysis of PhobPNHR is consonant with the phobyl moiety behaving as a bulky group. 12 The donor properties of L a and L b can be compared quantitatively with L 1 and L 2 from the ν CO values for their trans-[RhCl (CO)(L) 2 ] complexes. 14 The rhodium complexes were made in situ (see Scheme 2) and the recorded ν CO values ( Table 2) are consistent with L a and L b being slightly poorer σ-donors/better π-acceptors than their acyclic analogues L 1 and L 2 , as expected. 13 Ligands L a and L b form trans-dichloroplatinum(II) complexes 1a and 1b, and trans-tetracarbonylchromium(0) complexes 2a and 2b (Scheme 2). The crystal structures of 1b and 2b have been determined and are shown in Fig. 1 and 2. In addition, the crystal structure of trans-[PtCl 2 (L 2 ) 2 ] (3), an acyclic analogue of 1b has been determined (Fig. 3).
In aminophobane complex 1b and its acyclic analogue 3, the Pt metal centre is square planar. The Pt sits on a crystallographic inversion centre and the asymmetric unit consists of half of the complete molecule, consequently the N-P-P-N torsion angles are 180°in both cases, i.e. the anti conformer is adopted, as in other trans-[PtCl 2 (PhobPZ) 2 ] complexes. 6,12,15 The cone angle of L b in 1b is 111.8°and of L 2 in 3 is larger at 115.8°. In the structure of 2b, the asymmetric unit contains one complete molecule. The cone angle of L b in 2b is 109.2°w hich is smaller than in 1b, the compression probably reflecting the greater crowding in the octahedral complex. The N-P-P-N torsion angle in 2b is 108. 3(1)°indicating the amino   groups are gauche to each other, a conformation not previously observed in PhobPZ complexes.

Bidentate aminophobanes
The previously reported 16 diphosphinoamine L 3 is readily prepared from MeNH 2 and Cy 2 PCl in the presence of Et 3 N (Scheme 3). The intermediate in this reaction is presumably Cy 2 PNHMe (L 1 ) and indeed treatment of the isolated L 1 with Cy 2 PCl in the presence of Et 3 N in CH 2 Cl 2 gave L 3 quantitatively according to 31 P NMR spectroscopy. The spectrum of the reaction mixture also revealed a transient PPN species (as evidenced by a large J PP of 280 Hz) to which the tautomeric structure L′ 3 is assigned. The PPN species L′ 3 smoothly converted over 30 min to PNP ligand L 3 whose structure was confirmed by its conversion to the chelate complex 4 (Scheme 3), the crystal structure of which has been determined (Fig. 4).
The asymmetric unit consists of three independent molecules of 4 along with six chloroform molecules. Although the PtP 2 Cl 2 fragment is approximately planar (rms deviation ∼0.03 Å), the Pt has a distorted square planar geometry due to the constraints of the 4-membered PNP chelate. The three independent Pt-P-N-P rings are approximately planar with rms deviations of ∼0.03 Å.
In contrast to the ready reaction of L 1 with Cy 2 PCl to give PNP ligand L 3 (Scheme 3), the reaction of PhobPNHMe (L a ) with PhobPCl in the presence of NEt 3 or N-methylpyrrolidine did not give the expected diphosphinoamine L c . Instead, a PPN species ( J PP = 407 Hz) was the exclusive product; this was initially assigned structure L′ c but its 1 H NMR spectrum (which showed a multiplet at 7.01 ppm integrating for 1H) and mass spectrum (M + at [L′ c + 1]) led to its assignment as the HCl adduct L′ c ·HCl (Scheme 4). This was supported by its reaction with [PtCl 2 (NC t Bu) 2 ] which yielded crystals of the insoluble, zwitterionic complex [PtCl 3 (L′ c ·H)] (5) whose X-ray crystal structure is shown in Fig. 5. The conditions under which L′ c ·HCl was formed (Scheme 4) indicate that the iminophosphine L′ c is more basic than either NEt 3 or N-methylpyrrolidine.
The crystal structure of 5 has a square planar Pt with an rms deviation of the atoms from the square plane of ∼0.03 Å. The PPN ligand is rotated away from the PtCl 3 plane with torsion angles Cl1-Pt1-P1-P2 of −102.4(1)°and Cl3-Pt1-P1-P2 of 75.6(1)°.
Treatment of PhobPNMeH with n BuLi at −78°C followed by PhobPCl gave a PPN species with a J PP = 327 Hz (significantly smaller than the J PP of 407 Hz for L′ c ·HCl) that is assigned to the neutral L c ′ which has been isolated. No reaction occurred upon addition of PhobPCl to L′ c in CH 2 Cl 2 , conditions that might have been expected to tautomerise L′ c to L c . 10 It has previously been shown that some neutral PPN compounds rearrange when they react with [MCl 2 (cod)] (M = Pd or Pt) 17 or [NiBr 2 (dme)] 18 to give PNP chelate complexes. Reaction  of [PdCl 2 (cod)] with L′ c gave the chelate [PdCl 2 (L c )] (6) whose crystal structure has been determined and is shown in Fig. 6.
The asymmetric unit contains one molecule of 6, with the PdP 2 Cl 2 fragment being essentially planar (rms ∼ 0.07 Å) although the overall geometry is a distorted square planar due to the constraints of the 4-membered PNP chelate. As seen in the structure of analogue 4, the Pd1-P1-N1-P2 ring is also essentially planar with an rms deviation for the atoms of 0.01 Å.
From the homodiphos products obtained in the reactions of L a and L 1 with R 2 PCl (see Schemes 3 and 4), it appears that the PhobP group differs from Cy 2 P and Ar 2 P groups in promoting PPN over PNP formation; this raised the question of what would happen when the syntheses of the heterodiphos PNP The reaction between PhobPNHMe and Cy 2 PCl was followed by 31 P NMR spectroscopy and it was unambiguously shown that a PPN product was formed which, on the basis of its J PP of 358 Hz, was tentatively assigned to the protonated species L′ e ·HCl (Scheme 5); addition of Et 3 N led to multiple P-containing species but there was no evidence for the formation of the neutral PPN (L′ e ) or PNP (L e ) species. The reaction between PhobPCl and Cy 2 PNHMe was also monitored and in this case, 31 P NMR spectroscopy revealed that a PPN product was formed ( J PP = 403 Hz) which was assigned to the cationic species L′′ e ·HCl (Scheme 5), an isomer of L′ e ·HCl. It therefore appears that the PPN-promoting effect of the PhobP group dominates over the PNP-preference of the Cy 2 P group.
The unsymmetrical PNP ligands L f and L g ( J PP = 80 Hz in both) featuring PhobP groups were successfully prepared upon treatment of PhobPNHMe with Ar 2 PCl (Ar = Ph or o-Tol) in the presence of Et 3 N (eqn (2)). It therefore appears that the PPN formation promoted by the PhobP group is superseded by the PNP preference of the Ar 2 P groups.
The reaction of PhobPNHMe with Tol 2 PCl was monitored by 31 P NMR spectroscopy. A PPN species ( J PP = 331 Hz), tentatively assigned to L′ g ·HCl (Scheme 6) was formed rapidly which, upon treatment with NEt 3 , was transformed to L g ( J PP = 80 Hz).
Treatment of the bulky R 2 PNH i Pr (L b or L 2 ) with R 2 PCl (R 2 P = Cy 2 P or PhobP) under the conditions that converted R 2 PNHMe to the corresponding L 3 (Scheme 3) or L′ c ·HCl/L′ c (Scheme 4) gave, according to in situ 31 P NMR spectroscopy, mixtures of unidentified products as well as the reactants.
Under the conditions that smoothly led to the mixed PNP ligands L f and L g (eqn (2)), L b reacted with Ar 2 PCl to give PPN species whose structures were assigned to the protonated L′ h ·HCl and L′ i ·HCl (eqn (3)) on the basis of the large J PP values of 338 and 359 Hz respectively. Crystals of L′ h ·HCl were obtained and the crystal structure shown in Fig. 7 confirms the PPN assignment. The N⋯Cl distance of 3.101(1) Å indicates the presence of hydrogen-bonding between the N-H and Cl.

PPN versus PNP preferences
The N-and P-substituents determine whether PNP (A) or PPN (A′) species are formed in the reaction of amines with chlorophosphines (Scheme 7). In some cases, it has been shown 10, 19 that the PPN can be converted to the PNP tautomer using a R 2 PCl catalyst and we have observed PPN species as transients en route to the PNP products (e.g. Cy 2 PNMePCy 2 see Scheme 3) showing that the PNP is the thermodynamic product. In other cases (e.g. Cy 2 PN{SO 2 Ar}PCy 2 ) the neutral PPN tautomer appears to be the thermodynamic product. 17,18,20 An additional element observed in this work is the formation of a protonated A′·HCl product that is resistant to deprotonation by amines.
A pathway from chlorophosphine and primary amine to PNP/PPN products that encompasses these empirical observations is shown in Scheme 7. Nucleophilic attack by amine on chlorophosphine with loss of HCl would give the intermediate aminophosphine (step i). Reaction of a second chlorophosphine at the P site of the aminophosphine would give the salt A′·HCl (step ii) which can eliminate HCl to give the neutral A′ (step iii) and finally rearrangement to give PNP (step iv).
The formation of a PPN species when PhobPCl reacts with PhobPNHMe or PhobPNMeLi instead of PhobPNMePPhob Scheme 7 Proposed pathway for the conversion of an amine to a PNP ligand. A dotted line is used for step iv since this step is not observed with the aminophobanes.
contrasts with the smooth formation of Cy 2 PNMePCy 2 via a PPN intermediate; furthermore, PhobP(vNMe)PPhob does not isomerise to the PNP tautomer in the presence of PhobPCl. At present, it is not known whether these observations are due to PhobP(vNMe)PPhob being the thermodynamically preferred tautomer or slow kinetics of interconversion and therefore further investigation of this system is warranted.

Oligomerisation catalysis
The unsymmetrical PNP ligands L f and L g have been screened for Cr-catalysed ethene oligomerisation (see below) and it was therefore appropriate to explore their Cr coordination chemistry. The reaction of [Cr(CO) 4 (nbd)] with L f or L g gave the corresponding Cr(0) complexes 7 and 8 (eqn (4)) which have been fully characterised and their crystal structures have been determined ( Fig. 8 and 9).

ð4Þ
In combination with chromium, the ligands L f and L g gave moderate activities towards ethylene oligomerisation but the formation of polymer was high, as can be seen from Table 3. Within the liquid fraction, it is clear that a degree of selective oligomerisation to 1-hexene and 1-octene did occur ( particularly for L g ) but concurrently with Schulz-Flory selectivity (Fig. 10). The 1-octene to 1-hexene ratios obtained for both ligands is high.

Conclusions
The monodentate aminophobanes PhobPNHR (R = Me or i Pr) have been readily prepared and are more resistant to hydrolysis than their Cy 2 PNHR analogues consistent with the PhobP group having a greater effective steric bulk than Cy 2 P. Attempts to make the free ligand PhobPNMePPhob have been thwarted by formation of PPN species which resist tautomerisation although a rearrangement takes place in the presence of [PdCl 2 (cod)] to give the desired PNP-Pd chelate. The readily prepared mixed diphos ligands PhobPNMePAr 2 (Ar = Ph or o-Tol) in combination with Cr, catalysed the oligomerisation of ethylene with a partial selectivity to tri/tetramerisation, the remainder of the selectivity appearing to be Schulz-Flory in nature; the activities were moderate, but the polymer formation was high.

Experimental
Unless otherwise stated, all reactions were carried out under a dry nitrogen atmosphere using standard Schlenk-line techniques. Dry N 2 -saturated solvents were collected from a Grubbs system 21 in flame and vacuum-dried glassware. MeOH was dried over 3 Å molecular sieves, pentane was dried over 4 Å molecular sieves and both were deoxygenated by N 2 saturation. The starting materials PhobPCl, 13 [Cr(CO) 4 (η 4 -norbornadiene)], 22 [PtCl 2 (NC t Bu) 2 ], 23 [PdCl 2 (cod)], 24 were prepared by literature methods. All other reagents were used as received from Aldrich, Strem or Lancaster. The aminophosphines were stored under nitrogen at room temperature. NMR spectra were recorded on a Jeol Delta 270, Jeol Eclipse 300, Jeol Eclipse 400, Varian 400 or Lambda 300. Infrared spectroscopy was carried  suspension was stirred at room temperature for 16 h and the filtered to remove the isopropylammonium chloride. The solvent was removed under reduced pressure to give an oily residue which was dissolved in toluene (10 mL) to precipitate any remaining isopropylammonium chloride. This toluene solution was filtered to remove the methylammonium chloride and then the solvent was evaporated to dryness to give a colourless oil (0.57 g, 63%). 31

Preparation of PhobP(vNMe)PPhob (L′ c )
To the solution of L a (0.45 g, 2.6 mmol) in THF (3.0 mL), a 1.6 M solution of n BuLi (4.5 mL, 7.20 mmol) in hexane was added at −78°C over 5 min. The reaction mixture was stirred at −78°C for 40 min. PhobPCl (0.45 g, 2.6 mmol) in THF (2 mL) was added in portions to the cooled reaction mixture. The mixture was allowed to warm to room temperature and stirred for 2 h. The solvent was then removed under reduced pressure and the residue was triturated with diethyl ether (10 mL). The solid was filtered off and dried. satisfactory elemental analysis was not obtained and the product was used without further purification (0.52 g, 65%) 31

Preparation of [PtCl 3 (PhobP(vNHMe)PPhob)] (5)
A mixture of L′ c ·HCl (0.025 g, 0.070 mmol) and [PtCl 2 (NC t Bu) 2 ] (0.034 g, 0.070 mmol) were dissolved in CH 2 Cl 2 (5 mL) and stirred for 2 h to give a yellow solution. Warming this solution to 40°C led to the slow formation of yellow crystals of the product 5 suitable for X-ray crystallography. Satisfactory elemental analysis was not obtained and the crystals were insoluble in common organic solvents which precluded further characterisation by NMR spectroscopy.

Preparation of [PdCl 2 (PhobPNMePPhob)] (6)
To a suspension of L′ c (0.031 g, 0.099 mmol) in toluene (3 mL), [PdCl 2 (cod)] (0.031 g, 0.12 mmol) was added. The suspension was stirred at 50°C for 5 min. The clear reaction mixture was then cooled to room temperature and the resulting yellow precipitate was filtered off and washed with hexane (0.010 g, 20%). Crystals suitable for X-ray crystallography were grown from CDCl 3 although satisfactory elemental analysis was not obtained. 31

Oligomerisation catalysis
A rigorously cleaned autoclave was heated (130°C) under vacuum for 60 min, then cooled to reaction temperature and back-filled with Ar (1 bar). Solvent was then added via syringe.
The autoclave was pressurised with ethylene to 10 bar and vented. On a Schlenk line, a pre-activated catalyst solution was prepared by stirring the Cr source, ligand and modified methylaluminoxane (MMAO) together for 60 s, then transferred to the autoclave via syringe. The autoclave was pressurised and the pressure kept constant throughout the reaction by the continuous addition of ethylene, which was monitored via flow-meter. Once ethylene uptake had ceased or the autoclave was filled, the gas supply was closed and the reactor cooled to 5°C. The reactor was carefully vented. The reactor contents were treated with 1000 μL of nonane (GC internal standard) and 10% HCl (aq). A sample of the organic phase was taken for GC-FID analysis. Any solid formed was collected, washed repeatedly with EtOH, then acetone and dried overnight and weighed. GC-FID analysis was performed on an Agilent Technologies 6890N GC system equipped with PONA (50 m × 0.20 mm × 0.50 μm) and MDN-12 (60 m × 0.25 mm × 0.25 μm) columns. Catalysis was performed in a stainless steel 300 mL volume AE autoclave with Viton-ETP seals, equipped with a customised gas-entraining mechanical stirrer, internal cooling coil (tap water) and fluidised jacket (connected to an external thermostatic bath). Ethylene was passed through moisture and oxygen scrubbing columns prior to use and the flow measured using a Siemens Sitrans F C Massflo system (Mass 6000-Mass 2100) and the data logged.
Crystal structure determinations X-ray diffraction experiments for 1b, 3, 4, 5, 6 and L h ·HCl were carried out at 100 K and for 2b at 173 K on a Bruker APEX II diffractometer using Mo-K α radiation (λ = 0.71073 Å). 7 was collected at 120 K on a Bruker Nonius FR591 Rotating Anode using Mo-K α radiation (λ = 0.71073 Å) 25 and 8 was collected on EH1 of Station I19 of Diamond Light Source (λ = 0.71073 Å) at 120 K. 26 Data collections were performed using a CCD area detector from a single crystal mounted on a glass fibre. Inten- sities were integrated using SAINT with a multi-scan absorption correction preformed using SADABS. 27 All structures were solved using SHELXS and refined against all F o 2 using SHELXL and OLEX2. 28 All non-hydrogen atoms were refined anisotropically and hydrogen atoms were located geometrically and refined using a riding model. The structure of 4 was refined as a racemic twin and restraints were applied to the thermal displacement parameters to maintain sensible values. Crystal structure and refinement data are given in Table 4. The structures are shown in Fig. 1-8 with thermal ellipsoids drawn at the 50% probability level.