Pascal M.
Castro‡
a,
Henrik
Gulyás
ab,
Jordi
Benet-Buchholz
a,
Carles
Bo
ab,
Zoraida
Freixa§
a and
Piet W. N. M.
van Leeuwen
*a
aInstitute of Chemical Research of Catalonia (ICIQ), Av. Països Catalans 16, 43007 Tarragona, Spain. E-mail: pvanleeuwen@iciq.es; Fax: +34 977-920-221
bDepartament de Química Física i Inorgànica, Universitat Rovira i Virgili, Campus Sescelades, 43007 Tarragona, Spain
First published on 1st March 2011
The self-assembly of Secondary Phosphine Oxides (SPOs) into anionic bidentate chelates was used to construct unique systems for metal catalyzed transfer hydrogenation of ketones in isopropanol. Chelating bidentate or tridentate ligands were formed by assembly of secondary phosphine oxides through hydrogen bonding in the presence of rhodium trichloride as demonstrated by means of NMR spectroscopy and X-ray diffraction. When a chiral version of an SPO was used in asymmetric transfer hydrogenation of isopropanol and acetophenone, an enantiomeric excess of 89% was achieved. The presence of at least two ligands in the catalytically active species was confirmed by a positive non-linear effect. DFT calculations were applied to characterize several intermediates for the isopropanol dehydrogenation to produce a rhodium hydride complex and acetone. A transition state for the hydrogen-transfer was fully characterized, which revealed that the process occurs via a concerted outer-sphere mechanism.
![]() | ||
Scheme 1 SPOs tautomeric equilibrium, chelate formation, and heterolytic cleavage of H2. |
NMR spectra were recorded by using a Teflon capped or J. Young NMR tube with a Bruker Avance 400 Ultrashield NMR spectrometer or a Bruker Avance 500 Ultrashield NMR spectrometer for the variable temperature experiments. Electrospray ionization mass spectra (ESI-MS) were recorded with a Waters LCT Premier spectrometer in dry MeOH or MeCN. Gas chromatography analyses were carried out on an Agilent Technologies 6890N/G1530N spectrometer equipped with a FID detector and a Supelco Beta Dex 120 fused silica capillary chiral column (30 m × 0.25 mm diameter × 0.25 μm film thickness).
![]() | ||
Fig. 1 Ligands 1–3. |
In an exploratory screening intended to find an appropriate metal precursor, various metal salts were tested using diphenyl phosphine oxide (1) as the ligand, cyclohexanone and acetophenone as the substrates and isopropanol as the hydrogen donor. In general cyclohexanone was hydrogenated faster than acetophenone, many metal SPO complexes showed activity, but rhodium(III) or rhodium(I) gave the highest activities (see ESI†). The best combination involved RhCl3·3H2O (0.5 mM) for both substrates, using a metal to ligand ratio of 1∶
10 for cyclohexanone (92% conversion TOF = 1825 h−1) and 1
∶
3 for acetophenone (18% conversion, TOF = 353 h−1).
The nature of the species formed from the reaction of RhCl3·3H2O and 1 in a 1∶
3 ratio during the incubation process was first established by 31P{1H} NMR analysis in CDCl3, which revealed two doublets at 87 ppm and 82 ppm in a 3
∶
2 ratio (JRh–P = 124 Hz and 119 Hz) which correlates well with the data found in the literature for μ-Cl3 bridged binuclear rhodium complexes bearing anionic Ph2POH⋯−OPPh2 chelating ligands.4c At 213 K the signal at 87 ppm decoalesced to a doublet of triplets at 84 ppm and a double doublet at 89 ppm (Fig. 2). Its structure was assigned to 4 (Scheme 2) on the basis of the 2
∶
1
∶
2 ratio and multiplicity of the three signals.
![]() | ||
Fig. 2 Simulated (left) and experimental (right) variable temperature 31P{1H} NMR of complex 4 (CDCl3, 201 MHz) for the signal at 87 ppm. Best-fit rate constants k are shown in the simulated spectra. |
![]() | ||
Scheme 2 Formation of 4 by reaction of 1 with RhCl3·3H2O and generation of rhodium hydride species 5. |
The dynamic process is due to proton transfer and hydrogen bond exchange between the three SPOs4c (ΔH≠ = 56.9 kJ mol−1, ΔS≠ = 53.1 J mol−1 K−1, ΔG≠298 = 41.2 kJ mol−1). An electrospray mass spectrometry analysis (ESI-MS) in methanol corroborated the presence of 4. In addition a signal corresponding to [(1)4Rh2Cl5–2H] was observed (previously identified by X-ray analysis as the AsPh4+ salt).4c
The structure of 4 was elucidated unambiguously by single crystal X-ray structure analysis (Fig. 3).14 The central chloro-bridged core of the structure of 4 is similar to two symmetric structures reported in previous communications.4a,c,d The neutral complex 4 forms three intramolecular hydrogen bonds that do not interact with neighboring molecules. The two phosphinous acid protons located at the side of the Rh(III) atom linked to three phosphinous acids (P1–3) could be clearly located as Fourier peaks in the difference map. Additionally oxygen atoms O1 and O3 show larger P–O bond distances than oxygen O2, corroborating the expected positions where the protons were localized. Both hydrogen atoms are linked to the O2 atom via hydrogen bonds (distances: O1⋯O2: 2.57 Å [H1⋯O2: 1.90 Å uncorrected] and O3⋯O2: 2.49 Å [H3⋯O2: 1.65 Å uncorrected]), the three phosphorus atoms P1–3 thus forming a supramolecular analog to conventional tridentate anionic ligands.15
![]() | ||
Fig. 3 Ortep-plot (thermal ellipsoids shown at 50% probability level) of 4. Non-relevant hydrogen atoms have been omitted for clarity. Selected distances (Å) Exp./[Calc.]: Rh1–P1: 2.3025(4)/[2.261]; Rh1–P2: 2.3124(5)/[2.275]; Rh1–P3: 2.2999(5)/[2.260]; Rh1–Cl1: 2.4918(4)/[2.495]; Rh1–Cl2: 2.4355(4)/[2.457]; Rh1–Cl3: 2.4855(4)/[2.474]; P1–O1: 1.5903(13)/[1.611]; P2–O2: 1.5572(13)/[1.569]; P3–O3: 1.5828(13)/[1.610]; Rh2–P4: 2.2795(5)/[2.240]; Rh2–P5: 2.2703(5)/[2.257]; Rh2–Cl1: 2.5270(5)/[2.626]; Rh2–Cl2: 2.4788(4)/[2.557]; Rh2–Cl3: 2.3650(4)/[2.443]; Rh2–Cl4: 2.3121(4)/[2.334]; P4–O4: 1.5507(16)/[1.580]; P5–O5: 1.5454(13)/[1.546]. |
The proton corresponding to the moiety formed by the two phosphinous acids (P4 and P5) could not be easily localized, but its location was determined successfully by DFT full-geometry optimization of a model of 4, in which the phenyl substituents were replaced by hydrogen atoms. The geometry of the model indicates that this H atom is bonded to O4 and strongly hydrogen-bonded to O5 according to the computed distances (O4⋯O5: 2.457 Å [H4⋯O5: 1.356 Å]), the difference in the two H–O distances being 0.2 Å only. The strength of the other two hydrogen bonds is intermediate (O1⋯O2: 2.615 Å [H1⋯O2: 1.598 Å]; O3⋯O2: 2.620 Å [H3⋯O2: 1.607 Å]), in agreement with the X-ray structure. The trends in the Rh–P and Rh–Cl bond lengths are perfectly reproduced in the computed structure, in particular the Rh2–Cl1 bond length, which is the longest rhodium-chloride bond since this chlorine holds two anionic PO moieties in trans positions (see caption of Fig. 3).
The effect of the added base was investigated by adding 5 equivalents of potassium tert-butoxide to an isopropanol solution of 4. The solution was concentrated and the CD2Cl21H NMR spectrum of the resulting beige precipitate displayed two pseudo quadruplets around −20 ppm (JRh–H ≈ JP–H ≈ 25 Hz) which on account of NMR experiments were tentatively assigned to isomeric forms of a rhodium hydride species 5 featuring two SPOs on the rhodium nucleus bearing the hydride (Scheme 2 and ESI†).
An enantioselective version of the catalytic system was obtained using (R)-binaphthol-derived phosphorous acid 2, although it was realized that 2 might not survive the strong basic, alcoholic medium. Relatively low activities were obtained with the use of 2 (entries 2–5, Table 1), but enantiomeric excesses up to 89% were reached (entry 4, Table 1). When the catalyst concentration was lowered (entries 2, 4 vs. 3, 5 in Table 1) the ee dropped.
Run | Ligand | [Rh]/mM | Time/min | S/Rhb | Conv. (%) | TOFc | eed (%) |
---|---|---|---|---|---|---|---|
a Conditions: 1 hour incubation at 80 °C, reaction at 40 °C, [Rh]/[tBuOK] = 1![]() ![]() ![]() ![]() |
|||||||
1 | 1 | 3.3 | 240 | 1000 | 10 | 26 | <5 |
2 | 2 | 6.7 | 240 | 200 | 44 | 23 | 84 |
3 | 2 | 0.67 | 240 | 200 | 7 | 4 | 30 |
4 | 2 | 3.3 | 240 | 1000 | 23 | 53 | 89 |
5 | 2 | 0.67 | 240 | 1000 | 2 | 6 | 38 |
6 | 3 | 6.7 | 6 | 200 | 68 | 1364 | 8 |
7 | 3 | 6.7 | 240 | 200 | 79 | 42 | <5 |
8 | 3 | 0.67 | 6 | 200 | 19 | 376 | 14 |
9 | 3 | 0.67 | 30 | 200 | 97 | 387 | 13 |
10 | 3 | 3.3 | 6 | 1000 | 15 | 1490 | 10 |
11 | 3 | 3.3 | 240 | 1000 | 47 | 110 | 5 |
12 | 3 | 0.67 | 6 | 1000 | 10 | 971 | 26 |
13 | 3 | 0.67 | 240 | 1000 | 53 | 122 | 19 |
Indeed, 2 decomposes within 1 hour at 80 °C in the reaction medium as was shown by NMR and ESI-MS methods, which accounts for the low productivity. A test for a non-linear effect was undertaken by using 2 of varying degrees of enantiopurity in the catalytic reaction.16 The result was a positive non-linear effect, firmly establishing the presence of at least two chiral ligands at the active rhodium center.
To circumvent ligand decomposition, we synthesized 3, a “carbon analog” of 2 based on the dinaphthophosphepine moiety.17,18 Enantiopure R-3 (R-4,5-dihydro-3H-dinaphtho[2,1-c:1′,2′-e]phosphepine-4-oxide) was prepared by acidic hydrolysis of the corresponding diethylamino-phosphepine. Unlike 2, in situ formed complexes of 3 provided relatively high activities, with initial TOFs reaching 1490 h−1 (entry 10, Table 1) and almost quantitative conversions (entry 9, Table 1). Unfortunately, low enantiomeric excesses were obtained.
The nature of the precatalysts formed by reaction of 3 with RhCl3·3H2O during the incubation process was inspected in a similar manner as for 1. The ESI-MS analysis displayed the presence of complexes 6 (isostructural with 4) and 7 (Scheme 3).
![]() | ||
Scheme 3 Formation of 6 and 7 by reaction of 3 with RhCl3·3H2O. |
As mentioned in the introduction, it occurred to us that SPO complex catalyzed hydrogen transfer may involve a mechanism similar to that in Noyori and Shvo catalysts, in which an amido or an alkoxy anion participates in the hydrogen transfer.9,11 Therefore, we investigated theoretically the mechanism of isopropanol dehydrogenation to give acetone and considered two possible mechanisms: hydrogen transfer and β-H elimination. Starting from 4′, the PH2 analog of 4, we studied several ways of how isopropanol could approach the complex, and proposed first chloride/isopropanol exchange leading to species A, as shown in Scheme 4.
![]() | ||
Scheme 4 Proposed mechanism for the dehydrogenation of isopropanol. |
Evaluating the energy associated with this first step is not straightforward since isopropanol plays both the role of ligand and solvent, and the stability of a chloride ligand largely depends on solvation effects. By using a standard DFT GGA-based approach, the continuum solvent model method COSMO, and neglecting entropic effects, we evaluated an upper energetic limit of 20 kcal mol−1 needed to reach species A. From this point, hydrogen-transfer to the Rh–PO fragment turned out to be facile, and a transition state (TS) structure could be fully characterized (Scheme 4, Fig. 4). Note that, in the TS, although the hydroxy proton is even more strongly hydrogen-bonded to P
O, the proton has not yet been transferred. On the contrary, the C–H bond is elongated (from 1.102 to 1.479 Å) and the new Rh–hydride bond has almost formed. Recent DFT studies for the Shvo catalyst also confirmed that the hydride-proton transfer does occur in a concerted outer-sphere manner.19
![]() | ||
Fig. 4 Molecular structures of A, B and C intermediates, and the transition state TS. Selected interatomic distances in Å. Only the active part of the complex is shown. |
The reaction proceeds towards the formation of species B in which the ketone remains hydrogen-bonded. The final step, decoordination of the ketone, leads to species C, in which the protonated oxygen maintains the chelate through the intramolecular, now weakened, hydrogen bond. Note the evolution of the chelate hydrogen-bond from 1.577 in A to 1.901 Å in C. Deprotonation of C leads to the hydride complex 5 described above.
Fig. 5 plots the computed energy for each species along the reaction coordinate. The energy barrier for the hydrogen-transfer step, from A to B, is rather low and insignificant in free energy. Formation of the hydride complex appears to be strongly exothermic. In this case, entropic factors favour the separation of B in its two constituent parts, i.e. the acetone and the hydride complex, which lie 15 kcal mol−1 below the reactants. Activation barriers of the magnitude calculated are well in accord with the rate found at molar concentrations of the substrates. Thus, complex A is an efficient catalyst for the hydrogen-transfer process following the heterolytic mechanism.
![]() | ||
Fig. 5 Relative energy profiles (black: electronic energy, ΔE; red: free energy, ΔG, in kcal mol−1) for the dehydrogenation of isopropanol. |
Complex 5 (PH2 analog) could also be obtained from A by deprotonation followed by β-H elimination in an isopropoxide complex. This reaction pathway, which involves neutral complexes, was investigated as well and several species were characterized as shown in Fig. 6. Thus, starting from the isopropoxy complex A2, the reaction proceeds to B2 by opening one chloride bridge while the isopropoxide coordination position changes from axial to equatorial, while establishing a β-H agostic interaction (Rh–β-H = 2.019). In this rearrangement one of the SPO units of the chelate ligand moves from an equatorial to an axial position. Species C2 looks very much like B2, but the agostic interaction is not present, and the SPO proton is now bonded to an SPO unit in the equatorial position. Next, in the transition state TS2 the C–H bond breaks and the new hydride–rhodium bond is formed. Note also that the former isopropoxide moiety has been transformed into acetone, which remains coordinated as in D2. The final step corresponds to decoordination of acetone and skeletal rearrangement to form again the triple chloride bridging species complex 5. It was not possible to locate transition states for opening and closing the chloride bridge. The relative energy profile in Fig. 7 shows that, although the process is exothermic, high energy barriers for both the skeletal arrangement and the β-H elimination steps are needed, the last barrier being much higher than the energy required to perform direct hydrogen transfer. According to these results, the β-H elimination reaction pathway seems less favoured than the direct hydrogen transfer assisted by the SPO ligand.
![]() | ||
Fig. 6 Molecular structures of A2, B2, C2, D2 and 5, and the transition state TS2. Selected interatomic distances in Å. Only the active part of the complex is shown in the left side structures. |
![]() | ||
Fig. 7 Relative electronic energy profile (kcal mol−1) for the β-H elimination of an isopropoxide species. |
Footnotes |
† Electronic supplementary information (ESI) available. CCDC reference number 685489. For crystallographic data in CIF or other electronic format see DOI: 10.1039/c0cy00022a |
‡ Present address: Borealis Polymers Oy, Innovation Centre Porvoo, P.O. Box 330, FI-06101 Porvoo, Finland. |
§ Present address: Ikerbasque Research Professor, Departamento de Quimica Aplicada Facultad de Ciencias Quimicas Apdo. 1072, 20080 San Sebastian, Spain. |
This journal is © The Royal Society of Chemistry 2011 |