José A.
Fuentes
,
Piotr
Wawrzyniak
,
Geoffrey J.
Roff
,
Michael
Bühl
* and
Matthew L.
Clarke
*
School of Chemistry, University of St Andrews, St Andrews, Fife, KY16 9ST, UK. E-mail: mc28@st-andrews.ac.uk; Fax: +44 (0) 1334 463808
First published on 1st April 2011
In the course of studies on the tandem hydroformylation–reductive amination (hydroaminomethylation), fluorinated mono-phosphines were found to be more active than their more electron-donating counterparts in the enamine hydrogenation step of the reaction; this is in contrast to the widely held view that alkene hydrogenation activity increases with ligand donor strength. DFT calculations comparing the reaction pathways for a simple alkene and a representative enamine show that the rate-determining step changes from the first insertion into a Rh–H bond for but-2-ene to the final reductive elimination step from the Rh–hydride–alkyl species in the enamine hydrogenations.
We initially investigated Rh catalysed reactions of styrene, syngas and morpholine using ligands of contrasting electronic properties (Scheme 1). The donor properties of the ligands can be ordered as Cy3P > Bun3P > L2 > Ph3P > L3 > L1 by consideration of their carbonyl stretching frequencies in the complexes of type trans-[RhCl(CO)L2].10a Hydroaminomethylations are generally carried out above 100 °C, but lower temperatures were used in this study in order to analyse the reactions before complete conversion is reached.
Scheme 1 Products formed from a hydroaminomethylation of styrene and morpholine, and some of the ligands used in this study. |
A reaction time of 20 hours at 70 °C proved to be suitable conditions to differentiate the various catalysts; after 20 hours, the reactions were stopped and analysed using 1H NMR and GCMS (authentic samples of possible reaction products were prepared to confirm the identities of intermediates and products). The results are shown in Table 1.
Entrya | Ligand | Aldehyde (%) | Enamine (%) | Amine (L) | Amine (B) |
---|---|---|---|---|---|
a Conditions: 0.2% [RhCl(COD)]2 as pre-catalyst, 60 bar syngas (1∶1), toluene as solvent. Conversion of styrene >99% in all cases. Product composition determined using 1H NMR. Enamine and aldehydes yields are mixtures of isomers. b Temperature = 90 °C. c 92% isolated yield of amine as a 3.5∶1 mix of regio-isomers. d Time = 2 hours. e 0.2% [Rh(acac)(CO)2] used as a catalyst precursor. | |||||
1 | 0.4% L1b | 3 | 0 | 21 | 76c |
2 | 0.8% L1d | 13 | 71 | 0 | 16 |
3 | 0.8% L1 | 0 | 92e | 0 | 0 |
4 | 0.8% L1 | 0 | 0 | 8 | 92 |
5 | 0.8% L2 | 32 | 66 | 0 | 2 |
6 | 0.8% Ph3P | 21 | 72 | 0 | 7 |
7 | 0.8% Cy3P | 26 | 61 | 0 | 13 |
8 | 0.8% Bun3P | 23 | 71 | 0 | 6 |
9 | 0.4% L1 | <1 | 0 | 12 | 88 |
10 | 0.7% L1/0.1% L2 | 5 | 16 | 8 | 71 |
11 | 0.4% L1/0.4% L2 | 16 | 80 | 0 | 4 |
A number of useful conclusions can be made: under the conditions of hydroaminomethylation, even the poor hydroformylation catalysts derived from tris-(4-methoxyphenyl)phosphine, L2 completely convert styrene into aldehydes or subsequent products. The use of [Rh(acac)(CO)2] and (3,4,5-F3C6H2)3P, L1 results in essentially exclusive formation of the enamine, suggesting that the mono-hydrides formed from reaction of [Rh(acac)L] complexes with hydrogen and/or the resting states in hydroformylation [Rh(H)(PR3)n(CO)3−n] are not catalysts for enamine reduction. This also suggests that the use of [RhCl(COD)]2 as a pre-catalyst can form sufficient amounts of both type of catalysts necessary for hydroformylation and hydrogenation. Reduction of the enamines does not take place using the catalysts derived from more electron-donating ligands, PPh3, (4-MeO–C6H4)3P, (Bun)3P and Cy3P (entries 5–8). In addition, it is striking that using mixtures of (3,4,5-F3C6H2)3P and (4-MeO–C6H4)3P results in an inhibition of the enamine hydrogenation step of the reaction (Table 1, entries 10 and 11). A combination of (3,4,5-F3C6H2)3P, L1 and [Rh(COD)Cl]2 allows complete conversion to the desired amine with reasonable regioselectivity and high yield (entry 4). There are two main mechanistic scenarios for the observed reactivity: (i) ligand electronic effects on alkene hydrogenation using syngas as a hydrogen source are different to those using pure hydrogen, or (ii) the high activity of Rh complexes of electron rich phosphines normally employed in the hydrogenation of alkenes does not extend to enamine substrates that are rarely studied in alkene hydrogenation. This then provided us with impetus to investigate the rates of enamine hydrogenation as a function of the ligands.
A sample of 4-(2-phenylprop-1-enyl)morpholine was prepared by refluxing purified branched aldehyde, 2-phenyl-propanal with morpholine, and studied in hydrogenation. The results shown in Table 2 indicate that (3,4,5-F3C6H2)3P, L1 is a far more active catalyst for this reduction, regardless of whether the reactions are carried out using syngas or pure hydrogen.
The results described in Table 2 strongly support the suggestion that electron-deficient ligands are preferred in this class of enamine hydrogenation. In order to confirm this was not some special effect confined only to this substrate, we varied the secondary amine in hydroaminomethylation of styrene (pyrrolidine and N-methyl-N-benzylamine) using catalysts derived from (3,4,5-F3C6H2)3P and (4-MeO–C6H4)3P ligands. These results (Table 3) are again striking and in complete agreement with the findings in Tables 1 and 2. (3,4,5-F3C6H2)3P provides a much better catalyst than {(4-MeO)–C6H4}3P and in particular smoothly converts the enamine into the tertiary amine product, whereas the latter ligand stalls at the enamine stage. Reaction of cis-stilbene and morpholine was also carried out, although this gave poor conversion using {(4-MeO)–C6H4}3P and reduced chemo-selectivity using (3,4,5-F3C6H2)3P. None-the-less it is clear that, using the electron rich ligand, some enamine is present and there is no amine detected, whereas for (3,4,5-F3C6H2)3P, almost no enamine remains and a significant amount of amine has been produced, entirely consistent with the previous results. We would therefore suggest that electron-poor phosphines are the best-suited monodentate ligands for a significant number of enamine hydrogenations.
Entrya | Alkene | Amine | Ligand | T/°C | Conversionb % | Aldehyde % | Enamine d % | Amine % | Yieldc (%) |
---|---|---|---|---|---|---|---|---|---|
a Conditions: 0.2% [RhCl(COD)]2, 0.8% P ligand, toluene, 1 equiv. alkene, 1.1. equiv. secondary amine, initial pressure 60 bar CO/H2 (1∶1), reaction time 20 hours unless stated otherwise. b Conversion = alkene consumed against 2-methylnaphthalene internal standard using 1H NMR; the composition of the reaction mixture was calculated in the same way. <1 refers to barely detectable traces or undetectable amounts. c Yield of purified amine after column chromatography or acid/base extraction. d Mixture of E and Z isomers, and in the case of styrene branched and linear enamine. e b/l = 7∶1. f b∶l = 6.4∶1. g b∶l = 9∶1. h b/l = 7∶1. | |||||||||
1 | Styrene | Pyrrolidine | L1 | 70 | >99 | 1 | 15 | 59 | 53e |
2 | Styrene | Pyrrolidine | L2 | 70 | >99 | 6 | 82 | <1 | nd |
3 | Styrene | Bn(Me)NH | L1 | 90 | >99 | <1 | <1 | 95 | 87f |
4 | Styrene | Bn(Me)NH | L1 | 70 | >99 | 3 | 3 | 70 | 49g |
5 | Styrene | Bn(Me)NH | L2 | 70 | >99 | 31 | 65h | 2 | nd |
6 | cis-Stilbene | Morpholine | L1 | 80 | 72 | <1 | 3 | 40 | 38 |
7 | cis-Stilbene | Morpholine | L2 | 80 | 9 | <1 | 8 | <1 | nd |
These ligand electronic effects are quite contrasting with the accepted theory for alkene hydrogenation and are not easily reconciled with the early rate-determining step expected in such reactions. To gain further insights into mechanistic details of the hydrogenation step, some exploratory calculations were performed at a suitable level of density functional theory (DFT).9 The reaction between the widely studied and accepted active catalyst, Rh(PPh3)2Cl(H)2, and the (E)-4-(2-phenylprop-1-enyl)morpholine substrate was considered. To model the key intermediate resulting from the first Rh–H addition across the double bond a number of possible isomers was screened, in which Rh was attached either at the α or β position (with respect to the nitrogen atom in the morpholine group) and stabilised by secondary interactions with neighbouring groups. The most stable structure turned out to be an α-adduct with an additional Rh–N contact of 2.23 Å (1b in Fig. 1) with all other isomers at least 50 kJ mol−1 higher in energy.8b
Fig. 1 Salient stationary points involved in the Rh-catalysed olefin hydrogenation; generic scheme with a Rh–dihydride fragment (top), optimised structures involving (phenylpropenyl)morpholine (middle) and 2-butene (bottom) with the Rh(PPh3)2Cl(H)2 model catalyst, together with relative energies [kJ mol−1] computed at the B97-D/ECP2 level of DFT. Peripheral H atoms and phenyl groups shown as wireframe for clarity; C, H, N, O, P and Cl atoms are rendered in black, white, blue, red, orange and green, respectively. |
The transition state for the reductive elimination of the saturated product from 1b was located at 111 kJ mol−1 (TS1b, Fig. 1). This is a surprisingly high barrier, given that it is usually the first H-transfer step, rather than this second one, which is believed to be rate-limiting in Rh-catalysed hydrogenation reactions.8a A transition state for the first H-transfer could be located (TS1a), which, however, connects to the separated reactants, i.e. the olefin and a coordinatively unsaturated Rh(PPh3)2Cl(H)2 fragment (not shown). Starting from these reactants, the resting state before this first H-transfer will be stabilised, e.g. through coordination of solvent or the olefin. The only such olefin complex that could be located is 1a (Fig. 1). Taking this isomer as representative for the intermediate leading to 1b, a barrier of ca. 52 kJ mol−1 can be estimated for the first H-addition step. Thus, even allowing for this being an informed estimation, the second H transfer (1b → TS1b) clearly has a much larger energy barrier and is indeed indicated to be rate-limiting with this substrate.
When the computations were repeated for 2-butene, a model substrate without the N functionality, the expected, “normal” profile was obtained (lower part in Fig. 1). That is, the first H-transfer from 2aviaTS2a has a higher barrier (ca. 75 kJ mol−1) than the second one from 2b to TS2b (56 kJ mol−1). These results therefore indicate that the mechanistic details of homogenous olefin hydrogenation can change notably with an enamine as substrate. Key to this mechanistic change is the stabilisation of the organometallic intermediate 1b through an interaction with the lone pair of the adjacent nitrogen atom to form a Rh–C–N ring. This intermediate is lowered so much in energy that the barrier for the subsequent reductive elimination can become rate-limiting. Stable Rh–alkyl–hydrides have been detected in enamide hydrogenation before,10 but the reductive elimination in these cases has been a lower energy process relative to their formation.8b It is well established that electron-withdrawing ligands accelerate reductive elimination processes,11 and this therefore can be concluded to be the origin of the ligand effects we have observed. Quantitatively, our computed numbers may be associated with a large uncertainty because only a very limited region of the vast conformational space could be explored. However, the final step proposed to be rate-determining is calculated as approximately 60 kJ mol−1 greater in energy than the first hydride insertion. Thus we would estimate with some confidence that any subtle differences in phosphine conformation are extremely unlikely to stabilise any transition states to an extent that could alter the rate-determining steps from those calculated here, giving considerable certainty on the qualitative conclusions.
These two main mechanistic scenarios share a dependence on catalyst, substrate and hydrogen to deliver the species that undergoes rate-determining reactions. However, it was felt useful to collect several other pieces of data regarding the kinetic behaviour of these reactions, and these are archived in the ESI.† Measurements of turnover frequency under essentially identical conditions for L1 and L2 give initial average TOF of 515 mol mol−1 h−1 and 4 mol mol−1 h−1 for L1 and L2, respectively, and increased rates in going from 10 bar to 50 bar pressure of hydrogen. The reaction is slightly unusual in being promoted further by increasing phosphine/Rh ratios. The comparative studies described in the previous pages are done using low ratios, even more pronounced differences would be seen if using a significant excess of phosphine. This can be reconciled by either, (i) strong enamine binding and/or improved kinetics for phosphine loss from a saturated Rh species, or (ii) weak non-competitive product inhibition. In any case, these effects do not alter the likelihood that the rate-determining step changes when one moves from a simple alkene to an enamine.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c1cy00026h |
This journal is © The Royal Society of Chemistry 2011 |