Water Soluble Rhodium Catalysts
Water Soluble Rhodium Catalysts
A Hydroformylation System for the Manufacture of Aldehydes for the Fine Chemicals Market
The rhodium catalysed hydroformylation reaction is one of the most widely used industrial applications of homogenous catalysis to employ platinum group metals. Potential limitations in the application of this technology to molecules which are heat sensitive or have high boiling points are the stability of the catalyst and the ability to separate the catalyst from the products. A means of circumventing these limitations is described. This involves locating the catalyst in an aqueous liquid phase, and it enables viable reaction rates to be achieved at moderate temperatures and pressures.
Hydroformylation is the reaction of an alkene, hydrogen and carbon monoxide to generate an aldehyde, as shown in Figure 1. While the reaction is energetically favourable it only proceeds in the presence of a catalyst. A number of different metal complexes have been investigated as potential catalysts, but the only ones with sufficient activity to be useful industrially are those of cobalt and rhodium. The relative activities of these metals are given in Table I.
|Chromium, Molybdenum, Tungsten, Nickel||0|
The hydroformylation reaction was first reported by Roelen in 1938 during a study of Fischer Tropsch catalysts (1). Later it was realised that the active catalyst was a hydridocarbonylcobalt species, and in the period 1945 to 1951 a number of processes employing homogeneous cobalt catalysts were developed for the production of detergent and plasticiser alcohols.
It was demonstrated during the 1950s that rhodium compounds such as rhodium oxide were highly active as hydroformylation catalysts, although the ratio of normal:iso aldehydes was lower than those obtained with commercial cobalt systems. However in the 1960s Wilkinson’s group at Imperial College, London, and Pruett’s group at Union Carbide found, independently, that rhodium compounds containing organophosphines could convert alkenes to aldehydes at mild temperatures and pressures, and with high selectivity to linear aldehydes (2,3). An extention of this work resulted in the commercial exploitation of the rhodium Low Pressure Oxo (LPO) process by Union Carbide, Davy McKee and Johnson Matthey. The principal industrial application of this process is the hydroformylation of propylene to butyraldeyde. This
primary product can be converted to the plasticiser alcohol 2-ethylhexanol, via aldol condensation/hydrogenation, or to butanol. Since the LPO process was introduced commercially in the 1970s it has been immensely successful. It has been used for the production of over one million tonnes of butyraldehyde, and is licensed to 12 companies in 9 countries, Table II.
|Union Carbide||Texas, U.S.A.|
|Nan Ya Plastics||Taiwan|
In addition to the development being timely with respect to feedstock costs, the commercial success of the process can be attributed to a number of technical features:
The high activity of rhodium complexes.
The presence of a large excess of triphenylphosphine contributes to the high selectivity to aldehydes, in particular the desired product n-butyraldehyde, and it also suppresses hydrogenation.
In addition, the presence of triphenylphosphine renders the catalyst thermally stable and confers longevity on the catalyst, while the low volatility of the catalyst means that the product can be stripped from the reactor with a minimal rhodium loss, of less than 1ppm.
An efficient reactant purification system eliminates catalyst poisons and extends the catalyst lifetime.
Overall the process provides a graphic example of how a relatively expensive platinum group metal catalyst can successfully compete with a significantly cheaper cobalt catalyst.
On a commercial scale, the rhodium and cobalt catalyst systems are to some extent complementary in that rhodium catalysts are only suitable for the hydroformylation of low molecular weight alkenes, whereas the cobalt systems can be applied to the hydroformylation of high molecular weight feedstocks and are particularly attractive where alcohols are the desired products, Table III.
|Cobalt||Modified Cobalt||Rhodium Low Pressure|
|Normal : branched aldehyde ratio||2−4||7||>10|
|n-product yield, per cent||67||67*||90|
|Aldehyde yield, per cent||90||80*||98|
|Alkanes, per cent||1||15||0.9|
|Other by-products, per cent||9||5||1.0|
Aldehyde and Alcohol
Limitations to the rhodium process result from the availability of methods of separating the catalyst from the product. In the production of butyraldehyde the product is distilled out of the reactor at low temperatures which do not cause catalyst degradation. For higher molecular weight products higher temperatures are required, and at temperatures greater than 120°C severe catalyst degradation can occur, Table IV. Thus if rhodium catalysts are to be used commercially for the production of compounds of interest to the detergent and aroma industry, an alternative strategy is required.
|Catalyst Solution Degradation|
|Phosphido bridged rhodium species|
|Alkyl diphenylphosphine complexes|
An approach which has attracted a great deal of attention from research workers investigating hydroformylation and related reactions is the use of chemically anchored catalysts. After some preliminary work in this area Johnson Matthey decided against this approach, for a variety of reasons, including:
[a] The difficulty of producing a catalyst with reproducible activity and selectivity, which implied that catalyst costs would be extremely high.
[b] Many of the heterogenised catalysts were more air sensitive than their homogeneous analogues.
[c] Metal was slowly lost from the catalyst, and in many cases the metal could not be recovered in a cost effective manner.
[d] Neighbouring group effects such as chelation, arising from a high surface concentration of organophosphines, caused slower rates of reaction.
We were attracted to the idea of utilising a water soluble catalyst system, with a view to locating the catalyst in the aqueous phase and the feedstock and product in the organic phase. We concentrated on rhodium catalyst systems, with a few less successful diversions into ruthenium and platinum systems. The standard which we wished to match in terms of rate and selectivity was the triphenylphosphine-modified rhodium homogeneous system, and we chose dodec-1-ene as a convenient model substrate.
Our first task was to devise a water-soluble phosphine ligand which located the rhodium in the aqueous phase, yet was similar to triphenylphosphine. The sulphonated triphenylphosphine-rhodium systems which had been investigated by Rhône Poulenc and by Wilkinson gave rise to extremely slow rates of reaction at moderate pressures, and we selected carboxylated triphenylphosphines as ligands. We observed measurable rates at 80°C and about 6 atmospheres hydrogen-carbon monoxide with a ten-fold molar excess of carboxylated phosphine over rhodium.
At the conception of the project we had appreciated that we would need to use some “special” effect to enhance the rate in the two liquids system relative to the homogeneous system. For the system containing two liquid phases, mass transfer control is likely to dominate, which contrasts with the homogeneous hydroformylation system where kinetic control is operative.
At this time there was a burgeoning interest in the use of phase transfer agents in organic synthesis, and there were many reports that surfactant systems would induce micelle formation and promote liquid-liquid transport. Accordingly we set out to investigate the effect of a range of these amphiphilic reagents on the two liquid phase hydroformylation reaction, see Figure 2 and Table V. For those reagents with a high affinity for organic solvents, high rates of reaction were obtained, as measured by conversion to aldehydes and by hydrogen/carbon monoxide uptake. However, the reaction effectively took place in the organic phase, with high concentrations of rhodium observed in the organic. For quaternary ammonium salts (Quat.) of the type RN+Me3X− (R=C10-C16 alkyl, X=halide, acetate etc), under basic conditions reasonable reaction rates were obtained, as shown in Table VI. These rates were about half that of the homogeneous rhodium system at 80−100°C and about 6 atmospheres hydrogen-carbon monoxide (4).
|Surfactant/phase transfer agent||Selectivity n:i||Total aldehydes per cent||Colour of organic phase||Colour of aqueous phase|
|Aliquat 336||2||78||deep purple||colourless|
|Sodium dodecyl sulphate||4||80||yellow||brown|
Conditions: [Rh]aq = 300 ppm, 4-Ph2PC6H4COOH, P:Rh=10:1, C12H25NMe3Br:Rh = 20:1 in pH10 buffer at 80°C, 5.5 atms 1:1 H2:CO
In the more favourable cases rhodium loss to the organic phase was lower than ippm. Dependent on the R group, rapid phase separation could be achieved. Alternatively, a fairly stable emulsion could be obtained. Using a particular reactor conformation it was possible to achieve extremely high selectivities to normal aldehydes, with normal:iso ratios in the region of 300.
Rhodium loss to the organic phase, and selectivity both to total aldehydes and normal aldehydes were highly dependent on the organic phase composition, the reactor system and the rate of stirring. The selectivity to total aldehydes was dependent on the organic phase composition, which can be explained in terms of the relative solubility of carbon monoxide. High selectivity to normal aldehydes was promoted when relatively poor mixing of the liquid phases occurred, and poor mixing also minimised rhodium loss.
The fact that the mixing of the two liquid phases plays such an important part in the course of the reaction can be interpreted in terms of the formation of metastable aggregates. The pH of the aqueous phase is basic under reaction conditions, and in these circumstances the carboxylated phosphine is in an anionic form. It is possible to imagine an organised array containing catalyst, surfactant and substrate molecules (Figure 3). Under these highly organised conditions the approach of the polar alkene group will be favoured sterically for a 1-alkene, and likewise the formation of a linear alkyl and acyl will be favoured over the branched species for the two phase system relative to the homogeneous system. In this way it is possible to rationalise the extremely high normal:iso ratios that are obtained under poor mixing conditions, when the formation of highly organised assemblies is favoured.
Some work was carried out in elucidating the reaction mechanism. Through varying the hydrogen to carbon monoxide ratio it was found that at low partial pressures of carbon monoxide the normal:iso ratio is extremely high, with moderate selectivity to total aldehydes; increasing this partial pressure results in a decrease in normal:iso, but an increase in selectivity to total aldehydes. This indicates that it is not the initial attack of hydride to alkene which determines the ratio of linear to branched aldehydes, but that there is a pathway for the interconversion of iso-alkyl and normal-acyl compounds where the relative proportions of each are determinable by the partial pressure of carbon monoxide. The accepted hydrofor-mylation pathway is given in Figure 4.
In this mechanism there is a route for the interconversion of branched alkyl (a) and linear acyl (f) species, which is consistent with the observations made above. Thus by modifying the microstructure of the liquid assembly, through the addition of surface active agents and the polarity of the organic solvent, it is possible to effect high selectivities to total aldehydes.
Other features which make this system attractive are that high normal:iso ratios can be achieved at relatively low phosphine concentrations (less than 1/10 of that of the conventional triphenylphosphine-rhodium system). Minimal losses of phosphine, surfactant and rhodium occur, and the phosphine and surfactant are not degraded under the reaction conditions.
A related water soluble system has been developed by Rhone Poulenc and Ruhr Chemie (5). This system utilises sulphonated triphenylphosphine as a water solubilising ligand and operates at higher temperatures and pressures than the truly homogeneous rhodium-phosphine. A comparison of the sulphonated and carboxylated phosphine systems is given in Table VII.
The trisulphonated triphenylphosphinerhodium catalyst is operated commercially by Ruhr Chemie for the hydroformylation of propylene and the complexity of these reaction systems suggests that the design of an industrial plant for the hydroformylation of high molecular weight alkenes will not be trivial. However, the mild operating conditions which result from the use of the 4-diphenylphosphinobenzoate-rhodium catalyst, together with the high selectivities attained, mean that this system has considerable potential for the manufacture of aldehydes in the fine organic chemicals markets.
O. Roelen ., German Patent 849, 548 ; 1938
D. Evans, ., J. A Osborn . and G. Wilkinson ., J.Chem.Soc.A, 1968, (12), 3133
R. L Pruett . and J. A Smith ., J. Org. Chem., 1969, 34, (2), 327
M. J. H Russell . and B. A Murrer ., British Patent 2, 085, 874B
E. G Kuntz ., Chemtech, 1987, 17, (9), 570