Mixed Metal Catalysts for the Synthesis of Azelaic Acid
Mixed Metal Catalysts for the Synthesis of Azelaic Acid
Activity of a Palladium-Iron Complex
Azelaic acid is an important intermediate in the production of a speciality Nylon, and it is shown that a palladium-iron system may be used very selectively in its preparation. This work was described at the recent Chemical Congress at York by the author, who is now Manager of the Catalyst Department in the Johnson Matthey Research Laboratories.
There are numerous examples reported in the literature showing clear evidence that mixtures of metal oxides, or indeed of the metals themselves, produce increased catalytic activity by comparison with the individual metal oxides or metals respectively.
Bond and Webster (1), using mixed metal catalysts derived from mixed oxides using a modification of the Adams method, demonstrated that a mixed ruthenium-platinum catalyst is more effective in the reduction of organic nitrocompounds than either of the pure metals. (See Figure.) The mixed oxides are reduced by hydrogen under ambient conditions to yield what are thought to be finely divided alloys of the two metals. It was shown by Rylander (2) that catalysts made by fusing platinum and palladium, rhodium, iridium, or ruthenium salts with sodium nitrate varied widely in hydrogenation activity and showed marked synergistic effects. Platinum-rhodium catalysts were especially active for reduction of nitrobenzene, nitropropane, cyclohexanone oxime and xylenes, whereas platinum-palladium catalysts were most active for hydrogenation of butynediol.
On the industrial manufacturing scene there are examples of both heterogeneous and homogeneous mixed metal systems. A mixture of bismuth and molybdenum oxides is used as a catalyst for the partial oxidation of propylene to acrolein. The best results are obtained when the Bi:Mo ratio is 1:1. In the Haber ammonia synthesis the most widely used catalyst is α-iron containing an oxide such as molybdenum for widening the lattice and enlarging the active interface of the catalyst with the substrates. In biological systems both molybdenum and iron are necessary in the nitrogen fixation process.
Perhaps the most spectacular use of two metals in a heterogeneous catalyst is in the Ziegler-Natta titanium-aluminium systems for polymerising α-olefins. The important species involved is thought to be either a chlorobridged dinuclear molecule or a species having direct Ti-Al bonds. Some soluble Zieglers have been reported to be stereo-specific under certain conditions. Stereo-specific polymerisation is helped by the catalyst surface in the heterogeneous variety but the orientation of the monomer can also be controlled during the critical propagation step in the homogeneous systems provided that the monomer has at least two sites capable of co-ordinating to the catalyst (e.g. CH2=CHCN or CH3CH=CH2 at low temperatures). Butadiene has also been polymerised stereospecifically in homogeneous media. Wilke has shown that certain π-allylic nickel compounds such as π-allyl nickel chloride are active homogeneous catalysts for dimerising olefins, and it has now been shown that the rate of dimerisation can be greatly accelerated if aluminium alkyl is added to the nickel catalyst. The active species is probably a halogen bridged nickelaluminium compound.
Another homogeneous reaction carried out successfully in industry using two metals simultaneously is the Hoeschst-Wacker oxidation of ethylene to acetaldehyde in the presence of Pd(II). The Pd(O) thus produced can be reoxidised to Pd(II) using CuCl2, and the Cu(I) thus produced reoxidised in turn with oxygen. The presence of both palladium and copper simultaneously makes the reaction catalytic with respect to metal and it is quite possible that a discrete copper-palladium species is present in solution. A similar situation may be present in the catalyst system necessary to convert ethylene to vinyl acetate.
There is therefore considerable evidence that both in heterogeneous and in homogeneous systems and both in the laboratory and on the industrial scale mixed metal effects can be useful. We therefore investigated ways of synthesising mixed metal complexes.
Synthesis and Activity of a Palladium-Iron Catalyst
One of the most interesting compounds synthesised was a palladium-iron tetranuclear complex formed by reaction between π-allylpalladium chloride and diphenylphosphine-irontetracarbonyl in toluene (3):
This palladium-iron compound is a good selective hydrogenation catalyst (175°C/100 atm/1 h) for the hydrogenation of hex-1-yne in the presence of hex-1-ene. Under test conditions (10 p.p.m. moles metal/mole substrate), 93 per-cent of a sample of hex-1-yne in benzene was reduced to hexene and only 3 per cent to hexane. This is unexpected since palladium is normally an excellent catalyst for the hydrogenation of olefins. The effect probably results from easier formation of a palladium-acetylene coordinate bond than a palladium-olefin bond. The rate of hydrogenation of the olefin on its own is in fact faster than the acetylene on its own. The effect is analogous to those found with supported palladium catalysts (5 per cent Pd-BaSO4) and described as molecular queueing (4).
This Pd-Fe complex is also a selective isomerisation catalyst and in its presence oct-1-ene is isomerised to oct-2-ene but not to the 3- and 4-isomers. Most palladium complexes promote the isomerisation of Δ1-octene to a mixture of Δ1, Δ2, Δ3, and Δ4 octenes (5) and the reason for the selectivity in this case may be steric in that Δ1-octene may be more accessible to the palladium than the Δ2-octene.
Extension of this work into the field of carbonylation was prompted by the appearance of a review article by Bittler et al. (6), who described carbonylation of unsaturated hydrocarbons to carboxylic acids and esters under “mild” conditions (e.g. 60°C, 300−700 atm for 1,5-cyclooctadiene) using PdCl2 (PPh3)2 as catalyst. The use of the PPh3 ligands had considerably increased the yield of ester per mole of palladium compared with the use of non-liganded PdCl2 (7).
The carbonylation of oct-1-ene was studied under mild conditions using both the palladium-iron complex and PdCl2(PPh3)2 and the results of the two series of experiments compared:
At 75°C, 50 atm, the total yield of esters was ten times greater for the mixed metal catalyst than for PdCl2(PPh3)2. The threshold temperature for the reaction was ca. 30°C lower in the palladium-iron case than in the palladium. The isomer distribution of products is, however, ca. 1:1 in both cases, with a tendency for the iso to predominate.
Carbonylation of 1, 5–Cyclo-octadiene
Since the 50:50 n :iso ratio obtained with linear olefins had comparatively little industrial interest we next turned our attention to 1,5-cyclooctadiene (1,5-COD):
If the reaction could be stopped selectively at the first stage (I), it was known that this ester could be converted into azelaic acid in per cent overall yield by means of two straightforward hydrolysis steps:
Azelaic acid is of interest since it is an intermediate needed in the manufacture of the commercially important nylon-6,9. For the carbonylation step, a series of experiments using PdCl2(PPh3)2 and p -toluene sulphonic acid showed that the optimum temperature for producing the monoester (I) was ca. 110°C (8). At this temperature no diester is formed (not detectable by g.l.c. at high sensitivity). Lower temperatures gave lower rates according to the normal rate laws. Higher temperatures also gave lower rates, presumably due to catalyst decomposition. The optimum pressure seems to be ca. 100 atm. Higher pressures give little change in rate. Lower pressures reduce the rate and increase the amount of isomerisation of unchanged 1,5-COD to 1,3-COD. The pressure and temperatures are, however, interrelated as far as ideal conditions are concerned.
Optimisation of the reaction at 110°C/100 atm showed that yields of up to 7000 moles ester/g.atom Pd could be obtained using anhydrous ethanol (rather than MeOH), PPh3 ligand (rather than PBu3) and a Lewis acid such as FeCl3, TiCl4 or NbCl5. Highest yields were obtained when the catalyst was preformed rather than prepared in situ from PdCl2 and PPh3 and when a common chloride ion was used. Under these conditions, no 1,5-COD isomerised to 1,3-COD–unchanged starting material could therefore be recycled. The Pd-Fe complex was also an effective catalyst but not as good as PdCl2(PPh3)2/FeCl3, which could be giving [PdCl(PPh3)2]+[FeCl4]−.
Hydrolysis of Ethylcyclooct–4–ene–1–carboxylate
The atmospheric pressure, aqueous hydrolysis of the cyclic ester carbonylation product obtained from 1,5-COD and ethanol to give cyclic acid (step 2), was found to be complete after 0.5–1.0 h in refluxing solution. The mixture had become homogeneous after this time and isolation of the product gave yields of ca. 96 per cent. With the ethyl ester, as the temperature is raised to 90 to 95°C a strongly exothermic reaction sets in and the mixture becomes homogeneous during a few minutes.
It is acceptable to use “crude” ester product from step (1) in this stage of the reaction and then remove the ethanol and 1,5-COD at the end of this stage. The removal of the ethanol before stage (3) is essential. Most of the palladium is then precipitated during this stage (a very small quantity of palladium is also found on the walls of the glass liner in the carbonylation stage) and is removed by filtration. The black solid obtained dissolves readily in nitric acid, and would presumably dissolve in aqua regia to give H2PdCl6. Evaporation of the aqua regia would give PdCl2 which could be readily converted to the catalyst with PPh3 (via Na2PdCl4).
Conversion of CycIooct–4–ene–1–carboxylic acid to Azelaic Acid
The caustic cleavage step was performed in closed Inconel autoclaves under a self-generated pressure of up to 1000 p.s.i.g. water, and approximately 250 p.s.i.g. hydrogen. The use of ca. 320°C (internal temperature) for 1 hour was found to give higher yields of purer product than 300°C for 10 hours. Since the reaction is complete after 1 hour it is preferable to discontinue heating at this point to prevent discolorisation and condensation of the product. The shorter heating period may therefore account for the greater purity of this “crude” azelaic acid compared with samples obtained after longer heating periods. Two comparative g.l.c. traces for the esterified products from our work and the currently available acid derived from the ozonolysis of castor oil showed a single product in our case, whereas the commercial product contained 8 per cent of impurities. We therefore have an excellent, potentially commercial route to azelaic acid from 1,5-cyclooctadiene, the three stages of which proceed in almost theoretical organic yield to give a pure “crude” product.
Most of the practical work was performed by Mr. R. Jackson and I would like to express my thanks to him, and to Dr. D. E. H. Jones for discussions on aluminium chemistry.
G. C. Bond and D. E. Webster, Platinum Metals Rev. 1969, 13, (2), 57
P. N. Rylander et al., Engelhard Ind. Tech. Bull., 1967, 8, 25
B. C. Benson,, R. Jackson,, K. K. Joshi and D.T. Thompson, Chem. Commun., 1968, 1506
L. Crombie and P. A. Jenkins, Chem. Commun., 1969, 394
J. F. Harrod and A. J. Chalk, J. Am. Chem. Soc., 1964, 86, 1776
K. Bittler et al., Angew. Chem. Internat. Ed. En., 1968, 7, 329
J. Tsuji et al., Bull. Chem. Soc. Japan., 1966, 39, 141
R. Jackson and D. T. Thompson, British Patent 1,340,873, 1973