Process for the preparation of astaxanthin

The invention relates to an improved process for the preparation of the carotenoid astaxanthin by a double Wittig reaction of a 3-methyl-5-(2,6,6-trimethyl-3-oxo-4-hydroxy- l-cyclohexen-l-yl)-2,4-pentadienyltriphenylphosphonium salt(asta-Ci₅- triphenylphosphonium salt) with 2,7-dimethyl-2,4,6-octatriene-l , 8-dial (Ci_O-dial).

Astaxanthin is a natural colour which is very much in demand for colouring foods, salmon and trout. Accordingly, numerous methods for isolating or synthesising astaxanthin are known. Thus, for example, WO A 86/6082 discloses the isolation of astaxanthin by extraction of crustacean shells. Furthermore, astaxanthin can be obtained by fermentation processes, as described, for example, in EP A 329 754. However, these methods have significant disadvantages. Firstly astaxanthin is present in nature only in very low concentration and therefore has to be isolated by complicated processes. Secondly, only unsatisfactory yields are obtained which means that these known processes are of no interest from the economic point of view.

Among the processes for synthesising astaxanthin, the oxidation of canthaxanthin-bis-silyl- enol ethers with percarboxylic acids and subsequent hydrolysis (cf. EP A 101 597) and the oxidation of canthaxanthin enolates (EP A 440 037) may be mentioned. Disadvantages of these processes are the only moderate yields and purities of astaxanthin, incomplete conversions and undesired byproducts.

Finally, EP A 05 749 discloses a process for the preparation of astaxanthin by a Wittig reaction of a building block acylated on the hydroxyl group in the 4- position of the asta- Ci₅-triarylphosphonium salt with Cio-dial and subsequent hydrolysis. Inter alia, isopropanol is mentioned as a solvent for this Wittig reaction. The fact that protective groups have to be introduced into the Ci₅-triphenylphosphonium salts and eliminated again and that the yields obtained are moderate are disadvantages of this process.

The process according to the invention starts from a known Wittig process in which, in a preliminary stage, ketolylpentol is converted by partial hydrogenation into ketolyldienol, from which the asta-Cis-triphenylphosphonium salt also referred to below as astenyl salt subsequently forms as a result of bromination and phosphination with triphenylphosphine. In a continuous, quasicontinuous or batchwise subsequent stage (Figure 1) the astaxanthin is finally formed by a double Wittig reaction with the use of butenyl oxide as a base. In this case, for example, a mixture consisting of astenyl salt and Cio-dialdehyde in ethanol and 1 ,2-butylene oxide is reacted. After the reaction, the reaction solution is filtered. The still moist crude product is purified by solvent exchange (ethanol for methylene chloride). In this known process, the astaxanthin yield is 90.7%, based on Cio-dialdehyde, and 81%, based on astenyl salt.

The use of butylene oxide as a weak base has the advantage that the anion of the astenyl salt (for example bromide) is trapped and chemically bound. Thus, virtually no inorganic salts form. However, a disadvantage of the process is that butylene oxide is a relatively expensive and carcinogenic substance.

To date, all attempts successfully to replace the butylene oxide for economic and ecological reasons have failed. Thus, for example with the use of a stronger base_; in particular the formation of undesired byproducts presents most problems. There are substantially two reasons for the formation thereof. Firstly, the oxygen can have an oxidizing effect and secondly semiastacin or astacin may form as byproducts by rearrangement of unhydrogenated ketolylpentol. After the hydrogenation, minor amounts of ketolylpentol are in fact still present in the reaction solution. This can react in exactly the same way as ketolyldienol in the subsequent stages and form 9-trans-7,7-dehydroastenyl salt. This in turn can undergo the Wittig reaction to form two products, as shown in Figure 2. Finally, in the presence of a strong base, for example sodium methylate, these two compounds form astacin or semiastacin by rearrangement, as illustrated in Figure 3.

It is an object of the invention to improve the above mentioned Wittig process for the preparation of astaxanthin from an asta-Ci₅-triarylphosphonium salt and the Cio-dialdehyde in a Wittig reaction in such a way that, with the use of economical starting materials and at least identical operational safety, the formation of byproducts is substantially suppressed, the yield of astaxanthin, based on astenyl salt, is increased and hence the production costs are lowered.

It was surprisingly found that astaxanthin can be prepared in a technically simple manner and in very good yields if the butylene oxide is replaced by a more economical but stronger base during the Wittig reaction while maintaining specific process conditions.

In a particularly preferred embodiment of the process according to the invention, an alcohol mixed with a nonpolar solvent is additionally used as a solvent for the Wittig reaction.

Although EP A 0 733 619 has already disclosed an astaxanthin synthesis via a Wittig reaction in alcohol using a base, the process disclosed therein differs substantially from the parameters chosen according to the invention, such as, for example, from the profile for metering the base into the reaction solution and from the choice of the reaction temperature.

The invention therefore relates to a process for the preparation of astaxanthin of the formula I

by reacting 2 mol of the triphenylphosphonium salt of the general formula II

in which X represents chlorine, bromine or the (HSO₄)^" radical, preferably bromine, in a Wittig reaction with one mol of the Cio-dialdehyde of the formula III

which is characterized in that

a) the starting compounds of the formulae II and III are taken up in a solvent, the mixture is cooled to a temperature of not more than 10°C, preferably -18⁰C to +5°C,

b) about 0.9 to 1.5, preferably 0.9 to 1.2, mol of a base per mole of triphenylphosphonium salt are added to the resulting reaction mixture at a temperature of not more than 1O⁰C, preferably -18°C to +5°C,

c) the base is metered and mixed in over a predetermined reaction time T so that at least a

Vi base equivalent is added to the reaction mixture continuously or quasicontinuously within a timespan T' < %T and the remainder of the base within the remaining reaction time. For carrying out the reaction according to the invention in general alcohol and nonpolar solvents are used in an amount such that the concentration of the triphosphonium halides in the solvent mixture is 0.1 to 3 mol, preferably 0.8 to 1.5 mol, per litre of solvent mixture.

The solvent used is, for example, an alcohol, a mixture of different alcohols or a mixture consisting of an alcohol and a nonpolar solvent miscible with the alcohol. A particularly preferred solvent mixture consists of methanol and dichloromethane in a mass ratio of 1 : 1.2 to 1 :8, for example 1 : 1.48. In addition to methanol, it is also possible to use ethanol, 1-propanol, 2-propanol, n-butanol and acetone, tetrahydrofuran, dioxane or ethyl acetate as alcohols or polar solvents which can be used for the process according to the invention.

In addition to dichloromethane, it is also possible to use toluene, cyclohexane and hexane as nonpolar solvents which can be used for the process according to the invention

The base is introduced into the reaction mixture in general at temperatures of -18°C to +8°C, preferably at -18°C to 0°C, for example at -10°C to -5°C, according to the abovementioned metering profile specified below. The reaction mixture may be a homogenous solution or a suspension.

The reaction time T for the introduction of the bases and reaction are in general 0.5 to 30 hours, preferably 8 to 20 hours. In preferred working examples, a Vi to % base equivalent is added to the reaction mixture continuously or quasi continuously within a timespan T' < !4 T and the remainder of the base within the remaining reaction time.

The following may be mentioned as bases suitable for Wittig reactions: solutions of alkali metal or alkaline earth metal alcoholates in the corresponding alcohol, alkali metal or alkaline earth metal hydroxides or carbonates and butyl lithium, benzyltrimethylammonium hydroxide or methoxide and lithium amide. The process is particularly advantageous if a 20% to 40% solution of sodium methylate in methanol is used.

After complete addition of the base, it is advantageous to allow the reaction mixture to continue reacting for at least 15 min, preferably at least 30 min, in a subsequent reaction before the total mixture in the subsequent reaction is neutralized with an inorganic or organic acid, for example with acetic acid, glacial acetic acid or sulphuric acid.

The astaxanthin obtained according to the invention is generally thermally isomerized for conversion into the particularly desirable all-(E)-isomers. It is known that the thermal isomerization or the isomerization and purification can be readily carried out by heating in an alcohol. The process according to the invention is firstly characterized by the choice of the base, the method of metering and the composition of the solvent mixture. This novel procedure has two substantial advantages. Firstly, the novel process permits in particular the use of economical raw materials, and does so with at least identical operational safety and increased yield, which leads to an extremely economical production of astaxanthin, and secondly fewer byproducts, such as, for example, astacin and semiastacin, form in the process according to the invention.

A substantial obstacle which had to be overcome according to the invention in the choice of a stronger base than butylene oxide was the problem that the formation of astacin and semiastacin is promoted owing to the relatively strong base chosen, such as, for example, sodium methylate.

By using the novel, stronger base, semiastacin and astacin will be formed in a larger amount, but the amounts would be reduced to a minimum by the choice of the process conditions defined above, preferably to less than 2%, for example to less than 1 %.

The choice of the base metering profile according to the invention is to be explained in more detail below with reference to tests performed. Unless specially mentioned, all content data are stated as percentages by weight.

By monitoring the reaction, it was found that the formation of semiastacin begins from 0.75 equivalent of sodium methylate. The base can therefore be metered in relatively rapidly without particular disadvantages up to half an equivalent, but thereafter the metering of the base has to be reduced if it is intended to minimise the astacin/semiastacin formation.

Figure 4 in Table 1 shows five different base metering profiles over a reaction time T of 20 to 25 hours and at a preferred reaction temperature of -5°C. For the region of half an equivalent (10 ml) three different metering speeds were tested (profiles 2,4,5).

Tab. 1

For further optimization with respect to the choice of the amount of base to be used, profile 4 was chosen. Table 2 shows that, with a base excess of 1.5%, based on astenyl salt, the proportion of semiastacin can be reduced below the 1 % mark, and this can be achieved at a yield of at least 80%.

Tab. 2

Owing to the metering specification according to the invention, the speed of the base addition automatically decreases from the first to the second phase. On the laboratory scale, it is preferably 0.02 ml/min to 0.2 ml/min in the 1^st phase or 0.01 ml/min to 0.05 ml/min in the 2^nd phase and, on the production scale, 40 1/h to 500 1/h in the 1^st phase or 15 1/h to 25 1/h in the 2^nd phase.

From the optimization experiments carried out, the following preferred metering profile may be stated by way of example.

Tab. 3

At the end of the reaction, the reaction mixture is preferably neutralized with aqueous acetic acid (w=20%). In order to prevent any disadvantageous introduction of water into the reaction mixture, there is the possibility of using glacial acetic acid instead of aqueous acetic acid. It is also possible to use concentrated acetic acid.

The process according to the invention is explained in more detail below with reference to three examples.

Example 1 - laboratory batch

Batch table

At the beginning of the reaction, a 1000 ml double-jacket glass reactor was flushed for at least 10 min with nitrogen gas. Thereafter, 229.62 g of astenyl salt, 32.40 g of Cio- dialdehyde, 485.8 g (368 ml) of methylene chloride and 341.7 g (432 ml) of methanol were initially introduced into the reactor at 2O⁰C without flushing with N₂ gas and cooled to an internal temperature of -5°C in the course of about 30 min. During this procedure, stirring was carried out slowly and slight flushing with N₂ gas was effected. Thereafter, 71.78 g (74 ml) of sodium methylate (30% in methanol) were metered in at an internal reactor temperature of -5°C by means of a metering apparatus and with the following metering profile

1.) 0.15 ml/min : total 40 ml / 4.4 h (1^st phase) 2.) 0.04 ml/min : total 34 ml / 14.2 h (2^nd phase) after the addition of the base stirring was continued for a further 30 min at -5⁰C and neutralization was effected at constant internal temperature with 12.0 g (11.4 ml) of glacial acetic acid.

The neutralized reaction mixture was then heated in a reactor having a jacket temperature of 6O⁰C. At the boiling point (internal temperature about 45°/jacket temperature about 60⁰C) continuous solvent exchange was then effected over a timespan of 5 h, the reflux ratio being adjusted so that the level in the reactor was maintained. During the solvent exchange, 426.6 g (540 ml) of methanol were metered in continuously in the course of said time span and about 540 ml of solvent mixture were distilled off until an internal temperature of 65°C was reached.

After the solvent exchange, the reaction mixture was cooled at 25°C/h to an internal temperature of 20⁰C and stirred for at least a further 15 min and the suspension was filtered over a glass suction filter. The reactor was then washed first with the mother liquor and then with 158 g (200 ml) of methanol, and the wash solutions were filtered in succession over the crystals. Thereafter, the crystals were washed again with 2 x 158 g (200 ml) of methanol and dried at 55°C and < 60 mbar in a vacuum drying oven.

Yield: 95.05 g of astaxanthin (corresponding to 99.68% of astaxanthin, based on the ClO- dialdehyde used); according to HPLC, the all-E content was 80.9%.

Examples 2 and 3 - pilot batch and production batch

In general, the astaxanthin production can be divided up as follows, starting from the process according to the invention:

Wittiε reaction

Astenyl salt in a mixture of methanol and methylene chloride is reacted with Ci₀- dialdehyde ideally at -5°C throughout and under atmospheric pressure to give astaxanthin. The base used for the Wittig reaction is sodium methylate, the base being metered in in the course of 15 to 30 h, preferably 15 to 25 h, for example 20 h, on the production scale. Triphenylphosphine oxide (TPPO), sodium bromide and methanol are formed as byproducts.

Postreaction / neutralization

The postreaction takes place at -10°C to 5°C, preferably at -10° to -5°C, for example at -5°C, and takes from 30 min to 2 h, for example 1 h 40 min. Acidification with acetic acid is then effected.

Solvent exchange 1

After the transfer to a solvent exchanger and heating to the reflux temperature, a mixture comprising predominantly methylene chloride and methanol is distilled off while at the same time methanol is metered in so that the level in the reactor remains constant. The removal of distillate is complete at an internal temperature of > 64⁰C.

Crystallization / centrifugins

In a crystallizer for crude product, the distillate is then cooled to 20°C and the suspension is then centrifuged. The filter cake is washed with methanol. The amount of wash agent depends on the amount of TPPO present in the product.

Dissolution

Moist astaxanthin is finally made into a slurry with methylene chloride in a dissolution vessel and refluxed at an internal temperature of, for example, 40°C for 2 to 5 h.

Solvent exchange 2 / crystallization / centrifusins

The astaxanthin suspended in methylene chloride is heated to the reflux temperature, and the methylene chloride/methanol mixture is then distilled off. The distillation is stopped at an internal temperature of > 61°C, methanol being metered in up to this temperature. The suspension is then centrifuged. The filter cake is washed with methanol.

Moist astaxanthin is finally dried in a dryer at an internal temperature of 50-70°C and a final vacuum of < 20 mbar. The operations "dissolution", "solvent exchange 2", "crystallization" and "centrifuging" can alternatively also be omitted if, by a suitable choice of procedure, the residual content of byproducts already meets the requirements after the first centrifuging step.

I^s' experimental example

At the beginning of the experiment, a reactor vessel was evacuated to -0.8 bar and flushed with N₂ gas. 1151 kg of astenyl salt, 161 kg of Cio-dialdehyde and 9 kg of astaxanthin were then introduced into the reactor, hi order to eliminate influences of oxygen after the introduction, the vessel was evacuated again and flushed with nitrogen gas. Subsequently, 2457 kg of methylene chloride and 1700 kg of regenerated methanol (97% of methanol and 2% of methylene chloride) were added. It was then possible to begin the metering of sodium methylate (30% in methanol) at -5°C. In order to prevent the formation of too much semiastacin salt the metering was carried out as follows: 1^st phase metering: 44 1/h with a total time of 4 h 35 min (amount metered 194 kg) 2^nd phase metering: 11 1/h with a total time of 15 h 26 min (amount metered 165 kg) After the metering of sodium methylate, the reaction solution was stirred for a further 30 min before the neutralization of excess sodium methylate was begun. For this purpose, 60 kg of 100% glacial acetic acid were finally metered in.

The reaction solution thus formed was then transferred to a second reaction vessel and the subsequent solvent exchange was carried out so that the product formed was present in virtually pure methanol.

The astaxanthin suspension was then cooled down to an internal temperature of 20°C in the crystallizer. The crystallization time was about ¹A h. After the internal temperature had been reached, the suspension was stirred again for 15 min at 20°C. The crystal slurry could now be separated on a centrifuge and then dried in a dryer. The drying process was carried out with a constant jacket temperature of 60°C.

Yield: 510.1 kg of crystalline astaxanthin altogether. A yield of 85.6% could be achieved, based on the amount of astenyl salt used (1150 kg).

2^nd experimental example

2093 kg of astenyl salt, 292 kg Qo-dialdehyde and 9 kg of astaxanthin were introduced into a reaction vessel. In order to eliminate influences of oxygen after the introduction the vessel was evacuated to -0.8 bar and flushed with nitrogen gas. The metering of 1918 kg of methylene chloride and 1295 kg of methanol was then begun. The time in which the solvents were pumped in was about 45 min. After the reaction temperature of -5⁰C had been reached, the metering of sodium methylate (30% of methanol) was finally begun. In order to prevent the formation of too much semiastacin salt, the following metering profile was chosen: During the 1^st phase, the metering was on average about 450 1/h with a total time of 45 min. Thereafter, the metering was reduced stepwise over about 25 min from 450 1/h to 50 1/h (amount metered about 394 kg). The subsequent second phase was operated at 17 1/h and for a total time of 15 h 30 min (amount metered about 260 kg).

After the metering of sodium methylate the reaction solution was stirred again for 1 h 40 min and then neutralized with 105 kg of glacial acetic acid (100%). In this process, it was not necessary to maintain a metering ramp.

The reaction solution formed was then transferred to a further vessel in which the reaction solution was washed with 500 ml of methylene chloride. The subsequent solvent exchange and the drying process were carried out in each case analogously to the above mentioned first experimental example.

Yield: 946 kg of crystalline astaxanthin altogether. A yield of 87.4% could be achieved, based on the amount of astenyl salt used (2093 kg).