The present invention pertains to a process for the continuous production of polyesters, in particular aliphatic polyesters, from cyclic ester monomers.

Aliphatic polyesters based on cyclic ester monomers such as lactide (L-lactide, D-lactide, rac-lactide, also referred to as DL-lactide, meso-lactide) , glycolide, trimethylene carbonate (TMC) , epsilon-caprolactone, and p- dioxanone, and combinations thereof have many attractive properties. They often have nigh biocompat ibility and attractive resorbability properties, which makes them suitable for the preparation of scaffolds and implants for use in human or animal bodies, such as for example for fixation elements, films, membranes, suture thread or also for pharmaceutical drug delivery systems. Further, in particular, polylactide, also referred to as polylactic acid, is a promising material in the field of biobased polymers for, e.g., packaging material. The fact that it can be derived from renewable resources makes it particularly attractive as a sustainable alternative for polymers derived from oil.

Polymerisation processes for preparing aliphatic polyesters such as polyiacti.de are known in the art. They include ring-opening polymerisation processes and polycondensation processes. It has been found that the polylactic acid obtained through polycondensation processes is of limited value because polycondensation does not yield the required high molecular weight polyesters. Accordingly, (co) polyesters are preferably prepared by ring-opening polymerisation of the corresponding cyclic monomers, such as L-lactide, D-lactide, DL~-lactide or rac- lactide, meso-lactide, glycolide, trimethylene carbonate,, epsilon-caprolactone, and p-dioxanone, or mixtures thereof .

Therefore, most publications on processes for the production of polylactide disclose a first step wherein lactic acid is polymerised to form a pre-polymer through condensation, which pre-polymer is subsequently depolymerised by means of a catalyst to form crude lactide (i.e. the ring-closure reaction). The crude lactide is purified, and the purified lactide is used as monomer in the preparation of polylactide by ring-opening polymerisation. For the purpose of this description the terms polylactide and polylactic acid are used interchangeably.

Although the literature on the production of polyesters such as polylactide is abundant, most publications are silent on the specific equipment to be used on industrial scale. They mainly focus on laboratory scale. In most publications, the preparation of iactide from lactic acid and the subsequent purification of the lactide are described in detail, while for the ring-opening polymerisation of cyclic ester monomer (e.g. lactide} to form the corresponding polyester, e.g., polylactide, only temperature and catalyst are described. For instance, WO 94/06856 only describes that the polymerisation process of lactide to polylactide can be conducted in a batch, semi- continuous, or continuous manner. The reaction vessel may be equipped with a conventional heat exchanger and/or mixinσ device. Polymerisation processes for manufacturing polyesters can be divided into two groups, viz. polymerisation in the presence of a water-free solvent, e.g., suspension or emulsion polymerisation, and polymerisation in the substantial absence of solvent, e.g., melt polymerisation, carried out at a temperature above the melting temperature of the monomer and polymer, or mass polymerisation, carried out - in batch - at a temperature below the melting temperature of the polymer.

In general, polyesters are all made in the absence of solvent - bulk - by polycondensation and in some special cases, when the cyclic ester monomer is already dehydrated, by ring-opening polymerisation. Commonly used processes are performed batch-wise and the conversion is followed by monitoring increase of melt viscosity and reduction of carboxylic acid end group concentration. This classic approach is used for many polyesters, from rosin- based printing ink resins, via powder coating polyester resins to prepolymers for PET yarn. The latter are subjected after melt-polymerisation to solid-state post- polymerisation (SSP) in order to increase the average molecular weight to values that are not achievable in the molten state. Although SSP is a time-consuming process, it is inevitably applied broadly on industrial-scale.

A major problem which is often encountered in the polymerisation in the absence of solvent of cyclic ester monomers to polyesters, e.g., the polymerisation of iactide to polylactide, is the removal of the heat generated during the exothermic polymerisation reaction. Polyesters such as polylactide have a relatively low thermal conductivity. For example, the thermal conductivity of polylactic acid is 0.13 W/(m.K). For other polymers and rubbers, values of the same order have been reported. This means that the heat generated by the reaction cannot always effectively be removed, especially in large vessels, stirred tanks, and the like. This can lead to local overheating of the resulting polymer, causing chain degradation and discolouration of the polymer . In conventional processes the process settings are chosen such that reaction rates are low and residence times are long. This leads to voluminous and expensive equipment. Other conventional processes do not provide high quality, i.e. high molecular weight, polyester, e.g., polyldctide, with a low yellowness index with low residual lactide content in high yields.

A further problem which is encountered in the solvent-free ring-opening polymerisation of cyclic ester monomers is the difference in melt viscosity between the monomer melt and the polymer melt. In WO 99/50345 it is suggested to use a plug flow reactor or a series of plug flow reactors for the polymerisation process. We have found that when using a plug flow reactor, for example for lactide polymerisation, the difference in melt viscosity between the molten cyclic ester monomer (in this example the lactide) , and the resulting polyester (in this example the polylactide) is so high that plug flow conditions cannot be maintained and channelling occurs.

Since no appropriate solutions to the problems mentioned above have been disclosed, there remains a need for a continuous process for manufacturing high molecular weight polymers in high quality from cyclic ester monomers in an economically attractive manner on industrial scale. Tn accordance with the present invention, this neeα is addressed by the provision of a continuous process for the ring-opening polymerisation of cyclic ester monomers at a temperature between 100-?4Q°C which comprises the steps of: a) continuously providing cyclic ester monomer and polymerisation catalyst to a continuous mixing reactor, the reactor being operated at conditions effective for polymer.isati on to form a pre-polymensed reaction mixture, b) continuously removing pr e-polymcrised reaciion mixture from the continuous mixing reactor and continuously providing pre-polymerised reaction mixture Lo a plug flow reactor, the plug flow reactor being operated under polymerisation conditions, wherein trie reaction mixture is polymerised to a degree of polymerisation of at .least

90%, to form polymer c) continuously removing polymer from the plug flow reactor.

The number average molar mass (Mn) of the i nal produc j n gram/mole is generally at least 10 000 q/mo me :e m particular at least 30 000 g/mol, still more in particular at least 50 000 g/moi . The upper limit of he moiar mass is net critical to the process according to the invention. Generally it is below 500 000 g/moi, more specifically below 300 000 g/moi.

The ring-opening polymerisation is generally carried out at a temperature of at least 100 ⁰ C, in particular at. least 150 ⁰ C. The polymerisation temperature is generally at nost 240 ⁰ C, in particular at most PTO ⁰ C, still more in particular at most 200 ⁰ C. The temperature in bcth readers may be the same or different, Tt should in both cases be sufficiently high for the medium present in the reactor to be in the liquid phase.

The present invention pertains to polymerisation carried out in the substantial absence of solvent, namely to melt- polymerisation. If so desired minor amounts of solvent may be present in the process, e.g. added as a solvent for the catalyst or further reaction components. The process is intended to encompass situations where the reaction mixture contains less than 5 wt . % of solvent for the polymer, in particular less than 2 wt.%, more in particular less than 1 wt.%, still more in particular less Chan 0 , 5 wt . % .

Pressure is not critical to the reaction. The pressure applied will be appropriate to ensure flow of the liquid reactor components through the reactor, Tt is within the scope of the skilled person to determine suitable reaction pressure .

Both the pre-polymerisation in the continuous mixing reactor and the further polymerisation in the plug flow reactor are preferably performed in inert conditions, such as under a dry nitrogen or argon blanket.

It was found that with the process according to the invention residence times of less than three hours, more in particular less than two hours, still more in particular less than one hour, at temperatures below 200 ⁰ C with conventional amounts of catalyst sufficed to obtain a degree of conversion of at least 90%, in particular εit least 93%, more in particular at least 97%, still more in particular to within 1% of the chemical equilibrium conversion. This is significantly faster than the polymerisation rime of conventional processes where TO hours and more are mentioned. Reference is made, for example, to "Two step process for continuous polymerisation of polylactic acid" N. Okamoto, T, Matαso, b Journal of Chemical Engineering of Japan, Vol. 41, No. 6, pp. 474-484, 2008.

The residence time is calculated from the entry of the monomer in the continuous mixing reactor to the removal of me polymer from the plug flow reactor.

10

Tn the first step of the process according to the invention cyclic ester monomer and polymerisation catalyst are continuously provided to a continuous mixing reactor. Suitable continuous mixing reactors include continuous

15 εtirreα tank reactors and loop reactors, both of which εre known in the art: .

The use of a loop reactor may sometimes be preferred. Loop reactors are conventionally used for preparation of polyolefins such as polypropylene and other addition

/0 polyrrers like polystyrene, polymethyl methacrylate, or styrene acrylonitrile copolymers. Tn one embodiment, a loop reactor is used which comprises static mixing elements. This will ensure good homogeneity of the reaction mixture comprising polymerised products in

?5 combination with optimal removal of reaction heat. In the loop reactor, the added mononer and catalyst arc intimately mixed with partially polymerised product that is already present m the reactor. One beneficial result of this is tnat the rheological behaviour of the reaction

30 mixture on.!y changes gradually within the loop reactor.

This helps controlling the fluid flow within the reactor. Major jumps in viscosity over short distances are avoided and the heat released by the reaction is distributed uniformly in the reactor. Due to the increased flow rate and the mixing elements that are preferably present, the rate of heat removal from the reactor is significantly enhanced, further helping in the control of reaction conditions. More in particular, the combination of high flow rate and mixing elements results in enhanced temperature homogeneity, and thus a more even temperature distribution in the loop reactor. It also results in a narrow residence time distribution. Hot spots are avoided so there is less discoloration of the polymer. The intimate mixing in the loop reactor is also responsible for the reduced residence time necessary for this polymerisation reaction. Therefore this reactor design leads to a significant process intensification as compared to other techniques.

In the continuous mixing reactor, the reaction mixture is pre-polymerised. The degree of polymerisation in the continuous mixing reactor is generally at least 5 wt.%, more in particular at least 10 wt.%. The degree of polymerisation may be as high as 40 wt.%, or even 50 wt.%. The degree of polymerisation is generally below 90 wt.%, more in particular at most 85 wt.%. The degree of polymerisation aimed for in the continuous mixing reactor will, int. al . , depend on the viscosity of the reaction mixture.

Pre-polymerised reaction mixture is continuously withdrawn from the continuous mixing reactor and continuously provided to a plug flow reactor, where it is polymerised further to a degree of polymerisation of at least 90%. In the pϊucj flow reactor, which is preferably equipped with static mixing elements, the polymerisation can be completed up to high conversion rates. The static mixing elements used here provide for further intense mixing and homogeneous temperature distribution. DUP to this, the molecular weight distribution, degree of conversion, and residence Lime distribution can be tightly controlled. Further nore, the temperature profile of the reaction. 5 along the plug flow reactor can be controlled to a high degree, enabling optimisation of the polymerisation process .

In one embodiment, the loop reactor and/or the plug flow 10 reactor are static mixer reactors. That is, reactors equipped with static mixing elements. Suitable static mixing elements are described in US 4,314,606.

The reactants provided to the continuous mixing reactor 15 include cyclic ester monomer and polymerisation catalyst. If so desired, additional components such as co-catalyst, initiator for molecular weight control and/or additives may also be adαed. The components can be added to the reactor directly, either per se or in a solvent, or (some 20 of) the reactants may be combined prior to addition to the continuous mixing reactor,

Suitable cyclic ester monomers are monomers encompassing 5-7 coval ently linked carbon atoms in the ring, at least

?.b one, generally one or two, oxygen atoms in the ring, ana a carbonyl oxygen atom substituted onto the carbon atom adjacent to the oxygen atom in the ring (generating together the ester linkage). Where there are more than one oxygen atoms in the ring, oxygen atoms may be substituted

30 on the adjacent carbon atoms for all oxygen atoms, or just for one. The carbon atoms in the ring may be substituted with C1-C4 alkyl groups. Suitable monomers include lactide, glycolide, trimethylenc carbonate, epsilon- caprolactone, p-dioxanone and mixtures thereof. Where more than one type of monoirer is used, a polyester copolymer will be obtained. Where m the present specification mention :s made ol polymer, copolymer is also encompassed, unless specifically excluded, Lactide is the preferred cyclic ester monomer to be used as starting material, whether or not in combination with one or more of glycoliαe, trimethylene carbonate, and epsilon-caprolactone . The lactide used in the present invention nay be L-lactide (derived from two L-lactic acid molecules), D-lactidc (derived fron two D-lactic aciα molecules) , mesolaotide {derived from an L-lactic acid molecule and a D-lactic acid molecule), or a mixture of two or more of the above, A 50/bO mixture of L-lactiαe and D-lactide with a melting point of about 126 ⁰ C is often referred to in the literature as D, L- lactide or rac- 1 act ide .

Tn one embodiment of the present invention, the lactide used as starting material .is L-lactide with αp TO 50% of other lactides. For example, the lactide used as starting material may be L-lactide containing between b0 and 10% D- lactide. In another embodiment, the iactidc is substantially pure L-lactide, where the wording substantially pare means that it may contain up to 10 wt . % of other iactides, for example, up to 5 wt.%, or up to 1 wt , % , Tn a fur ^t her embodiment, the lactide is substantially pure D-lact ide, where the wording substantially pure means that it may contain up to 10 wt.% of oLner lactides, for example, up to b wt.%, or up to 1 wt . % . In a further embodiment, D, L lactide and/or D lactiαe are polymerised with L lactide to form polymers with an L/D monomer ration of at least 1:1. In a preferred embodiment high quality lactide is useα. PLi gh quality lactide Is defined here as lactide wii-h a low free acid content ar.d a low water content: the free acid content should be lower than 50 m:. lli-equivalents per kg (meq/'kg) , preferably lower than 20 ineq/kg and most preferably lower than 1C meq/kg. The water content should be lower than 2000 ppm, preferably lower than 500 ppm, more preferably lower than 200 ppm and even more preferably lower than 100 ppm. The use of high quality lactide ensures the economical production of polylactide with a low yellowness index with a high polymerisation rate and with good average molecular weight control. Depending on the embodiment, the monomers may be provided to the continuous mixing reactor m solid or liquid form. Where the continuous nixinq reactor is a loop reactor, the monomers should be provided ±r. liquid form. Where the continuous mixing reactor is a continuous εtirred tank reactor, providing the monomers in the solid phase may be considered. Wnore the monomers are to be provided to the continuous mixing reactor in the liquid state, they can be molten,, for example, in a melt tank or extruder, as in known in the art.

In addition to the cyclic ester monomer, polymerisation catalyst is added. Suitable catalysts for use in the invention are known m the art. In one embodiment of the present invention, lactide is polymerised with a catalyst of the formula

(M) (Xl, X2 . . - Xm) n

where M is selected from metals of Group 1, 2, 4, 8, 9, 10, 12, 13, 14, and 15 of the Periodic Table of Elements, (Xl, X2 . . . Xm) are independently selected from the group of alkyls, aryls, oxides, carboxylates, halides, alkoxides, alkylesters, in is an integer ranging from 1 to 6, and n is an integer ranging from 1 to 6, wherein the values for m and n depend on the oxidation state of the metal ion.

Within Group 1, the use of Na is preferred. Within Group 2 the use of Ca is preferred. Within Group 4 the use of Ti is preferred. Within Group 8 the use of Fe is preferred. Within Group 12 the use of Zn is preferred. Within Group 13 the use of Al, Ga, In, and Tl may be mentioned. Within Group 14 the use of Sn and Pb is preferred. Within Group 15 the use of Sb and Bi is preferred. In general, the use of metals of Groups 4, 14 and 15 is preferred. It is preferred for M to be selected from Sn, Pb, Sb, Bi, and Ti. The use of a Sn-based catalyst may be particularly preferred for cyclic ester monomers.

For halides, tin halides like SnC12, Sn3r2, SnC14, and SnBr4 may be mentioned. For oxides, SnO and PbO may be mentioned. Within the group of alkylesters, octoates (=2- ethyl hexanoates) , stearates, and acetates may be mentioned, e.g., in the form of Sn-octoate, (also known as Sn(II) bis 2-ethyl hexanoate) , Sn-stearate, dibutyltin diacetate, butyltin tris (2-ethyl hexanoate) , Sb (2-ethyl hexanoate) , Bi (2-ethyl hexanoate) , Sb triacetate, Na (2- ethyl hexanoate) , Ca stearate, Mg stearate, and Zn stearate .

Other suitable compounds include tetraphenyltin, Sb tris (ethylene glycoxide) , aluminum alkoxides, and zinc alkoxides.

The catalyst concentration is generally at least 5 ppm, calculated as metal weight, more in particular at least 10 ppm. Where larger than conventional amounts are used, the catalyst, concentration is at least 30 ppm, more in particular at least 50 ppm. The catalyst concentration is generally at most 200 ppm, in particular at most 150 ppm.

The use of Sn (Ii) -bis (2-ethylhexanoate) , also indicated as tin octoate, may be preferred, since this material is commercially available, liquid at room temperature and reaction temperature, and soluble in liquid lactide. Further, the compound has received FDA approval for many conventional end-uses.

If so desired, co-catalyst may be added to the lactide and the catalyst, that is, a compound that further increases the polymerisation rate. Suitable co-catalysts are known in the art. Reference is made, for example, to US

6,166,169. This reference describes co-catalysts of the formula

(Y) (Rl, R2 . ♦ . Rq)p

where Y is an element selected from group 15 or 16 of the periodic system, (Rl, R2 . . . Rq) is a substituent, selected from one of the compound classes of alkyls, aryls, oxides, halogenides, oxyalkyls, aminoalkyls, thioalkyls, phenoxides, aminoaryls, thioaryls, q is a whole number ranging from 1 to 6, and p is a whole number ranging from 1 to 6. The co-catalyst is generally used in an amount of the same order as the amount of catalyst, e.g., in a molar ratio of catalyst to co-catalyst of 50:1 to 1:50. US 6,166,169 indicates that the co-catalyst preferably includes a phosphorous compound, in particular P(Rl, R2, R3)3 with Rl, R2, R3 being independently selected from aryl and alkyl groups. P (Ph) 3 is indicated to be particularly preferred. Selection of an appropriate co-catalyst, initiator and optional additives such as anti-oxidants, phosphates, epoxidised vegetable oil, plasticisers etcetera, is within the scope of the person skilled in the art.

In step b) of the process according to the invention the polymerisation reaction is carried out further until a conversion of at least 90% is obtained, calculated on the starting lactide. More in particular, the polymerisation may be carried out until a conversion of at least 93% is obtained, more in particular at least 9 ^" ?%, still more m particular to within 1% of the chemical equilibrium conversion. Complete conversion of lactide is not possible for thermodynamic reasons. Maximum conversion will depend on reaction temperature and residence tirr.e. Lower residual monomer content, corresponding to higher conversion, will be obtained at lower polymerisation temperatures. The obtained conversion will generally be at most 98%. Conversion is to be determined directly after polymerisation.

The second part of rhe polymerisation can suitably take place m a plug flow reactor because the difference in viscosity of the pre-polymerised reaction product entering the plug flow reactor from the loop reactor and the polymer being removed from the plug flow reactor is small enough to avoid channelling. Tn addition, pronounced radial mixing will ensure a maximum driving force for the polymerisation reaction also where the conversion has advanced to cause a reduced concentration of reactants. As mentioned above, short residence tines can be reached with this two-step polymerisation.

In one embodiment, the plug flow reactor is placed vertically or tilted. This is preferred to decrease the chances of channelling or other type of flow maldistribution occurring .

The flow ratio of the continuous mixing reactor and the plug flow reactor is generally set between 1,5 and 50, more preferably between 1,5 and 15, most preferably between 1,5 and 12.

The polymer formed is continuously removed from the plug flow reactor.

In one embodiment the material thus obtained is subjected to a devolatilisation step to reduce the monomer content of the polymer. Prior to or simultaneously with the devolatilisation step the polymer may be subjected to a stabilisation step. Said stabilisation step comprises treatment of the polymer as obtained from the plug flow reactor with compounds that increase the stability of the compound against depolymerisation, discolouring and degradation in general. Examples of suitable compounds for stabilisation are organic peroxides, anti-oxidants such as phosphite-containing compounds, multi-functional carboxylic acids, hindered phenolic compounds, catalyst deactivating agents such as hindered alkyl, aryl and phenolic hydrazides, amides of aliphatic and aromatic mono- and dicarboxylic acids, cyclic amides, hydra∑ones and bishydrazones of aliphatic and aromatic aldehydes, hydrazides of aliphatic and aromatic mono- and dicarboxylic acids, bis-acylated hydrazine derivatives, heterocyclic compounds, endcapping with acetic anhydride, and mixtures thereof. The polymer is treated with the stabilisation compounds by admixing the stabilising compound with the polymer, e.g., at a temperature of the same order as the polymerisation temperature. This can be done by means of a static mixer, an extruder, or any other conventional way of mixing materials of which at least one is highly viscous.

If catalysts are αsoci which are based on metal ions such as Sn(IT), Pb(II), Sb(III), Bi (III), and T: (II), it is preferred to use organic peroxides as stabilising compound against depolymerisatior. and melt viscosity reduction. It is believed that the combination of a catalyst of this particular type and a peroxide leads to a highly stable polymer composition, in particular polylactide composition. Within this embodiment the use of a Sn(TI) catalyse may be particularly preferred, tor further specification of suitable catalysts within this embodiment reference is made to what has been stated above.

In general, organic peroxides of the following groups can be used, provided that they meet the requirements above: dialkyl peroxides of the formula R-O-O-R' , hydroperoxides of the formula R-O-O-H, peroxyesters of the formula R- (CO)-O-O-RS diacylperoxides of the formula R-(CO)-O-O-

(CO)-R', peroxycarbonates of the formula P-O-(CO)-O-O-R', peroxydicarbonates of the formula R-O-(CO)-O-O-(CO)-O-R', and pcroxyketals of the formula C (0-0-R) 2R' R". In all of the above, R, R ^f , and R" are organic alkyl or aryl groups with 1-20 carbon atoms, straight or branched, substituted or non-subs tituted, saturated or unsaturated, and whether or not provided with heteroatoros . Suitable peroxides are known in the art, and with the guidelines given above it iε within the scope of the skilled person to select ^" , a suitable material. Suitable peroxides include benzoyl peroxide, dilauroyl peroxide, ter t-butylperoxy- diethylacetate, tert-buty±peroxy-2-ethyl hexanoate, tert- butylperoxy-isobutyrate, tert-butylperoxy-acetate, tert- butylperoxy-benzoate, and dibenzoyl peroxide. The peroxide is generally added in an amount of at least 0.01 wt. %, in particular in an amount of at least 0.05 wt. %. The amount of peroxide added is generally at most 3 wt. %, in particular at most 2 wt . % . Tf peroxides are used for the stabilisation of the polymer, or to create long-chain branched PLA, as known in the art, higher amounts of catalyst may be used than conventionally, to increase che reaction rate even more. These higher amounts are mentioned above.

As the presence of substantial amounts of monomer in the polymer may detrimentally affect the mechanical properties and processing behaviour of the polymer, the polymer is generally also subjected to a devolatilisat ion step. The devolatilisation step is carried our to remove volatiles, in particular unreacted monomer from molten or solid polymer. The volatiles are removed at increased temperature under reduced pressure, e.g. under vacuum. In the product that is obtained after the devolatilisation step, monomer, in the case of polylactide, lactide monomer, is generally present in an amount of less than 2 wt.%, more in particular in an amount of less than 1 wt.%, still more in particular in an amount of less than 0.5 wt.%. Examples of devolatilisers include extruders, especially twin screw extruders, wiped film evaporators, falling film evaporators, rotary devolatilisers, rotary disk devolatilisers, centrifugal devolatilisers, flat plate devolatilisers, and static expansion chambers involving special distributors, e.g., Sulzer devolatilisation technology as described in EP1800724. The use of a static expansion chamber is considered preferred, Devolatilisation in various stages and/or a combination of various types of apparatus is also possible. Stripping gas such as nitrogen can be applied to one or several stages in order to facilitate devolatilisation. Devolati lisation may also be conducted in solid-state post-polymerisation equipment or by drying of solid pelletised product under vacuum or inert gas flow, e.g., in a tumble dryer. Optionally, a crystallisation step may be performed before the drying step.

After the devolatilisation step the polymer may be directly further processed to end-use by extrusion, blow- molding, film casting, film blowing, thermoforming, foaming, or fiber-spinning at elevated temperatures to form useful articles. If so desired, the polymer may be compounded with additives such as anti-oxidants nucleating agents, mineral fillers, glass or natural fibers, processing aids, UV-stabilisers, or other polymer- additives known to the skilled person.

It is also possible to allow the polymer to solidify to form a solid intermediate product. For example, it may be converted to particles such as beads, chips or other pelletised or powdered products in manners known in the art and then sold to end-users.

The ring-opening polymerisation process according to the invention provides polyesters, in particular polyesters, more in particular lactide-based polymers or copolymers, of controllable, consistent, and high quality in terms of molecular weight, polymer chain architecture, colour and residual monomer content, In the case that only optically pure lactides are used as cyclic ester monomer also a very high optical purity is obtained.

The present invention will be elucidated by the following Examples, without being limited thereto or thereby. Comparative Example 1

L-Lactide (Purasorb L, ex PURAC Biomaterials) was molten batch-wise in a 50 liter tank under a nitrogen atmosphere, The molten lactide with a temperature of 110-125 ⁰ C was pumped continuously into an oil-heated, straight tubular polymerisation reactor by means of a piston pump at a flow rate of 3-6 kg/h. The internal volume of the straight polymerisation reactor was equipped with static mixing elements designed to provide intimate mixing and plug flow behavior. Temperature and pressure were recorded on 4 different point equally distributed along the linear polymerisation reactor. The molten lactide was heated to the reaction temperature of 180-220 ⁰ C in the first part of the polymerisation reactor. Tin octoate polymerisation catalyst and dodecanol initiator were added to the molten lactide as a concentrated solution in toluene by means of an HPLC pump. Solution concentration and flow rate were adjusted to arrive at a typical catalyst concentration of 500ppm SnOcts and a dodecanol concentration of 0.15 wt.-%, corresponding with a theoretical average Mn of 122kg/mo1 of the PLLA at 100% conversion,

Ring-opening polymerisation started and melt viscosity increased accordingly while the mixture progressed through the linear tubular reactor.

The reaction mixture left the tubular reactor through a die plate with 1 circular hole mounted to the end of the reactor. The viscous strand was cooled in a water bath and pelletised by means of a strand pelletiser. During lactide polymerisation, a varying pressure drop with a maximum value of 80 bars developed over the reactor. The temperature increased slightly just after the position whore catalyst and initiator were added, indicating the onset: of polymerisation. The increase of temperature of the melt by approximately 5 ⁰ C was caused by the inability of the oil circuit encasing the tubular reactor to aosorb the heat generated by the exothermal polymerisation reaction.

Experiments demonstrated that this linear, tubular polymerisation reactor set-up with internal static mixers could not be operated in a steady-state condition. More m particular, the recorded nays temperature and pressure fluctuated unpredJ ctably. The output of the reactor changed from polymer to melted lactide and back, and lactide sprayed out of the die. Metal analysis of the lactiαe that came out of the die revealed that it contained the expecteα amount of tin polyner ipati on catalyst, so it should polymerise. The observation that molten lactide (with catalyse) left the die means that the residence time of the lactide monomer m the reactor was too short for polymerisation to polylactide. Apparently, low viscous lactide found a small channel through the high viscous reaction mixture and escaped fron the linear reactor prior to polymerisation.

Example 1 according to the Invention

L-lactide fPuralact L, ex. PURACj was molten batch-wise in a 100 liter stirred tank with external oil heating under a nitrogen atmosphere. The molten lactide with a temperature of 110-125 ⁰ C was transferred under gravity and nitrogen pressure to a 200 liter feed tank. The lactide monomer in the feed tank was pumped continuously into a first oil- heated static mixer of tne polymerisation reactor by means of a micro annular gear pump at a flow rate of 25k:σ/n. Α mixture of tin octoate polymerisation catalyst and hexanol initiator was continuously added to the molten lactide by means of a HPLC pump. The flow rate of the mixture was adjusted to arrive at a typical catalyst concentration of 200-300ppm SnOCt ₂ and a hexanol concentration corresponding with a theoretical average Mn of lOOkg/mol of the PLLA at 100% conversion.

This first static mixer mixed the lactide with the catalyst and initiator and heated it to 170-190 ⁰ C. The residence time in this first static mixer was 1-2 minutes, which is too short to build up significant melt viscosity.

Subsequently, the low viscous reaction mixture was provided to a further tubular polymerisation reactor with a loop configuration, The internal volume of the loop reactor was equipped with an array of internal static mixing elements, and a separate gear pump. The mass flow rate in this loop reactor was typically set at 250kg/h. In the loop-reactor, the viscosity of the melt increases and conversion develops to approximately 80-85% in a steady- state situation.

The loop reactor was operated at fairly constant polymerisation temperature chosen between 180 and 200 ⁰ C. No significant temperature increase was recorded upon addition of the catalyst/initiator to the lactide monomer.

At the opposite side of the feeding zone, a linear, tubular plug flow reactor equipped with internal static mixing elements was connected to the loop reactor. Pre- polymerised material was continuously withdrawn from the loop reactor and provided to the plug flow reactor by means of a gear pump. A die plate with two circular holes was mounted to the end of the plug flow reactor. Highly viscous polymer left the reactor through the dies, and the polymer strands were cooled in a water bath and subsequently pelletised by means of a strand pelletiser. The obtained pellets were transparent, virtually colorless amorphous PLLA.

During lactide polymerisation in the cascade of polymerisation reactors equipped with internal static mixers, a constant pressure drop over the reactor developed, with a maximum value of up to 60 bars in the loop reactor, a maximum pressure of 40 bars at the pump side of the plug flow reactor and a constantly decreasing pressure towards the end of the reactor. During polymerisation trials, temperature and pressure were stable for at least several hours and channeling of lactide was never observed. Thus, polymerisation of lactide could be performed consistently with this arrangement of polymerisation reactors equipped with internal static mixers,

The high-melting poly (L-lactide) collected from a stable operation of the process typically exhibited a D-isomer content of less than 0.4%, Molecular weight analysis of the PLLA by means of size-exclusion chromatography (SEC) using chloroform as the eluent and PS calibration standards, revealed Mn values of 130-150 kg/mol, Mw values of 190-210kg/mol and a polydi spersity of 1.4-1.5. Residual lactide content in the PLLA was 4-5% if the plug flow reactor was operated at a temperature of approx. 190 ⁰ C.

This example shows that the use of a combination of a loop reactor with internal static mixing element to effect pre- polymerisation and a plug flow reactor with internal static mixing elements to finalize the polymerisation results in a stable polymerisation process.