CA2197644A1 - Integrated process for the manufacture of lactide - Google Patents

Abstract

An integrated process for the manufacture of lactide from a solution of lactic acid in de-ionized water comprising the sequential steps: A) dehydrating the aqueous lactic acid in at least two stages to effect condensation polymerization of the lactic acid and the formation of oligomers in which the average number of monomer units is 8-25, the dehydration stages being conducted in equipment fabricated from low ferrous materials; B) thermally cracking the oligomers in the presence of depolymerization catalyst to form lactide vapor of which the average residence time within the cracking zone is less than 15 seconds; C) condensing the lactide vapor and fractionally distilling the condensate to remove concentrated lactide as a liquid sidestream; and D) subjecting the concentrated lactide sidestream to melt crystallization to separate purified lactide having an Acidity Potential less than 6.

Description

WO 9il06092 ~ ~ 9 17 ~ f P~ S ~3 .

INTEGRATED PROCESS FOR THE MANUFACTURE OF LACTIDE

Fi~ l~l o f lnveniif~n The invention is directed to an integrated process for the 10 ~ C; of purified lactide. In particular, the invention is directed to such process in which high purity lactide is produced ef~iciently with high yields.

~ orlcprounfl of th~ Jnventifln Because of its hir,fl~ 'ily~ polylactide Lnoly(l~tic acid)] has become of increasing c~Jl.llll.,l.,;al interest as a substitute for less readilydegradable materials such as polyolefins and pol~ urethanes. The basic technology for making polylactide extends b~k to as early as 1932 with the work of Carrothers et al. ["Studies of Pol y~ ;~iiUII and Ring Formation. X. The 20 Reversible Poly..l~l;~ii."l of Six M_.llb~lGd Cyclic Esters", American Chemical Society Journal, v. 54, pp 761-772]. N~ ,.lL~,k,~i" the C.JlIII11~ ' accept~mce of polylactide has been inhibited by the high cost of such pol~mers as compared to polyolefins and polrul, - However, as concerns for the ~ ilVIII~ . become greater, the urgency of using more ~.IV;II ' lly friendly materials such as 25 polylactide has become coll.,~lllhl~ly greater. To meet this urgency, there exists a substantial need for more eron~ miral ways of making poly(lactic acid) [PLA].

The basic process for making PLA involves the d~Lydlfltioll of 30 aqueous lactic acid to forrn a mixture of oligomers of lactic acid. The oligomers are then subjected to ll. ..,~ to make lactide, that is, the cyclic diester of lactic acid. The lactide is admixed with a ring-opening catalyst and subjected to heat and/or pressure to form PLA.

This general method for rnaking PLA is illustrated by the integrated process disclosed in U.S. 5,142,023 to Gruber et al. and related patents U.S.
5,247,058, U.S. 5,247,05g and U.S. 5,258,488.

Gruber et al. disclose an integrated process for making PLA from aqueous crude lactic acid comprising the following sequential steps:

WO96/0609~ ~ ~ q 7 ~ . ' 561 I . In two stages, c~ oli.Lin~: water from the crude iactic acid to forsn lacticacid ol;gomers having a molecular weight of 100-5,000 (n= 1.1-72);

2. Mixing the oligomers with d~ ulylh..,i~liu~l cataiyst and thermaiiy cracking the oligomer to form lactide vapor;

3. Removing lactide vapor from the thermai cracking zone, condensing it and fractiorlally distilling the lactide condensate to form a "purified" lactide; and

4. Reacting the "purified" lactide to form PLA.

It is recognized by those skilled in the art that water andlor its reaction products with materiais such as lactic acid and lactic acid oligomers 15 should be kept at very low ~ .... .1 . Al ;....~ in order that high molecular weight polymers can be made therefrom. For example~ see Rnti~h Polyrn~r Jollrn~AI Vol.
23, No. 3, p. 23~-240 (1990), which teaches that the content of free carboxyl;c groups should not e~ceed 0.8 meqlg (fO0 meqlkg). Ncv~ lLllcl. ~i., it has been found that even such modest amounts of acids are much too high for the 20 I~ ur~ ofhignpuritylacticacidpolymers.

Ilowever, the Gruber et al. process is based on the premise that the lactide from the oligomer cracking unit, if, ' by distillation, will be suitable for poly IlI.,II~LliiUII without further treatment such as crystallization or 25 solvent extraction, both of which have been suggested ~or this use in the prior art.
However, it has been found that the p~ocess taught by Gruber et al. has several UI L~UI~ which make it unsuitable for the lll~luLI~,L~ of high quality PLA.

In particular, following the teaching of the Gruber process results in 30 (I) a -..~ rA- u~.; ~ sequence which requires extensive time for polyl...,i~Liu", (2) excessive equipment costs and (3) produsts therefrom having inadequate properties for many A~ ;U"~ Manyy of these polymer quality concers~s arise from the fact that the process as taught fails to recognize the cause and effect of the many and varied side reactions which adversely affect polymer quality.
While the Gruber et al. process purports to advance the technology of making PLA, it n~ ~ ~l Lh~ .x, fall s Cul~;~l ~Iy short of teaching a practical process which can make PLA readily available in CU..~ II quantities with the W0 96/06092 2 1 ~ 7 6 4 ~ r~ 3 high quality that is needed for the widespread use of PLA in consumer ,1;. u ;. .ne Therefore, there remains a substantial unmet need for a practical route for making high quality lactide which is suitable for making PLA by ring-opening ~)oly~ uliuli thereof.

S-lmm~ty of the Invenrinn In its primary aspect, the invention is therefore directed to an integrated process for the nlull~u~ ; of purified lactide from an aqueous solution 10 containing at least S0% ~t. Iactic acid comprising the sequential steps:

A. Feeding a solution of crude lactic acid in de-ionized water to a first heatedzone in which free water is removed by e\i.llJUl~liUII and a molten mass of C' ~ ' ''"I ;C'Il polymer containing lactic acid and a minor amount of free water is 15 formed by r.... ~ ;.... polyll.~ uLiull to an extent that the average number of monomer units (n) in the ~....~I... -1;.... polymer is 2-8;

B. Feeding the molten mass of r nn I .- ~ ;.... polymer from step A to at least one furlher heated zone in which the diffusive surface area of the polymer is 20 increased, the residual lactic acid and ~;n.~ ;.... polymer are further c~n~ otir n-lluly"" ~i~d to an extent that n is 8-25 and bûth free water and bound water are removed by e . UIJUl.lliUII, steps A and B being carried out within equipment the surfaces of which in contact with the reactants are fabricated from low ferrous materials;

C. Contacting the mol~en ~ pol~mer with an alkali metal-free d~ IJolylll.,~;~uLioll catalyst in a cracking zone operated at a liquid h...~" ...lLi.~ no higher than 240C and pressure sufficient to effect cracking of the molten c~ ' polymerv.~iththe-....~u.,.;~- 'formationof(l)avaporousreaction 30 mixture containing water, lactic acid, lactide and entrained heavy oligomers and (2) molten liquid heavy ends containing heavy oligomers;

D. Removing the vaporous reaction mixture from the cracking zone at a rate such tbat the average residence time of the l~tide vapor within the cracking zone 35 is less than IS seconds;

E. Cn ~lr neing the vaporous reaction mixture and fi':lr'tin"~ting the condensate therefrom whereby lactic acid, water and minor amounts of lactide are removed as wo 9610609~ r~ u 21 '~7~ 4 vapor overhead~ 4- ~"' ~ ''0,-5 1 Iactide is removed as a liquid side stream and the heav5 ends are removed as molten liquid; and F. Subjecting the: ' lactide to melt ~ ~lli~liull by which a

5 purified lactide fraction having an Acidity Potential less than 6 meqlkg of lactide is separated from a residual lactide fraction having an Acidity Potential of at least 30 meq/kg.

In a second aspect, the invention is directed to a method of making 10 a -.,. .;. -l~l stream of lactide from lactic acid oligomers comprising the sequential steps:

A. Contacting molten lactic acid oligomers in which the average numher of monomer umits is 8-25 with an alkali metal-free d~l~oly~ iu~ catalyst in a 15 cracking zone operated at a lr. ~ r no higher than 240C and pressure sufficient to effect cracking of the molten oligomers with the .
formation of ( 1 ) a vaporous reaction mixture containing water, lactic acid, lactide amd entrained heavy oligomers and (2) molten cracker bottoms containing heavy oligomers;
B. Removing the vaporous reaction mi~ture from the cracking zone at a rate such that tne average residence time of the lactide vapor within the cracking zone is less than 15 seconds;

25 C. Condensing the vaporous reaction mixture and r. ~. 1;. . ;..g the condensate therefrom whereby lactic acid, water and minor amoUntS of lactide areremoved as vapor overhead, ~ ' lactide is removed as a liquid side stream and the condensate heavy ends are removed as molten liquid;

30 D. Cooling the molten liquid cracker bottoms to around 190C and remoi ing them from the cracking zone at a rate such that the average residence time of the molten cracker bottoms in the cracking zone is less than l~ minutes; and E. Subjecting the cooled molten liquid cracker bottoms to dehydration or to 35 hydrolysis followed by d~h,~ ;UII amd recycling the resultant dehydrate to the cracking zone.

W0 96106092 ~ 1 ~ 7 6 4 ~ , C ~3 Bri.lf D~r~pfi~n of th(~ Dm~

The Drawing consists of six figures as follows:

Figures la and I b together constitute a schematic flow diagram of a prefelred manner of carrying out the process of the invention; and Figures 2a, b and c, Figures 3a and b and Figure 4 are graphical cu,lel.lliu.l~ of various operating variables associated with the d.,h, dl~tivi~ step of the invention.

Figure 4 is a graphical correlation of the effect of alkali metal on Figure S is a graphical correlation of the effect of the C
of materials of CVlli~il U~tiV.. on product color.

Figure G is a graphical correlaùon of the effect of t~ Lul ~; on the amolmt of meso isomer in the product.

D~?finifinnc A. "Acidity Potential," when used with respect to a given process stream, refers to the content therein of acidic ~, ..~ ,I.~, ,.I~ or potential acid-forming 25 rnaterials such as water. Such materials are measured as meq/kg of material;

B. "Alkali metal-free" is used with reference to depolymerization catalysts and denotes that such materials contain no more than 10 ppm by weight alkali metal or alkaline eartb metal, basis lactide;
C. "Bound water" means water formed by the .~ reaction of t vo or more organic .~ containing a plurality of hydroxyl (-OH) groups, for example, water produced by the ~ lig. ,. ., S, ~: ;. ,. . of lactic acid;

35 D~ "De-ionized Water" means water which contains no more than 100 ppm by weight ions of allcali metals or alkaline earth metals;

2~ ~J16fi~4 wo 96/06092 - 6 - r~ 3 E. "Free water" means water physically associated with a body of reactants, e.g., as a solvent or as water of cry~tnlli7~til n F. "Heavy oligomers" means those oligomers of lactic acid formrd by 5 cnn~ polyl~. i~tivll of 3 or more lactic acid molecules; and G. "Non-ferrous" refers to the iron content of materials of n.~ I " .., In particular, non-ferrous materials are those which contain less than 50% by weight iron or iron ions. Lron content is of concern in the process ofthe invention where 10 it comes into contact with reactants, particularly acidic andJor aqueous reactants.

Df t~liled Dpcr~tion of thf- Invrn~i- n A. In General:
The pOIyJ~ .CLiion and ~iVIJOl~ aiiUll of aliphatic hydroxy carboxylic acids such as lactic acid are usually carried out by ring-opening polymeri_ation of their ~;VII~ .lOll lillg cyclic esters in the presence of a ring-opening catalyst such as divalent tin. Thus, as mentioned hereinabove, a ~,VII~,.IIiVlll~ll method for making 20 poly(lactic acid) [polylactide] involves the d~ hy~h~ivll of aqueous solutions of lactic acid to form oligomers of the lactic acid, which are then thermally cracked to form the cyclic diester (lactide) of lactic acid. The lactide is then mixed with a ring-opening catalyst and subjected to heat and/or pressure to effect ring opening of the cyclic diester and formation of polylactide. (As used herein, the terms 25 "polylactide," "poly~lactic acid)" and "PLA" are h,t~ f ) In ~/VI,ylll~,.l~.illg lactide, it has been found that the presence of even small amounts of acid and acid precursors, such as laclic acid, inhibits ring-opening of the lactide and increases the amount of . ~
30 in the polymer. Thus, pV~ iV~l time becomes excessive if high conversion levels are sought. The process of the invention is therefore directed to producing lactide which has low Acidity Potential and reduced l~ll lS. . . l . " i~ ll, by products from thermal drgrP~ ti-.n 35 B. Lactic Acid:

The aqueous lactic acid which can be used in the process of the invention cfsn be made synthetically or by 1.;1. h. .: ~I means such as the 21 S76~
WO 96106092 p~ .15 6 f~ " "F .~ n of various sugar sources such as whcy permeate, corn glycose, beet sugars and the like. It is, however, preferred that the lactic acid feed materials contain no more than about 50% by weight water. Even less water content is preferred since less energy must be expended to remove such smaller amounts of 5 water.

Lactic ~id, of course, exists in two forms which are optical , i.e., D-lactic and L-lactic acids. Either of these lactic acid forms can be polymerized to form oligomers and thus can be used in the invention to rnake 10 l~tides. The l~tides produced therefrom have three types of optical activity depending on tbe ~ o~ of the crude lactic acid. If the feed contains only L-lactic acid, the resulting product is L-lactide. If the feed contains only D-lactic acid, the resulting product is D-lactide. But if the feed contains a mixture of D-and L-lactic ~id, the resultant lactide is an L. D and meso-lactide mixture. The15 process of the invention can utilize either form of lactic acid as primary feed.
However, it will usually be preferred tbat either the L- or D-lactic acid be used in ll y pure form with respect to optical purity.

Despite their purity, the optical activity of lactic ac;d and lactides 20 made therefrom tends to change under certain conditions toward optical inactivity in which equal amounts of the D- and L-~ .. s are present. This tendency is aggravated by impurities and by exposure to high IC~ d~ C~ for long periods of time (thermal exposure). Furthermore, the rate of, ~ F 11;',~1;UIl is affected by the relative ~.. ..I, u;...,~ of D- and L-. - a;...,. ~. Tne invention process is 25 therefore oriented toward the ~ of those operating variables which adversely affect optical purity.

C. Oligomer Formation by Del..yd,dliu...

For the purposes of the invention, it is essential that the d~h~ livl.
of lactic acid to form oligomers be conducted in such manner that entire d~ l..,liu.. step is carried out at the lowest possible t~ Ul C consistent with obtaining economical yields of oligomer which has ,, . optical purity.
FIJ~ IVI~, to avoid .-- ~ ..;,. I;. .., and other side reactions which arise from tbe presence of alkali and alkaline earth metals, it is preferred that the Wo9GI06092 ~ t ~ i ~ 4 4 I~ 13 water in which the lactic acid is dispersed contain no more than lO0 ppm by weight of such metal ions and preferably still no more than ~0 ppm.

It is, of course, well known that water reacts with lactic acid 5 oligomers to reduce their chain length. Therefore, it is essential to remove as much water from the lactic acid as is practically possible in order toobtain oligomers hRving high enough molecnlar ~veight. In this regard, it is known thal if the feed acid is comprised of, say, 98~o L-lactic acid~ then the oligomers and ultimately the lactide can contain no less than 3.9~/0 meso-lactide. E'u~G~ vlc,10 depending upon the Ih~ t~ uu~ cd during d~,L.~dl~lliul~ and the nature of tlle particulamlG~UlY~ livll, the optical purity of the resultant lactide will be reduced even further. Therefore, the d~h~ tiull step of the invention isdirected to removing as much water as is possible while at the same time .,.;.,;".;,;,.g those factors which adversely affect optical purity.
From the standpoint solely of watecr removal rate, it would be desirable to operate at a t~,U~ tUI~ substantially above the boiling point of water.
Elowever. if the tc.~ luuG of d~ d~LIiUII is raised too high, excessive rRr,-mi7Rtinn of the formed oligomers takes place. The degree to which this is a20 problem in the dehydrator depends on the amount of, ~ ,..,i ~;.... which can be tolerated in the feed to the depol~ ,.i~ivll unit. However, it has been found that 240C is a practical rnaximurn t~ lUlc above which the degree of thermal n~rrmi~tirn becomesexcessiveformost~l.l,l;.~li...,~ At~ UI~ofnomore than 1 80C is further preferred. To be able to utilize lower ~ a ~ and 25 therebykeep-.--- .,;,~;..,.oftheoligomerstoalowvalue,itispreferredto operate the d~L.t~ , especially the second stage, at reduced pressure.

The preferred operating conditions for d~ hyll~ioll are detennined by balancing twû factors: (I) at lovver values of n, more water remains in the 30 system and lactide yield in the cracking step is reduced; and (2) at higher values of n, there is less water in the system, lactide yield is higher. but quality is poorer.

In the .L,I~ iou phase of the invention, the first dehydration stage is operated in such manner that sub~tantially only free water is removed 3~ from the lactic acid feed and the average number of molecules in the product therefrom is no more than 8 and preferably only 4-6. It ;s feasible to speed up water removal in the first stage by using a vacuum; however, it is not so critical as in the subscquent stage(s). Reduced pressure operation can, however, be used w0 96/06~92 2 ~ ~ ~ 6 ~ 4 ~ 5 ~ ~ ~3 throughout all stages of the d~,hylLdliull process. The first stage of d~,Lyll~liu..
can readily be carried out in a packed column.

It is definitely preferred to operate subsequent d~,hy.LdliOI, stages at 5 a vacuum in order Dul,l,~.~Dl'ully to remove the relatively small amounts of free water and boumd water in the lactic acid within the least practicable time and at the lowest practicable ~ ,ldtUI~. Again, depending on the nature of the depolyul.,.i~liull catalyst which will be used in later processing and in the amount of thermal ., .,~ which can be tolerated in the lactide product, it v~ill be 10 preferred to operate the second d~h.ydldliull stage and any subsequent d~llydl~liu stages at a substamtial vacuurn.

In general, it is preferred that the ll~.llydlllliUII phase of the invention be carried out in at least two stages. ln the first stage. essentially only 15 free water is removed and the average number of monomer units (n) in the product therefrom is 4-6 and preferably n is no higher than 8.

In the subsequent dehydration stage(s), the lactic acid is further ol ;~ " ;, J by removal of bound water until the average number of lactic acid' 20 molecules in the oligomeric mixture (n) is 8-25. Because ûf the high viscosity of ol;gomers in which n exceeds about 25, the rate of diffusiûn of bûund water fromthe r,/~n~r~ n reactions becomes limiting and makes the production of higher molecular weight oligomers even more difficult. That is, water removal becomes diffusion limited. Therefore, it is preferred that the oligomeric mixture in 25 subsequent dehydration stages be carried out in such manner that the surface area of the mixture is enhanced. Such enhanced surface area can be achieved, for example, by the use of wiped-film equipment or by the use of high pressure spraying devices. By these means, the rate of diffusion is raised and the overall t~ Lul~ exposure of the oligomer is kept to a minimum. It is preferred in 30 the practice of the invention that n not exceed 25 since the viscosity of higher molecular weight oligomers becomes so high that admixing the de catalyst without subjecting the oligomers to very high t~ ul~.~ becomes difficult.

Two stage dclly~ liul2 will ordinarily be ca2ried out so that n of the first product is 2-8 (preferably 4-6) and n of the second stage is 8-25. With three stage operation, n of the first stage product would preferably be 2. Product from WO 96/06092 Z ~ ~ 7 ~ 4 ll I ~_lru _I rt3 the second stage product would have n = ca. 10 and product from the last stage would have n = 15-25.

D. Thermal D-x~ n~ . Catalyst:

It is preferred in vhe process of the invention that the thermal d~,~,u~ ruD;Liul~ of vho lactic acid oligomers be facilitated by the addition of a d~,l,ol~lll.,.;~Liul. catalyst. Such catalysts include metal oxides, hydroxides,carbonates and wll~v~r.yl_t~,s of SnII, Snl~, SbIII, Znll and Billl. Examples of1n .such catalysts include tin octoate (tin 2-cLLy" ), t;n lactate, tin oxide, antimony octoate, bismuth octoate, zinc stearate, zinc octoate and the like. (Inmany instances, such depolylll.,.i~Lvl. catalysts can also function as a ring-opening catalyst for polylll~ iull of the lactide.) Either solid or solublecatalysts can be used and the catalyst can be dispersed in the oligomer feed tû the 15 depoly~ ..i~liu.. unit or the oligomers can ve contacted with a fixed bed of such catalyst within the Ih ,~nl - ~ ;,.g v essel. When the catalyst is dispersed in tbe oligomer cracking feed, it is preferred that the catalyst be used in amounts of 0.05~% by weigbt, and preferably 0.5-2.0% by weight.

It will be recognized that the use of a suitable d~,~,ol~ .,l;~l;w~
catalyst enables the 11l~l .. ,. ~ ;,.~ operation to be operated at milder conditions of t.,~ ....t~C. The use of milder ~ conditions, of course, reduces the amount of lactide ~ .,. caused by timc t~ rlu~l~ exposure.

Thûugh the precise c~ .. . of the cat.llyst is not critical, i.e., a wide variety o~ dcl~vly~.~i~liull catalysts can be used, it is 1~ ~ IhCI~
quite important that the catalyst be substantially free of alkali metal and alkaline earth metal ions. It is known that alk line substances have the adverse effect of causing, ,.. ; ~,a ;. ,l ~ ("lligh-Molecular, Particularly Optically Active Polyesters 30 of Lactic Acid, A C~ntrihu~inn to the SL.c ~ y of 1~ f ~. " ""..1....l ., Compounds",Mr~krnm~ hrm 30,no.1:23-38,April,1959~. However, applicants have found that limiting the amount of alkali metal and alkaline earth metal ions in the reaction system, in u" h:, ~ ;l ,. . with the use of rlon-ferrous materials of ~;u~ 1;". " greatly reduces the effect of other side reactions which 35 degrade both the quality and yields of the lactide.

Experience with such catalysts has shown that no single catalyst is superior from the standpoints of both quality and d~ ..~i~liul~ rate.

Wo96/06092 2 1 9 7 6 4 ~ r~ I/u~ u However, from the standpoint of depol~..l~;~.;uu rate, a preferred catalyst is antimony octoate. From the standpoint of optical quality, a preferred catalyst is tin lactate. ~ ,aa~ whatever catalytic metal may be used, it is preferred that it be present in its stable valence state lest oxidation of the catalyst cause formation 5 of color formers and other ~o ~Ih ..;.. ,t~ in the lactide product.

E. Lsctide Formation by Tl.. "... 1~ g As outlined L~ IlldllU ~'C, formation of lactide is ~ o. . .~ i by 10 thermally cracking the oligomers (n = 8-25) in the presence of a suitable catalyst.
This can be done by passing the heated oligomer through a fixed bed of catalyst witbin the cracking unit or it can be done by admixing the catalyst with the oligomer feed to the cracking unit. In the latter case, in order to minimize thetime- ~ ..at~..c exposure ofthe oligomers within the cracker, it is preferred that the catalyst be preheated and mixed with the oligomer feed. The oligomer is thenheated quickly as it passes into the upper end of the thermal cracking vessel to the topmost of a series of trays, each having liquid duwll~ ulll..a to the next lower tray.
The liquid oligomer is heated to its cracking t~ alulc as it passes down through the series of trays and the lactide cracking product becomes vaporized and 20 is removed from the vessel overhead. To assist the separation of lactide amd other vaporized materials from unreacted heavy oligomers and other residual liquid materials, a stream of heated dry nitrogen gas is passed upward through the column ~u~ t~ lolltly to the falling oligomer liquid. The nitrogen stream in t~nn l~in:~til~n with the rising lactide vapor effects more efficient stripping of the vapor from the liquid in the cracking vessel. The use of nitrogen-assisted stripping in this manner is disclosed in U.S. 5,023,84g and U.S. 5,091,544 to Bhatia.

Suitable t~ Lulcia for the d~olyl~ dtiùn vary widely, but will usually be within the range of 1 85-270C and preferably 200-220C. The 30 optimum t....l...,I...r for any particular oligomer feedstock will vary with the ~ u. . .~ of the feedstock, the catalyst and the pressure within the cracker. The pressure within the cracking unit can vary widely and thus can be either above or below hLIIIua~L~,I;C pressure. In some instances, it is desirable to operate at reduced pressure in order to lower the time ~ p.~.~lt~ exposure of the vapori~d 35 lactide product. Operating efficiency of the lactide recovery operation can be raised if the oligomer cracker is also operated under vacuum with reduced amounts of nitrogen for stripping.

WO9U06092 ~q 7~44 r l.. s6~3 Though the precise reactor ~ v l~ . is not critical, a palti.,lll~ly preferred design is a columnar reaction vessel having a vapor product outlet at the top of the column, an oligomer feed line near the top of the column and a series of distillation trays proceeding do~nward, each having liquid 5 dvw~,u~ to the next lower tray. Heavy liquid oligomers which bave not been converted flow do-wnward from the lowest tray and are collected and removed from the bottom of the column. Nitrogen gas for stripping is introduced behveen the liquid bottoms and the lowest tray and is directed upward mto the .lu....~,</~ g flow of ~ ull~dh,d oligomer. Sieve plate distillation trays, each bearing a liquid layer of IJII~,UII ~ ~. t.d oligomer, have been found to be very effective to mix the feed, vapor amd liquid intimately and thus to irnprove reaction efficiency.

In a preferred mode of operating tbe cracking function, solid inert packing is added onto the top of each of the sieve plate distillation trays by which the reaction effciency of the column is ~,ull~;dl,.Gbly improved. ln particular, the addition of packing to the sieve plates within the thermal cracking vessel adds greatly to the reaction efficiency of the thermal cracking operation, as follows:

1. The packing improves the heat transfer efficiency between the vapor and liquid;

2. An increased number of theoretical reaction stages is realized, 3. There is more intimate contact between the nitrogen and the VUWIll.,Ulllill~:, liquid, therefore, there is more effective stripping; and 4. For a given liquud height on the trays, the amount of liquid is reduced, thus reducing the average residence time of the react~mts and also reducing the degrçe of ".~ . ,. ,~I;i. .
Suitable packing materials are cvll~ iulldl column packing such as Berl saddles, Raschig rings, plain rings, spheres and the like. Suc,h packingmust, however, be made from matçrials which are chemically inert with respect tothe reactants.
Alternatively, instead of using a series of distillation trays, an . " " .1 J' zone in the column can be packed with a solid fnrllrninn--c bed of catalyst through which thç oligomer feed is passed downward. Such catalyst bed W096N60g2 ~ l 9 7 6 ~ 4 r.,l/u.,~'.J9~t3 would contain a series of heating coils to effect heating of tne oligomers to their cracking t~..",. ., before emerging from the catalyst bed.

Regardlessofthereactor~ l~L...rl;..,. itisimportantthatthe 5 average residence time of the vapor lactide product within the reactor be kept to a minimum in order to avoid ~,..rl. ~ Irl~lf side reactions. Therefore, it is preferred that the average residence time of vapor within the reactor be no more than l 5 seconds and preferably no more than l O seconds. Similarly. it is preferred that the average residence time of the unreacted oligomers within the reactor be no more 10 than 45 minutes and preferably no more tban 25 minutes in order to minimi~ the production of highly viscous residues which are not readily pumpable.

Tne bottoms product from the cracking unit consists mainly of heavy oligomers, crosslinked species, spent catalyst amd small amounts of 15 absorbed volatiles, and color bodies. It is therefore preferred that tne cracker bottoms be removed from cracking zone7 heated and subjected to vacuum flashing to remove the small amounts of volatiles therefrom. A portion of the bottoms canbe recycled to hydrolysis or to d~hrdl~Loll~ However, some will likely need to be removed from the process to prevent excessive buildup of inert spent catalyst.
F. Lactide Recovery:

The vapor overhead stream from tne oligomer cracker contains substantial quantities of lactide. However, the stream also contains substantial25 quantities of impurities such as water, light ~~ ;,... products and entrained liquids such as oligomers and other heavy ends, which must, of course, be separated from the lactide. In the process of the invention, this is preferably done in a two-stage operation which involves condensing tne cracker overhead to separate the more volatile materials and then fractionally distilling the 30 lactide-enriched ~ rm~ c~ltP The overhead from the distillation column consists mainly of lactic acid, water, nitrogen and entrained oligomers and l~tide. A
liquid stream containing principally lactide is withdrawn from the side of the column below the feed tray and heavy ends, consisting mainly of heavy oligomers,are collected and removed from the bottom of the column. The lactide recovery is35 carried out under vacuum in order to keep the ~ low. In operation of the cracker overhead condenser, it is important to optimize the t, ~ .,. That is, the t~ should be high enough to removed a maximum amount of water, but low enough to maximize lactide recovery.

Wo96106092 ;~ 7 ~ q ~ r~ ,3 .

Heavy ends from the lactide distillation can be recycled to tne lactide vacuum distillation columm or they can be recycled to one of the .1~".~ d~liu~ stages.

G. Lactide r~irl.,d~;ol..

In order to purify the ....,....~ .d lactide from the lactide recovery step, it is preferred to subject the ~ ' d impure lactide tnerefrom to melt10 w y~t~lli~tiull by which the lactide can be freed from impurities such as lactic acid, oligomeric residues, solvents and catalyst. This method of ,uul;r~ ,.. canbe comprised of one or more stages which involve the following steps:

I . Cooling molten ~ l.. u, 1~1 lactide at least to the freezing point of the lactide, thus partially crystallizing the melt and f'orming a solid crystalline phase having lower impurity content and a l;quid phase having a higher impurity content; and 2. Separating the crystallhle phase from the more impure liquid phase;
and 3. Optionally subjecting the separated crystalline phase to "sweating"
vvllereby it is warmed to a t.~ below the lactide melting point to melt selectively a portion of the remaining impurities and a minor amoumt of the lactide.

The impure liquid from the melt crystalli_ation process can be recycled for further separation of residual lactide.

Various types of equipment are known for melt crye~lli7~tinn For example, a simple batch method involves the use of a tank with multiple heat transfer elements equipped for heating and cooling at a controlled rate or with a controlkd L~ LUI~; differential with respect to the material in the tank. The product to be purified is simply melted, frozen, drained, sweated, re-drained and the purified product is melted. The equipment is easily h~L~u~ lt~,.l for automatic control. A preferred semi-continuous method involves the use of vertical coolingtubes and a method for pumping the melted crude material to the top of the tubesand drairling the liquid impurities from the bottom of the tubes. The method may W0 96,060g~ 2 1 ~ 7 6 4 4 P~ t l3 -be made fully continuous by using a number of the tube assemblies with various timing or control devices. A p}eferred type of apparatus for this function is disclosed in U.S. 3,621,664 and U.S. RE 32,241. Using such ~ . f ~, the melt ~,ly aL~lli~liull process can be carried out in either parallel or in series according to the purity, yield and economic criteria of the lactide product.

The use of melt crystallization processes for the ~ of impure lactide is disclosed in copending U.S. patent application S.N. 08/231,964, filed on April 22, 1994, by O'Brien et al.
H. Ring-opening Catalyst and Pol~ ;~Liul~.

The purified lactide product made by the process of this invention can be ring-opening polymerized by a wide variety of mrtl~lrnnt~/ininE catalysts, many of which are well know in the art. In particular metal oxides, carbonates and bu~y' of snIJ, Sblll, ZnlI and Billl are all effective to a substantial degree.
However, preferred catalysts are cnmpo~m~lC of trivalent lanthanum and rare earth metal ~ n,. ,I.u ,~1~ such as those disclosed in Ford et al., U.S. 5,208,297. Suitable catalysts include tin octoate (2-~Lhy-' ), tin lactate, tin oxide, amtimony octoate, bismuth octoate, zinc stearate, zinc octoate, lanthanum bis(2,3,6,6 t.,L~ yl-heptane~3,5-dionato) i-propoxide and the like.

The monomer-to-catalyst ratio (molar basis) is generally maintained below 20,000, but usually not less than 500. Good results have been observed when the IIIUIIUIII~ 04 .i ratio is within the range of 15,000 to I ,000, with the preferred ratio being from 10,000 to 4,000. By employing a ~L ratio of 1000 and ~ 1 .g a low Acidity Potential, very high l..U~ ;UllS can be l~chieved in as little as 2 minutes. I lowever, when monomer to catalyst ratios of 500 or less are used, the resultant polymer is frequently 30 discolored, contains r- ~ ~ and is unstable.

Lactide polymerization is generally carried out with both elevated t.lll~ and pressure in order to reduce the time required for getting high polymer conversion. For example, the pol~ l...,l ;~i;u.. can be carried out in one or more stages at a l- '' ~ '1'G of 120-220C and a pressure which may be as high asseveral tnousand pounds per square inch. In any event, it is preferred that the pol~ ..l.,l;~i;u., be carried out as rapidly as possible to reduce thermal tlP~otinn of the polymer. The purified lactide of the invention is ~ ,..!. Iy advantageous W0 96~06092 ~ t ~ 3 -l6 -in view of its very low Acidity Potential which enables a high degree of poly~ ;~iivu to be carried out in as little as two minutes. This compares to usual polyrnerization tirnes of two or more hours as are disclosed in the above-referred Gruber patent. It is preferred that the Acidity Potential of the 5 lactide from the process of the invention be no higher than 6. Even faster pOiylll.. ;~tiun rates can be obtained if the Acidity Potential is lower, e.g., an Acidity Potentiai of 2 or less is still further preferred. (See U.S. Pstent 5,310,599 to Ford.) The pol ~ ;Lal;0ll can be carried out in any standard equipment.
Continuous IJolyl~ lL~liull can be effected in a screw extruder~ or in a reactorvessel having good mixing capability so that 81~ . ' " U''~ polymer product can be obtained. For example~ a stirred tanic can be used for either batch or continuous pOlylll~ aliuLl~
Lactide ~ulyul~ l~LL;Ul~ can be carried out either with or without a solvent. It is, however, preferred to employ bulk POIYLII~1;LaiiUII~ that is~
pùlylll.l;,sal;oll without solvent. A ~~ lally good way of doing this is to carry out partial poly"...;~l;"l. in a stirred tank reactor and then complete the 20 conversion in a screw extruder. The ~ulyl~ LI;ull cataiyst cam be added to the lactide before andlor during poly.ll.~ ou.

A preferred method for carrying out the invention is illustrated by reference to the Drawing as follows:
With reference to Figure la, crude lactic acid is fed via feed line 1 lo mixing vessel 100 in which the fresh feed is mixed witn various recycle streams from later processing steps and the admixture is pre-heated to a t~ lal~ below which any substantial ~V.:yJUliliiUII of water talces place. The heated admixture of 30 crude lactic acid and recycle streams is pumped via line 3 to a first dehydrator stage 200 in which the mixture is heated further to evaporate free water from the feed and to begin ~ .;;u, polymerization of the lactic acid to an extent that the average number of monomer units in the admixture is from 2 to 4.

Free water amd water from the ~ poly,ll.li~tiul, are vaporized overhead. The p~lu~l~;l~llly aqueous overhead contains small amounts of entrained liquid lactic acid and is passed via vapor removal line S to separator 300 m which the water and entrained lactic acid are separated. The W096/06092 5 ~ 9 7 ~ ~ 4 r~ .. .'i5~13 separated water vapor is passed via line 7 to condenser 400 in which the water is condensed and then pumped through line 9 to hydrolysis unit 500. The separated lactic acid is then returned to dehydrator 200 via line l l . The aqueous solution of ;..., polymer in dehydrator 200 is optionally admixed with various 5 recycle streams from later steps of the process.

The ,,. ."~ ;"" polymer product from dehydrator 200, in which the average number of monomer units is 2-8, is passed via line 13 to second stage dehydrator 600. The feed is then heated further to effect water 10 removal and ~ ; pol~ dliu,l to a level such that the average number of monomer units is 8-25. It is preferred that the second stage dehydrator be operated at a reduced pressure nût lower than about 10 mm Hg in order to permit effective water removal at a lower t~."~,....tu.~.

The water vapor from second stage dehydrator 600, which contains very small amounts of entrained lactic ~id, is removed overhead and passed via line 1~ to separator 300 in which the separated water and lactic acid are disposed of the mar~ner described above with respect to the first dehydrator stage overhead.
The c;m~l~n-qti.~n polymer from the second stage dehydrator ~00 is also optionally ZO admixed with various recycle streams from later steps of the process.

The ~ , polymer (oligomers) from the second dehydrator stage is passed via line 17 to pre-heater 700 in which it is admixed with liquiddepolylllc.i~..iull catalyst which has been provided via line 19 from catalyst supply tank 800. The admixture of catalyst and oligomer is heated to about 21 SCand passed via line 21 into the upper end of cracking vessel 900 onto the topmost of a series of ~ perforated frqrti~nqtinn trays extending down the vessel, each having liquid dUV~ ,UIII~ connecting to the next lower tray in the series.

It is noted that the surfaces of the process lines both to and from the dehydrator vessels 2ûO arld 600, which come into contact ~ith the lactic acid feed and ,'.. 1~ .,- -~;.. polymer, are constructed of low ferrous metals m order to avoid the formation of corrosion products which might later cause ~licr~ r,qtinn of lactide product and also to reduce s;de reactions which would adversely afJ'ect the 35 purity of the product from the process of the invention.

Within the cracking vessel 900, thennally induced d. ~ol~ll.. .~Liun (cracking) of the oligomer to form cyclic diester takes place and WOg6106092 ~ t ~ s 6:3 the reaction mL.cture therefrom is f I. Each of the trays is equipped withheatiny coils ~AI-A53 which provide heat of reaction to the feed and enable the t~ alu~G of liquid on the trays to be controlled. In particular, by providing heat for the thermal cracking reaction within the cracking unit instead of preheating the 5 oligomer feed, the Ihll~, t~ G exposure of the feed and product is minimi_ed, ~ ,lc side reactions are lessened and the yield of lactide from the cracking unit is thereby m~imi7P~ By controlling the heat input to each trayaDd by adding more heat to the lower trays, the desired degree of cracking can be controlled throughout the cracking vessel.
A stream of heated nitrogen gas is introduced into the cracking unit 900 below the lowermost r. ~.. 1;l ,,, ~ ;i .. tray via line 23 to effect tnorough stripping of the du~ ulllh~t5 liquid. The vapor overhead, containing nitrogen gas, water, lactic acid and lactide vapors and small amounts of entrained heavy 15 oligomers, is remo~ed overhead through line 25 at a rate such that the av,erage residence time of the lactide vapor within the high t~ u~laLluG Gll~ ul~ of the cracking unit is less than about 15 seconds and preferably no more than 10 seconds.

Each of the h ~ trays in the cracking UDit 900 is equipped with a ser sing device Bl -B5 which measures the pressure drop between the top and bottom of each tray. This provides an indirect ~ of lactide viscosity and conversion which can be used for process control of the thermal cracking operation.
The liquid bottoms in the cracker, v. hich COIIsists mainly of heavy oligomers and complexes, is cooled by a series of cooling coils (A-6) within thebottom of the cracking vessel to a t~,...t,~ .~ltU~ below about 190C and the cooled liquid is removed from the cracker via line 27 at a rate such that the average residence time of the heavy ends within the cracking unit is less than about 45 minutes amd preferably less than 20 minutes. The heavy ends are removed from the process through line 29 or they are recycled via line 27 to the hydrolysis unit 500 andlor to one of the dehydrator stages 200 and 600 via lines 31 and 3 3 respectively. Materials recycled to hydrolysis unit 500 are passed v;a line 35 to scrubber 1400 from which the scrubbed l.~l.ul~ u is fed via line 37 either via line 39 to the first dehydrator stage 100 or removed from the process via line 41 to storage vessel 1500.

W0 96/06092 ~ ~ ~ 7 6 4 ~ 3 ~ h'ith reference to Figure Ib, the cracking unit overhead vapor is fed through line 25 to condensor 1000 in which the non-r~ ,.l ,Je ~Y I. I Ip. l. l "~ such as nitrogen, carbon monoxide, carbon dioxide, lactic acid, light ~1 v~ r.~
products and water are separated from entrained lactide. The t~ aiL~ and 5 pressure of conderlsor 1000 are carefully controlled to maximize the . ,.".1. . ,~ ;. "~
of lactide and minimize the c....-1 .~ of vater vapor.

Non-, I ' ' from condensor 1000 are passed via line 45 to scrubber 1100 in which they are scrubbed with lactic acid to remove residual 10 amounts of liquids in the stream. Scrubbed non~ , which comprise mainly nitrogen gas and small amounts of water vapor, are fed via line 47 to dryer 1200 in which water is removed from the nitrogen gas via line 49. The dry nitrogen gas is then fed via line 51 to heater 1300 in which the dry gas is heated to 150-250C and recycled via line 23 to the oligomer cracking vessel 900. Makeup 15 nitrogen gas is supplied to the process from nitrogen storage vessel 1350 via line 52 from which it is mixed with recycled nitrogen gas in line 51.

The condensate from condensor 1000, which is comprised of lactide and minor amounts of heavy oligomers and ., .~ t lactic acids, is 20 removed from the condensor via line 43 and fed to vacuum distillation column 1400. Optionally, liquid heavy ends from the distillation operation may be admixed with tbe condensate feed to the distillation column 1400. Distillation column 1400 is preferably a packed column in which light ends are taken overhead, ~ o. S d lactide is removed as a liquid side stream and liquid heavy 25 ends are removed as bottoms to vacuum flashing unit 1500.

As described above, the liquid heavy ends, which are comprised largely of heavy unreacted oligomers and small amounts of partially formed polymers, are heated and subjected to vacuum in order to flash off lighter 30 oligomers and other residual materials. The vacuum tlashing overhead from unit 1500 is then passed through line 55 and admixed with lactide feed to distillation column 1400. The residual heavy ends from the vacuum flashing unit are then recycled via lines 57 and 58 to hydrol~rsis unit 500, and/or dehydrator units 200 and 600 via line 60 and 62 ~ ly . Residual hea~ y ends can be removed 35 from the process through line 64.

The CUII~I ' ' i' lactide sidestream is removed from distillation column 1400 through line 59 to lactide heater l900 and then through line 61 to ~ ~ q~44 w0 96106092 ~ 3 melt ~,~y~ iUI~ unit 2000 in which the lactide is pumped to the top of the Imit and flows down in a thin film which cools and forms purified crystals of lactide.
The melt ~ clL;UII operation can consist of a single such unit or it can consistof two or more units arranged for either parallel or series operation Uncrystallized lactidc residue flows to ~e bottom of the unit amd is removed from the crystallizer via line 63. Crystallizer bottoms ~uncrystallized residue) can therefore be . ~ ' ' amd recycled to the crystallizer unit by itself or ;t can be~, recycled via line 65 in admixture with ~u ,1 5 d lactide feed in heater l900.
Alternatively, the crystallizer bottoms can be removed through line 57 and recycled to hydrolysis unit 500 and/or the d~_hydl~lliull units 200 and 600.

Purified lactide is collected by melting the crystals ' ' Oll the crystallizer walls and pumping the molten lactide through line 67 to storage or directly to puly~ ,.;~tiull.
The vapor overhead from the vacuum distillatiûn column 1400, which contains lactic acid, water amd small amounts of entrained lactide. is passed through line 69 to condensor 1500 in which the ~ ..l .L ~ are removed through line 73 to cold trap 1700 and thence through line 75 to vacuum pump 1800. The condensed liquid from condensor 1500 is recycled through line 71 to either the hydrolysis unit 500 or to one of the d~ L~liul~ stages 200 and 6ûO.

In both figures I a and I b of the Drawing a number of feasible locations for measuring ~ I ;u~ and properties of materials at ~arious stages of the process have heen indicated. These locations, which have been labeled Ml through M12, can be used to monitor andlor control process variablés. A partial list of such control and/or ~ ll points is as follows:

Ml (oligomer feed) color, oligomer chain length, catalyst C~.ll....l, ~a ;, ...
M (cracker hea~ ends) color, chain length, metal content, catalyst M3 (cracker vapor~ acidity potential, lactide optical purity, entrained catalyst;

M3 (cracker vapor) acidity potential, lactide optical purity, entrained catalyst;

W0 96/06092 ~~ i 9 7 6 4 4 ~ g~s3 M4 (hydrolysate) lactide optical purity, color, water content, metal content;

M5 (lactide feed) color, water content;

M6 (condensor overhead) water content, gas f ~ ", lactide content;
M7 (scrubbed gas) gas cl ~ yu~ lactide content;

10 M8 (dried nitrogen) nitrogen flow and purity;

M9 (c~ ~ lactide) lactide flow volume, acidity potential, optical purity, color;

15 M10 (purified lactide) acidity potential, optical purity, color; content; and M 11 (crystalli~r bottoms) acidity potential, optical purity, polymer content, and 20 M12 (crystallizer and/or distillation bottoms) color, optical purity, polymer content.

T~ct Plu~

25 Col-)r M ~ lr~ Color "-- ~ of both liquids and solids were Garried out using a Minolta Chroma Meter CR-131 ~ luG-utul~d by Minolta Camera Co., Ltd., Osaka, Japan. Color l ~ , were indicated using LYa~b* color notation.

30 T R~ tiflf A~ itY: Lactide (L) acidity is measured by titration of lactide dissolved in methylene-dichloride (CH2C12) with ,I~,l~di~d sodium methoxide (NaMeO) in anhydrous methanol using ~ :. indicator. Acidity is calculated as follows:

35 Acidity (meq/lcg) = (I/gL)(mL titrant - blank) (molarity factor).

Lactide ~C~mf r AnRlysis: Lactide isomers are measured by high pressure liquid ~11l. = ayL~ (HPLC) in a column with Chiracel~ OC packing using 70/30 WO 96/06092 ~? 7 f ~f 4 r ~ ~ . 13 .

(wt.) L..~ ,vl as the mobile phase. Lactide samples are dissol~ed in t-buLylul~,LII~l ether and filtered through 0.45 I~I;WUIII~ . syringe filter before loading. A 0.5 ~LL injection is used for each sample and 25 minutes analysis tirme is allowed. The W detector is set at 220 mn. Typical retention times for lactide5 isomers are as follows:

L 14.5 mimutes D 16.6 minutes Meso 17.g minutes.
Chiracel@) is a registered trademark of Diacel Chemical Industries, Tokyo, Japanfor tris(3,5-dimethyl-1JL~ lu.~.b~llat.) of cellulose and amylose packing.

Metals Ansllysic Analyses of metals in the products of the invention were carried 15 out by optical emission ~ ,Llu:,.,u~, on-line X-ray n..~ A .,. c of inductively-coupled plasma techniques.

Oli;~omer (~h~in r Pr~h Oligomer is dissolved in 80120 (wt.) anhydrous l.l.LI..yl.,l~ ,LIulidc/l--ethanol and titrated with 0.1M NaMeO in anhydrous 20 MeOH using 1 ' 'j ' ' ' indicator as foliows:

(]) Avg. M.W. = (10,000/mL 0.M titrant)x[lloligomer wt. (~z)~

(2) Chain length ~n) = [(AYg. M. Wt. - 90)/72] + I
Ortical Pnri~ Opt;cal rotation is a function of optical purity. Therefore optical purit,v was determined ~y measuring the optical rotation of various liquids using a SR-6 l~uk~lh~l~.t~,l made by PolyScience, a Division of Preston Industries, Inc., Chicago, IL.
Parti~ll poly~ ,n~ Solution cloudiness is a quantitative measure of lactide partial pol~ ,dliul~ using a Hoch Chernical tu.bid;lu.,~l No. IS900 (Hoch Chemical, Arnes, LA).

35 ~11 Anal~sis: Divalent tin is measured by redox titration with CeI~r using pl....--llll,..l..l~ ferrous sulfate as a ~n~l...;.,. :.;- indicator. The titration is preferably carried out under nitrogen gas. The rnaterial to be analy~d is dissolved in w~.t~ c~,lv~,. A 0.02~ M titrant solution of cerium (IV) sulfate in 2 N sulfuric W096/060~2 ~ ~ 9 7 6 4 ~ ,5,'1D5'13 acid was prepared. A O.O5 M ferrous ~ sulfate solution WdS used to sl~,~ the cerium sulfate. Ten mL of the ferrous s.trrnr~ninm sulfate solution was placed in a beaker with 25 mL of methanol and 25mL of acetone. Three drops of 0.025 M l,lO l ' ' ' ~ ferrous sulfate was added as an indicator.
5 The standard was titrated with the cerium until a light lemon yellow endpoint was reached. The cerium molality was calculated from the aveMge of three titrations as follows:
Molality = (10 mL x 0.05 mole/L)/mL of Ce(lV) titer Weighed 2 g samples of Sn(lI)-containing unknowns were dissolved in 4S mL of acetone and 5 mL of H2O was added to promote stable endpoint formation. Three drops of 0.025 M ],10 1.~ t ferrous sulfate indicator were added and the mixture titrated with cerium sulfate until a light lemon yellow endpoint was 15 reached. The percent Sn(II) in the sample was calculated by the follo~ing equation:

~/O wt. Sn(lI) = Molarity of Ce x ~mL titer - blank) :Y 5.9345 Sn(lI)/Sample weigbt in grams Wat~r ~n:llysi~ The moisture (water) content of materials is measured by titration of water with iodine in the presence of S02 in a suitable base. In particular, a 10%
wt. solution of lactide in methanol/water is prepared under dry nitrogen gas. The titration is carried out using a Karl Fischer Titrator, DL18, made by Mettler 25 I~l~t~ t~ AG, Griefensee, SU. The water content of the solution is then compared witn the known water content of HPLC-gMde CHC 13 EXAMPLES

30 E~le I

This example illustrates the effect of various operating variables du~ing the d~hy~Ldtiull step(s) ofthe invention In each of a series of four test runs, eight liters of ~,ul~ ,ial 88%
wt. aqueous lactic acid (98% L) were placed in a sealed stirred tank reactor. The liquid was well agitated by means of a centrally mounted impeller operating at 850-950 r.p.m. Heat was controllably added to the agitated liquid by means of two wall mounted 840 watt electric heaters which maintained the heat input so that ~I q7644 Wo 961060g2 ~ 13 .

tho ~ C difference behveen the wail and ~he buik liquid did not exceed 15~
C and the liquid t~ llulc did not exceeci 190C.

~o speed up water removai, the reaction system was operated a 5 reduced pressure of 50-200 mm Hg. To lessen loss of light lactic acid species~ the overhead vapor Was passed through a 5 tray column in which the lactic acid species were condensed and refluxed back into the buik lactic acid liquid. Each of the test runs exhibited similar water removal rates. (See Figure 2a.) In particular, water removal rates were rapid during the first 2.5-3 hours of operation. after 10 which the water removal rate slowed down cu-.s;d._..,l,ly. Ma~imum water removal was obtained at about 4 hours.

Figure 2b from the same four test runs shows that the maximurn Lu~c of 190C. was reached in about 2.5 hours, after which the t.~ lUIc 15 remained ~ub~ lly constant throughout the remainder of the runs (3-6 hours).
In this regard, an upper limit on h,..~.,,L.luic is required to avoid .i.
and . ~ ....i, ~: ;. " . problems as well as premature formation of lactide.

Figure 2c from the same four test runs is a graphicai correlation of 20 oiigormer chain len th as a function of time during the tests. These data show that durh~g the first 2 hours, substantially oniy free water was removed from the reaction system and the average chain length of the oligomers (n) was only 2.
When the process of the invention is carried out c~ fresh lactic acid feed should be introduced into this stage.
From 2 to 4.5 hours into the run, chain growth took place umtil n was about 20. Such high chain lengtb oligomers aTe very viscous. After 4.5 hours, very iittle chain further growth took place because of the high viscosity and thus low diffusivity of such heavy oligomers. It is clear that fresh lactic acid feed 30 should not be introduced into this stage since the free water therein wouid back-hydrolyze the oligomer chains.

After about 4.5 hours, the actuai amount of water remaining in the system was very small and the viscosity of the oligomers was quite high. At this35 point it is important to minimi~ further exposure to high Ll~ lu-c and thus to reduce ~ i.. ,. mrPmi7~tionamdprematurelactideformation.

WO 96/06092 ~ ~ ~ 7 6 4 4 1 ~ 3 1~

Upon l~O~ ;llg the important effect of viscosity in lirniting residual water diffusion from the oligomers, d~,hy~lLiO" tests were performed 5 using equipment in which the diffusive surface of the oligomers was enhanced in order to facilitate further diffusion of water from the oligomers.

In particular? separate quantities of oligomer (n=6) were dehydrated at 180C under vacuum (0.7 mm Hg and 3 mm Hg) in a twin screw devolatilizer 10 and an additional quantity of the same oligomer was dehydrated in a rotary flask evaporator (RotovapTM flask evaporator) at 125 mm Hg. The former apparatus creates a very thin wiped f Im of oligomer, while the latter apparatus increases the diffusion area of the oligomer by centrifugally swirling the oligomer around theinner surf~e of the flask. Both devices erihance the diffusive area of the 15 oligomer, but the f ..~ in the devolatilizer is c~ cts~lfi~lly higher. Figure 3 is a graphical ~ . of the oligomer chain length as a function of time during these tests.

The data in Figure 3 show that the rate of chain length growth in 20 the extruder thin film reactor was several times faster than in the rotary flask evaporator. For example, after only 15 minutes in the extruder, the value of n had reached 10-17. On the other hand, the oligomer dehydrated in the rotary flask evaporator reached n=l I only after 90 minutes. These data show very clearly the~dvantage of surface l ' to reduce the time needed to obt~in a given degree of olit ................... ;,~ . and thus cu~sid~ to reduce thermal exposure of the oligomer.

It is interesting to note that the diffusion area . . .1, - ,. . . ". . ,t combined with vacuum operation of the dehydrator enabled very high molecular 30 weight oligomers (n=30-35) to be formed quite easily.

When the data from the above-described 3mm Hg run were d on the data from Figure 2c, the advantage of enhanced diffusive surface was sbown more clearly still. ~or e~ample, oligomer in which n=18 was 35 obtained in only about 0.5 hour of dehydration as compared with over 2.5 hours being required to reach the same level of ..li~..l,...;,~;.,l, without surface . I , w0 96106092 2 I q 7 6 ~ 3 - 2~; -Fxs~mple 3 A further series of ~ ~ IV . ~ d~!hydl~lliull runs was carried out at 0.7 mm lHg using the abûve described equipment in which the ku~ lLu.~; was varied from 1 80C to 200C. Figure 4 shows the results of this series of tests.

The highest molecular weight oligomers (n=30-35) were obtained at 190C, while operation at 200C yielded oligomers in which the value of n was only 20-25 . Operation at 1 80C under these conditions yielded oligomers in which 10 n was higher than 30 after one hour's operation. These data, of course, show clearly the effect of h~ ; on oligomer molecular weight due to thermal ..",~ principally d~,~ul~ iuu and lactide fûrmation. Though still lower operating pressures are technically feasible and would result in higher n values at t~lll,u~,ldhY~a above 190C, such operating pressures are not at this time 15 ennnnm;r~l for cullu~ .ial scale operations.

E2Lample ~

The various process steps which are involved in the ~ uri~Lul~, of 20 purified lactide can be difficult to control because of many side reactions which occur because of the presence of various impurities. It has been found that the generation of color and its 1ll.~. .lu~,llL is a sensitive indicator by which the effect of such irmpurities can be obsen~ed. AccordinEly, a series of tests was conducted to determine the effect of several materials on lactide color. The tests wcre 25 conducted on the materials glass, Hastelloy, AL6XN, titamium, zirconium and 304 SS. Additional tests were slso conducted to observe the effccts of time-t.,ul~..aLul~, exposure and the effects of catalyst contained in the lactide.

In these tests a l,~mll~ 311y available heat stable lactic acid is 30 placed in a sealable container. A carefully cle;med piece of the material being tested is then half-immersed in the lactic acid, the head space is blanketed with nitrogen and the container sealed. The sealed container is then maintained at 1 20C
and changes in t_e color of the lactic are observed over the course of time. In addition the material is examined by optical emission a,u~ u~ to observe 35 corrosion incurred during the 5llhrn~r~inn Results of the corrosion tests are given hl Table I below:

WO96/06092 2 1 q76~4 ~ , s;3 Table 1 Cnrrocinn of M ~- ;n r ~ntic Acid ~ Cr.rrncinn Protl.1rtc prm Material % Fe Fe Cr Ni Gla.,s 0 12 Hastelloy 4 12 4 12 These data show the distinctly greater s~ptihi1ity to corrosion of high ferrous materials in the presence of lactic acid.

Example 5 A series of laboratory . ~ was conducted to observe the effect of iron in materials of cu--~Lluclivn on the d~ lA of lactide in contact with such materials. In this series of tests, high quality L-lactide was placed in a series of carefully dried glass containers. Into these containers were placed clean 20 pieces of 316SS, AL6XN, Hastelloy and glass and the containers were inerted with nitrogen and sealed. The sealed containers were then heated at 130C for 72 hours. Upon completion of the test period, the samples were analyzed by HPLC to determine the loss of lactide from each container and the color of the lactide was measured in the manner described L~,~c;~ buvc. The following data from these 25 ~ lt~ show that the loss of lactide was related directly to the arnount of iron contained in the material of culLSll u~,liv":

Table 2 Effect of Construction Materials on Lactide Color ~at~l 1ron ContrntLactide T.ncc (~o wt.) (% wt.) Glass None 0.0 Elastelloy 4 1.6 AL6XN 48 5.9 316SS 100 13.0 2 7 971~44 L*a*b eYAminAti~ n of the discolored lactide (Figure 5) shows that the formation of dark color (L) was a direct function of the iron content of the material in which the lactide was in contact.

5 Example 6 A lactic acid oligomer having a molecular weight of 738 (n=10~
was admixed with 1.5% by ~veight high purity tin octoate and fed at a nominal rate of 35 g/minute to a single sieve tray column. The tray was made of Hastelloy C
10 and had an adjustible weir set at 0.75 inch (1.9 mm) height.

The sieve tray was 8 inches (20.3 mm) in diameter, 0.125 inch (0.32 mm) thick and perforated with 216 eYenly spaced 0.067 inch (0.17 mm) holes. Inlet feed t..~ t~C was held at 1 90C maxirnum in order to lessen 15 catalyst ~ color formation, .,~ of oligomer and lactide produced therefrom. Actual liquid tC~ iL.C~ on the tray were controlled to 210-215C Heated nitrogen gas (215C) was passed upward through the sieve tray at a rate of 140 standard cubic feet (3.96 m3) per hour (SCFH) to strip out the formed l~tide very quickly, thus reducing the potential for ,1~ ~. .., ,1 ,,,~;l ;, . partial 20 pOlrll~ dtiWI and of the reactants and products tberefrom. Data from this ~ are given in Table 3 below:

Tsbie 3 Effect of Timc and Temperature on Resction snd r Time Feed RsteAcidity Totsl Isomer Distribution (~~3W) (e/hour~ ~ L~rti~ll- (~~3 W) 1 D ~SQ

450907 71.6 95.4 - 4.6 386907 88.4 92.4 - 7.6 135 322 802 85.1 87.2 1.3 11.5 35 165 210 256 93.4 78.3 4.1 17.5 The above data show a gradual decline in the rate of formation of L
lactide coupled with a gradual increase in meso lactide formation as a function of A~ENDED StlEEr IPE~JEP

~ 1 q764~
WO 96106092 P., 11 ~ , f' ~3 - 2C~ -time resulting from ~ ~Afl~tinn of the oligu~ ,dL~ mixture. In particular, It is believed that this ~)h.,llU~ U.. is due to the oxidation of SnII to SnrV with a Iûss of electrûns which reduce and thus degrade the lactide. These data cûnfirm the desirability of ~ ii,.,e/l~,...~..,.dtul, expûsure of catalyst 5 and oligomer in the cracking step.

F.xample 7 Toobservetheeffectofalkalimetal~ ,.;. IAf;(~.l onlactide formation by depolymerization, runs were carried out comparing liullllll~ ,;dllyavailable SnII octoate catalyst containing 3,000 ppm by weight, sodium c~nfPn~in ~inn with a tin octoate catalyst uhich contained only 50 ppm by weightsodium e.. ~ ~ . Both materials were used at a ~ ~ ... ...1., ~ ;...~ of 0.5% by weight to d.~,oly".~ oligomer having a molecular weight of 1460 (n=l9) at 290-335C in a thin film reactor operated at a pressure of 10 mm Hg and 6 secondsresidence time. The data frûm this test are given in Figure 6 which shows that the high-sodium catalyst yielded substantiarly higher amounts of ~,..~. . .~hl.
c u ~ ..u . ' ;....~ of the meso enantiomer even at the lower t~ ,IaLul~s and the 20 meso content rose s~7h~ ~ntiAlly to over 20% by weight as the reaction h,~
was raised to a level of 330C. On the other hnd, the low-sodium catalyst yieldedonly about 6% weight meso lactide at 330C. These data show clearly the desirability of using catalysts having lo-v alkali metal content and ~ the d~.~)ul,y~ iUII t~.lllJ.,~..L~
pxample 8 In tbis example, a series of three depolymerization catalysts was used to show further the adverse effect of alkali metal .. 1 .. 10 ~a ;.. n on the 30 ., r , .; ,~1 ;. ., . of lactide produced thereby. Three antimony catalysts were used--SbNa lactate, SbNH41actate and SbOctoate the last of which contained only 50 ppm by weight sodium. These three catalysts were mixed with lactic acid oligomers having a molecular weight of 450 (n=6) and were heated in an inerted small batch flask reactor at 1 f0-1 90C and the resulting lactide vapors were 35 collected and condensed over a 30 minutes period. The data from these three tests are given in Table 4 below.

W096/06092 ~ i 9 ;~ 4 1 "~ 13 Table 4 Effect of Alkali Metal C-r iqrninqfinn on Meso Lactide~ Formation Catalyst Total Lactide L D Meso ~/~1 wt. o/o wt. o/q wt. ~/~ ~vt.

SbNaLactate 83.7 36.5 27.2 36.3 SbNH4Lactate 78.6 8g.6 1.3 9.0 SbOctoate 71.5 95.7 0.0 4.3 These data again show the deleterious effect of alkali metal ions, such as sodium, on the ra~rmi7Otinn of lactide produced by cracking of lactic acid 15 oligomers in the presence of metal catalysts.

Examnle 9 Using the same type of equipment as in thc Example 8, a series of 20 tests was run in which the effect of different anions was observed. In particular, the runs were conducted on lactic acid oligomer having a molecular weight of 450(n=6) at l80-l90C using catalyst ~ .ns of 0,5~~q by weight, The resulting lactide vapors were swept from the f~ask with nitrogen, collected and condensed and the condensate was analyzed. The data from this, , are given in 25 Table 5 below.

Table 5 Effect of Catalyst Anion Volatility on Lactide Prodnction 30 ('~~~lysf % wt. Lqnfi~ ~ l ~ ~/O wt. M~cn SnC12 33.1 96.2 0.0 3.6 SnSO4 76.9 86.0 7.9 6.1 Analysis of the lactides showed that the one derived by cracking in the presence of the chloride-containing catalyst had a high acid level (2,000 ppm) and the above data show a very low yield of lactide W~AS obtained. Moreover, the W096106092 2 1 ~ 7 6 ~ 4 r~-~u~ ,i,3 product was dark in color and had a high odor level. Meso level was low, probably because of its relatively high instability. However. d~ul~ aLio using the sulfate catalyst resulted in more than hvice the amount of lactide recovery, the lactide had little odor and was white in color. These data indicate 5 that non-volatile acidic species such as H~SO4, which remain in the residue, are preferable to more volatile acidic species such as HCI, which remain in the lactide overhead vapor from the cracking operation.

EYsln~rle ~0 A further experiment was carried out in which pure lactide and a small weighed piece of carefully cleaned steel wool were placed in a glass bottle in which the head space was inerted with nitrogen gas and the bottle was sealed.The bottle was then placed in a 120C oven for 72 hours, after which the bottle was 15 opened and the steel wool was washed and reweighed.

Surprisingly, reweighing the steel wool revealed that there was no weight loss nor amy change in the properties of the steel wool. Thus, there was no corrosion of the steel wool. N., V~l Lh~,l..,:" the lactide had turned brown and the lactide acidihy had increased from 1.5 meq to 166 meq. Analysis by HPLC
revealed an 8.7% weight loss in lactide. These data therefore indicate that the loss of lactide was incurred by water generated internally during reaction of the lactide catalyzed by the presence of the iron.

Claims (23)

Claims:

1. An integrated process for the manufacture of purified lactide from an aqueous solution containing at least 50% wt. lactic acid comprising the sequential steps:

A. Feeding a solution of crude lactic acid in de-ionized water to a first heated zone in which free water is removed by evaporation and a molten mass of condensation polymer containing lactic acid and a minor amount of free water is formed by condensation polymerization to an extent that the average number of monomer units (n) in the condensation polymer is 2-8;

B. Feeding the molten mass of condensation polymer from step A to at least one further heated zone in which the diffusive surface area of the polymer is increased, the residual lactic acid and condensation polymer are further condensation-polymerized to an extent that n is 8-25 and both free water and bound water are removed by evaporation, steps A and B being carried out within equipment the surfaces of which in contact with the reactants are fabricated from low ferrous materials;

C. Contacting the molten condensation polymer with an alkali metal-freedepolymerization catalyst in a cracking zone operated at a liquid temperature nohigher than 240C and pressure sufficient to effect cracking of the molten condensation polymer with the concominant formation of (1) a vaporous reaction mixture containing water, lactic acid, lactide and entrained heavy oligomers and(2) molten liquid heavy ends containing heavy oligomers;

D. Removing the vaporous reaction mixture from the cracking zone at a rate such that the average residence time of the lactide vapor within the cracking zone is less than 15 seconds;

E. Condensing the vaporous reaction mixture and vacuum fractionally distilling the condensate therefrom whereby lactic acid, water and minor amountsof lactide are removed as vapor overhead, concentrated lactide is removed as a liquid side stream and the heavy ends are removed as molten liquid; and F. Subjecting the concentrated lactide to melt crystallization by whichpurified lactide having an Acidity Potential less than 6 meq/kg of lactide is separated from a residual lactide having an acidity potential of at least 30 meq/kg.

2. The process of claim 1 in which steps A-E are carried out continuously.

3. The process of claim 1 in which step B is carried out under vacuum at a temperature no higher than 200C.

4. The process of claim 1 in which water vapor removed from the condensation polymer in steps A and/or B is scrubbed with hot lactic acid to remove residual amounts of lactic acid contained therein.

5. The process of claim 1 in which steps A and B are carried out in a packed column.

6. The process of claim 4 in which the residual amount of lactic acid removed is recycled to steps A and/or B.

7. The process of claim 1 in which the molten condensation polymer is admixed with heated depolymerization catalyst and the admixture is introduced into the cracking zone.

8. The process of claim 7 in which the polymer-catalyst admixture is comprised of 0.1-5% wt. catalyst and is heated to a temperature of 150-240C.

9. The process of claim 8 in which the catalyst is a soluble organic compound of a metal selected from Sn, Sb, La, rare earth metals and mixtures thereof.

10. The process of claim 9 in which the catalyst is comprised of finely divided particles of an insoluble metal-containing material.

11. The process of claim 1 in which the molten condensation polymer is passed through a fixed foraminous bed of depolymerization catalyst contained within the cracking zone.

12. The process of claim 1 in which the cracking step is carried out at about atmospheric pressure and a liquid temperature of at least 200C.

13. The process of claim 1 in which the cracking step is carried out under vacuum.

14. The process of claim 1 in which nitrogen gas is introduced into the vapor space in the lower end of the cracking zone to effect stripping of thevaporous reaction mixture from the heavy ends.

15. The process of claim 1 in which the molten liquid heavy ends in the bottom of the cracking zone are cooled to a temperature below 190C and removed from the cracking zone at a rate such that the residence time of the liquid heavy ends within the cracking zone is less than 15 minutes.

16. The process of claim 1 in which the molten liquid heavy ends from the cracking zone in step C are recycled to step A and/or B.

17. The process of claim 1 in which the cracking zone is comprised of a plurality of spaced perforated trays having controllable heat exchange means for heating liquid contained on the trays and downcomers for transferring liquidbetween the trays.

18. The process of claim 1 in which fractionation of the condensate in step E is carried out under vacuum.

19. The process of claim 1 in which the lactide residue from step is fed to a heated vacuum zone in which lactide is vaporized from the residue, condensed and recycled to step A, B, C or E.

20. A process for making a concentrated stream of lactide from lactic acid oligomers comprising the sequential steps:

A. Contacting molten lactic acid oligomers in which the average number of monomer units is 8-25 with an alkali metal-free depolymerization catalyst in a cracking zone operated at a temperature no higher than 240C and pressure sufficient to effect cracking of the molten oligomers with the concomitant formation of (1) a vaporous reaction mixture containing water, lactic acid, lactide and entrained heavy oligomers and (2) molten cracker bottoms containing heavy oligomers;

B. Removing the vaporous reaction mixture from the cracking zone at a rate such that the average residence time of the lactide vapor within the cracking zone is less than 15 seconds;

C. Condensing the vaporous reaction mixture and fractionally distilling the condensate therefrom whereby lactic acid, water and minor amounts of lactide areremoved as vapor overhead, concentrated lactide is removed as a liquid sidestream and the condensate heavy ends are removed as molten liquid;

D. Cooling the molten cracker bottoms to below 190C and removing them from the cracking zone at a rate such that the average residence time of the molten cracker bottoms in the cracking zone is less than 15 minutes; and E. Subjecting the cooled molten cracker bottoms to either dehydration or hydrolysis followed by dehydration and recycling the resultant dehydrate to the cracking zone.

21. The process of claim 20 in which the concentrated lactide from Step C is subjected to melt crystallization by which purified lactide having an acidity potential less than 6 mek/kg of lactide is separated from a residual lactide having an Acidity Potential of at least 30 meq/kg.

22. A method for making oligomers of lactic acid which are suitable for depolymerization to form lactide comprising the sequential steps:

A. Feeding a solution of crude lactic acid in de-ionized water to a first heatedzone in which free water is removed by evaporation and a molten mass of condensation polymer containing lactic acid and a minor amount of free water is formed by condensation polymerization to an extent that the average number of monomer units (n) in the condensation polymer is 2-4; and B. Feeding the molten mass of condensation polymer from step A to at least one further heated zone in which the diffusive surface area of the polymer is increased, the residual lactic acid and condensation polymer are further condensation-polymerized to an extent that n is 8-25 and both free water and bound water are removed by evaporation, steps A and B being carried out within equipment the surfaces of which in contact with the reactants are fabricated from low ferrous materials.

23. A method for making poly(lactic acid) comprising:

A. admixing ring-opening catalyst with purified lactide made by the process of claim 1 or claim 21; and B. heating the admixture under pressure to a temperature of 100-200C to effect formation of poly(lactic acid) by ring-opening polymerization of the lactide.