Unlted States Patent [19] [11] Patent Number: 4,835,293

Bhatia [45] Date of Patent: May 30, 1989

[54] ATMOSPHERIC PRESSURE PROCESS FOR 3,784,585 l/ 1974 Schmitt et a1. .................... .. 260/861
PREPARING PURE CYCLIC ESTERS 4,727,163 2/1988 Bellis ................................. .. 549/267
[75] Inventor: Kamlesh K. Bhatia, Newark, Del. FOREIGN PATENT DOCUMENTS
[73] Assignee: E. I. Du Pont De Nemours and [European Pat 05- ---------- -- 549/518
W' ' , , apan .
Company’ llmmgton Del 1108720 4/1968 United Kingdom ..
21 A 1. No.: 17 418
[22] FT: F’b 24 1987 OTHER PUBLICATIONS
[ 1 1e ' e ' ’ Abstract of EP 47473 (03/82).
[51] Int° Cl.4 .......................................... .. CO0Z 319/00
[52] US. 01. .................................................. .. 549/274 Primww Examiner—Glennon H- Hollrah
[58] Field Of Search .............................. .. 549/274, 518 Assistant Examiner-Mark W- Russell
. Attorney, Agent, or Firm-Craig H. Evans
[56] References Cited
Us. PATENT DOCUMENTS [57] ABSTRACI
1095 205 5/1914 Gruter et a1. ................ .. 549/274 This invention relates to an improved Process for PR‘
2:163:109 6/1939 Spangel ......... __ 260/338 paring highly pure cyclic esters by heating a polymer of
2,163,268 6/1939 carothers er a1‘ __ 260/333 the corresponding a-hydroxy acid or its ester or a co
2,668,162 2/1954 Lowe ................ .. 549/274 polymer of a-hydroxy acid or its ester and a thermally
292K111“ et ai~ - stable polyether in the presence Of an inert gas at atmo
, , C mltt et a _ .. . ' ' ' '

3,457,280 7/1969 Schmitt et a1. .... .. 260/3402 Spher.‘° préjlslutr?‘ .The cychc esterl ‘5 Camed from the
3,597,449 8/1971 Deprospero at al' 549/274 reaction w1 e Inert gas to a so vent system.
3,597,450 8/1971 Schmitt et a1. .... .. 549/274
3,763,190 10/1973 Ross et a1. ......... .. 19 Claims, 1 Drawing Sheet
US. Patent May 30,1989 4,835,293

FIG.1
 1
4,835,293
2
and evolves water when polymerized to make the start
ATMOSPHERIC PRESSURE PROCESS FOR ing polymer. Any residual water is believed to react
PREPARING PURE CYCLIC ESTERS with the cyclic ester to form the linear dimer of a
hydroxy acid. The crude glycolide must be reheated to
DESCRIPTION OF THE INVENTION 5 remove it from the vessel that it condenses into. Impuri
1. Technical Field ties are then extracted in a suitable solvent and the eye
This invention relates to an improved process for lic ester is re?ned by one or more recrystallizations.
preparing highly pure cyclic esters by cracking a poly Considerable yield loss occurs during the puri?cation
mer of the corresponding o-hydroxy acid or its ester or steps.
a block copolymer of the corresponding a-hydroxy 10 The process of the present invention is distinguished
acid or its ester and a thermally stable polyether in the from these prior processes in that it can be run at atmo
presence of an inert gas at atmospheric pressure. The spheric pressure or above. It comprises cracking a poly
cyclic ester is carried from the reaction with the inert mer of a-hydroxy acid or its ester or a block copolymer
gas to a solvent system. made from a-hydroxy acid or its ester and a thermally
2. Background and Summary of the Invention 15
stable polyether (this polymer or block copolymer is
Cyclic esters of the general formula also referred to as prepolymer) in intimate contact with
an inert gas to form the cyclic ester. The cyclic ester
R10 vapors are carried from the cracking vessel by the inert
1 ll gas and are brought into intimate contact with a solvent
20
that preferentially removes impurities from the cyclic
O 0
ester and the product is recovered in an easily handle
able form that for most applications requires no further
if?“
0 R2
puri?cation. The process may be run continuously or
25
batchwise.
wherein R1 and R; can be independently hydrogen or The process of the present invention offers numerous
an aliphatic hydrocarbon having from 1 to about 6 advantages over the processes of the prior art. It elimi
carbon atoms, are a useful class of compounds that can nates the need for high vacuum equipment and atten
be polymerized into high molecular weight polymeric dant plugging and safety problems. The process is safer
plastic materials particularly useful in medical applica 30 to operate since the presence of inert gas eliminates the
tions such as wound closure devices, orthopedic im potential for explosive atmospheres that can result from
plants, and controlled release drug carriers. These ap air leaks in a vacuum process. Product removal is sim
plications require that the cyclic esters be highly pure so pli?ed since the cyclic ester forms into solid particles
that they can be polymerized into long chain polymers. dispersed in the solvent. Remelting is not needed. Me
In the past, these cyclic esters have been prepared by 35 chanical separation techniques such as ?ltration and
?rst condensing the corresponding a-hydroxy acid to a centrifugation can be employed. Also, since the impuri
brittle polymeric form of the a-hydroxy acid. For ex ties dissolve in the solvent, the mechanically separated
ample, if the desired product was glycolide, glycolic cyclic ester can for most applications be dried and used
acid would be condensed to form a brittle polymer, without the added puri?cation steps of the prior art.
polyglycolic acid. The polymeric material would then 40 The resulting product even without the solvent reslur
be ground to a ?ne powder and added slowly to a rying and recrystallization steps of the prior art is a pure
heated, evacuated vessel in which it would be depolym white, crystalline product having a low acid content.
erized to a crude material which had to be subjected to The process permits running the process of US. Pat.
an extensive and costly puri?cation operation. This No. 4,727,163 with less expensive technical grade 70%
process suffered from excessive tar formation, low 45 hydroxyacetic acid to produce glycolide without puri?
yields, and slow production rates. One attempt to im cation by repeated recrystallizations. The process can
prove upon that thermal cracking process and to pre also be used to produce pure cyclic esters without ?rst
pare relatively pure glycolide is described in US. Pat. forming the initial copolymer of US Pat. No.
No. 3,763,190 to Donald L. Ross. That process required 4,727,163. Higher yields can thus be obtained since
?rst making a salt of an O-haloacetylglycolic acid and product removal from the reactor is easier and the
then heating the salt to a suf?cient temperature under solids handling and solvent reslurrying steps of earlier
vacuum to effect ring closure. Mineral salts had to be processes are not needed. Also, as a result of high inter
removed and the resulting glycolide separated and puri facial area in the reactor facilitated by the intimate
?ed by sublimation. contact with the inert gas, cracking time is not expected
A recent development claimed in US. Pat. No. 55 to increase as reactor size is increased. The process thus
4,727,163, is a thermal cracking process that does not should be capable of being carried out on a larger scale
involve formation of a salt of a halogenated a-hydroxy then in the past. A continuous process is possible with
acid. That process involves ?rst making a block poly this atmospheric pressure process.
mer comprising a thermally stable polyether core with
a-hydroxy acid or its ester polymerized onto the core. 60 DETAILED DESCRIPTION OF THE
Upon heating under vacuum conditions, the chain ends INVENTION
are thermally degraded to form a cyclic ester which can The process of this invention can use the block co
be condensed under vacuum. The condensed cyclic polymer of US. Pat. No. 4,727,163 as its feed or it may
ester is a crude solid containing from about 1 to 12 use a polymer of the a-hydroxy acid. This block co
‘ weight percent acidic impurities. Other color-forming 65 polymer or polymer may be prepared from either the
oily or waxy impurities are also present. This is particu a-hydroxy acid or its ester. This block copolymer or
larly the case when starting with technical grade hy polymer is alternatively referred to herein as prepoly
droxyacetic acid which contains other acidic impurities mer. The a-hydroxy acid is of the form R1R2C(OH)
 4,835,293
3 4
COOH, wherein R1 and R2 can be independently hy The scrubber (4) contains a solvent in which the
drogen or aliphatic hydrocarbon groups having from 1 impurities are preferentially soluble over the cyclic
to 6 carbons. The preferred a-hydroxy acids are gly , ester at the temperature of operation. The proper sol
colic acid and lactic acid. The ester of the a-hydroxy vent can be determined by one skilled in the art based
acid is of the form R1R2C(OH)COOR3, wherein R1 and on knowledge of the solubilities of the impurities and
R2 are de?ned as for the a-hydroxy acid and R3 is an the cyclic ester. Preferably, the solvent should be one in
aliphatic hydrocarbon group having froml to 6 car which impurities are highly soluble and the cyclic ester
bons. The preferred esters are methyl and ethyl glyco is signi?cantly less soluble. Preferable solvents contain
late and methyl and ethyl lactate. When the starting polar groups to dissolve the acid impurities. Preferably,
acid is glycolic acid or the starting ester is methyl or the solvent would be chosen from the group comprising
ethyl glycolate, the resulting cyclic ester is glycolide. alcohols, ketones, ethers and esters and most preferably
When the starting acid is lactic acid or the starting ester would be isopropyl alcohol. By operating the scrubber
is methyl or ethyl lactate, the resulting cyclic ester is at a temperature lower than the freezing point of the
lactide. cyclic ester, preferably at room temperature (20° to 45°
The reactor and scrubbing equipment employed can C.), the cyclic ester will be solid particles dispersed in
be chosen from any suitable gas-liquid-contacting de the solvent. As such, the cyclic ester product may be
vice known in the art. For example, the reactor may be easily separated by mechanical means such as ?ltration
a sparged stirred tank, a bubble column, a spray reactor or centrifugation and the solvent may be treated as
or a film reactor. A reactor such as a sparged stirred waste, recycled to the scrubber or used in another pro
tank or a bubble column is preferred. Scrubbing equip 20 cess, depending on economic and environmental con
ment may be a sparged stirred tank, a spray column or siderations. ‘

a bubble column. When using a solvent in which the solubility of the

FIG. 1 depicts a typical reactor and scrubber ar cyclic ester is such that a signi?cant amount of the
rangement for a batch process. cyclic ester is retained in the solvent with the impuri
One skilled in the art can easily modify the batch 25 ties, the cyclic ester can be recovered by fractional
process into a continuous process. crystallization.
Referring to FIG. 1, the prepolymer is introduced to Suitable temperatures for the cracking step range
a well-mixed reactor (1) or is formed in the reactor and from about 215° to 290° C. with the preferred tempera
is heated to a suf?ciently high temperature to crack the ture range being from about 230° to 265° C. and the
prepolymer. In FIG. 1, the assuring that the reactor 30 most preferred range being 240“ to 255° C. There is
contents are well mixed may be used. Cracking in the theoretically no upper limit on the pressure at which the
reactor is carried out at about atmospheric pressure or process is run. Practical considerations, however, make
above. During this cracking, an inert gas, preferably it preferable to keep the pressure to a minimum since
nitrogen, is fed through line (2) which is positioned in higher pressures require a proportionately larger quan
the reactor so as to assure intimate contact with the tity of inert gas and result in a lower concentration of
prepolymer. Preferably the inert gas is preheated to the cyclic ester in the inert gas. Preferably, the pressure will
reaction temperature. The inert gas is injected below be from about 0 pounds per square inch gauge (psig.) to
the surface of the material in the reactor. In FIG. 1, the about 10 psig. Most preferably, the pressure will be
point of addition is shown to be below the agitator. The from about 0 psig. to about 5 psig. Operating at a slight
?ow rate of inert gas is set sufficiently high so as to not 40 vacuum would be within the equivalents envisioned so
limit the rate of production. If the inert gas ?ow is too long as the pressure in the scrubber is above the satura
low, the yield may be adversely effected and the rate of tion pressure of the solvent at the temperature of scrub
production will be limited since the inert gas is needed ber operation.
to carry the cyclic ester vapors out of the reactor. To run the process continuously, the reactor will be
Higher inert gas rates will not hinder the reaction but 45 periodically or continuously purged to remove tars and
will require larger downstream equipment and will fresh prepolymer will be added. The purge will either
result in higher cost the inert gas ?ow rate should be 5 be treated as waste, recycled or used in another applica
to 30 standard cubic feet per minute (scfm) per cubic tion depending on economic and environmental consid
foot of prepolymer and most preferably 10 to 20 scfm erations. The solids collected in the solvent will be
per cubic foot of prepolymer. The inert gas and the 50 periodically or continuously discharged and the solvent
cyclic ester, vapors exit the reactor through heated line will be replenished.
(3). The inert gas and the cyclic ester enter a well-mixed
scrubber (4) below the solvent in the scrubber. The EXAMPLES
cyclic ester preferably enters the scrubber as a vapor In order that the concept of the present invention
but may be partially liquid. The scrubber in FIG. 1 is 55 may be more fully understood, the following examples
equipped with a condensor (5) and a dry ice trap (6) to are set forth. They are intended for illustration and not
condense any solvent that is vaporized when the hot limitation.
cyclic ester and the inert gas are fed to the scrubber.
The dry ice trap would be replaced by a suitably sized Example 1
condensor in a larger scale facility. The scrubber is 60 A prepolymer of hydroxyacetic acid (glycolic acid)
equipped with an agitator to assure good contact of the on a thermally stable polyether core was prepared using
solvent with the incoming cyclic ester. Other means of the teachings of US. Pat. No. 4,727,163, using 70%
assuring that the scrubber solution is well mixed may be technical grade hydroxyacetic acid. Three hundred
used. Periodically or at the end of a production run the ninety grams (390 gms) of 70% technical grade hydrox
contents of the scrubber are discharged through line (7) 65 yacetic acid, 390 gms of Terethane 2000 ®, 7.4 gms of
to a ?lter (8) wherethe cyclic ester is collected as the calcium oxide and 2 gms of antimony oxide (SB2O3)
?lter cake and the solvent containing the impurities as catalyst were placed in a nominal 1 liter stirred reactor
the ?ltrate. (cylindrical reaction ?ask) equipped with a sparger
 5
4,835,293
6
positioned to permit the feeding of the nitrogen below applying a vacuum at a temperature of 195° C. Calcula
the level of the agitator. The reactor was heated ?rst at tions based on methanol recovered indicated that 47
atmospheric pressure to remove most of the water and weight % of the starting methyl glycolate was con
then at 10 mm mercury until no more Water was seen verted to the prepolymer.
condensing indicating that prepolymerization was com The prepolymer was then transferred to a nominal 1
plete. The temperature during this polymerization was liter reactor where it was heated in intimate contact
raised to 195° C. and controlled at that temperature with N2, as described in Example 1. Nitrogen flow was
until the prepolymerization was complete. controlled at 0.35-0.4 scfm. Temperature was con
After the prepolymerization was complete, a solvent trolled at between 250°-265° C. The reaction was
scrubber equipped with an agitator and containing 600 stopped after 4 hours. The starting isopropyl alcohol in
cc of isopropyl alcohol was connected to the reactor as the scrubber (as described in Example 1) was the ?ltrate
shown in FIG. 1. The temperature of the isopropyl from a prior experiment. Its starting temperature was
alcohol at the beginning of the run was room tempera room temperature and, as in Example 1, no attempt was
ture. No attempt to control temperature during the run made to control its temperature during the run. It was
was made. Dry nitrogen (N2) ?ow to the reactor was 5 not decanted at the end of the run as in Example 1. The
started at a rate of about 0.35 standard cubic feet per product slurry was ?ltered under a nitrogen blanket as
minute (scfm) and the reactor temperature was in in Example 1. The ?lter cake weighed 236 gms. The
creased gradually to about 255° C. A glass heat ex ?ltered product was pure white and crystalline. After
changer was used to prevent freezing of the glycolide in several days of storage at room temperature, the prod
the line to the scrubber. The reactor temperature during 20 uct was washed with isopropyl alcohol to remove any
the run ranged from about 250°‘to 260° C. The glycolide surface moisture that might have been picked up during
was recovered in the solvent scrubber as a slurry. storage and was ?ltered. The ?lter cake was found to
The reaction was stopped after 3.5 hours although contain no detectable acidic impurities.
glycolide was still coming over. Upon stopping the N2
?ow and the solvent scrubber stirrer, white crystalline Examples 3 to 12
glycolide particles quickly settled in the scrubber. The Examples 3 to 11 were run according to the proce
solvent layer, slightly discolored with yellowish impuri dure of Example 1, starting with 500 gms. of 70 weight
ties, was decanted off. The product was mixed with 200 % technical grade hydroxyacetic acid, 500 gms. Tere
cc of fresh isopropyl alcohol and discharged as a slurry thane 2000 ®, 9.5 gms. calcium oxide and 2.56 gms.
to a ?lter. Nitrogen was blown over the ?lter cake to antimony oxide in a nominal 1 liter stirred reactor. The
minimize atmospheric water contact. The ?ltration nitrogen ?ow during the cracking step was about 0.35
continued until no isopropyl alcohol droplets were scfm.
observed coming out of the ?ltercake under vacuum Example 12 was run according to the procedure of
conditions. The wet ?lter cake (containing from 10 to Example 1, starting with 5000 gms. of 70 weight %
15 weight % isopropyl alcohol) weighed 110 gms. The 35 technical grade hydroxyacetic acid, 5000 gmsl Tere
weight percent of acidic impurities (free acids as hy thane 2000 ®, 95 gms. calcium oxide and 25.6 gms.
droxyacetic acid) in the ?ltered product were measured antimony oxide in a nominal 12 liter stirred reactor
by dissolving l gramof the filtered product in 100 cc of (spherical reaction flask). The agitator used in this ex
a 1:1 mixture of methanol and acetone and titrating the ample was the same agitator as used in the l-liter exam~
solution with a standard sodium methoxide solution 40 ples. Thus, the inert gas contacting was not as effective
using an automatic titrator. The weight percent of as could have been obtained with a larger agitator. The
acidic impurities was found to be 0.03%. A portion of isopropyl alcohol condensor was also the same as that
the ?lter cake was dried by simply passing N2 through used in the l-liter experiments and could not handle the
it for 1% hours at room temperature. Gas chromato approximately 10 times greater gas rate that would be
graphic analysis of a portion of the dried product dis 45 expected. Cracking had to be interrupted several times
solved in acetone showed about 0.06% isopropyl alco during the experiment due to the condensor being
hol (IPA) in this dried product. The dried product was ?ooded. Nitrogen flow had to be limited to 2.5 to 3.0
characterized by Differential Scanning Colorimetry scfm. With a suitably sized condensor, increased pro
(DSC) and Nuclear Magnetic Resonance Spectroscopy duction rate would be expected.
(NMR). The DSC showed a sharp melting curve with a 50 The free acid content of the ?lter cake from each
peak at 85.3° C. and a melting point of 82.2° C., indicat example, measured as in Example 1, was determined
ing high purity. The proton NMR would predict that, and is shown in Table 1. Also shown is the weight of
except for the residual IPA from drying, the product ?lter cake and the cracking time of each experiment.
was more than 99.8% glycolide.
55 Example 13
Example 2
Filter cake from each of Examples 3 to 12 was stored
The prepolymer in this example was prepared from in capped glass bottles, which had been purged with dry
methyl glycolate and a thermally stable polyether. One nitrogen, in a refrigerator for up to several days. Part of
thousand grams (1000 gms.) methyl glycolate, 520 gms. the ?lter cake, wetted with isopropyl alcohol, from
of Terethane 2000 ®, 14.8 gms. calcium oxide and 4.9 60 Experiment 3 (50 grams) and all the ?lter cake, wetted
gms. Sb2O3 catalyst were heated in a separate ?ask (not with isopropyl alcohol, from Experiments 4 through 12
shown in FIG. 1) at atmospheric pressure to remove were combined. The total weight of the combined ?lter
methanol. Methanol was removed by means of a frac cake after certain sampling losses was 1,995 gms. To
tionating column so as to retain the methyl glycolate in facilitate mixing so as to assure a homogeneous compos
the reaction mass. The reaction temperature was main 65 ite slurry and to wash away any surface moisture picked
tained at 185° C. This polymerization reaction contin up in storage and handling, an approximately equal
ued until no more methanol was observed coming out. weight of isopropyl alcohol was added and the resulting
The unreacted methyl glycolate was then removed by mixture was agitated. The resulting slurry was ?ltered
 4,835,293
7 8
in two batches to remove the isopropyl alcohol. The polymer so as to create a large interfacial area between
acidities of the ?lter cakes from these batches were 0.01 the prepolymer and the inert gas, to evolve the cyclic
and 0.03 weight %. The ?lter cake was then dried in ester vapors.
four batches each at 50° C. and 5 mm mercury in a 2. The process of claim 1 comprising the further step
rotary vacuum drier for 3 hours. The acidities of the of scrubbing the cyclic ester vapors and inert gas in a
dried batches were 0.03, 0.013, 0.02 and 0.016 weight solvent.
%. The dried product contained no detectable isopro 3. The process of claim 1 wherein the temperature is
pyl alcohol by gas chromatographic analysis. DSC from about 215° to 290° C. and the pressure from about
analysis of the dried product showed a sharp melting 0 psig. to about 10 psig.
curve with a peak at 83.2“ C. and a melting point of 10 4. The process of claim 1 wherein the cyclic ester is
82.6° C. The purity of the glycolide was 99.92 mole %. glycolide or lactide.
5. The process of claim 3 wherein the temperature is
TABLE 1 from about 230° to 260° C. and the pressure is from
Nominal about 0 psig. to about 5 psig.
Reactor Filter Cracking 6. The process of claim 4 wherein the temperature is
Size Cake Free Acid Time from about 240° to 255° C. and the pressure is from
Exper. (liters) (grams) (weight %) (hours)
about 0 psig. to about 10 psig.
3 l 141 0.02 4
4 l 140 0. 12,003 3.5
7. The process of claim 1 wherein the inert gas ?ow
5 I 145 0.063 3.25 20 is 5 to 30 scfm per cubic foot of prepolymer.
6 1 135 0.065 3.5 8. The process of claim 4 wherein the inert gas ?ow
7 l 142 0.035 3.25 is 10 to 20 scfm per cubic foot of prepolymer.
S 1 141 ' 0.061 3.25
9 1 135 0.037 3.5 9. The process of claim 7 wherein the inert gas is
10 l 147 0.039 3.25 nitrogen.
ll l 165 0.044 3.5 10. The process of claim 8 wherein the inert gas is
12 12 825 0.0390056 4
nitrogen.
11. The process of claim 2 comprising the additional
I claim: step of separating the cyclic ester from the solvent.
1. An improved process for making a highly pure 12. The process of claim 2 wherein the solvent is a
cyclic ester of the form 30 polar solvent in which the impurities are more soluble
than the cyclic ester.
13. The process of claim 11 wherein the solvent is a
polar solvent in which the impurities are more soluble
than the cyclic ester.
O O 35 14. The process of claim 12 wherein the cyclic ester is
c-c-a, essentially insoluble in the solvent.
II I 15. The process of claim 12 wherein the solvent is
isopropyl alcohol.
wherein R1 and R2 can be independently hydrogen or 40 16. The process of claim 13 wherein the temperature
an aliphatic hydrocarbon having from I to about 6 car of the solvent is lower than the freezing point of the
bon atoms by cracking a prepolymer comprising a poly cyclic ester.
mer of an cL-hydroxy acid or its ester or a block copoly 17. The process of claim 16 wherein the temperature
mer of an a-hydroxy acid or its ester on a thermally of the solvent is between about 20° and 45° C.
stable polyether wherein the improvement comprises 18. The process of claim 17 wherein the cyclic ester is
cracking the prepolymer at a suitable temperature and separated from the solvent by mechanical means.
at a pressure at or above atmospheric pressure while 19. The process of claim 18 wherein the mechanical
maintaining a suf?ciently large ?ow of an inert gas, the means is ?ltration or centrifugation.
* it * ill it
inert gas being in intimate contact with the liquid pre
50

55

60

65