WWW.POLYMERCHEMISTRY.

ORG ARTICLE

Synthesis and Properties of Poly(butylene succinate): Efficiency of

Different Transesterification Catalysts

Nicolas Jacquel,1,2,3 Floriane Freyermouth,1,2 Françoise Fenouillot,1,2

Alain Rousseau,1,2 Jean Pierre Pascault,1,2 Patrick Fuertes,3 René Saint-Loup3
1
Université de Lyon, F-69003, France
2
INSA-Lyon, Ingénierie des Matériaux Polymères, IMP, CNRS UMR5223, F-69621, Villeurbanne, France
3
Roquette Frères, Division Recherche, F-62080 Lestrem, France
Correspondence to: A. Rousseau (E-mail: alain.rousseau@insa-lyon.fr)

Received 22 July 2011; accepted 14 September 2011; published online 5 October 2011
DOI: 10.1002/pola.25009

ABSTRACT: The efficiency of organometal- (Ti, Zr, Sn, Hf, and are less reactive during the transesterification reaction. The
Bi) and metal oxide- (Ge and Sb) based catalysts was investi- role of catalytic residues during the storage and processing of
gated during the transesterification step of the synthesis of pol- PBS was also studied. The reduced viscosity of the PBS sam-
y(butylene succinate) (PBS). PBS was prepared from succinic ples did not vary when processing at 190  C from 1 to 10
acid and 1,4-butanediol via a two-stage melt polycondensation minutes. However, when stored under ambient conditions, all
process. The catalytic efficiency of the organometal catalysts PBS samples were prone to significant hydrolytic degradation,
was as follows: Ti
Zr Sn>Hf>Sb>Bi. The germanium and especially those containing a titanium catalyst. This behavior
antimony metal oxides displayed desirable catalytic efficiency indicates that zirconium- and germanium-based catalysts could
when were associated with hydroxy acids (lactic acid or gly- be interesting substitutes for titanium-based catalysts. V
C 2011

colic acid), which acted as chelating agents. However, this cata- Wiley Periodicals, Inc. J Polym Sci Part A: Polym Chem 49:
lytic system exhibited lower efficiency compared to the 5301–5312, 2011
titanium system. Furthermore, at high concentrations of
hydroxy acids the overall transesterification rate decreased. KEYWORDS: catalysis; metal oxide; organometal; polybutylene
This effect can be explained by the substitution of PBS succinate; processing; synthesis
hydroxyl end groups by a lactic or glycolic unit, both of which

INTRODUCTION The accumulation of plastic waste and the over, 1,4-butanediol (BDO), the other monomer used in PBS
increase in oil prices over recent years has led to the ‘‘redis- synthesis, could be easily obtained by the reduction of SA,
covering’’ of several polymers (some biodegradable) that thus giving access to a fully bio-based PBS.3,4 Considering its
may be produced from renewable resources. With adequate mechanical properties, ductility, and desirable processing
assessment of the sustainable (or green) character of bio- characteristics, PBS could be a very interesting substitution
based polymers, these materials could give the polymer for polyolefins in some applications.
industry the ability to face criticism concerning petroleum- Nevertheless, the preparation of high molar mass aliphatic
based plastics. For this purpose, ‘‘old’’ polyesters (e.g., ali- polyesters from dicarboxylic acids and diols still remains dif-
phatics) are gaining increased interest. This is the case for ficult because of their low thermal stability.5 Researchers
poly(3-hydroxyalkanoates), which are synthesized directly by have, in part, been focused on the study of efficient catalysts.
microorganisms, and even more for poly(lactic acid), which It is well known in polyester polycondensation, especially for
is obtained from glucose fermentation and ring-opening po- poly(ethylene terephthalate) (PET) synthesis, that titanium
lymerization of the corresponding lactide dimer. The latter is compounds display the highest catalytic activity.6,7 However,
the more commercialized bio-based polyester of these two due to the high amount of side reactions and the yellowing
examples. Recent developments in fermentation technology generated by titanium-based catalysts, most commercially
have made it possible to produce new bio-based monomers. available aromatic polyesters are produced with antimony,
Succinic acid (SA) is one of the top twelve bio-based chemi- germanium oxides or a mixture of metals.8 These oxides are
cals that are produced from sugars and have a potential widely studied in the context of PET synthesis, but only a
market. This compound appears particularly interesting for few studies deal with the production of PBS.9,10 In fact, PBS
the production of poly(butylene succinate) (PBS).1,2 More- synthesis is mainly carried out with titanium (IV)

V
C 2011 Wiley Periodicals, Inc.

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312 5301
 ARTICLE WWW.POLYMERCHEMISTRY.ORG

isopropoxide,11–13 titanium (IV) isobutoxide14 or titanium alkoxides or carboxylates, such as titanium (IV) n-butoxide,
(IV) n-butoxide.15–17 Recently, new catalysts have been zirconium (IV) n-butoxide, tin (II) 2-ethylhexanoate, anti-
developed that are based on rare-earth metals, such as scan- nomy (III) n-butoxide, hafnium (IV) n-butoxide, or bismuth
dium [Sc(OTf)3 or Sc(NTf2)318–20]. These catalysts are capa- (III) neodecanoate, is presented. The activities of neat metal
ble of achieving the bulk polycondensation of methyl ester of oxides, such as germanium oxide (GeO2) or antimony oxide
SA with BDO under mild conditions (around room tempera- (Sb2O3), were also investigated. To increase the ability of
ture, 35  C). Moreover, these catalysts can be recovered and these oxides to catalyze PBS transesterification, a-hydroxy
reused by solubilization in chloroform and extraction with acids were added as chelating agents.
water. Polycondensation times at such temperatures remain
very long (>50 h), and the final number average molar mass To our knowledge, only a few publications have reported the
of Mn, ranging from 5000 to 12,000 g mol1, was relatively use and comparison of tin, antimony, hafnium, germanium,
low. Stannic components, such as distanoxannes5 or tin or zirconium compounds for PBS synthesis. These catalysts
salts,21 have also been studied for the synthesis of PBS in so- were compared in terms of reaction rates, and their effects
lution. Although relatively high molar masses were obtained on polymer characteristics were also analyzed.
(Mn ¼ 117,000 g mol1),5 the use of solvent and the long
reaction times (24–80 h) did not make this a favorable pro- EXPERIMENTAL
cess. More recently, Kircheldorf et al. reported the use of bis- Materials
muth-based compounds for the synthesis of aliphatic poly- Petrochemical-based succinic acid (SA, 99.72%) and 1,4-
esters from dimethyl ester and aliphatic acids22 using butanediol (BDO, 99.3%) were respectively purchased from
succinic anhydrides23 or from the direct polycondensation of Gadiv Petrochemical Ind. (Haifa, Israel) and Sigma-Aldrich
dicarboxylic acid.24 The values of Mn up to 30,000 g mol1 (France). Organometal catalysts used in this study were
were obtained after the bulk polymerization at 80  C for
(Fig. 1): (a) titanium (IV) n-butoxide (99%, Acros Organics),
over 48 h.24 Bismuth salts appeared to be particularly inter-
(b) zirconium (IV) n-butoxide (80 wt% in 1-butanol,
esting because of their very low toxicity characteristics.24
Aldrich), (c) tin 2-ethylhexanoate (95%, Sigma), (d) antin-
Industrially, several strategies have been proposed and are omy (III) n-butoxide (65 wt% in toluene, Aldrich), (e) haf-
reported in patents and articles. The most common strategy nium (IV) n-butoxide (99%, Aldrich), and (f) bismuth neode-
is to use an efficient catalyst, such as titanium, in a relatively canoate (technical grade, Aldrich). Metal oxides were
high amount (Ti ¼ 100–360 ppm, Sb ¼ 350–450 ppm, and germanium (IV) oxide (99.99%, Aldrich) and antimony (III)
Sn ¼ 450–1500 ppm)25,26 to limit the exposition of the poly- oxide (Honeywell). DL-Lactic acid (LA, 90% in water) and
mer to high temperatures as much as possible. This method glycolic acid solutions (GA, 70% in water) were respectively
prevents degradation inside the reactor and thus the colora- purchased from Fluka and Aldrich. All these chemicals were
tion of the material. However, the obtained polymer often used as received without further purification.
displays low thermal stability characteristics.13 To avoid
processing problems caused by these large amounts of General Procedure for PBS Synthesis
degrading metal atoms, some patents suggest the use of PBS samples were synthesized via a two-step melt polycon-
chain extenders, which is a different strategy. In these meth- densation reaction (Fig. 2). This reaction was carried out in
ods, the relatively low molar mass polyesters obtained after a 7.5-L stainless steel batch reactor equipped with a heating
a long transesterification stage (9 h at 190–210  C to reach system, a mechanical stirrer with torque measurement, a dis-
Mn  25,000 g mol1) using a low proportion of catalyst tillation column, a vacuum line and a nitrogen gas inlet.34
(10 ppm of Ti) are extended using diisocyanates to reach A typical operating procedure for this experiment was as
an Mn of  70,000 g mol1.27 Other coupling agents, such follows. The reactor was charged with 1889 g (16 mol) of
as diphenylcarbonates,28,29 biscaprolactamates30 or cyclosila- SA and 1513.7 g (16.8 mol) of BDO. The molar proportion of
zane,31 have been suggested. The molar masses from Mn ¼ [COOH]/[OH] groups was fixed at r ¼ 0.95, which corre-
30,000 to 70,000 g mol1 have been obtained, but most of sponds to a molar excess of butanediol of 5 mol %. The reac-
these modified polymers lose their desirable biodegradability tion mixture was then heated to 225  C (4  C min1) under
characteristics, which is an effect, that is, sometimes unde- a 2 bar nitrogen pressure and stirred at a constant speed
sired. The most promising alternative to chain extenders is (150 rpm). A mixture of water containing 1.5–2 mol % of
to select catalysts that display high efficiency combined with tetrahydrofuran (THF) produced by the dehydration of buta-
low degradation. Toward this aim, authors have recom- nediol was distilled off from the reactor. The proportion of
mended the use of germanium oxide.10,32 This metal is well THF formed was systematically measured and did not vary
known in PET synthesis for its good catalytic efficiency and
much depending on the nature of the catalyst. The formation
desirable thermo-oxidative stability.33 However, the low
of THF indeed modifies the stoichiometry of the reaction.
natural abundance of this element (0.0001%) makes it quite
The esterification yield was estimated with the quantity of
expensive to use.
water collected. In the second step, the pressure was
This study was initiated to better understand the polymer- reduced to 0.7 mbar in 120 min, and the temperature was
ization process of PBS, with a special focus on transesterifi- raised to 230  C to initiate the transesterification. Low-pres-
cation catalysis. Polymerization steps in this work were sure conditions were maintained until the desired torque, as
carried out in a pilot scale reactor. The efficiency of metal measured by the stirrer in DC (in N m), was reached.

5302 JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312
 WWW.POLYMERCHEMISTRY.ORG ARTICLE

FIGURE 1 Organometal catalysts used in this study.

A polymer strand was then withdrawn from the bottom drain were determined with a calibration method using poly-
valve of the reactor and quenched in a water bath. Crystalline methylmethacrylate standards. It is known that such method
PBS pellets were obtained after granulation. Using such a pro- overestimates the molar mass and does not permit to give
cedure avoids contact of the warm polymer with oxygen, thus access the absolute value. The data are used only for com-
reducing coloration and thermo-oxidative degradation. parison purpose. The reduced viscosity (gred) of the poly-
mers was determined with an automated Ubbelohde capil-
Characterization lary at 25  C. Polymer samples were dissolved at 135  C for
The polymer molar mass was assessed by size exclusion 15 to 25 min in a mixture of phenol/ortho–dichlorobenzene
chromatography (SEC) in 1,1,1,3,3,3-hexafluoro-2-propanol. (50:50 wt %, Aldrich France) at a concentration of C1 ¼ 0.5
Samples of 1 g L1 were eluted at a flow rate of 0.75 mL g L1. Subsequently, the reduced viscosity was calculated
min1. The signals were then detected with a RI detector according to eq 1 (t0 and ts refer to neat solvent and poly-
(Agilent-RI-1100a). The average molar masses (Mn and Mw) mer-solvent solution flow times, respectively):

FIGURE 2 General scheme of PBS synthesis from SA and BDO (a) esterification, (b) transesterification.

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312 5303
 ARTICLE WWW.POLYMERCHEMISTRY.ORG

ts  t0 The 1H NMR analysis of PBS was carried out with a 400

gred ðmL g1 Þ ¼ (1)
t 0  C1 MHz Bruker liquid state NMR spectrometer equipped with a
QNP probe. Deuterated chloroform was used as solvent.
The carboxylic acid end-groups content (CC) was determined
by a titration method using a Mettler Toledo automatic titra-
RESULTS AND DISCUSSION
tor. A value of 1.0 g of the polymer was dissolved in chloro-
form and then titrated with a NaOH solution in methanol Reaction Studies
(CNaOH ¼ 0.02 mol L1). From our experiments, it appears that the preparation of the
catalyst and its addition into the polymerization media has a
Veq ðmLÞ  CNaOH ðmol L1 Þ large impact on the efficiency of the transesterification step
CC ðleq =gpolym Þ ¼  103 (2)
mpolymer ðgÞ of polyesters. As the catalyst preparation depends on its
type, especially due to its sensitivity to water, the present
The coloration of the polymer was measured on polymer study of the reaction will be divided into two parts: (1) a
plates of comparable thicknesses (3 mm) with a DYK Gard- discussion of the efficiency of organometals and (2) an inves-
ner TCS II spectrophotometer (observation angle ¼ 10 , illu- tigation into the ability of metal oxides to catalyze the trans-
mination D65). The results were expressed using a yellow- esterification of PBS. The quantity of catalyst used for the
ness index (YI). Yellowness Index is a number calculated polycondensation was expressed in ppm of metal present in
according to ASTM E313 from spectrophotometric data that the final polymer.
describes the change in colour of a test sample from white
For practical reasons and to avoid the addition of oxygen
through to yellow for positive values or to blue for negative
during polymerization, we chose to not open the reactor to
values.
collect samples. In this study, torque variations measured at
The thermal stability of the synthesized polymers was eval- the stirrer were used to monitor the reaction evolution. This
uated using a DSM 15 twin-screw microextruder. Before method enabled a good comparison of the transesterification
extrusion, polymer pellets were dried for at least 5 h at 80 rates, assuming that the polymer chains were linear.

C to avoid hydrolytic degradation. The polymer was then
extruded at 190  C for different time periods, including 1, 2, Efficiency of Organometal Compounds
5, and 10 minutes. The degradation of extrudates was quan- It is well known that metal alkoxides in the presence of
tified by measuring the reduced viscosity. water form condensed species and agglomerates, which
Thermogravimetric analysis (TGA) was performed on a TA strongly affect the activity of these catalytic species.35,36 The
Instrument Q500 thermobalance. The samples were heated efficiency loss of titanium tetraalkoxides in water during the
from 25 to 500  C at a heating rate of 10 C min1 under a synthesis of PBS was confirmed by our own experiments, as
nitrogen atmosphere (10 mL min1). The 10% weight loss shown in Figure 3. In these two runs, titanium (IV) butoxide
temperature, Td10, was used to compare the thermal stability (400 ppm of Ti equivalent) were introduced before and after
of the different polymer samples. the esterification step. The results show an important
decrease in the reactivity when the catalyst was added
The thermal properties of the PBS were measured on a TA before esterification. The torque displayed an increase of
Instrument Q20 differential scanning calorimeter (DSC). The 16.6 Nm in 180 min when the catalyst was added after
polymer sample was first heated from 80 to 150  C (10  C esterification, whereas only 2.2 N.m were obtained in
min1), the temperature was then held for 2 min before 340 min for the other case.
cooling to -80  C (10  C min1). Temperature was also held
for 2 min before a second heating in the same conditions as
in the first step. The glass transition of the polymer was
taken at midpoint during the first heating. The melting tem-
perature, due to complex melting behaviour of PBS, was
taken at peak maximum during the first heating scan. The
melting enthalpy was calculated from the integration of that
last peak and by taking as a baseline reference the heat flow
curve after 120  C. The crystallinity (vc) of PBS was calcu-
lated according to the following equation:

DHm
vc ð%Þ ¼
DHPBS
0

where DHm is the melting enthalpy of the sample (J g1),

DHPBS
o
is the melting enthalpy of the fully crystalline PBS: FIGURE 3 The evolution of torque during transesterification
DHPBS
o
¼ 210 J g1.15 The crystallisation temperature with titanium (IV) n-butoxide (Ti ¼ 400 ppm) added before (l)
was measured at peak minimum point during the cooling and after (n) esterification (transesterification temperature ¼
scan. 230  C, butanediol excess ¼ 5 mol %).

5304 JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312
 WWW.POLYMERCHEMISTRY.ORG ARTICLE

Thus, due to the strong sensitivity of metal alkoxides to the assay was abandoned after 6.2 h of transesterification.
water, special attention was taken in the preparation of the For the other metals, the catalyst concentrations were
catalysts. First, organometals were diluted at 200  C in 30 g increased to get a torque variation value approaching 15 N
of butanediol. Subsequently, the mixture was added to the m (e.g., zirconium 1200 ppm, tin 900 ppm). Moreover, it was
reactor during the decompression step when the pressure not possible to obtain such a torque variation using catalysts
reached 20 mbar. based on antimony (1000 and 2000 ppm), hafnium (400
ppm) or bismuth (1000 ppm). Regarding degradation, zirco-
The curves depicted in Figure 4 show the influence of the
nium induced the least amount of yellowing and had the
various organometal contents on the rate of torque increase.
lowest carboxylic acid end-groups content, CC (see Table 1).
Among all catalysts tested in this study, titanium was defi-
nitely the most active catalyst. It appears that the fastest From the polymer molar mass measurements given in Table
transesterifications were observed for polyesters prepared 1, it appears that the polymer polydispersity index increased
with 200 to 400 ppm Ti. However, for 100 ppm Ti at 230  C, with tin (II) 2-ethylhexanoate content. This result suggested
that tin (II) 2-ethylhexanoate reacts differently from the
other components, in which the polydispersity index remains
constant between 1.8 and 2.1. Kricheldorf et al.22 reported
that the use of Sn(Oct)2 as a catalyst for the synthesis of
symmetric aliphatic polyester, poly(ethylene adipate) and
poly(ethylene sebacate), leads to a large fraction of cycles.
Polymer characteristics gathered in Table 1 show that car-
boxylic acid end-groups content increased with the quantity
of catalyst contained in the polymer. This indicated that deg-
radation was favored for samples containing higher catalyst
residues, regardless of the nature of the catalyst. Figure 5
gives a summary of the reduced viscosity variation for the
different catalysts studied as a function of the metal content.
The reduced viscosity variation in mL/g/min was obtained
by dividing the reduced viscosity measured on the final poly-
mer by the total duration of the transesterification step; this
value was chosen to highlight the trends. The efficiency of
metal-based catalysts was ranked as follows: Ti
Zr  Sn
> Hf > Sb > Bi.

Efficiency of Metal Oxides

In contrast to what is commonly observed for the synthesis
of PET, neat metal oxides, such as germanium oxide (GeO2)
or antimony oxide (Sb2O3), do not show sufficient activity to
synthesize PBS. The addition of hydroxy acids, especially lac-
tic acid (LA), is necessary and described in several patents
from Mitsubishi Chemical.32,37 An older Mitsubishi patent
reported the addition of other hydroxy acids, such as oxalic,
citric, or tartaric acid, that enhance the solubilization of
GeO2 in glycol by forming complexes.38
The use of lactic acid and other hydroxy-carboxylic acids has
been reported in several publications39,40 and patents41–43
with titanium chelates. These catalysts appear to be less
effective than alkoxides, but have the advantages of being
water stable and soluble in ethylene glycol at room tempera-
ture.39 Moreover, due to its convenience in use, this lactic
acid titanium chelate is produced on a commercial scale by
DuPont under the trade name TyzorVLA. In our case, this
R

technique was applied for the preparation of germanium and

antimony oxide catalytic systems.
FIGURE 4 The evolution of torque during transesterification for Germanium oxides solutions were prepared by dissolution in
various catalysts, (a) titanium (IV) n-butoxide, (b) zirconium (IV) boiling water, and then various amounts of lactic acid solu-
n-butoxide, (c) tin (II) 2-ethylhexanoate (transesterification tem- tion were added to the catalytic mixture. To reproduce the
perature ¼ 230  C, butanediol excess ¼ 5 mol %). conditions of dissolution before the addition of lactic acid,

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312 5305
 ARTICLE WWW.POLYMERCHEMISTRY.ORG

TABLE 1 Characterization of PBS Prepared with Various Organometal Catalysts (Transesterification Temperature 5 230 8C,
Butanediol Excess 55 mol %)

gred
gred variation Mn Yellowness
Catalyst ppma (mL g1) (mL g1 min1)b (g mol1)c PDId CC (leq g1)e index

Ti(OBu)4 100 99 0.27 21,400 2.0 23 5.0

200 209 0.60 45,100 1.9 39 8.1
300 198 0.79 35,400 1.9 78 7.9
400 190 1.05 31,700 2.1 86 10.5
Zr(OBu)4 200 109 0.29 – – 28 7.0
600 179 0.48 44,100 2.0 43 1.5
800 193 0.52 52,900 1.8 48 1.2
1,200 205 0.56 47,700 2.1 53 2.7
Sn(Oct)2 300 126 0.34 32,000 2.3 25 83.8
600 151 0.41 37,100 2.8 36 108.2
900 174 0.47 42,900 3.3 43 140.9
Sb(OBu)3 1,000 89 0.24 – – 17 0.2
2,000 75 0.20 – – 15 1.3
Hf(OBu)4 400 96 0.26 – – 25 5.1
Bi(ODec)3 1,000 65 0.18 – – 18 1.4
a c
ppm of metal present in the final polymer. Measured by SEC in HFIP with PMMA standards.
b d
Reduced viscosity of the final polymer reported on the transesterifica- Polydispersity index.
e
tion time. Carboxylic acid end-groups content.

the quantity of water used for the preparation was kept con- result was obtained when GeO2 was directly added into the
stant at 100 g of water for 1 g GeO2. The solution was added reactor without any further preparation. As a comparison,
to the monomers. This additional water was distilled off dur- using the same preparative conditions, germanium oxide
ing esterification. powder displayed a high efficiency in the transesterification
of PET.33
First, we noticed that lactic acid has a clear activation effect
on germanium dioxide (see Fig. 6). A comparison between The other data (Fig. 6) show the catalytic activity of increas-
the experiments performed with 600 ppm Ge was made ing quantities of a mixture made of germanium dioxide dis-
with and without lactic acid. In the absence of lactic acid, solved in a lactic acid solution containing a ratio of LA/Ge ¼
polymerization proceeded very slowly (see also the Table 2 105 mol %. The torque tended to increase linearly with time
sample with 600 ppm Ge and no lactic acid). The same in these results. However, for our previous work on metal
alkoxides, the torque increased much faster in comparison

FIGURE 5 Reduced viscosity variation during the transesterifi-

cation step for PBS synthesized with various catalysts. The
filled markers correspond to organometal compounds, and FIGURE 6 Evolution of torque during transesterification cata-
empty markers refer to the optimal combination of metal oxide lyzed by germanium-lactic acid mixtures: LA/Ge ¼ 105 mol %
with hydroxy acid (LA: lactic acid/GA: glycolic acid) (transesteri- (transesterification temperature ¼ 230  C, butanediol excess ¼5
fication temperature ¼ 230  C, butanediol excess ¼ 5 mol %). mol %).

5306 JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312
 TABLE 2 Properties of PBS Transesterified with Different Catalytic Mixtures

gred
Metal HA/Met HA feed HA content gred variation Mn CC Yellowness
(ppm)a (mol %) ratio (mol %) (mol %)b (mL g1) (mL g1 min1)c (g mol1)d PDIe (leq g1)f index

GeO2/lactic acid mixtures (Met ¼ Ge and HA ¼ LA)

100 105 1.2 1.9 97 0.26 26,800 1.8 46 4.0
200 52 1.2 1.6 118 0.32 35,700 1.8 21 2.9
105 2.4 5.1 173 0.47 48,100 1.9 27 3.8

WWW.MATERIALSVIEWS.COM
157 3.6 4.9 181 0.49 51,000 1.9 27 0.0
WWW.POLYMERCHEMISTRY.ORG

209 4.7 – 170 0.46 49,800 1.9 34 0.9

400 26 1.2 1.7 97 0.35 – – 25 1.2
52 2.4 3.5 173 0.47 – – 27 0.3
105 4.7 – 161 0.43 42,400 1.9 36 2.8
600 0 0.0 0 53 0.20 10,800 2.0 86 2.4
26 1.8 2.5 61 0.47 – – 30 0.8
52 3.6 5.2 126 0.49 – – 34 6.9
105 6.9 8.9 182 0.43 42,700 1.9 32 6.1
GeO2/glycolic acid mixtures (Met ¼ Ge and HA ¼ GA)
200 48 1.1 – 79 0.21 – – 34 7.2
96 2.2 – 97 0.26 30,700 1.8 23 6.2
193 4.4 – 138 0.37 – – 27 3.6
289 6.4 – 154 0.42 – – 29 2.6
Sb2O3/lactic acid mixtures (Met ¼ Sb and HA ¼ LA)
400 87 2.4 – 98 0.26 31,000 1.8 19 7.0
1000 0 0 0 92 0.25 – – 15 8.3
35 2.4 – 122 0.33 – – 24 9.3
87 5.8 – 123 0.33 38,000 1.8 32 26.4

Catalysts: GeO2, Sb2O3. Hydroxy acids (HA): lactic acid (LA), glycolic acid (GA) (transesterification temperature ¼ 230 C, butanediol excess ¼ 5 mol %).
a
ppm of metal present in the final polymer.
b
Polymer hydroxy acid content evaluated by 1H NMR spectroscopy.
c
Reduced viscosity of the final polymer reported on the transesterification time.
d
Measured by SEC in HFIP with PMMA standards.
e
Polydispersity index
f
Carboxylic acid end-groups content.

JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312

ARTICLE

5307
 ARTICLE WWW.POLYMERCHEMISTRY.ORG

With Ti(OBu)4, no activity enhancements were supposed to

occur in the presence of hydroxy acids. The objective was to
check if the insertion of lactic acid in the PBS chains slowed
down transesterification. Initially, lactic acid had a slightly
positive effect. However, the torque evolved linearly and
reached a maximum. The sample without lactic acid pro-
duced a torque result that increased much faster (Fig. 8).
This trend was also observed when the experiment was car-
ried out with the same amount of GA (6.9 mol %). These
experiments confirm that the incorporation of hydroxy acids
at concentrations of a few mole percent in the PBS chains
significantly slows down transesterification.
Finally, another metal oxide was tested in association with
FIGURE 7 Reduced viscosity increased during the transesterifi- lactic acid. In comparison with germanium, antimony was
cation step for PBS synthesized with different LA/Ge and GA/ found to be far less effective. This observation was consist-
Ge molar ratios (transesterification temperature ¼ 230  C, buta- ent with our organometal study, in which we showed that
nediol excess ¼ 5 mol %). antimony (antimony butoxide) had a very low activity for
PBS synthesis (Fig. 9). Nevertheless, with lactic acid it was
possible to reach a reduced viscosity of 122 mL g1 with
1000 ppm of Sb (Table 2); whereas, only 89 mL g1 was
(exponential shape). On the other hand, the activity obtained with the same Sb content using antimony butoxide
increased with mixture content for quantities of this mixture (Table 1).
with < 200 ppm Ge. Surprisingly, for higher Ge concentra-
tions, this activity decreased. This effect must be understood, As for germanium, the activity of antimony increased the
as more catalyst frequently accelerated the reaction. It transesterification efficiency for a lactic acid content of 2.4
should also be noted that when the germanium content mol %. The transesterification efficiency did not increase fur-
increased, the lactic acid content increased as well. Thus, it ther for higher lactic acid contents (e.g., 5.8 mol %).
is possible that a high proportion of lactic acid has a nega- As summarized in Figure 5, it appears that the optimal ger-
tive impact on the transesterification rate. This behavior was manium/hydroxy acid mixtures displayed a relatively good
confirmed by a systematic study carried out on a series of efficiency compared with the organometal compounds stud-
LA/GeO2 mixtures. Figure 7 depicts the efficiency of the ied in the previous portion of this work. Comparison of the
transesterification (represented by the reduced viscosity reduced viscosity variations over the transesterification step
variation during the transesterification step) for catalytic showed that the germanium/hydroxyl acid mixtures are
mixtures with variable germanium oxide and lactic acid even more efficient than zirconium and tin-based catalysts.
proportions.
For a given quantity of germanium, the addition of lactic Discussion on Metal Hydroxy Acid Catalytic Mixtures
acid increases the catalytic activity until a maximum amount To our knowledge, no publications have reported and
of activity is reached, after which the activity starts to explained how the chelation of germanium (or other metals)
decrease. To understand this behavior, it is useful to mention with hydroxy acids could modify the transesterification of
that lactic acid is indeed incorporated in the PBS chains dur- polyesters. From our experiments, both the lactic and
ing the polymerization, thus forming a PBS-co-LA copolymer.
Depending on the conditions, 1 to 7 mol % of lactic acid
units may be inserted. This result was assessed by 1H NMR
spectroscopy (Table 2). The hypothesis is that lactic acid
plays a positive catalytic role at low concentrations, while it
is responsible for the reduction of the reaction rate when
added at high concentrations. To elucidate this, both compo-
nents of the catalytic system were varied: lactic acid was
replaced by glycolic acid (GA) and GeO2 was replaced ini-
tially by Ti(OBu)4 and then by Sb2O3.
Figure 7 presents the results for 200 ppm of germanium and
variable proportions of glycolic acid. The substitution of lac-
tic acid by glycolic acid also showed a reduction in reactivity.
For a constant amount of germanium (Ge ¼ 200 ppm), the
maximum activity was achieved with glycolic acid contents FIGURE 8 Torque variation during transesterification for the
greater than 6.4 mol %; whereas, maximum activity was synthesis of PBS (n), P(BS-co-6.9 mol % LA) (l) and P(BS-co-
reached around 2.4 mol % for lactic acid. 6.9 mol % GA) (~). Catalyst: Ti(OBu)4; Ti ¼ 200 ppm.

5308 JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312
 WWW.POLYMERCHEMISTRY.ORG ARTICLE

decrease of the overall rate of chain growth as observed

experimentally. Indeed this hypothesis needs to be assessed
experimentally.

Polymer Properties
The polymer sample yellowness indexes were measured for
all samples, and the results are shown in Tables 1 and 2. As
shown in Table 1, tin-based catalysts led to the obtention of
highly colored polymers, going from yellowness (Sn ¼ 300
ppm; YI ¼ 83.8) to dark brown (Sn ¼ 900 ppm; YI ¼
140.9), despite their desirable activity. On the contrary,
experiments carried out with zirconium displayed very low
coloration characteristics. From the metal oxide study (Table
2), germanium appeared to be the least colored metal, de-
FIGURE 9 Comparison of viscosity increases during the trans-
esterification step for PBS synthesized with LA/Sb2O3 and LA/
spite the small increase in yellowness index that was
GeO2 (transesterification temperature ¼ 230  C, butanediol observed with increasing lactic acid content. This increased
excess ¼ 5 mol %). coloration caused by hydroxy acid is consistent with that
reported in the literature.41
It is known that discoloration results from the formation of
poly(ene) from vinyl groups formed by decomposition of
glycolic acids clearly had an important activating effect on polyester during the polycondensation reaction. Chung40
GeO2 and Sb2O3 catalysts. Nevertheless, larger amounts of studied the acid–base catalytic properties of metal com-
hydroxy acids tend to reduce the transesterification rate, as pounds in the preparation of PET and he related the forma-
shown in this study. tion of color in the final polyester to the reduction potential
of the catalytic metal ions. The reduction potential for Ti and
From experiments carried out with lactic acid, it was initially
Sn are near zero (Ti4þ/Ti3þ: 0.04V and Sn4þ/Sn2þ: 0.07V)
assumed that transesterification was limited because of the
and the final polymer synthesized using Ti(i-C3H7O)4 and
lower reactivity of the secondary hydroxyl group of lactic
Sn(CH3COO)2 exhibited high coloration from yellow to brown
acid when compared with the two primary hydroxyl groups
respectively. He also mentioned that the semiconducting
of butanediol. It is well known that primary alcohols are typ-
properties of a metal species affect the color characteristics
ically better nucleophiles than secondary alcohols due to
of the final polyester. In general, the insulator group (e.g., Zr-
their higher nucleophilicity. However, experiments carried
and Ge-based catalysts) does not induce any color formation
out with glycolic acid (a primary hydroxy acid) displayed a
on the contrary the semi conductor group (e.g., Ti-, Sb-, Sn-
similar reduced transesterification rate for higher
based catalysts) results in discoloration. Even if the work of
concentrations.
Chung concerns the PET, our work on PBS shows the some
Another explanation can be given by considering PBS trans- trends.
esterification reaction schemes. Since butanediol is in
Table 3 summarizes the thermal properties of the series of
excess, the PBS oligomers formed after the esterification
PBS synthesized with various catalysts. For PBS synthesized
step have a majority of hydroxyl chain-ends (hydroxyl/car-
with Ti, Zr and Sn based catalysts, glass transition tempera-
boxylic acid ratio is around 3 from our experiments). In
tures, crystallization temperatures, melting temperatures and
Figure 10(a) the mechanism of a typical transesterification
degree of crystallinity do not change much. However copoly-
reaction involving hydroxyl-terminated PBS and leading to
mers synthesized with GeO2/lactic acid based catalysts are
chain growth is shown (efficient trans esterification). It
statistical copolyesters which exhibit decreasing of both
consists in the nucleophilic attack on the carbon bearing
melting temperature (from 110  C for 2.4 mol % of lactic
ester function by the oxygen of a hydroxyl chain end.
acid to 102  C for 6.9 mol % of lactic acid), crystallization
Indeed other trans reaction pathways exist in the reactive
temperatures (from 56  C for 2.4 mol % of lactic acid to 29
medium but many do not lead to molar mass increase 
C for 6.9 mol % of lactic acid) and degree of crystallinity
(inefficient transreactions). Similar pathways have been
(from 31% for 2.4 mol % of lactic acid to 26% for 6.9 mol
applied to PBS that contains lactic acid units at the extrem-
% of lactic acid). The glass transition temperature remains
ity of the chain [Fig. 10(b,c)]. Lactic acid terminated chain
unchanged (approximately 30 C).
ends produce inefficient transesterification which do not
participate to the polymer chain growth [Fig. 10(b)]. The Thermo-gravimetric analysis of the PBS samples (Table 3),
reaction leads to an exchange of the lactic acid between reported as temperatures at 10% of weight loss (Td10),
the two polymer chains. Moreover, when transesterification showed that increasing quantities of catalysts are favouring
occurs on the middle of two (or higher), consecutives lactic polymer degradation in some cases. The thermal stability of
acid units as shown in Figure 10(c), two ‘‘dead’’ chain-ends the resulting polymers is not affected by the amount of Ge
similar to the one given in Figure 10(b) are obtained. The based catalyst (from 200 to 600 ppm of Ge) and a variation
two transesterification pathways described above involve a of 7 to 12  C for Td10 is observed when Ti (from 200 to 400

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312 5309
 ARTICLE WWW.POLYMERCHEMISTRY.ORG

FIGURE 10 PBS transesterification schemes involving lactic acid chain ends.

TABLE 3 Thermal Properties of PBS Prepared by Using Various Organometal Catalysts

at 230 8C with a Butanediol Excess of 5 mol %

LA/PBS gred Tg Tc Tm vc Td10

Catalyst ppmb (mol %) (mL g1) ( C) ( C) ( C) (%)c ( C)d

Ti(OBu)4 200 0 209 31 66 114 32 355

300 0 198 30 64 115 30 351
400 0 190 32 71 116 34 343
Zr(OBu)4 600 0 179 33 71 116 32 353
800 0 193 31 – 114 32 350
1200 0 205 31 69 114 32 351
Sn(Oct)2 300 0 126 32 66 117 33 347
600 0 151 31 68 115 34 343
900 0 174 33 67 114 33 340
LA/GeO2a 200 2.4 173 30 56 110 31 364
400 4.7 161 30 42 106 29 363
600 6.9 182 29 29 102 26 356
a
For a lactic acid/Ge molar ratio of 105.
b
ppm of metal present in the final polymer.

c
Degree of crystallinity calculated with DHPBS ¼ 210 J g1
d
Temperature recorded at 10 wt % loss in N2 atmosphere.

5310 JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312
 WWW.POLYMERCHEMISTRY.ORG ARTICLE

TABLE 4 Reduced Viscosity of PBS Samples Measured In our study, we noticed that this hydrolytic degradation was
Before and After Extrusion at 190 8C Under Air Atmosphere strongly dependent on the type of catalyst. As shown in Fig-
(Transesterification Temperature 5 230 8C, Butanediol ure 11, the reduced viscosity of samples stored under ambi-
Excess 5 5 mol %) ent conditions showed a significant decrease. This variation
was especially prominent for PBS containing titanium cata-
gred (mL g1) after lysts residues. This topic will be further analyzed in a forth-
various extrusion times
coming paper.
Catalyst ppmb 0 min 1 min 10 min
CONCLUSIONS
Ti(OBu)4 200 163 158 160
In this study, the use of various types of catalysts (either
400 155 155 152
organometal or metal-oxide compounds) was reported for
Zr(OBu)4 600 176 184 182
PBS synthesis. For these two types of catalysts, appropriate
1200 189 199 196 preparation procedures were set. Due to their sensitivity to
Sn(Oct)2 300 128 128 126 water, organometals were added after esterification, whereas
900 155 150 144 metal oxides were introduced with monomers at the begin-
GeO2a 200 170 170 169 ning of the reaction. The ability of metal oxides to catalyze
400 160 158 156 PBS transesterification was clearly improved by the presence
of hydroxy acids. Nevertheless, when the concentration of
a
For a lactic acid/Ge molar ratio of 105.
b these compounds became too high, the efficiency of the cata-
ppm of metal present in the final polymer.
lytic mixture began to decline. To gain a better understand-
ing of these effects, additional experiments should be carried
ppm of Ti) or Sn (from 300 to 900 ppm of Sn) are used.
out by substituting the metal and its ligand. As similar ten-
From these degradation temperatures it appears that PBS
dencies have been obtained, it seems that this decrease in ef-
prepared with germanium was more stable. On the other
ficiency is linked to the quantity of hydroxyl acid added. An
hand, the PBS prepared with tin catalyst was more sensitive
explanation of this phenomenon could be elucidated by con-
to thermal degradation. This increase in polymer degradation
sidering various transesterification reaction schemes. In fact,
with the catalyst content is consistent with the observations
the insertion of hydroxy acid units in PBS chains decreases
made by Yang et al.13 who carried out DSC and TGA meas-
the proportion of ‘‘active’’ hydroxyl PBS end-groups that play
urements on PBS synthesized with various titanium isoprop-
a dominant role in transesterification. In summary, with
oxide contents ranging from Ti ¼ 460 to 27,000 ppm (1/
regard to the reduced viscosity variation during the transes-
600 to 1/10 in catalyst: SA molar ratio).
terification step, the following metal ranking was elucidated:
To assess the thermal stability of PBS synthesized with the Ti > Ge > Zr  Sn > Hf > Sb > Bi.
different catalysts, extrusions were carried out at 190  C
Despite the fact that Ti-based catalysts are highly efficient,
under air. As shown in Table 4 the PBS extrudates collected
they should be used with caution due to the important deg-
at various times (from 1 to 10 min), did not show significant
radation reaction that they induce. That was particularly out-
changes and only a small decrease of the reduced viscosity
lined by the carboxylic acid end-groups contents, yellowness
was observed for the sample containing 900 ppm of tin.
indexes and reduced viscosities, which were measured upon
As reported by Kim et al.,44 PBS is prone to hydrolytic degra- ageing. From these aspects, zirconium- and germanium-
dation. Even if this aspect could be an advantage in some based catalysts appeared to be interesting substitutes for the
cases, it is clearly a problem for more durable applications. commonly-used titanium.
The authors thank Roquette Frères company at Lestrem in
France for providing funding in the context of the BioHub pro-
gram (Oséo).

REFERENCES AND NOTES

1 Caulier, B.; Laurent, S. (Roquette Frères). WO Patent
2,010,092,304 (A2), August 19, 2010.
2 Werpy, T.; Petersen, G.; eds. 2004; Top Value Added Chemi-
cals from Biomass. Volume I. Results of Screening for Potential
Candidates from Sugars and Synthesis Gas. US Department of
Energy, Offfiice of Scientific and Technical Information, OSTI
Id: 15008859.
3 Minh, D.; Besson, M.; Pinel, C.; Fuertes, P.; Petitjean, C. Top
Catal 2010, 53, 1270–1273.
FIGURE 11 Evolution of PBS reduced viscosity with the time of 4 Deshpande, R. M.; Buwa, V. V.; Rode, C. V.; Chaudhari, R. V.;
storage under ambient humidity and temperature. Mills, P. L. Catal Commun 2002, 3, 269–274.

WWW.MATERIALSVIEWS.COM JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312 5311
 ARTICLE WWW.POLYMERCHEMISTRY.ORG

5 Ishii, M.; Okazaki, M.; Shibasaki, Y.; Ueda, M.; Teranishi, T. 26 Chung, H. S.; Lee, J. W.; Kim, D. H.; Kim, D. Y.; Lee, S. W.
Biomacromolecules 2001, 2, 1267–1270. (IRe Chemical Ltd.). U.S. Patent 6,713,595 B2, March 30, 2004.
6 Otton, J.; Valerii, S. R.; Gali, A. V.; Kahirman, D. M.; Vladimir, 27 Imaizumi, M.; Kimura, H.; Fujihira, R.; Ichikawa, Y.; Suzuki,
M. N.; Lidiy, I. B.; Svetlana, I. K.; Vasilii, V. V.; Korshak, V. J J.; Moteki, Y.; Fujimaki, T.; Takiyama, E.; Showa Denko, K.K. EP
Polym Sci Polym Chem 1988, 26, 2199–2224. Patent 0,747,416, November 2004.
7 Spitz, R.; Lepage, J.; Pineau, Q.; Perez, G. (Tergal Fibres). WO 28 Kawai, R.; Miura, M.; Takakuwa, K.; Isahaya, Y.; Fujimori, T.;
Patent WO/2007/012,731, February 01, 2007. Suito, J.; Nakamura, M. (Mitsubishi Gas Chemical Company
8 Bersot, J. C.; Jacquel, N.; Saint-Loup, R.; Fuertes, P.; Rous- Inc). U.S. Patent 5,654,380, August 05, 1997.
seau, A.; Pascault, J. P.; Spitz, R.; Fenouillot, F.; Monteil, V. 29 Miuara, M.; Takakuwa, K.; Fujimori, T.; Ito, M. (Mitsubishi
Macromol Chem Phys, in press. Gas Chemical Company Inc.). EP Patent 0,823,449 A2, February
9 Mochizuki, M.; Mukai, K.; Yamada, K.; Ichise, N.; Murase, S.; 11, 1998.
Iwaya, Y. Macromolecules 1997, 30, 7403–7407. 30 Zhao, J. B.; Li, K. Y.; Yang, W. T. J Appl Polym Sci 2007,
10 Aoshima, T.; Miki, Y.; Kumazawa, K.; Katou, S.; Uyeda, T.; 106, 590–598.
Hoshino, T.; Shintani, N.; Yamagishi, K.; Isotani, A. (Mitsubishi 31 Zhao, J.-B.; Wu, X.-F.; Yang, W.-T. J Appl Polym Sci 2004,
Chemical Co.).WO/2006/115,226, 2006. 92, 3333–3337.
11 Pranamuda, H.; Tokiwa, Y.; Tanaka, H. Appl Environ Micro- 32 Aoshima, T.; Hoshino, T.; Uyeda, T.; Miki, Y. (Mitsubishi
biol 1995, 61, 1828–1832. Chemical Co.). WO Patent 2004/104067, December 2, 2004.
12 Han, Y.-K.; Kim, S.-R.; Kim, J. Macromol Res 2002, 10, 33 Aharoni, S. M. In Handbook of Thermoplastic Polyesters;
108–114. Fakirov, S. Ed.; Wiley-VCH:Weinheim, 2002; Vol. 1, Chapter 2,
13 Yang, J.; Zhang, S.; Liu, X.; Cao, A. Polym Degrad Stab pp 59–103.
2003, 81, 1–7. 34 Jéol, S.; Fenouillot, F.; Rousseau, A.; Monnet, C.; Masenelli-
14 Lim, J. S.; Hong, S. M.; Kim, D. K.; Im, S. S. J Appl Polym Varlot, K.; Briois, J.-F. J Polym Sci Pol Chem 2008, 46,
Sci 2008, 107, 3598–3608. 3985–3991.
15 Bikiaris, D. N.; Papageorgiou, G. Z.; Achilias, D. S. Polym 35 Fradet, A.; Marechal, E. J Macromol Sci Part A 1982, 17,
Degrad Stab 2006, 91, 31–43. 881–891.
16 Bikiaris, D. N.; Achilias, D. S. Polymer 2008, 49, 3677–3685. 36 Marechal, E. In Handbook of Thermoplastic Polyesters:
17 Bikiaris, D. N.; Achilias, D. S. Polymer 2006, 47, 4851–4860. Homopolymers, Copolymers, Blends, and Composites; Fakirov,
S., Ed.; Wiley-VCH: Weinheim 2002.
18 Takasu, A.; Oishi, Y.; Iio, Y.; Inai, Y.; Hirabayashi, T. Macro-
molecules 2003, 36, 1772–1774. 37 Aoshima, T.; Miki, Y.; Kumazawa, K.; Katou, S.; Uyeda, T.;
Hoshino, T.; Shintani, N.; Yamagishi, K.; Isotani, A. (Mitsubishi
19 Takasu, A.; Iio, Y.; Mimura, T.; Hirabayashi, T. Polym J Chemical Co.). WO Patent 2006/115226, November 02, 2006.
(Tokyo, Jpn) 2005, 37, 946–953.
38 Senmura, K.; Takashima, S.; Kamitani, R.; Ito, K.; Akio, N.;
20 Takasu, A.; Iio, Y.; Oishi, Y.; Narukawa, Y.; Hirabayashi, T.
Masao, K. (Mitsubishi Chemical Co.). JP Patent 46042493, De-
Macromolecules 2005, 38, 1048–1050.
cember 15 1971.
21 Zhu, C.; Zhang, Z.; Liu, Q.; Wang, Z.; Jin, J. J Appl Polym
39 MacDonald, W. A. Polym. Int. 2002, 51, 923–930.
Sci 2003, 90, 982–990.
40 Chung, J. S. J. Macromol. Sci. –Chem., 1990, 27, 479–490.
22 Kricheldorf, H. R.; Behnken, G.; Schwarz, G. Polymer 2005,
46, 11219–11224. 41 Hashimoto, S.; Okumura, N.; Kazama, K. (Teijin Ltd.). U.S.
Patent 4,131,601, December 26, 1978.
23 Yashiro, T.; Kricheldorf, H. R.; Huijser, S. Macromol Chem
Phys 2009, 210, 1607–1616. 42 Ahmadnian, F.; Reichert, K.-H. Macromol Symp 2007, 259,
188–196.
24 Buzin, P.; Lahcini, M.; Schwarz, G.; Kricheldorf, H. R. Macro-
molecules 2008, 41, 8491–8495. 43 Ahmadnian, F.; Velasquez, F.; Reichert, K.-H. Macromol React
25 Chung, H. S.; Lee, J. W.; Kim, D. H.; Kim, D. Y.; Lee, S. Eng 2008, 2, 513–521.
W. (IRe Chemicals Ltd.). U.S. Patent 6,399,716 B2, June 04, 44 Kim, H.-S.; Kim, H.-J. Polym Degrad Stab 2008, 93,
2002. 1544–1553.

5312 JOURNAL OF POLYMER SCIENCE PART A: POLYMER CHEMISTRY 2011, 49, 5301–5312