1. 321
Degradation of poly-L-lactide. Part 2: increased
temperature accelerated degradation
N A Weir1, F J Buchanan1*, J F Orr1, D F Farrar2 and G R Dickson3
1School of Mechanical and Manufacturing Engineering, Queen’s University Belfast, Belfast, UK
2Smith and Nephew Group Research Centre, Heslington, York, UK
3Department of Trauma and Orthopaedic Surgery, Queen’s University Belfast, Musgrave Park Hospital, Belfast, UK
Abstract: Poly-L-lactide (PLLA) is one of the most significant members of a group of polymers
regarded as bioresorbable. The degradation of PLLA proceeds through hydrolysis of the ester linkages
in the polymer’s backbone; however, the time for the complete resorption of orthopaedic devices
manufactured from PLLA is known to be in excess of five years in a normal physiological environ-
ment. To evaluate the degradation of PLLA in an accelerated time period, PLLA pellets were pro-
cessed by compression moulding into tensile test specimens, prior to being sterilized by ethylene oxide
gas (EtO) and degraded in a phosphate-buffered solution (PBS) at both 50 °C and 70 °C. On retrieval,
at predetermined time intervals, procedures were used to evaluate the material’s molecular weight,
crystallinity, mechanical strength, and thermal properties. The results from this study suggest that at
both 50 °C and 70 °C, degradation proceeds by a very similar mechanism to that observed at 37 °C
in vitro and in vivo. The degradation models developed also confirmed the dependence of mass
loss, melting temperature, and glass transition temperature (Tg) on the polymer’s molecular weight
throughout degradation. Although increased temperature appears to be a suitable method for
accelerating the degradation of PLLA, relative to its physiological degradation rate, concerns still
remain over the validity of testing above the polymer’s Tg and the significance of autocatalysis at
increased temperatures.
Keywords: poly-L-lactide, degradation, accelerated, molecular weight, crystallinity
Tg glass transition temperatureNOTATION
Tm melting temperature
DSC differential-scanning calorimetry
Ea activation energy DH
melt
enthalpy of fusion
EtO ethylene oxide gas
GPC gel-permeation chromatography
m
0
initial mass 1 INTRODUCTION
m
t
mass at time, t
Mn number average molecular weight Poly-L-lactide (PLLA) is a member of the aliphatic
polyester family of bioresorbable polymers, the mostMn
t
number average molecular weight at time, t
PBS phosphate-buffered solution attractive group of polymers that currently meet the
various medical and physical demands for safe clinicalPCL poly-e-caprolactone
PGA polyglycolide applications [1]. Other significant members of this family
include polyglycolide (PGA) and poly-e-caprolactonePLLA poly-L-lactide
(PCL). PLLA is a semicrystalline polymer and the most
common bioresorbable polymer used for orthopaedic
applications [2], due to its relatively high tensile strengthThe MS was received on 4 February 2004 and was accepted after revision
for publication on 17 June 2004. and low elongation [3]. Like all the members of the
* Corresponding author: School of Mechanical and Manufacturing
aliphatic polyester family PLLA degrades in vivo throughEngineering, Queen’s University Belfast, Ashby Building, Stranmillis
Road, Belfast BT9 5AH, UK. email: f.buchanan@qub.ac.uk simple hydrolysis of the hydrolytically unstable ester
H01204 © IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

2. 322 N A WEIR, F J BUCHANAN, J F ORR, D F FARRAR AND G R DICKSON
linkage in the polymer’s backbone, with the degradation 2 MATERIALS AND METHODS
products ultimately metabolized to carbon dioxide and
water and eliminated from the body [4]. 2.1 Materials
The evaluation of PLLA and other bioresorbable
The polymer studied in this investigation, poly-L-lactide
polymers’ properties throughout resorption is commonly
(PLLA) ResomerA L (batch number 26033) was supplied
determined through in vitro and in vivo experiments, with
in a sealed moisture-proof container by Boehringer
samples retrieved at predetermined time intervals and
Ingelheim (Ingelheim, Germany) in pellet form.
properties of mechanical strength, molecular weight,
crystallinity, and mass change monitored. Typically, the
in vitro test techniques developed to evaluate perform-
ance are conducted with the polymer fully submerged
2.2 Methods
in a pH 7.4 phosphate-buffered solution (PBS) and
incubated at 37 °C, mimicking the physiological environ- 2.2.1 Processing
ment [5]. Many studies have been conducted, investi-
The PLLA was processed by compression moulding into
gating the degradation of PLLA by this method [6–8].
plates 0.8 mm thick. ASTM D638-99 type-V tensile speci-
However, considering that the times for the complete
mens were then cut from the plates. The tensile specimens
resorption of PLLA orthopaedic devices, in a physio-
were annealed at 120 °C for a period of four hours in a
logical environment, have been reported to be in excess
preheated air-circulating oven prior to being sterilized
of five years [9], the impact this can have on product
using ethylene oxide gas (EtO) by Griffith Microscience
development periods is substantial. Accelerating this
(Derbyshire, UK) on their standard EtO cycle for
initial evaluation process would obviously benefit the
medical polymers, i.e. ‘Cycle 33’ [19].
development of this promising group of biomaterials. To
date a number of techniques, with varying success, have
2.2.2 Increased temperature degradationbeen investigated for accelerating the in vitro degradation
of bioresorbable polymers. These have included the The initial mass, m
0
, of each of the tensile specimens
introduction of applied strain [10], increasing the tem- was recorded. Individual specimens were then placed in
perature of the degradation medium [11–13] or varying 28 ml screw-top glass bottles and fully immersed in a
its pH [14] and [15], introducing organic compounds to pH 7.4 PBS in accordance with ISO 15814:1999 [5]. The
the polymer matrix [16], the application of ultrasound samples were divided into two separate groups and
[17], and introducing enzymes to the degradation medium placed in separate air-circulating ovens maintaining their
[18]. However, currently, limited practical insight has been temperatures at 50 °C and 70 °C. The pH of the solutions
gained from these accelerated test methods, with extra- was monitored at regular intervals.
polation of results back to service conditions proving Six PLLA tensile specimens were removed at each
very difficult. follow-up time (three mechanical test specimens and
This study investigates the potential of increasing the three mass-change specimens). Follow-up times for each
temperature of the degradation medium for accelerating of the 50 °C and 70 °C studies are given in Table 1.
the in vitro degradation of PLLA, relative to its physio-
logical degradation rate at 37 °C, and is a continuation
2.2.3 Characterization of retrieved material
of previous studies conducted by this research group,
that investigated the processing, annealing, sterilization Mechanical properties. The mechanical properties of
the PLLA tensile specimens were determined using a[19], and physiological degradation of PLLA in vitro
and in vivo [20]. The present study aims to model the JJ Lloyd EZ 50 tensile testing machine (Hampshire,
UK), equipped with a 1 kN load cell and tested at adegradation of PLLA at 50 °C and 70 °C and compare
the degradation mechanisms at these increased tempera- constant strain rate of 10 mm/min. Young’s modulus,
tensile strength, and extension at break were calculatedtures to those at 37 °C. Additionally, consideration is
given to the influence of testing at increased temperatures from each of the load versus extension curves.
In accordance with ISO 15814:1999 [5], the retrievedon the autocatalytic heterogeneous degradation mech-
anism reported for PLLA [21] and the effect of testing material underwent mechanical testing while ‘wet’, with
testing conducted within three hours of retrieval fromabove the polymer’s glass transition temperature (Tg)
on its degradation kinetics. the in vitro buffered solution.
Table 1 Accelerated degradation follow-up times
Follow-up Days
50 °C 1 3 5 11 21 32 50 72 88 115
70 °C 1 3 5 7 9 11 14 16 18 21 23
H01204 © IMechE 2004Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

3. 323DEGRADATION OF POLY-L-LACTIDE. PART 2
Mass change. On retrieval the tensile specimens were rate of the unstable ester linkages. Anderson [23] reported
a statistical method for relating molecular weight todried immediately with a paper towel to remove any
surface moisture before being weighed using an electronic hydrolysis rate; assuming that the extent of degradation
was not large, the following kinetic relationship basedbalance (Mettler Toledo, Fisher Scientific, UK) to deter-
mine the percentage swelling of the polymer and water on the polymers Mn was reported
uptake. The specimens were then dried in a vacuum oven
1/Mn
t
=1/Mn
0
+k
1
t (3)(Townson+Mercer, Altrincham, UK) at approximately
30 °C for 48 hours at a vacuum of 0.68 bar (20 in. Hg)
where Mn
t
=Mn at time, t; Mn
0
=Mn at t=0; k
1
=rate
and reweighed to obtain their mass at time, t (m
t
). The
constant; and t=time. If this theory holds true a linear
overall percentage mass change after drying was then
relationship should exist between 1/Mn versus time, up
calculated from equation (1)
until the point of mass loss.
However, a disadvantage of this statistical approach
percentage mass change=
m
t
−m
0
m
0
×100 per cent
is that it does not account for the possibility of auto-
catalysis accelerating the polymer’s degradation rate.(1)
Pitt and Gu [24] derived a relationship based on the
kinetics of the ester hydrolysis reaction, accounting forMolecular weight and thermal properties. Following
autocatalysis by the generated carboxylic acid endmass change measurements the dried PLLA specimens
groups, described by the rate equationwere reused for gel-permeation chromatography (GPC)
analysis to determine their weight and number average
d(E)/dt=−d(COOH)/dt=−k(COOH)(H
2
O)(E)
molecular weights (Mn) throughout degradation and
also for differential-scanning calorimetry (DSC) to deter- (4)
mine their thermal properties and percentage crystallinity.
where (COOH), (H
2
O), and (E) represent the con-
centrations of carboxyl end groups, water, and estersMolecular weight. The GPC analysis was conducted by
respectively.Rapra Technology Ltd (Shropshire, UK). Samples were
On further analysis of equation (4) and assuming thatprepared by adding 10 ml of choloform solvent to 20 mg
the ester and water concentrations remain constant andof sample taken through a cross-section of the material.
the concentration of acid end groups is equal to 1/Mn,A Plgel-mixed bed column with refractive index response
it can be shown thatdetector was used. The GPC system was calibrated
with polystyrene and all results were expressed as the
Mn
t
=Mn
0
e−k2t (5)‘polystyrene equivalent’ molecular weights.
If this relationship holds true, a linear relationshipThermal properties. The thermal properties of the dried
should exist between the ln Mn versus time up until theretrieved PLLA tensile specimens were analysed using a
point of mass lossPerkin Elmer DSC 6 (Beaconsfield, Buckinghamshire,
UK) testing machine, over a temperature range of 40 °C ln Mn
t
=−k
2
t+ln Mn
0
(6)
to 200 °C at a heating rate of 10 °C/min., providing
measurements of glass transition temperature, Tg, melt- Furthermore, the mechanical properties of polymers
ing point, Tm, and enthalpy of fusion, DH
melt
, J/g. The have also been shown to be related to their Mn through
DSC results were derived from this single heating cycle the Flory relationship [25] and [26]
to provide a true indication of changes in the polymer’s
thermal properties and morphology as a direct result of
s=s
2
−
B
Mn
t
(7)
degradation. The enthalpy of fusion, DH
melt
, was then
used to calculate the polymer’s percentage crystallinities
where s=fracture strength; s
2
=fracture strength atrelative to the enthalpy of fusion of a 100 per cent
infinite molecular weight; and B=constant. This relation-crystalline sample of PLLA, reported to be 93 J/g [22]
ship implies that a significant loss of molecular weight
percentage crystallinity=(DH
melt
/93)×100 per cent
can occur before any significant loss in mechanical
(2) properties is observed.
2.3.2 The Arrhenius relationship2.3 Theory
The Arrhenius relationship (equation (8)) represents a2.3.1 Modelling bioresorbable polymer degradation
common method used for extrapolating results from
Since Mn is directly related to the scission of the polymer higher temperatures back to service temperatures [27]
chains, a number of relationships have been derived
relating the changes in Mn with time to the hydrolysis k=Ae−Ea/RT (8)
H01204 © IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

4. 324 N A WEIR, F J BUCHANAN, J F ORR, D F FARRAR AND G R DICKSON
where k=rate constant; A=constant; Ea=activation were so brittle they simply disintegrated on handling. A
similar pattern was also observed at 50 °C, with the whiteenergy, J/mol; R=universal gas constant, 8.314 J mol−1
K−1; and T=temperature in Kelvins, K. If the relation- areas developing at 50 days and increasing with time.
ship holds true, a linear relationship should exist between
the ln k versus 1/T
3.2 Molecular weight versus time
ln k=−
A
Ea
R BA
1
TB+ln A (9) As expected, the molecular weight of the PLLA
tensile specimens at 50 °C and 70 °C decreased with time
(Table 2). After 115 days at 50 °C the tensile specimens’
Mn had decreased by approximately 93 per cent. At
70 °C a decrease of approximately 95 per cent in 23 days
3 RESULTS was observed, in comparison to an 82 per cent decrease
after 44 weeks at 37 °C, the final time point analysed.
3.1 Visual examination Additionally, the profiles of the molecular weight-loss
curves (Fig. 2), most notably at 70 °C, are characterizedInitially, at 0 weeks the annealed PLLA tensile specimens
by initial rapid loss in molecular weight, up untilwere opaque and off-white in colour (Fig. 1). After five
approximately five days at 70 °C, followed by a perioddays at 70 °C small areas of the specimens became more
where the molecular weight loss slowed considerably.intensely white, and as degradation time increased more
white areas became visible. At 23 days the specimens
Fig. 2 Comparison between Mw and Mn at 50 °C and 70 °CFig. 1 PLLA specimens degraded at 70 °C
Table 2 Molecular weight versus time summary at 37 °C, 50 °C, and 70 °C
Temperature Molecular weight
Days at 37 °C 0 28 70 140 182 224 266 308
37 °C Mw 424 000 339 000 309 000 199 000 199 000 159 000 133 500 74 900
Mn 158 500 143 000 120 000 72 500 93 850 65 800 53 050 22 500
Days at 50 °C 0 1 5 11 21 32 50 72 88 115
50 °C Mw 409 500 327 500 332 500 238 500 248 500 175 500 128 600 73 250 47 500 18 650
Mn 166 000 132 000 129 000 104 500 113 000 77 650 62 650 33 150 20 600 11 700
Days at 70 °C 0 1 3 5 9 14 18 21 23
70 °C Mw 409 500 276 500 127 500 86 200 52 550 56 950 25 000 19 350 11 350
Mn 166 000 119 000 64 450 39 300 24 900 29 200 13 750 11 700 9 090
H01204 © IMechE 2004Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

5. 325DEGRADATION OF POLY-L-LACTIDE. PART 2
Table 3 Thermal properties and crystallinity of PLLA3.3 Mass change versus time
samples degraded at 50 °C
Before drying a similar pattern was observed at both
Degradation50 °C and 70 °C (Fig. 3). Initially, after one day, increases
time, days % crystallinity Tm, °C Tg onset, °Cof approximately 0.5 per cent and 0.7 per cent were
observed at 50 °C and 70 °C respectively. Swelling then 0 47.0 181.7 67.4
1 45.7 181.8 66.3gradually increased to approximately 1 per cent after 72
11 47.1 181.8 65.1
days at 50 °C and 14 days at 70 °C. This 1 per cent swell- 21 43.4 181.5 65.6
ing appeared to coincide with the onset of polymer mass 32 45.3 181.7 66.0
50 56.4 181.6 64.3loss after drying, with specimens at both 50 °C and 70 °C
72 61.7 179.2 66.3
losing mass from this point onwards. A maximum mass 88 65.9 174.6 61.0
loss of 4 per cent was observed at 50 °C after 115 days 115 66.5 171.7 –
and 7.6 per cent after 23 days at 70 °C. The results
of the mass-change analysis are in agreement with the
general sequence of aliphatic polyester degradation, which
Table 4 Thermal properties and crystallinity of PLLAsuggests a time-lag before any mass loss is observed [3],
samples degraded at 70 °Cand have a very similar profile to the previous control
study that investigated the in vitro degradation of PLLA Degradation
tensile specimens at 37 °C [20]. time, days % crystallinity Tm, °C Tg onset, °C
0 57.5 180.5 65.8
1 59.9 180.9 63.9
3.4 DSC analysis versus time 3 57.5 180.5 64.0
5 77.1 178.8 63.7
At both 50 °C and 70 °C a general trend of increasing 9 85.7 174.5 59.7
14 76.2 176.7 60.3crystallinity and decreasing Tg onset temperature was
18 88.9 172.3 57.8observed with increasing degradation time (Tables 3
21 85.1 171.9 57.7
and 4), reflecting the findings of a similar study conducted 23 92.2 171.8 –
at 37 °C in vitro [20]. Additionally, a significant decrease
in the specimens’ melting point, Tm, was observed after
72 days at 50 °C and five days at 70 °C.
The DSC thermograms for the specimens degraded at At 0 weeks (Fig. 4(a)) a sharp peak commencing
at approximately 66 °C was observed, related to theboth 50 °C and 70 °C are shown in Fig. 4, along with a
control specimen at 0 days degradation. Again, the pro- polymer’s Tg and indicating unfreezing of main chain
segmental motion as weak secondary bonds in thefiles of the DSC thermograms reflect those observed at
37 °C [20], albeit in an accelerated time frame. amorphous regions are broken. As the temperature
increased further, a small endothermic dip was observed
just before melting commenced, followed by the main
melting peak. It is suggested that the dip before melting
was caused by some crystallization of the polymer.
Although the polymer was annealed prior to degradation,
with the aim of limiting crystallization throughout the
study, close to the polymer’s melting point the chain
mobility would have increased, allowing some of the
amorphous segments to order themselves into a more
crystalline structure.
At 50 days at 50 °C and five days at 70 °C (Figs 4(b)
and (e)) as degradation increased, presumably in the
amorphous regions, the initial dip before melting observed
at 0 weeks had disappeared, with less amorphous regions
remaining capable of crystallization, with a new peak
forming in its place after 50 days at 50 °C (Fig. 4(b)).
It is suggested that this new peak represents the melt-
ing of new crystallites formed by the crystallization of
the internal degradation by-products. As degradation
proceeds further, it is speculated that the newly formed
crystallite’s size also increases, moving it to higher tem-
peratures while the main melting peak is shifting to lower
Fig. 3 Percentage mass change at 50 °C and 70 °C temperatures. This results in the two peaks eventually
H01204 © IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

6. 326 N A WEIR, F J BUCHANAN, J F ORR, D F FARRAR AND G R DICKSON
Fig. 4 Comparison between DSC thermograms for samples degraded at 50 °C and 70 °C
Table 6 Change in the tensile properties of PLLA throughoutmerging, with the smaller, newly formed peak appearing
degradation at 70 °Cas a shoulder on the larger peak, before, finally, only one
sharp melting peak becomes visible (Figs 4(c) and (f)).
Degradation Young’s Tensile Extension at
time, days modulus (MPa) strength (MPa) break (mm)
0 668.4 64.3 1.62
3.5 Mechanical strength versus time 1 605.6 47.3 1.17
3 536.2 40.3 1.08
The tensile strength of the specimens reduced to zero 5 316.7 7.8 1.02
after approximately 50 days at 50 °C and 7 days at 70 °C 7 102.9 2.3 0.56
(Tables 5 and 6). A gradual decrease in Young’s modulus
was also observed, with the load versus extension curves
displaying a transition from a more ductile to brittle
Table 5 Change in the tensile properties of PLLA throughout failure mode after five days at 50 °C and one day at
degradation at 50 °C 70 °C.
Degradation Young’s Tensile Extension at
time, days modulus (MPa) strength (MPa) break (mm)
4 MODELLING DEGRADATION
0 668.4 64.3 1.62
1 625.0 61.4 1.81
4.1 Molecular weight models3 624.4 61.3 1.59
5 608.1 49.4 1.29
Molecular weight loss. Analysing the data presented in11 580.9 51.1 1.35
21 478.3 27.8 0.83 Table 2 in conjunction with equations (3) and (6), the
32 357.1 17.9 1.56 rate constants, k
1
, for the uncatalysed model and k
2
,
50 – 2.9 1.09
for the autocatalysed model, were determined by linear
H01204 © IMechE 2004Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

7. 327DEGRADATION OF POLY-L-LACTIDE. PART 2
regression, from plots of (1/Mn
t
−1/Mn
0
) versus time for Dependence of mass loss, Tm, and Tg on molecular
weight. Analysing the percentage mass loss results afterthe uncatalysed model and ln(Mn
t
/Mn
0
) versus time
for the autocatalysed model (Fig. 5) at 37 °C, 50 °C, and drying in conjunction with the molecular-weight-loss
results, it is apparent that a relationship exists between70 °C. The analysis was conducted up until the point of
mass loss at each of the temperatures investigated. It is the polymer’s Mn and percentage mass loss (Fig. 7),
with mass loss only observed after the polymer’s Mn hadapparent from Table 7 that as the degradation tem-
perature increased the correlation coefficient (R2 value) reduced to less than approximately 20 000. A similar
trend was also observed when considering the reductionalso generally increased.
Considering the rate constants, k
2
, for the auto- in Tm with Mn (Fig. 8). Only once the polymer’s Mn
had decreased to approximately 50 000, were significantcatalysed model further it is apparent that as the tem-
perature increased the rate of molecular weight loss decreases in the polymer’s melting temperature observed.
This suggests that once the polymer’s molecular weightappeared to increase exponentially, with an approximate
four-fold increase observed at 50 °C compared to a forty- had diminished sufficiently, and presumably with the
amorphous regions exhausted, the crystalline regionsfold increase at 70 °C, relative to the degradation rate at
37°C. Applying the Arrhenius relationship (equation (9)), were then predominantly attacked. A similar dependence
of Tg on Mn was also observed (Fig. 9).it is obvious that a linear relationship does exist (Fig. 6),
with a high linear regression correlation coefficient of
0.995. The activation energy (Ea) for the loss of Mn was
calculated from Fig. 6 as 100.5 kJ/mol over the 37 °C to
50 °C to 70 °C temperature range. For the uncatalysed
model an Ea of 102.7 kJ/mol was calculated.
Fig. 7 Percentage change after drying versus Mn
Fig. 5 Autocatalysed degradation model
Table 7 Uncatalysed and autocatalysed hydrolysis
rates
Model Uncatalysed Autocatalysed
Temperature k
1
R2 k
2
R2
37 °C 9×10−8 0.63 0.0052 0.85
50 °C 3×10−7 0.90 0.0196 0.96 Fig. 8 Relationship between melting temperature and
70 °C 4×10−6 0.99 0.2155 0.95 molecular weight
Fig. 9 Relationship between glass transition temperature and
Fig. 6 Arrhenius plot of ln k
2
versus 1/T molecular weight
H01204 © IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

8. 328 N A WEIR, F J BUCHANAN, J F ORR, D F FARRAR AND G R DICKSON
Mechanical strength and molecular weight. An effort the results of previous researchers [28–30], investigating
similarly semicrystalline PLLA at 37 °C in vitro andwas also made to model the mechanical properties
of the tensile specimens investigated in this study in vivo. They observed that as degradation increased, the
distributions of the GPC curves became bimodal andagainst molecular weight using the Flory relationship
(equation (7)). However, it is obvious from Fig. 10 that even multimodal in nature as a result of the selective
degradation of the amorphous regions.a relationship similar to the one predicted by Flory is not
derived from the results of the present study. However, The molecular weight models (Figs 7–9) also show
the dependence of the degradation of PLLA on itson further analysis it became apparent that an almost
linear relationship existed for the loss of tensile strength molecular weight, with mass loss, melting temperature,
and glass transition displaying a strong dependence onwith time, most notably at 50 °C and 70 °C.
molecular weight. An obvious benefit of these models is
the possibility of being able to predict the initial molecular
weight of PLLA required to produce devices with vary-5 DISCUSSION
ing resorption times. However, the failure of the Flory
model to predict mechanical strength is perhaps not5.1 Degradation mechanisms
surprising considering it only accounts for molecular
The results presented for the accelerated degradation
weight loss and not changes in morphology, such as
of PLLA at increased temperatures suggest that at
the increases in crystallinity observed for the PLLA
50 °C and 70 °C, degradation proceeded by the same
investigated in this study.
mechanism and followed the general sequence of bulk
degradation reported previously for semicrystalline
Significance of autocatalysis at increased temperatures.aliphatic polyesters. This suggests degradation occurs in
The degradation of aliphatic polyesters has been showntwo distinct stages [21] and is characterized by molecular
to occur more rapidly at the centre than at the surfaceweight loss being observed first, before loss of mechanical
[21], due to the hydrolytic cleavage of the ester bondsstrength and before any physical mass loss is observed
forming new acidic carboxyl end groups, resulting in[3]. Evidence from the results of the present study sup-
a higher internal acidity and a differentiation betweenporting this two-stage bulk degradation mechanism are
the surface and interior degradation rates. The resultsdiscussed below.
of the present study, modelling the loss of Mn to both
the uncatalysed and autocatalysed degradation models,Relationship between molecular weight loss and degradation.
showed no significant difference between activationThe general concensus that the degradation of semi-
energies calculated for the two models. Pohjonen andcrystalline aliphatic polyesters, like the PLLA investi-
To¨rma¨la¨ [11] found a similar value for activation energygated in this study, proceeds via random bulk hydrolysis
of self-reinforced PLLA of 101.4 kJ/mol using the auto-in two distinct stages is supported by considering the
catalytic model. Previous research conducted by Li andprofile of the molecular-weight-loss curves of Fig. 2.
McCarthy [32] on the in vitro degradation of poly-The curves show a faster initial loss of molecular weight
(DL-lactide) at 60 °C, showed that a hollow structureup until approximately five days at 70 °C, before the
was not obtained during degradation, in contrast torate of molecular weight loss slows considerably. This
degradation at 37 °C in vitro, where a hollow structuredegradation behaviour is typical where the rate of chain
was observed after 12 weeks due to autocatalysis [33].scission events is relatively constant with time. The
Li and McCarthy [32] attributed this to the rapidmolecular weight is therefore inversely proportional to
degradation at 60 °C and increase in the diffusiontime. However, interestingly, the GPC curves for both the
coefficient of water and water-soluble oligomers abovetensile specimens at 50 °C and 70 °C remained essentially
the polymer’s Tg. They concluded that although hetero-monomodal throughout degradation, contradictory to
geneous degradation was still detected at 60 °C, it was
much less significant and the relative importance of
internal autocatalysis diminished as temperature increased.
This does not appear to be the case in the current study,
with no observable difference between surface and core
degradation rate.
5.2 Suitability of increased temperature for accelerating
the degradation of PLLA
The results presented for the accelerated degradation of
PLLA at 50 °C and 70 °C indicate that the degradationFig. 10 Relationship between tensile strength and molecular
weight mechanisms were very similar to the in vitro and
H01204 © IMechE 2004Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

9. 329DEGRADATION OF POLY-L-LACTIDE. PART 2
in vivo degradation mechanism reported at 37 °C [20]. However, further work is needed to assess the validity
of this hypothesis with only semicrystalline PLLA, withThis suggests that increasing the temperature of the
degradation medium is a suitable method for accelerating identical initial crystallinities, investigated in this present
study.the degradation of PLLA relative to its physiological
degradation rate. However, in previous increased-
temperature studies caution has often been noted when
using test results performed at temperatures greater than 6 CONCLUSIONS
the polymer’s Tg for predicting degradation behaviour
below the polymer’s Tg [11] and [12]. Above the poly- This study suggests that increasing the temperature of
mer’s Tg it would be expected that, with main chain the degradation medium represents a powerful method
segmental motion unfrozen and the weak van der Waals for accelerating the degradation of PLLA, with the
forces holding the amorphous regions in place broken, degradation mechanisms at 37 °C, 50 °C, and 70 °C
water molecules would be able to access the amorphous proving to be very similar. However, concerns still
regions more easily, initiating hydrolytic chain scission, remain over the validity of testing above the Tg of PLLA
resulting in a further increased degradation rate. The when predicting results at service temperatures below its
compression-moulded PLLA tensile specimens investi- Tg and the significance of the autocatalytic phenomenon
gated in this study had a Tg (dry) of approximately 66 °C as the temperature is increased relative to that at 37 °C.
(Tables 3 and 4); two temperatures were investigated
below the polymer’s Tg (37 °C and 50 °C) and one above
the polymer’s Tg (70 °C). The high linear correlation
ACKNOWLEDGEMENTScoefficient (R2=0.995) achieved for the Arrhenius plot,
modelling the loss of molecular weight (Fig. 6), suggests
The authors would like to thank Boehringer Ingelheimthat the degradation kinetics were not greatly affected
(Ingelheim, Germany) for supplying the PLLA; Smithabove Tg. Ideally, to fully assess the implications testing
& Nephew Group Research Centre (York, UK) forabove the polymer’s Tg has on the degradation rate, a
their assistance with processing; Griffith Microsciencerange of temperatures would need to be investigated
(Derbyshire, UK) for the ethylene oxide sterilization;above and below Tg. However, this is not as straight-
and Rapra Technology Limited (Shropshire, UK) for theforward as it might appear. When determining the
molecular-weight characterization. Finally, the EPSRCpolymer’s Tg it must be realized that a small amount of
(Swindon, UK) for financial assistance.water can have a marked plasticizing effect, causing a
reduction in the polymer’s Tg [34]. Combine this with
the reduction in Tg due to the loss of molecular weight
(Fig. 9) and it is evident that defining the polymer’s Tg REFERENCES
and determining the test temperatures that remain above
and below the polymer’s Tg is an ever changing scenario. 1 Li, S. and Vert, M. Biodegradation of aliphatic polyesters.
In Degradable Polymers Principles & Applications (EdsTraditionally, as in this study and previous studies
G. Scott and D. Gilead) 1995, pp. 43–87 (Chapman & Hall,investigating the degradation of bioresorbable polymers,
London).for convenience Tg is commonly determined by DSC on
2 Barber, F. A. Resorbable materials for arthroscopicretrieved dried specimens; however, ideally for accurate
fixation: a product guide, Orthopedic Special Edn, 2002,
Tg measurements, representative of the polymers con-
8, 29–37.
dition in service, test regimes need to be developed that
3 Middleton, J. C. and Tipton, A. J. Synthetic biodegradable
can accurately monitor the polymers Tg while the polymers as orthopedic devices, Biomaterials, 2000, 21,
specimens remain ‘wet’. 2335–2346.
Although concerns still remain about the validity of 4 Hayashi, T. Biodegradable polymers for biomedical uses,
testing above the Tg of PLLA when predicting behaviour Prog. Polym. Sci., 1994, 19, 663–702.
5 Implants for surgery—copolymers and blends based onbelow its Tg, it is hypothesized from the results of the
polylactide—In vitro degradation testing. ISO 15814:1999.present study and from those of Pohjonen and To¨rma¨la¨
6 Duek, E. A. R., Zavaglia, C. A. C. and Belangero, W. D.[11] and Agrawal et al. [12] that the polymer’s
In vitro of poly(lactic acid) pin degradation, Polymer, 1999,morphology may also have a significant role to play in
40, 6456–6473.determining the validity of any relationships derived. It is
7 Matsusue, Y., Yamamuro, T., Oka, M., Shikinami, Y. et al.
speculated that for semicrystalline polymers, results from
In vitro and in vivo studies on bioabsorbable ultra-high-
tests performed above their Tg may be more valid for
strength poly(L-lactide) rods, J. Biomed. Mater. Res., 1992,
predicting degradation below their Tg, due to their greater 26, 1553–1567.
volume constrained in tightly packed, less-accessible 8 Kelloma¨ki, M., Paasimaa, S. and To¨rma¨la¨, P. Pliable poly-
crystalline domains, in comparison to polymers that are lactide plates for guided bone regeneration: manufacturing
intrinsically amorphous or likely to crystallize through and in vitro, Proc. Instn Mech. Engrs, Part H: J. Engineering
in Medicine, 2000, 214, 615–629.annealing as a direct result of the increased temperature.
H01204 © IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine

10. 330 N A WEIR, F J BUCHANAN, J F ORR, D F FARRAR AND G R DICKSON
9 Bergsma, J. E., de Bruijn, W. C., Rozema, F. R., 22 Fischer, E. W., Sterzel, H. J. and Wegner, G. Investigation
of the structure of solution grown crystals of lactideBos, R. R. M. and Boering, G. Late degradation tissue
response to poly(L-lactide) bone plates and screws, copolymers by means of chemical reactions, Kolloid-Z. u.
Z. Polymere, 1973, 251, 980–990.Biomaterials, 1995, 16, 25–31.
10 Miller, N. D. and Williams, D. F. The in vivo and in vitro 23 Anderson, J. M. Perspectives on the in vivo responses
of biodegradable polymers. In Biomedical Applications ofdegradation of poly(glycolic acid) suture material as a
function of applied strain, Biomaterials, 1984, 5, 365–368. Synthetic Biodegradable Polymers (Ed. J. O. Hollinger)
1995, pp. 223–233 (CRC Press, Boca Raton, FL, USA).11 Pohjonen, T. and To¨rma¨la¨, P. Hydrolytic degradation of
ultra-high-strength self-reinforced poly-L-lactide. A tem- 24 Pitt, C. G. and Gu, Z.-W. Modification of the rates of chain
cleavage of poly(e-caprolactone) and related polyesters inperature dependence study. In Biodegradable Implants in
Fracture Fixation (Eds L. K. Hung and P. C. Leung) 1994, the solid state, J. Control Release, 1987, 4, 283–292.
25 Ward, I. M. Mechanical Properties of Solid Polymers,pp. 75–88 (Department of Orthopaedics and Traumatology,
Chinese University of Hong Kong and World Scientific). 1st edn, 1971, p. 335 (Wiley-Interscience, Chichester).
26 Farrar, D. F. and Gillson, R. K. Hydrolytic degradation of12 Agrawal, C. M., Huang, D., Schmitz, J. P. and
Athanasiou, K. A. Elevated temperature degradation of a polyglyconate B: the relationship between degradation
time, strength and molecular weight, Biomaterials, 2002,50:50 copolymer of PLA-PGA, Tissue Eng., 1997, 3(4),
345–352. 23, 3905–3912.
27 Wise, J., Gillen, K. T. and Clough, R. L. An ultrasensitive13 Bucholz, B. Accelerated degradation test on resorbable
polymers. In Degradation Phenomena on Polymeric Bio- technique for testing the Arrhenius extrapolation assump-
tion for thermally aged elastomers, Polym. Degrad. Stab.,materials (Eds H. Planck, M. Dauner, and M. Renardy)
1992, pp. 67–76 (Springer-Verlag, Berlin). 1995, 49, 403–418.
28 Vert, M., Li, S. and Garreau, H. New insights on the14 Chu, C. C. The in-vitro degradation of poly(glycolic acid)-
sutures—effect of pH, J. Biomed. Mater. Res., 1981, 15, degradation of bioresorbable polymeric devices based on
lactic and glycolic acids, Clin. Mater., 1992, 10, 3–8.795–804.
15 Cam, D., Hyon, S.-H. and Ikada, Y. Degradation of high 29 Li, S., Garreau, H. and Vert, M. Structure-property
relationships in the case of the degradation of massivemolecular weight poly(L-lactide) in alkaline medium,
Biomaterials, 1995, 16, 833–843. poly(a-hydroxy acids) in aqueous media, Part 3 influence
of the morphology of poly(L-lactic acid), J. Mater. Sci.16 Cha, Y. and Pitt, C. G. The acceleration of degradation-
controlled drug delivery from polyester microspheres, Mater. Med., 1990, 1, 198–206.
30 Pistner, H., Bendix, D. R., Mu¨hling, J. and Reuther, J.J. Control Release, 1989, 8, 259–265.
17 Agrawal, C. M., Kennedy, M. E. and Micallef, D. M. Poly(L-lactide): a long-term degradation study in vivo.
Part III. Analytical characterization, Biomaterials, 1993,The effects of ultrasound irradiation on a biodegradable
50–50 per cent copolymer of polylactic and polyglycolic 14, 291–298.
31 Von Recum, H. A., Cleek, R. L., Eskin, S. G. andacids, J. Biomed. Mater. Res., 1994, 28, 851–859.
18 Gan, Z., Liang, Q., Zhang, J. and Jing, X. Enzymatic Mikos, A. G. Degradation of polydispersed poly(L-lactic
acid) to modulate lactic acid release, Biomaterials, 1995,degradation of poly(e-caprolactone) film in phosphate
buffer solution containing lipases, Polym. Degrad. Stab., 16, 441–447.
32 Li, S. and McCarthy, S. Further investigations on the1997, 56, 209–213.
19 Weir, N. A., Buchanan, F. J., Orr, J. F., Farrar, D. F. and hydrolytic degradation of poly (DL-lactide), Biomaterials,
1999, 20, 35–44.Boyd, A. Processing, annealing and sterilisation of poly-L-
lactide, Biomaterials, 2004, 25, 3939–3949. 33 Li, S. M., Garreau, H. and Vert, M. Structure-property
relationships in the case of the degradation of massive20 Weir, N. A., Buchanan, F. J., Orr, J. F. and Dickson, G. R.
Degradation of poly-L-lactide: Part 1—in vitro and in vivo aliphatic poly-(a-hydroxy acids) in aqueous media, Part 1:
poly(DL-lactic acid), J. Mater. Sci. Mater. Med., 1990,physiological temperature degradation, Proc. Instn Mech.
Engrs, Part H: J. Engineering in Medicine (article in press). 1, 123–130.
34 Siemann, U. The influence of water on the glass transition21 Li, S. Hydrolytic degradation characteristics of aliphatic
polyesters derived from lactic and glycolic acids, J. Biomed. of poly(dl-lactic acid), Thermochimica Acta, 1985, 85,
513–516.Mater. Res. (Appl. Biomater.), 1999, 48, 342–353.
H01204 © IMechE 2004Proc. Instn Mech. Engrs Vol. 218 Part H: J. Engineering in Medicine