You are on page 1of 12

Rheological Properties of Poly(lactides).

Effect of Molecular Weight and Temperature on the Viscoelasticity of Poly(l-lactic acid)


JUSTIN J. COOPER-WHITE, MICHAEL E. MACKAY Materials Characterisation and Processing Centre, and The Cooperative Research Centre for International Food Manufacture and Packaging Science, Department of Chemical Engineering, The University of Queensland, Queensland 4072, Australia

Received 25 August 1998; revised 29 March 1999; accepted 30 March 1999

ABSTRACT: The dynamic viscoelastic behavior of Poly(l-lactic acid) (PLLA), with molecular weights ranging from 2,000 to 360,000, have been studied over a broad range of reduced frequencies (approximately 1 103 s1 to 1 103 s1), using timetemperature superposition principle. Melts are shown to have a critical molecular weight, M c , of approximately 16,000 g/mol, and an entanglement density of 0.16 mmol/cm3 (at 25C). PLLA polymers are noted to require substantially larger molecular weights in order to display similar melt viscoelastic behavior, at a given temperature, as that for conventional non-biodegradable polymers such as polystyrene. The reason for this deviation is suspected to be due to steric hindrance, resulting from excessive coil expansion or other tertiary chain interactions. PLLA melts show a dependence of 0 on 0 chain length to the 4.0 power ( M 4.0 W ), whilst J e is independent of M W in the terminal region. Low molecular weight PLLA ( 40,000) shows Newtonian-like behavior at shear rates typical of those achieved during lm extrusion. 1999 John Wiley & Sons, Inc. J Polym Sci B: Polym Phys 37: 18031814, 1999

Keywords: poly(lactic acid); melt viscoelastic behavior; biodegradable polymer; terminal region properties; molecular weight effects; temperature effects

INTRODUCTION
Poly(lactic acid) (PLA) polymers and associated co-polymers are biocompatible, biodegradable polymers that are easily synthesized from renewable raw materials. The physical and mechanical properties of these polymers rival those of many conventional petrochemical plastics,1 yet they degrade via simple mechanisms to lactic acid and low molecular weight oligomers, all of which are metabolized by both soil and marine organisms.2 Although traditionally used for biomedical appli-

Correspondence to: M. E. Mackay (E-mail: mmackay@ stevens-tech.edu)


Journal of Polymer Science: Part B: Polymer Physics, Vol. 37, 18031814 (1999) 1999 John Wiley & Sons, Inc. CCC 0887-6266/99/151803-12

cations,35 PLA polymers are rapidly gaining recognition as biodegradable thermoplastics for general use applications, especially single-use packaging and consumer goods.4,6,7 Much research has been performed with biomedical applications in mind, including orthopedic applications and drug-release systems, by exploring thermal stability and biocompatibility, mechanical strength and rates of degradation in both in vitro and in vivo conditions.8 10 From these investigations, it has been observed that isotactic poly(l-lactic acid) (PLLA) polymers have favorable mechanical characteristics over their atactic, co-polymer relatives (poly(d,l-lactic acid)), largely due to the potential for high crystallinity within the isotactic materials. In addition to this observation, it has been noted that the use of high
1803

1804

COOPER-WHITE AND MACKAY

molecular weight PLLA chains alone, or to a substantial percentage, with co-polymer mixtures, also provides for a reduction in degradation kinetics.11 The thermal history of these polymers affects changes of crystalline-amorphous ratios, as well as large physical aging effects in the glassyamorphous phase.12 Polymerization residuals including catalyst (typically Stannous Octoate), monomer or oligomers, and water have been reported to substantially affect the rate of degradation (or more precisely, depolymerization) during exposure to elevated temperatures.3,1315 At temperatures in excess of 190C (and in some cases even at lower temperatures) chain scission reactions, along with other noted thermal degradation pathways, cause signicant reductions in molecular weight.3,5,11 Interestingly, although processing temperatures for PLLA in excess of 190C are commonly recommended,2,7 to date little attention has been given to understanding the fundamental melt ow behavior, that is a thorough rheological characterization, of PLLA at these temperatures. Temperature and molecular weight have signicant effects on the rheological properties of polymer melts. The quantication of these dependencies eliminates the need to perform rheological measurements at multiple temperatures and molecular weights. By measuring the rheological properties at one temperature and knowing the temperature dependence (in terms of ow activation energies or shift factors; a T , the horizontal and b T , the vertical) and molecular weight dependence (in terms of parameters K and a for the terminal viscosity, 0 K ( M w ) a , see, e.g., Fox and Loshaek16), the rheological properties at any other temperature and molecular weight can be determined for that polymer system. In addition, once a thorough understanding of temperature and molecular weight dependence is known, this knowledge can be applied in modeling applications, for example, screw extrusion and general ow simulations. In this article, the viscoelastic properties of PLLA melts of varying molecular weights have been measured over a range of temperatures, frequencies, and shear rates, utilizing the parallel plate geometry. The characteristic parameters of the terminal region, such as terminal viscosity 0, elasticity coefcient A G and equilibrium compliance J 0 e , and their dependence on molecular weight and temperature are discussed. Specic polymer chain parameters, including critical entanglement molecular weight M c , entanglement

density v e , characteristic ratio C , and the steric hinderance parameter are included in this discussion.

EXPERIMENTAL
Four commercial grades of PLLA were characterized (Boehringer Ingelheim, Ingelheim, Germany, Resomers L104, L206, L210, L214). Samples were dried, as discussed below, and otherwise used as received. Rheological measurements were performed utilizing a Rheometrics (Piscataway, NJ) RDSII torsional rheometer, with 7.9 mm stainless steel parallel plates. Samples were prepared for the RDSII in an 8 mm disc press, under vacuum, at temperatures between 120 and 140C. Various strains were used prior to dynamic shear tests to ensure testing was carried out within the linear viscoelastic region. Samples were checked for degradation and/or annealing after pressing by investigating any deviations in either molecular weight or thermal properties before and after this operation. Less than 5% deviation in weight average molecular weight, molecular weight distribution or thermal properties (including percent crystallinity) were noted, except the mid-range molecular weight material which experienced an increase in percent crystallization. It can only be assumed this material was received in a quenched state. This change was accompanied by a small decrease (5%) in molecular weight and a slight broadening of the molecular weight distribution. Gaps of 0.2 0.5 mm were used to ensure a reasonable aspect ratio (ratio of plate radius to gap) was maintained, minimizing edge effects (edge fracture) during testing. Dynamic frequencies ranged from 0.01 to 200 rad/s, whilst steady shear rates of 0.01 to 100 s1 were possible at temperatures from 170 230C. All tests were performed under a nitrogen atmosphere to reduce oxidation of the polymer samples during testing. The reported low thermal stability of PLLA5 required that high temperature testing be performed over small time intervals. Degradation under dynamic shear was investigated via long time tests at each frequency; refer to Figure 1 and Table I for denitions of sample codes. Consistency of storage modulus ( G ) and loss modulus ( G ) during these tests indicated that sample degradation had not occurred during testing time envelopes. It was noted that as molecular weight increased the range of applicable frequencies and

RHEOLOGICAL PROPERTIES OF POLY(LACTIDES)

1805

Figure 1. Changes in dynamic moduli with time for PLLA at 10rad s1 and 200C.

length of testing time, available prior to noticeable decreases (10%) in both storage and loss modulus, was shortened. In addition, repeat experiments were performed at all temperatures, especially those above 200C, to ensure reproducibility and checked once again for degradation. Good reproducibility was noted. Weight-average molecular weights ( M W ) and approximate molecular weight distributions (MWD, weight to number average molecular weight ratio) of the PLLA polymers investigated were measured by means of Gel Permeation Chromatography (Waters Corporation, Milford, MA). The measurements were carried out with dichloromethane as the solvent at room temperature (21C). Calibration was performed with polystyrene molecular weight standards. The weight average molecular weights and their associated MWDs are shown in Table I. Due to the reported effect of polymerization residuals (catalyst, monomer/oligomer, wa-

ter)3,1315 on the thermal degradation of these polymers, and to allow for comparison with previous work, analyses were performed to determine the respective amounts of residual catalyst and moisture in the polymers tested. Catalyst residual concentrations within each polymer were determined via Inductively Coupled Plasma Mass Spectrometry (ICPMS). Polymer samples were digested in concentrated nitric acid prior to dilution and vortexing (to ensure homogeneity). The ICPMS instrument used was a VG PlasmaQuad (VG Elemental, Winsford, UK) in its standard conguration with a Meinhard nebulizer and a double pass, water cooled, borosilicate spray chamber. All ICPMS measurements were determined in peak jumping mode by selection of mass 120Sn (32.6% natural abundance) together with 115In (95.7% natural abundance) as the internal standard. An external calibration technique using matrix matched standards coupled with internal standardization was employed to determine all polymer Sn concentrations. The limit of detection by this technique was determined as 0.2 g Sn/kg sample on a dry weight basis. Catalyst residuals (refer to Table I) exist in all samples. However, catalyst levels of less than 50 ppm have previously been noted to contribute minimally to the degradation pathways3,14 which our samples almost meet. Moisture analysis, for as received samples (refer Table I), was performed by measurement of the weight loss from a polymer sample after 3 h at 105C (under vacuum). Residual monomer has not been removed (or determined). As stated previously, all material results provided in this article are for dried samples. Glass transition ( T G ) and melting point ( T M ) temperatures, and percent crystallinity (%C) of the PLLA samples were determined utilizing a Perkin-Elmer (Norwalk, CT) Pyris 1 Differential Scanning Calorimeter (DSC) under a helium atmosphere. Materials were tested from 25 to 250C, at a scan rate of 10C/min. A heat of fusion

Table I. Weight Average Molecular Weight ( M W ), Molecular Weight Distribution (MWD), Tin Content (Sn), As Received Moisture Content, Glass Transition Temperature ( T G ), Melting Temperature ( T M ) and As Received Percent Crystallinity (%C) for the Three Highest Molecular Weight Samples Sample L206 L210 L214 M W (g/mol) 40,000 130,000 360,000 MWD 1.8 2.2 1.5 Sn (ppm) 38 52 53 Moisture (%) 0.38 0.40 0.40 T G (C) 54 56 60 T M (C) 178 179 186 %C 50 5 60

1806

COOPER-WHITE AND MACKAY

of 93.1 J/g for 100% crystalline PLLA (i.e., H m 93.1 J/g) 17 has been used to calculate percent crystallinity using eq. (1), % Crystallinity Hm Hc 100 H m (1)

where H m is the measured heat of fusion (melting) and H c is the heat of crystallization. The glass transition temperatures, melting point temperatures, and percent crystallinity of the as received polymer samples are shown in Table I. Densities were determined via dilatometry utilizing the fully automated GNOMIX (Boulder, CO) high-pressure dilatometer. The dilatometer measures volumes at controlled temperatures (25 400C) and pressures (10 200 MPa). The accuracy of the volume measurement is 0.002 cm3/g at 25250C, and the sensitivity is 0.0002 cm3/g. Only room pressure densities are reported here.

gent (tan ) remains invariant under a temperature shift, and for this reason we can dene the * in terms of the loss tangent and temperature. The temperature dependence of linear viscoelastic data is often discussed in terms of the temperature dependence of the relaxation times of polymer systems. The Generalized Maxwell Model19 adequately details the dynamic viscoelastic properties of polymer melts in terms of a discretized relaxation spectrum, where H i ( ) is the relaxation strength of the material at relaxation time i . This model has been tted to each polymer master curve to determine relaxation spectra for the PLLA polymers from our experimental viscoelastic data. The model takes the following forms, in dynamic shear, G

H 1
i i

2 2 i 2 2 i

(5)

DATA ANALYSIS
In an effort to map the ow behavior of PLLA melts over wide ranges of frequency and temperature, the Time-Temperature Superposition (TTS) principle18 was used to generate master curves for each polymer. The reference temperature, throughout this article for all polymers, is 200C. The equations used to calculate both vertical ( b T ) and horizontal ( a T ) temperature shift factors are shown below, b TG a T , T G , T r (2) (3) (4)

H 1
i i i

2 2 i

(6)

G * G 2 G 2

(7)

where G * is the complex modulus. The complex viscosity, *, is dened as

G* .

(8)

* T , tan

aT * T r, tan bT

The rheological behavior of a polymer melt in the long-time or terminal region is generally characterised by such parameters as the terminal viscosity 0, elasticity coefcient A G , and the equilibrium (or steady state) compliance J 0 e . These are dened respectively as shown,

b T a T , T , T r

where is frequency (rad s1), G , storage modulus (Pa), *, complex viscosity (Pa s), , shear stress (Pa), , steady shear rate (s1), tan , the loss tangent, T , temperature (C), and T r , reference temperature (C). Note that the equations above are given in terms of temperature shifts for both viscoelastic (e.g., G and *) and steady shear () data. It was shown by Markovitz18 that linear viscoelastic and steady-shear data must be superposable with the same shift factors for a thermo-rheologically simple polymer system. Note also that the loss tan-

0 lim *
30

(9)

A G lim
30

G 2

(10)

J0 e lim
30

AG G lim . G 2 2 0 30

(11)

It is worthwhile noting here that A G is a very sensitive parameter to the maximum relaxation time of the melt and hence a good indicator of

RHEOLOGICAL PROPERTIES OF POLY(LACTIDES)

1807

ature is shown in Figure 2. A consistent viscosity reduction of approximately 2/3 of the previous value is observed for an increase in temperature of 10C. This result is comparable with many polyolen melts, such as polyethylene,21,22 as well as polystyrene.23 To further quantify the dependence of the viscoelastic properties of PLLA on temperature, Arrhenius-type equations (shown below) were tted to the temperature shift factors a T and b T , described previously in eqs. (2) to (4), for all molecular weights. a T exp

EH 1 1 R T Tr EV 1 1 R T Tr

T T g 100C

(14)

b T exp
Figure 2. Effect of temperature on complex viscosity for PLLA ( M w 100,000) at temperatures of 170, 180, 190, 200, 210, and 220C.

T T g 100C

(15)

elasticity. In addition, within the terminal region, the variation of the storage and loss modulus with frequency, in the limit of zero frequency, tends to slopes of two and one, respectively.20

RESULTS AND DISCUSSION


The PLLA homopolymers investigated have melting temperatures of 178 186C, with glass transitions temperatures varying from 54 60C (refer to Table I). Glass transition and melting temperatures are seen to increase slightly with increasing molecular weight. Effect of Temperature Dynamic and steady tests were performed at individual temperatures for each molecular weight. An example for the variation of * with temper-

Here, the reference temperature, T r , is again 200C (473.15 K), E H and E V are the horizontal and vertical activation energies for viscous ow (J/mol), R is the gas constant (8.314 J/mol K), and T is temperature (K). A plot of loge ( a T ) vs. 1/ T shows that an Arrhenius-type behavior is certainly exhibited by all PLLA melts investigated ( r 2 0.998). The horizontal activation energy E H was shown to increase with increasing molecular weight, from 76 kJ/mol for the lowest molecular weight PLLA, up to 85 kJ/mol for the highest molecular weight investigated (refer to Table II). Porter and Johnson (1960) noted similar behavior for polyethylenes.24 Interestingly, Rudd (1960) reported that the horizontal activation energy for ow for polystyrene is independent of molecular weight above 50,000 g/mol.21 Similarly, a plot of loge ( b T ) vs. 1/ T shows that the values for b T vary little from unity and the calculated values for the vertical activation energy can be considered to be approximately zero (within the limits of the experiment); refer to Table II. Note that for b T 1.0, eqs. (2), (3), and

Table II. Activation Energies for Flow of PLLA Melts Polymer Horizontal activation energy E H (kJ/mol K) Vertical activation energy E V (kJ/mol K) L206 76.0 ( r 2 0.997) 4.0 ( r 2 0.999) L210 79.0 ( r 2 0.999) 4.0 ( r 2 0.999) L214 85.0 ( r 2 0.994) 4.0 ( r 2 0.999)

1808

COOPER-WHITE AND MACKAY

Figure 3. Effect of molecular weight on viscosity for PLLA melts at 200C.

(4) simplify to functions of a T only. Hence, as expected for a linear aliphatic polymer,22 only an horizontal activation energy E H (or the horizontal shift factor a T ) need to be considered to describe the temperature dependence of PLLA melts. Effect of Molecular Weight Figure 3 shows a plot of complex viscosity ( *), as a function of frequency and molecular weight for the series of PLLA polymers investigated. Steady shear experiments were also performed to determine the applicability of the Cox-Merz rule25 to the PLLA materials. The results of these tests have been included in Figure 3. Good agreement was noted between dynamic and steady viscosity for both L206 and L210. Agreement between dynamic and steady behavior was difcult to observe for the high molecular weight material (L214), even at very low frequencies, due to notable sample edge fracture and degradation under steady shear (not shown here). Newtonian-like behavior was observed for polymer L206, within the shear rates typical of those encountered during lm extrusion ( 100 s1). However, it was observed that this Newtonian-like behavior was noticeably shortened with increasing molecular weight. Comparison of the Master Curves for all molecular weights tested, shown in Figure 4, indicates an increasing melt elasticity ( G ) with mo-

lecular weight, even leading towards a possible rubber-like plateau for the very high molecular weight. In addition, we note the adherence of the storage and loss modulus data in the low frequency region (long time region) to slopes of two and one, respectively. This indicates that, for all molecular weights tested, the data is indeed within (and extending beyond) the terminal region. The lack of an apparent rubber plateau in the melt phase, especially with decreasing molecular weight, is in fact observed for many polyesters, for example, polycaprolactam and poly(ethylene terephthalate),26 and is thought to be due to the lower molecular weight materials crystallizing below their melting point (i.e., low temperatures extend the high frequency testing range and crystallization affects the rheological measurement). It is worthwhile noting that this plateau may also have been obscured by the relatively broad distribution of molecular weights of all materials tested (as noted by Masuda et al.27 with respect to broadly distributed polystyrenes). Higher frequencies could not be used due to limitations of the rheometer, thus, tests below T m and timetemperature superposition were needed to extend the frequency range in order to observe the start of the plateau region for the higher molecular weight PLLAs. It was conrmed negligible crystallization occurred over the short testing time. In the plateau region or rubbery zone, G for high molecular weight polymers is con-

Figure 4. Master curves of storage and loss modulus for PLLA melts at 200C. Solid lines are ts according to eqs. (5) and (6).

RHEOLOGICAL PROPERTIES OF POLY(LACTIDES)

1809

Table III. Relaxation Spectrum for PLLA Melts L206 T r 200C (Relaxation modes) 1 2 3 4 5 6 H() (Pa) 6,100 117,400 H() (Pa) 1,310 20,300 88,100 259,900 L210 H() (Pa) 5,840 11,200 42,800 107,000 186,000 293,000 L214

(s)
5.6 103 5.9 104

(s)
0.664 0.101 0.0235 4.65 103

(s)
68.0 10.6 2.22 0.562 0.144 0.0255

stant, independent of frequency and molecular weight, and is denoted as the plateau modulus G0 n . The absence of a plateau for these PLLAs negates the plateau modulus being observed effectively via this method. The Generalized Maxwell Model19 (GMM) has been tted to the storage and loss modulus data for all molecular weights with good agreement (refer to Fig. 4). Similar relaxation spectra was determined from both G and G data. Relaxation strengths and times are presented in Table III. With increasing molecular weight, the number of relaxation modes needed for a good t also increased. Maximum relaxation times ( 1 ) are shown in Figure 4. It is well understood that the viscoelastic properties in the terminal zone are dominated by the terminal relaxation time and is itself determined by long-range motions in which a molecule of high molecular weight may be constrained by the order of 100 entanglements.28 Considering processing effects, we see that the terminal relaxation time is directly related to the time required for internal stresses to relax during an annealing process. It is also a close approximation of the time required to attain steady-state ow under constant stress, or for elastic recoil to be accomplished after removal of stress (assuming linear viscoelastic behavior).28

Once a discrete relaxation spectra is known for a given polymer, the plateau modulus, G 0 n , may be estimated via simple extension of the frequency range of G . Alternative estimates of G 0 n can be evaluated27 from the integration of G vs. ln( ), i.e., 2 H ln G ln .
i i i i i

G0 n

(16)

The plateau modulus was estimated via the two methods described above, for the two highest molecular weight PLLAs, L210 and L214. For L210, 5 5 G0 n varied from 2.5 10 to 3.7 10 Pa, whilst 5 L214 showed values of 4.5 10 to 6.5 105 Pa. We note reasonable scatter of the results for G 0 n determined from these two methods. Further discussion as to the most appropriate value of the plateau modulus for this polymer is given below. Characteristic values of the terminal region, notably, terminal viscosity 0, elasticity coefcient A G , and equilibrium compliance J 0 e , have been determined from the data and are presented in Table IV at 200C. Values for 0 were obtained from the data at small frequencies in the terminal region. This calculation inherently assumes that

Table IV. Molecular Weight, Terminal Viscosity, Elasticity Coefcient, and Equilibrium Compliance for All Samples at 200C Sample L104 L206 L210 L214 M W (g/mol) 2,000 40,000 130,000 360,000

0 (Pa s)
0.03 100 6,200 7.0 105

A G (Pa s2) 0.23 840 7.1 107

1 J0 ) e (Pa

2.3 105 2.2 105 2.2 105

1810

COOPER-WHITE AND MACKAY

the steady viscosity, , is equivalent to the dynamic (complex) viscosity, *. This has been shown to be an accurate assumption for the shear ranges tested (refer Fig. 3). A G and J 0 e were determined according to eqs. (10) and (11). The possible effects of MWD on these measured constants will now be discussed. Both the elasticity coefcient ( A G ) and steady-state compliance ( J 0 e ) are quite sensitive to variations in MWD, and are normally seen to increase with increasing MWD.28 The terminal viscosity 0 is essentially dened by M w (except possibly at very high molecular weight28) and shows little effect from changes in MWD at constant M w . For all the samples tested, because of the small variation in molecular weight distribution (i.e., M w / M n of approximately two), we may assume that the relationships we have determined for 0 , A G , and J 0 e below are not inuenced by MWD and are representative of true molecular weight effects only. The empirical equation described by Fox and Loshaek,16 relating viscosity at zero shear rate ( 0 ) to molecular weight for linear amorphous polymers, is used here to compare PLLA melts with those of conventional polymers. The form of this equation, detailed below, has been applied to the elasticity coefcient, A G , to further quantify the effect of molecular weight on the elasticity,

Figure 5. Effect of molecular weight on zero shear viscosity and elasticity coefcient for PLLA at 200C.

shown by others to be around 7.5 for monodisperse polystyrene.23 From Figure 5, we see that for PLLA melts at T r (within the molecular weight range tested),

0 K M w a
A G K M w .
b

(17) (18)

0 3 10 17 Pa s M W 4.0
A G 2 10 38 Pa s 2 M W 8.0. The value for the exponent of molecular weight, with respect to terminal viscosity, is slightly higher than the generally accepted value of 3.4. The elasticity coefcient for PLLA melts shows a higher dependence on molecular weight, at a value of 8.0, than that observed for monodisperse polystyrene melts. Reference to Table IV indicates that J 0 e is independent of M W in the terminal region (this is also observed for other conventional plastics above M c ). 32 The fact that J0 e is independent of M w suggests that all samples were tested well into the entanglement region (see below). We note that only three data points were used in the regressions to determine the power law exponents for 0 and A G with M W which could have signicant error. However, we are encouraged by the nding of J 0 e being independent of M W showing that at least the data is consistent with expected trends. The value of M c cannot be explicitly calculated from the molecular weights tested. However, uti-

Within these equations, the constants K and K depend upon the polymer type, its number molecular weight29 ( M n ) and temperature. The power law factors a and b dene the slope of the log viscosity and log elasticity coefcient, respectively, vs. log molecular weight plots. The molecular weight exponent a has been predicted theoretically to have a universal value of 3.4 above M c , the critical molecular weight of entanglements, for linear exible polymers.28,30 In addition, many systems have been shown to follow this relation. This exaggerated dependence was described by Beuche30 to be a result of entanglement coupling, hence the label M c . De Gennes31 has also comprehensively described the molecular weight dependence of the terminal viscosity by his reptation theory. Below M c , the exponential factor a approaches unity.28,30 The dependence of the elasticity coefcient on molecular weight, described explicitly by the exponent b , has been

RHEOLOGICAL PROPERTIES OF POLY(LACTIDES)

1811

lizing the fact that below M c , the zero shear viscosity shows a near linear dependence of molecular weight,30 we may postulate a possible range for this parameter. With this in mind, the zero shear viscosity, at 200C, of a PLLA of molecular weight approximating 2,000 g/mol (refer to Table IV) has been included in Figure 5. Passing through this data point is a line of slope unity. From the intersection of these two regions, we note that the M C is of order 10,000 g/mol. Obviously, this method assumes that a polymer of molecular weight of 2,000 is below the critical molecular weight which seems a reasonable assumption based on the data. This range of values compare well with 1516,000 for poly(-caprolactone).33 However, considering that M C is 38,000 (i.e., M C / M r 360, where M r is molecular weight of repeat unit) for conventional polystyrene34 and 2,800 ( M c / M r 200) for linear polyethylenes,34 we see that if we compare the number of monomer segments which contribute to these critical molecular weights for each polymer, we now note that PLLA ( M C / M r 140) has a ratio close to that of polyethylene, and is far from that of conventional polystyrene. This is not the expected result, especially considering the rigid structure of the repeat unit for PLLA, as will be discussed later. An alternative estimate of M c can also be derived utilizing the following calculation sequence as outlined by Wu26 (see also references 26 and 34 for further explanation of the relevant calculation methods and associated theory). From our measurements of the dynamic moduli with frequency, we have determined a value for the crossover modulus, G c (the value of G and G at the crossover frequency), for the two higher molecular weight polymers, L210 and L214 (at T r 200C), of 1.25 105 Pa and 1.30 105 Pa, respectively. The plateau modulus ( G 0 n ) for PLLA may now be calculated to be approximately 8.0 105 Pa from Wus empirical relation,26 and using the data provided in Table I, 2.63 log p 1 2.45 log p

for over 40 other polymer species. Considering that the determination of G 0 n from the relaxation spectra is from an extension of the actual experimental data, and that Wus correlation is empirical, the agreement is reasonable, yet not precise. If the results from all methods are compared, we 5 nd an average value for G 0 n of 5.5 10 Pa. This value will be used in all calculations below. Further, when this chosen value is compared to that cited by Grijpma et al.35 of approximately 5 105 Pa for G 0 n at 200C, we note close agreement. The molecular weight between entanglements, M e , may now be calculated from the plateau modulus, utilizing eq. (20),

Me

RT G0 n

(20)

log G 0 N/ G c 0.380

(19)

where p is the polydispersity index ( M W / M N ). When these results are compared with those values obtained from our relaxation spectra, we see all results are of similar orders of magnitude. This agreement should be expected considering that eq. (19) was shown by Wu26 to be applicable

where is the melt density (kg/m3) at the chosen reference temperature. For a polymer melt, the molecular weight between entanglements is determined by the intrinsic stiffness of the polymer chain. From the melt through to the solid, even throughout crystallization, the molecular weight between entanglements remains constant.36 However, the entanglement density v e (discussed below) and morphology may vary via different crystallization catalysts and temperatures.37 Below their glass transition temperature, M e is one of the main parameters in determining the mechanical properties of polymer systems. For PLLA polymers, M e is estimated to be approximately 8,000 g/mol, for a melt density of 1,090 kg m3 at 200C determined with the GNOMIX PVT apparatus. M c is approximately equal to 2 M e , 28 hence, we see that from our results, that M c is expected to be of order 16,000 g/mol. This result supports our previous estimate via the terminal viscosity vs. molecular weight data. Little information is cited within the literature, excepting that indirectly provided by Grijpma et al.,35 as to a reference or agreed value of M c for PLLA polymers. Grijpma et al. calculated a molecular weight between entanglements ( M e ), for PLLA melts of 10,500 g/mol at 200C.35 This value was determined via a similar experimental procedure as used in this work, and using Wus calculation method.26 Considering the above relation between M c and M e , this provides for an M c value in the range of 21,000 g/mol. This result was determined using a melt density of 1,245 kg/m3 and, in fact, this difference is signicant in their determination of M e . The slight difference

1812

COOPER-WHITE AND MACKAY

between the G 0 n values determined from our dynamic experiments and those of Grijpma et al. produce similar deviations for M e when using eq. (20) when a melt density of 1,090 kg/m3 is used instead within their calculations (we nd M e from their investigation is around 9,000 g/mol, or an M c of 18,000 g/mol). It is well accepted that chain entanglement is a key factor controlling such polymer behaviors as melt rheology, and solid mechanical and adhesive properties.26 Thus, accurate determination of M e is critical and our experimental results give 8,000 g/mol, which is close to the corrected value of Grijpma et al. Characteristic Parameters We calculate two important polymer chain parameters based on M e , the characteristic ratio, C , and the entanglement density, v e . In addition, the steric hinderance parameter, , is discussed, in an attempt to further quantify the observed melt properties of PLLA. The characteristic ratio, C , is a measure of the intrinsic exibility and rigidity of a coiled chain. It is dened as
2 C lim R 2 0 / n v / v nv3

where a is the amorphous density. In addition, Wu26 has shown, via his topological model, that C can be predicted purely from the polymers chemical structure, once the molecular weight between entanglements, M e , is known, i.e., C


n vM e 3Mr

1/2

(23)

(21)

where R 2 0 is the mean-square end-to-end distance of an unperturbed coiled chain, and n v l 2 v the mean-square end-to-end distance of an equivalent random-ight chain. Essentially, this parameter is the inherent rigidity associated with the individual chain segments (i.e., bond angles and lengths) of the polymer species, which determines the overall capability of the chain to rotate around its own axis, resulting in either exible or rigid behavior of the bulk polymer. Bulk polymer behavior is, obviously, not only dependent on the inherent exible or rigid nature of these chains, but also, amongst other things, on how entangled they may be with other surrounding chains. The number of chain entanglements per unit volume or the entanglement density ( v e ) is a useful parameter to quantify this. Knowledge of these two parameters will provide insight into the expected chain conformation and entanglement behavior of poly(l-lactic acid) melts. The entanglement density may be simply calculated26 from M e , ve

a Me

(22)

where n v is the number of real or virtual repeating bonds per monomer unit around which the polymer chain is able to rotate, and M r , the molecular weight of the repeat unit. From our experimental results, we nd the entanglement density ( v e ) is approximately 0.16 mmol/cm3 (25C) and C is 10.5. This calculation assumes three real or virtual repeating bonds per monomer unit,38 M v is 72 g/mol and a is 1270 kg m3 at 25C (determined with the GNOMIX PVT apparatus) via extrapolation of the melt density. These are compared to the characteristic chain parameters for isotactic polystyrene [ C of 13.9, and v e of 0.036 mmol/cm3 (25C)], 34 and for polyethylene [ C of 5.7, and v e of 0.615 mmol/cm3 (25C)]. 34 We see that, according to Wus theories, solid PLLA should describe intermediate brittleductile behavior between that observed for isotactic polystyrene (brittle) and polyethylene (ductile). Grijpma et al. (1994) observed substantially brittle behavior for amorphous solid (bulk-polymerised) PLLA in solid tensile and impact tests,35 and reported characteristic ratios and calculated entanglement densities for their bulk-polymerised PLLA of 11.8 and 0.14 mmol/cm3 (25C),35 respectively. The differences between the estimates provided by Grijpma et al. and our work are again largely suspected to be due to differences in calculated amorphous density and melt density values for PLLA samples, and slightly different G 0 n values determined from our dynamic experiments. Despite these differences in v e and C , both results point to substantially brittle solid behavior. However, considering the similarity of these structural chain parameters to those of conventional polystyrene, we might expect PLLA polymers to show similar melt behavior at a given temperature, for equivalent molecular weight and polydispersity. Interestingly, this is not the case. PLLA polymers require substantially larger molecular weights to display similar melt behavior as that noted for polystyrene at a given temperature (unpublished). A possible reason for this difference between these two polymer

RHEOLOGICAL PROPERTIES OF POLY(LACTIDES)

1813

species, in particular, lies in the steric hindrance associated with the PLLA chain. The steric hindrance parameter, , is dened as the ratio of the mean-square end-to-end distance of an unperturbed coiled chain to the meansquare end-to-end distance of a freely rotating chain. That is, this ratio provides for a direct comparison between the movement of the real chain to that of a freely rotating chain, a chain in which the bond angles and lengths are now set, but the angle of rotation around each bond is not. This parameter highlights different polymer segment exibility or rigidity, specically due to the steric hindrance. For isotactic PLLA, Schindler and Harper39 found i to be 2.69, compared to that of isotactic PS of 2.30, and polypropylene (PP) of 1.66. In addition, the ratio of the steric hindrance parameters of isotactic ( i ) and atactic ( a ) PLA polymers has been shown to be higher than those recorded for many other polymers, including polystyrene. Whereas stereoregularity for polystyrene (and most other polymers studied by Schindler and Harper) seems to have little effect on the coil dimensions, PLA polymers show a ratio of i / a of 1.36, a signicant variation. This difference has been suggested by Schindler and Harper, to indicate a chain conformation that produces an expansion of the coils dimensions.39 This difference is not obvious via characteristic ratio comparisons. In light of this discussion, it is suggested that a predominance of helical sequences and possible chemical shift differences within and between isotactic PLLA polymer chains (including tertiary chain to chain interactions) may then provide possible reasons for this anomaly, although no evidence is provided in the present literature. However, the existence of this anomaly provides some reasoning as to the interesting behavior observed for this polymer melt. It may be that the above chain structure anomaly causes the larger power law exponents for the viscosity and elasticity coefcient with molecular weight. Interestingly, rigid polymers are noted for also showing greater power law exponents in eqs. (17) and (18).40

temperature superposition principle. PLLA melts have a critical molecular weight for entanglement, M c , of approximately 16,000 g/mol, and an entanglement density of 0.16 mmol/cm3 at 25C. PLLA melts show power law exponents for 0 and A G of 4.0 and 8.0 with M W , respectively, whilst J 0 e is independent of M W in the terminal region. These exponents are slightly higher than those determined for other polymers. This deviation is suspected to be due to steric hindrance, a result of excessive coil expansion produced by possible chemical shift differences within and between isotactic PLLA polymer chains (including tertiary chain to chain interactions). Further work is required to detail the exact nature of this anomaly. Low molecular weight PLLA ( 40,000) shows Newtonian-like behavior at shear rates typical of those achieved during lm extrusion.
Justin Cooper-White is grateful for a scholarship from the Cooperative Research Centre for International Food Manufacture and Packaging Science. Partial funding for this project was also provided by this center.

REFERENCES AND NOTES


1. Wehrenberg, R. H., II. Mat Eng 1981, 94, 63 66. 2. Spinn, M., et al. J M S Pure Appl Chem 1996, A33, 14971530. 3. Jamshidi, K.; Hyon, S.-H.; Ikada, Y. Polymer 1988, 29, 2229 2234. 4. Lipinsky, E. S.; Sinclair, R. G. C Eng Prog 1986, 82, 26 32. 5. Kopinke, F. D.; Remmler, M.; MacKenzie, K.; Moder, M.; Wachsen, O. Polym Deg Stab 1996, 53, 329 342. 6. Jopski, T. Kunstoffe 1993, 10, 748 751. 7. Sinclair, R. G. J M S Pure Appl Chem 1996, A33, 585597. 8. Vert, M.; Chabot, F. Makromol Chem 1981, 5, 30. 9. Tunc, D. C. Trans Soc Biomater 1983, 6, 47. 10. Vert, M.; Torres, A.; Li, S. M.; Roussos, S.; Garreau, H. In Studies in Polymer Science, Vol. 12: Biodegradeble Plastics and Polymers; Doi, Y. and Fukuda, K., eds.; Ezseuier: Amsterdam, 1994; p.11. 11. Migliaresi, C.; Cohn, D.; DeLollis, A.; Fambri, L. J Appl Polym Sci 1991, 43, 8395. 12. Celli, A.; Scandola, M. Polymer 1992, 33, 2699 2703. 13. Gogolewski, S.; Mainil-Varlet, P. Biomaterials 1996, 17, 523528. 14. Cam, D.; Marucci, M. Polymer 1997, 38, 1879 1884. 15. Sodergard, A.; Nasman, J. H. Ind Eng Chem Res 1996, 35, 732735.

CONCLUSION
The dynamic viscoelastic behavior of poly(l-lactic acid) (PLLA), with molecular weights ranging from 2,000 to 360,000, has been studied over a broad range of reduced frequencies (approximately 1 103 s1 to 1 103 s1), using time

1814

COOPER-WHITE AND MACKAY

16. Fox, T. G.; Loshaek, S. J Appl Phys 1955, 26, 1080 1082. 17. Fischer, E. W.; Sterzel, H. J.; Wegner, G. Kolloid-Z Polym 1973, 251, 980 990. 18. Markovitz, H. J Polym Sci: Symp 1975, 50, 431 456. 19. Verser, D. W.; Maxwell, B. Polym Eng Sci 1970, 10, 2. 20. Ferry, J. D.; Fitzgerald, E. R.; Grandine, Jr., L. D.; Williams, M. L. Ind Eng Chem 1952, 44, 703706. 21. Rudd, J. F. J Polym Sci 1960, 154, 459 474. 22. Mavridis, H.; Shroff, R. N. Polym Eng Sci 1992, 32, 1778 1791. 23. Onogi, S.; Kato, H.; Ueki, S.; Ibaragi, T. J Polym Sci: Part C 1966, 15, 481 494. 24. Porter, R. S.; Johnson, J. F. J Appl Polym Sci 1960, 3, 194 199. 25. Cox, W. P.; Merz, E. H. J Polym Sci 1958, 28, 619. 26. Wu, S. J Polym Sci B, Polym Phys 1989, 27, 723 741. 27. Masuda, T.; Kitagawa, K.; Inoue, T.; Onogi, S. Macromolecules 1970, 3, 116 125. 28. Ferry, J. D. Viscoelastic Properties of Polymers, Vol. 3; Wiley: New York, 1980.

29. Smith, O. W.; Koleske, J. V. J Paint Technol 1973, 45, 60 65. 30. Bueche, F. J Chem Phys 1952, 20, 1959. 31. De Gennes, P. G. J Chem Phys 1971, 55, 572579. 32. Graessley, W. W. Adv Polym Sci 1974, 16, 1. 33. Grosvenor, M. P.; Staniforth, J. N. Int J Pharmaceut 1996, 135, 103109. 34. Wu, S. Polym Eng Sci 1992, 32, 823 830. 35. Grijpma, D. W.; Penning, J. P.; Pennings, A. J. Colloid Polym Sci 1994, 272, 1068 1081. 36. Mandelkern, L. Physical Properties of Polymers; American Chemical Society: Washington, DC, 1984. 37. Nijenhuis, D. W.; Pennings, A. J. Macromolecules 1992, 25, 6419. 38. Joziasse, C. A. P.; Veenstra, H.; Grijpma, D. W.; Pennings, A. J. Macromol Chem Phys 1996, 197, 2219 2229. 39. Schindler, A.; Harper, D. J Polym Sci, Polym Chem Ed 1979, 17, 25932599. 40. Doi, M.; Edwards, S. F. The Theory of Polymer Dynamics; Oxford University Press: Oxford, 1986.

You might also like