Biomacromolecules 2004, 5, 397-406
397
Biodegradable Cellulose Diacetate-graft-poly(L-lactide)s: Thermal Treatment Effect on the Development of Supramolecular Structures Yoshikuni Teramoto and Yoshiyuki Nishio* Division of Forest and Biomaterials Science, Graduate School of Agriculture, Kyoto University, Sakyo-ku, Kyoto 606-8502, Japan Received November 5, 2003
Thermal transition property of cellulose diacetate-graft-poly(L-lactide) (CDA-g-PLLA) varies depending seriously on the molar substitution (MS) of lactyl unit, as represented by a drastic Tg depression with increasing MS (0 < MS e 8) and a subsequent crystallization of the PLLA side chains (MS g 14). To make clear the thermally induced development of supramolecular structures for this series of graft copolymers, physical aging and crystallization experiments were conducted under isothermal conditions at temperatures, respectively, comparable to and higher than their Tg’s. For aged copolymers with lower MSs of 4.7 and 22, an enthalpy relaxation was followed by differential scanning calorimetry. The analysis of time evolution of the relaxed enthalpy in terms of a Kohlrausch-Williams-Watts relation revealed that the overall relaxation time and the distribution of relaxation times were, respectively, rather longer and much narrower compared with the corresponding data for plain PLLA. For crystallized copolymers of MS ) 22-77, a spherulite formation was observed by polarized optical microscopy. The growth rate was much lower than that for PLLA per se, and the developed texture usually contained banded extinction rings unlike the homopolymer. The slower growth kinetics was analyzed quantitatively to estimate the interfacial free energy of PLLA crystals constituting the spherulites, by using a folded-chain crystallization formula expanded for a binary mixing system composed of a miscible crystalline/amorphous polymer pair. Discussion of these experimental results took into consideration the effect of the CDA backbone as “anchoring substrate” and “linked diluent” for the PLLA grafts. 1. Introduction Graft copolymerization of cellulose acetate (CA) is an alternative plasticization method to avoid a bleading-out problem attending the practical usage of low molecular weight plasticizers for the hardly processible cellulose ester,1,2 as well as blending CA compatibly with flexible polymers.3,4 CAs with a degree of substitution (DS) of 100 °C (for g-CDA22) or of >110 °C (for the others) were adopted to the fitting by taking account of a crystallization-regime transition (II T III)13 that is described below. The Teq m and φ values thus evaluated are compiled as a function of graft composition in Table 2. A value of Teq m ) 177 °C was obtained for PLLA-L, and the following were obtained for the graft copolymers: Teq m ) 173 °C (gCDA77), 162 °C (g-CDA58), and 195 °C (g-CDA22). The Teq m estimated for PLLA-L is rather lower compared with those (207-227 °C) found in the literature14,33 dealing with PLLA crystallization. A potential cause is that the detection of Tm′ was conducted with POM, differing from the previous studies by DSC. In addition, the lower molecular weight of PLLA-L may be responsible for the assessment of such a low Teq m value. The small φ value of 0.192 for PLLA-L suggests that the crystals formed isothermally are comparatively stable in this homopolymer, whereas the larger φ
402
Biomacromolecules, Vol. 5, No. 2, 2004
Teramoto and Nishio
Figure 6. Polarized optical micrographs of typical spherulites observed for CDA-g-PLLAs and plain PLLAs: (a) g-CDA22 at Tic ) 107 °C and t ) 4000 min; (b) g-CDA58 at Tic ) 119 °C and t ) 365 min; (c) g-CDA77 at Tic ) 110 °C and t ) 300 min; (d) PLLA-L at Tic ) 130 °C and t ) 30 min; (e) PLLA-H at Tic ) 134 °C and t ) 50 min. Tic denotes the isothermal crystallization temperature and t the elapsed time.
values 0.44-0.70 for the CDA-g-PLLAs indicate that the crystallization process of the PLLA graft chains was less stable. In particular, it should be noted for g-CDA22 that the slope of the Tm′ vs Tic plot is so steep (φ ) 0.703) that the line crosses the other three before the intersection with Tm′ ) Tic, although the observed Tm’s are usually lower than those of the other two graft copolymers (Figure 7). As is also suggested in the later discussion, there is a possibility that the crystallization of this sample of MS ) 22 (wPLLA ) 0.86) with shorter side chains progresses in a manner different from that of the other graft copolymers, for instance,
with respect to the exclusion of the CDA constituent from the crystalline regions of PLLA. 3.2.3. Growth Kinetics. The spherulitic growth rate G was measured for above-mentioned PLLA-L and three CDAg-PLLAs at the desired isothermal crystallization temperatures Tic’s. Figure 8 exemplifies plots of the radius r of growing spherulites as a function of elapsing time, all of the data bearing out a linear relationship. From the slope of each straight line, the growth rate G ) dr(dt)-1 was determined. The dependence of G values thus obtained on Tic is depicted by semilogarithmic representation in Figure 9 for
Supramolecular Structure Development of CDA-g-PLLA
Biomacromolecules, Vol. 5, No. 2, 2004 403
Figure 7. Hoffman-Weeks plots for isothermally crystallized CDAg-PLLAs and PLLA-L. See text for line drawings. Table 2. Equilibrium Melting Temperature (Teq m ) and Stability Parameter (φ) Estimated from the Hoffman-Weeks Plots and Nucleation Factor (Kg) and Surface Free Energy (σe) Estimated from the Quantitative Analysis of Crystallization Kinetics sample code
Teq m (°C)
g-CDA22 g-CDA58 g-CDA77 PLLA-L
195 162 173 177
φ
Kg × 10-5 (K2)
σe × 105 (J cm-2)
0.703 0.438 0.488 0.192
6.45 1.33 1.78 1.45
1.47 0.33 0.43 0.34
Figure 9. Spherulitic growth rate G as a function of crystallization temperature Tic for CDA-g-PLLAs and PLLA-L.
of the crystallization half-time measured by DSC,13 where the result was discussed in relation to a transition of crystallization regime (II T III), which is probably applicable to our present case. The temperature dependence of radial growth rate, G, estimated above was employed to calculate the surface energy, σe, of folded lamellar crystals in terms of a kinetic theory of polymer crystallization.35,36 In this theory, G is described in the following form37 G ) G0 exp(-∆ED*/(kTic)) exp(-∆Φ*/(kTic))
(4)
where G0 is a preexponential factor generally assumed to be constant or proportional to Tic, ∆ED* is an activation energy for the transport of crystallizing units across the molten liquid-solid interface, ∆Φ* is a free energy required to form a nucleus of critical size, and k is the Boltzman constant. For a polymer-diluent system,38-41 eq 4 was modified as follows: G ) V2G0 exp(-∆ED*/(kTic)) exp(-∆Φ*/(kTic))
Figure 8. Growth of spherulites of CDA-g-PLLAs and PLLA-L represented by radius vs time plots.
the four samples. It is found from the comparison that the incorporation of PLLA as grafting side chain gives rise to a marked drop of G by 1 order (g-CDA58 and g-CDA77) or by more than 2 orders (g-CDA22) in magnitude relative to the corresponding values for the uncombined PLLA. The lowering tendency of G for PLLA grafts with decreasing MS can be taken as being due to enhanced restriction in translational diffusion of the side chains dangling onto the CDA backbone. Miyata et al. formerly fitted their data of the Tic dependence of G for PLLA homopolymers to monomodal curves.34 However, the variations of G shown in Figure 9 appear to be essentially bimodal for all of the PLLA-L and CDA-gPLLAs, which became more unambiguous by plotting G in a normal coordinate system. If two line-curves are necessary to fit the data separately in higher and lower temperature regions for the respective samples, a “break” occurs at ∼110 °C (PLLA-L), ∼117 °C (g-CDA77), ∼115 °C (g-CDA58), and ∼100 °C (g-CDA22). For plain PLLA, a similar bimodal phenomenon has been noted for the temperature dependence
(5)
In eq 5, the preexponential factor is multiplied by a volume fraction V2 of the crystalline polymer because the rate of nucleation is proportional to the concentration of crystallizable units. For g-CDA22, g-CDA58, g-CDA77, and PLLAL, the V2 values were determined as 0.854, 0.938, 0.953, and 1, respectively, with the density data 1.34542 and 1.37843 for CDA and PLLA, respectively, together with wPLLA fractions. The transport term ∆ED* in eq 5 can be calculated by means of the Williams-Landel-Ferry relation:44 ∆ED*/(kTic) ) C1/[R(C2 + Tic - Tg)]
(6)
where C1 and C2 are constants, generally assumed as 1.724 × 10-4 J mol-1 and 51.6 K, respectively, and R is the gas constant. Concerning the other term in eq 5, ∆Φ* may be expressed as38,40 -∆Φ*/(kTic) ) -Kg/(Tic∆T) + (2σTeq m ln V2)/(b0∆Hu∆T) (7) with
404
Biomacromolecules, Vol. 5, No. 2, 2004
Teramoto and Nishio
Kg ) (Yb0σσeTeq m )/(k∆Hu)
(8)
where Kg is a nucleation factor for folded chain crystallization,35,36 ∆T is undercooling defined as Teq m - Tic, ∆Hu is the enthalpy of fusion per unit volume of the crystalline component, σ and σe are the interfacial free energies for unit area parallel and perpendicular, respectively, to the molecular chain axis, and b0 is the distance between two adjacent fold planes. In the present case, when the growth front is assumed to be the (110) plane in the PLLA lattice as in the case of high-density polyethylene, the layer thickness b0 should be taken as 0.53 nm.34 Y in eq 8 is a coefficient that depends on the regime of crystal growth: Y ) 4 for regimes I and III, and Y ) 2 for regime II.35,36 For the lateral surface energy σ of linear polymer crystals, the following relation holds.45,46 σ ) 0.23b0∆Hu
(9)
where we adopted 0.23 as the front factor, considering that the present sample is a high melting point polyester PLLA.47,48 Substituting eqs 6-9 into eq 5 and rearranging to facilitate the analysis of experimental data, we obtain the following expression:38,49 R ) log G0 - C(Teq m /(Tic∆T))
(10)
Figure 10. Plots of the quantity R versus Teq m /(Tic∆T) (see eq 10) for CDA-g-PLLAs and PLLA-L.
values of >2.2-24, >21-157, and >10-103 nm for PLLAL, g-CDA77, and g-CDA58, respectively, with an exception of a larger one 105-106 nm for g-CDA22. Judging from the Z test and taking into consideration that the samples were crystallized, as a whole, at relatively large undercoolings, regime I should be ruled out. Unfortunately, we have no way to quantitatively distinguish regime II and III crystallizations; however, there is a key prediction reflecting the nucleation regime aspects empirically:46 Kg(regime III) ≈ 2Kg(regime II)
where R ) log G - log V2 + 17240/[2.30R(51.6 + Tic - Tg)] (0.46Teq m log V2)/∆T (11) 2 C ) Kg/(2.30Teq m ) ) (Yb0 σe)/(10.0k)
(12)
Plotting R against Teq m /(Tic∆T) should make a linear relation, and hence, the nucleation constant Kg involving σe can be evaluated from the slope. Figure 10 shows the construction of such a plot for each of the four samples that participated in PLLA crystallization. As suggested by the deviation from a single straight line, however, two different growth regimes were observed for any sample in the range of crystallization temperatures explored here. The regimes should be assigned properly before the estimation of σe. Kg data listed in Table 2 were obtained from the line drawing in the side of larger Teq m /(Tic∆T). The Z test of Lauritzen35 is often available to judge whether the growth occurs in regime I or II. The Z is defined as Z ≈ 103(L/(2b0))2 exp(-X/(Tic∆T))
(13)
where L is the crystalline substrate length corresponding to an effective lamellar width. Regime I kinetics is obeyed if substitution of X ) Kg in the above equation leads to Z < 0.01; if eq 13 with X ) 2Kg results in Z > 1, then regime II growth is indicated. The regime is actually determined by deciding whether the range of L values calculated in each case is reasonable. Testing the Kg data in Table 2 for conformity to regime I led to the limits of L to be
