Article pubs.acs.org/Macromolecules
Isothermal Crystallization Kinetics of Poly(ε-caprolactone) Blocks Confined in Cylindrical Microdomain Structures as a Function of Confinement Size and Molecular Weight Ryota Kato,† Shintaro Nakagawa,† Hironori Marubayashi,†,‡ and Shuichi Nojima*,†,‡ †
Department of Organic and Polymeric Materials and ‡Department of Chemical Science and Engineering, Tokyo Institute of Technology, H-125, 2-12-1 Ookayama Meguro-Ku, Tokyo 152-8552, Japan S Supporting Information *
ABSTRACT: The isothermal crystallization kinetics of poly(ε-caprolactone) (PCL) blocks confined in cylindrical microdomain structures (nanocylinders) formed by the microphase separation of PCL-block-polystyrene (PCL-b-PS) copolymers were examined as a function of nanocylinder diameter D and molecular weight of PCL blocks Mn. Small amounts of polystyrene oligomers (PSO) were gradually added to PCL blocks in PCL-b-PS to achieve small and continuous decreases in D. The time evolution of PCL crystallinity during isothermal crystallization at −42 °C showed a first-order kinetic process with no induction time for all the samples investigated, indicating that homogeneous nucleation controlled the crystallization process of confined PCL blocks. The half-time of crystallization t1/2 (inversely proportional to the crystallization rate) of PCL blocks with Mn ∼ 14 000 g/mol showed a 140-fold increase (from 0.48 to 69 min) by a 16% decrease in D (from 18.6 to 15.6 nm). Another set of PCL-b-PS/PSO blends involving slightly longer PCL blocks with Mn ∼ 15 800 g/mol showed a similar result. It was found by combining the results of two PCL-b-PS/PSO blends that the small increase in Mn (from 14 000 to 15 800 g/mol) yielded an approximately 90-fold increase in t1/2 (from 0.76 to 67 min) for PCL blocks confined in the nanocylinder with D = 18.2 nm. It is possible from these experimental results to understand the individual contributions of D and Mn to the crystallization rate of block chains confined in nanocylinders.
1. INTRODUCTION
The spatial confinement of polymer chains using AAOs has several advantages over that using cylindrical microdomain structures in that it is possible to widely change D (e.g., 10 nm−1 μm) and any crystalline homopolymers with arbitrary molecular weights (M) can be infiltrated. In the case of cylindrical microdomain structures, on the other hand, D is inherently determined by the molecular characteristics of block copolymers (i.e., the total molecular weight and composition of block copolymers), so that we cannot change D and M independently. However, the microdomain structures have
The crystallization of polymer chains spatially confined in various nanodomains has been attracting much attention over the past years because they show unique crystallization behavior and crystal orientation.1−14 Several experimental methods are employed to confine the polymer chains: micelles or droplets in solution,2,10,12,14−18 anodic aluminum oxides (AAOs),4,8,10,11,14,19,20 nanolayered films,14,21−23 or microdomain structures formed in block copolymers.1−3,5−9,13,14,24 New experimental facts have been disclosed using these methods for the crystallization of confined polymer chains. For example, it is found using AAOs or cylindrical microdomain structures that the crystallization behavior of confined polymer chains is significantly influenced by the cylinder diameter D.4,7,8,10,13 © 2016 American Chemical Society
Received: April 27, 2016 Revised: June 23, 2016 Published: August 1, 2016 5955
DOI: 10.1021/acs.macromol.6b00877 Macromolecules 2016, 49, 5955−5962
Article
Macromolecules
longer crystalline blocks confined in nanocylinders using the same blending method (Figures 1c and 1d) to compare the crystallization rate of block chains with different Mn both confined in the nanocylinder with the same D (Figures 1a and 1d). From these results, we finally obtain individual contributions of D and Mn to the crystallization rate of block chains confined in nanocylinders.
some advantages over AAOs, which we have used in this study, as described below. Several experimental studies are reported on the crystallization behavior and crystal orientation of confined blocks as a function of nanodomain size.25−27 Chung et al.,27 for example, investigated the crystallization kinetics of poly(ε-caprolactone) (PCL) blocks confined in cylindrical microdomain structures (nanocylinders) formed in PCL-block-poly(4-vinylpyridine) (PCL-b-P4VP) copolymers with various D by changing the total molecular weight of PCL-b-P4VP. They obtained that the crystallizable temperature range of confined PCL blocks decreased drastically when the molecular weight of PCL blocks changed from 5400 g/mol (D = 17.8 nm) to 2700 g/mol (9.9 nm), and hence it was impossible to quantitatively compare the crystallization rate at the same crystallization temperature. In these studies, changing D requires a change in the total molecular weight of block copolymers (and eventually a change in the molecular weight of crystalline blocks Mn) as long as neat diblock copolymers are used. Therefore, the crystallization (e.g., crystallizable temperatures or crystallization rate) of confined blocks might be affected by the combined effects of D and Mn. It is an alternative method for changing D to blend a small amount of miscible (and amorphous) oligomers into block copolymers (blending method).28−31 However, the attention has mainly been paid to the alteration of crystallization mechanisms by changing the nanodomain shape (i.e., lamellae, cylinders, or spheres), which can be achieved by blending a large amount of oligomers into block copolymers. Therefore, no comprehensive study has been reported so far on the quantitative relationship between D and the crystallization rate of confined chains with a constant Mn. The blending method makes it possible to slightly and continuously change D, which is one of the advantages of microdomain structures when we study the confined crystallization. In this study, we investigate the crystallization behavior of cylindrically confined crystalline blocks with a constant Mn as a function of D, which can be achieved by gradual blending of amorphous oligomers into block copolymers (Figures 1a and 1b). Next, we examine the crystallization behavior of slightly
2. EXPERIMENTAL SECTION 2.1. Samples and Sample Preparation. The crystalline− amorphous diblock copolymers used in this study are poly(εcaprolactone)-block-polystyrene (PCL-b-PS). The synthesis method of this copolymer is already described.32,33 The composition of PCL-bPS was adjusted so as to form the cylindrical microdomain structure with the matrix of PS blocks. The melting temperature of PCL blocks is lower than 54 °C (Figure 5a), and the glass transition temperature of PS blocks is ca. 100 °C, so that the crystallization of PCL blocks is expected to proceed within glassy (i.e., hard) nanocylinders formed by the microphase separation of molten block copolymers. We gradually changed the nanocylinder diameter D by adding small amounts of commercially available polystyrene oligomers (PSO). That is, blending PSO homogeneously into PS blocks makes the distance between neighboring block junctions wider to bring about a substantial change in the conformation of PCL blocks, yielding a slight decrease in D.34,35 The results of molecular characterization of PCL-b-PS and PSO are shown in Table 1, where CS1, CS2, and CS3 have PCL blocks with Mn
Table 1. Characterization of the Samples Used in This Study Mn (g/mol) sample code
PCL blocka
PS blockb
totalc
Mw/Mnb
f PCLd
De (nm)
CS1 CS2 CS3 CS4 PSOf
13 900 14 100 14 200 15 800
26 200 33 000 46 200 39 900 1 800
40 100 47 100 60 400 55 700 1 800
1.03 1.03 1.04 1.04 1.09
0.345 0.297 0.233 0.282
19.8 19.8 18.6 20.6
a
Determined by 1H NMR. bDetermined by GPC calibrated using PS standards. cCalculated from Mns of PCL and PS blocks. dVolume fraction of PCL blocks at 100 °C calculated from specific volumes and Mns of PCL and PS blocks. eNanocylinder diameter calculated from f PCL and the long period measured using SAXS. fObtained from Scientific Polymer Products Inc., USA.
∼ 14 000 g/mol but different block ratios, whereas CS4 has PCL blocks with Mn ∼ 15 800 g/mol, slightly larger molecular weight than that of the others. However, this difference is physically meaningful because the experimental error in Mn is expected to be less than 200 g/ mol. The volume fraction of PCL blocks in PCL-b-PS (f PCL) is in the range between 0.23 and 0.35, from which we expect the neat PCL-bPS copolymers form cylindrical microdomain structures in the melt. Figure S-1 shows a double-logarithmic plot between the total degree of polymerization N and corrected nanocylinder diameter D′ for our copolymers (closed circles), together with several PCL-b-PS copolymers used in our previous studies (open squares).36−38 The data points make one composite line with a slope of 0.657, confirming the molecular characteristics (Mn and f PCL in Table 1) and resulting nanocylinder diameter D are consistent with the scaling law at the strongly segregated limit of diblock copolymers (D′ ∼ N2/3).39 The binary PCL-b-PS/PSO blends were prepared using a solutioncasting method with toluene as a common solvent. PSO is expected to completely mix with PS blocks (i.e., matrices of microdomain structures, Figure 1) to form wet brush structures, providing slight and continuous decreases in D with a constant Mn of confined PCL blocks. The final trace of toluene was completely removed under vacuum at ca. 120 °C for 3 h.
Figure 1. Schematic illustration showing the strategy to prepare crystalline blocks (orange chains) confined in nanocylinders with different diameters D. Small amounts of amorphous oligomers (dark blue chains) are blended into amorphous blocks (light blue chains) to obtain slightly smaller nanocylinders with the same crystalline block (a, b). This blending method is applied to another block copolymer (c, d) to get confined crystalline blocks with a higher molecular weight. Consequently, it is possible to prepare two crystalline blocks with different molecular weights both confined in the nanocylinder with the same D (a, d). 5956
DOI: 10.1021/acs.macromol.6b00877 Macromolecules 2016, 49, 5955−5962
Article
Macromolecules 2.2. Small-Angle X-ray Scattering (SAXS) Measurements. The microdomain structures formed in the blend were examined using SAXS with synchrotron radiation (SR-SAXS). The SR-SAXS experiments were performed at Photon Factory in High Energy Accelerator Research Organization, Tsukuba, Japan, with a small-angle X-ray equipment for solution installed at the BL-10C station. Details of the equipment and instrumentation were described elsewhere.40−42 The scattered intensity was detected using a two-dimensional detector PILATUS3 2M (Dectris Ltd., Switzerland), which had 1475 × 1679 pixels with 172 × 172 μm2 in each size. The accumulation time for each measurement was 60 s. The 2D-SAXS curves obtained by SR-SAXS experiments were azimuthally averaged to obtain 1D-SAXS curves as a function of wavenumber s defined as s = (2/λ)[sin θ], where 2θ is the scattering angle and λ the X-ray wavelength used (= 0.1488 nm). Silver behenate43 was used for the calibration of the SR-SAXS equipment. Subsequently, the 1D-SAXS intensity was corrected for background scattering and absorption by the samples. The nanocylinder diameter D was finally derived from the long period L evaluated from the primary peak position of 1D-SAXS curves after Lorentz correction D=
8f ′PCL L 3π
(1)
where f ′PCL is the volume fraction of PCL blocks existing in the system. 2.3. Differential Scanning Calorimetry (DSC) Measurements. A DSC Diamond (PerkinElmer Ltd., USA) was used to evaluate the melting temperature Tm,PCL and crystallinity χPCL of PCL blocks. The sample was isothermally crystallized at −42 °C for 1 day, and subsequently it was heated at a rate of 10 °C/min to obtain Tm,PCL and χPCL. The value of Tm,PCL was evaluated from the temperature of endothermic melting peaks of PCL blocks, and χPCL was calculated from the area under the peak ΔHPCL assuming that the heat of fusion for perfect PCL crystals ΔH°PCL was 135 J/g44 χPCL =
ΔHPCL ° wPCL ΔHPCL
Figure 2. 1D-SAXS curves obtained from various CS4/PSO blends, where the SAXS curve from neat CS4 is represented by a blue curve. The weight fraction of PSO in the system wPSO is indicated on each SAXS curve.
to that of crystallized samples, indicating that the microdomain structure existing in the melt is completely preserved after the crystallization of PCL blocks (i.e., confined crystallization) due to the vitrification of PS chains. The 1D-SAXS curves obtained from other crystallized blends are summarized in Figure S-2. Though every 1D-SAXS curve has several scattering peaks at the positions of 1:√3:2:(√7) in Figure S-2, the samples with higher f ′PCL (>0.29) had two crystallizable temperature rangesone at ∼10 °C and the other at ∼−50 °Cprobably because the cylindrical microdomain structure coexisted with the lamellar microdomain structure, the latter of which provided higher crystallizable temperatures due to the difference in the crystallization mechanism of confined PCL blocks. We did not use such blends for the following crystallization experiments. The nanocylinder diameter D was evaluated from the long period of microdomain structures L and f ′PCL using eq 1, and D is plotted against wPSO in Figure 3. The values of D decrease almost linearly with increasing wPSO, and make individual lines according to D of neat copolymers (Table 1). That is D changes between 15.6 and 18.6 nm for the blends including CS1, CS2, and CS3 with Mn, ∼ 14 000 g/mol (open symbols in Figure 3), and between 18.2 to 20.6 nm for the blends including CS4 with Mn ∼ 15 800 g/mol (closed circles). Therefore, it is possible to examine the crystallization kinetics of PCL blocks with Mn ∼ 14 000 or ∼15 800 g/mol as a function of D, where D changes by 16% (open symbols) or by 12% (closed circles), respectively. Furthermore, it is possible to compare the crystallization rate of PCL blocks with different Mn (14 000 or 15 800 g/mol) confined in the nanocylinder with the same D, for example, the nanocylinder with D = 18.2 nm (blue line in Figure 3). Therefore, we can examine the effects of D and Mn separately on the crystallization kinetics of PCL blocks confined in nanocylinders. It should be noted that
(2)
where wPCL is the weight fraction of PCL blocks existing in the system. The time evolution of PCL crystallinity χPCL(t) (t: crystallization time) during isothermal crystallization was also obtained from DSC measurements. The sample was first annealed at 60 °C (>Tm,PCL) to erase the thermal history, quenched into the crystallization temperature Tc (= −42 °C) at a rate of −500 °C/min, crystallized there for a prescribed time t, and then heated at a rate of 10 °C/min. The value of χPCL(t) was evaluated from the endothermic peak area using eq 2 as a function of t. The crystallizable temperature range of confined PCL blocks was moderately narrow, so that additional crystallization during quenching or heating would be negligibly small.
3. RESULTS AND DISCUSSION 3.1. Microdomain Structures. The microdomain structure formed in the system was examined for the samples after full crystallization of PCL blocks at −45 °C for a long time (>12 h) because the difference in electron density at amorphous states between PCL blocks (346 e/nm3 at 60 °C) and PS blocks (332 e/nm3 below 100 °C)36 is extremely small for detecting the higher-order SAXS peaks. Figure 2 shows typical 1D-SAXS curves of neat CS4 (blue curve) and CS4/PSO blends with various PSO weight fractions wPSO indicated (black curves). These 1D-SAXS curves have several scattering peaks, the positions of which exactly correspond to a ratio of 1:√3:2. This fact, together with the volume fraction of PCL blocks f ′PCL in the system, suggests that the cylindrical microdomain structures (nanocylinders) with the PCL block inside are formed in the system. It should be noted that the primary peak position in 1D-SAXS curves from molten samples was essentially identical 5957
DOI: 10.1021/acs.macromol.6b00877 Macromolecules 2016, 49, 5955−5962
Article
Macromolecules
Figure 4. DSC curves during heating at a rate of 10 °C/min for CS4/ PSO blends, where the DSC curve of neat CS4 is represented by a blue curve. The weight fraction of PSO in the system wPSO is indicated on each DSC curve.
Figure 3. Nanocylinder diameter D plotted against the weight fraction of PSO in the system wPSO for the blends including CS1 (open circles), CS2 (triangles), CS3 (squares), and CS4 (closed circles). The molecular weight of PCL blocks for open symbols is ca. 14 000 g/mol, and that for closed circles is 15 800 g/mol. The blue line represents D = 18.2 nm.
though the change in D or Mn seems to be considerably small, it is not difficult to suppose that such the small change in D yields a large change in the crystallization rate of confined blocks, as expected from previous studies on the crystallization of various polymer chains confined in nanocylinders with different D.19,27 3.2. Melting Temperature and Crystallinity of PCL Blocks. Prior to DSC heating experiments, the samples were cooled from the amorphous state at 60 °C with a rate of −10 °C/min to find the crystallizable temperature range of PCL blocks. The exothermic crystallization peak was observed between −46 and −57 °C depending on wPSO, indicating that the crystallization of confined PCL blocks occurs at extremely low temperatures as compared with that of bulk PCL homopolymers (∼38 °C45). This low crystallizable temperature is usually observed for crystalline blocks spatially confined in isolated nanodomains (i.e., nanospheres or nanocylinders) and is ascribed to the difference in the crystallization mechanism between confined blocks and bulk homopolymers (homogeneous nucleation or heterogeneous nucleation and growth, respectively). We decided from the results of DSC cooling experiments that the crystallization temperature of PCL blocks Tc employed in this experiment was −42 °C for all the samples by considering an experimentally accessible time range to examine the isothermal crystallization. The melting behavior of PCL blocks isothermally crystallized at −42 °C for 1 day was investigated using DSC with a heating rate of 10 °C/min, and typical DSC curves are shown in Figure 4 for CS4/PSO blends with various wPSO. The endothermic peaks for the melting of PCL blocks are rather diffuse as compared with those observed for bulk PCL homopolymers,45 and their peak position, slightly moving to the lower temperatures with increasing wPSO (dotted line in Figure 4), is extremely lower than that of bulk PCL homopolymers (∼62 °C45), suggesting that PCL blocks form immature lamellar crystals due to the space confinement imposed by nanocylinders. The melting temperature Tm,PCL and crystallinity χPCL of PCL blocks in all the blends crystallized at −42 °C are summarized in Figure 5 as a function of nanocylinder diameter D. The
Figure 5. Melting temperature (a) and crystallinity (b) of PCL blocks plotted against nanocylinder diameter D. For the symbols, see Figure 3. The molecular weight of PCL blocks for open symbols is ca. 14 000 g/mol, and that for closed circles is 15 800 g/mol.
values of Tm,PCL make one composite line and increase steadily with increasing D (Figure 5a). This fact indicates that the lamella thickness of PCL crystals is predominantly controlled by the space provided by nanocylinders. That is, the existing nanocylinder significantly disturbs the formation of usual lamellar crystals, the degree of which is larger as D is smaller. Consequently, thinner PCL lamellar crystals are formed with decreasing D. On the other hand, χPCL is almost constant (∼0.4) irrespective of D (Figure 5b) and significantly smaller than that of bulk PCL homopolymers (∼0.6). These results suggest that smaller and thinner PCL lamellar crystals are 5958
DOI: 10.1021/acs.macromol.6b00877 Macromolecules 2016, 49, 5955−5962
Article
Macromolecules formed under the spatial confinement, but the number of these immature crystals increases with decreasing D to keep the total PCL crystallinity almost constant. This conclusion is consistent with the general picture of polymer crystals in nanocylinders proposed on the basis of many results of crystallization in AAOs or cylindrical microdomain structures.19,27,31 3.3. Crystallization Behavior of PCL Blocks. The time evolution of PCL crystallinity χPCL is shown in Figure 6 as a
Figure 7. Plot of ln(1 − χ′PCL(t)) (χ′PCL(t): normalized crystallinity of PCL blocks at t) against crystallization time t for CS4/PSO blends with wPSO = 0 (neat CS4, blue circles), 0.05 (open circles), 0.14 (inverse triangles), and 0.21 (squares). The crystallization temperature is −42 °C for all samples.
found from Figure 7 that the crystallization rate decreases drastically with increasing wPSO or decreasing D. However, the data at the late stage of crystallization deviate moderately from each line. Actually, the conventional Avrami plot yields n < 1 for all the blends, which is sometimes observed for the crystallization of polymer chains confined in isolated nanodomains.5,10,20 The accelerated nucleation at the nanocylinder interface might be one of the reasons for n < 1.20 The half-time of crystallization t1/2 derived from k (or the slope of each line in Figure 7) was not substantially different from that directly evaluated from Figure 6 (as a time to give (1/2)(χPCL(∞)), so that we used eq 3 to evaluate t1/2. 3.4. Analysis of Crystallization Kinetics as a Function of D and Mn. The half-time of crystallization t1/2 is a measure for the crystallization rate of polymer chains, and it is larger as t1/2 is smaller. Figure 8a shows the logarithmic plot of t1/2 against D for all the blends investigated, where we find the data points make two separate lines; one consists of three blends including PCL blocks with Mn ∼ 14 000 g/mol (i.e., CS1/PSO, CS2/PSO, and CS3/PSO blends, open symbols in Figure 8a) and the other CS4/PSO blends including PCL blocks with Mn ∼ 15 800 g/mol (closed circles). The former composite line indicates that t1/2 depends only on D for PCL blocks with a constant Mn irrespective of wPSO and molecular characteristics of PCL-b-PS (i.e., total molecular weight and f PCL). That is, the crystallization rate at Tc = −42 °C is substantially driven only by two parameters: molecular weight of confined blocks and nanocylinder diameter. Furthermore, it is easily found from both lines in Figure 8a that the small decrease in D yields an extremely large increase in t1/2 (or large decrease in the crystallization rate) during isothermal crystallization at the same Tc (= −42 °C). That is, t1/2 exhibits approximately a 140-fold increase (from 0.48 to 69 min) by a 16% decrease in D (from 18.6 to 15.6 nm) for the PCL blocks with Mn ∼ 14 000 g/mol and a 260-fold increase (from 0.35 to 94 min) by a 12% decrease in D (from 20.6 to 18.2 nm) for the PCL blocks with Mn ∼ 15 800 g/mol. It is surprising and interesting that the small decrease in D yields such a drastic increase in t1/2, and this fact is the new finding of the present study. This extraordinarily large increase in t1/2 is, however, qualitatively anticipated from several experimental results reported for the nonisothermal crystallization of block chains confined in isolated nanodomains. For example, Xu et al. investigated the crystallization kinetics of poly(oxyethylene)
Figure 6. Time evolution of PCL crystallinity χPCL(t) plotted against crystallization time t for various CS4/PSO blends with wPSO = 0 (neat CS4, blue circles), 0.05 (open circles), 0.10 (triangles), 0.14 (inversetriangles), 0.18 (diamonds), and 0.21 (squares). The crystallization temperature Tc is −42 °C for all samples.
function of crystallization time t for various CS4/PSO blends isothermally crystallized at Tc = −42 °C, and that for CS1/ PSO, CS2/PSO, and CS3/PSO blends is summarized in Figure S-3. The crystallinity of PCL blocks increases abruptly at the initial stage of crystallization with no appreciable induction time, followed by an asymptotic increase at the late stage of crystallization. This crystallization behavior is substantially different from the sigmoidal time evolution of bulk crystalline homopolymers and usually observed for polymer chains spatially confined in isolated nanodomains,3,10,13 which is ascribed to the nucleation-controlled crystallization mechanism of confined chains. That is, once homogeneous nucleation occurs in the isolated nanodomain, the crystal growth finishes instantaneously because of the small size of isolated nanodomains. Furthermore, it is found from Figure 6 and Figure S-3 that the crystallization rate decreases considerably with increasing wPSO or decreasing D, from which it is intuitively anticipated that the molecular motion of PCL blocks intimately affects the confined crystallization in nanocylinders. The crystallization kinetics of PCL blocks shown in Figure 6 and Figure S-3 can be successfully analyzed by assuming that the crystallization rate of PCL blocks at crystallization time t is proportional to the volume fraction of uncrystallized PCL blocks existing at t. That is
ln{1 − χ ′PCL (t )} = kt
(3)
where χ′PCL(t) (= χPCL(t)/χPCL(∞)) is the normalized crystallinity of PCL blocks at t, and k (
