Andreas
Meyer
a,
Steffen M.
Weidner
b and
Hans R.
Kricheldorf
*c
aInstitute of Physical Chemistry, University of Hamburg, Martin-Luther-King-Platz 6, 20146 Hamburg, Germany
bFederal Institute for Materials Research and Testing – BAM, Richard Willstätter Str. 11, D-12489 Berlin, Germany
cInstitute of Technical und Macromolecular Chemistry, Bundesstr. 45, D-20146 Hamburg, Germany. E-mail: hrkricheldorf@aol.de
First published on 28th October 2019
Two kinds of cyclic poly(D- and L-lactide)s were synthesized, namely CI labeled samples mainly consisting of even-numbered cycles with low dispersity and CII, CIII or CIV-labeled ones consisting of equal amounts of even and odd-numbered cycles with high dispersity and higher molecular weights (Mw up to 300000). Furthermore, linear poly(L-lactide)s were prepared by initiation with ethanol and in both series the molecular weight was varied. The formation of stereocomplexes from cyclic poly(D-lactide)s and all kinds of poly(L-lactide)s was performed in dichloromethane/toluene mixtures. The stereocomplexes crystallized from the reaction mixture were characterized in the virgin state and after annealing at 205 °C. Stereocomplexes free of stereohomopolymers with crystallinities up to 80% were obtained from all experiments in yields ranging from 60 to 80%. Despite the high annealing temperature (maintained for 1 h), little transesterification was observed and the crystallinity slightly increased.
Since 1993 the work of Ikada and co-workers has been expanded by other research groups and properties of such StCs were explored in various directions, summarized and discussed by various authors in several review articles.16–20 In addition to StCs based on two separate enantiomeric chains,1–9,15,21–24 various types of block-copolymers were prepared and characterized. These block copolymers may be subdivided into two categories. The first category are the stereo-blockcopolymers which contain blocks of D- and L-units in the same chain. D/L-ratio and block lengths were varied over a broad range. Polylactides having a random or nearly random sequence of D- and L-units cannot crystallize and do not form stereocomplexes. However, above a certain block length StC formation can occur and both, Tm and melting enthalpy (ΔHm) increase with length and perfection of the D- or L-blocks.25–38 The second category consists of two- or three-block copolymers containing blocks of different chemical structure, such as blocks of poly(e-caprolactone), poly(ethylene glycol) or poly(dimethyl siloxane).
More recently, two research groups reported on attempts to prepare and characterize StCs of cyclic polylactides. Tezuka and coworkers synthesized the cycles by a ring-closing reaction of linear, two-block copolylactides consisting of one L-block and one D-block in combination with functional end groups.39,40 Yet, in this way guest–host copolymers and not cyclic homopolymers were obtained. Depending on whether the two blocks were arranged in a parallel or antiparallel fashion Tm values of 221 °C or 203 °C were found.
Waymouth and coworkers prepared cyclic and linear poly(L-lactide)s and poly(D-lactide)s by means of a heterocyclic carbine catalyst, which due to its basicity caused partial racemization.41 The optical rotation data suggest that the cycles contained approximately 40% of D-units and the linear chains approximately 25%. For the combination of enantiomeric cycles a Tm of 179 °C and ΔHm 15 J g−1 was determined, whereas a combination of cyclic with linear enantiomers resulted in Tm values of 178 or 185 °C with ΔHm of 20 or 41 J g−1.
In this context it was the purpose of the present work to elucidate to what extent optically pure cyclic polylactides prepared by various tin catalysts (Scheme 1) can form stereocomplexes with each other or with linear polylactides of similar molecular weight.42–47 Four different cyclic polylactides were used for this study that varied both by their weight average molecular weight (Mw) in the range of 10000 to 300
000 g mol−1, and by their dispersities.
CI-L was prepared analogously, but monomer and catalyst were dissolved in 42 mL of chlorobenzene (1 M solution).
MALDI TOF mass spectra were recorded on an Autoflex III Smartbeam (Bruker Daltonik, Bremen, Germany). Spectra were recorded in linear mode at 200 Hz. The matrix solution (DCTB – trans-2-[3-(4-tert-butylphenyl)-2-methyl-2-propenylidene] malononitrile, 10 mg mL−1) was doped with 5 μL of potassium trifluoroacetate (KTFA 2 mg ml−1) to selectively enable the formation of potassium adduct ions. 20 μL of dissolved samples (5 mg mL−1, chloroform) were mixed with 50 μL of the matrix solution. Finally, 1 μL of this solution was deposited on a stainless-steel target. The manufactures software (FlexControl and FlexAnalysis) was applied for recording and evaluation. Typically, 2000 single spectra recorded on different positions were accumulated for one spectrum.
SEC measurements were performed using a modular system consisting of an isocratic pump (Chromophor C 1130) running at 1 mL min−1 flow rate. The system was kept at 40 °C. Samples with a concentration of 2–4 mg mL−1 were manually injected by a Rheodyne valve. The sample volume was 100 μL. A triple detector combination was used consisting of a refractive index detector RI-501 (Shodex), a multi angle laser light scattering detector DAWN and a viscometer Viscostar (both Wyatt Germany, Dernbach). Clarity software (GPC extension, DataApex) was used for instrument control and molecular weight calculation, whereas ASTRA 6 (Wyatt) was used to perform the MHS calculations. The SEC instrument calibration was performed using commercially available polystyrene standard sets (Polymer Standards Service – PSS, Mainz, Germany). Due to peaks caused by solvent and low mass impurities, the calculation of the number average (Mn) and weight average (Mw) molecular weights and the dispersities Đ started from 1100 g mol−1 onwards. Therefore, probably not all oligomers were included in the calculation. This difference has only little influence on Mw, whereas Mn values may be too low, and thus, dispersities might be correspondingly underestimated.
Label | Even/odd | M n | M w | Đ | |
---|---|---|---|---|---|
a Measured after precipitation into ligroin. | |||||
CI-D | 9/1 |
|
12![]() |
1.1 | |
CIII-D | 1/1 | 26![]() |
82![]() |
3.1 | |
CI-L | 9/1 | 1100 | 12![]() |
1.1 | |
CII-L | 1/1 | 10![]() |
27![]() |
2.6 | |
CIII-L | 1/1 | 19![]() |
63![]() |
3.1 | |
CIV-L | 1/1 | 123![]() |
310![]() |
2.6 |
CI cycles were prepared from D- and L-lactide under identical conditions using SnNaph2 (see Scheme 1) as catalyst.41 These CI cycles are characterized by a weight average molecular weight around 12000 g mol−1 and a low dispersity. Furthermore, their MALDI-TOF mass spectra display a large predominance of even-numbered cycles (Fig. 1A) as already reported previously.41 This predominance has the advantage that any kind of transesterification yielding odd-numbered cycles can easily be detected by mass spectroscopy. In the case of the CII–CIV cycles even- and odd-numbered cycles are fully equilibrated as exemplarily illustrated by the mass spectrum of Fig. 1B (the mass spectra of the higher cyclic polylactides are given in the ESI Fig. S1†). Furthermore, the dispersities were much higher. CII and CIII type cycles were prepared by means of the BuSnOPF catalyst (Scheme 1) in chlorobenzene solution either with a monomer concentration 0.1 mol L−1 (CII) or with a concentration of 4 mol L−1 (CIII) according to a procedure published previously.44 The CIV-L sample was prepared by means of SnBiph in bulk at 170 °C (2 h) as described in a recent publication.47 All cyclic samples were precipitated from dichloromethane solutions into ligroin to get rid of most of the catalyst. In the case of CIV-L a Mark–Houwink–Sakurada plot was measured with triple detection in comparison with a linear commercial poly(L-lactide) to make sure that this sample still consists of cycles in the high molar mass range (>100
000 g mol−1) after dissolution and precipitation.
Two species of linear polylactides were synthesized using SnOct2 as catalyst and ethanol as cocatalyst which played in fact the role of an initiator (Table 2). The OH end groups were acetylated with acetic anhydride/pyridine to avoid transesterification with the cycles during annealing at high temperatures. As expected from previous publications, low dispersities and a predominance of even-numbered chains were obtained as demonstrated in Fig. 2.48,49
Since the hydrodynamic volume of linear (but not of cyclic) polylactides is higher than that of polystyrene, polystyrene calibrated SEC measurements overestimate the real molecular weights by approximately 50%. When the measured number average molecular weights (Mn's) were multiplied with the correction factor 0.68, they were quite similar to those of the CI cycles. The three linear polylactides were also precipitated into ligroin prior to use.
The approach used in the present work resembles the “acetonitrile method”. Solutions of the enantiomeric polylactides in dichloromethane were mixed, slowly evaporated by heating and the dichloromethane was substituted by portion-wise addition of toluene. In this way the solubility of polylactides in the reaction mixture steadily decreased over a period of approx. 2.5 h. The StCs precipitated slowly from the reaction mixture in the form of crystalline powders, when the molecular weights of the starting materials were low (CI, CI, LI, LII), or in the form of gel particles, when the molecular weights were high (CIII and CIV). Washing of the virgin reaction products with warm dichloromethane removed unreacted enantiomers, if present at all. As illustrated by the data listed in Tables 3 and 4, all combinations gave quite similar yields what indicates that this procedure favoured formation of StCs from any kind of starting materials. The yields together with the DSC measurements also demonstrate that the high molecular weight of CIV-L is no hindrance for an efficient formation of StCs.
Exp. no | Comp. | Yield (%) | T m1 (°C) | T m2 (°C) | ΔH (J g−1) | Cryst (%) DSCa | Cryst (%) WAXS |
---|---|---|---|---|---|---|---|
a Based on a ΔHm max of −142 J g−1 (ref. 8). b Based on 4 measurements with repeated calibration. | |||||||
1 | CI-D + CI-L | 77 | 202.5 | 231.5 | 88.8 | 62 | 62 |
2 | CI-D + CII-L | 71 | 206.5 | 229–231 | 87.2 | 61 | 59 |
3 | CID + CIII-L | 78 | 209.0 | 228.0–228.5 | 78.5–80.0b | 55–56 | 58 |
4 | CI-D + CIV-L | 69 | 218.5 | 226.0–227.0 | 72.8–75.0b | 51–53 | — |
5 | CIII-D + CIII-L | 72 | Shoulder | 228–229 | 72.7 | 51 | 56 |
6 | CIIID + CIV-L | 50 | — | 228.0 | 60.7–62.5b | 43–45 | 46 |
Exp. no. | Comp. | Yield (%) | T m1 (°C) | T m (°C) | ΔHm (J g−1) | Cryst (%) DSCa | Cryst (%) WAXS |
---|---|---|---|---|---|---|---|
a Based on a ΔHm max of −142 J g−1 (s. ref. 8). b Based on four measurements with repeated calibration. | |||||||
1 | LI-D-Ac + LI-L-Ac | 76 | — | 233, 0–233.5 | 109.0 | 76–77b | 70 |
2 | CI-D + LI-L-Ac | 78 | 204.5 | 229–230 | 101.6 | 72 | 64 |
3 | CIII-D + LII-L-Ac | 72 | 212.0 | 230–233 | 72.5 | 50 | 59 |
However, with exceptions of “CID + CIVL” (s. Fig. 1B), all DSC curves also displayed a weak endotherm between 200 and 220 °C (see Fig. 3 and 4B and Tm1 listed in Tables 3 and 4) which was also observed by other authors for StCs of linear polylactides. Furthermore, a lower weak melting endotherm accompanying the main endotherm is also known from numerous other polymers and usually interpreted as result of a small fraction of smaller and less perfect crystallites. If this interpretation is correct, Tm1 should vanishes upon short annealing at that temperature. Tsuji et al. were the first who studied this phenomenon in the case of a linear polylactide StC and came to the same conclusion.8 The annealing experiments performed in this work are discussed below. Characteristic for StCs of the high molar mass polylactides CIII and CIV are relatively broad endotherms as illustrated by Fig. 4.
The melting enthalpies fell into the range from −70 to −90 J g−1 for StCs exclusively consisting of cycles but reached a value of −101.6 J g−1 when CI-D was combined with its linear counterpart (Table 4). The highest ΔHm was found for the StC prepared from two linear, low molar mass polylactides (Table 4). Therefore, the melting enthalpies indicate two trends. First, ΔHm increases with decreasing molecular weight of the components. Second, linear chains are more favourable for high ΔHm's than cycles. These trends are reasonable, when the StCs are prepared under identical conditions, because long chains and the cyclic topology need more time to achieve optimum conformations for StC formation and crystallization. This conclusion is supported by the annealing experiments discussed below.
The ΔHm's may, in principle, be used to calculate crystallinities, but for such calculations a comparison with a reference value for 100% crystallinity is necessary.
At this point it should be mentioned that despite numerous publications, crystallinities of StCs derived from linear polylactides have rarely been reported. Only one reference value has been found by the authors, namely −142 J g−1 used by Tsuji et al. for the calculation of crystallinity of a sample annealed at 216 °C for 10 min.8 Those authors cited a paper of Loomis et al. (ref. 21 in this work) as source of this value, which is an abstract of a talk presented at the ACS meeting in 1990. It does not mention melting enthalpies and how they were determined. In 2005 Sarasua et al.15 reported crystallinities for numerous StC samples (after variation of the annealing conditions) calculated from DSC measurements with the same ΔHm max. Surprisingly, also, those authors cited the “Loomis paper” as source.15,24 Hence, origin, and margin of error of this ΔHm max are uncertain. Usually, a ΔHm max is calculated by extrapolation of ΔHm values of samples the crystallinity of which is determined by WAXS measurements. These WAXS measurements have a margin of error around 5% and thus, the extrapolated ΔHm max is also afflicted with a similar margin of error.
Nonetheless, crystallinities calculated with a ΔHm max of −142 J g−1 were listed in Tables 3–5 for comparison with the data of Tsuji et al.8 and Sarasua et al.15 The highest value found by Tsuji was 70% crystallinity after annealing at 216 °C for 10 min and by Sarasua, who found a value of 61.5% after annealing at 190 °C for 24 h. Those values agree well with the highest crystallinities calculated analogously in this work (see Tables 3 and 5).
Comp. | Annealing at 160 °C/1 d | Annealing at 205 °C/1 h | |||||
---|---|---|---|---|---|---|---|
T m (°C) | ΔHm (J g−1) | Cryst (%) DSCa | Cryst (%) WAXS | T m (°C) | ΔHm (J g−1) | Cryst (%) DSCa | |
a Based on a ΔHm max of −142 J g−1 (s. ref. 8). | |||||||
CI-D + CI-L | 214.0 | 96.5 | 68 | 73 | 220.5 | 80.8 | 57 |
CI-D + CII-L | 212.0 | 107.8 | 76 | 81 | 217.2 | 86.6 | 61 |
CI-D + CIII-L | 212.0 | 97.5 | 69 | 69 | 214.5 | 83.6 | 59 |
CID + CIVL | 230.5 | 85.5 | 60 | — | 225.5 | 85.7 | 60 |
CI-D + LI-L-Ac | 216.5 | 104.0 | 74 | 69 | 216 | 83.5 | 59 |
CIII-D + CIII-L | 217.5 | 103.0 | 73 | 69 | 232.0 | 95.5 | 67 |
CIII-D + LII-L-Ac | 215.5 | 103.5 | 73 | 74 | 220.0 | 97.5 | 69 |
The WAXS powder patterns of most all samples displayed a rather flat baseline indicating crystallinities above 50% (Fig. S3†). Only the combination of two high molar mass cycles (CIII-D + CIV-L) gave a curved baseline yielding a crystallinity below 50% (Table 3). The trends that can be derived from these WAXS crystallinities are the same as those extracted from the DSC measurements. High crystallinities are favoured by lower molecular weights and by the linear topology. In the case of the three StCs prepared by combination of cyclic and linear polylactides the crystallinities derived from DSC measurements vary over a broader range than those calculated from WAXS patterns. The reproducibility of both, DSC and WAXS measurements has been checked and an explanation for this difference cannot forwarded at this time, but it does not affect the conclusions and trends presented above.
SAXS measurements show only diffuse scattering and a lamellar long period signal cannot be detected (see ESI Fig. S3†). The lack of SAXS reflections may be explained by diffuse boundaries between the crystal lamellas and the interlamellar amorphous regions, and a short distance between the crystallites.
The annealing at 160 °C had the expected consequence that the degree of crystallinity increased as demonstrated by the ΔHm values listed in Table 5 compared with those listed in Table 3. This increase is particularly pronounced, when the initial crystallinity was relatively low (according to DSC), as it is typical for StCs including large cycles such as CIII-L and CIV-L. The WAXS data confirm this trend, although the absolute values were approximately 10% higher. At this point it is of interest to compare the melting enthalpies of the cyclic stereocomplexes (−85.5 up to −107.8 J g−1, Table 5) with the highest values reported for annealed StCs of linear polylactides. Tsuji found −98 and −102 J g−1 for samples annealed at 160 °C for 1 h or 216 °C for 10 min).8 Sarasua obtained a ΔHm of −87.3 J g−1 for a sample annealed at 190 °C for 24 h.15 These data demonstrate, that after annealing cyclic StCs can reach the same level of crystallinity as the best samples of linear StCs, even when high molar mass cycles are involved. This result needs to be underlined, because Tsuji et al. concluded from mixing linear polylactides of different molecular weights, that a large extent of StC formation requires Mw's below 4 × 104 g mol−1, and around 4 × 105 g mol−1 StC formation should become impossible.3 Sarasua later concluded: “Stereocomplexation hardly develops when mixing high molecular stereoisomers, a result related to the smaller overall mobility of high molecular weight chains compared to that of lower molecular weight ones”.15
The DSC traces of the six samples selected for the annealing experiments at 205 °C (see Table 5, last two columns) gave the following results. First, no glass transition step was detectable indicating that no large fraction of amorphous material was formed. Second, in all cases, where a weak melting endotherm below 220 °C (i.e. Tm1) was detected in the virgin stereocomplexes, this melting endotherm had vanished. Both observations are proved by the DSC curve of Fig. 5. Third, the melting enthalpies were higher than those of the virgin StCs, but they were lower than those obtained at 160 °C and thus, suggest that these results were either influenced by beginning random transesterification and/or thermal degradation. Hence, both, the DSC curves and the WAXS powder patterns carried the same message, namely absence of significant transesterification reactions generating a randomization of the homosteric D- and L-chains.
To confirm this conclusion by a more precise method, both, 1H and 13C NMR spectra of two StCs (CI-D + CII-L and CIII-D + CIII-L) annealed at 160 or 205 °C were recorded and compared with those of the virgin StCs (see ESI Fig. S4†). No difference was found between annealed and virgin samples. From polylactides it is well known, that the CO and CH signals are sensitive to different stereosequences, so that signals of different D-L/L-D containing triads or tetrads are detectable.50,51 Such signals were not observable in the spectra of the annealed samples (see ESI Fig. S5†). In this spectrum seven of the eight possible D-L/L-D tetrads are clearly observable, whereas only the isotactic tetrad shows up in the spectra of the annealed stereocomplexes. These NMR spectra were recorded with a 600 MHz NMR spectrometer and the 13C NMR scans were accumulated over a period of 12 h, so that a S/N ratio >150/1 was achieved for the CO signal of the stereocomplexes. Therefore, it may reliably be concluded, that the percentage of D-L/L-D diads (if any) did certainly not exceed 5%, a value which might be explained by transesterification in the amorphous phase.
This information is important, because the MALDI TOF mass spectrum of the CI-D + CI-L StC annealed at 205 °C told a somewhat different story (Fig. 6). As demonstrated by comparison of Fig. 6 with Fig. 1A (representing the starting materials), the predominance of the even-numbered cycles was still maintained but the fraction of odd-numbered cycles had increased from approx. 5% to 30%. Considering, that the 13NMR spectra proved almost total absence of random transesterification, the mass spectrum of Fig. 6 indicates that a second unusual transesterification mechanism was operating enabling selective ring–ring equilibration. This conclusion is supported by the fact that the intensity distribution of the “odd-peaks” parallels that of the “even peaks” and no broadening of the peak pattern has occurred as it may be expected for random transesterification. A satisfactory formulation of this unusual ring–ring equilibration cannot be forwarded at this time, but a detailed study of this phenomenon and other transesterification reactions in the solid state of cyclic polylactides is in progress and will be published in the near future.
The mass spectra of CID + CIL annealed at 160 and 205 °C also displayed peaks of linear chains resulting from thermal degradation. Since the crystallinity increased upon annealing these degradation reactions must be limited to the disordered phase between the crystallites and to the surface of the crystallites. Hence, these preliminary results demonstrate that the combination of NMR spectroscopy and MALDI TOF mass spectrometry is a powerful tool to study transesterification and other chemical reactions in the solid state of cyclic (and linear) polylactides, provided that these polylactides contain a large excess of either even or odd-numbered species.
The degradation reactions indicated by the mass spectra were confirmed by dynamic thermogravimetric measurements with a heating rate of 5 K min−1. In the case of CI-D + CI-L the loss of weight begins around 200 °C and reaches 5% at 222 °C (s. ESI Fig. S6†). This unexpectedly low thermal stability most likely results from the presence of the catalyst SnNaph2 which was used in large amounts for the preparation of the parent stereohomocycles and which is difficult to remove completely. When high molar mass cycles are prepared with small quantities of catalysts, the thermostability is considerably higher.
For example, CI-D + CIV-L shows 5% loss of weight at 242 °C (see ESI Fig. S7†) and CIV-L alone at 262 °C (see ESI Fig. S8†). This is the highest thermostability the authors have found for a polylactide so far, regardless if cyclic or linear. A detailed study of thermostabilities of cyclic and linear polylactides and their stereocomplexes is in progress and will presented in a future publication.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9py01236b |
This journal is © The Royal Society of Chemistry 2019 |