Hans R.
Kricheldorf
*a,
Steffen. M.
Weidner
b and
Jana
Falkenhagen
b
aUniversität Hamburg, Institut für Technische und Makromolekulare Chemie, Bundesstrasse 45, 20146 Hamburg, Germany. E-mail: kricheld@chemie.uni-hamburg.de
bBundesanstalt für Materialforschung und -prüfung, Richard-Willstätter-Strasse 11, 12489 Berlin, Germany
First published on 29th July 2021
L-Lactide was polymerized with tin(II)acetate, tin(II)2-ethyl hexanoate, diphenyltin dichloride and dibutyltin bis(pentafluorophenoxide) at 130 °C in bulk. When an alcohol was added as initiator, linear chains free of cycles were formed having a degree of polymerization (DP) according to the lactide/initiator (LA/In) ratio. Analogous polymerizations in the absence of an initiator yielded high molar mass cyclic polylactides. Quite similar results were obtained when ε-caprolactone was polymerized with or without initiator. Several transesterification experiments were conducted at 130 °C, either with polylactide or poly(ε-caprolactone) indicating that several transesterification mechanisms are operating under conditions that do not include formation of cycles by back-biting. Furthermore, reversible polycondensations (revPOCs) with low or moderate conversions were found that did not involve any kind of cyclization. Therefore, these results demonstrate the existence of revPOCs, which do neither obey the theory of irreversible polycondensation as defined by Flory nor the hypothesis of revPOCs as defined by Jacobson and Stockmayer. A new concept encompassing any kind of revPOCs is formulated in the form of a “polycondensation triangle”.
Critique of part of the JS-theory was recently published by the authors5,6 who demonstrated that formation of cyclic oligomers and polymers does not only result from “back-biting” but also from end-to-end (ete) cyclization, which is in contrast to the assumption of J + S. However, previous results and the results presented below demonstrate that ete-cyclization is a very efficient process. Moreover, it is unproven that the cycles indirectly detected by J + S (mass spectrometry and NMR spectroscopy did not exist at that time) mainly or exclusively result from back-biting. A consequence of ete-cyclization is that the JS theory does not provide a correct description of all revPOCs. Furthermore, it is necessary to emphasize that J + S did never study a stoichiometric polycondensation up to high conversions (>99.5%). Nonetheless, the widely accepted view and understanding of revPOCs is still such that the reversibility and thermodynamic control is mainly or exclusively the result of ring chain equilibration via “back-biting”.
Recently the authors have detected and defined a special kind of polymerization which combines ring-opening polymerization (ROP) with simultaneous polycondensation (ROPPOC).7,8 Such ROPPOC polymerizations arise from ROPs initiated with a compound that introduces a reactive end group, so that linear chains having two reactive chains will be formed. ROPPOC polymerization may be equivalent to a normal revPOC of a non-cyclic monomer as illustrated by syntheses of nylons from ω-amino acids alone or in combination with the corresponding lactams (Scheme 2).9
Scheme 2 Synthesis of nylon-6 by polycondensation of aminocapronic acid or equilibration of caprolactam with aminocapronic acid. |
From a preparative and theoretical point of view, ROPPOCs have the advantage that they allow for simulation of polycondensations with conversions >99.9%. Consider a monomer/Cat ratio of 1000/1 as used in this work, the ROP of the monomer corresponds to a polycondensation with 99.9% conversion. Any inter or intramolecular condensation following or accompanying the ROP will then raise the conversion above 99.9% and conversions of 99.999% or higher may be achieved, which can hardly be realized by a normal polycondensation starting out from linear monomers. Such extreme conversions are interesting, because the modern theories of reversible and irreversible polycondensations predict that the fraction of cycles increases with the conversion until all reaction products have a cyclic topology at 100% conversion. For ROPPOC syntheses of cyclic polyesters two classes of catalysts have been found, namely strong nucleophiles, such as pyridines10–14 heterocyclic carbenes15–22 on the one hand, and covalent tin(II) or tin(IV) compounds23–26 on the other. A simplified polymerization mechanism involving covalent end groups catalyzed by dibutyltin bis(pentafluorophenoxide) BuSnOPF is outlined in Scheme 3.24 These ROPPOCs of lactide or CL play an important role in the present work.
The present study was stimulated by an observation mentioned by several research goups,27–30 which reported that alcohol-initiated ROPs of lactide catalyzed by tin(II)2-ethylhexanoate (SnOct2) proceed without back-biting at temperatures around 130 °C or below. Yet, despite absence of back-biting a smaller or larger fraction of odd-numbered chains was detectable (depending on temperature) indicating occurrence of a transesterification reaction. In the absence of transesterification reactions, a clean polymerization of lactide must yield even-numbered species regardless if linear or cyclic (if the initiator does not incorporate an additional unit). Hereinafter, this type of transesterification will be called “single unit exchange mechanism” (SUE).
As a typical example, Fig. 1A presents a MALDI mass spectrum of an ethyl L-lactate-initiated polymerization of L-lactide catalyzed by SnOct2 at 130 °C with a lactide/initiator ratio of 40/1.
Fig. 1 MALDI TOF mass spectra of polyLA polymerized at 130 °C with SnOct2 as catalyst: (A) initiated with ELA (No.1A, Table 1), (B) initiated with pentaerythritol (No. 6, Table 1). |
This spectrum shows a predominance of odd-numbered chains (due to incorporation of an ethyl lactate end group) along with a considerable fraction of even-numbered species. These findings raise the question, if it is possible to perform ROPPOC syntheses of cyclic polyesters in the absence of back-biting, but in the presence of other transesterification/equilibration reactions. Since in this case the formation of cyclic polyesters will exclusively result from ete-cyclization, such a ROPPOC synthesis of cyclic polyesters will be outside the JS-hypothesis of revPOCs. To avoid misunderstanding, it should be clarified that the authors differentiate between two aspects of the JS-theory. On the one hand, the work of J + S is a theory of ring-chain equilibration, which allows for the calculation of equilibrium constants and their correlation with the conformational properties of the polyester chains under investigation. On the other hand, J + S have developed a hypothesis of revPOCs on the basis of ring chain equilibration with a mathematical approach predicting the composition of the reaction mixture with increasing conversion. Both aspects of the JS theory overlap, of course, but are not identical. The critique of the authors in their previous publications5,6 and in the present work is exclusively directed against the JS-hypothesis of revPOCs.
In this context, the present work was aimed at finding catalysts and reaction conditions allowing for the realization of revPOCs proceeding without equilibration via back-biting, which is the chemical basis of the JS concept.
When SnOct2 was used as catalyst, 0.1 mL of a 0.4 M solution in toluene was added to the lactide by means of a syringe.
An analogous experiment was performed with 0.1 mmol of SnOct2 and a duration of 24 h.
For the GPC experiments a modular system kept at 40 °C (isocratic pump, 1 mL min−1, refractive index detector, RI-501 -Shodex) was applied. Samples were manually injected (100 μL, 2–4 mg mL−1). For instrument control and data calculation Clarity software (GPC extension, DataApex) was used. The calibration was performed using polystyrene standard sets (Polymer Standards Service – PSS, Mainz, Germany). For the determination of the Mark–Houwink–Sakurada (MHS) relationship a viscometer (Viscostar, Wyatt, Germany) and a multiangle laser light scattering (MALS) detector (Dawn EOS, Wyatt, Germany) were used. Astra 6.1 software (Wyatt) served for calculating the MHS curves.
The LA/In ratio was fixed at the low value of 40/1 for all experiments of this work, for the following reason. It was found in a previous study of SnOct2-catalyzed ROPs that a high initiator/catalyst ratio (typically >10/1) is beneficial for the suppression of back-biting. Presumably, a high catalyst concentration has the consequence that the ROPPOC mechanism can compete with the alcohol-initiated ROP, so that cycles are formed by ete cyclization and appear to result from back-biting. Therefore, a LA/Cat ratio of 1000/1 was preferentially used, when SnOct2 served as catalyst (Table 1). However, for other catalysts, such as Ph2SnCl2 or BuSnOPF it was found that higher catalyst concentrations (LA/Cat = 100/1 up to 400/1) are beneficial to avoid side reactions including cyclization. Finally, it should be mentioned that most experiments were performed at 130 °C, because it was learned from ROPs catalyzed with SnOct2 that at this temperature and short reaction times (<3 h) formation of cycles by back-biting is avoidable. The following description of results is subdivided into sections according to the catalysts used for the polymerizations.
Exp. no. | Initiator | LA/Cat | Temp. (°C) | Time (h) | M n | M w | Đ | Cycles |
---|---|---|---|---|---|---|---|---|
1 | ELA | 400/1 | 130 | 1 | 11100 | 12400 | 1.1 | — |
1B | ELA | 400/1 | 130 | 3 | 11000 | 12300 | 1.1 | — |
2A | ELA | 1000/1 | 130 | 1 | 10600 | 11700 | 1.1 | — |
2B | ELA | 1000/1 | 130 | 3 | 10800 | 11900 | 1.1 | — |
3 | ELA | 1000/1 | 160 | 1 | 10900 | 15000 | 1.4 | Traces |
4A | HMBD | 1000/1 | 130 | 1 | 10700 | 12300 | 1.2 | — |
4B | HMBD | 1000/1 | 130 | 3 | 11000 | 12300 | 1.1 | — |
5 | PENT | 1000/1 | 130 | 1 | 8 900 | 10400 | 1.2 | — |
6A | — | 400/1 | 130 | 1 | 44000 | 76000 | 1.7 | + |
6B | — | 400/1 | 130 | 3 | 48000 | 84000 | 1.7 | + |
7A | — | 1000/1 | 130 | 1 | 118500 | 188000 | 1.6 | + |
7B | — | 1000/1 | 130 | 3 | 121500 | 217000 | 1.8 | + |
8 | — | 1000/1 | 160 | 1 | 91000 | 240000 | 2.6 | ++ |
Scheme 4 Alcohol-initiated ROP of L-lactide catalyzed by SnOct2 at low temperature and high In/lac ratio. |
Fig. 2 MALDI TOF mass spectra of polyLA polymerized at 130 °C/3 h with SnOct2 as catalyst (LA/Cat 1000/1): (A) initiated with ELA (No. 2B, Table 1), (B) without initiator (No. 7B, Table 1). |
These three trends were also valid for all alcohol initiated ROPs based on other catalysts (Tables 2–6). An even/odd equilibration around 50% (Fig. S1†), which increased with time, was also found for the ROPs initiated with the primary alcohol HMBD (No. 4A and 4B).
Exp. no. | Init. | Lac/Cat | Temp. (°C) | Time (h) | M n (meas.) | M w | Đ | Cycles |
---|---|---|---|---|---|---|---|---|
1 | ELA | 200/1 | 130 | 1 | 9200 | 10600 | 1.2 | — |
2 | ELA | 400/1 | 130 | 1 | 9100 | 10200 | 1.2 | — |
3 | ELA | 400/1 | 160 | 1 | 9100 | 14000 | 1.5 | — |
4 | ELA | 400/1 | 160 | 3 | 10000 | 15100 | 1.5 | — |
5 | — | 200/1 | 130 | 1 | 47000 | 105000 | 2.2 | + |
6 | — | 400/1 | 130 | 1 | 63000 | 153000 | 2.4 | + |
7 | — | 400/1 | 160 | 1 | 79500 | 205000 | 2.6 | ++ |
Exp. no. | Init. | Lac/Cat | Temp. (°C) | Time (h) | M n (meas.) | M w (meas.) | Đ | Cycles |
---|---|---|---|---|---|---|---|---|
1A | ELA | 200/1 | 130 | 1.0 | 9000 | 10700 | 1.2 | — |
1B | ELA | 200/1 | 130 | 2.0 | 10100 | 11200 | 1.1 | — |
1C | ELA | 200/1 | 130 | 4.0 | 10200 | 11300 | 1.1 | — |
2 | ELA | 400/1 | 130 | 1.0 | 8600 | 10800 | 1.2 | — |
3A | ELA | 400/1 | 160 | 1.0 | 8700 | 11000 | 1.3 | — |
3B | ELA | 400/1 | 160 | 3.0 | 9000 | 13800 | 1.4 | Traces |
4A | — | 200/1 | 130 | 2.0 | 44000 | 69500 | 1.6 | — |
4B | — | 200/1 | 130 | 4.0 | 43500 | 70500 | 1.6 | + |
5A | — | 400/1 | 130 | 2.0 | 41000 | 64000 | 1.6 | + |
5B | — | 400/1 | 130 | 4.0 | 37000 | 58000 | 1.6 | — |
6 | — | 100/1 | 160 | 1.0 | 42000 | 107000 | 2.5 | ++ |
7 | — | 200/1 | 160 | 1.0 | 54000 | 124000 | 2.3 | + |
8 | — | 400/1 | 160 | 1.0 | 55000 | 132000 | 2.4 | + |
Exp. no. | Init. | Lac/Cat | Temp. (°C) | Time (h) | M n (meas.) | M w (meas.) | Đ | Cycles |
---|---|---|---|---|---|---|---|---|
1 | ELA | 200/1 | 130 | 1 | 8500 | 12000 | 1.4 | — |
2 | ELA | 400/1 | 130 | 1 | 8300 | 11500 | 1.4 | — |
3 | ELA | 400/1 | 160 | 1 | 8800 | 15000 | 1.7 | Traces |
4 | ELA | 400/1 | 160 | 3 | 8500 | 14700 | 1.7 | Traces |
3A | — | 200/1 | 130 | 0.5 | 53000 | 111000 | 2.1 | ++ |
3B | — | 200/1 | 130 | 1 | 50000 | 103000 | 2.1 | ++ |
4 | — | 400/1 | 130 | 1 | 55000 | 139000 | 2.5 | ++ |
5A | — | 600/1 | 130 | 1 | 75000 | 181000 | 2.4 | ++ |
5B | — | 600/1 | 130 | 2 | 68000 | 157000 | 2.3 | ++ |
Exp. no. | Init. | CL/Cat | Time (h) | M n | M w | Đ | Cycles |
---|---|---|---|---|---|---|---|
1 | EHH | 200/1 | 1.0 | 8100 | 16000 | 2.0 | — |
2 | EHH | 400/1 | 1.0 | 8300 | 17000 | 2.0 | — |
3 | EHH | 1000/1 | 1.0 | 9000 | 18500 | 2.1 | — |
4A | HMBD | 400/1 | 1.0 | 6500 | 15700 | 2.5 | — |
4B | HMBD | 400/1 | 2.0 | 4100 | 15300 | 3.7 | — |
5A | HMBD | 600/1 | 1.0 | 6100 | 15300 | 2. | — |
5B | HMBD | 600/1 | 2.0 | 5400 | 13400 | 2. | — |
6 | — | 200/1 | 1.0 | 46000 | 92000 | 2.0 | ++ |
7 | — | 400/1 | 1.0 | 61000 | 127000 | 2.1 | ++ |
8 | — | 600/1 | 1.0 | 70500 | 143000 | 2.1 | ++ |
9 | — | 1000/1 | 1.5 | 85000 | 168000 | 2.0 | ++ |
Their mass spectra were quite similar to those of Fig. 1A. For a project dealing with biodegradable coatings, a ROP initiated with pentaerythritol was performed at 130 °C and in this case the odd-/even equilibrium was even complete as displayed in Fig. 1B. All these mass spectra have in common, that no cycles were detectable (s. Fig. S1B†). In the case of experiments 2B and 3, the low molar mass fraction was also subject to ESI† mass spectroscopy and again no cycles were found for No. 2B (130 °C) and only a trace for No. 3 (160 °C). The 160 °C experiment also revealed almost complete odd-/even equilibration along with broadening of the molecular weight distribution indicating random intermolecular transesterification. (Fig. S2A†). Hence these experiments demonstrate that under the given reaction conditions ROPs including intermolecular transesterification exist without contribution of back-biting.
With neat SnOct2 completely different results were obtained. The polymerizations were slower without addition of alcohol (a well-known phenomenon), and with a LA/Cat ratio of 1000/1 only a conversion of 91% was reached after 1 h, but after 3 h the conversion had reached the equilibrium level of 97%. With a LA/Cat ratio of 400/1, a conversion of 97% was reached within 1 h.
Regardless of the LA/Cat ratio, high molecular weights were achieved, and quite similar mass spectra were obtained which displayed strong peaks of cycles (Fig. 2B) along with a peak of an unidentified linear species.
Complete odd/even equilibration was not achieved, what indicates that the cycles were formed by ete-cyclization and not by back-biting. At 160 °C complete conversion, complete equilibration and a high molecular weight was already obtained after 1 h. MALDI TOF mass spectra exclusively displaying peaks of cycles (Fig. S2B†) and intrinsic viscosity measurements confirming a cyclic topology also for the high molar mass fraction were found, results that were recently published and thus, do not need extensive discussion at this point.26 All these properties were in agreement with the ROPPOC mechanism outlined in Scheme 3 as discussed in a recent publication.26
As outlined in Scheme 3, formation of linear chains having a COOH chain end may compete with the formation of cycles although to a low extent. This analytical problem does not exist for tin acetates. As published recently,26 the reactivity of SnAc2 is quite similar to that of SnOct2. At LA/In ratios <100/1 linear acetate terminated chains are predominantly formed, but at higher ratios formation of cycles is prevalent and at LA/In ratios >400 linear chains are not detectable in the mass spectra anymore. Furthermore, SnAc2 can catalyse alcohol-initiated ROPs, so that the average degree of polymerization (DP) parallels the LA/In ratio and low dispersities are obtained. The results listed in Table 2 are in line with previous results.26
The MALDI TOF mass spectra of the ELA-initiated poyLAs (No. 1–4, Table 2) were quite similar to those obtained with SnOct2, and Fig. 3 also illustrates the dependence of the even/odd equilibration on the catalyst concentration. Particular important is again the absence of cycles. The results obtained with neat SnAc2 were also analogous to those found with SnOct2, because a considerable fraction of cycles was formed after 1 h at 130 °C (Fig. 4A). At 160 °C the predominance of cycles was more complete (Fig. 4B). Regardless of temperature, again high molecular weights and high dispersities were obtained along with full odd/even equilibration. Therefore, the SnAc2 experiments were in full agreement with the SnOct2-catalyzed polymerizations and their interpretation.
Fig. 3 MALDI TOF mass spectra of polyLA polymerized at 130 °C with SnAc2 as catalyst and ELA as initiator (LA/In = 40/1): (A) LA/Cat = 200/1 (No. 1, Table 2), (B) LA/Cat = 400/1 (No. 2, Table 2). |
Fig. 4 MALDI TOF mass spectra of polyLA polymerized with SnAc2 as catalyst (LA/Cat = 400/1): (A) at 130 °C/1 h (No. 6, Table 2), (B) 160°/1 h (No. 7, Table 2). |
The (hitherto unpublished) ROPs initiated with ELA yielded polylactides of low dispersity and the molecular weights were almost independent on the LA/Cat ratio but corresponded to the LA/ELA ratio. At this point, it should be mentioned that all tin catalyst mentioned above yielded at 160 °C higher Mw values, whereas the increase of Mn was small. In agreement with the mass spectra this effect may be attributed to more efficient transesterification reactions resulting in higher dispersities. Condensation steps of the ethyl ester end groups (see discussion below) may be responsible for a slight increase of Mn. The MALDI TOF spectrum presented in Fig. 5 not only illustrates the relatively narrow MWDs, but also the absence of cycles.
Fig. 5 MALDI TOF mass spectra of polyLA polymerized at 130 °C/2 h with Ph2SnCl2 as catalyst (LA/Cat 200/1) and initiated with ELA: (A) full spectrum, (B) expanded segment (No. 1B, Table 4), the asterisk indicates the position of a hypothetical cycle. |
In contrast, all polymerizations conducted with neat Ph2SnCl2 yielded considerably higher molecular weights and higher dispersities. Furthermore, the mass spectra evidenced that mainly cyclic polylactides were formed as demonstrated in Fig. 6. In summary, the results obtained with Ph2SnCl2 were similar to those obtained from SnOct2 and SnAc2, and the predominant formation of high molar mass cyclic polylactides is also in good agreement with the previously published experiments. The results of the neat catalyst are also in perfect agreement with the hypothesis of a ROPPOC mechanism involving the intermediate formation of CO-Cl end groups as discussed previously.25
Fig. 6 MALDI TOF mass spectrum of polyLA polymerized with neat Ph2SnCl2 in bulk at 130 °C/2 h (No. 4A, Table 2). |
Fig. 7 MALDI TOF mass spectra of polyLA polymerized with BuSnOPF as catalyst (LA/Cat = 200/1) at 130 °C/1 h: (A) initiated with ELA (No.1, Table 4), (B) without initiator, (No. 3B, Table 4). |
Noteworthy is the absence of cycles in the 130 °C samples, whereas traces of cycles were detected in the 160 °C samples. The polymerizations catalyzed with neat BuSnOPF yielded polylactides with the four characteristics typical for a ROPPC process (see Scheme 4). First, the molecular weights were considerably higher, second the dispersities were higher, the odd/even equilibration was complete, and cycles were the predominant reaction products as demonstrated in Fig. 7B and S3B.†
A remarkable difference relative to analogous ROPs of lactide is the higher dispersity which is immediately evident from the MALDI TOF mass spectra (Fig. 8A and S4A†). In absence of an initiator higher molecular weights were achieved and the mass spectra exclusively displayed peaks of cycles (Fig. 8B).
Fig. 8 MALDI TOF mass spectra of polyCL polymerized with BuSnOPF as catalyst at 130 °C (LA/Cat = 200/1): (A) initiated with ethyl 6-hydroxyhexanoate (No. 1, Table 6), (B) without initiator (No. 6, Table 6). L – linear chains having an ethyl ester end group. |
Although information about transesterification yielding odd/even equilibration is, of course, lacking in the mass spectra of polylactones, the results obtained from the BuSnOPF catalyzed polymerizations listed in Table 6 clearly support the conclusions extracted from the polymerizations od L-lactide.
Therefore, the results presented in this work would not be new, if equilibration reactions were absent. As mentioned in the Introduction reversible back-biting was considered by J + S to be the only source of ring-chain, ring–ring and chain–chain equilibration. Therefore, it was a particular important aspect of the present study to demonstrate that equilibration reactions occur in the absence of back-biting. As discussed above, the SUE mechanism, which is perhaps confined to the chemistry of lactide is a first example of a transesterification/equilibration mechanism which operates at temperature below which back-biting sets out. However, intermolecular transesterification reactions that broaden the MWD occur more frequently and at lower temperatures than back-biting, which was already reported previously.36 That model reaction was performed at 120 °C in such a way that L-lactide was polymerized by means of SnOct2 + benzyl alcohol in the presence of preformed polylactide having blocked end groups (Ac-PLA-Et). Three series of additional experiments was performed in the present work. In the first series L-lactide was polymerized at 160 °C with neat SnOct2, SnAc2 or BuSnOPF, so that cyclic polylactides were formed. After 1 h the reaction was stopped by cooling. The polyLA was dissolved in dry dichloromethane and combined with diethyl succinate.
After homogenization the dichloromethane was evaporated at 80 °C and the remaining reaction mixture was thermostated at 130 °C for 1 or 2 h (see Table 6). Diethylsuccinate was selected as reaction partner of polyLA for three reasons. First its boiling point is high enough to avoid premature vaporization. Second it seemed to be partially miscible with molten poly(L-lactide) and the masses of reaction products expected from transesterification (Lc and Ld chains, Scheme 6) allowed for easy identification by mass spectrometry (Lc and Ld are isomers and give identical mass peaks). The mass spectra proved that Lc and Ld chains were formed in the experiment with BuSnOPF after 2 h (Fig. S6†).
A second series of experiments was performed in such a way that the potential polycondensation of ethyl L-lactate was studied at 130 °C with addition of SnOct2 or BuSnOPF as catalysts (Table 7).
Exp. no. | Catalyst | LA/Cat | Temp (°C) | Time (h) | La Chains | Cyclics |
---|---|---|---|---|---|---|
1 | SnOct2 | 1000/1 | 130 | 4 | + | — |
2 | SnOct2 | 400/1 | 130 | 24 | ++ | — |
3 | SnOct2 | 400/1 | 160 | 4 | ++ | — |
4A | BuSnOPF | 400/1 | 130 | 4 | + | — |
4B | BuSnOPF | 400/1 | 130 | 24 | ++ | Traces |
5 | BuSnOPF | 400/1 | 130 | 4 | ++ | — |
Due to the low reactivity of the ethyl ester group a first experiment was performed with the relatively low LA/SnOct2 ratio of 400/1 and La chains were indeed detectable after 4 h (Fig. 9A). Their concentration was higher after 4 h at 160 °C (Fig. 9B). Even with a LA/SnOct2 ratio of 1000/1 oligoLAs were formed, when the time was prolonged to 24 h and the La chains became detectable in the MALDI mass spectrum up to a DP of 25. In all three experiments peaks of cycles were absent.
Fig. 9 ESI† mass spectra of oligoLA prepared by polycondensation of ELA with SnOct2 (A) at 130 °C/4 h in, (B) 160 °C/4h. |
Similar results were obtained with BuSnOPF, which proved to be slightly more reactive than SnOct2 Cycles were again absent in the 4 h experiments (No. 4A and 5, Table 7 and Fig. S7†), but traces of cycles were observable in the ESI spectrum after 24 h at 130 °C (Fig. S8†). In the MALDI mass spectra La chains were detectable up to DPs around 40 (No. 5 (Fig. S9A†), and up to DPs around 140 (No. 4B, Fig. S9B†), whereas peaks of cycles were absent. These results include three remarkable aspects. First, they demonstrated the existence of an intermolecular transesterification of the ethyl ester end groups, which is reversible as long as the liberated alcohol is present in the reaction mixture. Second, they indirectly indicate that the active chain ends (Sn–O–CH) can also react with the lactyl-lactyl bonds in the polymer backbone, because those ester bonds are more reactive than the ethyl ester end groups. This conclusion is confirmed by previously published transesterification experiments.36 In other words, the polycondensations of ELA involve two types of intermolecular transesterification reactions. Third, the experiments of Table 7 present in this work the first examples of reversible polycondensations without back-biting outside the ROPPOC chemistry discussed before.
A third series of transesterification experiments was performed with Ac-PCL-Et (Scheme 7 and Table 8) as reaction partner of dimethyl succinate. Ac-PCL-Et having a DP of 40 was prepared by SnOct2-catalyzed and ethyl 6-hexanoate-initiated polymerization of CL at 130 °C followed by acylation of the OH–CH end group with acetic anhydride and pyridine.
Exp. no. | Catalyst | CL/Cat | Time (h) | Lf, Lg, Lh-chains |
---|---|---|---|---|
1A | SnOPF | 200/1 | 1 | + |
1B | SnOPF | 400/1 | 2 | ++ |
1C | SnOPF | 400/1 | 4 | ++ |
2A | SnOct2 | 200/1 | 1 | + |
2B | SnOct2 | 200/1 | 2 | ++ |
The MALDI TOF mass spectrum presented in Fig. 10A proved the formation of the expected structure. This model polymer was dissolved in dimethyl succinate and thermostated at 130 °C in the presence of BuSnOPF. The results listed in Table 8 and Fig. 10 demonstrate that the Lf, Lg and Lh chains indicating transesterification were detectable after 1 h, when a LA/Cat ratio of 200/1 was used or after 2 h with a LA/Cat ratio of 400/1. Analogous results, although with different intensity ratios of the Lf, Lg and Lh peaks were obtained with SnOct2 as catalyst.
Fig. 10 MALDI TOF mass spectra of (A) Ac-PCL-Et (average DP = 40), (B) reaction product of Ac-PCL-Et and dimethyl succinate catalyzed by BuSnOPF (LA/Cat = 200/1) after 1 h at 130 °C (No. 1A, Table 8) measured with Na doping. |
In summary, including previous results four different kinds of intermolecular transesterifications were performed at 120 or 130 °C which all proceeded under conditions where back-biting was absent.
Therefore, it is justified to conclude that the ROPPOCs presented in Tables 1–6 may be called reversible polymerizations and polycondensations despite absence of back-biting.
First: As already mentioned in the Introduction, the work of J + S has two aspects. On the one hand, it is a theory of ring-chain equilibria, on the other hand, it has been presented as a general theory of revPOCs. It is this second aspect, which is criticized by the authors. Characteristic for the JS-hypothesis of revPOCs is the assumption that equilibration exclusively results from back-biting. In contrast, the experiments presented above prove that revPOCs may proceed without back-biting. Any evidence is lacking that polycondensations exist, where according to J + S the reversibility is exclusively based on back-biting.
Second: J + S excluded that ete-cyclization plays a noteworthy role in the course of revPOCs. The ROPPOC experiments presented in this work (and previously5) demonstrate that ete-cyclization plays an important role in revPOCs, even when back-biting takes place. The results presented above demonstrate, that efficient formation of cycles may also occur in absence of back-biting. Any experimental evidence is lacking that polycondensations exist, which involve formation of cycles exclusively on the basis of backbiting. Therefore, the results described above are in total contradiction to the J + S hypothesis of revPOCs and thus, are considered to be outside the JS concept of revPOC, even when a ring-chain equilibration takes place.
Third: Even when back-biting is the main or only source of equilibration, the JS hypothesis does not provide a correct description of the course of revPOCs up to 100% conversion. For polycondensations at high conversions the JS theory predicts that a group of monomers yields 100% cycles at 100% conversion, whereas, for other groups of monomers only a few percent of cycles is predicted. This calculation is a total failure of chemical logic, because 100% conversion yields 100% of cycles for any kind of monomers and all reaction conditions.
Furthermore, it should be considered that all the experiments presented above have in common that they are based on cyclic esters and thus, indirectly on polycondensations of hydroxy acids. However, the technical production of most polyesters and the Jacobson–Stockmayer–Beckmann experiments3 are based on polycondensations of α,ω-diols and dicarboxylic acids or their dimethyl esters. Hence, such polycondensations need to be considered (and (re)investigated) to find out, to what extent the results and conclusions achieved in this work agree with a2 + b2 polycondensations. However, first pertinent experiments that shed light on this aspect have already been published by the first author in another context.38–40 For example, telechelic polyesters were prepared from dimethyl adipate or dimethyl sebacate and excess of ethane diol.38 With bismuth catalysts no cycles were formed up to temperatures of 240 °C. A typical MALDI TOF mass spectrum is displayed in Fig. S10.† Furthermore, telechelic polyesters were prepared from dimethyl terephthalate and excess 1,4-butane diol and with bismuth catalysts and no cycles were formed up to 240 °C. Moreover, 1,6-hexane diol or 1,10-decane diol were polymerized with dimethyl isophthalate in bulk catalyzed by Ti(OnBu)4 with variation of time and temperature. At temperatures up to 210 °C low molar mass polyesters free of cycles were obtained. All these syntheses of telechelic polyesters have in common, that at higher temperatures and longer times formation of small amounts cycles was observed. If these cycles were formed by backbiting, by ete-cyclization or by both reaction pathways remains an open question. Anyway, when those polycondensations are stopped because a sufficiently high molecular weight is achieved, they are certainly still far from complete equilibration, because perfect equilibration either requires extremely fast trans esterification reaction or long reaction times, which are usually not applied in preparative experiments.
In all those polycondensations, chain growth steps are reversible, because the liberated methanol in combination with catalyst can cleave any ester bond in the polymer chains. Hence, those experiments indicate that also in the field of a2 + b2 polycondensations reaction conditions exist that allow for polyester syntheses involving intermolecular transesterification in the absence of back-biting or with a negligible contribution of back-biting. Hence, also those polycondensations are outside the JS-hypothesis. All these facts and arguments together clearly demonstrate that the JS concept cannot serve as general theory of revPOCs.
As a consequence of this conclusion, the authors propose a new concept of revPOCs based on the definition of three classes of polycondensations:
Class I: Irreversible polycondensations that may also be labelled kinetically controlled polycondensations. This definition, in principle, agrees with Florýs polycondensation theory with the exception that ete-cyclization occur, which was denied by Flory.
Class II: RevPOCs exclusively involving back-biting as the only source of reversibility and including such a high rate of equilibration that thermodynamic control is given at any state of the polycondensation process.
Class III: RevPOCs involving rapid intermolecular trans-reactions in the absence of backbiting. Again, equilibration is so efficient that thermodynamic control is given at any stage of the polycondensation.
To avoid misunderstanding the term “thermodynamically controlled polycondensation” needs to be clarified. At long reaction times (i.e. weeks or months), all revPOC will end up with complete equilibration of all reaction products and thus, justify being called “thermodynamically controlled” polycondensations. Yet, as demonstrated by the ROPPOC experiments above and by the experiments in refs.38–40 in many real polycondensations the equilibration reactions will not be efficient enough to reach thermodynamic control at any stage of the process.
The three classes of polycondensations defined above should be considered as extreme cases which form the corner of a triangle. The vast majority of revPOCs will, in fact, be located somewhere inside this triangle. Consider, a series of class III-type polycondensations with decreasing rate constants of trans-reactions. When the rate constants approach zero the definition of Class I is reached and thus, this series of polycondensations is located on the line connecting Class III and Class I. An analogous thought experiment can be formulated for Class II polycondensations. The line connecting class I and II is formed by polycondensations that involve fast equilibration reactions, so that thermodynamic control is established at any time, but the contributions of back-biting and intermolecular equilibration reactions varies. If all the equilibration reactions are not fast enough to establish full thermodynamic control at any stage, this polycondensation will be positioned somewhere inside the triangle. In other words, revPOCs cover a wide area of kinetic and thermodynamic properties and exact knowledge of the effectiveness of all involved trans-reactions is required for a detailed description of an individual polycondensation process.
Finally, a publication of Montaudo et al.41 should be cited which, as early as 1997, described MALDI TOF mass spectrometric analyses of polylactides prepared by means of an Al complex containing one Al-OCH3 group. This methoxide group played the role of initiator and yielded methyl ester terminated polylactides. Despite the low temperature of 70 °C and short times complete odd/even equilibration was observed although cyclic oligomers were absent and only appeared after long polymerization times. Hence, these results demonstrate that a significant predominance of intermolecular transesterification reactions (entailing broadening of the Mw) over back-biting is not limited to the chemistry of tin catalysts. This conclusion and the results presented in ref. 39–41 also suggests that catalyst of other elements (e.g. Zn, Ti and Bi) may be found, that enable ROPPOC of cyclic esters obeying to the principles of Class III polycondensations.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1py00704a |
This journal is © The Royal Society of Chemistry 2021 |