Silicon‐Catalyzed Depolymerization of Polyethers: Pushing Scope, Practicability and Mechanistic Understanding

The depolymerization of polyethers is a sustainable yet challenging opportunity for a circular economy in materials processing. While we recently identified silicon Lewis superacids as promising catalysts for this transformation, limited scope (e. g., terminal OH groups not tolerated) and strict requirements for anhydrous conditions hampered wider applicability. In the present work, the impact of different polyether structures and reaction conditions were evaluated. By doing so, the generality for structural variations was confirmed and substantial improvements made the depolymerization feasible for large‐scale applications under ambient conditions. Based on systematic experimental screenings, a refined mechanistic model of the depolymerization process is developed.


Introduction
Metathesis of C(sp 3 )À X bonds gained increasing attention in past decades but is still much underdeveloped compared to metathesis reactions of other bonds. [1]As commonly employed CÀ X bond activations by oxidative addition encounter obstacles from high reaction barriers and side reactions like β-hydride elimination, Lewis acid catalysis emerged as a powerful alternative for this type of transformation. [2]Pioneering work by Enthaler et al. established transition metal based Lewis acids in the depolymerization of poly-THF, performing ring-closing metathesis (RCM) to yield the monomer THF under harsh conditions. [3]The Morandi group broadened the substrate scope by using iron(III) triflate in the RCM of bis-and tris-ethers, which the Liu group extended by using ionic liquids as reaction media (Scheme 1a). [4]However, these reports did not face the RCM of polyethylene oxide (pEO), a commodity polymer for which selective degradation would present a convenient entry into circular economy. [5]Enthaler und Cantat presented original approaches for pEO depolymerization, but using stoichiometric amounts of interception agents, while Jutzi was able to catalyze the CÀ O bond metathesis of short, methyl end-capped oligoethers with the low-valent Lewis acidic Cp*Si(II) + cation (Scheme 1b). [6]e Lewis acidity of bis(catecholato)silanes has gained importance in the field of catalysis during the past years, which lead us to probe bis(perchlorocatecholato)silane 1 in its oligomeric form [1] n as a CÀ O bond metathesis catalyst (Scheme 1a). [7]This system exhibited high yields in the depolymerization of pEO (Scheme 1b), and computational studies on model substrates allowed the postulation of a mechanism which confirmed preliminary experimental observations made by other groups.A low tendency of polydentate binding of substrates with > 1 donor groups and a high Lewis acidity were identified as key signatures for successful catalysis.Based on this knowledge, we employed the latest example of a silicon-based Lewis superacid, bis(nonafluoro-N-phenylamidophenolato)silane in this reaction, which even catalyzed the cross-metathesis of ethers. [8]More recently, the Inoue group successfully demonstrated bis(perfluoropinacolato)silane in the RCM of pEO, confirming our mechanistic model (Scheme 1b). [9]owever, all reported RCM reactions of pEO exhibited major problems that would render further applications, in particular on an industrial scale, as difficult.
All hitherto employed catalysts are sensitive to moisture and air, which necessarily required handling under inert conditions.In addition, terminal OH groups were found incompatible with the presented systems, which limited the application to protected ethers without protic functionalities.Further, the current scope of polyethers was minimal and focused on polymers with low molecular weight, lacking proof that the protocol can be extended to other polyether derivatives.
The present work aims to broaden this catalytic portfolio while investigating the influences of several structural features of the polyether on the reaction outcome (Scheme 1c).Based on the gathered information, reaction conditions are optimized to make this process feasible for wider applications, including a convenient in-situ generation of the catalyst from SiCl 4 and perchloro-and perbromocatechol (H 2 cat Cl/Br ).Unprotected polyethers, not applicable under previous conditions, are shown to easily undergo RCM, even if handled under ambient atmosphere.Based on the experimental insights, a refined version of our previously reported mechanism is given.

Results and Discussion
We launched our studies by proving the generality of the previously described protocol for depolymerizing a range of polyether derivatives.Therefore, nine polyethylene glycols with different features were investigated regarding the influence of end-capping groups and chain length (Scheme 2).
First, the effect of different end-capping groups was evaluated.While previously, only a single α,ω-methyl endcapped telechelic polyether was investigated, we extended the scope to allyl-, acetyl-and butyl-end-capped pEOs, and compared the results with the uncapped OH functionality (Table 1).Apolar end-capping groups are generally tolerated (entries 1 & 4), yielding comparable results under the established (ortho-dichlorobenzene, 115 °C), and milder conditions (dichloromethane, 60 °C).Even the strongly coordinating ester functionality is tolerated, although with decreased yields.The possibility of cleavage of the ester, resulting in stoichiometric amounts of acetate, which has a high affinity towards 1, could explain these findings. [10]To our surprise, even the unprotected ωÀ OH terminal polyether (PE4À H) was degraded, but only with a TON of 5, which underlines the severe catalyst deactivation.
Probing the behavior of varying molecular weights in αÀ BuÀ ωÀ OMe polyethers in the CÀ O bond metathesis, indicated beneficial effects of longer polyether chains (Table 2).
We reasoned that the higher local concentration of ether functionalities in close proximity to the catalytically active center of the Lewis acid-substrate complex, that occurs in the case of longer polyethers, has beneficial effects for the reaction outcome. [11]This can be explained by an overall increase in polarity around the catalyst, which stabilizes ionic intermediates in the mechanism and by close proximity of nucleophilic groups which are necessary to close the catalytic cycle via a subsequent Scheme 2. Overview about the heterochelechelic polyethers investigated in this study (n max = average length as determined by MS spectrometry).
Table 1.a] Entry Polyether ω-Capping  [b] Determined by 1 H-and 13 C-NMR spectroscopy and/ or GC analysis using cyclooctane as an internal standard.Yields in parentheses refer to reactions in DCM at 60 °C.
intermolecular S N 2 reaction (see Mechanism, below).To corroborate this hypothesis, reactions at different polyether concentrations but identical chain lengths were conducted (Table 3).Indeed, increased concentrations had a beneficial influence on the yield.Noteworthily, the reaction worked even solvent-free in bulk polyether on a 300 mg scale in 50 % yield.This information underlined the hypothesis that side reactions between the key intermediate oxonium ion and a nucleophile other than the alkoxy-silicate lead to the inhibition of the catalyst, another crucial piece for the mechanistic proposal (see below).
With the knowledge that an increased molecular weight and concentration improves the catalyst lifetime, we returned to the central problem, namely the limited compatibility with unprotected, (α),ωÀ OH polyethers, as observed in entry 3 of Table 1.To our delight, an increased molecular weight also shows beneficial effects for the catalyst activity in the degradation of polyethers with OH functionalities (Table 4), although the yields were still reduced compared to the results obtained with protected ethers (Table 2).
We were speculating, if the deactivation occurred from the cleavage of catechol from the silicon-based catalyst 1 by reaction with the terminal OH groups, which also renders this catalyst sensitive towards moisture.7d,12] To our surprise, the obtained yields were only slightly improved, indicating that OH functionalities cause similar catalyst deactivation for 2 (Table 4).For instance, when 2-sulf 2 was implemented as a catalyst for the degradation of the ωmethyl-protected polyether PE3, significantly higher yields were achieved (73 %) compared to the unprotected derivative (entry 3 & 4, Table 4).This observation led to the crucial conclusion that not alcoholysis of the catalyst seemed the major problem, but rather catalyst inhibition through formation of a catalytically inactive alkoxy-silicate/germanate via acidification of a coordinated alcohol, which requires subsequent reaction with an electrophile (oxonium-ion, H + , etc.) to revert to its active form.Before approaching the thus still remaining "OHproblem", another improvement was attempted.It was known that donor-adducts of 1 rapidly form by the reaction of HSiCl 3 or SiCl 4 with perchlorocatechol.Hence, we tried an in-situ generation of catalysts from these commercial precursors without isolating the moisture-sensitive Lewis acid.In a first attempt, we achieved promising results for the in-situ preparation of 1 in the presence of liquid α-allyl-ωÀ OMe pEO, PE4À Me (Table 5).
Table 3. Evaluation of the effect of concentration on the degradation of αÀ BuÀ ωÀ OMe heterochelechelic pEO PE1 (n max = 13).Control experiments with only SiCl 4 or perchlorocatechol resulted in no depolymerization, corroborating the role of 1 as the active species.The activity was not exclusive to allyl-endcapped polyether, but also αÀ Bu/ωÀ OMe pEO PE3 was degraded in higher yields compared to the use of oligomeric [1] n as catalyst (compare Table 2, entry 3).With this new approach, we returned to the "OH-problem".Much to our delight, unprotected polyether, αÀ Bu/ωÀ OH pEO PE1À H, which during our study proved to be the most challenging substrate, efficiently depolymerized when using the in-situ catalyst formation.The observed yield (62 %) is significantly higher than the results with [1] n (17 %, Table 4, entry 1).The in-situ formation is even feasible in neat PE1À H, although with diminished yields (300 mg scale, 31 %).While we are aware, that the in-situ generation of 1 requires an excess of the volatile SiCl 4 , which readily reacts with moisture, the thereby formed catalytic system turned out as the most effective and robust one, substantially improving and facilitating the process and making isolation and handling of the sensitive oligomer [1] n obsolete.
With the possibility of an in-situ generation of the catalyst in hand, we aimed to evaluate remaining factors.Since previous findings in our group revealed the increased Lewis acidity in the heavier bis(perbromocatecholato)silane, [7b] we tried to replace the more expensive and more toxic perchlorocatechol with the cheaper and non-toxic perbromocatechol.Indeed, a similar yield in the degradation of the unprotected PE1À H (αÀ Bu/ωÀ OH pEO) was achieved (60 %).We also investigated other silane precursors but found that SiCl 4 already performed superior (see ESI section 1.4 for details).
The fact that an in-situ catalyst generation, which liberates HCl, increases the catalyst lifetime and, therefore, the yields significantly, lead to the question whether chloride ions or acidic conditions have a beneficial influence on the catalytic activity of 1.While reactivity was quenched after the addition of 10 mol% PPh 4 Cl (see SI for details), the effect of acidic conditions was further investigated by adding bis(trifluoromethane)sulfonimide (HNTf 2 ) as an additive to the depolymerization catalyzed by [1] n .
As suspected, the addition of HNTf 2 increased the yields substantially (Table 6).Notably, control experiment with 5 mol% HNTf 2 in the absence of the Lewis acid also catalyzed the reaction in a moderate yield (entry 4, Table 6).This is in line with the findings of Morandi et al., [4a] but the Lewis superacid [1] n alone still shows superior reactivity in comparison to Brønsted catalysis (51 %, entry 3, Table 3).While the increased yields in the presence of the Brønsted acid could be explained by two independent catalytic cycles, the fact that the addition of an electrophile (H + in the case of HNTf 2 ) is beneficial for the lifetime of the catalyst is further supported by the finding that stoichiometric amounts of iodomethane increase the yield in the degradation of αÀ Bu/ωÀ OMe pEO PE3 by 11 %, while other additives like Et 3 SiH negatively affected the observed yields (see SI for details).This observation again underlines the hypothesis that the anionic silicate plays a major role in catalyst inhibition.By reaction with an external electrophile, a regeneration of the catalyst seems plausible.
To gain further insight into the processes during the depolymerization, we conducted a series of reactions in neat polyether (αÀ Bu/ωÀ OMe pEO PE2, n max = 25) and stopped the reaction after several timepoints, as seen in Table 7. Already after 30 minutes of reaction time, much less than the usually employed time margins for this reaction (18 h), substantial yields were observed (Entry 2).MALDI-MS analysis of the residual polyether indicated a statistical distribution of chain lengths with its maximum (n max ) shifting from 25 to 12 with increased reaction time.
Remarkably, a broadening of the chain length distribution can be observed for increased reaction time, which is indicating cross-metathesis occurring along with depolymerization.This observation introduces novel opportunities that could be achieved by chain redistribution to affect the properties of polyethers -a hitherto not considered aspect.
[b] Referring to 1,4-dioxane as the reaction product, determined by GC analysis using cyclooctane as an internal standard.
[c] Without [1] n .[d] Referring to 1,4-dioxane as the reaction product, determined by GC analysis using cyclooctane as an internal standard.
and terminal OH groups in unprotected polymers) and reacts with other impurities within the polyether, and 10 mol% perchlorocatechol, the gram scale degradation of αÀ Bu/ωÀ OH pEO (PE1À H, n max = 13) was achieved without purification of the used ingredients and in the presence of air and moisture (Scheme 4).Continuous removal of the product by trap-to-trap condensation was successfully managed.The conversion of 35 % after four hours of reaction time is similar to the yields obtained when using neat PE1À H (αÀ Bu/ωÀ OH pEO, n max = 13) under entirely inert conditions, which underlines the feasibility of this process on practical scale.Nevertheless, adding a highboiling solvent should be considered to facilitate stirring and increasing the yield, as already mentioned above.

Mechanistic Considerations
Based on all the above-mentioned experimental observations and our calculations on the degradation of diglyme, [7d] a refined catalytic cycle for the depolymerization was developed (Scheme 5).The bidentate cis-coordination of the polyether is identified as the resting state A. Only in the mono-adduct B, the intramolecular S N 2 type attack of the oxygen atom at the polarized α-carbon of the polyether can take place to form an oxonium ion and an alkoxy-silicate C. The subsequent S N 2 attack of the alkoxylate bound to silicon to liberate dioxane is an intermolecular step, explaining increased yields at higher concentrations.At lower concentrations, an intermolecular reaction attack of other nucleophiles, such as other polyethers, can compete, which gives rise to catalytically inactive silicate D.
The anionic silicate D can react with a different electrophile to regenerate the neutral, catalytically active silane E. Therefore, acidic conditions favor the reaction, decrease the number of nucleophiles in solution, and simultaneously enable catalyst regeneration.Competitive binding of the product 1,4-dioxane to the Lewis acid poses another major challenge to the reaction and adversely affects the reaction rate (product inhibition).The higher affinity towards the strongly donating and sterically less demanding dioxane leads to the inactive trans-dioxane adduct of 1. Constant removal of 1,4-dioxane through distillation accelerates the reaction by affecting this equilibrium.

Conclusions
The applicability of the depolymerization process for various pEO derivatives with different end-capping groups and chain lengths is demonstrated within this work.Problematic side reactivities were identified and the catalyst lifetime was increased by adjusting the reaction conditions.By combining SiCl 4 and perchloro-or perbromocatechol, we developed an easy-to-apply in-situ generation of the catalyst.We thereby did not only circumvent the problematics of working under an inert atmosphere but did also improve the reaction yields.On top, this approach also presents a way to efficiently depolymerize unprotected polyethers, thereby broadening the substrate scope of this process, giving rise to large-scale depolymerization without sample preparation and without the need of airfree manipulations.With the obtained experimental signatures, the previously proposed mechanism [7d,9] was refined and supported by experiment, corroborating intermolecular reactions of the intermediately formed oxonium ion with other nucleophiles as the major pathway for catalyst inhibition that can be circumvented by the presence of external electrophiles.We thereby pave the way for further improvements and applications of this reaction.

Scheme 4 .
Scheme 4. Gram scale degradation of αÀ Bu/ωÀ OH pEO (PE1À H, n max = 13) under air through in-situ generation of the catalyst, using SiCl 4 and H 2 cat Cl .

Table 5 .
Degradation of different pEOs by in-situ generation of 1 out of SiCl 4 and perchlorocatechol.[a] [a] Standard conditions: 100 mg polyether with 10 mol% perchlorocatechol and an excess of SiCl 4 in 2.5 mL oDCB, 18-21 h, 115 °C.[b]Referring to 1,4dioxane as the reaction product, determined by GC analysis using cyclooctane as an internal standard.[c] 20 mol% perchlorocatechol was used.