Damage of polyesters by the atmospheric free radical oxidant NO3 •: a product study involving model systems

Summary Manufactured polymer materials are used in increasingly demanding applications, but their lifetime is strongly influenced by environmental conditions. In particular, weathering and ageing leads to dramatic changes in the properties of the polymers, which results in decreased service life and limited usage. Despite the heavy reliance of our society on polymers, the mechanism of their degradation upon exposure to environmental oxidants is barely understood. In this work, model systems of important structural motifs in commercial high-performing polyesters were used to study the reaction with the night-time free radical oxidant NO3 • in the absence and presence of other radical and non-radical oxidants. Identification of the products revealed ‘hot spots’ in polyesters that are particularly vulnerable to attack by NO3 • and insight into the mechanism of oxidative damage by this environmentally important radical. It is suggested that both intermediates as well as products of these reactions are potentially capable of promoting further degradation processes in polyesters under environmental conditions.


Introduction
Polymers are without doubt the most important industrial materials, which have benefited our society in numerous ways. Improving the performance of polymers by making them long lasting and durable is therefore highly desirable not only for the consumer but also for the environment, because expensive waste removal strategies can be avoided or at least reduced. The most important way to improve polymer longevity is a detailed knowledge of the mechanism by which they undergo degradation upon exposure to the environment. It is quite surprising that, despite the heavy reliance of our society on polymeric materials, the chemical mechanism of polymer degradation is by far not fully understood. It has generally been assumed that polymer degradation involves a radical-mediated autoxidation mechanism, which propagates through hydrogen abstraction by an intermediate peroxyl radical ROO • . Although this autoxidation mechanism was initially proposed only for a limited number of polymers that contain activated allylic hydrogen atoms (for example rubber materials) [1][2][3][4][5], it has been universally adapted as general mechanism for polymer degradation. However, recent comprehensive high-level theoretical studies by Coote et al. clearly revealed that polymers possessing only saturated alkyl chains, for example polyesters, will not propagate autoxidation, particularly because the ROO-H bond-dissociation energy (BDE) is usually less than the BDE for unactivated R-H bonds [6].
Polymer surface coatings, which are widely used in the building, automotive and aircraft industries to protect the underlying material from degradation, are commonly highperforming polyesters, which are exposed to significant environmental stress, in particular high temperatures, humidity and UV irradiation. These materials are in direct contact with the troposphere, which is the lowest part of the atmosphere and a highly oxidizing environment. While the oxidation power during daytime can be assigned to the presence of hydroxyl radicals, HO • , the highly electrophilic nitrate radical, NO 3 • , is responsible for the tropospheric transformation processes at night. NO 3 • , which is formed through reaction of the atmospheric pollutants nitrogen dioxide, NO 2 • , with ozone, O 3 (Scheme 1a) [7,8], reacts with organic compounds through various pathways, such as hydrogen abstraction (HAT) and addition to π systems. Most importantly, NO 3 • is one of the strongest free-radical oxidants known [E(NO 3 • /NO 3 − ) = 2.3-2.5 V vs NHE] [9], and recent product studies by us revealed that NO 3 • readily damages aromatic amino acids and pyrimidine nucleosides through an oxidative pathway [10][11][12][13]. Thus, the ease by which model compounds of biologically important macromolecules are attacked by NO 3 • leads inevitably to the question, how resistant synthetic polymers are towards oxidative damage by this environmental free-radical species, in particular in conjunction with other atmospheric radical and non-radical oxidants, which are in direct contact with these materials. Is it possible that such reactions could lead to structural modifications in the polymer that may render the material more susceptible to further damage, for example through photodegradation and/or autoxidation? To our knowledge, the role of environmental free-radical oxidants as mediator of polymer degradation has barely been assessed so far.

Experimental conditions
The compounds that served in this work as models for substructures typically found in surface-coating polyesters are shown in Figure 1. These comprise aromatic moieties, such as phthalic and benzoic esters 1 and 3, respectively, as well as aliphatic diesters of type 2. The esters were used as methylates or neopentylates, where the latter provided a simplified model for diesters of neopentyl glycol, which is the commonly used diol component in such polyesters. All experiments were performed in solution, using two different methods to produce NO 3 • in situ in the presence of the respective substrate 1-3. In experiments where NO 3 • was used in the absence of other radical and non-radical oxidants, NO 3 • was generated at room temperature from cerium(IV) ammonium nitrate (CAN) through photo-induced electron transfer at an irradiation wavelength of λ = 350 nm (Scheme 1b) [11][12][13].
In a typical experiment, the polyester-model substrate and four equivalents of CAN were dissolved in acetonitrile and the solu-   [17].
In all experiments we have used NO 3 • in excess in order to obtain sufficient amounts of material to enable product separation by preparative HPLC using UV detection at wavelengths of λ = 214 and 230 nm and identification by spectroscopic characterization. Details are given in the Experimental section. HPLC chromatograms of the relevant raw reaction mixtures are shown in Supporting Information File 1. Although under natural conditions NO 3 • will be present in much lower concentrations compared to the polyester, our experimental procedure ensured that vulnerable sites in the polyester-model systems could be located with certainty. Due to the repeated purification by HPLC, yields could not be obtained for any of these reactions. However, since this study is aimed at obtaining insight into the nature of the products in order to qualitatively assess how such chemical modifications might affect polymer stability under environmental conditions, exact yields are not required. It is reasonable to assume that only very few damaged sites are initially required in the polyester to promote further degradation on a large scale through chain and other processes.

Reaction of polyester-model compounds 1-3 with NO 3 • from CAN photolysis
Study of the products formed in the reaction of NO 3 • obtained from CAN photolysis provides the opportunity to gain insight into the mechanism of oxidative damage in the absence of other radical and non-radical oxidants. In Scheme 2 the products of the reaction of the polyester-model compounds 1-3 with NO 3 • are shown.
It was interesting to note that no reaction occurred with the isomeric phthalates 1 and the adipic acid derivatives 2. In the case of the former this could be explained by the fact that the aromatic ring is very deactivated due to the two electron-withdrawing ester substituents, so that oxidative electron transfer (ET) by NO 3 • is not possible. Also, NO 3 • induced HAT from the ester, particularly the neopentyl moiety, which is a potential pathway that should most likely occur at the methylene groups α to the ester oxygen atom [18][19][20], is apparently not a feasible pathway. This finding is of potential relevance for the autoxidation mechanism, which proposes hydrogen abstraction by ROO • as propagating step. Thus, although NO 3 • is not only much more reactive than ROO • [7,8], and the BDE for the O 2 NO-H bond is with 427 kJ mol −1 also considerably higher than that of the ROO-H bond (which is about 360 ± 20 kJ mol −1 ) [21], the fact that no hydrogen abstraction from the ester units was observed in the reactions with NO 3 • demonstrates that saturated alkyl groups are quite inert to radical attack.
On the other hand, in the case of neopentyl ester of m-toluic acid (3), which differs from the phthalates by replacement of one ester group by a σ-donating methyl group, reaction with NO 3 • leads to selective oxidative modification of the methyl side chain, while a reaction at the ester moiety was, again, not observed. Analytical HPLC of the raw reaction mixture recorded at λ = 230 nm revealed besides unreacted starting material 3 (which was identified by comparison with an authentic sample but not isolated), nitrate 4, aldehyde 5 and carboxylic acid 6 as most important products (see Supporting Information File 1). Other products were formed in too minor amounts to enable isolation. Further, HPLC analysis revealed that shorter reaction times or a smaller excess of CAN shifted the product ratio towards the nitrate 4 at expense of the higher oxidized products 5 and 6 (data not shown).
The observed side-chain oxidation in 3 by NO 3 • is similar to the outcome of the reaction of thymidine nucleosides with NO 3 • , where oxidative transformation of the methyl substituent in the heterocyclic base occurs exclusively [13]. Concentration-time profiles revealed for the latter reactions that formation of a nitrate occurs first, which is converted to an aldehyde and subsequently into a carboxylic acid [13]. It is not unreasonable to assume that such a step-wise oxidation also occurs in the reaction involving 3, which could be rationalized by the mechanism shown in Scheme 3.
Because of the high oxidation power of NO 3 • , it is proposed that the reaction is initiated by ET at the aromatic ring through an addition-elimination pathway, as has been suggested from time-resolved transient spectroscopic studies for the reaction of NO 3 • with alkylaromatic compounds [22,23]. In the absence of any reactants the resulting radical cation 3 •+ undergoes deprotonation to give benzyl radical 7, in analogy to the mechanism of the NO 3 • -induced oxidation of aromatic amino acids and nucleosides [10][11][12][13]. This mechanism is supported by findings by Steenken et al., who showed that in the reaction of alkylaromatic compounds with NO 3 • ET and deprotonation can occur practically in a concerted fashion in the case of highly electronrich arenes, while in the case of less activated alkylaromatic compounds the intermediate radical cation has a lifetime on the nanosecond time scale [23]. It was further demonstrated that deprotonation of arylradical cations is accelerated by nitrate (NO 3 − ) that is present in the reaction system as 'byproduct' of the oxidation process and as ligand in CAN, and which acts as a Brønsted base [23]. It is important to note that the formation of radical intermediate 7 could principally also occur in one step through NO 3 • -induced benzylic HAT in 3 (not shown).
However, it appears from the outcome of the reactions with the neopentyl derivatives of 1 and 2 that HAT by NO 3 • is not competitive with NO 3 • -induced ET in these systems [7,8,24,25].  [27,28]. The latter is too unreactive to initiate a radical process in this system, which has been confirmed through independent control experiments.
Oxidation of aldehyde 5 to the carboxylic acid 6 under the experimental conditions could by initiated through abstraction of the aldehyde hydrogen atom by NO 3 • [29], followed by trapping of the resulting acyl radical 10 by NO 3 • to give the mixed anhydride 11, which could be hydrolysed to the acid 6 during aqueous work-up and/or purification by HPLC.
The mechanism in Scheme 3 shows that more than one equivalent of NO 3 • is required to produce the observed products 4-6.
Such multiple attacks seem unlikely under environmental conditions, where [NO 3 • ] is low [7,8]. However, from the previous work on NO 3 • -induced oxidative damage of biological molecules, it appears that an already damaged compound is more prone to attack by another NO 3 • than an undamaged substrate [11][12][13].
Reaction of polyester-model compounds The main reaction pathways lead to products possessing a nitroaromatic ring, such as the isomeric mono-nitroaromatic compounds 12a-d, the dinitrated product 13 and two isomeric species 14a,b, which carry both a nitro and a hydroxy substituent (Scheme 4). The nitro compound 15 appears to be the only product that results from oxidative modification of the methyl substituent at the aromatic ring. HPLC analysis indicated that additional products were formed in this reaction (see Supporting Information File 1), but their amounts were too small to enable isolation and identification. The proposed mechanism leading to the various products 12-15 is outlined in Scheme 5.
Similar to the mechanism shown in Scheme 3, initial ET should lead to the radical cation 3 •+ . However, in contrast to the reaction with NO 3 • in isolation, where benzylic deprotonation occurred exclusively, in the presence of excess NO 2 • the radical cation 3 •+ is trapped prior to deprotonation to form the isomeric σ-complexes 16 [30]. The aromatic ring is restored through loss of a proton, which leads to the nitroaromatic products 12a-d.  [31], which is followed by deprotonation to restore the aromatic system.

Conclusion
We have shown for the first time that certain aromatic moieties in commercial polyesters (e.g. alkylated benzoic acid derivatives of type 3) are vulnerable to damage by the environmental free-radical oxidant NO 3 • . The reaction is most likely initiated by ET to give a highly reactive aryl radical cation intermediate 3 •+ , whose fate depends strongly on the reaction conditions. In the absence of radical-trapping agents, in particular NO 2 • , benzylic deprotonation is the exclusive pathway that ultimately leads to oxidative functionalization of the alkyl side chain through formation of nitrates 4, aldehydes 5 and carboxylic acids 6. In this work we have not specifically explored the role of O 2 on the reaction outcome, but our recent studies on the NO 3 • -induced oxidative damage in thymidines showed that any residual O 2 present in the system solely accelerates production of the higher oxidized compounds 5 and 6, while no different products are formed [13]. It is reasonable to expect a similar outcome for the reaction of NO 3 • with 3 in the presence of O 2 .
On the other hand, when the reaction of NO 3 • with 3 is performed in the presence of NO 2 • , benzylic deprotonation in radical cation 3 •+ can hardly compete with trapping of the latter by NO 2 • , which leads to formation of the isomeric nitroaromatic compounds 12a-d as well as the dinitro and hydroxylated products 13 and 14, respectively, that result from further NO 3 • -induced oxidation of 12. An additional, however only minor pathway yields the nitromethylene compound 15, which is formed via benzyl radical 7. Although we have not studied the nature of the reactive intermediates formed in these reactions, it is difficult to rationalize formation of the ring-substituted products 12-14 through a mechanism that involves benzylic HAT by NO 3 • . The reaction must therefore be initiated by oxidation of the aromatic ring, which is in accordance with literature findings [22,23]. In contrast to the high reactivity of the aromatic ring in 3, phthalate-building blocks as well as ester moieties possessing only saturated alkyl chains appear to be inert to attack by NO 3 • through either ET or HAT, respectively, under the various conditions explored. Our observation that NO 3 • -induced HAT in the ester moieties does not occur, although NO 3 • is much more reactive than ROO • and the O 2 NO-H bond is considerably stronger than the ROO-H bond, could be taken as indication that an autoxidation mechanism involving ROO • as chain carrier cannot operate in intact polyesters with saturated alkyl chains, which is in support of the theoretical findings by Coote et al. [6].
None of the various polyester-model compounds explored in this work reacted with NO 2 • and O 3 in isolation. However, this outcome is not unexpected, since the reactivity of NO 2 • is much lower than that of NO 3 • . In particular, the oxidation power of NO 2 • is not sufficient to induce ET in deactivated aromatic compounds [31]. Likewise, although O 3 is a strong oxidant, it does not react via ET transfer. Rapid reactions are only expected for π systems, such as alkenes, which are not present in intact polyester materials (however, it should be noted that these structural motifs may be formed in the polymer through degradation processes).
What are the potential implications of NO 3 • -induced oxidative damage in aromatic building blocks for polyester stability? Although there are no experimental data available yet, it is possible to make some predictions from the outcome of this work, which can be used to guide future studies on polyester stability upon exposure to the environment. It is important to realize that under environmental conditions only few sites of initial damage are required to trigger degradation of the polymer material on a large scale. Identification of the reaction products using simplified model systems enables to obtain some general insight into the mechanism of radical-induced oxidative damage in these materials. Thus, in the reaction of the aromatic ester 3 with NO 3 • it could be speculated that both intermediates as well as products could principally promote further damage in the polymer. For example, the radical cation 3 •+ is itself a highly oxidizing intermediate, which could, when embedded in the polyester matrix, induce an ET cascade across the polymer involving aromatic moieties, where oxidative damage may end up at positions remote from the initial site of attack. The benzyl radical 7 resulting from deprotonation in 3 •+ on the other hand, could be trapped by O 2 and be involved as chain carrier in subsequent transformations that lead to degradation.
Of the various products formed in the reaction of NO 3 • with 3 under the different reaction conditions, in particular the aldehyde 5 and the nitroaromatic species 12-14 are expected to be photochemically active compounds. Exposure of the carbonyl or nitro moieties to UV light leads to photoexcited intermediates, which are strong hydrogen-atom abstractors in Norrishtype II photoreactions [32,33]. In the polymer matrix, where the various polyester chains are tightly packed, both intra-and interstrand reactions are likely to occur, such as photo-induced hydrogen abstractions, which could provide pathways to C-radicals in unactivated alkyl chains that would usually be inert to attack by peroxyl radicals.
To conclude, this work provides strong indications for a number of so far unexplored pathways that could promote degradation of high-performing polyesters under environmental conditions. It is obvious that detailed kinetic data and product analyses from exposure studies involving both simple as well as more complex model systems, including melamine cross-linker moieties, are required (for example from smog chamber experiments), to obtain further insight into the role of environmental free-radical oxidants, such as NO 3 • and HO • , in promoting polyester degradation.

Experimental General procedures
The irradiations were performed under a continuous gas flow (argon) in a Rayonet photochemical reactor (λ = 350 nm). Before the irradiations, residual oxygen was removed from the reaction mixture by bubbling argon through the solution while sonicating. 1