Linda E. Eijsinka,
Andy S. Sardjana,
Esther G. Sinnemaa,
Hugo den Bestena,
Keimpe J. van den Bergb,
Jitte Flapperc,
Rogier van Gemertb,
Ben L. Feringaa and
Wesley R. Browne*a
aStratingh Institute for Chemistry, Faculty of Science and Engineering, University of Groningen, Nijenborgh 4, Groningen, 9747AG, The Netherlands. E-mail: w.r.browne@rug.nl
bAkzo Nobel Car Refinishes N.V., Rijksstraatweg 31, Sassenheim, 2171 AJ, The Netherlands
cAkzo Nobel Decorative Coatings B.V., Rijksstraatweg 31, Sassenheim, 2171 AJ, The Netherlands
First published on 19th January 2022
The curing of bis-methacrylate–styrene resins initiated by the cobalt catalyzed decomposition of cumyl hydroperoxide is monitored at ambient temperatures in situ by EPR and Raman spectroscopy. EPR spectroscopy shows the appearance of organic radicals after ca. 1 h from initiation with an increase in intensity from both polystyrene and methacrylate based radical species over a further ca. 2 h period to reach a maximum spin concentration of ca. 2–3 mM. Alkene conversion to polymer was monitored by Raman spectroscopy in real time in situ with EPR spectroscopy and reveals that the appearance of the radical signals is first observed only as the conversion approaches its maximum extent (70% at room temperature), i.e., the resin reaches a glass-like state. The radicals persist for several months on standing at room temperature. Flash frozen samples (77 K) did not show EPR signals within 1 h of initiation. The nature of the radicals responsible for the EPR spectra observed were explored by DFT methods and isotope labelling experiments (D8–styrene) and correspond to radicals of both methacrylate and polystyrene. Combined temperature dependent EPR and Raman spectroscopy shows that conversion increases rapidly upon heating of a cured sample, reaching full conversion at 80 °C with initially little effect on the EPR spectrum. Over time (i.e. subsequent to reaching full conversion of alkene) there was a small but clear increase in the EPR signal due to the methacrylate based radicals and minor decrease in the signal due to the polystyrene based radicals. The appearance of the radical signals as the reaction reaches completion and their absence in samples flash frozen before polymerization has halted, indicate that the observed radicals are non-propagating. The formation of the radicals due to stress within the samples is excluded. Hence, the observed radicals are a representative of the steady state concentration of radicals present in the resin over the entire timespan of the polymerization. The data indicate that the lack of EPR signals is most likely due to experimental aspects, in particular spin saturation, rather than low steady state concentrations of propagating radicals during polymerization.
Currently, mainly cobalt catalysts are used in the activation of alkyl hydroperoxide initiators in resin curing, as they provide robust coatings and allow for control over curing profiles, in which the onset of polymerization is delayed sufficiently to allow time to apply the coatings. Concerns with regard to cobalt8–10 have prompted efforts to seek replacements for these catalysts focusing primarily on manganese and iron-based catalysts.11,12 ‘Drop-in’ replacements, where only the catalyst is changed, have met with only limited success, which is unsurprising given the complex kinetics of the radical based copolymerization that leads to curing. Hence, understanding of the mechanism of polymerization and the impact of catalyst replacement on radical generation is essential.
The curing of these resins by radical polymerization can be divided into four stages: inhibition, solution polymerization, auto-acceleration (Trommsdorff effect)13 and glassy state (Scheme 2). The inhibition stage is important to allow for mixing of the hydroperoxide with the resin before application on a surface and is controlled typically with the use of additives such as 2,6-tert-butyl-4-methylphenol (BHT).14 Once the radical polymerization is initiated the solution's viscosity increases due to formation of crosslinked polymer chains and eventually the reactive diluent (styrene and remaining methacrylate monomer) is gelated by the polymer. It is at this point that the kinetics of polymerization change and the onset of auto-acceleration is observed with rapid hardening of the coating, forming a mixed polymer that depends on monomer reactivity ratios15,16 (Table 1). The well-defined curing profile required of these resins (Scheme 2) challenges the development of catalysts that are suitable alternatives to those based on cobalt. Since the catalyst's role is primarily in initiation by decomposition of alkyl hydroperoxides, non-linear effects are to be expected and hence a detailed understanding of the radical polymerization and the radicals involved is essential.
Conditions | r1 (styrene) | r2 (BADGE-MA) | r2 (MMA) |
---|---|---|---|
30 °C, cobalt soap, CumOOH20 | 0.45 ± 0.00 | 1.54 ± 0.02 | |
90 °C, CumOOH20 | 0.47 ± 0.01 | 0.61 ± 0.1 | |
25 °C, cobalt soap, DMA, MEKP21 | 0.19 | 0.35 | |
60 °C, BPO21 | 0.19 | 0.50 | |
90 °C, BPO21 | 0.22 | 0.82 | |
140 °C, BPO21 | 0.23 | 0.88 | |
80 °C, BPO38 | 0.43 ± 0.03 | 0.41 ± 0.05 | |
35 °C (ref. 16) | 0.50 | 0.44 | |
15.2 °C, PLP39 | 0.523 | 0.421 |
Vibrational spectroscopy is particularly suited to determine changes in the concentration of reactive monomers over time as, although the resin changes from a mixture of small molecules to a polymer network, the fingerprint nature of these techniques allow for all major reaction compounds to be determined (semi)quantitatively.17–19 For example, cobalt/hydroperoxide initiated polymerizations have been studied with FTIR spectroscopy to study the effect of temperature (25 to 140 °C) on monomer reactivity and coating properties.7,20–22
The radical nature of the polymerization and the timescales over which curing takes place (e.g., 1 to 4 h) in principle lends it to be studied by EPR spectroscopy, as demonstrated in several studies of photochemical, thermal and redox-initiated systems. Photoinitiated radical reactions have seen most attention in regard to studies with EPR spectroscopy, aided by the possibility that optical access to the EPR cavity allows for at will initiation of polymerization and control over the rate of initiation (decomposition of photoinitiator) by the intensity of the actinic pulse of light used.
Photoinitiated radical polymerizations with phosphine oxides are used widely in dental resins, which typically consist of methacrylate based crosslinking monomers only. The stable methacrylic radicals observed in cured dental resins are typically a mixture of an allylic and a tertiary carbon radical (Scheme 1C: MA1 and MA2, 9 and 5-line species, respectively).23,24 In several studies it was noted that irradiation of the photoinitiator present in these resins results in an increase in the total concentration of radical over time,24,25 and the ratio between the allylic and tertiary carbon radical changes also.
In resins based on mixtures of methacrylate crosslinking monomers and styrene, the contribution of styryl based radicals to the EPR spectrum obtained after curing is not linearly dependent on the initial concentration of styrene and the total radical concentration in the sample decreases with increase in the fraction of styrene.25 A similar broad 4-line signal has been observed after radiooxidation of polystyrene and in heat/peroxide26 and photoinitiated styrene homopolymerization.27 Close inspection of the spectra obtained in photopolymerisation, with high intensity irradiation, of styrene, containing only the photoiniator, using long acquisition times28 or SP–PLP–EPR (single pulse – pulsed laser polymerization – EPR) spectroscopy29 showed the presence of an 18+-line spectrum assigned to the propagating styryl radical, whereas the 4-line signal was assigned to a mixture of tertiary benzyl and cyclohexadiene radicals (Scheme 1C: St1 and St2 resp.).26 Parallel monitoring of conversion by FT-NIR spectroscopy and radical concentration by EPR spectroscopy shows that the radical concentration increases as conversion increases, after which the concentration does not change.25 If the irradiation is halted prior to reaching the maximum extent of conversion, however, the radical concentration drops rapidly.30
In contrast to photoinitiated polymerizations, metal/hydroperoxide initiated polymerizations can take up to several hours, allowing for greater overall relative time resolution in spectroscopic studies. A key question arising, however, is to what extent the signals observed by EPR spectroscopy correspond to radicals participating in polymerization and especially the time dependence of spectral changes considering the specific time course of the polymerization (Scheme 2, e.g., lag period, autoacceleration and formation of glass-like state etc.). In this contribution we follow the curing of methacrylate styrene resins by cobalt-catalyzed alkyl hydroperoxide decomposition with EPR and in situ Raman spectroscopy. The latter technique allows for monomer conversion to be determined in real time and allows direct correspondence to be made between the changes observed by EPR spectroscopy, and reveals that the observed radicals appear only towards the end of the curing process as polymerization slows and even after it has halted. These radicals persist essentially indefinitely even with further curing at elevated temperatures and hence their actual involvement in polymerization is unlikely. However, their appearance is indicative of the radical concentration during polymerization, estimated to be in the millimolar range, and that the absence of signals during polymerization is most likely due to slow relaxation and hence saturation, rather than a low steady state concentration of observable radicals.
Fig. 1 X-Band EPR spectrum of cured resin at room temperature (20 °C). Note that the shape of the spectrum is highly dependent on microwave power (see Fig. S7†). Parameters: power 0.001 mW (50 dB), gain 70 dB, sweep width 200 G, sweep time 120 s, 10 accumulations, 1 G modulation amplitude, frequency 9.661781 GHz. |
Fig. 2 EPR spectra (77 K) of samples flash-frozen at various times following addition of cumyl hydroperoxide to the reaction mixture. Parameters: power 0.0001 mW (60 dB), gain 60 dB, sweep width 300 G, sweep time 60 s, 5 accumulations, 2 G modulation amplitude. A figure with x-axis in field (mT) is available in ESI (Fig. S8†). |
The EPR spectrum of the resin was recorded at room temperature over time from within ca. 1 min of addition of alkyl hydroperoxide to the resin containing catalyst. Within the first minutes, occasionally weak signals corresponding to the BHT radical were observed (Fig. S9†). Thereafter the spectrum was silent for typically 1 h, consistent with the absence of signals in the flash frozen samples. A weak broad signal appears after the first hour, the intensity of which increases initially rapidly and then more slowly over several hours to reach a maximum (Fig. 3). The growth of the signals across the spectrum was monotonic, and in situ spectra obtained during room temperature curing and samples flash-frozen at particular times were essentially identical (Fig. S10†). The spectrum does not undergo significant further change over several months standing at ambient temperature.
With D8–styrene (Fig. 3), the EPR spectrum shows the same time dependent evolution. The difference between an EPR spectrum of a perprotio and perduetero compound is a decrease in magnitude of hyperfine coupling due to the lower gyromagnetic ratio of the deuteron (0.411 × 108 s−1 T−1 vs. 2.67 × 108 s−1 T−1) and increased number of lines due to the increase in nuclear spin quantum number (1 vs. ½). Furthermore, the spectra with D8–styrene are potentially less affected by saturation due to nuclear quadrupole relaxation. The major part of the signal with D8–styrene does not show discernible hyperfine coupling and appears as a single broad line, which then dominates the spectrum, with the second signal unaffected (assigned to a methacrylate based radical, vide infra). The changes upon deuteration allow for assignment of the two signals, with the dominant signal due to a radical derived from styrene/polystyrene and the weaker signal with well resolved hyperfine coupling due to a BADGE-MA derived radical.
Fig. 5 Observed (blue) and fitted (orange) EPR spectra for BADGE-MA/ethylbenzene (left), BADGE-MA/H8–styrene (center) and BADGE-MA/D8–styrene (right). See ESI† for fit parameters. |
Three distinct structures were used for calculations: a BADGE-MA–styrene˙–BADGE-MA, a styrene–styrene˙–styrene and a styrene–styryl˙ molecule (˙ indicates location of the radical, Fig. 6). The spectra calculated for a terminal styryl radical show a high degree of variation in hyperfine coupling patterns and a weighted sum cannot be obtained that matches the experimental spectrum well (Fig. 6C). DFT calculations are consistent with the 18+-line signal observed by Buback et al.,29 including the complete collapse of hyperfine coupling upon deuteration. Interestingly, most (9/10) conformers of the BADGE-MA–styrene˙–BADGE-MA molecule show an isotropic signal without resolvable hyperfine coupling. Only one conformer shows a spectrum that is comparable to the H8–styrene EPR spectra. Upon deuteration coupling to protons from the BADGE-MA unit is retained (Fig. 6A, conformer 2).
Fig. 6 DFT calculations of (A) BADGE-MA–styrene˙–BADGE-MA, (B) a styrene–styrene˙–styrene and (C) a styrene–styryl˙ with structures used for DFT calculations (left) and the calculated spectra for 10 conformers with H8–styrene (middle) and D8–styrene (right) generated with EasySpin33 using DFT calculated parameters. The red part in A shows the part of the molecule derived from BADGE-MA which is unaffected by deuteration in the experimental spectra. Note that the spectra in C are stacked for clarity. |
In the case of styrene–styrene˙–styrene, most conformers show features similar to the experimental spectra and deuteration causes a complete collapse of the signal, as is also observed experimentally (Fig. 6B). Thus, the observed styrenic radical is referred to as a polystyrene based species. Although in-chain radical species are not observed frequently in styrene and methacrylate polymerization, it should be noted that similar EPR spectra are observed with divinyl benzene, used as a ‘difunctional styrene’,25 as well as after γ-irradiation of polystyrene.26 The rapid change in viscosity during the auto-acceleration stage restricts diffusion and may lead to reactivity not observed in non-crosslinking styrene/methacrylate polymerizations.
Fig. 7 Raman spectra obtained during room temperature curing (polymerization) of BADGE-MA with H8–styrene (A, top) and D8–styrene (B, bottom). Expansions (lower spectra, right) show the disappearance of the CC bonds of H8–styrene (1630 cm−1), BADGE-MA (1637 cm−1) and D8–styrene (1577 cm−1). The change in integral over time is shown (lower spectra, left). Note that the band at 1600 cm−1 (H8–styrene) and 1536 cm−1 (D8–styrene), arising from the styrene CC ring stretch is decreased due to a loss in conjugation. Spectroscopic assignment of observed bands can be found in ESI (Fig. S14 and Tables S5–S7†). |
The expected inhibition and acceleration phases of the curing are apparent in the time dependent Raman spectral data and it is clear that curing has reached completion within 1 h. This result contrasts with the time dependence of the EPR spectrum of a curing resin, which does not show the appearance of signals from polymer based radicals in the first hour following initiation. The difference in time dependence even between samples of the same resin mixture monitored with different techniques may, however, be due to the marked sensitivity of the polymerization to temperature (Fig. S16†) and therefore simultaneous recording of EPR and Raman spectra of the same sample is required for direct comparison of apparent reaction rates. Optical access to the cavity of the EPR spectrometer allows for combined in situ operando EPR and Raman spectroscopy (Fig. 8), which confirms that the sudden appearance of the EPR signals after ca. 1 h coincides with the rapid decrease in reaction rate apparent from the Raman spectra.
Fig. 8 Simultaneous in situ Raman (left) and EPR (center) spectroscopy during the room temperature reaction of BADGE-MA with the non-reactive diluent ethylbenzene. The EPR signal (orange) emerges as the alkene conversion (Raman, blue) reaches its limit (right). Integrations over time (EPR and Raman spectra) were a rolling average over 10 spectra. EPR parameters: 0.01 mW (40 dB), gain 50 dB, sweep width 300 G, sweep time 45 s, 1 accumulation, 1 G modulation amplitude. Fig. S17:† graph with EPR spectra in mT. |
The EPR spectra recorded in the present study for cured styrene/methacrylate resins are comparable with those reported by Scott et al. in the photopolymerization of similar resin compositions,25 as well as those reported by Dondi et al. in the post-cure irradiation of polystyrene.26 The spectra are distinct from those obtained by Yamada et al. and Buback et al. in their efforts to obtain the spectrum of the propagating α-styryl radical.28,29
The radicals appear relatively suddenly towards the end of the autoacceleration (Trommsdorff) phase. The current understanding of the Trommsdorff effect is that the initiation and propagation rates are unaffected during this phase, while the increase in chain length and viscosity reduces the diffusion and thus the termination rate.14 The decrease in termination events leads to a net increase in radical concentration during this phase. Iodometric studies of the concentration of oxidant in the polymerizing resin show that hydroperoxide consumption is slow beyond a rapid burst of activity at the start. Following the initial burst in decomposition, the steady and slower decomposition of the initiator should lead to a relatively constant rate of generation and termination of propagating radicals. In the present study the EPR signals due to radicals do not appear during the autoacceleration phase. Instead, it is only towards the end of this phase that radical signals due to in-chain polystyrene based and terminal BADGE-MA radicals appear and continue to increase in concentration well after polymerization had halted. The final concentration of radicals is ca. 2 mM, and hence an equal concentration of radicals capable of generating them is necessary. It is of note that although the methacrylate radical EPR signal is relatively unaffected by acquisition power (i.e., saturation effects) that of the styryl radical is highly sensitive to spectrometer power. It is therefore conceivable that the radicals responsible for propagation are present but not observed by EPR spectroscopy due to saturation, even with maximum power attenuation. The absence of signals in flash frozen samples at 77 K obtained during the acceleration phase indicates that broadening due to rapid relaxation is unlikely to play a role either. Furthermore, in the polymerization of the methacrylate based crosslinking monomer (BADGE-MA), in the absence of styrene, a similar conversion dependence for the appearance of non-propagating radicals is observed. Since this resin mixture does not form a hard glassy polymer, and indeed both styrene/BADGE-MA and ethylbenzene/BADGE-MA resins are flexible, stress-induced radical formation can be excluded. An alternative explanation for the absence of signals from propagating radicals is that they are subject to rapid relaxation by paramagnetic ions such as Co(II). UV-vis absorption spectroscopy and EPR spectroscopy indicate that the resting state of the cobalt catalyst is Co(III) and indeed the spectra with other catalysts, such as Mn(II) soaps, are identical in structure.
Heating a sample cured at room temperature to elevated temperatures (80 or 90 °C) induces a rapid increase in conversion. The observed radical concentration does not change in the early stages of the process and hence the observed styrene based species is thus not a resting state of the propagating radical that is responsible for the continuation of polymerization as the temperature is raised. After approximately 30 min and well after full conversion of alkene is observed, the concentration of methacrylate based radicals increases in samples held at higher temperatures. The total concentration of these radicals however remains lower than the concentration of styrene based radicals.
The sudden increase in observed radical concentration as the polymerization halts and then the continued increase in radical concentration for some time after polymerization has stopped is inconsistent with these radicals forming due to a late burst of initiator activation. Indeed, initiator decomposition is slow after an initial rapid burst in hydroperoxide consumption. Furthermore, the observed radicals are of a polystyrenic and methacrylic nature and do not arise from the initiator directly during polymerisation but may form either by reaction with propagating radicals or initiator radicals once the glassy state is reached and diffusion of monomer is no longer possible. The late appearance and growth of the signals as well as the obvious microwave power dependence of the polystyrene based radical signal suggests that propagating radicals are not responsible for the signals observed. Generation by mechanical stress developed within the cured resin was excluded based on experiments forming a non-glassy polymer. Similarly, Co(II) 2-ethylhexanoate does not influence the EPR spectra as the resting state of the catalyst is Co(III). Indeed, identical spectra are observed with other catalysts such as Mn(II) soaps. The absence of signals assignable to propagating radicals in the period leading up to and including the auto-acceleration phase was unexpected. The limit of detection for organic radicals under the conditions used was 0.1 to 0.5 μM and hence it is possible that concentration of propagating radicals was below this range. However, the rapid increase in radical concentration to 2 mM once conversion had ceased (i.e. in the glassy state) and the continued conversion seen when cured samples were heated to 80 °C, leads to the conclusion that radicals are present at higher concentrations but are simply not observable. Slow relaxation and saturation even at low power affects strongly the styrenic radical signal (but not the methacrylate based radical) and hence it is not unreasonable that the propagating radicals are affected by saturation even more so.
Efforts to replace cobalt from the initiator system rely on an understanding of the factors that control reaction kinetics, for example the use of inhibitors. The present study indicates that the radical concentrations during the polymerization are likely to be quite high and hence small variations due to the relative concentrations of propagating radicals may affect overall chain lengths in the cured resin, and hence performance of the final coating. Differences such as chain length are not apparent from Raman spectroscopy as this technique provides only a measure of conversion. However, although EPR spectroscopy does not reveal the concentration of propagating radicals directly, it may prove useful to estimate the total radical concentration during curing from the final observed spin concentration of the in-chain radicals. A change of initiator efficiency by a change in hydroperoxide decomposition catalyst will affect the concentration of radicals. This difference may in turn affect chain length and thus the physical properties of the material. Hence the focus on replacements for cobalt initiators may need to be on the pathways for peroxide decomposition rather than on the reaction kinetics themselves.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ra09386j |
‡ As a simple model H8–styrene was fitted with 2 interacting nuclei, D8–styrene was fitted as a singlet. |
§ The overall reaction time for the polymerization with D8–styrene is longer than with H8–styrene. The difference is likely due to the higher concentration of inhibitor present in this reaction mixture rather than a kinetic isotope effect since the inhibitor was removed from H8–styrene but not from D8–styrene. |
This journal is © The Royal Society of Chemistry 2022 |