Utku
Yolsal
a,
Thomas J.
Neal
a,
James A.
Richards
b,
John R.
Royer
b and
Jennifer A.
Garden
*a
aEaStCHEM School of Chemistry, University of Edinburgh, Joseph Black Building, David Brewster Road, Edinburgh, EH9 3FJ, Scotland, UK. E-mail: j.garden@ed.ac.uk
bSchool of Physics and Astronomy, University of Edinburgh, King's Buildings, Peter Guthrie Tait Road, Edinburgh, EH9 3FD, Scotland, UK
First published on 11th March 2024
Polymers with high molecular weights have superior properties, such as enhanced impact and chemical resistance. While these properties can be achieved by converting a thermoplastic into a thermoset, this can prevent polymers from further processing as thermosets cannot be melted and moulded without damaging their internal structure. Therefore, increasing the molecular weight of a polymer without losing the ability to process it is of utmost importance. Polyethylene (PE), the most commonly produced plastic in the world, is comprised of strong C–C and C–H bonds, which makes its controlled chain extension challenging to achieve. Herein, we report a novel, solution-based method for the modification of PE chains using commercially available, low-cost organic peroxides in solvents such as dichlorobenzene and tert-butylbenzene. To the best of our knowledge, this is the first solution-based methodology for PE modification through the incorporation of long chain PE branches by using organic peroxides. The effects of the modification reactions were extensively investigated using rheology, differential scanning calorimetry, small/wide angle X-ray scattering, size exclusion chromatography and NMR spectroscopy, and model studies were performed with n-dodecane to confirm the formation of branched moieties. The enhanced mechanical properties of the materials were demonstrated using rheology, where the modified polymers show significantly increased stiffness and higher viscosities. This is attributed to reactions between the PE chains to form branched structures, thus increasing both the molecular weight of the feedstock and the number of entanglements within the polymer microstructure. This methodology enables the properties of PE to be tailored, providing a shortcut for the development of new PE grades and formulations as its applications continue to grow in developing technologies such as 3D printing, artificial joints and soft robotics.
Polyethylene (PE) accounts for one third of the entire plastics markets,10 and has many applications as a high performance speciality polymer, including developing technologies such as implantable medical devices and advanced composites.11–13 The properties of PE can be modified during its synthesis, through careful choice of the initiator, and PE can also be modified post-synthesis, traditionally through reactive extrusion at high temperatures to achieve grafting on the PE surface.14–16 For example, the grafting of maleic anhydride can improve the polymer's adhesive properties.17,18 As the PE surface is made up of strong C–C and C–H bonds, it is usually necessary to use high-energy species to modify polyethylene, and a number of industrial processes already use radical-releasing peroxides.19,20 The organic peroxides employed are specifically chosen to be stable at room temperature and decompose when heated to PE extrusion temperatures, rapidly releasing radicals that react with the usually inert PE surface to form macroradicals. The macroradicals can subsequently undergo a variety of different reactions to modify the polymer chains (Scheme 1).21,22
![]() | ||
Scheme 1 A simplified reaction map for the post-polymerisation modification of PE from a generated macroradical. |
The reaction of PE with organic peroxides often leads to the formation of long chain branching (LCB) on the polymer chains. In LCB the side chains are over twice the molecular entanglement weight, and both the backbone and sidechains are entangled.23 Therefore, even a small component of LCB polymers can have a significant effect on the mechanical properties and improve the end properties.24–27 This arises from an increase in extensional strain hardening, e.g., preventing film rupture, while maintaining a low high-shear viscosity.28 Melt-state PE modifications with peroxides are already used in some commercial processes, yet this can generate polymers with gel fractions due to the poor dispersion of the organic peroxides.24,29 This can be attributed to poor mixing of the organic peroxide in the viscous polymer melt, resulting in “pockets” of radicals where certain parts of the polymer become highly cross-linked whilst others are sparsely modified. Solution-state modifications of PE have the potential to tackle some of the issues associated with the melt-state PE/peroxide modifications, by ensuring homogeneous LCB without introducing crosslinks into the polymer microstructure. Yet compared to the melt-state, solution-state modifications of PE are significantly underexplored. This is likely due to the challenge of modifying the strong C–C and C–H bonds of PE whilst avoiding competitive reactions with the solvent; this is especially difficult considering the solubility challenge associated with the crystallinity of high-density PE (HDPE).
To the best of our knowledge, solution-based peroxide modifications of PE have focussed on grafting,30,31 and polymer enhancements through LCB have not been explored. Herein, we report an unprecedented solution-state peroxide modification method for PE that can be used to enhance the material properties through LCB, and can alter the feedstock grade from an injection-moulding grade into a blow-moulding grade material.
Initially, a range of different conditions were tested to find the optimal conditions for the modification reactions. As HDPE is a highly crystalline polymer, heat was essential to dissolve the polymer in an organic solvent, which disrupts the crystalline domains. HDPE is commonly dissolved in high boiling point aromatic hydrocarbons or in halogenated solvents.20 As organic peroxides decompose to release radicals, we targeted solvents that would not compete with polyethylene C–H groups for hydrogen abstraction. This was particularly important as polymer concentrations of <10 w/v% were used, due to the viscosity of HDPE solutions. Dichlorobenzene (DCB) was selected as the main solvent for the initial investigations due to the decreased stability of sp2 radicals vs. sp3 radicals, which was expected to disfavour C–H abstraction from the solvent molecules compared to the CH2 units on PE. A range of other solvents were also studied to investigate compatibility with the organic peroxide modifications, including tert-butylbenzene (TBB), n-nonane and anisole.
Dilauroyl peroxide (DLP) and benzoyl peroxide (BPO) were both chosen as suitable organic peroxides. Dicumyl peroxide (DCP) and 2,5-bis(tert-butylperoxy)-2,5-dimethylhexane (BDH) are typically used for melt-phase PE modifications because they decompose rapidly at the extrusion temperatures (170–190 °C) with short half-times (<2 min).38,39 However, it was hypothesised that the reduced viscosity in solvent compared to the melt state would enable lower reaction temperatures to be used, and thus reaction temperatures in the range of 125–135 °C were selected as being well below the solvent boiling points. Therefore, DLP and BPO were selected for this temperature range because, at 125 °C, the half-life values were predicted to be 0.45 and 1.9 min in chlorobenzene, respectively.40
Following dissolution of HDPE in DCB at 125 °C, a solution of the peroxide in DCB was added to the polymer solution over a period of time (usually 20 min) and the reaction solution was stirred for a further 20 min (Scheme 2 and Table 1). All polymers were recovered by precipitation in hexane, and were filtered and dried before further characterisation (refer to Experimental for further details). A range of different conditions were tested to understand how each variable affects the modification reactions. Initial experiments were performed using DLP, and the peroxide loading was varied from 0 to 6 wt% relative to the polymer weight. All polymers remained fully soluble during and after the modification reactions, indicating that no crosslinks were introduced.
Exp. | Peroxide | Peroxide Loading (wt%) | Solvent | Peroxide addition duration (min) |
---|---|---|---|---|
a All reactions were performed at 125 °C, at 7.5 w/v% polymer concentration, under inert atmosphere unless otherwise stated. The experiment time was 20 min after the peroxide addition. b The reaction was performed at 135 °C. c The reaction was performed under air, instead of inert atmosphere. d The polymer concentration was 5 w/v%. | ||||
M1 | Control | 0 | DCB | 20 |
M2 | DLP | 1.5 | DCB | 20 |
M3 | DLP | 3.0 | DCB | 20 |
M4 | DLP | 4.5 | DCB | 20 |
M5 | DLP | 6.0 | DCB | 20 |
M6 | DLP | 4.5 | TBB | 20 |
M7 | DLP | 4.5 | n-Nonane | 20 |
M8 | DLP | 4.5 | Anisole | 20 |
M9 | DLP | 3.0 | DCB | 6 |
M10 | DLP | 3.0 | DCB | 1 |
M11 | DLP | 3.0 | DCB | 0.02 |
M12 | DLPb | 4.5 | DCB | 20 |
M13 | Dried BPO | 1.0 | DCB | 20 |
M14 | Dried BPO | 2.0 | DCB | 20 |
M15 | Dried BPO | 3.0 | DCB | 20 |
M16 | Dried BPO | 3.0 | TBB | 20 |
M17 | Wet BPO | 3.0 | DCB | 20 |
M18 | Wet BPO | 3.0 | TBB | 20 |
M19 | Wet BPOc,d | 0.5 | DCB | 20 |
M20 | Wet BPOd | 0.5 | DCB | 20 |
To provide a benchmark for the modification reactions, the rheological properties of IMPE were characterised. The tests were performed using a parallel-plate geometry, which restricted the sample surface exposed to air, and the temperature was set to 190 °C to mimic the commercial processing temperature of HDPE. IMPE had a melt flow index (MFI) of 4 g per 10 min (190 °C, @2.16 kg). For comparative purposes, a blow-moulding grade HDPE (BMPE) was tested and had an MFI of 0.2 g per 10 min (190 °C, @2.16 kg). Altering the PE grade from injection to blow-moulding grade would validate the solution-state modification methodology for increasing the molecular weight of the starting feedstock, and thus the rheological properties of BMPE were also characterised. At 190 °C, G′′ dominates over G′ at all the tested frequencies for IMPE, with G′ having a steeper slope than G′′ (0.9 vs. 0.75). This is indicative of the terminal regime,42 meaning that there is enough time for polymer chains to relax and diffuse through any entanglements (reptate) during the test timescales (Fig. 1 and S5–7†). However, these slopes are much weaker than expected for a linear, monodisperse polymer (2 vs. 1).43 Similar deviations have been observed for commercial PE samples, indicating a broad range of relaxation timescales from polydispersity and/or non-linear polymer architecture.28 However, a sharply different behaviour was observed with BMPE with G′ > G′′ at high frequencies and a crossover point (G′ = G′′) at 10 rad/s, which corresponds to a slower relaxation time of 0.5 seconds (also Fig. S8–11†). In addition, there was a significant increase in G′ values at low frequencies (0.1–1 rad/s), where it ranged from 600 to 2000 Pa for IMPE, compared to a range of 4000–10000 Pa for BMPE. The measurement of G′ is considered as one of the most reliable methods to determine the elasticity of a polymer melt.44 This indicates that there are significantly more entanglements with the blow-moulding grade polymer.
![]() | ||
Fig. 1 Frequency sweep experiment results obtained from the melt rheology of the model polymers and the modified samples M1–5. |
As the dosing of DLP was increased from 0 to 6 wt%, both G′ and complex viscosity values of the modified IMPE samples increased significantly at low frequencies (Fig. 2 and S12–30†). In addition to 190 °C, frequency sweep tests were also performed at 140, 160 and 180 °C for M5 (Fig. S30†), to evaluate the material properties under various processing conditions. These tests were run over 5 hours and confirmed the sample stability during the rheological measurements. This was in-line with literature where it was shown that rapid oxidative changes occur at much higher temperatures (230 °C).45 At 4.5 wt% DLP loading, the polymer viscosity profile is similar to BMPE, with a crossover frequency also observed in its frequency sweep tests. In addition, higher shear-thinning behaviours were observed with increasing DLP loadings, suggesting the presence of LCB.46 LCB is thought to particularly increase the low-shear viscosity as entangled branches dramatically slow down the reptation of the backbone. As frequency (shear rate) increases, polymer chains start to disentangle resulting in a shear thinning behaviour. However, higher molecular weight linear polymers are also more entangled and slower reptating, as such the two polymer properties are difficult to initially distinguish from oscillatory rheology data. Therefore, empirical rules from derived quantities have been developed based upon tests on well-characterised samples and extended to industrial polymers.28 The qualitative long chain branching index (LCBI = 1 − δ/90°) uses the phase angle at a given complex modulus, e.g., G* = 2 × 104 Pa < 105 Pa. With increasing DLP loading, the LCBI increases from 0.2 in injection-moulding grade PE, to 0.31 at 1.5 wt% peroxide loading, and through 0.41 at 3.0 wt% and 0.51 at 4.5 wt% to 0.58 at 6.0 wt%, where a higher LCBI indicates a greater degree of LCB, associated with extensional strain hardening. The peroxide-triggered polymer modification reactions result in random branches in a polymer sample, and this increases both the molecular weight of polymer chains and the entanglements associated with LCB. Promisingly, excellent control was also achieved over the melt flow properties of the modified samples (Fig. 2). While only a transition from injection to blow-moulding grade was desired, the results also demonstrated that the final flow properties of the polymers can be adjusted by simply altering the peroxide loading. This means that a feedstocks’ grade can be tailored for a particular application. It is worth noting that a control reaction performed with 0% peroxide loading also increased the complex viscosity profile of the starting feedstock, which was attributed to the removal of some lower molecular weight fractions from the sample during the precipitation steps.
![]() | ||
Fig. 2 Rheological characterisation of the model injection and blow-moulding grade polymers and the samples M1–5. |
The modified samples were further characterised using high temperature size exclusion chromatography (SEC) (Fig. 3 and Table 2), which was required due to the limited solubility of PE at room temperature. While the calculated number-average molecular weight (Mn) values remain relatively similar between the feedstock, control and the DLP modified samples, a trend can be observed with the Mw and Mp values, which correlate to the increased peroxide loadings. As the peroxide loadings increase from 3.0 to 4.5 and 6.0 wt% of DLP, the Mw values increase from 339 to 349 and 365 kg mol−1, respectively. In addition, investigation of the shape of the size exclusion traces shows the presence of higher molecular weight fronts at these peroxide loadings. While the Mw values of the control (M1) and 1.5 wt% DLP samples (M2) appear similar, this may be due to the broad molecular weight distributions which can mask small changes in polymer sizes. Although the dispersity values are broad, ranging from 3.2 to 3.5, this is not uncommon for industrial grade polymers such as IMPE, and broad dispersities were observed for the control as well as the modified samples. The molecular weight of the highest peak (Mp) also shifts towards shorter retention times with increased DLP loading, confirming the increase in the polymer molecular weights. Taken together, the Mw and Mp values show a clear trend of chain extension with increased peroxide loading, and provide further evidence that the solution-state reactions are a successful method of modifying PE. This is attributed to the formation of macroradicals that can react with each other to form higher molecular weight polymer chains, leading to a more significant change in Mw than Mn, as higher molecular weight chains have a greater contribution to Mw, giving a greater enhancement of the polymer mechanical properties.47,48
![]() | ||
Fig. 3 SEC viscometer traces recorded for M1–5, where VS-DP refers to the viscometer-differential pressure. |
Polymer | Peroxide, loading (wt%) | M p (g mol−1) | M n (g mol−1) | M w (g mol−1) | Đ |
---|---|---|---|---|---|
M1 | Control, 0 | 199![]() |
100![]() |
318![]() |
3.2 |
M2 | DLP, 1.5 | 202![]() |
87![]() |
307![]() |
3.5 |
M3 | DLP, 3.0 | 202![]() |
100![]() |
339![]() |
3.4 |
M4 | DLP, 4.5 | 211![]() |
108![]() |
349![]() |
3.2 |
M5 | DLP, 6.0 | 218![]() |
109![]() |
365![]() |
3.4 |
DSC analysis was also used to probe the changes in the modified polymer samples relative to the feedstock. HDPE is highly crystalline, due to the predominance of ethylene repeat units with minimal short chain branching and no LCB. Accordingly, M1 (control) was found to have a crystallinity of 67.8% (refer to the ESI for additional details†). This value was determined by calculating the enthalpy change associated with melting the polymer and then by dividing this by the theoretical enthalpy of melting for a 100% crystalline PE sample (290 J g−1).49 As the peroxide loading increased, the crystallinity of the resulting modified polymer samples decreased in a step-wise fashion, from 67.8% (M1) to 63.9% (M5, 6 wt% DLP). This trend provides further support for the modification of the starting feedstock and also hints at the formation of long chain branches, which are known to affect the packing of PE chains and thus the crystallinity.50,51 Taken together, the data from rheology, SEC and DSC analysis show that the solution-state modification of PE using DLP is effective and leads to an increase in polymer molecular weights. This may be due to LCB, as supported by the stronger G′ and complex viscosity values at low frequencies as well as the decreased polymer crystallinities.
Additionally, small-angle X-ray scattering (SAXS) and wide-angle X-ray scattering (WAXS) were performed on samples M1–M5 to investigate the polymer structure and further probe the effect of DLP. In all cases, sharp peaks related to the HDPE crystal structure were observed in the WAXS pattern (Fig. S109†). The observed crystalline peaks at scattering vector (q) values of ca. 15.2 nm−1 (0.415 nm), 16.8 nm−1 (0.374 nm), 21.2 nm−1 (0.296 nm) and 25.6 nm−1 (0.245 nm) correspond to the 110, 200, 210 and 020 reflections of an orthorhombic crystalline structure characteristic of HDPE.52–54 Additionally, a broad amorphous peak at 14.4 nm−1 (0.436 nm) is present, highlighting the semi-crystalline nature of HDPE. The degree of crystallinity for each sample was determined by comparing the integration of the crystalline peaks with the total integration of both the crystalline and amorphous peaks (refer to the SI for additional details).55 Similarly to DSC analysis, a general reduction in the degree of crystallinity was observed as the concentration of DLP is increased (Fig. 4). The range of crystallinity was found to be between 67.6% and 60.9%, which is concordant with the data collected by DSC. Conversely, a small increase in crystallinity was observed as the DLP content was increased from 0% to 1.5% (samples M1 and M2). However, both SEC and DSC show that the peroxide influence on the polymer structure between M1 and M2 is subtle and thus the difference in crystallinity between these two samples may be too minor to be assessed accurately via WAXS analysis. A broad peak with a local maxima at ca. q = 0.30 nm−1 is observed in Lorenz corrected SAXS patterns of samples M1–M5 (Fig. S110†) corresponding to an interlamellar spacing (d) of 21 nm (d = 2π/q).56 Small variations of d are observed between samples; however, these differences are extremely subtle and are likely to arise from small variations in preparation procedures between samples.
![]() | ||
Fig. 4 A plot to demonstrate the changes in % polymer crystallinity as the peroxide loading increases. |
Finally, characterisation of the LCB was attempted using previously reported 13C NMR spectroscopy methods.57–60 However, despite performing direct 13C NMR experiments over long relaxation times (t1 = 10 seconds) for 4k scans, it was not possible to identify any distinctive methine units originating from LCB (Fig. S112–S114†). This could be because of the broad molecular weight distribution of the polymer as well as the presence of short chain branching, which could result in sensitivity and resolution challenges where some proton environments are undetected/masked. In addition, LCB arising from the peroxide triggered modification of PE would result in either Y- or H-type branching (Scheme 1). In the near absence of olefinic end groups (as demonstrated for IMPE), peroxide-triggered modification reactions can be reasonably expected to form H-type branches via the reaction of two macroradicals. Identification of H-type branching is particularly challenging and, to the best of our knowledge, unequivocal 13C NMR observation of such CH–CH linking moieties has only been reported for the gamma irradiation of short n-alkanes or low molecular weight PE.61–63
Having established the concept of solution-state peroxide modification of PE, the impact of the solvent was subsequently investigated (Fig. 5). Due to the success of using DCB solvent, which features only aryl-CH protons that are challenging to abstract, TBB and anisole were also investigated, along with n-nonane (Fig. 5A). The chain extension reactions were performed at 4.5 wt% DLP loading, and showed a significant solvent dependence in the order DCB > TBB > anisole > n-nonane, as determined by the increase in complex viscosity. Interestingly, DCB is the best-performing reaction media and has no Csp3–H units, whereas all the other solvents do. This may indicate that the alkyl groups can undergo H-abstraction, i.e. can undergo side-reactions with the peroxide, especially when large quantities of solvent are present. It is worth noting that TBB has previously been used as a solvent for the peroxide-triggered grafting of maleic anhydride onto PE.30 For anisole, the enhancement in polymer viscosity was relatively weak compared to DCB and TBB, yet it still surpassed the control, hinting at some chain extension albeit with some interference from the solvent reacting with the radical species. Sakurai et al. demonstrated that abstraction of a methyl proton from anisole is only slightly less favoured than that of toluene,64 which has reactive benzylic protons due to resonance stabilisation of the resultant radical species. The readiness of anisole to undergo H-abstraction was attributed to the stabilisation of the resulting radical species through polar effects. Finally, the chain extension reactions performed in n-nonane did not show any enhancements in polymer melt viscosities. As this solvent can be considered as a polyethylene mimic, featuring multiple CH2 repeat units, it is likely that n-nonane reacted with the generated radicals as it is present in the reaction vessel at much larger quantities (vs. only 7.5 w/v% IMPE).
To understand the impact of the rate of radical generation, investigations were performed altering the peroxide addition time and the reaction temperature (Fig. 5B and C). Firstly, the rate of adding the peroxide solution was reduced below 20 minutes to increase the maximum radical concentration in the solution. The results showed that adding the peroxide solution too quickly can reduce the extent of the chain modification reactions, where the addition of all of the peroxide over 1 min or in one-shot (∼0.02 min) gave only a slightly enhanced viscosity profile compared to the control reaction (Fig. 5B). This could be because some of the generated radicals self-quench by reacting with each other. Increasing the reaction temperature by 10 °C did not result in a significant change in final polymer viscosity profiles at 4.5 wt% DLP loading (Fig. 5C).
To probe the influence of the peroxide, BPO was investigated as an alternative to DLP for the peroxide-triggered chain extension reactions. BPO is frequently used as an initiator in radical polymerisations; it is already produced in large quantities and is one of the most widely available and low-cost organic peroxides available. A series of experiments were performed with BPO in both DCB and TBB. As BPO is shock sensitive in its pure form, it is commonly sold as a 75% suspension in water, and it was dried at 40 °C under high vacuum overnight prior to these initial modification reactions (refer to Experimental for additional safety details). The results demonstrated that BPO is a strong chain modifier (Fig. 5D). Notably, both 2.0 and 3.0 wt% BPO loadings in DCB altered the viscosity profile of the injection-moulding grade polymer IMPE to blow-moulding grade PE. In addition, as BPO has a lower molecular weight (242.3 g mol−1) than DLP (398.6 g mol−1), lower quantities of the organic peroxide (in weight) are needed to achieve similar transformations. Given that DLP is 1.6 times heavier than BPO, DLP loadings of 3.0 and 4.5 wt% would correspond to ∼2.0 and ∼3.0 wt% BPO, respectively. Accordingly, it was shown that only 2.0 wt% BPO loading was enough to achieve the blow moulding grade transition, whereas 3.0 wt% DLP loading in DCB was not enough to obtain the same viscosity profile enhancement (Fig. 2, top). This difference hints that BPO may be a more suitable organic peroxide for the chain extension reactions. This may be because it releases radicals more slowly than DLP (the relative half-lives of DLP and BPO are 0.45 and 1.9 min in chlorobenzene), mirroring the reduced chain extension observed upon rapid addition of the peroxide (Fig. 5B). When a 3.0 wt% loading of BPO was used in TBB, instead of DCB, a strong viscosity enhancement was again observed by rheology. Interestingly, the viscosity enhancement was as significant (or even slightly more so) in TBB than DCB solvent, which differs from the result obtained using 4.5 wt% DLP loading (Fig. 5A), where the performance in TBB was significantly poorer than in DCB. This suggests that the reaction solvent can directly affect the organic peroxide decomposition and the subsequent chain extension reactions. These solvent effects are likely specific to each organic peroxide as the decomposition kinetics and/or mechanisms can be different.65 SEC analysis was also performed (Fig. 6 and Table 3) on polymers modified with 3 wt% BPO in both DCB (M15) and TBB (M16). For both M15 and M16, significantly higher Mw values of 411 and 459 kg mol−1 were obtained, respectively, compared to the control sample (M1, Mw = 318 kg mol−1). In addition, the dispersity values of both M15 and M16 were reported as 4.3, an increase from a dispersity of 3.2 for M1. The strong high molecular weight fronts are also clear in the overlay of the SEC chromatograms. Samples M15 and M16 were also tested using DSC (Fig. S105–108†) for the determination of their crystallinities. Similar to the earlier reported trends with DLP, the crystallinities of M15 and M16 decreased to 63.7% and 62.5% (Table S2†), respectively.
Exp. | Peroxide loading (wt%) | Solvent | M n (g mol−1) | M w (g mol−1) | Đ |
---|---|---|---|---|---|
M1 | 0 | DCB | 100![]() |
318![]() |
3.2 |
M15 | 3.0 | DCB | 95![]() |
411![]() |
4.3 |
M16 | 3.0 | TBB | 106![]() |
459![]() |
4.3 |
Finally, the sensitivity of the modification reactions towards water and oxygen was investigated. As BPO was obtained as a 75% suspension in water from the supplier, wet BPO was used to test the sensitivity of the chain extension reaction to water in both DCB and TBB under a nitrogen atmosphere. The results demonstrated no significant sensitivity to water as the obtained complex viscosity profiles were almost identical (Fig. 5E), irrespective of whether wet or dry BPO was used. On the other hand, when the reactions were performed under air (M19), but in the absence of water, a discolouration of the reaction solution from colourless to pale yellow was observed, which was also reflected in the final polymer product (Fig. 5F). A deterioration in the polymer viscosity profile relative to both M1 and M20 was also observed (see ESI, Fig. S75†) as well as a clear indication of oxygenated moieties on the polymer backbone by NMR spectroscopy (Fig. 7). Indeed, the 1H NMR spectra revealed proton environments often attributed to aldehydes (∼9.7 ppm) and other oxygenated moieties (3.0–5.0 ppm).57 As oxygen is known to react with active radicals, it is believed to inhibit the LCB modification reactions (Scheme 1). Similar observations have been reported for radical polymerisations, where dead chain ends are generated in the presence of oxygen.66 In support of these observations, where the system can tolerate water but not oxygen, atom-transfer radical polymerisations are often considered tolerant to alcohol groups and water as long as the reaction solution is deoxygenated prior to polymerisation.67 It is worth noting that both M19 and M20 were prepared the same way and purified by precipitation into hexanes twice, and that far fewer oxygenated moieties were observed when the reaction was performed in the absence of oxygen (M20). In addition, a higher peroxide loading sample (M5, 6 wt% DLP) was also characterised using NMR spectroscopy and no strong signs of oxygenated species were detected. Yet even using just 0.5% peroxide loading in air showed oxygenated moieties (M19), indicating that these functional groups arise from the presence of air. Some studies have shown that oxygenated functional groups can enhance the polymer properties,68 yet here, the presence of oxygenated moieties in M19 clearly decreases the viscosity profiles, providing further support that the enhanced viscosity profiles are due to the presence of long chain branching.
![]() | ||
Fig. 7 1H NMR (800 MHz, 100 °C, C2D2Cl4) spectra of M1, M5, M19 and M20. The intensity of the main CH2 peak at 1.4 ppm was normalised to 100 in all spectra to ensure a valid comparison. |
Characterisation of the reaction products was performed using SEC, where n-dodecane alone and the control experiment where no peroxide was present both gave a Mn value of 180 g mol−1 against polystyrene standards, which gives relatively good agreement with the molecular weight of n-dodecane being 170 g mol−1 (Fig. 8A and Table 4). Following the peroxide-triggered modification reaction, the product gave a peak front as well as an increase in the overall Mn value (230 g mol−1), Mw value (300 g mol−1) and the dispersity (Đ = 1.3). These results confirm that the modification reactions can extend the chain length of the starting material. In addition, analysing the product SEC trace as three separate peaks (Fig. 8B) showed that crude peak (CP) 1 has Mn (190 g mol−1) similar to n-dodecane, while CP2 and CP3 had Mn values of 340 g mol−1 and 650 g mol−1, respectively. CP2 has a more distinctive peak shape, with an Mn value approximately twice that obtained for n-dodecane (experimental values of 340 vs. 180 g mol−1), which suggests it could be two n-dodecane chains linked together. On the other hand, CP3 has a higher molecular weight hinting at multiple n-dodecane chains connected to each other, supported by the broad and poorly resolved peak shape. As the refractive index (RI) is a concentration detector, relative peak areas were also compared, and the results indicate that n-dodecane is the main species in the crude product, with higher molecular weight CP2 being the second most dominant species.
Sample | M n (g mol−1) | M w (g mol−1) | Đ | RI peak area (%) |
---|---|---|---|---|
n-Dodecane | 180 | 180 | 1.0 | 100 |
Control | 180 | 180 | 1.0 | 100 |
Model product | 230 | 300 | 1.3 | 100 |
CP1 | 190 | 200 | 1.0 | 64 |
CP2 | 340 | 350 | 1.0 | 22 |
CP3 | 650 | 680 | 1.0 | 14 |
Further studies were also performed using NMR spectroscopy, where the model experiment crude product was placed under high vacuum to remove some of the unreacted dodecane. Characterisation of this sample revealed a new aliphatic-CH 1H NMR resonance at 1.40–1.75 ppm (Fig. S116†). This is notably different from the 1H NMR spectra of linear alkanes, which show diagnostic CH2 resonances at 1.28–1.33 ppm, and CH3 resonances at 0.89 ppm across a range of chain lengths (e.g. heptane, hexane, pentane, propane and ethane).69,70 The new resonance at 1.40–1.75 ppm instead comes in a region associated with branched alkanes. Notably, the chemical shift of this CH resonance is significantly lower than that of oxygen-containing OCH units, which have chemical shifts around 3.5 ppm. Indeed, only trace resonances are present in the O–CH region of the spectrum. While there are other, aromatic resonances present in the 1H NMR spectrum, these are attributed to benzoic acid which is a common decomposition product of BPO, as both the 1H and 13C NMR spectra give good agreement with literature values.71
This study showcases the use of solution-state peroxide modification reactions as a simple method to enhance the material properties of HDPE from an injection moulding grade to a blow-moulding grade material. Furthermore, the data shows that the viscosity profiles can be tailored depending on the peroxide loading, potentially offering a facile route to modify the properties of PE for specific applications. This concept has the potential to be extended beyond PE, to the solution-state peroxide modification of other polymers. As higher molecular weight polymers with LCB are well-known to improve both the impact and chemical resistance of PE, this methodology could also offer a shortcut for the preparation of new PE grades and formulations with required specifications that may not currently be commercially available.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3py01399e |
This journal is © The Royal Society of Chemistry 2024 |