Rapidly Self‐Healable and Melt‐Extrudable Polyethylene Reprocessable Network Enabled with Dialkylamino Disulfide Dynamic Chemistry

Abstract

Catalyst‐free, radical‐based reactive processing is used to transform low‐density polyethylene (LDPE) into polyethylene covalent adaptable networks (PE CANs) using a dialkylamino disulfide crosslinker, BiTEMPS methacrylate (BTMA). Two versions of BTMA are used, BTMA‐S₂, with nearly exclusively disulfide bridges, and BTMA‐S_n, with a mixture of oligosulfide bridges, to produce S₂ PE CAN and S_n PE CAN, respectively. The two PE CANs exhibit identical crosslink densities, but the S₂ PE CAN manifests faster stress relaxation, with average relaxation times ∼4.5 times shorter than those of S_n PE CAN over a 130 to 160 °C temperature range. The more rapid dynamics of the S₂ PE CAN translate into a shorter compression‐molding reprocessing time at 160 °C of only 5 min (vs 30 min for the S_n PE CAN) to achieve full recovery of crosslink density. Both PE CANs are melt‐extrudable and exhibit full recovery within experimental uncertainty of crosslink density after extrusion. Both PE CANs are self‐healable, with a crack fully repaired and the original tensile properties restored after 30 min for the S₂ PE CAN or 60 min for the S_n PE CAN at a temperature slightly above the LDPE melting point and without the assistance of external forces.

LDPE is upcycled into polyethylene covalent adaptable networks (PE CANs) via simple, catalyst‐free, radical‐based reactive processing with dynamic crosslinker BTMA. The nearly exclusively disulfide‐based BTMA exhibits fast dynamic chemistry, rendering the PE CANs rapidly reprocessable, melt‐extrudable, and self‐healable, with full property recovery after self‐healing achieved slightly above the LDPE melting temperature within 60 min.

1. Introduction

Polyethylene (PE) is the most produced and utilized of all plastics due to its mechanical strength, toughness, and ductility, stemming from its semi‐crystalline structure.^{[ 1 ]} Thermoplastic PE, including low‐density PE (LDPE) and high‐density PE (HDPE), has extensive applications, including containers, building materials, and agricultural films. Although produced at a fraction of the level of thermoplastic PE, the manufacture of PE thermosets, or crosslinked PE (PEX), is commercially important.^{[ 2} , ^{3 ]} One method that is practiced on a major scale to produce PEX involves permanently crosslinking thermoplastic PE by radical‐based reactive processing assisted by a radical initiator, such as dicumyl peroxide (DCP).^{[ 4 ]} Because of the permanent covalent crosslinks, PEX has improved thermal stability, deformation resistance, and mechanical properties, rendering PEX suitable for pipework systems and high‐voltage cable insulation applications, among others.^{[ 4} , ^{5 ]} Nevertheless, sometimes under harsh use conditions, PEX can suffer damage due to mechanical or electrical stress.^{[ 6 ]} Conventional PEX is irreparable owing to the permanent crosslinks prohibiting the PE chains from undergoing the flow needed for healing.

Recent efforts have been exploring methods to create crosslinked PE with self‐healing capabilities, which can be categorized into extrinsic or intrinsic methods.^{[ 7} , ⁸ , ⁹ , ¹⁰ , ¹¹ , ^{12 ]} Extrinsic self‐healing methods involve doping additives such as microcapsules that contain healing agents in the material during fabrication.^{[ 10} , ¹¹ , ^{13 ]} Although extrinsic methods have been shown to be feasible in making self‐healing PEX,^{[ 10} , ^{11 ]} they still face challenges such as additional production costs, the requirement of specific design depending on the polymer matrix, and limited lifetimes due to the depletion of healing agents.^{[ 14} , ^{15 ]} Intrinsic self‐healing methods utilize covalent or non‐covalent dynamic bonds to make PE networks.^{[ 7} , ⁸ , ⁹ , ^{12 ]} Advancements in covalent adaptable networks (CANs) suggest a promising pathway for achieving self‐healing crosslinked PE through an intrinsic approach.^{[ 16} , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ^{21 ]} CANs are polymer networks with dynamic covalent bonds serving as crosslinks and are commonly classified into one of two types. Associative CANs are made with covalent crosslinks that undergo exchange reactions under proper stimuli, with the total number of crosslinks remaining constant;^{[ 22} , ²³ , ^{24 ]} dissociative CANs are made with covalent crosslinks that can reversibly dissociate and re‐associate under stimuli.^{[ 16} , ^{25 ]} In some CANs, associative and dissociative mechanisms co‐exist.^{[ 26 ]} In 2017, Leibler and coworkers reported the first research on PE CANs by covalently attaching dioxaborolane dynamic covalent crosslinks into PE using a radical‐based reactive processing approach.^{[ 27 ]} Since then, an array of dynamic covalent chemistries have been exploited to undertake research into PE CANs, including vinylogous urethane exchange,^{[ 28 ]} silyl ether exchange,^{[ 29 ]} and transesterification,^{[ 30} , ^{31 ]} among others.^{[ 32} , ^{33 ]}

However, relatively little research has focused on the self‐healing potential of PE CANs. Li et al. recently reported on self‐healing PE CANs which they produced by crosslinking maleic anhydride functionalized PE (PE‐MA) using multifunctional epoxies. Leveraging transesterification, their PE CAN exhibited reprocessability and healing properties.^{[ 9 ]} However, this method requires the use of PE‐MA rather than PE, limiting its versatility. Moreover, the “self‐healing” of their PE CAN in terms of restoring the original tensile properties was assisted by external pressure (2 MPa applied by a mold). Additionally, catalyst incorporation was necessary to enhance the transesterification, posing additional costs and environmental concerns. Another study by Wang et al. reported on dioxaborolane‐crosslinked CANs from ethylene/octene copolymers.^{[ 12 ]} However, an exceptionally long healing time (12 h) of damaged CANs was required to achieve full recovery of the original properties of the undamaged CANs.

We recently introduced a simple, commercially friendly method to prepare PE CANs and ethylene‐based copolymer CANs crosslinked with a bis(2,2,6,6‐tetramethyl‐piperidin‐1‐yl) disulfide (BiTEMPS) dynamic covalent crosslinker, which we call BiTEMPS methacrylate (BTMA).^{[ 34} , ³⁵ , ^{36 ]} At sufficiently high temperatures, the dialkylamino disulfide bond in BTMA reversibly dissociates into sulfur‐centered radicals and recombines without the assistance of any catalyst.^{[ 37} , ³⁸ , ^{39 ]} Due to the stabilization effect provided by the lone pair electron of the nitrogen atom next to the sulfur atom, the thionitroxide radicals have excellent stability in an air atmosphere and preferentially re‐associate with each other rather than undergoing side reactions with oxygen, olefins, or other substances.^{[ 37} , ^{40 ]} Capitalizing on the dissociative dynamic nature of dialkylamino disulfide chemistry, the PE CANs made from LDPE, HDPE, and ethylene‐based copolymers exhibited excellent reprocessability by compression molding, with full recovery of crosslink density after multiple recycling steps.^{[ 34} , ³⁵ , ^{36 ]}

Our recent study reveals that the existing synthesis procedure of BTMA gives a mixture of dialkylamino compounds containing various oligosulfide linkages (disulfide, trisulfide, and tetrasulfide) as products, here referred to as BTMA‐S_n.^{[ 41 ]} Although all of the dialkylamino oligosulfide compounds have shown dynamic covalent character, evidence also suggests that CANs crosslinked with different oligosulfide compounds can display different relaxation timescales.^{[ 42} , ⁴³ , ^{44 ]} We have recently improved the protocol of synthesizing BTMA, which leads to a BTMA product with ≥95% disulfide, referred to as BTMA‐S₂.^{[ 41 ]} The predominantly disulfide‐containing BTMA‐S₂ led to a hexyl methacrylate (HMA)‐based CAN showing remarkably faster stress relaxation, which was further translated into rapid melt‐reprocessing of the HMA‐based CAN. Accounting for such benefits, we aim to explore the potential fast reprocessability and self‐healing capabilities of PE CANs crosslinked by BTMA‐S₂.

For the first time, we demonstrate that the reprocessable, catalyst‐free PE CANs made using low levels of BTMA in conjunction with radical‐based reactive processing of low‐density PE (LDPE) with DCP also exhibit self‐healing capabilities, with the full healing and recovery of tensile properties occurring at 130 °C in the absence of external pressure. Two versions of BTMA, the predominantly disulfide‐containing BTMA‐S₂ and oligosulfide‐containing BTMA‐S_n, are employed separately as crosslinkers to produce S₂ PE CAN and S_n PE CAN, respectively. Although S₂ PE CAN exhibits comparable thermomechanical properties with S_n PE CAN, it manifests much more rapid stress relaxation. The S₂ PE CAN is able to be melt‐reprocessed by compression molding for 5 min at 160 or 180 °C with full restoration of its original crosslink density. In contrast, achieving reprocessability with full crosslink density recovery in the S_n PE CAN require as long as 30 min of compression molding at 160 °C. Importantly, both the S₂ PE CAN and the S_n PE CAN exhibit compatibility with twin‐screw melt extrusion at 180 °C and 5 rpm, yielding extrudates with fully recovered crosslink density in comparison with their original molds. Using microscopy and tensile testing, we demonstrate the outstanding self‐healing capabilities of S₂ PE CAN against mechanical damage. Specifically, the S₂ PE CAN fully heal and recover its original Young's modulus, tensile strength, and elongation at break after only 60 min of healing at 130 °C without the aid of external forces. In contrast, the S_n PE CAN requires 120 min under the same conditions to achieve full self‐healing, and PEX, made using a similar approach but lacking BTMA, exhibits no thermal reprocessability or self‐healing capability. To the best of our knowledge, we are the first to report such rapid self‐healing of PE CANs without the need for external pressure that also fully recovers their properties within experimental uncertainty.

2. Results and Discussion

2.1. Synthesis of PE CANs by Radical‐Based Reactive Processing and Basic Characterizations

Two versions of BiTEMPS methacrylate (BTMA) synthesized using different procedures were employed as dynamic crosslinkers in the preparation of PE CANs. The predominantly disulfide‐based BTMA‐S₂ was synthesized following a recently optimized method;^{[ 41 ]} BTMA‐S_n, containing a mixture of oligosulfides, was synthesized following a previous method.^{[ 34} , ^{38 ]} (See Figure S1, Supporting Information for the synthesis schemes and structures of BTMA‐S₂ and BTMA‐S_n.) Our S₂ PE CAN and S_n PE CAN were prepared with BTMA‐S₂ and BTMA‐S_n as dynamic crosslinkers, respectively. (See Supporting Information for the full details on the synthesis methods used for each of the BTMAs and for radical‐based reactive processing of LDPE to yield PE CANs.) Both S₂ PE CAN and S_n PE CAN were synthesized via a radical‐based reactive processing approach starting from commercially available LDPE in pellet form. For the synthesis, LDPE, dynamic crosslinker BTMA (S₂ or S_n), and radical initiator DCP were first non‐reactively homogenized in a lab‐scale melt‐mixer at 130 °C, a temperature at which DCP does not dissociate significantly into free radicals. After homogenization, the mixture underwent compression molding at 180 °C for 30 min to facilitate curing; the resulting films are considered 1st‐mold PE CANs. (In our 2022 study,^{[ 34 ]} the S_n PE CAN was made by curing at 160 °C instead of 180 °C. We observed no difference within experimental uncertainty in the thermomechanical properties obtained from dynamic mechanical analysis between the S_n PE CANs made in the current study and the S_n PE CANs in ref. [34]. See Figure S2, Supporting Information.) Building on our previous research,^{[ 34} , ³⁵ , ^{36 ]} we determined that loadings of 5 wt% BTMA and 1 wt% DCP generally give robustly crosslinked CANs with substantial crosslink density and full reprocessability. These loadings of BTMA and DCP were maintained in this work to prepare PE CANs. (The wt% values are calculated relative to the mass of the neat LDPE before synthesis.)

During the curing of PE CANs at 180 °C, DCP dissociates into free radicals. These free radicals can undergo chain transfer with PE chains, extracting hydrogen atoms from the backbone of PE and leaving PE macroradicals. The grafting of a BTMA molecule onto PE occurs when a PE macroradical reacts with the carbon‐carbon double bond on one end of BTMA. The successful formation of a BTMA crosslink involves another grafting reaction between the other carbon‐carbon double bond of the BTMA and the backbone of another PE chain. (See Figure 1 .) Of course, other reactions besides BTMA crosslinking can occur.^{[ 34} , ^{35 ]} First, carbon‐based PE macroradicals can terminate within each other by disproportionation or combination; the latter leads to an irreversible or permanent crosslink. Second, a DCP radical may react with one or more BTMA molecules, resulting in a BTMA oligomer radical. That radical could react with a second BTMA oligomer radical or a PE macroradical, with neither reaction resulting in a crosslink. Third, if an unreacted end of a grafted BTMA fails to undergo combination with a PE macroradical, it may result in a dangling BTMA that does not contribute to the overall network crosslink density.

Figure 1.

Scheme of synthesizing a PE CAN (taking S₂ PE CAN as the example) with an ideal structure by radical‐based reactive processing of LDPE with BTMA (taking BTMA‐S₂ as the example) as dynamic crosslinker and DCP as radical initiator. At high temperatures, BTMA crosslinks reversibly dissociate into stable thionitroxide radicals, enabling melt flow of the PE CAN.

For the synthesis of permanently crosslinked PE, or PEX, neat LDPE was first homogenized at 130 °C with only DCP (1 wt%). Then it was cured by compression molding at 180 °C for 30 min, which are the same molding conditions used to produce our PE CANs.

We performed FTIR spectroscopy to verify the successful incorporation of BTMA into our PE CANs. The black dashed curve in Figure S3 (Supporting Information) represents the spectrum of neat LDPE, which has no peak present associated with carbonyl units. Upon reactive processing that yields an as‐synthesized or 1st‐mold S₂ PE CAN as indicated by the red solid curve, a peak is present at ∼1720 cm⁻¹ corresponding to the carbonyl groups in BTMA. We also washed the 1st‐mold S₂ PE CAN by dissolving it in o‐xylene at 130 °C to remove any unreacted BTMA‐S₂ and PE chains that were not linked to a percolated network structure, resulting in “washed S₂ PE CAN”. As depicted in Figure S3 (Supporting Information), the ∼1720 cm⁻¹ carbonyl peak persists on the red dashed curve of washed S₂ PE CAN, confirming the covalent attachment, or grafting, of BTMA‐S₂ onto the PE backbone. The successful grafting of BTMA‐S_n was also confirmed in a similar manner based on the FTIR results on S_n PE CAN as shown by the blue curves in Figure S3 (Supporting Information).

The gel content of the as‐synthesized S₂ PE CAN was 62 ± 3%, in agreement with our previously reported result for an S_n PE CAN (61%) made from LDPE.^{[ 34 ]} Both FTIR and gel content experiments affirm the grafting of BTMA‐S₂ onto PE and the preparation of a percolated PE network, observations that are similar to our previous report on the S_n PE CAN.^{[ 34 ]}

We did further characterizations by differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA). DSC results indicate a slight reduction in crystallinity from 32% to ∼29% upon transforming neat LDPE into either the S₂ PE CAN or PEX. (See Figure S4 and Table S1, Supporting Information.) An identical slight reduction in crystallinity was reported in ref. [34] upon transforming LDPE into an S_n LDPE CAN. This slight crystallinity reduction can be attributed to the crosslinks, either dynamic or permanent ones, disrupting the formation of crystalline structures in LDPE. TGA results make evident the outstanding thermal stability of our S₂ LDPE CAN, with a T ₉₅ value (temperature at which 95% of the initial mass remains) of 420 °C. (See Figure S5, Supporting Information.) This value exceeds the 388 °C T ₉₅ value reported for the S_n LDPE CAN in ref. [34].

Thus, regardless of whether the BTMA is S₂‐based or S_n‐based, the same gel contents are obtained, and the same slight reductions in crystallinity result upon transforming LDPE into an LDPE CAN, consistent with highly similar crosslinking being achieved independent of the BTMA version. Both versions of BTMA yield high T ₉₅ values when incorporated in PE CANs, with an advantage to the S₂ PE CAN having a higher value by 32 °C.

2.2. Dynamic Mechanical Analysis and Stress Relaxation

We studied thermomechanical properties related to the crosslink density of the S₂ PE CAN using dynamic mechanical analysis (DMA). (See Figure 2 .) Between 101 and 113 °C, neat LDPE exhibits a factor of ∼300 reduction in tensile storage modulus (E’) which is associated with melting of crystalline domains. Above 113 °C, the neat LDPE exhibits a substantial flow‐like response, and E’ can no longer be characterized. In contrast, the S₂ PE CAN exhibits a quasi‐rubbery plateau in E’ at temperatures exceeding ∼110 °C, proving the formation of its crosslinked network structure. Like many other CANs with strictly dissociative dynamic covalent character,^{[ 34} , ³⁸ , ⁴⁵ , ^{46 ]} the S₂ PE CAN exhibits a slightly decreasing trend in E’ with increasing temperature in the quasi‐rubbery plateau. This can be attributed to the BTMA dissociative dynamic character being active at the temperatures associated with the plateau region, leading to a decrease in crosslink density with increasing temperature.

Figure 2.

Tensile storage modulus (E’) as a function of temperature of as‐synthesized, i.e. 1st‐mold S₂ PE CAN and S_n PE CAN in comparison with neat LDPE.

Across the tested temperature range, including in the quasi‐rubbery plateau above ∼110 °C, the S_n PE CAN exhibits an E’ curve that is nearly identical to that of the S₂ PE CAN. (See Figure 2 and Table S2, Supporting Information.) According to Flory's ideal rubber elasticity theory, the rubbery plateau tensile modulus, E, is proportional to the crosslink density of a polymer network.^{[ 47 ]} Under typical conditions of measurement, E’ and E are approximately equal. Therefore, the DMA results indicate that the S₂ and S_n PE CANs have identical crosslink densities within experimental uncertainty, in agreement with the identical gel contents discussed above. Thus, regardless of whether the BTMA is S₂‐based or S_n‐based, our reactive processing of LDPE with BTMA leads to the same level of crosslink density in the LDPE CANs.

We recently demonstrated that a hexyl methacrylate (HMA)‐based CAN made by free radical copolymerization of HMA and BTMA‐S₂ exhibited substantially faster stress relaxation at 130 °C than a HMA‐based CAN made in an identical manner but with BTMA‐S_n, with the CAN made with BTMA‐S₂ having a factor of ∼3.5 smaller average relaxation times than the CAN made with BTMA‐S_n.^{[ 41 ]} Here, we consider the effect of BTMA version on the stress relaxation behavior of CANs with identical crosslink density made from LDPE by radical‐based reactive processing. Figure 3a shows the tensile stress normalized by the initial stress (σ(t)/σ ₀) as a function of time (t) and temperature (T) for the 1st‐mold S₂ PE CAN after applying an initial small strain. At very short times, the S₂ PE CAN exhibits a sudden drop in normalized stress at each T owing to the relaxations of the fraction of PE chains left uncrosslinked after reactive processing. Neat LDPE relaxes stress on short time scales similar to those of the sudden drops in the S₂ PE CAN (between ∼3 to 7 s); see Figure S6 and Table S3 (Supporting Information) for shear‐mode relaxation experiments on neat LDPE at 130 and 160 °C. Ultimately, the relaxations of the S₂ PE CAN occurring at long times result from the dissociative dynamic covalent character of the BiTEMPS crosslinks, as permanent crosslinks in PEX prevent the material from relaxing stress. The S₂ PE CAN exhibits faster relaxation with increasing T, with 80% of the initial stress being relaxed within a timescale of ∼400 s at 130 °C and ∼70 s at 160 °C. In comparison, Figure 3b shows that the S_n PE CAN exhibits similar initial drops in normalized stress to the S₂ PE CAN from the relaxations of sol chains yet much slower relaxation than the S₂ PE CAN at every T, with the timescale to achieve 80% relaxation ranging from ∼1700 s at 130 °C to ∼300 s at 160 °C.

Figure 3.

Normalized stress relaxation (σ(t) relative to the initial σ) curves at various temperatures for a) S₂ PE CAN and b) S_n PE CAN. Arrhenius plots of the natural logarithm of average relaxation time (<τ>) as a function of inverse temperature with apparent stress relaxation activation energies (E _a) of c) S₂ PE CAN and d) S_n PE CAN.

We quantitatively analyzed the stress relaxation curves by fitting them to the Kohlrausch‐Williams‐Watts (KWW) stretched exponential decay function:^{[ 48} , ^{49 ]}

$$\frac{\sigma \left(t\right)}{{\sigma }_0}=\exp \left[-{\left(\frac{t}{{\tau }^{\ast }}\right)}^{\beta }\right]$$ (1)

where τ* is the characteristic relaxation time and β (0 < β ≤ 1) is the stretching exponent that accounts for the breadth of the relaxation distribution. Studies have shown that the stress relaxation of CANs typically, although not exclusively,^{[ 46 ]} exhibits a stretched exponential decay.^{[ 50} , ^{51 ]} The average relaxation time (<τ>) is calculated from the KWW fit as follows:^{[ 48 ]}

$$〈\tau 〉=\frac{{\tau }^{\ast }\mathrm{\Gamma }\left(1/\beta \right)}{\beta }$$ (2)

where Γ is the gamma function. Table 1 summarizes the values of τ*, β, and <τ> of our two PE CANs as functions of T.

Table 1.

Characteristic relaxation times (τ*), stretching exponents (β), and average relaxation times (<τ>) as a function of temperature for PE CANs obtained using KWW decay function fits.

Sample	T [°C]	τ* [s]	β	<τ> [s]	R2
S2 PE CAN	130	202	0.53	365	0.986
	140	111	0.55	189	0.992
	150	62	0.59	95	0.990
	160	39	0.71	49	0.993
Sn PE CAN	130	619	0.43	1706	0.993
	140	394	0.50	788	0.991
	150	242	0.54	424	0.992
	160	145	0.61	214	0.995

The low values of β, ranging from 0.53 to 0.71 for the S₂ PE CAN and from 0.43 to 0.61 for the S_n PE CAN, indicate substantial breadth to the distribution of relaxation modes in these PE CANs, which could include contributions from chain segments, branches, entanglements, dissociation, the subsequent diffusion of the dynamic bonds, etc.^{[ 52} , ^{53 ]} Notably, the β values are higher for the S₂ PE CAN than for the S_n PE CAN at all tested temperatures. This indicates that the S₂ PE CAN exhibits a narrower relaxation breadth, which is consistent with the fact that the S₂ PE CAN contains BTMA with nearly all disulfide bridges whereas the S_n PE CAN has substantial levels of disulfide, trisulfide, and tetrasulfide bridges.

As evident from Table 1, at any given temperature, the average stress relaxation time of the S₂ PE CAN is a factor of ∼4.5 times smaller than that of the S_n PE CAN. The substantially more rapid stress relaxation behavior of S₂ PE CAN, resulting from the faster dynamic chemistry of BTMA‐S₂ relative to BTMA‐S_n, is consistent with the observations in our recent study on hexyl methacrylate‐based CANs made with each version of BTMA.^{[ 41 ]} We performed fits of the <τ> values in order to assess the apparent Arrhenius activation energies for stress relaxation (E _a). Figure 3c,d shows that the data are well fit to an Arrhenius temperature dependence and yield identical E _a values within experimental uncertainty of 97 ± 2 and 100 ± 2 kJ mol⁻¹ for the S₂ PE CAN and S_n PE CAN, respectively. These values are in agreement with the BiTEMPS bond dissociation energy.^{[ 42} , ^{54 ]} Literature reports have also shown that CANs made with dialkylamino crosslinks containing different oligosulfide bridges exhibit similar E _a values for stress relaxation.^{[ 41} , ⁴² , ^{43 ]} However, in spite of the similar T dependence of the dynamic character of our S₂ PE CAN and S_n PE CAN, the more rapid dynamic character of the S₂‐based BTMA may make it a better choice for a dynamic covalent crosslinker in cases where rapid reprocessing is important or when efficient self‐healing character is desired.

While the relaxations of the S₂ PE CAN and S_n PE CAN are fit very well by a single KWW function (evidenced by the very high R² values (>0.985) corresponding to the fits), we further sought to account for the sudden drops in stress from sol PE chains in our assessment of the relaxation data, as these sudden drops could complicate our interpretation of the average relaxation times and Arrhenius activation energies of the CANs. Thus, as a means to isolate the short‐time relaxations from the sol PE chains and the long‐time relaxations from the dynamic covalent cross‐links, we refit the relaxation data seen in Figure 3 to a linear combination of two KWW functions:

$$\frac{\sigma \left(t\right)}{{\sigma }_0}=A\ast \exp \left[-{\left(\frac{t}{{\tau }_1^{\ast }}\right)}^{{\beta }_1}\right]+\left(1-A\right)\ast \exp \left[-{\left(\frac{t}{{\tau }_2^{\ast }}\right)}^{{\beta }_2}\right]$$ (3)

See Table S4 (Supporting Information) for the fitted parameters from our refits of the PE CAN relaxation data. τ*₁ and β ₁ represent the characteristic relaxation time and stretching exponent, respectively, of the short‐time relaxations, and τ*₂ and β ₂ represent the characteristic relaxation time and stretching exponent, respectively, of the long‐time relaxations. Average relaxation times of the short‐ and long‐time relaxations, <τ>₁ and <τ>₂, respectively, were calculated as follows:^{[ 55 ]}

$${〈\tau 〉}_i=\frac{{\tau }_i^{\ast }\mathrm{\Gamma }\left(\frac{1}{{\beta }_i}\right)}{{\beta }_i};i=1\mathrm{or}2$$ (4)

For both the S₂ PE CAN and S_n PE CAN, <τ>₁ values corresponding to short‐time relaxations are on the order of the <τ> values of neat LDPE after its relaxations were fit well to a single KWW function (Table S3, Supporting Information), further corroborating that the sudden drops in stress of the PE CANs are owed to the relaxations of sol PE chains. Additionally, we performed Arrhenius fits of the <τ>₂ values in order to quantify E _a values of the isolated long‐time relaxations (Figure S7, Supporting Information); we obtained 98 ± 7 and 94 ± 3 kJ mol⁻¹ for the S₂ PE CAN and S_n PE CAN, respectively. These values are either within experimental uncertainty or very close outside of experimental uncertainty of the E _a values of the S₂ PE CAN and S_n PE CAN determined from the <τ> values obtained from fits to a single KWW function (Figure 3c,d). Thus, our initial interpretations of the stress relaxations and associated E _a values hold when fully accounting for the sudden drops in stress due to the sol PE chains of the PE CANs. As a complementary analysis to average relaxation time determinations, we also conducted a tension‐mode frequency sweep at 150 °C on the S₂ PE CAN (Figure S8, Supporting Information) to determine the cross‐over time between E′ and E″, τ _cross‐over, a metric that is sometimes used to describe the relaxation time of CANs.^{[ 56 ]} We obtained a τ _cross‐over of ∼90 s from a cross‐over angular frequency of ∼0.07 rad s⁻¹ for the S₂ PE CAN at 150 °C, which corresponds very well with the <τ> of 95 s at this temperature.

2.3. Fast Thermal Reprocessing by Compression Molding and Compatibility with Melt Extrusion with Full Recovery of Crosslink Density

Motivated by the faster dynamic chemistry of the S₂ PE CAN as indicated by its more rapid stress relaxation, we investigated its thermal reprocessability by compression molding. A 1st‐mold S₂ PE CAN was reprocessed by a cut‐and‐remold method at 160 °C for 5, 10, or 30 min. We also attempted to reprocess the 1st‐mold S_n PE CAN under the same conditions.

We observed that the S₂ PE CAN requires only 5 min at 160 °C to yield a 2nd‐mold sample with full recovery of crosslink density (see DMA results in Figure S9b, Supporting Information). However, the 2nd‐mold samples had slight blemishes on the sample surfaces. A longer reprocessing time of 10 min yielded smooth, unblemished 2nd‐mold S₂ PE CAN samples. (See pictures in Figure S9a, Supporting Information.) In contrast, among the tested conditions, the S_n PE CAN exhibits full reprocessability based on the complete retention of crosslink density after a longer time of 30 min in a compression molder at 160 °C. The faster reprocessability exhibited by S₂ PE CAN compared to S_n PE CAN is consistent with the conclusion reached from our stress relaxation studies.

Next, we conducted successive reprocessing steps on the S₂ PE CAN using compression molding at 180 °C. At this higher reprocessing T, the S₂ PE CAN is able to be reprocessed into a transparent, intact, and smooth sample with no blemishes after only 5 min as shown in Figure 4a. Figure 4b shows the E’ curves as a function of T for the 1st‐mold, 2nd‐mold, and 3rd‐mold S₂ PE CANs, with each reprocessing step done by compression molding for only 5 min at 180 °C. Importantly, the S₂ PE CAN recovers its crosslink density within experimental uncertainty after each reprocessing step, as evidenced by the overlapping data in the quasi‐rubbery plateau.^{[ 47 ]} (See E’ data in Table S2, Supporting Information.) These results demonstrate the excellent reprocessability of the S₂ PE CAN, like that of previously studied S_n PE CAN,^{[ 34 ]} but achievable in only 5 min at 180 °C. In contrast, permanently crosslinked PE or PEX cannot be reprocessed at 180 °C, as shown in Figure S10 (Supporting Information).

Figure 4.

a) Reprocessing of S₂ PE CAN by cutting and remolding at 180 °C for 5 min using a hot compression molder. b) Tensile storage modulus (E’) as a function of temperature and molding of S₂ PE CAN.

The remarkably fast compression molding at 180 °C suggested that the S₂ PE CAN may have the potential to be (re)processed via continuous melt processing methods used in industry, such as melt extrusion. After being cut into small bits, the 1st‐mold S₂ PE CAN underwent successful twin‐screw melt extrusion at 180 °C with a rotation speed of 5 rpm. (Figure 5a) After extrusion, the S₂ PE CAN extrudate was simply cooled under ambient, room‐temperature conditions and then subjected to DMA characterization, yielding an E’ curve with a quasi‐rubbery plateau region that overlaps substantially with its 1st‐mold counterpart. (See Figure 5b) and E’ data in Table S2, Supporting Information.) Thus, the rapid dynamic chemistry associated with the S₂ PE CAN at a temperature of 180 °C allows for continuous melt extrusion with the cooled product having a crosslink density comparable to those of compression molded films. We observed that the S_n PE CAN also exhibits compatibility with melt extrusion under the same conditions (Figure S11a, Supporting Information), demonstrating full crosslink density recovery based on an E’ curve for the extruded sample overlapping with its 1st‐mold counterpart within small error. (Figure S11b and Table S2, Supporting Information.) Thus, an extrusion condition of 180 °C and 5 rpm is also sufficient for continuous melt‐reprocessing of the S_n PE CAN even though it has a slower dynamic character than the S₂ PE CAN.

Figure 5.

a) Picture of 1st‐mold S₂ PE CAN being successfully extruded using a twin‐screw extruder at 180 °C with a rotation speed of 5 rpm. b) Tensile storage modulus (E’) as a function of temperature of the extruded S₂ PE CAN in comparison with its 1st‐mold sample.

In comparison with reprocessing by compression molding, reprocessing by extrusion of CANs, including polyolefin‐based CANs, has seldom been reported.^{[ 28} , ³¹ , ³⁵ , ⁴¹ , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ , ⁶¹ , ^{62 ]} Furthermore, full property recovery of CANs after reprocessing by extrusion is even less common in academic literature; to the best of our knowledge, this report is among the first that demonstrates full crosslink density recovery of CANs after reprocessing via extrusion.^{[ 41} , ⁵⁷ , ^{62 ]} In previous cases of CANs in which the dynamic covalent chemistry was exclusively associative^{[ 57 ]} or both associative and dissociative but with significant associative character^{[ 62 ]} (and, thus, having constant or nearly constant crosslink density during reprocessing), special or complex circumstances were necessary to accelerate the dynamic exchanges to enable reprocessing via extrusion and recovery of properties. Taplan et al.^{[ 57 ]} demonstrated that their CANs containing vinylogous urethanes as dynamic covalent cross‐links required specific pendent amine and large para‐toluene sulfonic acid catalyst amounts to give relaxation times of <1 s and, thus, enable continuous reprocessing by extrusion. Serna et al.^{[ 62 ]} demonstrated that polythiourethane (PTU) CANs (with mostly associative and some dissociative dynamic covalent character) with twice the crosslink density of other PTU CANs counterintuitively underwent facile reprocessing by extrusion comparatively. A doubling in crosslink density was estimated to quadruple the rate of thiourethane associative dynamic covalent chemistry, enabling extrusion.^{[ 62 ]} Our CANs with exclusively dissociative cross‐links do not require additional accommodations such as catalyst incorporation or extremely high crosslink densities to allow for extrusion, lending credence to the idea that dissociative dynamic covalent chemistries may be advantageous for enabling eventual, large‐scale continuous (re)processing of CANs.

2.4. Self‐Healing

To investigate the potential self‐healing properties of our PE CANs, we prepared ∼0.65‐mm‐thick S₂ PE CAN sample films by compression molding and made ∼0.5‐mm‐thick crosscuts into the films using razor blades. Each crosscut film was then annealed at 130 °C without any external force. (We selected an annealing temperature of 130 °C, above the melt transition region of LDPE, to ensure the elimination of any PE crystalline domains and to provide sufficient mobility to the PE chain to accommodate self‐healing.) After 60 min at this T, the crosscut disappeared to the naked eye, leaving only barely discernible traces observed using a stereo microscope (see Figure 6a). We further used a 3D microscope to visualize the healing of the S₂ PE CAN as demonstrated in Figure 6b. At 130 °C, where dialkylamino disulfide bonds have some active dynamic character, the BTMA‐S₂ crosslinks can exchange at the interfaces and facilitate healing at the crack. Moreover, the reversible dissociation of BTMA‐S₂ crosslinks may temporarily release some PE chains from a crosslinked network topology. Once these chains gain sufficient mobility, they can diffuse across the interface and assist in crack healing.

Figure 6.

Images taken by a) a stereo microscope and b) a 3D microscope of an S₂ PE CAN film that was crosscut (∼0.50 mm deep) using a razor blade and healed at 130 °C for 60 min. c) An S₂ PE CAN tensile bar (∼0.65 mm thick) was cut in the middle using a razor blade. Glass slides (∼0.15 mm thick) were used as spacers to keep the cut consistent as ∼0.50 mm deep. d) Tensile stress‐strain curves of uncut S₂ PE CAN and PEX as well as cut S₂ PE CAN and PEX healed at 130 °C for various durations.

PEX sample films were also crosscut in a similar manner and annealed at 130 °C for 60 min. However, after annealing, they showed virtually no healing at the cut interfaces due to the inert nature of their permanent crosslinks (Figure S12, Supporting Information).

Tensile experiments were conducted to assess the extent of recovery of the original properties through self‐healing of a damaged sample. The stress‐strain curve of the 1st‐mold S₂ PE CAN sample (uncut) is represented by the green dashed curve in Figure 6d. Assisted by glass slide spacers (0.15 mm thick), S₂ PE CAN tensile bars (∼0.65 mm thick) were cut in the middle using a razor blade, leaving a consistent ∼0.50‐mm‐deep crack for each test specimen (see Figure 6c). Subsequently, the samples were placed flat in an oven and underwent self‐healing at 130 °C for various timeframes. As evidenced by the curves in Figure 6d for samples healed for 30, 45, and 60 min, the S₂ PE CAN exhibited progressively improved healing, as indicated by the increasing elongation at break, with complete healing occurring within 60 min.

Table 2 summarizes the Young's modulus, tensile strength, and elongation at break of the S₂ PE CAN samples. Within experimental uncertainty, the damaged S₂ PE CAN fully recovered its original tensile properties after 60 min of self‐healing at 130 °C, showcasing its outstanding, rapid self‐healing capability at a T only slightly above the LDPE melting point. The S_n PE CAN also exhibits self‐healing capability due to the dynamic chemistry of BTMA‐S_n, but it requires ∼120 min at 130 °C to fully recover the properties of the undamaged sample. (See Figure S13 and Table S5, Supporting Information.) Finally, after undergoing the same initial damage and then being annealed at 130 °C for 120 min, a PEX tensile sample snapped at the very initial stage of strain, displaying virtually no self‐healing due to its permanently crosslinked nature.

Table 2.

Summary of tensile properties of uncut S₂ PE CAN and S₂ PE CAN cut and then healed at 130 °C for various timeframes.

S2 PE CAN sample	Young's modulus [MPa]	Tensile strength [MPa]	Elongation at break [%]
Uncut	190 ± 25	14.0 ± 1.0	500 ± 10
Healed 30 min	160 ± 20	10.0 ± 0.5	170 ± 15
Healed 45 min	160 ± 20	10.0 ± 0.3	250 ± 20
Healed 60 min	170 ± 30	14.2 ± 1.0	510 ± 30

How do the self‐healing results for our S₂ PE CAN made from LDPE compare with the few literature results for self‐healing of PE CANs or ethylene‐based copolymer CANs? In 2021, starting from ethylene/octene random copolymers, Wang et al. reported ethylene‐based CANs that incorporated dioxaborolane dynamic covalent crosslinks.^{[ 12 ]} The self‐healing character was showcased in their study, but a long self‐healing time of up to 12 h was required to reach properties comparable to the original samples. In 2023, Li et al. prepared a PE CAN that relied on transesterification, achieving short healing times of as little as several minutes at 180 °C.^{[ 9 ]} However, to achieve full healing of the original tensile properties, Li et al. employed a mold at 180 °C, which applies substantial external pressure (2 MPa)^{[ 9 ]} to aid in healing, making it difficult to accept the description of their PE CAN as being self‐healing. Beyond CANs, a few other studies have reported intrinsic self‐healing of crosslinked PE leveraging non‐covalent dynamic bonds. Zou et al.^{[ 7 ]} reported on polar‐group‐functionalized ethylene‐based polyolefin crosslinked by metal coordination bonds, where a 12‐h healing time was needed to achieve full recovery of the original tensile properties. Kong et al.^{[ 8 ]} prepared PE/natural rubber composites with metal coordination bonds and hydrogen bonds serving as dynamic non‐covalent crosslinks. However, only 65–70% of tensile strength recovery was ultimately achieved in their study. In contrast, our S₂ PE CAN, enabled by fast dialkylamino disulfide dynamic chemistry via the incorporation of BTMA‐S₂, shows self‐healing with full recovery of properties within 60 min at a temperature only slightly above the melting temperature regime of LDPE and without the assistance of external force. Thus, to the best of our knowledge, this is the first demonstration of such rapid self‐healing of PE CANs without external pressure and allowing for the full recovery of properties within experimental uncertainty.

3. Conclusion

We achieved reprocessable, extrudable, and self‐healing PE CANs by incorporating BTMA as dynamic crosslinker into commercially available LDPE using a simple catalyst‐free, radical‐based reactive processing approach assisted by DCP as radical initiator. Two versions of BTMA, BTMA‐S₂, which is nearly exclusively disulfide‐based, and BTMA‐S_n, with a mixture of oligosulfide linkages, were used as dynamic covalent crosslinkers to produce S₂ PE CAN and S_n PE CAN, respectively. FTIR spectroscopy confirmed the grafting of BTMA onto the PE backbone during radical‐based reactive processing. Gel content experiments and dynamic mechanical analysis (DMA) revealed the robustly crosslinked nature of both S₂ and S_n PE CANs, achieving the same crosslink densities within a small uncertainty. Despite similar properties in the crosslinked state, the PE CANs exhibit drastically different dynamic behaviors. The S₂ PE CAN exhibited significantly faster stress relaxation than the S_n PE CAN, with ∼ 4.5 times smaller average relaxation times across a temperature range of 130 to 160 °C. This much faster relaxation is attributed to the faster dynamic chemistry of BTMA‐S₂, consistent with our previous findings in hexyl methacrylate‐based CANs.^{[ 41 ]} Moreover, the S₂ PE CAN exhibited short compression‐molding reprocessing times, with full recovery of crosslink density after only 5 min of molding at 160 °C. In contrast, the S_n PE CAN required 30 min at 160 °C for full crosslink density recovery. Both CANs are melt‐extrudable, resulting in extrudates with full recovery of crosslink density after extrusion. Both PE CANs are also self‐healable, as demonstrated by microscopy and tensile tests. Within only 60 min of self‐healing at 130 °C and without any external force, a damaged S₂ PE CAN exhibited full healing of a crack and 100% recovery of Young's modulus, tensile strength, and elongation at break within experimental error. The S_n PE CAN exhibited similarly excellent self‐healing, albeit with a longer full healing time of 120 min under the same conditions. In contrast, PEX exhibited no reprocessability or self‐healing capability. To our knowledge, this is the first report of PE CANs capable of self‐healing on these time scales without the use of external pressure. Our easily synthesized, rapidly reprocessable, and efficiently self‐healable PE CANs pose significant potential for a wide spectrum of applications, offering a viable substitution for traditional PE thermosets.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Chen B., Debsharma T., Fenimore L. M., Wang T., Chen Y., Purwanto N. S., Torkelson J. M., Rapidly Self‐Healable and Melt‐Extrudable Polyethylene Reprocessable Network Enabled with Dialkylamino Disulfide Dynamic Chemistry. Macromol. Rapid Commun. 2024, 45, 2400460. 10.1002/marc.202400460

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.