A broadly applicable cross-linker for aliphatic polymers containing C–H bonds

See allHide authors and affiliations

Science  15 Nov 2019:
Vol. 366, Issue 6467, pp. 875-878
DOI: 10.1126/science.aay6230

Cross-linking a range of alkyl polymers

Some alkyl polymers, such as polyethylene, can be cross-linked by using peroxides or high-energy radiation or through the addition of a radical forming agent. Others, like polypropylene, are likely to undergo chain scission, and this process tends to be uncontrolled in the distribution of the cross-links. Lepage et al. developed a widely applicable approach using bis-diazirine molecules as cross-linking agents (see the Perspective by de Zwart et al.). These molecules can be thermally or photochemically activated to form carbenes that readily insert into the polymer carbon-hydrogen bonds, thus leading to cross-linking. The bis-diazirine is nonexplosive, nonvolatile, and easily activated at relatively mild temperatures and thus could be used to fine-tune the properties of existing polymers through small chemical modifications.

Science, this issue p. 875; see also p. 800


Addition of molecular cross-links to polymers increases mechanical strength and improves corrosion resistance. However, it remains challenging to install cross-links in low-functionality macromolecules in a well-controlled manner. Typically, high-energy processes are required to generate highly reactive radicals in situ, allowing only limited control over the degree and type of cross-link. We rationally designed a bis-diazirine molecule whose decomposition into carbenes under mild and controllable conditions enables the cross-linking of essentially any organic polymer through double C–H activation. The utility of this molecule as a cross-linker was demonstrated for several diverse polymer substrates (including polypropylene, a low-functionality polymer of long-standing challenge to the field) and in applications including adhesion of low–surface-energy materials and the strengthening of polyethylene fabric.

Addition of cross-links to polymeric materials confers several important advantages to the final product. By converting a thermoplastic into a thermoset, a polymer’s impact resistance, tensile strength, and high-temperature performance are greatly enhanced, whereas material creep and unwanted thermal expansion are reduced (1). Cross-linked polymers also have increased resistance to solvents and electrical discharge as well as to chemical and biological effects. Although cross-linking can present challenges from the perspective of recyclability, it is advantageous in applications for which chemical, biological, or electrical degradation is a concern (2, 3). Cross-linked polyethylene, for example, is used for medical devices (4), insulation for electrical wires (5, 6), and containers for corrosive liquids (7). The principal disadvantage to cross-linking lies with an increase in brittleness, because the polymer chains are no longer free to slip across each other. Because these properties are highly correlated to the cross-link density, the control of the cross-linking process is key to the production of high-performance materials.

Cross-links can be established in polymers through various strategies. The most common method in the academic literature involves the use of copolymers wherein one of the monomer constituents incorporates a linkable fragment (1, 8). Alternatively, a monomer that has two functional groups may give rise to a linear prepolymer that can be thermally or photochemically cured (9). However, neither of the above strategies is appropriate when one needs to cross-link a polymer material that lacks functionality within its chemical structure. This includes important commodity plastics like polyethylene and polypropylene. Similarly, biomass-derived polymers (e.g., polylactic acid) and important biodegradable polymers (e.g., polycaprolactone) often lack any cross-linkable functional groups, even though they contain some functionality within their linear chains.

For these reasons, high-energy radical processes involving peroxides or electron or γ-irradiation are used industrially to produce cross-linked polyethylene (2, 3). However, the conditions required to initiate cross-linking through hydrogen abstraction are a limitation, and such methods are ineffective for polypropylene (1). The need to break a strong C–H bond (390 to 400 kJ/mol) in the vicinity of comparatively weaker C–C bonds (~350 kJ/mol) sets the stage for competing fragmentation and branching processes that can compromise the integrity of the material (fig. S1). Moreover, these methods do not allow for control over the type of molecular cross-link established, meaning that one cannot easily tune the mechanical properties of the final material.

We hypothesized that a superior cross-linking strategy could emerge from the use of low-barrier C–H insertions. Singlet carbenes are known to add directly to C–H, O–H, and N–H bonds through a concerted process that does not involve the generation of any new high-energy species (10). Carbene-generating reagents have been used for decades in the field of chemical biology to link small molecules to their protein targets, with the 3-trifluoromethyl-3H-diazirine motif (Fig. 1A) established as a particularly effective carbene precursor (11). Although a few records of multivalent diazirines exist, their occasional application to polymer cross-linking has remained limited to substrates with weak C–H bonds such as polyethylene glycol and highly functionalized materials in organic electronics (1215). The corresponding bis-azides (which function through nitrene insertion) have been somewhat better developed (16), but nitrenes are generally less reactive toward C–H insertion than carbenes and are more prone to undesirable rearrangement reactions (11). We envisioned that an optimally designed bis-diazirine could permit the cross-linking of unfunctionalized alkane polymers under mild conditions and without unwanted branching or fragmentation (Fig. 1B).

Fig. 1 A bis-diazirine strategy for polymer cross-linking.

(A) Mechanism of carbene formation from the light- or heat-promoted decomposition of diazirines, followed by C–H insertion. hν, ultraviolet light. (B) Cross-linking of nonfunctionalized polymers through double C–H insertion of bis-diazirines.

We began our search for an effective bis-diazirine cross-linker by preparing the known compound 1 (1215) and the pyridyl analog 2 (Fig. 2A). Both of these molecules were surprisingly volatile (fig. S2), and subsequent thermal analysis according to Yoshida’s correlations (17, 18) (Fig. 2B and eqs. S2 and S3) suggested that each possessed a substantial explosion risk. Although preliminary cross-linking trials demonstrated their ability to cross-link model substrates, both the volatility and the explosion risk negated the utility of these molecules for practical applications.

Fig. 2 Survey of cross-linkers 1 to 3.

(A) Compound structures and illustration of cyclohexane cross-linking. (B) Yoshida correlations showing that 1 and 2, but not 3, are potential explosion hazards. EP, explosive propagation; SS, shock sensitivity. (C) TGA and DSC analysis of 3, showing that the cross-linker is activated above 100°C and loses mass corresponding exactly to two equivalents of N2. (D) UV spectra collected during the photochemical and thermal cross-linking of cyclohexane with 3, showing that thermal initiation is faster and produces less diazo isomer. Asterisks indicate bands associated with each chromophore. (E) 1H and 19F NMR data for purified adduct 6, produced from cross-linking of cyclohexane with 3. 19F{1H} indicates a proton-decoupled experiment. (F) Physical properties for 1 to 3 and yields for purified cyclohexane adducts. Max temp., maximum temperature.

Stimulated by these observations, we designed and synthesized improved cross-linker 3 (Fig. 2A). Design features for 3 included (i) an increased molecular weight relative to 1 and 2, for reduced volatility and explosion risk; (ii) the absence of any labile C–O or C–N bonds (1215), which would limit the robustness of cross-linked products; (iii) the use of an electron-deficient linker para to the diazirine motif, for improved handling under ambient conditions (19); and (iv) the absence of any aliphatic C–H bonds, to reduce the risk of self-reaction.

Cross-linker 3 was found to have many desirable properties. It showed good solubility in a wide range of solvents (facilitating its dispersal into polymer matrices) and had a melting point conveniently just above room temperature (Fig. 2F), meaning that it could be handled either as a liquid or crystalline solid. Thermogravimetric analysis (TGA) revealed that it cleanly lost 2 equivalents of N2 upon gentle heating (Fig. 2C and eq. S1), whereas differential scanning calorimetry (DSC) and application of Yoshida’s correlations confirmed that 3 was not a likely explosive (Fig. 2B and table S1). Subsequent mechanical tests (movies S1 and S2) revealed no propensity for explosion with 3, at which point its synthesis was safely scaled up to afford multigram quantities (20).

Cross-linkers that are capable of inserting into the strong secondary (2°) C–H bonds of polyethylene should have equal or greater effectiveness against most other polymer substrates, because almost every other aliphatic polymer (aside from perhalogenated materials like Teflon) has C–H bonds of equal or lower strength (e.g., polypropylene or polystyrene) or contains O–H or N–H bonds that react more quickly with carbenes (e.g., polyalcohols or polyamides) (11). We therefore elected to first test 1 to 3 in models of polyethylene cross-linking, with the expectation that any successful cross-linkers identified in these trials would be broadly applicable to other systems. Seeking an initial substrate that would permit full spectroscopic characterization of cross-linked products, we first used cyclohexane as a molecular model for polyethylene, because it similarly contains only 2°C–H bonds.

The cross-linking of cyclohexane with 1 to 3 was studied under both thermal and photochemical activation conditions (Fig. 2A). Both long-wave ultraviolet (UV) irradiation (350 nm) and heating (110° or 140°C) were effective in activating all three bis-diazirines. A difference in the rate of photochemical conversion was observed: Whereas 3 was consumed within 1 hour, 1 and 2 required ~2 hours and ~4 hours, respectively, for complete conversion. In all cases, a small amount of linear diazo isomer (resulting from the known rearrangement of the diazirine group) was detected under photochemical conditions (Fig. 2D and figs. S5 to S11). These isomeric species persisted two to three times longer but can also participate in cross-linking (21). Under thermal activation at 140°C, the reaction was much faster (<20 min), and no linear diazo intermediate was observed.

Successful cross-linking was confirmed by careful isolation and characterization of products 4 to 6 (Fig. 2F). For all three adducts, 1H nuclear magnetic resonance (NMR) spectra showed a doublet of quartets at ~3.1 parts per million (ppm), and 19F NMR revealed a proton-coupled resonance at −63 ppm (coupling constant 3JH-F = 10 Hz), both indicating the presence of a hydrogen atom α to a trifluoromethyl group and at the foot of a new C(H)–C(H) bond (Fig. 2E). The modest isolated yields for 4 to 6, independent from the method of activation, should not be taken as an absolute measure of cross-linking efficacy, because several alternative cross-link structures (e.g., those in which 1, 2, or 3 oligomerize before cross-linking) would not be included within these yields. Indeed, observations of the spectroscopic signatures described above within the crude NMR spectra indicate that the overall C–H insertion efficacy in each case is >50% (20). Although the pyridine unit within 2 was added in the hopes of increasing cross-linking efficiency (19), this compound did not offer any advantages relative to 1 or 3.

With cross-linking of the molecular model substrate established, we turned our attention to cross-linking of relevant polymers, beginning with soluble, low–molecular-weight polyethylene (i.e., paraffin). Increasing amounts of bis-diazirine 3 [5 to 200 weight % (wt %)] (table S2) were readily dissolved in molten paraffin and activated at 110°C (figs. S15 and S16). Analysis by gel permeation chromatography (GPC) revealed a continuous increase in molecular weight with the amount of bis-diazirine added (Fig. 3A, blue arrow), providing evidence of cross-linking. Simultaneous UV detection confirmed that the chromophore from 3 was predominantly associated with higher-weight fractions (red arrow)—again, consistent with successful cross-linking. At 200 wt % of 3, cross-linking of paraffin afforded a tough gel with diminished solubility in tetrahydrofuran (THF) (fig. S16) (hence the decreased intensity in the GPC data), which supports the creation of a three-dimensional network. Subsequent studies also confirmed cross-linking in less-soluble, unbranched polyethylene (figs. S29 and S30).

Fig. 3 Cross-linking of soluble and insoluble polymers.

(A) Cross-linking of paraffin monitored by GPC. (B) Cross-linking of PDMS monitored by GPC. (C) Cross-linking of polypropylene increases the glass transition temperature (Tg) and decreases the fusion enthalpy (ΔHfus). (D) Structure of molecular control 7, used to validate mechanism. (E) Lap-shear data confirming adhesion for HDPE samples treated with 3 but not those treated with 7. Numbers indicate the total number of samples exhibiting sufficient adhesion for testing. (F) Drop-tower testing confirming reduced back-face signature and increased resistance to penetration upon cross-linking of UHMWPE fabric with 3. (G) Tear-testing data confirming increased mechanical strength for UHMWPE samples treated with 3 but not those treated with 7. Error bars correspond to standard deviations [N = 5 for (E) and N = 4 for (F); sample replicates for (G) are indicated in table S13]. *p < 0.05; **p < 0.01; ***p < 0.001. n.s., not statistically significant.

Cross-linker 3 was then tested on other polymer substrates. Experiments with polydimethylsiloxane (PDMS) provided similar results to those for paraffin: Low-viscosity PDMS exhibited an increased molecular weight upon thermal cross-linking with 5 wt % 3 (Fig. 3B), whereas high-viscosity PDMS was transformed into a rubbery solid with negligible solubility in THF (fig. S17). Photochemical cross-linking with 3 (20) likewise converted the liquid PDMS substrate into a stable gel (movie S3). Similar observations were made when cross-linking polycaprolactone (figs. S19 and S20), polystyrene (figs. S31 to S33), and polyisoprene (figs. S34 to S37). Polyvinyl alcohol cross-linked with increasing amounts of 3 progressively lost its aqueous solubility (figs. S24 and S25). The use of low concentrations of 3 for polyvinyl alcohol gave a product that floated atop the aqueous sample, whereas the use of higher concentrations gave a product that was heavier than water, demonstrating that cross-linker loading could control material density.

We next sought to demonstrate the efficacy of 3 for cross-linking commercial polypropylene samples. With increasing concentrations of cross-linker applied to low–molecular-weight polypropylene, we observed a monotonically increasing glass transition temperature (Tg) and decreased solubility (Fig. 3C and fig. S26). We also observed a consistent decrease in the enthalpy associated with the melting transition, while the actual Tm temperature remained constant. This is consistent with a model of polymer cross-linking where cross-linked regions of the polymer structure will be nonmelting [leading to a reduction in fusion enthalpy (ΔHfus)], whereas residual non–cross-linked regions will have a Tm similar to that of unmodified polypropylene. Even more profound effects were observed upon cross-linking of higher–molecular-weight polypropylene: The Tg was driven to a high of nearly room temperature, whereas the melting transition was almost completely lost at high–cross-linking density (fig. S27).

To demonstrate the utility of 3 for industrial processes, we were particularly interested to explore its effectiveness as an adhesive for high-density polyethylene (HDPE) and as a strengthening agent for polyethylene fabric. Adhesion of low–surface-energy materials like HDPE is an important problem in manufacturing (22). bis-Diazirine 3 can, in principle, connect two polymer surfaces through strong C–C bonds. We applied bis-diazirine 3 between bars of HDPE, cross-linked the assemblies at 110°C and then challenged them on a lap-shear experiment, along with appropriate controls (Fig. 3E). The cross-linked bars required far more load to be pulled apart than any of the controls, and analysis of separated samples by optical profilometry (figs. S40 and S41) indicated that residue derived from 3 was present on both faces, consistent with a cohesive rather than adhesive failure mechanism (23). Control samples prepared with no additives or with an equivalent weight (10 mg) of commercial Super Glue could not be measured, because they did not adhere. A set of samples coated with an equivalent weight of molecular control 7 (Fig. 3D) only barely adhered, proving that most of the adhesive force was due to cross-linking rather than simple surface modification. The use of a larger amount of 3 (25 mg) did not increase bonding strength.

To explore the effect of cross-linking ultrahigh–molecular-weight polyethylene (UHMWPE) fabric, we dissolved 3 in pentane and applied this solution to two different deniers of fabric (75 or 90 g/m2) from two different suppliers. The pentane was evaporated, and impregnated samples (or vehicle controls, treated with pentane but not 3) were cross-linked at 110°C. Samples treated with as low as 1 wt % 3 exhibited increased performance in both drop-tower and tear testing (Fig. 3, F and G). Increasing the cross-linker density to 10 wt % further improved material strength, but by a less marked increment. Surface sites on the UHMWPE fibers evidently become saturated, providing diminishing returns upon addition of more cross-linker. Fabric treated with molecular control 7 did not exhibit improved strength, once again confirming that the above results are due to authentic cross-linking and not surface modification. Cross-linking of aramid fabric likewise improved impact resistance, although the substantially increased rigidity in this case made the treated material easier to tear (figs. S43 to S46).

bis-Diazirine 3 is exceptionally stable (it can be recovered unchanged after dispersion in concentrated sulfuric acid at 70°C) but is readily activated by two complementary modes of activation: heating to >100°C or irradiation with ~350-nm light. Once activated, 3 is able to cross-link any aliphatic polymer containing C–H bonds, resulting in increased molecular weight, decreased solubility, increased Tg, and increased material strength: all well-known hallmarks of molecular cross-linking.

Supplementary Materials

Materials and Methods

NMR Spectra

Supplementary Text

Figs. S1 to S46

Tables S1 to S13

Movies S1 to S3

References and Notes

  1. See supplementary materials.
Acknowledgments: We thank the research groups of I. Manners and M. Moffit for assistance with TGA, DSC, and GPC. In particular, M.L.L. thanks L. MacFarlane for helpful discussions. We also thank P. Berrang, R. Mandau, and G. DiLabio for their collaboration on ongoing projects related to the use of the cross-linkers described here and R. Spontak for the low-MW polypropylene used in these experiments. Funding: Operating funds were provided by Mitacs Canada (grant IT11982) and Epic Ventures, Inc. Author contributions: J.E.W. conceived the study. M.L.L., C.S., C.L., and L.B. synthesized the cross-linkers and control compound. M.L.L., C.S., and L.B. carried out the cross-linking experiments. M.T. conducted the mechanical analyses with assistance from B.C. and supervision and infrastructure support from A.S.M. The manuscript was written by M.L.L. and J.E.W. with help from all authors. Competing interests: M.L.L., C.S., and J.E.W. are coauthors on U.S. Provisional Patent Application 62/839,062, which claims the use of cross-linkers described in this work. Data and materials availability: All data needed to reproduce the experiments described in the paper are available in the main text or the supplementary materials. Raw NMR data files for all synthesized compounds have been deposited with figshare (24).
View Abstract

Stay Connected to Science

Navigate This Article