Introduction

Polyacetylene, distinguished by its conductivity, has been a focal point of research since its discovery in the 1970s1,2,3,4,5,6,7,8,9, culminating in the 2000 Nobel Prize in Chemistry. However, the field has predominantly focused on 1,2-polyacetylene (1,2-PA) produced by 1,2-addition polymerization, while the polyacetylene polymerized by 1,1-addition of triple bonds has remained largely unexplored (Fig. 1a). Unlike 1,2-PA, 1,1-polyacetylene (1,1-PA) exhibits distinct 3D conformation and π-system distribution. These structural differences suggest that 1,1-PA could exhibit unique properties and applications as a polymer skeleton.

Despite the significant challenges associated with the synthesis of 1,1-PA, substantial progress has been made in synthesizing dendralenes by leveraging step-by-step approaches. For instance, Sherburn achieved the preparation of dendralenes with up to 12 vinylene units through cross-coupling of alkenyl halides and alkenyl metal species10,11,12 (Fig. 1b). Dong and Liu provided a homologation strategy for synthesizing similar cross-conjugated polyenes through an SNV mechanism by using (pseudo)alkenylidenes as feedstocks13 (Fig. 1c). Very recently, Zhu and Zheng developed a copper-catalyzed polymerization method by using in-situ-generated strained [3]cumulene as monomer14 (Fig. 1d).

Fig. 1: Background for the synthesis of 1,1-PA.
Fig. 1: Background for the synthesis of 1,1-PA.
Full size image

a Polyacetylene by 1,2-polymerization1,2,3,4,5,6,7,8,9 and 1,1-polymerization. b Synthesis of dendralenes by coupling reaction10,11,12. c Synthesis of dendralenes by SNV reaction13. d Synthesis of dendralenes from strained [3]cumulenes14. e Polyhomologation of methylidene sulfur ylide15. f Single step 1,1-carboboration of acetylene and terminal alkynes16,17,18, NHC N-heterocyclic carbene, BBN 9-borabicyclo[3.3.1]nonane. g Iterative 1,1-carboboration of acetylene.

With our long interest in acetylene chemistry, we sought to develop a polymerization strategy using cheap, abundant acetylene gas as the monomer to assemble 1,1-PA. Inspired by the polyhomologation reaction of methylidene sulfur ylide15 (Fig. 1e) and alkyne 1,1-carboboration process via tetracoordinated borane species in alkene synthesis reactions16,17,18,19,20,21,22,23,24,25,26 (Fig. 1f), we report here the 1,1-polymerization of acetylene through an iterative 1,1-carboboration process (The description of the reaction's discovery is included out of respect for the actual process of discovery and to emphasize that discoveries often stem from attention to detail. Providing a detailed account of the initial discovery would lead to unnecessary complications, as the reaction was first observed using Ni instead of Cd. Clearly, the mechanisms for Ni and Cd are likely significantly different.) (Fig. 1g). The reaction takes the C–B bond as the key reactive site and terminates upon protodeboration by transmetallation and protodemetalation. Acetylene is employed as a vinylidene equivalent, ensuring high atom- and step-economy. The key process of 1,1-carboboration in this design, which rearranges the alkynes to formal alkenylidenes, includes the deprotonation of acetylene, the formation of borane acetylide, the Lewis acid induced boron to carbon 1,2-shift (forming new alkenyl boranes), and protonation of the C-M bond. The formidable challenge of this work lies in facilitating the iterative occurrence of a single-step 1,1-carboboration to achieve a polymerization reaction. This is particularly arduous due to the exceptional instability of the C–B bond in the newly formed alkenyl borane under reaction conditions which include alkoxide, transition metal, and heat. Indeed, the freshly formed alkenyl boranes underwent direct protodeboration in the Lalic’s work with no polyene products reported16,17 (Fig. 1f). In addition, suppression of the thermodynamically more favored 1,2-polymerization also presents an additional challenge27,28.

Results

According to the proposed mechanism, selectivity between chain propagation and chain termination is controlled by competition between borane acetylide formation and transmetallation of C-B bonds (resulting in protodeboration). Given the strong metal-dependence of transmetallation rates, we screened transition and main group metals using BINAP (2,2’-Bis(diphenylphosphino)-1,1’-binaphthalene) or dtbbpy (4,4’-Di-tert-butyl-2,2’-bipyridyl) as standard phosphine or nitrogen ligands. However, this design proved to be hard. Initial trials showed only ds zone metals (Cu, Ag, Zn) yielded mono-vinylation products, while only Cd gave low degree of polymerization (DP ≤ 3) polymers. The better performance of ds zone metals is preliminarily attributed to their low migratory insertion tendency (preventing 1,2-PA formation; no reports exist for 1,2-PA synthesis with Cu/Ag/Zn/Cd catalysts to the best of our knowledge)1,2,3,4,5,6,7,8,9 and proper Lewis acidity strength (facilitating acetylene deprotonation and boron-to-carbon 1,2-shift).

Focusing on Cd, we evaluated ligands. All phosphine and NHC ligands evaluated gave DP < 3. Salen-type ligands were loosely coordinated to Cd and suffered from ligand dissociation during the synthesis. Nitrogen ligands performed best: ttbtpy (4,4’,4”-tri-tert-butyl-2,2’:6’,2”-terpyridine) and N,N’-dimethylcyclohexane-1,2-diamine (L1) achieved DP ≈ 10. Ligands with sulfur/arsenic showed no improvement. Among anionic ligands, 1,3-diketones outperformed halides, carboxylates, and multidentate ketones (Supplementary Table 2).

Taking all the results into account, we synthesized a hexacoordinated Cd complex with cyclohexan-1,2-diamine and hexafluoroacetylacetone ligands (cat I, Table 1). Its structure was confirmed by single-crystal diffraction (Deposition Number: 2433854), and synthesis was scalable to decagram quantities. Modifications of the cat I structure—removing the diamine cyclic ring, altering N-substituents, or changing hexafluoroacetylacetone derivatives—reduced yield and DP. Sterically hindered phenols (e.g., propofol, in situ deprotonated by NaOtBu to form conjugate bases) were essential for higher DP, likely by increasing steric repulsion and activation energy for transmetallation.

Table 1 Optimization of short-chain 1,1-PA synthesis†View full size image
Full size table

Finally, after systematically screening the other parameters, the optimal reaction conditions were established as 10 mol% cat I as the catalyst, 2.0 equivalents of propofol and 0.6 equivalent of NaOtBu as additives, and 2 bar (gauge pressure) acetylene as the monomer in a 0.1 M (based on boranes) 1,4-dioxane solution, reacted at 80 °C for 1 h. Boranes were generated in situ via alkene hydroboration. Under this optimized condition 60% yield and DP 15.4 were achieved with the model alkene substrate 1a. Due to the relatively small DP, products produced under this condition are denoted as short-chain 1,1-PA.

With the optimized reaction conditions in hand, the effects of key parameters are summarized in Table 1. Cu and Ag catalysts yielded no alkene products under standard conditions (Entries 2 and 3), although mono-vinylation products were observed in initial trials. In contrast, the Zn catalyst performed better than in preliminary screenings but still inferior to the Cd catalyst (Entry 4). A pressure of 2 bar was critical for good reproducibility (Entry 13). Further increases in pressure or decreases in temperature reduced short-chain 1,1-PA yields (Entries 10–12) and promoted long-chain 1,1-PA formation (discussed in Table 2).

Table 2 Optimization of long-chain 1,1-PA synthesisView full size image
Full size table

Figure 2 shows the 1H and 13C Nuclear Magnetic Resonance (NMR) spectra, and high-resolution mass spectrometry (HRMS) data of dodecyl 1,1-PA. The dendralene-type structure is distinguished by two distinct signal groups in the 13C NMR sp2 region (for 1,2-PA all sp2 carbons are similar and should give only one group of signals). The NMR spectra match the dendralenes synthesized by Sherburn12, confirming 1,1-PA’s identity. In addition, DEPT (Distortionless Enhancement by Polarization Transfer), HMBC (Heteronuclear Multiple Bond Correlation) and HSQC (Heteronuclear Single Quantum Coherence) spectra further support the dendralene-type constitution (Supplementary Figs. 1–5). The degree of polymerization was calculated by the vinylidene hydrogens (Fig. 2a in blue) integration divided by the alkyl -CH2- (Fig. 2a in black) integration which is 23.7 consistent with the HRMS data.

Fig. 2: Characterization of short-chain 1,1-PA.
Fig. 2: Characterization of short-chain 1,1-PA.
Full size image

a 1H NMR (CDCl3) spectrum of dodecyl 1,1-PA. b 13C NMR (CDCl3) spectrum of dodecyl 1,1-PA. c HRMS of dodecyl 1,1-PA.

Scope of short-chain 1,1-PAs

By changing the alkenes and alkynes used in the hydroboration (or other borane synthesis methods) a wide range of terminally functionalized 1,1-PAs were obtained in around 60% yield and DP about 14. Both aryl and aliphatic alkenes and all 1-substituted, 1,1-substituted and 1,2-substituted alkenes gave good yields (3a-3e). Chloromethyl, bromo, iodo, hydroxy, ester, amine, ferrocene and thiophene (3f-3m) are all compatible functional groups, which allows the 1,1-PA molecules to be further transformed. Phenyl borane, synthesized from the Grignard reagent, gave an all sp2 carbon polymer 3n, although with relatively lower yield. Terminal alkynes are not ideal substrates due to the formation of unwanted di-hydroborated byproducts during hydroboration29, however, by careful control of the hydroboration conditions a 22% yield and 20.7 DP were obtained (3o). Vinyl-BBN was also tested, but it led to no product. In contrast, an internal alkyne was a suitable substrate for this transformation and gave 3p in 57% yield. Bidirectional 1,1-PA made by using a α,ω-diene as the substrate is very unstable but was still able to be isolated in 53% yield (3q). Tri-directional 1,1-PAs synthesized from trienes formed membrane-like insoluble materials once the solvent was removed, which was proposed to be due to post-polymerization cross-linking10. Finally, ketones, aldehydes and the pyridine moiety are incompatible during the hydroboration. A carboxylic acid severely diminished the yield and DP (3s). The polydispersity index was calculated based on the time-of-flight mass spectra by a method introduced by McCullough30 (GPC failed to give Mn and PDI information for these small molecular weight products). The MS for all short-chain 1,1-PAs are provided in SI.

Acetylene is the only suitable monomer for this reaction, probably due to its small steric hindrance. Activated alkynes, such as phenyl acetylene and methyl propiolate, gave oligomers with DP < 5. No oligomers were detected with unactivated alkynes, such as 1-hexyne.

The reaction was scaled up to a 20 mmol scale with fewer equivalents of both catalyst and propofol additive (Fig. 3 bottom). The product was isolated either by chromatography, giving 3c in a 48% yield and 11.0 DP, or by recrystallization, giving a 4.9% yield and 31.4 DP. Compared with 1 mmol scale synthesis, the slightly decreased yield for 20 mmol scale synthesis (drop from 76% to 48%) was caused by the quick and severe heat release during polymerization which heated the reaction itself to 160 °C! The 1,1-PA products were stored in dilute toluene (0.1–0.2 mol/L) at 2 °C for at least two months without apparent decomposition (monitored by NMR). Storing pure 1,1-PAs should be avoided. Light, even 390 nm, did not decompose 1,1-PAs apparently.

Fig. 3
Fig. 3
Full size image

Synthesis of short-chain 1,1-PAs. †Ph-BBN was synthesized from PhMgBr and Cl-BBN and was isolated before use. ‡Hydroboration conditions: non-standard stoichiometry/temperature/handling. See SI. Isolated yields corrected for residual impurities/solvent.

Application of short-chain 1,1-PAs

Figure 4 demonstrates two applications for the 1,1-PAs. Telechelic decoration via a Diels-Alder reaction between ω-bromo substituted 1,1-PA 3g (DP = 31.4) and triazoledione was achieved with 45% yield. Byproducts formed by the Diels-Alder reaction in an internal region were also isolated but in only 7% yield. According to the difference in yields and given the significant difference in the number of the internal dienes and terminal dienes, the reaction rate of the terminal diene is about 100 times faster than that of the internal dienes. This “terminal first” phenomenon was also observed by Sherburn in dendralene chemistry but was much less apparent31. Furthermore, the wide range of functional groups that can be incorporated in the 1,1-PA makes the installation of 1,1-PA into other molecules easy. For example, estrone was decorated with 3g through a simple SN2 reaction in 72% yield.

Fig. 4
Fig. 4
Full size image

Application of short-chain 1,1-PAs. Isolated yields corrected for residual impurities/solvent.

Synthesis of long-chain 1,1-PAs

Having realized the synthesis of short-chain 1,1-PAs, we then proceeded to identify new conditions for the synthesis of 1,1-PAs with higher DP. When higher acetylene pressure (5-10 atm) was applied based on the standard conditions for the short-chain 1,1-PAs, we obtained a large amount of insoluble (No obvious dissolution of 0.5 mg of products (5a to 5e) was observed in 20 mL of common solvents (toluene, DCM, THF, or 1,2,4-trichlorobenzene), even after prolonged ultrasonication or heating at elevated temperatures (e.g., up to 145 °C for 1,2,4-trichlorobenzene).) white powdery precipitate, which exploded spontaneously during our first isolation attempt (Fig. 5). This white powder was shown, by the law of conservation of mass, to be polyacetylene. The solid state 13C NMR of the obtained polyacetylene 5a gives two peaks in the sp2 region (148 ppm and 118 ppm), which is in good agreement with the liquid state NMR of the short-chain 1,1-PA 3c (147-146 ppm and 117-113 ppm), demonstrating that they share the same dendralene-type constitution. The extremely weak aliphatic signal indicates a much larger degree of polymerization. In addition to instrumental analysis, 1,1-PAs and 1,2-PAs can simply be distinguished by the naked eye because 1,2-PAs are usually deep in color due to highly conjugated π system while 1,1-PAs are always white or colorless due to the incomplete conjugation of vinylidenes (will be discussed in the following section). These products, with their much larger DP, are long-chain 1,1-PAs. Characterization data is summarized in Supplementary Figs. 6–18 with 5a as the model compound.

Fig. 5: Long-chain 1,1-PAs.
Fig. 5: Long-chain 1,1-PAs.
Full size image

a Two photos of 5a before and after explosion. b Solid state 13C NMR of 5a. c Substrate scope: 5 mmol scale for 5a and 0.5 mmol scale for others. DP determined by end group elemental analysis. Isolated yields.

The influence of some parameters on long-chain 1,1-PA synthesis was investigated (Table 2). The best conditions are similar to those for the short-chain 1,1-PAs but at 40 °C, with 10 bar acetylene and ttbtpy + CdCl2 as the catalyst instead of cat I. The increased acetylene pressure was proposed to facilitate the formation of borane acetylide, which is the key intermediate of chain propagation. Under the best condition, a 100 mg alkene input led to a 780 mg 1,1-PA output with 100.2 DP (by Fe elemental analysis) and 59% yield (Entry 1). Temperatures higher than 40 °C gave more short-chain byproducts while temperatures lower than 40 °C caused low efficiency (Entries 2–6). Decreasing the acetylene pressure formed more short-chain 1,1-PA while maintaining a similar DP for long-chain 1,1-PA (Entry 7). Pressure >10 bar was not tested due to safety concerns. Halving the catalyst loading only led to a slight decrease in yield, but further reduction was unacceptable (Entries 8-9). DP information is not provided for all entries due to safety concerns when working with these highly explosive products during the sample digestion for elemental analysis.

Several examples, with a DP around 100 and a yield around 50%, are listed in Fig. 5. Though the thiophene moiety was successfully incorporated in the short-chain 1,1-PA synthesis, it significantly lowered the yield and DP of long-chain 1,1-PA. A 5 mmol scale synthesis of 5a was achieved with 4.81 g product output although the heat transfer problem still exists.

Property analysis of long-chain 1,1-PAs

For the stability of 1,1-PAs, a rule can be summarized as “the longer the 1,1-PA chain, the lower the stability”. For example, while the dendralene containing 5 vinylidene units can be separated by distillation in pure form11, short-chain 1,1-PAs need to be stored in dilute solution in refrigerators and long-chain 1,1-PAs prepared in this work explode easily. Preliminary experiments on the detonation of long-chain 1,1-PAs were conducted. Two modes of explosion were observed for 5a: oxidative explosion (Fig. 6a) and decomposition explosion (Fig. 6b). Under nitrogen, the decomposition explosion occurred at around 120 °C. Under air, 80 °C triggered the decomposition explosion. In fact, freshly synthesized long-chain 1,1-PAs is highly reactive and sometimes explode spontaneously under air without the need of any heating or shock (this property usually degenerates after 1 h of storage). In addition to heating, knocking with a hammer also triggered the decomposition explosion, whereas irradiation with an 8 W, 254 nm UV lamp did not. The oxidative explosion was triggered by high-temperature sources, such as open flame, and is very violent. The gaseous products of decomposition explosion were identified, by using gas chromatography, as acetylene and a trace amount of hydrogen. Quantitatively, explosion of 1.0 g of 5a produces about 100 mL acetylene, namely only about one tenth of the vinylidene unites gasifies into acetylene while the residues rearrange into aromatic compounds (determined by NMR, see Supplementary Fig. 19) like coal tar. Information regarding the safety procedures for handling long-chain 1,1-PA can be found in the Method section and SI.

Fig. 6: Explosion tests and SEM characterization of 1,1-PA.
Fig. 6: Explosion tests and SEM characterization of 1,1-PA.
Full size image

a Explosion of 100 mg 5a under air ignited by fire. b Explosion of 0.5 g 5a under nitrogen ignited by 120 °C heating. c SEM image of a particle of 5a. d SEM image of a thin layer on one particle of 5a.

SEM (Scanning Electron Microscopy) images reflected a loosely stacked multilayer structure of the long-chain 1,1-PA particles (Fig. 6c, d and Supplementary Figs.10–17). The thickness of one piece of long-chain 1,1-PA is about 500 nm. Two sharp peaks were detected by XRD (X-Ray Diffraction) which demonstrates that the long-chain 1,1-PA particles are at least partially crystalline (Supplementary Fig. S18).

According to calculations, 1,1-PA possesses a spiral like preferred conformation (see SI for details). As shown in Fig. 7, all internal vinylidenes only conjugate with one neighboring double bond while being nearly perpendicular to the other10 and every eight vinylidene units form a turn on the spiral. The energy of the spiral-1,1-PA with 24 vinylidene units is calculated to be 158 kcal·mol-1 higher than that of the trans-1,2-PA isomer, as the result of incomplete π conjugation.

Fig. 7
Fig. 7
Full size image

Computed preferential conformation of 1,1-PA. DFT calculations were performed at the B3LYP/6-31 G level of theory.

In summary, we have developed the 1,1-polymerization of acetylene via a Cd-catalyzed iterative 1,1-carboboration process. This work not only fills the long-standing void in alkyne polymerization chemistry but also introduces a class of materials with unique properties. The iterative 1,1-carboboration polymerization process provides ideas for the synthesis of other polymers.

Method

Representative method for synthesis of short-chain 1,1-PA 3c

A solution of BBN (23 mmol, 2.806 g) and 1-dodecene (20 mmol, 3.366 g) in anhydrous 1,4-dioxane (40 mL) was heated at 60 °C for 30 min, which was then transferred to high-pressure tank. Propofol (40 mmol, 7.131 g) and NaOtBu solution (12 mmol, 12 mL, 1 mol/L in THF) were then mixed with the borane solution. A syringe sealed with dotriacontane (m.p. = 70 °C) hung in the high pressure tank was loaded with a solution of cat I (1.2 mmol, 802.5 mg) in 10 mL of 1,4-dioxane. The system was filled with 2 bar acetylene gas, stirred at r.t. for 1 h, and then heated at 80 °C for 1 h. The product was isolated by column chromatography and stored in solution (9.6 mmol, 48% yield).

Representative method for synthesis of long-chain 1,1-PA 5a

A suspension of CdCl2 (0.509 mmol, 93.2 mg) and ttbtpy (0.560 mmol, 224.7 mg) in 40 mL of 1,4-dioxane was heated at 70 °C for 30 min. A solution of BBN (5.85 mmol, 713.7 mg) and 1-octene (5.09 mmol, 570.8 mg) in anhydrous 1,4-dioxane (5 mL) was heated at 60 °C for 30 min. Propofol (8.44 mmol, 1.505 g) and NaOtBu solution (3.05 mmol, 3.05 mL, 1 mol/L in THF) were then mixed with the borane solution. The above catalyst suspension and borane solution were transferred to the high-pressure tank. The system was filled with 10 bar acetylene gas and heated at 40 °C for 2 h. The white powdery product was isolated by filtration (4.81 g).

All molecular structure calculation was based on Gaussian 09W software with B3LYP/6-31G level of theory without any non-default settings.

Synthetic procedures for all compounds, experimental details, materials, Gaussian calculation details, and equipments are provided in the Supplementary Information.

Safety considerations

  1. 1.

    Long-chain 1,1-PA can be explosive. Appropriate personal protective equipment, including safety goggles, a certified respirator, gloves, and a lab coat, must be worn properly. Work should be conducted in a fume hood.

  2. 2.

    Reactor safety: The heat released by the reaction causes a significant increase in internal pressure during polymerization. Therefore, the reactor vessel must be rated for a maximum operating pressure at least 20 times higher than the initial acetylene pressure to accommodate pressure surges and the high-pressure reactor must be equipped with a pressure relief valve.

  3. 3.

    Explosion sensitivity of long-chain 1,1-PA:

    1. a)

      Critical factor: The primary factor determining explosion risk is the age/freshness of the sample and its exposure to air, not the use of non-metal spatulas, ground-glass joints, or light exposure.

    2. b)

      Fresh material (≤1 day old, especially immediately after filtration): Requires rigorous exclusion of air contact (e.g., via nitrogen purging or glovebox handling) to prevent self-oxidation. Heat accumulation from slow oxidation can sometimes trigger spontaneous explosion.

    3. c)

      Aged material (>1 day old) or material stored in a glovebox: Exhibits substantially reduced explosion sensitivity. Standard handling procedures (e.g., using metal spatulas for weighing), excluding grinding, are generally safe.

  4. 4.

    Despite its high impact sensitivity, long-chain 1,1-PA possesses characteristics enabling safe handling with proper precautions:

    1. a)

      Low temperature during decomposition explosion: Insufficient to ignite directly contacting materials like paper towels.

    2. b)

      Low explosive power: Insufficient to shatter non-sealed glass vials containing the sample. (This is only true for decomposition explosion. The oxidation explosion triggered by open flames is very violent.)

    3. c)

      Rapid activity loss: All documented accidental explosions occurred within 30 min of preparation.

  5. 5.

    Waste disposal: Waste 1,1-PA should be diluted with silica gel or sand and disposed of as solid waste.

  6. 6.

    Comprehensive step-by-step operating procedures, including photographs, are provided in the SI to maximize safety.