Cyclic Polyacetylene

Abstract

Cyclic polyacetylene (c-PA) or [∞]annulene is synthesized for the first time. Unique to the cyclic structure and evidence for its topology, the c-PA contains >99% trans-double bonds, even when synthesized at −94 °C. In contrast, linear polyacetylene (l-PA) contains cis-double bonds that requires additional thermolysis at >150 °C to isomerize. Absolute evidence for the cyclic topology comes from AFM images of bottlebrush derivatives generated from soluble c-PA. The tungsten catalyst demonstrates an acetylene polymerization activity of 620,000 g/mol/h, and the c-PA produced has low defect density (<1%, ¹³C NMR) and comparatively low free electrons (e⁻/14,000 (±1,400) carbons, EPR), making it one of the purest forms of PA. Raman spectroscopy is consistent with predominantly trans double bonds and high conjugation exhibiting a low energy absorption at 1458 cm⁻¹. Importantly, the high activity of the catalyst allows for the synthesis of temporarily soluble c-PA, and its UV-vis spectrum exhibits a broad absorption profile with a red-shifted λ_max at 670 nm. Lustrous thin films of c-PA acetylene, when doped with I₂, are conductive (398 (±76) Ω⁻¹cm⁻¹).

Introduction:

Credited with formulating the structure of benzene as a cyclic six-membered ring possessing alternating single and double bonds, Kekulé forever altered the way chemists think about the composition of molecules.^1,2 Kekulé used the term “affinity” to describe what we know today as bonds. In the following translated quote,² he expresses the composition of benzene and the nature of alternating single and double bonds:

“…they combine alternately by one and by two affinities. In fact, six carbon atoms by combining according to this law of symmetry will give a group, which, considered as a chain open, will still have eight unsaturated affinities. If we admit, on the contrary, that the two atoms which terminate this chain combine with each other, we will have a closed chain…”

This intellectual leap was significant since an understanding of the composition of atoms and the existence of electrons was not available to Kekulé at the time. It was not until 1929 that Lonsdale elucidated the structure,³ and that Hückel in 1931,⁴ formalized the concept that cyclic rings having 4n+2π electrons exhibit aromatic behavior. With the goal of exploiting the potential properties imparted by delocalized electrons, chemists have sought to synthesize large rings.⁵ Even after 150 years, new principles of this composition of matter continue to emerge, including recent discoveries of global 3D aromaticity.⁶

[n]Annulenes are ring compounds with the same empirical formula as benzene, C_2mH_2m, where n is the number of carbon atoms, 2m. Sodenheimer isolated [18]annulene in 1959⁷ and [30]annulene in 1960.^8,9 Synthetically impressive, it is remarkable that [30]annulene is isolable. Unable to challenge the limits of aromaticity, ring size, and the properties of large aromatic rings of basic formula (CH)_n, no larger annulene rings were reported over the past 60 years, though theoretical treatments of n = 42, 54, and 66 have been performed.¹⁰ Using Hartree-Fock and Density Functional Theory, Chi and Kertesz¹¹ determined that when the HOMO-LUMO gap decreases, a transition from delocalized to localized structures occurs at 4n + 2 = 30. An important property expected for an infinitely large annulene is electrical conductivity. [∞]Annulene (where n > 100) should be semi-conducting, just like its famous linear derivative polyacetylene, first synthesized by Natta in 1958.¹² Extensively studied,¹³ linear polyacetylene (l-PA) is a remarkable compound. Questioned recently, the true nature of the composition of the “as synthesized”¹⁴ linear polyacetylene remains.¹⁵ Berets and Smith first reported the doping of PA in 1968,¹⁶ but it was Shirakawa and Ikeda¹⁷ who synthesized free-standing films of cis-transoid polyacetylene at −78 °C using Ti(O^tBu)₄/AlEt₃. Heeger and MacDiarmid et al. demonstrated that its p- and n-doped derivatives exhibit high electrical conductivity.¹⁸ These collective discoveries paved the way for modern flexible electronics that exploit the semi-conducting properties of polymers containing delocalized electrons across alternating single and double bonds that now find application in sensors, electrochromic devices, optical and photonic devices, organic light emitting diodes (OLEDs), and photovoltaic (PV) cells.^19,20 With the potential of increasing the conductivity by orders of magnitude when doped, an infinitely large pure [∞]annulene should be semi-conducting. Herein, we report a number of important discoveries, the first and foremost being the catalytic synthesis of [∞]annulene or cyclic polyacetylene (c-PA) according to Fig. 2A. In addition, the high efficiency of catalyst 1, with measured initial rates of 620,000 g/mol_cat/h, allow the use of ultra-low (ca. ppm) catalyst loadings that produce highly conjugated c-PA with low (<1% by NMR) defects, low free electron density (EPR), and easy post-synthetic work-up. Free standing films, bulk polymer, and transiently soluble forms of c-PA are now attainable using catalyst 1.

Fig. 2.

Synthesis of c-PA. (A) Reaction scheme for the polymerization of acetylene to generate cyclic trans-transoid polyacetylene with catalyst 1. (B) Photograph of synthesized c-PA as a free standing film at 25 °C.

Results:

Evidence for a Cyclic Topology

Though newly developed catalytic ring-expansion^21–28 methods are rapidly growing the field, cyclic polymers are inherently difficult to synthesize. We recently reported the catalytic synthesis of cyclic polymers from alkynes.²⁹ Employing size-exclusion chromatography, dynamic light scattering, static light scattering, intrinsic viscosity, ozonolysis, rheology, and most recently, AFM images of cyclic bottlebrush polymers on a surface,³⁰ we confirmed the cyclic topology of the polymers produced using catalyst 1.^29,31–34 In 2016, c-PA was synthesized in our labs by bubbling acetylene into a solution of catalyst 1 according to Fig. 2A. However, due to the polymers’ well-known insolubility, air sensitivity, and lack of a melting point below its decomposition temperature, confirming a cyclic PA topology presented a significant challenge. In its free-standing film form, c-PA (Fig. 2B) has the same lustrous silvery appearance as l-PA. There is, however, an important difference between l-PA and the c-PA produced by catalyst 1: l-PA, as synthesized using Ti(OBu)₄/AlEt₃ at −78 °C, results in cis-transoid double bonds.¹⁷ The cis-double bonds form as a consequence of the mechanism of a coordinated acetylene insertion into a growing polymer chain. In contrast, catalyst 1 produces c-PA with >99% trans double bonds even at −94 °C, a temperature too low to induce isomerization.^35,36,37 Vetting of the mechanism of ring expansion with catalyst 1 continues in our laboratories, but the observation of pure trans c-PA appears to be unique to its cyclic topology (vide infra).

Heating cis l-PA at 180 °C isomerizes the polymer to the trans-transoid form.¹⁷ Figure 3 illustrates how the high barrier to isomerization is a consequence of having to rotate two C=C double bonds. Small annulenes undergo bond shift and configuration changes with barriers that permit their measurement in solution. For [16]annulene above −50 °C, planar bond shifting, degenerate conformational change, and configuration changes occur rapidly, resulting in a single resonance for the ring protons in its ¹H NMR spectrum.³⁸ Adding to this rapid exchange, tunneling of the 16-carbon atoms was recently proposed.³⁹ Having access to a low barrier π-bond shift differentiates linear and cyclic unsaturated compounds. Another important point is that as annulenes increase in size, the conformational constraint relaxes, and the double bonds adopt the thermodynamically more stable trans-form. For example, proposed for [30]-annulene (Fig 1), it has mainly trans bonds in its lowest energy configurations.^8,9 In short, cyclic polyacetylene can only exist in its trans-transoid configuration, an outcome consistent with >99% trans double bonds in the c-PA produced with catalyst 1.

Fig. 3.

Top: Isomerization of linear cis-transoid polyacetylene to trans-transoid polyacetylene. Bottom: Configuration change of [16]-annulene.

Fig. 1.

Historical development of annulenes.

Observation of an all trans-configuration is strong support for a cyclic topology, but it is not absolute evidence. Seeking to confirm the cyclic topology, we can exploit the properties of catalyst 1 that are distinctly different from conventional polymerization catalysts. Catalyst 1 permits polymerization at low temperatures and low concentrations, leading to temporarily soluble c-PA. Partial bromination of the backbone double bonds prior to precipitation followed by grafting from via atom-transfer radical polymerization (ATRP) produces cyclic bottlebrush derivatives, according to Fig. 4A. Figures 4B–4D depict AFM images of the cyclic bottlebrush samples cast onto a mica plate surface. Timing, concentration, equivalents of Br₂, and extent of conversion during ATRP are all crucial for creating cyclic bottlebrushes that conclusively reveal their cyclic topology (see ESI for synthetic details). Importantly, cyclic structures result from multiple samples using different conditions, and the structures appear throughout the surface. These are the first images of post-polymerization derivatives of polyacetylene, regardless of topology.

Fig. 4.

(A) Reaction scheme for the synthesis of cyclic bottlebrushes via partial bromination of temporarily soluble c-PA followed by grafting from via ATRP with styrene. (B) AFM height (left) and phase (right) images of cyclic bottlebrushes within a 5.0×5.0 μm range. (C) Focused AFM height (left) and phase (right) images of two cyclic bottlebrushes within a 2.0×2.0 μm range. (D) Focused AFM height (top) and phase (bottom) images of one single cyclic bottlebrush within a 900.0×900.0 nm range.

Spectroscopic characterization of trans-transoid cyclic polyacetylene (c-PA)

Catalyst 1 enables the synthesis of bulk, thin films, and as mentioned, temporarily soluble c-PA. For example, injecting a toluene solution (5 mg/mL) of catalyst 1 (400 μL) into an acetylene-saturated toluene solution (10 mL) at temperatures ranging from −94 to 65 °C produces c-PA as a black viscous gel that converts to a black powder after removing all volatiles and washing with pentane and THF. The formation of c-PA is rapid upon exposing acetylene to catalyst 1, with a measured initial activity of 620,000 g/mol_cat/h (see Supplementary Fig. 5) at 25 °C. Alternatively, exposing a vial coated with toluene solution (400 μL) containing catalyst 1 (5 mg/mL) with an atmosphere of acetylene for 15 min at different temperatures ranging from −78 °C to 65 °C produces free standing flexible and silvery films of c-PA (see Supplementary Fig. 2). Another approach involves injecting a dilute THF solution of 1 (25 μL, 1.0 mg/mL) into a highly dilute acetylene/THF solution (3.00 mL, 0.03 mg/mL) at −20 °C to produce temporarily soluble c-PA. At all temperatures ranging from −94 °C to 65 °C, all the methods, exclusively yield polyacetylene comprising >99% trans double bonds, as confirmed by IR spectroscopy (Fig. 5, Supplementary Fig. 1, and Supplementary Fig. 3). Figure 5 depicts the remarkably clean IR spectra of thin films of c-PA produced at −78 °C (blue), 25 °C (orange), and 65 °C (black). The significant features are the strong trans =C-H out-of-plane bending at 1010 cm⁻¹ and the weak =C-H stretching vibration at 3010 cm⁻¹. Importantly, the IR spectrum of c-PA does not exhibit the end-group terminal CH₂ or CH₃ stretches at 1458 cm⁻¹ and 1378 cm⁻¹ that are evident in the linear samples prepared using Ti(O^tBu)₄/AlEt₃ (see Supplementary Fig. 7).

Fig. 5.

Infrared spectra of c-PA films produced at −78 °C (blue), 25 °C (orange), and 65 °C (black).

Raman, absorption, CP MAS solid state ¹³C NMR, EPR, and morphology

Related to the discussion of conjugation and the very nature of the alternating double and single bonds in l-PA, Hudson¹⁵ suggests the true nature of linear trans-PA is not one of an extended structure but rather finite segments of polyenes. Hudson argues that the zero-point energy vibration is above the transition state barrier separating the Pierls distortion and leading to alternating single and double bonds.¹⁵ Thus, if polyacetylene could truly be synthesized, it would be metallic. Linear finite polyenes containing chain-ends and crosslinks give rise to Raman data showing C-C and C=C absorptions. An interesting phenomenon occurs for cyclic aromatics. As mentioned, for [n]annulenes where n ≥ 30, an alternating structure of single and double bonds is strongly favored. Raman spectra of thin films of c-PA feature C-C and C=C stretching vibrations at 1067 cm⁻¹ and 1458 cm⁻¹, respectively (excitation wavelength 785 nm). The Raman frequency of the C=C stretch in trans polyacetylene is markedly dependent on the excitation wavelength (Fig 6),⁴⁰ a result interpreted as a combination of the resonance Raman effect and the dependence of vibrational frequency and optical gap on the conjugation length of polyenes.^40–42 The dependence also suggests that the Raman signals arise from segments of different conjugation lengths in trans PA.⁴⁰ Samples of l-PA are challenging to characterize due to inherent polydispersity,⁴³ inhomogeneity,⁴⁴ and coexistence of ordered and disordered phases.^45,46

Fig. 6.

Raman spectra of trans-transoid cyclic polyacetylene films prepared from catalyst 1 at 25 °C and trans-transoid linear polyacetylene prepared from Ti(O^tBu)₄/AlEt₃ at 25 °C then isomerized at 180 °C. (A) Raman spectra of c-PA (blue) and l-PA (orange) excited at 785 nm. (B) Raman spectra of c-PA excited at 785 nm (blue), 633 nm (orange), and 532 nm (black). (C) Raman spectra of l-PA excited at 785nm (blue), 633 nm (orange), and 532 nm (black).

Several different methods, including extrapolations from oligomers^42,47–49 and calculations based on different theories,^41,50 were employed in previous studies to correlate the Raman frequencies and trans PA conjugation lengths. Substituting the value of 1458 cm⁻¹ from c-PA into those equations^47–50 yields average conjugation lengths and lower limits between 30–43 bonds, including ~40 based on the approach established by Schrock and coworkers.⁴⁸ However, using Lichtmann, Fitchen, and Temkin’s equation,⁴² the estimated conjugation length of c-PA is 114, similar to the linear PA produced by Xia et al. that exhibits a transition at 1463 cm⁻¹.⁵¹ According to Ikeda and Shirakawa et al., a C=C stretching vibration at 1466 cm⁻¹ (excitation wavelength 676.4 nm) corresponds to a conjugation length upper limit of ≥100 as well.⁴⁰ Considering that samples produced by 1 exhibit a red-shifted Raman absorption at 1458 cm⁻¹, an estimate for the upper limit of 100 conjugated units is not unreasonable. Eexciting c-PA at 532 nm does not produce a signal, while exciting at 633 nm results in significantly reduced absorption with minimal change in frequency (1460 cm⁻¹) or line shape (Fig. 6B), which is indicative of a lower limit on the conjugation.

A UV-Vis spectrum of c-PA reveals a broad absorption with a λ_max at 670 nm and allows for temporal monitoring of the polymerization in solution (Fig. 7). Notoriously insoluble, only a few experiments have directly measured the UV-Vis absorption of soluble PA. Xia et al. produced PA via sonication with the polymer exhibiting a λ_max at 636 nm,⁵¹ and most recently Choi et al. produced a related conjugated polyacetylene with a λ_max at 651 nm.⁵² Grubbs’ ring opening of substituted-cyclooctatetraene produces substituted PA with a range of λ_max between 302 – 634 nm.⁵³ For perspective, Grubbs⁵⁴ and Schrock^47,55 were able to produce oligomers of PA and observed absorptions between 355 – 540 nm with distinct transitions that correspond to 10–20 double bonds. c-PA synthesized with 1 at −20 °C under dilute conditions is soluble for ~1 h. The broadness of the UV-vis absorption indicates a distribution of segments with different conjugation lengths.^56,57 In the first 1 h, the increase in both λ_max and intensity suggests the continuous formation of c-PA and a corresponding increase in conjugation length. After 47 min, the peak intensity starts to fluctuate and gradually decreases after 67 min as c-PA precipitates (Fig 7B). c-PA produced by catalyst 1 exhibits the longest solution phase absorption recorded with a λ_max reaching 670 nm and a tail extending to 950 nm. The 950 nm absorption tail is unusual for PA, and while its origin is not clear at this time, it may be inherent to the cyclic topology. Xia observed a blue shift when PA precipitates from solution.⁵¹ However, as c-PA starts to precipitate after 67 min, the λ_max remains at ~670 nm with no significant hypsochromic shift as the intensity decreases, suggesting c-PA maintains its high conjugation as it precipitates from solution. The absorption profile of c-PA is broad, but not unusual, as other experimental and theoretical studies also report a broad profile.^40,58,59 Eliminating the possibility that the broad absorption is due to scattering from particles suspended in solution, a UV-Vis spectrum of a thin film of c-PA exhibits the same absorption profile (see Supplementary Fig. 8), thus confirming the absorption is molecular in origin.

Fig. 7.

(A) UV-Vis spectra recorded during polymerization of acetylene with 1 in THF at −20 °C for 5 h. (B) Plot of λ_max and intensity versus polymerization time.

In agreement with high trans content of >99%, the ¹³C CP-MAS NMR spectrum of c-PA in Fig. 8A exhibits a resonance centered at 136.49 ppm for trans (CH)_n.⁶⁰ Consistent with a cyclic topology, the spectrum does not exhibit the CH₂ or CH₃ chain end group resonances at 34.47 and 14.20 ppm,⁶¹ respectively, that are plainly evident in l-PA. The spectrum of l-PA contains a broad resonance between 15 and 70 ppm; attributed by others to defects, this resonance can arise from cross-linking or saturated bonds. Adopting a slower spinning speed, the resonances that overlap at 27.34 ppm with the spinning sideband separate and are observable in Supplementary Fig. 11. The minor impurity in the c-PA sample that resonates at 33.09 and 21.09 ppm can be due to saturation (−CH₂-CH=CH). As a rough estimate and evidence for low defects, the integration of sp³ hybridized carbons is <1% in c-PA compared to 24% in l-PA.

Fig. 8.

(A) CP-MAS ¹³C NMR spectra of trans c-PA (blue) and trans l-PA (orange) obtained at 25 °C. (B) MAS-DNP¹³C NMR spectra of trans c-PA (blue) and trans l-PA (orange) obtained at 25 °C.

Fig. 8B presents preliminary MAS-DNP ¹³C NMR (DNP = dynamic nuclear polarization)⁶² data revealing a clear difference between the two topologies of l-PA and c-PA. For the trans-C=C bond at 136 ppm, c-PA exhibits a significantly broader resonance (width at half-height ~ 5 ppm for c-PA vs 2 ppm for l-PA). This line broadening is independent of the measurement temperature. For model PA compounds, Schrock demonstrated that ¹³C resonances for C=C bonds in cis and trans l-PA rapidly converge to a value close to the one found for cis-l-PA (within ~ 5 units from the chain ends), indicating that the broadening in c-PA is not due to a distribution of small linear fragments.⁶³ In ¹³C CP-MAS-NMR experiments, Kaplan demonstrated that the trans-PA signal is sensitive to microstructural sequencing, while the cis is not.⁶⁴ During isomerization from cis l-PA to trans l-PA, the trans l-PA signal line width narrows with increasing trans content. Clearly, c-PA has less structural defects than l-PA: the median length of trans sequences in c-PA is at least 49, a 10-fold increase over the l-PA sample (determined from the ratio of the trans-PA signal vs cis-PA and sp³-hybridized carbons). The observed broadening for the more perfect c-PA sample is incompatible with a linear topology and points towards its cyclic topology, though more studies are required to confirm this phenomenon.

Solitons (neutral defects) are inherent in trans-PA regardless of topology due to two energetically equivalent states arising from bond alternation and give rise to X-band EPR signals that are well-studied.^65–67 Solitons are not present in thin films comprising linear cis-PA,⁶⁸ since bond alternation results in inequivalent states (trans-cisoid and cis-transoid). However, upon heating above 150 °C, cis-PA isomerizes to trans-PA, and an EPR signal develops that is characteristic of mobile free π-electrons (solitons). Figure 9 depicts the EPR spectra (26.9 °C) for trans l-PA and c-PA. Agreeing well with earlier studies on trans l-PA,^65,67,69 both l-PA and c-PA exhibit single narrow Lorentzian line shapes with a g factor of 2.00055, close to the free electron value. The peak-to-peak linewidth ΔH_pp of l-PA and c-PA are 1.58(±0.03) and 1.54(±0.04) G, respectively, a typical width for the trans-isomer of polyacetylene at ambient temperature.⁶⁷ Quantitative measurement relative to TEMPO reveals l-PA has one free electron for every 3,600 (±300) carbons, or 1.28×10¹⁹ e⁻/g, a close fit to previous studies.⁶⁵ In contrast, c-PA synthesized at 25 °C contains only one free electron for every 11,600 (±900) carbons (or 3.98×10¹⁸ e⁻/g).

Fig. 9.

EPR spectra (X-band) of trans c-PA (blue) and trans l-PA (orange) obtained at 26.9 °C with a g factor of 2.00055.

One exciting feature of c-PA produced by catalyst 1 in its all-trans form at low temperatures is the opportunity to measure the inherent soliton concentration without the need for heating, an experiment never performed before. Remarkably, an EPR spectrum of a c-PA sample prepared at −78 °C and kept cold reveals only one free electron for every 119,100 carbons. Moreover, demonstrating the difference between the two topologies, in parallel conditions, l-PA and c-PA were heated at 150 °C for 5 min before acquiring the EPR spectra. The linear sample results in 1 e⁻/3,600(±300) carbons; however, heating c-PA only results in 1 e⁻/6,000 (±500) carbons, or ~40% reduction in solitons. One explanation for this difference centers on the fact that chain ends repel solitons.^70,71 Without chain ends, solitons are free to diffuse over the entire length of a cyclic polymer, thus effectively lowering the inherent soliton concentration. Unfortunately, a direct comparison here is not possible since the composition of the two polymers (molecular weight and Ð) is unmeasurable. The preliminary result of a 40% reduction in soliton concentration for c-PA vs l-Pa is intriguing and will be the subject of future interrogations.

Not expected to change, the solid-state morphology of c-PA is very similar to that of l-PA.¹³ Figure 10 depicts the scanning electron microscopy (SEM) images of the lustrous and dull sides of the thin films prepared with catalyst 1 at 25 °C. Presented at the same scale, it is obvious that the dull side of the film is loosely packed with larger fibrils, whereas, the lustrous side contains dense fibrils with ~0.33 μm width. Energy Dispersive X-ray Spectroscopy (EDS) performed on c-PA reveals very low metal contamination (between 0.39 – 0.85 wt% W), whereas the l-PA sample contains 2.07 wt% Ti and 5.35 wt% Al (Supplementary Table 3,4,5 and Supplementary Table 6). Despite the high quality of the c-PA produced with catalyst 1, the stability of c-PA is identical to linear samples. Exposed to air, both l-PA and c-PA exhibit IR absorptions attributable to oxidation at the same rate, and thermal gravimetric analysis indicates both have similar onset decomposition temperatures of 361 °C (Supplementary Fig. 21 and 22).

Fig. 10.

SEM images of c-PA (A) SEM image of c-PA on the lustrous side (B) SEM image of c-PA on the dull side.

Electrical Conductivity

The effect of chain ends on electrical resistivity and soliton formation in polyacetylene has been discussed in the literature.^70,71 Unfortunately, probing chain end effects requires l-PA and c-PA featuring similar defects, conjugation lengths/distribution, and impurity content that is currently not possible. However, I₂ doped samples of c-PA exhibit conductivities at the higher end of the range (without chain alignment)^72,73 observed for l-PA prepared via Shirakawa’s method. Table 1 lists the results of several electrical resistance measurements.

Table 1.

Conductivity of c-PA films with different doping percentage of I₂.

	Doping time (h)	Composition	Conductivity (Ω−1cm−1)
trans c-PA-1	0	(CH)n	5.05 (±1.65) ×10−6
trans c-PA-2	1	(CHI0.16)n	234 (±50)
trans c-PA-3	3	(CHI0.20)n	398 (±76)
trans c-PA-4	24	(CHI0.23)n	363 (±58)
trans l-PA-1	0	(CH)n	3.02 (±1.50) ×10−6
trans l-PA-1	3	(CHI0.19)n	179 (±21)

Kekulé formulated the composition of aromatic compounds (benzene) without modern spectroscopy. Absolute proof of the composition of benzene was not available, but importantly, Kekulé’s hypothesis fit the available data. The insolubility preventing typical solution-phase size-exclusion chromatography and dynamic light scattering characterization, and the reactive nature of c-PA created a similar challenge to provide absolute proof of its cyclic topology. In a way, characterizing c-PA presented some of the same challenges Kekulé confronted 150 years ago. Not available to Kekulé, AFM images of bottlebrush derivatives of c-PA provide irrefutable evidence for the cyclic topology. However, evidence supporting c-PA comes from other methods and data too, in particular, the unusual all-trans form of the polymer. Moreover, for all other polyacetylenes produced by catalyst 1, where solubility is not an issue, size exclusion chromatography, dynamic light scattering, static light scattering, intrinsic viscosity, ozonolysis, rheology, and AFM images are consistent with a cyclic toppology.^29,31–33 l-PA produced using typical catalysts contains cis-double bonds in accordance with an insertion mechanism and requires relatively extreme conditions (150–200 °C) to isomerize. However, due to the cyclic structure and access to low barriers, even at −94 °C, a temperature far too low to induce isomerization, catalyst 1 produces a polymer with >99% trans-transoid double bonds. The exclusive trans-isomer fits with the fact that large annulenes undergo rapid configuration changes and double bond isomerization,³⁸ whereas linear polyenes such as β-carotene have much larger barriers (extrapolated experimental value = 22.4 kcal/mol;⁷⁴ computed model value⁷⁵ = 22.9 kcal/mol). Another important point is that the thermal barrier for isomerization of l-PA is not constant during the isomerization. Initially, the barrier is estimated to be 17 kcal/mol; but as more and more isolated cis-bonds remain, the barrier increase to 39 kcal/mol when 20% cis bonds remain. In contrast, cyclic polyacetylene isomerizes to greater than >99% at −94 °C. Eliminating the possibility of a radical-induced isomerization at low temperatures, polymerization in the presence of TEMPO does not change the trans-outcome during the synthesis of c-PA (see Supplementary Fig. 4). Finally, previously confirmed to have a cyclic topology, cyclic polyphenylacetylene produced with catalyst 1, in contrast to c-PA, contains 90% cis-double bonds.²⁷ The high cis-content suggests catalyst 1 inserts alkynes, as expected, via a traditional insertion mechanism, but the phenyl-substituent must raise the barrier to isomerization, thus effectively trapping the cis stereochemistry.

In addition to the unique topology produced by catalyst 1, c-PA has several properties that present exciting opportunities for further exploration. Catalyst 1 rapidly produces polyacetylene, with pristine films developing within seconds upon contact with 1 atm of acetylene gas and ppm catalyst loadings. Catalyst 1 is highly active towards the polymerization of acetylene, and importantly, does not require copious equivalents of AlEt₃ co-catalyst. Simply washing (3–5 times) with THF and pentane produces a film with 0.35–0.85 wt% W. The onset of polymerization even at low temperatures must limit defects (crosslinking/saturation) and permits the synthesis of temporarily soluble c-PA, which may be useful in device fabrication. Another unique feature of catalyst 1 is its exclusive reactivity with acetylene over ethylene. Exposing a benzene solution of catalyst 1 to a 4:1 mixture of acetylene:ethylene results in complete consumption of acetylene to produce c-PA. This result implies ethylene streams used for the mass production of polyethylene and containing acetylene as a contaminant can be used to produce c-PA, thus effectively purifying the ethylene stream while at the same time producing high purity cyclic polyacetylene. The high-quality low defect films of c-PA produced with catalyst 1 reveal high electrical conductivity when doped with I₂. Since the c-PA forms in the >99% trans isomer, there is no need to subject the polymer to high temperatures, and the electrical conductivity observed (398 (±76) Ω⁻¹cm⁻¹) is at the higher range of values reported for l-PA. The efficiency and selectivity of catalyst 1, the pristine films produced, the temporarily soluble polymer afforded by the synthetic methodology, and the inherent differences of a cyclic topology should reinvigorate studies on PA in general, with perhaps a new focus on the physical properties implicated by a semi-conducting cyclic polymer.