Carbon chains and carbyne: synthesis, predictions and stability

A one-dimensional carbon candidate

Carbyne is used for idealised extended linear carbon-chain structures. Predictions concern a defined chain, bonding pattern and environment. Finite chains, supported chains and a freestanding macroscopic solid are different physical systems. Their stability and properties cannot be identified by the name alone.

A chain can give a strong optical signal while contributing little to the force carried between two grips. Conversely, an ideal-chain calculation can predict a large tensile force without establishing that a long chain has been made or attached to a load. This lesson connects those apparently different observations through the actual chain, its host and its ends.

Predictions, targets and levels of evidence prepares the separation of a computed response, a measured specimen and an intended bulk target. Atomistic models, assumptions and validation prepares the energy-model and boundary distinctions. Here the named carbon-chain mechanisms and observations are the subject. All numerical examples labelled invented are teaching models, separate from the dated experimental results.

The bonding pattern is part of the candidate

In an ideal linear sp carbon backbone, two oppositely directed sp orbitals provide the axial σ framework. Two transverse p-orbital directions provide π bonding. The familiar cumulene representation uses successive double bonds, whereas the polyyne representation alternates nominal single and triple bonds. These drawings describe bonding patterns, not rigid classical sticks or independently measurable integer bond orders at every site.

Define bond-length alternation as b = d_long − d_short. A polyynic structure has b > 0; an equal-bond idealisation has b = 0. Alternation doubles the repeat of an equal-spacing chain. Distorting a partially filled electronic band can lower the energy of occupied states while costing elastic energy in the backbone: this is the physical competition behind a Peierls distortion. The preferred distortion requires both contributions, electron filling and the relevant constraints. An electronic gap calculation alone does not minimise the total structural energy.

A finite chain also has chemically terminated ends, possibly with different end groups, charge or attached electrodes. Their influence can be important for a short conjugated segment. A central region of a sufficiently long chain may have a response close to a long-chain limit under a specified environment. That convergence does not remove the ends from a mechanical load path, make every property length independent or turn a finite molecule into a three-dimensional solid.

A transparent electronic model explains a gap, not an entire material

Use a periodic two-site model to isolate the consequence of alternation. Each site supplies one orthogonal electronic state in one π channel. Keep only nearest-neighbour hopping, take real positive hopping energies v and w, set equal on-site energies to zero and neglect electron–electron interactions, spin-dependent terms and nuclear motion. Spin degeneracy would duplicate the bands; real sp chains have additional orbital and interaction physics. At half filling the lower band is occupied and the upper one empty.

Let a be the two-site repeat and θ = ka the dimensionless Bloch phase. Here k is the wave number, the phase change per distance along the chain; i is the imaginary unit with i² = −1, and exp(iθ) = cos θ + i sin θ. A hop across a cell boundary contributes a phase factor recording the phase difference between neighbouring cells. In the A, B basis the matrix is H(θ) = [[0, v + w exp(−iθ)], [v + w exp(iθ), 0]]. Its characteristic equation is E² = (v + w exp(−iθ))(v + w exp(iθ)) = v² + w² + 2vw cos θ. Thus E_± = ±√(v² + w² + 2vw cos θ). The sign of both hoppings can be changed by a basis convention without changing these energies.

For positive v and w the closest band edges occur at θ = ±π. The model gap is E_g = 2|v − w|; at θ = 0 the energies are ±(v + w). Equal hoppings close this particular gap. A change of bond geometry can change orbital overlap and hence hopping, but v and w are energies, not spring constants, bond lengths or measured bond strengths. No hopping-versus-length law is supplied here.

Invented set A has v = 2.80 eV and w = 2.20 eV; set B has v = 2.10 eV and w = 1.50 eV. Both have E_g = 1.20 eV, while their zone-centre energies are ±5.00 and ±3.60 eV. The same gap therefore leaves other electronic responses undetermined even in this restricted model. Neither set is fitted to carbyne, and the plotted curves are analytic model evaluations, not first-principles results or optical spectra.

This periodic single-particle gap omits finite-end states, many-electron corrections and electron–hole attraction. An optical excitation can create a bound electron–hole pair, so an optical onset is not automatically this single-particle gap. Nor does a positive gap prove mechanical stability, chemical persistence or useful electrical conduction through contacts. The model explains one consequence of a stated alternation; it does not establish the existence of a proposed specimen.

Confinement protects and perturbs the chain

A nanotube cavity separates a chain from nearby reactive species and from direct contact with other chains. Its walls restrict transverse configurations and provide a protected space in which a long chain can form. End termination offers a different way to protect finite molecules. These are changes to the physical system, not merely containers added to the name carbyne.

The host also supplies an interaction potential and an electronic environment. Tube geometry, noncovalent interactions, dielectric screening and possible charge redistribution can alter the chain’s bonding and spectroscopy. Their relative contributions require evidence for the particular hybrid. A shifted Raman line alone does not uniquely identify charge transfer, a bond-length change or a new chain length.

In a 2017 resonance-Raman study of chains inside double-walled nanotubes, excitation profiles were fitted to obtain optical transition energies from 1.848 to 2.253 eV. Those are hosted optical responses, not a universal isolated-chain electronic gap. The study explicitly found that environmental and end effects complicate a unique length interpretation. It also distinguished optical gaps, which include electron–hole binding effects, from other theoretical gaps.

Removing the tube changes shielding, available configurations and inter-chain encounters at the same time. The protected-chain response cannot be transferred unchanged to a freestanding assembly. Similarly, a conducting hybrid would not by itself show that an isolated neutral chain is metallic: the tube, contacts and electronic coupling participate in transport.

A specific synthesis result has a specific scope

Shi and colleagues reported long linear carbon chains confined inside double-walled nanotubes in Nature Materials in 2016. The confinement is part of the demonstrated system. That result should not be relabelled as an independently stable, freestanding bulk cable or a measured component strength.

The 2016 evidence combined complementary measurements. Transmission electron microscopy showed a chain-like line inside a tube over about 26 nm, corresponding to more than 200 atoms using the stated repeat. For the longest reported object, a near-field Raman intensity map gave a length of 797 nm, converted to more than 6000 atoms. That longest length was not an atom-by-atom TEM count. X-ray diffraction supplied an ensemble structural check, rather than a length for every object.

These observations support long confined chains with their respective spatial and structural limits. The reported optical signal does not measure tensile failure, prove that each illuminated chain bridges a macroscopic gauge or establish retention after extraction. The synthesis interpretation also leaves growth details unresolved; a successful protected product does not require that every proposed atomic growth step has been directly observed.

A Raman line is a vibration; its intensity needs a population model

Raman scattering measures a photon energy change associated with a vibration. For a polyyne-like backbone, the prominent chain C mode involves collective bond stretching and changing electronic polarisability. Its wavenumber is an energy difference divided by hc, usually expressed in cm⁻¹; it is not an optical wavelength or an atom count. Resonance occurs when the excitation couples strongly to an electronic transition and can greatly enhance the signal.

A 2020 study mapped 20 individual confined chains and separately measured their resonant Raman signals against a reference scatterer. Over that sample, the C-mode frequency showed no evident length dependence, while calibrated intensity increased with mapped length. The analysis accounted for wavelength-dependent detection and noted finite spatial resolution, detuning and illumination-size limits. A strong signal is therefore useful evidence, but its magnitude cannot be equated directly with total synthesis yield.

For an invented intensity comparison, suppose S_j = K β_j N_eff,j. S is a background-subtracted integrated count rate; K is a common photon-flux and collection factor; β is an effective per-atom differential cross section under the stated excitation, polarisation and environment; N_eff is the illuminated, weighted number of scattering atoms. Assume incoherent addition, a linear detector, no heating change and known equal K. This is an explicitly restricted signal model, not a universal Raman law.

Then N_eff,2/N_eff,1 = (S₂/S₁)(β₁/β₂). If an object extends beyond the illuminated region, N_eff need not equal its total atom count. Detuning from resonance changes β even without changing the population. An ensemble contains a sum KΣβ_jN_eff,j; one total intensity cannot in general separate the population distribution from the cross sections. Calibrating the relevant optical response and measuring representative spatial coverage are substantive parts of a yield claim.

An ideal tensile prediction has a loading and failure definition

A 2013 first-principles study estimated single-chain tensile forces of roughly 9.3–11.7 nN using two different criteria: a strain-dependent activated rupture estimate and a phonon-instability calculation. They are conditional theoretical results, not measured forces on the 2016 confined specimens. A thermal rupture criterion and a loss of local vibrational stability need not give the same endpoint. There is no protocol-free material strength in that range.

For an atomic chain, tensile force is often a cleaner starting quantity than a stress based on an assigned diameter. Define reference chain linear mass λ_c = 2m_C/a for a long chain with unstrained two-carbon repeat a; neglect end-group mass. Then F/λ_c has units N·m/kg. No arbitrary cross-sectional area is needed. Cross-sectional area, gauge length and test conditions develops that denominator issue in more detail.

For numerical bookkeeping only, take a = 0.2558 nm, m_C = 12u and u = 1.66053906660 × 10⁻²⁷ kg. This gives λ_c = 1.55797 × 10⁻¹⁶ kg/m. An illustrative single-chain force F₀ = 10.0 nN then gives F₀/λ_c = 6.41860 × 10⁷ N·m/kg. The repeat is a rounded structural input; the mass value is a stated carbon-12 teaching approximation. The result is not a measured strength, an uncertainty evaluation or an allowable load.

A stiff axial backbone does not specify resistance to a crack, cyclic damage or bending. Likewise, spectroscopically identifying a chain does not establish its end attachment. A force applied to the tube need not pass into the chain; that transfer requires a coupling or anchorage mechanism and its own response evidence.

Continuity, anchorage and host mass intervene before a bulk claim

Consider an invented assembly of n parallel hosted paths over the same reference gauge. Every path contributes chain mass λ_c and host mass hλ_c per length; h ≥ 0. Only a fraction p of the chains is continuous across the gauge and attached at both ends. The remaining chains still contribute mass. Every usable path has the same chain-force limit F₀ and the same end-transfer limit F_a. Neglect misalignment and assume simultaneous equal sharing up to those limits.

A usable chain can contribute at most min(F₀,F_a), so F_chain ≤ np min(F₀,F_a). The assembly linear mass is λ_tot = n(1 + h)λ_c. Its chain-attributable force-per-total-linear-mass ceiling is therefore R_chain ≤ p min(F₀,F_a)/[(1 + h)λ_c]. For an integer path count use np as the actual number of usable paths. Equality is the ideal simultaneous limit of this stipulated accounting model; variability or uneven transfer can reduce the chain contribution.

This is a bound on the chain contribution, not a predicted total hybrid breaking force. Real nanotube hosts can carry load themselves. Their force contribution, failure sequence and interaction with chains have not been specified here. No claim that hosts are mechanically useless follows from counting their mass in the denominator.

At p = 1 with perfect transfer, a mass ratio h = 5 alone divides the chain-only F₀/λ_c by six. With p = 0.80 and F_a = 4.00 nN the ceiling falls further to 3.42325 × 10⁶ N·m/kg. Improving the ideal chain limit beyond 10 nN cannot improve that number while anchorage still limits each usable chain to 4 nN. This explains why evidence for a better isolated backbone can leave a specimen-level bottleneck unchanged.

A change of support opens different chemical and mechanical pathways

Contact between reactive chains can permit cross-linking, changing the local coordination and interrupting the original conjugated backbone. The 2013 calculation examined a projected cross-linking route for finite chain pairs and found a barrier near 0.6 eV under its model. That barrier is neither a measured lifetime nor an established lowest barrier for every real assembly. Temperature, end chemistry, reactive species, contact geometry and alternative routes affect which events are accessible.

A protective host can suppress some encounters while also being needed for the observed electronic response. An extracted chain that changes its spectrum may have changed geometry, chemistry, environment or several at once. Before attributing the change to length alone, compare structural and optical evidence under matched support and exposure, with appropriate unfilled-host and background controls. These are evidence requirements, not a synthesis or extraction procedure.

For a load-bearing claim, distinguish chains that merely occur somewhere in a tube from chains connected between both load-introduction regions. Examine the intended gauge, interface response, distribution of defects and failure modes. For retention, identify what structure and useful response were observed, at what conditions and times, and with what detection limits. A finite observation can support persistence over that record without establishing equilibrium stability or safe service.

The scientific achievement survives these distinctions: long confined carbon chains can be studied as a specific hybrid material. What remains conditional is the transfer from that hybrid, or from an ideal periodic calculation, to a freestanding bulk material with reproducible performance. A useful next experiment is one that distinguishes the competing mechanism or bottleneck in that transfer, rather than simply repeating a strong signal.

Worked examples

Try the reasoning

1. Can a host alter stability or electronic response?

   Show the answer and reasoning

   Yes.

   The environment participates in the system.

2. Does a theoretical chain modulus establish cable toughness?

   Show the answer and reasoning

   No.

   The specimen, scale and failure question differ.

3. A new hypothetical map spans 420 nm of a chain whose long-repeat input is 0.2558 nm for two atoms. Estimate N. If end-to-end attachment was not observed, can this estimate establish a mechanically continuous 420 nm bridge?

   Show the answer and reasoning

   N ≈ 2 × 420/0.2558 = 3283.82, about 3300 atoms. It does not establish a loaded bridge.

   Repeat-based counting interprets the mapped extent; it does not demonstrate end anchorage, uninterrupted mechanical continuity or force transfer. The spatial and structural resolution still matter.

4. Set D has positive hoppings v = 3.40 eV and w = 2.80 eV. Find its gap and its upper energy at θ = 0 and θ = π/2. Can the common gap with A show that their full bands coincide?

   Show the answer and reasoning

   E_g = 1.20 eV; E_+(0) = 6.20 eV; E_+(π/2) = √(3.40² + 2.80²) = √19.40 ≈ 4.40454 eV. The full bands differ.

   The gap fixes the absolute difference of the positive hoppings, not their sum. A has E_+(0) = 5.00 eV and E_+(π/2) ≈ 3.56090 eV. None of these invented electronic energies determines a force constant or an optical resonance without additional physics.

5. In a new stipulated uniform-illumination example, population X has 6000 total chain atoms but only 2000 illuminated atoms. Y has 3000 total atoms, all illuminated. Both have equal K and β. Find S_X/S_Y. Would a naive intensity ranking recover the total atom ranking?

   Show the answer and reasoning

   S_X/S_Y = 2000/3000 = 2/3. X contains twice as many total atoms but gives the smaller signal.

   The signal model uses the illuminated weighted population. Changing to a real Gaussian beam would require the corresponding spatial weighting; the present counts stipulate a uniform illuminated region.

6. Keep B’s h = 3, F_a = 6.00 nN and F₀ = 10.0 nN, but improve p from 0.40 to 0.80 without changing any path mass. Calculate the new ceiling. Could a 10% larger F₀ achieve the same improvement at the original p?

   Show the answer and reasoning

   R_chain ≤ 7.70232 × 10⁶ N·m/kg, twice B’s original ceiling. Raising F₀ to 11.0 nN alone leaves the original ceiling unchanged.

   The model is linear in the usable-path fraction p, while min(F₀,F_a) stays equal to 6 nN. This is a transfer bottleneck, not evidence that a real continuity intervention can be achieved without other changes.

7. Two hypothetical hosted samples have the same C-mode wavenumber within resolution. One Raman map is twice as long. Must the longer one have a different electronic gap or twice the total chain mass in the batch?

   Show the answer and reasoning

   Neither conclusion is required.

   Long-chain vibrational response can be dominated by the environment and can approach a length limit. A mapped object’s length does not supply the number of other objects in a batch, their illumination or a calibrated population. Optical gaps and Raman frequencies also require the appropriate host-specific evidence.

8. A calculation finds a 0.6 eV barrier along one projected reaction route for two finite chains. An invented retention record then shows a protected specimen unchanged at two observations 24 hours apart. Does this give a universal 24-hour lifetime for a freestanding assembly?

   Show the answer and reasoning

   No. The calculation is a conditional route barrier; the observations are finite retention evidence for a protected specimen.

   Different contacts, ends, temperatures, reactive species or alternative paths can change the process. A rate model with suitable kinetic information and representative observations would be needed for a time prediction. Removing the host changes the system and the available routes.

What to carry forward

• Specify chain length, structure and support.
• Keep synthesis, stability and usable mechanics separate.
• Bond alternation can open a gap in a stated electronic model; that gap neither fixes every model parameter nor proves a stable real chain.
• Keep local microscopy, spatial Raman length inference, optical resonance and calibrated population as distinct observables.
• Host mass, chain continuity and end transfer can limit a chain’s specimen-level contribution before its ideal tensile limit is reached.