Carbon Nanosprings: Topology & Mechanics

• Carbon nanosprings are carbon nanostructures with engineered scroll or helical geometries that yield spring-like mechanical responses governed by curvature elasticity and van der Waals adhesion.
• Fabrication methods include CNT-initiated scrolling of graphene and controlled helical stacking of aromatic units, allowing precise tuning of stiffness and mechanical performance.
• Key findings demonstrate scaling laws for spring constants and oscillator frequencies in the GHz range, with implications for NEMS, energy harvesting, and force sensing.

Searching arXiv for the cited carbon nanospring and related carbon nanoscroll papers to ground the article in current records. Carbon nanosprings are carbon nanostructures whose geometry yields spring-like mechanical response, most prominently open carbon nanoscrolls formed by rolling graphene and chiral helical architectures built from stacked polycyclic aromatic units. In the literature summarized here, the term encompasses at least two distinct realizations: carbon nanoscrolls (CNSs), which can be fabricated by carbon-nanotube-initiated scrolling of graphene and subsequently treated as nanoscale coil springs, and single-layer helical macromolecules described as graphene helicoids or spiral nanoribbons derived from coronene or kekulene motifs (Zhang et al., 2011, Savin et al., 6 Aug 2025). Their behavior is governed by curvature elasticity, van der Waals adhesion, topology, and, in driven configurations, electromechanical coupling; these same ingredients determine their stiffness, dissipation, defect physics, and suitability for nanoelectromechanical systems.

1. Structural classes and topology

The primary structural distinction in this literature is between scroll-based nanosprings and intrinsically helical nanosprings. A carbon nanoscroll has an open “jelly-roll” topology produced by spiral wrapping of a graphene-like sheet, whereas the helicoidal class is generated by radially cutting a planar aromatic monomer and stacking it with a fixed axial shift and rotation. In both cases, the spring response emerges from a combination of curvature and reversible geometric rearrangement rather than from a bulk three-dimensional coil geometry (Júnior et al., 2020, Savin et al., 6 Aug 2025).

Class	Construction	Representative parameters
CNT-initiated CNS	Graphene monolayer scrolls around a CNT on a substrate	10 nm-long (10,10) SWCNT; 10 × 30 nm graphene sheet
Pristine or amorphous CNS	Spiral-wrapped sheet with open topology	Rolling angle $4\pi$ rad; $R_i \approx 5$ Å; $R_o \approx 8$ Å; $L = 120$ Å
Helicoid / spiral nanoribbon	Radially cut coronene or kekulene stacked with fixed shift and rotation	$\Delta z \approx 0.58$ Å; $\Delta\phi \approx 61^\circ$; $R \approx 4$–$5$ Å for $l=4$ coronene

For the helicoidal nanosprings, the continuum centerline can be written as

$$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$

with full pitch

$R_i \approx 5$0

A helix built from $R_i \approx 5$1 monomers has total length $R_i \approx 5$2, total rotation $R_i \approx 5$3, and number of turns

$R_i \approx 5$4

A recurring source of ambiguity is nomenclature. Some papers reserve “carbon nanospring” for explicitly chiral helical molecules, while others use the term for nanoscrolls once their axial compliance is characterized as spring-like. The two usages describe different topologies rather than incompatible results.

2. CNT-initiated scrolling and formation thermodynamics

For nanoscroll-based nanosprings, scrolling is seeded by placing a CNT at the edge of a graphene monolayer on a substrate. The CNT–graphene van der Waals attraction promotes initial wrapping; once overlap forms between graphene layers, graphene–graphene adhesion drives further scrolling into a multilayer CNS (Zhang et al., 2011). In a specific molecular-dynamics realization, a 10 nm-long (10,10) SWCNT is placed along one armchair edge of a 10 × 30 nm monolayer graphene sheet supported on a 34 × 14 × 1 nm SiO$R_i \approx 5$5 slab. C–C covalent interactions are described by the 2nd-generation Brenner (REBO) potential, non-bonded C–C interactions by a truncated Lennard-Jones pair potential, and graphene–substrate interactions follow Zhang and Li (APL 2010). The simulation is performed in LAMMPS in the NVT ensemble at $R_i \approx 5$6 K with a 1 fs timestep until spontaneous wrapping and scrolling occur in approximately 100 ps, accompanied by a potential-energy drop of about 600 eV (Zhang et al., 2011).

The energetics are resolved into four terms per unit width of graphene:

$R_i \approx 5$7

$R_i \approx 5$8

$R_i \approx 5$9

$R_o \approx 8$0

Here $R_o \approx 8$1 eV is the bending modulus of monolayer graphene, $R_o \approx 8$2 is the local curvature, $R_o \approx 8$3 eV/Å$R_o \approx 8$4 is the interlayer adhesion energy, and $R_o \approx 8$5–$R_o \approx 8$6 eV/Å$R_o \approx 8$7 for SiO$R_o \approx 8$8. In the initial stage, wrapping occurs if the decrease in $R_o \approx 8$9 outweighs the increase in $L = 120$0; after overlap forms, the gain in $L = 120$1 sustains continued scrolling.

A compact continuum expression for a scroll of $L = 120$2 turns, inner radius $L = 120$3, layer spacing $L = 120$4 nm, and axial width $L = 120$5 is

$L = 120$6

with $L = 120$7. A rough condition for spontaneous scrolling is

$L = 120$8

The simulation phase diagram distinguishes three regimes: Mode I, in which the CNT glides on flat graphene; Mode II, in which graphene wraps the CNT but does not form a full scroll; and Mode III, in which graphene rolls into a stable CNS. Two principal control parameters are the CNT diameter $L = 120$9 and the normalized C–C interaction strength $\Delta z \approx 0.58$0 for fixed substrate adhesion $\Delta z \approx 0.58$1. The approximate threshold is

$\Delta z \approx 0.58$2

At $\Delta z \approx 0.58$3 and $\Delta z \approx 0.58$4 eV, the Mode I $\Delta z \approx 0.58$5 II boundary lies near $\Delta z \approx 0.58$6 nm at $\Delta z \approx 0.58$7, whereas the Mode II $\Delta z \approx 0.58$8 III boundary lies near $\Delta z \approx 0.58$9 nm at the same $\Delta\phi \approx 61^\circ$0. Raising $\Delta\phi \approx 61^\circ$1 to 4 shifts both boundaries to larger $\Delta\phi \approx 61^\circ$2 or higher $\Delta\phi \approx 61^\circ$3, and below a critical $\Delta\phi \approx 61^\circ$4 no scroll forms for any CNT size.

3. Elastic response and spring-constant scaling

Once formed, a CNS can be modeled mechanically as a nanoscale coil spring. For small axial extension $\Delta\phi \approx 61^\circ$5, the axial spring constant follows from the change in bending energy as the layers wind or unwind. Neglecting interlayer friction, the closed-form approximation is (Zhang et al., 2011)

$\Delta\phi \approx 61^\circ$6

for $\Delta\phi \approx 61^\circ$7 and $\Delta\phi \approx 61^\circ$8. This immediately yields the principal scaling law $\Delta\phi \approx 61^\circ$9: longer sheets and more turns stiffen the spring, while a larger inner radius softens it strongly through the $R \approx 4$0 dependence.

The same work gives a constructive design rule. For a target spring constant $R \approx 4$1, one chooses graphene length $R \approx 4$2 and desired number of turns $R \approx 4$3, solves

$R \approx 4$4

selects a CNT of diameter $R \approx 4$5 to initiate scrolling, and ensures that

$R \approx 4$6

For $R \approx 4$7 nm, $R \approx 4$8, $R \approx 4$9 eV, and $5$0 N/m, the estimate is $5$1 nm, implying $5$2 nm and the use of $5$3 eV/Å$5$4, $5$5 eV/Å$5$6.

The directly helical nanosprings exhibit comparable spring-like scaling but in a different geometry. For a 4-coronene helicoid of length $5$7 nm, the small-strain stiffness is $5$8 N/m; for a 4-kekulene spiral nanoribbon, $5$9 N/m. Using a ribbon-like cross-section with width $l=4$0 nm and thickness $l=4$1 nm gives an effective modulus $l=4$2–$l=4$3 TPa, consistent with graphene’s Young modulus (Savin et al., 6 Aug 2025). This suggests that the spring response of these systems is not a low-modulus anomaly; rather, high in-plane carbon stiffness is being re-expressed through a compliant topology.

4. Axial nano-oscillators and dissipation control

A distinct use of nanoscroll-based nanosprings is as ultrafast axial oscillators. In the atomistic “mass-on-spring” picture, the CNT confined inside a CNS is treated as a point mass $l=4$4 attached to an effective spring of stiffness $l=4$5 arising from the axial van der Waals restoring force. The equation of motion is

$l=4$6

with natural frequency

$l=4$7

The oscillating CNT mass is

$l=4$8

where $l=4$9 kg/m$$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$0, $$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$1 is the CNT radius, and $$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$2 nm. A first-order expansion of the insertion energy,

$$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$3

with $$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$4 the number of turns and $$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$5 J/m$$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$6, yields

$$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$7

For $$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$8–$$\mathbf r(\theta)=\bigl(R\cos\theta,\;R\sin\theta,\;\tfrac{p}{2\pi}\theta\bigr),$$9 turns, $R_i \approx 5$00 nm, and $R_i \approx 5$01 nm, $R_i \approx 5$02 is on the order of $R_i \approx 5$03–$R_i \approx 5$04 N/m and $R_i \approx 5$05 kg, giving $R_i \approx 5$06 in the 10–50 GHz range, consistent with molecular dynamics (Zhang et al., 2011).

The dominant dissipation channels are interlayer sliding friction within the CNS, conversion of CNT translational energy into self-oscillation of the scroll, and increased friction from thermal roughness at elevated temperature. A specific dissipation-reduction strategy is interlayer bridging: vacancies are patterned along three parallel lines on the graphene before scrolling, and the assembled CNS is then heated from 300 K to 1300 K over 100 ps, held for 1600 ps, and cooled back over 100 ps so that covalent C–C bonds form between adjacent scroll layers. These bridges suppress internal sliding.

At 100 K, the unbridged CNS oscillator shows rapid decay and irregular coupling to CNS self-oscillation, with an FFT peak at $R_i \approx 5$07 GHz. The bridged CNS oscillator exhibits nearly purely harmonic CNT motion and slower amplitude decay. A DWCNT (10,10)/(15,15) axial benchmark oscillator behaves similarly but with modestly faster damping. Using the logarithmic decrement

$R_i \approx 5$08

the bridged CNS gives $R_i \approx 5$09 at 100 K, compared with $R_i \approx 5$10 for the DWCNT axial oscillator; at 300 K, both $R_i \approx 5$11 values drop modestly but the bridged CNS remains approximately 10–15% higher. An incommensurate SWCNT@CNS configuration using a (15,0) inner tube further reduces friction, with a friction rate of about 0.237 nm/ns versus 0.429 nm/ns for the commensurate case.

The same system can be driven externally by charging or polarizing the inner SWCNT while keeping the CNS neutral and applying an axial AC electric field. In molecular dynamics at 100 K, a square-wave field with $R_i \approx 5$12 GHz and effective per-atom force amplitude $R_i \approx 5$13 eV/Å, much larger than the intrinsic van der Waals restoring force of about $R_i \approx 5$14 eV/Å per atom, drives synchronous CNT oscillation at 125 GHz with negligible phase lag. The peak amplitude remains within $R_i \approx 5$15 over thousands of cycles, with no measurable decay under continuous drive; a slight DC offset of about 0.2 nm may arise from scroll asymmetry or non-uniform bridging.

5. Tensile mechanics, fracture, and amorphous analogues

The tensile behavior of scroll-based nanosprings depends strongly on structural order. A pristine CNS and an amorphous carbon nanoscroll (A-CNS) with the same overall dimensions, $R_i \approx 5$16 Å, $R_i \approx 5$17 Å, $R_i \approx 5$18 Å, and two full turns, were studied using fully atomistic reactive molecular dynamics. The A-CNS is derived from a monolayer amorphous-carbon sheet containing randomly distributed five-, six-, seven-, and eight-member carbon rings. Both structures were generated in Sculptor and relaxed via AIREBO-MD in LAMMPS with a 0.1 fs timestep; uniaxial tensile tests were then performed at 300 K in an NVT ensemble with strain rate $R_i \approx 5$19 fs$R_i \approx 5$20, using virial stress and von Mises stress to monitor loading and fracture (Júnior et al., 2020).

In the low-strain regime, the stress–strain relation is quadratic:

$R_i \approx 5$21

For the pristine CNS,

$R_i \approx 5$22

with fracture strain

$R_i \approx 5$23

For the amorphous A-CNS,

$R_i \approx 5$24

with

$R_i \approx 5$25

The corresponding linearized elastic limits are $R_i \approx 5$26 and $R_i \approx 5$27.

Young’s modulus is extracted from virial stress using the effective cross-section

$R_i \approx 5$28

with $R_i \approx 5$29 and $R_i \approx 5$30 Å, and the effective axial spring constant is

$R_i \approx 5$31

The key comparison is that $R_i \approx 5$32 GPa and $R_i \approx 5$33 GPa, only an approximately 0.7% reduction, while the tensile strength drops from $R_i \approx 5$34 GPa to $R_i \approx 5$35 GPa and the fracture strain from 33.8% to 23.6%. The pristine scroll fractures abruptly, whereas the amorphous scroll shows a pronounced non-elastic regime and the formation of linear atomic carbon chains before final rupture.

Thermal stability shows a related but not identical trend. Heating-ramp simulations from 0 to 10,000 K at approximately 200 K ps$R_i \approx 5$36 give melting points of 5100 K for A-CNS and 5900 K for pristine CNS. Thus, similar room-temperature stiffness does not imply similar load-bearing capacity or high-temperature viability. This directly counters a common simplification in nanospring discussions: elastic modulus alone is not an adequate proxy for resilience.

6. Defects, non-axial deformation, and thermal expansion

The helical nanospring literature extends beyond pure axial loading to bending, twisting, buckling, and chirality inversion. In the molecular-dynamics model, each monomer cell contains carbon atoms and edge united C–H pseudoatoms, with Hamiltonian

$R_i \approx 5$37

and

$R_i \approx 5$38

with $R_i \approx 5$39 eV and $R_i \approx 5$40 Å. Simulations use a Langevin thermostat with $R_i \approx 5$41 ps$R_i \approx 5$42 at 300 K, velocity-Verlet integration, and $R_i \approx 5$43 fs (Savin et al., 6 Aug 2025).

Under small axial strain, the spring energy is quadratic,

$R_i \approx 5$44

where $R_i \approx 5$45. Compression of a 4-coronene spring produces Euler buckling at $R_i \approx 5$46, followed by a first transverse crack at $R_i \approx 5$47 and a second at $R_i \approx 5$48. The corresponding critical load is

$R_i \approx 5$49

which implies

$R_i \approx 5$50

for a hinged-rod estimate. A 4-kekulene spring is softer by approximately 40% and breaks only once at $R_i \approx 5$51. Under bending, the critical lateral force is about 0.034 eV/Å (0.055 nN) for a 4-coronene spring and 0.012 eV/Å (0.019 nN) for a 4-kekulene spring; beyond this threshold the molecule irreversibly folds and is stabilized by inter-coil van der Waals adhesion.

Torsional behavior is likewise compliant:

$R_i \approx 5$52

From the parabolic part of the simulated curves, $R_i \approx 5$53 N m$R_i \approx 5$54 and $R_i \approx 5$55 N m$R_i \approx 5$56, one to two orders of magnitude smaller than the bending modulus $R_i \approx 5$57.

A characteristic defect of chiral nanosprings is the helix-reversal defect, which separates a right-handed segment from a left-handed segment and is localized on two consecutive coils. Reported defect energies span 1.8–12.4 eV, with corresponding angles between the axes of the two halves spanning 96.6°–164.8°. When a left-handed 4-coronene spring is untwisted by applying negative end rotation, the twist energy grows approximately quadratically until a critical angle $R_i \approx 5$58, at which point the energy drops by about 8 eV and a helix-reversal defect forms; continued rotation transports the defect along the spring until full chirality inversion.

These structures also display a relatively large axial thermal expansion coefficient,

$R_i \approx 5$59

with simulations over $R_i \approx 5$60 K giving

$R_i \approx 5$61

This exceeds the listed values for steel ($R_i \approx 5$62 K$R_i \approx 5$63), aluminum ($R_i \approx 5$64 K$R_i \approx 5$65), and copper ($R_i \approx 5$66 K$R_i \approx 5$67). The reported interpretation is that the large axial coefficient is a direct consequence of the soft anharmonicity of interlayer van der Waals forces in the coils.

7. Functional roles and engineering implications

Several application domains follow directly from the measured frequencies, stiffnesses, and thermal responses. For nanoscroll-based oscillators, the 10–50 GHz natural modes and greater-than-100 GHz driven modes support use in ultrafast NEMS oscillators for signal processing, GHz filtering, and nanoscale clocking (Zhang et al., 2011). The efficient coupling between electrical drive and mechanical motion also suggests routes for energy transduction and harvesting, including conversion between AC electrical or electromagnetic energy and mechanical motion. Low-dissipation bridged CNS oscillation has further been proposed as a molecular-scale mechanical “flywheel.”

Frequency and quality-factor shifts provide a sensing and metrology channel: local adsorbates, forces, or temperature can in principle be reported through changes in natural frequency or $R_i \approx 5$68. Because graphene and CNTs can be fabricated and modified separately, the platform is tunable through CNT chirality, CNT length, charge state, and patterned bridging sites. In the helicoidal class, the large axial thermal expansion coefficient and force response in the nN/m to N/m range support nanopositioners, temperature sensors over broad temperature ranges, and single-nN-scale force sensing (Savin et al., 6 Aug 2025).

The principal engineering limits are equally explicit. In CNS oscillators, dissipation is controlled not only by the interface with the oscillating core but also by interlayer sliding and excitation of scroll self-modes. In amorphous nanoscrolls, structural disorder preserves much of the stiffness while sharply degrading strength and fracture strain. In helicoidal nanosprings, irreversible folding, compression-induced cracking, and helix-reversal defects define the accessible loading envelope. Taken together, these results indicate that “carbon nanospring” is best understood not as a single morphology but as a family of carbon architectures in which spring functionality emerges from topology, adhesion, and nanoscale geometry.