3D CNT@rGO Composite for LiFePO4 Electrodes

• The paper demonstrates a covalently integrated CNT@rGO framework that boosts electron transport and cycling stability in LiFePO4 cathodes.
• The synthesis uses a two-step CVD process with melamine-assisted Cu distribution, achieving an ultra-high CNT growth yield with low catalyst residue.
• Mechanical studies reveal that CNT spacers effectively prevent rGO restacking, imparting bistability, hysteresis, and energy dissipation in the composite structure.

A three-dimensional CNT@rGO composite is a hierarchical carbon hybrid in which carbon nanotubes (CNTs) are grown directly on reduced graphene oxide (rGO), producing a three-dimensional conductive framework in which the CNT–rGO junctions are described as covalently integrated. In the reported lithium-ion battery context, the material is designed as a carbon conductive agent for LiFePO$_4$ (LFP) cathodes, with the explicit aim of addressing the limited conductive efficiency, poor percolation at low dosage, and inadequate stability under repeated cycling associated with conventional conductive additives such as carbon black (Super P) and graphite flakes (Tang et al., 6 Jul 2025). A related mechanical study treats a graphene/rGO–CNT hybrid unit as a local building block of a three-dimensional CNT@rGO architecture, showing how CNT spacers can suppress restacking of graphene-like sheets and generate bistability, hysteresis, and energy dissipation in stacked hybrid networks (Ding et al., 3 Jun 2026).

1. Architectural definition and functional concept

The defining structural feature of CNT@rGO is a hierarchical three-dimensional network in which rGO functions as a broad, conductive, sheet-like scaffold, while CNTs extend from the rGO surface into the third dimension. The CNTs and rGO are not described as a simple physical mixture; rather, the reported architecture is based on direct CNT growth on rGO and seamless or covalent connection at the junctions (Tang et al., 6 Jul 2025).

This geometry is functionally significant because it combines several conduction modes within one framework. The reported architecture provides in-plane conduction through rGO sheets, axial conduction along CNTs, and inter-sheet bridging by CNTs between graphene layers and between LFP particles. The consequence is a conductive network with multiple electron-transport pathways rather than a single dominant transport direction. In the battery-electrode setting, the same framework is described as uniformly dispersed among LFP particles, where it acts as a bridge between particles, an encapsulation network, and a structural stabilizer (Tang et al., 6 Jul 2025).

A related atomistic description treats a graphene–CNT hybrid as a stacked unit consisting of two graphene layers with two CNTs sandwiched between them. In that interpretation, the graphene layers correspond to rGO nanosheets and the CNTs to one-dimensional spacers or bridges. The study explicitly frames this unit as a local building block of a three-dimensional CNT@rGO composite, in which CNTs hold sheets apart or, under other conditions, permit collapse and restacking around deformed CNTs (Ding et al., 3 Jun 2026). This suggests that the electrical and mechanical roles of CNTs in CNT@rGO are coupled through geometry: the same spacers that create conductive bridges also regulate intersheet spacing and structural stability.

2. Synthesis route and compositional parameters

The reported synthesis of CNT@rGO is described as a simple two-step process in which CNT growth itself occurs in one CVD step. In the first step, melamine $(0.3\ \mathrm{g})$ and CuCl$_2$\cdot$2H$_2$O$(0.0027\ \mathrm{g})$$are dispersed in ethanol. After drying, Cu is uniformly dispersed in a carbon–nitrogen skeleton through melamine-assisted supramolecular interactions, which is reported to avoid Cu aggregation and enable nanoscale catalyst formation. In the second step, the melamine/Cu precursor is mixed with rGO powder$$(0.1\ \mathrm{g})$, placed in a CVD furnace, heated under flowing H$_2$/Ar, and then exposed to C$_2$H$_4$$at high temperature to grow CNTs. The final product is described as a black, fluffy CNT@rGO powder (<a href="/papers/2507.04296" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Tang et al., 6 Jul 2025</a>).</p> <p>Several quantitative synthesis outcomes are emphasized. The CNT growth yield is reported as$$(0.3\ \mathrm{g})$0, although the same paper is noted to give $(0.3\ \mathrm{g})$1 elsewhere; the intended result is therefore presented as ultra-high yield rather than a fully resolved single value. The residual Cu catalyst content determined by ICP is $(0.3\ \mathrm{g})$2, approximately reported as $(0.3\ \mathrm{g})$3, which is characterized as sufficiently low to reduce the need for post-synthesis decontamination. The route is further described as cost-effective, high-yield, and suitable for scaling (Tang et al., 6 Jul 2025).

In the electrode formulation used for LFP cathodes, the composition is given as

$(0.3\ \mathrm{g})$4

This explicitly places CNT@rGO at $(0.3\ \mathrm{g})$5 in the electrode formulation (Tang et al., 6 Jul 2025). A plausible implication is that the three-dimensional conductive framework is intended to function as a low-dosage performance enhancer rather than as the sole conductive additive.

A separate fabrication route has been reported for macroscopic rGO–CNT hybrid films used to study analogous structural units mechanically. In that system, a GO suspension $(0.3\ \mathrm{g})$6 prepared by a modified Hummers method and a CNT suspension $(0.3\ \mathrm{g})$7 using ZEONANO 02DS-WA-RD CNTs of average diameter $(0.3\ \mathrm{g})$8 are mixed at a mass ratio of $(0.3\ \mathrm{g})$9, assembled by vacuum filtration, transferred from the membrane with acetone to fused quartz substrates, and reduced at $_2$0 for $_2$1 at $_2$2 under ethanol $_2$3 and Ar $_2$4 (Ding et al., 3 Jun 2026). That route does not reproduce the covalent CVD-growth protocol of CNT@rGO, but it does realize a bulk rGO–CNT architecture relevant to the mechanics of three-dimensional hybrids.

3. Structural characterization and evidence for covalent integration

Morphological characterization of CNT@rGO shows dense and uniform CNT growth on rGO and a clear three-dimensional structure. The CNTs are described as dendritic, with diameters mainly in the $_2$5–$_2$6 range. HRTEM reveals the CNT base connected to rGO via a junction, indicating direct integration rather than loose attachment (Tang et al., 6 Jul 2025).

Raman spectroscopy provides complementary evidence regarding the carbon framework. The D band is reported at $_2$7 and the G band at $_2$8. The intensity ratio $_2$9 changes from $\cdot$0 for pristine rGO to $\cdot$1 for CNT@rGO-NM and $\cdot$2 for CNT@rGO. The reported interpretation is that the lower $\cdot$3 value in CNT@rGO indicates partial defect repair and improved graphitic order after CNT growth (Tang et al., 6 Jul 2025).

Thermogravimetric analysis shows a main weight-loss peak near $\cdot$4, with a residual mass around $\cdot$5 at the end of the measurement. That residual is attributed to ash and oxidized Cu species. XPS deconvolution of the C 1s spectrum gives components at $\cdot$6 for sp$\cdot$7, $\cdot$8 for sp$\cdot$9, $_2$0 for C–O, and $_2$1 for the $_2$2–$_2$3 satellite, consistent with a largely graphitic carbon framework containing some defects and residual oxygen groups (Tang et al., 6 Jul 2025).

The most direct evidence for the reported covalent coupling comes from STEM/EELS and atomically resolved imaging. These measurements reveal a seamless CNT–rGO interface, an open-ended CNT structure at the junction, and $_2$4-membered ring defects in the connection region. The interpretation given is that the CNTs are covalently integrated with graphene-like carbon rather than weakly physically attached. The same study further states that modeling shows the Dirac cone is preserved, indicating that the graphene-like electronic character is not severely destroyed by the connection (Tang et al., 6 Jul 2025).

In the mechanically focused rGO–CNT film study, SEM shows wrinkled surfaces for rGO films and fibrous features for rGO–CNT films. Raman spectroscopy with $_2$5 excitation gives $_2$6 and $_2$7 for rGO, and $_2$8 and $_2$9 for rGO–CNT. AFM imaging before indentation shows larger height variations in CNT-containing regions, which is taken to indicate local sheet separation induced by CNT insertion (Ding et al., 3 Jun 2026). These observations are relevant to three-dimensional CNT@rGO because they show, in a macroscopic hybrid assembly, the geometric effect of CNTs in maintaining noncollapsed intersheet spacing.

4. Electrochemical behavior in LiFePO$(0.0027\ \mathrm{g})$0 cathodes

When used as a conductive agent in LFP cathodes, CNT@rGO is reported to improve initial reversibility, electronic conduction, rate capability, lithium-ion diffusion behavior, and long-cycle stability (Tang et al., 6 Jul 2025).

At $(0.0027\ \mathrm{g})$1, the initial coulombic efficiencies are reported as $(0.0027\ \mathrm{g})$2 for CNT@rGO, $(0.0027\ \mathrm{g})$3 for rGO, $(0.0027\ \mathrm{g})$4 for MWCNT, $(0.0027\ \mathrm{g})$5 for rGO&MWCNT, and $(0.0027\ \mathrm{g})$6 for SP. The electrode resistivity of CNT@rGO is given as $(0.0027\ \mathrm{g})$7, identified as the lowest among the compared electrodes. Over the $(0.0027\ \mathrm{g})$8 to $(0.0027\ \mathrm{g})$9 range, CNT@rGO is stated to show the best performance, including the most stable cycling characteristics, the highest specific capacity at all tested rates, and the best capacity retention in the reported three-dimensional histogram comparison (Tang et al., 6 Jul 2025).

For long-cycle operation at $(0.1\ \mathrm{g})$0, the residual specific capacity after $(0.1\ \mathrm{g})$1 cycles is $(0.1\ \mathrm{g})$2, corresponding to $(0.1\ \mathrm{g})$3 capacity retention. The explanation given for this stability is structural as well as electrical: CNT@rGO is reported to maintain stable electron pathways during repeated volume changes, buffer mechanical stress through a flexible carbon network, keep LFP particles well connected and uniformly dispersed, reduce structural collapse or delamination, and lower interfacial resistance growth during cycling (Tang et al., 6 Jul 2025).

The paper expresses the lithium-ion diffusion coefficient as

$(0.1\ \mathrm{g})$4

where $(0.1\ \mathrm{g})$5 is the lithium-ion diffusion coefficient, $(0.1\ \mathrm{g})$6 is the gas constant, $(0.1\ \mathrm{g})$7 is temperature, $(0.1\ \mathrm{g})$8 is electrode area, $(0.1\ \mathrm{g})$9 is the number of electrons transferred, $_2$0 is Faraday’s constant, $_2$1 is Li-ion concentration, and $_2$2 is the Warburg coefficient. The stated interpretation is that a smaller Warburg slope $_2$3 leads to a larger $_2$4; CNT@rGO is reported to have the highest diffusion coefficient among the tested electrodes, consistent with its superior rate behavior (Tang et al., 6 Jul 2025).

Electrochemical impedance analysis is also used to support the cycling-stability argument. In the fitting discussion, the charge-transfer resistance $_2$5 for CNT@rGO is reported as $_2$6 before cycling and $_2$7 after cycling. The stated interpretation is that the conductive network remains effective and may become more optimized after formation or electrode wetting (Tang et al., 6 Jul 2025).

5. Mechanical unit model, bistability, and hysteresis

A distinct but directly relevant line of analysis models a graphene–CNT hybrid as the structural unit of a three-dimensional CNT@rGO composite. The atomistic system consists of two free-standing graphene layers with two CNTs sandwiched between them, the CNTs aligned parallel to the $_2$8-direction and separated by an intertube distance $_2$9. The graphene sheets have length $_2$0 in $_2$1, width $_2$2 in $_2$3, and periodic boundary conditions in-plane. The initial graphene–graphene spacing is defined as

$_2$4

where $_2$5 is the CNT diameter and $_2$6 is the graphite interlayer spacing (Ding et al., 3 Jun 2026).

The MD simulations use the AIREBO potential for C–C bonding and intramolecular carbon interactions, an LJ potential for van der Waals interaction between graphene layers and CNTs, an NVT ensemble, $_2$7, a $_2$8 timestep, and LAMMPS. The control parameters are the intertube distance $_2$9, CNT diameter $_4$0, and wall number. Reported examples include single-walled CNT diameters of $_4$1, $_4$2, $_4$3, $_4$4, and $_4$5, double-walled CNTs with outer-wall diameters of $_4$6 or $_4$7, and $_4$8 values of $_4$9, $(0.3\ \mathrm{g})$00, $(0.3\ \mathrm{g})$01, $(0.3\ \mathrm{g})$02, $(0.3\ \mathrm{g})$03, $(0.3\ \mathrm{g})$04, $(0.3\ \mathrm{g})$05, $(0.3\ \mathrm{g})$06, and $(0.3\ \mathrm{g})$07 depending on the case (Ding et al., 3 Jun 2026).

The relaxed state is characterized by the minimum interlayer distance $(0.3\ \mathrm{g})$08. Separation is identified when $(0.3\ \mathrm{g})$09, whereas adhesion is identified when $(0.3\ \mathrm{g})$10. The central mechanical result is that the system can exhibit two locally stable states. In the separation state, the graphene sheets remain apart and are held open by CNTs; in the adhesion state, the graphene sheets collapse to graphite-like spacing and the CNTs deform compressively, often becoming elliptical. For an intermediate range of geometric parameters, both states are stable, and the final state depends on the initial condition (Ding et al., 3 Jun 2026).

For a $(0.3\ \mathrm{g})$11 SWCNT with $(0.3\ \mathrm{g})$12, the reported thresholds are $(0.3\ \mathrm{g})$13 when starting in adhesion and $(0.3\ \mathrm{g})$14 when starting in separation, so the system is bistable for

$(0.3\ \mathrm{g})$15

For double-walled CNTs, bistability remains, but $(0.3\ \mathrm{g})$16 increases to about $(0.3\ \mathrm{g})$17, indicating that additional walls stabilize the open configuration (Ding et al., 3 Jun 2026).

The origin of bistability is attributed to competition between van der Waals attraction, which favors sheet adhesion, and elastic deformation energy in graphene and CNTs, which penalizes bending and collapse. Loading–unloading simulations show nonoverlapping paths, a smaller force required for separation on unloading than on loading, and, in some regimes, persistence of the adhered state even at zero applied force. At $(0.3\ \mathrm{g})$18, the system can remain adhered at zero force and require a negative peeling force for separation. The MD compression analysis uses

$(0.3\ \mathrm{g})$19

and

$(0.3\ \mathrm{g})$20

The area enclosed by the loading and unloading curves is identified as energy dissipation (Ding et al., 3 Jun 2026).

AFM nanoindentation on macroscopic rGO–CNT films shows clear hysteresis and higher dissipation energy than pure rGO, while pure rGO exhibits nearly overlapping loading and unloading curves. Statistical histograms and dissipation maps show consistently higher dissipation in CNT-containing regions (Ding et al., 3 Jun 2026). This provides an experimentally supported mechanism by which CNT-stabilized rGO architectures can display compressive resilience and dissipative mechanical response.

6. Comparative interpretation, misconceptions, and design limits

Within the battery study, CNT@rGO is positioned against carbon black (Super P), graphite-like conductive agents, rGO, MWCNT, and rGO&MWCNT. The stated comparison is that carbon black and graphite flakes typically require higher loading for good conductivity, often form less efficient and more tortuous conductive networks, can suffer from agglomeration, provide fewer long-range electron pathways, and are less effective at maintaining contact during long cycling. By contrast, CNT@rGO is described as achieving strong conductivity at low dosage, providing three-dimensional interconnected pathways, ensuring uniform dispersion, exhibiting covalent CNT–rGO integration, and delivering superior rate performance and cycle life (Tang et al., 6 Jul 2025).

A recurrent misconception is to treat all graphene–CNT hybrids as equivalent. The available evidence distinguishes at least two classes. One is the covalently integrated CNT@rGO obtained by direct CNT growth on rGO, for which seamless interfacial junctions, open-ended CNT bases, and $(0.3\ \mathrm{g})$21-membered ring defects are reported. The other is the mixed-dimensional rGO–CNT hybrid assembled from suspensions and reduced into films, which is used to study spacing, restacking, hysteresis, and dissipation. The latter is directly relevant to the mechanics of three-dimensional CNT@rGO, but it is not presented as the same synthesis pathway or the same interfacial chemistry as the covalently integrated CVD-grown material (Tang et al., 6 Jul 2025, Ding et al., 3 Jun 2026).

Another point requiring careful treatment is the scope of the mechanical model. The MD study does not directly simulate a fully three-dimensional CNT@rGO foam; it simulates a two-sheet/two-tube unit that the authors explicitly interpret as a local building block of a macroscopic stacked network. The macroscopic rGO–CNT films are then presented as assemblies of many such local units (Ding et al., 3 Jun 2026). This suggests a mechanistic bridge rather than a one-to-one structural identity between the atomistic model and any specific porous three-dimensional bulk architecture.

Finally, the reported CNT growth yield contains an internal inconsistency, appearing as both $(0.3\ \mathrm{g})$22 and $(0.3\ \mathrm{g})$23 in the same paper. The data support the qualitative conclusion of ultra-high yield, but not a fully harmonized single numerical value (Tang et al., 6 Jul 2025). That limitation does not alter the central picture of CNT@rGO as a low-cost, high-yield, covalently integrated three-dimensional conductive framework whose electrical performance derives from multiple conductive pathways and whose mechanical relevance includes suppression of rGO restacking, bistability of local sheet–tube units, and dissipative response under deformation.