Atomic-Scale Mechanisms of Catalytic Recombination and Ablation in Knitted Graphene Under Hyperthermal Atomic Oxygen Exposure

Abstract

Effective ablative thermal protection systems are essential for ensuring the structural integrity of hypersonic vehicles subjected to extreme aerothermal loads. However, the microscopic reaction mechanisms at the gas–solid interface, particularly under non-equilibrium high-enthalpy conditions, remain poorly understood. This study employs reactive molecular dynamics (RMD) simulations with the ReaxFF-C/H/O force field to investigate the atomic-scale ablation behavior of a graphene-based knitted graphene structure impacted by atomic oxygen (AO). By systematically varying the AO incident kinetic energy (from 0.1 to 8.0 eV) and incidence angle (from 15° to 90°), we reveal the competing interplay between catalytic recombination and ablation processes. The results show that the catalytic recombination coefficient of oxygen molecules reaches a maximum at 5.0 eV, where surface-mediated O₂ formation is most favorable. At higher energies, the reaction pathway shifts toward enhanced CO and CO₂ production due to increased carbon atom ejection and surface degradation. Furthermore, as the AO incidence angle increases, the recombination efficiency decreases linearly, while C-C bond breakage intensifies due to stronger vertical energy components. These findings offer new insights into the anisotropic surface response of knitted graphene structures under hyperthermal oxygen exposure and provide valuable guidance for the design and optimization of next-generation thermal protection materials for hypersonic flight.

1. Introduction

The continuing advancement of hypersonic technologies has markedly increased spacecraft operating velocities, exposing vehicles to increasingly extreme thermal and mechanical conditions. During atmospheric re-entry, intense aerodynamic heating arises from frictional and collisional interactions between the spacecraft and atmospheric particles, giving rise to high-temperature, thermochemical non-equilibrium flows [1,2,3,4]. These conditions induce severe surface heating, material ablation, and structural degradation, ultimately compromising vehicle integrity and complicating flight trajectory prediction [5,6,7,8,9]. Carbon fiber-reinforced materials have been extensively utilized in aerospace engineering [10,11,12] owing to their exceptional mechanical robustness [13], thermal stability [14], and resistance to high-temperature erosion [15]. As core ablative components in thermal protection systems (TPS), their structural durability is essential to ensuring mission reliability and longevity under high-enthalpy conditions [16,17,18].

To date, numerous experimental [19,20,21,22] and computational efforts [23,24,25,26] have been undertaken to elucidate the ablation mechanisms of carbon-based composites. Finite element modeling, such as the work by Wang et al., has revealed upstream migration of ablation fronts and intensified downstream erosion in C/SiC composites under hypersonic flow conditions [27]. Experimental investigations employing plasma jets have further clarified oxidation behavior at elevated temperatures, while the integration of SiBCN-derived ceramic coatings has been shown to dramatically reduce mass loss at 1500 °C [28]. Despite these advances, experimental techniques remain constrained by high cost and limited temporal-spatial resolution at the atomic scale. Similarly, conventional finite element approaches—although useful—typically rely on predefined parameters and lack the capability to resolve dynamic surface evolution and complex chemical kinetics during ablation.

To address these limitations, Reactive Molecular Dynamics—which integrates classical molecular dynamics with the ReaxFF reactive force field—has emerged as a powerful tool for probing atomic-scale ablation phenomena [29,30,31,32,33,34]. RMD enables direct visualization of bond dissociation, radical formation, and reaction-driven morphological evolution [35,36,37,38,39,40]. Previous RMD studies have shed light on oxidation processes in highly oriented pyrolytic graphite (HOPG), demonstrating that carbon removal preferentially initiates at defect edges [41]. Investigations into energy transfer during oxygen–graphite collisions [42], kinetic energy-dependent defect formation [43], and angle-sensitive erosion [44] have further emphasized the critical role of local impact conditions in dictating material degradation pathways. Nevertheless, the majority of prior research has centered on planar or multilayer graphene structures, while the ablation behavior of architected carbon configurations—such as woven or knitted graphene—remains largely unexplored.

In contrast to planar configurations, architected carbon systems—defined by mesoscale interlocking, anisotropic porosity, and intricate surface features—exhibit structural complexity that is hypothesized to yield fundamentally distinct ablation behavior. These features are particularly relevant under hyperthermal atomic oxygen (AO) exposure, where thermochemical non-equilibrium effects dominate, as in hypersonic flight environments [45]. The present study systematically investigates the atomic-level ablation behavior of knitted graphene under AO bombardment using ReaxFF-based RMD simulations. By modulating incident kinetic energies and angles of attack, we emulate the aerodynamic environments experienced by re-entering spacecraft. The simulations reveal how structural weaving modulates reaction front propagation, defect nucleation, and carbon atom ejection. These insights not only expand the current understanding of ablation mechanisms in architected carbon materials but also offer predictive guidance for the rational design of next-generation TPS for extreme aerothermal applications.

2. Model and Methods

2.1. Model Construction

To simulate the high-temperature gas–solid heterogeneous ablation of knitted graphene under AO impact, a reactive model was constructed, as illustrated in Figure 1a. In this setup, the gas-phase AO atoms collide with a solid-phase knitted graphene sheet composed of interwoven graphene nanoribbons (GNRs). Zigzag-edged GNRs were selected to eliminate chirality effects. Graphene nanoribbons (GNRs) were constructed using the Nanotube Builder module in Visual Molecular Dynamics (VMD) [35]. The bent GNRs were defined with dimensions of 1 nm (x-direction) and 6 nm (y-direction), while the straight GNRs, interwoven between the bent layers, were set to 2 nm × 2.5 nm. Complete atomic models were generated and exported, and their coordinates were subsequently processed using a custom Fortran script to assemble the knitted configuration. The out-of-plane deformation of the bent GNRs was described by a one-dimensional sinusoidal function.

$$w_1\left(x\right)=A_s\mathrm{s}\mathrm{i}\mathrm{n}({\displaystyle \frac{2\pi x}{\lambda }})$$

where $A_s$ and $\lambda$ denote the amplitude and wavelength of the curvature, respectively, which were set to set to 0.34 nm and 5.0 nm in this study.

The final structure consisted of two bent and two interlaced straight nanoribbons, yielding an effective model size of L_x = 27.02 Å and L_y = 55.56 Å. Previous studies have shown that thermal rectification is highly sensitive to nanoscale geometric features [46]. Therefore, parameters such as L_x, L_y, and $A_s$ can be tuned to modulate the morphology and thermal transport properties of the knitted system, thereby enhancing its utility in thermal protection applications. Due to the computational intensity of the ReaxFF reactive force field—particularly its low parallel efficiency—the model was constrained to a minimal representative unit containing 1040 carbon atoms to balance simulation accuracy and cost.

2.2. Simulation Details

RMD simulations were conducted to explore the high-temperature ablation behavior of knitted graphene under hyperthermal atomic oxygen (AO) impact. The ReaxFF-C/H/O force field, developed by Chenoweth et al. [45], was employed to model the reactive interactions between AO and carbon atoms. This potential has been widely validated in previous studies through comparisons with experimental data and first-principles calculations [40,41]. The knitted graphene structure consisted of 1040 carbon atoms and occupied a simulation cell measuring 27.02 Å × 55.56 Å × 13.99 Å in the x, y, and z directions, respectively. AO atoms were introduced sequentially at randomly selected lateral positions 5 Å above the highest surface atom, with a fixed time interval of 1.0 ps [47]. The z-direction height of the entire simulation box was extended to 240 Å to prevent unphysical recombination events at the gas–solid interface. Simulations were performed at a constant temperature of 1100 K, while systematically varying the incident kinetic energy and angle of incoming AO atoms to examine different gas-phase impact conditions. To ensure mechanical stability under continuous bombardment, carbon atoms located at the four corners of the graphene sheet were fixed to stabilize the system’s center of mass along the z-direction. Periodic boundary conditions were applied in the x- and y-directions, with a reflective boundary imposed along the z-direction. Prior to AO injection, energy minimization was performed using the conjugate gradient algorithm to relax the system to equilibrium. Temperature was then regulated using the Berendsen thermostat under the NVT ensemble [48]. Each simulation was run for 500 ps with a timestep of 0.10 fs. Atomic trajectories were saved every 1000 steps for post-simulation analysis. All reported results in this study were obtained from ten independent RMD simulations performed under identical thermodynamic and structural conditions, with variations introduced only through different initial atomic velocity distributions generated by distinct random seeds. All simulations were carried out using the LAMMPS package (lammps-2Aug2023) [49], and structural visualization was performed using OVITO (ovito-3.11.0) [50].

Figure 1. (a) Atomic structure of knitted graphene consisting of orthogonally interwoven graphene nanoribbons, presented from top, side, and 3D perspectives. The configuration displays periodic interlocking and directional porosity. (b) The breakup process of the knitted graphene due to successive atomic oxygen impacts (carbon atoms are indicated in red and oxygen atoms are indicated in cyan) with (c) a comparison of the result of Srinivasan and van Duin [48]. (d) Comparison of the ratio of carbon-carbon bonds broken between two-layer graphene and knitted graphene.

Figure 1. (a) Atomic structure of knitted graphene consisting of orthogonally interwoven graphene nanoribbons, presented from top, side, and 3D perspectives. The configuration displays periodic interlocking and directional porosity. (b) The breakup process of the knitted graphene due to successive atomic oxygen impacts (carbon atoms are indicated in red and oxygen atoms are indicated in cyan) with (c) a comparison of the result of Srinivasan and van Duin [48]. (d) Comparison of the ratio of carbon-carbon bonds broken between two-layer graphene and knitted graphene.

2.3. Model and Validation

To assess the validity of the simulation approach, our results were benchmarked against the reference data of Srinivasan and van Duin [48], which describe the ablation dynamics of monolayer graphene. As illustrated in Figure 1b, the graphene surface undergoes a typical degradation sequence: initial epoxide formation from 0 to 40 ps, emergence of nanoscale pores by 80 ps, and extensive fragmentation by 100 ps due to defect propagation. This progression closely matches the temporal evolution reported in their work, as further shown in Figure 1c [48], thereby confirming the reliability of the current simulation setup. Figure 1d presents a comparative analysis of C–C bond rupture ratios in bilayer planar graphene versus bilayer knitted graphene. In the planar structure, the upper layer exhibits rapid bond dissociation after 100 ps, reaching a maximum by 200 ps. The bottom layer remains largely intact until the upper layer is nearly ablated, reflecting a layer-by-layer ablation mechanism. Both layers eventually reach full bond breakage, indicating complete structural disintegration.

In contrast, the knitted graphene structure shows a markedly slower damage progression. By the end of the simulation, only ~60% of C–C bonds are broken, and the ablation front continues to evolve gradually. This enhanced resistance is attributed to the vertically interlaced architecture, which facilitates multidirectional energy dispersion and inhibits localized failure. Unlike planar graphene, where energy accumulates within discrete atomic layers, the knitted graphene architecture disrupts sequential ablation by promoting isotropic load sharing. These results underscore the critical role of three-dimensional interconnectivity in ablation resistance and highlight the potential of knitted graphene architectures for advanced thermal protection applications.

3. Results and Discussion

3.1. Effects of AO Incident Kinetic Energy on the Ablation of Knitted Graphene

As the flight speed of hypersonic vehicles increases, the associated surface heat flux density rises significantly, leading to high-enthalpy flows composed of reactive atomic species with elevated kinetic energy. To examine the effect of incident energy on the ablation dynamics of knitted graphene, six simulation models were constructed with vertically impinging atomic oxygen (AO) at kinetic energies of 0.1, 2.0, 4.0, 5.0, 6.0, and 8.0 eV. Although AO particles in low-Earth orbit typically exhibit hyperthermal kinetic energies exceeding 1 eV, the 0.1 eV case was deliberately included to represent a near-thermal condition corresponding to room-temperature translational motion (∼0.03–0.1 eV). This low-energy case serves as a non-reactive baseline, enabling clear differentiation between inert scattering behavior and chemically driven erosion observed at higher energies. Furthermore, this reference case provides a threshold benchmark to assess the onset of oxygen adsorption, molecular recombination, and ablation, thereby enriching the energy-dependent mechanistic understanding of AO–surface interactions. The morphological response of the graphene surface under AO bombardment was analyzed by tracking atomic trajectories, as illustrated in Figure 2a. A strong correlation was observed between the AO incident energy and adsorption behavior. At the lowest energy (0.1 eV), only 38 oxygen atoms were adsorbed, whereas at 8.0 eV, the number increased to 200—approximately a 5.3-fold rise. This monotonic increase indicates that higher kinetic energy enhances both the frequency of adsorption events and the accessibility of reactive surface sites. In the low-to-moderate energy regime (0.1–5.0 eV), adsorption differences are more pronounced, while in the higher range (5.0–8.0 eV), the increase becomes less significant, suggesting an approach to site saturation. Throughout the simulation, continuous AO bombardment progressively occupied surface sites, and a dynamic equilibrium was achieved during the final 50 ps of each run. Final structural snapshots for three representative incident energies are presented in Figure 2b. Clear differences in oxygen coverage and structural integrity can be observed, alongside the formation of gas-phase species such as O₂, CO₂, and CO. Notably, higher incident energies result in more severe lattice fragmentation, underscoring the critical role of energy magnitude in determining the extent of ablation-induced damage.

To elucidate the ablation mechanisms at high incident energy, the 8.0 eV case was selected for detailed analysis of gas-phase product evolution and structural changes over time, as shown in Figure 3. As shown in Figure 3a, three primary gas-phase products—O₂, CO₂, and CO—were formed during the thermal-chemical ablation of knitted graphene under 8.0 eV oxygen atom bombardment. Among them, O₂ appeared earliest and accumulated in the greatest quantity throughout the simulation, reaching over 70 molecules by 500 ps. This rapid and abundant generation is attributed to the catalytic nature of the graphene surface, which facilitates the recombination of incident oxygen atoms into O₂ during the early stages of exposure [51]. CO₂ production followed a slightly delayed but steady trend, ultimately reaching approximately 55 molecules. In contrast, CO formation exhibited a clear time lag and remained at significantly lower levels, totaling around 30 molecules by the end of the simulation. The delayed onset and lower yield of CO suggest that it is primarily generated through CO₂ decomposition, with additional contribution from the gas-phase oxidation of free carbon atoms in the later stages of ablation. Figure 3b illustrates the temporal evolution of the knitted graphene structure under 8.0 eV AO bombardment. As the simulation progresses, progressive degradation of the lattice can be observed, marked by the gradual loss of carbon atoms and the formation of surface vacancies and pores. These structural disruptions originate from the continuous interaction between high-enthalpy AO and the graphene framework, and they become increasingly pronounced with simulation time. The emergence and growth of pores provide direct visual evidence of ablation-induced damage and serve as a basis for quantifying the extent of material erosion over time.

Figure 4 presents a comprehensive evaluation of the catalytic recombination behavior of AO on knitted graphene under varying incident kinetic energies ranging from 0.1 eV to 8.0 eV. As shown in Figure 4a, the number of O₂ molecules formed increases with kinetic energy and reaches a maximum at 5.0 eV, after which it declines. The number of recombined O₂ molecules grows from 38 at 0.1 eV to 200 at 5.0 eV, representing a 5.3-fold enhancement. This kinetic-energy-dependent trend suggests that moderate AO energies are most conducive to O₂ formation. To explain this behavior, two surface-mediated reaction pathways are considered, as illustrated in Figure 4b,c. The Eley–Rideal (E–R) mechanism (Figure 4b) involves a direct reaction between a gas-phase O atom and a surface-adsorbed O atom, while the Langmuir–Hinshelwood (L–H) mechanism (Figure 4c) entails the surface diffusion and recombination of two adsorbed oxygen atoms. In both cases, the knitted graphene architecture acts as a catalytic platform that facilitates the formation of molecular oxygen through the stabilization of reactive intermediates. To quantify recombination performance, the surface catalytic recombination coefficient γ is introduced (Figure 4d) and defined as γ = N₁/N₂, where N₁ is the number of recombined O atoms and N₂ is the total number of incident O atoms. As shown in Figure 4d, γ increases with time and reaches a steady state after approximately 300 ps across all energy levels. The time-averaged γ value, γ_ave, calculated over the final 50 ps of simulation, is presented in Figure 4e. The data indicate a clear peak in γ_ave at 5.0 eV (~0.36), which is approximately nine times higher than that at 0.1 eV (~0.04). This trend highlights a non-monotonic relationship: while increasing energy promotes oxygen adsorption and recombination at moderate levels, excessive energy leads to new competing effects that reduce catalytic performance.

To further explore these effects, Figure 5a,b display the generation of CO₂ and CO as a function of kinetic energy. At energies above 5.0 eV, the production of CO and CO₂ becomes more prominent, indicating a shift in the dominant reaction pathway. Specifically, CO₂ is the major oxidation product at all energy levels, but CO production increases significantly at 6.0 eV and beyond. At 8.0 eV, both CO and CO₂ formation rates nearly double compared to 5.0 eV, accompanied by the appearance of significant lattice fragmentation. These results suggest that high-energy oxygen atoms not only drive recombination but also cause carbon-carbon bond breakage, enhancing gas-phase oxidation and surface erosion. This transition reflects the competing influence of adsorption-driven recombination and ablation-induced degradation. At lower energies, oxygen atoms are efficiently captured at surface-active sites, enhancing recombination via E-R and L-H pathways. As the energy increases beyond 5.0 eV, however, oxygen atoms acquire sufficient kinetic energy to break C–C and C–O bonds, reducing the density of available reactive sites and favoring the formation of CO_x; species instead. This competitive mechanism explains the observed peak in both γ and O₂ production at 5.0 eV. Similar energy-dependent behaviors have been observed by Vinogradov et al. [52] and Singh et al. [53], where enhanced CO and CO₂ evolution coincided with diminished recombination efficiency on carbon-based surfaces. Taken together, these findings demonstrate that 5.0 eV represents a critical threshold energy at which knitted graphene maximizes its catalytic function before ablation effects begin to dominate. The ability to tune incident energy for optimal recombination performance is crucial for designing heat shield materials capable of surviving in high-enthalpy, oxygen-rich reentry environments. To quantitatively assess this ablation behavior, two parameters are introduced: the carbon atom loss ratio and the average carbon atom loss rate k, defined as the slope of the carbon loss curve over time. A consistent bond breakage criterion—defined by a C–C bond length exceeding 1.8 Å–was applied throughout the analysis to identify ablation events. As shown in Figure 5c, carbon loss remains negligible during the early stages of AO exposure for all kinetic energy levels, indicating that initial interactions are governed predominantly by surface adsorption and catalytic recombination. In this phase, incident oxygen atoms are preferentially adsorbed onto reactive sites, facilitating O₂ formation through Eley–Rideal and Langmuir–Hinshelwood pathways, while the knitted graphene framework retains structural integrity. However, as the exposure continues and oxygen flux accumulates, the frequency and intensity of reactive collisions increase, particularly at higher energies. This leads to the progressive displacement of carbon atoms, transitioning the surface response from recombination-dominated to ablation-driven. Notably, at 8.0 eV, the carbon atom loss ratio approaches 0.09—approximately 22 times that observed at 0.1 eV—highlighting the pronounced influence of kinetic energy on ablation severity. Figure 5d further substantiates this trend by demonstrating a strong linear correlation between the incident kinetic energy and the average ablation rate k. This contrasts with the previously observed non-monotonic behavior of the catalytic recombination coefficient, indicating that ablation intensifies steadily with increasing energy. This divergence underscores a mechanistic shift: while moderate kinetic energies (≤5.0 eV) promote catalytic recombination by enhancing surface reactivity, excessive energies induce significant C–C bond breakage, leading to increased gasification and structural erosion. Collectively, these findings demonstrate that knitted graphene exhibits a dual functional response: initially facilitating exothermic recombination of reactive oxygen species, and subsequently undergoing energy-driven ablation. Clarifying the balance between these competing mechanisms provides valuable insight for the design of thermally robust, oxidation-resistant materials for hypersonic flight applications.

3.2. Effects of AO Incident Angle on the Ablation of Knitted Graphene

To further assess the influence of flow orientation on ablation behavior, a series of simulations were performed by varying the angle of atomic oxygen (AO) incidence (θ = 15°, 30°, 45°, 60°, 75°, and 90°), where θ is defined as the angle between the AO particle’s initial velocity vector and the x–y surface of the knitted graphene structure. Here, θ = 90° corresponds to normal incidence (along +z), while θ = 0° corresponds to grazing incidence (parallel to the surface). Oblique incidence was generated by rotating the AO velocity vector within the y–z plane, keeping the x-component zero to avoid lateral motion. This configuration ensures a well-defined incidence geometry. All cases were conducted at a constant surface temperature of 1100 K with an incident kinetic energy of 5.0 eV, a representative value for AO particles encountered in low Earth orbit [49]. The normal-incidence model (θ = 90°) served as the baseline for comparison. Figure 6 illustrates the structural evolution of the knitted graphene sheet at selected time intervals under oblique (θ = 15°, Figure 6a) and normal (θ = 90°, Figure 6b) AO impacts. Over the 500 ps simulation period, the normal-incidence case exhibits significantly more extensive surface damage, including the formation of larger and more numerous pores. In contrast, the oblique-incidence model retains more of its structural integrity, with fewer ejected atoms and smaller defect regions. These differences become especially pronounced after 300 ps, highlighting the cumulative effect of sustained vertical bombardment. This comparison underscores the directional dependence of ablation under hyperthermal AO exposure. While oblique impacts reduce the local energy deposition per site, normal impacts concentrate energy vertically, accelerating carbon dislodgement and defect propagation. Such angular effects are critical for optimizing the design of TPS, which must withstand anisotropic flow fields during hypersonic flight and atmospheric reentry.

To further elucidate the role of AO incidence angle on structural degradation, ablation behavior is quantitatively assessed by tracking C–C bond breakage over time and calculating the total number of broken bonds. Figure 7a displays the time evolution of C–C bond dissociation under six incidence angles (θ = 15° to 90°). The results reveal a clear positive correlation between the incidence angle and the number of broken bonds. Specifically, the θ = 90° (normal incidence) case shows the highest bond dissociation, while θ = 15° leads to the least. This trend arises from the increased normal component of the kinetic energy at higher angles, which enhances energy transfer into the graphene lattice and promotes bond rupture. Bond breakage initiates immediately after AO impact, and the slope of each curve reflects the ablation rate. The steeper the slope, the faster the rate of C–C bond loss, implying that higher incidence angles correspond to higher ablation rates. Among all cases, the θ = 90° model exhibits the fastest rate of degradation, driven by maximal vertical energy input and minimal lateral dissipation. To quantify the cumulative damage, Figure 7b summarizes the total number of C–C bonds broken at the end of the 500 ps simulation. The data show a near-linear increase in total bond dissociation with increasing incidence angle, reaffirming that more vertical AO impacts induce more severe lattice disruption. This is consistent with the energy deposition mechanism, whereby vertically incident atoms deliver their full kinetic energy into the graphene surface, maximizing the probability of structural failure. The catalytic recombination behavior of oxygen atoms at six different incidence angles is quantitatively assessed to evaluate the influence of angle on surface reactivity. As shown in Figure 7c, the surface recombination coefficient γ increases over time for all cases and reaches a fluctuating equilibrium after approximately 400 ps. Notably, the equilibrium level of γ varies significantly with angle: the lowest angle (θ = 15°) achieves the highest recombination efficiency, while the vertical incidence case (θ = 90°) exhibits the lowest value. This suggests that lower incidence angles favor surface-mediated catalytic activity. This trend is further quantified by the time-averaged recombination coefficient γ_ave, illustrated in Figure 7d. A clear linear decrease in γ_ave is observed with increasing angle θ, indicating that higher angles suppress catalytic recombination efficiency. This behavior arises from the trajectory dynamics of incident oxygen atoms: at lower angles (e.g., 15°), atoms tend to glide along the surface, increasing the probability of interacting with pre-adsorbed oxygen atoms and facilitating recombination via Eley–Rideal or Langmuir–Hinshelwood mechanisms. In contrast, at θ = 90°, the oxygen atoms impact the surface nearly perpendicularly, enhancing momentum transfer and bond breakage while reducing the chance of recombination. Taken together, these results reveal a trade-off between catalysis and ablation as a function of incidence angle. While oblique impacts promote surface-mediated recombination, vertical impacts enhance bond scission and gas-phase oxidation, shifting the dominant pathway from catalytic recombination to ablative degradation. This inverse correlation is critical for assessing the orientation-dependent performance of thermal protection systems in high-enthalpy, oxygen-rich environments, where surface geometry directly influences the balance between energy dissipation and material erosion. To isolate the intrinsic ablation behavior of the knitted graphene structure under AO impact, all simulations in this study were performed using fully periodic boundary conditions in the in-plane directions. This modeling approach eliminates free edges and grain boundaries, thereby avoiding boundary-induced artifacts and enabling a focused analysis of atomic-scale reaction mechanisms within the structurally repeating unit. While this setup allows for clear interpretation of energy-dependent reaction pathways, it also represents an idealized configuration. In real experimental systems, structural imperfections—such as edge terminations, grain boundaries, and vacancy defects—are known to serve as preferential initiation sites for AO-induced damage. Therefore, the damage resistance observed in our simulations may represent a lower bound of susceptibility.

4. Conclusions

This study employed reactive molecular dynamics simulations with the ReaxFF-C/H/O force field to investigate the coupled catalytic recombination and ablation behavior of knitted graphene under hyperthermal atomic oxygen (AO) impact. By varying the AO kinetic energy from 0.1 to 8.0 eV and the incidence angle from 15° to 90°, we revealed how structural geometry and energetic parameters modulate surface reactions at the atomic scale. The knitted graphene exhibited dual functionality-facilitating oxygen recombination at moderate energies while undergoing progressive degradation at higher energies. The catalytic recombination coefficient peaked at 5.0 eV with a time-averaged value of 0.36, but declined as surface-active sites were depleted by intensified C–C bond breakage and CO/CO₂ formation at higher energies. Similarly, increasing the incidence angle reduced catalytic efficiency (γ_ave dropped from 0.31 to 0.09) and amplified carbon loss, highlighting the directional dependence of ablation severity. Overall, the results demonstrate a clear competition between catalytic recombination and ablation, with 5.0 eV identified as a critical energy threshold. These findings provide mechanistic insight into the energy- and angle-dependent surface responses of architected carbon materials, offering a theoretical foundation for the optimization of thermal protection systems in high-enthalpy, oxygen-rich environments.