Taking ferrocenecarboxylic acid phosphate ester (P-Fc) as the reference structure, five phosphorus-iron-based fire-extinguishing agents with different P/Fe stoichiometric ratios (1:3, 1:2, 1:1, 2:1, and 3:1) were designed. Their molecular structures are shown in Figure 1, and the corresponding molecular compositions used in the simulation models are summarized in Table 1. All molecular structures were first geometrically optimized using Materials Studio and subsequently packed into simulation cells with a base dimension of 60 Å × 60 Å.

All reactive molecular dynamics simulations were performed using the LAMMPS package^[32], while visualization and post-processing analyses of structural evolution, species formation, and reaction trajectories were conducted using VMD software^[33].

After construction, each system was subjected to energy minimization followed by equilibration at 298 K for 80 ps. This procedure allowed sufficient diffusion and collision among electrolyte molecules, oxygen molecules, and fire-extinguishing-agent molecules, thereby establishing a relatively stable initial configuration and minimizing the influence of initial contact conditions on subsequent statistical analyses.

The simulations were carried out under the canonical ensemble with a time step of 0.25 fs. During the 1 ns simulation, the temperature was linearly increased from 298 to 2,500 K to induce thermal decomposition and combustion reactions. The equilibrated initial configurations of the five systems are shown in Figure 2.

To quantitatively evaluate the combustion inhibition performance of different P/Fe ratios, additional isothermal simulations were conducted over the temperature range of 1,600-2,000 K at intervals of 100 K. Each temperature point was independently repeated three times to reduce statistical uncertainty.

The consumption of EC and EMC molecules was analyzed based on the first-order kinetic model. The first-order dynamic equation is expressed as:

Here, x(t) represents the concentration of EC or EMC molecules at time t, k is the reaction rate constant. Calculate the reaction rate constant corresponding to each temperature by solving formula (Eq. 1). Then, the five rate constants are substituted into the Arrhenius formula (Eq. 2) and the activation energy is obtained by fitting the data. The Arrhenius formula is:

A is the pre-exponential factor (frequency factor); E_a is the apparent activation energy; R is the molar gas constant; T is the temperature.

For comparison, simulations of the blank EC/EMC/O₂ system and the Fe-only system were conducted under identical conditions, and the corresponding results are provided in the Supplementary Materials. The results [Supplementary Figure 1] showed that the Fe-containing system exhibited a clear inhibitory effect relative to the blank system, while the P/Fe = 1:1 system displayed a further improvement in suppression performance. Therefore, in the present work, a series of P/Fe-containing systems with different molar ratios were constructed to further clarify the effects of composition on electrolyte decomposition, radical evolution, and the microscopic suppression mechanism.

To evaluate the extent of combustion reactions in EC/EMC electrolytes under different P/Fe ratios, the consumption behaviors of O₂, EC, and EMC were monitored throughout the simulations. Meanwhile, the formation of major oxidation products, namely CO₂ and H₂O, was tracked. The remaining reactant concentrations and product yields provided direct insight into the combustion characteristics of each system.

Figure 3 shows the time evolution of O₂, EC, and EMC during the simulations. All systems exhibited a broadly similar overall evolution pattern, characterized by relatively slow reactant consumption during the early and intermediate stages, followed by accelerated decomposition and oxidation at the high-temperature stage. Despite this common behavior, clear differences were observed among the different P/Fe ratios. During approximately 150-300 ps, the phosphorus-rich systems (P/Fe = 2:1 and 3:1) showed earlier decreases in EC and EMC, indicating earlier initiation of electrolyte decomposition. Between approximately 300 and 700 ps, reactant consumption proceeded more gradually in all systems. After approximately 700 ps, the numbers of EC, EMC, and O₂ decreased rapidly, corresponding to extensive electrolyte decomposition and oxidation.

The onset and extent of rapid reactant consumption were strongly dependent on the P/Fe ratio. The phosphorus-rich systems entered the rapid-consumption stage earlier, whereas the P/Fe = 1:1 system exhibited the latest onset of pronounced EC and EMC decomposition. The iron-rich systems (P/Fe = 1:2 and 1:3) showed intermediate behavior. Consequently, the P/Fe = 1:1 system retained the largest amounts of both EC and EMC at the end of the simulation.

The evolution of the major oxidation products further supports this trend. As shown in Figure 4, CO₂ formation occurred earlier in the iron-rich systems, whereas H₂O formation began earlier in the phosphorus-rich systems. In contrast, the P/Fe = 1:1 system showed the latest onset of both CO₂ and H₂O formation and produced the lowest final amounts of these products. This indicated that the 1:1 composition most effectively delayed sustained oxidation and limited the conversion of electrolyte molecules into final combustion products.

The final residual fractions of EC and EMC are summarized in Table 2. The P/Fe = 1:1 system retained 98 EC molecules and 83 EMC molecules, corresponding to residual fractions of 28.3% and 28.4%, respectively, which were the highest among all investigated systems. In comparison, both Fe-rich and P-rich compositions showed substantially lower residual fractions. Taken together, the delayed reactant consumption, reduced CO₂ and H₂O formation, and higher final electrolyte retention consistently demonstrate that the P/Fe = 1:1 composition provides the strongest overall suppression of EC/EMC decomposition and oxidation.

To further clarify how the P/Fe ratio influences electrolyte decomposition at the molecular level, detailed reaction-pathway analysis was performed. The P/Fe = 1:1 system exhibited the strongest overall suppression performance; its EC and EMC decomposition pathways were selected for representative analysis, while the corresponding pathways of the other P/Fe systems are provided in Supplementary Figures 2-9 for comparison.

Figure 5 shows the main decomposition pathways of EC in the P/Fe = 1:1 system. The dominant initial pathway was identified as EC → C₂H₄ + CO₃, indicating that cleavage of the carbonate framework represents the primary decomposition mode. Secondary pathways included the formation of C₂H₄O + CO₂ and C₃H₃O₃ + H. Compared with these secondary channels, the C₂H₄ + CO₃ pathway retained a relatively stable carbonate-containing fragment, whereas the other pathways underwent further fragmentation and contributed to the formation of smaller products and radical species.

Figure 6 presents the decomposition pathways of EMC in the same system. In contrast to EC, EMC decomposition was dominated by side-chain cleavage. Two major reaction channels were identified, producing C₂H₃O₃ + C₂H₅ and C₄H₇O₃ + H, respectively, while a minor pathway generated C₃H₅O₃ + CH₃. The two dominant pathways subsequently converged to form the same intermediate combination, C₂H₃O₃ + C₂H₄ + H, which underwent further dehydrogenation and fragmentation reactions.

Comparison with the Fe-rich and P-rich systems shows that variation in the P/Fe ratio does not introduce fundamentally different primary decomposition pathways for EC and EMC, but mainly changes the relative contributions of competing reaction channels. As summarized in Figure 7, for the P/Fe = 1:1 system the fractions of EC and EMC that did not enter the first-step reaction were 28.3% and 28.4%. In addition, the contribution of the relatively stable EC pathway producing C₂H₄ + CO₃ was higher, whereas the contributions of several radical-generating EMC pathways were reduced.

These results indicated that the influence of the P/Fe ratio was expressed primarily through regulation of pathway branching and secondary decomposition rather than through the emergence of new primary reaction types. The P/Fe = 1:1 composition favored greater retention of parent molecules and intermediate species while reducing the contribution of pathways associated with further fragmentation and radical generation.

To further evaluate the influence of the P/Fe ratio on radical-related reactions during EC/EMC decomposition, representative radical species were analyzed based on the reaction-pathway statistics obtained from the ReaxFF-MD simulations. The radical counts reported here represent cumulative occurrence frequencies identified from EC/EMC/O₂ reaction pathways. These statistics therefore reflect the relative activity of radical-generating pathways under different P/Fe conditions.

Figure 8 compares the generation frequencies of representative radicals, including O·, H·, CH₃·, and C₂H₃·. A clear non-monotonic dependence on the P/Fe ratio was observed. The occurrence frequencies of O· generation reached its minimum at P/Fe = 1:2, whereas H· generation was lowest at P/Fe = 1:1. Further increasing the P/Fe ratio led to renewed increases in radical generation. The C₂H₃· generation frequency showed a similar minimum at P/Fe = 1:1, whereas CH₃· generation increased markedly under phosphorus-rich conditions. These results indicate that neither increasing nor decreasing the P/Fe ratio leads to a continuous improvement in the suppression of radical-generating pathways.

The oxygen-containing organic radicals shown in Figure 9 exhibited a similar composition-dependent trend. Most of the analyzed species, including CH₂O·, CH₃O·, and CHO₃·, showed their lowest occurrence frequencies in the P/Fe = 1:1 system, whereas CHO· reached its minimum in the P/Fe = 1:3 system. In contrast, the phosphorus-rich systems generated larger numbers of several oxygen-containing radicals, particularly CHO₃· and C₂H₃O·, with the P/Fe = 3:1 system exhibiting the highest values for several of these species.

Overall, the P/Fe = 1:1 system exhibited relatively low occurrence frequencies for most representative radical-generating pathways, whereas both strongly Fe-rich and P-rich compositions showed more extensive radical formation. This non-monotonic behavior highlights the importance of an appropriate P/Fe ratio for suppressing radical generation. Instead, an appropriate P/Fe ratio is required to limit the formation of key atomic, hydrocarbon, and oxygen-containing radicals, which is consistent with the enhanced electrolyte retention and reduced oxidation-product formation observed in the 1:1 system.

To quantify the synergistic effect associated with the P/Fe ratio, the reactions between extinguishing-agent-derived species and active radicals were further analyzed based on the reaction-pathway statistics of the extinguishing agents. Figure 10 summarizes representative reaction pathways of phosphorus-containing species in the P/Fe = 1:1 system. Phosphorus-containing species such as PO₂· and PHO₂· participated in reactions with H· and OH·, while precursor species including PO· and PH₂O₃ can react with H·, O·, and OH· to form further phosphorus-containing products. These reactions provide multiple pathways for the consumption of key radicals involved in chain propagation.

Figure 11 shows representative radical-consumption reactions involving extinguishing agent-derived species in the P/Fe = 1:1 system, while Table 3 summarizes the corresponding consumption counts for different P/Fe ratios. Considerable differences were observed among the systems. For example, the P/Fe = 1:3 system showed a relatively high number of H· consumption events, whereas the P/Fe = 3:1 system exhibited the highest OH· consumption count. In contrast, the P/Fe = 1:1 system did not show the maximum consumption frequency for every radical species.

It should be emphasized that the radical-generation and radical-consumption statistics were obtained from different reaction pathway datasets. Radical-generation counts were derived from EC/EMC/O₂ decomposition pathways, whereas the consumption counts represent reactions involving extinguishing-agent-derived species. Therefore, these two quantities cannot be directly subtracted as absolute concentrations.

Nevertheless, comparison of the two sets of pathway statistics provides insight into the overall reaction behavior. The strongly Fe-rich and P-rich systems exhibited relatively high radical-generation frequencies together with frequent radical-consumption events, indicating that a high consumption count does not necessarily correspond to stronger suppression. Instead, it may reflect more intensive formation and subsequent reaction of radicals within the system. In contrast, the P/Fe = 1:1 system maintained effective radical-consumption pathways while exhibiting relatively low generation frequencies for most key radicals.

These results indicate that the superior suppression performance of the P/Fe = 1:1 system is not associated with maximizing the absolute number of radical-consumption reactions. Rather, an appropriate P/Fe ratio limits radical generation during EC/EMC decomposition while preserving the ability of extinguishing-agent-derived species to participate in radical-consumption reactions. This combined effect provides a molecular-level explanation for the optimum combustion-suppression performance observed at P/Fe = 1:1.

To quantitatively evaluate the kinetic inhibition effect of different P/Fe ratios, the apparent activation energies of EC and EMC decomposition were calculated from the isothermal simulations using the Arrhenius relationship. The corresponding kinetic parameters are summarized in Table 4, and the Arrhenius fitting results are shown in Figure 12.

As summarized in Table 4, the apparent activation energies of EC ranged from 100.98 to 106.74 kJ/mol. The lowest value was obtained for the P/Fe = 1:3 system, whereas the highest value was observed for the P/Fe = 1:1 system. A more pronounced variation was found for EMC, whose apparent activation energies ranged from 88.68 to 105.57 kJ/mol. Again, the P/Fe = 1:1 system exhibited the highest activation energy, while the P/Fe = 3:1 system showed the lowest value.