^{Cheng Xinbing， Chen Kai， Zhang Dian， Shen Xin， Wu Yuping}

School of Energy and Environment， Southeast University， Nanjing 211189， China

Abstract

Significance To enhance battery safety， this study comprehensively reviews the evolution mechanisms of thermal safety characteristics in large-capacity lithium iron phosphate batteries， thereby facilitating their development in the field of electrochemical energy storage.

Progress The probabilities of thermal runaway in lithium iron phosphate （LFP） batteries and lithium nickel cobalt manganese oxide （NCM） batteries， as well as the deflagration probabilities of their thermal runaway gases， are comprehensively reviewed. First， the probabilities of thermal runaway in LFP and NCM batteries are compared. This comparison is based on characteristic parameters of thermal runaway， including onset temperature， triggering temperature， maximum temperature， temperature rise rate， accumulated heat， mass loss， and combustion possibility. Next， the hydrogen generation mechanisms during the thermal runaway of lithium-ion batteries are summarized. These mechanisms comprise trace water decomposition， protic species reduction， oligomer decomposition， lithium hydride decomposition， hydrogen fluoride reduction， and binder decomposition. These six hydrogen generation mechanisms are investigated according to their reaction locations and temperatures. All six hydrogen generation mechanisms are localized at the anode. Specifically， the simultaneous participation of both the cathode and anode is required for protic species reduction. Furthermore， five hydrogen generation mechanisms， including trace water decomposition， protic species reduction， oligomer decomposition， lithium hydride decomposition， and hydrogen fluoride reduction， occur simultaneously with the decomposition of the solid electrolyte interphase. Because these five hydrogen generation mechanisms occur at relatively low temperatures， they are classified into the first category. In contrast， binder decomposition takes place at a higher temperature stage. At this point， the solid electrolyte interphase is completely decomposed. Consequently， binder decomposition is assigned to a separate category. The process of hydrogen generation is influenced by the attack of reductive gases on the cathode and oxygen on the anode. As a result， a higher proportion of hydrogen is generated in the thermal runaway gases of LFP batteries than in those of NCM batteries， leading to an increased deflagration probability. Finally， the impact of battery capacity on the thermal safety characteristics of lithium-ion batteries is analyzed. As battery capacity increases， the maximum temperature during thermal runaway increases. Simultaneously， the volume of generated gases increases. These factors pose significant challenges to the application and development of large-capacity lithium-ion batteries.

Conclusions and Prospects Compared to NCM batteries， LFP batteries exhibit higher onset temperature and triggering temperature， and lower maximum temperature， temperature rise rate， accumulated heat， mass loss， and combustion possibility during thermal runaway. This indicates a lower probability of thermal runaway for LFP batteries. However， a higher hydrogen proportion is detected in the thermal runaway gases of LFP batteries， leading to a higher deflagration probability. For traditional small-capacity LFP batteries， the thermal runaway gases exhibit a lower temperature and a smaller volume. The gases are easily cooled and diluted by ambient air. Therefore， deflagration conditions are rarely reached. In contrast， thermal runaway gases with a higher temperature and a larger volume are generated by modern large-capacity LFP batteries. When released into the external environment， the gases deflagrate rapidly. Evidently， the high hydrogen proportion in the thermal runaway gases is the primary cause of the reduced safety of LFP batteries. Therefore， suppression measures based on the six hydrogen generation mechanisms need to be implemented. Regarding trace water decomposition， battery materials must be thoroughly dried. Furthermore， the water content in the assembly environment must be strictly controlled. For protic species reduction， functional separators can be designed as physical barriers. These barriers prevent the diffusion of protic species. Alternatively， the thermal stability of the solid electrolyte interphase can be enhanced by introducing electrolyte additives to prevent the reduction of protic species. Additionally， solid-state electrolytes can serve as physical barriers. However， their electrochemical performance and fabrication processes have certain limitations. To address oligomer decomposition and lithium hydride decomposition， the thermal stability of the solid electrolyte interphase needs to be improved. Regarding hydrogen fluoride reduction， lithium hexafluorophosphate should be replaced with more stable lithium salts. Furthermore， additives capable of scavenging hydrogen fluoride can be introduced into the electrolyte. For binder decomposition， the molecular structure of the binder can be redesigned. For instance， cyclic molecular structures can be used to increase the energy barrier for breaking hydrogen-containing chemical bonds. Alternatively， hydrogen-free binders， such as polytetrafluoroethylene， can be applied. Nevertheless， polytetrafluoroethylene suffers from poor adhesion and low conductivity. Beyond suppressing hydrogen generation， battery safety can also be enhanced through early detection and timely warning. Among the gases released during thermal runaway， hydrogen is the most easily detected. Moreover， its warning time is earlier than that of other gases. Therefore， reliable hydrogen sensors are crucial for the safe operation of battery systems.

Keywords：hydrogen； lithium iron phosphate； lithium nickel cobalt manganese oxide； lithium-ion battery； thermal runaway

Get Citation: Cheng Xinbing， Chen Kai， Zhang Dian， et al. Evolution mechanisms of thermal safety characteristics in large⁃capacity lithium iron phosphate batteries［J］. China Powder Science and Technology， 2027， 33（1）： 93-106.

Received: 2026-07-03， Revised： 2026-07-28， Online： 2026-09-15。

Funding：The research was supported by the National Natural Science Foundation of China (Grant No. 92372111).

DOI：10.13732/j.issn.1008-5548.2027.01.007

CLC No.: TB4； O646.21

Type Code: A

Serial No.: 1008-5548（2027）01-0093-14