Bio-inspired Organic Small-Molecule Memristor Enabled by Synergistic Electric-Thermal Field Modulation

Memristor-driven neuromorphic computing offers a promising path towards brain-inspired intelligence by emulating the multidimensional plasticity of biological synapses, thereby enabling energy-efficient parallel computation. Nevertheless, the attainment of robust environmental adaptability, particularly in response to fluctuating temperatures, continues to represent a substantial challenge for organic memristors in the context of dynamically modulating synaptic plasticity. In order to address this issue, a bio-inspired cobalt phthalocyanine (CoPc)-based memristor was developed, specifically designed for synergistic electric-thermal field modulation. The device employs the stable planar π-conjugated system of CoPc molecules and exploits dynamic oxygen vacancy (OV) migration at the CoPc/AlOx interface. A comprehensive electrical characterisation was conscucted, incorporating X-ray photoelectron spectroscopy (XPS), in-situ Raman spectroscopy, and temperature-dependent electrical measurements across a wide range (293–473 K). This was supported by physical modelling (SCLC, FNT, Arrhenius) to elucidate the underlying mechanisms. Evidence suggests that the apparatus is capable of effectively replicating essential synaptic plasiticy, encompassing short-term potentiation/depression (STP/STD), paired-pulse facilitation/depression (PPF/PPD), under the regulation of an electric field. The index rose to 151%, indicating a significant increase. Spike-amplitude-dependent plasticity (SADP, 45% weight increase), Spike-timing-dependent plasticity (STDP, ΔW = ± 90%), and learning-forgetting-relearning dynamics were revealed, unveiling cumulative memory effects linked to OV transport. It is crucial to note that the device demonstrates exceptional temperature resilience over the range of 293–473 K, characterised by a linear adaptive shift in its critical voltage (VCritical) from 8.7 V at 293 K to 4.5 V (dVCritical /dT = 0.023 V/K). Physical analysis attributes this adaptive threshold and stable operation to a dual-field synergistic mechanism based on trap-assisted carrier transport, elevated temperature thermally activates carriers, reducing the effective barrier for trap escape and OV migration activation energy (Ea = 0.073–0.312 eV), facilitating conduction via Fowler-Nordheim tunneling (FNT) at lower electric fields. Conversely, lower temperatures necessitate higher electric fields to enhance trap ionization efficiency via the Poole-Frenkel effect, compensating for reduced thermal energy. The exploitation of the linear VCritical-T relationship as a sensitive temperature transduction mechanism was validated through the construction of an intelligent fire warning system. This study incorporated a 6 × 6 CoPc memristor array integrated within household heaters, along with a deep learning model (20 × 16 + 16 × 8 + 8 × 1 fully connected network). The resultant model demonstrated a high abnormal temperature recognition accuracy of 96.54%. This work establishes a novel paradigm for environmentally adaptive neuromorphic devices through molecular/interface design and synergistic multi-field modulation, providing a physical realization of temperature-elastic synaptic operation and demonstrating its practical viability for robust next-generation brain-inspired computing platforms.