Research Progress on Synaptic Polymer Device Simulation in Neuromorphic Computing

As the artificial intelligence landscape transitions from the traditional von Neumann architecture toward neuromorphic computing paradigms, synaptic devices have emerged as the cornerstone of next-generation neural systems. Their ability to offer ultra-low power consumption, high parallelism, and non-volatile memory storage makes them indispensable for mimicking biological neural networks. Among the myriad candidate materials, functional polymers stand out due to their unique chemical tunability, mechanical flexibility, and superior electrochemical stability. These characteristics position them as ideal candidates for replicating complex biological phenomena such as Long-Term Potentiation (LTP) and Long-Term Depression (LTD). This review synthesizes the latest advancements in synaptic polymer devices, examining both their fundamental operating principles and their broader application prospects.

Core Mechanisms and Operational Principles

The heart of synaptic polymer devices lies in leveraging the electrochemical properties of polymers to induce reversible structural changes via external voltage or ion stimulation. These physical alterations directly modulate the device's resistance state, effectively simulating synaptic weight updates. The underlying processes typically involve redox reactions within polymer chains, the insertion and extraction of ions, and the swelling or shrinking of the polymer network.

Specific operational mechanisms include:

• Redox Switching Mechanism: Utilizing conductive polymers like polyaniline (PANI) or polypyrrole (PPy), the device toggles between oxidized and reduced states. This transition alters the carrier concentration, resulting in a step-like change in resistance that mimics discrete synaptic states.
• Ion Migration and Trap Mechanism: In insulating or semi-insulating polymer matrices doped with ion conductors, an electric field drives ions between trapping sites. This creates non-uniform electric field distributions that modulate the dielectric breakdown threshold, enabling the continuous tuning of synaptic weights.
• Swelling/Contraction Effect: By exploiting the polymer network's response to solvents or ions, the physical volume of the material changes. This geometric deformation alters the dimensions of conductive pathways, facilitating smooth, analog-like resistance transitions.

Compared to inorganic oxides (such as TaOx or HfOx), polymer devices offer a distinct advantage: their chemical diversity. Through side-chain modification and copolymerization, researchers can precisely engineer band structures and ion transport kinetics, bringing the device behavior much closer to the continuous plasticity observed in biological synapses.

Key Performance Metrics Comparison

Evaluating synaptic polymer devices requires a critical analysis of their performance against biological benchmarks and inorganic counterparts. The following comparison highlights the trade-offs and strengths of polymer-based systems:

• Weight Tuning Range: Polymer devices often exhibit a broader resistance ratio ($R_{on}/R_{off}$), with advanced systems reaching magnitudes between $10^4$ and $10^5$. Crucially, they support sub-linear regulation, more closely approximating the distributed nature of biological synaptic weights.
• Operating Speed: While traditional polymer devices have been limited by ion diffusion speeds (operating in the millisecond range), recent innovations in nano-composite structures have pushed performance into the microsecond regime, meeting the demands of real-time computation.
• Durability: This remains a primary challenge. Repeated cycling can lead to structural fatigue or ion depletion. Although cross-linking network designs and interface passivation techniques have significantly improved longevity, the cycle life (typically $>10^5$ cycles) generally lags behind mature inorganic memory technologies.
• Integration Density: Polymers possess exceptional flexibility and solution processability, enabling roll-to-roll manufacturing. This theoretically allows for ultra-high-density flexible array integration, a significant advantage over rigid inorganic devices that are constrained by lithography limits.

Prominent Material Systems and Application Cases

Current research is heavily concentrated on specific functional polymer systems that have demonstrated breakthrough capabilities in niche applications:

• Conductive Polymers and Composites: Represented by PANI and thiophene derivatives, these materials often incorporate nanotubes or metal oxide fillers to construct "ion-electron" synergistic transport channels. For instance, a study utilizing doped polypyrrole films successfully replicated STDP (Spike-Timing-Dependent Plasticity), demonstrating weight updates that strictly followed Hebbian learning rules under pulsed input sequences.
• Ion Gels and Solid Electrolytes: Serving as ion transport media, these materials address the leakage issues associated with traditional liquid electrolytes. Based on polyethylene oxide (PEO) or acrylate polymers, ion gels maintain high ionic conductivity while providing robust mechanical stability, making them ideal for flexible electronic skins and wearable neural interfaces.
• Self-Assembling Supramolecular Polymers: Constructed via non-covalent interactions such as hydrogen bonding and $\pi$-$\pi$ stacking, these dynamic networks possess intrinsic self-healing capabilities. This feature effectively mitigates device aging and enhances sensitivity to external stimuli, offering a promising path toward stable, long-lasting neuromorphic circuits.

Challenges and Future Outlook

Despite significant progress, the widespread adoption of synaptic polymer devices faces several hurdles. First, cycling stability must be further enhanced to address long-term ion exhaustion and electrode passivation. Second, process consistency remains a bottleneck; solution-based fabrication methods struggle to ensure microscopic structural uniformity at a large scale. Finally, system-level integration poses a complex engineering challenge: efficiently integrating individual devices into large-scale neural network architectures while maintaining compatibility with existing CMOS circuits.

Looking ahead, future research will focus on the deep integration of multi-scale simulation and experimentation. Leveraging machine learning to predict structure-property relationships will accelerate the discovery of new materials. Simultaneously, exploring "ion-electron" dual-mode synergistic mechanisms promises to break through the rate and precision bottlenecks of single-mode transmission. As material science advances, neuromorphic systems based on functional polymers are poised to transition from laboratory prototypes to commercial products within the next decade, laying a solid foundation for the next generation of intelligent hardware.