Biodegradable Neuromorphic Memristor Arrays with Enhanced Linearity and Synaptic Plasticity for Implantable Bioelectronics

Implantable bioelectronics have revolutionized healthcare, beginning with the cardiac pacemaker and advancing to innovations like brain-machine interfaces (BMI) and neuroprosthetic devices. These systems address previously intractable conditions, marking a shift in medical interventions for complex disorders. However, traditional devices rely on external systems for signal processing, leading to inefficiencies like high power consumption, latency, and privacy risks. Neuromorphic devices, inspired by the brain’s parallel processing capabilities, offer a solution by overcoming the von Neumann bottleneck. Mimicking biological synapses, they enable energy-efficient cognitive functions like learning and memory. Integrating neuromorphic technology allows implantable devices to achieve secure, autonomous, and efficient on-chip functionality, transforming personalized medical care.
In this study, a neuromorphic device with human brain-inspired biomimetic functionality, utilizing fucoidan (Fu) as the active layer material, is presented. Fucoidan, with its biocompatibility, biodegradability, anti-inflammatory, and antioxidant properties, is a highly promising material for implantable devices. These properties enable innovative interfacial functionalization for various medical applications, unlocking potential in transient neuromorphic systems for healthcare. Given these advantages, Fu was selected as the active layer material to develop biodegradable neuromorphic devices. The device was constructed as a two-terminal metal-insulator-metal memristor, mimicking the structure of a two-terminal synapse. Electrochemically active magnesium (Mg) serves as the top metal electrode, while biocompatible and bioresorbable polyhydroxybutyrate (PHB)/polyhydroxyvalerate (PHV) biopolymer is employed as the substrate. These components were integrated to fabricate a 6 × 6 transient memristor array (TMA) of neuromorphic devices (Mg/Fu/W/PHB/PHV), designed to replicate the structure and functionality of biological synapses.
To prepare the device, a Fu solution was synthesized by dissolving Fu powder in deionized (DI) water, stirring vigorously for 1 hour, and aging for 12 hours at room temperature. The bottom electrode (W) was deposited via radio frequency (RF) sputtering through a shadow mask onto the PHB/PHV substrate. During deposition, the Fu solution was filtered using a 0.2 μm polytetrafluoroethylene (PTFE) syringe filter. The Fu solution was then deposited as the active layer by spin-coating. The sample was subsequently annealed at 90 °C for 1 hour in ambient air. The concentration of the Fu solution, spin-coating conditions, and annealing temperature and environment were optimized to ensure uniform set and reset voltages for resistive switching. The I–V curve of the device exhibits non-volatile bipolar switching. The physically transient memristor achieves a high dynamic R_on/R_off ratio of 10³. The Fu-based neuromorphic device also demonstrates key synaptic behaviors, including a paired-pulse facilitation (PPF) index of 131.18%, transitions from short-term memory (STM) to long-term memory (LTM), long-term potentiation (LTP), and long-term depression (LTD). These behaviors are further enhanced by the presence of a MgO_x layer at the interface with the Mg top electrode. The TMA-based artificial neural network (ANN) simulation achieved a 93.91% recognition rate in Modified National Institute of Standards and Technology (MNIST) dataset-based digit classification, comparable to ideal devices. In addition, the device demonstrated complete degradation in a phosphate buffered saline (PBS) solution (pH 7.4) without producing harmful byproducts. Therefore, we expect that this paper will provide comprehensive engineering principles and detailed designs for degradable neuromorphic computing applications, paving the way for future integration of Fu-based neuromorphic devices into complex bioelectronic systems for real-time and in vivo applications.