Graphene In-Sensor Compute Device for Plant Hydration Monitoring

Analog sensing devices that exhibit artificial synaptic behavior offer a promising path toward scalable and energy-efficient environmental sensing. In this work, we report graphene-based, leaf-gated in-sensor compute devices for plant hydration monitoring that are capable of both hydration sensing and synapse-like conductance modulation. These devices measure plant water content through trends in channel conductance while simultaneously encoding memory-like states in response to electrical stimulation. Conductance changes track hydration-dependent ion mobility in the leaf with higher (lower) updates in hydrated (dehydrated) states. Devices show linear potentiation and depression and short-term memory retention, supporting their suitability for neuromorphic edge applications. When deployed on Monstera leaves, the devices maintain ultralow power operation (23 aJ/μS write energy/conductance update and 0.23 μW read power) and minimal weight (9 mg) and cause no disruption to plant physiology. By integration of computation and sensing into a single biocompatible platform, this approach minimizes data transmission requirements and enables real-time, long-term hydration monitoring.