A New Way to Benchmark Low-Dimensional Optoelectronic Synapses

The dynamic retention of low-dimensional optoelectronic synapses is conventionally modeled using the Kohlrausch–Williams–Watts (KWW) stretched exponential function. However, the systematic extraction of the energetic dispersion parameter (β) as a freely adjustable phenomenological fitting feature represents a fundamental physical oversight. By analyzing state-of-the-art transient photoresponse data from structurally diverse 1D (InAs, WS2) and 2D (SnSe2,WSe2) nanomaterials-based field effect transistors, we expose a profound thermodynamic anomaly: despite immense variations in geometric dimensionality and macroscopic retention times, β is strictly locked within a narrow window. To resolve this ambiguity, we import Phillips' diffusion-trap model, revealing a dimensionality collapse in 2D van der Waals planar networks. Spatial inhomogeneities and charge puddling force carriers into restricted, quasi-1D percolation pathways, structurally locking their thermodynamic relaxation into the same highly disordered topological regime as 1D nanostructures. Consequently, β is fundamentally insufficient to classify synaptic volatility. Recognizing this topological constraint, we introduce the dimensionless Adaptive Plasticity Index (AP), which takes into account the ratio between the relaxation time constant and the learning time window. By projecting the complex trap dynamics onto a Phase Map, AP establishes a robust, material-agnostic physical standard to unambiguously benchmark the transition from Short-Term Synaptic Plasticity (STSP) to Long-Term Memory (LTM).