Sungkyunkwan team builds light-controlled artificial neuron crystal

According to Bioengineer, a research team led by Professor Taesung Kim at Sungkyunkwan University developed an optoelectronic synaptic device built from a designable van der Waals (vdW) crystal that operates under optical stimuli and mimics neuron and synapse functions. Bioengineer reports the crystal was produced via a single-step sulfurization using Ar + H2S plasma applied to bulk ReSe2, converting the top layer into a nano-crystalline film while preserving an underlying bulk single-crystalline layer. Reporting by Asia Research News describes the device as enabling a structural analogy to light-sensitive ion channels and notes the process aims to address common vdW-material problems such as uncontrollable grain boundaries, intercalation, polymer residue, mechanical warpage, and nonuniform large-area crystallinity. Editorial analysis: For practitioners, the result highlights materials-level approaches to optically driven neuromorphic sensors, relevant for compact edge vision accelerators.
What happened
According to Bioengineer, a team led by Professor Taesung Kim at Sungkyunkwan University created an optoelectronic synaptic device built from a designable van der Waals (vdW) crystal that emulates neuronal and synaptic functions when stimulated with light. Bioengineer reports the material was synthesized by applying a single-step sulfurization process using Ar + H2S plasma to bulk ReSe2, producing a two-tier structure in which the surface converts to a nano-crystalline ReSe2 layer while the underlying bulk single-crystalline ReSe2 remains intact. Asia Research News frames the architecture as mimicking the biphasic nature of neuronal membranes, with the nano-crystalline surface layer acting analogously to light-sensitive ion channels, per reporting.
Technical details
Editorial analysis - technical context: Layered vdW materials are attractive for optoelectronic neuromorphic devices because of atomic-scale thickness and strong optical responses; optoelectronic synapses modulate conductance in response to light and are a common building block for neuromorphic vision systems. The reported Ar + H2S single-step sulfurization technique, as described in Bioengineer, produces a nanoscale grain network on the surface while preserving a crystalline substrate, which addresses material defects-reported problems include uncontrollable grain boundaries, intercalation, polymer residue, interface warpage, and inconsistent large-area crystallinity.
Context and significance
Industry context: Materials-level solutions that combine optical sensitivity with controllable nanoscale structure are an important path toward energy-efficient, compact neuromorphic sensors for edge AI. The device described in reporting could inform designs for stackable, 3D-integrated photonic-electronic neuromorphic arrays because it offers a structural route to combine light responsiveness and stable bulk crystalline support. For hardware practitioners, improvements in uniformity and interface integrity at the wafer scale are typical gating factors for moving lab devices toward larger-area arrays and system integration.
What to watch
For practitioners: follow peer-reviewed publication details and measured device metrics (switching energy, retention, endurance, optical-to-electrical responsivity) when they are released. Also monitor replication by independent groups, demonstration of pixelated arrays or 3D stacking, and any presentations that quantify variability and yield across wafer-scale samples. Reporting to date does not include standardized electrical performance benchmarks or commercialization timelines.
Scoring Rationale
The report describes a novel materials-process that could materially affect neuromorphic sensor hardware, a notable development for edge AI hardware research. Impact is limited until independent replication and benchmarked device metrics are published.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems

