Development of opto-electronic synapses based on graphene-like (sp2C)/diamond (sp3C) heterojunctions as “all carbon” ReRAM devices for neuromorphic computing

Electronic devices emulating biological functions, “artificial synapses and neurons,” are essential to advance brain-inspired neuromorphic computing. The integration of allotropic sp2-/sp3-bonded carbon platforms has evoked the realization of neuromorphic devices, that is, memristors. We report the fabrication of nonvolatile heterojunctions comprising graphene-like (sp2C-rich) layers on poly-diamond (sp3C-rich), which are significant for their reduced complexity and nanoscale footprint. Specific architectures include (1) boron-doped carbon nanowalls (BCNW/diamond), (2) diamond/BCNW/diamond, (3) CNW/diamond, and (4) reduced graphene oxide (rGO)/diamond, (5) nanodiamond/diamond, (6) diamond foil/BCNW, and flexible laser-induced graphene (FLM), all of them were synthesized by microwave plasma-assisted chemical vapor deposition, except rGO and LIG. The “all carbon” memristors exhibited lower switching voltages for nonvolatile operation with multiple states recognized via continuous high and low resistance, referring to adaptability in response to synaptic weight, a high switching ratio of ∼104–105 and retention of 104 s, indicating long-term potentiation under electrical (and optical) bias. In addition, devices (DUT 1–3) exhibited linearity and symmetry with high endurance under identical input pulses, which is essential for neural networks. The underlying switching originated from the redox of C–C double bonds at the reorganizational C–sp2–sp3 interface induced by H-terminated diamond paired with oxygen ion movement and local sp2C clustering, as revealed by I–V transport analyses and in situ Raman spectroscopy, respectively. They have potential in photodetection, logic operation, artificial neural networks, and multimodal perception.