Answers are grounded in this paper and cite their sources.
High-density three-dimensional crossbar arrays are attractive for data-intensive storage and neural-network computing, but their scalability is constrained by sneak currents through half-selected and unselected cells [1] . Suppressing these currents requires two-terminal selectors with high nonlinearity, where nonlinearity refers to the ratio of currents at the full (V op ) and half (V op /2) operating voltages. Such selectors should also offer sufficient drive current, fast response, weak temperature dependence, high endurance and low device-to-device variation. Existing selector technologies rarely satisfy all of these requirements. Phase-transition selectors can suffer from hysteresis and threshold-voltage drift, whereas diffusive memristive selectors rely on repeated filament formation and rupture that limits cycling stability [2] . Tunnel-junction selectors (TJSs) avoid structural switching and therefore offer fast, durable and temperature-stable operation, but their nonlinearity has typically remained approximately 10 5 [3] . In addition, the exponential sensitivity of tunneling to barrier thickness and composition makes device uniformity difficult to control. The central challenge is therefore to increase TJS nonlinearity without sacrificing the reliability and uniformity that make tunneling attractive.
No takes yet. Share an insight, caveat, or question.
Miao et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: