# Floating Gate **Entity class:** Semiconductor device structure **Domain:** Semiconductor memory / neural hardware **Doc Type:** Developed Technology Node **Maturity:** Developed ## Definition A **floating gate** is an electrically isolated conductor in a transistor that stores charge and thereby changes the device's behavior. Dawon Kahng and Simon Sze demonstrated the floating-gate MOS device at [[wiki/Bell Labs|Bell Labs]] in 1967. ## Machine-Intelligence Context The floating gate became the physical storage principle behind EPROM, EEPROM, and flash memory. Neural-hardware designers also used it as a persistent, adjustable synaptic weight. Intel's [[wiki/Intel 80170NX ETANN|80170NX ETANN]] and the 1996 single-transistor synapse developed by Carver Mead's Caltech group show the same device class serving memory, adaptation, and computation. ## Evidence Ledger - **Established:** Kahng and Sze demonstrated the floating-gate device in 1967. - **Established:** Floating gates became foundational to nonvolatile memory and adaptive neural circuits. - **Strongly indicated:** The device is a durable physical junction between memory and learning hardware. - **Unresolved:** The degree of direct knowledge transfer among the Bell Labs memory lineage, Intel ETANN, and Caltech synapse work; patent citations, personnel records, and laboratory notebooks would settle it. ## Sources / Provenance - [Computer History Museum, “Neural Network Chip Joins the Collection”](https://computerhistory.org/blog/neural-network-chip-joins-the-collection/). - [[articles/A History of Machine Intelligence|A History of Machine Intelligence]]. ## Relationships - **Originated at:** [[wiki/Bell Labs|Bell Labs]]. - **Enables:** [[wiki/NAND Flash|NAND Flash]], [[wiki/Intel 80170NX ETANN|Intel 80170NX ETANN]], and the single-transistor silicon synapse. - **Central to:** [[wiki/Substrate Intentionality|Substrate Intentionality]].