Methods for Fabricating Artificial Neural Networks (ANN) Based on Doped Semiconductor Elements
AssigneeIBM Corporation
InventorsAli AFZALI-ARDAKANI, Matthew Warren COPEL, James Bowler HANNON, Satoshi OIDA
Life of this patent
Estimating the term…
Abstract
A method of forming semiconductor elements in an artificial neural network, the method including forming a substrate including an oxide layer, forming a Silicon layer on the oxide layer, depositing a thin film dopant layer on the Silicon layer, and controlling a concentration of the dopant in the thin film dopant layer.
Claims21
A method of forming semiconductor elements in an artificial neural network, the method comprising:
- forming a substrate comprising an oxide layer;
- forming a Silicon layer on the oxide layer;
- depositing a thin film dopant layer on the Silicon layer;
- and
- controlling a concentration of the dopant in the thin film dopant layer.
A resistive element in an artificial neural network, the resistive element, comprising:
- a Silicon-on-insulator (SOI) substrate;
- and
- a Silicon layer formed on the Silicon-on-insulator substrate,
- wherein the Silicon layer includes dopants derived from a thin film dopant layer, and
- wherein the thin film dopant layer includes a programmed amount of dopant including at least one of Boron and Phosphorus.
(canceled)
A method of forming resistive elements, the method comprising:
- forming a plurality of resistive elements in an array, each of the resistive element formed by a method comprising:
- forming a Silicon layer on an oxide layer;
- depositing a thin film dopant layer on a Silicon layer;
- and
- controlling a concentration of the dopant in the thin film dopant layer.
Dependencies are inferred from the claim text.