IP Library Granted Patent US 12,417,802
Granted Patent B2
US 12,417,802 · App. 17/691,733 · Granted Sep 16, 2025

Semiconductor device and method of driving the same

Inventors: Takumi Mikawa (Shiga, JP); Koji Katayama (Nara, JP); Ryutaro Yasuhara (Hyogo, JP)
Assignee: PANASONIC HOLDINGS CORPORATION
G11C11/54G06N3/065G11C13/0007
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,417,802
App. No.
17/691,733
Granted
Sep 16, 2025
Kind
B2
Abstract

A semiconductor device includes variable resistance elements on a semiconductor substrate. Each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable. The variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable.

Claims (56)

1. A semiconductor device comprising:

variable resistance elements on a semiconductor substrate, wherein

each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable,

the variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable, and

an area of the filament in a plan view of the semiconductor substrate differs according to the neural network weight.

2. A semiconductor device comprising:

variable resistance elements on a semiconductor substrate, wherein

each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable,

the variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable, and

a length of the filament in a direction in which the first electrode and the second electrode face each other differs according to the neural network weight.

3. The semiconductor device according to claim 1 , wherein

the variable resistance layer includes:

a first resistance layer; and

a second resistance layer that has a resistance value higher than a resistance value of the first resistance layer and includes the filament.

4. The semiconductor device according to claim 2 , wherein

the first electrode is formed on the semiconductor substrate, and

the variable resistance layer has a resistance value that continuously or stepwisely increases from the second electrode toward the first electrode.

5. A semiconductor device comprising:

variable resistance elements on a semiconductor substrate, wherein

each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable,

the variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable, and

the resistance value of the variable resistance layer varies due to increase or decrease of oxygen defects with displacement or transfer of oxygen caused by application of an electric pulse.

6. The semiconductor device according to claim 5 , wherein

the variable resistance layer includes a transitional metal oxide.

7. The semiconductor device according to claim 5 , wherein

the filament has a current path through which a current flows via oxygen defects, and has redundant current paths that connect an end portion of the filament on a side of the first electrode to an end portion of the filament on a side of the second electrode.

8. The semiconductor device according to claim 7 , wherein

an oxygen defect density of the filament does not depend on the neural network weight.

9. A semiconductor device comprising:

variable resistance elements on a semiconductor substrate, wherein

each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable,

the variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable, and

the resistance value of the variable resistance layer varies due to metal ion transfer caused by application of an electric pulse.

10. A semiconductor device comprising:

variable resistance elements on a semiconductor substrate, wherein

each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable,

the variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable, and

each of resistance values of at least half the variable resistance elements is larger than a median value of analog values assumable by resistance values of the variable resistance elements.

11. A semiconductor device comprising:

variable resistance elements on a semiconductor substrate, wherein

each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable,

the variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable, and

each of resistance values of at least half the variable resistance elements is smaller than a median value of analog values assumable by resistance values of the variable resistance elements.

12. A semiconductor device driving method of driving a semiconductor device that includes variable resistance elements on a semiconductor substrate, wherein

each of the variable resistance elements includes a first electrode, a second electrode, and a variable resistance layer that is sandwiched between the first electrode and the second electrode and that stores a resistance value that is continuously variable, and

the variable resistance layer includes a filament whose shape differs according to a neural network weight, and stores, as an analog value, the resistance value that is variable, and

the semiconductor device driving method comprises:

determining a stress of an electric pulse associated with the neural network weight; and

in a forming process of forming a filament in the variable resistance layer, forming the filament whose shape differs according to the neural network weight by applying, as the stress, the electric pulse determined.

13. The semiconductor device driving method according to claim 12 , wherein

at least one of a voltage value, a current value, or an application time of the electric pulse differs according to the neural network weight.

14. The semiconductor device driving method according to claim 12 , wherein

the stress of the electric pulse includes a first electric pulse that is common to the variable resistance elements and a second electric pulse that is not common to the variable resistance elements and is associated with the neural network weight, and

in the forming process, the first electric pulse is applied to the variable resistance layer and the second electric pulse is further applied to the variable resistance layer.

15. The semiconductor device driving method according to claim 14 , wherein

at least one of a voltage value, a current value, or an application time of the second electric pulse differs according to the neural network weight.

Assignments (2)
CHANGE OF NAME Recorded May 9, 2022
From: PANASONIC CORPORATION
To: PANASONIC HOLDINGS CORPORATION
Reel/Frame 059909/0607 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Mar 23, 2022
From: MIKAWA, TAKUMI; KATAYAMA, KOJI; YASUHARA, RYUTARO
To: PANASONIC CORPORATION
Reel/Frame 059347/0159 →