# US 2021/0342682 A1 · Decoders for analog neural memory in deep learning artificial neural network

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0342682 A1
- Document ID: US20210342682A1
- Status: Published application
- Kind: A1
- Application number: 17/367,633
- Filed: 2021-07-06
- Published: 2021-11-04
- Earliest claimed: 2018-03-14
- Assignees: Silicon Storage Technology, Inc.
- Applicants: Silicon Storage Technology, Inc.
- Inventors: Hieu Van Tran; STANLEY HONG; ANH LY; THUAN VU; HIEN PHAM; KHA NGUYEN; HAN TRAN
- Web page: https://patentgrep.com/patent/US20210342682A1

## CPC

- G06N 3/065: Analogue means
- G06F 17/16: Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)}
- G06N 3/04: Architecture, e.g. interconnection topology
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G11C 11/54: using elements simulating biological cells, e.g. neuron
- G11C 16/0408: {comprising cells containing floating gate transistors (G11C16/0483, G11C16/0491 take precedence)}
- G11C 16/0425: {comprising cells containing a merged floating gate and select transistor}
- G11C 16/08: Address circuits; Decoders; Word-line control circuits
- G11C 16/24: Bit-line control circuits
- G11C 2216/04: Nonvolatile memory cell provided with a separate control gate for erasing the cells, i.e. erase gate, independent of the normal read control gate
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G11C: STATIC STORES (semiconductor memory devices H10B)

## Abstract

Numerous embodiments of decoders for use with a vector-by-matrix multiplication (VMM) array in an artificial neural network are disclosed. The decoders include bit line decoders, word line decoders, control gate decoders, source line decoders, and erase gate decoders. In certain embodiments, a high voltage version and a low voltage version of a decoder is used.

## Claims

### Claim 1

A bit line decoder circuit coupled to a vector-by-matrix multiplication array, the vector-by matrix multiplication array comprising an array of non-volatile memory cells organized into rows and columns, wherein each column is connected to a bit line, the bit line decoder circuit comprising:
a first circuit for enabling individual bit lines during a program-and-verify operation; and
a second circuit for enabling a plurality of bit lines during a read operation.

### Claim 2 (depends on 1)

The bit line decoder circuit of claim 1, wherein the second circuit comprises a select transistor and an activation function circuit coupled to each bit line.

### Claim 3 (depends on 2)

The bit line decoder circuit of claim 2, wherein the gate of each select transistor is coupled to the same control line.

### Claim 4 (depends on 1)

The bit line decoder circuit of claim 1, wherein a negative bias is applied to the word lines of each unselected memory cell during a program-and-verify operation or a read operation.

### Claim 5 (depends on 1)

The bit line decoder circuit of claim 1, wherein each of the non-volatile memory cells is a split-gate flash memory cell.

### Claim 6 (depends on 1)

The bit line decoder circuit of claim 1, wherein each of the non-volatile memory cells is a stacked-gate flash memory cell.

### Claim 7 (depends on 1)

The bit line decoder circuit of claim 1, wherein each of the non-volatile memory cells are configured to operate in a sub-threshold region.

### Claim 8 (depends on 1)

The bit line decoder circuit of claim 1, wherein each of the non-volatile memory cells are configured to operate in a linear region.

### Claim 9 (depends on 1)

The bit line decoder circuit of claim 1, wherein a negative bias is applied to the gates of each unselected bit line decoder during a program-and-verify operation or a read operation.

### Claim 10

A bit line decoder circuit coupled to a vector-by-matrix multiplication array, the vector-by matrix multiplication array comprising an array of non-volatile memory cells organized into rows and columns, wherein each column is connected to a bit line, the bit line decoder circuit comprising:
a multiplexed circuit, wherein in a first mode, the multiplexed circuit enables individual bit lines during a program-and-verify operation and in a second mode, the multiplexed circuit enabled all bit lines during a read operation.

### Claim 11 (depends on 10)

The bit line decoder circuit of claim 10, wherein the multiplexed circuit comprises a select transistor and an activation function circuit coupled to each bit line.

### Claim 12 (depends on 10)

The bit line decoder circuit of claim 10, wherein a negative bias is applied to the word lines of each unselected memory cell during a program-and-verify operation or a read operation.

### Claim 13 (depends on 10)

The bit line decoder circuit of claim 10, wherein each of the non-volatile memory cells is a split-gate flash memory cell.

### Claim 14 (depends on 10)

The bit line decoder circuit of claim 10, wherein each of the non-volatile memory cells is a stacked-gate flash memory cell.

### Claim 15 (depends on 10)

The bit line decoder circuit of claim 10, wherein each of the non-volatile memory cells are configured to operate in a sub-threshold region.

### Claim 16 (depends on 10)

The bit line decoder circuit of claim 10, wherein each of the non-volatile memory cells are configured to operate in a linear region.

### Claim 17 (depends on 10)

The bit line decoder circuit of claim 10, wherein a negative bias is applied to the gates of each unselected bit line decoder during a program-and-verify operation or a read operation.

### Claim 18

A word line driver coupled to a vector-by-matrix multiplication array, the vector-by matrix multiplication array comprising an array of non-volatile memory cells organized into rows and columns, wherein each row is coupled to a word line and each word line is coupled to the word line driver, the word line driver comprising:
a plurality of select transistors, each of the plurality of select transistors comprising a first terminal, a second terminal, and a gate, wherein the gate of each of the plurality of select transistors is coupled to a common control line and the first terminal of each of the plurality of select transistors is coupled to a different word line, and wherein the second terminal of each of the plurality of select transistors is coupled to one or more bias transistors;
wherein the bias transistors coupled to each of the plurality of select transistors are capable of providing a bias voltage to a single select transistor or to all of the select transistors.

### Claim 19 (depends on 18)

The word line driver of claim 18, where at least one bias transistor coupled to each of the plurality of select transistors is coupled to a common control line.

### Claim 20 (depends on 18)

The word line driver of claim 18, where each bias transistor coupled to each of the plurality of select transistors is coupled to a different control line.

### Claim 21 (depends on 18)

The word line driver of claim 18, wherein each of the bias transistors is coupled to a circuit for decoding a word line address

### Claim 22 (depends on 18)

The word line driver of claim 18, wherein the bias transistors are coupled to a shift register.

### Claim 23 (depends on 18)

The word line driver of claim 18, wherein each select transistor is coupled to a capacitor.

### Claim 24 (depends on 18)

The word line driver of claim 18, wherein each bias transistor is coupled to a comparator.

### Claim 25

A neural network device, comprising:
a first plurality of synapses configured to receive a first plurality of inputs and to generate therefrom a first plurality of outputs, wherein the first plurality of synapses comprises:
a plurality of memory cells, wherein each of the memory cells includes spaced apart source and drain regions formed in a semiconductor substrate with a channel region extending there between, a floating gate disposed over and insulated from a first portion of the channel region, a first gate disposed over and insulated from a second portion of the channel region, a second gate disposed over and insulated from the floating gate, and a third gate disposed over and insulated from the source region;
each of the plurality of memory cells is configured to store a weight value corresponding to a number of electrons on the floating gate;
the plurality of memory cells are configured to generate the first plurality of outputs based upon the first plurality of inputs and the stored weight values;
wherein the memory cells of the first plurality of synapses are arranged in rows and columns, and wherein the first plurality of synapses comprises:
a plurality of first lines each electrically connecting together the first gates in one of the rows of the memory cells;
a plurality of second lines each electrically connecting together the second gates in one of the columns of the memory cells;
a plurality of third lines each electrically connecting together the third gates in one of the rows of the memory cells;
a plurality of fourth lines each electrically connecting together the source regions in one of the rows of the memory cells;
a plurality of fifth lines each electrically connecting together the drain regions in one of the columns of the memory cells;
wherein the first plurality of synapses is configured to receive the first plurality of inputs as electrical voltages on the plurality of second lines or on the plurality of fifth lines, and to provide the first plurality of outputs as electrical currents on the plurality of fourth lines.

### Claim 26 (depends on 25)

The neural network device of claim 25, wherein the first plurality of synapses is configured to receive the first plurality of inputs as electrical voltages on the plurality of second lines.

### Claim 27 (depends on 25)

The neural network device of claim 25, wherein the first plurality of synapses is configured to receive the first plurality of inputs as electrical voltages on the plurality of fifth lines.

### Claim 28 (depends on 25)

The neural network device of claim 25, further comprising:
a first plurality of neurons configured to receive the first plurality of outputs.

### Claim 29 (depends on 28)

The neural network device of claim 28, further comprising:
a second plurality of synapses configured to receive a second plurality of inputs from the first plurality of neurons and to generate therefrom a second plurality of outputs, wherein the second plurality of synapses comprises:
a plurality of second memory cells, wherein each of the second memory cells includes spaced apart second source and second drain regions formed in the semiconductor substrate with a second channel region extending there between, a second floating gate disposed over and insulated from a first portion of the second channel region, a fourth gate disposed over and insulated from a second portion of the second channel region, a fifth gate disposed over and insulated from the second floating gate, and a sixth gate disposed over and insulated from the second source region;
each of the plurality of second memory cells is configured to store a second weight value corresponding to a number of electrons on the second floating gate;
the plurality of second memory cells are configured generate the second plurality of outputs based upon the second plurality of inputs and the stored second weight values;
wherein the second memory cells of the second plurality of synapses are arranged in rows and columns, and wherein the second plurality of synapses comprises:
a plurality of sixth lines each electrically connecting together the fourth gates in one of the rows of the second memory cells;
a plurality of seventh lines each electrically connecting together the fifth gates in one of the columns of the second memory cells;
a plurality of eighth lines each electrically connecting together the sixth gates in one of the rows of the second memory cells;
a plurality of ninth lines each electrically connecting together the second source regions in one of the rows of the second memory cells;
a plurality of tenth lines each electrically connecting together the second drain regions in one of the columns of the second memory cells;
wherein the second plurality of synapses is configured to receive the second plurality of inputs as electrical voltages on the plurality of seventh lines or on the plurality of tenth lines, and to provide the second plurality of outputs as electrical currents on the plurality of ninth lines.

### Claim 30 (depends on 29)

The neural network device of claim 29, wherein the second plurality of synapses is configured to receive the second plurality of inputs as electrical voltages on the plurality of seventh lines.

### Claim 31 (depends on 29)

The neural network device of claim 29, wherein the second plurality of synapses is configured to receive the second plurality of inputs as electrical voltages on the plurality of tenth lines.

### Claim 32 (depends on 29)

The neural network device of claim 29, further comprising:
a second plurality of neurons configured to receive the second plurality of outputs.
