A memory device includes a column of at least three memory cells and a source line coupled to the source terminal of each memory cell. A source line driver is coupled to the source line, a voltage terminal, and a program voltage source and is switchable between a program operation, an erase operation, and a read operation.

Numerous examples are disclosed for an output block coupled to a non-volatile memory array in a neural network and associated methods. In one example, a circuit for converting a current in a neural network into an output voltage comprises a non-volatile memory cell comprises a word line terminal, a bit line terminal, and a source line terminal, wherein the bit line terminal receives the current; and a switch for selectively coupling the word line terminal to the bit line terminal; wherein when the switch is closed, the current flows into the non-volatile memory cell and the output voltage is provided on the bit line terminal.

Numerous embodiments for performing tuning of a page or a word of non-volatile memory cells in an analog neural memory are disclosed. High voltage circuits used to generate high voltages applied to terminals of the non-volatile memory cells during the precision tuning process are also disclosed. Programming sequences for the application of the voltages to the terminals to minimize the occurrence of disturbances during tuning are also disclosed.

Numerous examples for performing tuning of a page or a word of non-volatile memory cells in an analog neural memory are disclosed. In one example, an analog neural memory system comprises an array of non-volatile memory cells arranged into rows and columns, each non-volatile memory cell comprising a word line terminal, a bit line terminal, and an erase gate terminal; a plurality of word lines, each word line coupled to word line terminals of a row of non-volatile memory cells; a plurality of bit lines, each bit line coupled to bit line terminals of a column of non-volatile memory cells; and a plurality of erase gate enable transistors, each erase gate enable transistor coupled to erase gate terminals of a word of non-volatile memory cells.

A semiconductor device includes a stacked structure, channel layers passing through the stacked structure, a well plate located under the stacked structure, a source layer located between the stacked structure and the well plate, a connection structure coupling the channel layers to each other and including a first contact contacting the source layer and a second contact contacting the well plate, and an isolation pattern insulating the source layer and the well plate from each other.

Numerous embodiments of power management techniques are disclosed for various operations involving one or more vector-by-matrix multiplication (VMM) arrays within an artificial neural network.

A memory device, and method of making the same, that includes a substrate of semiconductor material of a first conductivity type, first and second regions spaced apart in the substrate and having a second conductivity type different than the first conductivity type, with a first channel region in the substrate extending between the first and second regions, a first floating gate disposed over and insulated from a first portion of the first channel region adjacent to the second region, a first coupling gate disposed over and insulated from the first floating gate, a first word line gate disposed over and insulated from a second portion of the first channel region adjacent the first region, and a first erase gate disposed over and insulated from the first word line gate.

An artificial neural network device that utilizes one or more non-volatile memory arrays as the synapses. The synapses are configured to receive inputs and to generate therefrom outputs. Neurons are configured to receive the outputs. The synapses include 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 and a non-floating gate disposed over and insulated from a second portion of the channel 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 multiply the inputs by the stored weight values to generate the outputs.

An erasable programmable non-volatile memory includes a first select transistor, a first floating gate transistor, a second select transistor and a second floating gate transistor. A select gate and a first source/drain terminal of the first select transistor receive a first select gate voltage and a first source line voltage, respectively. A first source/drain terminal and a second source/drain terminal of the first floating gate transistor are connected with a second source/drain terminal of the first select transistor and a first bit line voltage, respectively. A select gate and a first source/drain terminal of the second select transistor receive a second select gate voltage and a second source line voltage, respectively. A first source/drain terminal and a second source/drain terminal of the second floating gate transistor are connected with the second source/drain terminal of the second select transistor and a second bit line voltage, respectively.

A method provided herein is adapted to a sense amplifier having a first cross-coupled latch and a second cross-coupled latch, each of which includes a first pair of transistors and a pair of coupling capacitors coupled to respective gate terminals of the first pair of transistors. The method includes, during a first phase, charging the pair of coupling capacitors of a first pair of transistors at a first cross-coupled latch to achieve zeroing and providing a first set of input voltages to a second cross-coupled latch, and, during a second phase following the first phase, discharging the pair of coupling capacitors to cancel a mismatch between the first pair of transistors and comparing the first set of input voltages provided to the second cross-coupled latch to generate a first set of output voltages.