PATENTSCOPE will be unavailable a few hours for maintenance reason on Tuesday 19.11.2019 at 4:00 PM CET
Search International and National Patent Collections
Some content of this application is unavailable at the moment.
If this situation persists, please contact us atFeedback&Contact
1. (WO2017138951) MEMORY DEVICES WITH VOLATILE AND NON-VOLATILE BEHAVIOR
Note: Text based on automatic Optical Character Recognition processes. Please use the PDF version for legal matters

WHAT IS CLAIMED IS:

1 . A memory device for neuromorphic computing, comprising:

first electrode;

an active oxide layer disposed on the first electrode to form and dissipate a conductive bridge in response to an input voltage based on interaction with the first electrode to provide ion mobility within the active oxide layer; and

a second electrode disposed on the active oxide layer;

wherein the first electrode and the active oxide layer enable the conductive bridge to dissipate spontaneously within a relaxation time to enable the memory device to self-refresh according to volatile behavior in response to the input voltage being below a threshold corresponding to disregarding sneak current and noise of a crossbar array in which the memory device is to operate; and

wherein the first electrode and the active oxide layer enable the conductive bridge to persist beyond the relaxation time to enable the memory device to retain programming for neuromorphic computing training according to non-volatile behavior of the memory device in response to the input voltage not being below the threshold.

2. The memory device of claim 1 , wherein active oxide layer comprises SiOx,

3. The memory device of claim 1 , wherein the active oxide layer includes a first layer having a first doping profile, and a second layer having a second doping profile different than the first doping profile.

4. The memory device of claim 1 , further comprising a passivation layer disposed on portions of the active oxide layer not in contact with the first electrode and the second electrode, to protect the active oxide layer.

5. The memory device of claim 1 , wherein operation based on the conductive bridge includes ionic migration of ions from the first electrode, and electron migration of electrons from the input voltage.

6. The memory device of claim 1 , wherein the conductive bridge is to dissolve in response to removal of the input voltage that is below the threshold, based on the second electrode serving as an ionic sink to absorb ions from the first electrode.

7. The memory device of claim 1 , wherein the relaxation time is substantially on the order of a microsecond.

8. The memory device of claim 1 , wherein the first electrode includes copper metal, and is to form and dissipate copper filaments in the active oxide layer.

9. The memory device of claim 1 , wherein the memory device is to retain programming in a first state according to non-volatile behavior of the memory device based on a first input voltage not lower than the threshold, and retain programming in a second state according to non-volatile behavior of the memory device based on accumulating a second input voltage, not lower than the threshold, applied to enhance the first state.

10. The memory device of claim 1 , wherein at input voltages not below the threshold, the memory device is to provide an analog mode of operation corresponding to output of the memory device being proportional to input according to an accumulated effect, and wherein at input voltages below the threshold, the memory device is to not provide the analog mode of operation, to disregard sneak current and noise and not exhibit the accumulated effect due to self-refreshing according to the relaxation time.

1 1 . A system for neuromorphic computing comprising:

a crossbar array;

a plurality of memory devices disposed in the crossbar array for programming according to training wherein a memory device targeted by an input voltage is to behave as non-volatile, and memory devices that are untargeted are to behave as volatile to self-refresh to disregard sneak current and noise;

wherein a given memory device is to self-refresh according to volatile behavior in response to the input voltage being below a threshold, corresponding to sneak current and noise of a crossbar array in which the memory device is to operate; and

wherein the given memory device is to retain programming for neuromorphic computing training according to non-volatile behavior in response to the input voltage not being below the threshold.

12. The system of claim 10, wherein the crossbar array is to employ a half-select scheme using full value voltage on a selected cell to be programmed, and using half value voltage on non-selected ceils, and wherein the threshold is set between the full value voltage and the half value voltage.

13. The system of claim 10, wherein the plurality of memory devices are programmed using the neuromorphic computing training and self-refreshing to disregard effects of sneak current and noise, without a need to apply a compensation afterward to undo effects of sneak current and noise.

14. A method for neuromorphic computing, comprising:

programming a memory device, of a crossbar array, that is to retain the programming, for neuromorphic computing training, beyond a relaxation time according to non-volatile behavior of the memory device in response to an input voltage not being below a threshold corresponding to sneak current and noise of the crossbar array; and

disregarding, by the memory device, sneak current and noise of the crossbar array based on the memory device self-refreshing within the relaxation time according to volatile behavior in response to the input voltage being below the threshold when the memory device is not being programmed.

15. The method of claim 14, further comprising forming and dissipating a conductive bridge inside an oxide layer of the memory device based on interaction between the oxide layer and a first electrode of the memory device in response to an input voltage.