Linear Phase Change Memory via RC Configuring Current
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Phase change memory (PCM) devices exhibit non-linear resistance/conductance changes with respect to the number of voltage pulses applied, leading to inaccurate data storage and retrieval in neuromorphic computing systems due to non-uniform performance and asymmetries in crossbar arrays.
Innovation Solution
Incorporating a resistance-capacitance (RC) circuit that produces a configuring current to modify the incremental changes in PCM property levels, ensuring linear changes with respect to the pulse number of identical voltage pulses, achieved by tuning the resistance and capacitive components of the RC circuit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If voltage pulses are applied to change PCM states, then resistance/conductance changes occur, but the changes are non-linear and non-uniform
Solution Approach 1:
An RC circuit is introduced as an intermediary component between the voltage pulse source and the PCM. The RC circuit generates a configuring current that mediates the interaction between the applied voltage and the PCM, transforming the non-linear resistance changes into linear changes with respect to the pulse number. This intermediary mechanism resolves the contradiction by adding controlled complexity to achieve uniformity in the PCM property changes.
Solution Approach 2:
The invention changes the parameter of current configuration by introducing an RC circuit that produces a configuring current. This configuring current modifies the way voltage pulses interact with the PCM, transforming the non-linear resistance/conductance changes into linear changes. By changing the current parameter through the RC circuit, the system achieves uniform property level changes across different pulse numbers.
2Measurement precision
If multiple voltage pulses are applied to SET/RESET PCM states, then property levels change incrementally, but the incremental changes are not uniform
Solution Approach 1:
The RC circuit serves as a mediator that processes the voltage pulses and generates a configuring current. This configuring current ensures that each voltage pulse produces a uniform incremental change in PCM property levels, regardless of the pulse number. The intermediary RC circuit adds the necessary control complexity to achieve precise, uniform measurements while maintaining ease of operation through a standardized pulse application method.
Solution Approach 2:
The RC circuit inherently provides feedback control by generating a configuring current that depends on the previous state of the PCM. This feedback mechanism ensures that each voltage pulse produces a consistent incremental change in property levels, making the system both precise and easy to operate through predictable, uniform changes.
3Stability of the object's composition
If identical voltage pulses are applied to PCM, then the same pulse characteristics are used, but the resistance changes are non-linear
Solution Approach 1:
The RC circuit is introduced as an intermediary that transforms the effect of identical voltage pulses on the PCM. While the voltage pulses remain identical and consistent, the RC circuit generates a configuring current that ensures linear, uniform resistance changes. This intermediary resolves the contradiction by decoupling the consistency of input pulses from the linearity of output changes.
Solution Approach 2:
The invention changes the current parameter through the RC circuit to transform the effect of identical voltage pulses. By introducing the RC circuit, the system maintains consistent voltage pulse characteristics while achieving linear resistance changes through the configuring current generated by the RC circuit's resistance and capacitance components.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach ensures accurate and linear changes in PCM property levels, improving the training accuracy and performance of neuromorphic computing systems by maintaining uniformity in data storage and retrieval.
Implementation Method 1
a resistance-capacitance (RC) circuit. The RC circuit produces a configuring current that flows through the PCM in response to one or more of the voltage pulses
Implementation Method 2
The configuring current has a configuring current characteristic. The configuring current modifies one or more of the incremental changes to one or more of the property levels
Implementation Method 3
a dielectric, which is normally insulating, can be made (SET) to conduct (in a low resistance state (LRS)) through one or more filaments or conduction paths formed by application of a sufficiently high voltage
Implementation Method 4
RESETTING a state of the PCM, e.g., with a series of voltage pulses, breaks one or more filaments for the respective resistance state and causes the total PCM resistance to increase incrementally
Data Source
AI summary
A phase change (PCM) memory device that includes a PCM and a resistance-capacitance (RC) circuit. The PCM has one or more PCM properties, each PCM property has a plurality of PCM property states. As the PCM property states of a given property are Set or Reset, the PCM property states each produce an incremental change to a property level of the respective/associated PCM property, e.g., PCM conductance. The incremental changes to property level of the PCM memory device are in response to application of one or more of a pulse number of voltage pulses. The RC circuit produces a configuring current that flows through the PCM in response to one or more of the voltage pulses. The configuring current modifies one or more of the incremental changes to one or more of the property levels so that the property level changes lineally with respect to the pulse number. The PCM memory device has use in a synapse connector, e.g., in a memory array. The memory array can be used to store and/or read memory values associated with one or more of the property levels. The memory values can be used as weighting values in a neuromorphic computing application/system, like a neural network.