Block Copolymer Lithography Mask for Resistive Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for manufacturing physical implementations of Reservoir Computing using recurrent neural networks are limited by the restricted choice of materials for resistive memory, particularly the Ag/Ag2S pair, which restricts flexibility and compatibility for nanowires and memory cells.
Innovation Solution
A method involving the use of block copolymers to create a substrate with random spatial distribution, allowing for the formation of resistive memory cells with random connections, latency, and weights, while providing freedom in material choice for the memory layer, enabling the physical implementation of recurrent neural networks trained by Reservoir Computing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If copper microspheres are used to form a seed layer for silver nanowires, then random spatial distribution is achieved, but the choice of materials is restricted to compatible pairs like Ag/Ag2S
Solution Approach 1:
The patent introduces an organic intermediary layer (block copolymer) between the substrate and the metal nanowires. This organic layer serves as a universal interface that enables the growth of various metal nanowires (Cu, Ag, Au, etc.) without requiring specific material compatibility pairs, thus resolving the material selection constraint while maintaining ease of manufacture
Solution Approach 2:
The patent replaces the inorganic copper microsphere seed layer with an organic block copolymer structure. This substitution allows for greater versatility in material choice as the organic polymer can serve as a universal template for different metal nanowires, eliminating the need for specific inorganic material compatibility pairs
2Adaptability or versatility
If block copolymers are used to create random spatial distribution, then freedom in material choice is achieved, but additional processing steps are required
Solution Approach 1:
The block copolymer performs multiple functions automatically: it self-assembles into the required random spatial pattern, serves as a lithography mask for structuring, and acts as a seed layer for nanowire growth. This self-service capability reduces the need for additional processing steps despite the added material flexibility
Solution Approach 2:
The patent merges multiple functions into the block copolymer layer: it combines the lithography mask function with the seed layer function. The same organic structure that provides random spatial distribution also serves as the template for nanowire growth, thereby reducing overall process complexity despite enabling material flexibility
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 introduces necessary randomness and flexibility in the formation of neural network connections and weights, enhancing the quality and versatility of the physical implementation of recurrent neural networks, allowing for improved performance in tasks like classification, object tracking, and recognition.
Implementation Method 1
block copolymers have the property of self-assembling into dense networks of nanometric objects with the ability to form a lithography mask
Implementation Method 2
etching the free areas of the dielectric layer so as to structure the dielectric layer
Data Source
Figure 1a~1d
Figure 1e~2
AI summary
This process comprises the following successive steps: a) providing a substrate having a dielectric layer; b) forming a first layer of block copolymers on a portion of the dielectric layer such that the dielectric layer has free areas with a random spatial distribution; c) etching the free areas so as to structure the dielectric layer; d) removing the first layer of block copolymers; d') forming a first electrode (20') on the structured dielectric layer (200); e) forming a memory layer (4), of the resistive memory type, on the first electrode; f) forming a second electrode on the memory layer; g) forming a second layer of block copolymer on a portion of the second electrode such that the second electrode has free areas with a random spatial distribution; h) etching the free areas so as to structure the second electrode (210); i) removing the second layer of block copolymers.