Memory Array Layout With Mixed Bit Storage for Higher Density

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing memory systems face challenges in achieving better memory performance, capacity, and power efficiency due to the slowing pace of Moore's Law, leading to reduced spatial efficiency and density.

Innovation Solution

The proposed memory structure incorporates two types of structures for storing bits: single-memory-element structures for most significant bits (MSBs) and multiple-memory-element structures for least significant bits (LSBs), optimizing memory-element-to-transistor ratio to enhance spatial efficiency and density.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If traditional memory structures are used, then memory capacity can be maintained, but spatial efficiency and density decrease due to the slowing pace of Moore's Law

Engineering Contradiction:
Improvememory capacityVSAvoidspatial efficiency
Core Design Contradiction:
Quantity of substanceVSArea of stationary object

Solution Approach 1:

The patent segments the memory array into distinct regions: first memory structures with one memory element per transistor, and second memory structures with multiple memory elements per transistor. This segmentation allows different spatial efficiencies in different regions, optimizing overall density while maintaining adequate capacity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from a uniform two-dimensional memory layout to a heterogeneous structure that effectively utilizes vertical dimension by stacking multiple memory elements in the second memory structures, thereby increasing spatial efficiency and density without proportionally increasing footprint area.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Area of stationary object

If memory-element-to-transistor ratio is increased to enhance density, then spatial efficiency improves, but manufacturing complexity increases

Engineering Contradiction:
Improvespatial efficiencyVSAvoidmanufacturing complexity
Core Design Contradiction:
Area of stationary objectVSDevice complexity

Solution Approach 1:

The patent divides the memory array into first and second memory structures with different memory-element-to-transistor ratios. The first memory structures use a 1:1 ratio for simplicity, while the second memory structures use higher ratios for density. This segmentation manages manufacturing complexity by isolating the more complex structures to specific regions.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different regions of the memory array are assigned different memory-element-to-transistor ratios based on local requirements. The second memory structures with higher ratios are placed in regions where density is prioritized, while the first memory structures with 1:1 ratio are used where manufacturing simplicity is more important, achieving local optimization.

Inventive Principle:
Principle #3Local quality

Data Source

PatentUS20250191628A1Memory array structure
Publication Date: 2025.06.12 TAIWAN SEMICONDUCTOR MANUFACTURING CO LTD
  • US20250191628A1 patent drawing
  • US20250191628A1 patent drawing
  • US20250191628A1 patent drawing

AI summary

In some aspects of the present disclosure, a memory array structure is disclosed. In some embodiments, the memory array structure includes a word array. In some embodiments, the word array stores an N-bit word. In some embodiments, the word array includes a plurality of first memory structures and a plurality of second memory structures. In some embodiments, each first memory structure includes a first transistor and a first memory element. In some embodiments, each second memory structure includes a second transistor and a plurality of second memory elements, each second memory element includes a first end and a second end, the first end of each second memory element is coupled to a corresponding bit line, and the second end of each second memory element is coupled to a first end of the second transistor.