3D Logic and Memory Stacking for Density Beyond 2D Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As semiconductor devices approach single-digit nanometer fabrication nodes, 2D transistor density scaling is limited by manufacturing variability and electrostatic device limitations, prompting a need for three-dimensional (3D) integration to increase transistor density in volume rather than area.
Innovation Solution
The technique involves stacking and bonding individual substrates to form higher density circuits, enabling 2 to 4 times increase in memory density for 3D NAND and SRAM, and allowing multiple cores on multiple 3D planes for high-speed computing by optimizing processing parameters and bonding methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If 2D transistor scaling is continued, then transistor density per unit area increases, but manufacturing variability and electrostatic device limitations prevent further scaling at single-digit nanometer nodes
Solution Approach 1:
The patent transitions from two-dimensional planar transistor arrangements to three-dimensional vertically stacked transistor configurations. By stacking multiple transistor layers vertically on the same footprint area, the design achieves higher transistor density without further reducing the contacted gate pitch, thereby overcoming the manufacturing precision limits of 2D scaling while maintaining compatibility with existing fabrication processes.
2Quantity of substance
If wire pitch scaling is reduced to increase transistor density, then more transistors can be wired into circuits, but resistance, capacitance, and reliability concerns limit further wire pitch scaling
Solution Approach 1:
The patent implements three-dimensional interconnect structures that route signals vertically through stacked layers rather than only laterally in planar layers. This vertical stacking of interconnect layers reduces the lateral wire pitch requirements and shortens signal path lengths, thereby decreasing resistance and capacitance while enabling higher wiring density for connecting increased transistor counts.
Solution Approach 2:
The patent employs nested interconnect structures where multiple conductive layers are stacked vertically with each layer nested within the vertical space of adjacent layers. This nested arrangement allows dense interconnection of vertically stacked transistors without increasing the lateral footprint, reducing wire length and improving electrical performance while maintaining high wiring density.
3Quantity of substance
If 3D vertical stacking of transistors is implemented, then transistor density in volume increases, but fabrication process complexity increases
Solution Approach 1:
The patent divides the fabrication process into separate sequential stages: first forming a initial substrate with a set of transistors, then bonding a second substrate with additional transistors stacked vertically. This segmentation of the manufacturing process into discrete bondable modules allows each substrate to be fabricated using standard processes, with the 3D integration achieved through wafer bonding rather than attempting to form all layers in a single complex process.
Solution Approach 2:
The patent performs preliminary fabrication of complete transistor stacks on separate substrates before final assembly. Each substrate is fully processed with transistors, interconnects, and passivation layers completed in advance, then bonded together to form the 3D structure. This preliminary action on individual substrates simplifies the overall fabrication by avoiding the need to perform complex 3D structuring operations during the main fabrication sequence.
Data Source
AI summary
Techniques herein include methods of forming higher density circuits by combining multiple substrates via stacking and bonding of individual substrates. High voltage and low voltage devices along with 3D NAND devises are fabricated on a first wafer, and high voltage and low voltage devices and/or memory are then fabricated on a second wafer and/or third wafer.


