Vertical Memory Defect Detection Using Multi-Wavelength Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor wafer inspection systems are inadequate for detecting defects in vertical memory structures like 3D or VNAND memory, as they struggle with depth penetration and distinguishing surface defects from those within the stack, leading to increased manufacturing costs and reduced memory density.
Innovation Solution
A method using incident light of different wavelength ranges (red-visible, ultraviolet, deep ultraviolet, blue-visible, infrared, and near-infrared) to detect defects on the surface and throughout the depth of vertical memory stacks, with specific implementations involving poly silicon and oxide layers, tungsten and oxide layers, and nitride and oxide layers, utilizing polarization and optical elements like dichroic beam splitters to optimize defect detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional inspection systems are used, then surface defects can be detected, but defects within the depth of vertical memory stacks cannot be detected
Solution Approach 1:
The inspection system segments the detection process by using multiple wavelength ranges (first wavelength range for surface defects, second wavelength range for deep defects) to separately detect defects at different depths within the vertical memory stack, allowing the system to overcome the limitation of single-wavelength inspection
Solution Approach 2:
The system changes the optical parameter (wavelength) of incident light to achieve different penetration depths. By using a first wavelength range that does not penetrate deeply and a second wavelength range that penetrates deeply into the stack, the system can detect defects at various depths within the vertical memory structure
2Reliability
If single wavelength inspection is used, then inspection process is simple, but inability to distinguish surface defects from buried defects reduces manufacturing yield
Solution Approach 1:
The system segments defect detection by spatial location (surface vs. buried) using multiple wavelength ranges. Defects detected only with the first wavelength range are identified as surface defects, while defects detected with both wavelength ranges are identified as buried defects within the stack, enabling reliable distinction and improving manufacturing yield
Solution Approach 2:
The inspection system performs multiple functions using a unified multi-wavelength approach: it detects surface defects, detects buried defects, and distinguishes between the two types, all within a single inspection process that uses both first and second wavelength ranges
3Quantity of substance
If vertical memory density is increased, then memory capacity improves, but defect detection becomes more difficult
Solution Approach 1:
As vertical memory density increases with more layers (16-24 bits currently, extending to 48-64 bits), the system adjusts the wavelength parameter to penetrate deeper into the stack. The second wavelength range is specifically selected to penetrate through increased numbers of layers, maintaining defect detection capability despite higher density
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
Enables effective detection of defects at various depths within vertical memory stacks, improving manufacturing yields by distinguishing surface and buried defects, and allowing for earlier detection before forming word-line trenches, thus reducing costs and enhancing memory density.
Implementation Method 1
incident light at the second wavelength range penetrates and, in response, output light is scattered or reflected back through the poly silicon and oxide layers to be detected by a detector of the inspection tool
Implementation Method 2
output light is scattered or reflected back through the poly silicon and oxide layers
Implementation Method 3
the second wavelength range is selected to result in a maximum intensity of light to reach a plurality of depths within the trench
Implementation Method 4
the method includes vertically or horizontally polarizing the incident light that has the second wavelength range
Data Source
AI summary
Disclosed are methods and apparatus for inspecting a vertical memory stack. On an inspection tool, incident light having a first wavelength range is used to detect defects on a surface of the vertical memory stack. On the inspection tool, incident light having a second wavelength range is used to detect defects on both the surface and throughout a depth of the vertical memory stack. The defects detected using the first and second wavelength range are compared to detect defects only throughout the depth of the vertical memory stack, excluding defects on the surface, as well as to detect defects only on the surface.