CAD Data Alignment to Optical Images via Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The semiconductor industry faces challenges in accurately aligning computer-aided design (CAD) data with optical images of integrated circuits due to limitations in image resolution and noise levels, particularly as device features shrink below 20 nm, making it difficult to observe and test electrical performance and propagate signals within ICs.
Innovation Solution
The solution involves image reconstruction techniques to enhance image resolution and signal-to-noise ratios, combined with processing CAD data to match expected optical images, enabling precise alignment between CAD data and optical images through methods like sparse imaging reconstruction and Boolean operations on CAD layers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If backside optical imaging is used to test completed ICs, then optical images can be obtained through the silicon substrate, but image resolution becomes poorer than device geometries making accurate alignment difficult
Solution Approach 1:
The patent applies preliminary action by performing image reconstruction on optical images before aligning them with CAD data. The system reconstructs images from multiple focal planes and performs deconvolution to enhance resolution and reduce blur, thereby improving alignment precision before the actual comparison with CAD data occurs
Solution Approach 2:
The patent introduces an intermediary processing step between optical imaging and CAD alignment. Image reconstruction algorithms act as an intermediary that transforms low-resolution optical images into enhanced images with improved resolution and contrast, facilitating more accurate alignment with CAD data
2Productivity
If device features are shrunk to below 20 nm to increase device density, then more devices can be packed on IC, but optical resolution limitations prevent observation of these small features
Solution Approach 1:
The patent applies dimensionality change by acquiring optical images at multiple focal planes (depth dimensions) and combining them through image reconstruction. This multi-dimensional approach allows recovery of sub-20 nm feature information that cannot be obtained from single-plane imaging, effectively overcoming optical resolution limitations
Solution Approach 2:
The patent changes imaging parameters by capturing images at multiple focal depths and using deconvolution algorithms with point spread function characterization. These parameter changes enable resolution enhancement beyond the diffraction limit, making sub-20 nm features observable
3Area of stationary object
If no fiducial marks are used to maximize device packing, then IC size is minimized, but alignment of CAD data to device image becomes imprecise
Solution Approach 1:
The patent applies self-service by enabling the optical image itself to provide alignment features through image reconstruction. The reconstructed images enhance natural features and edges within the device structure, allowing the image to serve its own alignment function without requiring external fiducial marks
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 significantly improves alignment precision, allowing for more accurate observation and testing of IC features, faster acquisition of ac device performance, and reduced noise in LVI data, overcoming the limitations of current optical imaging techniques.
Implementation Method 1
Optical methods may take advantage of minute ( ̃600 ppm) changes in the reflectivity of conduction regions (e.g., source-gate-drain) arising from voltage-induced changes in carrier concentrations
Implementation Method 2
optical imaging through the backside of the device ('looking' through the silicon substrate) may require using infrared (IR) light due to the silicon bandgap which causes absorption of light with shorter wavelengths
Data Source
AI summary
In one embodiment, a method for improving the alignment of CAD data to optical imaging data, such as LSM and LVI images of integrated circuits is disclosed. Image reconstruction techniques are applied to optical images, such as laser voltage images (LVI), laser scanning microscope (LSM) images, or emission images, to produce reconstructed images which may have higher resolution, increased signal-to-noise, or other enhancements. Multiple CAD pattern layers are processed to generate second CAD images more closely corresponding to the appearance of the reconstructed images. Alignment of the reconstructed images to the second CAD data may be substantially more accurate and precise than alignment of the initial optical images to the CAD data—in some cases this improvement may make the difference between a successful alignment and a failed alignment.