Wafer-Level Feed-Forward Dopant Dose Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As semiconductor devices become smaller and more complex, controlling critical dimensions and processing conditions across wafers and within wafers becomes increasingly difficult, leading to variations in transistor performance that affect electrical performance.
Innovation Solution
A method is developed to create a wafer-level semiconductor device electrical sensitivity map and a feed-forward computer model based on in-line physical measurements, allowing for variation of implant parameters by lot, by wafer, and by zone within a wafer to minimize electrical variation in transistor performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional uniform implant parameters are used across the entire wafer, then the fabrication process is simple and fast, but wafer-to-wafer and within-wafer electrical transistor variation increases
Solution Approach 1:
The wafer is divided into multiple zones (e.g., center, mid-ring, outer ring) and each zone receives customized implant parameters independently. This segmentation allows precise control of electrical characteristics in different wafer regions while maintaining overall process manageability through automated zoning algorithms.
Solution Approach 2:
The system performs preliminary mapping of the wafer surface topology and electrical characteristics before the implant process. This advance characterization enables the feed-forward control algorithm to pre-calculate optimal implant parameters for each zone, eliminating the need for iterative adjustments and reducing actual implant variation.
Solution Approach 3:
The system implements a feedback loop where electrical test data from completed wafers is fed back to refine the sensitivity map and update implant parameter recommendations for subsequent wafers. This continuous learning process improves manufacturing precision over time while the automated system manages the increasing complexity.
2Manufacturing precision
If wafer-level and zonal implant parameter variation is implemented, then electrical transistor uniformity improves, but measurement and control complexity increases
Solution Approach 1:
The system uses a single multi-functional measurement apparatus that performs multiple functions: topography mapping, electrical characterization, and parameter extraction. This universal tool reduces the need for multiple specialized measurement devices while maintaining the precision required for wafer-level and zonal control.
Solution Approach 2:
The sensitivity map acts as an intermediary that translates raw in-line physical measurements into meaningful implant parameter adjustments. This intermediate representation simplifies the control task by pre-computing the relationship between measured physical quantities and desired electrical outcomes, reducing the real-time measurement and control complexity.
3Manufacturing precision
If feed-forward computer model with sensitivity map is used, then implant parameter optimization is achieved, but computational complexity and processing time increase
Solution Approach 1:
The sensitivity map is pre-computed offline using historical data and computational models before production begins. This preliminary action stores the complex computational relationships in lookup tables or simplified mathematical forms, enabling rapid real-time queries during actual implant operations without repeating full computations.
Solution Approach 2:
The system computes implant parameters for all possible wafer zones and scenarios in advance (excessive action), then only retrieves and applies the relevant subset for each specific wafer (partial action). This approach trades increased pre-computation time for dramatically reduced real-time processing delays during production.
Data Source
AI summary
The present disclosure provides a method for adjusting implant parameter conditions in semiconductor processing by wafer and by wafer zone using in-line measurements from previous operations and a feed-forward computer model. The feed-forward model is based on a sensitivity map of in-line measured data and its effect of electrical performance. Feed-forward computer models that adjust implant parameters by wafer and by zone improve both wafer-to-wafer and within wafer electrical uniformity in semiconductor devices.