Beam Lithography Process Dose and Bias Determination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current beam lithography processes face challenges in determining an optimal process dose and process bias due to dependencies on various parameters, making it difficult to calibrate the isofocal dose and other optimum process doses, especially since most beam writing apparatuses cannot actively adjust beam focus and experimental data are correlated with multiple process-dependent factors.
Innovation Solution
A method is developed to determine the process dose and process bias by accessing a dataset associating measured feature dimensions with different exposure doses and feature densities, using a model parameterized with measured feature dimension, exposure dose, feature density, process dose, and process bias, allowing for the decoupling of process dose effects from bias effects, and optionally including process blur parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If experimental data are collected to determine optimal process dose, then calibration accuracy improves, but measurement precision deteriorates due to correlation with multiple process-dependent factors
Solution Approach 1:
The patent segments the determination of process dose and process bias into separate independent parameters. By using a model that treats process dose and process bias as distinct variables, the method separates the calibration task into independent optimizations, avoiding the confounding effects of correlated data. This segmentation allows each parameter to be determined without interference from the other, resolving the measurement precision issue caused by data correlation.
Solution Approach 2:
The patent introduces a computational model as an intermediary between the raw experimental data and the final process parameters. This model acts as a mediator that processes the correlated experimental data and outputs decoupled process dose and process bias values. The model transforms the complex correlated data into independent parameters through mathematical relationships, effectively using the intermediary to resolve the information loss problem.
2Manufacturing precision
If beam focus is actively adjusted to optimize patterning, then manufacturing precision improves, but device complexity increases due to apparatus limitations
Solution Approach 1:
The patent changes the approach from physically adjusting beam focus to computationally adjusting process parameters. Instead of modifying the physical state of the beam (which would require complex apparatus), the method changes the process dose and process bias parameters in the computational model to achieve optimal patterning. This parameter-based approach maintains manufacturing precision while avoiding the need for complex focus adjustment mechanisms.
Solution Approach 2:
The patent replaces the mechanical/physical beam focus adjustment system with a computational parameter optimization system. Rather than using mechanical focus adjustment mechanisms that would increase device complexity, the method uses software-based parameter optimization to achieve the same patterning improvement. This substitution of mechanical systems with computational methods resolves the device complexity issue while maintaining manufacturing precision.
3Productivity
If process dose and process bias are determined separately, then determination efficiency improves, but model complexity increases
Solution Approach 1:
The patent segments the model into distinct components for process dose determination and process bias determination. By creating separate model components or separate determination steps, the method enables independent optimization of each parameter, improving determination efficiency. The segmentation allows parallel or sequential processing without requiring complex interactions between the determination processes.
Solution Approach 2:
The patent creates a universal model framework that can determine both process dose and process bias using the same experimental data set. This multi-functional model handles multiple determination tasks within a unified structure, improving efficiency by avoiding redundant measurements while maintaining manageable complexity through modular design. The universal model can be adapted to determine different parameters without requiring completely separate systems.
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 enables efficient determination and setting of the process dose and bias, facilitating accurate pattern fabrication in the nanometer regime by isolating dose and bias effects, and allowing for size adaptation and proximity effect correction in beam lithography processes.
Implementation Method 1
The energy deposited by beam writing in the resist (i.e., the exposure dose) has an influence on the development process
Implementation Method 2
Electron scattering artefacts have been widely investigated in electron beam lithography. When a resist has been exposed by electron beam writing, electron scattering prevents that the developed resist regions mirror exactly the exposed resist regions. Electron scattering occurs within the resist itself as well as at the underlying substrate (in terms of backscattering).
Implementation Method 3
Moreover, some of the scattered electrons escape the resist towards the beam writing apparatus and are reflected back by an objective lens of the apparatus. This effect is called fogging.
Data Source
AI summary
A technique and method for determining a process dose for a beam lithography process includes accessing a data set that enables associating (i) a plurality of measured dimensions of features exposed by beam lithography with (ii) a plurality of different exposure doses, wherein the features were exposed with the different exposure doses, and with (iii) at least one of a plurality of different densities of the exposed features and a plurality of different nominal dimensions of the exposed features. The method also includes providing a model that is parameterized in at least the following parameters (i) measured feature dimension; (ii) exposure dose; (iii) at least one of feature density and nominal feature dimension; (iv) process dose; and (v) at least one process bias. In a further step, the method includes fitting the model with the data set to determine the process dose and the process bias.