SIMULATION METHOD AND SYSTEM BASED ON MACHINE LEARNING MODEL WITH REDUCED TRAINING REQUIREMENTS

Smart overview of the Invention

The patent application describes a method and system for simulating semiconductor device processes using a machine learning model. The approach involves acquiring a target parameter and initial state profile data to generate subsequent state profile data. This data reflects the attributes of semiconductor device states, offering an alternative to traditional technology computer-aided design (TCAD) simulations.

Technical Field

The focus is on improving simulation methods for semiconductor manufacturing by leveraging machine learning models. This technique aims to address the increasing complexity and accuracy challenges faced by conventional design simulators like TCAD, which require extensive calibration to predict semiconductor characteristics accurately.

System and Methodology

The system includes at least one processor and a non-transitory computer-readable storage medium that stores instructions for performing the simulation method. The method involves obtaining a target parameter and initial state profile data, then using a machine learning model to generate new state profile data that corresponds to the target parameter. Each set of state profile data represents an attribute profile of a semiconductor device's state.

Machine Learning Integration

The machine learning model can be constructed from various architectures such as artificial neural networks, decision trees, or support vector machines. The model processes initial state data to produce subsequent state data, effectively simulating the semiconductor process by calculating solutions to equations like the drift-diffusion equation, which models charged particle movement in semiconductors.

Application Example

An example illustrates using the system to simulate a transistor's electron and hole densities and electrostatic potential based on contact voltage and doping profiles. These profiles are predetermined by process parameters such as dopant type and application conditions. The method efficiently calculates these profiles, providing valuable insights into the semiconductor's behavior under specified conditions.