Study On The New Test Method For Sheet Resistance

With the high speed development of science and semiconductor industry,the integrate level of IC ,as the basic component ,becomes increasingly integrated.The constantly growing of silicon wafer radius,the minuter of graphics and the shrinking of circuits dimension, make stringent requirements for the machine intensity and the electronic character of the crystal.The measurement of sheet risistance becomes an important process through the IC's making.Now,the measurement of sheet risistance attract extensive attention and four-point probe measurement technique is widely used.But,for large area risistance measurement, four-point probe measurement has problems such as,big error for mechanical movement and low measurement speed. In this context, this article use EIT technology to test sheet resistance of semiconductor and use neural networks to reconstruct resistivity image.Electrical Impedance Tomography is a new non-invasive medical imaging technology developed in recent years.In EIT, small alternating currents are injected via an array of electrodes which is attached around the sample and corresponding voltages can be measured via some other electrodes.Using above current injections and collection voltages, an approximation for the distribution of impedance can be reconstructed. EIT has now become the research focus in bio-medical engineering, geological exploration areas for its non-invasive,inexpensive and safe. Common EIT image reconstruction algorithms such as back-projection reconstruction algorithm and Newton-Raphson algorithm have shortcomings of large amount of computation, low speed, and poor imaging accuracy. This article intends to use neural network to EIT image reconstruction. Ordinary neural network algorithm offen give slow convergence rate and is easy to fall into local minimum, so it is restricted in many engineering.Based on the theory of artificial life and evolutionary computation, particle swarm optimiszation inspired by social behavior of bird flocking or fish schooling. PSO is simple in concept, few in parameters, and easy in implementation. It was proved to be an efficient method to solve optimization problems, and has successfully been applied in the area of function optimization, neural network training and fuzzy control systems, etc. However, both theory and application of PSO are still far from mature. This article try to improve the PSO mechanism,in order to achieve increased speed and accuracy purposes,based on the theory and application of comprehensive analysis of the BP neural network and the PSO. At last, this article combines improved PSO and BP neural network to test sheet resistance of semiconductor, in which the BP network learning algorithm is replace by improved PSO.The main contents of this study are as follows:1,Discussing the importance of the sheet resistance semiconductor measurement,analyzing the basic principle of for-point probe measurement technique in details.2,Establish the EIT forward problem mathematical model for sheet resistance semiconductor measurement.Study the form of injection current.3,Study two commonly used algorithms for inverse problem of EIT, back-projection reconstruction algorithm and Newton-Raphson algorithm. Design simulation software of corresponding image reconstruction algorithm.4,Combines improved PSO ,BP neural network and EIT algorithm to test sheet resistance of semiconductor.5,The basic principles of particle swarm optimization and optimization of the mechanism,the current problems in particle swarm optimization algorithms. The oretical basis and practical way to improve the search performance of PSO.6,Design and development a 32-electrode electrical impedance imaging data acquisition hardware system. Design corresponding image reconstruction algorithm and the data communication software.The innovative main performances are:1,Using EIT technology to the measure of semiconductor sheet resistance, the silicon concentric circles model was established, the article laid the foundation of the measure of semiconductor sheet resistance using EIT and neural network technology.2,Proposed neural network for semiconductor conductivity distribution, solving the inverse problem of EIT, using PSO algorithm to optimize the establishment of neural network and achieved good results.3,Designed a new improved PSO algorithm, which was used to replace the BP learning algorithm of neural network, significantly improved the system's learning ability and prediction of evolutionary effects.4,Design and develop a set of resistivity data acquisition system using MCU and serial communication transfer data between PC and data acquisition system. The machine has adjustable injection current, stable, flexible, scalable, and other characteristics.5,ON matlab and Visual. Net platform, develop image reconstruction software on PC and successful completing the data transmission and the image reconstruction with neural network.