Thomas J. Sanders

Integrated circuit design for manufacturing through statistical simulation of process steps

The methodology, implementation, and results of a design for manufacturing (DFM) technique as applied to an integrated circuit boron base formation for an n-p-n transistor are presented. The primary purpose of the DFM technique is to achieve acceptable statistical prediction of the results by using the minimum number of variables and reducing the time required to perform physically based simulations. Excellent statistical results are achieved while the number of simulations is reduced by at least a factor of five if judicious statistical techniques are applied. >

Industrial perspective on research needs and opportunities in manufacturing assembly

Abstract This paper investigates whether investments in research and development (R&D) could have a significant impact on reducing the cost and/or enhancing the effectiveness of manufacturing assembly. It also recommends a research agenda that from an industrial perspective would result in the highest return on investment in assembly R&D. The assessment was conducted on 24 product lines across companies ranging in size from

Freeze-out characterization of radiation hardened N/sup +/ polysilicon gate CMOS transistors

10 million to

Silicon nitride deposition process for low cost microelectronics applications

2 billion in annual sales. Approximately 64% of the assembly dollar value surveyed corresponded to Department of Defense (DoD) products. A needs and trends analysis was conducted to determine the perception of industrial participants regarding their current investment, need for investment, and cost impact potential of assembly activities. Findings for government contractors versus commercial firms, for prime contractors versus subtier contractors, and for electrically intense versus mechanically intense operations are presented. The research methodology also included analysis of qualitative data from open-ended questions and interviews. This resulted in identification of three major assembly issues: R&D opportunities, R&D inhibitors, and technology transfer. These findings are also summarized.

Techniques for optimizing statistical simulations (IC processes)

Freeze-out behavior of radiation hardened N+ polysilicon gate CMOS devices is studied at liquid-nitrogen temperature before and after irradiation at different dose levels. Explanations are given for the threshold voltage variations and freeze-out effects both before and after irradiation. The degree of freeze-out of the counterdope implant is controlled by temperature, applied gate potential, and applied body-to-source potential. Based on experimental results of this work, freeze-out is not significantly affected by irradiation doses, or the magnitude of the gate bias during radiation. >

The use of STADIUM statistical TCAD simulation methodology in industrial environments

A new technique for Plasma Enhanced Chemical Vapor Deposition (PECVD) of silicon nitride is presented in this paper. This technique involves specially built equipment which allows the use of low temperature, low frequency processing. The results obtained show silicon nitride which has all of the positive characteristics of conventional silicon nitride but with equipment which is much less expensive and very easy to use and maintain.

IC yield modeling and statistical circuit simulation

The authors address the methodology developed for the Florida SEMATECH Center of Excellence (FSCOE) for performing statistical simulations of integrated circuit processes. This methodology involves doing a series of statistical simulations at the process, device, and circuit design levels. Various techniques to improve the amount of time it takes to obtain results from statistical simulations of processes using Suprem-IV are described. Methods the design engineer can use, rather than means of improving the models in the simulator are considered. The key to this methodology is that multiple simulations are performed in order to obtained the statistical results to adequately model the effect of the variation in the fab. Five possible techniques are outlined that reduce the simulation time and eliminate the need to buy expensive computer equipment or use less accurate simulation models.<<ETX>>

Modeling and test of pixel cross-talk in HgCdTe focal plane arrays

The software program called STADIUM has been developed by researchers at Florida Institute of Technology. The purpose of STADIUM is to facilitate the statistical design and simulation of integrated processes, devices and products. It is a software shell which uses existing simulators in a design of experiments statistical methodology, and which allows the estimation of the statistical response of the manufacturing environment. Some of the details of STADIUM are described together with activities of the Florida SEMATECH Center of Excellence to implement this software in several industrial environments.<<ETX>>

HgCdTe double-layer heterojunction detector device

An important concern whether designing a new process or maintaining an existing one is the cost of production and hence the chip yield. In order to maximize chip yield, the most significant process parameters need to be identified so that variations in these critical parameters can be minimized resulting in the highest possible chip yield. Presented in this paper is a software-based methodology for facilitating the identification of critical process parameters and relating them to circuit level performances using statistical analysis techniques and conventional simulators.

STADIUM CKT statistical simulation system

Mercury Cadmium Telluride focal plane arrays with well over 1000 pixels have been fabricated for a number of years. These FPAs have been built as large two-dimensional arrays of HgCdTe p-n junction diodes on a single CdTe or CdZnTe substrate. Sensitivity of each pixel to impinging radiation is one of the most important quality factors for these devices. However, material parameters, which give diode high sensitivity, are the same as those that cause cross talk between adjacent diodes in the array. This cross talk causes a blurred image and in general is a detrimental factor for the FPA system. The cross talk modeling is done in a two- dimensional simulation format to achieve high accuracy. In addition, the output information can be generated as a statistical function of the material and design parameter variations. Actual heterojunction FPA devices have been fabricated and tested for cross talk. In the paper, this data is compared to the simulation results. This design method and its algorithms are encapsulated in a software program called IRSIM. This physics-based simulator allows the engineer to use versatile geometries and material concentrations.