3.

4.

5.

6.

the present status of automation at various stages of production within the IC industry, here including both device manufacturers and manufacturers of processing equipment;

the impact which emerging process technologies are likely to have on automation in the future;

the status of development of the measurement technology needed for control of automated processes, and the availability of equipment to perform such measurements;

the business, economic, and technological factors which may inhibit or delay the advent of automated processes; and

the priorities for automating various key process and assembly functions and the criteria appropriate for developing these priorities.

The study identified certain areas of importance for further automation of IC production. The primary emphasis of this study of automation was its application to the processing of wafers for integrated circuits and the assembly of chips from these wafers into devices. Thus, other areas of automation also applicable to semiconductors or the circuit design of ICs were only given minor attention or excluded.

Information for the study was obtained from field visits, interviews, and analysis of in-house data on the basis of which subjects especially relevant to automated IC processing and assembly were identified. Discussions were restricted to information which

PROCESSING

respondents felt was not proprietary and not confidential. In this connection over 20 commercial IC manufacturers were visited regarding the automation and measurements used in IC production and how they intend to apply automation in the areas under study in the future; 15 other laboratory or pilot IC fabrication facilities were visited to discuss their special needs for automation and measurements in IC manufacture; and relevant equipment, such as processing equipment, production automation systems and services, measuring equipment, and test systems were reviewed.

Many IC manufacturers considered the automation of their production an important subject with certain details held proprietary. Nevertheless, in general respondents were quite willing to discuss information on measurements and automation which are normally shown to interested customer-visitors and where a description of capabilities on ongoing research might enhance their own marketing efforts, such as new research already published, processing equipment or systems for sale, controls included in detailed purchase specifications, and trends in automation on a broad and general basis. The helpful cooperation of many individuals in the industry was essential to the conduct of the study and is gratefully acknowledged.

Today, some 80 to 130 steps are used in manufacturing ICs. Many of these are repeated, so that about 20 to 30 basic unit processes or operations are used. The differing susceptibility of various unit operations has led to differing status in regard to the degree of

*

This study was conducted by Arthur D. Little, Inc., Cambridge, Mass., under NBS Contract No. 4-35807. The views and conclusions expressed herein are those of the author and should not be interpreted as necessarily representing the official position, either expressed or implied, of the Defense Advanced Research Projects Agency or the National Bureau of Standards. This summary is presented in preliminary form both to provide feedback to the industry and to elicit comments, pro or con, from the industry. Comments should be addressed to W. M. Bullis, National Bureau of Standards, Washington, D. C. 20234.

automation. It was found that automation has only been applied to those unit operations where positive results of control were demonstrable, technologies were feasible and available, automation appeared necessary or desirable, economics were favorable, and management interest and business climate were positive.

Although many process steps now are extensively mechanized, only a few steps are automated to a high degree of control. The increasing mechanization of handling has improved process consistency. This provides an approach toward a smooth flow of product wafers, chips, and devices and has aided efficient and rapid processing and facilitated measurement and data collection on a similar mechanized basis. The higher throughput per equipment also has improved the economics of measurement and processing.

There is a general trend towards automation of individual unit processes. The least automated process steps are etching and cleaning steps which are still difficult to control at all. In other processes the processing conditions are kept constant with sensitive analog measurements and feedback control, such as for temperature, gas flow, or bonding pressures. Here open-loop control is the best solution at this time. This allows one to freeze the process or assembly conditions with individual regulators or feedback controls. The final device characteristics cannot yet be measured accurately enough or results are still too inconsistent for use of parameter measurements in closedloop control. Only in a few instances has closed-loop automation been applied to control some parameter of the IC crystal, wafer, device, or contacts.

Automation is being applied to the sequencing and control of the process variables, or to the data collection and analysis of status. Automation of time and program sequence control is becoming prevalent, and a few automated data collection systems are now being introduced. At present, these two automation approaches have been coupled only in the case of crystal growing. Although there are only a few computer-controlled systems in use, there is large and widespread interest in computer control and many in the industry specify computer compatability for all new equipment.

It was found that highly automated electronic test equipment is available for the measurement of IC and test pattern characteristics.

Several IC manufacturers are now extensively automating the collection and analysis of electrical measurements from test pattern chips measured at the completion of wafer processing with the hope that results from these measurements will provide improved understanding and control of the processes as well as means for further automation.

Other in-process parameters which are now routinely measured on wafers include the sheet resistance or resistivity and thickness of silicon layers, the thickness of dielectric layers, and the electrical characteristics of simple transistors and MOS capacitors. The impossibility of measuring the wafer, chip, or device during many processes generally rules out closed-loop control except via post-process parameter measurements. The latter are limited by problems of electrical contacting, optical instrument complexity, the minute dimensions involved, and inability to profile a wafer surface in depth nondestructively.

APPENDIX F

automation are the lack of correlation between some process variable measurements and product parameters and characteristics, present difficulties in measuring the thickness of dielectric or oxide layers especially in windows to be etched, limited accuracy and range of the measurement of desired process variables, long throughput time of wafer and device production, which delays the feedback of measurement results for corrective action or closed-loop control, lack of measurement equipment providing suitable accuracy and reliability for use in a production environment, lack of standardization for sequence control and data acquisition, too long a delay between a process step and the measurements such as on test patterns, and the potential displacement by alternative emerging processes.

loads, narrow base widths, or line widths of critical patterns. These uses are generally rather limited; more widespread policies of measuring product parameters in-process primarily for monitor purposes and indications of error or alarm have caused an unwarranted stagnation in the interest for improved measuring equipment for production use. However, it is important to note that improved methods of measurements and rapid evaluation of measurement data are needed for process control in future automated production and that several emerging processes use different technologies and require other physical principles of measurement than were used in production heretofore. Among the quantities involved are ion beam currents, dopant vapor pressures, etching rates of new etching media, alignment errors, photoresist sensitivity, laser position, and bonding pressure.