Electron beam lithography for 0.13 μm manufacturing

Abstract
General requirements for the use of electron beam lithography in direct write manufacturing of silicon integrated circuits are discussed. 50 keV is suggested as an optimum beam energy, since this is the minimum beam energy that can achieve high aspect ratio structures (4:1) in single layer resists in a manufacturing environment. Higher beam energies result in an inefficient exposure process requiring larger currents; this combination will lead to excessive resist and wafer heating. Lower voltages will require the use of top surface imaging or multilayer resists, which have concerns of processing complexity, resist charging, and defects. At 50 keV, some form of proximity correction is required to achieve reasonable control of critical dimensions. While one of the principle arguments for low voltage lithography is that it avoids the need for proximity correction, proximity correction is a solvable problem for large chips and is therefore a less risky approach than developing a reliable surface imaging resist technology. From a quick review of available resists and recent resist progress, it appears that a sensitivity of 5 μC/cm2 at 50 kV is the best that will be achieved in the next several years. Neglecting overheads, for a design point of 40 8 in. wafers/h, a peak beam current of 13 μA for a raster scan or projection tool is required. One of the major challenges of designing a tool with such high beam currents is controlling space charge effects so that there is minimal impact on lithographic quality. After discussing the characteristics of various high speed electron beam writers that have been made to date, it will be concluded that there are two types of systems that have the best chance of meeting all of the requirements—a projection system such as SCALPEL, and a multibeam system with hundreds of independently blanked beamlets. These systems minimize space charge effects by spreading out the electrons through a larger volume of space, allowing a larger total beam current. However, in order to make these systems a commercial reality, a great deal of innovation, research, and development are still required.

This publication has 15 references indexed in Scilit: