One of the most difficult challenges in the semiconductor industry is developing a direct-write electron-beam lithography system that employs multiple columns and multiple beams within each column. William Parker and his colleagues at Ion Diagnostics Inc. (Santa Clara, Calif.) are up to that challenge. And they need all kinds of specialists to do the job, from mechanical engineers to scientists in optics and physics, and of course EEs. Such a team could enable inline e-beam inspection systems for checking 300-mm wafers at a clip of nine wafers per hour.
The potential is enormous: A maskless lithography system for both drawing circuits and inspecting them is the Holy Grail, particularly as the costs to write mask sets escalate to the point where low-volume chips may not be affordable, except in FPGA format.
And direct-write electron beams are capable of sub-100-nanometer resolutions.
|William Parker is developing a multicolumn e-beam system at Ion Diagnostics, for both drawing and inspecting
300-mm wafers at a fast clip. |
The stumbling block has been throughput-leading companies to propose multicolumn DWEB systems. As the pixel size decreases with process scaling and as wafer diameters move to 300 mm, throughput becomes an even bigger issue with single-column e-beam systems.
Ion Diagnostics employs cold-cathode arrays, which can be fabricated in parallel using the multibeam, multicolumn (MxM) approach.
The individual columns are spaced on 20 x 20-mm centers. A microfabricated cold field-emission array is used as the electron-beam source within each column. Each array can produce 32 independently controlled beams per column, writing at a high voltage of greater than 50 kilovolts to simplify proximity correction.
The beams would use common deflectors, an objective lens and a back-scattered electron detector for all 32 beams. Each microfabricated array would serve one column, with integrated electron optics to shape, align and blank each beam.
Ion Diagnostics currently proposes a system with 201 columns for 300-mm wafer production and 88 columns for a 200-mm wafer system.
William Parker is the son of a distinguished engineer, Norm Parker, who was a senior technology officer at Motorola Inc. William Parker also worked at Motorola from 1966 to 1972, and has a doctorate in physics from the University of Chicago in particle optics and materials analysis. He built an electron microscope while at Chicago, and then moved to Hughes Aircraft Co., where he headed up a group developing a focused ion-beam microprobe.
Parker said his company is backed by private investors and by Motorola Inc., which has a minority interest in Ion Diagnostics. Joe Nahas, manager of technology planning at Motorola, said Motorola has supported the technology during the incubation phase, with the goal of proving the system's feasibility. The challenge for Ion Diagnostics, Nahas said, is that an MxM system must deal with an extremely complex set of data, on the order of 10 to the 15th power number of bits. "The complexity of the data set is roughly equivalent to the mask data, which is presented to the tool, piped in as parallel data streams." Over the long range, such complex data-handling tasks-controlling some 6,000 sources in parallel-will be handled relatively easily by more powerful microprocessors, Parker said. His small staff of about 20 has a mix of skills in the electron-beam, optics, stage-positioning, system design and control subsystems fields.
In other ways, an MxM system is simpler than a single-column system. In a multicolumn system, the wafer stage travels the distance of the wafer, but in a multicolumn system the stage movement is minimal. And because the writing current is divided among many, there are no space-charge problems, he said. An MxM system can operate at high voltages, which improves the proximity correction as the fields are stitched together to make ICs.
Thus far, the group has created a prototype system with an electron-beam column capable of producing 32 individual electron beams. The group has developed source-conditioning technology that reduced the electron beam's temporal instability-that is, current fluctuations-which over time reduce the process latitude of the system.
Gerhard Gross, director of lithography at International Sematech, said he believes DWEB lithography has a future, particularly as mask costs continue to remain high. Semiconductor companies may be reluctant to bet on MxM systems early on, but they may be led to the concept as larger companies promote more conservative implementations of multicolumn DWEB systems.
Will Ion Diagnostics be able to create a prototype system that attracts support from other large commercial IC manufacturers besides Motorola? Will it gain enough funding, in a venture-capital environment that tends to shy away from IC equipment vendors? Those questions remain to be answered, but one thing is sure. Bill Parker and his team of scientists have put their energies and knowledge toward answering a complex set of challenges which, if solved, will change the course of the semiconductor industry.