ION implantation, yesterday and today

05/01/1997

Ion implantation, yesterday and today

Peter H. Rose, Krytek Corp., Danvers, Massachusetts

Ion implantation was among many processes identified and explored at the beginning of the semiconductor industry back in the1950s. Since then, ion implanters have developed to low- and medium-current implanters in the early 1970s and medium- and high-current machines at low energies in the 1980s. High-energy implanters are in growing demand now, and it is likely that the semiconductor industry will require a new class of high-performance, low-energy machines in the future.

This issue of Solid State Technology celebrates the publication`s 40th anniversary and also takes us back to the beginning of the semiconductor industry. In the 1950s, Bell Laboratories led the world in basic semiconductor science and technology, and ion implantation was among the many processes identified and explored at the laboratory then. The beginnings of ion-matter interactions, however, go back much further, arguably to the 1900s, when Aston developed the first ion mass spectrometer. This apparatus had some of the features of the modern ion implanter: an ion source, a means for accelerating the ions to tens of kV, and an analyzing magnet. It could not only identify nuclear isotopes (its primary purpose), but since it made available pure, energetic ion beams, it could be used to study such phenomena as sputtering and the penetration of ions into matter. The study of ions passing through gas and solid targets became an important field in physics, but not until the 1950s were ions recognized as a means to modify materials in useful ways. W.N. Shockley at Bell Laboratories discovered that ion implantation could be used to dope semiconductors. He also found that the damage preventing the technique from being exploited could be annealed away. The first patent describing all the basic features of an ion implanter was granted to R.S. Ohls, also of Bell Laboratories, in 1956.

For a long time, however, ion implantation was not considered a viable alternative to doping using diffusion furnaces. Furnaces are relatively simple tools that do not require a vacuum environment, and they satisfied the device requirements of the 1960s and 1970s. Mostek Corp. was the first to successfully commercialize an ion implanter using low doses to adjust threshold voltages.

By 1971, there were four small implanter companies in the US (KEV, Ortec, Extrion Corp., and Accelerators Inc.), one in the UK (Lintott Ltd.), and another in France. The market for ion implanters was small, and device manufacturers were only beginning to demand the more precise dose control, self-aligning gates, and lower process temperatures made possible by implantation. Fabs were naturally reluctant to replace diffusion furnaces with the much more complicated, and consequently less reliable, tool. In fact, all equipment in the fabs at that time was much simpler. Still, the technical drivers were rapidly requiring improved performance from equipment in the fab, and furnaces were clearly unable to supply the precise dose control necessary.

The accelerator technology needed by ion implanters was developed in various nuclear physics laboratories and at one or two key companies, such as High Voltage Engineering Corp. The first ion implanters looked like research accelerators. They had a long beam line, a large footprint, and were not packaged in a way acceptable to a manufacturing facility. Extrion Corp. became the industry leader in adapting the technology to meet customer needs. Today, the user and equipment supplier work together closely to define equipment specifications. For example, the "International 300 mm Initiative" (I300I) was held in Austin, TX, this year.

Figure 1 gives some idea of the dramatic change in the range of applications and performance of ion implanters since the early 1970s. The area outlined in black was served by the low- and medium-current implanters available at that time. In 1980, high-current implanters were developed to cover high dose applications (green). In the mid-1980s, megavolt implanters moved from the development stage to full production use (blue). From the mid-1980s, there has been a drive to increase the beam current of medium- and high-current machines at low energies, due to the increasing importance of shallow implants.

Figure 1. Dose vs. energy diagram showing the growth of applications with time; black: 1970, green: 1980, blue: 1985, red: 1998. *Some of these applications require high tilt angles and wafer repositioning. (Adapted from a graphic provided by R. Simington, Eaton Corp.)

This task is fundamentally difficult because the beam line of the conventional ion implanter is long and space charge causes the current to fall dramatically as the energy is lowered. To reach energies as low as 1 keV, specialized machines are designed to perform well at these energies (red). For example, Applied Materials Inc. and Eaton Corp. are offering new implanters with greatly improved low-energy performance. Applied Materials solves the beam-transport problem by decelerating the beam to low energies close to the target, while Eaton employs a very wide aperture beam analysis and transport system. Both machines use spinning disks for batch processing. Choosing between single-wafer or batch-wafer processing will have a significant influence on the design of implanters for larger wafer sizes (see "Batch size effects on 300-mm ion implant productivity," Solid State Technology, Sept. 1996, pp. 191-192). For low energies, though, disk target chambers (batch) give the shortest beam line. The best design choice requires consideration of a complex of issues, such as combining the lowest cost of ownership with the highest reliability.

Today, the high-energy machine is the fastest growing segment of the implanter market. High-energy implanters are large and expensive. A reduction in the cost and size of these machines would be most advantageous, and could be effected by accelerating higher charge state ions than are being used today.

Finally, two new ion implanters show the range and versatility of the implantation process. The first example, the ORion NV6072 (Fig. 2), demonstrates the use of an unanalyzed beam. It is designed to implant flat panels with the directly extracted beam from a large-area bucket ion source. Through an ingenious use of loadlocks and gas cooling, the machine is able to keep the panels from exceeding 100?C during implantation. The second example, the Ibis 1000 (Fig. 3), uses the highest currents employed for wafer implantation. It can implant 75 mA oxygen beams, at energies up to 220 keV, into wafers held at 600?C in a specially designed spinning disk target chamber.

Figure 2. A schematic illustration of the flat panel handling in the ORion NV6072. The machine is designed to process panels as large as 600 ? 720 mm in area and only 0.7-mm thick. Glass panels are electrically insulating and have poor thermal conduction, a formidable combination of difficulties affecting both charge-up and cooling. The panels are moved inside the machine by an internal robot. Since the panels are large, the implanter footprint has to be kept as small as possible. The robot picks up the panel from a horizontal loadlock, rotates it into an almost vertical position, and moves the panel up and down through the ion beam.

Figure 3. The Ibis 1000. For the first time, high-frequency 100-300 Hz magnetic fields are used to generate a parallel line scan. Parallel implantation of the ions across the whole wafer is achieved by a combination of the magnetic scan and a spinning disk.

In conclusion, it appears that innovations in implanter design will continue at a fast pace. In the immediate future, the most rapid growth will be seen in the demand for high-energy machines to satisfy the increased need for deep implants with better isolation of adjacent transistors. It is also likely that the semiconductor industry will require a new class of specialized low-energy machines for cost-effective shallow junction formation.n

PETER H. ROSE received his PhD degree in nuclear physics from the University of London in 1955. In 1971, he became founder and president of Extrion Corp., which was acquired by Varian Associates. In 1978, Rose founded Nova Associates, which was acquired by Eaton in 1981. Additionally, Rose was president of Ion Microfabrication Systems Inc. In 1993, he retired from Eaton and founded Krytek Corp. In 1986, Rose received the SEMMY Award for developments in semiconductor processing equipment. In July 1996, he was awarded the National Medal of Technology by President Clinton. Krytek Corp., 368 Cherry Hill Drive, Danvers, MA 01923; ph 508/524-9336; fax 508/524-9224.