New Technique 'Infuses' Wafer Surface

A new surface processing technique that has applications in areas as diverse as ultrashallow junction (USJ) doping, epitaxial strained silicon and low-k pore sealing is now commercially available from a company called Epion (Billerica, Mass.).

Based on gas cluster ion beam (GCIB) processing, the technique works in a way similar to ion implantation in that a charged mass is accelerated through a beam line and directed at the wafer surface; the wafer is moved beneath the beam. In the Epion approach, however, only a very small charge is added to the cluster (one ion per several thousand atoms). This allows for high total energy interactions (>30 keV) but very low energies per atom (<10 eV), so that penetration into the substrate is minimal. The composition of the cluster can be modified depending on the application to etch, deposit, dope, ash, seal, etc. (Fig. 1 ). As in an ion implanter, dosimetry is performed by measuring the ion beam with a Faraday cup.

1. Gas clusters can be neutral (left) or reactive (right). The ‘chemistry’ of the cluster can be selected to perform a wide variety of processes, including etching, deposition, ashing and doping. (Source: Epion)

Epion technologists believe this new approach will have some unique advantages at the 45 nm node. "The GCIB process is inherently a surface technique. This is a highly energetic chemical beam which has no process effects deeper than about 20 nm into any surface," said John Hautala, CTO at Epion. "We feel very well positioned for the 45 nm node, where the dielectrics are sensitive to plasma damage, interface control is critical, the junctions need to be ultrashallow, thermal budgets are low, and high-quality localized SiGe and strained silicon are needed."

GCIB processing was initially developed by Professor Isao Yamada at Kyoto University in the late 1980s, with the first results published in the early 1990s. Epion worked on commercial development during the 1990s and now has 20 systems in the field, used for surface acoustical wave (SAW) device production, data storage applications and UV maskmaking. Now the company has "refocused" on the semiconductor industry and is introducing a new 300 mm tool. Well-known entrepreneur Peter Rose is a key director and advisor, Allen Kirkpatrick is CEO, and Takashi Shimada is president of Epion's subsidiary operation in Japan.

The first application of the tool is likely to be in the formation of USJs. Epion presented a paper at the International Workshop on Junction Technology (IWJT) earlier this year titled, "Infusion Doping for USJ Formation," describing how doping was achieved with clusters made up of argon- and boron-containing molecules such as B₂ H₆ and BF₃, with an energy per atom of <10 eV. No channeling or end of range damage was observed and a 1 × 10¹⁸/cm³ junction depth (x_j) at 12 nm was achieved. This is much better than what can be achieved with traditional ion implantation techniques. Another significant advantage is that the process can be made to be self-amorphizing, which allows for the low-temperature diffusionless solid phase epitaxy (SPE) activation. Co-doping with germanium results in higher boron solid solubility as well.

The technique has also been used to thin SOI layers to create fully depleted devices, for low-k etching, ashing and pore sealing, and to infuse germanium into silicon for localized strained silicon structures.

"With etch chemistry added to the clusters, and the ability to control the processing at every square millimeter of the surface, we can thin SOI down to fully depleted values of <10 nm with improved uniformities across the wafer," Hautala said. "This same process also works very well for a non-damaging highly controllable low-k etch. Perhaps the most interest we have in the back end is ashing, where the infusion process shows no evidence of damage to the fragile porous materials. And with no measurable byproducts, no wet clean is required, which appears to be important to the IC manufacturers. The pore sealing also happens as a byproduct of the ash process."

As shown in Figure 2 , the technique relies on interactions generated when the gas cluster hits the wafer surface, creating a kind of mini-explosion, with high temperatures and pressures in the small impact infusion volume. The chemistry of the cluster determines the reaction.

2. Gas clusters slightly penetrate the wafer surface in the first 6 psec (left), creating a large amount of heat and, depending on the chemistry, a chemical reaction. After 16 psec (right), the cluster has broken up. These tiny reactions are repeated across the wafer. (Source: Epion)