Blast from the Past 3: FIB Applications Then (1999) and Now
by Cheryl Hartfield
FIB applications and developments for integrated circuits were a hot topic at ISTFA’99, as the acceptance and fan-out of FIB technology to the semiconductor industry were well underway in this time period. As documented in an ISTFA 1999 Trip Report, some of the FIB User Group discussion centered on circuit edit applications. Excerpts from this follow below:
“Microsurgery Overview. R.H. Livengood discussed Intel’s strategy for circuit edit/modifications and design debug procedures using FIB. For large areas, they prefer laser (LCVD) since it takes FIB >10h to mill a 100μm x 100μm box to 20μm depth. Initial steps are industry standard: thin a flip chip to 200μm, use the REVISE laser (Ar) chemical etcher to remove most of remaining Si over mm2 area of interest, then go to FIB to remove remaining material, stopping on oxide, then depositing blanket oxide with FIB, drilling through, and doing metal re-route. What is different is Intel’s strategy. Transistor densities present a challenge to backside analysis of flip chips. Therefore, design for debug needs to be incorporated into the chip design. Intel utilizes a PDFD (physical design for debug) process that creates physical hooks for cut and paste FIB operations, accessible from the backside to enable re-routing. These hooks are incorporated into every test die so circuit revisions can be easily confirmed before changing the Si at the fab level with new mask sets. Intel states this is a strategic advantage that enables them to speed development cycles.
Integration of laser and optical with FIB. Integration of either one of these two tools would improve throughput. One approach to add optical capability inside the FIB has been to attach an optical camera to the FIB column, spaced by a user-controlled distance “d”. This allows, with proper orientation, the ability to locate your FIB target area on an identical die, and the stage can be translated to view the actual milled area optically. This will allow more accurate and quicker identification of the target structures to be milled in the FIB. However, this approach will be difficult for packages due to surface flatness alignment issues. For laser, much work is required before a viable laser solution is identified. A Nd/YAG laser was tried but did not have good resolution and was not fast enough for backside work. An argon ion and excimer laser have been considered for material removal and metal deposition, but there are technical issues that need to be overcome (mainly, the environment created by the laser removal is not compatible with the FIB environment, but this could change with the introduction of variable pressure SEM columns).”
Removing large areas with FIB is still a challenge today, driving additional developments. The need for large area removal is driven not only by circuit edit, but also for packaging applications such as stacked die and new Si interconnect strategies such as through-Si vias. A new ion beam technology capable of very high sputter rates has been developed recently to meet this need and was commercially released in 2008 as a 25keV Xe inductively-coupled plasma beam. Reportedly, this technology can remove 16 million μm3 of Si in 1 hour (5000μm3/s). This is at least 12x faster than the broad ion “cross-section” technology developed by various manufacturers to remove large areas of material quickly (http://www.imec.be/efug/EFUG2008_05_Smith.pdf). Other commercial entities are engaged in a similar effort to introduce powerful ion beams.
The idea of a laser for large area removal is still alive and well. At EIPBN 2010, work was presented on XeF2 gas-assisted laser removal of material with a goal of milling large areas cleanly. Laser milling is 103 – 106 times faster than the standard Ga FIB and allows large power density, even at low pulse energy. In lab benchtop tests, 5 x 106 μm3/sec removal was demonstrated with a heat affected zone in the tens of nms. Challenges include redeposition on the lens, as has been reported by others. Another challenge, as referred to in the ISTFA 1999 FIB user group meeting, is that laser material removal creates a high local pressure that is not compatible with the FIB environment. Some researchers are taking an approach of bypassing the normal FIB environment to create a combined laser-FIB instrument that operates at atmosphere and applies a local vacuum (Yoshida et al., Rev. Sci. Instrum. 81, (2010) 02B702). This results in relatively high pressures for ion beam operation in the range of 10-3 mbar. The strategy is directed towards volume manufacturing and involves large area laser etching, followed by FIB clean-up.
Additional applications for lasers combined with electron and ion beam microscopes are arising. Dutch researchers have created the ILEM (integrated light electron microscope) for fluorescent imaging in the TEM, and Omniprobe has recently released the OptoProbe™, a new tool that introduces light into the FIB or SEM for a variety of applications. Stay tuned for further developments.
The views and opinions expressed herein represent those of the author and are not necessarily those of Omniprobe, Inc., or its employees. Any mention of product or company names are not endorsements. Trademarks mentioned in this blog are the exclusive property of their respective owners.