Blast from the Past 2: FIB Applications Then (1999) and Now
by Cheryl Hartfield
FIB user groups have been very active since the late 1990’s and are important networking venues to learn the latest and often unpublished information about industry and technology. The ISTFA’99 FIB User Group was no exception. Given the emergence of Cu metallization technology, flip chips, and backside analysis for IC’s in the late 90’s, it is not surprising the 1999 user group discussed gas assisted etching of Cu and the impact of FIB damage to sample preparation.
Excerpts below reveal the details of discussion from this meeting:
"Gas-assisted etching with Cu. Cu is more sensitive than Al to the common XeF2 enhanced etch, so it mills faster. As a result, lower pA are preferred to promote evenness of milling (100pA compared to 600pA). Gas assisted etching (GAE) with iodine gas is not recommended, because I2 shows no selectivity for Cu and there is conflicting data that corrosion and CuI by-products are formed (IMEC saw severe corrosion with/in one hour of exposure to I2 in FEI200 FIB; Gatan uses broad ion beam based technology with I2 and saw zero effect at any time). Also, iodine caused corrosion even if the gas was not turned on. As long as iodine had been used, if a sample was inserted in the FIB chamber within 24h of the iodine use, the corrosion was observed. Chlorine gas offers no enhancement, but also does not attack exposed Cu. A paper in J. of Vacuum Science (recent edition) shows Cl2 can form undesired non-volatile complexes with Cu, but Cl2 and NH4 together can work well (but tricky to get right conditions). For design debug purposes, it was noted that Pt or W deposited on Cu yields extremely low contact resistance (<1ohm).
Impact of FIB sample prep on TEM specimens. Fibics, Inc. presented data that for analysis purposes, in 99% of cases, the industry can live with the ion-induced damage caused by Ga implantation during TEM sample prep. High resolution TEM was performed on samples made on FIB systems with both 50kV and 30kV ion columns. Observed damage was minimal and did not interfere with analysis."
Today, Cu etching is still a hot topic. While the solutions developed in the past 10 years work well enough to enable circuit edit on Cu devices, there is still room for improvement, especially for Cu/low-K devices. Some solutions are patented and not available to the entire circuit edit market, and even in these cases, the success depends on the orientation and geometry of the area being milled. Formation of non-volatile Cu by-products, necessitating clean-up, continues to be an issue for GAE of Cu. New solutions are still being sought. At IPFA 2010, Makarov et al. presented a new Cu etch solution, using a challenging application for Cu etch as proof of concept – cutting Cu lines in deep narrow holes over low-k dielectrics. They show the solution offers an advantage of better imaging during gas flow, assisting FIB processing by better visualization of endpointing.
With respect to Ga+ damage, many studies have been done. In general, at 30kV, an amorphous damage region on the order of 20nm is created (in Si). This is shown by the image here, which displays a 16nm damage layer within a 143nm thick lamella (Langer et al., EFUG 2000). While not a problem for the typical 200nm thick TEM lamella, a 20nm thick damage layer becomes a problem for analysis that requires 20nm thin lamella, as is needed today for advanced semiconductor devices and samples destined for angstrom resolution imaging. However, a 5kV beam induces damage of only ~2.5nm and is now often used as a step in the final polish to clean the lamella. For most cases, Ga+ damage still does not present a problem for TEM inspection. While FIB damage is not an issue for most TEM samples, increasingly, FIB is used to create nanostructures for mechanical testing. Ga+ damage does impact the mechanical properties of materials, such as Mo-alloy single crystal, which exhibits FIB-induced hardening (Bei et al., Appl. Phys. Lett. 91, 111915-3, 2007). Mechanical testing of FIB-created nanostructures need to be interpreted with caution.
The next blog post will conclude this 3 part "blast from the past" series with additional information from the ISTFA 1999 FIB User Group.
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.