Protecting copper TEM specimens against corrosion via e‐beam induced carbon deposition

Abstract

Copper containing Transmission Electron Microscopy (TEM) specimens are vulnerable to corrosion during transfer from Focused Ion Beam (FIB) to TEM vacuum. The corrosion is an attack of the copper surface by sulfur from ambient laboratory air (present at ppb level). The sulfur attack leads to a CuS tarnish layer covering the specimen sidewalls, and holes are formed in the copper layers [1]. Hence the specimen has to be discarded for TEM analysis. Naturally, the native oxide of copper is protecting against sulfur attacks, but this oxide is removed during the specimen preparation by FIB. In this work, the protection efficiency of a carbon layer deposited in the FIB subsequently to the TEM specimen preparation, is studied. The carbon layers are e‐beam deposited in a FEI Helios460 dual beam FIB on a thinned copper specimen (Fig. 1). Sufficiently thin layers are achieved with a 15 pA, 5kV e‐beam scanning for 60 s/120 s/180 s in an area of 2x5 µm 2 with 300ns dwell time. For a full protection it is necessary to deposit carbon on both sidewalls of the specimen, which is done by rotating the FIB sample stage by 180 degree. The electron beam is incident to the TEM lamellae at an angle of 38º. The carbon layer quality and thickness are studied in a FEI Titan 3 60‐300 instrument on the sidewalls of a HfO 2 layer in a cone shaped specimen prepared through a Si/100 nm HfO 2 stack (Fig. 2). The HfO 2 cones give suitable contrast to the carbon layer and exhibit little ion beam sidewall damage from ion beam thinning. A small layer of redeposited material is covering the cone sidewall with a maximum thickness of 1nm. As the corrosion from the ambient laboratory air is not very reproducible from day to day and varies in strength from no corrosion to severe damage to the specimen, we designed a better controlled experiment by simulating the corrosion via a forced sulfur attack. This is done by storing the copper specimens after the ion beam milling and carbon deposition with a sulfur flake for 10 min in a gelatin capsule, which creates a sulfur enriched ambient. The effectiveness of the protection layer is shown on Fig. 1. The copper specimen has two areas where carbon was deposited on both sidewalls for 120 s and 60 s respectively. After the preparation, the specimen was stored for 10 min with sulfur and then immediately transferred to the TEM. This treatment resulted in complete corrosion of the uncapped areas, whereas the carbon protected regions remained in perfect condition for the 120 s deposition time. The 60 s deposition time shows little corrosion at the edge and thus does not offer a complete protection. The same carbon deposition conditions were used on the HfO 2 cones to measure the carbon layer thickness at the sidewall of the cones and to see the influence of an enduring beam interaction. The above mentioned deposition conditions lead to a thickness of 3/4/4.5 nm on the sidewall of the cones for a 60s/120s/180s deposition time, respectively (Fig. 3). A series of TEM images over 5 minutes did not show any increase of carbon layer thickness at the sample sidewall (Fig. 4). Henceforth the sample is suitable for enduring beam interactions in TEM mode. A 60s 25 % O 2 /75 % Ar plasma cleaning in a Fischione Instruments “Nanoclean 1070” system can remove the deposited carbon layer on the copper. It can be concluded that 4 nm of carbon can protect the surface sufficiently against sulfur attack from ambient laboratory air. The e‐beam deposition is done with low beam currents as used for SEM imaging during the specimen preparation and is suitable for e‐beam sensitive materials . Felix Seidel acknowledges the Institute for Promotion of Innovation by Science and Technology in Flanders (IWT) for his Ph.D. fellowship.