Electron beam drilling rates of silicon crystal measured on various accelerating voltages and probe currents

Abstract

The design rule of the semiconductor device has been miniaturized to be 22/20 nm or 16/14 nm recently. When we analyze a structure of these miniaturized devices by scanning transmission electron microscopy (STEM), it is necessary to prepare a lamella with a thickness less than 20 nm. And the thin lamella is analyzed by energy dispersive x‐ray spectroscopy (EDS) and/or electron energy‐loss spectroscopy (EELS), which require longer dwell time than that of imaging due to small signal intensity. This lamella may receive the electron beam damage if a STEM probe is not optimized. The damage could be categorized to be two types, one is structure deformation of sample and the other is etching and/or migration of sample atoms, resulting in beam drilling of sample [1]. The latter is crucial for elemental analysis, since it significantly affect to results of quantitative analysis. Such phenomena were proposed, since the field emission STEM was commercialized [2,3]. In this paper, we report beam drilling of Si crystal depending on electron probe currents and accelerating voltages. We used a field emission electron microscope (JEOL, JEM‐2800) equipped with a X‐ray detection system, which include two large‐sized silicon drift detectors (dual SDD, 100 mm 2 ). A lamella of a silicon device was prepared with Ar ion milling (JEOL, Ion Slicer). The thickness of the lamella was measured to be approximately 15 nm by the EELS ratio method. To measure electron beam etching rate, we measured a decay of Si X‐ray (Kα) count rates using a point analysis mode. Figure 1 shows the decay profiles on Si count rates for various probe currents at 200 kV. The decay rate ( R ) is express as R = R 0 *exp(‐ at ), where R 0 is initial decay rate, exp(‐ a ) is decay coefficient at a certain probe current and an accelerating voltage, and t is elapsed time. Thus, we estimate the decay coefficient by fitting the decay profiles. Si X‐ray count rates are decreased with the decay coefficients of 22.74 %/sec, 16.72 %/sec, 8.41 %/sec and 3.03 %/sec for probe currents of 1.50 nA, 0.96 nA, 0.50 nA and 0.22 nA. The measured decay coefficients are approximately proportional to probe current, through the electron densities under these probes are approximately constant to be 2.0 nA/nm 2 . We need to care to use a large probe current for analysis or imaging with long dwell time, since this result implies that the drilling rate is promoted when we uses larger current. Therefore, it is preferable to analyze with small current as possible for less damage of sample at 200 kV. Figure 2 plots decay of Si count rates for various accelerating voltages with probe current of 3.72 nA. The decay rates for 200 kV, 100 kV and 60 kV are measured to be 32.3 %/sec, 0.26 %/sec and 0.0 %/sec. The rates are not proportional to accelerating voltage. The drilling rate on accelerating voltage is related to threshold energy of sputtering of Si atoms. In addition to the low sputtering probability, the ionization cross section increases at lower energy, resulting in higher count rate at lower voltage. The initial count rates at 3.72 nA for 200, 100 and 60 kV were measured to be 12.7 kcounts/sec, 18.3 kcounts /sec, 29.0 kcounts /sec. In conclusion, we found out that elemental analysis in the low accelerating voltage is very effective for reduction of the electron beam damage as well as higher sensitivity due to larger ionization cross section of an element.