Thomas Pfeifer1,Eric Hoglund2,1,Patrick Hopkins1
University of Virginia1,Oak Ridge National Laboratory2
Thomas Pfeifer1,Eric Hoglund2,1,Patrick Hopkins1
University of Virginia1,Oak Ridge National Laboratory2
The detection of surface or near-surface abnormalities, whether in the form of contamination, oxide growth or crystalline defects, is of interest across a wide variety of research areas. Similarly, common tools for sample preparation may create surface defects without the user realizing. For example, we found that Focused Ion Beam (FIB) milling, used for Transmission Electron Microscopy (TEM) sample preparation or for the generation of fiducial markers, may create a large ion-affected region surrounding the milled hole.<br/><br/>To investigate the detectability and spatial extent of these defects, we mill a series of holes in a silicon wafer (Using a Helios UC G4 FIB-SEM with gallium ions. holes are 10μm square, created under a variety of beam current / energy / exposure time conditions). We then use two pump-probe laser metrology tools (Time Domain (TDTR) and Steady State Thermoreflectance (SSTR)) with samples mounted on motorized stages to explore detection of the spatial extent of damage. We also use atomic-resolution Scanning Transmission Electron Microscopy (STEM) and Energy-dispersive X-ray spectroscopy (EDX) (using a Thermo Fisher Scientific Themis system at 200 kV) on samples prepared at varying distances from the FIB hole.<br/><br/>As the pump-probe experiments typically require a metal film to be deposited on all samples, we first use TDTR and SSTR to measure the thermal boundary resistance (TBR) between a metal film (80 nm aluminum, e-beam evaporated) and silicon, observing changes measurable up to 300 μm from the edge of the hole in the most severe case. Comparing this to the STEM results, an unusually high gallium concentration can be found between the aluminum and silicon at 200 μm away (more than seen in a reference TEM sample prepared in an identical manner). Similarly, a slightly disordered region is seen within the first ~ 10 nm of silicon, suggesting low-energy errant gallium ions may be responsible.<br/><br/>When the aluminum capping layer is excluded however, SSTR was capable of measuring a difference out to > 600 μm. This signal is less easily interpreted as a change in thermal properties, but a notable change is visible nonetheless. By comparison, the only visible changes to the sample occur no more than 50 μm beyond the edge of the hole. Two more TEM samples were made at 500 μm and 1 mm away from the hole, and no substantial differences were found.<br/><br/>These findings suggests that laser-based pump-probe experiments are equally if not more capable at detecting surface damage as compared to EDX or atomic-resolution STEM. These laser based experiments are also substantially cheaper and faster to operate, not requiring the use of electron microscopy or extensive sample preparation. Furthermore, this is the first observation of extreme spatial effects due to FIB milling to our knowledge, and the precise nature of this damage warrants further study.