The APT met proposal consisted of several work packages, each covering a specific area underpinning an atom probe measurement.
WP1 focused on sample preparation. A key objective of this WP was to improve the sample preparation for ATP of III V and III nitride materials. As the region of interest (ROI) layer can be just nanometers below the sample surface, any protective capping layer employed for the needle sample shape preparation must not damage the underlying sample structure. The typical approach of using ion assisted platinum deposition (iPt) was found to etch our materials. A new method combining an initial electron beam deposited Pt (ePt) layer with a subsequent iPt layer was shown to be extremely successful. This solution meant that the benefits offered by a FIB SEM to locate and cover small regions was retained. Additionally, high failure rates associated with poor adhesion to the sample carrier using iPt were reduced by reconfiguring the sample/carrier geometry and resulted in a significant measurement yield.
For WP2, the impact of the laser power/energy and wavelength combined with electric field to stimulate surface atom emissions were studied. The objective was to establish the link between electric field (at evaporation), local and temporal heating from the laser-tip interaction, and stoichiometric emission and/or detection of the constituents of the compound. A significant finding was that III V semiconductor materials not only emit singular ions but cluster ions too, with the single/cluster ion ratio being dependent on the experimental conditions. Cluster ions have a strong impact on quantification since their formation leads to mass spectrum peak overlaps for the same element but with a different charge state. Unambiguous assignment and accounting becomes impossible unless the species of interest has multiple isotopes and correction procedures based on expected abundancies can be developed. For single isotope compounds, stoichiometric analysis cannot be readily achieved while experimental conditions leading to a complete absence of clusters was never found for all the III V materials tested. Based on our extensive exploration of experimental conditions, operational parameters for an accurate stoichiometric analysis of GaN was determined. An important conclusion of the data analysis is that the apparent composition is predominantly influenced by the electrical field at the tip apex with a correlation virtually invariant between different instruments and independent of the laser conditions (different wavelengths, laser power,..).
In WP3 the shape of the tip was investigated to evaluate its impact on the detection probabilities and composition analysis. For higher laser energy/power, significant variations in the III V stoichiometry across the tip was observed which indicates a spatially varying tip shape and varying electric field. The tip shape arises from a localized absorption (and heat diffusion) within the tip apex, causing spatially variant heating and locally varying evaporation probabilities. Since the detailed tip shapes are a result of the laser-tip interaction, they do have a wavelength and power/energy dependence. As the detailed tip controls the local electric field, the emission probability of single ions/clusters and the concentration quantification is affected. The non hemispherical shape invariably impacts on the ion emission trajectories with deleterious consequences on spatial resolution. Lower laser energies/powers resulted in a more homogenous tip emission, minimizing such inaccuracies. For GaN, the materials crystallographic nature was observed in the form of poles and zone lines. These features are associated with low field regions leading to nanoscale field variations over the tip surface. Investigation suggested a moderate dependence on their presence to the ion trajectories and spatial resolution attainable.
WP4 and WP5 focused on data quantification and reconstruction of real device structures, building upon the understanding learnt from WP1, WP2 and WP3. A GaN/AlGaN/GaN device structure with a ROI layer <70 nm below the surface was prepared, measured, and analyzed. The objective was to establish a correlation between variations in composition and device performance. Repeat measurements exploring APT reproducibility were also conducted. Analysis of the measured data was performed using the latest software and the sensitivity of the reconstruction parameters studied. Accurate 3D reconstructions were obtained while nanometer scale voxel analysis yielded information about the 3D atomic distributions not currently achievable using alternative metrologies.
Data and results from the APT-Met project have been presented at International workshops/conferences as well as to IMEC stakeholders through the biannual industrial partner weeks, while journal publications are in progress.
