Interaction of femtosecond laser pulses with nanoscaleSi-tips for atom probe tomography

Sammanfattning: The atom probe is an analysis technique based on the emission of ionized species from a needle-shaped sample (apex radius < 100 nm) under the influence of a very strong electric field ( 10-50 V/nm). A DC-voltage is applied on the sample in order to generate a field slightly below the one necessary to remove atoms (in the form of ions) from its surface. An ultrashort (femtosecond) laser pulse is used to trigger the emission. The evaporated ions are accelerated in the electric field and projected onto a position sensitive detector where a magnified image of the surface is formed (magnication from 10^6 to 10^7). Time of flight mass spectrometry is used to chemically identify the evaporated atoms. The technique thus allows to analyze the composition of a 3D volume with sub-nanometer resolution. Imec conducts research in order to introduce the 3D characterization with quasi-atomic resolution capabilities of the technique to the semiconductor industry. It became quickly apparent that a detailed understanding of the laser interaction with the nanoscale samples is crucial in order to interpret the analysis results. In this work, we briefly introduce the principles of the technique and review some of its applications. We then summarize some of the currently unexplained experimental observations, taken from the literature or from experiments conducted at Imec. Based on those observations, we introduce a thermally assisted model of field evaporation that includes the electromagnetic nature of light and the semiconducting character of silicon. The optical absorption of the nanoscale sample is computed by numerical simulations using the FDTD algorithm. The temperature evolution at the tip apex is obtained by solving a coupled thermal conduction-carrier recombinations problem and the shape of the mass spectrum is deduced. We discuss the model and confront its results to experimental data. We show that the model qualitatively explains many experimental aspects of the characterization of silicon by means of an atom probe analysis. Nevertheless, we show that at this stage the model lacks quantitative accuracy and we suggest several ways to improve it.