Electroquasistatic sensors for surface and subsurface nano-imaging of integrated circuit features

Abstract

The following thesis relates to the design, simulation, and testing of electroquasistatic (EQS) sensors to be used for feature/defect location and imaging. The aim of this thesis is to launch an investigation into the use of EQS sensor arrays for non-destructive evaluation and quality control purposes for integrated circuit industry applications. Research into three specific areas serves as the primary focus of the thesis: 1.) The use of EQS sensors to penetrate the surface of doped silicon and locate p-n junctions and doped wells. Arrays of coplanar EQS sensors are scanned laterally over the surface of a doped silicon bulk at a fixed scan height. Electric fields from the driven EQS sensor array are capable of penetrating the surface of the semiconductor when sensors are operated at a frequency comparable to its charge relaxation break frequency. It is demonstrated through finite element method (FEM) simulations that voltage-driven EQS electrodes can couple into the p-n junction without making any direct electrical contact with the semiconducting bulk. A new methodology for locating p-n junctions is presented where the currents on these voltage-driven sensors are monitored for harmonic distortion due to the junction's nonlinear drift/diffusion carrier dynamics. With sensors located over a p-n junction at a scan height of 200 nm and driven at 1GHz, the ratio of the second harmonic current to the fundamental current on a sensor is shown to exceed 9%. Such an IC imaging technique could prove to be useful for verification and detection of fabrication errors, externally monitoring current flows, as well as detecting hidden Trojan circuits that might be present. 2.) The use of EQS sensors to locate and image surface features and contaminant objects on photomasks. Motivation for research into this area comes from the desire to be able to locate and remove contaminant particles that might be present on extreme ultraviolet lithography (EUVL) photomasks used in the mass production of next-generation integrated circuits. FEM simulation results demonstrate the sensitivity of EQS sensor arrays in detecting various contaminant particles located in a 100nm wide by 70nm deep gap in the absorber layer of an EUVL photomask. A millimeter-scale, in-lab experiment using capacitive sensors is performed with sensors and materials having similar aspect ratios and electrical properties to those simulated. Experiments demonstrate both the capabilities and limitations of sensors in detecting various objects located in a trenches milled out of aluminum. Additionally, a discussion of the need for low-noise pickup circuitry to interface with sensors is presented. 3.) An investigation into the inversion of sensor transimpedance response signals into predicted feature/defect profiles. In this case study, an inverse electromagnetic sensor problem is solved by training a radial basis function artificial neural network (RBF-ANN) to accurately approximate the forward mapping of the physical dimensions (width and depth) of a high aspect ratio trench in doped silicon into a sensor's transimpedance response as the sensor array scans past the trench at a fixed scan height. This is an example of the type of inverse problem that might be encountered in an EQS array microscope and one possible approach to its solution. The function-approximation network is then inserted into an iterative signal inversion routine which converges to a prediction for the trench's dimensions, given a measured transimpedance response. The routine is capable of predicting trench dimensions to within 1% of their actual value. In all three cases, research involves extensive finite element method (FEM) simulations of sensor performance using COMSOL Multiphysics.