Mehrdad Nikoonahad

Defect detection algorithm for wafer inspection based on laser scanning

A defect detection algorithm for wafer inspection based on laser scanning is presented. Microscopic anomalies, contaminants, and process induced pattern defects result in a two-dimensional (2-D) laser scattering signature, which closely resembles the coherent point-spread-function of the scanning laser beam. This point-spread-function is a 2-D Gaussian in the majority of cases and can be characterized by four parameters. The algorithm fits Gaussian surfaces to sampled data points. Events are accepted or rejected on the basis of how similar the Gaussian parameters are to that of the point-spread-function, known a priori. It is shown that the algorithm achieves a 95% capture for submicron particles and pattern defects on typical logic and array wafer regions. Results demonstrating the algorithms performance relative to mechanical and electronic noise and to signal resolution are presented.

Picosecond photoacoustics using common-path interferometry

A pump-probe technique based on the detection of optical phase changes at the surface is described. A common-path interferometer in which the reference and probe pulses interrogate the surface immediately (∼500 ps) before and after the pump pulse is demonstrated. The primary application of this system is film thickness measurement in integrated circuit processing. Pulses 60 to 100 fs wide, from a Ti-sapphire laser at 800 nm, are brought to a 5 μm focus on the surface, resulting in picosecond acoustic pulses in the film. Echoes from film interfaces, upon arrival at the surface, lead to a phase change in the probe beam which is measured. Results from tungsten, aluminum, and copper film structures are presented.

New laser scanning techniques for wafer inspection

A laser scanning system designed for inspection of patterned wafers is described. This system addresses the inspection needs for 64 Mb (0.35 micrometers ) and 256 Mb (0.25 micrometers ) DRAM process technologies. The system is capable of detecting contaminant particles and planar pattern defects on memory and logic devices. The throughput of the system is designed for 30 wafers (200 mm in diameter) per hour. The beam at 488 nm is brought to a focal spot and is scanned on the wafer surface using an acousto-optic deflector (AOD). The entire wafer is scanned under oblique illumination in narrow strips in a serpentine fashion. The specular beam is collected and processed in, what we have named, the autoposition sensor (APS) to servo- lock the height position of the wafer during the scan. The system utilizes multiple independent collection channels positioned around the scan line and it is possible to select the polarization of the collected light for enhanced signal-to-background ratio. The engineering tradeoffs for realizing a system with high throughput and sensitivity are formulated and discussed. Calculations ilustrating scattering from submicron size particles under various polarization conditions are shown. These results lead to optimum design for collection optics. The APS channel is described and illustrated by results indicating that it is possible to keep the surface height of the wafer constant to within 0.4 micrometers in the presence of large changes in topography and wafer reflectivity. Results obtained from a range of production wafers demonstrating detection of 0.1 micrometers anomalies on bare wafer, 0.3 micrometers on memory devices, and 0.4 micrometers on random logic structures are presented.

In-situ height correction for laser scanning of semiconductor wafers

An in situ technique for servo control of surface height during laser scanning of semiconductor wafers is described. The scheme corrects any macroscopic height changes due to tilt and bow in the wafer and rejects local and pattern-dependent changes of height and reflectivity. The waist of the scanning beam is imaged on a slit aperture placed in front of a position-sensitive photodiode, leading to an ac signal at the scanning frequency. This ac signal then undergoes synchronous detection using a reference signal at the scanning frequency. This detection scheme leads to a reduced sensitivity to low-frequency electronic and thermal drifts. Normalization circuitry provides means for excluding the effects of reflectivity which can vary over four orders of magnitude on patterned wafers. The height signal, so obtained, is used to drive a PZT stage to a nominal height position in closed loop. On patterned wafers, an rms height accuracy of better than 0.1 μm has been achieved in 20-Hz bandwidth.

Laser-induced fluorescence spectroscopy and imaging of semiconductor wafers

Fluorescence spectra of selected films used in microelectronic fabrication have been recorded. We have used a 0.125-m focal length spectrophotometer and a 400-line/mm grating resulting in 4.2-nm spectral resolution. The optical setup employs a laser at 364 nm for excitation and a dark-field collection configuratiow-a geometry that we routinely use for laser scanning for inspection purposes. A simple, though thorough, analysis and methodology for the removal of the system spectral response is presented. Results show that films used in microelectronic fabrication, in general, yield a broadband fluorescence spectrum under 364-nm excitation. Further, a scanning system that bases the image contrast on laser-induced fluorescence from the wafer surface is described and demonstrated. It is shown that this is a particularly useful inspection/review modality when the wafer is at poly/metal process level and the contaminant is a fall-on or residue of an organic material.

Extendibility of laser scanning tools for advanced wafer inspection

This paper describes a broad range of design issues that influence the performance of optical equipment for in-line inspection of random (logic) and repetitive (memory) patterns. In particular, we describe the angular distribution of signals from defects on a patterned wafer illuminated by a focused optical beam. We analyze the configuration of both illumination and collection optics to maximize the signal to background ratio for the detection of submicron defects on pattern. In addition, we analyze the distribution of the scattered light as a function of pattern periodicity and orientation with respect to the illuminating beam. The advantages of polarization selection and spatial filtering techniques are explored to enhance the detection sensitivity on repetitive and random pattern wafers. From these results we have developed a new patterned wafer inspection system that offers increased sensitivity and improved defect capture.