Graphics Script Provides Quick Classification of GaAs Wafers

By Dr. Millard G. Mier, Research Physicist
U.S. Air Force Research Laboratory
Wright Patterson Air Force Base, Ohio

Infrared transmission topography has long been used to detect variations in EL2 density that can cause dark-line defects that limit lifetime of lasers and other problems. In the past, a program was used to measure infrared transmission, calculate absorption across the surface of the wafer and produce color coded plots that allow the wafer's characteristics to be determined at a glance. Due to memory limitations, accuracy was lower than desired and the program ran on VAX/VMS computers that are being taken out of service due to obsolescence. To overcome these problems, the author developed a script using a state-of-the-art data analysis program that removes the memory limitation problem, runs on inexpensive personal computers and, as a bonus, produces bit-map plots that can be cut and pasted into Windows word processing and presentation software.

Polished wafers of semi-insulating undoped GaAs or doped conducting GaAs are important for the manufacturer of semiconductors that operate at very high frequencies. The advantage of GaAs is that it is capable of operating at 5 to 10 times the maximum frequency level of silicon circuits. These devices are currently used in three major consumer markets: wireless including PCS and cellular, fiber optic communications, and television including cable and direct broadcast satellite TV. The EL2 defect is an abnormal crystalline state that is not yet well understood but which can cause problems in GaAs, particularly if it is concentrated in one area of the wafer.

Wafer scanning
Wright Laboratories has developed an automated method of accurately measuring the density of the EL2 defect at all locations across a GaAs wafer. The wafer is mechanically scanned past a beam from a tungsten-halogen light source. The light source focused through a monochromater that eliminates all but 1.1 wavelength micron light that is absorbed by the EL2 defect and an electromechanical chopper that limits the light to a 0.5 mm square spot on the wafer. Infrared light passing through the sample is detected by a germanium diode detector operating in the low-noise zero-bias mode. A commercial amplifier digitizes the detector output and stores it in a file along with on-wafer coordinates. Measurement of the 16,597 locations required to map a three inch wafer takes about an hour. For 100 mm wafers, the resolution is typically reduced to 0.6 mm and 19,287 locations are measured.

The author has found that ranking the amplitude data into fourteen bins and assigning a color to each location, then plotting that color at each location where the measured value corresponds to a bin range provides an easily interpreted colormap of the measured values keyed to the values location on the wafer. This method provides an excellent method of investigating relatively obscure correlations between materials properties and device properties. The plotted dataset can easily be compared to the properties of semiconductors produced with the wafers such as Hall-effect measured free carrier density, source-drain resistance, gated source-drain saturation current, pinchoff voltage and associated gate voltage. Visual inspection of the colormaps quickly reveals any rough correlations and more detailed mathematical correlations can be carried out as desired.