Temperature Mapping of Localized Hot Spots on Microelectronic Chip Surfaces

Abstract

This paper describes the use of digital image processing in conjunction with an infrared imaging apparatus to locate and quantify “micro” hot spot temperatures on the surface of energized microelectronic chips. Briefly, the temperature mapping/processing procedure creates emissivity maps for the surface of the chip at different isothermal conditions. The emissivity map images are digitized and stored as a 512 × 512 pixel array, of which 400 lines contain IR information. Apparent temperature measurements are then collected with the chip energized in its normal operating environment. These apparent temperature data are digitized and stored as a 512 × 512 integer array using the same format as the digitized emissivity data. Before correcting for emissivity variations, the apparent temperature images are rectified using digital image processing to precisely overlay the spatial coordinates of the emissivity map. Finally, actual temperature maps are obtained by correcting the apparent temperature data for the local emissivity variations and background reflections. The computer driven measurement technique has been applied to the task of measuring localized temperatures on areas as small as 30 μm on the surface of an energized chip to an accuracy of ±1°C once the surface emissivity is accurately known. The infrared equipment, image processing hardware and supporting software are used to measure the temperature distribution on the surface of a 4.7 mm × 4.7 mm energized chip. IR measured temperatures at isolated locations on the chip are compared with results obtained by the resistance-temperature technique. Since the resistance-temperature technique provides an area-averaged temperature for the energized region, the result obtained from the high resolution IR measurements yields higher localized temperatures. Results are presented for peak surface temperatures up to 100°C and maximum heat flux values of 7.9x106 W/m2. A separate set of infrared measurements are used to predict the influence of surface emissivity on the accuracy of the temperature measurements.