The thermal characteristics analysis, prediction and simulation of IC packaging

• Main Authors: Kun-Fu Tseng, 曾昆福
• Other Authors: Luke,Su Lu
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/53680662136856157333

博士 === 中正理工學院 === 國防科學研究所 === 86 === Abstract The purpose of this dissertation is to analyze the thermal characteristics and to predict the steady state temperature of Integrated Circuit (IC) packages. To obtain the thermal characteristics of IC packages, two sets of test chips were designed: one was for thermal resistance measurements which is conformed to the Semiconductor Equipment and Material International (SEMI) standard, and the other was for the temperature distribution measurements inside the chips. To obtain the temperature profile of the second set of chips, twelve temperature sensors were distributed at different locations on the chips. For temperature prediction, a thermal R-C (resistance-capacitance) model was proposed to predict the steady state temperature from transient state. To accurately predict the steady state temperature with this new model, accurate RC parameters were needed. To this end, two methodologies are developed in this dissertation: the first methodology uses an approximated RC value which was obtained from transient data; and the second methodology employed curve fitting to extract the RC values from transient data. In the first method, a heating pulse was applied to the heater on the test chips and the RC values were obtained from the sampled temperatures during power off stages. After comparing experimental data from the conventional methodologies, the steady state temperature can be fast obtained by the new methodology without losing accuracy if the test chips were in oil bathed and forced air convection environments. By the second methodology, the heating power was continuously applied to the heater on the chips from beginning to the steady state and temperatures were measured every few seconds. Then, necessary parameters can be extracted from curve fitting processes so that the predict steady state temperature were obtained with only one sampled datum (OSD) or two sampled data (TSD). It is concluded that the new methodologies, which makes the over-temperature protection function more reliable, efficient, sensible and faster, can be used as an assistance for over-temperature protection of a circuit or a system.