The Investigation Of Thermal Spectrum Analysis Method Of Microelectronic Devices

It has always been a scientific and technological puzzle in microelectronics field since several decades ago, using electrical method to detect peak temperature and non-uniform property & non-uniform degree of temperature distribution of microelectronic devices. In microelectronic devices, for PN junction barrier devices, Schottky barrier devices, field-effect devices, semiconductor light-emitting devices, et al, their uniform property of generating heat and temperature distribution is very important. And parameter stability, quality reliability of semiconductor devices, and life time of devices, IC chips, and complete machines & systems, all are crucially influenced by the property. To aviation and military field, the significance of the property is not fluctuated. For semiconductor devices, it is an important factor whether the temperature distribution is uniform or not, which influences the thermal performance, the electrical performance, the reliability and the life time. The study is more and more extensive in this field.Currently, thermal infrared method, simply signed as TIM is the most accurate method to detect temperature distribution of microelectronic devices, i.e. taking infrared images of chips. However, the method is generally used in reliability analysis of semi-finish devices. For the measurement of encapsulated finish devices, images must be taken after the devices are dissected. Thus, the measurement is destructive, even the measurement could not be implemented after destruction. Therefore, the thermal infrared method is not suitable to be a conventional measurement. Electrical method, used to measure both finish and semi-finish devices, is nondestructive and without damage in measurement to be extensively used as technology standard method. The electrical method, adopted in international IEC standard to detect devices temperature, is called standard electrical method, simply signed as SEM. In the standard method, uniform temperature distribution of device is hypothesized, and only single temperature is used to characterize temperature information of the whole device, which is called standard-electrical temperature. Due to good operability, convenient measurement, and simple comparison & estimation based on only one parameter, standard electrical method is always popular in the world. However, its demerits are obvious, i.e. that it obliterates and covers up the non-uniform property of temperature distribution, and easily causes misjudging and leaked judgment.Therefore, it is very significant and gets huge applied value to explore a new method to detect temperature distribution both like the thermal infrared method and without destruction like standard electrical method. From 1970s, D. L. Blackburn referred an electrical method to measure peak junction temperature, to 1990s, Lv Changzhi, Wang Mingzhu of Beijing University of Technology went on the research based on Blackburn's paper. In the period, there are many other scholars did the research in the field, but they could not went on deeply to gave up the research because of all kinds of limitations in their methods: In the middle of 1990s, Professor Miao Qinghai of Shandong University found the excessive thermotaxis effect of low current, then based on the effect, the sally port to the puzzle emerged. The research in this paper, just aiming to the international scientific and technological puzzle, is developed at the background. The thermal spectrum analysis method, represented in this paper, is just the method which uses complete electrical method to detect non-uniform property and non-uniform degree of temperature distribution of devices.The thermal spectrum analysis method, simply signed as TSAM, inherits the merits from electrical method, i.e. that the measurement is nondestructive and without damage, not only to semi-finish devices but also to finish devices. Also, it gets the merits like the thermal infrared method, i.e. that it could obtain information on peak temperature and non-uniform property & degree of devices. Further more, the thermal spectrum analysis method is belter than thermal infrared method in detecting temperature information. TIM can obtain the temperature distribution information of whole chip, but the reliability analysis and judgment based on the method is restrictedly only from sense and experience, and lacks qualitative and quantitative analysis, specially aiming at the active region of devices. However, TSAM calculates temperature distribution and gives temperature values and corresponding effective area, via the suitable mathematics & physics model, based on the measured parameters on the active region. It's the result due to qualitative and quantitative analysis specially aiming at the active region of devices. On the hand of devices reliability analysis and judgment, TSAM gets more advantages than TIM, much more advantages than SEM, because it provides more exact and reliable information.In the first chapter "introduction", it's introduced that the research background, the current research status and the purpose and significance of choosing subject, about detecting temperature and temperature distribution using complete electrical method. It's the first time to investigate and explore the physical significance of the junction temperature measured by SEM in the paper. Moreover, an original and special experiment method is used to verify the inequality relationship among several junction temperature values.In the second chapter "thermal spectrum curves", based on the thermal infrared image of devices, aiming at the active region, via qualitative and quantitative analysis, the temperature spectrum curves are obtained, and firstly named thermal spectrum curves via the analogy with optical spectrum. In thermal infrared images of devices, the temperature distribution information of the whole chip is given, however, the qualitative and quantitative analysis is absent. Except the peak temperature, there are not analysis result and directly applied data of the active region. Therefore, the thermal spectrum analysis for thermal infrared images software was coded independently based on colorimetric method, to be used to statistically obtain temperature and its distribution information of selected region. Via using the templates corresponding to different devices to select the active region, the selected region is qualitatively and quantitatively analyzed based on the colorimetric method to obtain statistically all temperature values and corresponding ratio each temperature value occupied. Thus, the spectrum curve and one-dimension temperature distribution curve are given from these data. The junction-temperature distribution of emitter region of transistors, peak junction temperature and minimum junction temperature are briefly shown and the average junction temperature is calculated through the one-dimension temperature distribution curves. The thermal spectrum curve of transistor is the novel method to character non-uniform property of junction temperature of transistor which is simpler and briefer than thermal infrared images. The thermal spectrum curves in this chapter are obtained via the optical method, i.e. that all temperature information and data come from the thermal infrared images, however, it provides experiment basis and direction for the complete electrical method, i.e. that the thermal spectrum curves from thermal infrared images should be the final result of the electrical method.In the 3^rd chapter "the excessive thermotaxis effect of low current", respectively based on the simulative and the actual temperature distribution, the effect was verified in this paper, via theoretical calculation and experiment data, which is one of the most essential theories. The excessive thermotaxis effect of low current was found in the middle of 1990s, and finally applied to judge non-uniform property of junction temperature through two measuring currents. However, the effect has not been verified roundly and completely. In this chapter, it's the first time to completely verify the effect based on the actual temperature distribution from theoretical calculation and experiment.Using the research method in the 2^nd chapter, the thermal spectrum curve is obtained from the actual thermal infrared image of the sample. Based on the sub-transistor parallel connection model which is created according to temperature and corresponding normalized area, the experiment had a good agreement with the theoretical calculation, therefore, the actual existence of the effect in PN junction barrier is verified completely. When transistors are dissipating power, the junction temperature distribution generally is non-uniform.Based on the sub-transistor parallel connection model, from theoretical calculation and experiment, in the simulative and the actual temperature distribution, it is investigated that on non-uniform junction temperature distribution, current density of high-temperature region is higher than that of low-temperature region; the ratio of current density of high-temperature to that of low-temperature, increases with measuring current decreasing. This phenomenon is called excessive thermotaxis and concentration effect of low current. Base on the characteristics, uniformity and uniform degree of junction temperature distribution can be studied.The 4^th chapter "the thermal spectrum analysis method of transistor" is the most important research in this thesis. It is investigated that the thermal spectrum analysis method based on the sub-transistor parallel connection model and the transistor isothermal circle model. The excessive thermotaxis effect of low current, MQH Rule and Ebers-Moll model are the basic theories of the arithmetic of the thermal spectrum analysis method. MQH Rule is the key of the arithmetic of TSAM, and verified from theory and experiment, which was established by Professor Miao Qing-Hai, Shandong University, in many years research working about TSAM. Ebers-Moll model is the key to simplify the complicated dual PN junction barrier to simple single junction barrier to be analyzed.The background data, I-V-T data of transistors, is investigated in the chapter, which is the most essential experiment data in the TSAM. In the high accurate constant temperature apparatus, the temperature sensitive parameters (TSP), corresponding to several staircase constant currents, are collected in high rate to obtain the I-V-T characteristic curve clusters. It is the most essential physical property and the most important thermotics character for transistors. The background data is the synthetical result which contains thermal conductivity, thermal capacitance, series residence, injection ratio, et al. Therefore, it has much higher degree of credence than that in D. L. Blackburn's method.In this chapter, it is introduced in detail that the basic arithmetic of TSAM and the more correct regression of zero-time temperature sensitive parameter. Through-combined experiment among three instruments: Model BJ2984 Transient Thermal Residence Testing Instrument based on the standard electrical method (SEM), RPY-1 Military Reliability Analysis Apparatus for semiconductor devices based on the thermal spectrum analysis method (TSAM), thermal infrared imager InfraScope II based on the thermal infrared method (TIM), the temperature measured on the three methods is compared. The combined experiment result indicates that with the power increasing, the result of SEM gets more and more large difference with result of TIM, i.e. that the result of SEM deviate the actual status more and more far, however, the result of TSAM (the peak temperature and the effective area) always has a good agreement with the result of TIM. The result of the combined experiment presented that the thermal spectrum analysis method can exactly and reliably measure peak temperature and non-uniform property & degree of temperature distribution. Relative to TIM and SEM, the thermal spectrum analysis method has a big ascendancy to replace the two methods finally to be the routine measuring method.In the 5^th chapter, real-time measurement and real-time thermal spectrum analysis method of transistors are investigated. It is the first time to solve authentically the real-time measurement method for semiconductor devices. Moreover, it is also the first time to apply TSAM into the real-time measurement of transistors to real time analyze non-uniform property & degree of temperature distribution. The method of real-time measuring junction temperature of transistors is that the heating current is directly treated as the measuring current to measure junction temperature as power being dissipated, without changing the power condition.In this chapter, the puzzle presented in thermal resistance measurement part of IEC 60747-7, the indeterminacy puzzle of temperature sensitive parameters corresponding to high currents, has been solved. In the real-time measuring process, the heating current is fleetly changed to staircase current in small amplitude, and the corresponding temperature parameters are collected to be analyzed in TSAM to obtain the non-uniform property & degree information of temperature distribution. The method can be applied in non-traumatic detecting chip weld quality and safety operation area (SOA) of semiconductor devices, and junction temperature controlling in steady-state lifetime experiment. The method still can be use in advancing thermal resistance measurement and apparatus, on-line testing, evaluating reliability and lifetime of apparatuses and systems, et al. It has very significant impact and application values for microelectronic devices.In the 6^th chapter, the thermal spectrum analysis method for temperature sensitive elements and devices is investigated. It is the first time to extend TSAM to temperature sensitive elements and devices field, such as temperature sensors, Shocktty barrier devices, field effect transistors, LED light-emitting devices, via the transverse research and exploration, based on the successful application of TSAM on transistors. In this chapter, the feasibility of TSAM is investigated, specially aiming at Platinum resistance in temperature sensors and Shocktty barrier devices in microelectronic devices. It's the first time to verify the physical significance of the single temperature value returned in each measurement. It's not the average temperature of measured region, and the significance is not definite now. Based on the sub-resistance series connection model, the feasibility of TSAM is verified via the theoretical deduce.For the Shocktty barrier devices in microelectronic devices, the excessive thermotaxis effect of low current in this kind of barrier was verified in this chapter from theoretical deduce. Thus, so long as that the appropriate model is created to be used in the TSAM, the method for the kind of devices should be successfully applied.It is the first time that the equivalent transistor model is represented in this thesis. The model has so good universal property and applicability that it provides a good guarantee for the application of TSAM extended into microelectronic devices field.The thermal spectrum analysis method can consumedly advance the resolution and accuracy of temperature sensors. Moreover, the method provides more abundant, more scientific, more reliable, and more accurate temperature distribution information of the active region for reliability analysis of semiconductor devices, with more convenient method.In the 7^th chapter, the scientific effort and achievements in the whole under graduate period are generalized. The foreground and trend of development in future are expected and evaluated in this chapter, and the research work proposal in future is presented finally.