CN-121997886-A - Distributed temperature real-time monitoring method for integrated chip micro-module

CN121997886ACN 121997886 ACN121997886 ACN 121997886ACN-121997886-A

Abstract

The invention discloses a method for monitoring the distributed temperature of an integrated chip micro-module in real time, which belongs to the field of semiconductor device degradation monitoring and comprises the following steps of extracting the distributed temperature map of the integrated chip micro-module; modeling by mapping the actual working temperature with the micro-module test vector, and realizing the method for monitoring the distributed temperature of the integrated chip and early warning the power consumption based on ATE. The invention has no invasiveness to the integrated chip, has no influence on the performance and time sequence of the micro-module of the integrated chip, measures the distributed temperature value of the device which is difficult to obtain or is obtained by the invasive temperature sensor through a test vector, obtains the temperature and transient power consumption value of the device by using an artificial intelligent model, effectively solves the problems of interference, additional cost or power consumption of an additional time sequence path caused by the design of the integrated chip by the invasive temperature sensor, and can feed back the heat distribution condition of the micro-module in the integrated chip in real time on line and finally measure the current degradation level of the device.

Inventors

JI WEIWEI
CHEN YUXUAN
SONG GUODONG
Shou Kaiyuan
Shao Jiakang

Assignees

中国电子科技集团公司第五十八研究所

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (8)

1. The distributed temperature real-time monitoring method for the integrated chip micro-module is characterized by comprising the following steps of: Extracting a distributed temperature map of an integrated chip micro module; Step two, modeling of mapping actual working temperature through a micro-module test vector; and thirdly, realizing a method for monitoring the distributed temperature of the integrated chip and early warning the power consumption based on the ATE.
2. The method for monitoring the distributed temperature of the integrated chip micro-module according to claim 1, wherein the step of extracting the distributed temperature map of the integrated chip micro-module comprises the steps of generating a three-dimensional temperature characteristic map through thermal distribution simulation software when the integrated chip runs a specific micro-module to input excitation, and extracting the distributed temperature map corresponding to the micro-module test vector by using python or other scripting languages.
3. The method for monitoring the distributed temperature of the integrated chip micro-module in real time according to claim 1, wherein the modeling mode from the micro-module test vector to the actual working temperature of the device is a multi-target optimization modeling method, input data are divided into two parts, namely a micro-module type and a test vector corresponding to the micro-module, and the input data are output as distributed temperature values of each module or region of the device.
4. The integrated chip micro-module distributed temperature real-time monitoring method of claim 3, wherein the micro-module category comprises a processor block, a memory block, a transmission interface block and a signal conversion block.
5. The method for monitoring distributed temperature of integrated chip micro-modules in real time according to claim 3, wherein the flip frequency of 0 and 1 in the test vector corresponding to the micro-modules, the number of micro-modules activated at the same time and the interconnection state also need to be considered in modeling.
6. The integrated chip micro-module distributed temperature real-time monitoring method according to claim 1, wherein the integrated chip is connected with an automatic test system through a device to be tested test board, and test vectors are input into corresponding micro-modules through the automatic test system according to specified pins and routing paths, so that device specific module performance and parameters are tested.
7. The integrated chip micro-module distributed temperature real-time monitoring method according to claim 6, wherein the performance and parameters of the specific module of the device are tested, and meanwhile, input variables are collected and uploaded to an upper computer through a serial port, and the distributed temperature and the transient power consumption of the device are estimated in real time according to the established model.
8. The integrated chip micro-module distributed temperature real-time monitoring method according to claim 7, wherein the real-time estimated distributed temperature and the transient power consumption value are substituted into a historical data set, and the current reliability state of the device is judged to realize power consumption early warning.

Description

Distributed temperature real-time monitoring method for integrated chip micro-module Technical Field The invention relates to the technical field of semiconductor device degradation monitoring, in particular to a real-time monitoring method for integrated chip micro-module distributed temperature. Background With the continuous increase of the complexity of integrated chips, the superposition of inherent deviation brought by nanoscale technology and severe working environment accelerates the aging process of chips, wherein HCI (hot carrier injection), NBTI (negative bias temperature instability) and TDDB (time dielectric breakdown) are core causes of transistor aging, NBTI becomes a dominant factor under advanced technology, and the degradation degree of the NBTI is closely related to parameters such as voltage, temperature, signal Probability (SP) and the like. Monitoring the degradation level and life prediction of an integrated chip have become key to guaranteeing the reliability of a system, but the prior art still has a plurality of limitations. The traditional monitoring method mainly contradicts the precision and invasiveness trade-off, namely the failure prediction based on the aging characterization circuit can only realize basic early warning, has poor stability and is easy to misjudge, and the failure sign monitoring method can improve the precision, but needs to go deep into the chip to set detection nodes, has obvious interference on the performance of the chip, and is especially not suitable for complex structures such as multichannel digital intelligent computing chips. The most used model-based prediction method belongs to the mainstream, but the method is also in the bottleneck that a non-life loss model relies on ageing characteristic parameters such as path delay, threshold voltage and the like as input, the process of acquiring the parameters and failure symptom monitoring have the same invasive problem, the life loss model is mostly constructed by adopting machine learning, and although the life loss model can be cut in from an ageing source, a special detection circuit is required to be implanted for directly detecting key input parameters such as SP and the like, the performance of a chip is obviously influenced, and the power consumption burden is brought. Burn-in testing of advanced package level devices also faces the problems of high cost and long cycle time, and Chiplet architecture makes it difficult to cover the complete logic and interconnect degradation conditions for post-package testing. The aging nature of the chip is a time-cumulative function of temperature, voltage, workload, etc., where temperature is strongly related to the NBTI effect and directly determines the internal node pressure state. Dynamic regulation of thermal management techniques causes voltage and load to fluctuate dramatically over time, relying only on static physical parameters or pessimistic condition hypothesis modeling, severely degrading prediction accuracy. The research shows that the real-time heat distribution monitoring is a core foundation for realizing active heat management and avoiding thermal runaway and accelerated aging, and directly determines the reliability and the service life evaluation accuracy of the integrated chip. However, the distributed temperature monitoring in the integrated chip by the built-in temperature sensor is difficult to break through the limit brought by the power consumption wall and the temperature wall, and a rapid and accurate distributed temperature real-time monitoring method for the micro-modules of the integrated chip is needed, so that the complete logic and interconnection degradation condition evaluation of the integrated chip is realized under the guarantee of non-invasive measurement. Disclosure of Invention The invention aims to provide a real-time monitoring method for distributed temperature of an integrated chip micro-module, which aims to solve the problems that the traditional method for obtaining the distributed temperature value of a device is difficult to obtain or the distributed temperature value of the device is obtained through an invasive temperature sensor and the like, meanwhile, the distributed temperature condition of the integrated chip micro-module can be fed back on line in real time, and finally, the real-time monitoring of the power consumption and the degradation level estimation of the integrated chip are realized. In order to solve the technical problems, the invention provides a distributed temperature real-time monitoring method for an integrated chip micro-module, which comprises the following steps: Extracting a distributed temperature map of an integrated chip micro module; Step two, modeling of mapping actual working temperature through a micro-module test vector; and thirdly, realizing a method for monitoring the distributed temperature of the integrated chip and early warning the power consumption based on the ATE.