A brief discussion on automotive-grade power module testing

In recent years, after new energy vehicles (starting with hybrid vehicles) have been accepted by capital, government and consumers, power semiconductor devices, as the core of power conversion, have also become the darling of the capital market.

Especially after Tesla launched SiC power devices. From Infineon, Mitsubishi, ON Semiconductor to STMicroelectronics, CRRC Times Electric has trained many heroes for many years, and each has shown its capabilities in the current automotive-grade power semiconductor device field. To a certain extent, we want to borrow Infineon's Hybridpack module packaging form to enter the standardized electric drive application field;

The other part takes a different approach, introducing double-sided cooling design and epoxy resin packaging processes, hoping to reach the pinnacle of technology and achieve the goal of reducing the reliability and performance of the module.

The device reliability test standard has become the only indicator to check whether the above products are suitable for use.

What are the automotive-grade reliability testing standards?

There are currently two reliability testing standards for power semiconductor devices that are recognized by customers, namely AQG-324 and AEC-Q101, which represent the testing requirements for modules and single-tube devices respectively.

As for other reliability testing standards proposed by domestic units, groups and other institutions, they are basically derived from the contents of the above two standards.

When customers ask about the reliability verification standards corresponding to the device, I generally recommend AQG-324 because it strictly limits the scope of application of the standard in the preamble: "This document defines requirements, test conditions and tests for validating properties, including the lifetime of power electronics modules and equivalent special designs based on discrete devices, for use in power electronics converter units (PCUs) of motor vehicles up to 3.5 t gross vehicle weight."

This also confirms the concept that I have always promoted, that the "test standards for automotive-grade power semiconductor devices" use "car" as the subject and "semiconductor devices" as the means.

Any changes in chips (application of SiC or GaN) and packaging structures and materials (using double-sided cooling structure, etc.) can only add test items or replace test methods, but cannot change the system's requirements for equipment reliability.

What does the AQG-324 standard say?

The AQG-324 standard includes four types of tests: module test (QM), characteristic test (QC), environmental test (QE) and life test (QL).

The test items in the standard simulate various types of stress that the device is subjected to in actual use, age the weak points of the device package and chip, and thus achieve accelerated verification of the actual service life of the device.

Module testing (QM) mainly tests the static electrical parameters, interconnection layers and appearance defects of the equipment to ensure that the equipment meets the basic prerequisite requirements for the next test and the functional indicators after the test.

Characteristic testing (QC) is used to verify the dynamic electrical characteristics, short circuit characteristics, thermal resistance and insulation withstand voltage characteristics of the equipment to evaluate the quality of the equipment product.

Environmental testing (QE) verifies the environmental adaptability of power devices in vehicles by applying temperature change conditions and mechanical vibration/shock.

The 2021 version of QE-01 Thermal Shock (TST) no longer requires a two-chamber/basket-type temperature shock test solution. A one-chamber temperature cycle solution or a three-chamber temperature shock solution is considered equivalent in terms of effect. Its basic testing requirements are listed in the table below.

If the customer has concerns about the performance of the product in the ambient temperature test (mostly plastic-encapsulated devices), I would recommend that they refer to "QC/T 1136-2020 Environmental Test Requirements and Test Methods for Insulated Gate Bipolar Transistor Modules for Electric Vehicles".

In this standard, temperature shock and temperature cycle tests are included in the assessment indicators.

In the QE-01 Thermal Shock (TST) device that I operate, a thermocouple is attached to the bottom of the device (or inside the module heat sink) to detect whether the device temperature meets the basic requirement of tdwell>15min during the temperature change time of the temperature chamber.

Since some customers strongly require that the number of high and low temperature shock cycles be completed within one hour, the high/low limit temperature setting of the box will be much higher than the highest or lowest storage temperature of the equipment, which can easily lead to damage to the equipment during 1000 cycles. Therefore, the author recommends that customers should rather extend the constant temperature time rather than increase the box temperature to meet the experimental requirements in TST experiments.

In the life test (QL), the failure of the device under extreme temperature (high temperature/low temperature), long-term withstand voltage (HTRB/HTGB/H3TRB) and multiple switching cycles (power cycling) are considered. Among them, the power cycling test (power cycling) is the most difficult test item in the AQG-324 standard and has the most bizarre judgment requirements.

What is power cycling?

The principle of power cycling is very simple: introduce current to heat the chip, then cool the chip when the current is turned off, and so on. The uneven temperature distribution in the device combined with the different physical properties of the device packaging materials (such as the coefficient of thermal expansion CTE) causes the interconnect structure in the device to gradually age, eventually leading to device failure.

The figure below is an example of current conduction and chip temperature changes during a power cycling test.

The figure below shows the temperature distribution inside the device and the deformation of the device during power cycling test.

During the power cycling test, the device thermal resistance parameters and forward conduction voltage are collected to determine whether the device is damaged. During the test, if a sudden increase in the forward conduction voltage of the device is found, it indicates that the bonding wire inside the device has fallen off or broken. If the thermal resistance parameter increases, it indicates that the device is damaged in the heat dissipation path.

The failure judgment in the power cycling test is shown below, which includes multiple failure modes such as bonding wire failure and solder layer failure.

Power cycling is a must-have item in all types of test standards. However, the biggest difference between other standards and AQG-324 is that AQG-324 does not specify the life requirements for power cycling testing. The AEC-Q101 standard requirements are as follows:

Among them, the AEC-Q101 standard is the one that I least recommend as a power cycle test judgment indicator, because when the standard was established, power semiconductor devices had not yet been used in the main drive part of new energy vehicles.

China's automotive industry standards establish test indicators for second-level and minute-level power cycles, with the minute-level test drawing on the test requirements of AEC-Q101.

The AQG-324 standard does not use a specific number of cycles as a criterion for power cycling, but rather requires verification of the lifetime model proposed by the product under different time and temperature parameters.

Most customers are unfamiliar with the establishment of life models, and have no experience in comparing life models between different products. Therefore, the setting of power cycle test schemes has become the focus of the AQG-324 standard.