What tests are required to ensure the reliability of IGBT products?

Infineon / Mitsubishi / Fuji / Semikron / Eupec / IXYS

What tests are required to ensure the reliability of IGBT products?

Posted Date: 2024-01-29

In today's semiconductor market, two important factors for a company's success are product quality and reliability. The two are interrelated, and reliability is reflected in long-term quality performance over the expected life of the product. For any manufacturer to remain in business, it must ensure that its products meet or exceed basic quality and reliability standards. As a semiconductor supplier, ON Semiconductor provides products that can operate in harsh environments for demanding applications and achieve high quality and reliability. Previously we shared how to conduct reliability testing on IGBTs. Today we will introduce how to ensure the product reliability of IGBTs through reliability audit procedures.

ON Semiconductor ensures reliability through strict implementation of reliability audit procedures. All IGBT products are divided into different series based on process technology and packaging type. Each quarter, stock from these series is sampled during final testing and then submitted for audit testing. By subjecting each product to real-time extreme stress testing, some process anomalies may be discovered that can be detected by process control. Typical reliability audit tests include high temperature reverse bias, high temperature gate bias, intermittent operating life, temperature cycling and autoclave testing. In order to discover any hidden failure modes, reliability testing is designed to test conditions beyond those of conventional quality and reliability testing.

Detected faults are sent to the product analysis laboratory for real-time evaluation. This highly specialized laboratory is equipped with a variety of analytical capabilities, including electrical characterization, wet chemistry and plasma techniques, metallurgical cross-sections, scanning electron microscopy, dispersive X-ray, Auger spectroscopy and micro/macro photography. Together, these capabilities allow for rapid and accurate analysis of failure mechanisms, ensuring that the results of the assessment can be translated into corrective actions and directed to the appropriate areas of responsibility.

ON Semiconductor's Reliability Audit Program provides a powerful method to detect potential signs of process anomalies in the IGBT product line. It is this rigorous and continuous focus on reliability audits that provides a good guarantee for achieving customer satisfaction.

IGBT Reliability Audit Procedure

Reliability Points

Semiconductor users are most concerned about the relationship between device performance and time. Once the suitability of a particular device has been determined, the device's effectiveness depends on the length of time it can provide trouble-free service. The reliability of a device reflects how well it will serve its customers. Reliability can be redefined as the probability of failure-free performance for a given period of time under a given manufacturer's specifications. Typically, semiconductor failure rates exhibit what is known as a "bathtub curve" over a long period of time.

Figure 3. Semiconductor failure rate

Reliability Mechanics Analysis

Since reliability assessment usually involves only a subset of the population of devices, the concept of the central limit theorem applies and the failure rate can be calculated using the λ2 distribution via the following formula:

Confidence limits are the degree of conservatism required in a calculation. The central limit theorem states that the values ​​of any unit sample in a large population will produce a normal distribution. The 50% confidence limit is called the best estimate and is the mean of this distribution. The 90% confidence limit is a very conservative value that results in a higher λ, indicating the point at which 90% of the distribution lies to the left of this value.

(2r + 2) is called the degree of freedom and represents the number of defective products in a form suitable for the λ2 table. The number of nonconforming items is a critical factor because the definition of nonconforming items often varies from manufacturer to manufacturer. Since there is an increasing likelihood that the test is not representative of the entire population as sample size and test time decrease, the λ2 calculation can produce surprisingly high λ values ​​during short test periods even though the true long-term failure rate may be very low. . To do this, relatively large amounts of data must be collected to demonstrate true long-term failure rates. Because this required years of testing thousands of devices, accelerated testing methods were developed.

Figure 4. Confidence limits and distribution of sample failure rates

Years of testing of semiconductor devices have shown that temperature accelerates the onset of failure, and this behavior is of the form Arrhenius' formula:



#tests #required #ensure #reliability #IGBT #products