On July 1st, Toshiba Corporation's Semiconductor Company and Storage Products Company consolidated to form Semiconductor & Storage Products Company.This page describes reliability information of semiconductor products.

Estimating Failure Rates Using Accelerated Lifetime Tests

[As of April, 2011]

With the rapid technological advances in the electronics industry, accelerated lifetime testing has become increasingly important as a practical and efficient means of evaluating reliability. This is essential for bringing new products to market quickly and for ensuring semiconductor device reliability. Whereas described the accelerated lifetime test in detail, this chapter describes how to estimate field failure rates using the accelerated lifetime test.
When estimating field failure rates from test data, the actual number of failures is often very small or sometimes zero.
In such cases, the failure rate must be estimated based on a certain confidence level. With semiconductor components, the upper reliability limit is often used, assuming that the failure distribution is an exponential distribution. This method is specified in JIS C5003 (General rules for determining the failure rate of electronic components during tests) and uses the following equation:
Total test time: T (Number of test samples n × Test time t)
r : Number of failures

This is multiplied by a number (coefficient) from Table 1 according to the set confidence level (degree of data accuracy) and number of failures r.
When r = 0, the failure rate can be obtained using the statistical calculation that was used to obtain the values in Table 1. In this case:

The coefficient a is 0.92 for a confidence level of 60% and 2.30 for 90%.

Table 1 Failure Rate Reliability Limit Coefficients
Number of Failures (r)	Confidence Level		Number of Failures (r)	Confidence Level
Number of Failures (r)	60%	90%	Number of Failures (r)	60%	90%
1	2.02	3.89	6	1.22	1.76
2	1.55	2.66	7	1.20	1.68
3	1.39	2.23	8	1.18	1.62
4	1.31	2.00	9	1.16	1.58
5	1.26	1.85	10	1.15	1.54

The following describes how to calculate the failure rate based on a specific example.
Assume that 100 semiconductor devices are subjected to high-temperature testing (T_a = 125°C, at rated operating voltage) for 2,000h with zero faults.
To find the failure rate, first calculate the acceleration A_F (voltage acceleration coefficient A_V × temperature acceleration coefficient A_T) to obtain the total component hours.
The voltage acceleration coefficient A_V can be obtained from the failure rates for actual applied voltage and test voltage conditions. It is assumed that the actual usage voltage is within the rated specification.
Given that the actual usage temperature is 50°C and the typical activation energy of the expected failure mode is 0.8 eV, the temperature acceleration coefficient A_T is obtained using the Arrhenius equation as follows:

This is [the equation for the temperature acceleration coefficient].

Acceleration coefficient A_F = Voltage acceleration coefficient A_V × Temperature acceleration coefficient A_T
= 1 × 225
= 225
Given the number of samples n and test time tn for sample number n, the total test time (component hours) is:

This is [the equation for the total test time].

Because the number of failures r = 0, given a reliability level of 60%,

This is [the equation for the failure rate, λ].

Therefore, in this case the estimated failure rate in the field is 20 FIT.
In a similar manner, the failure rate of a device under actual usage conditions can be estimated from the accelerated test data provided that the major failure modes and failure mechanisms for the device are understood.