Page 145 - NEW Armstrong Book - 2

The SiC Production Process
he intermittent operating life (IOL) test per- T formed in accordance with AEC-Q101 according to MIL-STD-750 Test Method 1037 aims to investigate the reliability of devices when they are subjected to thermo-
mechanical stress (stress phase).
The test consists of applying power to the device (DC) for a necessary time (TON) to reach a case jump ΔTCASE AMBIENT of about 100˚C with a tolerance of 15˚C/–10˚C. These operat- ing conditions will result in an increase in the junction tem- perature (TJ), followed by a shutdown period (TOFF) necessary for the junction to cool.
This sequence is repeated for a specific number of cycles, and the test duration (IOL) is set in accordance with MIL-STD-750 Test Method 1037. The test duration can be set by the user. Any readouts are managed by the operator, in a completely automatic way, by programming their duration in advance.
Eles developed a methodology and IP hardware for reli- ability testing on discrete MOS in
SiC/GaN/Si technology. The
approach integrates Eles’s ART
system and its firmware/software development platform to perform the IOL test in an automatic way in terms of controllability, reachabil- ity, and observability of the system device under test (DUT).
The test is performed on reliability cards with a high parallelism, thus providing the ability to stress and monitor up to N DUTs per board. The number of devices that can be allocated depends on the mechani- cal dimensions of any sockets/con- nectors used to house the device itself.
The methodology allows the device to reach, starting from an
By Antonio Zaffarami
initial state T /T /T , a desired stress state AMBIENT CASE(MIN) CASE(MAX)
ΔTCASE AMBIENT/ISOURCE (controllability) by checking its state upon reaching it (reachability) and monitoring, during the “on” phase, its output quantities (observability).
The thermomechanical stress is induced by electrical stress, and the electrical quantities are monitored during the exe- cution of the test itself (runtime), overcoming the limits that this type of test previously presented. Traditionally, a blind stress was performed, without any control over or observ- ability of the progress of the test itself, and with the electrical and characterization measurements requested only on dedi- cated post-cycle stress testbenches.
The solution for the IOL test allows a “sample size” of N devices per board to define:
TON, TOFF, VGATE, VDRAIN
and observe/measure runtime for each test run cycle:
131