HBM Trailing Pulse Current Leakage
Summary
One of the unwanted stresses that have been discovered when using some HBM testers is current leakage occurring after an HBM pulse, which is often called a Trailing Pulse. For very high input impedance pins and this small current occurs long enough to cause a voltage ramp that can cause unintended damage such as forcing charge into the CMOS gate dielectric producing a shift the threshold voltages of input transistors.[i]
The JEDEC/ESDA joint standard JS-001 describes a method to measure the leakage level using a 10 kΩ resistor test DUT and suggests a safe level of 1 µA/kV of HBM stress.[ii]
The design of GTS HBM test system avoids this and other tester parasitics and anomalies.[iii]
Detail of HBM tester operations
A basic HBM tester schematic from JS-001 Fig. 1 can be used to understand the cause of the trailing pulse stress.
The HBM capacitor C1 is charged, before the pulse, to the HV Supply output voltage through a current limiting resistor R1 and the deactivated relay S1. When the HBM relay S1 is activated, its armature (the moving part) moves from its resting normally closed contact, as shown in this figure, toward the normal open contact. A spark occurs been the moving armature and the normally open contact while the armature is still approaching the normally open contact, generating the HBM pulse. The entire HBM pulse current is conducted by this spark. The relay spark lasts until the current flow decreases to a level that the conducting spark is no long maintained. When the current flow is extinguished, the HBM capacitor C1 is typically not fully discharged, and may have about ten percent of the starting HBM voltage remaining. This voltage is not usually high enough to produce another spark as the armature continues its motion. When the motion is completed and the armature touches the normally open contact, the remaining energy in the capacitor is discharged. This produces a secondary pulse which is described in the standards as normally less than 15% of the HBM pulse.
A common type of relay used in older HBM testers is a mercury reed relay. When the spark occurs in a mercury wetted relay, the mercury on the contacts is sputtered into a conductive cloud by the high-energy spark. This results in a cloud of mercury atoms and ions being produced. This conductive cloud is available to conduct for up to hundreds of microseconds after the HBM pulse. During that period until the mercury cloud dissipates the high voltage at the normally closed contact can produce a current sourced from the HV supply through R1. While S1 remains in a conductive state between the normally closed and normally open contacts, there is a small, almost constant current flow to the DUT. If the pin under test is of a very high impedance, this small current will cause a voltage ramp at the pin under test that can reach hazardous levels.
Testing for Trailing Current
The test for after pulse leakage uses a relatively high value resistor in place of the DUT and measures the voltage across it to determine the tester current flow. JS-001 Annex B1 Fig. 5 shows the test load of a parallel 10 kΩ resistor and Zener of 6 to 15 V.
A high impedance oscilloscope probe is used and a Zener diode is added to protect the scope probe from the high HBM voltage. The Zener diode is protection for the 10x scope probe and scope input channel. The scope probe is connected on the Terminal A side of the 10k and the scope probe ground connection is the Terminal B side of this resistor. The scope can be triggered from the voltage prober channel or by using an inductive current probe (CT1 or CT2) in series with the 10k.
What is measured on the scope is: first, the HBM pulse going off scale, then decaying to zero; second, typically the secondary pulse from the relay closing; and finally, a trailing pulse after the secondary pulse if it exists. Since a trailing pulse leakage is a relatively constant current, it will be measured as voltage plateau until all the mercury vapor dissipates and the leakage current stops. While there is no formal after pulse leakage specification, 1 uA/kV is a recommended maximum value. This is equivalent to R1 >1 GΩ , the current limiting resistance between the HV supply and the relay S1. Using the recommended 10 kΩ resistor for a leakage current-to-voltage transducer, the after pulse current should not produce more than 40 mV/4kV (4kV is the recommend test voltage).
Conclusion
When testing devices with high impedance I/O pins, a tester with significant trailing pulse currents can cause false failures. The advanced design of GTS HBM test equipment does not use mercury relays, and will not damage devices under test with after pulse leakage or other unwanted stresses.
[1] Charvaka Duvvury, et. al., "Gate Oxide Failures Due to Anomalous Stress from HBM ESD Testers", EOS/ESD Assoc. Symposium Proceedings, 2004, 3A.1, pp132- 140.
[1] ANSI/ESDA/JEDEC JS-001-2014 - Joint Standard for Electrostatic Discharge Sensitivity Testing – Human Body Model (HBM) – Component Level, ANNEX B.1 – HBM, Test Equipment Parasitic Properties; Optional Trailing Pulse Detection Equipment / Apparatus. Available from www.esda.org or www.jedec.org.
[1] Evan Grund, et. al., “Methods to Remove Anomalies from Human Body Model Pulse Generators”, EOS/ESD Assoc. Symposium Proceedings, 2006, 5B.5, pp 335-342.