12
\$\begingroup\$

Background:

I am using a Si7456CDP N-channel MOSFET in a switching power supply. The power supply and load are housed in a plastic enclosure. Yesterday, the power supply and load were working perfectly. This morning, when I came to power it on, nothing worked. No power. Eventually, I discovered that the MOSFET's source and drain had shorted together. Replacing the MOSFET fixed the problem.

Question:

What could cause an N-channel MOSFET to suddenly fail with a source-drain short?

\$\endgroup\$
5
  • 3
    \$\begingroup\$ Bad karma? Seriously, you have enough experience here to know that this a bad question. How is the MOSFET being used in the circuit? Where's the schematic diagram? \$\endgroup\$ – Dave Tweed Mar 27 '13 at 15:08
  • 4
    \$\begingroup\$ While this form of question might be bad in general, in this specific type of device there's one class of failure that should be an automatic suspicion. \$\endgroup\$ – Chris Stratton Mar 27 '13 at 15:12
  • 5
    \$\begingroup\$ @DaveTweed - No, the point is to keep the question general and useful to more people than just myself. There must be a limited number of ways in which MOSFETs fail with this condition. The particulars of my circuit should not be relevant. \$\endgroup\$ – Rocketmagnet Mar 27 '13 at 15:15
  • 1
    \$\begingroup\$ @ChrisStratton - And that is ... ? \$\endgroup\$ – Rocketmagnet Mar 27 '13 at 15:18
  • 2
    \$\begingroup\$ As an initial cause, failure of the gate oxide leading to the device turning itself itself halfway on, at which point other fun things may happen. \$\endgroup\$ – Chris Stratton Mar 27 '13 at 15:18
11
\$\begingroup\$

There are two principal mechanisms but first a diagram:

enter image description here

Body and source are tied together, and several features are removed for simplicity.

Scenario 1:

  • Over voltage spike on Drain, causing filaments and the contacts and drain implants to be spiked. IT may or may not have causes the contacts to fail/melt but very high currents can cause breakdown of the D/B junction. Once the junction is spiked it is connected to the well drain and source are now shorted. This only requires breakdown at one location in the transistors

Scenario 2:

  • HIgh voltage on drain, causing EOS (Electrical Over stress) on the GOX (Gate Oxide) particularly on the gate nearest the drain. Very likely this is a LDMOS structure with an extended drain structure (which means the gate voltage need not get to the same voltage as the drain ever). Breakdown at that end of the gate can cause the gate to short to drain. Once it is shorted, it now is essentially on always, but also, the gate now gets driven to levels it wasn't intended to be at and the failure runs away. THis still only requires one fault in the transistor.

There are other scenarios but they all require two faults.

This device is fairly large and will be visible under a microscope. De-lidding this might be instructive.

\$\endgroup\$
5
\$\begingroup\$

That is actually a MOSFET. Drain-source shorts are the usual failure mode in MOSFETs and are usually caused by transients on the gate.

\$\endgroup\$
1
  • \$\begingroup\$ Thanks Leon. That's what I was wondering, could a transient at the gate or something cause a S-D short. \$\endgroup\$ – Rocketmagnet Mar 27 '13 at 15:17
3
\$\begingroup\$

Anything that damages the die could lead to a drain-source short. (Sometimes the die blows itself to smithereens.)

This includes:

  • Excessive over/under voltage at the gate
  • Poor / improper gate drive causing thermal runaway
  • Thermal runaway in general (loss of cooling / forced air)
  • Avalanche-induced EOS

Without more specific application information, it's hard to judge which mode could be the culprit.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.