Why does a MOSFET or IGBT's surge current rating stop increasing with decreasing pulse width beyond a point, unlike thyristors?

Question

Thyristor

The surge current rating of a thyristor increases as the width of the pulse decreases. For a single non-repetitive pulse, it's not uncommon for a thyristor to have a surge rating that is 10x or even 100x higher than its average or RMS current rating.

For example, a TYN640 thyristor has an average current rating of 25 A for a half-cycle sinusoidal current. However, when the width of this current is decreased to just 10 milliseconds, the absolute maximum surge rating increases to 480 A. For a 1-millisecond pulse, 1500 A. And for a 200-microsecond pulse, 2500 A.

Eventually, the surge rating starts decreasing again after shorter pulses start hitting the di/dt limitation of the thyristor. But in general, the shorter the pulse, the higher the surge current. This property makes thyristors extremely useful in pulse and surge generators.

MOSFET

However, the surge ratings of MOSFETs do not have this property.

For nearly all MOSFETs, the pulsed drain current is given as a single number in the datasheet, and it's not a function of the pulse width. Instead, "pulse width tp limited by Tjmax". In other words, the pulse width is determined by temperature rise, but it must always be lower than the fixed pulsed rating.

As far as I'm aware of, MOSFETs are rated this way regardless of their voltage rating, Rds(on), packaging, silicon or SiC. But I'm open to correction if anyone can find an example to the contrary.

For example, using the IPB65R050CFD7A MOSFET, although the 211 A pulsed rating is significantly higher than the 45 A DC rating (with an infinite heat sink), but it cannot be increased further by reducing the pulse width.

The Safe Operating Area shows that 200 A is the upper limit under any circumstances.

IGBT

The surge current ratings of IGBTs are similar to MOSFETs. Using the STGW75M65DF2 as an example - its DC current rating is 120 A, but the pulsed rating is only 225 A, and again, "pulse width limited by maximum junction temperature".

As far as I'm aware of, nearly all IGBTs are rated this way regardless of their voltage rating or packaging.

Again, the Safe Operating Area shows the surge current is maxed out around 200 A.

Question

I believe localized hotspot formation in the silicon set the ultimate limitation of surge current rating of all semiconductor switches, as heating under a short pulse is an adiabatic condition. I know this is the physical origin of the di/dt limitation of thyristor, and I guess it's also why MOSFETs and IGBTs have a maximum pulsed current rating.

But why does the hotspot create only a di/dt limitation in thyristors, but a pulsed current limitation in MOSFETs and IGBTs? Why does thyristor's surge current rating keep increasing with decreasing pulse width, but MOSFET or IGBT's surge current rating stop increasing with decreasing pulse width beyond a point? What is the basis in semiconductor physics behind setting a static pulsed rating regardless of its width for MOSFETs and IGBTs?

Answer 1

Notice the left edge of the transistor curve: it is slanted because the device simply cannot draw more current at those (low) voltages. It's not a matter of capacity, but sheer ability. The same may be true of the top (horizontal segment) of the curve as well.

Now, there can be other limitations -- especially of very high current types, the bondwires or component leads can burn up in this region. There may be other physical limitations too, like electromigration limiting the current particularly over long periods of time (seconds to decades), and at high temperatures. (Some of these are also time-dependent, so you might expect to see the rating change with pulse width. This may or may not be included, though.)

But it can also be the case that drain/collector current simply can't be any higher than that, due to the chip itself having maximum conductivity over its whole [active] area.

Current saturation*, I think depends on the type. I'm not real familiar with high-Vgs operation of older MOSFET types, but I have tested newer generation SuperJunction types (these characteristically have extremely low C_oss and especially C_rss, above V_DS ~ 20-50V) and found that some simply don't carry any extra current for V_GS > 7V or thereabouts.

*In the sense of, current saturates, it cannot go any higher, no matter what you do with V_GS. Alternately, y_fs drops towards zero for I_D ~ I_DM.

The same applies to IGBTs, though current flow has the added factor that there is a bipolar enhancement effect taking place, which not only multiplies the MOSFET-mode current flow, but can have additional positive tempco (read: tendency to run away) because h_FE generally increases with temperature.

In either case, even if I_D keeps increasing with V_GS, you can set an effective, if crude, limit by simply setting V_GS(on) to some reasonable value -- this is relevant for IGBTs, at least of the older generation modules I last worked with, where, in combination with a desat detector, the fault current needs to be limited so that the desat protection circuit is able to shut down operation in time to avoid destruction.

A final note: notice again the horizontal region is not the low-voltage sloped region. It's above the point where R_DS(on) is limiting current flow. It's high voltage and high current, very high power. The die heats up rapidly in this region.

In contrast, thyristors operate wholly on minority carriers (bipolar effect), the voltage drop is small (a couple volts), and essentially all the bulk participates. It's like having a nearly unlimited "R_DS(on) limited" region: put more current into it and it just keeps getting more and more conductive -- the voltage drop stays low the whole time! So the power dissipation isn't very high, and the surge capacity is just immense.

But, do take note of the dI/dt limitation: they can handle a lot of current, but it can't start up all at once. It takes time for the current to spread out over the die, which is to say, the rate of change must be limited to avoid overheating any spot. This is different from MOSFETs and IGBTs, which can commutate in hundreds of nanoseconds, or less. That's kind of your analogous limitation: trading unlimited current for limited (above some minimum) time, versus limited current for unlimited (no minimum) time.