They exist and are made by the likes of IXYS, Semikron, IR, maybe Fuji and Toshiba, plenty others too. Try Google.

MOSFETs parallel nicely so you can use multiple smaller devices to behave like one large one. Large FETs used to be made in this way, paralleling several smaler dice inside one package. Watch the gate capacitance - the peak drive current can be quite high. Be sure you can get current to and from the device - PCB traces become awkward at this level although the device manufacturers claim they can get the current in and out of the package.

Would you consider an IGBT instead? The losses are likely to be lower because the saturation voltage is likely to be lower than the ohmic drop through the FET channel.
 

----------------------------------
  Sometimes I only open my mouth to swap feet...

Thanks ScottyUK,
I haven't considered any IGBT yet, but it sounds a very good choice. 120Amps is defintely a high current, I don`t know if the package can handle that top cuerrent for many cycles. I think the most important thing that I should consider in this case is the heat dissipation, since I only have a small place available.
Once again, thanks for your help!

There are lots of Mosfets that claim astounding currents, the who problem is that these currents are often the product of major specsmanship games.  They represent only switched values that last short periods.  Steady state currents are going to be lower. There was a recent thread almost exactly your question.  I'll look for it.

Keith Cress
Flamin Systems, Inc.- http://www.flaminsystems.com

Ah yes.

For example:

Id(on) 150A (note 2)

note 2: Measured for 300us at a 2% duty cycle sort of thing...

Yes! Exactly.

BTW I spent a ridiculuos amount of time searching for that thread of about a month and a half ago and could not find it.  Sorry.

Generally you don't want more than about 50A trying to make it into a standard TO-220 package as the leads will be a major loss all by themselves.  I'd stay below that if I wanted the product to be reliable.  You also don't just slap these type designs together.  You start with an iterative method: power->heatsink->package->trace widths->layers->number of devices in parallel->back to the beginning.  Think about the connectors to bring this type of power onto the board!  Oh and if you're smart you'll bring fans to the par-tee.

Keith Cress
Flamin Systems, Inc.- http://www.flaminsystems.com

I'm not sure I'd term it specsmanship really.

It's more about being able to make a repeatable measurement of a parameter without something melting or searing the operator's fingers...

Unfortunately it also means that you can't directly transfer the datasheet values to your average application... not many applications apply the power for 300us at a 2% duty cycle in reality.

Thanks for the replys. In fact I thought that split the current in two IGBT`s was better, even for circuit protections. In fact I need this High current values for a short period of time (Time enough to heat up some Plugs).
Thanks for your help!

Sounds like the prepackaged IGBT modules are the best fit then.  No circuit board layouts required.

Keith Cress
Flamin Systems, Inc.- http://www.flaminsystems.com

IGBTs don't parallel particularly well - use one big one with a semiconductor fuse upstream. 120A isn't that big for an IGBT.
 

----------------------------------
  Sometimes I only open my mouth to swap feet...