Stressed silicon

Yes ?:-)

...Jim Thompson

Perhaps i can help you address the issue yourself. It is less an issue of atomic radii than crystal lattice pitch. Temperature by itself, and especially in gradients will aggravate the problems. Also consider more smoothly variegated Si/Ge boundaries for less abrupt stress.

Sorry, not the best question ever posed. Is it a big deal in terms of electron mobility (high-frequency performance) and how ubiquitous is it?

The real benefit of SiGe is fT... try fT > 35GHz if you really want a nice plaything... makes single-chip RF systems a piece-a-cake ;-)

...Jim Thompson

SiGe transistors are fairly common nowadays, and moderately low cost. They also have other advantages like generally higher Beta (100-400 instead of typically 30-100 for RF Si transistors), and often very good 1/f noise characteristics. Good choice for VHF or higher oscillators. Generally, real world SiGe transistors have fairly low breakdown voltage. SiGe transistors are practical in low noise, multi-octave wide bandwidth preamps with moderately high IP3 (e.g. 20-30dBm.)

NEC has some good examples of SiGe components, but so does Infineon. There is an interesting component from Infineon that is SiGe:C, with generally higher yet beta and even lower noise. (

Googling for SiGe leads to "some claims" of f sub T almost an order of magnitude beyond what you posted. My confusion is that the electron mobility enhancement resulting from the stretched silicon layer does not explain what strained silicon (SiGe) seems to be claiming these days? I sense some over hyping here.

Anyway, thanks for your help.

Guys in my building have working 65 GHz single-chip transceivers, and are working on things above 90 GHz.

Cheers,

Phil Hobbs

Thanks Phil! I am trying to understand the mechanism (as Curly Joe said ... I'm trying to think, but nothing is happening!).

Sirenza has some nice SiGe wideband MMICs, very low noise, great for stuff like MCP delay-line detector preamps and other exotica. Some of them, shockingly, actually have 50 ohm input impedances!

John

I'm not really a solid-state guy, but I think it's just that changing the interatomic spacing changes the band structure, which flattens out the bands near zero momentum, which means lower effective mass and higher mobility. The strain is just the fractional difference between the lattice constants of strained silicon and bulk silicon.

This would of course be easier if we could put more choices in the periodic table, but that takes *way* too much standards work to be feasible on a commercial timescale.

Cheers,

Phil Hobbs

Yeah, quantum mechanics won't be applied science until it can predict the periodic table from first principles. I am one who believes that time is nearing. I am also one who lost a chunk of my retirement on Enron! So much for me and my acumen.

Thanks again.

Yep, you are right. For some reason, I left the fT out of the list. I believe that I was listing those 'features' that are often overlooked, and forgot about the 'up front' advantage. The high beta (and usually low rbb) adds up to some very interesting advantages -- incl good 1/f performance for oscillators. (That isn't to claim that there aren't other possible problems for oscillators.)

Another 'advantage' of current technology SiGe transistors (over and above TYPICAL FETS) is that they can have a fantastic noise match for 50ohms along with a good return loss at the same time. Of course, some FETS also have good noise match for 50ohms. (For BJTs, often a 5-20ma Ic produces a 'best' noise match at 50ohms.) On the other hand, a good noise match for an HP PHEMT f54143 might be at 20-60ma, perhaps providing a slightly better IP3 or output power than a noise matched SiGe BJT. (I am not sure if the P1dB is definitely better, but would initially guess so because of the higher current.)

John