<<So being an ex semiconductor employee in your opinion is LSI a good company - technically i.e., R&D wise? Is LSI considered a has-been or a gonna-be?>>
Let's see, what do I know about LSI? Regarding their process technology, my impressions are favorable. It is not easy for a company of their size to be on the leading edge like they are. Mostly it is the giants like Intel, Motorola, TI, and AMD who are able to run their own fabs with the leading edge processes. I don't have any idea of what percentage of their products are manufactured with the leading edge processes or what their yields are like. Does anyone know what process technology the Gresham fab is supposed to run (e.g. 8", 0.35 micron)? The second thing that I like about LSI is that they seem to have gotten the infrastructure where they can run the large designs (>100k, 200k gates). My understanding from EDA vendors was that they had gone to full-scan designs, which greatly improves design for testability. Also, I had a co-worker who had worked in the quality organization at LSI who seemed to be impressed with them. These are some of the reasons I have invested in LSI. I am not as familiar with their market position, where I have tended to rely more on comments on this thread as well as the company's reports for information.
Warning: The remainder of this post is an OT physics discussion, read at your own peril:
<<Perhaps I quoted Mr. Gordon Moore incorrectly. My subscription to Red Herring has lapsed so I took no more than a quick peak at the issue at the magazine stand and based what I wrote on memory. (My guessing now that you made me think about it!: At around 0.1 microns with an atomic radius at about 1 angstrom or .0001 microns, when the line-widths go to 0.1u we are "just" a 1000 times bigger than the size of an atom? That's awfully close to the absolute limit of the atom's size aren't we? Tough to get things to work and be separate and act "predictably" when there could be quantum mechanical "interference" from the atoms when they are so close together?)>>
I would not be surprised if you quoted Mr. Moore correctly.
The 0.1 micron dimension you refer to is the width of the gate of the transistor. Alternatively it can be seen as the length of the channel between the source and the drain. I've included a crude little ASCII representation of a cross-sectional view of a transistor at the bottom of my message, in hopes it will make what I am trying to say at least as clear as mud. (Sorry if it the dots make it look strange, I did my best with what SI offers.) The way a MOS field-effect transistor works is that it has three connections: the source, gate, and drain. The voltage on the gate controls whether or not there is a connection between the source and drain. The gate acts as a switch turning off or on the channel connecting the other two terminals of the transistor. You are right that at 0.1u we are about 1000 times bigger than an atom. But that is not as close as you might think. According my scratch paper calculations, the wavelength of a free electron here is roughly 10 angstroms. We won't see quantum effects until our structures start getting on the order of this dimension. Below the wavelength of the electrons, we would start seeing even more quantum effects as we got close to the spacing of the silicon atoms in the crystal lattice (~3 angstroms) and the size of the atoms themselves (~1 angstrom). That means we have to get at least 20 to 50 times smaller before we start seeing electrons tunnel from the source to the drain in significant amounts because linewidths are getting too small. However, there is a different place that quantum mechanics will start to affect us sooner. That is in electrons tunneling between the gate and the channel. Between the two is a thin gate oxide as an insulator. Right now the thickness is about 80-100 angstroms (0.008-0.01 microns) in leading-edge stuff. That is a lot closer to the wavelength of the electrons and hence a lot closer to being affected by quantum mechanics. The gate oxide thickness is scaled down as the other dimensions of the transistor are scaled down. But we still have some breathing room there, too, before we run into quantum problems.
MOS Field-Effect Transistor (Cross-Section):
. . . . . . . . . . . . ___________
. . . . . . . . . . . . |......................|
. . . . . . . . . . . . |.......Gate........|
. . . . . . . . . . . . |___________|
----------------|---------------|----------------
. .|.....Source.....|. . .channel. . . |......Drain.....|
. .\__________/ . . . . . . . . . . . \_________/
.
.
.
.
. . . . . . . . . . Silicon substrate
.
----------------------------------------------
.
G.P. |
