JMD, << insulating films >>
As you may know, zirconium, has been widely used in jewelery as
a diamond look-alike. It has good qualities as an insulator and
is also very hard. Titanium, also a very hard matal, probably has
equally good properties when oxidized to become an insulator.
Silicon dioxide, very compatiable to chips on silicon wafers,
has a very disappointing dialectric coeffecient. I suspect the
article disclosed revolves around building better capacitors, the
key component in translating analog information into digital codes.
The dielectric coeffecient is directly relative to the size of
a capacitor. Literally thousands are used on analog to digital chips.
Often we used deposited nitride, about 2.5 times better than silicon
dioxide, but creates more processing steps due to the nature in
some etching steps. (the removal of unwanted material after exposing
the wafer at photo-lithogophry.-- DPMI ring a bell here?)
The article says it will be 7 times better suggesting to me that
somewhere near 1/7 of the area on a chip will be needed to build
a capacitor compared to silicon dioxide. Capacitors can take up
50% of a chips area depending on the chip function. Especially
when flash conversions are made, a key factor in moving modems
from 28k to 56k in speed. Chip size was a tradeoff, increasing processing cost. As chip size goes down, chips per wafer
go up, significantly reducing the cost of the chip. Given the same
yield ( good chips vs possible good chips) an increase of 50% in chip
count reduces the cost of the chip by 50%. It cost the same to
process an entire wafer regardless of chip count, all other factors
remaining constant. Hope this helps. Now if the innovation is about
interlevel insulation, this is another entire discussion, but I suspect it's both. Better interlevel isolation is critical to
reducing unwanted parasitics, very critical with the reduction of
actual transistor size. In short, the circuits can run faster.
Larry ps -- numerous mispellings. Deal with them. Back to stocks. |
