Loadstar 242/242.d81/t.mos4

Text File | 2022-08-26 | 18KB | 574 lines

u
            MOS Technology

                 from
             On The Edge:
       The Spectacular Rise and
          Fall of Commodore
           by Brian Bagnall

               Part IV

    With the mask complete, mass
fabrication of the microchips could
begin. Fabrication occurred in an
alien-like environment on the second
floor of the MOS Technology building
called the clean rooms. These
hermetically sealed rooms produced a
nearly dust free environment. The
precautions were extreme, since a
single grain of dust during the
etching process could cause a
miniature short circuit.

    To enter the clean rooms, lab
technicians were required to don
hairnets, beard nets, moustache
guards, gloves, paper booties, and
white jumpsuits. "It makes you look
like a bunny," says Peddle. "We used a
lot of them." As a final measure, the
technicians walked over a sticky-mat
to remove the last traces of dust
before stepping into the airlock.

    Within a crimson-tinted darkroom,
technicians replicated print after
print of the 6502 circuit. They coated
the round silicon wafers with thin
layers of metallic substances. After
each layer, technicians placed the
wafer into a special machine that
copied the circuit from the metal mask
to the surface of the silicon wafer.
Electrons flowed through the mask,
causing a thin layer to harden in the
shape of the circuit. Each wafer had
the chip pattern imprinted
approximately fifty times.

    In another room, bathed in yellow
light, technicians developed the
microchips. This process was almost
exactly like developing a photograph.
A studious technician carefully washed
each wafer with chemical solutions
that removed all but the hardened
circuitry. The industrial strength
solvents went by names like
trichloroethene, trichloroethane,
dichloroethene, dichloroethane, and
vinyl chloride.

    The technicians repeated the
process six times for the six layers,
each time using a different set of
chemicals and metallic substances.
According to Peddle, "You put this
mask on the device and do whatever
step you are going to do, and then you
take it off, put another mask on, and
do another step."

    The top layer was aluminum, which
was the best conductor. Beneath the
aluminum were various semiconductors
such as Germanium. Each layer went by
a different name, such as diffusion
layer, buried contact, depletion
layer, polysilicon, poly-metal
contact, and metal. With all six
layers applied, the wafers entered an
oven to bond the circuitry.

    Technicians then added a
passivation layer(6) to protect the
fragile metallic circuitry from
oxidation. After applying the
passivation layer, a machine sliced
the wafers into individual chips, each
smaller than a fingernail.

    The chemical etching process used
dangerous industrial solvents.
Inevitably, the solvents evaporated
into the air, which worried some of
the staff. Robert Russell, an early
Commodore employee, chuckles about the
general indifference regarding this
threat. "MOS had a little cafeteria at
the back alongside the production
line," he explains. "They had a
chemical release in the production
line that turned all your blueprints
that were hanging on the walls
different colors. You would come in
and they would all be yellow or green.
You kind of hoped that wasn't
happening when you were breathing it."

    "The production of semiconductors
produces all kinds of nasty
byproducts," says engineer Bob Yannes.
Inevitably, accidents occurred. "I
remember things happening like
occasionally we'd have a gas leak in
the front end and you'd have people
walking through the building saying,
'Hurry! Get out of the building!'"

    Most people were ignorant of the
dangers posed by the semiconductor
industry. "This is a time in history
when everybody looked at the clean
rooms and the guys all wearing their
bunny suits, and how sterile it was,
and everybody wanted a semiconductor
company in their hometown," says
Peddle. "It was high tech, big money,
and clean as opposed to a foundry or
something like that. What they didn't
realize was these guys were dealing
with the most poisonous, noxious stuff
in the world, and they had to put it
somewhere."

    The semiconductor industry was
still new in 1970 when John Pavinen
and his partners created MOS
Technology. "Nobody in the
semiconductor industry had a clue
about how to deal with the stuff they
were making for years," says Peddle.
"John did the best he could and he
actually did pretty well."

    The industrial solvents drained
from the chip fabrication line into a
250-gallon concrete holding tank.
"They built these double tanks and
they stored it underground. But you
know, we just didn't have the
technology," says Peddle. "Let me just
make a point; John Pavinen was a very
meticulous guy, and he absolutely
designed his tanks the best he could
given the environment at the time."

    In early 1974, a serious disaster
occurred. Technicians monitoring the
tank realized the tank was emptier
than it should have been. During the
cold Pennsylvania winters, the
concrete tank developed a small crack.
"Some of their storage tanks leaked
and it leached into the ground,"
recalls Yannes.

    Pavinen kept the spill quiet, even
from Peddle. "We didn't join him until
the summer of 1974, and they wouldn't
have told us about it anyhow," says
Peddle. "With all due respect, they
keep that stuff a lot quieter in
Silicon Valley. There's been a whole
bunch of stories about breast cancer
being much higher in Silicon Valley,
and there's a bunch of other
anomalies."

    As the Environmental Protection
Agency later determined, the leak was
the source of groundwater
contamination in the area.(7) The
Valley Forge Corporate Center bordered
a residential development that relied
on well water, so there was cause for
concern. Fortunately, water tests at
the time indicated the solvent had not
yet entered the water table. Pavinen
replaced the faulty tank with an
unlined steel tank.

    After the chemical solvents etched
the chips, the technicians inserted
the flecks of silicon and metal into
an easy to handle package. Today,
semiconductor companies typically
place their chips in black plastic
shells with silver pins. Back in 1975,
MOS Technology placed their
microprocessors in distinctive white
ceramic shells with forty gold plated
pins.

    As if a price drop from $300 to
$25 was not radical enough, Peddle and
his team planned to release an
ultra-low cost microprocessor called
the 6507. "Our goal was to do a $5
processor," Peddle states flatly. "The
6507, which was a subset of (the
6502), could be made at a cheaper
price. It was designed to be a really
small package."

    The packaging determined how
cheaply Peddle could sell his chips.
"Packaging costs money and pin outs
cost money," explains Peddle. "Back in
those days, those big 40-pin packages
were very expensive." The 6507
contained only 28 pins.

    In a perfect world, every single
chip would work. If they fabricated
10,000 chips, they would ideally have
10,000 working chips. However,
imperfections snuck in from every
imaginable source. Inconsistencies in
the etching process caused flaws.
Small particles of dust getting in the
way of the mask caused flaws. Even
impurities in the silicon wafer
produced flaws. The number of flaws
the engineers could defeat determined
the chip yield.

    Technicians methodically tested
every single chip to determine if it
worked. In 1975, most chipmakers
considered a 30% yield to be quite
successful. The industry simply
discarded the remaining 70%. The
process was inherently inefficient and
resulted in monumental chip prices. If
Pavinen wanted to achieve low cost
microprocessors, he would have to use
every trick available to raise the
yield.

    In the seventies, most
semiconductor houses tested their
chips with a Fairchild Century system.
The huge machine occupied almost an
entire room and cost almost a million
dollars. As Bill Mensch explains, "We
couldn't afford them at MOS
Technology." Instead, Mensch
constructed a small handheld chip
tester that resembled a computer
motherboard covered in IC chips. Every
single chip from MOS Technology was
hand tested by the homebrew devi