This section is from a paper entitled "Recollections of Early Chip Development at Intel", Andrew M. Volk, Peter A. Stoll, and Paul Metrovich.
It appeared in the Intel Technology Journal - 1st Quarter 2001. It gives a good idea of exactly what was involved in making the first photomasks.
Our thanks to Intel and the authors for this excellent paper. We have used quotations and an image from this paper to tell the story of how the first photomasks were generated.
The first masks were made by transferring the drawings on the Mylar to "rubylith." Rubylith is a two-layered material, which comes in huge sheets. The base layer is heavy transparent dimensionally stable Mylar. A thin film of deep red cellophane-like material covers the base layer. The first chips at Intel used a machine called a "Coordinatograph" to guide cutting of the ruby layer. The coordinates and lengths had to be measured and transferred by hand to the cutter. Later, a Xynetics plotter with knives, instead of pens, was used to cut more quickly and precisely."
"When the cutting was finished, the technicians had to peel away only the desired geometries that made the mask layers. The design engineer and mask designers would spend days hand-checking the rubylith for peeling errors, nicks, and unintended cuts. A final check was made for design rule violations. The rubylith was sent to the mask vendor to be made into masks for fabricating the silicon die."
"Missing a cut or forgetting to peel a geometry would mean a bad part. Ted Jenkins remembers working on the first Intel product, the 3101 64-bit RAM. Actually, the first version was only a 63-bit RAM due to a simple error peeling one layer on the rubylith."
"The rubylith sheets had to be handled very carefully so they were not damaged. Small areas of ruby could be rubbed off. Andrew remembers a call from the 8080A mask vendor saying that they had found a "floater," an unexplained piece of ruby stuck in a random place on the Mylar. They feared that a piece had come off somewhere. A several hour check against the layout found no missing bits and the mask was taken as is. Fortunately, the dice made with that mask were okay."
"Adding or removing transistors and interconnect on rubylith was definitely a manual task, not unlike surgery. In fact, the technician who did the edits used a surgical scalpel and a metal ruler (scale). Adding transistors or interconnect involved cutting and peeling away bits of ruby. Removing objects involved adding ruby-red tape to the back of the heavy Mylar. Cuts had to be precise so as to leave no nicks or cut marks on the Mylar that might show on the mask. Verification was done with the metal scale and a 7X-magnifying eyepiece with a calibrated scale on the bottom."
Rubylith is still manufactured and used to make lithographic images Ulano