1960 · Integrated Circuit Mask Layout · Work 842 of 1,871
A large drawing on a tilted drawing table in an engineering office in Mountain View, a wide sheet of pale paper ruled in a fine green grid, in its middle a symmetrical figure forty centimetres across of rectangles and bars shaded in blue-grey hatching and outlined in red dashed lines, three identical groups of small diamond shapes inside it, eight small empty squares around its edge, the margins of the sheet bare paper, a hand in a white shirt sleeve drawing a red dashed line with a coloured pencil along a plain wooden straightedge, blue and red pencils beside it, an angled desk lamp, cool fluorescent light from above.
What it was used for
Before the mask came the drawing. For Fairchild's first planar integrated circuit, the "F" flip-flop element with 4 transistors and 5 resistors, Lionel Kattner drew the mask design by hand; the original is held by the Computer History Museum as item 102658656. The drawing fixed one mask layer for each production step. The pattern was then photographically reduced to mask size. Robert Noyce had asked Jay Last to begin the work in August 1959; the first working devices were produced on 26 May 1960, and the public introduction as Micrologic followed in March 1961.
What was new
What marks it out is the jump in scale: drawn large, built small, with photographic reduction in between. That made the design an object you could check by hand - by the Computer History Museum's account early circuits were designed by 1 or 2 engineers with slide rules, while later chips with millions of transistors needed design software and whole teams. The cost stayed manual at first: in the later process operators cut the pattern into two-layer Rubylith film on light tables; at Intel, digitising the drawings in the early 1970s was only the first step towards computer-aided design, which later made the cutting unnecessary.
What the sources leave open: Kattner as the draughtsman and the item numbers come from the Computer History Museum's "Semiconductor Artifacts" exhibit text, which contradicts itself on whether the displayed piece is the original or a copy. It does not name the drawing medium used in 1960; Rubylith is documented there only for a later mask (Mostek, 200x). The Rubylith procedure comes from the Intel Museum and describes Intel's practice, not that of 1960.