A mask store in a chip factory in Sunnyvale under amber yellow light, on a white work surface a flat case of milky white plastic with strongly rounded corners, twenty centimetres across, opened flat on its hinge, beside the empty lid a square glass plate thirteen centimetres across, almost black and mirror-bright, filled by a large pale circle of a fine even grid of tiny rectangles with a bright cross-shaped group of alignment marks at its centre, two hands in white nylon gloves lifting the plate by its edges, behind them a shelf of identical closed plain white cases standing in a row.
What it was used for
A mask carries the pattern of one single layer of the circuit and is reused at every exposure. The reconstruction of the Dresden megabit memory at richis-lab gives concrete dimensions for its mask set: 125 by 125 millimetres, chrome on glass, 5:1 imaging ratio, written with the ZBA21 electron-beam writer; the device needed, depending on the source, about 16 masks and 6 implantation steps, 430 part-steps in all. A mask is therefore not handled in the open but moved and stored in a case of its own.
What was new
The pellicle - a thin film stretched on a frame above the pattern, holding dust out of the focal plane - came into general use in the early 1980s according to chiphistory.org. Who invented it is unresolved: Chris Mack asks the question in his post and names Ron Hershel's 1981 SPIE paper (volume 275, pages 23 to 28) as the earliest reference he has found. In the reader comments below, one signed Anne Miller calls her the inventor of the JASCO pellicle, and Ray Winn writes that his firm Advanced Semiconductor Products was the first to produce practical pellicles.
What the sources leave open: chiphistory.org names an IBM patent of 1978; a commenter on Mack's post names the same technique in IBM patent 4131363 of 1977. Neither says which kind of date is meant - filing or grant - and the patent itself was not opened here. The mask dimensions come from a 1988 device, not from 1980.