1988 · East German Megabit DRAM · Work 879 of 1,871
A panelled conference room in Berlin, on a polished dark wood table a small pyramid of clear blue glass about twelve centimetres high set out as a gift, inside it on a small base a ceramic chip two and a half centimetres long with a dark violet-brown body and a plain smooth unmarked gold lid, nine gold legs on each side, across the table and over its edge a long paper strip showing only a fine grid of grey lines, flat and soft in focus, hands in dark suit sleeves with white cuffs, one holding the end of the strip, plain glass water carafes, pale wood panelling, daylight through tall net curtains.
What it was used for
The U61000 was East Germany's 1 Mbit memory, developed at the microelectronics research centre in Dresden. On 12 September 1988 representatives of VEB Carl Zeiss Jena handed Erich Honecker a working specimen; the decision behind it had been taken in February 1986, with the aim of reaching world standard within three years. The reconstruction at richis-lab gives the U61000-1 a die area of 12.8 by 5.1 millimetres, 2.3 million components, 1.2 percent spare cells, about 16 masks and 430 part-steps; exposure used the AÜR stepper at 436 nanometres and 25 wafers per hour, on 5-inch wafers carrying 90 chips.
What was new
Technically there was nothing in the device that did not already exist in the West; it was proof that it could be done at all. What is characteristic is the price of that proof. Yield in pilot production was 9.5 percent, and around the turn of 1989/90 3 percent without the redundancy circuit and 10 percent with it; in all, on the figures collected at richis-lab, about 30,000 devices were made, by the site's own estimate from some 2,500 wafers, while the HNF gives around 35,000 up to 1989. The barb: the world market price of a 1 Mbit DRAM was, by richis-lab, a factor of 100 below the manufacturing cost of the U61000 - and the chip's package had to be imported.
What the sources leave open: The unit figures diverge: richis-lab gives about 30,000 devices in all, the HNF around 35,000 in pilot production up to the fall of the Wall; both are therefore stated side by side. What the title calls a glass pyramid - the mount in which the sample was handed over - is described by none of the sources opened, so the text names only the handover itself. richis-lab gives no feature size for the U61000.